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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
94be68c4-9e19-47b7-b119-1c9194d36728 | music-mood-detection-based-on-audio-and | 1809.07276 | null | http://arxiv.org/abs/1809.07276v1 | http://arxiv.org/pdf/1809.07276v1.pdf | Music Mood Detection Based On Audio And Lyrics With Deep Neural Net | We consider the task of multimodal music mood prediction based on the audio
signal and the lyrics of a track. We reproduce the implementation of
traditional feature engineering based approaches and propose a new model based
on deep learning. We compare the performance of both approaches on a database
containing 18,000 tracks with associated valence and arousal values and show
that our approach outperforms classical models on the arousal detection task,
and that both approaches perform equally on the valence prediction task. We
also compare the a posteriori fusion with fusion of modalities optimized
simultaneously with each unimodal model, and observe a significant improvement
of valence prediction. We release part of our database for comparison purposes. | ['Jimena Royo-Letelier', 'Francesco Piccoli', 'Rémi Delbouys', 'Romain Hennequin', 'Manuel Moussallam'] | 2018-09-19 | null | null | null | null | ['multimodal-emotion-recognition', 'music-emotion-recognition', 'multimodal-emotion-recognition'] | ['computer-vision', 'music', 'speech'] | [ 2.26494744e-02 -2.31858239e-01 -8.71311650e-02 -5.09407341e-01
-1.12450707e+00 -7.74497867e-01 6.28004611e-01 3.25769395e-01
-1.73184380e-01 4.81110811e-01 5.77385306e-01 7.60949016e-01
-2.93807566e-01 -3.41153949e-01 -2.91647702e-01 -6.02616847e-01
-2.09621698e-01 3.70807946e-01 -4.38704908e-01 -3.93351734e-01
4.16783690e-01 1.91074610e-01 -1.97820556e+00 1.00033438e+00
1.19883884e-02 1.70747221e+00 -4.16303307e-01 8.18596184e-01
2.02835694e-01 5.66324234e-01 -5.20770371e-01 -5.87446451e-01
1.46486834e-01 -3.12787116e-01 -5.83702683e-01 -2.69315362e-01
4.87985164e-01 3.64189923e-01 -9.94485244e-02 7.44712710e-01
9.18492138e-01 3.77535820e-01 6.59939408e-01 -1.33658516e+00
4.25547361e-02 6.76544666e-01 -3.54944140e-01 -1.65733889e-01
7.04184294e-01 -3.19572508e-01 1.58697248e+00 -9.00585949e-01
4.80110973e-01 9.02464032e-01 1.04226363e+00 3.01588178e-01
-1.37179828e+00 -5.15486598e-01 -1.59504905e-01 6.03971422e-01
-1.24393284e+00 -6.93931699e-01 1.08399069e+00 -4.99722630e-01
9.56856489e-01 5.11150539e-01 9.23293471e-01 1.06281412e+00
-1.56784505e-01 7.63398111e-01 1.15592670e+00 -3.46123546e-01
1.70213088e-01 1.57265007e-01 1.79862604e-01 2.22396925e-01
-4.35804278e-01 5.22907218e-03 -1.32677364e+00 -3.63602400e-01
-1.21621817e-01 -5.58756709e-01 -1.15067042e-01 -2.19874799e-01
-1.31239688e+00 6.37031078e-01 -2.98661808e-03 1.30400941e-01
-6.71227157e-01 1.03104874e-01 7.57253885e-01 5.05861163e-01
2.69266188e-01 6.02584362e-01 -5.31969488e-01 -6.18470430e-01
-1.24521172e+00 4.38190222e-01 9.46847916e-01 3.93505752e-01
6.11498713e-01 8.47312156e-03 -3.02017033e-01 1.03716457e+00
1.21662788e-01 2.44718432e-01 4.54811156e-01 -1.37048805e+00
-1.54498667e-01 4.09850150e-01 -1.26947183e-02 -8.17009509e-01
-6.71983719e-01 -4.04847652e-01 -5.09308994e-01 1.29619867e-01
1.56908914e-01 -1.45856723e-01 -3.51991266e-01 1.95606744e+00
8.44869241e-02 1.68930367e-01 2.28745937e-02 7.98377752e-01
1.03446603e+00 6.66226983e-01 -1.06930494e-01 -7.08804727e-01
1.22306752e+00 -6.49590373e-01 -1.08533382e+00 2.05974042e-01
3.05757318e-02 -9.57787216e-01 1.00079954e+00 1.25242543e+00
-1.38552439e+00 -6.71956718e-01 -1.21405292e+00 1.58750147e-01
-2.82552749e-01 3.49970490e-01 7.10929036e-01 8.74993443e-01
-9.30684090e-01 9.19658482e-01 -2.56174535e-01 -7.55925998e-02
7.26494566e-02 7.38610923e-01 -2.37492695e-01 7.38300622e-01
-1.17693460e+00 7.36582458e-01 3.04336846e-01 -6.48301840e-02
-7.72963226e-01 -5.35534501e-01 -6.08830392e-01 6.62714019e-02
-3.68034959e-01 -4.17879254e-01 1.31269348e+00 -1.27954197e+00
-1.79838574e+00 9.36034322e-01 -7.17241764e-02 -2.09058866e-01
-4.08450281e-03 -2.79916316e-01 -5.98764241e-01 2.50019412e-02
-4.65923935e-01 8.21689427e-01 7.73941040e-01 -1.07448947e+00
-4.44463402e-01 -1.87729791e-01 -2.15151027e-01 4.86765504e-01
-6.44008458e-01 1.10294364e-01 -2.82193720e-01 -5.26508272e-01
-1.29944518e-01 -1.04210508e+00 1.54869258e-01 -6.08960807e-01
-2.76510835e-01 -1.63911313e-01 1.91623479e-01 -5.08849323e-01
1.51889145e+00 -2.45949674e+00 4.57096279e-01 4.71587628e-01
-2.31331140e-01 -4.14444774e-01 -2.02509895e-01 4.28028107e-01
-4.12440091e-01 -4.34846312e-01 6.90741884e-03 -5.28732181e-01
4.78904188e-01 -8.08782578e-02 -4.19544399e-01 4.34809983e-01
-3.14832479e-01 6.28368556e-01 -2.20219716e-01 -4.16815758e-01
6.72044680e-02 6.19467616e-01 -6.32669210e-01 1.97877824e-01
-4.12155986e-02 1.77234858e-01 2.95771718e-01 7.46497691e-01
2.42279932e-01 2.13754445e-01 4.58967775e-01 -5.74515820e-01
-4.23266478e-02 4.83595729e-01 -1.36178303e+00 1.90072680e+00
-3.25712353e-01 9.03582752e-01 -3.09756286e-02 -6.01858258e-01
1.12761509e+00 7.07453430e-01 8.84349883e-01 -4.23434466e-01
2.52575696e-01 1.46861732e-01 1.70161165e-02 -2.26333886e-01
6.90143645e-01 -6.05549037e-01 -4.74369287e-01 4.90480095e-01
4.50534850e-01 -5.52575588e-02 -2.51901280e-02 -4.65148780e-03
7.37682045e-01 1.48808554e-01 2.69170195e-01 -5.77144995e-02
4.49722707e-01 -3.24213237e-01 4.88820344e-01 4.13769364e-01
-2.14616925e-01 6.65243447e-01 6.96493745e-01 -2.56088346e-01
-7.24765241e-01 -1.03938305e+00 -2.32260704e-01 1.65627885e+00
-5.44282012e-02 -1.09912789e+00 -5.69329917e-01 -2.28276074e-01
-1.51758254e-01 5.87979198e-01 -9.92334545e-01 -1.63646936e-01
-1.38152659e-01 -9.87276495e-01 6.86861098e-01 4.26238894e-01
-2.58981049e-01 -1.25569534e+00 -3.92542690e-01 -2.43182257e-02
-4.78466034e-01 -6.28566265e-01 1.58662438e-01 6.29098237e-01
-6.77324653e-01 -6.17580831e-01 -1.34751365e-01 -5.67716062e-01
-4.46153045e-01 -6.12455308e-01 1.33459079e+00 -3.69610131e-01
-1.23654999e-01 6.99317992e-01 -1.63971901e-01 -5.76214314e-01
-2.60764182e-01 8.66364241e-02 2.72158414e-01 3.30442637e-01
4.88619506e-01 -8.33209932e-01 -4.50079590e-01 -5.31950444e-02
-5.51676452e-01 -2.75265068e-01 1.70192868e-01 7.22872853e-01
7.32333064e-01 -2.34968200e-01 8.46663892e-01 -2.96172738e-01
7.52027452e-01 -4.13976729e-01 -5.49466014e-02 -6.96522221e-02
-6.51676297e-01 -2.16246873e-01 5.08483313e-02 -4.18442249e-01
-7.98752666e-01 7.14243829e-01 -4.63249981e-01 -4.50337857e-01
-2.10928485e-01 4.07999486e-01 -3.16110522e-01 5.64021319e-02
6.70607209e-01 -1.26923203e-01 -2.22372562e-01 -6.53280556e-01
3.58751416e-01 7.70464480e-01 8.62524986e-01 -4.07662719e-01
2.20289633e-01 3.26986760e-01 1.14800073e-01 -5.70678592e-01
-7.94378340e-01 -5.45618713e-01 -5.59227943e-01 -7.78108895e-01
7.51848578e-01 -1.15443873e+00 -1.20758438e+00 1.43157676e-01
-8.61041188e-01 -3.97245362e-02 -3.92700344e-01 6.57957077e-01
-1.19346273e+00 1.94144547e-01 -6.51492596e-01 -9.97775614e-01
-4.39407885e-01 -6.68624640e-01 1.11759591e+00 2.41886415e-02
-8.04977775e-01 -8.01099539e-01 9.72723603e-01 2.52426207e-01
1.09052770e-02 2.18184754e-01 8.73657882e-01 -7.74820328e-01
2.73977220e-01 -3.75456005e-01 3.39648724e-01 3.03145498e-01
-2.54422367e-01 -9.78457704e-02 -1.80673075e+00 1.21637933e-01
-5.75129651e-02 -7.31372654e-01 1.12923610e+00 4.62374896e-01
9.99100208e-01 1.57552883e-01 2.99342781e-01 4.79741067e-01
1.25352764e+00 -1.91206500e-01 6.29468381e-01 4.16659325e-01
3.64633948e-01 7.60531366e-01 8.47921073e-01 9.61227536e-01
2.13207752e-01 8.88278484e-01 6.36620700e-01 1.91781178e-01
1.06965519e-01 1.20577611e-01 7.75701761e-01 9.47279394e-01
-3.35337609e-01 1.02011248e-01 -4.15939331e-01 4.87456888e-01
-2.00216913e+00 -1.23753190e+00 -2.40508988e-01 2.14471745e+00
8.13895464e-01 -1.27683297e-01 6.96393728e-01 6.51704669e-01
4.82016623e-01 8.84367749e-02 -1.09786667e-01 -8.74432862e-01
-4.33250815e-01 4.66615289e-01 -1.74471900e-01 5.11901498e-01
-1.37408090e+00 5.04241109e-01 8.24174595e+00 6.81788862e-01
-9.10780668e-01 7.89052621e-02 2.77689099e-01 -8.35989416e-01
-2.66422868e-01 -3.25770557e-01 -3.78951132e-01 1.69245377e-01
1.24736488e+00 1.24116018e-01 6.89374924e-01 6.56983137e-01
1.06648415e-01 -5.30866124e-02 -1.36713350e+00 1.52769053e+00
4.12019759e-01 -9.97066081e-01 -2.38685608e-01 -1.45566657e-01
7.25172639e-01 -6.64765835e-02 2.68029988e-01 4.35407162e-01
-6.35326445e-01 -9.46880639e-01 8.65189314e-01 9.40638006e-01
4.30073172e-01 -1.15659606e+00 6.24801815e-01 -1.15527928e-01
-1.03972650e+00 -2.57230610e-01 3.13302800e-02 -3.88066381e-01
6.44874573e-02 4.05627698e-01 -3.58938873e-01 4.87394571e-01
7.50342727e-01 9.02491570e-01 -7.34560132e-01 9.91052926e-01
4.13610227e-02 5.70324540e-01 -3.05317104e-01 1.70624480e-01
-2.45265439e-01 3.49607542e-02 7.13356733e-01 1.44252694e+00
3.76219898e-01 -3.19577157e-01 -1.26860052e-01 4.58327919e-01
-7.86437839e-03 5.19883275e-01 -2.63247550e-01 6.62025288e-02
1.41829252e-01 1.74079013e+00 -4.77005720e-01 -4.58612651e-01
-1.41176894e-01 7.99919248e-01 -3.65983993e-02 -8.62642303e-02
-8.17039132e-01 -3.88303846e-01 7.66916752e-01 -3.81477654e-01
2.60790914e-01 3.53123188e-01 -7.93863058e-01 -1.00632358e+00
-2.68617243e-01 -1.05559790e+00 6.66284561e-01 -9.96615112e-01
-1.29857433e+00 5.82931519e-01 -3.31104815e-01 -1.39038837e+00
-4.86490697e-01 -5.36166489e-01 -5.63203216e-01 5.96023798e-01
-1.02796435e+00 -7.60385215e-01 -1.01753339e-01 6.68306291e-01
1.29396379e-01 -3.68573040e-01 1.24652004e+00 6.15148246e-01
-1.62288055e-01 5.42291403e-01 -1.24492839e-01 -3.66538316e-01
1.21371818e+00 -1.49311900e+00 -4.03711885e-01 1.06162643e-02
6.68590128e-01 1.15340874e-01 1.04653013e+00 -1.39214754e-01
-1.28774393e+00 -6.03824914e-01 9.78151023e-01 -7.05657721e-01
4.95554894e-01 -3.51896018e-01 -5.02959013e-01 4.85259652e-01
7.05761135e-01 -7.60868967e-01 1.46250820e+00 9.08012152e-01
-2.46070847e-01 -2.63463616e-01 -7.46463895e-01 1.75193354e-01
4.68663186e-01 -6.97597623e-01 -8.40703905e-01 1.07389897e-01
-1.35256033e-02 -7.30771348e-02 -1.18661308e+00 6.84515774e-01
1.19316256e+00 -1.23780012e+00 8.67444098e-01 -4.72255737e-01
3.33877027e-01 -2.19221771e-01 -7.25112498e-01 -1.15044141e+00
-4.04690087e-01 -6.14492416e-01 -3.82966697e-01 1.10175347e+00
4.58295465e-01 2.55259126e-01 6.09842062e-01 9.83706117e-02
-1.61732540e-01 -5.58283389e-01 -8.39058340e-01 -3.72951359e-01
-3.38337421e-01 -9.41035688e-01 3.23863983e-01 1.00548840e+00
7.29505658e-01 7.24474490e-01 -8.24568272e-01 -2.26118967e-01
3.40628088e-01 5.88144720e-01 6.27174020e-01 -1.54734445e+00
-6.82643116e-01 -6.96131408e-01 -5.90757608e-01 -1.50394931e-01
2.66702116e-01 -8.85846138e-01 1.11473419e-01 -1.11546683e+00
4.49720770e-01 3.52429718e-01 -1.03347826e+00 6.26776218e-01
2.29940623e-01 1.17426944e+00 2.31750980e-01 7.47328531e-03
-9.99255836e-01 5.53501487e-01 4.02564138e-01 -7.96544626e-02
-5.00336766e-01 -6.82977512e-02 -7.05847561e-01 8.55841517e-01
7.80682981e-01 -3.10588092e-01 -2.18739226e-01 3.16469133e-01
9.78910148e-01 1.39455320e-02 2.04133347e-01 -1.18807101e+00
-1.75116286e-01 2.19069615e-01 7.30035007e-01 -7.84427226e-01
1.08285165e+00 -4.32453394e-01 3.31709057e-01 7.10141063e-02
-6.75124049e-01 -4.47953939e-02 3.67655843e-01 3.13518643e-01
-4.97640699e-01 -5.76444007e-02 6.71056688e-01 3.51900309e-01
-5.31118751e-01 -2.75680661e-01 -4.86036032e-01 -4.35626715e-01
5.91012836e-01 -4.99032624e-02 2.36171782e-01 -8.50854874e-01
-1.54232788e+00 -3.64344805e-01 3.01262856e-01 5.39450943e-01
3.97851676e-01 -1.75373948e+00 -6.01505578e-01 1.21248439e-01
2.70340860e-01 -1.29154050e+00 4.24329251e-01 1.29621112e+00
2.39561051e-01 1.18571959e-01 -6.16225481e-01 -5.47801375e-01
-1.72934675e+00 5.07899344e-01 3.90563488e-01 -4.63773832e-02
7.94503838e-02 5.48127949e-01 6.72957525e-02 -2.11178720e-01
4.42683160e-01 2.75979161e-01 -6.94860578e-01 8.68184268e-01
5.66643715e-01 4.25472260e-01 3.10258120e-01 -8.04982960e-01
-4.78958488e-01 4.55977708e-01 5.76556861e-01 -6.20660484e-01
1.51789570e+00 -9.04736891e-02 -3.01781535e-01 1.34795678e+00
1.01321471e+00 2.00612843e-01 -6.90200269e-01 1.86991230e-01
-9.42918956e-02 -1.11790992e-01 2.26290107e-01 -1.21349180e+00
-7.89620996e-01 9.47584510e-01 1.08064032e+00 2.53797323e-01
1.59283876e+00 -2.81934887e-02 2.98243999e-01 4.38483179e-01
-2.20145062e-01 -1.56953907e+00 1.34813756e-01 4.38433319e-01
6.77408814e-01 -9.43473220e-01 1.70572266e-01 1.69044673e-01
-9.16314483e-01 1.22206020e+00 2.09941119e-01 -1.44283473e-01
6.53944492e-01 1.87526390e-01 2.21339017e-01 -2.94773877e-01
-1.33060181e+00 -2.90923506e-01 8.18350673e-01 3.27472627e-01
8.77685428e-01 1.38342917e-01 -4.44607019e-01 1.27253234e+00
-4.67905790e-01 -2.07971230e-01 2.91519105e-01 4.89649743e-01
-4.03895110e-01 -1.13233733e+00 -5.41428804e-01 2.35144362e-01
-7.29010224e-01 -2.71320269e-02 -9.90893900e-01 3.29071939e-01
3.35569173e-01 9.14476335e-01 8.44934024e-03 -8.99350286e-01
1.97814152e-01 7.28929579e-01 7.71225810e-01 -2.57158130e-01
-1.12490106e+00 6.74464405e-01 4.99423802e-01 -5.62746286e-01
-7.58795381e-01 -1.04834056e+00 -9.47051346e-01 -7.06270486e-02
2.01197597e-03 1.57567471e-01 9.38880920e-01 6.88829541e-01
1.56403944e-01 4.46880281e-01 9.77396011e-01 -1.19670415e+00
-1.55486733e-01 -1.04849148e+00 -9.65999544e-01 4.83810097e-01
1.12847887e-01 -6.39139473e-01 -1.54859975e-01 2.36188143e-01] | [15.84627628326416, 5.171200752258301] |
6218b855-7bd7-4be4-9efc-80746911b4ee | bop-challenge-2022-on-detection-segmentation | 2302.13075 | null | https://arxiv.org/abs/2302.13075v1 | https://arxiv.org/pdf/2302.13075v1.pdf | BOP Challenge 2022 on Detection, Segmentation and Pose Estimation of Specific Rigid Objects | We present the evaluation methodology, datasets and results of the BOP Challenge 2022, the fourth in a series of public competitions organized with the goal to capture the status quo in the field of 6D object pose estimation from an RGB/RGB-D image. In 2022, we witnessed another significant improvement in the pose estimation accuracy -- the state of the art, which was 56.9 AR$_C$ in 2019 (Vidal et al.) and 69.8 AR$_C$ in 2020 (CosyPose), moved to new heights of 83.7 AR$_C$ (GDRNPP). Out of 49 pose estimation methods evaluated since 2019, the top 18 are from 2022. Methods based on point pair features, which were introduced in 2010 and achieved competitive results even in 2020, are now clearly outperformed by deep learning methods. The synthetic-to-real domain gap was again significantly reduced, with 82.7 AR$_C$ achieved by GDRNPP trained only on synthetic images from BlenderProc. The fastest variant of GDRNPP reached 80.5 AR$_C$ with an average time per image of 0.23s. Since most of the recent methods for 6D object pose estimation begin by detecting/segmenting objects, we also started evaluating 2D object detection and segmentation performance based on the COCO metrics. Compared to the Mask R-CNN results from CosyPose in 2020, detection improved from 60.3 to 77.3 AP$_C$ and segmentation from 40.5 to 58.7 AP$_C$. The online evaluation system stays open and is available at: \href{http://bop.felk.cvut.cz/}{bop.felk.cvut.cz}. | ['Jiri Matas', 'Carsten Rother', 'Bertram Drost', 'Eric Brachmann', 'Gu Wang', 'Yann Labbe', 'Tomas Hodan', 'Martin Sundermeyer'] | 2023-02-25 | null | null | null | null | ['2d-object-detection', '6d-pose-estimation'] | ['computer-vision', 'computer-vision'] | [-1.75764501e-01 1.75957292e-01 2.63996631e-01 3.14940065e-02
-1.25278723e+00 -6.72930181e-01 4.73665029e-01 -9.51550603e-02
-6.66580200e-01 5.93318582e-01 -4.08367723e-01 2.08964609e-02
2.16880534e-02 -6.71965718e-01 -1.09711218e+00 -4.77695912e-01
-1.57806426e-01 9.11041558e-01 3.20098549e-01 -9.48601365e-02
2.21103862e-01 8.30273926e-01 -1.49573350e+00 -1.43981069e-01
3.22900087e-01 1.61659837e+00 8.75839218e-02 8.03124905e-01
8.28216448e-02 8.32380429e-02 -7.28230119e-01 -5.11419892e-01
6.97132111e-01 2.15941742e-02 -6.60763860e-01 -1.56687886e-01
8.27687144e-01 -2.93763578e-01 -1.73068002e-01 7.38531590e-01
7.68169641e-01 -1.68976575e-01 6.62038743e-01 -1.25984824e+00
-8.56295526e-02 1.07003443e-01 -9.63866770e-01 7.55966306e-02
5.59023559e-01 2.67426878e-01 9.55808699e-01 -1.14941251e+00
7.66819119e-01 1.04890931e+00 8.83896053e-01 5.49531579e-01
-1.08501327e+00 -9.93065655e-01 -2.15041880e-02 3.20668072e-02
-1.60336018e+00 -1.47452772e-01 5.49255550e-01 -5.74025393e-01
1.06558847e+00 3.12995762e-01 8.04080546e-01 8.30253422e-01
-2.01737732e-01 9.83072221e-01 1.09478748e+00 -2.40231812e-01
-8.26737285e-02 -1.73309490e-01 -2.89344609e-01 7.38238275e-01
2.03670204e-01 2.03993335e-01 -5.15713215e-01 3.69844377e-01
1.04215229e+00 -5.53930521e-01 5.20278513e-03 -5.05452931e-01
-1.20552814e+00 6.11280322e-01 7.55902469e-01 -3.56242359e-02
-1.88241705e-01 3.63976866e-01 3.51310343e-01 3.23650637e-03
4.65010464e-01 6.67750478e-01 -6.67668104e-01 -2.54774988e-01
-1.00868297e+00 6.32131279e-01 4.40896630e-01 1.15390027e+00
5.13518214e-01 -1.45111710e-01 1.94722518e-01 7.26354420e-01
3.58608365e-01 8.46389353e-01 7.82846957e-02 -1.24884677e+00
5.37425399e-01 4.76856083e-01 4.08070356e-01 -9.11246836e-01
-7.25829184e-01 -6.11375928e-01 -5.09501278e-01 4.24360454e-01
6.27699375e-01 3.51731316e-03 -1.17343867e+00 1.25106287e+00
3.43821913e-01 -3.89046937e-01 -3.80303711e-01 9.28289652e-01
1.11867261e+00 5.58047950e-01 -1.96660042e-01 1.50487497e-01
1.19928038e+00 -8.43658745e-01 -1.80259645e-01 -2.37912446e-01
3.70101959e-01 -9.75425303e-01 8.63486826e-01 7.06017792e-01
-1.45812559e+00 -6.77942932e-01 -1.08312035e+00 3.40419337e-02
-3.13909292e-01 1.80490270e-01 5.52618682e-01 5.92555225e-01
-1.03791904e+00 4.43821877e-01 -7.47411132e-01 -3.13174039e-01
7.89684594e-01 5.13270140e-01 -3.03536803e-01 -9.70172957e-02
-8.75923753e-01 1.01632690e+00 2.41423115e-01 1.50950447e-01
-9.71190929e-01 -9.11335409e-01 -6.27719820e-01 -4.58069801e-01
4.65237588e-01 -3.30722541e-01 1.47270548e+00 -3.30496788e-01
-1.29479802e+00 1.48030412e+00 4.00305957e-01 -5.80445766e-01
1.13259125e+00 -5.29495537e-01 -2.36716405e-01 3.71413052e-01
3.25016707e-01 1.29423714e+00 4.37285334e-01 -1.33337498e+00
-9.15916502e-01 -5.20700276e-01 6.70254007e-02 2.74499416e-01
4.52045858e-01 3.13157104e-02 -9.07778561e-01 -4.66664046e-01
3.51473123e-01 -1.12775064e+00 -2.14626864e-01 3.22873116e-01
-4.51442271e-01 -2.01310620e-01 7.08658755e-01 -6.85880721e-01
4.55580413e-01 -1.90161359e+00 -1.83972027e-02 -3.37368920e-02
2.44359881e-01 3.35950553e-01 2.44472530e-02 1.29614279e-01
-4.32383753e-02 1.26407156e-02 -2.84138739e-01 -6.21656239e-01
2.00204775e-01 -1.92767620e-01 -3.07064690e-02 7.09709406e-01
6.48964345e-02 1.00906205e+00 -7.22157359e-01 -5.34630060e-01
6.30028546e-01 3.76593709e-01 -2.07518116e-01 -7.06233904e-02
-1.87003747e-01 5.81379056e-01 -1.62172660e-01 9.66066062e-01
1.10416734e+00 -4.83484603e-02 -4.63144779e-01 -4.17602181e-01
-2.40937129e-01 -5.36267925e-03 -1.26450968e+00 1.92140090e+00
-3.36276054e-01 7.63983488e-01 2.37785801e-01 -9.48374987e-01
1.00226665e+00 6.10905439e-02 1.05783486e+00 -7.19106853e-01
1.91389054e-01 4.47481304e-01 -1.65392324e-01 -9.94995311e-02
4.69848543e-01 -9.05179158e-02 -3.10257107e-01 -1.34878963e-01
1.03810251e-01 -9.77604628e-01 2.73501247e-01 5.01645729e-03
8.92448962e-01 5.53617299e-01 -1.30653912e-02 -1.55795813e-02
2.69685298e-01 3.03831935e-01 3.62790555e-01 7.32844710e-01
-3.64894539e-01 1.16862881e+00 2.33721644e-01 -3.97997141e-01
-9.45258081e-01 -1.43583798e+00 -4.12386060e-01 4.93443459e-01
3.45394850e-01 -2.69831926e-01 -7.69778311e-01 -5.92832386e-01
1.02856852e-01 4.56199050e-01 -5.52114904e-01 1.93281144e-01
-7.39567995e-01 -5.91007054e-01 5.92520773e-01 7.69300818e-01
8.44072878e-01 -1.00343442e+00 -8.79588723e-01 1.74638405e-01
-1.38231441e-01 -1.32137954e+00 -1.04603544e-02 1.16359137e-01
-7.39234805e-01 -1.10344708e+00 -1.21098876e+00 -3.86086375e-01
3.03562164e-01 -1.84504360e-01 1.30210662e+00 -4.47516620e-01
-7.82144904e-01 4.54795420e-01 -3.60189080e-01 -8.69204223e-01
2.29642600e-01 6.48013949e-02 -3.09437942e-02 -5.07803082e-01
1.65302575e-01 -1.43575177e-01 -9.80799198e-01 6.53910458e-01
-5.08855283e-01 -1.23787753e-01 6.54918849e-01 3.53570938e-01
8.48925829e-01 -4.35442477e-01 1.31668672e-01 -2.99164504e-01
-2.62568384e-01 1.36752129e-01 -8.42583716e-01 -2.60738283e-01
-4.91258204e-01 -4.82051641e-01 1.83060858e-02 -7.64190555e-02
-6.57973647e-01 3.15809757e-01 -5.33224285e-01 -6.14757061e-01
-2.88089365e-01 -2.13574506e-02 2.44287997e-02 -2.15104505e-01
7.31714070e-01 -1.30909413e-01 -1.49813429e-01 -6.80657566e-01
3.48491967e-01 3.63046944e-01 6.75547838e-01 -4.31508124e-01
8.51684153e-01 7.23619759e-01 1.09773248e-01 -6.01642966e-01
-9.01711226e-01 -5.01742542e-01 -6.70748651e-01 -3.70006472e-01
1.06621039e+00 -8.99402797e-01 -8.00427616e-01 7.29480088e-01
-1.13856995e+00 -3.53073567e-01 -6.19591653e-01 5.57014644e-01
-7.56187677e-01 5.01663843e-03 -3.76658827e-01 -7.05118477e-01
-2.01801032e-01 -1.33491981e+00 1.44609892e+00 9.28260610e-02
-1.33051097e-01 -4.63378489e-01 -4.23384793e-02 6.62432969e-01
1.62285149e-01 8.35722685e-01 6.52489439e-02 -2.66938537e-01
-4.90587205e-01 -6.57184422e-01 -5.81133604e-01 3.81763071e-01
-2.20305786e-01 -6.26085252e-02 -1.00711811e+00 -1.69505820e-01
-2.41612718e-01 -2.89570630e-01 6.57979131e-01 6.81753814e-01
1.16485858e+00 3.04133266e-01 -5.09054303e-01 5.56995571e-01
1.35356104e+00 3.21531236e-01 6.92390382e-01 4.98567045e-01
6.67696595e-01 3.80982339e-01 8.94654870e-01 3.82845730e-01
1.25303283e-01 1.06814885e+00 9.27772880e-01 -8.18359926e-02
-4.15470064e-01 -1.76414907e-01 5.02599701e-02 3.45701277e-01
-5.00047684e-01 5.52152507e-02 -1.23478854e+00 4.35230255e-01
-1.31537294e+00 -4.58516359e-01 -4.34286147e-01 2.10599637e+00
4.45533693e-01 5.39490402e-01 4.00575101e-01 1.67700365e-01
5.90821803e-01 6.17691837e-02 -5.68662345e-01 -9.71822217e-02
-2.15010829e-02 5.58951676e-01 8.75666618e-01 3.44543219e-01
-1.28549838e+00 9.07603800e-01 5.54412937e+00 1.00028396e+00
-1.01626563e+00 7.02782869e-02 7.31973231e-01 -4.50495005e-01
4.37703162e-01 -3.18382680e-01 -9.75565970e-01 4.17306691e-01
6.80919826e-01 3.07050079e-01 -1.04031749e-01 9.72691357e-01
-2.10149989e-01 -3.99548084e-01 -9.96874511e-01 1.42997837e+00
9.55458432e-02 -1.33049083e+00 -5.08795381e-01 5.83710484e-02
6.27764225e-01 5.14702618e-01 2.03613490e-01 4.12670761e-01
3.94272096e-02 -1.21822131e+00 1.14422107e+00 3.90119404e-01
1.10611761e+00 -7.79588759e-01 6.73133492e-01 2.00358510e-01
-1.29842246e+00 1.98052466e-01 -2.11976707e-01 1.46011353e-01
4.05447394e-01 5.39802492e-01 -9.44677651e-01 7.02722549e-01
1.30699384e+00 4.99425560e-01 -6.40512943e-01 1.15729070e+00
-2.47108296e-01 1.64918214e-01 -6.38482094e-01 -6.28237752e-03
2.49159351e-01 8.06519538e-02 7.23685980e-01 1.24075997e+00
2.99288750e-01 4.12882306e-02 -1.16205238e-01 6.42938912e-01
-1.26692817e-01 -1.67731509e-01 -2.52552330e-01 2.75561899e-01
2.31160998e-01 1.26093388e+00 -1.00069332e+00 -3.06871403e-02
1.98126659e-02 9.16872621e-01 -1.18597753e-01 -1.92774553e-02
-1.31461465e+00 -4.83834296e-01 3.78084451e-01 4.76655900e-01
5.46774924e-01 -3.01760942e-01 -3.30083191e-01 -7.29415953e-01
1.27369747e-01 -5.07799685e-01 2.21316189e-01 -1.02692735e+00
-1.07636416e+00 5.73997796e-01 3.74103785e-01 -1.29380596e+00
2.30418034e-02 -1.14795280e+00 3.01162549e-03 7.06791759e-01
-1.10646880e+00 -1.22063529e+00 -4.09073412e-01 1.23501703e-01
5.62245607e-01 1.03682533e-01 6.10537946e-01 4.06963825e-01
-3.59679490e-01 7.19233215e-01 2.55783554e-02 2.61742353e-01
4.92356062e-01 -1.24856794e+00 5.56222498e-01 3.10037047e-01
1.76550552e-01 -9.73223671e-02 6.72668636e-01 -4.31200415e-01
-1.11891079e+00 -8.50506246e-01 5.17344713e-01 -8.53150129e-01
4.08821523e-01 -4.91921693e-01 -2.91660190e-01 5.38526058e-01
-8.79479572e-02 3.68410468e-01 1.47387356e-01 -3.41301382e-01
-2.24971190e-01 -2.30494842e-01 -1.47850013e+00 3.31324279e-01
1.20702720e+00 -1.02689326e-01 -2.44703636e-01 4.98563826e-01
6.68178856e-01 -1.05324686e+00 -9.37389016e-01 8.29360664e-01
6.71140015e-01 -1.08490372e+00 1.34648776e+00 -2.62007047e-03
3.45895320e-01 -2.85471708e-01 -3.73690486e-01 -7.87019849e-01
1.20220669e-01 -4.86337513e-01 -1.36054635e-01 8.83710563e-01
4.63090390e-01 -4.43839371e-01 1.24496925e+00 3.63461941e-01
-3.58584195e-01 -9.97609913e-01 -1.31450868e+00 -1.02076793e+00
3.12402099e-01 -8.30222428e-01 2.35710010e-01 4.54106390e-01
-7.64279962e-01 -6.31240383e-02 2.09985301e-01 -3.90141681e-02
7.00732768e-01 6.37808815e-02 8.67604434e-01 -1.07321548e+00
-2.43257105e-01 -6.57083392e-01 -6.95322216e-01 -1.19757867e+00
-4.07073647e-01 -6.39910221e-01 2.24976353e-02 -1.75979304e+00
-1.70554653e-01 -5.87560534e-01 -6.57737628e-02 2.38577992e-01
3.22368294e-01 1.00809395e+00 4.03533578e-01 2.41179615e-01
-4.72442746e-01 1.27668038e-01 1.27242875e+00 -5.09550348e-02
1.01213753e-02 1.22516349e-01 -3.03026110e-01 9.50726271e-01
8.50819826e-01 -2.58707255e-01 1.87330216e-01 -3.20926815e-01
1.20406248e-01 -1.85201690e-02 6.79493785e-01 -1.30994380e+00
-1.68818414e-01 2.30819196e-01 7.31101394e-01 -1.23501277e+00
9.93073404e-01 -7.06814349e-01 1.64260477e-01 5.77916265e-01
2.05163881e-01 6.54462725e-02 5.27068257e-01 1.94572181e-01
-6.53833896e-02 -1.29928485e-01 9.23114359e-01 -4.35880482e-01
-8.20355713e-01 3.78078222e-01 1.03387982e-01 1.67179689e-01
1.32621038e+00 -6.68461621e-01 -5.35908788e-02 -2.35348195e-01
-1.09825432e+00 1.71790004e-01 3.04072082e-01 2.35782519e-01
5.24874270e-01 -1.22092426e+00 -7.18442380e-01 -1.86213091e-01
6.35734499e-02 5.81617713e-01 2.18752831e-01 9.83763039e-01
-9.85353827e-01 5.45409322e-01 -1.24274030e-01 -9.20778811e-01
-1.02113962e+00 1.89169288e-01 4.42656338e-01 -2.88024127e-01
-4.85132307e-01 1.48799777e+00 -2.99416259e-02 -5.18249452e-01
3.74240041e-01 -2.91941702e-01 1.94517329e-01 3.17137331e-01
1.63015258e-02 6.79617763e-01 4.33495343e-01 -7.02116251e-01
-7.23294854e-01 1.01217186e+00 5.44455387e-02 -4.95907754e-01
1.40767682e+00 2.12878942e-01 2.38256916e-01 3.00545275e-01
1.38508737e+00 -1.06407024e-01 -1.54372263e+00 2.64217824e-01
-1.18391916e-01 -5.35932004e-01 -1.08621523e-01 -1.12734401e+00
-1.24579227e+00 9.39237237e-01 1.08446932e+00 2.17876378e-02
9.00936425e-01 5.15395820e-01 7.16392875e-01 -6.08852059e-02
5.48641443e-01 -1.10323536e+00 2.47379854e-01 3.56735200e-01
1.21080279e+00 -1.31931317e+00 2.94293731e-01 -4.98471618e-01
-4.29646045e-01 8.37900102e-01 8.12717259e-01 -4.33322787e-01
5.52874506e-01 2.18266830e-01 -6.67466745e-02 -4.40141827e-01
2.09662303e-01 -1.37582868e-01 4.05530304e-01 6.32350683e-01
2.14541793e-01 5.96657395e-02 -1.13765970e-01 2.40281641e-01
-4.93313015e-01 -2.96479613e-01 5.14214486e-02 9.46432412e-01
-2.17066333e-01 -8.10280919e-01 -5.62077820e-01 4.23852354e-01
-2.71674037e-01 2.89757788e-01 -4.38106328e-01 1.25623488e+00
4.91931528e-01 6.12971723e-01 3.51513594e-01 -2.99316168e-01
7.39701629e-01 -1.84204057e-01 8.75293076e-01 -6.20892346e-01
-4.41639066e-01 1.49635017e-01 1.29003540e-01 -9.25372183e-01
-4.55508590e-01 -9.04519141e-01 -1.34241104e+00 -7.62678832e-02
-4.06316549e-01 -6.97858408e-02 1.04393399e+00 4.63338882e-01
-1.29404655e-02 3.88641387e-01 3.31836879e-01 -1.40553093e+00
-3.34933251e-01 -8.56941581e-01 -4.50455248e-01 1.17233498e-02
1.64350882e-01 -9.20012712e-01 -3.67378980e-01 -1.84428811e-01] | [7.626652717590332, -2.666656255722046] |
a8394675-fec6-491d-9c8e-18de2e301267 | online-temporal-calibration-for-monocular | 1808.00692 | null | http://arxiv.org/abs/1808.00692v1 | http://arxiv.org/pdf/1808.00692v1.pdf | Online Temporal Calibration for Monocular Visual-Inertial Systems | Accurate state estimation is a fundamental module for various intelligent
applications, such as robot navigation, autonomous driving, virtual and
augmented reality. Visual and inertial fusion is a popular technology for 6-DOF
state estimation in recent years. Time instants at which different sensors'
measurements are recorded are of crucial importance to the system's robustness
and accuracy. In practice, timestamps of each sensor typically suffer from
triggering and transmission delays, leading to temporal misalignment (time
offsets) among different sensors. Such temporal offset dramatically influences
the performance of sensor fusion. To this end, we propose an online approach
for calibrating temporal offset between visual and inertial measurements. Our
approach achieves temporal offset calibration by jointly optimizing time
offset, camera and IMU states, as well as feature locations in a SLAM system.
Furthermore, the approach is a general model, which can be easily employed in
several feature-based optimization frameworks. Simulation and experimental
results demonstrate the high accuracy of our calibration approach even compared
with other state-of-art offline tools. The VIO comparison against other methods
proves that the online temporal calibration significantly benefits
visual-inertial systems. The source code of temporal calibration is integrated
into our public project, VINS-Mono. | ['Tong Qin', 'Shaojie Shen'] | 2018-08-02 | null | null | null | null | ['time-offset-calibration'] | ['miscellaneous'] | [-1.03752151e-01 -5.33558190e-01 -1.94005996e-01 -3.15735966e-01
-4.32465345e-01 -6.12756371e-01 6.02430940e-01 2.65623271e-01
-5.78727961e-01 7.22991288e-01 -2.66375691e-01 -2.22925738e-01
-7.13626519e-02 -3.55540693e-01 -8.49564016e-01 -5.49485862e-01
-8.28940868e-02 2.89245754e-01 3.05514127e-01 -2.55393803e-01
3.95866275e-01 6.96943045e-01 -1.50343680e+00 -9.60861981e-01
7.38810956e-01 1.16522443e+00 3.89950693e-01 7.46033967e-01
2.12918490e-01 2.67203242e-01 -5.33187747e-01 5.99207319e-02
3.01498413e-01 1.75055951e-01 -2.37595998e-02 -1.42964527e-01
2.62621850e-01 -3.48952860e-01 -3.98570269e-01 1.20977008e+00
4.14238065e-01 1.45553902e-01 8.32763165e-02 -1.61286962e+00
4.90100719e-02 -1.31854555e-02 -7.10673749e-01 4.14250866e-02
5.56749642e-01 5.15601598e-02 2.95194119e-01 -7.64499724e-01
6.18349671e-01 8.44129801e-01 8.11097443e-01 -6.19696192e-02
-8.09229434e-01 -7.55984902e-01 1.61381751e-01 3.39989036e-01
-1.60080981e+00 -6.69809043e-01 6.36557817e-01 -5.35945177e-01
7.46831238e-01 1.86744690e-01 7.23967493e-01 7.30284810e-01
5.58360875e-01 3.53017151e-01 7.67154396e-01 -1.02016784e-01
1.77643612e-01 9.57159922e-02 1.68116838e-02 4.80527848e-01
5.84643364e-01 3.68155837e-01 -8.11878026e-01 -8.69026780e-02
7.09547997e-01 2.83205181e-01 -2.51268446e-01 -7.81121194e-01
-1.68103850e+00 3.48478705e-01 3.74588311e-01 -6.61740750e-02
-2.98325390e-01 4.14949059e-01 3.32408607e-01 2.72182912e-01
1.02650084e-01 1.26510020e-02 -2.90430754e-01 -5.69579482e-01
-6.51413918e-01 6.73615709e-02 4.84838277e-01 1.32896268e+00
8.08117986e-01 1.98359996e-01 3.75860244e-01 1.79405540e-01
5.27401388e-01 1.04114556e+00 5.29497504e-01 -7.01376498e-01
6.14592910e-01 3.56709957e-01 5.71453512e-01 -1.17797935e+00
-6.40902996e-01 -2.34773785e-01 -8.42472494e-01 1.26107499e-01
1.39322653e-01 -1.97465792e-01 -5.44217706e-01 1.55947292e+00
7.35733986e-01 6.14000797e-01 1.54684246e-01 1.04575467e+00
3.07612062e-01 3.65946352e-01 -4.72739518e-01 -4.38492268e-01
1.14246750e+00 -5.61161101e-01 -1.14831567e+00 -6.07481658e-01
3.53972048e-01 -1.13478374e+00 3.40513140e-01 1.48384094e-01
-5.58371365e-01 -6.00876033e-01 -1.60395277e+00 9.01430026e-02
-3.30106676e-01 2.24960729e-01 5.62116623e-01 4.96007830e-01
-8.40992630e-01 3.84788752e-01 -1.17245650e+00 -5.52821517e-01
-5.44452071e-01 4.21330184e-01 -3.81131887e-01 2.01392040e-01
-1.17546773e+00 1.19627512e+00 3.27477574e-01 5.06995201e-01
-4.37099725e-01 -2.90467739e-01 -1.02909184e+00 -4.58438009e-01
5.13807297e-01 -6.33271158e-01 1.42504060e+00 -3.00291151e-01
-1.70172346e+00 3.42549354e-01 -6.14921629e-01 -6.96985126e-01
5.64077795e-01 -4.94464129e-01 -6.45967841e-01 -3.17597419e-01
1.11175589e-01 7.10894614e-02 1.01355338e+00 -1.16817057e+00
-7.79367208e-01 -4.60645884e-01 -1.39484361e-01 4.02084529e-01
2.37867907e-01 -5.70796251e-01 -6.60354018e-01 -1.73771054e-01
7.67783642e-01 -1.19892728e+00 -5.39182544e-01 -1.20574608e-01
-2.35467046e-01 5.17342567e-01 8.28377485e-01 -2.72267371e-01
1.14583457e+00 -2.17393661e+00 4.72887680e-02 2.77209550e-01
-5.67530170e-02 -5.29630445e-02 4.59535152e-01 5.42023659e-01
2.24626020e-01 -5.66973150e-01 2.34704942e-01 -5.99061549e-01
-1.41810318e-02 3.24783474e-01 -4.96388406e-01 1.02649331e+00
-3.67380738e-01 5.43274939e-01 -1.01372588e+00 -1.53397784e-01
8.83002043e-01 5.03944337e-01 4.20322977e-02 5.73895406e-03
3.03504050e-01 7.62813449e-01 -3.42579842e-01 6.01760983e-01
7.11946249e-01 2.16416214e-02 6.16504140e-02 -4.37877357e-01
-5.79626679e-01 2.03276560e-01 -1.58583081e+00 1.97733748e+00
-4.87277538e-01 7.24383533e-01 9.12353173e-02 -5.12439311e-01
8.93043458e-01 1.91634461e-01 5.09543300e-01 -7.33118474e-01
4.62006241e-01 3.93193781e-01 -3.57961595e-01 -2.44915187e-01
1.30240834e+00 3.61472428e-01 -3.31751823e-01 2.82467697e-02
-3.05958480e-01 -3.65943104e-01 1.11963108e-01 1.53890803e-01
7.85411358e-01 2.54524887e-01 7.36759186e-01 2.08853409e-01
5.55642545e-01 -1.93164572e-02 7.76026249e-01 5.55462480e-01
-3.88051003e-01 3.00966680e-01 -1.26302689e-01 -6.91886693e-02
-9.86916006e-01 -8.41979325e-01 -6.03030510e-02 2.40900770e-01
1.00365770e+00 -4.68172342e-01 -1.27225712e-01 9.31292176e-02
4.12273794e-01 2.27816507e-01 -2.42161393e-01 -2.13395864e-01
-5.41183889e-01 -3.03167313e-01 2.30292410e-01 4.71374869e-01
3.38897496e-01 -6.67538075e-03 -9.72892284e-01 4.65619028e-01
-1.41963184e-01 -1.44776332e+00 -4.41526473e-01 1.14024110e-01
-1.03742218e+00 -9.89585757e-01 -2.85456240e-01 -1.86947197e-01
6.30548894e-01 9.06948090e-01 5.79378009e-01 -1.48772717e-01
2.91782822e-02 6.21359050e-01 -2.18042836e-01 -4.99659866e-01
-2.09931517e-04 -2.25420058e-01 6.88823283e-01 5.50229810e-02
6.24010228e-02 -4.26207542e-01 -5.22813082e-01 5.63307583e-01
-5.53265810e-01 7.62838796e-02 4.07160848e-01 5.69651306e-01
7.49683082e-01 -3.82061839e-01 1.35810468e-02 -3.25197995e-01
3.64227593e-01 -4.15943444e-01 -1.44517422e+00 -2.92445421e-02
-8.07063341e-01 5.81483766e-02 3.67139310e-01 -3.25379342e-01
-7.35570252e-01 5.33881366e-01 2.17242405e-01 -4.69711363e-01
2.46644273e-01 6.69082761e-01 6.88260272e-02 -4.23916638e-01
3.35815579e-01 1.52347103e-01 1.71114534e-01 -2.21731454e-01
3.01103115e-01 6.14919960e-01 1.01879466e+00 -2.55364329e-01
1.04569173e+00 7.36963511e-01 3.29084069e-01 -7.62060821e-01
-2.67999232e-01 -9.57734227e-01 -5.28676510e-01 -4.48120713e-01
3.03415477e-01 -1.25755394e+00 -9.74465370e-01 5.76734364e-01
-1.21275282e+00 2.66899735e-01 1.20778963e-01 8.82028162e-01
-4.71608281e-01 5.72038054e-01 -1.13117233e-01 -9.53325689e-01
-1.57607421e-02 -1.43066192e+00 1.21773863e+00 5.32241642e-01
-2.07517836e-02 -8.71757925e-01 3.10606837e-01 -3.57158780e-02
2.41185710e-01 4.38275546e-01 -3.61253023e-01 7.71634728e-02
-8.80151749e-01 -5.59882879e-01 -1.09174578e-02 -2.40781695e-01
3.78171839e-02 1.01789199e-01 -7.36979723e-01 -6.24235272e-01
8.05698708e-02 3.13755840e-01 3.24654132e-01 2.98074245e-01
1.71618387e-01 5.57653792e-02 -7.32473314e-01 8.15804303e-01
1.69301641e+00 3.83911699e-01 4.00998652e-01 6.49046957e-01
7.19718874e-01 -1.46959508e-02 1.38356411e+00 6.48639321e-01
6.33878648e-01 1.00174820e+00 7.90762722e-01 1.76585376e-01
3.67572963e-01 -1.16575085e-01 4.97694522e-01 1.06293666e+00
7.14447163e-03 1.15259804e-01 -7.26656854e-01 5.35435975e-01
-2.23954582e+00 -5.48262239e-01 -3.44458789e-01 2.73070979e+00
3.13719034e-01 1.13284416e-01 -4.82824504e-01 1.34513870e-01
8.37039053e-01 3.38536531e-01 -7.57429063e-01 -9.61431712e-02
1.55462980e-01 -4.32436496e-01 1.37889314e+00 6.71319485e-01
-9.78807449e-01 7.44439840e-01 5.99082613e+00 1.71394095e-01
-1.39811993e+00 1.08619548e-01 -4.50977802e-01 -5.94640262e-02
1.52006567e-01 3.65042269e-01 -1.09052122e+00 5.51193237e-01
1.12345338e+00 -5.22904575e-01 4.18771893e-01 8.38249087e-01
3.67068499e-01 -5.98689139e-01 -1.02084029e+00 1.34235835e+00
2.36212481e-02 -1.14913762e+00 -7.62712002e-01 1.38613492e-01
6.50795460e-01 3.55904847e-01 -6.14199750e-02 -4.68815602e-02
1.48813263e-01 -2.75096178e-01 1.07190597e+00 4.28379744e-01
6.84606671e-01 -6.98078990e-01 8.15028846e-01 2.99393892e-01
-1.68103206e+00 1.43672928e-01 -3.37114066e-01 -3.14083070e-01
7.29233742e-01 8.17166924e-01 -9.58983123e-01 1.06228280e+00
4.80412781e-01 8.46059978e-01 -4.53259856e-01 1.23596931e+00
-2.72845566e-01 -6.43678084e-02 -6.61392808e-01 -2.85728667e-02
-8.46698061e-02 -3.78720254e-01 8.07965040e-01 5.36900461e-01
7.08972216e-01 -1.04142956e-01 1.54166847e-01 1.52643248e-01
4.26692277e-01 -3.12729746e-01 -9.39361870e-01 2.47865289e-01
9.42081988e-01 1.03021777e+00 -3.85143816e-01 -4.16201711e-01
-4.14953709e-01 7.40908027e-01 -5.70075735e-02 2.00626791e-01
-1.12318587e+00 -4.13588345e-01 1.01362956e+00 -4.14013535e-01
6.51094094e-02 -1.04230845e+00 -2.48278454e-01 -1.32441807e+00
3.05114120e-01 -4.54972506e-01 -8.07827860e-02 -8.04872930e-01
-5.19616365e-01 3.63835722e-01 -6.15393408e-02 -2.16049480e+00
-5.45847774e-01 -2.56344914e-01 2.15340015e-02 6.39611602e-01
-1.51725101e+00 -8.33056808e-01 -6.38261497e-01 5.95948637e-01
1.85102776e-01 7.77874216e-02 4.49570149e-01 4.90331054e-01
-5.19137383e-01 1.43490016e-01 5.05166292e-01 -3.10840905e-01
9.91645634e-01 -8.60386968e-01 6.87475681e-01 1.38872719e+00
7.92751834e-02 7.71613002e-01 1.22263372e+00 -7.24518180e-01
-2.05835700e+00 -7.62707829e-01 8.39850187e-01 -3.59236568e-01
8.72145951e-01 -1.69776976e-01 -4.67000246e-01 9.38385129e-01
-2.75604669e-02 -1.15586808e-02 1.18057877e-01 -1.34319812e-01
8.93394556e-03 -4.44614649e-01 -8.01327825e-01 5.61673224e-01
8.06033969e-01 -4.91359740e-01 -4.41238016e-01 2.26549968e-01
7.34191418e-01 -1.32459736e+00 -7.74535358e-01 4.42569911e-01
7.89488018e-01 -9.84490275e-01 9.03865635e-01 2.83305138e-01
-5.39182365e-01 -1.05424285e+00 -2.04990596e-01 -1.21378660e+00
5.25386818e-02 -7.78023303e-01 -3.89303386e-01 1.03933120e+00
-7.38884434e-02 -1.20681381e+00 4.73327577e-01 4.06309515e-01
-3.49259265e-02 3.57217900e-03 -1.21176839e+00 -9.53889012e-01
-1.06550324e+00 -6.88589811e-01 8.23196232e-01 7.31925249e-01
-1.07399836e-01 1.75793916e-01 -6.32028937e-01 9.71022725e-01
7.73452282e-01 1.36233032e-01 1.43370116e+00 -1.11512816e+00
1.63058518e-03 -1.02355226e-03 -9.22860026e-01 -1.36104071e+00
-2.73714691e-01 -1.50754780e-01 2.69567668e-01 -1.22954261e+00
-5.30479670e-01 -4.15692687e-01 -1.72232017e-01 -6.98893741e-02
2.28390880e-02 7.43604600e-02 7.41799325e-02 3.53670478e-01
-4.49552268e-01 5.08395851e-01 7.59171844e-01 -6.34550862e-03
-2.23922133e-01 1.55589789e-01 3.67939994e-02 6.30578518e-01
5.53207099e-01 -3.74033570e-01 -4.01758224e-01 -6.66366577e-01
3.79587620e-01 4.22007322e-01 3.71835411e-01 -1.45913398e+00
8.09561014e-01 -1.89925969e-01 1.09084316e-01 -1.06141281e+00
5.99581242e-01 -1.33777118e+00 7.50930309e-01 6.08838618e-01
5.59918046e-01 5.63435256e-01 2.42207363e-01 7.78645813e-01
-5.19833565e-01 3.82671282e-02 3.41995299e-01 2.84689993e-01
-1.15812850e+00 2.89178967e-01 -1.46664530e-01 -5.88132977e-01
1.19134057e+00 -4.15271193e-01 -3.98517191e-01 -5.06217778e-01
-1.71575725e-01 4.05380428e-01 8.79692435e-01 6.31579161e-01
4.46009725e-01 -1.48521173e+00 8.51531252e-02 2.12137103e-01
3.50392818e-01 1.43468142e-01 3.16596627e-01 1.29451036e+00
-5.72144449e-01 6.05086803e-01 -1.05501153e-01 -1.14428413e+00
-1.19316947e+00 4.41631675e-01 5.79176061e-02 1.13583617e-01
-2.93074191e-01 3.18966180e-01 -3.85293722e-01 -2.92031199e-01
2.79569387e-01 -6.24718666e-01 2.49164011e-02 -6.58932775e-02
4.21117127e-01 4.62855607e-01 1.76240444e-01 -9.60987270e-01
-6.71408236e-01 9.82159734e-01 2.65540540e-01 -3.84953409e-01
8.20418477e-01 -8.86033297e-01 7.29371011e-02 8.57646525e-01
1.09166014e+00 1.64624795e-01 -1.35523498e+00 -2.16144338e-01
5.91657013e-02 -6.28847361e-01 -2.61093602e-02 3.49991918e-02
-7.70882487e-01 4.29599673e-01 8.80412459e-01 2.16194615e-02
9.00634885e-01 -5.93539357e-01 7.73000240e-01 4.32459474e-01
1.03236008e+00 -1.13075936e+00 -4.84656811e-01 7.55267024e-01
4.66174543e-01 -1.38959503e+00 4.65559661e-01 -4.14508462e-01
-3.60495985e-01 8.47618699e-01 3.76576751e-01 -9.19103697e-02
3.87752116e-01 4.34154958e-01 3.48191261e-01 2.98414439e-01
-5.19825220e-01 -2.23668873e-01 6.82564229e-02 4.80833203e-01
5.39856702e-02 1.30339473e-01 -3.95683646e-01 1.66571420e-02
-2.28977054e-01 -9.26550478e-02 5.55615127e-01 1.40236032e+00
-3.43882948e-01 -1.04065192e+00 -8.56930554e-01 -2.23542005e-01
9.82486159e-02 2.69786626e-01 1.14170916e-01 8.63101006e-01
-1.54404610e-01 9.23082232e-01 -9.73523222e-03 -6.59246683e-01
4.66716409e-01 -3.35619986e-01 3.75394374e-01 -5.43149449e-02
-2.83146590e-01 4.15520482e-02 4.01524194e-02 -1.09731936e+00
-4.59947795e-01 -7.73672521e-01 -1.34672570e+00 -3.91508281e-01
-4.93049264e-01 -2.82811336e-02 1.45163250e+00 8.92389476e-01
5.66421092e-01 4.70052004e-01 6.70479417e-01 -1.17776549e+00
-3.73302549e-01 -6.52782142e-01 -3.73565763e-01 5.81798609e-03
7.92698741e-01 -1.03529775e+00 -3.16433072e-01 -3.57771851e-02] | [7.436136722564697, -1.9954513311386108] |
fdbb1de0-888f-4176-bc74-7f8c105dd433 | learning-semantic-role-labeling-from | 2305.14600 | null | https://arxiv.org/abs/2305.14600v1 | https://arxiv.org/pdf/2305.14600v1.pdf | Learning Semantic Role Labeling from Compatible Label Sequences | This paper addresses the question of how to efficiently learn from disjoint, compatible label sequences. We argue that the compatible structures between disjoint label sets help model learning and inference. We verify this hypothesis on the task of semantic role labeling (SRL), specifically, tagging a sentence with two role sequences: VerbNet arguments and PropBank arguments. Prior work has shown that cross-task interaction improves performance. However, the two tasks are still separately decoded, running the risk of generating structurally inconsistent label sequences (as per lexicons like SEMLINK). To eliminate this issue, we first propose a simple and effective setup that jointly handles VerbNet and PropBank labels as one sequence. With this setup, we show that enforcing SEMLINK constraints during decoding constantly improves the overall F1. With special input constructions, our joint model infers VerbNet arguments from PropBank arguments with over 99% accuracy. We also propose a constrained marginal model that uses SEMLINK information during training to further benefit from the large amounts of PropBank-only data. Our models achieve state-of-the-art F1's on VerbNet and PropBank argument labeling on the CoNLL05 dataset with strong out-of-domain generalization. | ['Vivek Srikumar', 'Martha Palmer', 'Susan W. Brown', 'Ghazaleh Kazeminejad', 'Tao Li'] | 2023-05-24 | null | null | null | null | ['semantic-role-labeling'] | ['natural-language-processing'] | [ 4.82264400e-01 7.45524704e-01 -6.30835593e-01 -7.21240222e-01
-1.21724808e+00 -1.06714761e+00 4.47230905e-01 2.65302122e-01
-6.88753188e-01 1.06787229e+00 4.19098914e-01 -4.94951427e-01
1.34479225e-01 -4.02239621e-01 -9.16436195e-01 -4.13536817e-01
1.36503130e-01 6.78710580e-01 4.45003688e-01 -2.81179965e-01
-1.47721007e-01 -1.29037350e-01 -1.35816371e+00 7.20254719e-01
7.13147879e-01 5.17546356e-01 2.94709504e-01 3.81430984e-01
-2.18240350e-01 1.01429701e+00 -7.01291323e-01 -7.46893406e-01
5.16482890e-02 -3.66303384e-01 -1.12765563e+00 -2.48429820e-01
3.30079883e-01 -7.13777989e-02 2.06846714e-01 8.95962596e-01
4.40513879e-01 -1.38568521e-01 4.38879341e-01 -1.12783694e+00
-4.67322528e-01 1.39747274e+00 -2.85382777e-01 1.33336797e-01
3.62773955e-01 -5.90250604e-02 1.91512585e+00 -6.58055782e-01
8.46768856e-01 1.60311675e+00 6.21805966e-01 7.53875971e-01
-1.55088985e+00 -5.51001012e-01 4.99380738e-01 1.09980337e-01
-9.45700824e-01 -5.02980173e-01 6.43080473e-01 -1.99534670e-01
1.22077119e+00 1.46945760e-01 2.14726001e-01 1.11769724e+00
-4.28414881e-01 9.62038994e-01 9.16765869e-01 -6.78149998e-01
1.27540961e-01 -1.43114626e-01 5.52212059e-01 6.87320113e-01
3.49804133e-01 -2.86679417e-01 -6.28096879e-01 -1.54501542e-01
3.26853096e-01 -8.10658872e-01 -2.38551065e-01 1.41201373e-02
-1.08921599e+00 9.52171504e-01 9.22970474e-02 3.01594824e-01
1.12836190e-01 4.47321773e-01 5.00477195e-01 2.71855682e-01
4.46092606e-01 5.25183201e-01 -9.61308122e-01 -4.07371372e-02
-3.60028625e-01 2.91343629e-01 9.22156513e-01 8.81610870e-01
5.14169216e-01 -2.52859116e-01 -2.19691500e-01 1.07316160e+00
4.60151285e-01 2.52413839e-01 1.16430230e-01 -1.22758663e+00
7.75170088e-01 3.40342432e-01 1.56954333e-01 -2.33022362e-01
-4.41486478e-01 -1.78583324e-01 1.07531473e-01 -1.92603573e-01
9.46104646e-01 -2.09195077e-01 -7.25873828e-01 2.55496454e+00
3.59651089e-01 5.62634356e-02 2.00504869e-01 6.07799947e-01
7.90731609e-01 4.64427471e-01 8.51025522e-01 -2.14407325e-01
1.85781527e+00 -8.24955642e-01 -8.20350707e-01 -6.50673211e-01
1.17780483e+00 -5.04834592e-01 1.39338756e+00 2.31515840e-02
-1.28200102e+00 -2.59202033e-01 -9.60822046e-01 -4.44315225e-01
9.10900533e-03 4.59398739e-02 9.15365279e-01 5.21746516e-01
-8.34019542e-01 4.68067467e-01 -8.02087486e-01 -4.71301959e-04
4.28492755e-01 3.20453346e-01 -3.55879486e-01 -2.48539224e-01
-1.55142856e+00 1.01822114e+00 7.55502284e-01 -3.01409662e-01
-5.87271571e-01 -8.68255794e-01 -1.21682966e+00 7.27313980e-02
5.43099403e-01 -5.86400867e-01 1.66158199e+00 -8.05912912e-01
-1.08391488e+00 1.30482376e+00 -3.54460984e-01 -4.96022612e-01
3.57845694e-01 -3.48238707e-01 6.02876842e-02 -2.94898869e-03
4.42352414e-01 9.46953475e-01 4.84457701e-01 -1.27230108e+00
-7.39546657e-01 -1.53386354e-01 5.06715596e-01 3.87962818e-01
-9.20758769e-02 3.95954639e-01 -2.40328625e-01 -5.52908063e-01
1.03589274e-01 -8.71865392e-01 1.19672537e-01 -1.05497397e-01
-3.28272581e-01 -9.19794321e-01 1.94094852e-01 -5.55549860e-01
9.73000407e-01 -2.12079501e+00 4.80724275e-02 -1.85381144e-01
-1.67580754e-01 1.91515148e-01 -2.40697831e-01 9.86602083e-02
-2.94251889e-01 3.80908251e-01 -4.81357932e-01 -6.28897309e-01
2.82304615e-01 8.69335234e-01 -2.82911837e-01 2.47823998e-01
3.39316219e-01 1.08084071e+00 -1.01628697e+00 -6.49462044e-01
-2.45093137e-01 1.29723728e-01 -7.78029978e-01 8.90196636e-02
-7.59502411e-01 3.74615520e-01 -1.13377146e-01 5.49055576e-01
5.77599704e-01 -3.68250608e-01 1.03290200e+00 -2.91528463e-01
2.23416612e-01 1.23955500e+00 -1.06418180e+00 1.84411788e+00
-4.76390392e-01 1.14264272e-01 2.37538457e-01 -1.11571598e+00
2.56434977e-01 4.51347083e-01 8.40052739e-02 -5.61686039e-01
6.65792599e-02 5.20310938e-01 4.15680945e-01 -2.38004625e-01
8.86072963e-02 -5.89143157e-01 -4.47489768e-01 5.61674535e-01
3.42193425e-01 -4.29405421e-02 4.89039034e-01 1.36984318e-01
1.05341947e+00 4.70369220e-01 5.45677960e-01 -4.68179494e-01
4.65242326e-01 7.16728494e-02 7.84216762e-01 6.67688370e-01
1.00139104e-01 3.05777818e-01 8.70752275e-01 5.29416720e-04
-8.05234969e-01 -1.18546581e+00 -2.93685228e-01 1.60142601e+00
3.13244872e-02 -4.32092786e-01 -6.01084054e-01 -1.32980812e+00
4.02493924e-02 1.05995500e+00 -3.33599120e-01 2.74745047e-01
-1.34129739e+00 -9.43193555e-01 7.50547945e-01 7.37492263e-01
7.00685456e-02 -1.20012414e+00 -4.27502006e-01 4.19097334e-01
-3.13253731e-01 -1.54225922e+00 -9.30562541e-02 6.00028872e-01
-5.57829797e-01 -1.28157341e+00 -2.45991647e-02 -1.05351496e+00
6.12979412e-01 -2.65432894e-01 1.66758823e+00 4.20284659e-01
1.89352691e-01 -2.08867528e-02 -3.86475056e-01 -1.89985499e-01
-6.21286094e-01 2.12083049e-02 -3.16966981e-01 -5.12590230e-01
1.94115698e-01 -5.47542751e-01 1.42014893e-02 8.51967037e-02
-6.22338295e-01 1.64874956e-01 2.37865686e-01 9.89672005e-01
5.97474277e-01 -2.12721482e-01 7.07802534e-01 -1.57330930e+00
3.44168156e-01 -4.92047638e-01 -6.05579436e-01 3.23547184e-01
-2.06391603e-01 3.95518333e-01 6.20263159e-01 -1.99905694e-01
-1.35903752e+00 4.59233038e-02 -5.82609832e-01 4.53749925e-01
-3.79305780e-02 3.90660256e-01 -6.72503650e-01 5.11429846e-01
6.12854064e-01 -3.22134495e-01 -1.64595753e-01 -6.32100344e-01
6.06386125e-01 3.19891632e-01 4.78481323e-01 -1.17619574e+00
6.23410761e-01 3.66183490e-01 -4.10209857e-02 -4.19571429e-01
-1.71734667e+00 -2.99704850e-01 -4.28141415e-01 4.81901079e-01
1.03363740e+00 -1.22309017e+00 -6.48107171e-01 1.35000780e-01
-1.60711515e+00 -7.08536029e-01 -3.88912827e-01 2.12236002e-01
-4.87073123e-01 3.75466913e-01 -9.06520545e-01 -4.95308936e-01
-6.42824471e-02 -8.66687596e-01 1.13254106e+00 -2.52779663e-01
-5.69991231e-01 -1.22034848e+00 -2.45223656e-01 3.63119721e-01
-1.36523619e-01 -1.15434490e-01 1.42904758e+00 -1.10091031e+00
-4.63874847e-01 3.25053185e-01 -2.93390572e-01 4.29786652e-01
6.75157905e-02 -7.32382536e-01 -1.14003110e+00 -5.71069606e-02
4.84385900e-02 -6.30364478e-01 9.39725161e-01 9.47574526e-02
1.00553453e+00 -3.78460765e-01 -2.52948284e-01 3.55059564e-01
1.29169619e+00 -1.04449980e-01 3.89707804e-01 2.69065857e-01
7.47466385e-01 8.29437613e-01 8.08237374e-01 -2.90682409e-02
5.46217918e-01 6.48228228e-01 3.17349762e-01 5.57843335e-02
-4.73298758e-01 -4.36569661e-01 4.62012917e-01 3.92052293e-01
4.29201275e-01 -6.06726408e-01 -7.28855193e-01 6.10001981e-01
-1.91331708e+00 -6.38382018e-01 -2.86213726e-01 1.71696210e+00
1.51458383e+00 4.37838376e-01 -3.76740396e-02 2.97686726e-01
5.28921604e-01 3.31326574e-01 -2.54786134e-01 -1.58859104e-01
-2.75131673e-01 5.95660508e-01 5.00316620e-01 1.06038678e+00
-1.21460962e+00 1.36220407e+00 6.02859259e+00 9.39336419e-01
-4.68747377e-01 6.17636800e-01 3.99765581e-01 2.09623843e-01
-8.54976952e-01 3.11844766e-01 -1.26514578e+00 4.81654257e-01
8.05634558e-01 2.67496914e-01 1.77079991e-01 7.56283641e-01
-1.76108509e-01 -1.33521914e-01 -1.49252295e+00 5.53688586e-01
-7.13974908e-02 -1.28686047e+00 -1.53974816e-01 -2.92071879e-01
4.19536352e-01 -1.48284985e-02 -3.84017140e-01 3.90430182e-01
8.51607859e-01 -9.65250671e-01 1.05503154e+00 -4.35135424e-01
1.02729309e+00 -3.65833044e-01 6.70296431e-01 3.83180350e-01
-1.06274366e+00 -1.00270160e-01 -1.73195571e-01 -2.62651473e-01
5.07553220e-01 5.63485980e-01 -9.54990387e-01 3.15713227e-01
1.24241665e-01 7.49811888e-01 -3.69463295e-01 3.65292728e-01
-1.35538006e+00 1.03649974e+00 -4.21723485e-01 1.71532661e-01
2.45445762e-02 3.01576585e-01 4.83449399e-01 1.24260128e+00
-1.54778793e-01 1.37257785e-01 3.97419900e-01 8.27096641e-01
-2.92708218e-01 -6.76114559e-02 -3.28432977e-01 1.12549469e-01
7.59292781e-01 1.00382757e+00 -7.69609928e-01 -4.19120997e-01
-4.36197996e-01 7.19832301e-01 6.60330892e-01 -2.80976444e-02
-8.69772077e-01 2.69959241e-01 7.68926919e-01 1.16848201e-01
2.77194321e-01 -1.90032050e-01 -5.57468832e-01 -1.02588701e+00
2.54015718e-02 -7.76684642e-01 8.22577715e-01 -4.22243029e-01
-1.39399266e+00 6.40589148e-02 3.80807370e-01 -5.89233160e-01
-4.90773916e-01 -8.51611197e-01 -2.50409186e-01 8.12293708e-01
-1.91610980e+00 -1.36760235e+00 2.79100329e-01 1.62670344e-01
6.91371500e-01 2.90992975e-01 8.30642343e-01 3.15718859e-01
-2.23277032e-01 7.65593529e-01 -6.28650665e-01 2.39262074e-01
6.90752923e-01 -1.50265837e+00 6.04505002e-01 9.01006937e-01
1.48560166e-01 6.09537899e-01 6.25676990e-01 -7.25420415e-01
-9.14510906e-01 -8.96805346e-01 1.44988418e+00 -9.26553965e-01
7.34939754e-01 -6.97870374e-01 -9.93975282e-01 9.60724473e-01
7.43526220e-02 1.74238846e-01 7.53803551e-01 4.60594088e-01
-6.26664519e-01 3.39369386e-01 -1.09441984e+00 4.81814444e-01
1.59514117e+00 -5.03691375e-01 -1.07385349e+00 4.53232765e-01
1.37061870e+00 -5.01564085e-01 -4.30209845e-01 4.01092082e-01
2.10497513e-01 -5.45623183e-01 1.01928306e+00 -7.79045999e-01
3.35774064e-01 -3.63091171e-01 -5.30407429e-01 -1.08048868e+00
-1.46457002e-01 -4.00670528e-01 1.18395472e-02 1.36425734e+00
8.02415907e-01 -5.90438485e-01 6.48760736e-01 5.45302272e-01
-4.01562929e-01 -4.29954618e-01 -9.85528290e-01 -7.86666572e-01
1.83565915e-01 -6.80661917e-01 3.75256360e-01 8.47428501e-01
-2.64468696e-02 9.43087101e-01 1.09765334e-02 6.07846640e-02
5.87350428e-01 2.44365316e-02 2.84776270e-01 -1.34580672e+00
-6.88919485e-01 -5.35025299e-02 2.82864064e-01 -1.06759214e+00
9.87191260e-01 -1.34878778e+00 3.14623147e-01 -1.56688368e+00
4.98013236e-02 -8.96080971e-01 -9.66107547e-02 1.09900081e+00
-5.03702700e-01 2.34549254e-01 3.24606985e-01 1.49283896e-03
-6.75015807e-01 1.76457465e-01 9.58378613e-01 1.79416701e-01
5.23298234e-02 -4.01903838e-01 -1.01504028e+00 1.06120479e+00
6.88653231e-01 -9.01222825e-01 -5.68474174e-01 -7.54368603e-01
5.09996176e-01 -3.78055964e-04 2.52287835e-01 -4.13296193e-01
-2.46452138e-01 -5.57321794e-02 -1.10260524e-01 -9.55911428e-02
4.11959976e-01 -5.66753149e-01 -2.85310835e-01 3.68863076e-01
-6.19616747e-01 -1.52116209e-01 2.37737477e-01 4.09502655e-01
-1.83113702e-02 -6.28621459e-01 6.72276616e-01 -3.52659762e-01
-7.37594485e-01 -2.88366020e-01 1.21138357e-02 8.32928717e-01
6.04839504e-01 1.66284636e-01 -5.53021789e-01 5.42519055e-02
-7.55104363e-01 2.81478554e-01 3.60171676e-01 1.09079614e-01
9.68645215e-02 -9.53849852e-01 -8.81191313e-01 7.40086436e-02
-4.02581990e-02 2.94655114e-01 -1.08171567e-01 5.74532032e-01
-1.40761405e-01 2.43959755e-01 2.24431813e-01 -4.65763748e-01
-1.33292770e+00 1.58944979e-01 6.93874136e-02 -7.03621209e-01
-4.13827896e-01 1.40746140e+00 2.95936882e-01 -6.12200201e-01
3.96860689e-02 -2.13932544e-01 -2.19749764e-01 3.17683637e-01
2.69305408e-01 -2.24570855e-01 7.13475645e-02 -5.01494944e-01
-5.74671090e-01 2.33306155e-01 -4.00179140e-02 -4.38912094e-01
1.32400560e+00 7.18001723e-02 -2.98289210e-01 2.08014041e-01
1.07054341e+00 4.60718960e-01 -1.24067104e+00 -1.59429193e-01
5.62312484e-01 -2.69067228e-01 -5.32101333e-01 -9.59735632e-01
-4.29852605e-01 6.23042166e-01 -2.56480612e-02 3.48347835e-02
6.26892149e-01 5.74183106e-01 7.74580181e-01 2.54930735e-01
3.35334718e-01 -9.42605257e-01 9.18292254e-02 8.89668822e-01
6.03045881e-01 -1.10320807e+00 -1.75135717e-01 -1.08556151e+00
-6.33214951e-01 4.83271211e-01 5.37305534e-01 2.20273193e-02
2.71892995e-01 4.34606880e-01 7.74165941e-03 -1.36369780e-01
-9.38841879e-01 -4.59365964e-01 4.56867069e-02 5.37086368e-01
8.66374075e-01 1.69440180e-01 -8.27183723e-01 8.67243409e-01
-3.24581623e-01 -4.30161327e-01 2.43862689e-01 9.91546988e-01
-4.16928917e-01 -1.70840240e+00 -4.39268462e-02 1.30943015e-01
-9.12986696e-01 -5.72922230e-01 -1.39708668e-01 1.00284576e+00
2.89567471e-01 9.07232761e-01 1.22814447e-01 2.18204603e-01
1.64161831e-01 5.93481481e-01 8.08624387e-01 -1.29684532e+00
-4.79741722e-01 -1.50560707e-01 1.02985883e+00 -4.05487001e-01
-5.44025421e-01 -6.47213280e-01 -1.85081136e+00 2.76109487e-01
-1.57260969e-01 2.59851724e-01 4.95684683e-01 1.24090528e+00
1.98614858e-02 5.08122385e-01 1.09030502e-02 -3.79370868e-01
-8.55003178e-01 -8.38632643e-01 -3.14569116e-01 6.93318665e-01
1.55753329e-01 -7.29901254e-01 -4.67430413e-01 2.85364017e-02] | [10.382769584655762, 9.42599868774414] |
82d674d5-896c-4e29-a86e-64238bcb36fa | enhancing-retrieval-augmented-large-language | 2305.15294 | null | https://arxiv.org/abs/2305.15294v1 | https://arxiv.org/pdf/2305.15294v1.pdf | Enhancing Retrieval-Augmented Large Language Models with Iterative Retrieval-Generation Synergy | Large language models are powerful text processors and reasoners, but are still subject to limitations including outdated knowledge and hallucinations, which necessitates connecting them to the world. Retrieval-augmented large language models have raised extensive attention for grounding model generation on external knowledge. However, retrievers struggle to capture relevance, especially for queries with complex information needs. Recent work has proposed to improve relevance modeling by having large language models actively involved in retrieval, i.e., to improve retrieval with generation. In this paper, we show that strong performance can be achieved by a method we call Iter-RetGen, which synergizes retrieval and generation in an iterative manner. A model output shows what might be needed to finish a task, and thus provides an informative context for retrieving more relevant knowledge which in turn helps generate a better output in the next iteration. Compared with recent work which interleaves retrieval with generation when producing an output, Iter-RetGen processes all retrieved knowledge as a whole and largely preserves the flexibility in generation without structural constraints. We evaluate Iter-RetGen on multi-hop question answering, fact verification, and commonsense reasoning, and show that it can flexibly leverage parametric knowledge and non-parametric knowledge, and is superior to or competitive with state-of-the-art retrieval-augmented baselines while causing fewer overheads of retrieval and generation. We can further improve performance via generation-augmented retrieval adaptation. | ['Weizhu Chen', 'Nan Duan', 'Minlie Huang', 'Yelong Shen', 'Yeyun Gong', 'Zhihong Shao'] | 2023-05-24 | null | null | null | null | ['multi-hop-question-answering', 'fact-verification'] | ['knowledge-base', 'natural-language-processing'] | [ 1.64050609e-01 2.12868303e-01 -3.48362088e-01 6.44484758e-02
-1.36763728e+00 -6.78264558e-01 8.85665059e-01 4.10423130e-01
-4.63585436e-01 6.98299527e-01 7.89553940e-01 -3.85576874e-01
-2.38161922e-01 -1.07098329e+00 -7.06440151e-01 -6.54723793e-02
4.46097851e-01 9.70628440e-01 1.84086919e-01 -6.95128620e-01
4.81252491e-01 1.64883837e-01 -1.36368251e+00 6.87948287e-01
9.28480029e-01 5.14205694e-01 2.23731220e-01 6.49061322e-01
-4.37071472e-01 1.28668344e+00 -6.75675809e-01 -6.11079276e-01
-9.08371955e-02 -4.10401553e-01 -1.34542632e+00 -3.72875154e-01
4.15421963e-01 -5.39974034e-01 -3.41970086e-01 6.84196353e-01
5.29546738e-01 3.71500999e-01 5.60415745e-01 -7.72811353e-01
-1.29743803e+00 1.06667173e+00 -2.69127451e-02 2.94400901e-01
7.71378994e-01 1.96433380e-01 1.31211960e+00 -9.22939181e-01
7.98592746e-01 1.51813960e+00 3.38852316e-01 6.03438675e-01
-1.24914825e+00 -5.34289181e-01 1.30989775e-01 3.65611136e-01
-1.25359213e+00 -5.88956475e-01 5.60087502e-01 3.18159945e-02
1.48971343e+00 4.06144798e-01 5.52513897e-01 9.66853857e-01
-2.18477443e-01 1.21476185e+00 7.00348675e-01 -6.51698947e-01
1.37225911e-01 2.15097833e-02 2.60192662e-01 4.55438733e-01
3.16568166e-01 -1.05571717e-01 -6.70187354e-01 -4.22253191e-01
5.81160545e-01 -1.42161012e-01 -3.48551840e-01 1.04687914e-01
-1.23541927e+00 9.05219674e-01 4.25425887e-01 2.24123865e-01
-6.89700603e-01 2.46607482e-01 2.12434381e-01 4.38385278e-01
2.07455635e-01 1.09428144e+00 -4.53741550e-01 -1.45331353e-01
-1.01572859e+00 6.58958912e-01 1.08495927e+00 9.67102051e-01
7.69131124e-01 -9.85661894e-02 -6.48572922e-01 8.43750298e-01
2.35743657e-01 8.36838841e-01 8.03214192e-01 -1.21162701e+00
5.75057149e-01 7.22497344e-01 1.28159896e-01 -7.78502047e-01
-1.79625541e-01 -5.09823084e-01 -3.74416232e-01 -4.69999939e-01
7.21982569e-02 4.50569317e-02 -8.93478096e-01 1.82124126e+00
1.46168992e-02 -2.44946927e-02 6.11722589e-01 7.76111543e-01
9.34474468e-01 7.44247675e-01 2.25968108e-01 -4.80532870e-02
1.39660990e+00 -9.79505837e-01 -6.47720516e-01 -6.03356123e-01
8.62409234e-01 -8.40063751e-01 1.35273325e+00 2.93497890e-01
-1.27369583e+00 -1.72408611e-01 -7.85832942e-01 -4.19230223e-01
-3.71221513e-01 -3.82232249e-01 9.00355220e-01 3.75215530e-01
-1.23009038e+00 2.95455337e-01 -4.76515323e-01 -2.56565124e-01
2.40224659e-01 3.01998314e-02 1.28593650e-02 -4.57081765e-01
-1.61013186e+00 1.11270595e+00 6.64074004e-01 -1.16996229e-01
-7.19464898e-01 -8.07997823e-01 -9.12273824e-01 1.41035065e-01
8.07850122e-01 -1.29233217e+00 1.54713011e+00 -4.14954394e-01
-1.30889964e+00 3.71634632e-01 -3.20362747e-01 -4.88782644e-01
5.92365526e-02 -4.74606186e-01 -4.47325766e-01 2.14772627e-01
1.81617394e-01 7.56715477e-01 5.54012895e-01 -1.17429483e+00
-2.08637610e-01 -2.32818410e-01 5.36672235e-01 6.02895379e-01
-3.50269377e-01 1.61516182e-02 -8.19817781e-01 -6.25535548e-01
1.36375934e-01 -7.10437775e-01 -1.38862297e-01 -4.96208251e-01
-2.68913299e-01 -4.18257654e-01 4.59245533e-01 -5.67915142e-01
1.46268451e+00 -1.60674989e+00 1.02389991e-01 2.29752168e-01
2.17910454e-01 4.21107531e-01 -4.81876493e-01 6.83691978e-01
2.39338264e-01 4.24476027e-01 5.21893948e-02 -8.87725800e-02
1.75658107e-01 2.33015582e-01 -8.47251952e-01 -4.23098534e-01
2.00306073e-01 1.57634902e+00 -1.23023641e+00 -5.28791606e-01
-1.96864754e-01 3.06471169e-01 -8.05128753e-01 6.03562742e-02
-7.18555629e-01 -2.80205369e-01 -6.44263864e-01 6.65708303e-01
1.55371308e-01 -6.89227998e-01 -4.90771830e-02 5.94811328e-02
6.19175196e-01 7.63414919e-01 -1.11023164e+00 1.80829608e+00
-7.04533517e-01 3.32700610e-01 -3.35603952e-01 -3.98272544e-01
6.60755932e-01 3.21585387e-01 -7.05231726e-02 -1.00278270e+00
-2.88614780e-01 2.57374585e-01 -1.17690466e-01 -5.63877344e-01
1.12649691e+00 -3.97039980e-01 -3.45631987e-02 9.21506345e-01
-1.13994144e-01 -5.36616504e-01 3.60727757e-01 1.00319517e+00
1.23579943e+00 8.79015177e-02 9.86525565e-02 1.46329030e-01
2.76548326e-01 1.17146246e-01 3.19033787e-02 1.11710763e+00
4.57949549e-01 2.44010612e-01 1.81120738e-01 8.04450065e-02
-6.86513603e-01 -9.44071949e-01 3.12646866e-01 1.21226573e+00
1.75177604e-01 -7.37558305e-01 -3.34829688e-01 -4.89616603e-01
3.56060714e-02 1.35145533e+00 -2.99622387e-01 -6.09982371e-01
-6.15570366e-01 -5.66854954e-01 8.33018363e-01 8.12786281e-01
2.95177668e-01 -1.48129225e+00 -4.61716026e-01 4.90733743e-01
-7.11134791e-01 -8.86546195e-01 -3.84613067e-01 8.75919461e-02
-1.00807333e+00 -7.62771010e-01 -4.09132421e-01 -3.66025269e-01
3.05199355e-01 3.50375056e-01 1.72938657e+00 3.43576819e-01
5.09088822e-02 6.83780551e-01 -4.01802093e-01 -2.95548767e-01
-5.80241263e-01 3.66969556e-01 -2.71230608e-01 -6.49616778e-01
3.75924230e-01 -5.05069077e-01 -4.92597520e-01 -1.98263507e-02
-1.38634491e+00 4.36973497e-02 8.49942327e-01 8.76379013e-01
3.93421620e-01 -2.45038018e-01 1.03871477e+00 -1.06221342e+00
1.30798924e+00 -6.51774466e-01 -1.88716814e-01 6.42745912e-01
-8.90474916e-01 6.96594298e-01 5.24550676e-01 -4.67556924e-01
-1.15574765e+00 -6.74255848e-01 -7.99253210e-03 -2.86178499e-01
3.07281405e-01 9.99364257e-01 7.75017357e-03 4.19234335e-01
1.19036949e+00 3.70247751e-01 -2.13319719e-01 -2.55110502e-01
1.03776300e+00 5.48080325e-01 5.43858469e-01 -1.07163310e+00
7.09299505e-01 1.42214775e-01 -3.83835047e-01 -2.76780516e-01
-1.01033580e+00 -6.85743630e-01 4.28339690e-02 2.32629389e-01
2.25731298e-01 -9.60918605e-01 -4.94825572e-01 7.08372444e-02
-1.22945130e+00 -3.91483337e-01 -6.18587315e-01 2.06351742e-01
-4.79545712e-01 3.33312839e-01 -8.01427364e-01 -7.02321708e-01
-7.58214176e-01 -7.26466894e-01 1.14140534e+00 9.23238397e-02
-4.12384927e-01 -1.16981256e+00 5.52833732e-03 6.34242952e-01
7.50275671e-01 -3.66571546e-01 1.16884971e+00 -8.49989235e-01
-9.30322170e-01 -2.59169877e-01 -2.01919213e-01 1.63882673e-01
-1.83887884e-01 -6.31854653e-01 -8.43450189e-01 1.26427785e-03
-4.99507189e-02 -9.20690477e-01 1.03252244e+00 -2.26130635e-01
8.22780073e-01 -6.45849168e-01 -1.70047805e-01 2.11736113e-01
1.30453789e+00 -2.79520784e-04 8.87554884e-01 3.45970243e-01
4.92579520e-01 3.49467695e-01 4.61391091e-01 2.27383271e-01
6.61130607e-01 5.18028319e-01 4.15314436e-02 1.24285854e-01
-2.86344796e-01 -5.31750977e-01 2.92579859e-01 6.91235602e-01
1.03893697e-01 -3.65467578e-01 -9.10275340e-01 7.42889762e-01
-1.98310483e+00 -1.19494569e+00 2.42150575e-01 2.04627371e+00
1.48276842e+00 6.43350780e-02 -3.63706499e-01 -1.83354512e-01
1.80478081e-01 2.73546651e-02 -7.28642166e-01 -2.35635132e-01
-2.83541322e-01 5.63804388e-01 1.35479167e-01 7.96026289e-01
-3.75710070e-01 1.35908055e+00 6.70598173e+00 1.05959511e+00
-8.10065866e-01 -4.53820527e-02 1.42989680e-01 -2.06635639e-01
-1.05302477e+00 3.28650087e-01 -8.25089037e-01 -1.00173458e-01
8.95781219e-01 -5.58505952e-01 7.97732472e-01 6.24045849e-01
-3.94625038e-01 -8.89987573e-02 -1.26173878e+00 8.12294185e-01
4.68800485e-01 -1.49444771e+00 7.46513128e-01 -2.04641461e-01
6.87621415e-01 -3.28569449e-02 -1.09734058e-01 8.56177449e-01
5.87681293e-01 -1.09113526e+00 5.95647931e-01 6.57838702e-01
4.20199424e-01 -5.41922271e-01 6.62633359e-01 6.33399963e-01
-8.44959557e-01 -8.67160335e-02 -3.27957600e-01 -1.05957270e-01
1.90712914e-01 3.56819153e-01 -1.05236578e+00 5.47512114e-01
2.73729712e-02 2.71001279e-01 -7.95541823e-01 7.06430376e-01
-5.48754215e-01 5.18797040e-01 -3.14289808e-01 -2.08691254e-01
1.28350079e-01 3.74189734e-01 4.66151536e-01 1.14593863e+00
1.01059504e-01 2.59150952e-01 2.64986694e-01 1.16549003e+00
-4.39289063e-01 1.51868284e-01 -5.26985347e-01 -2.46415526e-01
7.63736010e-01 1.02850187e+00 -2.77074873e-01 -9.27362740e-01
-1.47704765e-01 8.17625582e-01 5.73133767e-01 5.35674095e-01
-4.36168998e-01 -3.42949659e-01 7.55514652e-02 4.90333214e-02
1.65535405e-01 2.33087838e-02 -8.42583477e-02 -1.38046396e+00
2.30351642e-01 -1.23309493e+00 7.11585402e-01 -1.16652751e+00
-1.29234266e+00 4.84210342e-01 2.46453986e-01 -5.37375808e-01
-9.16041434e-01 -2.35245153e-01 -6.97808638e-02 1.11185944e+00
-1.85193896e+00 -1.20125771e+00 3.32566425e-02 7.70253241e-01
6.02630615e-01 7.90645257e-02 1.10761464e+00 -4.98563722e-02
3.73512432e-02 5.84430814e-01 -2.05804169e-01 -6.92932457e-02
7.67841458e-01 -1.24409807e+00 3.37422371e-01 7.87245095e-01
4.15247172e-01 1.28636038e+00 4.06584263e-01 -9.04779732e-01
-1.75445902e+00 -9.16927755e-01 1.20369577e+00 -1.00246978e+00
8.01377296e-01 1.10392431e-02 -1.09939945e+00 7.30426669e-01
1.40582979e-01 -3.00165951e-01 6.35610223e-01 5.47318935e-01
-8.55145514e-01 1.15965024e-01 -8.11016083e-01 8.71793747e-01
1.06906426e+00 -9.57710445e-01 -1.30829036e+00 4.01626587e-01
1.01276922e+00 -3.62337083e-01 -6.31501019e-01 2.02899799e-01
3.39180887e-01 -4.55583423e-01 1.21698713e+00 -7.58354127e-01
4.82370019e-01 -2.44290739e-01 -2.50093788e-01 -1.19967532e+00
-3.50210994e-01 -7.65182376e-01 -6.23563945e-01 9.86367166e-01
7.12915897e-01 -5.71234286e-01 3.61402631e-01 1.06368208e+00
-1.86392367e-01 -5.84841728e-01 -4.84422684e-01 -7.80600250e-01
2.10573137e-01 -7.24974811e-01 8.72503817e-01 8.07058573e-01
3.23921144e-01 9.06318724e-01 -2.22586505e-02 -6.23514727e-02
2.24703699e-01 3.30895305e-01 7.18604803e-01 -9.24448729e-01
-5.94165266e-01 -5.60085714e-01 9.05042663e-02 -1.35925102e+00
2.29276597e-01 -1.21611071e+00 2.95617506e-02 -2.01058912e+00
4.69952345e-01 -4.85406846e-01 -2.92871177e-01 9.02653515e-01
-7.19677746e-01 8.99576470e-02 2.82484353e-01 4.28471506e-01
-9.50802863e-01 4.72715467e-01 1.33448458e+00 -2.84061939e-01
-3.89748156e-01 -3.96747738e-01 -1.56464136e+00 3.54257911e-01
6.11497879e-01 -2.90491998e-01 -8.43356490e-01 -5.65100372e-01
8.07659686e-01 2.93121457e-01 5.24610817e-01 -6.95140302e-01
6.24962091e-01 -1.24791875e-01 2.08946809e-01 -3.89215469e-01
4.93204087e-01 -4.51504141e-01 -6.29878342e-02 8.12108517e-02
-8.93440783e-01 1.75389186e-01 3.41402799e-01 5.10666430e-01
-2.04817817e-01 -4.33771133e-01 3.35196592e-02 -4.49007571e-01
-6.58556283e-01 -6.57814369e-02 -1.39278710e-01 6.11259222e-01
1.84959918e-01 9.29787233e-02 -8.04276943e-01 -7.03702807e-01
-3.77615601e-01 3.63124073e-01 3.26641709e-01 4.64144737e-01
8.13666284e-01 -1.31759405e+00 -8.88468385e-01 -9.42362994e-02
4.37729508e-01 3.44038382e-02 5.82873411e-02 4.85908031e-01
-2.72483025e-02 7.15987206e-01 4.24846619e-01 -2.37369776e-01
-8.54848564e-01 6.96865141e-01 3.22303176e-02 -7.81903803e-01
-4.93098587e-01 7.24106669e-01 -6.47911802e-02 -3.89406681e-01
9.11620483e-02 -4.02115256e-01 -1.32148847e-01 -6.04131334e-02
8.15096498e-01 6.30478486e-02 2.71962970e-01 -9.49353725e-02
-1.57263711e-01 1.53431818e-01 -4.66183484e-01 -5.42070627e-01
1.01849997e+00 3.29591110e-02 -2.92648107e-01 2.93664455e-01
8.04577291e-01 2.44503215e-01 -5.27987421e-01 -7.72713184e-01
7.01266453e-02 -4.11576062e-01 1.36836186e-01 -1.52899814e+00
-6.45317793e-01 5.48650563e-01 -8.86211842e-02 3.84550765e-02
1.11345494e+00 2.64332414e-01 9.95014310e-01 1.24338317e+00
4.28885460e-01 -9.41548586e-01 4.44566220e-01 8.60059500e-01
1.10463107e+00 -9.86198187e-01 2.90559251e-02 -6.35423809e-02
-7.17088223e-01 9.01506305e-01 5.39176643e-01 3.84687632e-01
1.41047895e-01 4.69511151e-02 9.39319283e-02 -4.02840674e-01
-1.15669966e+00 -3.35661232e-01 4.89308387e-01 4.66155380e-01
4.42615509e-01 -1.04475953e-02 -3.58139239e-02 6.31664038e-01
-3.87663722e-01 2.95066983e-01 2.87934899e-01 1.08227670e+00
-5.11850297e-01 -1.32248402e+00 -4.47369665e-01 6.75317109e-01
-1.97191745e-01 -8.90596926e-01 -3.55598986e-01 6.41475677e-01
-2.79146194e-01 1.17324209e+00 -4.51652795e-01 -1.07638337e-01
2.60848314e-01 4.49690819e-01 5.13140798e-01 -9.53526556e-01
-7.96446383e-01 -1.10356927e-01 4.60615307e-01 -6.06606007e-01
-8.53030756e-02 -3.62507612e-01 -1.43206525e+00 -1.61864042e-01
-5.83273172e-01 4.00666207e-01 3.78212452e-01 9.78245139e-01
6.89671874e-01 3.35884988e-01 8.22785683e-03 -1.52004778e-01
-8.86642992e-01 -9.74348187e-01 -3.62786762e-02 5.55775821e-01
1.05962031e-01 -2.96468288e-01 -2.14345679e-01 -3.68398689e-02] | [11.154084205627441, 7.929203033447266] |
d4612894-89cf-4810-983c-5e20bdafdba8 | halp-hallucinating-latent-positives-for | 2304.00387 | null | https://arxiv.org/abs/2304.00387v1 | https://arxiv.org/pdf/2304.00387v1.pdf | HaLP: Hallucinating Latent Positives for Skeleton-based Self-Supervised Learning of Actions | Supervised learning of skeleton sequence encoders for action recognition has received significant attention in recent times. However, learning such encoders without labels continues to be a challenging problem. While prior works have shown promising results by applying contrastive learning to pose sequences, the quality of the learned representations is often observed to be closely tied to data augmentations that are used to craft the positives. However, augmenting pose sequences is a difficult task as the geometric constraints among the skeleton joints need to be enforced to make the augmentations realistic for that action. In this work, we propose a new contrastive learning approach to train models for skeleton-based action recognition without labels. Our key contribution is a simple module, HaLP - to Hallucinate Latent Positives for contrastive learning. Specifically, HaLP explores the latent space of poses in suitable directions to generate new positives. To this end, we present a novel optimization formulation to solve for the synthetic positives with an explicit control on their hardness. We propose approximations to the objective, making them solvable in closed form with minimal overhead. We show via experiments that using these generated positives within a standard contrastive learning framework leads to consistent improvements across benchmarks such as NTU-60, NTU-120, and PKU-II on tasks like linear evaluation, transfer learning, and kNN evaluation. Our code will be made available at https://github.com/anshulbshah/HaLP. | ['Rama Chellappa', 'Anoop Cherian', 'David Jacobs', 'Shlok Kumar Mishra', 'Ketul Shah', 'Aniket Roy', 'Anshul Shah'] | 2023-04-01 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Shah_HaLP_Hallucinating_Latent_Positives_for_Skeleton-Based_Self-Supervised_Learning_of_Actions_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Shah_HaLP_Hallucinating_Latent_Positives_for_Skeleton-Based_Self-Supervised_Learning_of_Actions_CVPR_2023_paper.pdf | cvpr-2023-1 | ['action-recognition-in-videos'] | ['computer-vision'] | [ 6.44514918e-01 1.95865706e-01 -3.87509972e-01 -3.85011792e-01
-1.19847906e+00 -4.81135070e-01 5.02326667e-01 -3.06436896e-01
-3.14357013e-01 6.92984700e-01 2.86159247e-01 3.84196360e-03
7.07066283e-02 -4.11865920e-01 -1.20641792e+00 -4.69223082e-01
-1.08734764e-01 5.74052215e-01 1.15593404e-01 -1.47413909e-01
5.13176359e-02 3.37282002e-01 -1.28624046e+00 4.30935532e-01
5.82551181e-01 9.43689823e-01 -3.46201845e-02 7.80856311e-01
2.94529498e-01 9.85752702e-01 -4.50535178e-01 -3.64017397e-01
6.02872431e-01 -5.84290981e-01 -8.98128033e-01 5.06515086e-01
7.09829628e-01 -5.95056117e-01 -3.33858848e-01 7.72138178e-01
4.65794712e-01 1.70831472e-01 6.78427637e-01 -1.44718230e+00
-1.42262235e-01 4.36008751e-01 -7.30416477e-01 -1.09473713e-01
4.04510647e-01 4.14141148e-01 1.15368080e+00 -9.67932403e-01
7.24140227e-01 1.31354809e+00 6.46101356e-01 7.59032130e-01
-1.36931586e+00 -6.39729142e-01 1.49464771e-01 2.67583847e-01
-1.18145287e+00 -4.81248528e-01 7.19994724e-01 -4.06013489e-01
8.11305702e-01 2.36121371e-01 7.46261179e-01 1.32843280e+00
-1.14807114e-01 1.10314047e+00 1.02097499e+00 -4.34242785e-01
2.22761750e-01 -2.97753811e-01 -4.77117784e-02 8.06696117e-01
-5.99349895e-03 -6.27103299e-02 -6.88605428e-01 -9.23092216e-02
7.81859398e-01 -5.81585355e-02 -2.03049481e-01 -7.88344860e-01
-9.48268116e-01 7.80322373e-01 3.42348933e-01 -2.02704251e-01
-3.00719887e-01 5.78040659e-01 4.73325282e-01 7.21254200e-02
5.04930019e-01 5.51949799e-01 -5.36361754e-01 -4.48724210e-01
-6.45370722e-01 4.36731040e-01 5.67215681e-01 7.89899528e-01
6.28711343e-01 -4.73691225e-02 -9.26362425e-02 9.02641535e-01
1.72870815e-01 4.07273442e-01 2.93946683e-01 -1.29641664e+00
6.40105724e-01 4.79961753e-01 7.61417905e-03 -8.14698637e-01
-2.45232329e-01 -4.46600497e-01 -4.21100050e-01 3.57945234e-01
5.45681775e-01 -1.30567789e-01 -1.01822090e+00 1.87184560e+00
3.80103946e-01 4.00695980e-01 -3.18133421e-02 8.04163039e-01
1.86898708e-01 6.27020299e-01 -7.48161599e-02 1.39645562e-01
1.06935906e+00 -1.22274733e+00 -3.55286211e-01 -4.55525547e-01
8.03419352e-01 -6.34660840e-01 1.32080114e+00 4.49117988e-01
-1.22700727e+00 -3.20441246e-01 -1.07889628e+00 -1.73053801e-01
2.85277665e-01 2.49495566e-01 4.70042497e-01 2.82891691e-01
-7.87710011e-01 7.84575641e-01 -1.34507751e+00 -2.27959067e-01
6.58275604e-01 4.25125062e-01 -3.34493101e-01 -1.18565261e-01
-8.08689594e-01 7.41054773e-01 5.12112141e-01 4.59181145e-02
-1.15177679e+00 -5.24819911e-01 -9.43592846e-01 -1.52688608e-01
8.09629440e-01 -5.78042567e-01 1.57185340e+00 -1.09677958e+00
-1.53262115e+00 5.48583567e-01 -3.81237268e-03 -4.73680258e-01
7.97232330e-01 -6.05504334e-01 3.37904036e-01 2.26770371e-01
8.15389305e-02 1.13796389e+00 8.38912725e-01 -1.12938392e+00
-4.17633533e-01 -1.30933940e-01 2.74717212e-01 4.20114279e-01
-5.09686351e-01 -7.22123757e-02 -6.41797721e-01 -8.15249741e-01
2.13280451e-02 -1.32463992e+00 -3.27872962e-01 4.32427615e-01
-3.20158392e-01 -1.09770149e-01 7.42217064e-01 -7.32593954e-01
6.48607790e-01 -2.00293708e+00 3.37053001e-01 4.87335958e-02
2.72078719e-02 2.90348619e-01 -4.02724028e-01 2.80629039e-01
-1.80420190e-01 -1.47457227e-01 -4.38815236e-01 -5.43430805e-01
-2.63671409e-02 5.50690830e-01 -1.94870681e-01 3.51005435e-01
5.44282675e-01 9.44624782e-01 -9.62975264e-01 -4.12392676e-01
1.97611660e-01 4.46479589e-01 -9.63965416e-01 2.40457609e-01
-4.72264916e-01 4.78934705e-01 -2.43742645e-01 5.94592452e-01
2.03926161e-01 -4.54955518e-01 2.21322730e-01 -1.30979910e-01
3.18443686e-01 3.82733852e-01 -1.23169494e+00 1.75114095e+00
-3.90787929e-01 4.45538670e-01 -1.77678972e-01 -9.66496527e-01
5.57060540e-01 1.33613512e-01 6.36797369e-01 -3.96690905e-01
4.42212261e-02 2.09805086e-01 -1.08977705e-01 -3.49448830e-01
2.82152534e-01 -1.13891929e-01 1.32576838e-01 4.56154823e-01
7.82140940e-02 -1.44322038e-01 4.08222407e-01 1.02801926e-01
1.31702292e+00 7.13865578e-01 2.41263017e-01 2.50715226e-01
2.57840723e-01 6.92352653e-02 6.89466298e-01 3.68849933e-01
-4.99771861e-03 7.35264599e-01 5.43850303e-01 -2.19447330e-01
-1.12819874e+00 -9.21125650e-01 2.21192554e-01 1.00114691e+00
-2.16281310e-01 -5.45390844e-01 -7.03579009e-01 -9.45113301e-01
-3.23777288e-01 4.94814515e-01 -4.69538122e-01 -3.42673838e-01
-9.27139521e-01 -5.72640300e-01 5.21722794e-01 9.28095341e-01
3.75913292e-01 -1.02631366e+00 -6.08409882e-01 4.26452458e-02
-2.66092956e-01 -1.19611704e+00 -6.42604172e-01 2.15397403e-01
-8.65144908e-01 -1.15219104e+00 -8.56432319e-01 -6.96744442e-01
7.61992991e-01 -2.42705960e-02 9.98199224e-01 -5.73086143e-02
-3.74769866e-01 4.67440575e-01 -4.37464416e-01 -2.34113201e-01
-4.96155560e-01 1.04620114e-01 -7.70348087e-02 1.30928792e-02
-2.86670099e-03 -5.67707241e-01 -5.08527517e-01 3.52590501e-01
-9.90983605e-01 3.67797017e-01 7.73255765e-01 9.18318987e-01
7.41674840e-01 -3.95498216e-01 2.24043593e-01 -6.83042109e-01
1.89648151e-01 -2.25876898e-01 -3.80660415e-01 1.41415400e-02
-3.08647901e-01 3.02355349e-01 5.16387105e-01 -5.49225032e-01
-7.92377412e-01 6.69296026e-01 -3.29145104e-01 -7.38982499e-01
6.27178699e-03 4.20357168e-01 -2.12324485e-01 -6.20092489e-02
7.83169210e-01 1.93383321e-01 1.63610712e-01 -4.26938325e-01
3.99877995e-01 3.20229888e-01 4.72176582e-01 -8.30809772e-01
8.96199882e-01 4.92442131e-01 1.08381562e-01 -7.37749577e-01
-1.05100310e+00 -3.23802173e-01 -5.84481955e-01 -2.61241645e-01
5.96725047e-01 -8.74844491e-01 -4.09595698e-01 4.18073744e-01
-1.03928101e+00 -7.65481412e-01 -5.22518039e-01 5.70848942e-01
-1.02614045e+00 5.98383784e-01 -6.97336078e-01 -5.75258732e-01
-1.18135661e-01 -1.26498747e+00 1.20606220e+00 -2.86936849e-01
-5.23663998e-01 -6.04035556e-01 2.76394218e-01 6.81474268e-01
-6.22722171e-02 3.58124703e-01 5.42368770e-01 -4.78214592e-01
-7.28369057e-01 -4.96722385e-02 3.35697830e-02 8.15745354e-01
8.57014433e-02 -1.29998952e-01 -7.66107619e-01 -3.56287330e-01
-2.96046376e-01 -1.01419711e+00 8.43161881e-01 1.25739843e-01
1.22896194e+00 -5.16620457e-01 -1.00292586e-01 5.67403555e-01
9.87739444e-01 -4.84226570e-02 6.43910646e-01 2.83944458e-01
8.41860533e-01 4.79081243e-01 7.63196468e-01 4.84438539e-01
1.57928988e-01 9.93955612e-01 5.02569616e-01 8.44974164e-03
-2.63028353e-01 -3.71367991e-01 6.66202784e-01 5.34625411e-01
-1.95112288e-01 -3.66071425e-02 -9.58802760e-01 3.83156568e-01
-1.95096707e+00 -8.25606823e-01 1.46865174e-01 2.08471847e+00
1.05132329e+00 2.54351795e-01 3.24115962e-01 2.66690254e-01
4.36414391e-01 9.41744074e-02 -6.19245052e-01 1.00765422e-01
2.66975462e-01 4.14691389e-01 4.67777729e-01 5.60343206e-01
-1.19723320e+00 9.24534440e-01 5.14843512e+00 7.81258345e-01
-1.06469870e+00 -2.38853972e-02 6.26589239e-01 -3.43934983e-01
1.35717839e-02 1.10884160e-01 -7.32194960e-01 2.90001273e-01
7.37147868e-01 2.35325590e-01 3.08287829e-01 9.32251215e-01
2.40471512e-01 1.92145050e-01 -1.32350075e+00 8.75429630e-01
1.40384123e-01 -1.10108232e+00 6.29187748e-02 5.13947941e-02
7.49699295e-01 -2.24542078e-02 9.02940184e-02 3.22697848e-01
2.39015296e-01 -8.90265882e-01 8.05350304e-01 2.86380500e-01
6.67955816e-01 -6.50496304e-01 3.10428292e-01 3.39941829e-01
-9.66270864e-01 -4.45934013e-02 -2.00966746e-01 -1.97977096e-01
3.31193805e-01 3.67477983e-01 -1.04026568e+00 2.60584593e-01
3.97796690e-01 8.61301303e-01 -5.56936443e-01 9.72428441e-01
-5.34670532e-01 7.51783371e-01 -3.08306873e-01 1.66333735e-01
3.31947863e-01 2.77560744e-02 4.31260079e-01 9.13393021e-01
1.89632893e-01 -2.02560738e-01 5.03348649e-01 6.33360922e-01
-1.97202474e-01 9.02478620e-02 -4.46242958e-01 -2.01267406e-01
6.35738894e-02 1.13725412e+00 -6.15568578e-01 -2.70036280e-01
-2.79578865e-01 1.10015714e+00 5.51057577e-01 7.75689781e-02
-1.03679371e+00 1.36712655e-01 5.49075127e-01 3.55895817e-01
2.32179046e-01 -3.93286705e-01 -1.26315117e-01 -1.09898269e+00
4.53935087e-01 -1.33108473e+00 2.62680560e-01 -5.96703708e-01
-1.05427134e+00 2.37428322e-01 2.02726901e-01 -1.32810438e+00
-4.89760399e-01 -5.91549158e-01 -1.73849657e-01 4.16613489e-01
-1.19189060e+00 -1.09911942e+00 -2.20006034e-01 3.84199321e-01
7.93038845e-01 1.27306044e-01 6.36656582e-01 3.83908033e-01
-5.83723843e-01 6.20945632e-01 -3.00882548e-01 2.31255203e-01
6.99108005e-01 -1.21161449e+00 5.28888226e-01 7.76254117e-01
3.54661286e-01 2.85917580e-01 7.41751254e-01 -7.26076603e-01
-1.25038922e+00 -1.25059128e+00 4.79425400e-01 -3.76437038e-01
5.87372899e-01 -3.38679433e-01 -7.32756674e-01 9.57796693e-01
-8.14309791e-02 1.42134607e-01 4.51956749e-01 -1.03939608e-01
-3.89510870e-01 3.84605750e-02 -6.28664911e-01 7.19403923e-01
1.29131365e+00 -2.78732657e-01 -3.07801515e-01 7.36512721e-01
4.96592790e-01 -6.63287461e-01 -5.85584283e-01 5.63864529e-01
5.92782736e-01 -6.14258111e-01 9.69813526e-01 -7.34613359e-01
8.16364646e-01 -2.53394276e-01 -4.52741385e-02 -1.20769966e+00
2.68316977e-02 -5.90916097e-01 -3.12990636e-01 8.00563097e-01
3.99492949e-01 -3.32407326e-01 1.33504701e+00 4.74904090e-01
-2.67907083e-01 -1.04565716e+00 -7.58730352e-01 -9.97044802e-01
-3.30132395e-02 -4.87856925e-01 1.09169893e-02 7.20638573e-01
-2.20873252e-01 4.61716533e-01 -7.31684983e-01 -7.33713359e-02
5.69276989e-01 -3.03820908e-01 1.19858134e+00 -8.14845324e-01
-9.67833817e-01 -1.78136364e-01 -4.93256032e-01 -1.47493660e+00
2.37846404e-01 -8.80670965e-01 2.43531525e-01 -1.45396006e+00
3.64379615e-01 -3.07300419e-01 1.86941624e-01 9.39944625e-01
-2.34554499e-01 4.02299285e-01 4.23842430e-01 1.50888473e-01
-6.02294564e-01 7.38994122e-01 1.24748921e+00 -2.10081190e-01
-9.83457193e-02 6.26981258e-02 -3.64275903e-01 9.12559390e-01
1.03585637e+00 -5.35350382e-01 -6.11961663e-01 -3.70774835e-01
2.66242027e-01 3.89381908e-02 5.50728917e-01 -1.22089207e+00
-4.41824123e-02 -1.04694709e-01 2.30441287e-01 -4.86595720e-01
8.06455791e-01 -6.65260255e-01 -1.26288626e-02 5.99660397e-01
-6.35133624e-01 -1.62275974e-02 3.37712131e-02 5.21840036e-01
-5.04280925e-02 -3.83163303e-01 7.42898822e-01 -1.46508381e-01
-5.01603246e-01 3.07568669e-01 -8.27125460e-02 3.50337505e-01
1.04308963e+00 -1.11665912e-01 2.53072903e-02 -5.41836798e-01
-7.07348883e-01 1.71093956e-01 4.63902503e-01 2.97759026e-01
5.75748503e-01 -1.30137503e+00 -7.40857482e-01 -1.11486852e-01
8.82709995e-02 1.29078820e-01 -6.74462616e-02 8.13802540e-01
-6.36604130e-01 9.34850723e-02 -2.87334979e-01 -6.51116610e-01
-1.47246003e+00 3.15661132e-01 3.28783840e-01 -4.77428377e-01
-6.49906695e-01 8.73780370e-01 2.09430307e-01 -3.94078195e-01
5.42864561e-01 -2.09406227e-01 1.71287373e-01 -1.39570847e-01
3.42101842e-01 5.43255627e-01 -1.92723479e-02 -5.01472414e-01
-1.84398726e-01 4.27949876e-01 -2.65905559e-01 -2.26112604e-01
1.36203718e+00 2.70242095e-01 3.19877148e-01 3.03421378e-01
1.32027936e+00 -1.50549740e-01 -1.68694651e+00 -9.66503099e-02
1.33830264e-01 -4.55009937e-01 -4.04432118e-01 -5.81689715e-01
-1.06556177e+00 7.92434156e-01 4.85033751e-01 -3.86016518e-01
8.85592341e-01 -2.35062931e-02 1.00498950e+00 5.91540813e-01
3.03838760e-01 -1.11802006e+00 8.81540477e-01 4.68867689e-01
1.02310264e+00 -1.14917803e+00 -1.29507538e-02 -6.39567316e-01
-6.14686728e-01 9.85480726e-01 7.68202603e-01 -2.45544463e-01
1.93608478e-01 2.74126917e-01 -2.48774420e-02 -9.40749571e-02
-6.45814240e-01 -1.70986012e-01 2.63736397e-01 2.24502981e-01
4.24411893e-01 -5.70605546e-02 -2.44501770e-01 3.19758579e-02
-1.15980253e-01 1.68951839e-01 4.89667714e-01 1.29113543e+00
-2.92828232e-01 -1.49729049e+00 -3.14901024e-01 3.50275338e-01
-4.45556223e-01 8.10928717e-02 -5.67070484e-01 7.23379195e-01
-3.45770060e-03 6.31710410e-01 -2.68486083e-01 -3.92317116e-01
3.57630581e-01 1.77897409e-01 7.29687214e-01 -9.08090055e-01
-2.02810630e-01 8.69897977e-02 3.40236783e-01 -8.02456558e-01
-4.80760872e-01 -8.51170957e-01 -1.36859798e+00 2.20956564e-01
-1.83793858e-01 -1.79501787e-01 4.04453665e-01 9.65795636e-01
1.95744917e-01 5.52441001e-01 3.66374820e-01 -9.44734573e-01
-1.09968901e+00 -8.77166986e-01 -1.28033340e-01 4.96839643e-01
6.13458008e-02 -8.66360545e-01 -2.12276205e-01 1.70382932e-01] | [8.469904899597168, 0.6107240915298462] |
259c46f7-19da-4ec5-bfee-ad3db6f98ff4 | learning-visual-affordance-grounding-from | 2108.05675 | null | https://arxiv.org/abs/2108.05675v1 | https://arxiv.org/pdf/2108.05675v1.pdf | Learning Visual Affordance Grounding from Demonstration Videos | Visual affordance grounding aims to segment all possible interaction regions between people and objects from an image/video, which is beneficial for many applications, such as robot grasping and action recognition. However, existing methods mainly rely on the appearance feature of the objects to segment each region of the image, which face the following two problems: (i) there are multiple possible regions in an object that people interact with; and (ii) there are multiple possible human interactions in the same object region. To address these problems, we propose a Hand-aided Affordance Grounding Network (HAGNet) that leverages the aided clues provided by the position and action of the hand in demonstration videos to eliminate the multiple possibilities and better locate the interaction regions in the object. Specifically, HAG-Net has a dual-branch structure to process the demonstration video and object image. For the video branch, we introduce hand-aided attention to enhance the region around the hand in each video frame and then use the LSTM network to aggregate the action features. For the object branch, we introduce a semantic enhancement module (SEM) to make the network focus on different parts of the object according to the action classes and utilize a distillation loss to align the output features of the object branch with that of the video branch and transfer the knowledge in the video branch to the object branch. Quantitative and qualitative evaluations on two challenging datasets show that our method has achieved stateof-the-art results for affordance grounding. The source code will be made available to the public. | ['DaCheng Tao', 'Yang Cao', 'Jing Zhang', 'Wei Zhai', 'Hongchen Luo'] | 2021-08-12 | null | null | null | null | ['video-to-image-affordance-grounding'] | ['computer-vision'] | [ 1.26286075e-01 6.21209070e-02 -2.49867737e-01 -2.55975127e-01
-6.60990775e-02 -1.69261634e-01 2.67524570e-01 -3.61117452e-01
-3.35220337e-01 3.17588359e-01 3.37116510e-01 9.57521349e-02
-8.93975198e-02 -3.89311701e-01 -8.08721721e-01 -5.75519919e-01
1.30786806e-01 2.17137828e-01 5.32525957e-01 -1.84198543e-01
1.80548981e-01 5.30143201e-01 -1.50827563e+00 5.29864848e-01
8.30727875e-01 1.08931792e+00 7.82956600e-01 2.46150851e-01
-3.49634215e-02 8.20464075e-01 -3.79399836e-01 1.46581784e-01
3.89093518e-01 -4.40715313e-01 -8.40508640e-01 4.64841425e-01
5.03860354e-01 -9.69574749e-01 -7.65933990e-01 1.09656465e+00
2.05644190e-01 3.87122542e-01 4.17803466e-01 -1.64500475e+00
-5.25938570e-01 4.08035666e-01 -7.30437160e-01 1.52258888e-01
5.16070604e-01 5.36859155e-01 9.22733426e-01 -7.14491367e-01
6.60471439e-01 1.59799492e+00 5.99587858e-02 6.03103220e-01
-8.15030932e-01 -6.76987410e-01 6.75308287e-01 5.62741458e-01
-1.13993907e+00 -2.36536622e-01 8.48208547e-01 -4.55292135e-01
8.84878635e-01 -6.08723238e-03 9.08360422e-01 9.22670603e-01
-5.16816899e-02 1.36455846e+00 6.37838483e-01 -2.45300189e-01
-2.16236904e-01 -2.42374048e-01 -1.42228650e-02 9.06557560e-01
-1.07962087e-01 1.77139655e-01 -4.43277508e-01 3.53550822e-01
1.17655563e+00 6.78975821e-01 -5.88470638e-01 -5.60322285e-01
-1.60152602e+00 5.18052518e-01 9.51957464e-01 2.12677300e-01
-6.37644053e-01 1.94762960e-01 1.13935977e-01 -6.39875382e-02
-5.12192436e-02 3.04946065e-01 -4.14294630e-01 1.46524563e-01
-4.60388690e-01 4.41122293e-01 4.47443992e-01 1.06504023e+00
8.03824365e-01 -3.00337434e-01 -4.44909185e-01 5.15137792e-01
4.89227355e-01 2.73940176e-01 2.11962149e-01 -1.19604218e+00
9.47555542e-01 1.06459236e+00 2.53099501e-01 -9.44111586e-01
-4.06883746e-01 1.13735951e-01 -3.82737070e-01 3.32306623e-01
6.27270162e-01 9.91121083e-02 -1.12931812e+00 1.67190135e+00
6.36441886e-01 6.41989186e-02 -1.80169627e-01 1.49373972e+00
9.16568577e-01 5.01696706e-01 6.59171194e-02 1.46875352e-01
1.42126131e+00 -1.51224959e+00 -6.28366709e-01 -4.33982223e-01
4.36732084e-01 -4.30563122e-01 1.23669887e+00 -2.63471510e-02
-8.33983123e-01 -6.93210959e-01 -9.93030250e-01 -3.28208476e-01
-1.47970036e-01 4.61775005e-01 6.07530296e-01 -1.39731765e-01
-4.94056463e-01 5.83975375e-01 -9.37664509e-01 -2.89262682e-01
4.35597867e-01 2.88272083e-01 -5.04083157e-01 -3.12400967e-01
-9.89455402e-01 6.97400570e-01 5.32629490e-01 4.57424581e-01
-9.91489708e-01 -2.90517241e-01 -1.15477037e+00 1.26172319e-01
7.95237064e-01 -6.54115140e-01 9.59463179e-01 -1.22189987e+00
-1.09984195e+00 4.44953293e-01 -9.49040204e-02 -4.24279794e-02
6.01437271e-01 -5.29852569e-01 4.97620031e-02 4.85729247e-01
2.07306370e-01 1.05443144e+00 7.48756409e-01 -1.28316534e+00
-8.96351278e-01 -5.24295807e-01 4.47255433e-01 4.92807120e-01
-3.07532158e-02 1.33874174e-02 -6.94174111e-01 -8.53509426e-01
3.92471999e-01 -8.26913536e-01 -1.59294195e-02 3.82031083e-01
-7.10801065e-01 -5.74113607e-01 1.13187182e+00 -9.52283919e-01
8.52547705e-01 -2.29327941e+00 5.54495692e-01 -5.18138818e-02
1.96257100e-01 1.56360686e-01 -3.73166561e-01 6.25481755e-02
1.49150938e-02 -1.87881574e-01 1.27339838e-02 -2.30521068e-01
-1.03792645e-01 3.22030127e-01 -1.46406740e-01 4.37411696e-01
1.42622098e-01 1.01529050e+00 -9.38414037e-01 -5.01230657e-01
5.14694929e-01 4.05029118e-01 -5.38834095e-01 5.62892318e-01
-3.77988696e-01 5.82133114e-01 -6.32933438e-01 7.42141306e-01
3.29657644e-01 -1.15812778e-01 -8.92096087e-02 -5.09167373e-01
-2.64340341e-02 1.82311058e-01 -1.27780426e+00 1.88443983e+00
1.19018788e-02 4.35917825e-01 1.10748768e-01 -7.28614151e-01
4.51863945e-01 1.55890480e-01 4.78208482e-01 -4.12259340e-01
1.74180388e-01 1.06227852e-01 3.72715920e-01 -9.96211290e-01
1.06155962e-01 3.13725382e-01 3.14413071e-01 4.34644550e-01
1.91929117e-01 2.19156772e-01 1.16592273e-01 1.00375444e-01
7.39889324e-01 5.15795112e-01 8.05442259e-02 1.34343997e-01
3.45770597e-01 -2.15234175e-01 5.22902727e-01 4.59878772e-01
-3.98637354e-01 5.72912812e-01 3.63076389e-01 -5.60493469e-01
-9.19655859e-01 -8.76114070e-01 4.74936754e-01 1.05321646e+00
7.78655648e-01 -1.67263195e-01 -7.18994796e-01 -8.10031056e-01
-3.38752121e-02 3.62796992e-01 -7.86977291e-01 -9.40769985e-02
-7.53488779e-01 4.01572473e-02 -1.89760864e-01 9.00997996e-01
7.63442576e-01 -1.65805626e+00 -9.94823813e-01 -1.10562056e-01
-5.39856970e-01 -1.21504235e+00 -9.04100597e-01 -1.20679818e-01
-6.01964831e-01 -1.35843325e+00 -6.82328343e-01 -8.89890194e-01
9.28144097e-01 5.62534451e-01 3.65695477e-01 3.80137891e-01
-2.75800139e-01 2.73129940e-01 -4.24019605e-01 -3.92160453e-02
9.11739990e-02 -1.11508101e-01 4.61738817e-02 1.20298065e-01
2.16263413e-01 -2.06558079e-01 -7.90090024e-01 4.04390544e-01
-6.39969587e-01 3.05787325e-01 6.88940942e-01 8.60407352e-01
3.44241738e-01 -5.05593717e-02 1.69452414e-01 5.28089218e-02
1.40832931e-01 -2.25188449e-01 -1.75306037e-01 3.62708211e-01
3.25115323e-02 3.30157429e-02 2.25646555e-01 -6.71724856e-01
-8.24550271e-01 2.90127069e-01 1.12797990e-01 -9.28400457e-01
-2.08633617e-01 1.62468895e-01 -7.02201009e-01 1.64926857e-01
1.69578239e-01 -6.15211613e-02 1.33592740e-01 -3.28076631e-01
2.26420686e-01 5.40488183e-01 5.14878571e-01 -4.98117030e-01
4.09504712e-01 4.78040546e-01 -2.57297784e-01 -5.48464477e-01
-8.37393761e-01 -5.02169609e-01 -8.12040389e-01 -5.08096039e-01
1.31218112e+00 -6.78595662e-01 -8.59630346e-01 4.27577168e-01
-1.35218620e+00 -4.84738737e-01 -9.69375372e-02 6.70253813e-01
-4.99079883e-01 3.83586526e-01 -5.62050462e-01 -6.82873487e-01
-1.36234388e-01 -1.51752126e+00 1.22068477e+00 5.01704454e-01
-1.53858140e-01 -3.91600132e-01 -7.69123316e-01 3.88376653e-01
-2.07739800e-01 1.07910648e-01 8.38282108e-01 -5.21418571e-01
-9.13387775e-01 -7.59649575e-02 -5.17082989e-01 2.62534738e-01
3.83718729e-01 -1.71310112e-01 -5.54829478e-01 -2.55224615e-01
-2.13396877e-01 -3.18298697e-01 9.47831094e-01 4.56066340e-01
1.07959974e+00 -4.22539681e-01 -5.60894668e-01 4.82013047e-01
8.32024455e-01 3.23357821e-01 6.06009245e-01 2.81087458e-01
1.15050280e+00 1.06383812e+00 8.50979030e-01 1.82953060e-01
2.99134195e-01 8.57080877e-01 6.78229392e-01 -5.17291054e-02
-2.45338142e-01 -4.82496113e-01 4.48544651e-01 3.00193392e-02
-3.17566454e-01 -1.32789630e-02 -6.13923192e-01 4.23623055e-01
-2.16810346e+00 -9.94872570e-01 7.53968358e-02 1.99533951e+00
4.49168593e-01 -3.88108902e-02 2.53554314e-01 -6.73931092e-02
8.44457209e-01 6.70377091e-02 -8.55185211e-01 2.22022295e-01
1.94362909e-01 -5.11311054e-01 4.89934683e-02 3.26551825e-01
-9.82625186e-01 8.99891734e-01 5.00651217e+00 5.41484594e-01
-1.04129004e+00 -1.09433204e-01 2.57731259e-01 -2.40867838e-01
8.65107477e-02 -2.97101978e-02 -6.78688467e-01 4.90978897e-01
-2.37303600e-01 4.21860129e-01 5.92691779e-01 7.68158078e-01
3.12088996e-01 -2.41429731e-01 -1.36567378e+00 9.26830828e-01
1.50004283e-01 -8.26542199e-01 1.44102633e-01 1.28492601e-02
4.67012316e-01 -2.42097639e-02 -2.28015453e-01 1.59650519e-01
-1.20196581e-01 -8.52583647e-01 9.57822740e-01 4.65889573e-01
4.70184833e-01 -4.96745408e-01 5.59043527e-01 4.85330582e-01
-1.30269384e+00 -4.92529154e-01 -1.18601963e-01 -1.71990052e-01
2.82487690e-01 -1.30412117e-01 -4.40102011e-01 3.49637300e-01
1.02406633e+00 7.07898140e-01 -3.27018499e-01 9.67621624e-01
-5.69552243e-01 -7.44271129e-02 -3.06432158e-01 6.40801862e-02
3.39365661e-01 -1.20695040e-01 6.98364198e-01 5.88505089e-01
2.79288322e-01 2.57730991e-01 5.72137475e-01 1.06597161e+00
1.49274170e-01 -1.48141786e-01 -4.11242187e-01 2.08859127e-02
3.88935119e-01 1.03327227e+00 -7.73828506e-01 -3.30529898e-01
-4.93380398e-01 1.22556746e+00 3.84516597e-01 7.02848196e-01
-7.54084826e-01 -4.98861879e-01 5.23942888e-01 2.14428484e-01
6.01105571e-01 -2.75963217e-01 5.23122475e-02 -1.13070583e+00
4.72142339e-01 -7.98323691e-01 3.72438699e-01 -1.13388216e+00
-9.28279161e-01 4.60073858e-01 1.79274127e-01 -1.29497874e+00
-1.32987264e-03 -8.49684894e-01 -5.23466170e-01 8.05213213e-01
-1.11042011e+00 -1.37909091e+00 -6.29909277e-01 5.96332848e-01
7.50205576e-01 -2.52232924e-02 2.94547677e-01 -3.79383564e-02
-4.99937296e-01 2.09972113e-01 -6.06173337e-01 4.33882475e-01
4.50805068e-01 -9.76750135e-01 4.49279137e-03 8.70625854e-01
5.75228035e-02 8.26617241e-01 4.57749993e-01 -7.69750595e-01
-1.08276594e+00 -8.93437743e-01 6.49606049e-01 -3.00344884e-01
3.20857733e-01 -3.43726188e-01 -8.12410951e-01 8.41356456e-01
1.15661032e-01 1.41244963e-01 3.84227931e-03 -1.75250143e-01
-2.76736498e-01 1.04114644e-01 -9.76557672e-01 8.67451727e-01
1.30414116e+00 -4.09915805e-01 -9.10299182e-01 2.08791524e-01
8.40089858e-01 -6.09614134e-01 -3.64085019e-01 5.03187895e-01
8.30514431e-01 -7.98727393e-01 9.83057618e-01 -6.83507204e-01
7.02777684e-01 -6.83230221e-01 -1.60611451e-01 -1.10060549e+00
-1.40150934e-01 -2.60107934e-01 -3.81705910e-01 9.58438873e-01
2.96975923e-04 -3.54051024e-01 6.04716599e-01 7.10692942e-01
-2.59055257e-01 -9.57010388e-01 -8.41864407e-01 -6.01937652e-01
-4.33793604e-01 -1.34037822e-01 5.95521390e-01 6.37680650e-01
1.60304829e-01 2.82882184e-01 -2.96277434e-01 1.51716366e-01
3.31912786e-01 3.36132675e-01 8.65218341e-01 -9.34816420e-01
-2.11180553e-01 -4.69351798e-01 -3.13695073e-01 -1.56543326e+00
2.96488255e-01 -6.02711856e-01 4.02001619e-01 -1.83110368e+00
4.04004991e-01 -2.99970418e-01 -1.76992029e-01 8.83268118e-01
-4.77854073e-01 2.17422396e-02 5.89894652e-01 4.00740355e-01
-5.82899988e-01 6.25133812e-01 1.91522217e+00 -3.06681991e-01
-3.79230857e-01 -1.56680956e-01 -3.36579800e-01 7.84289658e-01
4.83998269e-01 -1.55090079e-01 -3.68388474e-01 -4.43413138e-01
-2.48675957e-01 1.07114352e-01 8.69266748e-01 -8.89930964e-01
1.69594899e-01 -3.34230691e-01 5.70914984e-01 -6.28548443e-01
4.93192703e-01 -9.64781702e-01 -2.88870692e-01 6.48375392e-01
-2.89951533e-01 -5.30200042e-02 -3.26877050e-02 5.91492295e-01
-4.25736904e-02 -2.01343924e-01 5.15842974e-01 -2.74932057e-01
-9.89511907e-01 5.24968624e-01 -3.19018029e-02 -2.51472503e-01
1.24459589e+00 -3.03961635e-01 -2.51176685e-01 -3.14042687e-01
-6.94053352e-01 6.56220555e-01 4.83736604e-01 8.25347185e-01
8.17855835e-01 -1.22459948e+00 -3.74648213e-01 3.38317275e-01
1.02837041e-01 3.84883076e-01 5.17953098e-01 1.04268372e+00
-2.20285580e-01 1.82937548e-01 -4.17897224e-01 -6.87132716e-01
-1.28420663e+00 8.08811724e-01 5.27309775e-01 2.19652563e-01
-8.97376597e-01 8.58769953e-01 7.46542752e-01 -2.80728638e-01
4.60881054e-01 -6.22880876e-01 -2.96558738e-01 -2.12340325e-01
5.45631528e-01 2.91514426e-01 -5.77930450e-01 -9.00864661e-01
-3.07444185e-01 4.74716693e-01 -8.61603841e-02 7.51422942e-02
1.07788396e+00 -1.85671166e-01 -1.43203914e-01 2.32053861e-01
1.07254863e+00 -4.98779535e-01 -1.97560716e+00 -2.92198539e-01
-4.16221172e-01 -5.86635768e-01 -1.32242396e-01 -7.42515147e-01
-1.21451950e+00 9.41615760e-01 3.92771125e-01 -2.61918753e-01
9.34356272e-01 3.18138123e-01 7.85567522e-01 4.03152406e-01
3.01475495e-01 -1.03446758e+00 5.11355579e-01 3.54845881e-01
1.20591962e+00 -1.26545405e+00 -1.28740910e-02 -5.26644289e-01
-7.42752075e-01 1.04308748e+00 1.13642716e+00 -1.66557042e-03
3.35688889e-01 -1.25923201e-01 -7.37594143e-02 -2.70297945e-01
-2.55027533e-01 -2.91112155e-01 5.27944267e-01 5.33085525e-01
-2.63070259e-02 -5.29108830e-02 -8.82693604e-02 7.58245647e-01
2.35931993e-01 -1.08760551e-01 7.30980784e-02 8.93368244e-01
-4.65819091e-01 -6.39600933e-01 -2.69065768e-01 4.21874106e-01
-9.50591639e-02 1.24975279e-01 -4.14498210e-01 7.95525849e-01
4.27235126e-01 9.25223649e-01 1.00292251e-01 -4.55362439e-01
4.54621106e-01 -1.39014408e-01 7.15119779e-01 -6.58801854e-01
-2.85677314e-01 1.81983516e-01 -1.46489024e-01 -8.98593128e-01
-6.19158268e-01 -7.84873068e-01 -1.72033417e+00 8.74175504e-02
-3.86567920e-01 -1.58587158e-01 3.31377387e-01 1.34092593e+00
2.13472992e-01 6.64530098e-01 1.78074196e-01 -1.38428223e+00
-3.24044675e-01 -1.05579841e+00 -4.81982529e-01 5.46705484e-01
5.40286481e-01 -1.21849954e+00 -1.28551975e-01 -5.07225581e-02] | [5.17734956741333, -0.0705491453409195] |
59b92c2e-7b7e-4403-bc56-1c0d4e4d4531 | recognising-cardiac-abnormalities-in-wearable | 1807.04077 | null | http://arxiv.org/abs/1807.04077v1 | http://arxiv.org/pdf/1807.04077v1.pdf | Recognising Cardiac Abnormalities in Wearable Device Photoplethysmography (PPG) with Deep Learning | Cardiac abnormalities affecting heart rate and rhythm are commonly observed
in both healthy and acutely unwell people. Although many of these are benign,
they can sometimes indicate a serious health risk. ECG monitors are typically
used to detect these events in electrical heart activity, however they are
impractical for continuous long-term use. In contrast, current-generation
wearables with optical photoplethysmography (PPG) have gained popularity with
their low-cost, lack of wires and tiny size. Many cardiac abnormalities such as
ectopic beats and AF can manifest as both obvious and subtle anomalies in a PPG
waveform as they disrupt blood flow. We propose an automatic method for
recognising these anomalies in PPG signal alone, without the need for ECG. We
train an LSTM deep neural network on 400,000 clean PPG samples to learn typical
PPG morphology and rhythm, and flag PPG signal diverging from this as cardiac
abnormalities. We compare the cardiac abnormalities our approach recognises
with the ectopic beats recorded by a bedside ECG monitor for 29 patients over
47.6 hours of gold standard observations. Our proposed cardiac abnormality
recognition approach recognises 60%+ of ECG-detected PVCs in PPG signal, with a
false positive rate of 23% - demonstrating the compelling power and value of
this novel approach. Finally we examine how cardiac abnormalities manifest in
PPG signal for in- and out-of-hospital patient populations using a wearable
device during standard care. | [] | 2018-07-11 | null | null | null | null | ['photoplethysmography-ppg'] | ['medical'] | [ 5.24602652e-01 -2.73208432e-02 2.74869680e-01 -1.91267967e-01
-6.94699645e-01 -7.13755667e-01 -2.51803368e-01 2.28648245e-01
4.60983030e-02 7.84373283e-01 -2.67873015e-02 -6.47534192e-01
2.09553838e-02 -3.46494317e-01 -4.24178779e-01 -6.27586246e-01
-6.77899778e-01 1.54458597e-01 -4.00992453e-01 4.06156123e-01
-1.15884148e-01 5.32722235e-01 -6.38069749e-01 1.12802200e-01
6.31580412e-01 1.22853136e+00 -6.10864580e-01 1.20509565e+00
3.58924747e-01 3.30937147e-01 -1.14451730e+00 1.21682711e-01
4.32071030e-01 -1.05760157e+00 -2.52122194e-01 1.08275525e-02
6.62944615e-01 -4.69434798e-01 -4.56854291e-02 3.81287992e-01
1.39422166e+00 -6.15561008e-01 2.54051089e-01 -7.82989085e-01
-1.12228408e-01 1.88299358e-01 -5.19228935e-01 8.52579892e-01
3.79004151e-01 5.18381774e-01 4.40575182e-01 -4.93521810e-01
7.32938573e-02 2.81981558e-01 1.49088955e+00 4.42826986e-01
-1.33070195e+00 -4.32169378e-01 -6.58741474e-01 -1.97296381e-01
-9.87330377e-01 -5.48529565e-01 8.79856586e-01 -2.98011452e-01
1.10150909e+00 6.20251179e-01 1.32370758e+00 1.03439546e+00
7.37586200e-01 7.45387226e-02 1.02069581e+00 -3.21077079e-01
-2.71126460e-02 -2.83984423e-01 -9.81684681e-03 4.86151606e-01
5.37474215e-01 1.00430109e-01 -4.56000924e-01 -6.63913488e-01
9.47962999e-01 2.92446792e-01 -6.53017342e-01 2.23747537e-01
-1.37309933e+00 1.70084283e-01 -2.20827058e-01 3.45317036e-01
-6.68759286e-01 4.10021186e-01 6.13049507e-01 5.78900456e-01
2.07059219e-01 6.26726806e-01 -4.33871925e-01 -6.06547832e-01
-8.64994109e-01 -3.03030133e-01 8.71526778e-01 4.21233922e-01
1.17853675e-02 4.64152694e-01 -3.17834795e-01 6.99050605e-01
1.17318124e-01 5.71003854e-01 7.15459168e-01 -9.20100987e-01
-5.82224615e-02 4.24633980e-01 4.50090095e-02 -1.08858955e+00
-6.51575267e-01 -6.63653851e-01 -1.14408445e+00 -1.44367948e-01
4.05888945e-01 -5.36759555e-01 -8.40378881e-01 1.24719477e+00
8.43279734e-02 4.37142313e-01 -3.84263471e-02 7.42001474e-01
9.23183024e-01 1.88438520e-01 -2.42598411e-02 -5.81903517e-01
1.21740341e+00 -1.18014090e-01 -6.38660073e-01 -3.01521301e-01
5.58680415e-01 -4.22357172e-01 6.60986543e-01 4.14608181e-01
-1.08996606e+00 -2.80890882e-01 -1.08895326e+00 4.92784500e-01
2.38094360e-01 1.75044820e-01 2.63530314e-01 1.08017433e+00
-1.16795421e+00 9.53920245e-01 -8.67375672e-01 -4.47224796e-01
5.86099148e-01 3.89466107e-01 -1.13630533e-01 5.10057092e-01
-1.09402061e+00 6.63783371e-01 -2.01611482e-02 4.67651576e-01
-4.36204672e-01 -9.60361481e-01 -6.48167729e-01 3.58578227e-02
2.60966495e-02 -7.71263301e-01 9.66314912e-01 -5.94769955e-01
-1.23591793e+00 1.02501166e+00 -1.82546601e-01 -5.04682899e-01
5.00838637e-01 -1.73303798e-01 -8.16590846e-01 2.12964892e-01
-1.90198615e-01 -2.26839006e-01 1.02490103e+00 -6.39689684e-01
-2.13568643e-01 -3.45706433e-01 -6.05506957e-01 -9.91235226e-02
2.37484589e-01 -8.11140537e-02 2.50646502e-01 -5.28201997e-01
4.40855026e-01 -7.55410016e-01 1.09128456e-03 -9.29232836e-02
-3.62068504e-01 4.60767984e-01 5.72532177e-01 -8.88563514e-01
1.24661827e+00 -2.09332490e+00 -8.09009075e-01 5.44937193e-01
6.42656744e-01 5.50535023e-01 3.29948395e-01 2.81440109e-01
-1.85725048e-01 4.47715700e-01 -4.71966237e-01 2.44875073e-01
-3.79259795e-01 1.99943855e-01 -3.11626673e-01 7.86226034e-01
1.33132666e-01 1.27830338e+00 -7.10943580e-01 -3.14844251e-02
2.57179916e-01 5.77576101e-01 1.06940813e-01 2.66179472e-01
5.12296319e-01 7.29454696e-01 1.01411968e-01 9.00233626e-01
3.57500374e-01 -4.64557737e-01 3.93913746e-01 -1.51208401e-01
2.63793737e-01 3.89879048e-01 -9.44220185e-01 1.38475227e+00
-4.32254560e-02 7.51381516e-01 -1.97723314e-01 -9.72074747e-01
1.14806175e+00 9.10581172e-01 4.69232231e-01 -5.95876813e-01
9.84830931e-02 4.59740579e-01 7.04602957e-01 -7.45545268e-01
-5.17732263e-01 -6.30551875e-01 3.05804282e-01 7.88158536e-01
-3.41943145e-01 9.80634913e-02 -2.37237602e-01 -2.53153652e-01
1.47154284e+00 -1.37542278e-01 7.60610461e-01 -2.79366583e-01
3.33189309e-01 -5.25873661e-01 7.74107218e-01 1.15186667e+00
-4.98530954e-01 1.01710999e+00 5.57144344e-01 -1.20023656e+00
-6.62769735e-01 -9.87677932e-01 -2.00008780e-01 1.95663691e-01
-3.18600923e-01 -1.75926596e-01 -2.10775778e-01 -4.08008069e-01
4.87683378e-02 -4.43130219e-03 -3.64309579e-01 -1.90486521e-01
-8.92784774e-01 -1.03397024e+00 1.11433816e+00 7.39256024e-01
3.32062602e-01 -1.35226417e+00 -1.47779810e+00 8.16580176e-01
-4.60260287e-02 -7.61300147e-01 -2.37540469e-01 5.49907833e-02
-1.09647262e+00 -1.10240626e+00 -9.87871885e-01 -4.78381336e-01
3.82151365e-01 -4.39186215e-01 1.36665165e+00 2.87779987e-01
-1.08946371e+00 5.73770165e-01 2.18182653e-02 -7.06024885e-01
-4.62707639e-01 -3.93215895e-01 2.99065262e-01 1.61139607e-01
4.37005103e-01 -1.06621981e+00 -1.03064215e+00 4.48395573e-02
-1.50262997e-01 -4.67690676e-01 6.41235292e-01 5.48867941e-01
4.11410302e-01 -6.93367481e-01 9.61715817e-01 -7.26324856e-01
1.09282207e+00 -3.21670800e-01 -3.53520364e-02 -1.04045309e-02
-8.84363234e-01 -6.57573223e-01 4.97459054e-01 -3.39510411e-01
-2.87378609e-01 -2.94375539e-01 -2.64324769e-02 -4.16034132e-01
-5.98972559e-01 2.69919008e-01 4.56687599e-01 -3.73937637e-02
8.52023125e-01 3.84389788e-01 1.38776109e-01 -1.46462604e-01
-5.19892514e-01 5.52824795e-01 9.73969162e-01 -4.61557209e-01
4.78518039e-01 3.41073692e-01 2.21037135e-01 -1.03309870e+00
-3.37118149e-01 -4.57534790e-01 -3.98261398e-01 -1.60078347e-01
5.04207075e-01 -6.56553745e-01 -8.31883311e-01 4.55449432e-01
-1.11994016e+00 -1.66812062e-01 -5.67267179e-01 3.89030963e-01
-1.47231713e-01 6.29544914e-01 -5.83586931e-01 -1.12915981e+00
-1.05415714e+00 -4.51126933e-01 7.45991230e-01 1.17561601e-01
-8.76068234e-01 -1.02751744e+00 2.94689238e-01 -3.06903064e-01
7.18352795e-01 1.14156651e+00 5.98208487e-01 -6.84605539e-01
-2.61686556e-02 -5.21209300e-01 3.52404416e-02 4.25047129e-01
7.02435493e-01 -2.04194799e-01 -1.16142428e+00 -3.81134033e-01
2.99408317e-01 2.66759872e-01 2.80189097e-01 8.12900186e-01
9.46547329e-01 -2.16540620e-01 -2.16997877e-01 8.78438473e-01
1.11881638e+00 8.89177442e-01 7.64929950e-01 -1.25924125e-01
7.51485765e-01 2.59897718e-03 -2.89164394e-01 4.35309887e-01
-1.65952176e-01 -1.62376035e-02 -1.38593614e-01 -6.14260316e-01
1.04003735e-01 2.12808639e-01 2.87213475e-01 6.22705519e-01
-5.13058126e-01 9.74006876e-02 -1.17613029e+00 5.75047016e-01
-1.36590254e+00 -8.81477714e-01 -4.87858057e-01 2.41454244e+00
7.39461720e-01 -1.14417551e-02 3.52729142e-01 4.09309775e-01
7.74023831e-01 -1.05391294e-01 -8.29984426e-01 -8.52277577e-01
-1.53942347e-01 5.87950170e-01 2.69413888e-01 -2.89830510e-02
-8.31960022e-01 -1.88450977e-01 7.16786432e+00 -4.42248225e-01
-1.76757729e+00 -1.00022733e-01 9.20725226e-01 -1.95167914e-01
9.02579501e-02 -3.36272359e-01 -2.25726604e-01 6.74736321e-01
1.16881251e+00 -1.70471728e-01 -2.55722404e-02 3.21668118e-01
3.11473727e-01 1.82075918e-01 -1.38435793e+00 1.67238367e+00
2.02189267e-01 -1.34481144e+00 -5.59429526e-01 -9.46719721e-02
1.07195996e-01 -1.95233859e-02 -2.78167874e-01 -1.20308712e-01
-7.77960181e-01 -1.23123062e+00 4.73571792e-02 6.63652718e-01
1.51463306e+00 -2.25474924e-01 7.69253552e-01 1.24267610e-02
-8.24401200e-01 1.60770476e-01 7.21765682e-02 -2.49994263e-01
8.89825597e-02 9.23752964e-01 -9.91333663e-01 1.17779754e-01
6.86799645e-01 7.24330306e-01 -3.29568982e-01 1.27241027e+00
-1.81227133e-01 1.24742985e+00 -6.93866789e-01 3.97627473e-01
-3.22762549e-01 -2.48395205e-02 9.23721015e-01 1.17593908e+00
6.34179354e-01 4.01220322e-01 -1.27788901e-01 1.10019636e+00
2.31656194e-01 -1.24769382e-01 -1.02612889e+00 -2.76665040e-03
2.15380207e-01 1.19629920e+00 -7.72446215e-01 -5.17979801e-01
-3.13799798e-01 6.20590210e-01 -6.38612986e-01 3.51148874e-01
-6.37761176e-01 -6.67538345e-01 2.21851170e-01 4.91530001e-01
-2.80277073e-01 3.70467812e-01 -7.37687528e-01 -9.92746472e-01
6.18253112e-01 -1.10893559e+00 4.65151340e-01 -4.01223898e-01
-1.16635799e+00 6.41079724e-01 -5.80831409e-01 -1.38217974e+00
-6.74856782e-01 -2.42930368e-01 -1.33831978e+00 1.12808251e+00
-1.06756759e+00 -4.86531109e-01 -4.73573446e-01 2.96776086e-01
-7.68481046e-02 2.85942927e-02 1.28110051e+00 2.41540685e-01
-3.25936496e-01 4.85178441e-01 -5.91378987e-01 2.19468817e-01
7.15270758e-01 -1.52892184e+00 6.84891760e-01 8.34990084e-01
-1.72050983e-01 7.99300134e-01 4.07142282e-01 -6.28722310e-01
-1.13357794e+00 -8.06315958e-01 1.02187622e+00 -4.22268301e-01
7.27311075e-02 -5.28887175e-02 -9.69693840e-01 5.39712310e-01
1.65543169e-01 5.40551841e-01 9.15782928e-01 -1.91509679e-01
1.56671911e-01 -2.84188956e-01 -1.17669892e+00 1.70708358e-01
6.89499736e-01 -4.59765047e-01 -8.31515312e-01 4.03947420e-02
-7.82463476e-02 -5.43527305e-01 -1.18529344e+00 6.63715243e-01
9.74918544e-01 -1.08846903e+00 7.97705293e-01 -4.00721639e-01
6.29957169e-02 -8.48023146e-02 6.17111206e-01 -1.31535506e+00
-1.99178010e-01 -1.62819386e+00 -3.19731921e-01 9.06582117e-01
2.61553258e-01 -1.24256301e+00 7.69823015e-01 7.77698994e-01
-2.24389598e-01 -8.31463516e-01 -7.19428778e-01 -5.39009333e-01
-3.57035398e-01 -2.61826545e-01 2.08936140e-01 1.03659105e+00
2.44496152e-01 2.29836032e-01 -4.31202590e-01 -8.45520124e-02
6.19297206e-01 7.39474297e-02 2.44023055e-01 -1.47433972e+00
-5.52968562e-01 -1.18458323e-01 -5.29227614e-01 -1.60157830e-01
-6.83068871e-01 -6.88724041e-01 -1.27933398e-01 -1.48296154e+00
-1.20096534e-01 -2.93010473e-01 -5.80829084e-01 6.52059197e-01
-2.28892177e-01 7.40281641e-01 -2.32546940e-01 -6.22769184e-02
1.20204791e-01 -3.62930983e-01 7.91164875e-01 2.48426702e-02
-7.27041006e-01 1.20287664e-01 -6.98885381e-01 8.74886692e-01
8.88216436e-01 -4.72877353e-01 -3.02173048e-01 4.28001545e-02
3.76384199e-01 4.26385820e-01 4.88665134e-01 -1.19187701e+00
-8.92213359e-02 4.97258633e-01 9.01264906e-01 -2.32349843e-01
-2.09183265e-02 -4.60022390e-01 3.53679329e-01 7.74467051e-01
6.98641762e-02 4.59936112e-01 4.47014123e-01 1.59335285e-01
-1.93562016e-01 2.49368072e-01 7.12556958e-01 -4.70293462e-01
-1.05085010e-02 3.09782922e-01 -7.04503536e-01 5.39882660e-01
7.14086413e-01 -6.42497122e-01 -2.36656085e-01 -6.10611677e-01
-1.07618868e+00 -2.52887428e-01 -8.83888230e-02 -6.07303865e-02
8.34500134e-01 -1.05354774e+00 -9.16590631e-01 6.02513373e-01
-7.41348788e-02 -9.37312003e-03 2.72911042e-01 1.64426017e+00
-8.30436349e-01 1.13742091e-01 -2.66532034e-01 -1.03407121e+00
-1.26045728e+00 -1.55808315e-01 8.82379711e-01 1.76066652e-01
-1.40510213e+00 6.16435587e-01 -1.60915241e-01 2.99137294e-01
1.83174238e-01 -5.29353559e-01 1.33650184e-01 -1.10237136e-01
6.28803670e-01 2.85688430e-01 4.39696431e-01 1.73104573e-02
-5.42418242e-01 5.55627346e-01 2.63126731e-01 3.66663903e-01
9.90698040e-01 -5.85904866e-02 -1.76652759e-01 6.27129138e-01
8.02314103e-01 -6.27410710e-02 -6.12523198e-01 2.19772890e-01
-1.85538962e-01 -2.64550030e-01 -2.77032971e-01 -1.11427128e+00
-1.12089586e+00 1.05060506e+00 1.12158847e+00 6.01846337e-01
1.16354167e+00 -3.50098789e-01 9.86863136e-01 2.87816495e-01
6.51383325e-02 -5.55653512e-01 -2.75839508e-01 -1.46376818e-01
6.38434589e-01 -7.75188863e-01 -6.94464967e-02 1.42925948e-01
-4.53871161e-01 1.26598573e+00 9.46583077e-02 -3.50881368e-01
5.00533938e-01 3.94808650e-01 6.69024050e-01 -4.17157233e-01
-4.88292783e-01 3.74875933e-01 1.38601288e-01 7.82667398e-01
7.68162191e-01 5.96449226e-02 -3.23009610e-01 4.17962223e-01
1.04435654e-02 3.15540493e-01 9.06422794e-01 1.00213444e+00
-8.00646376e-03 -6.95144594e-01 -1.84194833e-01 1.09720588e+00
-1.19998395e+00 -2.28542566e-01 -4.23607141e-01 3.96244496e-01
9.09021646e-02 8.18388402e-01 6.45153895e-02 1.15881003e-02
4.56155390e-01 7.02349663e-01 5.01516104e-01 -5.76544344e-01
-8.68950307e-01 4.83801752e-01 2.59318084e-01 -5.65257192e-01
-2.60284752e-01 -6.44220591e-01 -9.10632730e-01 3.07241291e-01
1.95308238e-01 -1.62929162e-01 3.22098255e-01 7.67431974e-01
7.06122458e-01 7.00914145e-01 2.36571953e-01 -4.32277173e-01
-4.17591274e-01 -1.08979952e+00 -9.53034699e-01 3.87181163e-01
1.01995361e+00 5.70338927e-02 -6.63224638e-01 3.80590767e-01] | [14.170363426208496, 3.1665008068084717] |
61fe9101-6857-48ef-8106-3b7e5fd99edd | modern-gpr-target-recognition-methods | 2211.01277 | null | https://arxiv.org/abs/2211.01277v1 | https://arxiv.org/pdf/2211.01277v1.pdf | Modern GPR Target Recognition Methods | Traditional GPR target recognition methods include pre-processing the data by removal of noisy signatures, dewowing (high-pass filtering to remove low-frequency noise), filtering, deconvolution, migration (correction of the effect of survey geometry), and can rely on the simulation of GPR responses. The techniques usually suffer from the loss of information, inability to adapt from prior results, and inefficient performance in the presence of strong clutter and noise. To address these challenges, several advanced processing methods have been developed over the past decade to enhance GPR target recognition. In this chapter, we provide an overview of these modern GPR processing techniques. In particular, we focus on the following methods: adaptive receive processing of range profiles depending on the target environment; adoption of learning-based methods so that the radar utilizes the results from prior measurements; application of methods that exploit the fact that the target scene is sparse in some domain or dictionary; application of advanced classification techniques; and convolutional coding which provides succinct and representatives features of the targets. We describe each of these techniques or their combinations through a representative application of landmine detection. | ['Maria Antonia Gonzalez-Huici', 'Kumar Vijay Mishra', 'Fabio Giovanneschi'] | 2022-11-02 | null | null | null | null | ['landmine', 'gpr', 'gpr'] | ['computer-vision', 'computer-vision', 'miscellaneous'] | [ 7.30862975e-01 -3.07908863e-01 5.79504907e-01 -6.31078899e-01
-8.04005206e-01 -8.76006708e-02 5.01767516e-01 2.69422472e-01
-4.23742175e-01 4.76846099e-01 4.30623978e-01 -1.09502554e-01
-4.35998112e-01 -9.96398211e-01 -2.75153488e-01 -7.15217054e-01
-4.02747631e-01 3.12009215e-01 3.12372029e-01 -5.43898284e-01
4.66953576e-01 9.61149335e-01 -1.69188082e+00 2.65674174e-01
6.26353979e-01 1.23262703e+00 3.06374252e-01 5.17625391e-01
-4.06197459e-02 5.78851938e-01 -4.85445082e-01 2.77166128e-01
1.99008465e-01 -9.42519233e-02 -1.25044405e-01 -1.79356933e-01
-8.67188945e-02 -8.16115141e-02 -3.54430884e-01 8.69251847e-01
7.39673734e-01 1.46205932e-01 7.77890921e-01 -3.22282463e-01
-2.01287761e-01 3.92361671e-01 -7.09062994e-01 5.04352450e-01
4.33284611e-01 -4.68838274e-01 2.38313004e-01 -1.19636011e+00
-4.83047999e-02 9.60451603e-01 1.37955725e+00 2.26701703e-02
-1.03475761e+00 -4.56689298e-01 -2.70480841e-01 2.62788564e-01
-1.26976299e+00 -6.99970424e-01 6.84749544e-01 -4.14225250e-01
8.03268969e-01 2.55562454e-01 3.94939274e-01 7.36176729e-01
1.03068255e-01 3.37583333e-01 9.35781300e-01 -8.05843711e-01
2.90400386e-01 -2.42393121e-01 3.92005026e-01 3.13919812e-01
3.80359590e-01 6.14988029e-01 -3.72479230e-01 -4.97312814e-01
3.06751221e-01 -1.18349262e-01 -4.93632019e-01 -6.81292713e-02
-5.76523900e-01 9.35802519e-01 2.29338095e-01 4.14562136e-01
-8.31203640e-01 -2.16209769e-01 3.40009630e-01 1.50175139e-01
3.64706457e-01 2.76608974e-01 -2.27736577e-01 1.93757325e-01
-1.23744214e+00 2.64272302e-01 7.87306726e-01 3.87588024e-01
8.37870419e-01 5.55496514e-01 3.52605172e-02 8.30151558e-01
5.82228243e-01 7.51332402e-01 3.96831244e-01 -2.60663003e-01
1.12456322e-01 8.56721178e-02 1.39154390e-01 -9.97031569e-01
-6.26380026e-01 -4.94315267e-01 -7.32493341e-01 1.77028626e-01
-4.39149560e-03 -3.26297522e-01 -1.23324668e+00 1.11954355e+00
2.81227648e-01 2.77339131e-01 4.41958755e-01 5.57964385e-01
9.46254849e-01 5.73623478e-01 4.09322185e-03 -4.26879153e-02
1.02731347e+00 -1.38207719e-01 -6.95662379e-01 -5.10684729e-01
2.95257270e-01 -9.09677565e-01 1.23264596e-01 4.05283600e-01
-4.59035784e-01 -6.55801237e-01 -1.10062957e+00 8.65116417e-01
-4.69193339e-01 2.08922923e-01 6.39606774e-01 1.09077108e+00
-6.73482418e-01 5.12604296e-01 -7.21414804e-01 -3.27357411e-01
1.48812950e-01 3.32570255e-01 -6.22131005e-02 -3.58546764e-01
-1.17131686e+00 1.23423469e+00 1.59061551e-01 6.46714091e-01
-6.17844105e-01 -7.44212508e-01 -9.77321982e-01 -1.75592333e-01
-2.79789954e-01 -1.84597760e-01 8.41713250e-01 -7.86885321e-01
-1.25169086e+00 4.26266700e-01 2.44516775e-01 -5.86085677e-01
1.75689146e-01 -4.94600356e-01 -8.56958747e-01 7.21019208e-02
-3.46222296e-02 -2.90013790e-01 1.03961551e+00 -1.01697075e+00
-7.00210989e-01 -5.02847016e-01 -6.72634780e-01 5.36154211e-02
2.75957495e-01 1.08738057e-01 1.78959444e-01 -5.54869533e-01
6.27069950e-01 -5.26573539e-01 -4.74801391e-01 -3.80158603e-01
1.77068129e-01 6.06909811e-01 9.32920277e-01 -8.60951304e-01
6.80784345e-01 -2.31990623e+00 -6.08579457e-01 5.34768820e-01
-2.88576543e-01 3.87861133e-01 8.80163256e-03 7.62657344e-01
-1.71461746e-01 -7.88011968e-01 -4.89663541e-01 2.58154478e-02
-4.01090682e-01 1.37665287e-01 -4.50195730e-01 8.25302660e-01
-2.79209428e-02 2.33895019e-01 -5.27861118e-01 -2.26243120e-02
5.12419343e-01 7.62925208e-01 -2.25690708e-01 5.51419631e-02
4.06788588e-01 4.07527477e-01 -5.75666428e-01 6.94395304e-01
1.21260130e+00 6.47623420e-01 -2.26773202e-01 -6.49170756e-01
-3.78828168e-01 1.02314793e-01 -1.50682271e+00 9.50777352e-01
-4.24043298e-01 4.78134692e-01 5.52521408e-01 -1.50710702e+00
1.35984182e+00 1.20380700e-01 6.13191903e-01 -5.93211174e-01
1.63718253e-01 4.91556525e-01 -1.07300460e-01 -6.24946654e-01
4.41905886e-01 -3.03306133e-01 3.54485326e-02 1.15099899e-01
-8.75690728e-02 2.78596883e-03 -4.09187943e-01 -3.50322962e-01
1.24610531e+00 -8.49188268e-02 2.81611949e-01 -2.12784156e-01
4.79116499e-01 4.01052758e-02 3.81290287e-01 1.08252895e+00
1.92449242e-01 6.74102187e-01 -5.68608522e-01 -3.84673238e-01
-4.55709368e-01 -8.70074153e-01 -4.00004685e-01 8.27888310e-01
-1.01347588e-01 3.22668344e-01 -4.02577892e-02 8.56578946e-02
1.39995098e-01 6.89313829e-01 -6.71225071e-01 -3.38887751e-01
-6.58227623e-01 -1.50895405e+00 7.37376988e-01 5.09066761e-01
6.49053395e-01 -1.07871032e+00 -8.73498499e-01 7.12090135e-01
2.04743464e-02 -1.07051504e+00 5.32556117e-01 7.09286094e-01
-1.00582016e+00 -9.14061904e-01 -5.47938943e-01 -5.36382675e-01
2.73011744e-01 3.21721613e-01 6.64162219e-01 -1.57040194e-01
-5.36910176e-01 6.31906748e-01 -7.99198925e-01 -5.33261657e-01
-1.18384317e-01 -4.88775671e-01 -9.22185183e-02 1.72058523e-01
5.14258265e-01 -8.49390805e-01 -1.72037140e-01 -1.19756791e-03
-5.67899764e-01 -4.63305146e-01 8.63016725e-01 9.10284162e-01
2.28829071e-01 1.63195774e-01 5.12827635e-01 -6.93911970e-01
6.61924660e-01 -4.76995856e-01 -4.57224935e-01 7.11471215e-02
-1.51618809e-01 -1.47208199e-01 2.20034122e-02 -2.73951381e-01
-1.23077917e+00 3.52377445e-01 -4.86759424e-01 2.55165827e-02
-2.83395439e-01 1.00779092e+00 -3.06828860e-02 -7.23868549e-01
1.00899231e+00 6.01934731e-01 4.33662646e-02 -6.26258671e-01
-1.19604040e-02 7.82285273e-01 7.27798343e-01 -4.75336254e-01
6.85664356e-01 6.32130325e-01 3.54928561e-02 -1.38044810e+00
-7.47553587e-01 -7.27764010e-01 -4.96244073e-01 -3.00381809e-01
2.39415795e-01 -7.73355842e-01 6.32644817e-02 6.46477222e-01
-8.02664101e-01 -1.18224703e-01 -2.76311934e-01 1.02346861e+00
-2.41611928e-01 5.72950900e-01 -3.80499065e-01 -1.21148896e+00
-6.78576767e-01 -5.40917337e-01 7.55579948e-01 2.63998926e-01
9.31228101e-02 -6.89364672e-01 2.51379967e-01 -1.37928575e-01
8.46826375e-01 3.66129309e-01 5.37754476e-01 -6.00896060e-01
1.45600334e-01 -5.63052952e-01 -2.23821640e-01 2.40759894e-01
2.20980883e-01 -3.56196851e-01 -1.12312567e+00 -2.32924938e-01
4.73603785e-01 5.40137589e-02 1.07843852e+00 7.57481217e-01
6.53906167e-01 2.57613391e-01 -6.42738163e-01 7.99767017e-01
1.60910606e+00 2.75976181e-01 8.17300737e-01 3.26088101e-01
2.13744462e-01 5.97057581e-01 8.64478052e-01 6.69919312e-01
-1.97454482e-01 4.20661569e-01 5.58336616e-01 -1.13804244e-01
8.49228129e-02 3.57897133e-01 1.92831933e-01 2.57936209e-01
-2.99464971e-01 2.57494569e-01 -8.41769397e-01 5.37002861e-01
-1.57108557e+00 -9.14431334e-01 -4.04271454e-01 2.10209751e+00
2.25832383e-03 1.03869416e-01 -2.11562544e-01 3.87846053e-01
8.50199521e-01 -2.63836924e-02 -1.82100251e-01 -1.99627265e-01
-1.88400343e-01 6.68726206e-01 1.01505196e+00 4.43440825e-01
-1.10843551e+00 5.06483197e-01 7.01656437e+00 4.41414148e-01
-1.29066491e+00 4.16523255e-02 -3.01916420e-01 6.12257957e-01
-3.14928708e-03 -1.25694051e-01 -6.98592365e-01 1.95359662e-01
7.63459623e-01 4.34861511e-01 -3.60115096e-02 6.78965509e-01
9.84742716e-02 -4.77977574e-01 -3.35011005e-01 9.26576614e-01
1.78013712e-01 -9.57028925e-01 -4.64573711e-01 -1.66817278e-01
1.11216567e-02 4.81290251e-01 -1.85507044e-01 3.86496425e-01
2.56095409e-01 -7.32010663e-01 6.73108101e-01 7.18763769e-01
4.26662832e-01 -5.41549504e-01 9.87475574e-01 2.85821736e-01
-1.20503795e+00 -3.03157121e-01 -6.53707623e-01 -2.65500695e-01
3.37282538e-01 1.18612838e+00 -6.35783315e-01 8.77190888e-01
8.28786075e-01 3.86835426e-01 -2.40821987e-01 1.49367201e+00
-1.59445070e-02 5.85604548e-01 -8.94470096e-01 1.90399602e-01
1.84627846e-01 -1.00452498e-01 7.15163767e-01 1.62156522e+00
8.61767828e-01 5.04004359e-01 2.56670654e-01 3.46793830e-01
8.97278368e-01 -7.09061399e-02 -5.93316555e-01 4.47236270e-01
4.40909266e-01 1.14112735e+00 -4.27854359e-01 1.08002342e-01
-3.53265643e-01 3.04407865e-01 -3.48633856e-01 2.45186165e-01
-4.88343567e-01 -4.64445084e-01 3.06561917e-01 2.89636701e-01
6.73563898e-01 -2.77819932e-01 -2.59897739e-01 -5.47842681e-01
-2.11452112e-01 -6.58562303e-01 5.52192867e-01 -6.46533012e-01
-1.24426520e+00 7.26397812e-01 1.30337566e-01 -1.06690073e+00
-1.10844776e-01 -5.37544072e-01 -8.41998816e-01 1.05796707e+00
-1.54122782e+00 -1.06874585e+00 -5.22601128e-01 6.73767388e-01
1.52756229e-01 -4.84965146e-01 9.85416770e-01 5.68567991e-01
1.23279259e-01 1.19722057e-02 2.27562994e-01 -4.44379598e-02
4.07740206e-01 -5.42166114e-01 1.68657497e-01 8.57343078e-01
-2.49427930e-01 4.68095802e-02 1.12247527e+00 -7.92080343e-01
-1.16596687e+00 -8.98598313e-01 5.75333059e-01 2.90192902e-01
3.55469644e-01 -9.35894921e-02 -1.01633489e+00 3.73400480e-01
-3.69650602e-01 1.21753074e-01 6.91408575e-01 1.10475555e-01
-1.87884301e-01 -2.37892568e-01 -1.44806433e+00 4.22383808e-02
5.14362633e-01 -1.20755009e-01 -9.36838925e-01 5.16575016e-02
-2.88476348e-01 -3.98270279e-01 -4.14443612e-01 8.30537617e-01
5.79854250e-01 -9.14296806e-01 1.03929567e+00 2.43588872e-02
-3.84253114e-01 -3.23683053e-01 -7.14438796e-01 -1.14164937e+00
-5.59215248e-01 -8.00895914e-02 1.42146245e-01 9.46062148e-01
2.07078859e-01 -7.67174125e-01 6.37819886e-01 3.22394930e-02
-5.19226015e-01 -2.57996678e-01 -8.24621916e-01 -4.62137491e-01
-6.21012509e-01 -5.38701832e-01 2.63395607e-01 6.89439595e-01
-2.67997980e-01 -3.44649404e-02 -4.02688265e-01 7.18156695e-01
9.53281045e-01 1.52738944e-01 4.85277146e-01 -1.48967218e+00
-2.10040465e-01 1.27915695e-01 -6.55317426e-01 -6.69683397e-01
-2.52738863e-01 -3.81734133e-01 5.02088666e-01 -1.58720386e+00
-3.66870344e-01 -6.72198057e-01 -1.44827753e-01 2.96529353e-01
3.04155022e-01 2.40945727e-01 -3.74783248e-01 1.59038290e-01
1.88254654e-01 4.17396486e-01 4.28960204e-01 -3.60124297e-02
-2.55549878e-01 4.76291358e-01 -5.87755919e-01 8.34888756e-01
8.18640947e-01 -7.38670170e-01 -6.82662183e-04 -3.04896206e-01
1.24044985e-01 3.06780394e-02 3.92512649e-01 -1.66460049e+00
3.07289660e-01 1.20769367e-01 7.41037309e-01 -9.09665644e-01
6.20643675e-01 -7.04773664e-01 4.31701690e-01 5.78401506e-01
2.68144518e-01 -2.29630649e-01 3.25547814e-01 7.54549921e-01
-3.50868046e-01 -5.85751116e-01 1.10955703e+00 -9.94029865e-02
-1.04777765e+00 -1.96996957e-01 -9.28426802e-01 -5.29564798e-01
5.34949362e-01 -5.78637183e-01 2.25497231e-01 -4.38456684e-01
-9.04460132e-01 -2.84739196e-01 -3.11340064e-01 -3.52403112e-02
7.27580965e-01 -8.15641880e-01 -9.73538578e-01 3.53891402e-01
-1.04957752e-01 -3.67963254e-01 5.51289678e-01 8.95964086e-01
-4.16126221e-01 4.38561514e-02 -4.30797130e-01 -4.65918809e-01
-1.14700294e+00 1.77554607e-01 6.58899128e-01 -2.37622887e-01
-8.69400620e-01 7.02348471e-01 -6.72056079e-02 -3.79618764e-01
-9.54516418e-03 8.39413628e-02 -8.13407421e-01 9.13019925e-02
6.30661845e-01 2.96444446e-01 6.44149959e-01 -7.66463518e-01
-5.66312551e-01 7.77426898e-01 9.94739030e-03 -1.57091036e-01
1.73347318e+00 1.68130081e-02 1.58936337e-01 5.53682186e-02
6.70935512e-01 -5.26393950e-03 -7.95643508e-01 -4.54792857e-01
-1.81940272e-02 -3.33039463e-01 4.54937696e-01 -6.11177921e-01
-9.17556405e-01 7.16628313e-01 9.68537569e-01 6.15598410e-02
1.35453916e+00 -4.23841566e-01 2.46714741e-01 4.43894267e-01
4.64693129e-01 -9.92334545e-01 -3.32762778e-01 9.18168604e-01
8.07343721e-01 -6.43802702e-01 2.72087455e-01 -5.57017624e-01
-9.25323814e-02 1.35016680e+00 -8.08798615e-03 -5.71555138e-01
1.37102640e+00 5.69332659e-01 1.23004980e-01 -4.68845367e-01
7.22298324e-02 -5.10294199e-01 -9.60765928e-02 1.30685043e+00
8.50284472e-02 -1.45776480e-01 -2.99651504e-01 6.58148348e-01
-1.79834768e-01 -8.50857645e-02 4.32036012e-01 1.27757716e+00
-9.91500139e-01 -7.75726736e-01 -1.02070236e+00 7.83713400e-01
-5.45299649e-01 -9.25566554e-02 3.24807912e-02 5.94867349e-01
3.67999636e-02 8.17018211e-01 -1.78253099e-01 -2.04959393e-01
5.47700286e-01 -2.58534163e-01 6.22129619e-01 -6.80022359e-01
-5.00304818e-01 1.68872312e-01 3.27409565e-01 -1.23374723e-01
-5.88606894e-01 -8.19895029e-01 -8.91478002e-01 1.04248784e-01
-6.52397931e-01 3.28222007e-01 9.34884429e-01 1.07774734e+00
5.35680912e-02 4.39097553e-01 2.75695592e-01 -1.29953337e+00
-6.45847559e-01 -1.19058311e+00 -9.76935029e-01 -2.94945270e-01
2.29235619e-01 -9.82467771e-01 -4.84281421e-01 -2.50771672e-01] | [6.877942085266113, 1.2386753559112549] |
43b2eefc-2509-41e1-b1f7-06dcaa228eff | static-fuzzy-bag-of-words-a-lightweight | 2304.03098 | null | https://arxiv.org/abs/2304.03098v1 | https://arxiv.org/pdf/2304.03098v1.pdf | Static Fuzzy Bag-of-Words: a lightweight sentence embedding algorithm | The introduction of embedding techniques has pushed forward significantly the Natural Language Processing field. Many of the proposed solutions have been presented for word-level encoding; anyhow, in the last years, new mechanism to treat information at an higher level of aggregation, like at sentence- and document-level, have emerged. With this work we address specifically the sentence embeddings problem, presenting the Static Fuzzy Bag-of-Word model. Our model is a refinement of the Fuzzy Bag-of-Words approach, providing sentence embeddings with a predefined dimension. SFBoW provides competitive performances in Semantic Textual Similarity benchmarks, while requiring low computational resources. | ['Vincenzo Scotti', 'Licia Sbattella', 'Roberto Tedesco', 'Matteo Muffo'] | 2023-04-06 | null | null | null | null | ['sentence-embeddings', 'sentence-embeddings', 'semantic-textual-similarity'] | ['methodology', 'natural-language-processing', 'natural-language-processing'] | [ 6.92126453e-02 6.93385974e-02 -7.95399621e-02 -4.65438545e-01
-4.47691530e-01 -3.57893765e-01 8.17966282e-01 1.41049170e+00
-8.04387927e-01 2.97283620e-01 5.65255523e-01 -1.59940332e-01
-4.66726393e-01 -1.12343812e+00 9.55663025e-02 -4.40032899e-01
-5.45481592e-02 3.30508798e-01 3.84051889e-01 -6.31904781e-01
6.96252644e-01 3.24238032e-01 -1.82905304e+00 5.68470359e-01
6.30492806e-01 1.07000566e+00 6.26266301e-02 5.46274126e-01
-7.46415198e-01 4.87666190e-01 -4.13578153e-01 -6.51215315e-01
6.11910410e-02 -5.49682751e-02 -7.89655924e-01 -2.47798666e-01
4.13631707e-01 2.19694212e-01 -1.64501369e-01 1.17793226e+00
3.92241448e-01 3.08627725e-01 5.37806571e-01 -8.17473412e-01
-1.05131650e+00 7.12208152e-01 -2.39211038e-01 3.71446371e-01
6.88772440e-01 -4.25565928e-01 1.46798491e+00 -1.18283057e+00
3.70418936e-01 1.27683127e+00 5.73573947e-01 2.88028210e-01
-1.17655706e+00 1.43062934e-01 5.25040962e-02 5.95473528e-01
-1.17259967e+00 8.25324729e-02 7.63310730e-01 -3.67438316e-01
1.54318583e+00 3.53486151e-01 5.39047241e-01 8.03928196e-01
3.62825632e-01 4.27609533e-01 9.29352760e-01 -7.68139660e-01
4.88157302e-01 3.81302685e-01 8.53664458e-01 3.05685192e-01
6.66528463e-01 -4.40001845e-01 -2.91547418e-01 -3.22742432e-01
-2.53196526e-02 1.54544845e-01 -2.43214201e-02 -3.24908227e-01
-7.37460375e-01 1.17998588e+00 3.09654474e-01 1.20873928e+00
-2.90471852e-01 6.83826581e-02 1.00636661e+00 4.61956918e-01
6.99440241e-01 7.34869301e-01 -1.86941579e-01 -1.38910890e-01
-8.34742904e-01 3.96280348e-01 6.90891385e-01 6.32996976e-01
6.07383609e-01 -1.02124669e-01 -2.28981629e-01 6.90897524e-01
2.45947033e-01 -3.47101204e-02 1.08274412e+00 -3.65991861e-01
4.79270816e-01 9.26575541e-01 -3.38159710e-01 -1.62084365e+00
-3.31512630e-01 -2.96911538e-01 -6.96148157e-01 -2.70456020e-02
-1.80383861e-01 4.97015357e-01 -4.43005919e-01 1.24762261e+00
1.62185147e-01 1.04573019e-01 2.68376827e-01 6.25005484e-01
7.01289296e-01 6.24097049e-01 5.39920293e-02 -4.74726707e-02
1.61246824e+00 -6.89643681e-01 -1.02509916e+00 2.26894971e-02
8.88163209e-01 -6.30517066e-01 1.20594203e+00 4.13017750e-01
-1.10803318e+00 -6.59161568e-01 -1.23212588e+00 -1.04675442e-01
-1.30082738e+00 -4.89882857e-01 5.92529774e-01 1.02589428e+00
-1.43606400e+00 7.76744068e-01 -3.27931345e-01 -7.35149622e-01
9.23336148e-02 2.84399360e-01 -4.69129145e-01 1.98284775e-01
-1.61009777e+00 1.22979856e+00 7.90205717e-01 -3.78695935e-01
1.58153445e-01 -6.30489767e-01 -1.12496281e+00 4.58090067e-01
1.12233706e-01 -5.49491465e-01 7.02951789e-01 -6.08242333e-01
-1.12028861e+00 8.50450337e-01 1.68773085e-02 -8.91228914e-01
-9.92307290e-02 -1.03850141e-01 -6.08091354e-01 2.67285526e-01
-1.74506903e-01 4.36825067e-01 6.80935264e-01 -9.39215124e-01
-3.03251475e-01 -4.92063314e-01 3.06829333e-01 3.82287055e-02
-1.30609512e+00 2.00817853e-01 3.04058224e-01 -8.08361232e-01
-1.45414636e-01 -4.00180668e-01 -2.56118357e-01 -1.71510339e-01
4.52867746e-01 -7.74743676e-01 6.81862772e-01 -1.82744399e-01
1.66637516e+00 -2.18048334e+00 3.02318007e-01 1.20570019e-01
2.02636167e-01 5.67983627e-01 -8.59487429e-02 1.14735496e+00
-2.09950000e-01 3.23273748e-01 -3.87391180e-01 -6.59887433e-01
1.66672185e-01 3.16033632e-01 -4.34307247e-01 1.02769025e-01
2.66543984e-01 6.40098631e-01 -1.10743999e+00 -7.79507697e-01
3.75878781e-01 4.34330195e-01 -7.45842040e-01 -1.61508024e-01
1.16274148e-01 -6.67783082e-01 -3.27998340e-01 7.54455179e-02
6.40874267e-01 1.31938860e-01 1.66592505e-02 -8.99322778e-02
-1.59223273e-01 2.53944457e-01 -1.09105444e+00 1.96435344e+00
-5.04209876e-01 4.45896536e-01 -4.07617807e-01 -1.29525876e+00
1.06498992e+00 3.25646996e-01 6.69750273e-01 -6.05445981e-01
2.56475687e-01 2.93178707e-01 -7.62954429e-02 -4.74812299e-01
1.31905639e+00 -2.78439283e-01 -4.37142253e-01 1.89529181e-01
3.49893928e-01 -2.14282945e-02 3.82132888e-01 3.06359917e-01
1.20771348e+00 -4.00038183e-01 6.33569360e-01 -5.04180610e-01
1.10194004e+00 -5.77057526e-02 -2.49092922e-01 4.24139470e-01
-2.18001798e-01 5.30364513e-01 2.27373779e-01 -3.96007687e-01
-9.86404061e-01 -9.74578381e-01 -4.05200154e-01 1.00507641e+00
3.36817540e-02 -9.18212414e-01 -7.40739465e-01 -4.46627498e-01
2.21838444e-01 7.13153660e-01 -6.75141871e-01 -5.04214764e-01
-4.40819263e-01 -4.47716117e-01 6.67634547e-01 5.46185315e-01
9.46392938e-02 -8.67213964e-01 -7.19100833e-01 4.43122625e-01
1.89407632e-01 -9.94240701e-01 -2.67929882e-01 2.38857299e-01
-1.01596427e+00 -5.26627421e-01 -5.06887197e-01 -9.10147488e-01
3.19613487e-01 1.67041048e-01 1.07052612e+00 1.38567509e-02
-4.70893115e-01 4.78024304e-01 -1.14464331e+00 -2.68973559e-01
-2.66356379e-01 -3.72356884e-02 1.78005755e-01 2.86395937e-01
8.98210466e-01 -5.57429671e-01 -4.06868160e-01 -4.16159719e-01
-1.42411184e+00 -6.55397475e-01 1.98235199e-01 8.02761793e-01
1.88409671e-01 1.46687105e-01 6.61518335e-01 -5.72622657e-01
1.29520380e+00 -4.04604912e-01 2.40714252e-02 2.11750880e-01
-8.96483302e-01 3.66043627e-01 9.90527630e-01 -1.06230542e-01
-6.81888819e-01 -4.68942225e-01 -3.41925532e-01 -1.49015158e-01
-1.98779479e-01 7.07729995e-01 7.30567649e-02 1.03792146e-01
4.94136602e-01 3.41699392e-01 -7.30337277e-02 -5.99930406e-01
7.10560501e-01 1.24082124e+00 1.80444807e-01 -4.03584838e-01
4.77197856e-01 3.84677589e-01 -1.39494771e-02 -8.43454599e-01
-3.89709234e-01 -7.78669417e-01 -7.63758898e-01 -8.60356018e-02
9.22497690e-01 -3.69293958e-01 -3.97899449e-01 -6.12422377e-02
-1.40714085e+00 6.86052978e-01 -6.55742526e-01 3.96734148e-01
-3.44585866e-01 7.60217547e-01 -3.75705868e-01 -7.54244328e-01
-5.21852195e-01 -7.48904288e-01 9.48725939e-01 1.69685110e-02
-5.19805491e-01 -1.30115354e+00 3.93779039e-01 3.40877920e-02
6.11036599e-01 4.35623853e-03 1.23793721e+00 -1.16107023e+00
2.24374816e-01 -6.73973560e-01 -1.08672865e-02 7.70880163e-01
1.80523738e-01 -2.17807040e-01 -9.33057725e-01 -2.66288340e-01
3.82768124e-01 -2.70234607e-02 9.87286925e-01 -1.53567567e-01
1.08803678e+00 -1.88377142e-01 -4.68899831e-02 1.90144498e-02
1.74567270e+00 3.43645103e-02 5.28200924e-01 5.15812635e-01
1.08493589e-01 6.80502534e-01 5.41694701e-01 5.85520983e-01
2.52296448e-01 7.41876960e-01 3.29162955e-01 3.69956166e-01
-3.55413221e-02 -3.74040753e-02 2.02276826e-01 1.31020510e+00
2.52462029e-01 -2.69191474e-01 -8.97449315e-01 6.47775769e-01
-1.83845949e+00 -1.20154047e+00 -1.19536310e-01 2.02095270e+00
5.07200837e-01 2.63235480e-01 -5.58753265e-03 8.91556859e-01
5.23587763e-01 6.01678848e-01 2.77035773e-01 -1.41548359e+00
-1.39772100e-02 5.48298717e-01 2.61022806e-01 6.36368871e-01
-8.63270164e-01 6.34234309e-01 5.70060253e+00 9.11422133e-01
-9.88343477e-01 1.78050831e-01 -1.24625608e-01 2.50437766e-01
-6.18833959e-01 -1.35627836e-01 -6.24206483e-01 5.98468661e-01
1.08041060e+00 -5.44388056e-01 -4.98329848e-02 6.97154284e-01
-6.99184984e-02 1.59290899e-02 -1.00663018e+00 9.28651869e-01
7.22267330e-01 -1.30311632e+00 5.60479760e-01 -1.23375386e-01
1.67319268e-01 -4.88317192e-01 1.61164373e-01 2.67090678e-01
-5.54671466e-01 -7.02947259e-01 3.99692595e-01 5.22094846e-01
2.03169540e-01 -8.57806623e-01 1.06462014e+00 1.36416316e-01
-1.39543211e+00 -2.56985813e-01 -6.90798700e-01 -2.48880580e-01
1.96251601e-01 6.57688320e-01 -2.78248817e-01 1.14180100e+00
5.58742285e-01 7.73088396e-01 -8.29010248e-01 8.01313043e-01
5.01930475e-01 1.85958296e-01 -4.21371609e-02 -5.28351068e-01
5.75659394e-01 -3.06218207e-01 4.96380806e-01 1.55368006e+00
3.33244890e-01 -1.44357011e-01 5.73054701e-03 5.75954914e-01
2.69575447e-01 6.86572075e-01 -7.63109505e-01 -2.79222757e-01
3.64031613e-01 1.30005229e+00 -6.73556447e-01 -4.54639882e-01
-4.88995254e-01 9.03638899e-01 6.54826835e-02 -4.41900879e-01
-6.73266947e-01 -9.71767962e-01 7.97117770e-01 -2.61636153e-02
1.99060604e-01 -3.67588192e-01 -3.90500158e-01 -9.51243818e-01
2.15130255e-01 -4.44532096e-01 4.48521048e-01 -4.50967550e-01
-1.49920845e+00 8.35680783e-01 2.16868550e-01 -1.20338011e+00
-1.51199862e-01 -9.98810112e-01 -5.36581337e-01 6.94685102e-01
-1.31407177e+00 -5.80558360e-01 -5.86976996e-03 2.72767812e-01
6.32957757e-01 -2.03537777e-01 1.33207762e+00 5.63665569e-01
-4.38930243e-02 4.85009462e-01 2.03449488e-01 -7.04593882e-02
4.84958231e-01 -1.42181814e+00 4.01648313e-01 4.47741508e-01
4.98905659e-01 8.08441520e-01 7.74650872e-01 -2.46254638e-01
-1.33990705e+00 -6.88598514e-01 1.58792770e+00 -4.28125978e-01
9.59092259e-01 -5.45391262e-01 -1.10219896e+00 -7.30981901e-02
6.44122660e-01 -1.52645901e-01 9.59856093e-01 1.03878332e-02
-4.80763882e-01 -4.21812266e-01 -1.41074383e+00 6.05578125e-01
7.78377712e-01 -7.57221937e-01 -1.55873346e+00 5.58164455e-02
1.04614282e+00 2.89141119e-01 -1.14869094e+00 2.43935034e-01
1.40575975e-01 -8.99093390e-01 1.16415429e+00 -7.33289421e-01
3.39890450e-01 -1.65000081e-01 -4.91457105e-01 -1.33718538e+00
-4.22413200e-01 -2.70423889e-01 -1.50376737e-01 1.33601856e+00
5.79297356e-02 -7.94322133e-01 5.21849275e-01 2.73277611e-02
-2.29290724e-01 -9.51511621e-01 -1.15927780e+00 -1.07534599e+00
2.54013091e-01 -3.99058968e-01 6.73263311e-01 7.49474347e-01
8.20992231e-01 2.44502172e-01 2.01793715e-01 -4.15592700e-01
4.06760067e-01 -1.17410325e-01 -1.12934574e-01 -1.29890168e+00
-1.41051486e-01 -9.37167048e-01 -1.33177638e+00 -6.44783795e-01
2.45970219e-01 -1.42064869e+00 -4.19696033e-01 -1.72752261e+00
-1.12181425e-01 2.23456532e-01 -6.61401391e-01 1.76369809e-02
9.85308457e-03 1.48069739e-01 2.52943426e-01 -2.86766350e-01
-4.64677721e-01 6.17241323e-01 5.24833977e-01 -1.61499217e-01
3.91667485e-01 -7.50401735e-01 -4.64147389e-01 5.69513619e-01
8.83292258e-01 -4.45442051e-01 -4.74604726e-01 -4.82343614e-01
5.04247606e-01 -5.99582136e-01 8.13352838e-02 -1.29388285e+00
3.59961540e-01 2.25179508e-01 -1.53663591e-01 -4.51980591e-01
6.36034131e-01 -1.22390914e+00 -2.86426067e-01 5.59258521e-01
-6.68093383e-01 5.64463675e-01 7.92782083e-02 6.86042726e-01
-6.72365010e-01 -7.49478281e-01 5.11079311e-01 -3.56442370e-02
-5.96680641e-01 -1.34845570e-01 -6.02099597e-01 -8.74158740e-02
1.18855047e+00 -5.78628898e-01 -9.33841914e-02 1.87700719e-01
-5.11536777e-01 -1.12496644e-01 3.80234450e-01 7.31089950e-01
8.55479002e-01 -1.46128166e+00 -4.45945740e-01 4.72846925e-02
5.74611306e-01 -6.79471374e-01 3.71463485e-02 5.22728741e-01
-4.31421816e-01 8.72334778e-01 -2.12650448e-01 -1.14038303e-01
-1.32857203e+00 9.88746226e-01 3.86547223e-02 -2.22369954e-01
-5.78040540e-01 5.15550494e-01 -3.74915004e-01 -2.09720626e-01
1.86625659e-01 -5.94338000e-01 -7.87813485e-01 7.20829725e-01
6.46988034e-01 6.03917360e-01 3.82591993e-01 -6.31762922e-01
-5.56332529e-01 7.06398666e-01 2.24918537e-02 -1.48106083e-01
1.43957663e+00 -7.47037828e-02 -5.33128798e-01 8.71157885e-01
1.51745653e+00 -1.80756792e-01 1.05358571e-01 -1.03931546e-01
7.42401242e-01 -5.74312031e-01 8.82139653e-02 -2.59451449e-01
-2.97480077e-01 1.19032729e+00 5.56372166e-01 9.65822995e-01
8.87457728e-01 -1.98434323e-01 8.74602318e-01 4.64996338e-01
6.87159359e-01 -1.24086654e+00 9.98491198e-02 6.44754529e-01
7.49209940e-01 -8.61831486e-01 -1.27644166e-01 -7.37922862e-02
-3.51837665e-01 1.44239402e+00 2.17231512e-01 -4.34042037e-01
7.99258888e-01 3.00703317e-01 -3.50569010e-01 -1.39477268e-01
-7.23013401e-01 -3.52668941e-01 4.14626837e-01 4.01624650e-01
7.65847862e-01 -4.70024459e-02 -1.19970512e+00 7.66629279e-01
-3.37092504e-02 -1.87537536e-01 5.13788283e-01 9.99403954e-01
-9.34822381e-01 -1.66616678e+00 -2.14315578e-01 5.08472204e-01
-3.73146772e-01 -4.18560743e-01 -5.73207200e-01 3.63598585e-01
2.89602220e-01 1.01933599e+00 3.28335613e-01 -4.76549327e-01
4.53055114e-01 4.01579291e-01 4.55530673e-01 -9.89239037e-01
-1.08100581e+00 -7.79396355e-01 -7.82090649e-02 -3.91831666e-01
-4.46198314e-01 -5.93778849e-01 -1.35022211e+00 -2.65733749e-01
-3.17542464e-01 6.44515157e-01 6.85198486e-01 7.92120814e-01
4.75508720e-01 4.15913820e-01 5.63364983e-01 -5.39934099e-01
-8.59554410e-01 -1.05315566e+00 -6.27300382e-01 4.37091559e-01
1.17539175e-01 -3.91559392e-01 -5.07920504e-01 -3.99354130e-01] | [10.589359283447266, 8.663086891174316] |
0399ec85-0433-4ea8-8f96-8c05f72a7a61 | on-the-unlikelihood-of-d-separation | 2303.05628 | null | https://arxiv.org/abs/2303.05628v1 | https://arxiv.org/pdf/2303.05628v1.pdf | On the Unlikelihood of D-Separation | Causal discovery aims to recover a causal graph from data generated by it; constraint based methods do so by searching for a d-separating conditioning set of nodes in the graph via an oracle. In this paper, we provide analytic evidence that on large graphs, d-separation is a rare phenomenon, even when guaranteed to exist, unless the graph is extremely sparse. We then provide an analytic average case analysis of the PC Algorithm for causal discovery, as well as a variant of the SGS Algorithm we call UniformSGS. We consider a set $V=\{v_1,\ldots,v_n\}$ of nodes, and generate a random DAG $G=(V,E)$ where $(v_a, v_b) \in E$ with i.i.d. probability $p_1$ if $a<b$ and $0$ if $a > b$. We provide upper bounds on the probability that a subset of $V-\{x,y\}$ d-separates $x$ and $y$, conditional on $x$ and $y$ being d-separable; our upper bounds decay exponentially fast to $0$ as $|V| \rightarrow \infty$. For the PC Algorithm, while it is known that its worst-case guarantees fail on non-sparse graphs, we show that the same is true for the average case, and that the sparsity requirement is quite demanding: for good performance, the density must go to $0$ as $|V| \rightarrow \infty$ even in the average case. For UniformSGS, while it is known that the running time is exponential for existing edges, we show that in the average case, that is the expected running time for most non-existing edges as well. | ['Devansh Arpit', 'Caiming Xiong', 'Weiran Yao', 'Juan Carlos Niebles', 'Shelby Heinecke', 'Huan Wang', 'Itai Feigenbaum'] | 2023-03-10 | null | null | null | null | ['causal-discovery'] | ['knowledge-base'] | [ 2.30432704e-01 7.72806644e-01 -4.33516234e-01 -1.57725215e-02
-7.74643898e-01 -8.52116525e-01 -1.67355016e-01 2.81268120e-01
-1.37296066e-01 8.38037014e-01 -2.46838465e-01 -6.87993348e-01
-7.40169942e-01 -1.40074790e+00 -9.78285074e-01 -9.21731591e-01
-1.08603156e+00 7.78949082e-01 1.21571928e-01 2.49721944e-01
7.50346994e-03 3.84785712e-01 -1.21005976e+00 -5.47319829e-01
6.82930768e-01 6.64167762e-01 -2.14254022e-01 7.65667379e-01
-8.52551758e-02 2.83990830e-01 -3.81698936e-01 -5.38035870e-01
5.70375025e-01 -8.24183047e-01 -9.18190897e-01 -3.61718349e-02
1.69992581e-01 3.21841873e-02 -5.71445227e-01 1.47874415e+00
1.31783605e-01 -1.01334333e-01 6.11423671e-01 -1.54536128e+00
-2.67706573e-01 1.25800633e+00 -1.09399951e+00 1.12580679e-01
3.32603753e-01 -1.08441874e-01 1.36493063e+00 -2.29077429e-01
7.43692875e-01 9.59770858e-01 4.80045825e-01 3.17245185e-01
-1.56750345e+00 -1.08572423e+00 2.11171776e-01 -3.94571811e-01
-1.54467690e+00 -1.60997778e-01 6.59345031e-01 -3.72363031e-01
5.58661401e-01 5.83991945e-01 3.82383019e-01 4.25472409e-01
-2.42796078e-01 3.19850862e-01 9.81732905e-01 -5.38268209e-01
4.40813303e-01 -3.74213099e-01 3.28259408e-01 1.00014162e+00
9.08515453e-01 3.27248365e-01 -5.23229241e-01 -3.93007576e-01
6.19867742e-01 -2.62755543e-01 -4.27503228e-01 -4.70099777e-01
-7.34243095e-01 9.55595970e-01 3.71801704e-01 2.66714483e-01
-5.56486100e-02 6.41236722e-01 1.03611104e-01 4.47449207e-01
-9.62144360e-02 1.72865883e-01 -5.06605387e-01 5.22016287e-02
-9.61263359e-01 1.31385043e-01 9.65900242e-01 1.36986017e+00
7.24485219e-01 -1.01173379e-01 2.51811624e-01 1.55016646e-01
2.24877819e-02 7.35143363e-01 -5.77682555e-01 -1.03268862e+00
5.57194233e-01 6.28484190e-01 1.58566937e-01 -9.84955311e-01
-3.82984459e-01 -3.03791493e-01 -1.00626850e+00 2.84480285e-02
8.73321116e-01 -2.03275189e-01 -7.99580097e-01 2.35883212e+00
1.10412046e-01 1.06624857e-01 -3.19407910e-01 6.87504530e-01
2.45345339e-01 5.24860144e-01 -2.58065443e-02 -8.84626329e-01
1.02347684e+00 -8.68404135e-02 -3.12391609e-01 -1.64716989e-01
6.11492217e-01 -4.11794722e-01 7.90418923e-01 3.17164302e-01
-1.30740309e+00 2.76584271e-02 -9.24981594e-01 4.37243700e-01
3.38165425e-02 -4.84780073e-01 9.09480929e-01 1.11345375e+00
-9.44963872e-01 3.44298512e-01 -7.22595036e-01 -1.83274135e-01
3.58038664e-01 4.81151849e-01 -4.27691966e-01 -6.15346134e-01
-9.83723044e-01 9.70394090e-02 1.35463223e-01 -2.58984417e-01
-1.13752365e+00 -4.83740330e-01 -7.42399395e-01 2.45852709e-01
7.64868796e-01 -3.97432864e-01 6.30517840e-01 -5.47385395e-01
-4.96624351e-01 8.84248018e-01 -3.05766344e-01 -3.91872406e-01
1.89281598e-01 5.80322206e-01 -2.29005083e-01 2.73456931e-01
3.17918897e-01 -2.67486237e-02 4.62848425e-01 -1.29003906e+00
-6.58240557e-01 -9.98531222e-01 3.38735133e-01 -1.37139648e-01
-1.47181321e-02 3.31537984e-02 -4.81463462e-01 -1.94392949e-01
6.76833391e-01 -9.15505230e-01 -4.96673942e-01 -1.91698909e-01
-8.32706630e-01 -2.02579707e-01 2.65621841e-01 -1.99242264e-01
1.35147893e+00 -2.08877873e+00 -5.12925349e-02 1.03764522e+00
7.69549370e-01 -2.27511674e-01 8.84266272e-02 4.98596072e-01
-2.40322992e-01 6.02631629e-01 -5.72865784e-01 9.63387340e-02
1.13535665e-01 2.26543009e-01 -2.33599730e-03 8.58892560e-01
-3.67561817e-01 2.49637410e-01 -8.56482744e-01 -2.30877489e-01
-3.11870992e-01 -1.73471197e-01 -6.87974632e-01 -1.98898111e-02
-3.27822059e-01 -2.43425183e-02 -4.08796251e-01 5.11511326e-01
9.46389258e-01 -5.40872693e-01 9.89074051e-01 3.30602795e-01
1.21729923e-02 1.43287286e-01 -1.81266749e+00 1.39670074e+00
8.40122104e-02 2.18517900e-01 7.41473138e-01 -1.50072896e+00
6.37502253e-01 2.87499577e-01 6.53844953e-01 -4.61939305e-01
1.08147413e-01 3.34878922e-01 1.65852159e-01 -1.76013023e-01
-4.82480116e-02 -4.76337969e-01 -4.69689220e-01 7.53183782e-01
-2.50611246e-01 2.53135324e-01 3.72763604e-01 8.03374290e-01
1.97990239e+00 -4.58857536e-01 -9.73867532e-03 -4.62691575e-01
-8.99428725e-02 4.34002280e-02 7.90931463e-01 1.00829446e+00
-1.54984191e-01 2.43604049e-01 1.37802136e+00 1.97447076e-01
-7.23464668e-01 -1.30064023e+00 -2.71543581e-02 7.53705740e-01
4.64394122e-01 -5.81802368e-01 -7.06203341e-01 -7.95564711e-01
1.14183664e-01 6.30147755e-01 -7.20158219e-01 -7.73496479e-02
-3.80859792e-01 -9.72701550e-01 3.57004404e-01 3.45127195e-01
1.92247525e-01 -5.09347320e-01 -3.85362625e-01 1.52288243e-01
1.59404859e-01 -8.30769181e-01 -4.09496248e-01 6.22759640e-01
-8.26727688e-01 -1.50541031e+00 -1.12694070e-01 -5.43943286e-01
1.09435356e+00 1.44419566e-01 1.11477995e+00 9.53382477e-02
-3.23148459e-01 2.08241925e-01 -1.29268080e-01 -1.94471888e-02
1.07610449e-02 -2.61382520e-01 -1.07206747e-01 -5.81416488e-01
1.89038813e-01 -8.43285978e-01 -6.33233547e-01 1.61247730e-01
-7.95849144e-01 -5.51303804e-01 3.48758310e-01 6.51642919e-01
7.21922874e-01 7.56879210e-01 2.87532657e-01 -1.27746677e+00
1.62717149e-01 -7.23621905e-01 -9.55637395e-01 1.13256320e-01
-8.08965802e-01 7.12105781e-02 5.15163660e-01 8.83672461e-02
-4.20977086e-01 -1.92923024e-02 1.37997687e-01 -2.80893117e-01
2.63237581e-02 7.90379465e-01 -4.32443291e-01 2.54903436e-01
6.25383615e-01 -2.15511441e-01 -2.50083089e-01 -4.54067558e-01
4.74846303e-01 9.42520350e-02 4.18649077e-01 -7.05158770e-01
8.91052067e-01 6.11225307e-01 5.56271672e-01 -5.54863155e-01
-3.98202688e-01 -1.20536156e-01 -1.64944738e-01 1.12170398e-01
4.99410450e-01 -6.17525697e-01 -1.25652361e+00 -3.28232110e-01
-7.58772314e-01 -3.64393562e-01 -3.19093227e-01 3.83519977e-01
-3.04534048e-01 1.92695677e-01 -1.96744412e-01 -1.18729067e+00
7.18796030e-02 -8.18790376e-01 4.11580920e-01 -1.46767378e-01
-1.50830775e-01 -7.01838434e-01 -1.51373431e-01 1.52823105e-01
-1.33813336e-01 4.68603879e-01 1.43267608e+00 -5.69076478e-01
-7.77070582e-01 -3.62555593e-01 -5.02345562e-01 -1.04496397e-01
-2.07779527e-01 -7.03784302e-02 -3.02056730e-01 -3.79697382e-01
-3.38602126e-01 2.07460448e-02 9.02369201e-01 6.34560287e-01
1.15450656e+00 -6.53601170e-01 -9.54615891e-01 3.29830617e-01
1.74160159e+00 2.93767452e-01 4.92921889e-01 -4.24619317e-01
4.37957615e-01 4.44235116e-01 2.57751167e-01 4.40779209e-01
1.96750626e-01 1.69528514e-01 6.61865175e-01 -9.07017514e-02
6.35385364e-02 -1.44873843e-01 8.87769535e-02 2.94605911e-01
1.66003536e-02 -5.83347321e-01 -9.07466292e-01 7.98172832e-01
-1.44721258e+00 -7.20828116e-01 -5.07012069e-01 2.83356380e+00
8.96648109e-01 3.99718195e-01 3.08609545e-01 2.56570518e-01
8.77180934e-01 -1.59142390e-01 -4.77396160e-01 -2.74494946e-01
-8.74580890e-02 1.06653106e+00 1.03777921e+00 6.43782794e-01
-4.79519904e-01 5.20414650e-01 5.89448833e+00 7.61415064e-01
-5.47618568e-01 1.90209880e-01 6.35904551e-01 -4.56064463e-01
-7.58627772e-01 5.89670479e-01 -5.52359760e-01 5.18621683e-01
9.88914371e-01 -3.35722357e-01 4.47818190e-01 7.95071423e-01
-1.15577728e-02 -4.58140135e-01 -1.16278303e+00 6.84237897e-01
-2.45808437e-01 -1.09208345e+00 -5.55142701e-01 5.89399815e-01
7.95543432e-01 -3.39465857e-01 -2.41403669e-01 -6.10605627e-02
1.11134684e+00 -1.23101962e+00 3.18910569e-01 -3.00687611e-01
1.13872898e+00 -1.09019196e+00 2.98494220e-01 4.02819484e-01
-1.22614908e+00 -1.26250714e-01 -1.72106981e-01 -2.92580966e-02
1.89607784e-01 1.20057154e+00 -4.27291304e-01 7.34660625e-01
6.84215307e-01 -5.49031720e-02 1.17744617e-01 6.27045214e-01
-1.91499233e-01 7.64829934e-01 -6.52321398e-01 1.99463621e-01
-1.81975681e-02 -2.88109869e-01 6.82907641e-01 8.38155687e-01
4.41327959e-01 6.80866122e-01 2.95265496e-01 7.13740468e-01
-4.84928876e-01 -4.78474237e-02 -7.57327259e-01 -2.65106738e-01
8.06264222e-01 7.15877414e-01 -1.02895176e+00 -9.27892551e-02
-2.13558048e-01 5.44662356e-01 2.64872730e-01 2.33301252e-01
-7.99438953e-01 -5.37799180e-01 6.98330581e-01 5.00342309e-01
3.94342929e-01 -3.67221236e-01 -4.26717162e-01 -7.84098387e-01
-4.13323194e-02 -6.29903257e-01 1.01198125e+00 -2.90644825e-01
-1.10686147e+00 1.70523301e-01 -5.48080951e-02 -5.23364961e-01
9.26192254e-02 -3.44612926e-01 -2.81985313e-01 9.12385762e-01
-7.64542341e-01 -4.21442211e-01 2.10132584e-01 7.67027497e-01
-2.96967745e-01 3.72862458e-01 7.71921635e-01 1.89879030e-01
-4.92019683e-01 6.60755277e-01 -1.57492068e-02 1.60111129e-01
2.50049829e-01 -1.21737742e+00 -2.87927598e-01 1.25948262e+00
1.79797977e-01 7.74735689e-01 9.29061174e-01 -7.99008131e-01
-1.75564420e+00 -7.99701095e-01 8.69853020e-01 7.77121857e-02
4.12435472e-01 -3.61890465e-01 -4.77452487e-01 9.84682679e-01
1.13942772e-02 1.50446445e-01 7.27345288e-01 5.87323427e-01
-4.35345471e-01 -1.95114478e-01 -1.42115164e+00 4.99876350e-01
1.67966080e+00 -2.95303106e-01 -4.27118279e-02 3.52638185e-01
5.57591081e-01 -1.60745010e-01 -1.02216280e+00 4.89984095e-01
2.32627913e-01 -1.02814758e+00 7.52914488e-01 -6.04831457e-01
2.47523770e-01 -2.79941291e-01 -4.68613982e-01 -8.66986334e-01
-2.47495666e-01 -8.52729678e-01 -6.49392158e-02 1.09165871e+00
7.02269912e-01 -6.06590927e-01 1.11597836e+00 7.43860185e-01
1.56688571e-01 -3.48464370e-01 -1.33664000e+00 -8.93536270e-01
2.99314886e-01 -6.45699322e-01 5.13153315e-01 1.09443390e+00
2.17416793e-01 4.57506359e-01 -1.28535032e-02 6.43000901e-01
1.07445657e+00 4.64050770e-01 6.31207764e-01 -1.41877151e+00
-5.65660894e-01 -5.03520846e-01 -1.41430497e-01 -9.19250190e-01
3.01243979e-02 -1.06926465e+00 -9.19780061e-02 -1.37143886e+00
5.54408669e-01 -1.16113913e+00 -6.52825013e-02 4.00661796e-01
5.78388721e-02 -5.52716963e-02 -1.64268747e-01 -1.29006699e-01
-3.83208901e-01 -2.49614250e-02 1.14046252e+00 -1.69504378e-02
-1.71546593e-01 -7.11050034e-02 -9.76418853e-01 6.23898149e-01
5.17136157e-01 -7.66047120e-01 -4.54558223e-01 -1.73917890e-01
5.98896325e-01 9.06063974e-01 3.01228374e-01 -4.26646441e-01
9.08232555e-02 -2.95710653e-01 -9.60123464e-02 -4.04398680e-01
-4.16335948e-02 -7.23489821e-01 6.08055592e-01 6.90141737e-01
-4.04367685e-01 -1.28566921e-01 -1.59157380e-01 9.48820233e-01
1.84844196e-01 -2.10264161e-01 6.97923839e-01 -1.54934406e-01
-4.20624800e-02 6.01371884e-01 -8.03804025e-03 2.95924872e-01
1.05844629e+00 -1.70336321e-01 -4.03748572e-01 -4.71208334e-01
-8.66346121e-01 2.44716272e-01 3.91665131e-01 -3.74945790e-01
2.95292079e-01 -1.00385416e+00 -5.95456243e-01 -1.95289962e-02
-1.79429557e-02 3.77076775e-01 3.51965904e-01 9.36151683e-01
-8.27032104e-02 -1.61678828e-02 3.84800732e-01 -3.45814586e-01
-1.09839964e+00 8.39984775e-01 6.83501735e-02 -3.12511712e-01
-3.27904105e-01 1.28916669e+00 2.94647008e-01 1.80157408e-01
1.43186793e-01 -7.75578395e-02 6.05996192e-01 -2.29194909e-01
9.44850966e-02 4.98993725e-01 -1.85693949e-01 -1.87124670e-01
-4.29538339e-01 1.97966188e-01 1.80350140e-01 -4.42832679e-01
1.23019397e+00 -1.02316648e-01 -5.30080736e-01 -3.93968672e-02
1.14625883e+00 5.33127606e-01 -9.66621280e-01 -9.84363481e-02
-9.04791877e-02 -6.08935893e-01 -2.37718105e-01 -6.00503564e-01
-1.41739774e+00 6.52119160e-01 2.29605719e-01 8.87631834e-01
1.23436511e+00 5.35571516e-01 4.04681712e-01 -9.57844034e-02
7.75705397e-01 -8.41346502e-01 -2.68463880e-01 8.16326886e-02
4.71136361e-01 -7.61356831e-01 2.74554268e-02 -8.17894697e-01
3.55051309e-02 6.39731944e-01 2.59476066e-01 -1.87354028e-01
8.15026879e-01 4.65407193e-01 -7.85866559e-01 -5.69159448e-01
-6.05751812e-01 -1.87979102e-01 -4.03137296e-01 4.98565763e-01
2.11087808e-01 4.63355720e-01 -6.34604871e-01 8.30831647e-01
-2.33185992e-01 -2.02673361e-01 4.79912192e-01 8.12165141e-01
-4.50790048e-01 -1.26915514e+00 -3.39788824e-01 7.70044327e-01
-5.69981933e-01 -1.84338763e-01 -2.48063520e-01 9.07268882e-01
2.34323859e-01 1.16059351e+00 8.47134963e-02 -1.62985012e-01
1.09236665e-01 -1.02308869e-01 7.69169569e-01 -5.56783676e-01
-1.55962095e-01 1.97795406e-01 3.14690143e-01 -5.04635394e-01
-1.99718848e-01 -7.30754972e-01 -1.68378139e+00 -8.78667533e-01
-3.53667915e-01 3.56982082e-01 3.76481920e-01 7.02794135e-01
2.16455996e-01 1.48032710e-01 7.48294353e-01 2.77085632e-01
-3.09439093e-01 -5.59237897e-01 -1.35419917e+00 6.62653670e-02
2.00653356e-02 -5.84799886e-01 -7.95494616e-01 1.97019987e-03] | [6.755875587463379, 4.993382453918457] |
f4f5baec-b7b9-4a95-b7f7-6ebe09facbad | exploring-high-order-structure-for-robust | 2203.11492 | null | https://arxiv.org/abs/2203.11492v1 | https://arxiv.org/pdf/2203.11492v1.pdf | Exploring High-Order Structure for Robust Graph Structure Learning | Recent studies show that Graph Neural Networks (GNNs) are vulnerable to adversarial attack, i.e., an imperceptible structure perturbation can fool GNNs to make wrong predictions. Some researches explore specific properties of clean graphs such as the feature smoothness to defense the attack, but the analysis of it has not been well-studied. In this paper, we analyze the adversarial attack on graphs from the perspective of feature smoothness which further contributes to an efficient new adversarial defensive algorithm for GNNs. We discover that the effect of the high-order graph structure is a smoother filter for processing graph structures. Intuitively, the high-order graph structure denotes the path number between nodes, where larger number indicates closer connection, so it naturally contributes to defense the adversarial perturbation. Further, we propose a novel algorithm that incorporates the high-order structural information into the graph structure learning. We perform experiments on three popular benchmark datasets, Cora, Citeseer and Polblogs. Extensive experiments demonstrate the effectiveness of our method for defending against graph adversarial attacks. | ['Fengxiang He', 'Liu Liu', 'Baosheng Yu', 'Jinlong Li', 'Yibing Zhan', 'Guangqian Yang'] | 2022-03-22 | null | null | null | null | ['graph-structure-learning'] | ['graphs'] | [ 1.86148994e-02 1.77965105e-01 -3.50633007e-03 -3.92483128e-03
1.66943818e-01 -1.02053547e+00 5.01755953e-01 1.12494506e-01
1.65301621e-01 2.88097382e-01 2.78687656e-01 -7.23592341e-01
-2.01397941e-01 -1.11680150e+00 -1.00707018e+00 -8.68752182e-01
-3.03870231e-01 -1.53351784e-01 5.30070484e-01 -7.63279557e-01
3.27755362e-01 6.47945821e-01 -5.43781102e-01 -8.97517353e-02
8.20196867e-01 5.36308050e-01 -4.06036437e-01 5.71304023e-01
1.26135275e-01 8.45942378e-01 -7.27581859e-01 -6.04446411e-01
5.61915755e-01 -4.24081206e-01 -6.84251964e-01 -2.30477706e-01
2.02784166e-01 -9.80200991e-02 -1.03876245e+00 1.78632021e+00
3.58055681e-01 3.44975553e-02 4.13178951e-01 -1.35886347e+00
-9.05621171e-01 9.64227140e-01 -5.48352778e-01 3.61809522e-01
3.14959437e-01 4.64567393e-01 9.34983909e-01 -4.53731716e-02
2.39755690e-01 1.72778821e+00 5.44233143e-01 5.14619052e-01
-8.11768830e-01 -8.35383117e-01 5.64400017e-01 2.17425108e-01
-1.07879615e+00 4.26969565e-02 1.17788887e+00 -1.37144014e-01
2.69168586e-01 5.04706740e-01 3.28147084e-01 1.25778031e+00
4.81138796e-01 4.76127565e-01 9.48850274e-01 -6.97556734e-02
-6.47693649e-02 -1.70123562e-01 1.20872699e-01 8.16343963e-01
6.10749424e-01 4.24354881e-01 1.55020997e-01 -3.01237643e-01
5.92187881e-01 8.82909968e-02 -5.80253601e-01 -1.88188210e-01
-6.96009159e-01 9.28807318e-01 1.16059554e+00 1.62850052e-01
7.98040777e-02 1.15094259e-01 5.30418158e-01 7.73402035e-01
1.98215917e-01 5.74910462e-01 -2.07921267e-01 5.14257729e-01
8.82647783e-02 -5.24014123e-02 8.84583652e-01 7.86746740e-01
5.08747339e-01 4.24589574e-01 9.22231302e-02 3.54633987e-01
4.25664514e-01 5.43377697e-01 2.05399737e-01 -2.17667937e-01
5.47038317e-01 7.75976896e-01 -5.50622404e-01 -1.88319719e+00
-3.42306972e-01 -5.37858307e-01 -1.36530197e+00 1.15480579e-01
3.38687927e-01 -3.25311452e-01 -8.39578986e-01 1.68199515e+00
3.41381043e-01 4.42780703e-01 1.15010872e-01 7.74772048e-01
8.71814668e-01 6.03512943e-01 -4.34417650e-02 3.61170471e-02
9.85391676e-01 -8.12078714e-01 -4.73299801e-01 -1.97130710e-01
5.15049517e-01 -4.16915774e-01 1.05991459e+00 2.25337639e-01
-5.35603046e-01 -3.29698771e-01 -1.17163968e+00 4.60006505e-01
-5.32307804e-01 -8.39500010e-01 7.48729050e-01 9.04104233e-01
-5.81209004e-01 9.57439780e-01 -5.78662276e-01 -1.56542122e-01
4.84166920e-01 3.64417911e-01 -4.79356557e-01 -8.91602784e-03
-1.71511197e+00 4.83611107e-01 6.81504488e-01 1.36212064e-02
-9.89739239e-01 -3.85026306e-01 -8.01062644e-01 2.13962391e-01
6.36541009e-01 -3.54749799e-01 6.18737280e-01 -9.58107233e-01
-1.24122858e+00 3.96326631e-01 5.87512732e-01 -7.14268625e-01
4.59500551e-01 1.19634300e-01 -8.47211123e-01 2.02725723e-01
-5.11257887e-01 -1.90659136e-01 9.89172637e-01 -1.11743021e+00
-8.59692767e-02 -4.09686387e-01 5.98333597e-01 1.06381871e-01
-6.78150833e-01 -4.66720723e-02 -8.66932124e-02 -8.43395233e-01
-3.26245390e-02 -9.19146001e-01 -4.59998131e-01 -3.95665765e-01
-9.97115314e-01 -1.91857014e-02 8.32956553e-01 -4.43125516e-01
1.47942901e+00 -2.26869893e+00 8.94989073e-02 6.98051572e-01
5.52239358e-01 6.01767898e-01 -1.37704015e-01 6.73972189e-01
-4.10487980e-01 7.43086398e-01 1.69410128e-02 6.03650033e-01
-1.60285890e-01 4.16539162e-02 -5.22480369e-01 7.25266755e-01
-5.95180951e-02 9.69273210e-01 -9.67694104e-01 -1.28848597e-01
-6.50659278e-02 1.35793656e-01 -4.73099470e-01 1.21985644e-01
-1.52358532e-01 4.99675244e-01 -8.81847978e-01 4.34204459e-01
9.68531430e-01 -1.85753778e-01 6.94029927e-02 -1.42138228e-01
4.81661499e-01 1.95717942e-02 -9.48900402e-01 1.00533032e+00
1.43023357e-01 9.99868959e-02 1.25521094e-01 -9.98250008e-01
1.07557714e+00 -4.45888266e-02 -1.80634394e-01 -5.33221483e-01
5.29045999e-01 7.05609005e-03 5.22313058e-01 -2.60870069e-01
-8.54626149e-02 2.22573459e-01 -1.46867707e-01 1.72065556e-01
-4.65315074e-01 2.40815759e-01 -4.82234955e-02 5.88739574e-01
1.40336227e+00 -6.08717084e-01 1.61933318e-01 -4.45315510e-01
1.01294756e+00 -4.44604576e-01 4.89204526e-01 8.90891016e-01
-1.28795415e-01 6.30317107e-02 9.85110521e-01 -5.07850587e-01
-6.13779902e-01 -1.04788435e+00 4.42078084e-01 8.38067770e-01
5.20160258e-01 -4.57112789e-01 -9.30533588e-01 -1.23078072e+00
8.76316950e-02 4.73872811e-01 -7.75556266e-01 -1.06099641e+00
-6.72736049e-01 -5.41495621e-01 7.75078356e-01 1.92551792e-01
5.47923625e-01 -1.09751630e+00 5.48015535e-01 -7.78016821e-02
2.80041605e-01 -9.84939218e-01 -8.32040548e-01 -2.52170235e-01
-7.23470569e-01 -1.36518407e+00 -6.32888302e-02 -7.75192320e-01
8.85992348e-01 4.50786620e-01 8.67771745e-01 8.22618008e-01
7.76346624e-02 3.06031294e-02 -5.06777167e-01 -3.80407751e-01
-6.79836869e-01 2.54877985e-01 1.94455341e-01 1.34768203e-01
3.20503265e-02 -9.99352574e-01 -4.15129721e-01 3.91719937e-01
-1.10068500e+00 -4.09954548e-01 6.77868664e-01 6.46839917e-01
2.80506760e-01 5.42107344e-01 4.59646314e-01 -1.40553570e+00
9.86855030e-01 -6.39333487e-01 -5.55449247e-01 9.83805656e-02
-5.67934692e-01 2.85569489e-01 1.41427994e+00 -6.10546112e-01
-4.53097522e-01 -3.59970778e-01 -3.20299596e-01 -5.65115809e-01
-4.63399217e-02 4.89098698e-01 -8.61079037e-01 -7.68952549e-01
8.02143097e-01 3.69313926e-01 -9.83541482e-04 -3.31144363e-01
4.39989865e-01 2.38685287e-03 5.11651158e-01 -5.43278694e-01
1.79350817e+00 2.27555707e-01 5.76459646e-01 -6.59169912e-01
-7.97230303e-01 8.99638087e-02 -1.80109814e-01 1.09451585e-01
4.51245755e-01 -3.91722649e-01 -1.00225925e+00 6.00141644e-01
-8.42249691e-01 -2.77231988e-02 2.21975461e-01 9.19724405e-02
-7.37119094e-02 9.61835563e-01 -5.41656017e-01 -3.83910626e-01
-6.07022762e-01 -8.99580419e-01 2.93973774e-01 4.45746928e-01
4.72786903e-01 -1.34319246e+00 3.51075418e-02 -1.92399770e-01
2.50254929e-01 8.43476534e-01 1.10295188e+00 -1.25446522e+00
-7.13216007e-01 -3.76828402e-01 -1.24458313e-01 3.92626673e-01
6.13459907e-02 3.42547521e-02 -5.12679994e-01 -6.15485311e-01
8.18452090e-02 -1.16026448e-02 7.40979970e-01 -3.81263718e-02
1.49738026e+00 -8.91862094e-01 -1.78048417e-01 1.12357450e+00
1.36341405e+00 9.09704715e-02 8.92152131e-01 3.26530427e-01
1.30151796e+00 2.10523844e-01 3.16330314e-01 -8.74772370e-02
-8.19010437e-02 1.43929243e-01 1.10013807e+00 -8.71681944e-02
2.55758077e-01 -6.61820471e-01 3.94152880e-01 8.30375075e-01
7.82685727e-02 -6.34017706e-01 -7.00577199e-01 -2.88026705e-02
-1.51570582e+00 -8.41313303e-01 -2.93187529e-01 1.88771546e+00
4.32643294e-01 7.32244134e-01 -4.86874245e-02 1.82630971e-01
9.41957533e-01 6.03442430e-01 -6.37202680e-01 -4.05952245e-01
-2.47841805e-01 6.31852522e-02 8.71280432e-01 4.28425729e-01
-1.13193715e+00 1.12955117e+00 4.86272383e+00 1.17225182e+00
-8.77175689e-01 -3.92010063e-01 5.04239082e-01 5.72788596e-01
-4.11104918e-01 2.25367442e-01 -4.55513030e-01 6.35753095e-01
7.17870653e-01 -5.00257432e-01 7.48341560e-01 9.77807879e-01
-1.37852371e-01 8.55692089e-01 -7.15120971e-01 5.16954780e-01
-5.52676879e-02 -1.15898418e+00 5.44475198e-01 2.37569302e-01
4.68420893e-01 -1.65100098e-01 1.55727550e-01 3.51555735e-01
7.78006971e-01 -1.25378537e+00 1.60825789e-01 3.72529417e-01
4.34987873e-01 -1.14118946e+00 5.56742966e-01 4.51133192e-01
-1.33240795e+00 -3.50809209e-02 -6.94608331e-01 1.25625923e-01
-3.01150799e-01 2.05662042e-01 -4.98660564e-01 1.01047993e+00
3.22192729e-01 7.91682303e-01 -9.92415607e-01 8.11354697e-01
-6.27222240e-01 9.03454423e-01 -7.39882737e-02 -1.73244581e-01
4.54164803e-01 -2.87516207e-01 9.66092348e-01 7.59655058e-01
-1.21962056e-01 1.94062322e-01 5.42956769e-01 5.12858868e-01
-3.96453142e-01 9.21170935e-02 -9.02758658e-01 -3.32518399e-01
5.98230541e-01 1.23104703e+00 -6.59097493e-01 2.19463915e-01
-1.13072522e-01 7.05993533e-01 4.19624269e-01 2.71406025e-01
-8.66594493e-01 -8.35269928e-01 5.88133395e-01 1.37602672e-01
9.96876508e-02 -6.15391769e-02 8.06386024e-02 -1.09560609e+00
-8.88464004e-02 -1.32192636e+00 6.33495331e-01 -3.20840538e-01
-1.69233620e+00 6.94543302e-01 -4.67652649e-01 -1.12759674e+00
4.01988119e-01 -6.69510365e-01 -1.28979039e+00 7.53910244e-01
-1.24195552e+00 -1.02742040e+00 -2.54115850e-01 1.00966644e+00
-5.56839891e-02 -3.80130291e-01 5.82682192e-01 -5.37246838e-03
-8.08994889e-01 1.08394098e+00 -4.87007312e-02 7.97272861e-01
4.71449733e-01 -1.20722258e+00 1.02293813e+00 1.25554287e+00
1.44833118e-01 9.43065882e-01 7.65018106e-01 -9.01390553e-01
-1.69631457e+00 -1.23801053e+00 2.79306364e-03 -4.05964851e-01
1.16006410e+00 -2.92873174e-01 -1.17972708e+00 6.88595712e-01
5.92476055e-02 1.92369789e-01 4.13942277e-01 -7.70183131e-02
-7.65814900e-01 -1.15010932e-01 -1.20176828e+00 9.27739203e-01
1.24415338e+00 -4.31839913e-01 -4.13864702e-01 5.96196353e-01
1.34312057e+00 -3.79624099e-01 -6.81655943e-01 4.51219678e-01
4.08495180e-02 -8.35074544e-01 1.10867035e+00 -1.10016310e+00
2.71868795e-01 -3.14992130e-01 2.18919903e-01 -1.47732961e+00
-7.15358973e-01 -1.14131010e+00 -3.89817625e-01 1.05303574e+00
2.66797096e-01 -1.00420582e+00 7.50930309e-01 -4.91163284e-02
-6.86947927e-02 -7.34086990e-01 -4.89632100e-01 -9.09784079e-01
2.63447791e-01 1.43801138e-01 9.26907718e-01 1.07016289e+00
-3.72488409e-01 2.78644651e-01 -5.75504243e-01 8.28863740e-01
5.89497507e-01 -1.65228382e-01 1.01838291e+00 -1.34423363e+00
-3.48699689e-01 -5.19378960e-01 -8.02924573e-01 -8.46380055e-01
2.47059062e-01 -1.10099328e+00 -4.54176128e-01 -1.17406690e+00
-2.83121258e-01 -2.12517306e-01 -4.80498910e-01 1.90360844e-01
-6.21705174e-01 -8.35833475e-02 7.38603473e-02 1.94139257e-02
-5.42538524e-01 3.57090503e-01 1.40416825e+00 -2.22916782e-01
1.34630844e-01 1.71494946e-01 -1.24775195e+00 9.05154943e-01
9.44782317e-01 -6.32756412e-01 -6.75601542e-01 -2.00583488e-01
2.50177115e-01 -3.19684774e-01 3.17277700e-01 -9.22762871e-01
6.47790655e-02 -3.30154061e-01 1.75394773e-01 -1.82651415e-01
-3.39799315e-01 -9.57553029e-01 -5.28509393e-02 8.31921875e-01
-1.99592769e-01 1.36946410e-01 -3.09536960e-02 1.14837193e+00
1.04550764e-01 -6.43974766e-02 1.05582976e+00 -1.46091849e-01
-3.86874706e-01 1.00827968e+00 1.05207846e-01 5.93470037e-01
9.73466516e-01 1.41547590e-01 -5.93574286e-01 -4.90472913e-01
-5.68956077e-01 3.45467299e-01 5.13987422e-01 5.40626407e-01
4.66396093e-01 -1.44032073e+00 -6.94776058e-01 2.63709694e-01
-1.17183737e-01 -1.86458036e-01 2.40008637e-01 4.23229486e-01
-6.61998689e-01 -1.57865584e-01 -2.40024418e-01 -1.30673930e-01
-1.30188334e+00 1.13956749e+00 4.26428616e-01 -5.38501084e-01
-7.49069810e-01 9.92785335e-01 3.77554357e-01 -3.12346488e-01
2.57746518e-01 1.91720650e-01 -3.99606466e-01 -6.08932853e-01
5.18304348e-01 2.14937434e-01 -3.01349521e-01 -5.15528798e-01
-2.64205158e-01 3.72700125e-01 -4.76722747e-01 6.39818072e-01
1.04984832e+00 1.10992566e-01 -3.77241671e-01 -1.79855406e-01
1.13989484e+00 4.52945679e-01 -8.96604478e-01 -2.88915157e-01
-1.74603447e-01 -5.88065863e-01 -4.16310728e-01 -3.35228145e-01
-1.32316375e+00 7.77779818e-01 1.88978359e-01 9.78788674e-01
1.13361144e+00 -3.29672903e-01 8.31252694e-01 6.40857577e-01
2.26661086e-01 -3.87793452e-01 2.58232325e-01 7.28268802e-01
7.55984306e-01 -8.89975905e-01 2.05571838e-02 -6.84434772e-01
-4.41322297e-01 1.04233515e+00 7.41785824e-01 -7.53454089e-01
8.35061133e-01 -1.25928774e-01 -2.52926648e-01 -3.35348934e-01
-2.85470188e-01 2.01676264e-01 3.84509414e-01 6.97543979e-01
-2.10809782e-01 6.80981725e-02 -1.80853501e-01 6.51484311e-01
-5.02413511e-01 -1.03561044e+00 5.97525358e-01 4.84062552e-01
-6.02506280e-01 -1.09691846e+00 -1.99329361e-01 3.37542355e-01
-7.87824988e-01 -2.75577873e-01 -9.36614990e-01 6.70161903e-01
-3.20977390e-01 9.91330028e-01 -7.32447267e-01 -9.41884875e-01
4.73897040e-01 -3.92487109e-01 -3.68808731e-02 -4.53469455e-01
-7.35960186e-01 -4.00842369e-01 -8.22338089e-02 -6.01030529e-01
2.62450576e-01 -1.10589981e-01 -1.03100169e+00 -7.54100502e-01
-3.50060821e-01 5.07453024e-01 1.69645101e-01 6.10238314e-01
2.82513320e-01 7.49358654e-01 1.27971697e+00 -1.70529917e-01
-8.40613663e-01 -7.66484320e-01 -5.40351272e-01 6.95242643e-01
2.55094826e-01 -4.43509340e-01 -9.85453904e-01 -5.69996953e-01] | [6.139091491699219, 7.311561107635498] |
17e40a04-0120-4854-9c12-1fc8e9d01a5a | hyperbolic-hierarchical-knowledge-graph | 2204.13704 | null | https://arxiv.org/abs/2204.13704v1 | https://arxiv.org/pdf/2204.13704v1.pdf | Hyperbolic Hierarchical Knowledge Graph Embeddings for Link Prediction in Low Dimensions | Knowledge graph embeddings (KGE) have been validated as powerful methods for inferring missing links in knowledge graphs (KGs) since they map entities into Euclidean space and treat relations as transformations of entities. Currently, some Euclidean KGE methods model semantic hierarchies prevalent in KGs and promote the performance of link prediction. For hierarchical data, instead of traditional Euclidean space, hyperbolic space as an embedding space has shown the promise of high fidelity and low memory consumption; however, existing hyperbolic KGE methods neglect to model them. To address this issue, we propose a novel KGE model -- hyperbolic hierarchical KGE (HypHKGE). To be specific, we first design the attention-based learnable curvatures for hyperbolic space to preserve rich semantic hierarchies. Moreover, we define the hyperbolic hierarchical transformations based on the theory of hyperbolic geometry, which utilize hierarchies that we preserved to infer the links. Experiments show that HypHKGE can effectively model semantic hierarchies in hyperbolic space and outperforms the state-of-the-art hyperbolic methods, especially in low dimensions. | ['Yanping Zhang', 'Shu Zhao', 'Fulan Qian', 'Wenxue Wang', 'Wenjie Zheng'] | 2022-04-28 | null | null | null | null | ['knowledge-graph-embeddings', 'knowledge-graph-embeddings'] | ['graphs', 'methodology'] | [-7.86084116e-01 6.96148098e-01 -2.71995425e-01 -1.58382475e-01
-1.41095966e-01 -3.00194055e-01 2.31234014e-01 3.63997400e-01
7.96009973e-02 3.54482174e-01 4.30214107e-01 -3.39751035e-01
-6.35810494e-01 -1.38024032e+00 -7.29429543e-01 -4.19340372e-01
-2.16632038e-01 4.27000582e-01 6.81822121e-01 -4.61250514e-01
-1.19567141e-01 2.89084494e-01 -1.12116134e+00 -2.61160522e-03
1.09614635e+00 6.55302525e-01 -3.46761346e-01 2.79614300e-01
-6.77373633e-02 8.12659323e-01 4.40637209e-02 -8.79720867e-01
6.31935373e-02 -2.39857659e-02 -1.02995861e+00 -5.34069121e-01
2.05219418e-01 -1.82323650e-01 -1.30953455e+00 9.12139893e-01
2.66322434e-01 1.61055714e-01 5.56269348e-01 -1.73279631e+00
-1.53873599e+00 9.47900176e-01 -2.22428709e-01 -6.06197072e-03
2.58884996e-01 -6.52461946e-01 1.47522545e+00 -1.15558505e+00
6.55920267e-01 1.31855309e+00 1.08806026e+00 1.57148719e-01
-1.10225892e+00 -6.65160716e-01 1.07413799e-01 7.12333679e-01
-1.85071754e+00 1.69482455e-01 7.44201362e-01 -4.84220386e-01
6.35774076e-01 1.62390277e-01 8.68420243e-01 6.51333749e-01
6.48991913e-02 8.13356400e-01 5.41451514e-01 -7.69265369e-03
-4.79979739e-02 1.52117461e-01 3.87387663e-01 1.25389421e+00
6.46051586e-01 -3.07819009e-01 -6.51376545e-01 -1.26311466e-01
8.77217829e-01 -9.33953077e-02 -3.37072968e-01 -8.37827384e-01
-1.12865794e+00 1.06351352e+00 1.08635390e+00 1.07393628e-02
-5.23416325e-02 2.18796864e-01 3.49816054e-01 6.95033744e-02
2.78451592e-01 5.77355981e-01 -3.25886756e-01 3.91941935e-01
-2.85190195e-01 3.20827365e-01 7.00419247e-01 1.33457077e+00
7.12459624e-01 -2.82030553e-01 4.20739837e-02 6.40679181e-01
5.35471916e-01 4.19947654e-02 5.35743535e-02 -6.50130272e-01
4.46536660e-01 1.25128853e+00 -1.24364197e-01 -1.83552802e+00
-6.32089496e-01 -4.62196708e-01 -8.22568774e-01 -6.23099327e-01
1.07590724e-02 4.28123832e-01 -3.79387856e-01 1.60235858e+00
6.63737416e-01 2.97849178e-01 2.11283386e-01 8.03609192e-01
1.08722448e+00 7.11684942e-01 -1.27441555e-01 2.76062399e-01
1.25533354e+00 -9.03068841e-01 -7.80437529e-01 5.29384732e-01
9.50918078e-01 -5.91294803e-02 1.15492690e+00 -6.07047677e-02
-8.96602213e-01 -2.78127372e-01 -1.21500385e+00 -6.29528344e-01
-1.09002709e+00 -2.25670874e-01 9.92627621e-01 5.05146742e-01
-1.06243503e+00 4.92547244e-01 -8.64415169e-01 -3.82865608e-01
2.92168558e-01 2.17476189e-01 -4.10327226e-01 -1.45566612e-01
-1.90762794e+00 7.03713715e-01 8.25275302e-01 2.08876982e-01
-2.67892122e-01 -1.09475780e+00 -1.25412917e+00 3.09631556e-01
3.54147971e-01 -8.74084175e-01 5.28968275e-01 3.75199467e-01
-1.03212154e+00 4.84629631e-01 3.70068789e-01 -3.05041134e-01
3.24332744e-01 -1.32125676e-01 -5.94930232e-01 2.00149387e-01
6.87410682e-03 5.42781711e-01 2.15848014e-01 -1.32726300e+00
-3.56826425e-01 -5.98745227e-01 4.76037234e-01 1.79125428e-01
-9.77823913e-01 -6.21509373e-01 -5.64452767e-01 -5.46466589e-01
6.26297891e-01 -9.49024737e-01 2.40465403e-01 -2.70101409e-02
-7.43185401e-01 -4.70970839e-01 9.26142812e-01 -9.29598868e-01
1.70680702e+00 -2.01792789e+00 3.77032697e-01 4.38858747e-01
7.30154335e-01 5.42104542e-02 2.04983845e-01 5.21625221e-01
2.81166524e-01 2.42379725e-01 -4.93254699e-02 -2.56228521e-02
4.03810471e-01 3.26700777e-01 -1.31223872e-01 3.71591181e-01
-1.14251040e-01 1.24790430e+00 -1.16126704e+00 -6.95872784e-01
1.22025132e-03 6.91826582e-01 -8.20562243e-01 -5.96408825e-03
6.37625530e-02 -8.52978602e-02 -6.05484068e-01 4.08921123e-01
6.76798165e-01 -5.27147710e-01 4.08028245e-01 -7.32651770e-01
1.83452711e-01 -2.22886056e-02 -1.15875983e+00 1.50770462e+00
-2.06189618e-01 4.62006271e-01 -4.21744347e-01 -7.91765034e-01
9.95594859e-01 -2.10401528e-02 5.26688516e-01 -4.82841611e-01
-1.67213872e-01 1.19113870e-01 -3.32614779e-01 -4.42368686e-01
7.37132072e-01 2.59471536e-01 -4.37791608e-02 -1.75740123e-01
-9.28200781e-02 1.49974793e-01 -1.23922564e-01 8.95531237e-01
1.13858879e+00 -9.46746580e-03 2.65352484e-02 -3.56545508e-01
4.16410655e-01 -5.63442521e-02 7.08585203e-01 2.26094782e-01
4.34471667e-02 3.17956358e-01 7.19757676e-01 -3.10959369e-01
-1.18378556e+00 -1.33545125e+00 -2.45347634e-01 9.08085406e-01
7.94483721e-01 -1.00033784e+00 -6.59413636e-01 -7.97451079e-01
3.96407127e-01 5.58463097e-01 -9.07953262e-01 -5.70715666e-01
-3.80469114e-01 -6.54835999e-01 6.87636733e-01 8.03103507e-01
6.96018994e-01 -3.65454286e-01 1.80526555e-01 4.02190574e-02
-3.72844003e-02 -1.30818617e+00 -5.85226417e-01 -4.29893613e-01
-5.08296371e-01 -1.54682279e+00 -2.44557038e-01 -9.32095170e-01
6.06512606e-01 3.01909089e-01 7.31886923e-01 8.69888738e-02
-3.32353711e-01 5.65468431e-01 -7.23421514e-01 1.97351143e-01
2.77823173e-02 3.10828507e-01 1.65520892e-01 -3.27259451e-02
1.98045999e-01 -6.66305125e-01 -8.82936358e-01 4.13383543e-01
-9.21201706e-01 3.68213177e-01 4.41494823e-01 7.28256762e-01
4.43139881e-01 4.79164779e-01 6.40914261e-01 -1.10261250e+00
5.74157000e-01 -6.93336248e-01 -3.69695038e-01 5.04430175e-01
-9.16822612e-01 2.68275619e-01 6.76312268e-01 -6.52874932e-02
-8.05648506e-01 -4.69092935e-01 1.38073236e-01 -5.30191779e-01
6.07698739e-01 8.51265490e-01 -2.83042997e-01 -2.64811784e-01
3.64812136e-01 3.06094084e-02 -3.51951152e-01 -4.70591307e-01
7.69977868e-01 3.84567261e-01 4.15143222e-01 -5.99558175e-01
8.74748468e-01 4.12360013e-01 2.77899474e-01 -6.69300020e-01
-8.92004132e-01 -4.08427179e-01 -7.35060334e-01 1.69858038e-01
1.01348031e+00 -7.63628423e-01 -1.01848638e+00 -8.32135975e-02
-9.88268256e-01 -2.82971617e-02 1.59336887e-02 4.36490476e-01
-2.86270618e-01 4.54953223e-01 -8.54681671e-01 -4.62380588e-01
-3.10881436e-02 -7.46229649e-01 9.26117718e-01 3.64347361e-02
9.19767767e-02 -1.34457612e+00 3.07715498e-02 5.82058609e-01
8.94557387e-02 3.74019653e-01 1.61750460e+00 -4.13393646e-01
-7.85780609e-01 -1.20137017e-02 -7.09459186e-01 -1.39602795e-01
-5.86687438e-02 -9.32912901e-02 -4.69377995e-01 -1.80120364e-01
-8.21604311e-01 9.32532027e-02 7.27726042e-01 -1.75744593e-01
1.22318208e+00 -5.61722338e-01 -5.53956330e-01 9.07701313e-01
1.49440777e+00 -2.10202008e-01 6.77578032e-01 4.54921991e-01
1.39545953e+00 5.42445600e-01 4.08502012e-01 2.59010285e-01
1.18338084e+00 6.93912745e-01 4.60566968e-01 -1.40789542e-02
-2.32522059e-02 -9.52647626e-01 -2.77353544e-02 1.41737366e+00
2.92654596e-02 1.31790310e-01 -9.81569648e-01 6.20215416e-01
-2.14393902e+00 -6.10629976e-01 -5.24436891e-01 1.95179009e+00
7.42018282e-01 -1.03862345e-01 1.11373499e-01 5.76364249e-02
8.59793305e-01 -3.03266533e-02 -4.62670386e-01 7.65004456e-02
-1.42556474e-01 -3.91996235e-01 5.68453908e-01 6.27293527e-01
-9.58810151e-01 1.11911285e+00 5.54205084e+00 5.79541326e-01
-3.44150573e-01 2.35979244e-01 -1.32168874e-01 4.09521043e-01
-7.06469297e-01 7.15041086e-02 -8.98248672e-01 3.43582004e-01
5.86768031e-01 -3.23671937e-01 2.71279007e-01 8.78452241e-01
-2.08807483e-01 6.03766501e-01 -1.08516228e+00 1.05318820e+00
-1.57987252e-01 -1.34519911e+00 3.18384051e-01 1.94160029e-01
9.36487973e-01 -5.04889488e-01 1.30744010e-01 6.46994650e-01
4.68518108e-01 -9.69979107e-01 2.96579242e-01 4.50606167e-01
5.50065756e-01 -8.59302104e-01 7.28059113e-01 -6.96678311e-02
-1.75273716e+00 -6.76294714e-02 -6.94490433e-01 4.78881717e-01
-9.61305946e-03 3.70213389e-01 -8.32284570e-01 1.07130671e+00
6.55822039e-01 8.47389817e-01 -8.65679145e-01 9.16130722e-01
-2.21736401e-01 2.69384027e-01 -1.01775751e-01 1.60272822e-01
2.70995766e-01 -3.99337292e-01 2.34278545e-01 1.02013338e+00
3.84010613e-01 4.14564818e-01 1.77262112e-01 9.15751636e-01
-3.61064941e-01 2.76854277e-01 -6.10361755e-01 -1.60717249e-01
8.52945685e-01 9.89654005e-01 -4.82349098e-01 -5.09991050e-02
-6.53042138e-01 8.77028167e-01 6.92861557e-01 4.59766418e-01
-1.03216612e+00 -6.95858061e-01 6.43764615e-01 4.12265390e-01
3.04217841e-02 -4.66916114e-01 -2.92879809e-02 -1.22419119e+00
7.19904080e-02 -2.58025229e-01 8.28451395e-01 -8.20001662e-01
-1.24323237e+00 2.91364044e-01 4.73535918e-02 -7.76218534e-01
4.71462786e-01 -6.21490717e-01 -6.20780885e-02 2.70664394e-01
-1.43380916e+00 -1.54533458e+00 -6.01643562e-01 7.05124617e-01
-5.16077392e-02 9.94744971e-02 6.23741388e-01 4.62755114e-01
-6.09616518e-01 8.13624263e-01 4.25062925e-01 4.26653683e-01
3.08756888e-01 -1.58390296e+00 2.55307764e-01 2.99860507e-01
1.12283468e-01 8.29328120e-01 4.33209658e-01 -7.29099333e-01
-1.72939777e+00 -1.35485137e+00 8.08370471e-01 -4.54138905e-01
1.09477448e+00 -4.32115436e-01 -1.29671657e+00 1.01008129e+00
-5.25039911e-01 2.18683481e-01 7.43815243e-01 4.57242697e-01
-8.21954906e-01 -2.21806094e-01 -8.82858872e-01 7.44241178e-01
1.58595300e+00 -8.04600060e-01 -5.38272142e-01 3.89079928e-01
1.34500587e+00 -2.85279512e-01 -1.74810851e+00 6.41501427e-01
6.15358770e-01 -8.11666608e-01 1.17684281e+00 -7.80362368e-01
2.49248400e-01 -4.55279797e-01 -2.72440612e-01 -1.30077851e+00
-5.48145533e-01 -3.15031797e-01 -7.07006574e-01 1.17976606e+00
4.10177469e-01 -7.47884393e-01 9.68567669e-01 5.81917524e-01
-2.22900093e-01 -1.05755866e+00 -5.03550351e-01 -9.60816324e-01
2.56636981e-02 9.49180219e-03 8.56344581e-01 1.47476220e+00
4.94141847e-01 3.36649328e-01 -2.31498688e-01 7.49475658e-01
7.62156308e-01 5.66184223e-02 6.03593349e-01 -1.47588921e+00
3.08271404e-02 -2.42699608e-01 -1.06152606e+00 -9.04627025e-01
3.70103061e-01 -1.25849175e+00 -4.54837978e-01 -1.94662154e+00
2.28888378e-01 -7.05915809e-01 -3.29861820e-01 3.31462145e-01
-2.52289206e-01 7.79542997e-02 -8.15804228e-02 2.31002301e-01
-8.33147526e-01 1.15544975e+00 1.28619981e+00 -8.71714279e-02
-1.92341730e-01 -5.51287174e-01 -7.55505383e-01 5.38908958e-01
4.86209333e-01 -6.93646893e-02 -8.73438537e-01 -3.81644040e-01
7.49296784e-01 -8.47566500e-02 3.41355681e-01 -7.11189687e-01
7.15290308e-01 6.50065318e-02 1.85940154e-02 -4.45749730e-01
3.29261124e-01 -8.25049818e-01 2.38633692e-01 6.36809096e-02
-2.62140572e-01 -1.31633468e-02 -2.18223140e-01 9.01279271e-01
-3.47501278e-01 2.42099181e-01 4.68707711e-01 3.70540977e-01
-9.13837254e-01 6.71941996e-01 4.45045173e-01 2.22948715e-01
1.04706550e+00 -1.17474489e-01 -5.30308068e-01 -1.82727635e-01
-9.91250455e-01 6.13250732e-01 3.28984588e-01 4.73848432e-01
7.55194068e-01 -1.88143718e+00 -4.59123731e-01 -1.91486433e-01
4.11125660e-01 2.74247855e-01 2.84380823e-01 9.26644742e-01
-6.65215075e-01 6.57226324e-01 1.91373512e-01 -1.19060338e-01
-9.69743013e-01 9.02882457e-01 2.08505139e-01 -8.97976309e-02
-9.65602398e-01 5.99980056e-01 4.23352629e-01 -7.43507922e-01
2.06047252e-01 -1.90265909e-01 -2.12019160e-01 5.21473847e-02
6.96386993e-02 7.36672819e-01 -2.35887598e-02 -4.69832063e-01
-2.98894763e-01 6.12231314e-01 -4.32447977e-02 2.41193742e-01
1.16520751e+00 -2.00622067e-01 -2.33393103e-01 4.38941598e-01
1.44827628e+00 -9.37944576e-02 -8.56737971e-01 -4.01191413e-01
1.73160702e-01 -5.68683028e-01 6.87507614e-02 -1.62225991e-01
-8.49236906e-01 6.17917001e-01 1.14649452e-01 5.90655327e-01
5.15412807e-01 2.54477412e-01 9.41008806e-01 4.99446332e-01
5.33321500e-01 -1.06101370e+00 8.24768245e-02 3.94945920e-01
7.01966524e-01 -8.40327799e-01 -6.88186586e-02 -1.18222284e+00
-3.84786844e-01 9.53684330e-01 9.24456000e-01 1.12238564e-01
9.09693599e-01 -3.47827435e-01 -6.15806520e-01 -4.99127805e-01
-3.99973065e-01 -2.14679375e-01 4.32024360e-01 6.54308379e-01
2.66174763e-01 2.42773801e-01 -2.42666692e-01 8.40283334e-01
-4.72011626e-01 -3.68957400e-01 2.74998397e-01 4.22666520e-01
-4.03559208e-01 -6.92776620e-01 -6.71264753e-02 1.61972612e-01
1.57933295e-01 -1.47852555e-01 -5.08383036e-01 1.15842581e+00
-1.86120179e-02 6.56657100e-01 -1.35988310e-01 -7.42262363e-01
5.41868031e-01 -3.17533016e-02 4.76492822e-01 -4.04521972e-01
1.40874848e-01 -7.03173697e-01 -2.00138539e-02 -4.61781710e-01
1.59399167e-01 -1.63966194e-01 -1.57727861e+00 -8.20995450e-01
-3.84427249e-01 7.24822104e-01 3.96327674e-01 6.46441877e-01
5.49761951e-01 6.50196612e-01 5.44298887e-01 -1.61742777e-01
-3.84610772e-01 -5.27123451e-01 -1.10763836e+00 6.84957981e-01
-1.22722872e-01 -1.05385911e+00 -1.89180344e-01 -2.20000058e-01] | [8.647019386291504, 7.787686347961426] |
575f4b99-3674-4542-968c-d5bfc4f5fd74 | knowledge-distillation-for-action | 2004.07711 | null | https://arxiv.org/abs/2004.07711v2 | https://arxiv.org/pdf/2004.07711v2.pdf | Knowledge Distillation for Action Anticipation via Label Smoothing | Human capability to anticipate near future from visual observations and non-verbal cues is essential for developing intelligent systems that need to interact with people. Several research areas, such as human-robot interaction (HRI), assisted living or autonomous driving need to foresee future events to avoid crashes or help people. Egocentric scenarios are classic examples where action anticipation is applied due to their numerous applications. Such challenging task demands to capture and model domain's hidden structure to reduce prediction uncertainty. Since multiple actions may equally occur in the future, we treat action anticipation as a multi-label problem with missing labels extending the concept of label smoothing. This idea resembles the knowledge distillation process since useful information is injected into the model during training. We implement a multi-modal framework based on long short-term memory (LSTM) networks to summarize past observations and make predictions at different time steps. We perform extensive experiments on EPIC-Kitchens and EGTEA Gaze+ datasets including more than 2500 and 100 action classes, respectively. The experiments show that label smoothing systematically improves performance of state-of-the-art models for action anticipation. | ['Giovanni Maria Farinella', 'Pasquale Coscia', 'Guglielmo Camporese', 'Antonino Furnari', 'Lamberto Ballan'] | 2020-04-16 | null | null | null | null | ['action-anticipation'] | ['computer-vision'] | [ 3.11509818e-01 5.19831300e-01 -3.18021178e-01 -9.56487179e-01
-3.19587708e-01 1.58155158e-01 7.97417223e-01 -2.46439740e-01
-6.45932257e-01 9.84553695e-01 6.44678295e-01 4.68379147e-02
7.40059987e-02 -3.45794350e-01 -4.57249612e-01 -3.35221976e-01
-3.35950494e-01 5.60901046e-01 7.06116855e-02 -8.62199664e-02
2.57934898e-01 1.70336857e-01 -1.83686984e+00 6.85289264e-01
6.65036798e-01 8.24973404e-01 3.11621219e-01 6.65197313e-01
-4.38636914e-02 1.40974903e+00 -1.16795242e-01 -3.47831607e-01
-4.52567637e-02 3.78638469e-02 -1.01906931e+00 2.22948760e-01
8.35251715e-03 -3.69328558e-01 -3.32487971e-01 6.79895699e-01
2.49963090e-01 6.31700873e-01 8.36341858e-01 -1.58546400e+00
-5.93221843e-01 4.40407187e-01 -3.72377098e-01 -8.22131261e-02
6.16259515e-01 5.47632933e-01 5.13085008e-01 -6.17174089e-01
5.53290784e-01 1.61585462e+00 6.13865614e-01 1.06041992e+00
-8.17927837e-01 -4.45953310e-01 7.51314938e-01 1.06939614e+00
-1.06363213e+00 -5.16979158e-01 7.46606469e-01 -6.19559050e-01
1.27258587e+00 -2.85497885e-02 4.83117551e-01 1.44506121e+00
3.39570016e-01 1.26013827e+00 8.60321283e-01 -3.66644174e-01
9.42237228e-02 1.67670906e-01 3.14940393e-01 5.01615703e-01
-3.01486880e-01 2.85346746e-01 -8.61409843e-01 3.54535222e-01
5.60180768e-02 3.41988891e-01 1.44625023e-01 -6.97735175e-02
-1.47377384e+00 5.80133915e-01 4.22072500e-01 -1.85012415e-01
-7.35558093e-01 3.23326916e-01 5.31477332e-01 5.78427278e-02
4.83922869e-01 1.61211446e-01 -3.78512412e-01 -4.46850628e-01
-2.72710353e-01 2.58420885e-01 4.56535459e-01 1.00567544e+00
8.61629128e-01 -9.49768126e-02 -5.20668626e-01 6.16370618e-01
5.81080139e-01 4.17510927e-01 5.94374537e-01 -1.23824418e+00
5.16299427e-01 6.30284607e-01 5.00685871e-01 -5.57572126e-01
-8.90889883e-01 2.18785539e-01 -7.73270965e-01 3.76630962e-01
1.96123794e-01 -4.10651177e-01 -1.11105335e+00 1.81479895e+00
3.27522278e-01 5.48180521e-01 4.53894585e-01 7.17278302e-01
6.70340776e-01 6.34126246e-01 9.61036623e-01 -2.61282086e-01
1.28245389e+00 -1.28399849e+00 -9.45171833e-01 -9.75219548e-01
1.08760703e+00 -3.01477611e-01 8.44691992e-01 3.10797393e-01
-7.07279682e-01 -8.81276071e-01 -6.52576983e-01 -2.08117127e-01
-4.85545248e-01 1.71954840e-01 7.95298636e-01 1.19415693e-01
-7.98872709e-01 5.97913504e-01 -8.13790023e-01 -5.13232946e-01
5.06673992e-01 3.01719695e-01 -4.70253468e-01 -3.53437774e-02
-1.42610586e+00 1.39687788e+00 6.72861874e-01 3.30828846e-01
-7.94106305e-01 -3.26741606e-01 -1.19563556e+00 -1.26797274e-01
3.69300127e-01 -6.86333060e-01 1.59736502e+00 -9.26799655e-01
-1.49849129e+00 6.69813991e-01 -5.21185100e-01 -9.12145734e-01
4.25714463e-01 -6.51218653e-01 -7.03870058e-01 -3.11087996e-01
1.41167015e-01 1.29684150e+00 7.09999859e-01 -8.38687003e-01
-1.05426455e+00 -2.89967358e-01 -3.20188776e-02 6.40145779e-01
2.86709927e-02 7.40280375e-03 1.40669271e-01 -1.12491529e-02
-2.87731051e-01 -1.21812284e+00 -3.31125349e-01 -3.02689135e-01
-3.18734765e-01 -8.81759048e-01 9.63324547e-01 -5.26194394e-01
1.00015152e+00 -1.97618747e+00 -1.70227706e-01 -3.78725946e-01
3.72089557e-02 3.05684417e-01 -1.99072689e-01 3.52428079e-01
-2.05744460e-01 -3.45725209e-01 2.16071770e-01 -7.74508715e-01
3.20265263e-01 2.57725000e-01 -4.20651227e-01 2.51128018e-01
2.00573534e-01 1.10782433e+00 -9.77707326e-01 -5.60046911e-01
6.59793377e-01 3.77417266e-01 -5.61966114e-02 3.72051448e-01
-4.69393343e-01 5.98466694e-01 -4.50910330e-01 4.21286970e-01
2.92896807e-01 -2.45172992e-01 -1.45727351e-01 2.03674436e-01
-1.86022267e-01 2.66344190e-01 -8.01425159e-01 1.76425123e+00
-4.39508528e-01 7.90006280e-01 -6.95915222e-01 -7.72371709e-01
6.93333089e-01 5.10524452e-01 2.14238256e-01 -7.06917763e-01
-5.46106584e-02 -3.39504510e-01 -1.79179177e-01 -1.07007205e+00
5.53344071e-01 -1.54916540e-01 -1.37759104e-01 4.06688154e-01
-2.95156240e-01 4.32775319e-01 8.35791454e-02 6.88976347e-02
7.60399759e-01 5.35079539e-01 3.76165122e-01 3.67616296e-01
5.85764349e-01 1.08207315e-01 8.55069220e-01 6.53466582e-01
-7.05596507e-01 2.00061634e-01 1.62331477e-01 -8.59070241e-01
-5.89799047e-01 -7.10118592e-01 3.95781696e-01 1.40318274e+00
7.89310113e-02 -1.81952417e-01 -3.06434751e-01 -7.74016619e-01
-2.97471166e-01 1.55617809e+00 -7.60916352e-01 -4.80804622e-01
-5.44684947e-01 -3.12184185e-01 -3.00350897e-02 8.66070390e-01
6.85382247e-01 -1.81841469e+00 -1.05449593e+00 1.41463310e-01
-4.48074996e-01 -1.00672948e+00 -1.45481646e-01 -1.12213276e-01
-4.56800163e-01 -8.04403186e-01 -5.11340916e-01 -6.39394522e-01
3.48071992e-01 1.68352693e-01 1.01721644e+00 -4.14156467e-01
-1.28394677e-04 6.04032695e-01 -2.70059884e-01 -8.24169159e-01
-2.21917704e-01 -2.68225759e-01 3.73013079e-01 2.98745707e-02
1.09727848e+00 -4.00580317e-01 -6.00403488e-01 1.27946943e-01
-2.12775603e-01 6.54120326e-01 6.09722018e-01 7.35866189e-01
2.77717352e-01 -1.78941280e-01 8.01804245e-01 -7.55669296e-01
3.82656038e-01 -5.97089648e-01 -1.19775906e-01 5.58398247e-01
-4.39751685e-01 1.54619917e-01 2.64234751e-01 -6.18289173e-01
-1.86319935e+00 5.00505090e-01 1.23096509e-02 -1.83321193e-01
-7.92712510e-01 3.94225985e-01 -8.07326734e-02 4.45185155e-01
4.78801817e-01 5.00297211e-02 -1.73937291e-01 -1.58085540e-01
6.37968838e-01 8.12701643e-01 5.99158466e-01 -6.65539205e-02
9.96763185e-02 4.65298027e-01 -3.70446034e-02 -6.21092260e-01
-1.24066031e+00 -6.04630828e-01 -8.22867811e-01 -8.06180179e-01
1.07497823e+00 -1.06887519e+00 -1.25811887e+00 5.83829641e-01
-1.48825812e+00 -6.44803107e-01 -2.14634940e-01 6.86437309e-01
-9.52289939e-01 2.00165033e-01 -1.95148870e-01 -1.29553759e+00
-1.32594973e-01 -8.53028834e-01 9.31752384e-01 5.67979395e-01
-5.50396025e-01 -9.95340049e-01 1.37966648e-01 3.54404688e-01
8.79273266e-02 2.16326758e-01 4.91646290e-01 -6.16414607e-01
-4.93831903e-01 -2.04667121e-01 -1.33677632e-01 7.76367038e-02
-6.20057210e-02 -5.44208407e-01 -1.15645719e+00 2.00669140e-01
-2.53201704e-02 -6.30437076e-01 1.07121730e+00 4.95011181e-01
8.69692028e-01 -2.77862012e-01 -7.14922547e-01 8.80305190e-03
6.98411107e-01 3.02848279e-01 8.61626208e-01 3.36161286e-01
8.06385636e-01 1.25180399e+00 1.30901563e+00 5.53029656e-01
1.00386047e+00 5.37800014e-01 4.75188732e-01 3.54463339e-01
-9.75313783e-02 -2.73826629e-01 5.86047232e-01 1.36070386e-01
-4.02903445e-02 -2.82533318e-01 -1.06568706e+00 6.78027749e-01
-2.42620444e+00 -1.35553241e+00 -2.08115458e-01 2.00304937e+00
5.59755445e-01 1.36288390e-01 5.13659697e-03 -2.75974482e-01
7.19585180e-01 2.11054757e-01 -8.42353463e-01 -3.80753726e-01
3.45567197e-01 -7.05464065e-01 2.51843899e-01 5.39942980e-01
-1.44619632e+00 1.22718894e+00 5.99838638e+00 4.72165555e-01
-8.09178829e-01 1.36535883e-01 7.14444458e-01 -1.28794126e-02
2.80598253e-01 -1.51436895e-01 -1.08190191e+00 4.26054835e-01
1.35317516e+00 -1.58307895e-01 8.61957818e-02 1.07004845e+00
4.12995964e-01 -5.05842388e-01 -1.36199999e+00 1.13204443e+00
-2.23144935e-03 -9.46576178e-01 -2.34249949e-01 -3.66058946e-01
6.54390693e-01 1.48456484e-01 1.67960152e-01 7.72681534e-01
5.47922850e-01 -1.08477056e+00 6.83311045e-01 1.37438023e+00
5.66286564e-01 -6.93531513e-01 6.01102889e-01 7.55405307e-01
-1.09623981e+00 -5.26673377e-01 -3.14631552e-01 -5.98957121e-01
8.46941471e-01 1.50589094e-01 -1.14164472e+00 1.62954926e-01
5.49997687e-01 1.15662396e+00 -5.13140559e-01 8.55093837e-01
-4.99378175e-01 2.94754624e-01 -5.06174527e-02 -1.01258464e-01
2.37927318e-01 -2.27077641e-02 4.09203768e-01 9.15717125e-01
2.99213350e-01 2.72209257e-01 1.80211172e-01 4.31254625e-01
3.22547048e-01 -3.97105098e-01 -8.47259521e-01 4.47907932e-02
2.65608937e-01 9.55446720e-01 -1.16194405e-01 -6.08032763e-01
-6.69498861e-01 1.04753196e+00 5.30124068e-01 5.23398638e-01
-8.85369420e-01 -1.09897114e-01 7.64137447e-01 -1.15399227e-01
-1.17066987e-01 -1.32698640e-01 -3.13490659e-01 -8.33929241e-01
-1.85405672e-01 -3.47663581e-01 5.57820320e-01 -1.17476320e+00
-1.06412852e+00 5.68465829e-01 1.12100005e-01 -1.43843484e+00
-1.05977452e+00 -4.86130953e-01 -7.08455861e-01 7.19607830e-01
-1.70263648e+00 -1.48294377e+00 -4.63473141e-01 3.72033894e-01
1.00456393e+00 -1.01378389e-01 8.51680577e-01 -1.26259282e-01
-4.55558121e-01 -6.49292441e-03 -2.97682941e-01 -3.35098654e-01
9.13401246e-01 -1.13897860e+00 4.88411427e-01 7.14505494e-01
-1.76617354e-01 9.65854675e-02 8.73070836e-01 -1.02520406e+00
-6.25810564e-01 -1.42939782e+00 1.54557443e+00 -7.04565525e-01
4.25518572e-01 -4.05878313e-02 -8.52346063e-01 1.36902261e+00
4.37856615e-01 -2.65786231e-01 6.67236209e-01 2.77036339e-01
1.35700777e-01 1.57118440e-01 -8.34894598e-01 8.60162556e-01
1.24072921e+00 -3.22289884e-01 -9.06822622e-01 6.27244055e-01
7.86881268e-01 -1.80088639e-01 -3.05171221e-01 3.91232729e-01
4.75696325e-01 -1.08495998e+00 8.91771436e-01 -8.27250957e-01
2.15171650e-01 -1.47573695e-01 2.57058859e-01 -1.32887125e+00
-4.09151018e-01 -5.76939642e-01 -3.63729298e-01 1.04522121e+00
3.60027045e-01 -4.95612234e-01 8.53210986e-01 1.41162789e+00
-3.57550979e-01 -4.65306431e-01 -7.36867309e-01 -5.74850500e-01
-5.20126700e-01 -7.34943986e-01 6.15396380e-01 5.08632600e-01
3.02893311e-01 4.08238918e-01 -1.01570261e+00 -3.72176012e-03
4.64716196e-01 -2.37109452e-01 7.93237627e-01 -1.40542793e+00
3.01780164e-01 6.07778803e-02 -3.23627472e-01 -1.18155468e+00
7.11019576e-01 -4.31362808e-01 4.64181930e-01 -1.77794802e+00
7.14880154e-02 -8.54897946e-02 -2.83887953e-01 7.56272018e-01
-2.14530215e-01 -2.41721675e-01 1.22126766e-01 1.10493451e-01
-1.29071677e+00 1.05330682e+00 9.11034882e-01 -6.95081353e-02
-3.68809938e-01 3.54139894e-01 -3.32253516e-01 1.22111154e+00
7.39549220e-01 -3.75397205e-01 -5.97353220e-01 -3.85352165e-01
2.04736710e-01 1.38586566e-01 4.21561837e-01 -1.28574085e+00
5.30562639e-01 -5.59177220e-01 3.78296107e-01 -1.00118363e+00
9.02240574e-01 -7.81125247e-01 1.56328902e-02 3.81235629e-01
-7.11690664e-01 -1.88938662e-01 2.29105614e-02 9.43756878e-01
-3.20504941e-02 -1.32972464e-01 4.05559540e-01 -1.15572631e-01
-1.66914165e+00 2.72030592e-01 -7.37184107e-01 -3.84484619e-01
1.39104283e+00 -2.39951521e-01 -4.08792317e-01 -6.95035696e-01
-1.08738148e+00 9.20578778e-01 -6.30335063e-02 7.04710066e-01
7.06391573e-01 -1.45259845e+00 -5.78894019e-01 -1.75581649e-02
3.95598978e-01 -1.03732683e-01 8.63253295e-01 8.44903231e-01
1.44728780e-01 7.47023225e-01 -4.51981723e-01 -4.34028834e-01
-1.33919489e+00 7.60383844e-01 -4.47394885e-02 -3.10769498e-01
-6.37086451e-01 9.57482278e-01 4.69155639e-01 -3.24953467e-01
4.98159975e-01 -1.18951714e-02 -8.97573531e-01 8.84146094e-02
8.58759701e-01 5.76859057e-01 -6.00363553e-01 -8.33948493e-01
-4.37367260e-01 9.82330367e-02 -2.09370971e-01 -1.45711051e-02
1.28050423e+00 -6.94403827e-01 3.34107459e-01 8.08916032e-01
7.85066247e-01 -9.96067762e-01 -1.84614646e+00 -4.12112176e-01
3.43006968e-01 -2.42371127e-01 -1.54097974e-01 -8.91707599e-01
-3.82781595e-01 1.20405829e+00 8.07778478e-01 -1.30211502e-01
9.06035960e-01 -4.65411916e-02 7.49382436e-01 7.39033878e-01
3.09971392e-01 -1.45729542e+00 1.77037492e-01 6.84691727e-01
9.97229278e-01 -1.74374914e+00 -1.50931209e-01 -9.37825516e-02
-1.50796497e+00 9.11874354e-01 1.04703057e+00 2.61309892e-01
4.43087786e-01 -3.66690397e-01 2.29144216e-01 -7.13227838e-02
-1.14969766e+00 -4.95438367e-01 5.00160038e-01 8.82404983e-01
4.63099070e-02 2.36127749e-01 -1.18422527e-02 5.97579181e-01
1.51460961e-01 2.25155041e-01 4.17826265e-01 8.06375861e-01
-6.73763216e-01 -7.95963168e-01 -1.60271838e-01 4.64833379e-01
1.44449607e-01 2.03529999e-01 -4.00013775e-02 4.34924275e-01
2.50930399e-01 1.14212751e+00 1.25637442e-01 -4.65068549e-01
1.33790776e-01 6.40129864e-01 1.96201000e-02 -6.87214136e-01
1.75096709e-02 -4.70151871e-01 4.07875985e-01 -8.32865536e-01
-8.18400323e-01 -9.62477028e-01 -1.46808946e+00 -9.88883451e-02
5.80475107e-02 -2.53389210e-01 4.38930452e-01 1.39573586e+00
5.55826485e-01 5.83640814e-01 2.46781871e-01 -1.10673368e+00
-2.52614796e-01 -1.32728124e+00 -4.26719546e-01 4.28107083e-01
2.19527528e-01 -1.08365333e+00 -7.75153190e-02 3.86152327e-01] | [7.921581268310547, 0.4516451060771942] |
a58d5a2e-12b4-4a26-af08-a8e1b984da42 | efficient-vertical-federated-learning-with | 2305.11236 | null | https://arxiv.org/abs/2305.11236v1 | https://arxiv.org/pdf/2305.11236v1.pdf | Efficient Vertical Federated Learning with Secure Aggregation | The majority of work in privacy-preserving federated learning (FL) has been focusing on horizontally partitioned datasets where clients share the same sets of features and can train complete models independently. However, in many interesting problems, such as financial fraud detection and disease detection, individual data points are scattered across different clients/organizations in vertical federated learning. Solutions for this type of FL require the exchange of gradients between participants and rarely consider privacy and security concerns, posing a potential risk of privacy leakage. In this work, we present a novel design for training vertical FL securely and efficiently using state-of-the-art security modules for secure aggregation. We demonstrate empirically that our method does not impact training performance whilst obtaining 9.1e2 ~3.8e4 speedup compared to homomorphic encryption (HE). | ['Nicholas D. Lane', 'Pedro Porto Buarque de Gusmão', 'Chenyang Ma', 'Wanru Zhao', 'Heng Pan', 'Xinchi Qiu'] | 2023-05-18 | null | null | null | null | ['fraud-detection'] | ['miscellaneous'] | [-8.64127055e-02 9.76992697e-02 -1.98509097e-01 -5.08339465e-01
-8.14731121e-01 -8.68148148e-01 2.47218147e-01 5.02981246e-01
-5.81251800e-01 9.12165999e-01 6.93842545e-02 -6.97465181e-01
-1.12360418e-01 -1.03655803e+00 -9.52377141e-01 -8.35528016e-01
-4.21502203e-01 8.61561149e-02 -1.73184380e-01 1.82125315e-01
-8.81244093e-02 5.49162805e-01 -1.17851281e+00 8.86926830e-01
4.66476202e-01 1.14361250e+00 -7.80067801e-01 4.20659959e-01
-4.72851433e-02 7.76548624e-01 -6.06271148e-01 -1.14538133e+00
9.14403737e-01 -9.30469334e-02 -1.03360987e+00 -2.06464514e-01
6.26066029e-01 -6.88020825e-01 -3.19379568e-01 7.61976242e-01
3.21719974e-01 -2.86273122e-01 8.99720713e-02 -1.50409675e+00
-2.34747231e-01 7.03284860e-01 -6.23006523e-01 -2.07083553e-01
4.35980596e-03 6.71538636e-02 1.01997662e+00 -3.23406547e-01
7.11861849e-01 7.94848621e-01 7.93295979e-01 5.94245732e-01
-1.38286567e+00 -1.09531283e+00 -1.17850244e-01 -9.08997804e-02
-1.12169671e+00 -2.90426135e-01 2.12605298e-01 -3.70771959e-02
8.99176717e-01 7.50717461e-01 3.57653201e-01 6.60148442e-01
1.49375543e-01 7.25311160e-01 1.11157477e+00 -2.20109567e-01
3.10511291e-01 7.18111277e-01 2.30993498e-02 5.68150461e-01
8.36807132e-01 2.49822866e-02 -6.21917486e-01 -1.09878802e+00
2.77023941e-01 5.51621795e-01 -2.06779435e-01 -7.55533576e-01
-8.83234501e-01 1.08826911e+00 4.86588866e-01 -8.06198493e-02
-9.94528607e-02 1.59578353e-01 7.42971241e-01 8.50285888e-01
3.27650189e-01 1.83875412e-01 -7.97811031e-01 4.21307445e-01
-8.27048957e-01 5.68792343e-01 1.15253079e+00 8.21152568e-01
9.10024464e-01 -5.23084462e-01 1.06242396e-01 1.15944758e-01
1.38478085e-01 -1.38005819e-02 2.90911764e-01 -1.03415525e+00
9.38605905e-01 8.27237606e-01 1.30674183e-01 -7.72627890e-01
1.10703453e-01 -1.72979787e-01 -8.68849993e-01 2.46832788e-01
5.16325355e-01 -6.19634032e-01 -2.70075381e-01 1.68131447e+00
6.11853659e-01 -2.44649768e-01 4.54682499e-01 3.59755784e-01
5.78638501e-02 2.46060938e-01 5.41485548e-02 -1.76319443e-02
1.32365191e+00 -6.77822709e-01 -4.43747818e-01 8.68823230e-02
1.04235613e+00 -3.28516215e-01 5.36186457e-01 3.07094157e-01
-1.11515200e+00 3.67629737e-01 -9.24507439e-01 -1.81165174e-01
-5.25220931e-01 -2.84970850e-01 1.09185565e+00 1.19791210e+00
-9.47231591e-01 5.88572741e-01 -9.21365321e-01 -1.09997891e-01
1.04086089e+00 9.08660233e-01 -1.00186515e+00 -1.05509266e-01
-9.29155767e-01 2.40504846e-01 2.58433938e-01 -5.57382345e-01
-6.00566089e-01 -9.68798637e-01 -5.80957115e-01 3.17115575e-01
-2.13266965e-02 -8.49145353e-01 1.17299032e+00 -6.03235722e-01
-8.20417583e-01 9.60308313e-01 1.22435607e-01 -1.06606638e+00
8.87368262e-01 -1.36929587e-01 -1.26753941e-01 6.14806488e-02
-3.72355968e-01 2.74216056e-01 3.87484252e-01 -1.10756862e+00
-9.40442562e-01 -9.11347926e-01 1.59659415e-01 -1.03727341e-01
-1.01657832e+00 2.20551163e-01 2.22424150e-01 -3.64011288e-01
-2.04950765e-01 -8.13983023e-01 -3.91777962e-01 5.00385046e-01
-7.57922754e-02 4.83528227e-02 1.35539949e+00 -4.93406862e-01
1.05008924e+00 -2.21514702e+00 -6.24816775e-01 4.04058903e-01
3.24864745e-01 2.28063285e-01 4.38001692e-01 7.20980704e-01
4.45730001e-01 4.28800970e-01 -3.70280802e-01 -6.28097713e-01
6.95855618e-02 1.14155345e-01 -5.98222196e-01 7.17486858e-01
-2.78803229e-01 8.04476202e-01 -4.37348574e-01 -4.76703912e-01
-2.18460634e-01 3.70265514e-01 -9.36126888e-01 1.45248935e-01
-4.71275374e-02 1.07347600e-01 -3.62461746e-01 4.56421256e-01
1.12475479e+00 -4.49158907e-01 5.91079712e-01 3.37892979e-01
8.21817368e-02 3.38608980e-01 -1.13535774e+00 1.57795298e+00
-4.88538176e-01 2.36148208e-01 5.51251769e-01 -5.92042923e-01
5.25591671e-01 6.89755559e-01 5.29767156e-01 -2.89969981e-01
5.99611513e-02 2.47598320e-01 -4.51127470e-01 -6.86759576e-02
1.65648639e-01 1.65302102e-02 -2.16914296e-01 1.10406458e+00
-6.82656839e-02 6.16421700e-01 -5.03049850e-01 1.63897708e-01
1.35031712e+00 -4.78456706e-01 1.54296771e-01 -1.55447885e-01
4.46177840e-01 -1.68552101e-01 6.61957085e-01 7.57677257e-01
-4.12673876e-02 3.11063915e-01 4.56758022e-01 -8.02899182e-01
-7.85855472e-01 -8.46679151e-01 -2.66961873e-01 9.38313127e-01
-1.67968467e-01 -6.51082277e-01 -8.71396065e-01 -1.15735984e+00
6.94331646e-01 3.30207735e-01 -4.11792338e-01 8.69793743e-02
-5.59204221e-01 -8.27243686e-01 8.88884604e-01 3.36522311e-01
7.32358754e-01 -6.48117423e-01 -1.08693171e+00 -6.05540723e-02
3.38214412e-02 -7.33401120e-01 -3.74734491e-01 2.87028551e-01
-1.09969079e+00 -1.08407760e+00 -1.18398540e-01 -4.58556205e-01
8.71862948e-01 4.05863672e-01 8.77884030e-01 5.75479269e-02
-3.76065701e-01 1.67624950e-01 1.09661929e-01 -6.06164336e-01
-1.86747432e-01 1.64117381e-01 -8.88399631e-02 3.36358339e-01
6.34566963e-01 -5.32892644e-01 -1.00060499e+00 1.68985620e-01
-9.97066021e-01 -2.89223641e-01 3.14285934e-01 7.24960148e-01
3.57592583e-01 1.63868740e-02 1.87141344e-01 -1.58446717e+00
4.98088181e-01 -5.43563128e-01 -6.83777153e-01 4.09452498e-01
-9.07907426e-01 8.72256532e-02 9.31796551e-01 -1.13968283e-01
-9.87664282e-01 3.36897701e-01 2.02547804e-01 -4.23648655e-01
-2.04595819e-01 -5.05262613e-02 -2.55845159e-01 -5.85691631e-01
5.91722071e-01 -2.51415037e-02 2.40524590e-01 -5.33974409e-01
2.78598934e-01 9.48941171e-01 1.02613725e-01 -5.12302220e-01
6.93747282e-01 9.73030567e-01 1.68626070e-01 -9.03331265e-02
-3.08193564e-01 -3.21977586e-01 -2.49223545e-01 7.83885181e-01
3.90856653e-01 -9.37035501e-01 -1.28708494e+00 3.20724905e-01
-8.60182583e-01 7.57473484e-02 -6.35955572e-01 1.53755844e-01
-3.22355151e-01 4.25242454e-01 -7.97314763e-01 -6.69843912e-01
-1.02437663e+00 -8.25891256e-01 6.20951116e-01 2.57311072e-02
-4.03547697e-02 -8.65368128e-01 2.63837352e-02 6.84441745e-01
6.04892015e-01 4.69757080e-01 7.91160047e-01 -9.15352345e-01
-8.67620885e-01 -6.41662240e-01 6.49001375e-02 2.86461890e-01
1.10337146e-01 -6.60593390e-01 -1.13644135e+00 -9.44480598e-01
3.08481514e-01 -6.13276839e-01 6.49724066e-01 -2.49656036e-01
1.55122995e+00 -1.26006687e+00 -3.43116581e-01 1.10117364e+00
1.57140708e+00 -9.19393599e-02 2.90968299e-01 3.04737419e-01
3.47218394e-01 8.62208903e-01 2.17873812e-01 9.10153925e-01
3.27514082e-01 1.72855854e-01 3.39798331e-01 -1.30912244e-01
6.57134116e-01 -2.94220567e-01 1.89227715e-01 -1.16771720e-01
3.80847543e-01 -1.44235454e-02 -6.99797690e-01 5.13339639e-01
-1.77821219e+00 -8.96207333e-01 1.38743639e-01 2.57181668e+00
9.60245967e-01 -3.69381100e-01 8.25096294e-02 7.36729875e-02
3.62054169e-01 -5.20370044e-02 -5.39773345e-01 -7.27308989e-01
-4.06284072e-02 3.55290622e-01 1.23933077e+00 1.49615735e-01
-1.06481564e+00 3.53024334e-01 5.42351389e+00 4.15680200e-01
-1.01766634e+00 3.62783611e-01 1.02997983e+00 -6.48445249e-01
-3.74337196e-01 1.45256087e-01 -8.42515886e-01 2.75482595e-01
1.05739248e+00 -5.42846799e-01 3.96779269e-01 9.54220593e-01
-5.13457179e-01 3.10834736e-01 -1.25806212e+00 8.44092429e-01
-4.28885043e-01 -1.58402431e+00 -1.31902993e-02 7.18312442e-01
8.53730559e-01 4.87677269e-02 3.32371324e-01 -6.10884912e-02
6.42769217e-01 -1.08667338e+00 2.61089772e-01 -1.62806883e-01
8.00014198e-01 -1.13678885e+00 7.28213787e-01 5.49568653e-01
-7.98673332e-01 -6.35146916e-01 -3.64978641e-01 3.08342218e-01
-3.60855877e-01 1.61307007e-01 -7.79280901e-01 6.59272432e-01
1.01689041e+00 -1.25058860e-01 -3.38229090e-01 7.46695757e-01
2.59074807e-01 5.75693667e-01 -6.20944262e-01 2.72479653e-01
2.87742108e-01 2.10222155e-02 -4.33959067e-02 1.16980278e+00
2.09857866e-01 1.06881984e-01 -1.83709621e-01 5.31207740e-01
-7.54593313e-01 2.83771664e-01 -7.87062705e-01 2.13252470e-01
4.09123808e-01 1.14693058e+00 -7.46884942e-02 -2.23820992e-02
-7.36796260e-01 9.20023561e-01 4.07613963e-01 -4.06922027e-02
-2.93780923e-01 -3.49587351e-01 1.20479679e+00 4.10342008e-01
4.72898036e-01 1.41324922e-01 -4.81134981e-01 -1.27667582e+00
2.23987490e-01 -1.01559424e+00 1.10505641e+00 5.00554800e-01
-1.46920967e+00 3.21875215e-01 -2.59683073e-01 -9.51480806e-01
-1.40308082e-01 -1.68393552e-01 -5.02167702e-01 8.66093278e-01
-1.30464518e+00 -1.41153419e+00 4.28771786e-02 1.17416418e+00
-4.20130014e-01 -2.15280011e-01 1.17502654e+00 2.87797719e-01
-2.10390672e-01 1.34499681e+00 6.25426352e-01 8.39527026e-02
6.35807335e-01 -1.14217973e+00 5.65656960e-01 7.48859882e-01
1.27696767e-01 1.12814498e+00 3.14317673e-01 -4.88653898e-01
-1.66767132e+00 -1.35111010e+00 1.26094294e+00 -4.08684611e-01
1.62363872e-01 -8.52123201e-01 -9.05543804e-01 1.10029364e+00
2.59369850e-01 7.46585965e-01 1.28136027e+00 3.32243107e-02
-7.52853930e-01 -5.51756918e-01 -2.10876060e+00 2.69216985e-01
8.59054387e-01 -6.95222855e-01 2.25227863e-01 4.93569881e-01
6.78574324e-01 -3.36739928e-01 -1.01297450e+00 1.60935163e-01
6.56966329e-01 -1.20242012e+00 9.33592081e-01 -9.57843542e-01
-1.81706533e-01 2.11697780e-02 -1.60504103e-01 -5.51865518e-01
8.66201147e-02 -9.87094104e-01 -6.18678153e-01 1.24754834e+00
2.85863310e-01 -1.14822614e+00 1.49183989e+00 1.25361478e+00
7.99224377e-01 -7.63685107e-01 -1.20847821e+00 -5.15928447e-01
3.67583036e-01 5.78011833e-02 1.17923951e+00 1.07668364e+00
6.16021380e-02 -5.93658626e-01 -3.68109047e-01 3.50104511e-01
1.08235836e+00 4.59952652e-01 9.76081550e-01 -9.83135283e-01
-3.82846266e-01 1.48870438e-01 -3.77253801e-01 -3.81333381e-01
4.43296134e-02 -9.72307503e-01 -7.95542181e-01 -9.69746768e-01
3.02068174e-01 -6.66402698e-01 -6.87816143e-01 9.04174447e-01
2.29669228e-01 8.18958431e-02 1.72097117e-01 1.04333580e-01
-4.51340318e-01 1.80734321e-01 4.31949973e-01 -2.27215495e-02
-3.07273753e-02 3.44536930e-01 -1.10167885e+00 7.21849427e-02
1.00184393e+00 -7.19884694e-01 -5.02959967e-01 -4.15778935e-01
8.07653554e-03 9.19956788e-02 4.10885870e-01 -8.76584172e-01
5.67979038e-01 -1.11793749e-01 3.67245972e-01 -2.84609795e-01
-3.90535630e-02 -1.30676496e+00 5.91574728e-01 6.79034472e-01
-3.79990578e-01 2.28856415e-01 9.14089009e-02 6.39918029e-01
-1.89022541e-01 1.33827642e-01 6.78545177e-01 -3.28064740e-01
1.25191256e-01 6.63957715e-01 1.58194363e-01 -5.23655266e-02
1.44349027e+00 -6.72572851e-02 -5.00326335e-01 -1.71819866e-01
-2.17215434e-01 4.42731917e-01 9.98382390e-01 -1.28073865e-04
4.16801512e-01 -8.49886537e-01 -6.32293522e-01 5.71979582e-01
-1.11409210e-01 1.47958174e-01 2.95684725e-01 3.82296085e-01
-5.49044311e-01 6.36776090e-01 -1.80890888e-01 -3.21661425e-03
-1.75035644e+00 6.47298276e-01 2.68789947e-01 -5.78585207e-01
-8.01292539e-01 9.02540028e-01 1.26047000e-01 -6.26278043e-01
5.45072973e-01 9.47331637e-02 5.72971225e-01 -8.23440328e-02
5.76040328e-01 3.96127224e-01 3.56306970e-01 -1.36631742e-01
-3.60799015e-01 -2.00138435e-01 -5.89528263e-01 -7.26824347e-03
1.48225343e+00 1.22061625e-01 -3.92249554e-01 -2.93616503e-01
1.63181162e+00 1.72307760e-01 -1.23511410e+00 -3.62584114e-01
5.73225021e-02 -9.48169589e-01 -8.91621560e-02 -5.23717582e-01
-1.32475662e+00 7.48191059e-01 7.80910254e-01 -1.29898805e-02
1.16218328e+00 -2.00414583e-01 1.17495275e+00 4.98221338e-01
8.88519466e-01 -6.66865885e-01 -6.74449801e-01 -2.51283109e-01
2.13865325e-01 -9.97459710e-01 2.78916746e-01 -2.06330672e-01
-5.74271858e-01 7.97339499e-01 5.08289814e-01 1.37891442e-01
6.86010957e-01 5.85150361e-01 -7.91339960e-04 -1.72472578e-02
-1.19929028e+00 6.97169960e-01 -5.22755086e-01 2.97426164e-01
9.81250405e-02 6.09472878e-02 -2.62870193e-01 6.49271011e-01
-2.09516391e-01 2.57840622e-02 1.64339423e-01 1.62917614e+00
2.28765368e-01 -1.69192028e+00 -1.78289637e-01 4.64473128e-01
-1.29175782e+00 9.60801095e-02 -3.50812554e-01 4.69221085e-01
1.24873899e-01 5.67148387e-01 1.24775302e-02 -1.39656410e-01
7.50599056e-02 1.83256119e-01 1.88358992e-01 -2.83158094e-01
-1.57074106e+00 -5.65163136e-01 -1.72959447e-01 -6.15924239e-01
2.90351547e-02 -8.73077035e-01 -1.12726986e+00 -8.46052885e-01
-1.70890629e-01 3.98469776e-01 7.15587616e-01 1.36899874e-01
8.63134086e-01 -4.23086464e-01 1.06712365e+00 4.24471758e-02
-1.19567156e+00 -1.55013964e-01 -7.35460579e-01 6.42473161e-01
6.47266269e-01 1.82096064e-01 -4.95439380e-01 -1.73806623e-01] | [5.883149147033691, 6.6911845207214355] |
e655a5dc-b06d-4ca6-bf13-4f0a0e2a912b | curriculum-knowledge-switching-for-pancreas | 2306.12651 | null | https://arxiv.org/abs/2306.12651v1 | https://arxiv.org/pdf/2306.12651v1.pdf | Curriculum Knowledge Switching for Pancreas Segmentation | Pancreas segmentation is challenging due to the small proportion and highly changeable anatomical structure. It motivates us to propose a novel segmentation framework, namely Curriculum Knowledge Switching (CKS) framework, which decomposes detecting pancreas into three phases with different difficulty extent: straightforward, difficult, and challenging. The framework switches from straightforward to challenging phases and thereby gradually learns to detect pancreas. In addition, we adopt the momentum update parameter updating mechanism during switching, ensuring the loss converges gradually when the input dataset changes. Experimental results show that different neural network backbones with the CKS framework achieved state-of-the-art performance on the NIH dataset as measured by the DSC metric. | ['Xueming Wen', 'Saisai Wang', 'Mingxuan Zhang', 'Zhibo Tian', 'Kun Zhan', 'Yumou Tang'] | 2023-06-22 | null | null | null | null | ['pancreas-segmentation'] | ['medical'] | [-7.55643025e-02 1.02044448e-01 -3.93097818e-01 -2.94799775e-01
-4.43462938e-01 -6.14112794e-01 1.63771808e-01 1.38432370e-03
-5.21299243e-01 5.77570617e-01 9.19683650e-02 -2.40602463e-01
-2.05785334e-01 -5.09302974e-01 -7.21121848e-01 -7.92136669e-01
-9.21423435e-02 3.39935243e-01 5.47520995e-01 1.01674534e-01
-1.78224757e-01 1.39101461e-01 -6.04040146e-01 -1.83933675e-01
1.39563620e+00 8.06271195e-01 2.75788248e-01 3.65950853e-01
-1.92238644e-01 8.46513748e-01 -1.75031081e-01 -4.21846658e-01
4.92411196e-01 -5.07239938e-01 -7.48425066e-01 2.17949506e-02
3.67421061e-01 -2.92912245e-01 -6.15622699e-01 1.37542975e+00
3.99076730e-01 -5.71490601e-02 4.06558067e-01 -1.13482094e+00
-6.12744153e-01 9.25391018e-01 -3.89185518e-01 4.20529872e-01
8.10643956e-02 2.84559250e-01 5.16789734e-01 -4.23099846e-01
6.06208980e-01 7.67764330e-01 9.11050916e-01 5.69315374e-01
-1.30080914e+00 -6.70839310e-01 7.65670657e-01 2.58500785e-01
-1.03575480e+00 1.22068733e-01 9.16639566e-01 -4.33614165e-01
2.79492378e-01 -2.42657177e-02 1.27293813e+00 1.15523148e+00
1.04381002e-01 1.20654738e+00 1.22643173e+00 -2.55889492e-03
1.63017854e-01 -5.70077896e-02 5.94216347e-01 1.02797294e+00
4.36176270e-01 1.04179040e-01 -1.20198548e-01 1.14695787e-01
1.01937389e+00 -2.09988817e-03 -7.23092973e-01 -8.36668551e-01
-1.43776011e+00 7.31526077e-01 5.40532470e-01 -3.16966102e-02
-5.00598967e-01 -7.90209472e-02 5.17571390e-01 4.57855463e-01
-1.20786525e-01 2.70874470e-01 -6.37288332e-01 4.32671607e-02
-7.67708480e-01 -3.35848667e-02 1.10466433e+00 1.08890605e+00
8.22530687e-02 -8.23511332e-02 -1.74549907e-01 5.90244710e-01
4.33297366e-01 1.24574445e-01 6.52193487e-01 -9.35139120e-01
1.88922793e-01 7.35489786e-01 -3.98986995e-01 -5.53454161e-01
-7.11201072e-01 -7.64293134e-01 -1.12325871e+00 -7.06131160e-02
5.71170866e-01 -3.61415505e-01 -1.41610098e+00 1.79997003e+00
5.10370731e-01 5.67744136e-01 1.12901092e-01 1.10971248e+00
1.30996108e+00 1.72290996e-01 3.43444079e-01 -2.37164557e-01
1.25369787e+00 -1.36137462e+00 -5.67021310e-01 1.96190178e-01
2.52706170e-01 -3.23779017e-01 1.04826677e+00 3.97569239e-01
-1.12773037e+00 -2.36935154e-01 -1.02181876e+00 2.37030193e-01
-2.63218128e-04 -6.41287118e-02 8.42321873e-01 4.57770318e-01
-1.06218207e+00 4.78971839e-01 -1.20945275e+00 -2.91502357e-01
6.09017909e-01 2.57938683e-01 6.85375631e-02 9.51360632e-03
-1.08301079e+00 7.44872451e-01 5.75038970e-01 7.21859336e-02
-1.07266188e+00 -1.21018732e+00 -7.96132147e-01 1.02285914e-01
5.38786232e-01 -1.13155532e+00 1.40692973e+00 -9.66182649e-01
-2.07365155e+00 8.02787006e-01 4.46231455e-01 -4.35353249e-01
1.16870177e+00 3.94818783e-02 -1.35564238e-01 3.06029171e-01
-2.02053979e-01 7.57800102e-01 6.02320850e-01 -1.22411263e+00
-5.84943771e-01 -1.40892923e-01 -1.15385540e-02 3.98074180e-01
-5.17205372e-02 -2.87580729e-01 -8.37394953e-01 -7.38736212e-01
5.21267116e-01 -1.08971143e+00 -5.84115922e-01 2.71174222e-01
-3.69709790e-01 -1.61888033e-01 4.52188432e-01 -7.27272391e-01
9.95256126e-01 -1.98766410e+00 5.17318249e-01 3.10086936e-01
4.42076057e-01 4.13558744e-02 -7.66623244e-02 -4.29149210e-01
-5.37766889e-02 -3.01790744e-01 -3.70682567e-01 3.44548047e-01
-1.18127160e-01 3.86319727e-01 -3.74556668e-02 2.56603539e-01
-1.64154053e-01 9.98374164e-01 -1.10189950e+00 -7.56872952e-01
-2.32257135e-02 2.77264833e-01 -4.70833957e-01 1.72956631e-01
2.68267300e-02 7.59739995e-01 -5.72924733e-01 9.69007492e-01
6.71830475e-01 -7.05800653e-01 3.51889253e-01 -2.73838103e-01
-2.75402088e-02 -1.77664474e-01 -1.02662706e+00 1.94122410e+00
1.28131837e-01 6.44995049e-02 3.40864152e-01 -1.01027524e+00
4.70491081e-01 3.02679271e-01 5.52201331e-01 -6.23630285e-01
1.26776397e-01 1.39580548e-01 1.68317080e-01 -5.23618579e-01
-3.91533710e-02 -1.06882244e-01 1.22491479e-01 -1.93246529e-01
2.08737209e-01 9.88016352e-02 2.18820930e-01 1.13079898e-01
8.72685552e-01 3.10411304e-01 1.06881455e-01 -6.96838677e-01
5.61556578e-01 3.02673280e-01 1.05324244e+00 6.30661786e-01
-8.72341096e-01 3.87979001e-01 6.14046097e-01 -5.66366851e-01
-5.47314405e-01 -1.59561348e+00 -2.71298826e-01 8.69499505e-01
7.07192957e-01 1.98750257e-01 -8.61449718e-01 -1.20475364e+00
2.13234946e-01 1.93789899e-01 -6.63335502e-01 -2.44072869e-01
-9.67624485e-01 -1.10223567e+00 2.98240632e-01 7.63552248e-01
7.14619875e-01 -1.01993799e+00 -5.54578304e-01 3.15821826e-01
-1.17044717e-01 -9.00536418e-01 -6.70545697e-01 3.38372409e-01
-1.38278627e+00 -1.35232592e+00 -1.24868643e+00 -1.36052871e+00
9.93832052e-01 3.62630398e-03 1.15399468e+00 -1.22602150e-01
-1.55785516e-01 4.67190623e-01 -1.25960827e-01 -7.75470957e-02
-2.66031474e-01 3.52563709e-01 -3.14847052e-01 -5.02084732e-01
2.81110518e-02 -5.51156700e-01 -9.89733815e-01 2.34952450e-01
-7.65802741e-01 1.32424027e-01 8.07085454e-01 8.05648685e-01
8.33537281e-01 -3.64466459e-01 5.38731158e-01 -6.91297591e-01
5.83965778e-01 -5.29738486e-01 -7.02813208e-01 4.75414217e-01
-7.39831626e-01 -3.46666053e-02 6.70589209e-01 -8.71632695e-01
-9.14421916e-01 1.63678914e-01 6.36451244e-02 -5.21446884e-01
-9.30342674e-02 5.39488137e-01 2.83956885e-01 -4.76402521e-01
1.78186104e-01 6.24888182e-01 1.76044051e-02 -4.78107631e-01
1.92230985e-01 -8.84473622e-02 7.80299127e-01 -6.75533354e-01
5.61485171e-01 3.28508228e-01 -1.49625808e-01 -2.71670729e-01
-6.94032669e-01 -2.47451946e-01 -5.49914181e-01 -3.07270765e-01
8.02274227e-01 -9.40677464e-01 -7.48419225e-01 8.26492012e-01
-5.72097242e-01 -6.30347967e-01 -3.79545152e-01 7.77729750e-01
-3.07094902e-01 6.35719538e-01 -1.14794147e+00 9.62981954e-02
-6.36137068e-01 -1.34991336e+00 2.52956212e-01 6.58706427e-01
2.41358116e-01 -9.83777523e-01 1.13974497e-01 1.98148891e-01
5.41461468e-01 4.45928305e-01 9.36919928e-01 -3.80717725e-01
-8.02657187e-01 2.25936696e-01 -3.95430773e-01 1.83264166e-01
-4.76636663e-02 -1.83198527e-01 -2.10808679e-01 -5.89460492e-01
-5.72191365e-02 -2.31056243e-01 1.06398785e+00 6.86796606e-01
1.25330329e+00 -5.86276781e-03 -2.69462913e-01 1.16573071e+00
1.36696875e+00 3.36324334e-01 2.99217224e-01 5.32836437e-01
6.89951599e-01 2.30389331e-02 1.71383724e-01 3.39615755e-02
7.13004708e-01 2.45681956e-01 2.49167934e-01 -4.12526995e-01
-2.48633146e-01 -8.05743039e-02 2.16383010e-01 1.45017862e+00
6.51447624e-02 3.21502805e-01 -9.01975811e-01 5.18880010e-01
-1.86719096e+00 -4.30521280e-01 8.08037594e-02 1.80233264e+00
1.29020619e+00 1.78744316e-01 1.56817004e-01 -5.92140913e-01
6.13016486e-01 -1.27465382e-01 -9.67667282e-01 7.51309916e-02
1.45592257e-01 -2.57255375e-01 6.64089382e-01 2.84086674e-01
-1.24078095e+00 8.29818964e-01 6.98111916e+00 4.02478337e-01
-1.29155433e+00 7.17517510e-02 6.31027460e-01 6.92121983e-02
-8.49135369e-02 -2.41449326e-01 -6.26836777e-01 8.17658484e-01
3.26508075e-01 -2.29686931e-01 4.26034212e-01 7.87165642e-01
-3.16236138e-01 7.64053315e-02 -1.03563130e+00 7.91486502e-01
-3.70901078e-02 -9.92490053e-01 1.13858864e-01 -4.34342921e-01
8.27481508e-01 3.19432467e-01 7.97276013e-03 7.01648176e-01
7.46177495e-01 -6.40118480e-01 5.49795747e-01 5.83059371e-01
4.54273731e-01 -4.37883228e-01 6.39506757e-01 1.30380824e-01
-1.23263311e+00 3.89398355e-03 7.62506723e-02 4.15746480e-01
1.50981173e-01 3.80640209e-01 -5.61570466e-01 3.70844215e-01
7.77443528e-01 5.44230103e-01 -3.86516750e-01 1.57116318e+00
-3.29317778e-01 6.65240109e-01 -4.58819568e-01 1.44409999e-01
3.72657776e-01 -5.57794094e-01 6.89915657e-01 1.24094343e+00
3.66420969e-02 8.10690671e-02 7.36687720e-01 8.55386913e-01
-2.52314150e-01 1.46284103e-02 1.43097937e-01 4.78488475e-01
1.71927541e-01 1.18530619e+00 -1.02119017e+00 -4.94576871e-01
-3.91875803e-01 9.65365648e-01 3.27202767e-01 5.65025449e-01
-1.01941788e+00 -2.32628748e-01 3.22031885e-01 -4.26874191e-01
5.30093014e-01 9.80613288e-03 -1.89589366e-01 -1.45680785e+00
1.48971707e-01 -6.72714233e-01 8.02899063e-01 -2.09891215e-01
-1.42410123e+00 4.07396466e-01 -1.07562482e-01 -9.97067332e-01
4.31020528e-01 -3.92229438e-01 -6.81256413e-01 3.75746936e-01
-2.25156641e+00 -1.03292143e+00 -8.00921977e-01 5.97914100e-01
4.48988736e-01 1.80274248e-01 4.50860679e-01 4.29488957e-01
-9.29792881e-01 8.86294067e-01 1.77941576e-01 4.25058693e-01
6.79134548e-01 -1.53349876e+00 -1.56774260e-02 5.95075369e-01
-5.86713672e-01 5.08921802e-01 2.42144853e-01 -5.68799973e-01
-1.34228921e+00 -1.11037290e+00 2.74592042e-01 9.28457081e-03
6.20290756e-01 1.42351896e-01 -9.69545960e-01 7.71405995e-01
1.71244517e-01 1.82424039e-01 5.87647140e-01 -1.00911044e-01
-1.66832536e-01 -6.89390898e-02 -1.35458410e+00 7.27014303e-01
9.85093057e-01 1.42984703e-01 -8.83235991e-01 1.53111100e-01
9.54477429e-01 -1.03713381e+00 -1.26777911e+00 7.06864476e-01
5.70376277e-01 -3.95478368e-01 1.03245234e+00 -6.62404656e-01
2.27945507e-01 -1.87045261e-01 3.11555445e-01 -1.29332483e+00
-6.55436277e-01 -5.46466768e-01 -5.28538644e-01 6.84695423e-01
3.42567503e-01 -6.56064630e-01 1.03041565e+00 4.51693922e-01
-2.87920833e-01 -1.15786052e+00 -7.88278639e-01 -6.94964409e-01
4.37676728e-01 1.94614306e-01 4.19240743e-01 1.60667384e+00
5.70496693e-02 -3.21652666e-02 1.71337634e-01 2.85785973e-01
9.92262423e-01 4.03673679e-01 2.88731992e-01 -1.22956228e+00
-1.95688888e-01 -8.73410106e-01 -2.05902219e-01 -1.36467183e+00
-2.65486926e-01 -1.18681788e+00 -9.36163887e-02 -1.43903708e+00
6.55989826e-01 -4.32900280e-01 -9.65410352e-01 4.91871506e-01
-5.50292969e-01 -1.90826789e-01 2.57746220e-01 2.56263286e-01
-1.03396201e+00 5.34235597e-01 1.79183960e+00 -1.90383837e-01
-4.60017264e-01 -4.16826494e-02 -7.14940727e-01 8.24062407e-01
7.73286402e-01 -2.40305170e-01 -2.49008551e-01 -4.04986978e-01
-1.66788667e-01 2.04112008e-01 3.21031958e-01 -9.18956399e-01
4.40452367e-01 -9.89228040e-02 5.69607437e-01 -5.03198802e-01
-4.51083779e-01 -8.54800344e-01 -8.83311033e-02 1.01767337e+00
-3.79833221e-01 -3.55839655e-02 2.74600059e-01 5.51362991e-01
-2.59265043e-02 -2.04449669e-02 1.04577684e+00 -2.85605460e-01
-8.71131599e-01 5.15881956e-01 -8.96629915e-02 5.12084007e-01
1.15013146e+00 -3.01012516e-01 -4.37832661e-02 2.31498569e-01
-1.04003656e+00 9.53223526e-01 2.94013619e-01 3.78400147e-01
3.94233584e-01 -1.25083053e+00 -6.40463769e-01 1.23364121e-01
-2.16760263e-01 4.02879447e-01 2.99213588e-01 1.29604876e+00
-7.87626028e-01 2.54311562e-01 -2.32805088e-01 -9.14064109e-01
-1.01064634e+00 4.24404860e-01 5.56640863e-01 -5.44985473e-01
-1.24321818e+00 9.16190147e-01 1.98374927e-01 -8.10253441e-01
9.75773811e-01 -8.93253326e-01 -2.04982013e-01 -1.12324119e-01
1.81274787e-01 3.10532808e-01 -3.16006064e-01 2.58465707e-02
-2.77202368e-01 5.68251848e-01 -4.74986732e-01 6.30230606e-01
1.20677447e+00 5.22696748e-02 1.61918938e-01 -2.96439826e-02
9.15130615e-01 -6.46596789e-01 -1.68895233e+00 -3.70177269e-01
7.52941985e-03 1.33475689e-02 2.55408108e-01 -1.33067751e+00
-1.53255868e+00 2.21916437e-01 8.93195510e-01 -2.47568592e-01
1.19283402e+00 -7.85647407e-02 1.23091996e+00 3.71170163e-01
2.76944816e-01 -1.06690586e+00 -5.81585765e-02 6.18392289e-01
4.83032227e-01 -1.23631668e+00 -2.62705702e-02 -4.95177478e-01
-6.60880923e-01 1.01726091e+00 8.31010520e-01 -2.36856773e-01
7.45849252e-01 3.57859612e-01 4.10414875e-01 -1.90396965e-01
-4.13467497e-01 1.99245542e-01 1.05840921e-01 3.53300989e-01
1.03277326e-01 1.10618770e-01 -4.24470395e-01 8.17562521e-01
1.16306089e-01 3.31827343e-01 2.40117654e-01 9.16518271e-01
-3.53614002e-01 -7.23432064e-01 -9.12037417e-02 5.08161545e-01
-3.50282460e-01 7.04660267e-02 -6.48472682e-02 9.87650990e-01
-1.23599393e-03 3.06830555e-02 -2.15744451e-01 7.59902373e-02
5.57376504e-01 8.95108506e-02 6.96101308e-01 -2.20485497e-02
-8.83396387e-01 6.08851947e-02 -4.50167626e-01 -6.68506622e-01
-3.16053450e-01 -7.12089062e-01 -1.56003428e+00 2.61367764e-02
-2.71809638e-01 8.97512659e-02 2.73845077e-01 6.23248339e-01
1.86569646e-01 1.00245285e+00 4.38898206e-01 -4.46754396e-01
-1.01881528e+00 -6.23574257e-01 -2.94810444e-01 3.81549716e-01
3.56148392e-01 -5.64523280e-01 -2.28271008e-01 6.99075498e-03] | [14.495694160461426, -2.6002581119537354] |
75faa026-df6a-4509-a614-98dde42811f1 | deep-hashing-learning-for-visual-and-semantic | 1909.04614 | null | https://arxiv.org/abs/1909.04614v1 | https://arxiv.org/pdf/1909.04614v1.pdf | Deep Hashing Learning for Visual and Semantic Retrieval of Remote Sensing Images | Driven by the urgent demand for managing remote sensing big data, large-scale remote sensing image retrieval (RSIR) attracts increasing attention in the remote sensing field. In general, existing retrieval methods can be regarded as visual-based retrieval approaches which search and return a set of similar images from a database to a given query image. Although retrieval methods have achieved great success, there is still a question that needs to be responded to: Can we obtain the accurate semantic labels of the returned similar images to further help analyzing and processing imagery? Inspired by the above question, in this paper, we redefine the image retrieval problem as visual and semantic retrieval of images. Specifically, we propose a novel deep hashing convolutional neural network (DHCNN) to simultaneously retrieve the similar images and classify their semantic labels in a unified framework. In more detail, a convolutional neural network (CNN) is used to extract high-dimensional deep features. Then, a hash layer is perfectly inserted into the network to transfer the deep features into compact hash codes. In addition, a fully connected layer with a softmax function is performed on hash layer to generate class distribution. Finally, a loss function is elaborately designed to simultaneously consider the label loss of each image and similarity loss of pairs of images. Experimental results on two remote sensing datasets demonstrate that the proposed method achieves the state-of-art retrieval and classification performance. | ['Jon Atli Benediktsson', 'Shutao Li', 'Weiwei Song'] | 2019-09-10 | null | null | null | null | ['semantic-retrieval'] | ['natural-language-processing'] | [ 1.14950538e-01 -5.58505774e-01 -1.61919191e-01 -6.39830112e-01
-1.09969878e+00 -2.72418380e-01 4.80847985e-01 1.85318723e-01
-4.65889841e-01 3.75723839e-01 -1.59166470e-01 -6.07986934e-02
-3.21812332e-01 -1.11202240e+00 -4.94545460e-01 -9.18688416e-01
-6.06291220e-02 8.99249241e-02 -4.45044748e-02 4.22207545e-03
1.48436308e-01 5.01039565e-01 -1.63723540e+00 -5.22903837e-02
6.82780325e-01 1.44107795e+00 4.12080407e-01 1.81272745e-01
1.29579604e-01 6.25276327e-01 -3.97044122e-01 -1.15272917e-01
4.17775214e-01 -2.32542172e-01 -8.35440516e-01 -5.21332063e-02
3.00011069e-01 -7.32885659e-01 -5.57809234e-01 1.31091881e+00
8.53632927e-01 2.91692346e-01 5.50452530e-01 -1.10889351e+00
-1.15761411e+00 3.44489127e-01 -5.49966156e-01 2.84968108e-01
-8.34744647e-02 -2.96826847e-02 1.15597785e+00 -9.60918546e-01
1.71569750e-01 1.23313260e+00 2.74472475e-01 1.09705091e-01
-7.72803843e-01 -8.13836038e-01 -5.34835346e-02 4.35936898e-01
-2.16628528e+00 1.90871861e-02 5.73823690e-01 -1.49750307e-01
5.12906075e-01 2.18968272e-01 5.08148193e-01 2.62052536e-01
-1.32073209e-01 8.41054380e-01 6.45578980e-01 -4.48376089e-02
1.56514302e-01 1.09495893e-02 2.67342515e-02 4.40540373e-01
6.55137300e-02 -7.12198541e-02 6.63017482e-02 -4.47716005e-02
5.28856933e-01 8.38579595e-01 -3.65959257e-01 -1.57695059e-02
-1.03899276e+00 1.18328941e+00 1.27976906e+00 2.89844841e-01
-5.36841810e-01 3.29808921e-01 4.18022484e-01 2.64392644e-01
5.19886971e-01 4.70171906e-02 -1.02614291e-01 6.33371532e-01
-9.48903739e-01 2.58578628e-01 3.21912199e-01 7.41640866e-01
1.27861869e+00 -3.32720459e-01 -2.96479374e-01 1.06923854e+00
5.29382408e-01 8.78781915e-01 5.54875314e-01 -4.96501148e-01
2.79898852e-01 6.43963158e-01 -1.34345457e-01 -1.47182322e+00
-1.56140059e-01 -2.40788430e-01 -1.13572884e+00 -1.61528811e-01
-3.73446345e-01 3.13294858e-01 -8.46540511e-01 1.27058303e+00
2.52942622e-01 1.42237991e-01 5.91307096e-02 1.54156542e+00
1.02590013e+00 1.28296614e+00 4.14254554e-02 2.01595366e-01
1.30362833e+00 -8.60397577e-01 -3.87084484e-01 -5.64170144e-02
7.39400461e-02 -6.19241476e-01 9.06073451e-01 -1.48330301e-01
-7.40113974e-01 -7.11538672e-01 -1.07517517e+00 -2.58799016e-01
-6.13121986e-01 1.58926889e-01 5.43890178e-01 1.40551329e-01
-9.77616906e-01 3.39773834e-01 -5.86786330e-01 -2.26283833e-01
4.30427492e-01 3.27171266e-01 -2.88964957e-01 -3.84963989e-01
-1.51961362e+00 4.63351399e-01 6.37678862e-01 7.11831868e-01
-9.94476736e-01 -3.23374808e-01 -9.86792207e-01 4.52683955e-01
1.20099336e-02 -2.79847980e-01 1.01553607e+00 -7.54366636e-01
-9.85320270e-01 1.02134144e+00 2.68280149e-01 -1.52417839e-01
1.17269717e-01 -7.34364381e-03 -4.49465573e-01 4.45470095e-01
1.99850440e-01 8.89343500e-01 8.19712281e-01 -1.13147914e+00
-7.16214836e-01 -4.56655413e-01 3.30621167e-03 1.77745834e-01
-5.54352820e-01 8.28083158e-02 -5.30816674e-01 -6.47988796e-01
3.85579526e-01 -8.33462715e-01 -1.50131196e-01 3.24497372e-01
-2.32413858e-01 -3.22393745e-01 8.21859419e-01 -4.45313543e-01
1.13857591e+00 -2.30901384e+00 2.62734550e-03 4.35554981e-01
2.34021738e-01 4.41439539e-01 -4.42985922e-01 3.89212817e-01
1.05421364e-01 1.98538959e-01 -4.04479831e-01 -6.86574280e-02
1.18010163e-01 2.44752586e-01 -5.71026921e-01 7.07670450e-01
2.69456267e-01 1.08603883e+00 -8.44661474e-01 -6.04964912e-01
2.27392435e-01 7.81151652e-01 -1.99537039e-01 4.33155000e-01
5.26592210e-02 7.48882592e-02 -8.49698067e-01 8.91041517e-01
1.05107534e+00 -5.91585159e-01 -3.12396556e-01 -1.74389154e-01
-3.74945886e-02 -2.05873355e-01 -1.05870211e+00 1.44580221e+00
-2.19852850e-01 4.10511225e-01 -5.92215061e-02 -1.26330626e+00
1.32068145e+00 -1.02275759e-02 4.09099579e-01 -9.45813835e-01
6.41599670e-02 5.04502118e-01 -5.96849263e-01 -5.58546007e-01
6.28441453e-01 -3.41374427e-01 -1.95820913e-01 4.05225903e-01
-3.41293812e-01 -1.40325114e-01 -3.17826748e-01 -1.76507488e-01
5.06133318e-01 -4.57239479e-01 4.03123908e-02 1.02693655e-01
7.59521484e-01 -6.97547123e-02 4.30721521e-01 6.60506368e-01
-1.86529562e-01 8.05823326e-01 -3.53392124e-01 -8.81855905e-01
-9.61994171e-01 -7.86435068e-01 -4.27069962e-01 9.99572337e-01
7.20610797e-01 7.42297918e-02 -2.71504045e-01 -2.27158383e-01
2.18834996e-01 -2.60683328e-01 -4.87131506e-01 -3.95727843e-01
-2.84396350e-01 -7.32589841e-01 5.49999177e-01 2.62667745e-01
1.07822943e+00 -1.37794185e+00 -6.98341906e-01 1.28810659e-01
-3.07936519e-01 -7.51621068e-01 -3.68257046e-01 -8.32570642e-02
-5.27493536e-01 -8.49988103e-01 -1.18521059e+00 -1.12934649e+00
4.41845268e-01 9.38926280e-01 7.11277962e-01 4.81201231e-01
-4.83416647e-01 -1.85282305e-02 -7.19573677e-01 -5.64067289e-02
1.36506870e-01 3.01379293e-01 -4.25412148e-01 1.63204685e-01
6.69023812e-01 -3.16319823e-01 -1.15097094e+00 3.03770423e-01
-1.62324739e+00 -3.78418505e-01 8.39391172e-01 9.21823144e-01
7.72760808e-01 2.63187140e-01 4.68085140e-01 -1.29419759e-01
4.12782848e-01 -7.09362090e-01 -7.23675668e-01 5.52064180e-01
-4.72598433e-01 3.41974385e-02 5.29549479e-01 -1.54372171e-01
-4.49147582e-01 -1.46243602e-01 -4.80862893e-02 -5.87402642e-01
3.85131985e-02 9.98690248e-01 -3.35010774e-02 -1.63459331e-01
1.94323257e-01 5.87091386e-01 1.55678783e-02 -4.41417247e-01
3.56876343e-01 1.27743006e+00 3.47905874e-01 -1.68395340e-01
8.60150695e-01 5.23974478e-01 -2.39910126e-01 -6.20560110e-01
-9.04233098e-01 -7.99403846e-01 -2.66791314e-01 5.22558019e-02
8.77207041e-01 -1.26403582e+00 -8.66560400e-01 5.98116517e-01
-9.51391160e-01 2.69822474e-03 9.85341445e-02 5.32252967e-01
-2.35968128e-01 3.93353939e-01 -5.12051642e-01 -6.43953860e-01
-8.45810473e-01 -1.16601634e+00 1.61423361e+00 5.45152128e-01
6.35344565e-01 -6.09080315e-01 6.46681413e-02 2.33885333e-01
7.66610801e-01 4.66140881e-02 7.93923616e-01 -4.65635359e-01
-8.61653268e-01 -3.98560047e-01 -9.97422338e-01 4.63034809e-01
1.13034904e-01 -2.56545633e-01 -9.48876381e-01 -5.11879742e-01
-2.85446942e-01 -7.35814095e-01 1.20116079e+00 2.92465240e-01
1.39434397e+00 -4.52207953e-01 -2.77521521e-01 8.35378706e-01
1.78479183e+00 8.37791618e-03 7.60967851e-01 5.41624784e-01
7.13187993e-01 4.53350335e-01 5.80571830e-01 6.65911555e-01
5.37867546e-01 5.07616580e-01 6.02998853e-01 -2.54896015e-01
3.49573284e-01 -2.36490518e-01 -1.18796788e-01 4.70054090e-01
2.75891006e-01 -2.13729173e-01 -7.81717896e-01 4.45264876e-01
-1.78635263e+00 -1.16325712e+00 3.92565519e-01 2.07865381e+00
7.68423080e-01 -5.30132830e-01 -3.52434307e-01 1.32175311e-01
9.93600905e-01 6.28010154e-01 -6.67960346e-01 2.59189010e-01
-2.10017681e-01 -7.30020693e-05 4.16490465e-01 2.78715283e-01
-1.41084278e+00 9.73558486e-01 5.21357059e+00 1.00474811e+00
-1.72482479e+00 -8.21117982e-02 6.33775234e-01 1.66887671e-01
-3.08246821e-01 -1.01974621e-01 -7.01653063e-01 4.50428098e-01
4.41554964e-01 -1.75801501e-01 4.47292924e-01 7.81162262e-01
8.88965353e-02 2.30252340e-01 -5.26878059e-01 1.17469251e+00
-6.78216387e-03 -1.12682080e+00 5.47218204e-01 1.97364353e-02
6.09143913e-01 3.16454202e-01 3.11901361e-01 3.58276963e-01
-5.66468425e-02 -1.11649346e+00 6.30991042e-01 5.94791114e-01
8.77452493e-01 -7.90931940e-01 8.87309372e-01 2.24934638e-01
-1.56850839e+00 -3.31062496e-01 -1.05668223e+00 2.30476707e-01
-1.21282279e-01 4.71248835e-01 -5.02392113e-01 5.64357638e-01
8.91246676e-01 1.06220341e+00 -5.86417913e-01 1.36644363e+00
1.09850932e-02 2.30549887e-01 -2.48224229e-01 -6.89541101e-02
6.42468631e-01 -2.70756721e-01 1.45945307e-02 9.66616809e-01
4.40918297e-01 3.68539691e-01 4.72567081e-01 8.44667971e-01
-2.05006450e-01 1.52754590e-01 -6.33176804e-01 -5.16946316e-02
7.22081423e-01 1.54960454e+00 -4.21122581e-01 -2.43700951e-01
-1.37425661e-01 9.44720030e-01 2.09413201e-01 3.97772640e-01
-5.96791506e-01 -7.20272064e-01 4.91130829e-01 -1.88952580e-01
4.27523464e-01 1.39156967e-01 3.87048781e-01 -1.18334568e+00
1.24784388e-01 -6.45684600e-01 5.99266708e-01 -1.06651199e+00
-1.39795244e+00 6.09795809e-01 -2.59267449e-01 -1.59254742e+00
1.03806011e-01 -2.13328466e-01 -4.76921558e-01 9.85516667e-01
-2.37304020e+00 -1.35298324e+00 -9.01295125e-01 7.42825150e-01
2.27938890e-01 7.31179193e-02 7.58910477e-01 6.35443985e-01
-3.87619615e-01 5.61383367e-01 4.07883614e-01 4.72470313e-01
5.07544219e-01 -7.66883850e-01 -1.84779614e-02 5.17023683e-01
-1.58695504e-01 5.06779671e-01 1.45919070e-01 -2.49588937e-01
-1.53194773e+00 -1.54671228e+00 7.98250675e-01 4.47664261e-01
5.04118323e-01 4.54530679e-02 -1.17526197e+00 3.19826603e-01
-1.44450575e-01 5.25006890e-01 5.91614068e-01 -6.21351004e-01
-5.23686409e-01 -5.11753380e-01 -1.03158152e+00 1.16647832e-01
5.18546164e-01 -8.23611140e-01 -4.42389816e-01 5.27123690e-01
6.71445131e-01 -2.79145818e-02 -8.48838687e-01 4.09681439e-01
4.89203870e-01 -6.44321144e-01 1.22654366e+00 -3.77356410e-01
5.34973025e-01 -5.13412058e-01 -5.93661964e-01 -9.19645667e-01
-3.24910849e-01 -8.35738182e-02 6.37908280e-01 9.21069026e-01
-2.20033675e-02 -4.99165297e-01 4.11316335e-01 3.28079283e-01
1.25392035e-01 -6.11223578e-01 -7.36035287e-01 -7.71338224e-01
7.79009461e-02 2.29061782e-01 8.51794302e-01 9.86989021e-01
-2.54094541e-01 2.08420962e-01 -4.71971244e-01 4.55814511e-01
6.55184805e-01 7.02849805e-01 4.57781762e-01 -1.38083315e+00
1.73647091e-01 -4.49621558e-01 -5.84145486e-01 -1.16667855e+00
4.19720858e-01 -1.18447459e+00 3.43926728e-01 -1.61551154e+00
5.90248585e-01 -8.52215409e-01 -5.57333171e-01 7.38350987e-01
-3.85966182e-01 5.61745882e-01 2.04019621e-01 7.75843084e-01
-8.47136676e-01 1.07416475e+00 1.06263685e+00 -6.73419654e-01
6.33539036e-02 -1.54329538e-01 -6.13248706e-01 4.52943072e-02
6.25026584e-01 -5.36000848e-01 -1.10532425e-01 -7.52062321e-01
3.99377733e-01 1.30541265e-01 7.13006675e-01 -9.20893192e-01
4.55378711e-01 3.34581640e-03 2.99709469e-01 -6.95480645e-01
1.56298995e-01 -9.27045882e-01 5.00599630e-02 5.23706377e-01
-5.19182563e-01 -6.43153535e-03 -1.46694779e-01 5.07884681e-01
-8.58549833e-01 -2.06126809e-01 7.19805837e-01 -7.12104142e-02
-9.02559996e-01 8.83770347e-01 3.43602337e-02 -1.93773076e-01
8.89341354e-01 -9.56899393e-03 -3.80030662e-01 -3.59963238e-01
-1.96699858e-01 5.85595071e-01 1.94377542e-01 4.72950757e-01
1.03235924e+00 -1.58419061e+00 -9.74946141e-01 2.44312227e-01
6.20994270e-01 2.00846016e-01 3.44169945e-01 3.22157115e-01
-7.33746469e-01 4.79587495e-01 -1.55125856e-01 -5.83991766e-01
-8.76810670e-01 5.56089103e-01 4.15018082e-01 6.09161258e-02
-5.34114063e-01 7.07011580e-01 9.75525007e-03 -5.18715382e-01
1.64887652e-01 -4.37560417e-02 -2.22296432e-01 2.07464695e-01
9.77335691e-01 6.87303543e-02 -1.12284184e-01 -7.37831652e-01
-4.68876004e-01 8.39401782e-01 -1.75273102e-02 2.40327284e-01
1.56177080e+00 -2.85913736e-01 -3.78055573e-01 4.19424959e-02
1.88013148e+00 -6.66676700e-01 -1.04069459e+00 -6.68494523e-01
-2.56819248e-01 -4.73383486e-01 3.82760555e-01 -3.94276410e-01
-1.49839783e+00 1.22101462e+00 9.17800665e-01 3.18671912e-01
1.20197368e+00 1.61484852e-01 1.05470538e+00 7.14945555e-01
2.40572035e-01 -8.15484166e-01 -4.17766757e-02 5.16057491e-01
7.70535946e-01 -1.65884900e+00 -9.00290161e-02 1.37972161e-01
-3.27178240e-01 1.00182188e+00 1.75720796e-01 -1.53100416e-01
8.70284736e-01 -4.56207931e-01 2.66462773e-01 -4.01308864e-01
-1.41384557e-01 -3.92301440e-01 2.87796825e-01 1.99200034e-01
2.87872106e-02 -8.60389322e-03 -2.08642617e-01 2.95199335e-01
2.22295910e-01 7.41780251e-02 5.25868358e-03 8.73624504e-01
-9.53140676e-01 -6.24353826e-01 -3.21663380e-01 2.85885960e-01
-3.60827357e-01 -3.91027659e-01 -7.00745312e-03 2.74326861e-01
-1.29634187e-01 8.36081386e-01 1.99941605e-01 -4.25257325e-01
1.37320191e-01 -4.27791059e-01 -2.35774726e-01 -4.98848736e-01
-5.12643933e-01 -2.24571805e-02 -8.72686565e-01 -2.63212353e-01
-5.58418870e-01 -6.32606298e-02 -1.30768907e+00 -2.72511750e-01
-4.10482526e-01 4.18125957e-01 7.09411442e-01 8.00424516e-01
3.89967978e-01 -8.42899829e-02 1.19545472e+00 -9.46999252e-01
-8.81229997e-01 -7.89592981e-01 -9.12945509e-01 4.56131250e-01
4.91404563e-01 -3.58785748e-01 -4.62378144e-01 -1.21953599e-01] | [11.247323036193848, 0.8570356965065002] |
2a119fad-80cd-49f7-9ebc-95fe7c6ee85b | subtitles-to-segmentation-improving-low | null | null | https://aclanthology.org/2020.clssts-1.11 | https://aclanthology.org/2020.clssts-1.11.pdf | Subtitles to Segmentation: Improving Low-Resource Speech-to-TextTranslation Pipelines | In this work, we focus on improving ASR output segmentation in the context of low-resource language speech-to-text translation. ASR output segmentation is crucial, as ASR systems segment the input audio using purely acoustic information and are not guaranteed to output sentence-like segments. Since most MT systems expect sentences as input, feeding in longer unsegmented passages can lead to sub-optimal performance. We explore the feasibility of using datasets of subtitles from TV shows and movies to train better ASR segmentation models. We further incorporate part-of-speech (POS) tag and dependency label information (derived from the unsegmented ASR outputs) into our segmentation model. We show that this noisy syntactic information can improve model accuracy. We evaluate our models intrinsically on segmentation quality and extrinsically on downstream MT performance, as well as downstream tasks including cross-lingual information retrieval (CLIR) tasks and human relevance assessments. Our model shows improved performance on downstream tasks for Lithuanian and Bulgarian. | ['Kathy Mckeown', 'Chris Kedzie', 'Zhengping Jiang', 'David Wan', 'Peter Bell', 'Elsbeth Turcan'] | 2020-05-01 | null | null | null | lrec-2020-5 | ['speech-to-text-translation', 'cross-lingual-information-retrieval'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.96717620e-01 4.32125717e-01 -1.92097351e-01 -4.27411407e-01
-1.90556800e+00 -9.12516057e-01 4.38255519e-01 -8.28534439e-02
-6.52651310e-01 4.74465668e-01 6.70323074e-01 -7.86303222e-01
5.34571230e-01 -2.42542356e-01 -7.57473707e-01 -2.22291365e-01
5.22252679e-01 8.57519150e-01 3.54946256e-01 -3.94570649e-01
4.27691638e-02 -1.09978877e-02 -8.49684715e-01 6.95198298e-01
9.34828103e-01 4.77634132e-01 6.13456011e-01 1.06713641e+00
1.14613764e-01 5.37713766e-01 -8.41614187e-01 -5.23799777e-01
2.51880437e-01 -7.81184494e-01 -1.15274239e+00 -8.88927728e-02
2.81195849e-01 -5.77947758e-02 -2.29939178e-01 6.89742327e-01
6.95453525e-01 1.14008747e-01 7.38937497e-01 -5.25653780e-01
-7.18511581e-01 1.30472529e+00 -1.70257121e-01 4.01971012e-01
3.47639948e-01 2.26326674e-01 1.34816885e+00 -9.61249173e-01
7.32418776e-01 1.29836082e+00 2.30054528e-01 8.26852202e-01
-1.41017997e+00 -2.25937679e-01 1.47794470e-01 2.96553946e-04
-1.14824235e+00 -1.10447061e+00 3.46842319e-01 -1.41142204e-01
1.24467933e+00 4.75642830e-01 1.63881138e-01 1.15367317e+00
-1.70618683e-01 1.26068199e+00 8.80340576e-01 -7.35459268e-01
-1.11997478e-01 1.84954315e-01 -1.46064907e-01 8.01390707e-02
-5.82915306e-01 -3.80985402e-02 -6.59633875e-01 4.18750197e-01
5.60187578e-01 -7.92532921e-01 -3.51189584e-01 6.17122233e-01
-1.41357565e+00 6.45056188e-01 1.29013121e-01 4.49509263e-01
1.07417218e-02 1.91192329e-01 3.62705082e-01 6.35910332e-01
5.87448835e-01 7.12952077e-01 -7.16182888e-01 -4.33241993e-01
-1.26852965e+00 -1.76428005e-01 6.42000079e-01 1.18825102e+00
4.42148089e-01 1.48308039e-01 -6.68460906e-01 1.32274795e+00
3.09622347e-01 7.95329094e-01 6.11896574e-01 -1.21363902e+00
1.06134832e+00 -1.28795505e-01 -4.87993024e-02 -1.26188189e-01
-1.26761515e-02 -4.34606761e-01 -2.02041373e-01 -4.94421571e-01
5.79376101e-01 -3.97212178e-01 -1.32776415e+00 1.69667888e+00
-1.32465526e-01 -5.51468432e-01 4.30291384e-01 9.94315207e-01
7.53640652e-01 1.02681220e+00 4.73594405e-02 -3.66977036e-01
1.21454024e+00 -1.24318647e+00 -7.89112568e-01 -5.15823901e-01
1.06822932e+00 -1.31579292e+00 1.34475589e+00 2.68373609e-01
-1.48031056e+00 -5.66679955e-01 -6.33280933e-01 -3.71312737e-01
2.93417815e-02 4.31028694e-01 -7.70495236e-02 4.38451141e-01
-1.42729878e+00 4.42938566e-01 -9.89441633e-01 -3.23950619e-01
-7.38009587e-02 2.86679089e-01 -1.39193190e-02 -1.88622549e-02
-1.37262440e+00 9.63836074e-01 1.65594921e-01 3.74427661e-02
-7.41886973e-01 -3.42804134e-01 -9.87221122e-01 -1.75969712e-02
2.37761915e-01 -2.85348713e-01 1.97083116e+00 -1.12971389e+00
-1.80075574e+00 9.37009931e-01 -5.07983625e-01 -4.10948128e-01
4.64958787e-01 -2.16921434e-01 -2.12621629e-01 2.84469098e-01
7.03795254e-02 1.10977018e+00 5.75360715e-01 -9.29094970e-01
-7.35945702e-01 -2.22051561e-01 -3.51514280e-01 6.54020190e-01
-9.13102999e-02 4.32014674e-01 -6.12709761e-01 -8.01087439e-01
2.11149886e-01 -1.20562947e+00 -2.07091734e-01 -9.51207697e-01
-5.19518316e-01 -2.68346637e-01 3.92080665e-01 -1.22557569e+00
1.20255375e+00 -1.96362925e+00 2.24133506e-01 -2.57160552e-02
-5.42931318e-01 9.94786769e-02 -3.71689737e-01 3.39034438e-01
3.36387128e-01 4.13293302e-01 -8.20529982e-02 -4.19879138e-01
-1.39169423e-02 1.07867666e-01 -3.80300492e-01 -1.61454473e-02
4.72497523e-01 1.21641552e+00 -7.96181083e-01 -6.48840606e-01
7.32182786e-02 2.72723734e-01 -3.40030253e-01 1.81844637e-01
-2.80558884e-01 8.30068469e-01 -3.24702650e-01 4.68865335e-01
-1.56051412e-01 1.54868916e-01 -9.25202072e-02 2.36521170e-01
-2.95992941e-02 1.18977726e+00 -4.34864908e-01 1.78700483e+00
-8.03914845e-01 7.37353623e-01 5.18412963e-02 -8.41818929e-01
6.83992684e-01 8.04381788e-01 1.16339557e-01 -9.90430057e-01
8.89777243e-02 4.98704404e-01 3.21240366e-01 -2.38450721e-01
8.35945845e-01 -5.03079817e-02 -3.95645767e-01 4.44647938e-01
2.31598273e-01 -5.90152383e-01 1.59385845e-01 1.89044282e-01
9.90611851e-01 2.86751211e-01 -1.92626715e-01 -1.70341045e-01
2.85117418e-01 4.54736054e-01 3.23154926e-01 7.94541419e-01
-8.08191746e-02 1.27043080e+00 1.98093235e-01 3.85050982e-01
-1.07500625e+00 -9.99521792e-01 5.55789750e-03 1.50205922e+00
-1.81000069e-01 -2.73662388e-01 -1.13257134e+00 -6.54572129e-01
-6.75609112e-01 1.14293885e+00 4.60885428e-02 5.00727333e-02
-1.11059737e+00 -4.09141988e-01 8.31628740e-01 5.01470864e-01
-1.42401025e-01 -1.22154248e+00 -2.37583891e-02 5.63277185e-01
-6.57403231e-01 -1.42851460e+00 -8.47123384e-01 3.32461536e-01
-9.18520749e-01 -2.90930182e-01 -1.18552327e+00 -1.00169230e+00
2.59063661e-01 3.03782858e-02 1.25023234e+00 -1.60092637e-01
1.94390029e-01 2.37307236e-01 -6.33733928e-01 -2.07472920e-01
-1.15008068e+00 6.09647810e-01 -1.46870792e-01 -3.84975284e-01
4.01849806e-01 -1.65517971e-01 -2.60980636e-01 4.83665049e-01
-4.54811722e-01 1.90223470e-01 7.39109159e-01 6.50585830e-01
6.38848305e-01 -7.89941430e-01 7.25657463e-01 -9.08801377e-01
5.51755905e-01 -6.03578798e-02 -3.61025959e-01 3.17523628e-01
-4.21690106e-01 1.02053486e-01 5.96398294e-01 -3.61677259e-01
-1.17460430e+00 1.65801316e-01 -5.70544064e-01 -2.17473850e-01
-1.60356313e-01 4.49512303e-01 -2.59646922e-01 5.87083161e-01
7.66320705e-01 2.17783764e-01 -2.93080866e-01 -4.62474376e-01
5.65558791e-01 1.24527478e+00 6.29633725e-01 -6.91581190e-01
4.74542886e-01 -2.55618572e-01 -5.54968655e-01 -8.86169672e-01
-8.26682687e-01 -6.26902461e-01 -5.98271966e-01 -8.08808431e-02
9.76001501e-01 -1.08683264e+00 -2.98988018e-02 2.97812261e-02
-1.34967959e+00 -7.23480642e-01 -2.72587240e-01 6.07432306e-01
-8.71913552e-01 9.08889920e-02 -1.11506367e+00 -7.23836064e-01
-3.03315520e-01 -1.47504807e+00 1.41579235e+00 -6.61188737e-02
-5.52133381e-01 -7.48123407e-01 -4.90437523e-02 8.06371570e-01
3.00884306e-01 -6.32694900e-01 7.31877863e-01 -1.07010674e+00
-5.76454937e-01 1.21558905e-01 1.07333146e-01 4.47893709e-01
-1.05787851e-01 -1.52959108e-01 -1.12071478e+00 -5.19827046e-02
-1.51546165e-01 -3.94204736e-01 9.41835701e-01 5.33504844e-01
5.51379800e-01 -4.58515018e-01 -2.08384722e-01 1.19700611e-01
8.40994835e-01 1.43826604e-01 6.41421556e-01 1.16578573e-02
7.51156688e-01 1.02319276e+00 8.88973415e-01 -2.47060195e-01
3.03940147e-01 8.02717209e-01 -2.99652398e-01 -4.90928851e-02
-5.64945698e-01 -3.11773717e-01 8.96899641e-01 1.47914362e+00
1.75794780e-01 -6.99081004e-01 -1.04212081e+00 9.90879416e-01
-1.64139938e+00 -5.31828046e-01 -3.63176912e-01 2.12938571e+00
1.36385846e+00 1.79422438e-01 8.18671733e-02 -1.53093770e-01
9.85656857e-01 -2.01325387e-01 -1.34718463e-01 -7.41792977e-01
-1.62286252e-01 3.17447692e-01 5.91187060e-01 1.01521456e+00
-7.98401773e-01 1.62078679e+00 6.23167753e+00 1.17954421e+00
-9.55947936e-01 5.13748586e-01 9.77168083e-01 -1.25045866e-01
-5.31150401e-01 5.28070086e-04 -1.01582313e+00 3.01402986e-01
1.55648756e+00 8.17042068e-02 4.04518336e-01 3.98066849e-01
4.85360086e-01 3.93744820e-04 -1.46859860e+00 6.27573848e-01
-1.81155980e-01 -8.78911734e-01 -1.13483898e-01 -1.52750432e-01
7.76732266e-01 4.94877040e-01 1.43303409e-01 5.02424955e-01
4.16503340e-01 -1.10869634e+00 1.00436974e+00 -1.03119172e-01
9.06248093e-01 -6.08063996e-01 6.54309094e-01 3.88052702e-01
-7.99718261e-01 3.85301232e-01 -1.24679826e-01 1.69893548e-01
5.11547804e-01 2.15857565e-01 -1.23805165e+00 2.83808738e-01
2.98841894e-01 3.86300534e-01 -4.97975200e-01 6.59240425e-01
-5.98554909e-01 1.23041153e+00 -3.89327794e-01 -3.59362289e-02
3.58428806e-01 -4.41814363e-02 7.10295916e-01 1.74010253e+00
3.41642439e-01 4.38060835e-02 1.87646404e-01 6.37250304e-01
-1.39663637e-01 4.02313977e-01 -4.81378496e-01 -2.73927420e-01
5.31461477e-01 9.30776358e-01 -8.31259131e-01 -4.18866307e-01
-3.43395680e-01 1.21361041e+00 1.01077175e-02 6.16677046e-01
-3.91277313e-01 -3.21555793e-01 3.95839542e-01 4.79970537e-02
1.90268874e-01 -2.51467288e-01 -5.08445323e-01 -1.07372725e+00
-5.52506410e-02 -1.04829884e+00 2.28775457e-01 -6.49252713e-01
-1.05504680e+00 7.48227894e-01 -3.26338083e-01 -1.05698526e+00
-8.07232082e-01 -3.43883634e-01 -2.49587476e-01 1.04128897e+00
-1.45746267e+00 -1.10512948e+00 5.89645624e-01 1.31417707e-01
1.26582050e+00 2.86715990e-03 6.60978138e-01 2.71457255e-01
-1.49307922e-01 8.53733540e-01 4.07751510e-03 3.50649595e-01
9.08851564e-01 -1.30286205e+00 9.55719650e-01 1.05189824e+00
5.95686853e-01 2.76577413e-01 7.41826892e-01 -6.34691417e-01
-1.07788444e+00 -1.10102534e+00 1.29480720e+00 -8.71313453e-01
5.99223256e-01 -5.27923167e-01 -7.44841039e-01 7.92032123e-01
3.80891293e-01 -4.55179095e-01 4.96281892e-01 1.91994637e-01
1.23824589e-01 2.39375085e-01 -7.27964818e-01 8.24371099e-01
1.09049511e+00 -7.71708071e-01 -7.86089122e-01 3.40355247e-01
1.42249811e+00 -5.58924794e-01 -8.16520035e-01 3.18039268e-01
2.05181524e-01 -2.87598461e-01 6.56539798e-01 -3.93723875e-01
4.27225500e-01 -3.68538983e-02 -3.18895370e-01 -1.45722151e+00
2.64402735e-03 -1.05568528e+00 3.19034249e-01 1.45023060e+00
1.29242444e+00 -6.45834059e-02 5.24759114e-01 5.94835460e-01
-6.47159457e-01 -2.18600020e-01 -1.02507401e+00 -9.40794885e-01
2.24452123e-01 -6.56122446e-01 9.23281088e-02 5.41895151e-01
1.79779798e-01 9.43957269e-01 2.00217813e-02 6.70061931e-02
1.23095654e-01 -1.72912911e-01 3.50214899e-01 -7.66074002e-01
-4.57092285e-01 -4.67471808e-01 4.69742976e-02 -1.47799861e+00
1.96018219e-01 -1.05686486e+00 8.77862871e-01 -1.54126954e+00
-1.65099889e-01 -5.31658053e-01 -1.30806863e-01 4.76737976e-01
-1.74980834e-01 5.01396000e-01 3.82539958e-01 2.54833013e-01
-6.77097082e-01 3.77370417e-01 1.23333645e+00 -1.24484554e-01
-5.18448114e-01 2.46047422e-01 -5.54800153e-01 3.31675768e-01
7.67431617e-01 -7.24249661e-01 -2.86242634e-01 -1.02735639e+00
-1.25606850e-01 5.42723298e-01 -2.88803101e-01 -6.80794001e-01
4.86212522e-02 -1.13344714e-01 1.26581833e-01 -5.64359963e-01
4.14146423e-01 -3.79547983e-01 -3.06266844e-01 5.32512218e-02
-8.47927809e-01 -4.20953333e-02 -2.22629271e-02 9.80906785e-02
-3.33455175e-01 -5.46944380e-01 5.87375045e-01 -3.54410052e-01
-2.02095553e-01 -8.15126449e-02 -7.93031096e-01 5.12005031e-01
3.13541681e-01 -1.44619718e-01 -1.21347316e-01 -6.16629124e-01
-9.44787383e-01 2.59836853e-01 4.20329422e-01 5.78951538e-01
3.34222436e-01 -9.19314623e-01 -1.03772986e+00 -4.34723906e-02
8.68882313e-02 4.39410424e-03 -3.94089788e-01 8.50079596e-01
-2.44381547e-01 7.68671155e-01 4.15565789e-01 -8.16571414e-01
-1.20179200e+00 9.30983275e-02 -1.07512223e-02 -1.79005116e-01
-2.91248202e-01 1.04557431e+00 2.31324032e-01 -5.59241831e-01
3.46322149e-01 -4.09949362e-01 -2.03975979e-02 -2.17567831e-02
2.71997720e-01 1.13106027e-01 3.18432808e-01 -8.49431872e-01
-9.43430290e-02 3.52939636e-01 -2.44658604e-01 -1.17075861e+00
9.53996956e-01 -5.61163425e-01 2.69031405e-01 5.46981931e-01
1.12270260e+00 2.22170562e-01 -1.03339469e+00 -1.95558831e-01
4.31790918e-01 -9.31052789e-02 1.34000570e-01 -1.03004193e+00
-5.11928380e-01 1.20682216e+00 1.32122323e-01 2.36592256e-03
8.27571094e-01 3.39293003e-01 1.17934859e+00 5.75399280e-01
2.63094991e-01 -1.65476561e+00 -3.03177815e-02 9.61221993e-01
6.17747784e-01 -1.30180347e+00 -6.48859203e-01 -2.93837488e-01
-9.85431552e-01 8.81306648e-01 4.45958912e-01 2.66814888e-01
2.52000988e-02 1.16316020e-01 5.00817060e-01 3.04403365e-01
-7.52390683e-01 -5.32894552e-01 5.14102757e-01 5.65489352e-01
8.10908794e-01 1.59398049e-01 -3.33949387e-01 5.94235182e-01
-6.63085878e-01 -4.08582062e-01 6.10234559e-01 7.09946394e-01
-6.04677439e-01 -1.33141434e+00 -2.37694085e-01 2.74755865e-01
-1.08176136e+00 -5.63398302e-01 -5.35535455e-01 2.97867566e-01
-4.62184638e-01 1.35440052e+00 1.34855472e-02 -1.57324046e-01
1.53468713e-01 2.81510413e-01 4.11367089e-01 -1.25934029e+00
-6.51412547e-01 7.53107488e-01 5.48392296e-01 -2.57000536e-01
-5.63134030e-02 -6.78294122e-01 -1.39757872e+00 3.44750255e-01
-5.53685308e-01 3.63379478e-01 9.40069199e-01 1.00293565e+00
1.52129680e-01 4.38718855e-01 5.94363749e-01 -7.84373820e-01
-5.12746811e-01 -1.35149908e+00 -1.41555564e-02 1.75788864e-01
3.63299251e-01 3.12857211e-01 -2.91724563e-01 4.25159633e-01] | [14.486135482788086, 7.135370254516602] |
27e6e1d1-85fb-40dd-99b0-ee3aadee1135 | optimal-target-assignment-and-path-finding | 1612.05693 | null | http://arxiv.org/abs/1612.05693v1 | http://arxiv.org/pdf/1612.05693v1.pdf | Optimal Target Assignment and Path Finding for Teams of Agents | We study the TAPF (combined target-assignment and path-finding) problem for
teams of agents in known terrain, which generalizes both the anonymous and
non-anonymous multi-agent path-finding problems. Each of the teams is given the
same number of targets as there are agents in the team. Each agent has to move
to exactly one target given to its team such that all targets are visited. The
TAPF problem is to first assign agents to targets and then plan collision-free
paths for the agents to their targets in a way such that the makespan is
minimized. We present the CBM (Conflict-Based Min-Cost-Flow) algorithm, a
hierarchical algorithm that solves TAPF instances optimally by combining ideas
from anonymous and non-anonymous multi-agent path-finding algorithms. On the
low level, CBM uses a min-cost max-flow algorithm on a time-expanded network to
assign all agents in a single team to targets and plan their paths. On the high
level, CBM uses conflict-based search to resolve collisions among agents in
different teams. Theoretically, we prove that CBM is correct, complete and
optimal. Experimentally, we show the scalability of CBM to TAPF instances with
dozens of teams and hundreds of agents and adapt it to a simulated warehouse
system. | ['Sven Koenig', 'Hang Ma'] | 2016-12-17 | null | null | null | null | ['multi-agent-path-finding'] | ['playing-games'] | [-7.00787902e-02 5.84753394e-01 3.00108418e-02 -7.78156146e-02
-3.52759510e-01 -1.13956261e+00 4.12522554e-02 7.67677724e-01
-4.49275851e-01 1.30741763e+00 -2.21386254e-01 -6.48938343e-02
-1.11789632e+00 -1.43836093e+00 -5.22879422e-01 -6.48727179e-01
-8.74028206e-01 1.81574368e+00 6.33715928e-01 -6.45276129e-01
1.66267514e-01 4.74860847e-01 -9.54162836e-01 -1.32724941e-01
5.78522623e-01 5.05562305e-01 9.01174396e-02 6.20315909e-01
-3.20878118e-01 3.09205890e-01 -7.15745389e-01 -5.24898648e-01
9.00520861e-01 -2.38140196e-01 -1.08444953e+00 3.01540256e-01
-2.71641374e-01 5.83389141e-02 1.01013273e-01 9.40118551e-01
5.25527541e-03 7.80271888e-02 5.01342714e-01 -2.37071419e+00
-1.20395742e-01 1.23220134e+00 -9.83981252e-01 -2.03698650e-01
5.66662431e-01 -3.78971726e-01 9.85431671e-01 -7.27094263e-02
8.88228655e-01 1.33356643e+00 3.99237841e-01 2.32221916e-01
-1.04887414e+00 -2.89652795e-01 3.22577536e-01 -9.46943164e-02
-1.07379591e+00 4.77229059e-02 -1.69782788e-01 -9.07434002e-02
7.00484991e-01 4.07798499e-01 6.25916004e-01 -8.71617422e-02
2.40875781e-01 5.48161007e-02 8.72029126e-01 -5.07885993e-01
3.83379966e-01 7.95596763e-02 -4.13104333e-02 5.22147596e-01
8.57631624e-01 -5.78664849e-03 -5.69943726e-01 -4.87193257e-01
5.48551023e-01 -1.34533569e-01 -1.51029497e-01 -7.72367179e-01
-1.43152988e+00 1.18535686e+00 4.33401048e-01 -7.99285714e-03
-6.19886339e-01 1.84030548e-01 3.24314177e-01 7.82258332e-01
-1.89569220e-02 7.52164304e-01 -5.13294458e-01 4.13652003e-01
-3.42708901e-02 4.38149005e-01 1.36800849e+00 1.39026809e+00
8.00545335e-01 -4.46958661e-01 1.02861948e-01 1.79352164e-01
2.64194369e-01 5.30688643e-01 -6.22262299e-01 -1.47565186e+00
7.78930783e-01 7.97506750e-01 7.71113098e-01 -1.18721282e+00
-8.35824668e-01 -1.11884721e-01 -3.58315945e-01 8.13341796e-01
7.75427997e-01 -1.27564579e-01 -4.46618050e-01 1.71855044e+00
8.85376692e-01 -5.15040934e-01 3.39997530e-01 9.33260381e-01
1.08493760e-01 7.97286272e-01 -3.74256551e-01 -7.28201270e-01
1.22139859e+00 -1.36157990e+00 -3.39769453e-01 -9.39507503e-03
7.63989866e-01 -8.08269262e-01 9.65925008e-02 3.04726988e-01
-1.39463019e+00 3.21903169e-01 -6.70008361e-01 7.10151553e-01
-5.65214276e-01 -8.70587468e-01 4.36156541e-01 6.72803402e-01
-1.24214232e+00 3.56864512e-01 -3.76419008e-01 -6.95499778e-01
-1.79894641e-01 7.76984870e-01 -6.76980436e-01 -4.70237970e-01
-8.85544658e-01 7.73360848e-01 4.38917965e-01 -1.67188004e-01
-8.56749892e-01 -3.61520261e-01 -7.38763392e-01 -4.81253043e-02
1.29833758e+00 -8.89519751e-01 1.12300622e+00 -5.58363140e-01
-1.07715046e+00 4.38898504e-01 -1.68567002e-02 -3.57577652e-01
4.32145804e-01 8.05342913e-01 -2.93595307e-02 2.23764673e-01
9.73579049e-01 4.84431088e-01 4.80046123e-02 -1.74531460e+00
-1.53979170e+00 -4.35742855e-01 4.98769075e-01 4.26723361e-01
-4.61953543e-02 9.73081067e-02 -1.83074147e-01 2.54255056e-01
2.77972668e-01 -1.00883973e+00 -7.87700295e-01 -2.25661740e-01
-6.09902322e-01 -3.32876027e-01 2.93342590e-01 2.46291533e-01
5.29770851e-01 -1.55148828e+00 4.37062651e-01 9.07978892e-01
6.10604644e-01 -3.84837449e-01 -5.33401370e-01 1.22489369e+00
3.41820806e-01 7.36894906e-02 7.41383135e-02 -1.70553789e-01
3.72673899e-01 6.31901622e-01 1.92057848e-01 5.39816797e-01
-5.91292441e-01 3.64744842e-01 -1.05996954e+00 -2.87758112e-01
-2.47836262e-01 -5.65470636e-01 -2.78194398e-01 -2.39281133e-01
-3.21785867e-01 -1.06409654e-01 -4.69398469e-01 6.33639038e-01
1.09238040e+00 1.40107676e-01 5.92745543e-01 6.05301738e-01
-5.36088407e-01 -1.79033801e-01 -1.64420712e+00 1.02251709e+00
-3.84617150e-02 1.52143259e-02 8.61897707e-01 -6.01716459e-01
8.27619255e-01 -1.80198904e-02 7.80007839e-01 -4.64482605e-01
7.08989054e-02 6.39687538e-01 5.64615540e-02 -1.01753799e-02
6.78021371e-01 -1.44337207e-01 -5.51438212e-01 1.11043155e+00
-5.01621366e-01 2.40814790e-01 7.22625256e-01 7.83171177e-01
1.43672419e+00 -5.42654932e-01 -4.33529653e-02 -1.39406726e-01
3.55397761e-01 8.84481966e-01 7.76505828e-01 1.05723870e+00
-1.92653641e-01 -2.42490485e-01 6.38850152e-01 -4.55273241e-01
-6.97911143e-01 -1.20536518e+00 6.82327688e-01 1.13060665e+00
7.91637540e-01 -3.64379257e-01 -4.66124654e-01 -4.03220773e-01
5.01385272e-01 4.87475246e-01 -6.28378332e-01 3.51239175e-01
-2.72101671e-01 -3.89947265e-01 -2.61545002e-01 -5.01739979e-03
1.75145492e-01 -6.71049297e-01 -6.45889997e-01 7.19575822e-01
-1.95845321e-01 -8.90216589e-01 -7.65601993e-01 2.91045547e-01
-2.23298460e-01 -1.51204705e+00 -2.16580376e-01 -9.46006835e-01
1.00267124e+00 8.40862155e-01 7.40637422e-01 1.17418647e-01
-3.77687253e-02 4.39827681e-01 -6.53596520e-01 -5.62790751e-01
-1.80796593e-01 8.36338326e-02 2.88844079e-01 -2.73443848e-01
4.82737646e-02 -2.49595359e-01 -1.71378225e-01 9.83341336e-01
-3.86178285e-01 -5.01052737e-01 1.25056580e-01 4.14378285e-01
6.76145375e-01 9.81867909e-01 5.92558086e-01 -7.76189387e-01
7.48026967e-01 -6.40463948e-01 -9.47771490e-01 4.58123356e-01
-3.09537351e-01 -6.47166669e-01 6.10595644e-01 -5.75909689e-02
-4.60264027e-01 1.69177949e-01 9.50541854e-01 8.81318152e-02
4.77363579e-02 6.50988162e-01 -2.66936690e-01 -4.29357767e-01
9.63777378e-02 -2.78560460e-01 3.52378905e-01 8.99030343e-02
1.71700880e-01 8.54749084e-02 3.05049598e-01 -5.65712273e-01
1.12582183e+00 5.85444689e-01 7.26262748e-01 -1.34382933e-01
-2.56560117e-01 -3.10153902e-01 -4.72256392e-01 -4.52769488e-01
4.77728516e-01 -4.62003261e-01 -1.61809051e+00 1.33150086e-01
-1.08776188e+00 -2.78859615e-01 -2.91219234e-01 3.27147514e-01
-6.75468802e-01 1.42036811e-01 -2.81904012e-01 -1.12226725e+00
1.77339584e-01 -8.37517381e-01 2.31813177e-01 2.08465427e-01
-1.77548513e-01 -9.97877061e-01 6.43572137e-02 2.57131994e-01
3.25365424e-01 6.34739518e-01 8.55231881e-01 -9.39659059e-01
-9.51756477e-01 -2.41428241e-02 -4.22370210e-02 -9.23074365e-01
2.05374151e-01 -2.98521906e-01 3.86401713e-01 -6.39029980e-01
-4.78968471e-01 -9.83178690e-02 1.65440902e-01 4.45443988e-01
-2.21473202e-01 -5.07117271e-01 -1.03055167e+00 -1.34773985e-01
1.52856445e+00 6.50238991e-01 1.96340472e-01 1.07009780e+00
2.32540831e-01 1.33524644e+00 1.08663678e+00 5.66175282e-01
1.05399239e+00 5.15102684e-01 9.59362268e-01 -1.56078245e-02
7.83078611e-01 1.59367591e-01 1.70244619e-01 1.54116973e-01
-6.09873571e-02 -7.60397494e-01 -7.35221505e-01 7.54197538e-01
-2.26450777e+00 -8.58515978e-01 -7.16978669e-01 2.36106277e+00
2.30130583e-01 5.50722964e-02 7.83436716e-01 1.63274497e-01
9.71729398e-01 -6.99125588e-01 -3.66994053e-01 -8.50211561e-01
-1.74876392e-01 -4.80452240e-01 9.75559950e-01 8.79010379e-01
-6.06366873e-01 6.50218129e-01 5.76691437e+00 2.61977285e-01
4.82202359e-02 8.83914307e-02 2.51259297e-01 -1.72581494e-01
-3.30679357e-01 2.08864823e-01 -7.19743609e-01 -1.14597008e-01
7.23331988e-01 -8.27132344e-01 8.32109928e-01 4.05282855e-01
3.16061646e-01 -4.76265073e-01 -1.08555639e+00 2.02900946e-01
-2.30969176e-01 -1.32194102e+00 -2.21012935e-01 5.78997612e-01
8.21972609e-01 -3.28334808e-01 -2.84711152e-01 -1.93220273e-01
1.11465073e+00 -7.00775087e-01 8.69480729e-01 -4.62767258e-02
6.61629811e-02 -1.42529106e+00 7.72399306e-01 4.25640136e-01
-1.55716085e+00 -5.63425124e-01 -4.93663549e-01 -3.39667290e-01
7.56396651e-01 1.80539832e-01 -9.67605472e-01 1.27707922e+00
8.05297256e-01 -3.64799201e-02 4.24849778e-01 1.51427138e+00
4.02699918e-01 -5.11264145e-01 -6.41384363e-01 1.19092520e-02
6.14194572e-01 -8.51817966e-01 7.08246887e-01 3.61586154e-01
3.83989513e-01 3.81643862e-01 1.05113876e+00 4.61610019e-01
2.68117785e-01 1.78073898e-01 -6.27686560e-01 1.85277626e-01
1.00587285e+00 1.33650947e+00 -1.34558892e+00 8.92860442e-02
-1.62921250e-01 5.20927012e-01 9.45624113e-02 1.49913371e-01
-5.98728001e-01 -7.28778839e-01 7.66904354e-01 2.39881635e-01
2.64930993e-01 -1.97290882e-01 -9.10534337e-03 -3.63002717e-02
-2.82610059e-01 -6.15960240e-01 9.36935306e-01 -6.10920727e-01
-1.47076845e+00 6.20117962e-01 8.47188160e-02 -1.00538039e+00
2.79187206e-02 -2.69613206e-01 -8.02695632e-01 8.16722333e-01
-1.53496659e+00 -8.59394729e-01 1.72561742e-02 9.81388927e-01
1.20789008e-02 -3.16548198e-01 7.03544497e-01 -1.81863621e-01
-2.19862908e-01 1.70988217e-01 9.08419415e-02 -3.73682499e-01
3.26128840e-01 -1.14334369e+00 -1.47860825e-01 6.00485265e-01
-6.48732066e-01 3.89102578e-01 7.24136233e-01 -8.51447046e-01
-1.66274464e+00 -1.12716949e+00 1.01371956e+00 -6.41413629e-02
7.74621487e-01 -1.68262795e-01 -1.97464809e-01 9.49885190e-01
5.30043960e-01 -5.93604684e-01 6.75212801e-01 -5.08732200e-02
1.78078070e-01 -2.64346659e-01 -1.59657049e+00 6.05235815e-01
1.06103957e+00 5.49275160e-01 -1.11859016e-01 7.41114259e-01
1.00783145e+00 -3.31039160e-01 -8.07432473e-01 -1.57336786e-01
-7.65117481e-02 -8.49484503e-01 5.49419999e-01 -4.65587735e-01
-1.98638946e-01 -8.55287194e-01 -2.67645627e-01 -1.74897695e+00
-7.81298459e-01 -9.94826555e-01 9.15567935e-01 1.17360640e+00
7.49188960e-01 -1.38988388e+00 8.19315314e-01 6.12563729e-01
-2.28637561e-01 -4.43441391e-01 -1.26777589e+00 -1.12368107e+00
8.28187093e-02 2.93404043e-01 9.75459218e-01 8.39742184e-01
5.21291673e-01 3.49275589e-01 -1.77653566e-01 7.25044727e-01
1.24710631e+00 4.77736354e-01 9.57382858e-01 -1.50513208e+00
-2.03112453e-01 -1.09558396e-01 3.63406278e-02 -2.51492709e-01
1.46804914e-01 -8.56749535e-01 -5.69029413e-02 -2.31051326e+00
3.70725207e-02 -1.09084570e+00 4.71860141e-01 8.52261245e-01
6.86472714e-01 -1.81030542e-01 6.52111173e-01 8.94725099e-02
-9.33258832e-01 -1.43819023e-02 1.39606571e+00 -2.22874567e-01
-4.92275417e-01 2.99705207e-01 -1.00000906e+00 2.36821979e-01
9.25758243e-01 -8.04165006e-01 -5.20961404e-01 -3.23050976e-01
4.21698421e-01 8.22522700e-01 -1.33614168e-01 -3.02051246e-01
7.32451081e-01 -7.90078640e-01 -1.45375088e-01 -6.27126753e-01
3.21801722e-01 -1.23396933e+00 7.56599009e-01 9.89507854e-01
-5.33307157e-02 5.98915219e-01 -7.43164867e-02 8.08875561e-01
1.08424790e-01 -3.44583094e-01 1.65906549e-01 -4.51506019e-01
-3.53410304e-01 2.56953090e-01 -7.25212216e-01 -3.59729856e-01
1.94349718e+00 -5.56446314e-01 -9.59860504e-01 -6.68146551e-01
-7.18025565e-01 1.37771213e+00 5.78358293e-01 -4.46278565e-02
4.92302746e-01 -1.02604306e+00 -9.44248319e-01 -3.60506564e-01
1.58421755e-01 1.17930532e-01 5.06435297e-02 8.83717358e-01
-3.99713635e-01 2.09592178e-01 -5.34888387e-01 -1.16832376e-01
-1.28172421e+00 9.87182021e-01 3.19721103e-01 -3.19141299e-01
-2.13346347e-01 7.60928392e-01 1.22923199e-02 -6.70995414e-01
2.02485621e-01 3.71096373e-01 -1.06081039e-01 9.71684158e-02
4.68662798e-01 7.63858855e-01 -3.57250899e-01 -5.74995935e-01
-5.78565419e-01 5.88667870e-01 -2.51030654e-01 -4.47828710e-01
1.46557450e+00 -5.68046212e-01 -9.74640369e-01 -2.64127791e-01
3.08174461e-01 7.52714649e-02 -5.61427236e-01 -1.01248892e-02
-1.07906351e-03 -8.87956262e-01 -6.59264207e-01 -8.11818600e-01
-1.27525687e+00 -2.05476463e-01 -3.52510214e-01 8.15818071e-01
1.07645106e+00 5.84896430e-02 4.11610335e-01 4.12202954e-01
1.31229126e+00 -1.08683336e+00 4.00392860e-02 4.93891388e-01
4.89941567e-01 -4.99750614e-01 -1.89541563e-01 -1.13892567e+00
-6.64524674e-01 1.11687958e+00 7.40239799e-01 2.25511089e-01
3.34110826e-01 4.96113867e-01 2.03617601e-04 -3.06595713e-01
-1.00257015e+00 -2.78291702e-01 -1.12586188e+00 1.27269447e+00
-6.71621978e-01 4.39161807e-01 -3.80631655e-01 2.80976772e-01
-2.80694783e-01 -3.80830586e-01 1.52566159e+00 1.21292651e+00
-9.01560068e-01 -1.56172407e+00 -1.05981326e+00 2.81332642e-01
3.49959552e-01 4.79793578e-01 -6.31578922e-01 8.95239055e-01
2.75221914e-01 1.73806190e+00 3.50613981e-01 -2.33864725e-01
7.58818984e-01 -6.18945718e-01 4.22777742e-01 -5.62606335e-01
-6.72037184e-01 1.46224052e-01 5.41556776e-01 -2.78602034e-01
-3.92909825e-01 -6.46509945e-01 -1.68064284e+00 -1.04197264e+00
-2.30069801e-01 8.23611379e-01 2.92441517e-01 4.39441442e-01
1.49211749e-01 2.31270701e-01 1.07561839e+00 -5.93919158e-01
-2.55702496e-01 -1.62575170e-01 -1.20956826e+00 -2.42961943e-01
-1.19908348e-01 -7.98619270e-01 -3.39758277e-01 -7.26065218e-01] | [4.956528186798096, 1.7517321109771729] |
586eac6f-e2b5-42c7-a546-5d9a35467019 | a-new-manifold-distance-measure-for-visual | 1605.03865 | null | http://arxiv.org/abs/1605.03865v1 | http://arxiv.org/pdf/1605.03865v1.pdf | A New Manifold Distance Measure for Visual Object Categorization | Manifold distances are very effective tools for visual object recognition.
However, most of the traditional manifold distances between images are based on
the pixel-level comparison and thus easily affected by image rotations and
translations. In this paper, we propose a new manifold distance to model the
dissimilarities between visual objects based on the Complex Wavelet Structural
Similarity (CW-SSIM) index. The proposed distance is more robust to rotations
and translations of images than the traditional manifold distance and the
CW-SSIM index based distance. In addition, the proposed distance is combined
with the $k$-medoids clustering method to derive a new clustering method for
visual object categorization. Experiments on Coil-20, Coil-100 and Olivetti
Face Databases show that the proposed distance measure is better for visual
object categorization than both the traditional manifold distances and the
CW-SSIM index based distances. | ['Hong Qiao', 'Fengfu Li', 'Xiayuan Huang', 'Bo Zhang'] | 2016-05-12 | null | null | null | null | ['object-categorization'] | ['computer-vision'] | [-2.67874271e-01 -6.07601702e-01 -7.44572580e-02 -4.99505550e-01
-1.94537923e-01 -3.99763286e-01 4.37487334e-01 2.68160045e-01
-3.96750748e-01 2.94800252e-02 -1.36361957e-01 -1.36298165e-01
-5.90559959e-01 -6.61621034e-01 -1.50540918e-02 -7.49213576e-01
-2.73553669e-01 9.27290842e-02 3.19912016e-01 1.32372146e-02
6.50310874e-01 6.95477784e-01 -1.79566896e+00 6.79448852e-03
8.38410497e-01 1.17859554e+00 3.11052471e-01 2.97227234e-01
-3.39464962e-01 2.98578471e-01 -5.12570322e-01 2.12248147e-01
2.74338663e-01 -5.75421214e-01 -6.16429448e-01 2.94996381e-01
3.22345048e-01 2.49895886e-01 -3.19797695e-02 1.46204114e+00
3.98009360e-01 3.37364584e-01 1.10850775e+00 -1.40648210e+00
-9.17770743e-01 1.79665402e-01 -8.18829834e-01 4.49976385e-01
2.11335137e-01 -3.82664412e-01 5.54693878e-01 -9.36280370e-01
5.38595676e-01 1.58068955e+00 5.87188601e-01 7.95010254e-02
-1.12512386e+00 -4.15038168e-01 -1.36801168e-01 9.10558283e-01
-1.86657429e+00 1.12922899e-01 9.59083140e-01 -5.37185073e-01
5.63874304e-01 3.45747799e-01 2.20697835e-01 8.86693522e-02
8.67678970e-02 4.08282161e-01 1.26931334e+00 -5.55554628e-01
1.76031262e-01 1.66001096e-01 2.41711676e-01 7.43997395e-01
6.87838569e-02 -2.95260310e-01 1.78300336e-01 -9.98092592e-02
5.31464338e-01 1.87693983e-01 -3.02795529e-01 -5.91505468e-01
-1.28146589e+00 8.59638751e-01 7.11654425e-01 9.90768671e-01
7.08767399e-02 -1.81611031e-01 6.92927018e-02 3.34778368e-01
2.04446524e-01 1.43735364e-01 9.56837237e-02 3.78551722e-01
-7.75289357e-01 -1.61465555e-01 2.57346988e-01 8.06632161e-01
8.77641439e-01 -3.18709910e-02 1.01457253e-01 9.38400328e-01
6.32710516e-01 4.61742014e-01 7.93954372e-01 -8.74553025e-01
-1.81802176e-02 8.43433380e-01 -2.20506519e-01 -1.93503296e+00
-5.58667481e-01 -9.76212439e-04 -8.33603442e-01 4.17249829e-01
3.63423705e-01 4.72890347e-01 -6.16698503e-01 1.46721458e+00
4.80127841e-01 1.90768465e-01 -8.21449906e-02 8.44082415e-01
1.04961681e+00 6.64138436e-01 -2.16798797e-01 -4.58284140e-01
1.19134712e+00 -4.73573804e-01 -8.08726311e-01 3.39382887e-01
5.46149015e-01 -1.01329899e+00 8.55788052e-01 2.89063781e-01
-8.39528501e-01 -9.31733251e-01 -1.21530378e+00 2.66015917e-01
-5.16504884e-01 3.77982184e-02 8.07619244e-02 6.85604393e-01
-9.49739516e-01 6.81404412e-01 -5.98582804e-01 -6.72004402e-01
1.53557345e-01 1.29082069e-01 -4.22993243e-01 -7.70853609e-02
-5.70310056e-01 9.27911103e-01 5.52320600e-01 -2.17079386e-01
-2.24820063e-01 -1.63702860e-01 -6.01936936e-01 -6.16366528e-02
-1.80442721e-01 5.97540140e-02 4.90041226e-01 -7.18581378e-01
-1.29180753e+00 8.63172770e-01 -1.25984877e-01 5.50352223e-02
2.46786639e-01 6.37248576e-01 -8.35231543e-01 6.37731314e-01
-1.26360014e-01 5.58885276e-01 1.18520117e+00 -1.33781898e+00
-5.66437662e-01 -6.39520168e-01 -5.69115460e-01 1.38595134e-01
-5.14078617e-01 7.01237246e-02 -2.25483179e-01 -6.46133780e-01
7.46514618e-01 -8.11135590e-01 1.25140905e-01 1.02064706e-01
-1.29106998e-01 -5.84991217e-01 1.35134828e+00 -5.35941005e-01
1.13411331e+00 -2.51343322e+00 9.79708880e-02 7.17212141e-01
2.02792615e-01 5.63279212e-01 -3.33013386e-01 3.31465513e-01
-3.82158130e-01 -1.43673643e-02 -1.48383975e-01 1.76040888e-01
-2.55763888e-01 3.21169943e-02 2.77146876e-01 7.26553261e-01
-3.06188781e-03 3.02017778e-01 -7.18481064e-01 -9.41491127e-01
4.15531129e-01 4.80872631e-01 -2.53685623e-01 -1.52158633e-01
3.88186872e-01 2.52048641e-01 -2.16975480e-01 3.70155245e-01
9.27448452e-01 8.16995800e-02 6.01663105e-02 -5.15035272e-01
6.34368882e-02 -6.94573760e-01 -1.51063740e+00 1.33924854e+00
8.55435655e-02 6.04143500e-01 -4.24772166e-02 -1.53320646e+00
1.37420607e+00 1.99115306e-01 6.49028540e-01 -5.03743589e-01
4.35660571e-01 7.27018490e-02 1.94143176e-01 -5.07160366e-01
7.44826421e-02 -5.23617528e-02 3.50188553e-01 3.12169522e-01
1.26438692e-01 -6.48974851e-02 1.80426970e-01 1.30706340e-01
5.98022223e-01 -2.50208050e-01 2.15112984e-01 -7.79841065e-01
1.05805731e+00 -2.77268827e-01 5.17670512e-01 6.89363405e-02
-4.15956706e-01 5.21870673e-01 2.14995772e-01 -1.90200850e-01
-8.67708743e-01 -1.42814612e+00 -5.60443163e-01 4.69186455e-01
5.89829922e-01 -3.40379536e-01 -9.88776326e-01 -5.74171662e-01
1.42843306e-01 3.66256267e-01 -3.65716845e-01 -2.92406708e-01
-2.17138499e-01 -6.11562729e-01 2.63410985e-01 2.96811312e-01
6.77561879e-01 -5.60663462e-01 -2.86673129e-01 -6.67011663e-02
-2.94273168e-01 -7.76781917e-01 -7.77386963e-01 -5.24156272e-01
-1.10970581e+00 -1.21594656e+00 -6.24441803e-01 -1.20293319e+00
9.17892814e-01 8.60293269e-01 5.34933567e-01 2.89007127e-01
-7.63121843e-01 5.22643626e-01 -6.33504689e-01 -2.23975122e-01
-4.55698013e-01 -6.29122674e-01 3.28171372e-01 4.98716354e-01
5.22229671e-01 -4.35583651e-01 -6.79919600e-01 7.58592486e-01
-1.02296877e+00 -6.42697930e-01 3.47960174e-01 6.08452260e-01
5.52431345e-01 5.55169165e-01 5.44577360e-01 -5.86490370e-02
6.76511884e-01 -1.33961543e-01 -4.51209366e-01 2.02664942e-01
-9.31775391e-01 1.13809034e-01 4.48049903e-01 -6.94950640e-01
-7.27316022e-01 -2.03625187e-01 4.18487817e-01 -5.69840252e-01
-1.99210793e-01 3.25760245e-01 -2.62287617e-01 -4.66750830e-01
7.77898073e-01 1.70475245e-01 5.11947870e-01 -5.64743459e-01
4.14541781e-01 9.82968092e-01 4.84411597e-01 -1.00088261e-01
8.02907348e-01 7.48600781e-01 2.12959483e-01 -1.33505416e+00
-1.70832090e-02 -7.25548029e-01 -8.38573992e-01 -2.52616435e-01
1.08793914e+00 -3.93609107e-01 -8.83040428e-01 4.59130913e-01
-8.44313979e-01 4.71441060e-01 -1.09674837e-02 9.10018146e-01
-5.72971404e-01 1.16729593e+00 -4.41544026e-01 -6.04156971e-01
-7.42670596e-02 -1.11065388e+00 5.96031845e-01 1.71555713e-01
-1.25468895e-01 -9.86204624e-01 -1.90442294e-01 2.13407636e-01
2.01493308e-01 3.46819431e-01 1.20979524e+00 -3.52674663e-01
-1.03582248e-01 -2.53960520e-01 -3.58163506e-01 5.97219110e-01
3.91874045e-01 2.65261471e-01 -4.14409935e-01 -4.31543618e-01
3.89664054e-01 1.62676618e-01 5.93750179e-01 3.49208623e-01
1.07215595e+00 -8.86977687e-02 -3.56566548e-01 5.21370828e-01
1.56461239e+00 7.79200554e-01 7.90039003e-01 2.38836229e-01
5.74181974e-01 7.40723252e-01 7.94681966e-01 1.02366187e-01
1.61312461e-01 5.50296128e-01 2.40217969e-01 -1.12007029e-01
1.09570660e-01 2.27164894e-01 1.92568213e-01 1.27334189e+00
-1.52705923e-01 3.67555439e-01 -7.66074896e-01 4.22068149e-01
-1.70917821e+00 -1.16964710e+00 -4.23380822e-01 2.31139827e+00
3.21704060e-01 -3.14358801e-01 2.62159079e-01 6.71979666e-01
1.13392532e+00 -3.66049982e-03 -2.02039197e-01 -5.40184557e-01
-1.30557194e-01 -1.10250730e-02 2.56319314e-01 4.48735505e-01
-9.95880127e-01 5.92470527e-01 6.11547422e+00 1.09230614e+00
-9.26878452e-01 1.12499304e-01 1.44015670e-01 3.30699384e-01
1.81511670e-01 -1.56705648e-01 -3.68928850e-01 5.99915326e-01
4.95400041e-01 -2.22280189e-01 2.72809982e-01 7.07526624e-01
9.50214826e-03 -1.15037180e-01 -8.85879755e-01 1.53790605e+00
4.73523438e-01 -7.73811400e-01 9.52521190e-02 6.96563944e-02
4.53980565e-01 -3.99995208e-01 2.27498889e-01 -2.98601151e-01
-1.79166317e-01 -9.13716912e-01 2.66680032e-01 5.50108790e-01
4.65615034e-01 -9.54095006e-01 4.79311645e-01 2.41558254e-01
-1.43690062e+00 -5.69136366e-02 -6.63517356e-01 3.60392123e-01
-2.31135026e-01 2.17697322e-01 -5.84050238e-01 5.97461760e-01
8.29294741e-01 7.54971206e-01 -7.83799708e-01 1.11533451e+00
4.07791704e-01 1.16954390e-02 -1.36717781e-01 -1.13128982e-01
2.76890010e-01 -7.25303590e-01 4.85857934e-01 9.47408319e-01
6.34257913e-01 1.11865968e-01 5.75155132e-02 6.87758446e-01
5.93055487e-01 5.08782029e-01 -6.18641078e-01 8.90788585e-02
4.66981083e-01 1.29042590e+00 -1.04934311e+00 -3.38527709e-01
-4.91138160e-01 9.18507636e-01 -1.69603705e-01 1.88324034e-01
-5.67805052e-01 -8.72489512e-01 6.83543980e-01 3.22035030e-02
4.52912211e-01 -4.86719400e-01 9.69845206e-02 -1.01305425e+00
-1.46338627e-01 -7.78823674e-01 4.38192636e-01 -6.88738406e-01
-1.35081840e+00 5.79663217e-01 3.05396557e-01 -1.76958942e+00
3.45158279e-02 -7.98875153e-01 -6.01793766e-01 6.45873606e-01
-1.13573611e+00 -6.40302360e-01 -4.01783586e-01 1.13258123e+00
6.92395791e-02 -3.21459651e-01 7.18085706e-01 2.16797575e-01
-2.11380541e-01 4.29899275e-01 5.45948207e-01 2.80243307e-01
7.52863705e-01 -1.12247872e+00 -3.12515050e-01 5.86577773e-01
1.69950351e-01 4.82243627e-01 6.23222232e-01 -3.19767684e-01
-1.11808860e+00 -9.05866861e-01 5.28121352e-01 -2.70461768e-01
4.34105039e-01 1.82618096e-01 -1.01633346e+00 2.44123697e-01
1.26067251e-01 -1.06718183e-01 8.74205291e-01 -4.06729013e-01
-4.80297416e-01 -4.94247347e-01 -1.57901156e+00 3.89409423e-01
7.50144660e-01 -3.99379104e-01 -7.88070917e-01 3.37765783e-01
2.09666431e-01 4.00108308e-01 -1.38842142e+00 4.23434675e-01
3.79068643e-01 -1.02471912e+00 1.19040275e+00 -3.28610927e-01
-3.57761592e-01 -7.96679735e-01 -4.45815146e-01 -1.34477890e+00
-7.37179339e-01 -1.14742287e-01 4.81485099e-01 1.23158073e+00
-1.02699809e-01 -6.81068659e-01 4.06386107e-01 -1.20307483e-01
2.66832173e-01 -1.83818057e-01 -1.05463278e+00 -1.24226725e+00
6.39885142e-02 -3.26807499e-02 3.70344400e-01 1.31803179e+00
4.61584449e-01 2.88134545e-01 1.17782898e-01 7.37443790e-02
1.16775119e+00 6.48654476e-02 5.12210011e-01 -1.52664304e+00
-1.27559051e-01 -8.14130664e-01 -1.37427354e+00 -3.55769515e-01
1.48358457e-02 -1.11704755e+00 -1.13801986e-01 -1.42075574e+00
7.47563392e-02 -3.40599209e-01 -3.53687882e-01 -1.46018833e-01
-1.39547959e-01 3.27443510e-01 4.24916953e-01 5.51502883e-01
-2.77054727e-01 4.59021389e-01 1.08417082e+00 -2.09051788e-01
-1.13022007e-01 -2.73944974e-01 -2.45339587e-01 8.57613802e-01
8.20633531e-01 -1.64221093e-01 -4.90241170e-01 -2.13637739e-01
-5.07483661e-01 -1.99572310e-01 3.24180759e-02 -1.15521836e+00
8.11849833e-02 -5.99052496e-02 2.87333906e-01 -5.89210451e-01
2.07144901e-01 -1.09720719e+00 1.28331602e-01 6.89806342e-01
1.39275510e-02 2.04724446e-01 -2.05342293e-01 7.04945564e-01
-5.28420091e-01 -2.67308265e-01 1.35364199e+00 -7.32064843e-02
-7.75613964e-01 1.92569092e-01 -3.77185881e-01 -2.80974478e-01
1.57099974e+00 -5.85739076e-01 -2.28400096e-01 -2.79584467e-01
-7.88358986e-01 -1.91108305e-02 4.01834607e-01 4.88109112e-01
9.17344034e-01 -1.77338755e+00 -5.36086440e-01 2.51912147e-01
2.83228934e-01 -4.30612564e-01 4.95429114e-02 1.04143703e+00
-7.24275112e-01 2.69598842e-01 -5.85929155e-01 -1.01122808e+00
-1.87250459e+00 1.13666546e+00 1.86274052e-01 3.18435997e-01
-2.66734958e-01 3.78060311e-01 3.03578023e-02 -2.46721119e-01
2.14654207e-01 -3.96006316e-01 -5.42204857e-01 2.19960824e-01
6.37513638e-01 8.22246194e-01 -1.30949616e-01 -1.19548690e+00
-4.64202046e-01 1.37697220e+00 3.79671395e-01 -1.32747009e-01
9.23685849e-01 -2.82638848e-01 -4.61017489e-01 6.09024405e-01
1.80250716e+00 -3.85846883e-01 -5.93840241e-01 -5.15701294e-01
3.03967386e-01 -7.24688470e-01 -1.19464554e-01 -1.07011221e-01
-9.93065655e-01 1.15891886e+00 1.26021445e+00 4.63586748e-01
1.25027215e+00 -3.53225954e-02 3.59871030e-01 3.47262174e-01
4.10238773e-01 -1.31869316e+00 3.38134229e-01 -1.50846317e-02
8.12656939e-01 -1.06881714e+00 -3.98841910e-02 -5.05553007e-01
-5.01600921e-01 1.09052753e+00 3.57216507e-01 -2.77282715e-01
1.12539959e+00 -1.74968496e-01 2.35170886e-01 3.99528742e-02
6.69674426e-02 -3.00147980e-01 4.95992005e-01 8.56896520e-01
2.58075088e-01 2.93915141e-02 -7.98032880e-01 -1.69572294e-01
1.48338854e-01 -3.85665357e-01 1.47159547e-01 7.96544969e-01
-8.96913767e-01 -9.85046744e-01 -7.39109933e-01 2.20475703e-01
-2.99007565e-01 3.74646664e-01 -3.30005914e-01 6.71211779e-01
1.94172636e-01 1.36269307e+00 3.54901582e-01 -7.45604217e-01
1.43610403e-01 9.89672020e-02 6.78056419e-01 -2.33132094e-01
8.15762486e-03 2.89840072e-01 -5.88654816e-01 -4.23014909e-01
-8.57765615e-01 -6.19358301e-01 -1.37714589e+00 -2.69985884e-01
-3.88536483e-01 3.76667559e-01 8.90269995e-01 6.19346559e-01
2.58943737e-01 -4.44422290e-02 1.07058597e+00 -6.58993661e-01
-5.53019941e-01 -1.05116534e+00 -1.01957309e+00 8.74607265e-01
-7.95487091e-02 -7.94620216e-01 -4.86766964e-01 1.22894265e-01] | [7.863219261169434, 4.245777130126953] |
463cc164-fd9b-4b6a-83ef-84d9d23441f3 | semantic-binary-segmentation-using | 1805.00138 | null | http://arxiv.org/abs/1805.00138v2 | http://arxiv.org/pdf/1805.00138v2.pdf | Semantic Binary Segmentation using Convolutional Networks without Decoders | In this paper, we propose an efficient architecture for semantic image
segmentation using the depth-to-space (D2S) operation. Our D2S model is
comprised of a standard CNN encoder followed by a depth-to-space reordering of
the final convolutional feature maps. Our approach eliminates the decoder
portion of traditional encoder-decoder segmentation models and reduces the
amount of computation almost by half. As a participant of the DeepGlobe Road
Extraction competition, we evaluate our models on the corresponding road
segmentation dataset. Our highly efficient D2S models exhibit comparable
performance to standard segmentation models with much lower computational cost. | ['Ian Stavness', 'Shubhra Aich', 'William van der Kamp'] | 2018-05-01 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 5.35164893e-01 6.45960748e-01 -1.21570013e-01 -6.96099758e-01
-8.41914475e-01 -5.87957084e-01 5.27349949e-01 -3.35251577e-02
-5.78262687e-01 3.24618667e-01 1.07003562e-01 -6.22069716e-01
3.98020208e-01 -1.02849293e+00 -9.98500586e-01 5.68129569e-02
3.45199823e-01 2.69840389e-01 8.40845823e-01 -1.41006187e-01
4.92478579e-01 5.61571419e-01 -1.24336612e+00 1.89181045e-01
7.43418038e-01 1.29731953e+00 1.23517901e-01 9.04732645e-01
-3.91792238e-01 5.08285582e-01 -4.01591212e-01 -4.75685060e-01
5.04916072e-01 -2.86261499e-01 -1.37912130e+00 5.12916744e-02
6.44989610e-01 -7.11599290e-01 -5.73903799e-01 8.92451108e-01
1.63607478e-01 -1.44225031e-01 3.65473986e-01 -1.00939906e+00
-3.64173830e-01 4.92925912e-01 -6.24772251e-01 5.98633736e-02
-1.03719786e-01 -2.02291980e-01 8.35977256e-01 -8.83997142e-01
8.41970861e-01 1.20418048e+00 8.03113997e-01 3.25728327e-01
-1.17262220e+00 -2.50799268e-01 5.05536757e-02 -3.00194994e-02
-1.38196993e+00 -3.42149764e-01 4.02673155e-01 -3.98758084e-01
1.13280022e+00 -1.54827058e-01 7.06065774e-01 4.03015167e-01
-5.24032079e-02 1.08114719e+00 5.97117662e-01 -1.17291123e-01
2.54578471e-01 -4.86190528e-01 3.32919329e-01 9.66802657e-01
1.89595580e-01 -8.08572844e-02 -2.79650331e-01 3.75149071e-01
1.11106157e+00 -2.31706485e-01 2.48291045e-01 -3.41900676e-01
-9.68754053e-01 8.83145094e-01 7.56433904e-01 -1.11702241e-01
-8.04417953e-02 8.45512927e-01 4.82027948e-01 -2.92089935e-02
4.90920782e-01 2.50483602e-01 -7.20085680e-01 -2.09805846e-01
-1.25643408e+00 2.80916512e-01 7.07399428e-01 1.15329516e+00
1.14077759e+00 -2.26318508e-01 8.50032270e-02 5.08343637e-01
2.37342685e-01 4.46956120e-02 9.06949639e-02 -1.52734160e+00
5.04972219e-01 5.96432328e-01 -1.76360637e-01 -5.63889325e-01
-4.38918561e-01 -1.43620491e-01 -4.74881917e-01 1.70427382e-01
3.80140305e-01 -2.59354949e-01 -1.33172560e+00 1.19124544e+00
2.22638175e-01 1.34370968e-01 5.38517535e-02 6.57810688e-01
7.48111010e-01 5.57114124e-01 -1.09295845e-02 6.87611818e-01
1.28134358e+00 -1.44493103e+00 -3.66204977e-01 -5.85858762e-01
9.32670474e-01 -6.87662780e-01 7.28861094e-01 -3.70060429e-02
-1.31673801e+00 -3.20968360e-01 -1.08800483e+00 -8.52931917e-01
-3.88656050e-01 1.95410937e-01 7.71412253e-01 4.31037724e-01
-1.46383536e+00 8.87407124e-01 -9.49604034e-01 -2.43604660e-01
1.04469132e+00 4.62639421e-01 -4.51801181e-01 1.36400387e-01
-9.79545951e-01 7.42353022e-01 5.81178129e-01 1.82669535e-01
-6.20484531e-01 -5.34837067e-01 -1.15728426e+00 -5.18377386e-02
3.40710878e-01 -6.53961182e-01 1.65189338e+00 -6.92446351e-01
-1.53430796e+00 1.17172933e+00 -3.95254761e-01 -7.57223129e-01
4.45121676e-01 -4.24848020e-01 2.62481660e-01 3.85378063e-01
2.72619247e-01 1.42173517e+00 5.22109389e-01 -9.30041015e-01
-8.31605256e-01 -1.24719314e-01 2.21094251e-01 1.37945667e-01
3.22669804e-01 -2.90492773e-01 -9.29887950e-01 -5.18710852e-01
4.77307379e-01 -8.12478304e-01 -6.10498846e-01 3.46715808e-01
-7.42442727e-01 1.56473238e-02 8.36417377e-01 -7.24442065e-01
1.03070533e+00 -2.15559292e+00 -5.22785224e-02 2.82875866e-01
4.07554448e-01 2.40805417e-01 -1.25643373e-01 1.82177424e-01
1.49462208e-01 4.54802424e-01 -7.28754044e-01 -4.73343849e-01
-1.76947191e-01 3.38012815e-01 -8.41877982e-02 1.39554054e-01
3.85794282e-01 1.22008371e+00 -6.58306360e-01 -6.84388220e-01
1.91753596e-01 3.37906748e-01 -7.10716009e-01 3.64638939e-02
-1.55728877e-01 8.62198621e-02 -3.37122291e-01 5.84744751e-01
7.39869297e-01 -3.44066560e-01 -1.92455560e-01 -8.96209925e-02
-3.55269283e-01 7.20194280e-01 -7.02475488e-01 2.21244860e+00
-3.75607699e-01 1.16665220e+00 4.81306650e-02 -9.69396353e-01
8.21064115e-01 -2.43422419e-01 3.85217905e-01 -8.59818876e-01
1.17252089e-01 3.76471668e-01 -2.52123982e-01 -5.29166795e-02
6.49625421e-01 3.70194823e-01 -2.35605448e-01 8.94498006e-02
1.39601259e-02 -4.62375879e-01 3.62740546e-01 3.09983701e-01
1.01846099e+00 4.15946186e-01 1.51888505e-01 -3.23725730e-01
2.63087422e-01 4.40584332e-01 3.22305292e-01 5.89913726e-01
-1.11787945e-01 8.27484787e-01 9.39443648e-01 -5.45000732e-01
-1.28200436e+00 -1.07422388e+00 -2.08857507e-01 6.79915667e-01
5.12126684e-01 -5.32485545e-01 -1.37560320e+00 -6.97311580e-01
-8.97736996e-02 4.33156669e-01 -5.56228220e-01 1.49768382e-01
-9.14081573e-01 -3.42709035e-01 1.01220393e+00 9.85323489e-01
1.09423208e+00 -7.77181745e-01 -9.17253554e-01 2.35189319e-01
-6.39117360e-02 -1.38650215e+00 -4.72568005e-01 3.59329790e-01
-1.23119092e+00 -1.12669456e+00 -6.63777113e-01 -1.10151017e+00
6.08776212e-01 2.36751184e-01 1.05430627e+00 4.36555035e-02
-4.68860716e-01 -1.99991748e-01 -9.25749466e-02 -1.52393922e-01
-3.74345630e-02 5.12686133e-01 -1.05611360e+00 -6.07757449e-01
3.29308361e-01 -2.50376523e-01 -9.25355971e-01 1.97156504e-01
-6.39124095e-01 5.08796871e-01 5.04876435e-01 5.16756237e-01
8.61678243e-01 -2.63324916e-01 1.23898298e-01 -9.44711983e-01
7.84085467e-02 -2.65441328e-01 -8.95180762e-01 -2.38277614e-01
-5.26917100e-01 1.31594509e-01 2.37725809e-01 3.47791880e-01
-8.00976634e-01 4.31822330e-01 -6.74992144e-01 1.17430836e-01
-1.33352801e-01 2.55612075e-01 4.47136303e-03 -1.30868047e-01
8.76882300e-02 -7.35705718e-02 -2.82624494e-02 -6.62917972e-01
6.31410003e-01 5.49311161e-01 7.20829606e-01 -9.84234959e-02
7.39013731e-01 5.07889569e-01 1.31973311e-01 -8.33009541e-01
-1.01417470e+00 -3.54328394e-01 -1.06687880e+00 2.44315118e-01
1.31457663e+00 -1.01728308e+00 -4.49876279e-01 6.50735140e-01
-1.49779141e+00 -7.32641459e-01 -4.65482593e-01 5.25227077e-02
-6.67446971e-01 2.48103872e-01 -8.57295394e-01 -9.22410414e-02
-2.69291073e-01 -1.38714206e+00 1.48269749e+00 3.65804106e-01
-1.50777906e-01 -8.54178965e-01 -2.02021107e-01 4.87944335e-01
4.02758986e-01 2.29929298e-01 8.32147062e-01 -3.50028187e-01
-1.05564511e+00 -5.77473566e-02 -8.61816406e-01 3.63934100e-01
-2.63596743e-01 -3.04659139e-02 -8.27179432e-01 4.17784423e-01
-5.99267662e-01 -1.79701909e-01 1.39068532e+00 5.68003774e-01
1.57094646e+00 -4.97731008e-02 -3.70729804e-01 1.08216119e+00
1.53150320e+00 2.22954974e-01 1.06965959e+00 5.59505582e-01
1.06872809e+00 4.92701143e-01 4.27965969e-01 4.98990938e-02
8.07457507e-01 3.55140656e-01 5.73665202e-01 -6.45637751e-01
-4.57727849e-01 -4.96183276e-01 1.33724939e-02 4.67841059e-01
3.06497753e-01 -2.82132357e-01 -1.09354329e+00 9.21904802e-01
-1.66934097e+00 -4.97815728e-01 -5.59312046e-01 1.66443217e+00
6.64511561e-01 2.58942187e-01 -4.47360761e-02 4.31516692e-02
5.88343561e-01 3.31862658e-01 -4.51576442e-01 -9.38144922e-01
9.05921832e-02 7.34572351e-01 1.23789728e+00 7.76577652e-01
-1.40609121e+00 1.62077427e+00 7.50231028e+00 7.29019403e-01
-7.87152886e-01 -2.31876113e-02 6.80445313e-01 1.23296611e-01
-2.59626061e-01 2.89478421e-01 -8.08189154e-01 2.06576392e-01
8.88797700e-01 1.58013165e-01 1.38358012e-01 9.14274693e-01
-1.00427046e-01 -4.35154647e-01 -9.27839935e-01 8.73354971e-01
-2.15438142e-01 -1.81677306e+00 -5.52216992e-02 1.56610221e-01
6.38474524e-01 4.31692004e-01 -1.80662259e-01 -1.10049419e-01
3.91915709e-01 -1.24977231e+00 7.97930956e-01 1.69945449e-01
9.67889845e-01 -7.64458179e-01 6.69417083e-01 -1.53312176e-01
-1.41362226e+00 1.37976900e-01 -2.27809206e-01 4.53041270e-02
5.46818078e-01 5.11220336e-01 -5.54378510e-01 1.58112124e-01
7.05237627e-01 8.08990955e-01 -4.74542201e-01 7.93659210e-01
-3.41492862e-01 5.26271760e-01 -4.03297931e-01 4.07255679e-01
7.97619462e-01 -2.01740324e-01 -7.02347308e-02 1.31968129e+00
1.59445167e-01 7.51271378e-03 -2.31388360e-02 7.46308744e-01
-4.52636898e-01 -2.35145107e-01 -5.45877755e-01 3.06985099e-02
4.07965809e-01 8.05862069e-01 -1.21404195e+00 -6.09368086e-01
-3.77076715e-01 1.34886038e+00 1.59824520e-01 2.20199093e-01
-7.83123672e-01 -9.50159132e-01 8.51308882e-01 2.45405838e-01
6.20503902e-01 -3.55917215e-01 -7.66666234e-01 -7.44532824e-01
-1.78179830e-01 -1.72464579e-01 -1.32366106e-01 -5.92337310e-01
-5.94171584e-01 4.57538784e-01 -5.64280525e-02 -6.40917122e-01
1.22710215e-02 -4.90816295e-01 -4.09629405e-01 7.22390950e-01
-1.89510810e+00 -1.04504418e+00 -4.00547355e-01 4.09171969e-01
7.24255621e-01 2.52281427e-01 4.32640761e-01 3.69488627e-01
-5.93741894e-01 3.84204715e-01 -6.85071647e-02 5.57213187e-01
1.39779046e-01 -1.38301456e+00 1.57083488e+00 9.01111960e-01
-1.97577477e-01 3.85016888e-01 1.16120137e-01 -6.36143565e-01
-7.99194515e-01 -1.25803709e+00 1.08911097e+00 -7.46783391e-02
5.09530485e-01 -3.54892433e-01 -6.45118713e-01 8.83938670e-01
1.07811652e-01 4.73277457e-02 3.16712290e-01 -3.41708511e-01
-2.89367825e-01 1.77333489e-01 -9.79260564e-01 6.34761393e-01
1.37514937e+00 -5.63418090e-01 -5.42476296e-01 2.36985763e-03
1.03743160e+00 -6.63770616e-01 -7.25132823e-01 2.85146505e-01
6.02570713e-01 -9.39463735e-01 1.08618903e+00 -1.25190005e-01
9.06847894e-01 -3.28531504e-01 -1.12556517e-01 -8.00597966e-01
-8.97028577e-03 -5.11706829e-01 8.66642594e-02 7.33945608e-01
5.07247090e-01 -3.80769968e-01 1.13853741e+00 3.22570682e-01
-5.53493261e-01 -9.58040893e-01 -9.02833998e-01 -5.20455420e-01
1.76495701e-01 -4.21921879e-01 5.58239400e-01 3.67420048e-01
-5.17175138e-01 1.57188445e-01 1.37005031e-01 -9.08348188e-02
6.81925714e-01 1.00516118e-01 7.59170890e-01 -1.21030629e+00
1.61022544e-01 -5.93130410e-01 -4.87992823e-01 -1.90509057e+00
-6.20496867e-04 -9.65509474e-01 2.82290787e-01 -2.20282435e+00
-2.22935095e-01 -2.43516132e-01 2.20020235e-01 6.49220526e-01
2.03452542e-01 6.58330262e-01 9.37764719e-02 -3.27471197e-02
-5.19954920e-01 4.04443502e-01 1.29903436e+00 2.29299068e-03
-1.46513879e-01 -3.17126393e-01 -7.58994818e-01 1.15554225e+00
9.23453808e-01 -4.54029709e-01 -4.58030879e-01 -8.77286196e-01
1.59652531e-01 -1.25160143e-01 4.34349328e-01 -8.46207261e-01
1.37438491e-01 6.32695109e-02 2.28200600e-01 -1.03752804e+00
2.48194247e-01 -5.57910323e-01 -3.97891313e-01 4.97583508e-01
-2.21778587e-01 -5.31443954e-03 2.30998337e-01 2.19745263e-01
-4.46318299e-01 -1.73242748e-01 8.43881786e-01 -2.56430119e-01
-1.13757885e+00 2.87523508e-01 -6.31651103e-01 1.67489022e-01
1.06379604e+00 -8.30016851e-01 -3.66360635e-01 1.91365238e-02
-6.55606151e-01 3.82058591e-01 5.48730850e-01 2.31428564e-01
6.73848450e-01 -8.82789612e-01 -2.85897106e-01 1.66324928e-01
-1.64724320e-01 7.28366613e-01 1.24708116e-01 5.47972560e-01
-1.51155317e+00 6.72876179e-01 -1.53614745e-01 -5.42764246e-01
-9.11545217e-01 -3.47197917e-03 5.18344581e-01 -6.98521808e-02
-1.05343676e+00 1.17227995e+00 3.96479577e-01 -3.10726911e-01
6.71764044e-03 -4.50615257e-01 8.89388993e-02 3.26479636e-02
1.53433532e-01 5.13531923e-01 -5.80079183e-02 -5.70106387e-01
-5.04869044e-01 8.27873349e-01 -1.25949845e-01 -2.22190827e-01
1.17333126e+00 -9.35286209e-02 -6.85123727e-02 8.63756165e-02
1.49253178e+00 -5.09558141e-01 -1.67110193e+00 1.32332265e-01
3.10319245e-01 -2.53385931e-01 3.66602778e-01 -5.94709814e-01
-1.28112447e+00 1.09698510e+00 2.73694992e-01 -2.73653030e-01
1.09530759e+00 7.60066211e-02 1.65176904e+00 2.51955211e-01
2.13850155e-01 -1.45377696e+00 -2.08094150e-01 8.32806230e-01
3.87575835e-01 -1.05513871e+00 -1.79929584e-01 -8.65179121e-01
-4.56060231e-01 1.18928289e+00 5.19144714e-01 -4.81799215e-01
8.13035488e-01 5.84717929e-01 1.88761149e-02 -3.66035908e-01
-2.94125199e-01 -6.31979227e-01 -4.24949825e-02 4.72722262e-01
2.86234945e-01 -1.14156716e-01 -2.98945338e-01 1.00485697e-01
-3.33707124e-01 6.80455044e-02 5.61202645e-01 9.90814924e-01
-5.12718141e-01 -1.16201317e+00 1.40687212e-01 2.84562886e-01
-2.51016289e-01 -2.63820648e-01 -4.48671222e-01 9.45464432e-01
1.83980331e-01 6.98388517e-01 6.11128509e-01 -2.72232115e-01
4.21040624e-01 -8.71541426e-02 3.03127408e-01 -6.91261172e-01
-4.99713391e-01 -6.49848860e-03 2.03518227e-01 -1.16606236e+00
-2.76666135e-01 -3.12118590e-01 -1.80938685e+00 -4.53564316e-01
-2.70224288e-02 -3.78565609e-01 1.01174510e+00 9.18715537e-01
5.80207884e-01 6.73414171e-01 4.16099727e-01 -8.55472207e-01
1.20406114e-02 -3.72667044e-01 -3.18445146e-01 -7.35410396e-03
4.27519083e-01 -2.86151081e-01 3.75334173e-02 1.09968759e-01] | [9.483966827392578, 0.10613437741994858] |
27b4e6f5-a92f-4247-b43a-bd344a2c8cb5 | disassemblable-fieldwork-ct-scanner-using-a | 2011.06671 | null | https://arxiv.org/abs/2011.06671v1 | https://arxiv.org/pdf/2011.06671v1.pdf | Disassemblable Fieldwork CT Scanner Using a 3D-printed Calibration Phantom | The use of computed tomography (CT) imaging has become of increasing interest to academic areas outside of the field of medical imaging and industrial inspection, e.g., to biology and cultural heritage research. The pecularities of these fields, however, sometimes require that objects need to be imaged on-site, e.g., in field-work conditions or in museum collections. Under these circumstances, it is often not possible to use a commercial device and a custom solution is the only viable option. In order to achieve high image quality under adverse conditions, reliable calibration and trajectory reproduction are usually key requirements for any custom CT scanning system. Here, we introduce the construction of a low-cost disassemblable CT scanner that allows calibration even when trajectory reproduction is not possible due to the limitations imposed by the project conditions. Using 3D-printed in-image calibration phantoms, we compute a projection matrix directly from each captured X-ray projection. We describe our method in detail and show successful tomographic reconstructions of several specimen as proof of concept. | ['Oliver Cossairt', 'Michael Rubenstein', 'Tobias Würfl', 'Andre Aichert', 'Thomas Bochynek', 'Florian Schiffers'] | 2020-11-12 | null | null | null | null | ['tomographic-reconstructions'] | ['medical'] | [ 4.27058548e-01 -1.18260004e-01 5.34565330e-01 -2.56437868e-01
-5.70183814e-01 -3.05931926e-01 1.92989081e-01 2.67075747e-01
-5.76257825e-01 6.00514889e-01 -4.24615324e-01 -4.80155379e-01
-1.37167037e-01 -8.40932727e-01 -4.98515278e-01 -5.42099297e-01
1.13001075e-02 8.76810014e-01 5.93666553e-01 1.22902438e-01
1.26728237e-01 9.64996636e-01 -1.24625957e+00 -2.23983839e-01
7.15132952e-01 7.64985621e-01 7.26066291e-01 2.48880446e-01
-1.48905143e-01 6.64042458e-02 -3.26250076e-01 -3.54705513e-01
3.39872032e-01 -6.95772707e-01 -4.46358144e-01 5.41348040e-01
-1.27159879e-01 -6.28060520e-01 2.20187828e-01 7.77617812e-01
3.32249790e-01 -1.60085142e-01 5.05159557e-01 -4.54528868e-01
1.22200847e-01 1.11307651e-01 -5.07746398e-01 -2.02049360e-01
3.41269940e-01 5.29090054e-02 1.98787838e-01 -4.36762035e-01
7.20864654e-01 6.01481080e-01 3.85923594e-01 4.52599764e-01
-1.49742377e+00 -4.03211445e-01 -4.11166579e-01 -2.23321334e-01
-1.07160091e+00 -2.02295020e-01 6.12751842e-01 -4.82398987e-01
3.66102546e-01 2.77294517e-01 8.39744568e-01 7.53140569e-01
4.25939888e-01 9.89609864e-03 1.35992169e+00 -5.86825609e-01
5.21917045e-01 3.67281139e-01 -4.92050231e-01 6.54460549e-01
4.53036189e-01 -2.72204667e-01 1.50141073e-02 3.25459428e-02
1.25187218e+00 1.09748103e-01 -5.67499995e-01 -6.25603497e-01
-9.69999909e-01 3.31257045e-01 2.20414653e-01 6.96891129e-01
-3.24989676e-01 -5.36104143e-02 9.55360010e-02 -2.17143223e-02
2.02654913e-01 4.70674187e-01 -4.09719162e-02 -2.20262989e-01
-8.99670601e-01 -2.19816625e-01 4.80405957e-01 5.85532784e-01
3.46379817e-01 -5.79676107e-02 8.14763427e-01 7.29034007e-01
1.95948184e-01 3.22817087e-01 3.47543746e-01 -5.47067761e-01
7.59366378e-02 4.31335449e-01 5.54508902e-02 -6.55454516e-01
-2.57699609e-01 -2.70380586e-01 -7.16640413e-01 3.81802022e-01
4.98001754e-01 2.24843338e-01 -7.95653224e-01 1.08656156e+00
6.06623173e-01 -1.63089305e-01 -2.22339451e-01 1.03611112e+00
3.69401693e-01 4.47711587e-01 4.77328449e-02 -5.10441244e-01
1.18182981e+00 -2.13781018e-02 -4.68074828e-01 -1.41737416e-01
4.61156607e-01 -9.98242199e-01 1.10383928e+00 5.14473081e-01
-1.12917805e+00 -2.78281700e-02 -1.08588600e+00 5.10776162e-01
2.36677587e-01 7.04685319e-03 3.07663471e-01 8.41166854e-01
-4.38820601e-01 8.00742805e-01 -1.28977740e+00 -6.50275469e-01
8.72324631e-02 4.46304888e-01 -6.04296148e-01 -2.70900816e-01
-4.02234346e-01 9.42026377e-01 -9.80306044e-02 2.23055825e-01
-4.66716647e-01 -4.38628107e-01 -4.65365142e-01 -2.39669811e-02
1.62674144e-01 -3.71713996e-01 9.35652018e-01 -3.94516051e-01
-1.74750566e+00 9.80996490e-01 3.73369634e-01 -6.06438853e-02
9.47176278e-01 3.86267938e-02 -2.80236512e-01 4.59294945e-01
8.00986290e-02 1.55976504e-01 4.19913352e-01 -1.37828577e+00
1.97285730e-02 -3.71284485e-01 -9.42638740e-02 -6.48197979e-02
-1.35930046e-01 1.33354157e-01 -5.04463434e-01 -2.28509367e-01
7.70304799e-01 -9.91607547e-01 -3.93718213e-01 3.97969663e-01
-3.12996566e-01 6.87263489e-01 8.32003057e-01 -6.67035520e-01
7.37260640e-01 -2.03965163e+00 -7.43244067e-02 3.86917919e-01
-2.87355334e-01 2.51469649e-02 4.59532976e-01 6.04825497e-01
2.30285581e-02 -1.29838005e-01 -5.81558466e-01 -8.15886706e-02
-4.51175869e-01 8.83161947e-02 1.39322907e-01 9.16264057e-01
-3.48847508e-01 2.11079180e-01 -5.97089410e-01 -7.76316285e-01
4.77214247e-01 4.11027938e-01 -4.00597841e-01 1.78734720e-01
-1.45964287e-02 1.05304921e+00 -6.82860613e-01 5.88031590e-01
6.30047917e-01 -1.91051915e-01 5.09214342e-01 -1.88591257e-01
-3.96016598e-01 3.08387596e-02 -1.10478771e+00 1.70329094e+00
-7.07515061e-01 2.24821299e-01 3.82900745e-01 -6.65104508e-01
5.69353938e-01 5.03896594e-01 7.58410394e-01 -8.58513653e-01
2.19510674e-01 7.09232032e-01 2.97014900e-02 -6.40800476e-01
2.43072778e-01 -8.63679290e-01 2.35511944e-01 6.95329726e-01
-5.26102662e-01 -7.22378552e-01 -2.16421127e-01 -3.57622616e-02
8.07090819e-01 2.88981766e-01 8.02670792e-02 -2.55571663e-01
2.73435801e-01 1.86360702e-01 2.29610443e-01 4.91732098e-02
2.24530488e-01 1.02667761e+00 4.70446199e-02 -3.44624728e-01
-1.40969384e+00 -9.31182861e-01 -5.39384186e-01 -4.81416322e-02
1.39208391e-01 2.96042841e-02 -7.14176178e-01 -1.60223112e-01
-3.70043099e-01 3.59896749e-01 -1.37158126e-01 2.14087769e-01
-7.92272091e-01 -6.07918739e-01 5.43845482e-02 1.69695660e-01
4.65166420e-01 -6.87598348e-01 -1.37766969e+00 5.47965586e-01
-5.65543817e-03 -1.02061796e+00 4.53905985e-02 2.88653702e-01
-1.19064665e+00 -8.39969575e-01 -9.77062881e-01 -4.61542159e-01
1.02135623e+00 1.10974759e-01 7.90761828e-01 3.79882663e-01
-5.09178579e-01 5.76192260e-01 -2.81064004e-01 -2.14003418e-02
-5.04663885e-01 -1.95674524e-01 1.49299186e-02 -2.88258493e-01
-3.16216320e-01 -9.29473877e-01 -7.71607876e-01 6.39485657e-01
-1.19049752e+00 1.49070201e-02 4.89301562e-01 7.18353391e-01
7.67532468e-01 3.96022946e-02 1.07992873e-01 -8.60651135e-01
2.43324459e-01 4.45446873e-04 -9.51046109e-01 1.59348309e-01
-1.95666432e-01 -1.24482624e-01 5.09831131e-01 -3.14144701e-01
-1.09034050e+00 9.44884196e-02 -4.92286831e-01 -9.70481858e-02
-3.86433482e-01 4.93851423e-01 -1.91548467e-01 -3.60443264e-01
5.74604452e-01 9.02908891e-02 4.55703773e-02 -3.96716416e-01
-3.96147996e-01 6.13010049e-01 3.91282201e-01 -5.47603488e-01
8.21857452e-01 6.88092053e-01 3.45427185e-01 -1.25483453e+00
-6.36446327e-02 -3.18698674e-01 -6.53113604e-01 -6.69034719e-01
8.53687644e-01 -4.22689825e-01 -6.56955779e-01 1.22691549e-01
-9.26965654e-01 -2.65382558e-01 -1.05580837e-01 9.24573600e-01
-4.78532791e-01 5.20796955e-01 -4.75726515e-01 -7.01352894e-01
6.06852844e-02 -1.48777068e+00 1.01313365e+00 1.77001670e-01
-2.10534379e-01 -7.14117467e-01 -9.29078013e-02 3.93916726e-01
4.98973846e-01 5.64102650e-01 9.41291213e-01 4.11217064e-01
-9.17095065e-01 -3.12695712e-01 3.37290734e-01 -1.44078853e-02
1.79559276e-01 -1.14504099e-02 -7.88773954e-01 -1.71684265e-01
3.90404969e-01 -4.92274731e-01 9.76742283e-02 4.58926648e-01
1.12096214e+00 2.75573075e-01 -4.41835165e-01 5.78887939e-01
1.71121025e+00 5.37176609e-01 8.99534106e-01 4.13944811e-01
1.86229646e-01 8.09213042e-01 6.71131909e-01 2.63363183e-01
-1.18661143e-01 9.05584514e-01 2.74644822e-01 5.23922592e-02
1.92792311e-01 1.89628862e-02 -1.64429218e-01 7.75558591e-01
-4.45369154e-01 -1.23522662e-01 -1.01592004e+00 2.51410604e-01
-9.46479559e-01 -7.21368492e-01 -5.29326379e-01 2.71147656e+00
6.15839243e-01 -1.58166792e-02 -3.21992040e-01 1.63103893e-01
3.45088959e-01 -7.41328955e-01 -2.03665644e-01 -1.44615009e-01
2.00780362e-01 3.54021579e-01 3.56087446e-01 2.62770951e-01
-5.29838324e-01 2.10272267e-01 5.63195419e+00 2.98556358e-01
-1.40385640e+00 2.21896186e-01 4.98353779e-01 1.32086828e-01
-4.15047377e-01 2.13531643e-01 -1.95585862e-01 3.80713135e-01
6.36412203e-01 2.55421668e-01 -1.10406458e-01 5.76937199e-01
2.52937108e-01 -1.01088274e+00 -8.74080956e-01 8.67869556e-01
-3.61059248e-01 -7.69758999e-01 -5.14979661e-01 5.09954929e-01
2.59551167e-01 -2.89285451e-01 -1.65909559e-01 -4.71294612e-01
-1.85871556e-01 -7.57455289e-01 7.49934793e-01 1.87829379e-02
9.24284279e-01 -5.09383202e-01 4.63945001e-01 4.84030396e-01
-8.40788960e-01 4.12459344e-01 -3.47607046e-01 3.67593408e-01
9.24801350e-01 7.86238313e-01 -8.52071702e-01 5.27203977e-01
7.04324067e-01 -3.15335184e-01 -1.74150705e-01 1.33925903e+00
1.63932905e-01 5.03046095e-01 -7.15464175e-01 7.61116520e-02
9.97614674e-03 -6.67082310e-01 2.62184978e-01 6.83275104e-01
7.83752680e-01 4.94518638e-01 -1.74481109e-01 6.34557068e-01
2.81232536e-01 9.75083262e-02 -4.84814107e-01 1.18699111e-01
-4.23512757e-02 1.15013671e+00 -1.53116071e+00 9.15702805e-02
-2.84028590e-01 1.08212352e+00 -1.65986836e-01 -5.80876917e-02
-7.33077705e-01 -4.74533774e-02 3.49582033e-03 8.79722536e-01
1.54619530e-01 -6.30209744e-01 -2.61620402e-01 -1.05736077e+00
7.87766725e-02 -4.51153100e-01 -4.54976521e-02 -6.98504925e-01
-7.75273621e-01 5.59883058e-01 3.65286767e-01 -1.66887951e+00
-2.21648470e-01 -3.32235873e-01 -3.76501262e-01 7.92221069e-01
-9.71665859e-01 -7.75306106e-01 -3.42074603e-01 2.44355798e-01
1.40555233e-01 3.93218428e-01 9.63847458e-01 5.87583005e-01
-1.82322785e-01 -4.73333262e-02 2.66976386e-01 -2.76593775e-01
5.00285685e-01 -7.46975005e-01 -1.56570464e-01 6.12465739e-01
-9.67740715e-02 6.59990549e-01 8.90367627e-01 -7.00885653e-01
-1.42464614e+00 -4.08078313e-01 5.09425402e-01 5.27251437e-02
3.61344963e-01 -4.27433252e-01 -8.49993348e-01 5.66757023e-01
5.80374524e-02 -9.23512802e-02 7.54790425e-01 -3.11587930e-01
3.81719768e-01 -1.64541956e-02 -1.46926367e+00 5.25861502e-01
6.12161100e-01 -2.47817472e-01 -1.44269541e-01 3.03929478e-01
-6.70038685e-02 -5.80896676e-01 -7.74681389e-01 7.67794400e-02
8.28664184e-01 -1.15769768e+00 7.31044769e-01 1.52160272e-01
3.95084739e-01 -3.38382155e-01 4.82418537e-02 -1.12220144e+00
1.20159060e-01 -5.17035313e-02 9.19242084e-01 9.87621069e-01
4.26672667e-01 -6.76520884e-01 8.79124403e-01 1.05414534e+00
-1.29295856e-01 -4.13134694e-01 -1.17353559e+00 -7.99192309e-01
-1.90500587e-01 -4.00630444e-01 2.45198086e-01 7.25291908e-01
-4.23753448e-02 -2.27053612e-02 -1.62950337e-01 2.86928505e-01
6.83608174e-01 3.85888517e-01 5.89299679e-01 -1.18350279e+00
-7.28057623e-01 1.42944485e-01 -5.60718417e-01 -7.63004065e-01
-5.31420410e-01 -3.15428704e-01 1.64248914e-01 -1.59655821e+00
-1.87189959e-03 -9.28414226e-01 6.49799466e-01 -1.19696490e-01
5.08611917e-01 4.76594269e-01 5.48661202e-02 4.17246342e-01
2.31960937e-01 4.08591062e-01 1.53043771e+00 1.44374460e-01
-7.15494007e-02 8.16179588e-02 5.06735966e-02 6.72095418e-01
6.33164465e-01 -5.91381729e-01 -3.98337573e-01 -5.88305593e-01
1.40912384e-01 4.73515779e-01 1.84519753e-01 -1.33119547e+00
2.16700315e-01 -1.21056296e-01 4.64112908e-01 -5.64989567e-01
6.03065014e-01 -1.49336851e+00 1.06546140e+00 7.80127466e-01
2.94016182e-01 1.97998844e-02 -4.86521572e-02 3.26564968e-01
-1.18296839e-01 -6.49448037e-01 1.09645724e+00 -5.68013370e-01
-1.77473407e-02 -5.73219098e-02 -4.69351679e-01 -5.82022369e-01
1.21365273e+00 -4.41562951e-01 1.39934301e-01 -3.05646062e-01
-6.18677676e-01 -2.49032781e-01 9.98635650e-01 -5.75853050e-01
7.20935524e-01 -9.67912078e-01 -1.52578935e-01 2.15124309e-01
-1.42923281e-01 3.08984548e-01 5.40991604e-01 1.02797484e+00
-1.35891533e+00 2.69303590e-01 -3.81931603e-01 -7.87074804e-01
-1.18222630e+00 1.80454403e-01 2.94643134e-01 -1.58023909e-01
-1.01686144e+00 4.19902861e-01 1.21266142e-01 -1.52266234e-01
-3.76007438e-01 -3.32745582e-01 1.41378686e-01 -4.49481219e-01
3.25993150e-01 7.86397792e-03 5.81192136e-01 -5.22045791e-01
-2.40270793e-01 7.16542423e-01 2.08648309e-01 -5.38302064e-01
1.59819996e+00 -1.48488358e-01 9.56371427e-02 2.53088266e-01
7.10029304e-01 4.36634012e-02 -9.62420404e-01 2.24676669e-01
-2.72773951e-01 -7.52434194e-01 -5.57751656e-02 -4.79430288e-01
-1.02568150e+00 9.25347984e-01 5.40001512e-01 9.80131924e-02
1.22361588e+00 7.91956037e-02 6.57961011e-01 9.60278139e-02
1.13805830e+00 -9.18529272e-01 -1.48928881e-01 -2.37385482e-01
7.04663038e-01 -8.92234683e-01 3.81069809e-01 -6.95364296e-01
-2.72771448e-01 1.15468788e+00 1.86266914e-01 9.08543393e-02
7.20452666e-01 5.14097095e-01 9.65405777e-02 -2.96665519e-01
-1.75860003e-01 7.99103677e-02 -3.02785546e-01 4.81788814e-01
6.67091966e-01 -4.93885167e-02 -7.28200674e-01 -2.12337092e-01
-7.98699036e-02 -1.22103828e-03 8.74875844e-01 1.47521651e+00
-2.47326866e-01 -1.36855793e+00 -8.51039231e-01 3.94902855e-01
-6.93060398e-01 3.64481598e-01 -3.71510126e-02 8.60045433e-01
-2.11714417e-01 5.70627332e-01 -3.27728800e-02 6.89846650e-02
4.07871246e-01 -2.17144683e-01 8.62815976e-01 -6.41641378e-01
-4.88482416e-01 5.70985615e-01 2.13335026e-02 -1.57488465e-01
-7.43660450e-01 -7.17041850e-01 -1.19080031e+00 -2.09362283e-01
-4.68947649e-01 7.82034919e-02 1.20581794e+00 7.34260857e-01
-2.06916273e-01 2.70111591e-01 3.96563768e-01 -7.58751035e-01
-1.83902398e-01 -7.35084414e-01 -1.04424536e+00 1.69053018e-01
-1.98115200e-01 -7.33991563e-01 -5.20896390e-02 2.57274598e-01] | [12.951370239257812, -2.762744665145874] |
bfcb0991-7ad4-46e9-8b90-826520d83f00 | mode-seeking-generative-adversarial-networks | 1903.05628 | null | https://arxiv.org/abs/1903.05628v6 | https://arxiv.org/pdf/1903.05628v6.pdf | Mode Seeking Generative Adversarial Networks for Diverse Image Synthesis | Most conditional generation tasks expect diverse outputs given a single conditional context. However, conditional generative adversarial networks (cGANs) often focus on the prior conditional information and ignore the input noise vectors, which contribute to the output variations. Recent attempts to resolve the mode collapse issue for cGANs are usually task-specific and computationally expensive. In this work, we propose a simple yet effective regularization term to address the mode collapse issue for cGANs. The proposed method explicitly maximizes the ratio of the distance between generated images with respect to the corresponding latent codes, thus encouraging the generators to explore more minor modes during training. This mode seeking regularization term is readily applicable to various conditional generation tasks without imposing training overhead or modifying the original network structures. We validate the proposed algorithm on three conditional image synthesis tasks including categorical generation, image-to-image translation, and text-to-image synthesis with different baseline models. Both qualitative and quantitative results demonstrate the effectiveness of the proposed regularization method for improving diversity without loss of quality. | ['Ming-Hsuan Yang', 'Hung-Yu Tseng', 'Hsin-Ying Lee', 'Qi Mao', 'Siwei Ma'] | 2019-03-13 | mode-seeking-generative-adversarial-networks-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Mao_Mode_Seeking_Generative_Adversarial_Networks_for_Diverse_Image_Synthesis_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Mao_Mode_Seeking_Generative_Adversarial_Networks_for_Diverse_Image_Synthesis_CVPR_2019_paper.pdf | cvpr-2019-6 | ['multimodal-unsupervised-image-to-image'] | ['computer-vision'] | [ 7.15703130e-01 2.35359207e-01 -3.75297256e-02 -2.05786869e-01
-9.88543689e-01 -6.45022690e-01 8.69520366e-01 -2.80850798e-01
-6.63230196e-02 9.10332620e-01 2.95594305e-01 -1.76386476e-01
3.89263362e-01 -7.85311043e-01 -8.31305206e-01 -9.86127198e-01
4.53288794e-01 1.42622247e-01 -1.26306370e-01 -1.18071273e-01
2.96435267e-01 3.04688066e-01 -1.22181904e+00 2.44687736e-01
1.22760582e+00 6.53965592e-01 4.22136515e-01 6.37293577e-01
4.26889583e-02 7.10206389e-01 -9.24913406e-01 -4.76573914e-01
2.69127637e-01 -1.02973986e+00 -4.30748224e-01 2.38174379e-01
3.88747722e-01 -1.63331524e-01 9.87916533e-03 1.24549484e+00
6.56671166e-01 1.07198097e-01 1.02129579e+00 -1.22193396e+00
-8.16041768e-01 4.43410337e-01 -4.09971625e-01 -1.53575376e-01
2.56888568e-01 4.16040063e-01 8.93119872e-01 -9.79360938e-01
7.61873782e-01 1.28238964e+00 3.51694822e-01 7.60194242e-01
-1.62020004e+00 -7.07695246e-01 7.90998191e-02 -2.86572337e-01
-1.41043186e+00 -6.10402346e-01 9.02650416e-01 -4.37955290e-01
6.92306995e-01 1.90030113e-01 1.14189036e-01 1.37806547e+00
1.21583916e-01 6.11514866e-01 1.26857805e+00 -6.25125587e-01
2.01557919e-01 3.58972967e-01 -6.95222437e-01 5.11677980e-01
4.99051288e-02 2.06947044e-01 -4.83201653e-01 7.22101033e-02
9.03505862e-01 -3.85387391e-01 -3.80540788e-01 -3.06231916e-01
-1.23525834e+00 9.13378894e-01 2.66853154e-01 2.52441108e-01
-1.93067536e-01 1.95282176e-01 8.39464180e-03 1.51478812e-01
3.52548629e-01 5.03371119e-01 -1.81166474e-02 7.67870769e-02
-1.07466638e+00 2.59913325e-01 5.50842524e-01 1.18500662e+00
7.57354438e-01 6.59784198e-01 -5.31856000e-01 7.99166501e-01
2.08565712e-01 5.87263823e-01 4.83630240e-01 -8.46886754e-01
6.81951582e-01 3.36401731e-01 9.34411306e-03 -1.11886251e+00
2.72599131e-01 -5.76145887e-01 -1.08040428e+00 3.80286008e-01
1.19072244e-01 -4.55902934e-01 -1.13276434e+00 2.06989074e+00
6.51810020e-02 -4.39997353e-02 1.09221257e-01 6.91055655e-01
6.31645083e-01 7.98947990e-01 9.59480554e-02 -2.54100591e-01
7.64018774e-01 -8.94905448e-01 -8.02177191e-01 -1.89325705e-01
4.51667875e-01 -1.14423001e+00 1.20424902e+00 4.17991094e-02
-1.23052692e+00 -6.47346556e-01 -1.16289210e+00 1.80222049e-01
-1.89248361e-02 4.79707092e-01 3.25477630e-01 6.85323834e-01
-1.05492926e+00 4.78745878e-01 -7.09139824e-01 -5.95074371e-02
1.95892900e-01 2.04959810e-01 -1.05566166e-01 1.29248232e-01
-1.18867111e+00 7.23147273e-01 4.64350700e-01 1.07163921e-01
-1.02751946e+00 -5.01231611e-01 -8.17759216e-01 6.40198439e-02
1.10395305e-01 -7.25768447e-01 9.42153156e-01 -1.31468606e+00
-1.80607641e+00 5.16148150e-01 -1.75498962e-01 -2.30035633e-01
6.86694980e-01 -8.65209997e-02 -3.01367968e-01 -6.64764829e-03
1.65735692e-01 1.03107727e+00 1.29931307e+00 -1.59175146e+00
-1.38511673e-01 2.23595992e-01 -3.16958539e-02 4.14105475e-01
-2.88107485e-01 -2.76147932e-01 -3.21698934e-01 -1.13645649e+00
4.88162078e-02 -9.97846484e-01 -2.21454754e-01 -1.63087606e-01
-6.89251065e-01 1.76497027e-01 6.49331450e-01 -5.68652093e-01
1.02560711e+00 -2.22221756e+00 4.89061087e-01 1.41388059e-01
-2.15787143e-01 8.60507712e-02 -2.13985354e-01 4.01612759e-01
-1.48166448e-01 1.99832246e-01 -5.32274604e-01 -4.68744963e-01
-2.13458240e-02 1.06636457e-01 -4.64324802e-01 1.22775793e-01
4.98228222e-01 8.87761354e-01 -6.81578457e-01 -6.87650025e-01
1.03044853e-01 4.86430228e-01 -8.01277339e-01 4.83055919e-01
-5.08223832e-01 6.94523811e-01 -2.60969251e-01 4.84419197e-01
6.42024398e-01 -1.76298544e-01 1.70457184e-01 -1.49460167e-01
1.10017523e-01 1.66917786e-01 -1.12319613e+00 1.71738267e+00
-5.78572750e-01 6.89384878e-01 -9.97012705e-02 -8.52514446e-01
9.33924913e-01 3.90593886e-01 3.88272628e-02 -4.40635234e-01
1.15778223e-02 1.56972244e-01 2.96763331e-02 -1.14934951e-01
6.06908143e-01 -4.38736796e-01 -2.62920614e-02 3.09265167e-01
2.59622902e-01 -3.55964273e-01 2.06452176e-01 3.19595486e-01
5.07736981e-01 5.01103520e-01 1.16094977e-01 -1.66985571e-01
4.81140286e-01 -3.23174924e-01 5.42691767e-01 5.91647863e-01
1.52368143e-01 1.04989171e+00 5.73344469e-01 1.68413326e-01
-1.05779040e+00 -1.27332139e+00 8.26704577e-02 7.44297862e-01
9.13541112e-03 -2.29622602e-01 -9.04074728e-01 -4.79369253e-01
-4.97960180e-01 9.74955022e-01 -5.67495644e-01 -2.21877709e-01
-5.15133679e-01 -7.21107721e-01 7.02399850e-01 3.44933122e-01
4.84894931e-01 -1.13951516e+00 -1.85980454e-01 1.47487283e-01
-5.76705337e-01 -1.04096997e+00 -6.01646066e-01 -2.73028649e-02
-9.70021546e-01 -4.33559120e-01 -1.04183078e+00 -1.00057614e+00
1.07936525e+00 -1.48319915e-01 1.06762803e+00 -9.22749639e-02
-9.25238281e-02 -2.16733590e-02 -1.69161975e-01 -5.02474681e-02
-9.14676428e-01 1.81512862e-01 -2.14247480e-01 1.00801215e-01
-3.25517654e-01 -7.59957254e-01 -5.73069274e-01 2.29042947e-01
-1.08602083e+00 4.43468094e-01 7.22237825e-01 1.02384901e+00
5.52875340e-01 -1.00767761e-01 6.90484643e-01 -8.89737964e-01
8.58751595e-01 -2.01137543e-01 -4.53802764e-01 2.29749575e-01
-5.89730740e-01 3.67276549e-01 8.71113181e-01 -6.12990856e-01
-1.55763555e+00 8.33092183e-02 -6.79361820e-02 -3.98168623e-01
-1.39848068e-01 3.73484254e-01 -3.10453117e-01 1.08416148e-01
8.50189686e-01 5.18440187e-01 -2.66680866e-01 -2.00085804e-01
5.97215831e-01 2.29917005e-01 5.91726065e-01 -7.43520379e-01
1.09357846e+00 1.65919781e-01 6.39573932e-02 -5.63328624e-01
-4.46037620e-01 2.04883024e-01 -4.36277181e-01 -8.02863389e-02
9.69512939e-01 -1.07632279e+00 -8.46738070e-02 4.40869242e-01
-1.14848924e+00 -2.79991865e-01 -2.18418941e-01 2.65685916e-01
-6.91590548e-01 3.50413382e-01 -5.20262241e-01 -6.20807171e-01
-3.13037157e-01 -1.24023414e+00 9.41520214e-01 1.24768786e-01
-1.24944277e-01 -9.64457810e-01 5.13659008e-02 1.97091728e-01
4.24713790e-01 4.60622013e-01 1.17318833e+00 -1.33910298e-01
-7.47101605e-01 -7.48513937e-02 -1.63618729e-01 5.63578188e-01
3.61062527e-01 7.57339448e-02 -8.28051150e-01 -3.46691459e-01
-4.51781601e-02 -4.71994758e-01 8.22993517e-01 1.99996889e-01
8.73423576e-01 -4.35179025e-01 -9.65382680e-02 6.38688862e-01
1.52743757e+00 1.22479446e-01 8.51599991e-01 -6.18438423e-02
8.66692841e-01 3.80460292e-01 3.62283170e-01 3.80992264e-01
-1.50897130e-01 6.20412469e-01 2.74652213e-01 -2.43431583e-01
-4.24920768e-01 -5.42611718e-01 5.75898409e-01 9.87917483e-01
-5.60301393e-02 -5.30961514e-01 -5.32623589e-01 6.99451208e-01
-1.48534119e+00 -8.15913081e-01 1.40060022e-01 2.07908034e+00
1.20106161e+00 2.71122098e-01 -4.40953434e-01 -1.05492190e-01
9.23274100e-01 3.07912230e-01 -3.58698785e-01 -4.62490410e-01
-2.33268335e-01 3.13952476e-01 2.16047227e-01 5.32053947e-01
-8.48209143e-01 1.00817573e+00 6.26835346e+00 1.03310013e+00
-1.15168834e+00 -1.14392206e-01 6.99524522e-01 -1.13275647e-02
-6.87323272e-01 1.72749758e-01 -5.26255071e-01 4.98281687e-01
6.32566571e-01 -7.62082115e-02 3.57961804e-01 6.99572444e-01
1.71388462e-01 -6.66773121e-04 -1.03799117e+00 7.16611803e-01
1.10650532e-01 -1.29018760e+00 4.65191513e-01 -1.82358772e-01
1.34270191e+00 -5.33427119e-01 4.76133198e-01 1.66507050e-01
2.70912498e-01 -1.21823299e+00 7.69830823e-01 4.31523502e-01
1.17233741e+00 -7.72515953e-01 4.66440827e-01 3.08398992e-01
-9.80923176e-01 3.05505991e-01 -1.70691147e-01 1.40666842e-01
2.99620748e-01 3.82745653e-01 -9.14740562e-01 4.55739081e-01
1.87359348e-01 3.72186989e-01 -6.56525433e-01 4.13825125e-01
-5.10906100e-01 6.70454323e-01 -6.13455363e-02 1.57961875e-01
2.19296664e-01 -3.07594270e-01 7.03613281e-01 1.31678486e+00
5.41229784e-01 -2.06308961e-01 -8.26973394e-02 1.28864360e+00
-2.22525939e-01 -8.01632702e-02 -6.47861600e-01 -2.60218382e-01
2.72946954e-01 9.43247914e-01 -6.99385285e-01 -2.42420673e-01
-2.00711146e-01 1.25219989e+00 1.66515425e-01 6.67282343e-01
-9.19273496e-01 -4.98038948e-01 3.49683613e-01 -2.04851683e-02
3.81135464e-01 -2.64271021e-01 -5.46829164e-01 -1.20533991e+00
8.44536871e-02 -1.15705407e+00 -1.06308125e-01 -7.36798823e-01
-9.60037768e-01 7.04017282e-01 -4.99512218e-02 -1.39132762e+00
-7.13065684e-01 -4.48753312e-02 -6.97729111e-01 1.10436881e+00
-1.25193667e+00 -1.18306506e+00 -1.68191850e-01 7.52732515e-01
6.41737163e-01 -3.55623782e-01 8.80411685e-01 2.41336390e-01
-4.52211827e-01 9.05677974e-01 1.04643732e-01 3.32005285e-02
9.32709813e-01 -1.19321811e+00 3.23471069e-01 1.27216697e+00
4.99745272e-02 7.76226997e-01 9.50601757e-01 -6.47906721e-01
-9.38875496e-01 -1.08010542e+00 8.81338418e-01 -1.25594392e-01
2.08580449e-01 -4.96229738e-01 -6.39799953e-01 5.24353325e-01
5.76674700e-01 -3.64486665e-01 4.27295715e-01 -2.69273818e-01
-3.54841381e-01 1.83266163e-01 -1.11914849e+00 8.90284419e-01
8.75258088e-01 -4.89437342e-01 -2.88215607e-01 1.57989919e-01
8.67288768e-01 -3.86078686e-01 -5.95913768e-01 4.60570306e-01
2.75692731e-01 -8.70938897e-01 8.55663717e-01 -2.98053861e-01
9.28943396e-01 -4.02755052e-01 -1.82042748e-01 -1.41917872e+00
-2.76914984e-01 -7.27481663e-01 7.69430697e-02 1.42805839e+00
6.55222416e-01 -2.47078851e-01 8.03684473e-01 3.51895094e-01
-1.20249555e-01 -4.34468120e-01 -6.54927731e-01 -6.32615328e-01
2.58159399e-01 -8.79466683e-02 3.00060868e-01 8.87173355e-01
-4.18442845e-01 3.95450503e-01 -8.14387202e-01 3.42562012e-02
5.83388686e-01 2.66105771e-01 7.91154802e-01 -5.64338088e-01
-6.16078556e-01 -2.69042581e-01 -1.04989290e-01 -1.07447529e+00
8.76539797e-02 -8.77358198e-01 3.36655796e-01 -1.39491737e+00
9.65923220e-02 -2.78321654e-01 -1.87278986e-02 2.03110427e-01
-3.65344673e-01 2.88533986e-01 3.52844268e-01 3.07502061e-01
-1.20811127e-01 8.62831593e-01 1.57884502e+00 -1.53669953e-01
-2.16115803e-01 -9.74101052e-02 -6.00015223e-01 5.21659493e-01
8.90479922e-01 -5.96441448e-01 -8.53673697e-01 -5.02878368e-01
2.00495780e-01 5.71674332e-02 3.40659827e-01 -1.03429711e+00
7.53581803e-03 -3.57788920e-01 4.99739468e-01 -3.44464213e-01
3.12883586e-01 -5.72043598e-01 4.07533258e-01 3.37351203e-01
-5.53971589e-01 -5.80885075e-02 1.37597680e-01 5.84467173e-01
-4.28136617e-01 -3.00161988e-01 9.83596563e-01 -1.25426695e-01
-2.72581935e-01 2.80564297e-02 -3.74890983e-01 2.02931583e-01
7.51298726e-01 -1.00772247e-01 -8.41637254e-02 -6.72091901e-01
-6.62871540e-01 -2.39652723e-01 4.97332990e-01 3.77059072e-01
6.58760130e-01 -1.54417551e+00 -8.50467503e-01 3.03350747e-01
1.69903841e-02 -4.87101860e-02 1.19884849e-01 3.88050735e-01
-5.53484917e-01 2.88380980e-01 -3.79599273e-01 -5.12203217e-01
-1.01256955e+00 3.83717716e-01 1.80619732e-01 -4.33130860e-01
-2.84525514e-01 8.68100166e-01 5.48422873e-01 -3.78762722e-01
1.33879170e-01 -2.02644125e-01 8.02088678e-02 -1.04700781e-01
1.35657098e-02 1.25798047e-01 -2.26120263e-01 -6.37270033e-01
-6.50672195e-03 3.71452838e-01 -1.70433503e-02 -5.15961111e-01
8.99140179e-01 -1.23190261e-01 7.97600597e-02 1.58435240e-01
1.17385197e+00 4.35620770e-02 -1.47116864e+00 -4.46565486e-02
-4.01369989e-01 -4.49990988e-01 -2.22702354e-01 -8.24439347e-01
-1.17825592e+00 9.73455489e-01 5.29873073e-01 -7.66400844e-02
1.33070064e+00 -3.85654986e-01 6.21121109e-01 6.68992028e-02
1.32936716e-01 -9.87111568e-01 3.95761609e-01 2.77198613e-01
1.03838563e+00 -1.12571907e+00 -2.16142833e-02 -4.12494719e-01
-7.86295116e-01 7.64769971e-01 7.06693590e-01 -1.14676781e-01
3.15464526e-01 1.37913123e-01 8.02099705e-02 3.37889612e-01
-7.15777814e-01 8.98996815e-02 2.77031392e-01 7.64390111e-01
5.95377207e-01 4.03092764e-02 -5.28882146e-01 1.78997621e-01
-2.16840446e-01 -1.98898271e-01 4.29117113e-01 7.50273168e-01
-4.08229940e-02 -1.33071065e+00 -3.39489520e-01 1.00207128e-01
-5.66633523e-01 -4.28954512e-01 -3.36001724e-01 5.43445706e-01
2.80283928e-01 9.42035735e-01 -2.03553498e-01 -2.09259957e-01
6.14334606e-02 1.37710094e-01 5.13095737e-01 -4.59740341e-01
-4.37582850e-01 4.20697182e-01 -2.31299661e-02 -2.53533363e-01
-4.99476731e-01 -5.89983344e-01 -1.08754969e+00 -9.46134999e-02
-3.96038234e-01 -2.63970029e-02 5.88023365e-01 6.70050919e-01
3.79747391e-01 6.52323604e-01 7.05698371e-01 -7.37787962e-01
-5.34877837e-01 -1.03906298e+00 -2.78605878e-01 6.44673467e-01
2.74615526e-01 -3.52755040e-01 -4.63589877e-01 5.39085507e-01] | [11.509248733520508, -0.2809615731239319] |
dc7c2399-9cd2-4380-ba50-587c7630216b | oss-net-memory-efficient-high-resolution | 2110.10640 | null | https://arxiv.org/abs/2110.10640v1 | https://arxiv.org/pdf/2110.10640v1.pdf | OSS-Net: Memory Efficient High Resolution Semantic Segmentation of 3D Medical Data | Convolutional neural networks (CNNs) are the current state-of-the-art meta-algorithm for volumetric segmentation of medical data, for example, to localize COVID-19 infected tissue on computer tomography scans or the detection of tumour volumes in magnetic resonance imaging. A key limitation of 3D CNNs on voxelised data is that the memory consumption grows cubically with the training data resolution. Occupancy networks (O-Nets) are an alternative for which the data is represented continuously in a function space and 3D shapes are learned as a continuous decision boundary. While O-Nets are significantly more memory efficient than 3D CNNs, they are limited to simple shapes, are relatively slow at inference, and have not yet been adapted for 3D semantic segmentation of medical data. Here, we propose Occupancy Networks for Semantic Segmentation (OSS-Nets) to accurately and memory-efficiently segment 3D medical data. We build upon the original O-Net with modifications for increased expressiveness leading to improved segmentation performance comparable to 3D CNNs, as well as modifications for faster inference. We leverage local observations to represent complex shapes and prior encoder predictions to expedite inference. We showcase OSS-Net's performance on 3D brain tumour and liver segmentation against a function space baseline (O-Net), a performance baseline (3D residual U-Net), and an efficiency baseline (2D residual U-Net). OSS-Net yields segmentation results similar to the performance baseline and superior to the function space and efficiency baselines. In terms of memory efficiency, OSS-Net consumes comparable amounts of memory as the function space baseline, somewhat more memory than the efficiency baseline and significantly less than the performance baseline. As such, OSS-Net enables memory-efficient and accurate 3D semantic segmentation that can scale to high resolutions. | ['Heinz Koeppl', 'Özdemir Cetin', 'Tim Prangemeier', 'Christoph Reich'] | 2021-10-20 | null | null | null | null | ['liver-segmentation'] | ['medical'] | [ 3.48838776e-01 6.24375403e-01 -2.80914128e-01 -1.95718795e-01
-8.58997822e-01 -3.47676486e-01 4.39312667e-01 4.64488149e-01
-5.02570927e-01 4.87476975e-01 1.89086586e-01 -7.56238997e-01
-1.59454474e-03 -9.78370547e-01 -7.48007655e-01 -4.26359504e-01
-3.82754207e-01 8.83133471e-01 5.17527580e-01 3.10074270e-01
-2.01568395e-01 7.50966191e-01 -1.00513160e+00 2.76673526e-01
4.69420731e-01 1.30953896e+00 -2.97939181e-02 7.69296110e-01
-3.88495922e-01 4.05411154e-01 -3.44209641e-01 2.35384911e-01
3.01464766e-01 -1.73323467e-01 -1.16115618e+00 -8.54324028e-02
3.35125864e-01 -3.78758609e-01 -2.99988717e-01 7.95030951e-01
5.20926237e-01 -9.01639536e-02 8.53061438e-01 -8.47169161e-01
-2.75333166e-01 3.64984691e-01 -3.65947664e-01 3.33137423e-01
-3.75998951e-03 2.72421509e-01 5.41964650e-01 -5.35920799e-01
6.65978730e-01 1.10118127e+00 1.28057230e+00 7.17463613e-01
-1.35202312e+00 -2.48577520e-01 -1.28100112e-01 -5.50398350e-01
-1.33349943e+00 -1.41925991e-01 1.06287301e-01 -4.34166461e-01
1.58567643e+00 4.40659493e-01 1.11932003e+00 7.59511828e-01
5.60019433e-01 8.92452419e-01 9.47576523e-01 -1.32003516e-01
4.95148093e-01 -2.93840557e-01 1.31703377e-01 1.04258060e+00
7.49940351e-02 1.19233444e-01 -1.91823747e-02 -2.46379137e-01
1.39806366e+00 9.51743498e-02 -1.54993385e-01 -3.83906633e-01
-1.21999741e+00 1.00205135e+00 8.46832693e-01 3.23782414e-01
-2.59965539e-01 5.95690668e-01 8.01900148e-01 -1.37737826e-01
8.24924409e-01 4.66998428e-01 -5.26398122e-01 7.79901892e-02
-1.36916280e+00 2.33360231e-01 7.08570182e-01 8.16543639e-01
4.63561058e-01 4.02953997e-02 -5.01014411e-01 4.34747726e-01
5.68213277e-02 2.90687740e-01 4.73948210e-01 -9.01993155e-01
8.22056457e-02 6.71015143e-01 -3.43819529e-01 -5.02002597e-01
-1.07588494e+00 -5.73258042e-01 -1.17157114e+00 1.88355580e-01
3.63168210e-01 5.89845031e-02 -1.63240170e+00 1.53314531e+00
3.13412756e-01 1.86913386e-01 -1.63178295e-01 7.58490384e-01
1.10727203e+00 4.00566339e-01 1.49688289e-01 -5.03358245e-02
1.54094541e+00 -8.63776743e-01 -2.65016407e-01 -1.05792023e-01
1.06054294e+00 -2.88867623e-01 8.47293675e-01 -3.79860550e-02
-1.34986734e+00 -2.60471076e-01 -9.01925504e-01 -2.34864980e-01
-4.15560424e-01 -5.95214248e-01 8.90046418e-01 8.21080983e-01
-1.57618213e+00 7.08789468e-01 -1.38345397e+00 -2.67664045e-01
1.11039436e+00 7.36204565e-01 -1.37439698e-01 2.26793006e-01
-1.04808784e+00 1.02882421e+00 5.88577807e-01 -2.44826868e-01
-1.08847702e+00 -1.29209518e+00 -1.00188076e+00 -4.51813824e-02
1.19542092e-01 -9.89711642e-01 1.29570067e+00 -6.22666955e-01
-1.22621596e+00 1.19240463e+00 2.04677403e-01 -9.76371467e-01
7.13371396e-01 3.64641994e-01 1.46153584e-01 2.82144874e-01
4.79067154e-02 1.32267535e+00 3.38250548e-01 -7.73222685e-01
-2.87095219e-01 -3.38436872e-01 -1.62467156e-02 1.69691637e-01
1.44145161e-01 -4.53029364e-01 -4.64575440e-01 -5.12942672e-01
2.43207261e-01 -7.06189752e-01 -7.18675852e-01 2.85544991e-01
-3.68789881e-01 -6.28951192e-02 7.89067388e-01 -6.64291739e-01
8.33342731e-01 -1.88234150e+00 -4.18949053e-02 3.63061875e-01
6.85805798e-01 9.58509743e-02 4.92530987e-02 -4.66863036e-01
9.67053920e-02 2.78714001e-01 -6.99629962e-01 -2.42121145e-01
-6.53197765e-02 4.94263738e-01 2.71428883e-01 6.49556875e-01
1.94173440e-01 1.48837209e+00 -7.92083323e-01 -7.27902353e-01
3.29751134e-01 7.09693193e-01 -8.22805285e-01 -1.13369934e-01
-4.29442436e-01 4.34436023e-01 -3.47482383e-01 7.15990186e-01
5.02582669e-01 -6.51949763e-01 -9.16375592e-02 -2.22132966e-01
4.24180217e-02 1.39707163e-01 -6.40898168e-01 1.87558675e+00
-5.13894200e-01 3.80633026e-01 2.30142117e-01 -1.13365817e+00
6.85401201e-01 3.24170381e-01 8.33817244e-01 -9.01633382e-01
3.40374142e-01 3.47293437e-01 -2.56608039e-01 -1.48418233e-01
1.94296435e-01 -4.14097667e-01 -2.39766449e-01 3.67014319e-01
1.41172074e-02 -5.53138673e-01 -1.72179163e-01 -5.72212450e-02
1.39231253e+00 9.36571509e-02 3.69261771e-01 -6.27395451e-01
1.53761983e-01 2.75895119e-01 2.60607988e-01 8.29720616e-01
-3.84867519e-01 5.53238213e-01 6.47547603e-01 -8.16933274e-01
-1.40123844e+00 -1.19089448e+00 -5.72691798e-01 7.06493080e-01
1.67787040e-03 -1.47784382e-01 -1.12986052e+00 -7.21965551e-01
1.36569545e-01 2.75690526e-01 -7.97208190e-01 2.32398182e-01
-8.17164183e-01 -8.44680309e-01 9.01618302e-01 9.01862621e-01
5.60235083e-01 -1.03740752e+00 -1.22023201e+00 3.49264979e-01
1.41692132e-01 -1.01331794e+00 -3.81475121e-01 5.70085347e-01
-1.41364312e+00 -9.30193424e-01 -9.86056864e-01 -8.00417304e-01
7.39854872e-01 -4.48258817e-01 1.38591659e+00 2.71545976e-01
-8.34077656e-01 2.71503299e-01 1.01398669e-01 -4.12221223e-01
-4.82622623e-01 2.70908892e-01 -1.94996119e-01 -8.87281775e-01
1.58738434e-01 -4.51024473e-01 -8.94957900e-01 -1.11236321e-02
-9.09234226e-01 4.31035101e-01 5.23271620e-01 9.31174934e-01
1.03462231e+00 -1.11672238e-01 2.34302655e-01 -1.00380790e+00
2.30440795e-01 -4.09116954e-01 -4.86150503e-01 -5.51714934e-02
-4.64040369e-01 1.15868054e-01 5.84172189e-01 -1.77796528e-01
-3.91731560e-01 2.18499124e-01 -4.34575289e-01 -6.23026311e-01
-2.19207540e-01 2.99082071e-01 4.76227790e-01 -1.18021816e-01
7.57128835e-01 1.92807280e-02 3.31707954e-01 -2.56387979e-01
2.75673419e-01 2.75419772e-01 5.72989643e-01 -5.08396506e-01
1.95035636e-01 6.44815445e-01 3.97875965e-01 -5.72631598e-01
-7.11847544e-01 -2.63875276e-01 -7.47681320e-01 1.26561150e-01
1.31196737e+00 -6.15414381e-01 -5.96644998e-01 2.32304156e-01
-1.07604277e+00 -8.25471580e-01 -6.62416637e-01 1.15091182e-01
-8.42997432e-01 6.82143345e-02 -9.31706071e-01 -3.97275180e-01
-8.06527197e-01 -1.50869346e+00 1.32595944e+00 -1.51415423e-01
-4.07230556e-01 -1.35605323e+00 -3.43393654e-01 1.31859049e-01
6.98896945e-01 7.94534206e-01 1.35364830e+00 -6.15994930e-01
-5.55472374e-01 5.25927953e-02 -5.41757286e-01 -7.90794939e-03
-1.99647352e-01 -5.94669461e-01 -7.99689949e-01 -3.17280948e-01
-1.13594554e-01 -2.41832376e-01 8.63231540e-01 1.03120446e+00
1.66096747e+00 -2.35086650e-01 -6.73707068e-01 9.23102856e-01
1.39367449e+00 7.32467100e-02 4.60704803e-01 1.66520700e-01
5.62522173e-01 1.26325265e-01 -1.66708633e-01 8.94859657e-02
1.07884847e-01 4.23622817e-01 4.75446790e-01 -6.95014417e-01
-4.02606606e-01 1.37962148e-01 -2.81246722e-01 7.14076579e-01
1.32466227e-01 1.61984935e-01 -1.35779572e+00 6.44737065e-01
-1.60138786e+00 -4.77610737e-01 5.61532862e-02 1.89638746e+00
8.38257015e-01 4.12990063e-01 2.18085259e-01 -2.29870062e-02
4.88133103e-01 9.74197984e-02 -8.35773170e-01 -6.29719973e-01
2.52981782e-01 7.76253462e-01 9.07595932e-01 6.03912532e-01
-1.20275927e+00 6.78185046e-01 7.04945230e+00 9.52745140e-01
-1.08115065e+00 6.36986673e-01 1.20652950e+00 -3.17278713e-01
-2.41276130e-01 -4.43316847e-01 -6.24957919e-01 1.25815034e-01
9.68754470e-01 3.11845034e-01 2.28927195e-01 7.80540049e-01
-1.46024391e-01 -1.31630823e-01 -1.30135679e+00 9.94898677e-01
-1.89687163e-01 -1.89674592e+00 -1.45750076e-01 1.97763115e-01
5.87439299e-01 4.25253838e-01 -4.27585244e-02 3.74657989e-01
3.10900450e-01 -1.72075033e+00 5.20118654e-01 1.89577311e-01
1.34360802e+00 -7.80131996e-01 8.49879444e-01 2.37560421e-01
-1.21685505e+00 2.70725340e-01 -2.06883073e-01 2.25409359e-01
1.47105873e-01 5.69779158e-01 -1.05967438e+00 2.87196845e-01
7.28418529e-01 2.67003119e-01 -2.11217061e-01 8.29726875e-01
4.67633396e-01 3.74837518e-01 -6.17449999e-01 -1.88784506e-02
7.92541683e-01 3.35262805e-01 3.48330826e-01 1.51191032e+00
5.85468002e-02 1.05686970e-01 5.43547392e-01 1.12990355e+00
-1.21648356e-01 -2.87853293e-02 -4.82479334e-01 2.08217755e-01
1.76401034e-01 1.00924742e+00 -1.37323081e+00 -4.68734920e-01
-1.71193659e-01 8.51583302e-01 2.33416662e-01 -3.71619016e-02
-9.32599127e-01 -1.11685634e-01 4.78906482e-01 3.90793234e-01
2.70631075e-01 -1.93432905e-02 -7.84106612e-01 -4.73827630e-01
-4.11794811e-01 -3.52504104e-01 4.26250666e-01 -4.28031623e-01
-1.18069911e+00 7.01324284e-01 4.60863560e-02 -6.93109751e-01
4.39187251e-02 -8.08526993e-01 -1.80135474e-01 8.27325583e-01
-1.43612111e+00 -1.03023922e+00 -2.73257047e-01 4.76648241e-01
4.13552612e-01 1.69988036e-01 1.14400709e+00 8.18717182e-02
-1.69455543e-01 4.08011734e-01 -1.98549405e-01 3.91265452e-01
-1.22645125e-01 -1.37642729e+00 6.69566333e-01 2.98749655e-01
-2.56450921e-01 3.19206804e-01 2.20541626e-01 -7.44347513e-01
-1.22260725e+00 -1.29993081e+00 3.42814088e-01 -4.00747240e-01
4.01443601e-01 -4.08239186e-01 -7.49006987e-01 6.16589427e-01
-1.52466714e-01 6.04147911e-01 6.01939738e-01 -6.73955008e-02
-1.13477558e-01 4.31977868e-01 -1.63918114e+00 4.14842159e-01
1.13409770e+00 -4.75959063e-01 -5.00635862e-01 5.18220603e-01
9.89173353e-01 -1.16187966e+00 -1.38685763e+00 6.85103953e-01
3.63254488e-01 -8.54984105e-01 1.18786287e+00 -3.66045594e-01
3.90612036e-01 1.26073146e-02 -5.71266189e-02 -9.24030006e-01
-3.37687284e-01 -2.84028500e-01 -2.79195726e-01 2.17912853e-01
3.25712085e-01 -5.38153172e-01 1.19984794e+00 4.96116251e-01
-4.72078413e-01 -1.41838050e+00 -1.40924752e+00 -6.14144266e-01
5.30510843e-01 -5.02230287e-01 7.22164035e-01 8.22032452e-01
-2.16606125e-01 -1.25399873e-01 3.25535417e-01 -1.71363413e-01
7.12273955e-01 -9.99814048e-02 1.96975172e-01 -1.01145768e+00
-2.50331610e-01 -8.81026864e-01 -6.43425822e-01 -1.04692256e+00
-7.66037544e-03 -1.42988575e+00 -7.61783049e-02 -1.86615336e+00
3.34599167e-01 -7.34669864e-01 -2.71826565e-01 8.04403603e-01
2.45762810e-01 6.69147611e-01 -3.96754891e-02 6.81530610e-02
-4.15142536e-01 1.00577995e-01 1.50636840e+00 -2.01253474e-01
-3.18307400e-01 -2.98343807e-01 -3.07005882e-01 6.76572502e-01
6.89391315e-01 -3.02653164e-01 -3.01503986e-01 -3.47270042e-01
-1.74256757e-01 4.58767772e-01 6.90747142e-01 -1.01339352e+00
1.59340948e-01 3.67904782e-01 8.42714250e-01 -8.76814783e-01
2.38319650e-01 -6.36462092e-01 7.42350072e-02 9.65644836e-01
-1.98789805e-01 -2.54069418e-02 5.85290313e-01 1.32112533e-01
1.92120343e-01 4.98630553e-02 9.37519014e-01 -6.81000531e-01
-4.52830046e-01 7.62684822e-01 -3.40728194e-01 2.84058452e-01
9.07206416e-01 -7.04929948e-01 -1.99238643e-01 1.85012966e-01
-1.03492725e+00 1.24448732e-01 4.04470295e-01 1.39794964e-02
5.53367019e-01 -1.23640597e+00 -5.33754289e-01 2.96968549e-01
-2.70428419e-01 7.75014877e-01 3.14394593e-01 1.08767343e+00
-1.12272406e+00 8.08568299e-01 -2.63947517e-01 -1.20894480e+00
-8.80378842e-01 3.97040814e-01 8.29329431e-01 -6.55449390e-01
-1.21303213e+00 9.12191629e-01 3.89270842e-01 -7.23548830e-01
3.27486157e-01 -8.79371762e-01 2.79163718e-01 -3.55641961e-01
3.69815826e-01 1.39266327e-01 2.69858867e-01 -2.06229582e-01
-5.18519938e-01 3.51395160e-01 1.63006335e-01 1.07093312e-01
1.40178597e+00 3.54527116e-01 -1.46871731e-01 6.83016106e-02
1.36267602e+00 -8.64734471e-01 -1.26014102e+00 -1.37866378e-01
-1.17252499e-01 -5.23232818e-02 4.75766808e-01 -8.87016356e-01
-1.22592914e+00 7.94252634e-01 7.67980099e-01 1.31571934e-01
9.37545240e-01 2.33976945e-01 1.04462457e+00 -7.23055154e-02
3.36172193e-01 -9.01499510e-01 -7.33774304e-02 5.79690516e-01
5.00674188e-01 -8.73308718e-01 6.22861907e-02 -3.44797254e-01
-8.37351382e-02 1.06489587e+00 5.49139380e-01 -1.64847434e-01
8.04550469e-01 9.48150694e-01 -1.38500243e-01 -5.99887073e-01
-3.01647574e-01 -2.78166737e-02 3.94100189e-01 6.07420981e-01
4.59704071e-01 3.29996377e-01 2.41322950e-01 3.11306328e-01
-2.36365795e-01 1.48353176e-02 -6.12620749e-02 9.28382635e-01
-3.12571436e-01 -6.19598031e-01 -3.71185720e-01 9.22472775e-01
-5.02743483e-01 -2.93492258e-01 1.59374431e-01 9.16450679e-01
1.80108860e-01 1.43551797e-01 6.49985015e-01 9.21975300e-02
1.63122550e-01 1.03507750e-01 8.32140326e-01 -6.34658873e-01
-8.56504440e-01 1.33764550e-01 -5.77519089e-02 -8.58175576e-01
-2.96412677e-01 -3.91581595e-01 -1.72285354e+00 -3.68981630e-01
-1.47851765e-01 -2.12017462e-01 7.06611216e-01 9.22951162e-01
2.28761122e-01 8.85128260e-01 -1.62319224e-02 -1.10585463e+00
-3.41225386e-01 -6.05399668e-01 -3.70174825e-01 1.67881280e-01
3.17662895e-01 -4.91121382e-01 -1.74686834e-02 -3.79507422e-01] | [14.51259994506836, -2.523165225982666] |
b5e50014-3b1c-4611-97a5-0660389cbd69 | styleipsb-identity-preserving-semantic-basis | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Jiang_StyleIPSB_Identity-Preserving_Semantic_Basis_of_StyleGAN_for_High_Fidelity_Face_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Jiang_StyleIPSB_Identity-Preserving_Semantic_Basis_of_StyleGAN_for_High_Fidelity_Face_CVPR_2023_paper.pdf | StyleIPSB: Identity-Preserving Semantic Basis of StyleGAN for High Fidelity Face Swapping | Recent researches reveal that StyleGAN can generate highly realistic images, inspiring researchers to use pretrained StyleGAN to generate high-fidelity swapped faces. However, existing methods fail to meet the expectations in two essential aspects of high-fidelity face swapping. Their results are blurry without pore-level details and fail to preserve identity for challenging cases. To overcome the above artifacts, we innovatively construct a series of identity-preserving semantic bases of StyleGAN (called StyleIPSB) in respect of pose, expression, and illumination. Each basis of StyleIPSB controls one specific semantic attribute and disentangles with the others. The StyleIPSB constrains style code in the subspace of W+ space to preserve pore-level details. StyleIPSB gives us a novel tool for high-fidelity face swapping, and we propose a three-stage framework for face swapping with StyleIPSB. Firstly, we transform the target facial images' attributes to the source image. We learn the mapping from 3D Morphable Model (3DMM) parameters, which capture the prominent semantic variance, to the coordinates of StyleIPSB that show higher identity-preserving and fidelity. Secondly, to transform detailed attributes which 3DMM does not capture, we learn the residual attribute between the reenacted face and the target face. Finally, the face is blended into the background of the target image. Extensive results and comparisons demonstrate that StyleIPSB can effectively preserve identity and pore-level details. The results of face swapping can achieve state-of-the-art performance. We will release our code at https://github.com/a686432/StyleIPSB. | ['Min Tang', 'Ruofeng Tong', 'Dan Song', 'Diqiong Jiang'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['face-swapping'] | ['computer-vision'] | [ 2.78640538e-01 8.07506368e-02 4.29723002e-02 -4.07705277e-01
-3.67369682e-01 -5.95506907e-01 4.69198942e-01 -1.12252033e+00
2.46444479e-01 6.32907450e-01 3.66420358e-01 2.26517007e-01
1.19798817e-01 -8.40751648e-01 -6.81417644e-01 -9.43030536e-01
3.96186918e-01 1.11958027e-01 -5.14183402e-01 -3.60080421e-01
-1.10128827e-01 6.01062953e-01 -1.32829642e+00 3.15891802e-01
7.45642185e-01 8.75983477e-01 -2.45216954e-02 3.02708566e-01
-5.21925613e-02 2.90509373e-01 -4.13045496e-01 -7.60693729e-01
6.77630246e-01 -9.21456456e-01 -5.32095194e-01 1.23854496e-01
6.85809672e-01 -2.50858098e-01 -6.33188307e-01 1.26960075e+00
5.26872516e-01 -2.20552310e-01 4.75130647e-01 -1.63103485e+00
-1.17702782e+00 5.04199415e-02 -7.57267356e-01 -4.33967590e-01
3.71576786e-01 2.99285233e-01 5.67494750e-01 -1.06686950e+00
7.22439826e-01 1.62920153e+00 6.78401709e-01 1.11577868e+00
-1.29880476e+00 -1.35602832e+00 -1.55023515e-01 -4.96839136e-02
-1.54836094e+00 -7.01628923e-01 1.12051594e+00 -2.54357368e-01
-6.63327873e-02 5.88861465e-01 7.49310315e-01 1.30377460e+00
2.05017418e-01 4.50824618e-01 1.38333261e+00 -1.13615282e-01
-2.16513127e-01 1.21161409e-01 -5.94760001e-01 9.75216627e-01
1.24408968e-01 3.72996837e-01 -7.37610817e-01 -1.89454272e-01
1.15247631e+00 9.88867879e-02 -6.21725857e-01 -4.00281280e-01
-1.11069214e+00 6.86011493e-01 4.15896744e-01 2.39202064e-02
-1.12377167e-01 1.23228684e-01 -1.70312896e-01 3.79461497e-01
2.51286119e-01 2.64258951e-01 -6.42727017e-02 2.79447854e-01
-6.83113694e-01 2.00285494e-01 4.84197736e-01 1.22804856e+00
9.97605503e-01 1.17495783e-01 -2.15423346e-01 8.97677302e-01
4.85453367e-01 8.62904549e-01 2.13589564e-01 -1.21752214e+00
1.37202546e-01 3.73462886e-01 -2.02220857e-01 -1.28186333e+00
2.04272076e-01 -5.15294299e-02 -1.10822058e+00 2.94122040e-01
-1.01419240e-02 -1.32509694e-01 -9.73119617e-01 1.97534823e+00
3.95625591e-01 3.66987675e-01 -1.34887278e-01 8.27671945e-01
8.96316707e-01 7.21219957e-01 -2.49419942e-01 -1.10101409e-01
1.42855120e+00 -8.85026991e-01 -9.14400518e-01 -1.20547540e-01
3.85570638e-02 -9.81526673e-01 1.11700606e+00 -5.56863062e-02
-1.20083284e+00 -5.81529379e-01 -9.75596607e-01 -8.22364166e-02
4.02193666e-02 3.57938595e-02 3.76126856e-01 7.49383628e-01
-1.25876606e+00 3.80599320e-01 -4.60965544e-01 -9.29776952e-02
6.84012175e-01 4.30822581e-01 -8.34632337e-01 -1.31946281e-01
-1.22990227e+00 6.59562826e-01 -6.99078292e-02 2.70848513e-01
-8.26746404e-01 -1.00995624e+00 -9.76722479e-01 -1.74004614e-01
1.92550030e-02 -9.62318063e-01 6.97516263e-01 -1.14277875e+00
-1.66046679e+00 1.14638054e+00 -2.75477886e-01 4.24049497e-01
5.81603587e-01 1.19722188e-01 -4.91937101e-01 5.94293606e-03
1.49395585e-01 9.76523221e-01 1.22297120e+00 -1.54903960e+00
-3.04512054e-01 -3.77800792e-01 -2.03700081e-01 8.77892226e-02
-4.93639529e-01 -6.49992451e-02 -6.79447412e-01 -9.90263283e-01
1.11345559e-01 -1.11955249e+00 1.72788724e-01 4.70169783e-01
-5.03521204e-01 4.54999357e-01 1.18141258e+00 -7.18749285e-01
8.24147880e-01 -2.40647006e+00 3.74057502e-01 1.89286649e-01
2.15291083e-01 2.91769177e-01 -5.66105425e-01 2.56009698e-01
-3.34810466e-01 2.37091750e-01 -4.34847802e-01 -5.07643342e-01
-8.68921801e-02 2.73348033e-01 -3.38692635e-01 4.82212603e-01
2.98681736e-01 1.14693022e+00 -6.95625961e-01 -4.95555878e-01
1.50404304e-01 1.02285421e+00 -7.13164270e-01 3.37293416e-01
1.74541578e-01 9.93453979e-01 -4.39147502e-01 7.11732686e-01
1.27113128e+00 7.82933012e-02 1.04187550e-02 -5.56924582e-01
2.57528603e-01 -3.54693741e-01 -9.65470135e-01 1.74071872e+00
-2.55871564e-01 5.17419279e-01 2.65827298e-01 -3.98085117e-01
1.05020452e+00 2.12053567e-01 4.33069766e-01 -6.27422750e-01
1.89480424e-01 2.44257689e-01 -2.65268326e-01 -3.67098421e-01
8.61013755e-02 -5.26855350e-01 6.01618141e-02 2.10466802e-01
1.48642242e-01 -3.22983325e-01 -4.56329882e-01 -8.77975449e-02
5.90445697e-01 2.16464892e-01 -7.29021579e-02 -4.08527941e-01
7.55306661e-01 -6.25230432e-01 8.14710557e-01 1.06271774e-01
-8.26332420e-02 1.05133557e+00 3.95551383e-01 -4.46106613e-01
-1.05755210e+00 -1.14994800e+00 -6.97639510e-02 5.56522071e-01
3.27869862e-01 -2.63395756e-01 -1.08143580e+00 -6.04399204e-01
4.88679558e-02 3.09854120e-01 -8.19414556e-01 -5.12014508e-01
-6.74615800e-01 -4.60703075e-01 6.62228584e-01 1.24057293e-01
1.00823748e+00 -9.54232514e-01 5.31896092e-02 -2.21745685e-01
-3.11524451e-01 -9.57447469e-01 -1.14785397e+00 -7.78105438e-01
-5.30122697e-01 -1.10568225e+00 -8.77407312e-01 -1.00225639e+00
1.02899086e+00 2.42718458e-01 8.20222616e-01 2.66172290e-01
-1.96744189e-01 2.23585799e-01 -9.35296342e-02 -1.52439475e-01
-3.88821930e-01 -2.57928878e-01 1.54967383e-01 5.34765959e-01
-6.18105829e-02 -8.24318171e-01 -7.62987196e-01 5.47401965e-01
-9.60678697e-01 3.27898830e-01 3.91878873e-01 8.98095071e-01
4.63943511e-01 -7.29735866e-02 2.90282041e-01 -8.24325025e-01
3.37808043e-01 -1.49973363e-01 -3.49461675e-01 2.45012373e-01
-3.37263525e-01 -7.81414956e-02 4.77110982e-01 -4.08543348e-01
-1.10656047e+00 -1.04069849e-02 -2.76509076e-01 -7.43638635e-01
1.50589272e-01 -2.46084929e-01 -8.31771255e-01 -5.98935843e-01
2.14209452e-01 3.04315090e-01 4.49743837e-01 -4.25197154e-01
4.10976380e-01 3.96244943e-01 6.27695203e-01 -6.41188622e-01
1.24390543e+00 7.11482882e-01 7.63647258e-02 -5.06594002e-01
-6.46412671e-01 3.50841403e-01 -4.21046942e-01 -1.46959677e-01
7.31509030e-01 -9.57592607e-01 -7.49835908e-01 8.45932901e-01
-1.03930020e+00 -9.60956812e-02 -3.68324965e-01 7.86728188e-02
-5.78978479e-01 1.63583308e-01 -4.26646829e-01 -3.46419096e-01
-2.52039045e-01 -1.23862100e+00 1.11477053e+00 4.40842748e-01
4.08565300e-03 -8.60561132e-01 -6.23540208e-02 4.36768472e-01
4.99610364e-01 5.46721458e-01 8.16682935e-01 2.13438496e-01
-5.52949965e-01 7.72330612e-02 -1.56599194e-01 4.15521890e-01
7.11819470e-01 -1.07206302e-02 -9.92466211e-01 -5.68490565e-01
2.65996754e-01 3.32377665e-02 6.14725292e-01 1.10514045e-01
1.18585074e+00 -5.29849470e-01 -3.48198354e-01 1.37170887e+00
1.18337965e+00 1.98490873e-01 8.37720096e-01 -2.02511530e-02
1.05702055e+00 7.16424525e-01 3.61159533e-01 2.05532745e-01
2.66321957e-01 7.30696142e-01 3.87971252e-01 -3.56020033e-01
-4.82791990e-01 -7.36433506e-01 5.23004711e-01 7.17822790e-01
-3.90207693e-02 2.61290353e-02 -3.84038925e-01 2.94354409e-01
-1.44608939e+00 -8.69409144e-01 2.81450897e-01 1.93896306e+00
9.86580133e-01 -4.72187042e-01 -1.93987757e-01 -1.66108847e-01
1.00633061e+00 3.80964547e-01 -4.85777438e-01 -5.02184443e-02
-2.92303920e-01 2.94128031e-01 2.76848793e-01 6.34647787e-01
-7.66082585e-01 9.83332992e-01 5.89931107e+00 9.00465071e-01
-1.19596219e+00 1.70369565e-01 7.72260964e-01 -1.70970231e-01
-7.79880941e-01 -2.70196591e-02 -6.47115827e-01 8.44979942e-01
3.47943157e-01 -1.57829955e-01 6.79233313e-01 4.77344334e-01
-3.30682769e-02 6.18015110e-01 -9.69360709e-01 1.32304561e+00
3.03833693e-01 -1.33860540e+00 5.10692477e-01 2.88580507e-01
1.01378822e+00 -7.55406916e-01 4.66563076e-01 -9.02928784e-02
3.17154154e-02 -1.37711370e+00 7.19834626e-01 6.22851968e-01
1.45486367e+00 -8.36569965e-01 4.27432925e-01 -1.71597362e-01
-1.18398714e+00 1.06394060e-01 -2.60849744e-01 3.06957692e-01
1.26680732e-01 2.70253390e-01 -1.99711666e-01 4.59499240e-01
6.24236941e-01 7.21353948e-01 -2.75141686e-01 4.08094615e-01
-3.80725205e-01 9.55952257e-02 -2.32912018e-03 5.76359332e-01
-1.15283467e-01 -6.17435038e-01 6.88874364e-01 6.88423097e-01
7.27905869e-01 3.16154450e-01 -2.73016363e-01 1.20638323e+00
-5.22477686e-01 -2.17813030e-01 -8.21793139e-01 1.14511646e-01
6.58781111e-01 1.28078187e+00 -1.92992941e-01 -8.70602950e-02
-1.61345944e-01 1.39006114e+00 -1.09611757e-01 3.68700504e-01
-9.70261395e-01 -2.25574031e-01 1.31865299e+00 1.94581091e-01
1.57304779e-01 8.01975131e-02 -2.19191954e-01 -1.23647547e+00
1.29053310e-01 -1.21390879e+00 -2.64602881e-02 -8.53656828e-01
-1.20158517e+00 7.16400385e-01 -2.62449443e-01 -1.09102643e+00
3.46786141e-01 -4.78748024e-01 -6.58175886e-01 1.10098767e+00
-1.44013476e+00 -1.58453834e+00 -6.11949563e-01 8.47481489e-01
2.71263599e-01 -2.97035843e-01 9.02413726e-01 3.85251701e-01
-5.11092365e-01 1.02541614e+00 -7.95591474e-02 2.09215611e-01
9.49848413e-01 -6.66906416e-01 4.90552813e-01 6.58213854e-01
-3.35855298e-02 6.77568734e-01 4.89357620e-01 -6.07950449e-01
-1.60748935e+00 -1.20471871e+00 6.37196898e-01 -5.13542593e-01
6.84154406e-02 -5.22822678e-01 -7.63740122e-01 7.37727165e-01
2.30658859e-01 2.23422140e-01 6.09565139e-01 -5.19296825e-01
-5.41653216e-01 -3.77641022e-01 -1.50068855e+00 7.82056153e-01
1.46312797e+00 -7.22861886e-01 -2.64972657e-01 7.50257075e-02
7.52787232e-01 -4.54524964e-01 -8.30315590e-01 5.60543656e-01
7.54418731e-01 -1.03620970e+00 1.06211483e+00 -4.34455872e-01
5.13185441e-01 -5.03822327e-01 -1.00038096e-01 -1.37265325e+00
-4.54776764e-01 -1.07143474e+00 1.58436581e-01 1.63815165e+00
1.49455876e-03 -8.44915807e-01 7.73222804e-01 5.38288057e-01
1.07769519e-01 -6.61394596e-01 -8.90148103e-01 -5.81381798e-01
7.86326751e-02 3.09127778e-01 1.19772232e+00 1.22378695e+00
-5.33654034e-01 -8.43931455e-03 -7.89181530e-01 1.26981556e-01
7.40918338e-01 1.36228099e-01 7.89313138e-01 -8.17879140e-01
6.55470639e-02 -3.17858666e-01 -4.55461860e-01 -8.42747986e-01
3.91614169e-01 -8.20957422e-01 -1.64601058e-01 -8.88986468e-01
3.27885330e-01 -5.72884023e-01 -8.35909098e-02 5.96886575e-01
-2.13921681e-01 8.43144357e-01 3.59481037e-01 2.98820198e-01
2.36861959e-01 9.33243036e-01 1.95798206e+00 -1.76330522e-01
8.02869797e-02 -3.63923311e-01 -1.12003338e+00 5.82099438e-01
7.01171696e-01 -3.49211365e-01 -5.19808292e-01 -5.72828829e-01
-1.80897176e-01 5.47452420e-02 5.15491307e-01 -7.96776533e-01
-7.91874230e-02 -3.52394581e-01 6.12408221e-01 6.43090392e-03
5.25330007e-01 -8.45311165e-01 8.66730273e-01 3.94629508e-01
-6.89816549e-02 4.68724929e-02 6.97159618e-02 2.80814350e-01
-3.09222430e-01 1.67870447e-01 1.16333127e+00 -5.33769690e-02
-4.28993404e-01 8.22909176e-01 3.19359720e-01 -8.21282491e-02
1.04021752e+00 -3.55762452e-01 -2.18527108e-01 -4.75954145e-01
-4.25139189e-01 -2.18231641e-02 9.83343244e-01 6.69846833e-01
8.12925458e-01 -1.96633112e+00 -8.92204821e-01 9.41119552e-01
-8.16506818e-02 -1.76171251e-02 3.89053881e-01 5.08049250e-01
-6.28901482e-01 -6.33107647e-02 -7.46322036e-01 -3.71357352e-01
-1.31622791e+00 4.58297282e-01 4.70670998e-01 2.97965795e-01
-5.29167235e-01 1.09591591e+00 7.78096914e-01 -5.76178610e-01
-2.34742180e-01 3.66881073e-01 1.80861369e-01 -1.06474496e-01
5.75560987e-01 5.26118279e-02 -3.69071126e-01 -1.05607831e+00
-3.43241483e-01 1.16167390e+00 5.20477518e-02 -1.25444576e-01
1.25155866e+00 -2.42929682e-01 -4.81958658e-01 -1.70999959e-01
1.51663780e+00 2.66295731e-01 -1.56879532e+00 -3.83386239e-02
-7.10266113e-01 -1.12591875e+00 -2.11018652e-01 -4.82265025e-01
-1.71725106e+00 7.60868251e-01 6.57076359e-01 -3.61909986e-01
1.39984334e+00 -1.87014252e-01 9.87465799e-01 -3.42645258e-01
3.62769186e-01 -5.96703649e-01 1.33554146e-01 2.20820576e-01
1.22480357e+00 -9.32212889e-01 -2.12896615e-01 -6.60737395e-01
-5.71557939e-01 7.77435064e-01 7.03408897e-01 -1.46638766e-01
7.03308344e-01 1.40874356e-01 2.26445258e-01 -1.58387452e-01
-3.16292793e-01 3.86558443e-01 3.56133848e-01 7.82748938e-01
9.05649289e-02 7.08988681e-02 8.04303139e-02 3.35084051e-01
-5.66970587e-01 -1.81427240e-01 2.03329802e-01 5.76791883e-01
1.00240134e-01 -1.43469608e+00 -5.19929290e-01 -1.78102423e-02
-1.79718554e-01 -7.62303025e-02 -4.75939721e-01 7.13251173e-01
3.93003345e-01 6.12638950e-01 6.05214462e-02 -6.83960021e-01
3.51478010e-01 -3.61505970e-02 7.24406242e-01 -3.21805745e-01
-2.28954494e-01 -8.28239769e-02 -3.60552728e-01 -7.21941113e-01
-3.20026666e-01 -4.20049310e-01 -9.04883802e-01 -9.28334534e-01
-1.69594064e-02 6.42466843e-02 3.14455152e-01 4.47784126e-01
6.86320424e-01 1.88773751e-01 1.10557628e+00 -7.78073072e-01
-1.01492241e-01 -5.60826659e-01 -6.27868712e-01 6.73964143e-01
4.99460131e-01 -6.76269174e-01 -3.54438394e-01 2.50651896e-01] | [12.663434982299805, -0.09527446329593658] |
6c1c6e13-a511-4239-bfd9-e9c8c379d666 | contraction-mapping-of-feature-norms-for | 2007.13406 | null | https://arxiv.org/abs/2007.13406v2 | https://arxiv.org/pdf/2007.13406v2.pdf | Contraction Mapping of Feature Norms for Classifier Learning on the Data with Different Quality | The popular softmax loss and its recent extensions have achieved great success in the deep learning-based image classification. However, the data for training image classifiers usually has different quality. Ignoring such problem, the correct classification of low quality data is hard to be solved. In this paper, we discover the positive correlation between the feature norm of an image and its quality through careful experiments on various applications and various deep neural networks. Based on this finding, we propose a contraction mapping function to compress the range of feature norms of training images according to their quality and embed this contraction mapping function into softmax loss or its extensions to produce novel learning objectives. The experiments on various classification applications, including handwritten digit recognition, lung nodule classification, face verification and face recognition, demonstrate that the proposed approach is promising to effectively deal with the problem of learning on the data with different quality and leads to the significant and stable improvements in the classification accuracy. | ['Xiabi Liu', 'Weihua Liu', 'Ling Ma', 'Murong Wang'] | 2020-07-27 | null | null | null | null | ['handwritten-digit-recognition', 'lung-nodule-classification'] | ['computer-vision', 'medical'] | [ 2.95640051e-01 -1.75810620e-01 -3.77664953e-01 -7.49055445e-01
-4.67000008e-01 1.50079459e-01 1.57007828e-01 -1.51407257e-01
-5.63855052e-01 7.42415905e-01 -2.69098848e-01 -5.83679788e-02
-5.06142676e-01 -6.94897950e-01 -4.44569796e-01 -1.07516646e+00
-2.40886956e-02 2.81365141e-02 -1.18830793e-01 2.29674384e-01
2.65065104e-01 4.80597913e-01 -1.46032059e+00 1.71335995e-01
7.22029328e-01 1.65130258e+00 1.83720767e-01 1.97649971e-01
-1.56845421e-01 8.74821186e-01 -5.68246663e-01 -2.21645892e-01
3.50061327e-01 -2.29788810e-01 -6.63221240e-01 2.42082685e-01
2.92507350e-01 -1.86014399e-01 -3.97299379e-01 1.22181618e+00
6.33727610e-01 8.01830664e-02 6.38611078e-01 -1.18000829e+00
-7.23960936e-01 2.85214633e-01 -6.42087519e-01 2.55510181e-01
-2.00229257e-01 -1.08375363e-01 8.09700429e-01 -9.33860838e-01
1.65303260e-01 1.08946300e+00 6.52419329e-01 4.67807233e-01
-8.15532148e-01 -7.33277678e-01 -6.17336147e-02 5.48334002e-01
-1.45334339e+00 -2.32929602e-01 8.97157252e-01 -3.00206661e-01
4.45513815e-01 1.73324376e-01 3.00983965e-01 8.86033654e-01
2.85434783e-01 8.80188346e-01 1.07440233e+00 -3.07489574e-01
-4.16767374e-02 4.65936810e-01 4.11938690e-02 8.35582018e-01
1.96872219e-01 5.60447164e-02 -4.32278097e-01 2.10956454e-01
5.65562904e-01 1.28602773e-01 -5.37313044e-01 -7.15923235e-02
-8.33617926e-01 9.16760027e-01 6.97634637e-01 5.86079001e-01
-2.69695938e-01 -8.94413739e-02 4.00567174e-01 4.98657346e-01
4.85952556e-01 2.56793618e-01 -4.67358261e-01 8.04880187e-02
-7.37238646e-01 -9.97261405e-02 5.77279627e-01 5.54615498e-01
5.84894300e-01 1.87359020e-01 -1.46041229e-01 1.17613661e+00
4.01997894e-01 3.07946742e-01 8.85399759e-01 -6.30127966e-01
4.26015288e-01 5.85838497e-01 -4.06247407e-01 -1.37193501e+00
-4.30308342e-01 -7.65513241e-01 -1.21582150e+00 5.31467617e-01
2.86525816e-01 -8.94290656e-02 -7.39073217e-01 1.72268546e+00
3.30639303e-01 1.65591151e-01 3.90923619e-02 9.27689910e-01
1.05756867e+00 5.93364418e-01 -1.57906547e-01 -3.44782025e-01
1.09995079e+00 -6.57393038e-01 -8.87353063e-01 -4.98399697e-02
4.41592991e-01 -7.21367776e-01 1.10746169e+00 6.96061671e-01
-9.13594961e-01 -9.64848995e-01 -1.29590356e+00 3.19069698e-02
-1.25297651e-01 4.81064171e-01 4.72578734e-01 5.35123050e-01
-6.21678829e-01 6.71199262e-01 -7.65335083e-01 1.39088884e-01
8.21499407e-01 5.34373403e-01 -4.76331413e-01 -1.79162219e-01
-1.06473076e+00 8.24257076e-01 6.76551223e-01 4.40794230e-01
-4.88618016e-01 -6.33231938e-01 -8.41446638e-01 1.23563826e-01
2.12324858e-01 -3.47260714e-01 9.43136454e-01 -1.32700074e+00
-1.33922064e+00 8.51719618e-01 2.65198588e-01 -3.48846853e-01
5.36477685e-01 1.18182130e-01 -4.21398669e-01 -1.37265325e-01
-2.46788368e-01 4.88290101e-01 1.08923686e+00 -7.03167677e-01
-7.06020653e-01 -4.85127509e-01 -3.04646492e-01 1.50446862e-01
-9.37375605e-01 -2.06637084e-01 -5.45314364e-02 -6.93305135e-01
8.29357952e-02 -5.80446303e-01 4.41957600e-02 4.33188647e-01
-1.64669782e-01 -3.97695243e-01 1.13047755e+00 -5.28595448e-01
1.03672540e+00 -2.40301991e+00 2.24572152e-01 3.03483963e-01
1.34635434e-01 3.86017501e-01 -2.92115778e-01 -2.63811707e-01
-1.07154027e-01 -6.05643243e-02 -3.73277903e-01 -8.37006569e-02
-1.83647811e-01 3.81957501e-01 1.14512302e-01 5.56042314e-01
4.10334527e-01 6.94812655e-01 -5.64852834e-01 -6.15894496e-01
1.67452931e-01 6.60171688e-01 -2.11809918e-01 2.31689081e-01
1.15876876e-01 2.39439681e-01 -4.97048408e-01 5.28136373e-01
8.38353813e-01 -2.96365738e-01 -1.45867214e-01 -4.37269270e-01
3.40814799e-01 -1.33846819e-01 -1.19221020e+00 1.48072994e+00
-3.77690464e-01 7.21954286e-01 2.28452772e-01 -1.52380443e+00
1.20892763e+00 2.20045269e-01 6.01172805e-01 -8.56157064e-01
3.08232903e-01 2.90649801e-01 1.03821203e-01 -8.31775784e-01
-3.65288220e-02 -4.37077433e-01 3.51695716e-01 1.46073669e-01
3.63336317e-02 9.05491784e-02 8.18925500e-02 -4.71438259e-01
7.81075656e-01 -3.07177335e-01 2.57132918e-01 -1.12520300e-01
9.78706539e-01 -5.45440018e-01 6.72020197e-01 3.29013407e-01
-3.67792100e-01 3.72927338e-01 3.68052334e-01 -5.40826797e-01
-7.68486738e-01 -6.87932432e-01 -6.12846315e-01 7.23015010e-01
7.51986820e-03 1.15602486e-01 -5.89884043e-01 -1.00395894e+00
4.43229377e-02 1.12521954e-01 -6.34778857e-01 -6.05222046e-01
-5.12284994e-01 -1.03717554e+00 4.65955973e-01 5.29827833e-01
8.83243680e-01 -1.10082042e+00 -3.60631555e-01 -3.37940566e-02
1.41308546e-01 -8.71984482e-01 -3.58326316e-01 1.87414646e-01
-9.17818129e-01 -1.11602306e+00 -7.55067229e-01 -1.25534928e+00
8.43659282e-01 5.97521141e-02 6.86859846e-01 2.71722972e-01
-4.95225668e-01 -7.58967176e-02 -2.82462567e-01 -2.78567851e-01
-4.31331992e-01 6.45736605e-02 2.02686191e-02 2.26926163e-01
2.00939953e-01 -3.07990223e-01 -4.63572174e-01 3.38640243e-01
-1.31488240e+00 -2.49198422e-01 7.53092349e-01 1.17318451e+00
6.60931587e-01 6.12433314e-01 7.29208171e-01 -6.88166201e-01
8.37860584e-01 -3.82682979e-01 -5.11474133e-01 2.24609211e-01
-8.76583457e-01 1.84249461e-01 7.12921202e-01 -5.00205398e-01
-9.91645873e-01 -9.61463451e-02 -3.38619888e-01 -2.93748111e-01
4.13289852e-02 9.18090999e-01 -1.65997952e-01 -3.79842609e-01
5.80993354e-01 2.38765731e-01 3.38633865e-01 -4.53728944e-01
-1.57256037e-01 7.07266033e-01 3.92164648e-01 -3.18577170e-01
6.43773377e-01 1.45577148e-01 2.80511111e-01 -7.31971800e-01
-8.47160280e-01 -1.60151407e-01 -3.54292065e-01 -1.06176630e-01
7.50908971e-01 -5.11324584e-01 -6.86772764e-01 7.32328832e-01
-9.03580964e-01 1.98701322e-02 -1.93220228e-01 7.63562381e-01
-3.45827371e-01 4.42494303e-01 -5.53939223e-01 -4.99548256e-01
-5.32754362e-01 -1.26745963e+00 5.83600044e-01 3.48626703e-01
3.04221928e-01 -1.00796723e+00 -1.29443377e-01 2.23265499e-01
4.54102546e-01 1.92091092e-01 9.93645787e-01 -6.26097918e-01
-3.72753382e-01 -4.48729903e-01 -4.05410528e-01 1.19046926e+00
4.13250804e-01 -9.71784368e-02 -7.68272996e-01 -5.81176698e-01
5.51020324e-01 -5.92515349e-01 9.00264263e-01 4.46583331e-01
1.86045492e+00 -3.84823710e-01 2.88051385e-02 9.19026256e-01
1.41559756e+00 3.45772296e-01 5.31656146e-01 1.93881154e-01
4.43729758e-01 4.63282257e-01 5.29616952e-01 3.78586590e-01
-2.30101675e-01 5.05183578e-01 5.04958868e-01 -3.19982946e-01
-7.17415884e-02 1.96391687e-01 2.03691095e-01 9.15892422e-01
-3.38086151e-02 1.91491656e-02 -4.84772712e-01 9.59376395e-02
-1.56549072e+00 -8.70784819e-01 1.84011966e-01 2.13544106e+00
9.30061400e-01 2.14086711e-01 -2.74003267e-01 4.93524373e-01
6.29853547e-01 1.19202867e-01 -6.56890929e-01 -2.81877190e-01
-1.38283819e-01 3.07173848e-01 2.83396035e-01 2.34729320e-01
-1.17707050e+00 3.74129921e-01 5.80625725e+00 1.13503039e+00
-1.66186702e+00 -1.65760983e-02 1.03811145e+00 -1.44196758e-02
-1.51591608e-02 -6.48596466e-01 -6.36997640e-01 6.04480863e-01
4.60906476e-01 -9.79705825e-02 3.00591677e-01 8.09965730e-01
8.37065503e-02 1.18766516e-01 -1.03545892e+00 1.31197238e+00
1.33207664e-01 -1.16358244e+00 -5.00650518e-02 -5.60731515e-02
7.41370559e-01 -1.39060810e-01 4.88629818e-01 3.06045264e-01
-4.20278400e-01 -1.23462117e+00 4.18852210e-01 5.17735720e-01
7.52133787e-01 -9.31493342e-01 1.08953571e+00 4.16552275e-01
-8.64940763e-01 -3.59948635e-01 -6.13050044e-01 1.16039932e-01
-3.29850733e-01 7.41259813e-01 -7.77603149e-01 3.88018578e-01
6.21845245e-01 9.11018133e-01 -4.53754216e-01 1.10417008e+00
-5.01500396e-03 4.91148919e-01 -9.57283303e-02 -8.95707011e-02
2.15874374e-01 -1.11223243e-01 2.02076703e-01 9.27861691e-01
3.71529073e-01 -5.05136997e-02 1.75669804e-01 7.82432973e-01
-5.16541541e-01 3.21309566e-01 -2.89740205e-01 4.53872718e-02
-3.61581636e-03 1.24472511e+00 -5.27682662e-01 -1.55666083e-01
-4.15069491e-01 6.47850335e-01 2.78342992e-01 1.94527179e-01
-7.93596506e-01 -6.00168884e-01 4.14569646e-01 -1.43072248e-01
3.04010540e-01 1.47302091e-01 -4.35998559e-01 -9.16998923e-01
4.13005590e-01 -9.04282629e-01 4.68726039e-01 -3.03365320e-01
-1.30977225e+00 7.24884868e-01 -2.45610192e-01 -1.37699091e+00
4.48794812e-02 -8.93897116e-01 -5.35490036e-01 7.20106900e-01
-1.74725211e+00 -6.83293045e-01 -4.61413234e-01 6.86081707e-01
5.50773919e-01 -4.97623980e-01 4.76040840e-01 6.92027688e-01
-6.98319852e-01 9.20366049e-01 3.49454135e-01 1.92751080e-01
5.54591358e-01 -9.28565383e-01 -4.39928949e-01 5.38818061e-01
8.36830772e-03 1.14399165e-01 4.31657940e-01 -1.33508340e-01
-9.59064662e-01 -9.45428848e-01 5.88698924e-01 2.76864797e-01
2.20479622e-01 1.22647420e-01 -1.15115666e+00 1.38414845e-01
9.89230797e-02 3.89306933e-01 5.54810047e-01 -3.12559664e-01
-1.83172286e-01 -7.77607262e-01 -1.43563640e+00 -1.26920333e-02
6.41600251e-01 -3.48523289e-01 -2.64679581e-01 4.53971684e-01
3.77077818e-01 -1.97668567e-01 -9.67296779e-01 8.01422536e-01
5.62430739e-01 -7.23926187e-01 9.20420170e-01 -6.73358440e-01
6.85757637e-01 -1.04507945e-01 -1.01506226e-01 -1.27637446e+00
-2.66237825e-01 7.56516308e-02 -6.41205013e-02 9.62417245e-01
2.81769991e-01 -5.64703882e-01 8.50720644e-01 3.25452060e-01
-6.63059428e-02 -1.26275671e+00 -1.17852771e+00 -5.83187282e-01
3.79773080e-02 -1.96314201e-01 3.36150408e-01 9.50169802e-01
-2.70167232e-01 4.19410020e-02 -4.56487507e-01 -1.09339543e-01
6.35325670e-01 -9.80610251e-02 2.94613242e-01 -1.31626797e+00
-4.28639650e-01 -6.45908833e-01 -6.72628760e-01 -8.31323624e-01
2.89804935e-01 -1.09465289e+00 2.66071819e-02 -1.20074821e+00
2.19957799e-01 -6.44531667e-01 -7.46781886e-01 5.23738265e-01
-9.78302583e-02 2.54027098e-01 -5.08211367e-02 1.41707212e-01
-3.46958697e-01 7.51512945e-01 1.47626066e+00 -5.36931753e-01
1.97973400e-02 4.32578593e-01 -6.83250785e-01 6.94467545e-01
7.24232674e-01 -5.24942935e-01 -4.56293821e-01 -4.93119925e-01
3.55253220e-02 -5.86475357e-02 1.31876364e-01 -1.06105399e+00
3.55306044e-02 -1.19279690e-01 6.41139567e-01 -2.52285659e-01
1.85669601e-01 -1.03607321e+00 -2.40671471e-01 7.10523129e-01
-6.73126161e-01 -2.38942996e-01 2.59830542e-02 3.70470524e-01
-5.44780910e-01 -5.06251693e-01 1.15828693e+00 -8.80298298e-03
-6.27853870e-01 6.39156103e-01 3.59660201e-02 -9.38286334e-02
9.73636448e-01 -1.85714811e-01 5.18133119e-02 -2.35637218e-01
-6.58511281e-01 2.85597518e-02 -1.93037301e-01 4.26737398e-01
1.07919705e+00 -1.60929120e+00 -8.41705978e-01 6.48319840e-01
-9.79084074e-02 2.91739125e-02 7.17911124e-02 9.50413227e-01
-4.71386075e-01 1.83065683e-01 -4.51768041e-01 -6.18351638e-01
-1.45060599e+00 3.18006098e-01 6.12784803e-01 -2.31707662e-01
-2.87169248e-01 8.92793298e-01 -3.17784436e-02 -2.46157736e-01
7.80438423e-01 -4.99583989e-01 -2.66699791e-01 -6.94862679e-02
6.72503829e-01 3.26955199e-01 3.59167814e-01 -4.70225602e-01
-2.68915385e-01 5.13840258e-01 -1.28251299e-01 4.30624068e-01
1.45046091e+00 1.61261439e-01 -1.32789657e-01 2.15287939e-01
1.88635516e+00 -5.57411671e-01 -1.15342951e+00 -5.81193924e-01
-2.70515233e-02 -6.68594003e-01 3.32971156e-01 -6.31602585e-01
-1.74836802e+00 9.31430638e-01 1.09942734e+00 1.66014940e-01
1.39036298e+00 -1.69680610e-01 8.50980997e-01 7.06855357e-01
3.14140841e-02 -1.14885437e+00 3.58953089e-01 1.51561916e-01
8.97388399e-01 -1.62113094e+00 -1.16200168e-02 6.15926571e-02
-2.06693962e-01 1.42820072e+00 6.70531034e-01 -4.45134006e-02
9.05077696e-01 2.46815979e-01 5.40907085e-02 8.13527033e-02
-4.51555312e-01 1.93066627e-01 4.60107058e-01 3.59538406e-01
4.00240839e-01 -7.63585716e-02 -6.00990772e-01 4.99747366e-01
1.70432121e-01 2.07623839e-01 1.05882592e-01 7.40684509e-01
-5.89644730e-01 -1.13342369e+00 -3.92223388e-01 6.48266375e-01
-6.10511899e-01 1.55829683e-01 2.74458081e-02 5.29039264e-01
3.44175756e-01 6.95864141e-01 3.75931151e-02 -4.43832636e-01
2.31188968e-01 -1.74863711e-01 4.98645157e-01 -2.92709857e-01
-2.91052938e-01 -1.92584828e-01 -3.62735748e-01 -2.79371858e-01
-4.43135321e-01 -2.47121871e-01 -1.26440930e+00 -7.24430084e-02
-4.79368627e-01 2.87481379e-02 7.95056522e-01 9.91106451e-01
-1.06334664e-01 4.99168634e-01 1.13442612e+00 -4.86834168e-01
-1.08412051e+00 -8.86867702e-01 -7.71946073e-01 6.34531915e-01
5.42836726e-01 -6.42168760e-01 -4.23344940e-01 -1.14204161e-01] | [9.431173324584961, 2.3407845497131348] |
59cba718-a0fd-4c65-82f8-f441eb067a06 | mwp-bert-numeracy-augmented-pre-training-for | null | null | https://openreview.net/forum?id=1c5HGUkc1SL | https://openreview.net/pdf?id=1c5HGUkc1SL | MWP-BERT: Numeracy-Augmented Pre-training for Math Word Problem Solving | Math word problem (MWP) solving faces a dilemma in number representation learning. In order to avoid the number representation issue and reduce the search space of feasible solutions, existing works striving for MWP solving usually replace real numbers with symbolic placeholders to focus on logic reasoning. However, different from common symbolic reasoning tasks like program synthesis and knowledge graph reasoning, MWP solving has extra requirements in numerical reasoning. In other words, instead of the number value itself, it is the reusable numerical property that matters more in numerical reasoning. Therefore, we argue that injecting numerical properties into symbolic placeholders with contextualized representation learning schema can provide a way out of the dilemma in the number representation issue here. In this work, we introduce this idea to the popular pre-training language model (PLM) techniques and build MWP-BERT, an effective contextual number representation PLM. We demonstrate the effectiveness of our MWP-BERT on MWP solving and several MWP-specific understanding tasks on both English and Chinese benchmarks. | ['Anonymous'] | 2022-01-16 | null | null | null | acl-arr-january-2022-1 | ['math-word-problem-solving', 'math-word-problem-solving', 'math-word-problem-solving'] | ['knowledge-base', 'reasoning', 'time-series'] | [ 1.06010400e-01 3.03437978e-01 -5.13071954e-01 -1.96091518e-01
-4.20575917e-01 -6.53697670e-01 2.47127891e-01 3.84026140e-01
-3.28238726e-01 7.23655224e-01 2.25597173e-01 -9.45433557e-01
-1.74469918e-01 -1.58177495e+00 -9.00254250e-01 -1.01307549e-01
1.53146505e-01 3.92102122e-01 9.71403942e-02 -5.06534696e-01
5.80011368e-01 4.52351272e-01 -1.19829357e+00 5.91466606e-01
9.27383959e-01 6.40105605e-01 -1.17634840e-01 3.73980165e-01
-9.67482984e-01 1.38137567e+00 -7.03808904e-01 -6.15176022e-01
2.15318441e-01 -1.27428532e-01 -1.12480295e+00 -7.13611424e-01
1.84575081e-01 -3.57356071e-01 -1.29491270e-01 9.91482913e-01
1.07059656e-02 2.51858085e-01 6.79779351e-01 -1.63170242e+00
-7.25460529e-01 1.56878603e+00 -5.31768918e-01 -1.88983493e-02
5.69244981e-01 2.83644527e-01 1.29820859e+00 -4.49334353e-01
5.91177940e-01 1.62203610e+00 7.80214608e-01 5.70604861e-01
-1.06819844e+00 -5.83425701e-01 1.49203151e-01 4.14948106e-01
-1.33422089e+00 6.92280829e-02 8.53088140e-01 -2.01934904e-01
1.40083778e+00 4.43553954e-01 9.17802036e-01 5.05072117e-01
-5.39059117e-02 7.18167245e-01 8.47533882e-01 -8.19434643e-01
1.76259894e-02 5.68643510e-02 5.55384040e-01 7.97401607e-01
3.14732850e-01 -3.03069413e-01 -3.52104545e-01 1.89154577e-02
9.99786139e-01 -2.72477925e-01 -1.40194982e-01 -2.63371080e-01
-1.25390184e+00 1.05702341e+00 3.81287545e-01 3.21217299e-01
3.11077714e-01 8.00250292e-01 5.30950725e-01 4.58003610e-01
-2.37472206e-01 9.60310042e-01 -6.50564194e-01 -3.94550979e-01
-7.81283498e-01 4.88926649e-01 1.05389333e+00 8.83935034e-01
6.88630164e-01 1.04573570e-01 -2.00167492e-01 4.39850658e-01
2.36595318e-01 2.99399406e-01 2.74814576e-01 -1.07158029e+00
9.48464453e-01 1.00993145e+00 -5.57652153e-02 -1.23472238e+00
-4.19315845e-01 5.63264117e-02 -4.73297477e-01 5.88039458e-02
8.06884408e-01 1.40061662e-01 -3.51947248e-01 1.85229862e+00
2.32083648e-02 2.42408022e-01 2.36846671e-01 5.56001961e-01
8.86018395e-01 1.13381350e+00 1.63162962e-01 -6.04663938e-02
1.33933103e+00 -9.96307671e-01 -6.12344265e-01 -1.17248133e-01
1.32777619e+00 -1.58420950e-01 1.50478041e+00 2.94107884e-01
-1.13846469e+00 -2.39297107e-01 -1.18732166e+00 -4.70231444e-01
-7.35757291e-01 -8.61100554e-02 1.30286324e+00 7.78500021e-01
-7.71092415e-01 6.75491095e-01 -3.83799851e-01 1.67486250e-01
3.93741518e-01 3.30829680e-01 -6.32115304e-02 -1.87277630e-01
-1.56800091e+00 1.14831853e+00 7.61217058e-01 1.21242472e-03
-1.48251399e-01 -1.15765882e+00 -1.15440392e+00 4.46439952e-01
8.05394292e-01 -6.43388867e-01 1.12764621e+00 -8.13758969e-01
-1.56429446e+00 5.69146991e-01 1.30245676e-02 -7.36743331e-01
3.98068994e-01 -6.24880344e-02 -7.66111016e-02 -1.11377567e-01
-2.71259546e-01 7.27585495e-01 4.78658050e-01 -8.06325793e-01
-7.03281686e-02 6.02575168e-02 6.72288835e-01 2.90030781e-02
-3.65819745e-02 8.53480548e-02 -2.96556894e-02 -6.87573791e-01
7.58035779e-02 -4.25451338e-01 -2.42698997e-01 -1.12876808e-02
-3.13432664e-02 -7.83960283e-01 2.74273634e-01 -3.83484334e-01
1.54582012e+00 -2.00545692e+00 3.35845441e-01 5.44290960e-01
3.33492517e-01 2.34099433e-01 -1.82858050e-01 3.80993426e-01
-2.33016685e-01 5.12134075e-01 4.66630012e-02 4.18206662e-01
4.63920325e-01 3.61721307e-01 -7.04631925e-01 1.16194105e-02
4.56781268e-01 1.39216089e+00 -9.63332474e-01 -7.30713487e-01
1.73976868e-01 -1.24056116e-01 -9.51327384e-01 -2.75223583e-01
-9.52750921e-01 -2.16692701e-01 -3.08457971e-01 5.72611690e-01
8.71206284e-01 -3.02721858e-01 4.50790614e-01 -4.62145768e-02
1.12714566e-01 4.94837761e-01 -1.52993286e+00 1.71637249e+00
-6.78462386e-01 3.29434514e-01 -5.18354237e-01 -1.02718258e+00
7.12600768e-01 -1.47351965e-01 -3.81388180e-02 -6.88372076e-01
1.15183592e-01 2.69718498e-01 1.80581957e-01 -2.71089435e-01
8.27048302e-01 -1.66910455e-01 -3.44738185e-01 4.64715868e-01
-1.26624167e-01 -6.48230731e-01 3.58767778e-01 3.84048015e-01
8.65708232e-01 5.49726903e-01 6.53981566e-01 1.37357293e-02
8.21899414e-01 2.19623864e-01 5.40559173e-01 7.09686935e-01
1.88139275e-01 1.57919191e-02 1.21403635e+00 -5.44775486e-01
-7.73883641e-01 -8.85434568e-01 1.71941489e-01 9.83482480e-01
-1.21143751e-01 -9.93848562e-01 -5.47400713e-01 -4.78674382e-01
-1.30924480e-02 1.09067214e+00 -2.13554919e-01 -2.80739814e-01
-1.20248997e+00 -2.12633342e-01 9.84311461e-01 9.06334877e-01
4.57826674e-01 -1.23250175e+00 -6.68551505e-01 3.22418511e-01
-2.88840830e-01 -9.87526357e-01 2.75854040e-02 1.93601251e-01
-8.46442103e-01 -9.15642142e-01 -2.22175777e-01 -6.87564015e-01
5.25839448e-01 -1.14696741e-01 1.40430999e+00 5.02345264e-01
-1.21723503e-01 1.16670050e-01 -2.69496590e-01 -3.79044384e-01
-4.97034281e-01 -7.56448135e-02 -4.20124322e-01 -8.38926733e-01
4.99109119e-01 -4.48602229e-01 1.13582378e-02 -3.54726285e-01
-7.15124726e-01 3.60597342e-01 2.14419410e-01 5.78952312e-01
1.77442297e-01 4.28805023e-01 4.47602391e-01 -8.15516531e-01
5.50933123e-01 -1.81906477e-01 -8.73802304e-01 5.97380757e-01
-1.37414500e-01 4.57173258e-01 9.53375518e-01 -5.07804215e-01
-6.11013412e-01 -2.92828381e-01 7.56851137e-02 -1.89911515e-01
4.62084651e-01 7.52312839e-01 -3.84793907e-01 3.34125347e-02
6.11359179e-01 1.36879891e-01 -1.20526493e-01 2.49834478e-01
8.65105152e-01 -1.48336425e-01 2.62679815e-01 -1.52441692e+00
8.78189147e-01 -3.81081775e-02 3.79809946e-01 -2.53021419e-01
-6.88623667e-01 -1.71366781e-02 -3.51292014e-01 1.66528270e-01
5.46439111e-01 -6.97801888e-01 -1.18468761e+00 1.87742174e-01
-1.40228999e+00 -5.44520915e-01 -5.82120240e-01 1.50326803e-01
-7.37962365e-01 4.21577096e-01 -5.81358016e-01 -9.87633884e-01
-5.61743937e-02 -1.15364647e+00 5.58813930e-01 1.05566129e-01
-6.53556645e-01 -8.44727874e-01 -2.95498759e-01 1.61021203e-01
3.24209809e-01 1.82654738e-01 2.02596831e+00 -5.26548982e-01
-7.55111337e-01 -2.22819485e-03 -5.62394023e-01 2.67336369e-01
-1.97721303e-01 -1.27682939e-01 -5.26055157e-01 4.25786674e-01
-9.09463838e-02 -4.14106727e-01 6.71616435e-01 8.48579332e-02
1.52612662e+00 -6.89627945e-01 6.94027320e-02 5.06678402e-01
1.62951982e+00 -1.31762758e-01 1.13817084e+00 5.19591570e-01
5.51183641e-01 4.36612159e-01 4.48967397e-01 3.67688835e-01
5.36587834e-01 3.57351691e-01 1.32279009e-01 3.88967335e-01
-1.15401611e-01 -5.23868561e-01 2.30157420e-01 3.47266197e-01
-1.84462503e-01 6.96613491e-02 -1.11688340e+00 2.44928166e-01
-1.80262625e+00 -9.73338664e-01 2.45823320e-02 1.92812788e+00
1.50948620e+00 1.40593052e-01 4.20711674e-02 5.38066566e-01
1.79908499e-01 2.49451578e-01 -3.30126584e-02 -8.39809835e-01
2.24571619e-02 7.54030168e-01 3.51607621e-01 5.72464406e-01
-6.31362915e-01 1.22146142e+00 5.09902239e+00 1.14017403e+00
-8.55391324e-01 -3.55397373e-01 3.03642988e-01 1.63435161e-01
-6.24195039e-01 7.56132230e-02 -7.22781718e-01 -1.22626463e-03
7.16422021e-01 -2.76668787e-01 8.29631507e-01 1.04752469e+00
-3.76024991e-01 -4.50350270e-02 -1.53036547e+00 1.16892862e+00
-2.35710964e-01 -1.65711522e+00 4.16823596e-01 -5.55383325e-01
2.46165171e-01 -9.10541654e-01 6.89003393e-02 8.23338747e-01
3.51365358e-01 -1.72117591e+00 8.77519250e-01 4.36238050e-01
5.47391236e-01 -9.33718324e-01 5.85692823e-01 2.95945108e-01
-1.18304336e+00 -2.65373170e-01 -5.47178984e-01 -6.30449831e-01
-1.85398757e-01 2.35790357e-01 -6.55859828e-01 6.27011478e-01
1.67215481e-01 5.97821414e-01 -4.97787207e-01 6.57882631e-01
-5.91873229e-01 1.66084379e-01 -2.81037629e-01 -4.55909163e-01
3.82513732e-01 -2.57849485e-01 1.31413639e-01 1.02414382e+00
2.01459303e-01 4.02820528e-01 -1.03089467e-01 1.42694998e+00
-1.55772597e-01 1.79502159e-01 -4.86393929e-01 -4.12200481e-01
5.07905960e-01 8.19276035e-01 -8.68868947e-01 -5.15866220e-01
-5.22963226e-01 3.00567210e-01 5.38385153e-01 2.26243392e-01
-1.20210886e+00 -4.25145447e-01 2.68593401e-01 9.40291137e-02
1.98899865e-01 -5.10761380e-01 -7.03467846e-01 -1.15950298e+00
1.32876024e-01 -1.25781906e+00 3.00952941e-01 -7.62835979e-01
-8.77278388e-01 -1.38772383e-01 4.19887424e-01 -8.69469106e-01
-1.17883638e-01 -9.98317957e-01 -6.12604022e-01 9.25030649e-01
-1.64597201e+00 -1.00392735e+00 5.87456487e-02 3.77915472e-01
2.46601030e-01 8.31412449e-02 1.00501132e+00 -1.81696042e-01
-2.17697442e-01 7.08974540e-01 -6.06824458e-01 2.74401098e-01
1.40638486e-01 -1.14918065e+00 1.94715038e-01 5.84169388e-01
7.66836181e-02 1.19321597e+00 5.03961384e-01 -3.61828834e-01
-1.93379080e+00 -6.61469519e-01 9.87438023e-01 -5.25746167e-01
1.02487886e+00 -2.49033853e-01 -9.06965137e-01 6.84248030e-01
-7.07579181e-02 -2.68062800e-01 5.27464867e-01 1.63945243e-01
-1.00911748e+00 9.78168100e-02 -1.14118648e+00 9.96219814e-01
9.26728427e-01 -4.07192349e-01 -1.10437334e+00 3.06767315e-01
1.20007622e+00 -6.67870402e-01 -6.41012669e-01 3.14510137e-01
4.39389378e-01 -4.63120818e-01 1.36043608e+00 -7.94404507e-01
8.60538483e-01 -4.33009088e-01 -3.35683316e-01 -9.69176054e-01
1.23054748e-02 -5.29421449e-01 -4.73436832e-01 1.18830585e+00
3.59185010e-01 -7.14289904e-01 7.48951077e-01 8.65601897e-01
1.02210499e-01 -7.92841613e-01 -8.71290803e-01 -7.58096695e-01
8.05653274e-01 -9.09542799e-01 1.16040957e+00 9.38093543e-01
4.59455311e-01 1.57910243e-01 1.36634838e-02 -1.46611154e-01
3.92261222e-02 4.16882932e-01 7.38144577e-01 -9.51270401e-01
-5.54343760e-01 -7.33039141e-01 -2.49167874e-01 -8.39427471e-01
6.64798260e-01 -1.33187091e+00 -4.80773330e-01 -1.62113023e+00
-1.44083455e-01 -6.24419749e-01 -1.34840906e-01 8.42404246e-01
2.52553653e-02 -3.29713911e-01 4.85284358e-01 -2.91757971e-01
-3.94859254e-01 1.61148846e-01 1.35817695e+00 -3.72419119e-01
-1.75427020e-01 -3.88052970e-01 -7.40776718e-01 8.54314864e-01
8.60720098e-01 -2.88845628e-01 -5.46493888e-01 -4.05228615e-01
1.20729423e+00 2.91164637e-01 5.49161375e-01 -8.74763966e-01
2.15736628e-01 -6.77092016e-01 9.99340862e-02 -5.38069487e-01
1.01665586e-01 -8.73397410e-01 -3.49464983e-01 6.45166159e-01
-3.66405308e-01 1.36591658e-01 5.04529417e-01 4.56648245e-02
-5.20851798e-02 -7.52952039e-01 3.77136081e-01 -5.97604752e-01
-1.01286936e+00 -3.54775250e-01 -2.10596710e-01 3.45310330e-01
7.19176233e-01 -8.39027241e-02 -6.32130563e-01 -7.21465796e-02
-3.32412094e-01 2.86922038e-01 2.69906104e-01 2.37980913e-02
7.40320861e-01 -1.20994759e+00 -2.40585104e-01 -2.24786382e-02
-3.75412367e-02 2.31075197e-01 8.08902904e-02 6.38566911e-01
-9.60517049e-01 6.34437621e-01 -1.55408069e-01 9.73547101e-02
-9.30342376e-01 9.01700795e-01 3.04819167e-01 -7.76098311e-01
-7.50115156e-01 8.56982112e-01 -2.23381072e-01 -6.59867465e-01
4.23541099e-01 -1.29135776e+00 -2.63885856e-01 -4.81512956e-02
5.61304390e-01 2.99189746e-01 -1.27298310e-01 2.33329386e-01
-3.33609462e-01 5.55425644e-01 1.45262942e-01 1.04821794e-01
1.22976553e+00 4.59927827e-01 -7.28807271e-01 4.36199069e-01
7.45340705e-01 2.17032433e-01 -4.34360921e-01 -1.97622240e-01
2.97108322e-01 -3.47658753e-01 -5.73625505e-01 -7.74669886e-01
-6.26831412e-01 7.85732090e-01 -2.58243144e-01 -1.25405416e-01
8.67719233e-01 -3.31401303e-02 7.53312886e-01 9.56076860e-01
6.85604870e-01 -9.19571459e-01 1.10560179e-01 9.08136010e-01
7.87693799e-01 -8.68493378e-01 2.38627851e-01 -7.23061919e-01
-4.88788426e-01 1.51226330e+00 6.68204129e-01 -2.43485644e-01
2.17772782e-01 5.41493535e-01 -3.22530389e-01 3.95965427e-02
-6.54114008e-01 1.76965460e-01 -1.97063331e-02 5.05058348e-01
5.81126451e-01 3.94420177e-02 -3.59113902e-01 1.09764051e+00
-7.81148612e-01 3.02961469e-01 7.11342633e-01 9.51028645e-01
-3.00716698e-01 -1.29975259e+00 -4.53766763e-01 2.47259170e-01
-1.66429847e-01 -7.10028172e-01 -1.31872352e-02 1.11391366e+00
8.77914056e-02 4.18116719e-01 -8.91405642e-02 8.28172266e-02
7.73574645e-03 4.78219599e-01 1.10363114e+00 -7.29080737e-01
-6.58837020e-01 -6.83933437e-01 2.80449480e-01 -5.72388411e-01
-2.24328130e-01 -1.20226607e-01 -1.93800402e+00 -6.52487278e-01
-3.62235568e-02 2.25726187e-01 2.07566723e-01 9.43917990e-01
-2.83542514e-01 8.38342309e-01 -7.43476972e-02 -1.28553286e-01
-8.27853024e-01 -4.56880927e-01 -2.93300569e-01 1.47245759e-02
-2.23783888e-02 -4.08521295e-01 -1.92044258e-01 -3.94284070e-01] | [9.503569602966309, 7.4110107421875] |
5e0e1452-2682-4752-8fdf-f4d016641ed8 | context-aware-selective-label-smoothing-for | 2303.06946 | null | https://arxiv.org/abs/2303.06946v1 | https://arxiv.org/pdf/2303.06946v1.pdf | Context-Aware Selective Label Smoothing for Calibrating Sequence Recognition Model | Despite the success of deep neural network (DNN) on sequential data (i.e., scene text and speech) recognition, it suffers from the over-confidence problem mainly due to overfitting in training with the cross-entropy loss, which may make the decision-making less reliable. Confidence calibration has been recently proposed as one effective solution to this problem. Nevertheless, the majority of existing confidence calibration methods aims at non-sequential data, which is limited if directly applied to sequential data since the intrinsic contextual dependency in sequences or the class-specific statistical prior is seldom exploited. To the end, we propose a Context-Aware Selective Label Smoothing (CASLS) method for calibrating sequential data. The proposed CASLS fully leverages the contextual dependency in sequences to construct confusion matrices of contextual prediction statistics over different classes. Class-specific error rates are then used to adjust the weights of smoothing strength in order to achieve adaptive calibration. Experimental results on sequence recognition tasks, including scene text recognition and speech recognition, demonstrate that our method can achieve the state-of-the-art performance. | ['Yongpan Wang', 'Mengchao He', 'Jin-Gang Yu', 'Zhenzhou Zhuang', 'Yu Luo', 'Shuangping Huang'] | 2023-03-13 | null | null | null | null | ['scene-text-recognition'] | ['computer-vision'] | [ 6.27855837e-01 -4.78015661e-01 -1.95311487e-01 -9.27717865e-01
-1.00839150e+00 -2.80248135e-01 6.82380557e-01 1.37461051e-01
-5.80458760e-01 5.73953152e-01 9.61281583e-02 -4.41206902e-01
7.70088881e-02 -3.05318713e-01 -6.04308188e-01 -9.30777371e-01
4.69895601e-01 1.88585173e-03 3.18686068e-01 3.11302632e-01
4.20998901e-01 2.49766186e-01 -1.27156270e+00 9.90178734e-02
1.32581985e+00 1.28287661e+00 5.55306852e-01 6.01677179e-01
-2.60435373e-01 9.19199884e-01 -5.87068856e-01 -4.23902929e-01
-1.59732223e-01 -4.37779069e-01 -3.34634125e-01 2.68292487e-01
2.72720009e-01 -2.95023113e-01 -3.63377124e-01 1.43497336e+00
3.99463922e-01 3.65902930e-01 7.58708358e-01 -9.45372462e-01
-4.27788854e-01 6.55663490e-01 -3.71432811e-01 1.69631422e-01
-1.54501721e-01 -1.26607241e-02 8.23564529e-01 -1.11559093e+00
-3.20272855e-02 1.15222847e+00 6.44132435e-01 4.30840194e-01
-1.04319358e+00 -6.71747208e-01 4.79303390e-01 4.48176771e-01
-1.35402858e+00 -5.99292338e-01 6.42162740e-01 -3.80358279e-01
7.09471881e-01 -4.98753935e-02 1.28628150e-01 1.42437053e+00
-8.74215662e-02 1.06663120e+00 9.57354903e-01 -5.34517407e-01
3.66816789e-01 6.98198229e-02 1.36627316e-01 4.03397650e-01
-6.45039603e-02 6.29756898e-02 -6.02486670e-01 1.65550664e-01
5.82202911e-01 -1.57193225e-02 -3.58215153e-01 -2.85144925e-01
-1.05974102e+00 6.33277953e-01 2.39953443e-01 1.79325685e-01
-7.32471198e-02 -1.07221641e-01 5.53827703e-01 2.61346791e-02
5.08828342e-01 3.56113277e-02 -5.17392397e-01 -3.55324864e-01
-9.80950177e-01 -3.28728378e-01 4.45802063e-01 9.58265722e-01
5.16029537e-01 4.16141123e-01 -3.69904220e-01 1.27705920e+00
2.44004592e-01 4.45618570e-01 6.77927315e-01 -4.50437725e-01
8.64575326e-01 4.03024971e-01 -8.77677798e-02 -8.03687751e-01
-2.33835399e-01 -5.86357832e-01 -1.06897283e+00 -1.03754744e-01
3.61179948e-01 -8.07245448e-02 -1.13766265e+00 1.89833057e+00
1.61027834e-01 5.64599454e-01 1.77816883e-01 8.11666489e-01
4.42835808e-01 5.81753850e-01 2.61660397e-01 -2.70803839e-01
9.86489892e-01 -1.07106471e+00 -8.85428071e-01 -4.86245483e-01
5.77045441e-01 -6.52441025e-01 1.19496930e+00 2.61518627e-01
-3.85625541e-01 -7.23315299e-01 -1.24712205e+00 1.61459655e-01
1.18703663e-03 4.17180240e-01 2.75696456e-01 7.39031732e-01
-4.30297315e-01 5.76649487e-01 -1.00794137e+00 -1.73164546e-01
3.66372138e-01 1.03467152e-01 -4.74929884e-02 -1.38214946e-01
-1.29016888e+00 8.02537084e-01 7.13363409e-01 3.29807520e-01
-8.28155935e-01 -5.20335376e-01 -7.97079027e-01 1.64511338e-01
7.92808712e-01 -1.32851511e-01 1.42842817e+00 -1.02788758e+00
-1.75547314e+00 2.62434006e-01 -1.27967492e-01 -6.05591953e-01
6.19162202e-01 -3.48403305e-01 -6.56819642e-01 -1.11203149e-01
-1.47838175e-01 3.01772505e-01 9.19698358e-01 -1.09702003e+00
-8.84245276e-01 -1.37444198e-01 -2.83751696e-01 3.34536344e-01
-5.39426029e-01 -1.20131537e-01 -7.07263827e-01 -9.09635365e-01
1.83269620e-01 -7.15921700e-01 -1.95626110e-01 -2.41499394e-01
-4.54235315e-01 -1.03793949e-01 7.72502959e-01 -7.74915457e-01
1.45430529e+00 -2.26042175e+00 2.03918628e-02 1.40190095e-01
-2.56844848e-01 4.50430989e-01 -1.05972797e-01 1.66864470e-01
7.94208273e-02 -1.12502657e-01 -4.97719198e-01 -5.59409857e-01
-3.11353914e-02 2.76748240e-01 -3.81141007e-01 3.60807270e-01
5.11353314e-01 4.35417861e-01 -8.21348310e-01 -5.21755457e-01
5.42740762e-01 6.22334659e-01 -3.08022648e-01 3.63671869e-01
-2.73357213e-01 4.69110817e-01 -2.75938213e-01 5.62507033e-01
6.82375252e-01 -2.84379363e-01 3.19080591e-01 -2.41819263e-01
6.21029362e-03 4.07558501e-01 -1.09816778e+00 1.49620044e+00
-4.99991834e-01 6.25412345e-01 -2.74989992e-01 -1.20337498e+00
1.06257057e+00 1.69207007e-01 1.30648419e-01 -5.76309085e-01
5.43696992e-02 1.99485809e-01 -8.78607854e-02 -3.68488044e-01
5.19892573e-01 8.65954533e-02 9.30991173e-02 -5.83395325e-02
9.66532622e-03 4.93756980e-02 -3.60622592e-02 1.22435847e-02
7.45423317e-01 2.10138768e-01 2.33636931e-01 -7.32829124e-02
7.32232571e-01 -5.06624758e-01 8.85143936e-01 8.35660338e-01
-3.56921583e-01 6.77609801e-01 3.05438548e-01 -2.62026545e-02
-1.21008766e+00 -8.14761996e-01 -2.92569011e-01 8.61227393e-01
1.74453706e-01 -3.75122875e-01 -9.53000128e-01 -7.48753488e-01
-2.23533392e-01 9.58872318e-01 -4.56297815e-01 -4.10740674e-01
-3.54417831e-01 -6.54149532e-01 4.12837744e-01 8.32749307e-01
8.19860101e-01 -7.25473940e-01 -1.95791647e-01 3.24048191e-01
-3.30274612e-01 -1.50323081e+00 -7.05094516e-01 4.10654992e-01
-8.55365098e-01 -8.46205592e-01 -6.21313810e-01 -5.21583080e-01
5.38543165e-01 1.51082903e-01 5.92338443e-01 -8.94922316e-02
2.98591584e-01 1.74667701e-01 -4.16153312e-01 -1.69251487e-01
-5.21349967e-01 5.55430911e-02 2.19168723e-01 3.04131001e-01
5.34859717e-01 -2.41170228e-01 -5.09397089e-01 4.30894226e-01
-9.56481159e-01 1.48034677e-01 6.68130696e-01 1.27748036e+00
5.49172401e-01 5.71701005e-02 6.07298851e-01 -6.30317748e-01
4.96518105e-01 -2.52645940e-01 -7.40864158e-01 6.08298123e-01
-8.34336817e-01 2.09580734e-01 6.97526813e-01 -7.26487994e-01
-1.32665682e+00 -1.59487082e-03 -2.87471235e-01 -4.77349609e-01
-1.35962069e-01 7.95581162e-01 -2.67338902e-01 1.03246346e-01
4.04326200e-01 5.03821969e-01 -2.45112404e-01 -4.01554972e-01
-6.45494387e-02 9.43720102e-01 6.52771115e-01 -4.75018561e-01
2.06683293e-01 1.48026496e-02 -1.78706899e-01 -9.02115226e-01
-1.16095448e+00 -4.05780733e-01 -5.64555883e-01 -2.50452727e-01
6.46748424e-01 -8.50629151e-01 -4.31633174e-01 8.03017318e-01
-1.07383049e+00 -4.08404738e-01 1.09914333e-01 6.83208466e-01
-3.36887717e-01 8.55975270e-01 -3.94586354e-01 -1.25461555e+00
-2.19094679e-01 -1.28148174e+00 8.72394860e-01 2.47338042e-01
1.03316016e-01 -8.30967724e-01 -2.42833808e-01 2.41586000e-01
3.53864133e-01 -1.99894920e-01 7.05572546e-01 -7.58045793e-01
-4.62584645e-01 -2.68604577e-01 -3.94860119e-01 8.25400949e-01
1.30193338e-01 1.57805979e-01 -1.07549584e+00 -1.81803524e-01
7.48013109e-02 -4.13735509e-01 9.93752420e-01 3.94939989e-01
1.61304462e+00 -1.10967748e-01 -1.37198240e-01 5.62172472e-01
1.16093206e+00 4.44556922e-01 6.59743428e-01 2.42613494e-01
6.16829395e-01 4.19908017e-01 8.92232239e-01 6.50582790e-01
1.80049151e-01 8.18642974e-01 1.91963837e-01 3.97471249e-01
-3.82420272e-02 -3.23291600e-01 2.94759870e-01 1.06402969e+00
3.05310637e-01 -4.74275351e-01 -9.51573253e-01 2.53288418e-01
-1.97870016e+00 -7.60609865e-01 6.98712766e-02 2.37452984e+00
1.07117379e+00 3.42788458e-01 -3.26689780e-01 2.49636948e-01
1.03010023e+00 1.16517596e-01 -7.72938609e-01 1.45632952e-01
-3.10352772e-01 -2.34058216e-01 4.29793000e-01 4.47173178e-01
-1.31321096e+00 9.02805090e-01 5.45886993e+00 1.34193230e+00
-1.13755000e+00 -6.10399731e-02 8.12845349e-01 3.39194924e-01
-3.03533636e-02 -2.60823201e-02 -1.02423644e+00 6.93512499e-01
1.11905313e+00 2.05244049e-01 2.26945266e-01 8.72934282e-01
5.81864901e-02 -1.76979899e-01 -9.27169800e-01 1.13859379e+00
-3.56212221e-02 -1.02541566e+00 -1.52607962e-01 -4.53763425e-01
7.76758611e-01 -2.53284723e-02 1.73223764e-01 4.37336743e-01
2.92146593e-01 -7.30863810e-01 8.92995059e-01 5.38901925e-01
9.53557253e-01 -8.47137392e-01 7.35610008e-01 4.63196844e-01
-1.03067982e+00 -2.03162655e-01 -4.32787061e-01 4.05467838e-01
1.60218298e-01 8.10342014e-01 -9.38321471e-01 3.28520954e-01
5.99357426e-01 7.21232891e-01 -3.41353327e-01 9.67072904e-01
-3.31276894e-01 9.70130980e-01 -3.12923610e-01 -2.65163988e-01
2.55062670e-01 -2.68581361e-01 2.94068962e-01 1.53844547e+00
3.77959520e-01 -6.17495179e-03 4.84156832e-02 5.65320432e-01
-4.39378917e-02 3.17311324e-02 -4.14427407e-02 -2.28093818e-01
7.10413098e-01 8.53932023e-01 -6.11206472e-01 -4.70043242e-01
-4.24400657e-01 9.58651245e-01 4.89342332e-01 5.60954034e-01
-8.43581915e-01 -3.84475440e-01 5.52445948e-01 -5.16620219e-01
5.58388829e-01 -4.85509306e-01 -5.25741518e-01 -1.36652172e+00
1.01397686e-01 -8.26460063e-01 2.98838228e-01 -5.87750316e-01
-1.36855412e+00 5.79548895e-01 -2.32626319e-01 -1.32776046e+00
-1.86423451e-01 -6.14237607e-01 -3.01658899e-01 7.34595299e-01
-1.44911587e+00 -8.66330802e-01 -3.31789821e-01 4.09519374e-01
8.86109710e-01 -3.13879907e-01 6.06965363e-01 3.47502470e-01
-8.94726396e-01 9.75132048e-01 5.46503603e-01 1.41404018e-01
9.10399973e-01 -9.11538541e-01 4.37518060e-01 1.22995305e+00
3.40320636e-03 4.92858052e-01 6.22474730e-01 -8.24431598e-01
-9.78781223e-01 -1.10230112e+00 6.94886208e-01 -2.24295005e-01
7.72464275e-01 -4.65229899e-01 -1.16822541e+00 2.83192992e-01
-1.83861285e-01 -4.14158441e-02 7.06923544e-01 1.71673764e-02
-6.25413895e-01 -2.60415494e-01 -8.22390974e-01 8.35613251e-01
1.01503062e+00 -8.11159849e-01 -2.54937828e-01 1.33677840e-01
8.50484073e-01 -5.43279290e-01 -6.86396480e-01 6.48942649e-01
5.05270600e-01 -8.73790979e-01 7.29339480e-01 -3.69615525e-01
3.71677279e-01 -2.84475744e-01 -5.30421853e-01 -1.23785174e+00
-1.38506576e-01 -2.34366730e-01 -1.27387896e-01 1.42582750e+00
4.06019330e-01 -4.92840111e-01 6.21850550e-01 7.99560070e-01
-4.51426148e-01 -5.07588208e-01 -1.15472174e+00 -7.95628667e-01
-2.61306614e-01 -9.53688920e-01 4.99357313e-01 8.47597182e-01
-3.26280296e-01 2.85289168e-01 -6.54029727e-01 2.98736602e-01
6.57923877e-01 -2.00122222e-01 5.86018622e-01 -9.46471334e-01
-3.70936841e-01 -3.20758492e-01 -2.85049051e-01 -1.44900858e+00
1.89494759e-01 -5.09840786e-01 5.71630180e-01 -1.06969249e+00
1.18968651e-01 -5.49076974e-01 -4.79110032e-01 2.49798089e-01
-5.16053736e-01 -1.96710780e-01 -1.38656190e-03 2.14077681e-01
-6.57424569e-01 9.69625175e-01 8.97228003e-01 -1.85030669e-01
-1.58927962e-02 1.07975833e-01 -4.03286964e-01 5.31657159e-01
8.22926819e-01 -2.94604778e-01 -4.28421438e-01 -3.63757312e-01
-5.67194447e-02 5.63069098e-02 1.95325345e-01 -1.15784383e+00
4.15274352e-01 -3.28523636e-01 3.25473219e-01 -8.61992180e-01
2.89839834e-01 -9.42114353e-01 -8.08488801e-02 1.61941290e-01
-6.55401409e-01 -4.73924130e-01 2.18206689e-01 1.08471286e+00
-3.67244154e-01 -3.69648635e-01 9.33631241e-01 4.14034605e-01
-7.98185289e-01 2.89282888e-01 -2.79155195e-01 1.35764861e-02
5.50494432e-01 -2.11121410e-01 -2.08121225e-01 -4.45859671e-01
-3.71643543e-01 1.98299915e-01 2.10620597e-01 4.54930216e-01
6.52489722e-01 -1.42094159e+00 -5.62398076e-01 2.00257391e-01
2.97080606e-01 6.76602945e-02 4.94269580e-01 7.92561173e-01
-2.05144268e-02 4.63092715e-01 2.50417084e-01 -9.67055619e-01
-1.14817321e+00 5.41639566e-01 2.68232048e-01 -7.48910010e-02
-3.72440875e-01 9.01685417e-01 2.32354745e-01 -2.92861521e-01
7.41719723e-01 -4.78765666e-01 -1.14527836e-01 -2.23255664e-01
5.37104189e-01 1.11295484e-01 2.12225899e-01 -4.02231395e-01
-2.63579279e-01 3.91652972e-01 -3.41513067e-01 -7.16385245e-02
1.14837778e+00 -5.39881885e-01 4.05303478e-01 5.91495693e-01
1.08239841e+00 -4.08164114e-01 -1.91368735e+00 -7.00510502e-01
2.83382297e-01 -6.32196426e-01 1.25959575e-01 -7.31040001e-01
-9.31980371e-01 1.14402187e+00 5.66496432e-01 -1.28320843e-01
1.09358716e+00 -4.16367263e-01 5.91324687e-01 6.61176622e-01
1.37680486e-01 -1.53644943e+00 2.78007865e-01 7.93266356e-01
6.51106060e-01 -1.48531032e+00 -2.63340473e-01 -1.87894642e-01
-9.23302233e-01 1.25476575e+00 6.33440971e-01 3.13655198e-01
4.64761734e-01 2.94151843e-01 -2.83633675e-02 5.16818762e-01
-6.20210111e-01 4.94555943e-02 2.49076143e-01 5.70587754e-01
3.51280600e-01 3.73121491e-03 -9.45082530e-02 6.72719836e-01
2.64078170e-01 -9.24360976e-02 2.85689950e-01 7.75187731e-01
-4.44778979e-01 -9.37951624e-01 -2.72918373e-01 4.51425672e-01
-3.90615970e-01 -3.10553133e-01 -2.01668013e-02 2.93054879e-01
-2.74879247e-01 1.10999143e+00 -3.55424061e-02 -5.72482705e-01
2.21823335e-01 1.28323942e-01 1.74012870e-01 -2.10580125e-01
-1.50864512e-01 2.23057747e-01 1.54027998e-01 -2.98323601e-01
-4.86876875e-01 -7.87468255e-01 -1.13217640e+00 -1.52815863e-01
-7.59727001e-01 1.62994172e-02 9.07296777e-01 1.04614019e+00
1.71148866e-01 4.74891990e-01 6.08695924e-01 -6.18242919e-01
-1.05414903e+00 -1.21941781e+00 -4.35811579e-01 4.11369979e-01
4.50204223e-01 -7.65292943e-01 -4.54476237e-01 2.15115189e-03] | [11.949773788452148, 2.275439977645874] |
96f4bce6-6c79-4f04-97a2-a1388d4beec7 | pranet-parallel-reverse-attention-network-for | 2006.11392 | null | https://arxiv.org/abs/2006.11392v4 | https://arxiv.org/pdf/2006.11392v4.pdf | PraNet: Parallel Reverse Attention Network for Polyp Segmentation | Colonoscopy is an effective technique for detecting colorectal polyps, which are highly related to colorectal cancer. In clinical practice, segmenting polyps from colonoscopy images is of great importance since it provides valuable information for diagnosis and surgery. However, accurate polyp segmentation is a challenging task, for two major reasons: (i) the same type of polyps has a diversity of size, color and texture; and (ii) the boundary between a polyp and its surrounding mucosa is not sharp. To address these challenges, we propose a parallel reverse attention network (PraNet) for accurate polyp segmentation in colonoscopy images. Specifically, we first aggregate the features in high-level layers using a parallel partial decoder (PPD). Based on the combined feature, we then generate a global map as the initial guidance area for the following components. In addition, we mine the boundary cues using a reverse attention (RA) module, which is able to establish the relationship between areas and boundary cues. Thanks to the recurrent cooperation mechanism between areas and boundaries, our PraNet is capable of calibrating any misaligned predictions, improving the segmentation accuracy. Quantitative and qualitative evaluations on five challenging datasets across six metrics show that our PraNet improves the segmentation accuracy significantly, and presents a number of advantages in terms of generalizability, and real-time segmentation efficiency. | ['Geng Chen', 'Ge-Peng Ji', 'Deng-Ping Fan', 'Huazhu Fu', 'Jianbing Shen', 'Tao Zhou', 'Ling Shao'] | 2020-06-13 | null | null | null | null | ['camouflage-segmentation', 'camouflaged-object-segmentation', 'video-polyp-segmentation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 1.84116021e-01 6.19887859e-02 -1.43008843e-01 -1.65494978e-01
-5.78406274e-01 -5.17070293e-01 1.89769566e-01 6.13087416e-01
-3.26286554e-01 1.34314477e-01 3.06578487e-01 -3.27681154e-01
4.38484326e-02 -9.31211770e-01 -6.80976689e-01 -7.08911777e-01
-1.04565278e-01 1.37508199e-01 5.73909521e-01 -1.23521544e-01
1.67088866e-01 2.39862442e-01 -8.96648586e-01 3.93433750e-01
1.39799261e+00 9.59638000e-01 5.90063930e-01 4.12510276e-01
1.68966763e-02 2.83972949e-01 -2.04684258e-01 -2.06325248e-01
2.15951145e-01 -4.39518631e-01 -6.26315117e-01 1.35710478e-01
-3.20621915e-02 -1.15108997e-01 -2.03254119e-01 1.44828761e+00
2.99822599e-01 -1.22864015e-01 5.34689844e-01 -3.24671060e-01
-4.20475036e-01 6.41902089e-01 -6.87245250e-01 1.63579464e-01
1.91272154e-01 2.79120743e-01 8.28867793e-01 -3.90903294e-01
1.14740863e-01 8.87404740e-01 6.62001193e-01 1.79187432e-01
-8.66378486e-01 -3.73360783e-01 4.25386488e-01 -2.15880647e-01
-1.20433986e+00 1.52030155e-01 4.36084926e-01 -5.83421588e-01
3.32739621e-01 2.77743787e-01 1.09585059e+00 5.82729220e-01
5.36964655e-01 9.20026958e-01 6.77716076e-01 -1.44062713e-01
-1.04462482e-01 3.99444113e-03 -9.71534401e-02 7.86256969e-01
3.59966636e-01 2.52053980e-02 2.46148735e-01 1.61139399e-01
1.06087530e+00 3.12830627e-01 -5.39120615e-01 -3.13958377e-01
-1.56010449e+00 6.88999116e-01 1.31086302e+00 4.63434428e-01
-3.94353032e-01 -1.82779357e-01 2.38039702e-01 -1.70161143e-01
-6.76074112e-03 9.19001460e-01 -3.97844613e-02 3.28291327e-01
-6.96937978e-01 -2.59699881e-01 6.86226547e-01 6.21253192e-01
4.22515988e-01 -5.90710104e-01 -3.02692413e-01 8.16176832e-01
5.49999416e-01 1.97785363e-01 9.51426148e-01 -3.27638030e-01
4.68594104e-01 1.25008178e+00 9.34490655e-03 -1.26145244e+00
-6.22976303e-01 -8.47707391e-01 -1.16579223e+00 -1.62844017e-01
3.62091601e-01 -1.15108844e-02 -9.38388526e-01 1.36800826e+00
2.14188531e-01 7.59440958e-02 -1.62806943e-01 1.15686524e+00
1.00340700e+00 4.25128728e-01 8.76035839e-02 2.26094767e-01
1.69341552e+00 -1.28261518e+00 -3.06700498e-01 -7.23977506e-01
5.35169423e-01 -6.75015271e-01 8.34822774e-01 1.89507455e-01
-9.26502705e-01 -5.28433621e-01 -1.22267735e+00 -1.76736206e-01
-6.16378337e-02 5.77063501e-01 6.21404409e-01 3.74488980e-01
-8.12426925e-01 2.58282542e-01 -1.07083964e+00 -2.07878634e-01
3.79570603e-01 4.16675538e-01 -1.46752864e-01 -9.82979313e-02
-1.04897821e+00 4.78176385e-01 4.79455084e-01 4.72808987e-01
-5.41155040e-01 -5.03391981e-01 -1.16501236e+00 2.30790779e-01
1.61197990e-01 -7.66475201e-01 1.02083647e+00 -9.82362688e-01
-1.30386913e+00 7.87551403e-01 6.81484714e-02 -3.15453202e-01
5.83179533e-01 5.38003072e-02 -1.79878935e-01 1.80673674e-01
1.14251852e-01 7.01406717e-01 4.48313415e-01 -9.22840893e-01
-8.58759642e-01 -3.58343929e-01 1.74834892e-01 5.34513175e-01
-2.04124659e-01 -5.44620872e-01 -9.21514630e-01 -6.08857036e-01
5.14436364e-01 -9.37128425e-01 -7.92673409e-01 1.59178093e-01
-5.97105324e-01 3.37290883e-01 1.31520540e-01 -8.41347337e-01
1.42065573e+00 -2.48468351e+00 1.65103078e-01 3.23793143e-01
4.58400100e-01 4.35221136e-01 9.48573500e-02 -2.30293006e-01
1.85379848e-01 2.42322639e-01 -4.54408705e-01 9.46849361e-02
-4.37917918e-01 -5.79338074e-02 1.03210591e-01 3.38207453e-01
1.80922925e-01 1.20952713e+00 -1.19660771e+00 -5.31072438e-01
3.55503201e-01 2.71451682e-01 -5.64900935e-01 1.52120799e-01
-1.64090276e-01 6.40129030e-01 -6.79962993e-01 4.97777164e-01
5.57165027e-01 -6.79020107e-01 2.46799067e-01 -4.04637933e-01
-1.77996680e-01 1.16083413e-01 -9.01171565e-01 1.71588767e+00
-5.47105670e-01 3.87771279e-01 3.70187834e-02 -7.31374621e-01
8.61437261e-01 2.04526186e-01 3.80107105e-01 -6.07632697e-01
3.02422553e-01 5.13690948e-01 3.33763897e-01 -6.08483732e-01
1.51393041e-01 2.74374276e-01 -1.49199679e-01 2.22540963e-02
-4.76323754e-01 -2.40207538e-01 2.27921903e-01 -1.39224812e-01
8.87218654e-01 -2.97221780e-01 6.21560812e-01 -2.91734904e-01
7.69088864e-01 7.01487111e-03 6.56187415e-01 4.38170165e-01
-3.33259523e-01 8.69589329e-01 6.08824313e-01 -4.78472263e-01
-6.48517191e-01 -8.64306629e-01 -1.42744467e-01 2.96188474e-01
7.56337047e-01 7.14371502e-02 -5.73262513e-01 -6.70408785e-01
-1.51133209e-01 1.08716093e-01 -8.23577583e-01 -2.08606780e-01
-6.32285297e-01 -9.91200447e-01 9.91444066e-02 4.60932136e-01
7.57630229e-01 -9.47976649e-01 -6.77785635e-01 3.84985328e-01
-5.36300063e-01 -7.83046365e-01 -8.57628524e-01 -4.14547324e-02
-9.02087092e-01 -1.34293699e+00 -9.65589583e-01 -1.26921129e+00
1.18088925e+00 6.22633636e-01 9.98481810e-01 4.89076912e-01
-3.64496827e-01 -2.93954104e-01 -2.62827694e-01 -1.22572370e-01
-3.74126285e-01 2.67437994e-01 -8.07768524e-01 3.57393585e-02
1.69904947e-01 -1.91990510e-01 -1.16576982e+00 5.18307984e-01
-9.25943255e-01 3.98241788e-01 1.06830180e+00 9.92604375e-01
8.23430300e-01 4.21400964e-02 1.84840545e-01 -8.06448340e-01
5.38798988e-01 -5.93191862e-01 -7.32247233e-01 2.49683499e-01
-6.13683090e-02 -1.81880191e-01 5.30978084e-01 -2.17473105e-01
-8.83280158e-01 2.61887014e-01 -2.44747564e-01 1.24281935e-01
7.84607977e-02 7.56618083e-01 -2.76791938e-02 -1.42417386e-01
6.05262816e-01 1.47630766e-01 9.40168127e-02 -1.68179885e-01
-8.65617394e-02 5.46085238e-01 4.06202465e-01 -8.32781866e-02
2.84915358e-01 4.41993326e-01 -3.40929538e-01 -5.34378171e-01
-8.62967134e-01 -6.74316883e-01 -4.71239567e-01 9.83973220e-02
9.90555882e-01 -8.95119190e-01 -5.84455490e-01 4.42135781e-01
-9.25507188e-01 -3.20140392e-01 -1.34484291e-01 7.60056794e-01
-4.78508845e-02 4.99682724e-01 -8.94669294e-01 -5.95947914e-03
-4.31439072e-01 -1.73649895e+00 9.71364975e-01 7.42483675e-01
5.81855923e-02 -1.14107406e+00 -7.54372478e-02 2.01081261e-01
3.60738695e-01 3.53783846e-01 9.39107060e-01 -4.09146309e-01
-7.81356335e-01 -2.99343467e-01 -6.13228500e-01 9.40044373e-02
2.55990684e-01 -1.95076466e-02 -4.76104140e-01 -3.21938574e-01
-1.23161770e-01 1.48102120e-01 1.01858926e+00 7.84352660e-01
1.25377369e+00 4.35493402e-02 -7.36184061e-01 9.34438705e-01
1.41215241e+00 2.80204654e-01 4.83257592e-01 2.81537473e-01
5.08971810e-01 4.34072465e-01 5.68857670e-01 1.38818845e-01
4.52672482e-01 4.97440904e-01 7.11401761e-01 -5.87917089e-01
-2.39593536e-01 -1.88510314e-01 -3.51963453e-02 8.24809611e-01
2.83273429e-01 -1.53105944e-01 -9.71279383e-01 8.27972233e-01
-1.76292133e+00 -3.46684664e-01 -1.62706286e-01 2.31626225e+00
8.01942885e-01 9.34165120e-02 -2.10431531e-01 -2.82788396e-01
7.55108476e-01 -1.01883389e-01 -5.92098236e-01 -1.71095785e-02
1.58362299e-01 -3.55236202e-01 4.45203573e-01 4.91159737e-01
-1.27125585e+00 3.99235904e-01 5.59922934e+00 5.27119696e-01
-1.45345974e+00 -2.60278583e-01 9.12895858e-01 3.10987681e-01
-2.59350717e-01 -3.72413009e-01 -6.04821265e-01 6.19886279e-01
-5.45189297e-03 -3.71031761e-02 1.67040303e-01 5.97607851e-01
-1.13475323e-02 -1.64311334e-01 -9.82880235e-01 5.84329307e-01
-7.22449943e-02 -1.20667589e+00 -7.46017769e-02 1.89015102e-02
8.69625092e-01 2.07573876e-01 1.40904874e-01 1.26645237e-01
2.41905823e-01 -9.85025644e-01 2.40571305e-01 3.77377033e-01
6.77038312e-01 -4.68772620e-01 1.17742705e+00 3.40099633e-01
-1.41680539e+00 -1.00799069e-01 -2.96071976e-01 4.26055133e-01
-6.44489145e-03 8.24944854e-01 -9.14152205e-01 5.61814606e-01
4.23496991e-01 8.43600631e-01 -6.51088834e-01 1.90642238e+00
-4.27131057e-01 2.75079906e-01 -3.01911801e-01 8.38142559e-02
3.99117619e-01 -2.88818896e-01 4.27246332e-01 1.17605710e+00
5.41105628e-01 -1.78606018e-01 3.82741570e-01 8.66217613e-01
1.68679878e-02 2.46228173e-01 -3.95725258e-02 2.04340249e-01
2.81643629e-01 1.35871732e+00 -1.18319368e+00 -2.28022248e-01
-3.36237907e-01 7.38269925e-01 1.14892073e-01 1.14865646e-01
-7.24373817e-01 -3.71472925e-01 2.14974239e-01 -9.38342884e-02
1.78441569e-01 1.46022096e-01 -3.52539420e-01 -1.16656029e+00
1.11579359e-01 -7.76702225e-01 4.36774015e-01 -2.83574462e-01
-1.03399515e+00 8.06828856e-01 -5.98483860e-01 -1.53264701e+00
1.85613856e-01 -4.76396203e-01 -6.80525839e-01 8.61733735e-01
-1.66820526e+00 -9.97343361e-01 -9.56354856e-01 1.16310246e-01
3.90443236e-01 3.11042905e-01 7.62198150e-01 3.94286007e-01
-7.03221023e-01 4.73913819e-01 -1.20130204e-01 4.08327997e-01
5.65893948e-01 -1.41786766e+00 4.82825816e-01 8.07438314e-01
-1.45935535e-01 6.22694671e-01 1.59624264e-01 -6.47830129e-01
-9.00911748e-01 -1.22475338e+00 5.73525727e-01 8.66590887e-02
3.95835131e-01 4.51381877e-03 -9.20885384e-01 4.73467767e-01
-1.74850255e-01 1.24263614e-01 5.56685328e-01 -2.17679352e-01
6.90886527e-02 -6.47641793e-02 -9.44030702e-01 7.19781876e-01
5.12215912e-01 1.72793102e-02 -3.59599680e-01 4.20801163e-01
7.81616151e-01 -1.03321123e+00 -8.90529633e-01 4.39596206e-01
4.91462290e-01 -1.05123973e+00 9.45494533e-01 2.39413872e-01
5.98281503e-01 -4.21240360e-01 4.75100696e-01 -1.37839282e+00
-4.83700424e-01 -2.39590302e-01 4.31909859e-01 5.81764042e-01
6.45593405e-01 -8.10834289e-01 7.49543667e-01 3.56808662e-01
-4.23828185e-01 -1.06949341e+00 -5.48858166e-01 -1.57167152e-01
-1.84258386e-01 1.13236025e-01 7.09484816e-01 7.24847436e-01
2.81168371e-02 -7.76224360e-02 2.27024853e-01 3.71941596e-01
1.60268754e-01 3.73547375e-01 3.57550174e-01 -9.56058800e-01
-3.35861832e-01 -8.02251458e-01 -3.30875218e-01 -1.56824589e+00
-5.64392328e-01 -8.88217926e-01 2.66505808e-01 -1.99072409e+00
1.95385769e-01 -8.12430441e-01 -4.48521405e-01 2.55032301e-01
-6.24830902e-01 2.33274847e-01 4.25226986e-03 2.86247611e-01
-4.60371673e-01 1.25162527e-01 1.93490624e+00 -2.36611128e-01
-4.30464417e-01 5.16581535e-01 -1.00119185e+00 8.40305626e-01
6.89417601e-01 -1.97090328e-01 -2.34849766e-01 -5.98751605e-01
2.74650872e-01 2.53406674e-01 3.51397425e-01 -1.01552701e+00
3.89615953e-01 9.12724286e-02 3.54794830e-01 -4.62471992e-01
-6.03353716e-02 -7.91598797e-01 -1.53607637e-01 1.03514981e+00
-2.89977014e-01 -3.22302729e-01 1.75731570e-01 5.76769412e-01
-6.72751546e-01 -3.14076215e-01 8.70319963e-01 -4.42622632e-01
-4.67507541e-01 3.82046461e-01 -1.75516516e-01 -1.91757139e-02
1.04893720e+00 -1.88436896e-01 -1.64933890e-01 -1.78304091e-02
-5.55937111e-01 6.84904635e-01 5.66427708e-01 2.82203913e-01
5.24554670e-01 -1.06273150e+00 -7.18580425e-01 4.67810005e-01
1.32656693e-01 7.64694989e-01 4.92011160e-01 1.34488416e+00
-9.95631099e-01 4.45721805e-01 -2.35693213e-02 -9.09624398e-01
-8.43115330e-01 3.64008009e-01 7.62315154e-01 -6.95418119e-01
-7.45369494e-01 1.00353122e+00 7.34468520e-01 -3.15175205e-01
1.23405732e-01 -8.77445161e-01 -5.49818873e-01 -8.29144046e-02
6.06365442e-01 -9.78979468e-02 7.37920925e-02 -2.58595109e-01
-9.71154273e-02 6.44419909e-01 -2.69914478e-01 4.46180344e-01
9.80112076e-01 -1.11245826e-01 -1.80286974e-01 -4.24524955e-02
8.62536252e-01 1.34610176e-01 -1.35329127e+00 -2.95764267e-01
-2.17722595e-01 -5.13098240e-01 4.18151841e-02 -8.65471303e-01
-1.30081475e+00 8.80667090e-01 4.45109814e-01 2.49951199e-01
1.15671921e+00 -2.31177643e-01 9.70296741e-01 -1.87411681e-01
7.25405142e-02 -4.93472964e-01 3.33245136e-02 2.09713474e-01
8.55533063e-01 -1.34460449e+00 -4.65091951e-02 -6.92877412e-01
-5.60728252e-01 1.14986372e+00 6.04221523e-01 -2.24594325e-01
5.27497411e-01 5.14929742e-02 2.12642059e-01 -1.26885533e-01
-1.76814914e-01 7.59757683e-02 5.22665679e-01 2.65988588e-01
5.48345804e-01 2.40350038e-01 -3.50689232e-01 6.86844468e-01
-5.71053103e-02 -7.82748163e-02 2.86546826e-01 4.93970037e-01
-4.82438654e-01 -8.52125585e-01 -2.15343937e-01 6.21803820e-01
-4.68758166e-01 -1.49701148e-01 1.76312421e-02 5.69394648e-01
2.28526190e-01 7.43441284e-01 6.27138689e-02 -1.44768134e-01
2.39245489e-01 -8.31115603e-01 1.27338648e-01 -8.28505933e-01
-9.11294043e-01 2.88361162e-01 -2.11738288e-01 -3.73072535e-01
-9.36298445e-02 -4.53848779e-01 -1.29059863e+00 3.94740909e-01
-4.46860582e-01 1.52034014e-01 5.62608838e-01 7.36206889e-01
3.14779252e-01 9.78509784e-01 5.88791311e-01 -6.23806000e-01
-4.21599865e-01 -6.53210759e-01 -2.73429632e-01 4.03046638e-01
3.60105485e-01 -3.10225219e-01 -2.53394395e-01 -1.71103939e-01] | [14.588854789733887, -2.707505941390991] |
59d3bcc3-0cff-4926-aa04-eafe5b1024c3 | enhanced-detection-of-the-presence-and | 2303.09440 | null | https://arxiv.org/abs/2303.09440v2 | https://arxiv.org/pdf/2303.09440v2.pdf | Enhanced detection of the presence and severity of COVID-19 from CT scans using lung segmentation | Improving automated analysis of medical imaging will provide clinicians more options in providing care for patients. The 2023 AI-enabled Medical Image Analysis Workshop and Covid-19 Diagnosis Competition (AI-MIA-COV19D) provides an opportunity to test and refine machine learning methods for detecting the presence and severity of COVID-19 in patients from CT scans. This paper presents version 2 of Cov3d, a deep learning model submitted in the 2022 competition. The model has been improved through a preprocessing step which segments the lungs in the CT scan and crops the input to this region. It results in a validation macro F1 score for predicting the presence of COVID-19 in the CT scans at 93.2% which is significantly above the baseline of 74\%. It gives a macro F1 score for predicting the severity of COVID-19 on the validation set for task 2 as 72.8% which is above the baseline of 38%. | ['Robert Turnbull'] | 2023-03-16 | null | null | null | null | ['covid-19-detection'] | ['medical'] | [ 8.55922699e-02 3.02116513e-01 -4.05284995e-03 -2.01197088e-01
-1.11522663e+00 -2.98480064e-01 2.75189310e-01 2.45288134e-01
-4.66906756e-01 3.70309979e-01 2.69734859e-01 -4.75968748e-01
-1.76230714e-01 -2.30851740e-01 -3.50591928e-01 -5.89373827e-01
-3.74002367e-01 9.91503775e-01 3.50532800e-01 4.49427336e-01
-3.85764003e-01 5.68056464e-01 -6.93613112e-01 9.27904904e-01
1.05516247e-01 1.11687958e+00 2.62161046e-01 1.17457962e+00
4.00839806e-01 9.92830992e-01 -4.65766907e-01 -6.02212735e-02
2.19981879e-01 -4.53080803e-01 -9.13551390e-01 -2.11642280e-01
6.20278001e-01 -4.62386250e-01 -3.32339108e-01 5.53421080e-01
7.40062296e-01 -4.40611452e-01 9.82640684e-01 -1.02001870e+00
1.37909606e-01 5.37492275e-01 -6.43472970e-01 9.96252060e-01
-4.77856211e-03 3.92700821e-01 8.44379544e-01 -7.87758291e-01
7.74059474e-01 8.08216631e-01 1.06482267e+00 5.55440366e-01
-8.32266450e-01 -4.76812392e-01 -5.59451282e-01 5.63717373e-02
-1.17091751e+00 1.65121436e-01 -1.07697710e-01 -1.00116062e+00
1.02563679e+00 4.71292943e-01 6.54857457e-01 8.68022501e-01
6.85425103e-01 8.37234795e-01 6.80135369e-01 -1.49243265e-01
-1.21219745e-02 -2.61357784e-01 2.62936026e-01 8.63356590e-01
2.11798310e-01 4.37595136e-03 3.02702487e-02 -2.28113666e-01
7.50429153e-01 -2.79205292e-01 3.91643271e-02 -1.27303720e-01
-1.45209777e+00 9.11536694e-01 5.46241403e-01 3.74262661e-01
-7.31036186e-01 6.17451072e-02 7.80292094e-01 -1.41414866e-01
2.79494196e-01 4.04570103e-01 -4.92142111e-01 6.59767911e-02
-9.33683872e-01 4.52620238e-01 2.62631476e-01 3.35590452e-01
-4.34413314e-01 1.95618682e-02 -6.90907359e-01 6.95158184e-01
2.02538699e-01 5.63246906e-01 8.33038628e-01 -9.67366040e-01
3.78120303e-01 4.02329236e-01 -1.16228908e-01 -5.30683279e-01
-1.08420551e+00 -6.92406774e-01 -8.57570529e-01 1.81439191e-01
3.73123318e-01 -5.28029323e-01 -1.33213162e+00 1.33892906e+00
1.00507952e-01 1.18592486e-01 -1.73767388e-01 9.94083285e-01
1.11758149e+00 2.60070026e-01 2.71177113e-01 -1.67653151e-02
1.43854010e+00 -8.21012378e-01 -3.42867583e-01 1.15407102e-01
8.77324760e-01 -4.75073814e-01 8.37730348e-01 5.05765021e-01
-1.13298428e+00 -4.32439834e-01 -1.01086271e+00 3.89262885e-01
-5.75033017e-04 5.96118607e-02 2.50097185e-01 6.74906135e-01
-1.06185508e+00 2.98534989e-01 -9.61862206e-01 -1.08566910e-01
6.10540092e-01 3.72234076e-01 -2.82782853e-01 -4.60789874e-02
-1.03153694e+00 1.24893367e+00 2.88804203e-01 -2.97523826e-01
-1.44310081e+00 -1.18026102e+00 -5.16580045e-01 -1.38708219e-01
2.03483477e-01 -8.62204492e-01 1.53539598e+00 -4.08735812e-01
-4.37346965e-01 1.22717857e+00 3.02456737e-01 -8.91632020e-01
1.07869935e+00 -1.46673806e-02 -4.70310301e-01 3.72736067e-01
2.36384839e-01 7.55937159e-01 5.99051654e-01 -7.44705200e-01
-8.56734157e-01 -2.92564034e-01 -3.72032493e-01 4.06020805e-02
4.72132415e-01 1.33997023e-01 -3.21716338e-01 -6.61833465e-01
-2.46782407e-01 -1.03311610e+00 -4.53729451e-01 -2.03405797e-01
-3.10797662e-01 -1.77603751e-01 8.14244866e-01 -7.91353464e-01
9.12827194e-01 -2.08710051e+00 -3.58144939e-01 1.87399164e-01
9.11131799e-01 5.39286256e-01 1.40861971e-02 -2.94618428e-01
-5.76561928e-01 1.88483626e-01 -2.07349524e-01 -8.59322771e-02
-3.60993207e-01 1.19370446e-02 2.11253256e-01 4.75448400e-01
2.27336690e-01 1.00168693e+00 -7.00317562e-01 -5.70990741e-01
2.56725967e-01 2.82829225e-01 -3.29434484e-01 2.02322006e-01
-1.10101895e-02 5.73964477e-01 -4.08865362e-01 4.29211855e-01
4.44432914e-01 -5.76851249e-01 -7.27213025e-02 -5.66668995e-02
2.85959303e-01 -1.82741597e-01 -6.04593396e-01 1.32870936e+00
-2.09989995e-01 7.03344941e-01 4.40934822e-02 -8.54017735e-01
3.05484712e-01 7.14571297e-01 1.26151943e+00 -6.45693123e-01
2.41312131e-01 2.15468302e-01 5.80627382e-01 -7.93030083e-01
-1.07133470e-01 -2.08227634e-01 -5.23475558e-02 4.14892226e-01
-4.99478281e-02 -2.93709636e-01 1.10154748e-01 3.05382460e-01
1.59162664e+00 -4.80392992e-01 3.00778091e-01 -4.27025944e-01
5.76117218e-01 2.14441746e-01 3.42417449e-01 8.50428343e-01
-6.08168662e-01 8.96483958e-01 3.69298488e-01 -7.34772205e-01
-1.06742167e+00 -1.19023561e+00 -4.83146578e-01 4.69144762e-01
-4.23422515e-01 -1.02777109e-01 -7.83025861e-01 -1.07708097e+00
4.87311706e-02 6.13943458e-01 -9.54421461e-01 -8.00120160e-02
-7.31416047e-01 -8.77670467e-01 8.15227687e-01 8.32125187e-01
4.21645381e-02 -1.12534881e+00 -1.07690501e+00 4.85376231e-02
-3.59315395e-01 -1.20723403e+00 -3.13497454e-01 1.35457307e-01
-8.04929256e-01 -1.36944437e+00 -1.05086517e+00 -6.88427806e-01
2.56161630e-01 -2.53975928e-01 1.25798249e+00 1.80858165e-01
-8.56332660e-01 2.73540586e-01 -1.87325507e-01 -8.42652798e-01
-5.67582846e-01 -1.78480092e-02 -8.77025276e-02 -5.60070574e-01
1.12563662e-01 8.22024867e-02 -6.55092955e-01 1.34620309e-01
-6.50782883e-01 2.11663932e-01 5.57508588e-01 7.30129898e-01
7.10860908e-01 -1.70317933e-01 5.34716010e-01 -8.94877255e-01
3.80009145e-01 -6.52546763e-01 -3.54829013e-01 -5.08994572e-02
-6.60929799e-01 -2.74302930e-01 2.57796347e-01 1.34887649e-02
-5.16537488e-01 2.49413833e-01 -2.92844474e-01 -5.61402857e-01
-1.61516443e-01 3.71937007e-01 6.56843603e-01 2.66200095e-01
7.00614393e-01 -3.16703230e-01 2.45065559e-02 -4.40075397e-01
-8.01630542e-02 5.30508339e-01 7.74905264e-01 -6.43024221e-02
3.17763627e-01 3.46723706e-01 4.23584461e-01 -5.14798224e-01
-9.85479951e-01 -6.97498918e-01 -6.92203939e-01 -3.25580269e-01
1.37110460e+00 -7.40506887e-01 -2.55941004e-01 3.79133046e-01
-7.79592276e-01 -3.92020434e-01 -3.54438812e-01 6.80141807e-01
-4.00874466e-01 7.87093043e-02 -5.67984760e-01 -2.79989839e-01
-7.59264767e-01 -1.53711188e+00 1.00115490e+00 -1.51663914e-01
-5.05641639e-01 -8.76771569e-01 2.73688227e-01 3.80895793e-01
4.77329344e-01 6.00648463e-01 1.14801943e+00 -1.19975996e+00
7.24344775e-02 -3.30333114e-01 -4.71424192e-01 3.01993728e-01
-1.12498194e-01 -2.10035145e-01 -7.91406691e-01 -1.59590691e-01
-2.45546356e-01 -2.86514431e-01 9.32512283e-01 8.98310065e-01
1.35193002e+00 3.73782367e-01 -3.95195544e-01 3.87269944e-01
1.19575071e+00 5.26233494e-01 2.64419824e-01 1.71224415e-01
5.50493538e-01 8.69011432e-02 4.35361803e-01 4.70552474e-01
5.24115153e-02 6.46906972e-01 7.23092139e-01 -2.61084914e-01
-6.04344845e-01 3.47177804e-01 -1.84818149e-01 4.42920804e-01
-2.34919954e-02 -2.17333168e-01 -1.68204832e+00 6.99405134e-01
-1.45036030e+00 -4.97904122e-01 -7.78456569e-01 1.67484319e+00
4.97665793e-01 1.53051391e-01 3.50352496e-01 -4.50373106e-02
5.86114526e-01 -1.73678696e-01 -4.66925949e-01 -6.16231799e-01
2.81156510e-01 5.20079732e-01 5.12640178e-01 3.37461621e-01
-1.34975517e+00 2.93516129e-01 7.58297014e+00 5.93671083e-01
-1.14942253e+00 3.92394751e-01 8.67756248e-01 -1.68295443e-01
2.35683605e-01 -7.78499305e-01 -5.70990860e-01 4.53026593e-01
1.26136398e+00 -3.27415168e-02 -2.12547660e-01 6.92856073e-01
4.24881220e-01 -1.50812805e-01 -1.08892345e+00 8.83235037e-01
7.08087087e-02 -1.31745040e+00 -1.54442579e-01 1.17246665e-01
7.61405230e-01 6.62304282e-01 1.69222057e-01 3.85481030e-01
2.63631880e-01 -1.25940979e+00 4.22951341e-01 4.29148257e-01
1.15378606e+00 -6.09640419e-01 1.16945934e+00 3.66490066e-01
-7.58422196e-01 6.56188875e-02 -3.09009347e-02 4.36486781e-01
2.28953287e-01 5.50063968e-01 -1.78489089e+00 3.00754011e-01
7.65681028e-01 2.36349300e-01 -7.51596093e-01 1.30973077e+00
1.46256313e-01 9.03989315e-01 -1.60416007e-01 3.84362876e-01
4.18340713e-01 6.29494965e-01 8.05494010e-01 1.55346477e+00
2.03214958e-01 2.02465847e-01 1.92024171e-01 3.49234045e-01
-1.34010509e-01 1.76698603e-02 -4.04755652e-01 2.25842863e-01
-8.80511999e-02 1.04405570e+00 -6.58460140e-01 -7.56854713e-01
-9.67823491e-02 5.87619245e-01 -3.55803728e-01 -2.84698099e-01
-1.19266582e+00 7.35986009e-02 7.09972158e-02 2.92777359e-01
2.49209613e-01 2.74695814e-01 -5.29969931e-01 -6.46627903e-01
-4.95518178e-01 -9.20233250e-01 8.72800827e-01 -9.00811493e-01
-1.01385844e+00 8.31558704e-01 2.79030893e-02 -1.10185814e+00
-6.57677650e-01 -7.88940609e-01 -5.44188559e-01 8.27349007e-01
-1.01515639e+00 -9.26244676e-01 -5.26378214e-01 4.70574141e-01
5.73013663e-01 -4.00350749e-01 9.08727348e-01 3.22547644e-01
-3.44177693e-01 6.49721920e-01 -1.12391599e-01 3.12078774e-01
6.59039855e-01 -1.42834067e+00 3.80090237e-01 6.99571252e-01
-2.53707826e-01 -5.58903776e-02 3.63865286e-01 -7.21972883e-01
-7.35016644e-01 -1.25866294e+00 7.64318585e-01 -7.09469020e-01
4.37523037e-01 3.15481365e-01 -6.95744514e-01 5.87481916e-01
1.34617969e-01 1.88988790e-01 5.62650323e-01 -5.44338286e-01
-3.35426666e-02 3.15313250e-01 -1.48979819e+00 2.98488196e-02
5.48995912e-01 -1.11466691e-01 -7.28943765e-01 5.01919568e-01
7.49811888e-01 -6.94309056e-01 -1.08527875e+00 9.76674557e-01
5.85610747e-01 -6.79812253e-01 1.06939483e+00 -9.50134099e-01
6.17782056e-01 6.18216805e-02 -1.27874061e-01 -1.05015469e+00
-4.70018983e-01 9.85407457e-02 3.06695998e-02 2.73693681e-01
6.94063485e-01 -4.25507650e-02 8.74363899e-01 4.47841495e-01
-2.30681032e-01 -9.31834996e-01 -9.97560143e-01 -4.51301038e-01
3.41643959e-01 -7.96730101e-01 5.12138456e-02 9.29539561e-01
-4.68319386e-01 -2.90514762e-03 1.04428127e-01 -6.63213730e-02
5.17909586e-01 -3.23591262e-01 3.46232474e-01 -1.19232237e+00
-4.40826893e-01 -6.98463678e-01 -4.50483590e-01 -1.64502338e-01
-4.97875482e-01 -9.83112991e-01 -1.06615342e-01 -1.69038045e+00
6.72991276e-01 -2.92276949e-01 -7.09031165e-01 5.14498591e-01
-1.10713765e-01 5.44585884e-01 3.34429383e-01 2.68992752e-01
-3.35079521e-01 -3.72058988e-01 1.35026813e+00 -1.85922131e-01
1.61418974e-01 2.84562469e-01 -2.85921305e-01 9.36643600e-01
8.04251730e-01 -5.67615509e-01 -2.28544354e-01 -4.07442629e-01
-1.49563953e-01 5.16174316e-01 4.35189396e-01 -1.15824056e+00
-3.56408507e-01 3.96759920e-02 7.22374260e-01 -1.08306110e+00
1.69285879e-01 -7.80362010e-01 3.22710425e-02 1.21903419e+00
-4.78976667e-01 4.41680878e-01 3.39148641e-01 2.64425606e-01
3.25813442e-02 -2.05529779e-01 1.12231231e+00 -4.46960300e-01
-6.52938426e-01 2.72431731e-01 -8.93826604e-01 4.56995189e-01
1.16443253e+00 6.30525574e-02 -1.94518968e-01 -2.09938526e-01
-1.16070521e+00 2.54514396e-01 -2.65832067e-01 2.69897074e-01
5.29041111e-01 -9.92762327e-01 -1.19479918e+00 3.92907821e-02
-5.30854389e-02 1.43446764e-02 3.03812861e-01 1.30960155e+00
-1.01950359e+00 7.73488700e-01 -1.81530744e-01 -9.16681409e-01
-1.54947853e+00 3.05838943e-01 8.08740199e-01 -8.51964593e-01
-8.38572323e-01 9.73218560e-01 2.65233040e-01 -5.01479991e-02
2.21257195e-01 -4.76537257e-01 -2.76464283e-01 -1.07463263e-01
6.12320483e-01 3.82408500e-01 5.47858000e-01 -6.81967020e-01
-7.11917222e-01 2.23781958e-01 -2.77670562e-01 -1.13620952e-01
1.38872504e+00 2.69608200e-01 2.80795455e-01 5.69020696e-02
1.31443322e+00 -2.02609390e-01 -5.79144120e-01 -3.84243242e-02
1.60285935e-01 -1.37652725e-01 4.15817797e-01 -1.59389603e+00
-1.26486862e+00 9.29820716e-01 1.24727392e+00 -1.07608736e-01
8.74532342e-01 1.35242730e-01 8.78955424e-01 -5.68428598e-02
-2.29155153e-01 -6.61432803e-01 2.02410936e-01 4.96504724e-01
8.25525284e-01 -1.19483614e+00 4.10499983e-02 -4.16326784e-02
-9.21221852e-01 9.34373438e-01 6.15981102e-01 -2.05171958e-01
7.09331930e-01 4.55906749e-01 2.59893268e-01 -6.09461069e-01
-7.55906701e-01 8.37044194e-02 5.88904798e-01 6.13049865e-01
4.57432687e-01 5.70278525e-01 -1.75384223e-01 8.43720436e-01
-1.17166992e-02 5.99448755e-02 4.46556866e-01 6.08358920e-01
-4.18029934e-01 -6.65129066e-01 -3.27084303e-01 8.63503695e-01
-1.08767796e+00 -3.24106626e-02 -4.04613227e-01 8.98384035e-01
4.35052484e-01 4.78954405e-01 1.27270911e-02 -3.90248448e-01
4.08862919e-01 1.63635567e-01 3.23162705e-01 -6.20524049e-01
-7.23263443e-01 1.83025092e-01 2.04137057e-01 -5.64097285e-01
-1.00155711e-01 -7.19867349e-01 -1.48906028e+00 1.28516227e-01
-1.57271754e-02 -4.16318849e-02 7.21345901e-01 8.15434039e-01
1.11177817e-01 1.05095983e+00 2.58972824e-01 -4.22227323e-01
-6.10705376e-01 -9.62276816e-01 -3.82710338e-01 2.52296835e-01
5.95001519e-01 -5.89859545e-01 -7.01230839e-02 8.89327750e-02] | [15.340465545654297, -1.9230289459228516] |
8e6d1f1c-fa7b-4271-b7cd-e10c82e6f726 | cellvit-vision-transformers-for-precise-cell | 2306.15350 | null | https://arxiv.org/abs/2306.15350v1 | https://arxiv.org/pdf/2306.15350v1.pdf | CellViT: Vision Transformers for Precise Cell Segmentation and Classification | Nuclei detection and segmentation in hematoxylin and eosin-stained (H&E) tissue images are important clinical tasks and crucial for a wide range of applications. However, it is a challenging task due to nuclei variances in staining and size, overlapping boundaries, and nuclei clustering. While convolutional neural networks have been extensively used for this task, we explore the potential of Transformer-based networks in this domain. Therefore, we introduce a new method for automated instance segmentation of cell nuclei in digitized tissue samples using a deep learning architecture based on Vision Transformer called CellViT. CellViT is trained and evaluated on the PanNuke dataset, which is one of the most challenging nuclei instance segmentation datasets, consisting of nearly 200,000 annotated Nuclei into 5 clinically important classes in 19 tissue types. We demonstrate the superiority of large-scale in-domain and out-of-domain pre-trained Vision Transformers by leveraging the recently published Segment Anything Model and a ViT-encoder pre-trained on 104 million histological image patches - achieving state-of-the-art nuclei detection and instance segmentation performance on the PanNuke dataset with a mean panoptic quality of 0.51 and an F1-detection score of 0.83. The code is publicly available at https://github.com/TIO-IKIM/CellViT | ['Jens Kleesiek', 'Jan Egger', 'Barbara Grünwald', 'Jens Siveke', 'Selma Ugurel', 'Giulia Baldini', 'Julius Keyl', 'Constantin Seibold', 'Lukas Heine', 'Moritz Rempe', 'Fabian Hörst'] | 2023-06-27 | null | null | null | null | ['panoptic-segmentation', 'cell-detection', 'instance-segmentation', 'cell-segmentation', 'classification-1'] | ['computer-vision', 'computer-vision', 'computer-vision', 'medical', 'methodology'] | [ 2.28726104e-01 2.41618484e-01 -4.31769667e-03 -1.01180032e-01
-1.29657137e+00 -6.72687590e-01 2.98842400e-01 2.71088630e-01
-8.39536965e-01 5.98566234e-01 -1.69678748e-01 -2.05343708e-01
3.01821202e-01 -6.33732736e-01 -5.33170044e-01 -1.09387481e+00
1.45736173e-01 7.81464875e-01 4.69535977e-01 2.18136813e-02
-6.05751984e-02 5.14074564e-01 -9.37274158e-01 3.52803677e-01
6.59127772e-01 9.53274727e-01 2.27565303e-01 9.82701957e-01
7.66370147e-02 5.07701814e-01 -2.77152061e-01 -4.22509581e-01
4.22857143e-02 -4.34398890e-01 -1.01565468e+00 -1.03197098e-02
4.63407934e-01 -3.30803841e-01 -2.05824926e-01 1.02614534e+00
6.93382263e-01 -4.37243462e-01 1.01614130e+00 -7.61616290e-01
-4.00700152e-01 5.65769017e-01 -4.73512858e-01 6.94236934e-01
-5.89378238e-01 5.22333205e-01 8.93727660e-01 -4.45960104e-01
9.31236207e-01 4.34414953e-01 1.02390313e+00 7.39450812e-01
-1.50779462e+00 -6.21264219e-01 -8.46789062e-01 4.57278714e-02
-1.42807722e+00 -2.14365631e-01 1.98054969e-01 -8.12605143e-01
9.48055029e-01 5.97724319e-02 1.03526008e+00 5.29526949e-01
4.52334970e-01 9.69904840e-01 1.23475325e+00 -1.06712721e-01
2.16224566e-01 -4.88021635e-02 7.79424757e-02 6.32370472e-01
1.07991166e-01 -3.25861543e-01 2.29380667e-01 1.42678395e-01
9.93467391e-01 -6.25810539e-03 -4.08765733e-01 -8.36876482e-02
-1.59568381e+00 7.92187989e-01 7.83447146e-01 6.69279993e-01
-2.26665050e-01 4.50027287e-01 8.61472368e-01 -1.00881740e-01
4.98809367e-01 4.42174494e-01 -2.57055700e-01 -5.49130626e-02
-1.21479750e+00 -3.31980884e-02 4.62088257e-01 3.97925079e-01
5.12829363e-01 -2.70565331e-01 -4.45216060e-01 8.14967871e-01
2.81709675e-02 3.61928642e-01 5.67903697e-01 -9.76703346e-01
-3.74978870e-01 7.39430666e-01 -3.86812329e-01 -3.00527722e-01
-6.73200846e-01 -6.01866126e-01 -1.21320248e+00 1.98685244e-01
8.08261573e-01 8.70632231e-02 -1.36614251e+00 1.37985313e+00
5.89695990e-01 1.59176260e-01 -1.95819303e-01 1.00487041e+00
1.04238343e+00 3.62804264e-01 2.35474631e-02 -1.30538583e-01
1.64638746e+00 -8.93143535e-01 -6.43500566e-01 7.40989000e-02
1.03898692e+00 -4.64119285e-01 8.33431244e-01 1.54326126e-01
-8.40567291e-01 -1.47103712e-01 -8.56698573e-01 -4.64540392e-01
-4.76105273e-01 2.67013043e-01 5.17169535e-01 3.80784243e-01
-1.47184396e+00 3.36969912e-01 -1.13676274e+00 -6.96605980e-01
1.19498265e+00 5.86354733e-01 -5.84879994e-01 -1.91362444e-02
-6.11521959e-01 7.68108964e-01 2.67643988e-01 2.40014374e-01
-1.00722313e+00 -1.10436630e+00 -4.14336830e-01 -2.22213604e-02
-6.32414296e-02 -7.97811210e-01 1.40246820e+00 -6.41303420e-01
-9.88129675e-01 1.51189959e+00 4.69762385e-02 -5.05225658e-01
5.97770035e-01 5.59431612e-01 2.02582881e-01 5.78131676e-01
2.05383420e-01 1.18167770e+00 2.06267014e-01 -8.17282021e-01
-5.10817647e-01 -3.76934707e-01 -4.19896066e-01 -1.11956127e-01
-1.02082580e-01 -2.73472786e-01 -4.95881051e-01 -4.29731160e-01
-2.72429883e-01 -8.87449205e-01 -4.12180156e-01 3.87281507e-01
-4.00465578e-01 -9.11839455e-02 6.17971718e-01 -8.41428459e-01
6.56686246e-01 -2.14759493e+00 -5.32039627e-02 1.47203639e-01
6.99361920e-01 2.81931579e-01 -9.78363305e-02 -5.02158888e-02
9.91717502e-02 2.24524871e-01 -2.14646280e-01 -3.19688857e-01
-6.11239523e-02 -5.63015826e-02 2.36677751e-01 8.83403540e-01
-1.06202094e-02 1.27316499e+00 -8.34673464e-01 -8.40800822e-01
1.82662442e-01 6.91070616e-01 -4.15191770e-01 1.05929919e-01
-7.35623613e-02 4.44225848e-01 -8.29457119e-02 9.66634154e-01
4.94777471e-01 -5.64896166e-01 1.36722356e-01 -4.72536027e-01
2.22893685e-01 -3.17270756e-01 -3.30374569e-01 1.38864934e+00
2.39652954e-03 9.04349446e-01 3.78262192e-01 -9.70437348e-01
3.79098743e-01 5.09379208e-01 7.67081261e-01 -5.48174918e-01
5.55041134e-01 5.42038560e-01 2.34209821e-01 -3.92340600e-01
1.28393071e-02 -5.48114121e-01 1.53939471e-01 2.79587626e-01
5.23260176e-01 -5.02747416e-01 6.75259709e-01 1.27192393e-01
1.40835059e+00 -5.42557359e-01 2.74571210e-01 -3.29302341e-01
2.31650501e-01 3.31731290e-01 5.04934072e-01 3.69196713e-01
-6.24882817e-01 9.37475741e-01 8.55104148e-01 -3.18157166e-01
-1.30950880e+00 -8.75786006e-01 -5.12466848e-01 5.93747199e-01
-2.30111614e-01 -4.88748737e-02 -9.27475214e-01 -7.84552455e-01
-1.33084550e-01 4.03522216e-02 -1.26180840e+00 2.16659904e-01
-1.95088863e-01 -8.76811743e-01 9.12608683e-01 5.25964916e-01
4.21646804e-01 -1.06551898e+00 -5.60737014e-01 2.77993858e-01
-2.80497760e-01 -1.07013273e+00 -5.44903040e-01 4.98847067e-01
-6.11488521e-01 -1.26651382e+00 -1.29599905e+00 -1.04193091e+00
8.52044702e-01 -1.15045056e-01 1.11076522e+00 1.76829442e-01
-8.80899906e-01 2.73273617e-01 -6.14897907e-02 -3.12437087e-01
-5.01958013e-01 3.27154130e-01 -5.01364291e-01 -3.85625631e-01
3.03510159e-01 -3.19843382e-01 -1.01311743e+00 1.05171785e-01
-9.94918287e-01 1.82065517e-01 7.91447163e-01 1.30673873e+00
1.25056159e+00 -1.73421785e-01 3.37316662e-01 -1.14388037e+00
1.70183495e-01 -2.00176403e-01 -5.38861215e-01 1.89912528e-01
-2.26507217e-01 -5.42579174e-01 3.80213052e-01 -1.92077622e-01
-3.81924957e-01 1.55935913e-01 -4.12874520e-01 -3.97466362e-01
-1.37585774e-01 5.01238346e-01 4.84357625e-01 -2.67773092e-01
6.24690890e-01 2.47497275e-01 4.63733763e-01 2.54701346e-01
-1.44620268e-02 4.58845675e-01 7.74040878e-01 1.82423852e-02
3.98555785e-01 7.50189066e-01 1.20354652e-01 -7.13384569e-01
-6.67513371e-01 -8.96362662e-01 -5.17348647e-01 -3.43068510e-01
1.18736267e+00 -7.75268137e-01 -6.47937596e-01 8.23520660e-01
-7.63184428e-01 -9.78478312e-01 -4.03414458e-01 1.99357122e-01
-6.57774627e-01 1.62024096e-01 -1.20065176e+00 -9.70136672e-02
-9.07323301e-01 -1.32099652e+00 1.19972360e+00 3.16615969e-01
-2.07836404e-01 -1.07262492e+00 4.19684082e-01 5.40423214e-01
4.25504506e-01 5.15574336e-01 1.01947463e+00 -7.21138000e-01
-4.00960892e-01 -2.96136260e-01 -4.05357271e-01 3.86662334e-01
1.39640108e-01 2.77549922e-01 -9.41259623e-01 -4.53441322e-01
-5.11470735e-01 -6.50423765e-01 1.26564109e+00 9.18594062e-01
1.02326882e+00 -1.31190894e-02 -7.33662009e-01 8.30771446e-01
1.50687850e+00 -1.22952461e-01 7.80781448e-01 2.86884129e-01
6.13432586e-01 4.64047819e-01 3.41260582e-01 -6.47712275e-02
2.81297952e-01 2.64222831e-01 4.01941776e-01 -9.43081498e-01
-3.27472687e-01 3.23906571e-01 -3.11787933e-01 5.27080595e-01
1.08026206e-01 -1.24701083e-01 -1.24715185e+00 1.18474948e+00
-1.46665013e+00 -8.10459137e-01 -1.18189543e-01 1.50962055e+00
1.18233156e+00 -9.82579738e-02 2.62638181e-01 9.16195735e-02
6.57816589e-01 -3.28248799e-01 -7.01023340e-01 -1.01367004e-01
1.47902757e-01 3.82854313e-01 5.42795777e-01 1.03340410e-01
-1.29737484e+00 7.97713578e-01 6.20400858e+00 1.18237007e+00
-1.43382454e+00 2.03796476e-01 1.37700713e+00 -1.99379120e-02
2.28383839e-02 -4.82771039e-01 -5.45283914e-01 3.97591144e-01
5.87858379e-01 4.17628437e-02 6.91878423e-02 4.91757601e-01
6.24295883e-02 -2.15976700e-01 -9.89342570e-01 8.10120106e-01
-2.10146278e-01 -1.82075381e+00 -2.49207601e-01 4.48500782e-01
7.04914033e-01 5.48103333e-01 9.61448401e-02 3.28374624e-01
1.04888543e-01 -1.28078532e+00 3.35691333e-01 2.58101821e-01
1.40240705e+00 -5.38940966e-01 1.33894861e+00 3.12961545e-03
-8.75477970e-01 3.79872978e-01 -3.49323273e-01 6.49935305e-01
-1.37161657e-01 8.21290374e-01 -1.36570036e+00 -8.11695158e-02
7.42446721e-01 7.78599441e-01 -6.40244842e-01 1.49096978e+00
2.13549197e-01 9.33193386e-01 -4.40393299e-01 -3.15326154e-02
3.40995997e-01 8.17685500e-02 1.03101783e-01 1.49015951e+00
3.02737713e-01 9.78995562e-02 -1.76871195e-01 9.43640113e-01
-3.13399643e-01 -7.84193277e-02 -1.56953499e-01 -2.63014704e-01
1.50865659e-01 2.05213881e+00 -1.56962502e+00 -3.08681488e-01
-3.33930887e-02 5.37151933e-01 3.00031781e-01 2.98156053e-01
-8.75294566e-01 -3.11197013e-01 3.26657385e-01 2.45303690e-01
5.07279634e-01 3.03758085e-01 -3.07227373e-01 -6.10228002e-01
-6.13252163e-01 -6.06937408e-01 3.96077454e-01 -4.93262112e-01
-1.43351316e+00 4.39780056e-01 -5.78571975e-01 -1.04785275e+00
6.54928088e-02 -6.84166431e-01 -6.13994896e-01 5.05499482e-01
-1.42390656e+00 -1.50276828e+00 -3.61275047e-01 2.90960699e-01
2.37325743e-01 2.18539864e-01 7.72400200e-01 1.57523096e-01
-4.97400135e-01 6.29827619e-01 4.48705822e-01 4.65303063e-01
6.14180863e-01 -1.71628428e+00 3.49411219e-02 4.01800603e-01
-2.57303476e-01 3.79211873e-01 4.39987123e-01 -2.16530427e-01
-9.47883785e-01 -1.33576453e+00 5.84419310e-01 -2.40327746e-01
6.45414531e-01 -3.60045701e-01 -8.08664739e-01 7.39732027e-01
3.70248884e-01 5.52103877e-01 1.06812334e+00 -2.99120456e-01
9.63295847e-02 3.83592024e-02 -1.50272024e+00 5.41016817e-01
3.30404490e-01 -4.68130082e-01 -1.32192954e-01 4.41304922e-01
2.41790980e-01 -6.70938611e-01 -1.33467829e+00 3.21141601e-01
5.12752175e-01 -8.77400756e-01 6.29803598e-01 3.05025935e-01
5.18828392e-01 -3.12601566e-01 2.52800584e-01 -1.21080017e+00
-4.96514767e-01 -1.00563005e-01 3.77308577e-01 1.01716805e+00
3.86554509e-01 -6.33157194e-01 1.02033460e+00 1.71964958e-01
-4.76220191e-01 -1.23820412e+00 -1.30304527e+00 -1.08585089e-01
4.82238114e-01 7.78744966e-02 1.43298477e-01 8.16632271e-01
1.08227812e-01 2.38607869e-01 4.07570899e-01 -3.59845400e-01
5.68400264e-01 -9.32805017e-02 5.77223897e-01 -8.62902701e-01
-1.20006211e-01 -8.25201094e-01 -6.59281373e-01 -5.14006436e-01
5.02962023e-02 -1.21857536e+00 1.10524476e-01 -1.92679679e+00
7.02066541e-01 -2.97402322e-01 -3.83140475e-01 8.41928959e-01
-8.95728245e-02 1.04384887e+00 -2.10198089e-01 2.20926791e-01
-7.39516139e-01 2.45109990e-01 1.65175223e+00 -6.95715308e-01
3.83081615e-01 -3.34141523e-01 -6.01269782e-01 3.76976788e-01
6.42925739e-01 -3.36500764e-01 -1.44814163e-01 -1.39026446e-02
-1.20375399e-02 -7.14660361e-02 5.39583921e-01 -1.27318466e+00
2.80166566e-01 2.97485977e-01 6.38052762e-01 -8.22302759e-01
1.79673821e-01 -5.15942037e-01 1.47607133e-01 7.47361064e-01
-3.57115269e-01 -4.93234992e-01 3.84858966e-01 3.21372032e-01
-2.60446310e-01 1.36764959e-01 1.02529347e+00 -2.72992879e-01
-3.32685977e-01 4.41529989e-01 -7.26585388e-01 6.48278464e-03
1.09235489e+00 -6.62757814e-01 -8.16051185e-01 5.71395159e-02
-6.05639875e-01 4.29717064e-01 6.20664060e-01 -4.45226341e-01
4.38445300e-01 -1.02817404e+00 -8.07085156e-01 -1.55996501e-01
2.54866898e-01 3.99503142e-01 6.43488228e-01 1.48932207e+00
-1.07214749e+00 3.57467830e-01 -2.86836803e-01 -1.23598826e+00
-1.21379101e+00 1.39050901e-01 9.34684098e-01 -7.19377279e-01
-5.16433179e-01 1.16412997e+00 5.80568969e-01 -6.02672577e-01
-2.93103158e-01 -7.41473556e-01 -2.18212858e-01 -6.66126460e-02
2.55090326e-01 1.96777992e-02 2.74214894e-01 -5.27578652e-01
-3.05394918e-01 4.75642651e-01 -3.81456584e-01 2.73027629e-01
1.21560264e+00 2.76710093e-01 -5.02767801e-01 3.20909709e-01
1.26935291e+00 -4.95894343e-01 -1.30644703e+00 1.01135932e-01
-2.84979373e-01 6.20928928e-02 3.14823866e-01 -9.50604200e-01
-1.43168044e+00 6.77535117e-01 7.98743725e-01 2.70231999e-02
9.93796527e-01 2.53816128e-01 9.97934699e-01 -6.66227145e-03
1.11553133e-01 -8.54184806e-01 -4.35826369e-02 4.48183745e-01
3.69825572e-01 -1.20334971e+00 -1.87975779e-01 -3.60528827e-01
-3.73269469e-01 9.20388758e-01 5.94802916e-01 -5.43370843e-02
4.40519691e-01 5.83773196e-01 3.49023789e-01 -5.00142515e-01
-7.77528822e-01 -1.91336006e-01 -8.76944221e-04 7.17709005e-01
5.87021589e-01 1.00480482e-01 -1.46872893e-01 5.51965773e-01
-4.02096622e-02 3.21534932e-01 5.50082445e-01 6.54050589e-01
-1.89722687e-01 -5.02132058e-01 3.42980772e-03 9.20761585e-01
-8.09403896e-01 -1.25054864e-03 -4.11997914e-01 8.33693027e-01
7.63843507e-02 3.01136643e-01 2.42326677e-01 -1.01287663e-01
-3.97916622e-02 -5.15217841e-01 5.64728796e-01 -4.87872422e-01
-9.64141011e-01 3.86130273e-01 -1.39250651e-01 -2.34873563e-01
-3.94923061e-01 -5.38893342e-01 -1.57620239e+00 -2.74001569e-01
-2.44893774e-01 2.64067091e-02 2.07546636e-01 8.03090453e-01
1.57632828e-02 7.53240466e-01 2.01036662e-01 -8.34675431e-01
-4.14391458e-02 -1.07628262e+00 -9.67170000e-01 2.20709011e-01
4.16150570e-01 -2.99554199e-01 -3.28541607e-01 2.85996050e-01] | [15.01533317565918, -3.1032397747039795] |
b5697012-d4f1-4143-9488-af9348367da8 | a-learning-exploring-method-to-generate | null | null | https://aclanthology.org/2020.coling-main.209 | https://aclanthology.org/2020.coling-main.209.pdf | A Learning-Exploring Method to Generate Diverse Paraphrases with Multi-Objective Deep Reinforcement Learning | Paraphrase generation (PG) is of great importance to many downstream tasks in natural language processing. Diversity is an essential nature to PG for enhancing generalization capability and robustness of downstream applications. Recently, neural sequence-to-sequence (Seq2Seq) models have shown promising results in PG. However, traditional model training for PG focuses on optimizing model prediction against single reference and employs cross-entropy loss, which objective is unable to encourage model to generate diverse paraphrases. In this work, we present a novel approach with multi-objective learning to PG. We propose a learning-exploring method to generate sentences as learning objectives from the learned data distribution, and employ reinforcement learning to combine these new learning objectives for model training. We first design a sample-based algorithm to explore diverse sentences. Then we introduce several reward functions to evaluate the sampled sentences as learning signals in terms of expressive diversity and semantic fidelity, aiming to generate diverse and high-quality paraphrases. To effectively optimize model performance satisfying different evaluating aspects, we use a GradNorm-based algorithm that automatically balances these training objectives. Experiments and analyses on Quora and Twitter datasets demonstrate that our proposed method not only gains a significant increase in diversity but also improves generation quality over several state-of-the-art baselines. | ['Yufeng Chen', 'Jinan Xu', 'Changjian Hu', 'Yao Meng', 'Yujie Zhang', 'Deyi Xiong', 'Erguang Yang', 'Mingtong Liu'] | 2020-12-01 | null | null | null | coling-2020-8 | ['paraphrase-generation', 'paraphrase-generation'] | ['computer-code', 'natural-language-processing'] | [ 3.58608842e-01 -1.28028274e-01 -3.00014228e-01 -4.96035486e-01
-1.16398752e+00 -3.71875316e-01 6.55006707e-01 -8.25890824e-02
-2.99161971e-01 1.10835123e+00 5.23527801e-01 5.47338836e-03
-2.04605777e-02 -9.01371181e-01 -7.58050740e-01 -5.98643959e-01
2.74863750e-01 3.67350787e-01 -1.18159287e-01 -5.16206503e-01
5.39909184e-01 1.31515801e-01 -1.43845952e+00 4.47189778e-01
1.50343239e+00 8.09344888e-01 4.79237765e-01 7.02070177e-01
-2.40384877e-01 7.08575845e-01 -8.63396585e-01 -4.95778352e-01
1.65132582e-01 -1.07433319e+00 -7.63997138e-01 -3.75531524e-01
2.78390627e-02 -3.29121202e-02 -8.40765540e-04 9.55723584e-01
1.10325456e+00 2.74705201e-01 7.61476994e-01 -9.37719882e-01
-9.48496342e-01 8.57075274e-01 -2.96788841e-01 2.53545821e-01
4.32840198e-01 4.83045638e-01 1.25777483e+00 -8.35691810e-01
6.12207651e-01 1.28385532e+00 4.72029179e-01 9.00493622e-01
-1.13819075e+00 -6.05716944e-01 -3.49701643e-02 1.69196859e-01
-1.03564143e+00 -5.12308836e-01 8.13530505e-01 -7.10356161e-02
9.26308692e-01 3.58668536e-01 6.12075806e-01 1.41711545e+00
3.78426015e-01 1.19215333e+00 9.43968058e-01 -3.40911210e-01
4.81110454e-01 1.85949609e-01 -1.89605862e-01 3.45346987e-01
-1.24329649e-01 2.80810028e-01 -8.35318744e-01 1.26321018e-02
3.20681572e-01 -3.75083178e-01 -2.56360918e-01 8.25872496e-02
-1.12724817e+00 9.78778899e-01 5.61479092e-01 1.78275287e-01
-4.06474352e-01 4.79351021e-02 3.72831821e-01 2.25662023e-01
3.21877658e-01 9.71025348e-01 -1.81123734e-01 -4.66683209e-01
-9.73765314e-01 5.20546854e-01 6.40961170e-01 8.86425495e-01
4.96135384e-01 1.96674526e-01 -1.01693630e+00 1.27203512e+00
5.81579879e-02 6.06511116e-01 1.02030134e+00 -8.45068634e-01
8.07072103e-01 4.69650388e-01 -1.37008175e-01 -7.76421726e-01
2.63876375e-03 -7.63224185e-01 -9.37936127e-01 -1.69160679e-01
-5.97741678e-02 -2.00795054e-01 -5.65836489e-01 1.89035261e+00
1.06760807e-01 -3.81873064e-02 2.94463187e-01 7.46658683e-01
8.51527631e-01 9.18523133e-01 1.01940371e-01 -2.12993726e-01
8.88188183e-01 -1.14168596e+00 -6.89253509e-01 -1.74058065e-01
5.69318533e-01 -4.99505520e-01 1.48068190e+00 2.32179791e-01
-1.34710813e+00 -7.03043282e-01 -9.91622448e-01 6.21834844e-02
-9.09967050e-02 5.75294311e-04 1.21331595e-01 5.26072264e-01
-8.33160877e-01 9.10176694e-01 -4.34791833e-01 1.85439773e-02
6.27895594e-01 1.44407213e-01 2.28886858e-01 1.98268920e-01
-1.62548852e+00 9.47869003e-01 6.46658063e-01 -2.61048675e-02
-8.19653034e-01 -7.10978329e-01 -8.24625909e-01 2.82358736e-01
-4.26902324e-02 -1.09052253e+00 1.26437163e+00 -1.09586430e+00
-2.02703977e+00 4.25869912e-01 -1.77420422e-01 -6.29210949e-01
6.59309030e-01 -1.71006992e-01 -2.53545940e-01 3.46960537e-02
1.21011138e-01 1.00339997e+00 7.57733226e-01 -1.19915700e+00
-4.82921153e-01 7.54233897e-02 -1.09636247e-01 5.54847062e-01
-6.81043923e-01 -2.23063052e-01 -4.63014208e-02 -8.54937017e-01
-4.55376625e-01 -5.92557907e-01 -2.82964706e-01 -3.35960567e-01
-5.22838950e-01 -1.44704685e-01 2.86788404e-01 -6.68837309e-01
1.63319421e+00 -1.58356190e+00 3.32427859e-01 5.90957664e-02
-1.99382663e-01 6.23971581e-01 -4.10879105e-01 5.81889093e-01
4.04283941e-01 2.08297089e-01 -5.57442605e-01 -3.74704629e-01
5.93812279e-02 -1.83387101e-02 -3.92796308e-01 -1.94453016e-01
5.15548110e-01 1.36019778e+00 -1.21594310e+00 -5.48434019e-01
-1.15797646e-01 2.84387797e-01 -8.18411350e-01 4.80723888e-01
-5.33542693e-01 5.85722625e-01 -5.35151541e-01 4.71657753e-01
4.47056144e-01 -1.24957047e-01 -1.43491969e-01 2.81352520e-01
1.01764157e-01 2.67646253e-01 -6.47242486e-01 1.76773739e+00
-8.97702932e-01 2.46599033e-01 -4.96298134e-01 -9.17268276e-01
1.27649522e+00 -6.18506782e-02 -2.70808730e-02 -9.95578170e-01
4.87824231e-02 3.44065577e-01 1.14653707e-02 -6.91552699e-01
7.08869755e-01 -2.69477457e-01 -8.98008272e-02 3.45659882e-01
-1.37352616e-01 -4.63139981e-01 3.69225681e-01 9.07931104e-02
8.72191787e-01 2.55528837e-01 2.11863220e-01 -1.43245965e-01
7.79633045e-01 -2.49085188e-01 5.94821572e-01 7.83632457e-01
-9.99157205e-02 7.45776236e-01 5.39423525e-01 1.24223277e-01
-1.10204399e+00 -1.11369145e+00 1.56023279e-01 9.19221759e-01
9.82123390e-02 -1.45899609e-01 -7.93329358e-01 -6.43352985e-01
-1.31212145e-01 1.16467965e+00 -4.25398290e-01 -7.10725963e-01
-6.28620327e-01 -7.83222198e-01 6.78695679e-01 4.05668795e-01
6.86811090e-01 -1.50298679e+00 -3.21420997e-01 2.80069321e-01
-4.29050177e-01 -6.57229900e-01 -7.47325301e-01 -2.35639051e-01
-7.99058795e-01 -4.60758299e-01 -9.57134604e-01 -8.40960503e-01
4.54833448e-01 -7.55249932e-02 1.13050961e+00 -1.19994104e-01
2.05069110e-02 -2.41153121e-01 -4.67179954e-01 -3.97283971e-01
-8.75463486e-01 5.08402407e-01 -2.51948982e-01 3.30688991e-02
8.33481550e-02 -6.03164196e-01 -7.54849374e-01 3.70400250e-02
-1.00381136e+00 1.81663275e-01 8.86947334e-01 1.26910985e+00
5.62343299e-01 -4.16881859e-01 1.36948466e+00 -7.70142674e-01
1.36922693e+00 -7.01349080e-01 -1.45334959e-01 4.32106793e-01
-6.79179251e-01 3.84777278e-01 1.22523165e+00 -3.79575878e-01
-1.17830801e+00 -1.66724220e-01 -3.88699979e-01 -4.25325274e-01
1.31251678e-01 6.58451498e-01 -2.57501245e-01 3.91012311e-01
9.47208524e-01 7.83891916e-01 2.32846409e-01 -3.25691886e-02
4.99039948e-01 9.16651905e-01 1.54158011e-01 -7.06932306e-01
4.99802113e-01 -4.01154682e-02 -1.71844065e-01 -6.60053253e-01
-8.57126296e-01 -1.57039732e-01 -2.39239335e-01 -1.26694351e-01
6.04847372e-01 -8.23861420e-01 -5.11895895e-01 3.85577083e-01
-1.07905316e+00 -4.15187448e-01 -3.73502582e-01 2.67557412e-01
-8.66773605e-01 2.94779360e-01 -3.01804632e-01 -7.73670614e-01
-8.51536155e-01 -1.07237065e+00 1.12128353e+00 4.54105824e-01
-2.94361353e-01 -9.37528968e-01 3.32226068e-01 4.64694500e-01
7.09166467e-01 1.02810778e-01 7.38871753e-01 -6.68977082e-01
-4.32985067e-01 1.49695575e-01 3.14497836e-02 5.71178734e-01
8.62363204e-02 -1.42496198e-01 -7.41110325e-01 -1.86699897e-01
-1.33851692e-01 -6.27919853e-01 8.72925520e-01 2.70371616e-01
1.42424119e+00 -5.59295297e-01 -1.03102393e-01 6.05232298e-01
1.19092298e+00 1.47431985e-01 6.93293273e-01 3.04159641e-01
3.35182786e-01 6.38945758e-01 7.90535867e-01 5.16626954e-01
1.99458167e-01 6.61076725e-01 9.82183591e-02 2.59211183e-01
-1.69623435e-01 -7.04537570e-01 5.82911134e-01 8.67075026e-01
3.94223124e-01 -5.00714004e-01 -6.70035958e-01 5.91618776e-01
-1.92047524e+00 -1.35220480e+00 2.88039923e-01 2.02457356e+00
1.39221334e+00 2.12666094e-01 2.04799324e-01 -1.81991816e-01
8.74348938e-01 1.40621409e-01 -8.74415994e-01 -6.83776975e-01
-2.14008525e-01 3.01940620e-01 -7.09534883e-02 4.78117377e-01
-7.18450606e-01 1.02959538e+00 5.83248663e+00 1.29402602e+00
-1.12330759e+00 -2.60179430e-01 9.16411877e-01 -4.96687829e-01
-8.03166807e-01 -3.04532588e-01 -9.23980117e-01 8.99728119e-01
1.06045651e+00 -5.95468223e-01 5.23071647e-01 7.23442912e-01
7.21437931e-01 2.84379780e-01 -9.26168740e-01 7.97580540e-01
1.12449229e-01 -1.47242177e+00 4.66107756e-01 -2.90421605e-01
1.09031832e+00 -3.16268355e-01 2.14948401e-01 6.21185482e-01
2.88196981e-01 -1.18102491e+00 7.03322530e-01 6.89136505e-01
6.31265759e-01 -9.23569977e-01 6.03538334e-01 7.01344311e-01
-7.31829882e-01 -2.64884233e-01 -4.33817923e-01 6.23367876e-02
4.70971584e-01 6.98523760e-01 -9.04326200e-01 6.13370836e-01
1.09466024e-01 7.97471762e-01 -4.65081662e-01 9.82407451e-01
-3.82528186e-01 6.27316892e-01 6.52853698e-02 -8.10386956e-01
2.52408266e-01 -2.96406001e-01 6.93804145e-01 1.35104430e+00
4.84134942e-01 -1.71564609e-01 -1.46555930e-01 1.20045483e+00
-3.33635449e-01 3.56919259e-01 -6.02707565e-01 -8.39740336e-02
7.83985078e-01 9.33633924e-01 -1.20998569e-01 -3.62013370e-01
8.42178091e-02 1.02770102e+00 5.50828576e-01 1.67997524e-01
-9.00465667e-01 -7.28766680e-01 4.83871520e-01 -2.24737495e-01
2.43859634e-01 1.36926368e-01 -5.28684020e-01 -1.11292791e+00
1.73135743e-01 -1.12871182e+00 1.13577865e-01 -4.92892146e-01
-1.54467452e+00 6.93878770e-01 -7.02585578e-02 -1.49836290e+00
-6.08649135e-01 1.88280419e-02 -8.31960261e-01 1.09132063e+00
-1.69064569e+00 -1.07825971e+00 -1.17394030e-01 2.47053891e-01
9.09357309e-01 -4.60178465e-01 4.72999215e-01 8.26055277e-03
-5.18009186e-01 9.78713632e-01 4.18391734e-01 -2.25441754e-01
6.19380951e-01 -1.10119987e+00 4.89799500e-01 7.07549572e-01
-2.89090909e-02 6.27748787e-01 6.76176429e-01 -6.22042060e-01
-1.04751754e+00 -1.38303912e+00 1.07646143e+00 -2.24289685e-01
3.05471420e-01 -2.94029176e-01 -7.02577949e-01 6.80937394e-02
3.04059774e-01 -4.90366101e-01 7.03940809e-01 -2.58709431e-01
-1.14337936e-01 -1.78458944e-01 -1.14982677e+00 8.80849481e-01
1.26262534e+00 -2.03398094e-01 -5.38178980e-01 2.53944546e-01
9.02323902e-01 -4.07926351e-01 -6.32388234e-01 4.05695438e-01
4.94946897e-01 -9.31568205e-01 8.55706036e-01 -7.47119486e-01
1.11270666e+00 3.60428505e-02 2.10690349e-02 -1.90855026e+00
-2.40460724e-01 -7.38954902e-01 -1.59029573e-01 1.43887758e+00
6.09528899e-01 -5.09298146e-01 8.43875766e-01 2.17100754e-01
-3.80116731e-01 -1.27779222e+00 -7.42478132e-01 -1.03695321e+00
5.14962137e-01 5.63579388e-02 8.02051544e-01 6.83466196e-01
1.73638746e-01 6.01892531e-01 -3.70758563e-01 -3.57891649e-01
3.18356186e-01 1.88574821e-01 7.08963335e-01 -5.95571160e-01
-6.05644524e-01 -8.96347761e-01 5.23310602e-02 -1.29887080e+00
3.97921205e-01 -1.23666573e+00 2.11355299e-01 -1.51838732e+00
3.13193381e-01 -3.71881604e-01 -8.25893059e-02 -8.06920230e-03
-7.00745463e-01 -1.67234436e-01 2.57584214e-01 6.95405975e-02
-5.44331849e-01 1.34663045e+00 1.51012242e+00 -1.81671008e-01
-4.59440619e-01 2.45391890e-01 -9.72454786e-01 2.42083035e-02
1.07870102e+00 -2.95641094e-01 -6.86685085e-01 -4.26415503e-01
2.35764682e-01 1.23357326e-01 -2.91256397e-03 -9.42131162e-01
-4.38651592e-02 -3.08563143e-01 1.78667620e-01 -3.44537526e-01
2.14602947e-01 -1.70080677e-01 -5.98470755e-02 5.91630936e-01
-1.01050663e+00 -1.07456921e-02 -8.50571785e-03 5.89851260e-01
-4.00216043e-01 -5.66126168e-01 8.49839926e-01 -2.57979572e-01
-3.04600626e-01 2.68430829e-01 -9.91407931e-02 4.74150747e-01
9.64744449e-01 -1.24670900e-01 -2.58182019e-01 -5.17972827e-01
-2.97757417e-01 6.62649393e-01 1.62181154e-01 6.83386505e-01
7.31067240e-01 -1.52850592e+00 -1.16609454e+00 2.48183981e-02
1.18586369e-01 -1.01975992e-01 2.57350206e-01 2.88151652e-01
-5.27759612e-01 4.23516810e-01 -2.03715175e-01 -4.68013078e-01
-1.05736983e+00 2.40215898e-01 2.70230949e-01 -7.08108485e-01
4.41005267e-02 9.40930545e-01 -6.07713275e-02 -7.39654124e-01
1.30824931e-02 -2.68706307e-02 -3.42813253e-01 1.14021562e-02
4.68697757e-01 4.59963292e-01 7.67472684e-02 -2.89425462e-01
2.99538728e-02 2.53087789e-01 -1.05758503e-01 -2.58566648e-01
1.10550308e+00 -7.89824575e-02 2.77690917e-01 2.21180767e-01
1.42648411e+00 -1.78292125e-01 -1.19878328e+00 -1.29434541e-01
-1.99847370e-01 -4.68307644e-01 -3.75570297e-01 -9.39923763e-01
-7.63961315e-01 9.90791976e-01 2.12805733e-01 6.69685304e-02
1.04181266e+00 -2.74464011e-01 1.16081560e+00 3.40628892e-01
1.53807193e-01 -1.23613107e+00 6.33305490e-01 6.39201462e-01
1.18130136e+00 -1.16382980e+00 -3.00213844e-01 -1.57764535e-02
-1.09803760e+00 8.89860392e-01 8.30257893e-01 7.83193915e-04
1.93385556e-01 -2.23245367e-01 -3.70339677e-02 3.80523324e-01
-9.53245640e-01 2.11776793e-03 1.67726427e-01 5.62670469e-01
4.44581687e-01 -1.37546718e-01 -7.63320565e-01 5.71484923e-01
-6.69659615e-01 8.41851160e-02 2.46939912e-01 6.52918398e-01
-6.00491583e-01 -1.36149120e+00 -2.10273992e-02 6.35185599e-01
-2.78885007e-01 -3.86963189e-01 -3.34227473e-01 2.36787215e-01
-1.03096537e-01 9.12817180e-01 -1.37530074e-01 -4.59706932e-01
3.62002164e-01 -2.58909445e-02 2.03216329e-01 -5.86784899e-01
-9.18017745e-01 -3.85440946e-01 1.82048112e-01 -3.33768338e-01
-7.02493787e-02 -6.41172528e-01 -1.20182085e+00 -1.29097462e-01
-3.75404537e-01 2.71698952e-01 5.24963260e-01 7.80239582e-01
7.84779429e-01 6.31749809e-01 1.19153798e+00 -6.31789505e-01
-1.22869813e+00 -1.05240011e+00 -2.15438694e-01 5.93369842e-01
1.03470176e-01 -1.73182413e-01 -3.55653733e-01 -8.36471170e-02] | [11.866119384765625, 9.228275299072266] |
daa99916-6562-4784-a737-927e5250fdd4 | binary-coding-for-partial-action-analysis | null | null | http://openaccess.thecvf.com/content_cvpr_2017/html/Qin_Binary_Coding_for_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Qin_Binary_Coding_for_CVPR_2017_paper.pdf | Binary Coding for Partial Action Analysis With Limited Observation Ratios | Traditional action recognition methods aim to recognize actions with complete observations/executions. However, it is often difficult to capture fully executed actions due to occlusions, interruptions, etc. Meanwhile, action prediction/recognition in advance based on partial observations is essential for preventing the situation from deteriorating. Besides, fast spotting human activities using partially observed data is a critical ingredient for retrieval systems. Inspired by the recent success of data binarization in efficient retrieval/recognition, we propose a novel approach, named Partial Reconstructive Binary Coding (PRBC), for action analysis based on limited frame glimpses during any period of the complete execution. Specifically, we learn discriminative compact binary codes for partial actions via a joint learning framework, which collaboratively tackles feature reconstruction as well as binary coding. We obtain the solution to PRBC based on a discrete alternating iteration algorithm. Extensive experiments on four realistic action datasets in terms of three tasks (i.e., partial action retrieval, recognition and prediction) clearly show the superiority of PRBC over the state-of-the-art methods, along with significantly reduced memory load and computational costs during the online test.
| ['Bingbing Ni', 'Jie Qin', 'Chen Chen', 'Yunhong Wang', 'Li Liu', 'Fumin Shen', 'Ling Shao'] | 2017-07-01 | null | null | null | cvpr-2017-7 | ['action-analysis'] | ['computer-vision'] | [ 7.25139081e-01 -5.60464799e-01 -5.01057744e-01 -1.44495666e-01
-8.77985060e-01 -1.05996870e-01 5.09441972e-01 3.31762768e-02
-2.97061265e-01 6.07449651e-01 2.19341710e-01 8.05227309e-02
-3.97196412e-01 -3.46108735e-01 -4.71609622e-01 -8.57212961e-01
6.10647351e-02 3.00788283e-01 1.45704076e-01 2.90032983e-01
4.01596397e-01 6.35303259e-01 -1.90215945e+00 4.16721374e-01
7.11768985e-01 1.28537714e+00 2.82956988e-01 5.27602732e-01
1.46672383e-01 1.24621713e+00 -5.24663329e-01 -6.46484643e-02
1.03938602e-01 -4.22887266e-01 -5.03660619e-01 7.54782259e-01
-1.11134395e-01 -7.47264624e-01 -6.71080053e-01 7.25087762e-01
2.88682103e-01 2.77908921e-01 4.05629456e-01 -1.32897794e+00
-1.68705091e-01 7.61433467e-02 -5.81437886e-01 2.85175532e-01
7.27650046e-01 1.71128497e-01 9.38514948e-01 -6.63281262e-01
3.77976805e-01 1.07772231e+00 2.08953008e-01 3.35851550e-01
-9.49202061e-01 -4.72066224e-01 2.31289864e-01 8.15731823e-01
-1.51019740e+00 -7.29159117e-01 7.63827503e-01 -2.71923393e-01
7.08928585e-01 4.44305599e-01 6.61515832e-01 9.96035457e-01
-8.93356353e-02 1.54153228e+00 8.29420924e-01 -2.98438042e-01
2.45507330e-01 -3.96649241e-01 -1.02132790e-01 5.36264002e-01
-2.16946881e-02 7.72665143e-02 -7.62561321e-01 -1.33639231e-01
5.55259109e-01 6.64121747e-01 -4.36575204e-01 -3.16801339e-01
-1.20887780e+00 3.28561991e-01 -2.57100135e-01 7.43303299e-02
-7.67721117e-01 7.17569888e-02 4.85126317e-01 1.56408891e-01
1.70582041e-01 -2.36494929e-01 -2.04815477e-01 -8.54507864e-01
-9.81856287e-01 1.25636548e-01 4.91773993e-01 8.78196895e-01
6.34822607e-01 -4.16732915e-02 -5.07452905e-01 7.63238847e-01
1.12237751e-01 5.84578216e-01 6.25489831e-01 -1.00023568e+00
5.38537562e-01 7.57277250e-01 2.80009627e-01 -1.21074808e+00
-6.97579980e-02 -7.33321980e-02 -7.94979334e-01 -2.45951355e-01
2.37909481e-01 2.84614563e-01 -6.38824821e-01 1.30068290e+00
4.08802480e-01 5.55186987e-01 6.98464513e-02 9.41583395e-01
2.17676923e-01 5.06436229e-01 -1.22508973e-01 -6.15149915e-01
1.04071116e+00 -9.17169631e-01 -8.21880221e-01 -1.26671851e-01
7.56212950e-01 -6.56813383e-01 7.79198349e-01 7.35677600e-01
-7.11668372e-01 -6.39810562e-01 -9.46397126e-01 1.54728234e-01
1.83450505e-01 7.32810140e-01 7.64072180e-01 5.13746142e-01
-3.81872445e-01 5.19552946e-01 -1.30130506e+00 5.87023944e-02
4.81433213e-01 3.01560849e-01 -3.99782300e-01 -5.13731420e-01
-6.57952428e-01 4.20442075e-01 3.69166613e-01 2.51791775e-01
-1.08142126e+00 -1.05890162e-01 -7.07009256e-01 6.41511530e-02
9.60095763e-01 8.59513134e-02 1.13088226e+00 -8.55154216e-01
-1.49561107e+00 3.63404512e-01 -3.33073050e-01 -5.87279797e-01
4.60287452e-01 -6.68978512e-01 -3.92518342e-01 4.84929979e-01
-1.18000522e-01 1.66051567e-01 1.04406619e+00 -7.18060076e-01
-8.09109271e-01 -4.06053036e-01 1.07062206e-01 3.22539419e-01
-5.21552503e-01 6.75359592e-02 -7.86976397e-01 -4.54983711e-01
2.29003429e-01 -8.64506960e-01 -3.08753792e-02 2.94851176e-02
-2.10665599e-01 -2.36226186e-01 1.00047660e+00 -9.39743042e-01
1.39768982e+00 -2.21935606e+00 2.33061299e-01 9.58871618e-02
-1.58428669e-01 3.33621949e-01 4.20114547e-02 3.81618202e-01
2.55123936e-02 -6.29718006e-01 -6.02248348e-02 -3.76828492e-01
9.18268636e-02 3.48187447e-01 -4.02805865e-01 6.72818303e-01
-1.02674991e-01 7.49463379e-01 -9.22875702e-01 -7.38025606e-01
5.01315534e-01 3.29116762e-01 -2.92890310e-01 4.30534482e-01
-1.00450203e-01 5.21228313e-01 -7.38090992e-01 1.10466874e+00
3.33311617e-01 -2.36320600e-01 2.74701685e-01 4.59436746e-03
1.48326876e-02 9.55769978e-03 -1.29000127e+00 1.77802908e+00
-3.51931304e-01 3.79221022e-01 -9.27914083e-02 -1.47459412e+00
7.62135863e-01 4.67123955e-01 9.02049303e-01 -9.12476718e-01
-6.94909990e-02 5.95898591e-02 -4.20451522e-01 -6.27177000e-01
4.24409509e-01 4.48293418e-01 -6.79969266e-02 5.71994722e-01
-2.42674932e-01 4.61784959e-01 3.49422991e-01 -1.87420137e-02
1.45641756e+00 4.03755218e-01 4.63910341e-01 5.84908962e-01
7.57249236e-01 -1.81376487e-01 7.56322682e-01 6.27169013e-01
-2.99447894e-01 4.42503572e-01 4.49206859e-01 -4.19339538e-01
-6.52187288e-01 -4.96432394e-01 2.86132574e-01 9.55110729e-01
2.69828677e-01 -5.84686875e-01 -5.77708662e-01 -7.11603343e-01
-2.99387038e-01 4.34330881e-01 -2.85406053e-01 -3.62281591e-01
-6.25369549e-01 -3.94388139e-01 3.93197238e-01 4.82216001e-01
6.33306563e-01 -1.18944263e+00 -1.14471948e+00 3.14456820e-01
-4.32676286e-01 -1.31643534e+00 -4.08451349e-01 -1.50241956e-01
-8.18226516e-01 -1.33133399e+00 -8.20204794e-01 -3.08609545e-01
5.15950739e-01 7.61397183e-01 7.28110254e-01 3.61633271e-01
-4.22128975e-01 6.49568081e-01 -8.48766863e-01 1.28947034e-01
-3.27317603e-02 -2.94199914e-01 -2.91205849e-03 5.54159999e-01
4.89406526e-01 -5.77094555e-01 -7.10360646e-01 4.94566262e-01
-1.10105872e+00 1.84502870e-01 1.04395115e+00 8.35104346e-01
9.18681026e-01 4.10719633e-01 -7.18421722e-03 -3.91683370e-01
2.68620491e-01 -3.52414697e-01 -5.16114712e-01 6.93329632e-01
-3.03019553e-01 2.13455446e-02 4.15562868e-01 -5.59473753e-01
-9.79049087e-01 2.71404982e-01 2.12524027e-01 -8.07854950e-01
-3.22401404e-01 4.38456446e-01 -3.51585984e-01 1.39971301e-01
1.19358346e-01 9.75063980e-01 -1.06239125e-01 -5.53883851e-01
1.48096345e-02 9.09420550e-01 4.59369034e-01 -4.97244626e-01
4.49883103e-01 6.45306289e-01 -5.57200350e-02 -8.29105496e-01
-6.75543010e-01 -6.79639757e-01 -6.25897706e-01 -4.46246922e-01
5.66885591e-01 -9.50066805e-01 -9.06634390e-01 5.52097082e-01
-8.81254911e-01 -7.60365054e-02 -1.55574337e-01 7.20828772e-01
-8.22762251e-01 9.65360403e-01 -2.94864237e-01 -1.17185414e+00
-1.02251053e-01 -1.17642999e+00 1.36032867e+00 3.15554515e-02
4.91815358e-02 -2.98205286e-01 -8.76782089e-02 6.73034906e-01
-7.23710060e-02 8.32048133e-02 3.57977897e-01 -4.11045074e-01
-9.10141647e-01 -5.65094054e-01 -1.45944819e-01 4.34424788e-01
2.40031019e-01 -1.65754288e-01 -4.84055877e-01 -3.13615024e-01
7.72633702e-02 -6.19784653e-01 7.57174432e-01 2.32808795e-02
1.52111650e+00 -2.57914782e-01 -2.87590623e-01 2.73430139e-01
1.04606652e+00 4.24756140e-01 8.85471344e-01 1.56388938e-01
5.02008915e-01 1.92451909e-01 1.47189665e+00 1.16179478e+00
6.22516759e-02 9.95642841e-01 4.33994174e-01 4.48393434e-01
2.54475698e-02 -1.41401619e-01 5.37221849e-01 6.25858843e-01
-2.66420096e-01 -1.27637684e-01 -5.79545140e-01 3.84690315e-01
-2.16125941e+00 -9.84881461e-01 2.90234149e-01 2.41508746e+00
5.53489268e-01 9.68642905e-02 1.24811735e-02 6.21719122e-01
6.20854855e-01 4.55096334e-01 -6.57738924e-01 2.71966398e-01
7.16973841e-02 6.94191679e-02 2.83387870e-01 -2.25234516e-02
-1.17922926e+00 6.22580111e-01 4.99493742e+00 1.30848086e+00
-8.41351867e-01 2.35929061e-03 4.06578600e-01 -1.30206555e-01
2.92817205e-01 8.42153206e-02 -7.01098919e-01 7.09409356e-01
7.38583088e-01 2.13151425e-01 5.62023640e-01 1.04635859e+00
2.18638033e-01 -4.85353887e-01 -9.56525087e-01 1.42223072e+00
3.24550956e-01 -9.38385546e-01 -5.10328934e-02 4.11655270e-02
4.64694113e-01 -3.81725430e-01 -2.09490612e-01 2.78292835e-01
-1.14992961e-01 -6.47236288e-01 6.80023193e-01 8.05291772e-01
7.52319217e-01 -6.49887085e-01 4.98907417e-01 7.18214869e-01
-1.28350377e+00 -4.20100093e-01 -3.80222112e-01 -1.01242803e-01
2.28315353e-01 4.25817281e-01 -2.80049860e-01 6.19678378e-01
4.78926122e-01 1.08732224e+00 -3.91940325e-01 1.08522999e+00
-3.63217622e-01 5.39658487e-01 -2.65206814e-01 1.08066842e-01
-1.31532652e-02 -1.62034422e-01 3.61144960e-01 7.32548773e-01
4.23306078e-01 4.40131128e-01 4.55522209e-01 1.85558751e-01
3.09794992e-01 -2.89605316e-02 -3.00901353e-01 -3.72065246e-01
3.33920270e-01 8.70313168e-01 -6.78673685e-01 -3.59399855e-01
-5.46918809e-01 1.17930949e+00 1.99519888e-01 3.27353656e-01
-1.01780033e+00 -8.71407166e-02 4.41815168e-01 -2.85318941e-01
6.25517726e-01 -4.37082738e-01 2.34439269e-01 -1.35478055e+00
4.30271327e-01 -1.20336652e+00 4.45407212e-01 -5.75311303e-01
-5.50276697e-01 2.60193408e-01 -2.49445648e-03 -1.83970094e+00
-2.82746136e-01 -1.37342602e-01 -1.71832114e-01 2.47411907e-01
-1.41600573e+00 -8.42338979e-01 -4.88766640e-01 8.32668066e-01
8.24531674e-01 -6.37824535e-02 8.22809041e-01 4.74248290e-01
-7.95370460e-01 3.99508655e-01 2.58258492e-01 -7.56113753e-02
3.23806494e-01 -6.54169202e-01 -2.64363497e-01 8.53629053e-01
5.24649978e-01 1.13418072e-01 4.51488703e-01 -6.79604650e-01
-1.75977290e+00 -8.16835523e-01 5.26588321e-01 5.28488420e-02
4.56376046e-01 -4.42171730e-02 -7.92050481e-01 5.31777263e-01
-3.58982414e-01 6.76076561e-02 5.80211878e-01 -2.41992861e-01
7.01600499e-03 -2.45960295e-01 -5.69074571e-01 4.03958499e-01
1.07234156e+00 -5.94309747e-01 -4.35208380e-01 6.33428395e-01
2.61235178e-01 -4.47963059e-01 -5.83902121e-01 5.15748799e-01
6.81179285e-01 -9.78355348e-01 9.42062378e-01 -4.32131052e-01
2.09240854e-01 -2.42233381e-01 -3.40224773e-01 -5.01446366e-01
5.77692091e-02 -6.50484622e-01 -7.61545658e-01 9.37073410e-01
-2.78715402e-01 -1.45318717e-01 9.91638601e-01 6.35226130e-01
1.34981334e-01 -7.58342803e-01 -1.20305669e+00 -8.61285269e-01
-1.18748724e+00 -7.32970953e-01 4.55375046e-01 4.35160011e-01
-9.25926566e-02 -1.27838060e-01 -9.54557240e-01 1.52436912e-01
5.49043000e-01 5.41754007e-01 1.03078067e+00 -8.02046478e-01
-7.06458628e-01 5.82914390e-02 -7.22756803e-01 -1.69401681e+00
1.28914833e-01 -1.07652932e-01 2.20084518e-01 -1.19919753e+00
4.22621667e-01 -2.73844451e-01 -4.96038646e-01 5.53113878e-01
-9.22718868e-02 5.71860522e-02 1.88321069e-01 6.47130251e-01
-1.34495413e+00 8.79549265e-01 1.09230947e+00 -2.49417305e-01
-7.26821870e-02 3.25941741e-01 -1.42119840e-01 6.07793868e-01
4.28337008e-01 -4.98156130e-01 -5.49421430e-01 -3.11363280e-01
-1.21331625e-01 7.10441053e-01 3.48273844e-01 -1.37053955e+00
2.48701394e-01 -4.29845840e-01 1.01006001e-01 -8.65808368e-01
7.56696939e-01 -1.00255179e+00 2.77180731e-01 4.62476850e-01
-3.23913068e-01 -2.90054351e-01 -2.36212924e-01 1.03753281e+00
-5.06701231e-01 -2.40070403e-01 2.04009667e-01 -3.80040295e-02
-9.74740088e-01 3.96218210e-01 -2.71346152e-01 -3.34979981e-01
1.15348983e+00 -5.00796258e-01 1.22494828e-02 -4.06603247e-01
-8.28552663e-01 1.91777751e-01 2.54383385e-01 3.31620723e-01
9.02419150e-01 -1.36571789e+00 -4.11224753e-01 1.40222102e-01
1.74313650e-01 -2.29366079e-01 6.53818488e-01 1.17805481e+00
-2.99668849e-01 5.83205640e-01 -2.84772310e-02 -5.63390613e-01
-1.66135716e+00 5.94033718e-01 -1.85649067e-01 -5.94414651e-01
-7.22522020e-01 4.89577234e-01 -1.11661479e-01 3.40348184e-01
5.48673153e-01 -4.80979718e-02 -3.50864023e-01 5.20669483e-03
8.31044734e-01 6.00425839e-01 -7.77788982e-02 -7.01589942e-01
-3.05518836e-01 3.28774393e-01 -1.56464636e-01 1.24167018e-01
1.31435680e+00 -2.17453152e-01 1.30471751e-01 3.16994041e-01
9.60271478e-01 -4.38782662e-01 -1.64598799e+00 -5.83164990e-01
1.05831132e-04 -1.01782370e+00 2.40911450e-02 -4.75823432e-01
-1.10571599e+00 8.31665218e-01 5.31491876e-01 -1.12335309e-01
1.49281836e+00 -3.21655929e-01 1.10354650e+00 7.17355669e-01
6.47303045e-01 -1.33363330e+00 3.36944699e-01 1.29963979e-01
7.51172960e-01 -1.09628987e+00 2.64714181e-01 -2.17766598e-01
-6.16004527e-01 9.83545244e-01 3.44159573e-01 4.58927676e-02
4.51750755e-01 -2.15202212e-01 -4.01848167e-01 3.46076908e-04
-8.05042922e-01 -2.38987982e-01 1.40583262e-01 2.34900162e-01
-2.56372631e-01 5.08649200e-02 -4.02464449e-01 3.77982199e-01
5.58945537e-01 2.85097301e-01 1.56826928e-01 1.52472973e+00
-4.07082498e-01 -1.02664864e+00 -4.55883473e-01 4.98786271e-01
-3.37655634e-01 1.97042003e-01 -1.56196490e-01 4.43071604e-01
-1.00497290e-01 1.04634142e+00 -1.85237423e-01 -4.20994788e-01
2.21110970e-01 -3.15951928e-02 4.80664909e-01 -3.58210832e-01
1.62767604e-01 2.13515669e-01 -1.25965491e-01 -1.17947268e+00
-7.99123287e-01 -1.06232655e+00 -9.99706209e-01 -1.22127207e-02
-4.42794800e-01 1.59600377e-01 4.30708557e-01 1.31564617e+00
4.09435362e-01 3.82651359e-01 7.89792657e-01 -8.56833220e-01
-7.57023633e-01 -8.76152039e-01 -7.41658032e-01 5.59766889e-01
1.99471742e-01 -9.68789876e-01 -1.76973343e-01 1.11184649e-01] | [8.446305274963379, 0.614672839641571] |
b6c16312-b351-45db-b7d2-bd9efc35cd40 | chemical-protein-interaction-extraction-via | 1911.09487 | null | https://arxiv.org/abs/1911.09487v2 | https://arxiv.org/pdf/1911.09487v2.pdf | Chemical-protein Interaction Extraction via Gaussian Probability Distribution and External Biomedical Knowledge | Motivation: The biomedical literature contains a wealth of chemical-protein interactions (CPIs). Automatically extracting CPIs described in biomedical literature is essential for drug discovery, precision medicine, as well as basic biomedical research. Most existing methods focus only on the sentence sequence to identify these CPIs. However, the local structure of sentences and external biomedical knowledge also contain valuable information. Effective use of such information may improve the performance of CPI extraction. Results: In this paper, we propose a novel neural network-based approach to improve CPI extraction. Specifically, the approach first employs BERT to generate high-quality contextual representations of the title sequence, instance sequence, and knowledge sequence. Then, the Gaussian probability distribution is introduced to capture the local structure of the instance. Meanwhile, the attention mechanism is applied to fuse the title information and biomedical knowledge, respectively. Finally, the related representations are concatenated and fed into the softmax function to extract CPIs. We evaluate our proposed model on the CHEMPROT corpus. Our proposed model is superior in performance as compared with other state-of-the-art models. The experimental results show that the Gaussian probability distribution and external knowledge are complementary to each other. Integrating them can effectively improve the CPI extraction performance. Furthermore, the Gaussian probability distribution can effectively improve the extraction performance of sentences with overlapping relations in biomedical relation extraction tasks. Availability: Data and code are available at https://github.com/CongSun-dlut/CPI_extraction. Contact: yangzh@dlut.edu.cn, wangleibihami@gmail.com Supplementary information: Supplementary data are available at Bioinformatics online. | ['Jian Wang', 'Cong Sun', 'Yin Zhang', 'Leilei Su', 'Zhihao Yang', 'Lei Wang', 'Hongfei Lin'] | 2019-11-21 | null | null | null | null | ['chemical-protein-interaction-extraction'] | ['medical'] | [ 4.13733423e-01 -1.02527410e-01 -4.94416058e-01 -3.45480651e-01
-8.09466839e-01 -2.76895463e-01 1.66526571e-01 5.92614591e-01
-4.57121849e-01 1.16601801e+00 2.50922769e-01 -3.36905807e-01
-1.26968384e-01 -6.83448970e-01 -7.09310293e-01 -9.29466486e-01
3.03586066e-01 2.82317489e-01 5.99712878e-02 1.59737781e-01
2.52044350e-01 2.50026762e-01 -8.63649011e-01 4.29063201e-01
1.22148204e+00 7.31908441e-01 6.13386214e-01 1.80421919e-01
-3.40811789e-01 5.58167219e-01 -6.47087991e-01 -2.52949744e-01
-3.98324162e-01 -3.06644559e-01 -8.83001804e-01 -3.69591415e-01
-4.64835107e-01 7.90306181e-02 -3.77828062e-01 1.21979272e+00
8.03193152e-01 -3.08056846e-02 6.74761772e-01 -8.29228163e-01
-6.34807229e-01 7.21928298e-01 -5.46582282e-01 2.45927274e-01
2.73237675e-01 3.43696261e-03 9.23529804e-01 -7.68427134e-01
7.45200455e-01 9.44736242e-01 1.39163688e-01 3.58127892e-01
-7.67292798e-01 -8.33447337e-01 2.58660968e-03 4.34276164e-01
-1.42185879e+00 -2.78541386e-01 7.15893447e-01 -2.12608129e-01
1.15807271e+00 1.25425225e-02 1.91795900e-01 1.00942969e+00
2.73881316e-01 1.11023188e+00 6.49830639e-01 -2.11877257e-01
-1.92834035e-01 1.09116614e-01 2.98765242e-01 5.94473302e-01
2.35909358e-01 -2.01575249e-01 -3.80743712e-01 -1.70640796e-01
2.26209357e-01 9.18948203e-02 -5.12444198e-01 4.03903335e-01
-1.17344868e+00 5.19446790e-01 4.30325329e-01 5.63136458e-01
-5.81998646e-01 -3.03846568e-01 5.91984451e-01 -1.86841220e-01
2.57705539e-01 3.34677190e-01 -8.51297855e-01 -2.22913995e-02
-5.90760887e-01 5.60109057e-02 7.45410264e-01 1.15801454e+00
5.21054029e-01 -4.25217092e-01 -3.16044033e-01 9.23685074e-01
1.74641803e-01 3.93138379e-01 7.15359092e-01 -2.68326581e-01
7.11438000e-01 5.99758863e-01 -1.28866866e-01 -1.20390928e+00
-6.24875963e-01 -4.82351452e-01 -9.13572192e-01 -9.03649509e-01
3.57273147e-02 -4.50184971e-01 -7.24895120e-01 1.57891333e+00
4.87250119e-01 3.01127732e-01 3.72865736e-01 6.65977478e-01
1.44083285e+00 8.61853838e-01 3.55612218e-01 -4.27785903e-01
1.68427229e+00 -7.49113917e-01 -1.15869689e+00 3.81926447e-02
6.10802472e-01 -9.54075158e-01 4.01450008e-01 1.30523071e-01
-9.31753576e-01 -3.13656896e-01 -9.93792832e-01 -1.40522018e-01
-3.67006391e-01 5.08480430e-01 4.18127030e-01 -4.19647060e-02
-1.16664656e-01 5.61137736e-01 -8.70287657e-01 -1.57632381e-02
7.15066671e-01 4.50212896e-01 -3.80442888e-01 -1.45362347e-01
-1.61914170e+00 8.72040808e-01 9.72171605e-01 3.77026081e-01
-1.39924258e-01 -7.76207685e-01 -8.11170459e-01 1.54598221e-01
4.84568954e-01 -6.76956475e-01 1.05637908e+00 -2.74340123e-01
-1.13096964e+00 5.71495116e-01 -4.48904276e-01 -2.36018121e-01
-7.47230053e-02 -2.06548303e-01 -3.93672109e-01 2.32364103e-01
1.82147279e-01 3.43629062e-01 5.89459166e-02 -7.06783354e-01
-5.56239903e-01 -4.75213766e-01 -3.14487278e-01 1.20510638e-01
-2.13797450e-01 2.24145696e-01 -7.66988754e-01 -5.29277802e-01
-6.11890666e-02 -5.82167506e-01 -3.06764930e-01 -6.37964725e-01
-7.89050460e-01 -3.23322266e-01 2.87041485e-01 -1.08007479e+00
1.45702147e+00 -1.97774220e+00 1.84945285e-01 2.30315179e-01
1.23432413e-01 5.93909621e-01 -7.67767280e-02 5.77086985e-01
-1.90435559e-01 3.58499646e-01 -3.10066760e-01 3.30236137e-01
-1.52508140e-01 1.09400610e-02 2.62028128e-01 3.58053565e-01
6.08773947e-01 1.02604473e+00 -8.49563360e-01 -7.55963326e-01
-1.70583785e-01 5.82787156e-01 -6.82323202e-02 1.62986442e-01
-2.30537489e-01 4.44814563e-01 -9.02779400e-01 5.74670792e-01
7.58780599e-01 -4.03209269e-01 3.26812923e-01 -5.16071856e-01
2.75594950e-01 5.57470381e-01 -8.40007722e-01 1.37904382e+00
-2.74954475e-02 -4.69777510e-02 -2.19408363e-01 -1.23442614e+00
9.36055303e-01 6.45250261e-01 5.92768610e-01 -3.57461005e-01
3.78916740e-01 2.88925380e-01 2.28615284e-01 -8.79224896e-01
-1.43710732e-01 -1.22796632e-01 1.34752363e-01 -4.19127233e-02
1.09710775e-01 3.31809670e-01 3.93655270e-01 9.84783545e-02
9.12356973e-01 2.50727773e-01 7.60401130e-01 6.16488233e-02
9.68791246e-01 -1.19056910e-01 1.19293773e+00 2.32498378e-01
-1.20203078e-01 1.30212873e-01 6.33254945e-01 -1.53885745e-02
-7.22311199e-01 -6.00958407e-01 -4.45069075e-01 3.81121516e-01
2.89476160e-02 -5.41255414e-01 -6.16961062e-01 -6.91670418e-01
-1.97761014e-01 5.21378696e-01 -4.08370137e-01 -2.48730510e-01
-4.99755085e-01 -1.13851917e+00 5.59563935e-01 5.38809538e-01
4.58627373e-01 -1.13495624e+00 7.18209371e-02 3.97774756e-01
-6.98359787e-01 -1.23587167e+00 -5.84103286e-01 1.81492090e-01
-6.64318562e-01 -1.15912402e+00 -7.39952326e-01 -7.31007040e-01
6.51639402e-01 -7.62146488e-02 6.86405241e-01 9.55191404e-02
-3.77248228e-01 -4.45608735e-01 -3.74520302e-01 -6.85984135e-01
-2.85192579e-01 1.89155698e-01 -1.22756273e-01 -3.38489711e-01
9.12602186e-01 -4.21936393e-01 -5.87509334e-01 -1.96151342e-02
-9.54806626e-01 1.70264930e-01 9.20146346e-01 1.03780222e+00
7.70479858e-01 1.62714049e-01 7.05164552e-01 -1.10149324e+00
6.81282520e-01 -6.84773862e-01 -3.64218146e-01 2.57804424e-01
-2.90154994e-01 2.62724370e-01 5.15759885e-01 -4.46501404e-01
-1.11211526e+00 9.56547633e-02 -5.60437441e-01 2.68861800e-02
-3.46313804e-01 1.20409536e+00 -6.64062083e-01 4.75205332e-01
1.48327678e-01 3.62162292e-01 -2.72006005e-01 -4.85416323e-01
-6.33037612e-02 8.76560330e-01 2.90562630e-01 -6.44464076e-01
2.00735927e-01 -1.67698026e-01 6.35230616e-02 -7.84947753e-01
-9.29918528e-01 -5.67102611e-01 -4.86367673e-01 3.62611592e-01
8.48594904e-01 -7.60446072e-01 -9.72629905e-01 3.13635349e-01
-1.29973125e+00 2.90174663e-01 5.22498131e-01 8.02554905e-01
-2.20735058e-01 6.95243657e-01 -8.83489192e-01 -5.33952773e-01
-8.03915739e-01 -1.29559183e+00 7.93053806e-01 5.38531423e-01
-2.52865888e-02 -6.61824286e-01 -7.62223899e-02 5.16479254e-01
-1.68696150e-01 2.15928555e-01 1.09541214e+00 -1.06694627e+00
-2.89926022e-01 -2.35535622e-01 -3.88213664e-01 9.04918984e-02
3.89618605e-01 7.69925341e-02 -6.98569059e-01 2.16342181e-01
-1.55853093e-01 9.98270288e-02 8.21595907e-01 4.52807128e-01
1.22314358e+00 -4.65433687e-01 -7.02055633e-01 3.57397556e-01
1.26088846e+00 5.27071536e-01 5.61573565e-01 2.17673346e-01
7.49844372e-01 6.06128097e-01 8.64644170e-01 4.17730987e-01
2.28095964e-01 6.36710227e-01 -3.02713476e-02 -1.23946657e-02
3.00989598e-01 -1.28747270e-01 1.61237359e-01 8.96021605e-01
-1.45361826e-01 -1.28272906e-01 -1.02608669e+00 4.55040038e-01
-1.89336193e+00 -8.98542702e-01 -2.13221371e-01 1.89515758e+00
1.51947510e+00 2.88609117e-02 -1.62749901e-01 1.51072862e-02
7.33119965e-01 -3.12621444e-01 -5.87604284e-01 -7.66652226e-02
-4.74706776e-02 2.41996124e-01 4.34943080e-01 3.57468575e-01
-1.13423049e+00 8.46655846e-01 3.50460982e+00 1.03633344e+00
-8.37556839e-01 -7.15325475e-02 5.06682992e-01 2.09436007e-02
-9.09423642e-03 -2.83097535e-01 -1.13329840e+00 7.60040104e-01
1.05931127e+00 -3.98738384e-01 6.08296618e-02 4.92125601e-01
2.60070771e-01 2.80439019e-01 -7.63558388e-01 7.93636143e-01
-2.71319807e-01 -1.48176014e+00 -1.72987938e-01 6.78925663e-02
3.77960145e-01 1.32639080e-01 -3.39601815e-01 1.45421937e-01
-1.46310916e-02 -8.44753504e-01 -8.68580565e-02 7.53067911e-01
4.62658614e-01 -9.82716680e-01 1.44230485e+00 4.73625422e-01
-1.13145626e+00 1.06526203e-01 -4.46479172e-01 5.97825229e-01
6.32325336e-02 1.03122663e+00 -9.15554404e-01 1.19745052e+00
4.10825074e-01 7.92634964e-01 -3.07366699e-01 1.08790684e+00
-6.30484819e-01 5.80614388e-01 -2.22093359e-01 -2.73970991e-01
1.18509956e-01 -1.82765469e-01 3.39600205e-01 1.45727670e+00
1.20176263e-01 4.37991917e-01 2.40291744e-01 7.91401267e-01
-2.41622671e-01 6.56684160e-01 -3.21297258e-01 -4.95484352e-01
4.84540612e-01 1.22872293e+00 -4.24041390e-01 -3.98413539e-01
-5.03390610e-01 5.77003002e-01 3.68124157e-01 3.95076364e-01
-7.89028227e-01 -7.49587595e-01 3.98591787e-01 -3.90579551e-01
2.61407554e-01 9.80460793e-02 -3.07877492e-02 -1.24698520e+00
7.52400607e-02 -9.37619686e-01 5.10670364e-01 -4.76654589e-01
-1.40170443e+00 6.11639798e-01 -4.72637676e-02 -9.70115900e-01
4.81390841e-02 -5.72625101e-01 -3.40229273e-01 1.14868522e+00
-1.45828807e+00 -1.18540525e+00 1.63483873e-01 2.34719232e-01
2.76065260e-01 9.76888835e-02 9.11826074e-01 4.49804693e-01
-1.09530604e+00 6.55161679e-01 3.09232950e-01 5.01614988e-01
8.78922999e-01 -9.79316890e-01 1.24500573e-01 5.75971723e-01
-2.42781997e-01 1.19378555e+00 4.62123036e-01 -9.20878410e-01
-1.07596111e+00 -1.13959014e+00 1.20821893e+00 4.39801365e-02
5.03197730e-01 -3.93661112e-02 -1.22814739e+00 4.07384902e-01
5.67413680e-02 -2.77530462e-01 1.14023578e+00 -8.68531317e-03
1.05281007e-02 1.00062437e-01 -9.75570619e-01 4.72575277e-01
5.39820015e-01 -3.20201755e-01 -5.83606422e-01 5.73455632e-01
8.75333786e-01 -6.30341768e-01 -1.30687582e+00 5.88659346e-01
4.66323406e-01 -7.24536926e-02 1.02504659e+00 -9.11081076e-01
7.73203135e-01 -4.66737837e-01 1.01929769e-01 -9.99568582e-01
-2.93077379e-01 -1.77542612e-01 -1.77584752e-01 1.41547465e+00
7.59499371e-01 -7.43687868e-01 4.65508044e-01 4.03324366e-01
-1.17752373e-01 -1.24336708e+00 -6.45158112e-01 -4.03450549e-01
8.89577121e-02 -4.22074571e-02 6.40920937e-01 8.74221921e-01
4.82973099e-01 7.81688094e-01 -1.96801603e-01 3.75201911e-01
2.45784476e-01 1.92931712e-01 3.01527619e-01 -9.28577006e-01
-4.20225322e-01 -3.26738983e-01 -2.32465118e-01 -7.85986304e-01
2.64993191e-01 -1.09550500e+00 9.21822153e-03 -1.67523026e+00
5.25074780e-01 -8.34668130e-02 -6.67569637e-01 7.07649052e-01
-8.12276185e-01 -3.44239771e-01 -1.06843963e-01 -4.28382121e-03
-1.79718673e-01 5.35137475e-01 1.34472632e+00 -2.93910682e-01
-2.11816981e-01 3.12288590e-02 -8.34160089e-01 4.61107999e-01
1.13339674e+00 -5.36367595e-01 -1.05091184e-01 -2.28548665e-02
3.26302499e-02 1.36145085e-01 -4.08104695e-02 -5.46658456e-01
3.00690830e-01 -3.10745567e-01 5.23196518e-01 -7.67859817e-01
1.36673674e-01 -6.04103446e-01 2.48454697e-02 5.37782848e-01
-4.15204346e-01 -3.96440923e-01 4.09332335e-01 5.40642440e-01
-2.38233700e-01 -5.24122477e-01 5.05440950e-01 -1.98189080e-01
-1.27341375e-01 4.02746171e-01 -2.39020303e-01 -8.21259320e-02
6.77112758e-01 3.11605573e-01 -5.17691553e-01 3.78933921e-02
-7.58060575e-01 5.08058667e-01 -1.26077846e-01 2.85542399e-01
6.22616708e-01 -1.06445217e+00 -8.58018756e-01 -1.29816681e-01
1.00058541e-01 1.22550167e-01 4.22485024e-01 1.11478794e+00
-1.93380326e-01 7.38297105e-01 5.55929057e-02 -2.48508513e-01
-1.59836650e+00 8.72775376e-01 6.62517846e-02 -4.42550927e-01
-3.56670648e-01 6.85343206e-01 2.18774021e-01 -4.30325687e-01
1.10103175e-01 -2.72116214e-01 -5.42457223e-01 -7.88884088e-02
6.35206521e-01 -1.43775761e-01 6.33141994e-02 -6.22447550e-01
-5.68721950e-01 2.80250847e-01 -5.04786909e-01 4.28689539e-01
1.42365599e+00 2.41353273e-01 -4.35105920e-01 9.53783318e-02
1.26609874e+00 -1.06737256e-01 -3.12834144e-01 -5.34810007e-01
1.55689701e-01 -8.35979953e-02 -3.56252603e-02 -8.76727760e-01
-9.16592360e-01 9.65689838e-01 9.97001305e-02 -4.90533262e-01
1.04188037e+00 3.46556269e-02 8.65274429e-01 4.25073832e-01
-1.07410222e-01 -8.81798923e-01 -4.37920243e-01 4.57931876e-01
7.51834631e-01 -1.09241903e+00 2.89009839e-01 -7.24092603e-01
-6.61427796e-01 9.88544583e-01 4.66675073e-01 4.10332888e-01
6.68783844e-01 2.70307958e-01 -1.82143316e-01 -2.38244489e-01
-6.76318884e-01 -1.38759911e-01 4.15858209e-01 3.27322215e-01
7.87763536e-01 1.66065935e-02 -9.14113224e-01 1.29353499e+00
6.74851462e-02 2.81242073e-01 1.67146727e-01 9.15762484e-01
-2.19151646e-01 -1.37072945e+00 -1.10820420e-01 5.47846138e-01
-8.78100693e-01 -6.08856618e-01 -3.46381247e-01 3.31244111e-01
1.26793444e-01 1.04773510e+00 -6.13713145e-01 -2.43153512e-01
2.72503495e-01 2.50039041e-01 2.91292429e-01 -4.98992085e-01
-4.46151972e-01 4.29349452e-01 2.91548997e-01 -2.31854528e-01
-5.69797158e-01 -5.85247278e-01 -1.67834961e+00 3.74936759e-02
-7.21196234e-01 5.27533531e-01 5.49049139e-01 1.01405025e+00
5.62768161e-01 8.94268095e-01 4.33858424e-01 -1.95727453e-01
-2.35638633e-01 -1.20456040e+00 -1.71187103e-01 1.03628814e-01
-6.88158721e-02 -5.71448684e-01 1.05998799e-01 2.48709634e-01] | [8.417847633361816, 8.838783264160156] |
1557dde8-9eb7-4817-aa72-1a802aecbae7 | policy-learning-and-evaluation-with | 2202.07808 | null | https://arxiv.org/abs/2202.07808v2 | https://arxiv.org/pdf/2202.07808v2.pdf | Policy Learning and Evaluation with Randomized Quasi-Monte Carlo | Reinforcement learning constantly deals with hard integrals, for example when computing expectations in policy evaluation and policy iteration. These integrals are rarely analytically solvable and typically estimated with the Monte Carlo method, which induces high variance in policy values and gradients. In this work, we propose to replace Monte Carlo samples with low-discrepancy point sets. We combine policy gradient methods with Randomized Quasi-Monte Carlo, yielding variance-reduced formulations of policy gradient and actor-critic algorithms. These formulations are effective for policy evaluation and policy improvement, as they outperform state-of-the-art algorithms on standardized continuous control benchmarks. Our empirical analyses validate the intuition that replacing Monte Carlo with Quasi-Monte Carlo yields significantly more accurate gradient estimates. | ['Fei Sha', 'Yi-fan Chen', 'Liyu Chen', "Pierre L'Ecuyer", 'Sebastien M. R. Arnold'] | 2022-02-16 | null | null | null | null | ['policy-gradient-methods'] | ['methodology'] | [-4.23349231e-01 -4.46031317e-02 -6.35395229e-01 4.87035140e-02
-1.06594241e+00 -5.30106843e-01 6.07834101e-01 1.78025626e-02
-8.59791636e-01 1.76258326e+00 1.02317184e-01 -5.90429068e-01
-1.02016322e-01 -5.89466453e-01 -7.96519339e-01 -4.50595587e-01
-2.79951066e-01 6.06841683e-01 -1.71866622e-02 -1.26312464e-01
3.06857735e-01 1.37019739e-01 -1.17253864e+00 -3.11091214e-01
1.49646306e+00 8.95872176e-01 -1.70488521e-01 6.67225242e-01
2.03948393e-01 5.62904298e-01 -6.11252248e-01 -2.78032780e-01
4.15045172e-01 -5.91005921e-01 -2.60870159e-01 -2.26097181e-01
1.07828736e-01 -7.37789273e-01 1.98235605e-02 1.25323176e+00
6.89589262e-01 5.63041210e-01 6.94704950e-01 -1.02178741e+00
-2.50515670e-01 5.77706277e-01 -5.91029763e-01 1.74680263e-01
2.94714689e-01 7.71237552e-01 7.91330934e-01 -4.50203806e-01
3.22254032e-01 1.65236199e+00 7.81427860e-01 3.88201356e-01
-1.28590989e+00 -5.43288052e-01 3.81246924e-01 3.57783511e-02
-9.55157101e-01 -1.77822948e-01 3.69592279e-01 -4.51093674e-01
8.72769952e-01 -1.32581234e-01 1.00022662e+00 1.10106552e+00
4.69068259e-01 6.80101931e-01 1.48718107e+00 -1.33310303e-01
8.96200716e-01 -1.70638226e-02 -5.05037665e-01 5.49824655e-01
7.09417164e-01 1.01373458e+00 5.31715713e-02 -3.91830564e-01
1.11247838e+00 -1.78476214e-01 -6.54485673e-02 -3.67588490e-01
-1.00913906e+00 1.11909306e+00 6.92534894e-02 -2.48735249e-01
-8.57148290e-01 7.14875519e-01 6.27044618e-01 1.88186422e-01
5.08635879e-01 7.26826310e-01 -3.74796987e-01 -7.08769977e-01
-7.08807230e-01 8.35603654e-01 8.97362888e-01 5.31260192e-01
3.92350435e-01 6.31500065e-01 -6.40922368e-01 5.64122617e-01
3.24878067e-01 9.48231995e-01 4.77759391e-01 -1.37749517e+00
5.11525393e-01 -2.48671398e-02 1.04101801e+00 -4.54224706e-01
-2.38746151e-01 -5.34576118e-01 -4.28375810e-01 7.05908358e-01
8.54772985e-01 -8.47467065e-01 -6.50753140e-01 1.66643858e+00
5.54931641e-01 1.75755084e-01 4.71133254e-02 9.13547516e-01
-3.79477799e-01 5.48887193e-01 2.15263605e-01 -5.54063380e-01
9.26170409e-01 -1.04870296e+00 -9.65868771e-01 -2.23145485e-01
4.82445210e-01 -5.67166686e-01 1.40470374e+00 2.58844733e-01
-1.30192983e+00 -3.62638772e-01 -8.70889783e-01 7.44070590e-01
4.02688794e-02 1.55245706e-01 5.26491106e-01 6.60446584e-01
-6.55576169e-01 1.17998552e+00 -9.36315000e-01 3.07304502e-01
5.66013813e-01 1.19894954e-04 5.04746377e-01 4.70150769e-01
-1.16476989e+00 1.31147516e+00 4.56180751e-01 -2.45508879e-01
-1.22977090e+00 -8.87021661e-01 -6.75609410e-01 -1.72981635e-01
8.61791253e-01 -6.04570627e-01 1.90849710e+00 -6.30766869e-01
-2.37101460e+00 -2.94826459e-02 1.49842560e-01 -9.95051801e-01
9.96699989e-01 -4.27447170e-01 -1.19882613e-01 -8.94736052e-02
-2.07744539e-01 3.18991095e-01 1.23515677e+00 -1.02200341e+00
-6.13601804e-01 -1.44292146e-01 2.29538009e-02 3.86840343e-01
2.02119008e-01 -3.02296758e-01 3.59482110e-01 -5.14969587e-01
-5.73270679e-01 -8.46316457e-01 -7.29033947e-01 -1.80259272e-01
-1.28528565e-01 -5.29860854e-02 3.11022192e-01 -4.12368298e-01
1.37111080e+00 -1.61786187e+00 -2.86110759e-01 2.62078822e-01
-1.44757986e-01 6.07081771e-01 8.94818678e-02 1.74696550e-01
2.84513503e-01 6.71359990e-03 -1.15928568e-01 1.34461135e-01
5.33332765e-01 2.82365680e-01 -6.14394248e-01 7.10107386e-01
-4.47250865e-02 1.02119088e+00 -1.21665990e+00 -3.26779634e-01
4.24242020e-01 6.04655817e-02 -8.75924408e-01 4.17842455e-02
-7.40106046e-01 6.47048235e-01 -9.03081894e-01 2.15228230e-01
4.01681244e-01 -2.12746158e-01 2.68897295e-01 3.03265274e-01
-3.40747476e-01 2.77071089e-01 -1.27680957e+00 1.31919277e+00
-5.31768382e-01 1.55188933e-01 -2.17517558e-03 -9.37942624e-01
6.53307080e-01 1.83750659e-01 3.81814837e-01 -6.58161342e-01
3.21234256e-01 2.83375204e-01 -2.20159411e-01 -2.91808099e-01
3.20237428e-01 -2.95819491e-01 2.21195668e-01 5.54488719e-01
-2.75244325e-01 -6.64195895e-01 5.63794911e-01 -3.75427961e-01
6.79476142e-01 3.35931659e-01 6.48012400e-01 -2.91625112e-01
3.24077129e-01 9.74148810e-02 5.38992047e-01 1.08215153e+00
-5.47323585e-01 -2.07624719e-01 7.21256554e-01 -3.57832253e-01
-1.18511403e+00 -1.27416432e+00 -1.25663251e-01 7.90492833e-01
-7.01081231e-02 -1.74010664e-01 -8.44616592e-01 -6.83321357e-01
5.08411348e-01 1.02669919e+00 -3.51960421e-01 -2.23333135e-01
-4.43027347e-01 -8.01741242e-01 1.73554555e-01 4.53889877e-01
3.75635743e-01 -7.86144435e-01 -8.90132904e-01 7.64709055e-01
3.57536167e-01 -7.55671501e-01 -4.80017960e-01 -1.98611453e-01
-1.06666005e+00 -1.03168762e+00 -9.70127165e-01 1.61187083e-01
4.36773628e-01 -4.79976296e-01 1.17460024e+00 -4.69700962e-01
5.33755943e-02 4.42367554e-01 1.48620000e-02 -5.81923366e-01
-5.94191015e-01 -2.62550414e-01 4.57939953e-01 -3.94692272e-01
-2.21053183e-01 -4.75260377e-01 -8.94850075e-01 3.85092199e-01
-4.20990914e-01 -4.82822061e-01 5.11975825e-01 1.23281920e+00
8.03781211e-01 -3.19265664e-01 7.27012694e-01 -4.95185465e-01
1.50793397e+00 -1.74372166e-01 -1.46208370e+00 1.53586522e-01
-9.46447253e-01 4.37527895e-01 9.36728954e-01 -8.24324369e-01
-1.05779564e+00 -3.25540960e-01 -3.45939514e-03 -4.95160580e-01
1.71338171e-01 1.89492330e-01 6.92658544e-01 -1.75007150e-01
8.96793842e-01 -2.15663716e-01 3.01744133e-01 -2.64070332e-01
4.35908616e-01 1.71557814e-01 1.86105877e-01 -1.12251747e+00
4.13072288e-01 2.94821709e-01 -2.53705475e-02 -1.37579083e-01
-9.44145918e-01 1.29721656e-01 2.59044737e-01 -2.38906160e-01
6.09019578e-01 -6.85177088e-01 -1.25224936e+00 1.64409623e-01
-8.31985474e-01 -8.48815620e-01 -9.75771189e-01 1.08698201e+00
-1.18162417e+00 1.50122344e-01 -4.73617226e-01 -1.33706856e+00
-2.49568745e-01 -1.29325747e+00 7.61289775e-01 6.05700254e-01
3.40384617e-03 -1.16859889e+00 4.87987608e-01 -4.56560552e-01
5.46018481e-01 3.31313610e-01 4.19555753e-01 -2.71394998e-01
-1.20948009e-01 -5.60935028e-02 1.12675220e-01 4.51673210e-01
-1.98132664e-01 -5.01333550e-02 -6.25558794e-01 -5.21161437e-01
-5.82462065e-02 -5.16209960e-01 5.52832425e-01 9.89082277e-01
1.08755410e+00 -6.16411328e-01 -1.13049760e-01 3.66526961e-01
1.32229233e+00 3.76965016e-01 3.55500221e-01 2.10273340e-01
9.99382883e-02 1.21947415e-02 1.01338589e+00 1.17941201e+00
7.16101304e-02 3.46607774e-01 2.60973841e-01 2.48652026e-01
4.25555885e-01 -5.41218221e-01 4.97206032e-01 2.49818459e-01
-1.07648239e-01 5.62499650e-02 -8.03766847e-01 2.49569535e-01
-2.03114748e+00 -1.02499211e+00 4.15461093e-01 2.35036182e+00
1.26666236e+00 3.32823575e-01 3.81298214e-01 -3.26683760e-01
7.28765905e-01 4.09777686e-02 -1.02623904e+00 -5.71927547e-01
4.16880250e-01 3.66357684e-01 9.84053910e-01 7.07997859e-01
-8.40659022e-01 9.35116470e-01 7.92836809e+00 1.15433228e+00
-7.60741234e-01 1.41750947e-01 5.87123215e-01 -2.60798067e-01
-7.69433239e-03 4.36432054e-03 -8.91983688e-01 6.85485899e-01
1.03811932e+00 -5.51454365e-01 8.97105336e-01 1.13186085e+00
7.48538613e-01 -5.31963468e-01 -7.85070837e-01 7.37462580e-01
-7.42058158e-01 -1.41792417e+00 -3.71139318e-01 5.38454466e-02
1.30579126e+00 -8.94487575e-02 2.16574848e-01 7.54728913e-01
1.08138096e+00 -9.00463462e-01 6.40753806e-01 6.00653410e-01
4.90711838e-01 -9.02831256e-01 5.17884076e-01 1.87640414e-01
-6.91411972e-01 -2.78162956e-01 -5.36047757e-01 -2.37973794e-01
4.70004886e-01 7.92203665e-01 -8.24529588e-01 7.49414861e-02
1.87674403e-01 3.22416127e-01 1.07037902e-01 1.19443393e+00
-4.51885670e-01 6.88319027e-01 -5.50505996e-01 -6.69226825e-01
6.91480398e-01 -7.48061419e-01 7.41430283e-01 6.80709064e-01
2.71228611e-01 -3.58798802e-01 4.34398293e-01 1.14131331e+00
4.01247561e-01 1.74209718e-02 -5.18699050e-01 -1.91322193e-01
5.48834741e-01 7.07947671e-01 -4.32173967e-01 -6.34394228e-01
-2.20051166e-02 3.43926072e-01 3.02839071e-01 7.13390887e-01
-1.10259640e+00 -4.00485098e-01 9.54115391e-01 -3.07486176e-01
4.62581456e-01 -4.76658106e-01 -1.03099130e-01 -8.42966378e-01
-9.58781689e-02 -1.13057101e+00 2.18729407e-01 -2.67530352e-01
-1.26244593e+00 9.53917801e-02 3.19545150e-01 -1.18429184e+00
-1.06092894e+00 -5.97846568e-01 -5.86236656e-01 6.47360384e-01
-1.41882384e+00 3.58324535e-02 3.51521641e-01 4.10081953e-01
2.61482507e-01 4.76089232e-02 5.26304126e-01 2.16384567e-02
-6.38205171e-01 5.50441325e-01 8.15128446e-01 -3.48237336e-01
5.16399384e-01 -1.29035485e+00 3.53448361e-01 4.03142273e-01
-3.94712567e-01 3.30029637e-01 1.04237592e+00 -7.20408201e-01
-1.34718287e+00 -7.67437637e-01 -1.49425432e-01 1.25920758e-01
9.47081208e-01 2.73479313e-01 -5.13305545e-01 3.14231247e-01
-4.08042595e-02 1.66005418e-01 9.10131820e-03 -1.48868397e-01
1.58317536e-01 -2.33151447e-02 -1.33922577e+00 9.23848569e-01
7.13894725e-01 -2.99861401e-01 -3.73987943e-01 5.34689903e-01
5.87067485e-01 -6.78333998e-01 -9.82534766e-01 3.77399683e-01
4.61334437e-01 -8.10518920e-01 9.80873466e-01 -8.98920655e-01
-2.94395369e-02 -1.48328349e-01 1.62696213e-01 -2.07882977e+00
2.97232985e-01 -1.16134119e+00 -5.62069714e-01 4.23727542e-01
3.20885569e-01 -1.08109093e+00 5.91564119e-01 2.64818877e-01
7.13646114e-02 -9.58677471e-01 -1.11086464e+00 -1.33761752e+00
3.82639349e-01 -4.91728336e-01 5.94548762e-01 5.58751106e-01
1.23274080e-01 -1.97883531e-01 -3.46127838e-01 -2.73232043e-01
8.68674755e-01 5.02892658e-02 6.81095243e-01 -5.80661952e-01
-6.64020121e-01 -7.89984941e-01 9.21397284e-02 -1.06701803e+00
3.62575203e-01 -2.87332416e-01 1.17852069e-01 -1.39778411e+00
-4.34436053e-01 -2.10218757e-01 3.98121960e-02 1.04505762e-01
-4.39548045e-01 -3.38509738e-01 -8.15472752e-02 -3.49235088e-01
-7.61074066e-01 1.22256827e+00 1.90695906e+00 1.47460207e-01
-4.50745672e-01 3.12416881e-01 -1.20683156e-01 8.29284132e-01
1.15474272e+00 -5.04325688e-01 -5.01707017e-01 1.50161341e-01
1.57611385e-01 3.13016057e-01 2.98071355e-01 -9.99375880e-01
-1.90489113e-01 -6.73785210e-01 2.73880899e-01 -4.76185799e-01
1.58308372e-02 -3.30415696e-01 -2.55020648e-01 7.83530891e-01
-4.04390484e-01 2.46755853e-01 3.81997347e-01 8.11019659e-01
2.12607011e-02 -2.96592146e-01 9.56449926e-01 -3.07240546e-01
-3.09071332e-01 2.87246853e-01 -6.70448005e-01 7.77779341e-01
9.91258144e-01 8.72940272e-02 -2.97790140e-01 -6.79891169e-01
-6.07834637e-01 4.87689137e-01 1.39435768e-01 -1.81168526e-01
4.66167361e-01 -1.38381982e+00 -5.26788831e-01 -1.22213326e-01
-5.17604530e-01 -3.64752024e-01 -1.17995948e-01 8.36544752e-01
-2.99886495e-01 3.20579201e-01 -1.82900026e-01 -3.72297317e-01
-3.03085178e-01 4.45548415e-01 6.94997668e-01 -7.61503100e-01
-3.69262099e-01 3.00015271e-01 -3.75002980e-01 -4.54242289e-01
4.24199462e-01 -8.13833058e-01 3.05234969e-01 -2.20054999e-01
5.19055009e-01 8.55288446e-01 -4.27021444e-01 3.19499224e-01
2.08547935e-02 3.34916204e-01 1.79624513e-01 -7.31312335e-01
9.25374806e-01 1.48871183e-01 4.93155837e-01 2.81563461e-01
7.91916430e-01 -4.72166240e-01 -1.98992527e+00 8.19575414e-02
-2.98354570e-02 -5.74821055e-01 2.80521840e-01 -7.79974401e-01
-7.14372993e-01 5.52185535e-01 7.84762740e-01 1.04571536e-01
6.05620146e-01 -6.86290443e-01 5.81230521e-01 5.77988386e-01
5.08909106e-01 -1.77845860e+00 -2.39186529e-02 8.62051129e-01
7.22270012e-01 -1.23096490e+00 3.48880023e-01 3.45778048e-01
-9.30839837e-01 6.56769097e-01 5.69601655e-01 -5.26510179e-01
6.14110172e-01 2.75970727e-01 -2.39612356e-01 2.29963258e-01
-8.17722738e-01 -3.20594400e-01 1.78743497e-01 4.74736571e-01
6.76758066e-02 3.93273979e-01 -8.93770218e-01 2.20375985e-01
-8.10822546e-02 2.25326508e-01 4.17517498e-02 1.11369789e+00
-5.71893692e-01 -1.03661239e+00 -4.46461469e-01 5.11554778e-01
-4.98320669e-01 3.96822356e-02 2.73955494e-01 8.46095741e-01
-6.09701216e-01 8.06988716e-01 6.15298375e-02 1.12059250e-01
3.54813129e-01 1.25076212e-02 8.27299833e-01 2.72175688e-02
-5.79145730e-01 7.30060413e-02 2.97761202e-01 -9.72596526e-01
4.23812270e-02 -5.66846132e-01 -1.36743605e+00 -4.17038560e-01
-2.12278977e-01 4.09376353e-01 8.47295940e-01 8.67427528e-01
3.45479965e-01 5.44061840e-01 6.47631586e-01 -1.02698064e+00
-1.97682178e+00 -9.12675381e-01 -5.01694143e-01 1.31345972e-01
3.20567131e-01 -1.10000741e+00 -4.77907211e-01 -6.51543021e-01] | [4.106679439544678, 2.371483325958252] |
4f2efe4c-627a-41ec-8117-f881c4bb33fa | a-least-square-approach-to-semi-supervised | 2202.02904 | null | https://arxiv.org/abs/2202.02904v2 | https://arxiv.org/pdf/2202.02904v2.pdf | A Compressed Sensing Based Least Squares Approach to Semi-supervised Local Cluster Extraction | A least squares semi-supervised local clustering algorithm based on the idea of compressed sensing is proposed to extract clusters from a graph with known adjacency matrix. The algorithm is based on a two-stage approach similar to the one in \cite{LaiMckenzie2020}. However, under a weaker assumption and with less computational complexity than the one in \cite{LaiMckenzie2020}, the algorithm is shown to be able to find a desired cluster with high probability. The ``one cluster at a time" feature of our method distinguishes it from other global clustering methods. Several numerical experiments are conducted on the synthetic data such as stochastic block model and real data such as MNIST, political blogs network, AT\&T and YaleB human faces data sets to demonstrate the effectiveness and efficiency of our algorithm. | ['Zhaiming Shen', 'Ming-Jun Lai'] | 2022-02-07 | null | null | null | null | ['stochastic-block-model'] | ['graphs'] | [ 1.93301171e-01 3.04397732e-01 -7.63925165e-02 -2.16947809e-01
-7.35541821e-01 -3.70440334e-01 2.90818065e-01 1.27406686e-03
-1.45587221e-01 7.06946850e-01 -1.05549723e-01 -1.68657273e-01
-6.37878776e-01 -7.04421401e-01 -6.42221272e-01 -1.08459604e+00
-3.83898795e-01 4.99058187e-01 6.69083968e-02 4.14886773e-02
3.12137932e-01 2.05460399e-01 -8.75708342e-01 -7.85911083e-02
6.73590064e-01 8.19585025e-01 3.03558588e-01 4.54199433e-01
4.33912754e-01 4.82782423e-01 -1.53271453e-02 6.30551428e-02
6.34331822e-01 -5.38927794e-01 -7.99750805e-01 6.78930581e-01
-5.30577600e-02 2.50430644e-01 -4.58192527e-01 1.34244204e+00
3.58722627e-01 2.04368643e-02 7.17213154e-01 -1.20909154e+00
-1.64046228e-01 7.84881949e-01 -1.12343121e+00 1.60608158e-01
3.70646566e-01 -4.34230179e-01 7.79285848e-01 -6.48316085e-01
6.96527183e-01 9.35219705e-01 7.77037203e-01 7.71554485e-02
-1.25859165e+00 -1.00267887e+00 -3.90410647e-02 -4.23074216e-02
-1.94362760e+00 -4.98220414e-01 8.59427571e-01 -4.55506682e-01
8.96026269e-02 2.42948487e-01 3.23935926e-01 4.52456743e-01
-1.68103144e-01 5.81459105e-01 1.55436468e+00 -5.72323561e-01
4.57106739e-01 -5.43083139e-02 1.97163120e-01 8.88612270e-01
4.42304820e-01 -4.68868352e-02 -4.54646289e-01 -5.76945841e-01
6.75138116e-01 3.60719860e-02 -2.62266785e-01 -6.32119477e-01
-1.09211302e+00 1.02702677e+00 2.13617459e-01 5.28049350e-01
-3.06447923e-01 1.42068371e-01 -8.24401304e-02 2.60481656e-01
4.53593701e-01 -2.64633805e-01 4.52177934e-02 1.82678148e-01
-1.36388123e+00 -1.80018485e-01 8.51372302e-01 1.08869350e+00
8.93076122e-01 -7.60481805e-02 5.60402572e-01 5.16466558e-01
6.66293144e-01 6.16879880e-01 1.97660416e-01 -1.13262963e+00
3.88003945e-01 3.74922276e-01 8.02976936e-02 -1.57505131e+00
-3.79767925e-01 -5.01126826e-01 -1.39342582e+00 1.02769220e-02
2.21515879e-01 -4.42365319e-01 -6.01356208e-01 1.60953188e+00
2.78791726e-01 4.76845145e-01 -6.26526028e-02 7.07563221e-01
2.73767620e-01 5.44396460e-01 -6.01729453e-01 -8.60400498e-01
8.42977166e-01 -5.44934809e-01 -5.01281500e-01 5.71073852e-02
3.81323516e-01 -6.75315022e-01 2.02574566e-01 5.92112184e-01
-8.80842388e-01 -3.69968623e-01 -9.16484892e-01 6.94396794e-01
1.41316220e-01 6.74943775e-02 5.55706382e-01 9.27455604e-01
-1.33442140e+00 6.16024494e-01 -9.44023848e-01 -8.61493826e-01
2.38415211e-01 4.09606665e-01 -2.59964943e-01 -2.38225758e-01
-7.57643759e-01 7.04515576e-02 3.22670102e-01 1.38244241e-01
-8.21837366e-01 7.35469162e-02 -6.18079245e-01 -1.24421939e-01
4.29128408e-01 -2.21264094e-01 4.39589024e-01 -8.92834842e-01
-1.15508687e+00 6.93490565e-01 -3.40102196e-01 -5.69339335e-01
4.34920549e-01 3.15464199e-01 -3.66718948e-01 4.94362146e-01
3.47699136e-01 2.53032595e-01 8.12696576e-01 -1.42933929e+00
-3.97147596e-01 -4.56951767e-01 -3.93085033e-01 3.82504836e-02
-1.33607060e-01 -2.37043407e-02 -4.08518612e-01 -5.90648711e-01
7.40895987e-01 -1.34057593e+00 -6.95486665e-01 -3.05669546e-01
-6.43145502e-01 -4.47041765e-02 8.24727118e-01 -5.41851699e-01
1.12258172e+00 -2.39965677e+00 1.64715171e-01 1.01076126e+00
3.52616489e-01 -1.37587115e-01 2.09585488e-01 8.64624143e-01
-2.68058717e-01 6.90281913e-02 -6.18017554e-01 -2.03113288e-01
-2.76509553e-01 8.23243335e-02 2.70989507e-01 1.17798352e+00
-6.99621499e-01 1.66874424e-01 -8.19750607e-01 -7.33605325e-01
5.35730682e-02 -4.49767523e-03 -2.71482974e-01 -3.79603833e-01
2.54836798e-01 4.49755311e-01 -6.70001388e-01 4.48369741e-01
6.25982583e-01 -7.16738462e-01 5.76117814e-01 -1.49955833e-02
8.01444426e-02 -5.22923172e-01 -1.76542604e+00 1.52441728e+00
1.60540789e-01 3.86486053e-01 7.18156993e-01 -1.56591940e+00
8.21102023e-01 6.55056417e-01 9.75951970e-01 -1.01733826e-01
1.98027730e-01 1.06001481e-01 -1.19726561e-01 -1.38543144e-01
4.78096679e-02 -3.00180882e-01 -2.14365795e-01 6.95521533e-01
-3.75958055e-01 2.34832227e-01 2.51697212e-01 7.59157419e-01
1.29949236e+00 -2.78713316e-01 4.69391108e-01 -8.02957714e-01
3.19958717e-01 1.10834613e-01 6.36574328e-01 8.09202313e-01
-4.33214068e-01 5.96186638e-01 2.16686457e-01 2.00433776e-01
-6.76778674e-01 -9.06510293e-01 -7.76095912e-02 4.16063547e-01
3.31720412e-01 -5.23586392e-01 -6.64272010e-01 -5.04614890e-01
-1.90814033e-01 7.23062530e-02 -5.10063469e-01 1.94037721e-01
-7.42310509e-02 -8.42669308e-01 4.06141490e-01 -2.91340202e-02
6.77610755e-01 -5.66876292e-01 -2.98953831e-01 -1.53635433e-02
-2.76627898e-01 -1.19799197e+00 -3.97779077e-01 3.31046917e-02
-8.09224546e-01 -1.36924946e+00 -5.80739617e-01 -9.82582748e-01
1.00461471e+00 5.61903954e-01 6.90896988e-01 -1.64755955e-02
-1.42658204e-01 4.61920291e-01 -4.85333264e-01 3.76322567e-02
-1.54435635e-01 -5.84349409e-02 3.29091698e-01 3.82956505e-01
1.97278321e-01 -8.12727094e-01 -6.32155359e-01 3.62828404e-01
-7.07671583e-01 -4.65035066e-02 6.46376848e-01 6.66390598e-01
5.49654543e-01 1.07395911e+00 4.23828483e-01 -1.10196662e+00
4.22940254e-01 -8.99586678e-01 -6.54175162e-01 -1.57925799e-01
-8.66336226e-01 -4.42369759e-01 5.90040386e-01 4.86844629e-02
-8.42596889e-01 5.20445764e-01 5.06634355e-01 -5.01698673e-01
-9.72995311e-02 6.08727455e-01 1.86598785e-02 -7.75271729e-02
4.18615550e-01 3.01422536e-01 1.01779245e-01 -5.14027297e-01
3.68499398e-01 9.06503677e-01 3.01066041e-01 -3.71035635e-01
1.25190771e+00 1.02493465e+00 1.82164848e-01 -1.02398038e+00
-3.23560655e-01 -8.87587488e-01 -6.94956958e-01 -1.90824583e-01
6.33973062e-01 -1.03085303e+00 -7.86211193e-01 3.32386732e-01
-2.90399492e-01 -8.55058804e-02 8.92473664e-03 6.38900399e-01
-6.14772916e-01 7.96575367e-01 -6.27196968e-01 -1.13180053e+00
1.00564599e-01 -7.88623214e-01 6.88770533e-01 -1.24556315e-03
-9.61565003e-02 -1.01802492e+00 1.00932010e-01 5.26927352e-01
-6.14103302e-02 6.00422502e-01 5.77292562e-01 -6.05960727e-01
-4.75789070e-01 -2.87120968e-01 -2.25249588e-01 1.51519775e-01
4.17861313e-01 -1.09502904e-01 -3.48278195e-01 -8.94945383e-01
1.27175421e-01 1.97615139e-02 5.51269352e-01 4.59769666e-01
7.96042144e-01 -5.38189709e-01 -8.44453990e-01 3.10243428e-01
1.74404907e+00 3.38291407e-01 3.75476927e-01 -2.59550035e-01
4.82975185e-01 4.97391790e-01 5.62651336e-01 6.67931795e-01
3.52677852e-01 3.09435904e-01 2.45565623e-01 -2.11192630e-02
2.42287666e-01 -8.16013664e-03 3.25746864e-01 1.18774939e+00
4.61976901e-02 -1.29626393e-01 -8.81370246e-01 9.78709161e-01
-2.11234808e+00 -1.15587604e+00 -3.86394948e-01 2.13473916e+00
4.95827973e-01 -1.05377577e-01 3.05642426e-01 3.55413377e-01
1.16966176e+00 1.70251638e-01 -2.60079235e-01 1.47713616e-01
-1.22033857e-01 1.09445751e-01 7.15364099e-01 5.44924498e-01
-9.71715748e-01 6.64806128e-01 6.04092216e+00 9.56105113e-01
-5.56763947e-01 7.86823407e-02 8.53547752e-01 1.33102253e-01
1.87601056e-03 4.89207894e-01 -3.23974133e-01 6.21097445e-01
7.63505399e-01 -1.75804853e-01 5.52012682e-01 4.47317898e-01
4.74245399e-01 -4.39306766e-01 -5.84359050e-01 1.09294248e+00
1.22355901e-01 -1.07331181e+00 -5.50384223e-01 4.19115275e-01
9.25615549e-01 1.22200169e-01 -2.06112802e-01 -4.37167764e-01
9.24290538e-01 -9.72025156e-01 2.15495050e-01 4.90784317e-01
6.63302779e-01 -7.91683614e-01 4.66116548e-01 7.72409618e-01
-1.30012405e+00 -2.61272825e-02 -1.18261531e-01 -1.89631209e-01
1.24940366e-01 9.88492191e-01 -6.30923688e-01 8.57606411e-01
8.81232381e-01 7.45788038e-01 -3.57637554e-01 9.10735965e-01
2.07218170e-01 9.82767642e-01 -7.15723395e-01 2.78396774e-02
2.56001770e-01 -7.98973560e-01 7.22357273e-01 8.93374979e-01
4.50520247e-01 6.16552532e-01 4.72658306e-01 4.47462142e-01
1.84588477e-01 1.18509777e-01 -6.79535031e-01 -8.30905437e-02
6.00881219e-01 1.23038900e+00 -1.24689698e+00 -2.69767195e-01
-2.31777102e-01 9.07981038e-01 4.29598652e-02 4.98755187e-01
-5.02120852e-01 -2.28542820e-01 -1.33989200e-01 3.99930209e-01
5.73870063e-01 -4.46713656e-01 -5.08533977e-02 -8.98109317e-01
-2.01463401e-01 -8.92945588e-01 4.89675850e-01 -4.08520788e-01
-1.21759331e+00 3.08249921e-01 1.75759941e-01 -1.19269896e+00
-1.17569165e-02 -1.43385127e-01 -4.10401791e-01 4.08917010e-01
-7.44884849e-01 -8.35546076e-01 -7.28979260e-02 9.61081624e-01
1.25609204e-01 -1.70060351e-01 4.43501770e-01 2.44619027e-01
-5.71412325e-01 1.50408044e-01 8.52877617e-01 3.04675996e-01
3.73634964e-01 -1.18028998e+00 -3.15562695e-01 1.12325275e+00
1.40592635e-01 5.43817341e-01 8.71739268e-01 -8.08319390e-01
-1.15654445e+00 -1.00147939e+00 7.13517785e-01 2.23460812e-02
6.39443874e-01 -3.26200992e-01 -3.17565411e-01 8.45420480e-01
4.63239700e-01 -2.62092147e-02 7.23757148e-01 -2.07562208e-01
1.32084727e-01 -1.26184374e-01 -1.37106287e+00 3.22895408e-01
1.18733072e+00 -2.38438636e-01 -2.13668555e-01 7.88698614e-01
1.74597040e-01 5.39188311e-02 -9.43539858e-01 2.41143733e-01
4.39935587e-02 -1.08853710e+00 7.89021909e-01 -1.20444357e-01
-1.12177566e-01 -5.22234499e-01 -4.86517698e-01 -1.05906463e+00
-5.76210380e-01 -1.13430619e+00 8.93725976e-02 1.19295311e+00
1.52877361e-01 -7.56547093e-01 1.02964222e+00 1.17545691e-03
5.04403353e-01 -4.50163186e-01 -1.28495026e+00 -7.79841542e-01
-6.62526488e-02 -1.11895829e-01 1.90441132e-01 1.18478703e+00
2.36610457e-01 2.72824794e-01 -4.20368344e-01 4.03257519e-01
1.32900667e+00 3.14686537e-01 6.60296321e-01 -1.44136500e+00
-3.69990408e-01 -6.07544892e-02 -3.26167285e-01 -9.89389062e-01
3.26766670e-02 -7.67401993e-01 -1.23155816e-02 -1.39462423e+00
4.58943188e-01 -9.36080754e-01 -2.87957281e-01 1.47950947e-01
3.29700291e-01 4.42273319e-01 -3.07803182e-03 5.69471180e-01
-8.21426868e-01 1.86393440e-01 7.02850461e-01 5.91167174e-02
1.65466920e-01 2.01782197e-01 -7.46398807e-01 7.66858459e-01
6.55870438e-01 -5.69437921e-01 -5.49379587e-01 1.72985002e-01
1.63677752e-01 6.26093268e-01 1.49808809e-01 -1.09857368e+00
5.32203138e-01 -1.24775037e-01 -1.55273462e-02 -5.66847801e-01
3.70400161e-01 -1.05847764e+00 5.52703202e-01 7.17057586e-01
2.49792859e-02 1.48416385e-01 -4.77585971e-01 1.26524115e+00
-1.34807333e-01 -1.50293618e-01 7.47312069e-01 -3.00483525e-01
-3.99886012e-01 3.05260569e-01 -4.28420246e-01 -6.30172566e-02
1.22908914e+00 -2.36315832e-01 5.46557717e-02 -9.62074161e-01
-9.60843980e-01 3.71400982e-01 5.82463264e-01 -2.06844240e-01
3.30696642e-01 -1.22951281e+00 -7.71140397e-01 6.38851710e-03
-2.09132925e-01 -1.44255543e-02 1.72421932e-01 1.29103982e+00
-2.92031825e-01 5.09143353e-01 3.84676576e-01 -6.76135957e-01
-1.33126783e+00 6.75180793e-01 9.59700719e-02 -2.83514977e-01
-4.55548912e-01 6.47316039e-01 -6.68404624e-02 -3.59167159e-01
-1.08007118e-01 2.53396600e-01 1.52740866e-01 -1.91013828e-01
-1.27389207e-01 6.76836908e-01 -2.39884168e-01 -1.04471695e+00
-4.63536024e-01 6.44893348e-01 2.97569215e-01 -3.78917098e-01
1.37861526e+00 -8.56812060e-01 -3.52446586e-01 4.05102402e-01
1.15123439e+00 4.16263640e-01 -8.77093077e-01 -5.26144385e-01
5.34475781e-02 -3.99437398e-01 -9.91894975e-02 -2.12673903e-01
-1.34255624e+00 3.48577350e-01 6.13051355e-01 6.73516452e-01
9.85715270e-01 1.04081310e-01 6.13192618e-01 2.96924412e-01
8.68196726e-01 -1.22372651e+00 -1.39346480e-01 -1.57257468e-01
3.13927799e-01 -1.20902956e+00 2.99852103e-01 -6.01917505e-01
-3.80797565e-01 7.13174999e-01 1.00539468e-01 -5.27717531e-01
1.19936025e+00 1.30890712e-01 -1.64822459e-01 -4.30767834e-01
-4.09345120e-01 -1.36438802e-01 -1.86782748e-01 4.63301808e-01
6.91915378e-02 1.65358469e-01 -4.54253018e-01 2.40355343e-01
-4.51630242e-02 -4.31747884e-02 6.47146165e-01 1.00542784e+00
-4.86711115e-01 -8.11051667e-01 -6.38721943e-01 5.86284697e-01
-3.49870384e-01 4.25745025e-02 -4.44631428e-01 8.27963948e-01
3.48941535e-02 1.42208064e+00 -3.42327237e-01 -3.12659770e-01
-3.75489324e-01 -1.17987163e-01 2.65713364e-01 -5.90576828e-01
-1.74736783e-01 5.12464702e-01 -5.12742661e-02 -4.36833024e-01
-9.39250112e-01 -9.81849313e-01 -1.24415100e+00 -3.40824366e-01
-5.62535882e-01 6.69751942e-01 3.00140530e-01 1.03274906e+00
3.33181232e-01 1.27069978e-02 1.04888058e+00 -3.60954583e-01
-2.17822850e-01 -8.45866501e-01 -1.17131674e+00 3.04192454e-01
-3.68965082e-02 -3.13364685e-01 -6.40893340e-01 1.64452881e-01] | [7.155086994171143, 5.018365383148193] |
dd6a470f-50a4-4f1c-bf76-cb248d35b943 | human-centered-prior-guided-and-task | 2204.12729 | null | https://arxiv.org/abs/2204.12729v1 | https://arxiv.org/pdf/2204.12729v1.pdf | Human-Centered Prior-Guided and Task-Dependent Multi-Task Representation Learning for Action Recognition Pre-Training | Recently, much progress has been made for self-supervised action recognition. Most existing approaches emphasize the contrastive relations among videos, including appearance and motion consistency. However, two main issues remain for existing pre-training methods: 1) the learned representation is neutral and not informative for a specific task; 2) multi-task learning-based pre-training sometimes leads to sub-optimal solutions due to inconsistent domains of different tasks. To address the above issues, we propose a novel action recognition pre-training framework, which exploits human-centered prior knowledge that generates more informative representation, and avoids the conflict between multiple tasks by using task-dependent representations. Specifically, we distill knowledge from a human parsing model to enrich the semantic capability of representation. In addition, we combine knowledge distillation with contrastive learning to constitute a task-dependent multi-task framework. We achieve state-of-the-art performance on two popular benchmarks for action recognition task, i.e., UCF101 and HMDB51, verifying the effectiveness of our method. | ['Gaoang Wang', 'Zhanhao He', 'Yang Zhou', 'Keyu Lu', 'Guanhong Wang'] | 2022-04-27 | null | null | null | null | ['human-parsing', 'self-supervised-action-recognition'] | ['computer-vision', 'computer-vision'] | [ 4.45657670e-01 -2.24799514e-01 -5.91007948e-01 -3.65629673e-01
-8.09924781e-01 -2.04458311e-01 6.03180528e-01 -3.18272829e-01
-3.89642000e-01 7.85906434e-01 5.28925478e-01 3.13918531e-01
1.30275292e-02 -3.12389106e-01 -5.76043189e-01 -7.93663085e-01
3.33456546e-01 2.23591685e-01 5.53667486e-01 -3.31558324e-02
6.67060092e-02 -2.41542328e-02 -1.61123359e+00 6.72023356e-01
9.45503473e-01 1.04224300e+00 2.14868531e-01 1.89323246e-01
-1.61494970e-01 1.20986700e+00 -6.20143831e-01 -4.16993439e-01
2.82978211e-02 -6.13532126e-01 -1.17022359e+00 4.17686343e-01
3.74767900e-01 -3.81575048e-01 -3.37267518e-01 1.06132531e+00
2.95237422e-01 2.44066119e-01 5.94877064e-01 -1.34544468e+00
-7.18829513e-01 3.08225960e-01 -8.20999801e-01 2.11877093e-01
3.51645738e-01 3.06231648e-01 9.48779762e-01 -5.78118145e-01
5.93495548e-01 1.42452419e+00 3.19965452e-01 6.94276631e-01
-9.08583224e-01 -6.88946664e-01 6.04704678e-01 6.23791397e-01
-1.10364664e+00 -4.72515553e-01 9.34075952e-01 -4.30660903e-01
7.26363480e-01 6.97726943e-03 5.16808748e-01 1.59179163e+00
5.31019010e-02 1.29805255e+00 1.24108458e+00 -1.34902567e-01
4.65738475e-02 -1.58698663e-01 -4.17529494e-02 5.15885174e-01
1.83242992e-01 -1.03402331e-01 -5.00578463e-01 1.93317786e-01
8.61484945e-01 1.30576029e-01 -2.75661886e-01 -5.71112335e-01
-1.31979787e+00 4.84789878e-01 1.32098496e-01 5.56289017e-01
-3.09967518e-01 2.37047933e-02 8.37398052e-01 1.48465514e-01
3.64048660e-01 2.31962353e-01 -4.37550455e-01 -2.05236122e-01
-5.17742157e-01 9.34262350e-02 2.82714486e-01 7.16668129e-01
6.72150254e-01 7.48398229e-02 -5.42683423e-01 9.29519176e-01
3.74876648e-01 3.71579438e-01 8.78418088e-01 -9.48632598e-01
5.85993826e-01 6.27279818e-01 4.21821512e-02 -9.22440052e-01
-2.32215956e-01 -2.32794181e-01 -8.96457732e-01 -2.72816233e-03
4.19861436e-01 2.08085869e-02 -8.16577315e-01 1.83121216e+00
3.33885789e-01 4.92379695e-01 3.94341975e-01 9.10427332e-01
9.60811138e-01 4.89665061e-01 5.05011022e-01 -1.85436621e-01
1.24146855e+00 -1.55621779e+00 -9.49933410e-01 -5.81830263e-01
6.39455259e-01 -4.91865069e-01 1.03745234e+00 3.09757829e-01
-8.37063491e-01 -8.41197670e-01 -8.85106385e-01 6.31179735e-02
-1.58981860e-01 2.29358375e-01 6.13896310e-01 2.64161348e-01
-4.86626595e-01 4.49309260e-01 -7.89200783e-01 -2.87892938e-01
7.13111401e-01 -5.73534379e-03 -6.24301791e-01 -3.19497555e-01
-1.08062398e+00 8.89912188e-01 7.21821547e-01 -2.63989400e-02
-1.03690267e+00 -2.88574815e-01 -8.67624938e-01 -2.21077755e-01
8.21468472e-01 -5.72675228e-01 1.22214341e+00 -1.57485092e+00
-1.54938436e+00 9.13631797e-01 6.02569152e-03 -1.55757099e-01
6.02645814e-01 -4.97454792e-01 -6.85432673e-01 2.28279203e-01
1.80984348e-01 6.49538577e-01 1.04311824e+00 -1.27271879e+00
-7.18988240e-01 -3.41857791e-01 2.51181543e-01 3.51913631e-01
-3.76915574e-01 1.00010730e-01 -6.93838656e-01 -8.61052632e-01
-1.11784056e-01 -8.40204775e-01 -1.69226453e-01 -4.95796204e-02
-1.59083053e-01 -3.81210744e-01 9.43126321e-01 -6.62506461e-01
1.09672642e+00 -2.28162885e+00 3.83173972e-01 -4.49989855e-01
-1.13307342e-01 5.19989789e-01 -5.47631264e-01 -5.51476609e-03
-1.27644002e-01 -1.38295129e-01 -1.66746005e-01 -3.61853123e-01
-2.99544603e-01 5.01108944e-01 -1.58014089e-01 3.61732244e-01
3.55234861e-01 9.21908975e-01 -1.11487830e+00 -8.11894238e-01
3.09680432e-01 2.74719775e-01 -2.73662806e-01 3.35591853e-01
-3.58240336e-01 8.60671043e-01 -7.10328162e-01 7.66072929e-01
5.50917804e-01 -4.01140034e-01 2.87984312e-01 -4.87004966e-01
2.13758051e-01 4.42761444e-02 -1.20435345e+00 2.21315694e+00
-3.12524259e-01 1.82790801e-01 -2.29015440e-01 -1.37181556e+00
7.19157338e-01 2.06791550e-01 7.15614438e-01 -7.91205525e-01
9.50527489e-02 1.87531188e-02 -1.19749472e-01 -8.71830046e-01
2.31603071e-01 5.35856187e-03 3.90770696e-02 2.95160830e-01
2.12323293e-01 2.49726519e-01 2.35128924e-01 -1.78128883e-01
1.00049865e+00 8.15279126e-01 5.70881307e-01 -1.34829404e-02
7.65125990e-01 -1.47658065e-01 1.00665808e+00 4.17746633e-01
-6.02437317e-01 6.08408034e-01 3.59225780e-01 -4.60454017e-01
-5.48653424e-01 -7.32059658e-01 2.00962692e-01 1.33727086e+00
4.24331814e-01 -2.84197897e-01 -4.42837983e-01 -1.22948301e+00
-2.10421160e-01 4.13233787e-01 -8.00577939e-01 -2.48703077e-01
-6.83707595e-01 -7.17547715e-01 3.92886907e-01 8.13319445e-01
9.90297258e-01 -1.20717061e+00 -7.10829556e-01 1.53814033e-01
-4.81501043e-01 -1.55911994e+00 -2.87909180e-01 -1.57033414e-01
-9.10855234e-01 -1.27375472e+00 -8.34943295e-01 -6.49302244e-01
6.07617438e-01 5.10471642e-01 1.08288836e+00 3.08929402e-02
8.33633542e-03 4.63720322e-01 -6.95025802e-01 8.01878721e-02
-1.70240089e-01 -2.32733250e-01 -1.36824235e-01 2.69007802e-01
2.96269029e-01 -3.64066005e-01 -5.15675247e-01 5.54170728e-01
-9.63353574e-01 2.40398556e-01 1.00962567e+00 8.88825178e-01
6.29075289e-01 -9.17458683e-02 5.68773985e-01 -7.15325892e-01
1.99148968e-01 -4.21309531e-01 -1.21664196e-01 7.20659912e-01
-1.96105450e-01 2.19189391e-01 5.14159203e-01 -6.59265816e-01
-1.46477270e+00 4.40565437e-01 9.45499241e-02 -5.90570569e-01
-3.67555857e-01 2.34513685e-01 -5.00283539e-01 -1.36008356e-02
3.53115469e-01 4.40859586e-01 6.77830800e-02 -4.07736987e-01
3.70960653e-01 4.20269787e-01 6.32041097e-01 -7.35547423e-01
4.47206914e-01 5.34933388e-01 -1.09894633e-01 -4.96028304e-01
-1.32442820e+00 -3.35184366e-01 -7.82218277e-01 -2.37543702e-01
1.09421480e+00 -1.07002556e+00 -4.76341486e-01 7.82252192e-01
-1.14769042e+00 -3.49344403e-01 -1.54126823e-01 5.71217179e-01
-6.50810421e-01 7.85589635e-01 -3.09923202e-01 -5.95512509e-01
-1.27690479e-01 -1.21033633e+00 1.07693934e+00 2.00605154e-01
-2.96522286e-02 -8.12477648e-01 1.18722968e-01 7.71269619e-01
2.22552806e-01 2.42759019e-01 6.83296621e-01 -6.39008999e-01
-3.92770439e-01 1.57883152e-01 -4.56778556e-01 5.19530773e-01
3.62594604e-01 -1.53657615e-01 -9.41254497e-01 -1.56514883e-01
-1.02465704e-01 -9.01511312e-01 1.07156956e+00 1.99024498e-01
1.42533314e+00 -1.81848988e-01 -3.31707597e-01 4.18372273e-01
1.12044871e+00 2.13907808e-01 8.95017087e-01 4.20394033e-01
7.90950716e-01 6.53933823e-01 9.74126756e-01 4.96245921e-01
5.47895551e-01 7.91341662e-01 3.64099056e-01 1.53644243e-02
-2.16168970e-01 -2.66737849e-01 4.86554712e-01 5.53487062e-01
-3.76446009e-01 -2.21743509e-02 -6.42453551e-01 4.05126899e-01
-2.31408906e+00 -1.07662976e+00 1.64476424e-01 1.97324824e+00
8.20708096e-01 9.96045470e-02 2.78967172e-01 -1.79497451e-01
6.56734407e-01 6.49140060e-01 -7.25154519e-01 3.08092922e-01
-1.70199826e-01 -1.78157181e-01 2.54033327e-01 7.61998296e-02
-1.58930874e+00 1.06737125e+00 5.93133497e+00 8.79663110e-01
-9.57778931e-01 2.83882171e-01 5.19903839e-01 8.55855420e-02
-3.69093269e-02 -1.89326346e-01 -6.55463934e-01 5.10888278e-01
3.02009821e-01 -1.10917568e-01 -7.32493401e-02 1.00651729e+00
-6.35375306e-02 -1.15525529e-01 -1.06619096e+00 1.25701165e+00
4.49103206e-01 -1.06038463e+00 2.51917154e-01 -2.71458626e-01
6.97409093e-01 -2.09999338e-01 -1.12382077e-01 7.47793257e-01
2.60002106e-01 -7.75332451e-01 5.44067860e-01 4.04920220e-01
4.99236405e-01 -4.82152432e-01 7.05217898e-01 2.27475777e-01
-1.37903774e+00 -1.58822075e-01 -2.74426103e-01 6.07762858e-03
1.19287863e-01 2.31744066e-01 -2.42428377e-01 9.62728441e-01
6.70476377e-01 1.23333681e+00 -7.46209085e-01 8.71135592e-01
-3.85058045e-01 2.50960618e-01 1.54429659e-01 2.24231020e-01
3.04641068e-01 -6.97538406e-02 1.80292159e-01 1.10837579e+00
8.89335573e-02 3.16649079e-02 5.81607342e-01 4.83574301e-01
9.12529156e-02 9.75627825e-02 -5.02566755e-01 -1.32739767e-01
1.93925545e-01 1.12095916e+00 -5.54383755e-01 -3.86725545e-01
-8.07627916e-01 1.29037178e+00 4.44478571e-01 4.39891785e-01
-9.90051985e-01 1.92434806e-02 6.49764657e-01 -2.66957700e-01
3.72808307e-01 -2.13075846e-01 8.59078299e-03 -1.48330534e+00
9.10045430e-02 -1.16328907e+00 7.25453436e-01 -6.03872359e-01
-1.25666344e+00 4.54566926e-01 9.79892984e-02 -1.51153302e+00
-2.82829702e-01 -5.00208974e-01 -5.12769401e-01 2.33172849e-01
-1.75016809e+00 -1.31603515e+00 -5.10159016e-01 6.60179377e-01
9.45873857e-01 -2.81228513e-01 6.94830239e-01 4.79333967e-01
-8.88118088e-01 4.61070985e-01 -2.35821098e-01 1.61127046e-01
9.21683431e-01 -8.80759358e-01 7.29686320e-02 9.01558518e-01
1.00807801e-01 1.50476560e-01 3.77989650e-01 -6.23497546e-01
-1.35652483e+00 -1.17672014e+00 4.13053811e-01 -3.81249934e-01
5.67660451e-01 2.50608400e-02 -1.14899492e+00 7.46061921e-01
2.18899012e-01 2.83505529e-01 7.26026595e-01 1.38293535e-01
-5.62590182e-01 -1.27229974e-01 -7.67829299e-01 4.23386216e-01
1.31990647e+00 -3.61714721e-01 -7.67094731e-01 4.28203374e-01
4.78688836e-01 -3.15658540e-01 -6.27606273e-01 6.60051346e-01
4.78023469e-01 -8.85550618e-01 1.00746799e+00 -9.60596502e-01
6.56121492e-01 -3.13186496e-01 -2.11092159e-01 -1.16009283e+00
-4.55133885e-01 -3.17133516e-01 -2.92858899e-01 1.23838198e+00
7.99744502e-02 -2.43561357e-01 6.93367064e-01 4.66201395e-01
-9.83035862e-02 -6.81745768e-01 -8.69525969e-01 -1.11578238e+00
-1.42021418e-01 -2.10120231e-01 3.46651077e-01 1.10961401e+00
-1.24908850e-01 4.37821567e-01 -9.09819245e-01 -1.00213297e-01
5.58139682e-01 2.53225923e-01 8.99131358e-01 -1.09053457e+00
-4.43874449e-01 -3.76968235e-01 -3.27495813e-01 -1.25911331e+00
5.47508836e-01 -4.86986905e-01 1.84793755e-01 -1.56719708e+00
5.20629704e-01 -1.37272775e-01 -5.53054869e-01 8.97889018e-01
-3.91188115e-01 -5.23295254e-02 2.31780916e-01 3.67650926e-01
-1.35353696e+00 8.44950199e-01 1.51533532e+00 -1.88050017e-01
-2.19623223e-02 -1.99062273e-01 -6.48167491e-01 8.70383978e-01
6.35531425e-01 -3.50272655e-01 -5.51426768e-01 -6.72658980e-01
-2.73570627e-01 -6.59728348e-02 3.35096776e-01 -1.03757584e+00
-7.74093298e-03 -5.96280098e-01 3.52805495e-01 -3.61784875e-01
4.11950707e-01 -8.37573886e-01 -4.71214466e-02 4.65712070e-01
-3.19940776e-01 -2.21308574e-01 1.11991495e-01 7.62995481e-01
-5.39744914e-01 -2.27228656e-01 7.99794257e-01 -2.52977461e-01
-1.45763755e+00 3.78882468e-01 4.63060616e-03 3.03000003e-01
1.24791932e+00 -1.28320724e-01 -5.00072360e-01 -2.15520397e-01
-4.92812812e-01 3.08068156e-01 1.97348163e-01 7.51679540e-01
4.71790373e-01 -1.50571299e+00 -6.56024754e-01 -3.44231054e-02
3.83150697e-01 -1.51945397e-01 6.57435179e-01 8.21559548e-01
2.86080930e-02 1.80768654e-01 -6.93858564e-01 -6.13230109e-01
-1.30484998e+00 5.62524378e-01 1.28704146e-01 -4.05845761e-01
-6.74036503e-01 6.50242031e-01 4.02263045e-01 -9.24215540e-02
3.95553172e-01 -2.47439116e-01 -4.25108433e-01 1.82983428e-01
6.67062402e-01 2.96475083e-01 -3.47885072e-01 -7.19071984e-01
-6.46327496e-01 6.32974565e-01 -2.10900962e-01 1.48223609e-01
1.21339178e+00 1.44311050e-02 2.68271476e-01 3.56218010e-01
1.08652532e+00 -5.31970859e-01 -1.56275392e+00 -4.51854020e-01
5.50506115e-02 -8.16917658e-01 -1.29185766e-01 -7.47274935e-01
-1.23832679e+00 7.85846829e-01 5.07644236e-01 -2.38444477e-01
1.23746371e+00 1.81059800e-02 7.57965863e-01 5.42571545e-01
2.30843037e-01 -1.37314355e+00 7.37129629e-01 4.23821002e-01
9.54031169e-01 -1.66089106e+00 1.72318295e-01 -4.16811973e-01
-1.25988650e+00 1.00403619e+00 1.14271843e+00 1.39399558e-01
2.04611838e-01 -1.65654257e-01 2.76128650e-02 2.22398881e-02
-5.89348257e-01 -4.26654041e-01 4.24168766e-01 6.80120766e-01
3.49346399e-01 -1.69990391e-01 -3.27735484e-01 7.69264340e-01
6.61362171e-01 3.27396274e-01 1.07268102e-01 1.09782183e+00
-4.39275712e-01 -1.13394940e+00 -3.31543130e-03 1.13669135e-01
-3.22541177e-01 3.91704440e-01 -3.50012183e-01 7.20063984e-01
1.12791426e-01 7.56539047e-01 -1.68676317e-01 -4.30445284e-01
2.62466550e-01 7.65059218e-02 6.99765921e-01 -5.92829227e-01
-1.68279067e-01 5.21763926e-03 2.00032502e-01 -8.47158074e-01
-1.05188751e+00 -6.90892160e-01 -1.13101900e+00 3.00011635e-01
-8.50013942e-02 -1.57886237e-01 1.41270936e-01 1.27931547e+00
4.57985044e-01 6.51451409e-01 4.61618423e-01 -6.46984577e-01
-6.10250294e-01 -8.60400200e-01 -3.78048956e-01 7.87167668e-01
-1.98831875e-02 -1.07373750e+00 -1.80651203e-01 1.74046531e-01] | [8.517196655273438, 0.779137909412384] |
b55ce6d8-9cda-4054-a96f-dde1ca3c1c9c | 3d-mesh-processing-using-gamer-2-to-enable | 1901.11008 | null | http://arxiv.org/abs/1901.11008v3 | http://arxiv.org/pdf/1901.11008v3.pdf | 3D mesh processing using GAMer 2 to enable reaction-diffusion simulations in realistic cellular geometries | Recent advances in electron microscopy have enabled the imaging of single
cells in 3D at nanometer length scale resolutions. An uncharted frontier for in
silico biology is the ability to simulate cellular processes using these
observed geometries. Enabling such simulations requires watertight meshing of
electron micrograph images into 3D volume meshes, which can then form the basis
of computer simulations of such processes using numerical techniques such as
the Finite Element Method. In this paper, we describe the use of our recently
rewritten mesh processing software, GAMer 2, to bridge the gap between poorly
conditioned meshes generated from segmented micrographs and boundary marked
tetrahedral meshes which are compatible with simulation. We demonstrate the
application of a workflow using GAMer 2 to a series of electron micrographs of
neuronal dendrite morphology explored at three different length scales and show
that the resulting meshes are suitable for finite element simulations. This
work is an important step towards making physical simulations of biological
processes in realistic geometries routine. Innovations in algorithms to
reconstruct and simulate cellular length scale phenomena based on emerging
structural data will enable realistic physical models and advance discovery at
the interface of geometry and cellular processes. We posit that a new frontier
at the intersection of computational technologies and single cell biology is
now open. | [] | 2019-12-17 | null | null | null | null | ['physical-simulations'] | ['miscellaneous'] | [ 3.62362266e-01 -1.87324271e-01 1.00254178e+00 2.94267964e-02
-3.61080825e-01 -4.37410772e-01 4.58925605e-01 5.25030673e-01
-7.65599966e-01 9.55035329e-01 -4.26730663e-01 -5.41747868e-01
1.24452271e-01 -9.57165480e-01 -5.04908264e-01 -7.08998859e-01
-1.28021166e-01 1.22337902e+00 5.13908207e-01 -1.37118787e-01
3.70492458e-01 1.09196448e+00 -1.46955049e+00 -2.70804521e-02
4.72695231e-01 3.72527748e-01 5.54196835e-01 9.55543697e-01
-1.16800651e-01 2.03360450e-02 -8.07352588e-02 -1.15995526e-01
-2.11518295e-02 -1.48868188e-01 -7.09173322e-01 1.33394524e-01
5.37908189e-02 1.43438995e-01 3.00135732e-01 6.66811228e-01
5.37744820e-01 -7.03755720e-03 6.36186540e-01 -5.29801428e-01
-2.53815353e-01 -1.93904772e-01 -1.03052028e-01 2.40777448e-01
1.75258681e-01 2.18479946e-01 2.75762260e-01 -8.43505681e-01
1.21725404e+00 1.27648771e+00 8.53215575e-01 5.03329813e-01
-1.75303543e+00 -1.64244637e-01 -4.15012807e-01 -3.92663151e-01
-1.25693715e+00 -1.94670752e-01 1.61356017e-01 -7.93162405e-01
9.43605423e-01 1.63017750e-01 1.11631966e+00 6.46908998e-01
9.74217415e-01 -2.43407592e-01 1.62820101e+00 -2.63613611e-01
6.93842769e-01 -4.46272105e-01 -1.85050339e-01 5.16223311e-01
4.58037645e-01 -7.04111485e-03 -1.59805939e-02 -3.66541415e-01
1.45646393e+00 4.81207184e-02 -1.29639313e-01 5.60861267e-02
-1.54230845e+00 3.58007133e-01 -1.53995648e-01 4.89584506e-01
-2.52481133e-01 2.93393642e-01 3.22784901e-01 -8.53410289e-02
5.07536471e-01 5.20329177e-01 -4.46814388e-01 -1.36184886e-01
-7.89067984e-01 5.90654671e-01 7.80481696e-01 3.49891514e-01
6.67755902e-01 -1.81493551e-01 6.26375675e-01 3.14290434e-01
5.68789840e-01 4.07532692e-01 3.82600948e-02 -1.24911654e+00
-4.67749953e-01 7.00133145e-01 2.30740905e-01 -6.59412503e-01
-5.25283575e-01 9.73523781e-02 -7.80311048e-01 8.88581276e-01
6.33989096e-01 -5.17703891e-02 -8.66205156e-01 1.01945806e+00
7.82565475e-01 2.06119642e-01 -2.68634260e-01 7.52048075e-01
4.42811191e-01 4.28301781e-01 1.07932337e-01 -2.90874243e-01
1.29804802e+00 6.11583032e-02 -4.02856588e-01 4.24372256e-01
8.16894531e-01 -6.93639457e-01 6.16792738e-01 2.04373181e-01
-1.41774499e+00 -2.93861598e-01 -7.30404317e-01 -2.44510313e-03
-4.89152551e-01 -5.27825117e-01 4.60077554e-01 2.50869602e-01
-1.27865767e+00 1.11138773e+00 -1.28593957e+00 -6.55498683e-01
4.43390518e-01 3.06607902e-01 -5.30072570e-01 3.48730326e-01
-7.26553977e-01 1.03409803e+00 -1.86110631e-01 -1.85639970e-02
-6.71156943e-01 -7.85593927e-01 -4.36660588e-01 -2.08096683e-01
-3.32762241e-01 -1.09567070e+00 8.97447526e-01 -7.11195096e-02
-1.39960349e+00 1.41776371e+00 -1.77219763e-01 -3.24291140e-01
6.89293981e-01 5.96572816e-01 3.12402304e-02 2.77252585e-01
-1.44760475e-01 5.82665443e-01 1.08657517e-01 -1.44614494e+00
-1.63420498e-01 -5.07764995e-01 -4.07096505e-01 -7.86156356e-02
5.40819347e-01 5.52789755e-02 2.21481062e-02 -1.61552310e-01
2.15410694e-01 -8.71209085e-01 -7.45609939e-01 2.97808498e-01
-2.24805623e-02 2.34536141e-01 7.53810942e-01 -3.95948321e-01
4.22277272e-01 -1.67196906e+00 2.62749165e-01 1.63095355e-01
3.38863999e-01 -7.60374544e-03 4.71640557e-01 8.93048882e-01
2.28550479e-01 2.95720607e-01 -3.73836577e-01 -5.77819884e-01
-2.89762467e-01 1.57330483e-01 6.78388309e-03 7.50259817e-01
-2.01029986e-01 7.21952856e-01 -7.51749814e-01 -5.00491261e-01
4.77708846e-01 9.67004836e-01 -3.71268004e-01 -8.10593218e-02
-3.18003595e-01 1.09456265e+00 -4.17735100e-01 4.26608860e-01
6.30186260e-01 -4.07657713e-01 3.90383989e-01 8.87306929e-02
-8.95642042e-01 -3.94524783e-02 -1.00098872e+00 1.50431967e+00
-1.81831673e-01 2.47809574e-01 8.31789017e-01 -3.78553867e-01
8.29843163e-01 4.11857814e-01 4.12392020e-01 -2.16261595e-01
4.65428144e-01 3.62169296e-01 -3.60041787e-03 -7.48693105e-03
5.29504493e-02 -7.62637496e-01 2.96135783e-01 4.90965605e-01
-2.51073956e-01 -7.90155470e-01 1.41785011e-01 1.05300218e-01
1.08861101e+00 3.90551090e-01 3.90679613e-02 -1.00453365e+00
3.51719260e-01 2.92750180e-01 1.72965065e-01 3.96366388e-01
-3.22363712e-02 7.92792439e-01 1.80670410e-01 -8.13011110e-01
-1.84185469e+00 -9.52139914e-01 -7.46912122e-01 2.65088022e-01
3.24128270e-02 -7.95600489e-02 -9.28104758e-01 4.46965754e-01
7.08593354e-02 -5.49782664e-02 -7.97751486e-01 5.00856221e-01
-4.68799025e-01 -9.53734994e-01 4.27420884e-01 -9.60825458e-02
-1.48489863e-01 -1.12740207e+00 -1.10568404e+00 5.86530268e-01
4.66204554e-01 -1.05190098e+00 3.38576764e-01 1.64929509e-01
-1.14508116e+00 -1.05002725e+00 -6.19971991e-01 -6.87321484e-01
8.30106616e-01 1.05336234e-02 9.23103750e-01 7.11364985e-01
-8.46893072e-01 3.38246673e-01 -3.91816162e-02 -3.82327110e-01
-7.58483231e-01 -3.27806264e-01 1.42612711e-01 -5.74271798e-01
-2.55348533e-01 -1.30713916e+00 -8.50725412e-01 3.54711264e-01
-1.03886569e+00 5.88480473e-01 5.02027497e-02 3.09316754e-01
1.33419716e+00 -1.39476806e-01 4.70557570e-01 -9.18795526e-01
4.87715304e-01 -3.23075294e-01 -7.23357677e-01 -1.20092608e-01
-2.35114485e-01 -2.57354885e-01 8.62579048e-01 -6.78084716e-02
-7.73365974e-01 -1.17288597e-01 -5.64325809e-01 -6.98406398e-02
-5.54063976e-01 3.56066227e-01 8.70377645e-02 -3.15442771e-01
3.00851256e-01 1.41884714e-01 4.74435866e-01 -4.38303322e-01
-3.78905326e-01 1.54707313e-01 2.68418938e-01 -7.77104497e-01
3.67186308e-01 1.39073861e+00 7.64808297e-01 -1.03162658e+00
1.25642613e-01 -2.56798714e-01 -7.03300297e-01 -4.31424379e-01
9.81080472e-01 -4.03849095e-01 -1.10734117e+00 6.77642345e-01
-9.53110278e-01 -7.86833107e-01 -3.09871793e-01 2.54585534e-01
-8.09972823e-01 3.04877698e-01 -8.45584273e-01 -7.64179111e-01
-3.53794456e-01 -1.20715952e+00 1.33563030e+00 1.11094110e-01
-3.74439359e-01 -1.34294915e+00 6.08219206e-01 8.59610885e-02
4.21385348e-01 7.05001891e-01 9.81780469e-01 -3.53014059e-02
-6.12417758e-01 -1.22104748e-03 1.85481325e-01 -4.09061939e-01
-8.46211240e-02 5.82889915e-01 -7.47263968e-01 -2.96997815e-01
4.80699651e-02 -7.30806217e-02 4.30005491e-01 5.71348011e-01
7.18643248e-01 4.86728400e-01 -7.29433537e-01 6.41034424e-01
1.78911698e+00 5.07099703e-02 8.86821210e-01 2.80594558e-01
5.92301786e-01 6.19673610e-01 9.42359865e-02 4.01951939e-01
1.55790254e-01 4.88449067e-01 3.15539569e-01 -9.96713862e-02
4.72052582e-03 2.15914533e-01 -3.37653011e-01 9.02494013e-01
-5.25069475e-01 6.78249374e-02 -1.33221960e+00 3.73377353e-01
-1.45192647e+00 -1.00788438e+00 -7.67017126e-01 2.06660414e+00
7.82356024e-01 1.57874733e-01 1.52509287e-02 7.82560855e-02
6.65990770e-01 -6.23315215e-01 -4.33706522e-01 -5.27351916e-01
3.46872695e-02 4.40895617e-01 3.02156866e-01 8.30370665e-01
-5.45955241e-01 8.33378017e-01 6.66417599e+00 2.32076094e-01
-1.07514203e+00 7.76224211e-02 6.26776278e-01 1.37962386e-01
-3.76814276e-01 2.42911518e-01 -8.42795312e-01 6.15257859e-01
1.17867482e+00 -1.24597140e-01 4.10160929e-01 2.34324425e-01
8.25530946e-01 -5.21722317e-01 -8.40761542e-01 4.36853647e-01
-6.53442919e-01 -1.95592654e+00 1.34517863e-01 6.49186194e-01
7.27970123e-01 3.52324486e-01 -3.45822811e-01 -4.74179000e-01
5.16372502e-01 -1.20349133e+00 5.41899621e-01 8.12702835e-01
8.77073824e-01 -5.65251708e-01 4.92864996e-01 5.41516006e-01
-1.07108140e+00 7.59961307e-01 -5.54157257e-01 -3.90029162e-01
8.24114084e-01 7.78797567e-01 -9.80114341e-01 2.55275875e-01
6.17845953e-01 2.96408683e-01 -1.60758302e-01 1.06985843e+00
7.61815310e-01 2.13823453e-01 -4.84938473e-01 1.49790913e-01
-2.16488186e-02 -7.18310535e-01 4.42300707e-01 1.14896846e+00
2.75622964e-01 2.32997954e-01 -1.71048135e-01 9.86465096e-01
2.15768144e-01 -4.82045673e-02 -7.23279655e-01 9.54081342e-02
4.14762020e-01 1.50333083e+00 -1.65194309e+00 -1.89069495e-01
-1.03439152e-01 4.57786113e-01 4.92004335e-01 -6.39403462e-02
-5.25060117e-01 2.54032284e-01 6.96440518e-01 7.06806719e-01
5.29710278e-02 -8.49514067e-01 -5.06376743e-01 -7.23233342e-01
-6.35375142e-01 -2.93210000e-01 -2.50104338e-01 -1.04138744e+00
-1.11922371e+00 2.45020837e-01 -1.92327648e-01 -4.22460616e-01
3.59121293e-01 -7.42226303e-01 -7.58286715e-01 9.81338859e-01
-9.07450855e-01 -9.29804564e-01 -1.18318684e-01 1.09059148e-01
9.96603966e-02 6.75572038e-01 1.13855529e+00 -2.48604324e-02
-2.05313072e-01 -5.96882582e-01 5.78679204e-01 -4.13658112e-01
1.68636411e-01 -1.24607635e+00 8.49314809e-01 2.89296597e-01
-3.55636597e-01 7.10670114e-01 1.05307174e+00 -1.05483329e+00
-1.46815014e+00 -9.92955208e-01 6.40245140e-01 -5.93223155e-01
7.15248883e-01 -4.88582551e-01 -1.17142570e+00 5.01100898e-01
4.90625668e-03 1.47991151e-01 8.55980277e-01 -5.13279021e-01
5.27566314e-01 7.14788377e-01 -1.36701572e+00 7.55476594e-01
1.02906215e+00 -3.68613839e-01 -2.33033061e-01 2.72167444e-01
1.67558923e-01 -1.97166532e-01 -1.36135697e+00 4.47385311e-01
7.12441206e-01 -9.33738828e-01 7.64294147e-01 -3.73398811e-01
2.23756716e-01 -3.48392069e-01 1.13905728e-01 -1.13325667e+00
-1.02054752e-01 -7.42473841e-01 5.55518210e-01 8.97375762e-01
1.28180087e-01 -6.57284558e-01 8.16803753e-01 6.44318938e-01
-3.29788834e-01 -1.03099084e+00 -1.23474360e+00 -3.45589221e-01
6.77811384e-01 -1.24104992e-01 3.17774892e-01 7.18471825e-01
1.17086537e-01 -6.31548464e-02 6.30556941e-01 -2.34717041e-01
8.21555734e-01 6.08452186e-02 4.28819090e-01 -1.73272300e+00
-6.67304173e-03 -1.50003314e-01 -5.79666555e-01 -4.17886853e-01
1.17990367e-01 -6.35321438e-01 -3.64355370e-02 -1.79992902e+00
2.14043066e-01 -4.43958342e-01 5.36157906e-01 -2.60170698e-01
3.51855487e-01 4.25638735e-01 -1.83309346e-01 4.02797431e-01
-2.69017994e-01 1.36838004e-01 1.74353635e+00 5.52774072e-01
1.36937484e-01 -6.25251055e-01 -1.30231932e-01 9.73078370e-01
7.65963852e-01 -3.27140212e-01 -1.01087131e-01 -1.87691316e-01
4.83612686e-01 3.19146365e-02 5.26738226e-01 -1.13810146e+00
3.59015048e-01 -9.18136239e-02 3.99042815e-01 -4.40024048e-01
3.68355900e-01 -8.44367146e-01 8.53096366e-01 4.27905500e-01
-4.27030697e-02 2.00996071e-01 2.90144503e-01 4.63496447e-01
3.41556609e-01 -5.16858809e-02 1.15572083e+00 -8.02268445e-01
-6.10076711e-02 1.91404492e-01 -1.10753882e+00 -9.87842008e-02
1.23497081e+00 -5.38200319e-01 -4.55349147e-01 2.11627543e-01
-1.11543238e+00 -2.56413728e-01 1.58009171e+00 -7.28284538e-01
6.34417892e-01 -9.84767079e-01 -5.30633986e-01 5.49566559e-02
-5.74704051e-01 3.58431786e-01 4.06010985e-01 8.50915253e-01
-1.36461306e+00 2.77284294e-01 -5.73490858e-01 -5.87368965e-01
-1.14358115e+00 2.25677267e-01 5.55451453e-01 -2.61837184e-01
-7.69316614e-01 5.93335211e-01 2.51395106e-01 -5.23237765e-01
-6.16135657e-01 -3.49343777e-01 -3.51070501e-02 -4.72704321e-01
4.67404276e-01 4.15109009e-01 1.46041051e-01 -7.42602944e-01
-3.00017178e-01 7.20355868e-01 1.70676708e-01 -3.02739680e-01
1.66016662e+00 -3.54530215e-01 -4.96623605e-01 7.80666232e-01
7.67999589e-01 -3.97602916e-02 -1.56623673e+00 6.16180420e-01
-2.40524843e-01 -1.54385149e-01 -1.42360643e-01 -3.81886959e-01
-5.36807716e-01 7.58587122e-01 1.32488325e-01 2.78861582e-01
4.12555516e-01 1.12099402e-01 8.14914107e-01 1.22020645e-02
9.07503605e-01 -8.11028779e-01 -4.92312014e-01 5.45691967e-01
5.32271564e-01 -6.16891682e-01 2.53864266e-02 -6.57019913e-01
3.00910175e-01 1.14564860e+00 3.13773274e-01 -5.90341449e-01
7.52879202e-01 1.08653617e+00 -1.41105369e-01 -7.26827979e-01
-1.06879008e+00 1.17831543e-01 -5.67259848e-01 5.77304304e-01
6.01966381e-01 -1.59597427e-01 -4.03077245e-01 9.20390561e-02
8.74324366e-02 1.86721295e-01 8.53998184e-01 1.19248104e+00
-7.22885430e-01 -1.19973314e+00 -5.42371929e-01 3.96112829e-01
-6.46962821e-01 2.62181982e-02 -4.28031772e-01 7.34157920e-01
5.24343401e-02 3.55313778e-01 2.90534049e-01 7.61193559e-02
1.08297523e-02 1.15957312e-01 8.40258956e-01 -8.09536934e-01
-5.32901943e-01 -6.46871403e-02 -6.66532367e-02 -2.02948749e-01
-4.11625922e-01 -8.94156694e-01 -1.90955091e+00 -6.82356358e-01
-1.39507651e-01 2.55274385e-01 9.17102456e-01 9.69527364e-01
4.90642458e-01 3.41888696e-01 -7.53981769e-02 -1.64882052e+00
1.60149902e-01 -5.67282081e-01 -8.88030171e-01 3.71654809e-01
-1.31027952e-01 -5.50293446e-01 -4.40461069e-01 4.77130532e-01] | [13.635560035705566, -3.0808465480804443] |
34c4479f-cb67-4674-8b31-6ecba00c620e | a-critical-re-evaluation-of-benchmark | 2307.01231 | null | https://arxiv.org/abs/2307.01231v1 | https://arxiv.org/pdf/2307.01231v1.pdf | A Critical Re-evaluation of Benchmark Datasets for (Deep) Learning-Based Matching Algorithms | Entity resolution (ER) is the process of identifying records that refer to the same entities within one or across multiple databases. Numerous techniques have been developed to tackle ER challenges over the years, with recent emphasis placed on machine and deep learning methods for the matching phase. However, the quality of the benchmark datasets typically used in the experimental evaluations of learning-based matching algorithms has not been examined in the literature. To cover this gap, we propose four different approaches to assessing the difficulty and appropriateness of 13 established datasets: two theoretical approaches, which involve new measures of linearity and existing measures of complexity, and two practical approaches: the difference between the best non-linear and linear matchers, as well as the difference between the best learning-based matcher and the perfect oracle. Our analysis demonstrates that most of the popular datasets pose rather easy classification tasks. As a result, they are not suitable for properly evaluating learning-based matching algorithms. To address this issue, we propose a new methodology for yielding benchmark datasets. We put it into practice by creating four new matching tasks, and we verify that these new benchmarks are more challenging and therefore more suitable for further advancements in the field. | ['Themis Palpanas', 'Peter Christen', 'Nishadi Kirielle', 'George Papadakis'] | 2023-07-03 | null | null | null | null | ['entity-resolution'] | ['natural-language-processing'] | [-1.00681223e-01 -1.33239534e-02 -3.19740504e-01 -4.29966211e-01
-1.10301316e+00 -5.50580919e-01 6.65846467e-01 6.85876131e-01
-5.81049562e-01 7.69164264e-01 -6.48382157e-02 -1.65779218e-01
-6.14126801e-01 -9.51808691e-01 -8.95633101e-01 -3.57402295e-01
-2.40723789e-01 7.54995823e-01 1.23027675e-01 -1.42269984e-01
1.24964684e-01 4.87808555e-01 -1.62736845e+00 3.77107322e-01
9.46506977e-01 1.01446605e+00 -2.26998091e-01 2.43789062e-01
-1.46004111e-01 6.17664516e-01 -4.77765769e-01 -1.01597476e+00
4.31688070e-01 -1.34621888e-01 -9.17048573e-01 -6.01607740e-01
6.83468759e-01 7.32768793e-03 -4.31650817e-01 1.03384578e+00
6.61450028e-01 -1.59863040e-01 5.02395332e-01 -1.27557588e+00
-7.77446687e-01 9.17274475e-01 -2.11127266e-01 3.03494215e-01
7.20977247e-01 -2.52052665e-01 1.42999887e+00 -8.68417501e-01
8.17317307e-01 1.00744998e+00 1.02773654e+00 2.72699207e-01
-1.07724977e+00 -7.31778860e-01 -3.34852159e-01 5.07068038e-01
-1.66342366e+00 -6.34798884e-01 4.15661663e-01 -4.30979401e-01
8.16601098e-01 4.75281358e-01 1.94406733e-01 9.43776011e-01
4.43769433e-02 7.40871668e-01 9.35876727e-01 -5.62316477e-01
1.02164648e-01 4.33457404e-01 2.50214875e-01 4.21319425e-01
5.96445918e-01 2.38171816e-01 -3.76886785e-01 -3.46004754e-01
3.21323663e-01 -1.92614093e-01 -2.38956228e-01 -6.85543656e-01
-1.08028162e+00 6.51751578e-01 4.60598797e-01 7.82758176e-01
-2.55958855e-01 -2.34582663e-01 4.62148041e-01 5.99509180e-01
3.34203243e-01 8.49932134e-01 -4.67295855e-01 7.42454752e-02
-9.49019194e-01 6.31540239e-01 1.05873132e+00 9.96157229e-01
7.08859384e-01 -5.94570756e-01 -2.43047297e-01 7.51878977e-01
9.68805607e-03 1.29318282e-01 2.63910949e-01 -6.08742535e-01
7.49649107e-01 8.96059394e-01 3.64365667e-01 -1.25316525e+00
-5.35023630e-01 -4.14994895e-01 -7.41499305e-01 -3.00391108e-01
6.60203457e-01 1.59016684e-01 -3.53351921e-01 1.73369789e+00
1.43215716e-01 1.42840073e-01 1.77512556e-01 7.02972114e-01
1.06265450e+00 2.35216647e-01 1.04864910e-01 -4.98839319e-02
1.28274024e+00 -7.32941091e-01 -6.98537409e-01 -2.13108826e-04
9.21776533e-01 -6.95256054e-01 9.41106379e-01 1.18214682e-01
-1.31563783e+00 -5.89602172e-01 -1.08487642e+00 5.24248295e-02
-7.97186017e-01 1.81848541e-01 6.72218502e-01 5.98280311e-01
-9.43472385e-01 7.75870085e-01 -4.72520202e-01 -3.48946154e-01
2.21526548e-01 3.70442480e-01 -6.25688910e-01 -1.72931418e-01
-1.53718972e+00 1.25953436e+00 4.32175845e-01 6.90301657e-02
-3.28655243e-01 -6.63307071e-01 -6.32935166e-01 3.07867795e-01
2.67196357e-01 -6.07635081e-01 1.21274090e+00 -7.61627972e-01
-7.32186854e-01 1.31466353e+00 9.78035554e-02 -6.76231384e-01
7.69966006e-01 -1.88621193e-01 -5.79217434e-01 -3.66288722e-01
6.67541176e-02 1.86825529e-01 1.46412343e-01 -1.21993911e+00
-8.53864372e-01 -3.87664735e-01 2.74957329e-01 -1.23900041e-01
-3.49402457e-01 1.69816285e-01 -3.82891893e-01 -4.95987922e-01
-9.07465741e-02 -8.55000794e-01 -1.02379583e-01 -1.82611808e-01
-3.37215036e-01 -5.00852406e-01 6.84957802e-02 -5.02175748e-01
1.41806197e+00 -2.07107782e+00 8.81592408e-02 1.53117985e-01
2.45362639e-01 4.77496982e-01 -8.31137449e-02 5.76603115e-01
-1.70300171e-01 1.87511742e-01 -1.54811993e-01 -3.11580628e-01
3.93301755e-01 4.43441533e-02 -2.90007859e-01 4.04925525e-01
9.03024152e-02 1.00940084e+00 -8.29838097e-01 -5.52818179e-01
-1.28911242e-01 2.91866511e-01 -2.10043952e-01 3.79439920e-01
9.42926556e-02 2.50951033e-02 -2.62726158e-01 6.62822664e-01
7.57339239e-01 -1.93311036e-01 3.06538284e-01 -3.64226967e-01
2.05450542e-02 5.22834480e-01 -1.55057251e+00 1.37804818e+00
-2.70288050e-01 4.13442284e-01 -2.85382152e-01 -1.27083492e+00
9.63962734e-01 3.30628306e-01 5.00727832e-01 -1.03788328e+00
-9.05667916e-02 6.11057580e-01 -4.55560684e-02 -5.19634724e-01
6.47482455e-01 2.13258788e-01 -1.15047224e-01 3.31175476e-01
3.75816319e-03 4.62726653e-01 4.67657626e-01 -2.58737832e-01
1.14197195e+00 -1.63025558e-01 4.49081272e-01 -1.54625967e-01
6.43222332e-01 -4.14653979e-02 6.02041364e-01 9.11032975e-01
-9.81709734e-02 4.40398067e-01 2.60390073e-01 -7.38867402e-01
-1.11786377e+00 -9.35713053e-01 -3.86136144e-01 1.07460308e+00
2.79866457e-01 -4.26012874e-01 -7.26625860e-01 -6.13576591e-01
2.93037355e-01 3.19788694e-01 -6.30306840e-01 -2.04121009e-01
-7.25382209e-01 -8.36798131e-01 9.14001167e-01 6.38576388e-01
3.35195184e-01 -9.76035655e-01 -4.69123334e-01 1.82850212e-01
-3.39860708e-01 -1.14565778e+00 -1.28975362e-01 7.21597001e-02
-7.03416169e-01 -1.33781409e+00 -5.95261395e-01 -7.76547551e-01
3.76739979e-01 5.68982884e-02 1.52334964e+00 2.78923869e-01
-7.09296837e-02 1.02288969e-01 -2.66734689e-01 -2.42086202e-01
-5.29469550e-01 5.46708345e-01 6.11214619e-03 -6.07947931e-02
7.40582407e-01 -3.11449736e-01 -4.06403720e-01 4.92831588e-01
-8.29636216e-01 -3.68057847e-01 8.22318852e-01 6.87663496e-01
5.07061958e-01 -1.47449076e-01 6.23860121e-01 -1.01283526e+00
7.13757038e-01 -6.87826872e-01 -6.56695724e-01 6.61886036e-01
-1.04835343e+00 2.77729869e-01 5.63663960e-01 -2.87416846e-01
-6.47970557e-01 -1.92216039e-01 -1.81424156e-01 -1.74843475e-01
-1.06326766e-01 6.28224254e-01 -2.08426684e-01 -1.62841111e-01
7.94942260e-01 -2.99834684e-02 -5.42138577e-01 -6.71777368e-01
1.59706116e-01 6.43104017e-01 6.53581679e-01 -6.11077547e-01
9.00190532e-01 1.45816743e-01 -1.38435870e-01 -7.44360387e-02
-8.36254060e-01 -4.37990814e-01 -6.05717301e-01 2.02441722e-01
3.89857203e-01 -8.03967595e-01 -5.73306441e-01 2.65168697e-01
-1.13308322e+00 1.30999893e-01 -1.57857642e-01 4.14498657e-01
-3.56574416e-01 3.09204549e-01 -6.34736538e-01 -5.09354711e-01
-5.18104494e-01 -1.14770365e+00 9.09065247e-01 2.00215027e-01
-3.32537055e-01 -9.68321741e-01 4.20242727e-01 2.00089902e-01
5.68862259e-01 3.47353578e-01 1.01938987e+00 -1.09322143e+00
-5.59605598e-01 -4.19844478e-01 -3.59590530e-01 -3.98760587e-02
-1.07355267e-01 -7.70557523e-02 -9.26480711e-01 -5.28743029e-01
-3.59411657e-01 -1.30499586e-01 7.32307553e-01 -7.42662400e-02
9.35326040e-01 -1.41901731e-01 -5.96512914e-01 6.08069956e-01
1.63046527e+00 2.29331240e-01 7.05557585e-01 9.48158860e-01
4.18874502e-01 6.14867747e-01 7.59818912e-01 1.41742602e-01
6.40634298e-01 1.07207906e+00 2.13679045e-01 -2.36766621e-01
2.61555821e-01 -2.53390521e-01 -8.80905241e-02 6.20008290e-01
-1.10693000e-01 -2.90042371e-01 -1.14186800e+00 6.77458465e-01
-1.93796778e+00 -9.02433097e-01 -7.58275017e-02 2.40030432e+00
8.32129419e-01 2.11385757e-01 6.05967529e-02 2.99661338e-01
6.41797185e-01 -7.57054016e-02 -3.76396567e-01 -2.00653896e-01
-2.68208474e-01 2.71383941e-01 4.33027416e-01 6.10755496e-02
-1.33227587e+00 5.56012154e-01 6.66789436e+00 5.52242994e-01
-8.25007379e-01 -1.04993343e-01 3.62998217e-01 4.21668202e-01
-3.94787043e-01 1.43347457e-02 -8.96203518e-01 4.87817794e-01
9.34481859e-01 -3.95448267e-01 2.52265275e-01 8.94880533e-01
-3.76687676e-01 2.78812081e-01 -1.75848877e+00 1.09810269e+00
-1.60528049e-01 -1.30216610e+00 -4.97388244e-02 -1.80329103e-02
5.70366800e-01 -4.30820743e-03 3.32643423e-04 5.53135753e-01
5.87232895e-02 -1.13974512e+00 5.23934424e-01 7.46469676e-01
4.67291415e-01 -5.73024571e-01 1.21860957e+00 1.47100419e-01
-1.23152399e+00 -1.45420119e-01 -4.56240833e-01 1.85365543e-01
-3.08388412e-01 4.91396248e-01 -6.24912441e-01 1.00765479e+00
9.82504368e-01 4.33188766e-01 -7.85897315e-01 1.37859321e+00
1.55851305e-01 2.50306785e-01 -2.49887303e-01 2.01836765e-01
-2.55992785e-02 1.25092760e-01 3.33343357e-01 1.46198165e+00
3.39606941e-01 -3.99345517e-01 -2.42553204e-02 8.38910758e-01
-4.34101343e-01 3.11754912e-01 -5.81764698e-01 1.43890947e-01
8.20551574e-01 1.13075852e+00 -2.79623121e-01 -1.89202338e-01
-5.40471613e-01 4.76148307e-01 6.80821836e-01 2.27743052e-02
-8.96223187e-01 -4.28846389e-01 5.96788943e-01 3.44099663e-02
1.25510126e-01 1.22880615e-01 -1.66914150e-01 -1.27575541e+00
4.91641343e-01 -1.22633600e+00 8.85924757e-01 -3.00465167e-01
-1.50913358e+00 6.98630631e-01 3.25352252e-02 -1.24129796e+00
-2.53310353e-01 -3.68319035e-01 -3.90796393e-01 6.84202611e-01
-1.78775597e+00 -8.18574429e-01 -4.90294278e-01 3.14833820e-01
6.04664981e-02 -2.92606860e-01 9.64673638e-01 1.17148495e+00
-5.90088546e-01 1.25060415e+00 3.24325800e-01 2.29439780e-01
9.68243718e-01 -1.33381486e+00 6.53204858e-01 6.75175190e-01
3.83401841e-01 6.46393061e-01 7.03795671e-01 -8.92116874e-02
-1.54503620e+00 -8.38241041e-01 1.35020530e+00 -6.85222805e-01
4.62581694e-01 -2.48412371e-01 -1.25911963e+00 5.97698867e-01
-9.97027978e-02 1.59720834e-02 6.55838847e-01 2.60927111e-01
-4.75076973e-01 -3.85089070e-01 -1.17414820e+00 2.80484617e-01
1.10144448e+00 -5.69367349e-01 -7.59683788e-01 1.54274061e-01
4.45311844e-01 -4.65345681e-01 -1.35258055e+00 8.44363987e-01
6.46626890e-01 -9.95968938e-01 1.14007235e+00 -8.26438427e-01
3.66501153e-01 -4.92988527e-02 -2.60200739e-01 -1.06932092e+00
-2.65673816e-01 -2.38406882e-01 -2.45394439e-01 1.61812735e+00
5.86784303e-01 -7.25206256e-01 7.71619260e-01 7.10215867e-01
1.78006440e-01 -9.50126112e-01 -9.02132452e-01 -7.41568625e-01
1.68283224e-01 -1.02688782e-01 1.02649736e+00 1.40689600e+00
-1.59178138e-01 9.11235511e-02 -2.47333273e-01 2.63876975e-01
3.69473338e-01 4.75880355e-01 9.36533570e-01 -1.47610319e+00
-9.16078016e-02 -6.70791090e-01 -5.48081756e-01 -6.41484737e-01
9.66710225e-02 -1.03609419e+00 -3.11901987e-01 -1.39075863e+00
3.67053986e-01 -8.32905889e-01 -6.36742234e-01 3.94307554e-01
-2.70473212e-01 -1.23116188e-02 4.99831624e-02 4.71279860e-01
-6.90197289e-01 1.63100615e-01 5.77428520e-01 -2.48001203e-01
5.40566863e-03 8.76213163e-02 -7.14941978e-01 4.38289851e-01
6.17074728e-01 -6.57668412e-01 -5.44579625e-02 -4.65990037e-01
5.91034174e-01 -6.99073747e-02 2.32951269e-01 -1.03296626e+00
4.57477897e-01 1.03417620e-01 2.07254961e-01 -4.50186700e-01
-1.55434400e-01 -8.12676191e-01 4.68540341e-01 3.30446392e-01
-5.94841957e-01 4.31340247e-01 8.71434510e-02 3.39813024e-01
-4.72166419e-01 -2.06014991e-01 6.26598418e-01 -6.39601797e-02
-6.16549730e-01 2.72829592e-01 2.22985893e-01 2.79372364e-01
8.50246727e-01 -9.62690711e-02 -3.47197860e-01 -6.47507235e-02
-4.85729307e-01 3.64666432e-01 4.78700697e-01 5.88824749e-01
3.01811665e-01 -1.39752471e+00 -9.31184411e-01 -3.73454578e-02
4.42121923e-01 -3.27694833e-01 -1.85627639e-01 7.86842585e-01
-4.19439971e-01 6.85943663e-01 -2.97432721e-01 -4.35600668e-01
-1.16938365e+00 7.84132183e-01 6.85000658e-01 -7.48539150e-01
-3.55332315e-01 4.55172658e-01 -1.40368730e-01 -8.41531575e-01
6.35636687e-01 -2.34663844e-01 -4.10310805e-01 1.71420991e-01
3.86198342e-01 5.36567807e-01 4.77852851e-01 -4.63050067e-01
-4.09860075e-01 4.50353473e-01 -9.70741659e-02 5.00153422e-01
1.33780658e+00 2.08147950e-02 -1.31829575e-01 3.04908544e-01
1.14347136e+00 -1.14960268e-01 -4.44045871e-01 -6.38870776e-01
8.53536725e-01 -4.82746899e-01 -3.56210947e-01 -6.54747903e-01
-9.55479562e-01 5.36902905e-01 8.06407869e-01 6.42499566e-01
1.01087201e+00 -1.08245738e-01 8.28499138e-01 6.55426800e-01
5.04383087e-01 -9.38644290e-01 -4.58974004e-01 2.91990250e-01
6.24398589e-01 -1.40001535e+00 -9.97968093e-02 -2.34231159e-01
-1.76326320e-01 1.01117957e+00 6.17777348e-01 1.89750809e-02
3.88314992e-01 2.66983360e-01 -1.89610347e-02 -3.08082193e-01
-7.26184785e-01 -1.77549303e-01 4.79620993e-01 2.63003200e-01
6.07540667e-01 -6.57749921e-02 -7.37494230e-01 7.20542908e-01
-1.68908685e-01 5.52994572e-02 8.77977163e-02 7.08304167e-01
-9.31739807e-03 -1.30638742e+00 -1.88009068e-01 3.64784688e-01
-6.17095530e-01 1.16133288e-01 -5.83975196e-01 1.05401099e+00
1.29431978e-01 6.80437505e-01 -2.46690542e-01 -4.74910587e-01
8.90164793e-01 1.12490356e-01 3.77480686e-01 -2.33089343e-01
-7.72039950e-01 -7.88262725e-01 2.76231349e-01 -5.11073112e-01
-5.24877250e-01 -5.26250124e-01 -8.34997058e-01 -4.16857272e-01
-5.88151276e-01 4.17746991e-01 4.41103846e-01 8.83018494e-01
3.70701730e-01 1.75921321e-01 5.44377863e-01 -4.79368180e-01
-8.32206547e-01 -7.71143377e-01 -2.08485171e-01 7.71515727e-01
1.53894097e-01 -7.90757179e-01 -2.23570973e-01 -4.33507621e-01] | [9.456721305847168, 8.457087516784668] |
c8340226-8335-4304-b38a-f191d4e20ee3 | symmetry-breaking-for-k-robust-multi-agent | 2102.08689 | null | https://arxiv.org/abs/2102.08689v2 | https://arxiv.org/pdf/2102.08689v2.pdf | Symmetry Breaking for k-Robust Multi-Agent Path Finding | During Multi-Agent Path Finding (MAPF) problems, agents can be delayed by unexpected events. To address such situations recent work describes k-Robust Conflict-BasedSearch (k-CBS): an algorithm that produces coordinated and collision-free plan that is robust for up to k delays. In this work we introducing a variety of pairwise symmetry breaking constraints, specific to k-robust planning, that can efficiently find compatible and optimal paths for pairs of conflicting agents. We give a thorough description of the new constraints and report large improvements to success rate ina range of domains including: (i) classic MAPF benchmarks;(ii) automated warehouse domains and; (iii) on maps from the 2019 Flatland Challenge, a recently introduced railway domain where k-robust planning can be fruitfully applied to schedule trains. | ['Peter J. Stuckey', 'Jiaoyang Li', 'Daniel Harabor', 'Zhe Chen'] | 2021-02-17 | null | null | null | null | ['multi-agent-path-finding'] | ['playing-games'] | [ 2.00372770e-01 4.66760635e-01 -1.38724580e-01 -2.94818014e-01
-9.59786355e-01 -1.12291276e+00 5.27226210e-01 4.97793347e-01
-3.53672415e-01 1.37771380e+00 1.92485064e-01 -2.88551331e-01
-1.24594462e+00 -8.34594250e-01 -5.73727787e-01 -4.51995075e-01
-1.04511011e+00 1.52346766e+00 9.52997267e-01 -8.98924589e-01
4.81439292e-01 8.16912591e-01 -9.61719871e-01 2.12871790e-01
4.36763495e-01 3.57433587e-01 2.99445689e-01 6.86973512e-01
5.29244393e-02 5.28291941e-01 -6.11957669e-01 9.83304009e-02
5.10966241e-01 -1.73860326e-01 -1.47268379e+00 -2.84018158e-03
-5.36171138e-01 -9.49274972e-02 5.18821664e-02 5.55824816e-01
1.31798327e-01 2.19846278e-01 2.68945277e-01 -2.17783308e+00
3.19678247e-01 1.08666146e+00 -6.36988282e-01 3.03488821e-01
9.61780369e-01 -5.61726242e-02 7.24804521e-01 -1.24891967e-01
1.29394770e+00 1.22977388e+00 5.14716625e-01 2.07392514e-01
-1.15624690e+00 -2.55720556e-01 4.62514788e-01 4.15577054e-01
-1.19387555e+00 -3.32875162e-01 3.53857934e-01 2.05442026e-01
1.62994480e+00 5.83809197e-01 4.20525193e-01 4.28625315e-01
7.08459318e-01 4.84289676e-01 1.05144215e+00 -3.17172825e-01
4.24760938e-01 -4.22407955e-01 -4.97865677e-02 2.81918854e-01
1.87992811e-01 3.48458201e-01 -5.52628696e-01 -5.12010932e-01
4.43293661e-01 -5.84466159e-01 2.38414146e-02 -4.02927130e-01
-1.81976593e+00 7.31881738e-01 1.25708625e-01 1.76986620e-01
-5.37380040e-01 2.67084330e-01 4.67638522e-01 6.27554238e-01
-3.45179260e-01 8.51389587e-01 -6.29197061e-01 -2.77138591e-01
-2.96188414e-01 1.07714379e+00 1.01815319e+00 1.52303267e+00
5.37782431e-01 -3.82225364e-01 1.18021868e-01 1.36760473e-01
7.36785159e-02 2.30151460e-01 -7.40359426e-01 -1.44283247e+00
8.26193213e-01 2.71538258e-01 8.35943222e-01 -8.98723245e-01
-1.17125130e+00 3.58162075e-01 -1.94181129e-01 5.04582286e-01
2.94096798e-01 -1.22659095e-01 -6.18439257e-01 1.34976768e+00
5.53923845e-01 -8.90434384e-02 4.58546638e-01 7.70835221e-01
2.17768312e-01 7.27625012e-01 -3.86264950e-01 -7.11639881e-01
1.04051888e+00 -1.18423474e+00 -7.90507197e-01 -1.41545340e-01
7.63012528e-01 -7.96111524e-01 8.95070434e-02 5.88278472e-01
-1.69192779e+00 3.36563736e-01 -1.10731530e+00 5.72423100e-01
-4.35762584e-01 -1.23750353e+00 8.27549517e-01 3.23799312e-01
-1.29190207e+00 3.05409700e-01 -7.80332088e-01 -4.97837603e-01
-9.93838236e-02 5.90252101e-01 -6.85412288e-01 -6.62628829e-01
-9.30472612e-01 1.21018481e+00 4.38241571e-01 -8.87507871e-02
-9.98405218e-01 -4.23709929e-01 -6.01520836e-01 -2.50782669e-01
1.33026922e+00 -2.57035315e-01 1.44578159e+00 -5.31382747e-02
-1.23373485e+00 2.04499453e-01 2.21969321e-01 -2.39022672e-01
4.21316534e-01 2.49386832e-01 -6.12015486e-01 8.51161480e-02
5.06598830e-01 4.17413086e-01 -1.89026326e-01 -1.48375535e+00
-1.22504628e+00 -2.13959627e-02 3.28549832e-01 3.27185303e-01
7.24372387e-01 5.09312034e-01 -6.66027889e-02 -8.46035555e-02
3.06124657e-01 -1.21822381e+00 -1.06734836e+00 -6.47765577e-01
-5.48295379e-01 -4.03464019e-01 2.96398968e-01 5.50446883e-02
8.09977233e-01 -1.49011004e+00 3.90748888e-01 8.80328000e-01
-2.86277711e-01 -3.96619022e-01 -6.57311916e-01 1.45453489e+00
1.30657718e-01 -4.95437980e-02 -1.18551575e-01 3.02469373e-01
3.04509342e-01 7.41706371e-01 -9.12506357e-02 3.82852554e-01
1.38219103e-01 6.90797567e-01 -1.10723031e+00 -3.53863478e-01
-9.21165124e-02 -5.97917378e-01 -3.85672241e-01 -1.81008130e-01
-4.07827467e-01 1.32905409e-01 -5.85158288e-01 7.90141165e-01
7.80420363e-01 4.75847840e-01 6.47636235e-01 4.99263823e-01
-7.73076534e-01 8.33086967e-02 -1.61876059e+00 1.99398971e+00
3.04561675e-01 -2.70599369e-02 4.56000865e-01 -4.87738729e-01
6.99817061e-01 2.71131337e-01 8.72846961e-01 -8.77345979e-01
-1.92390308e-01 3.91012192e-01 -7.82279968e-02 -2.77170241e-01
8.62569153e-01 1.03694618e-01 -7.93121994e-01 8.39576840e-01
-6.03190899e-01 -4.00980592e-01 6.80802166e-01 3.51487398e-01
1.82332039e+00 -8.69773030e-02 2.26636872e-01 -5.75391829e-01
1.92246273e-01 1.08946407e+00 8.21802139e-01 1.01074934e+00
-3.07099730e-01 -5.48109561e-02 4.22771305e-01 -8.92601252e-01
-6.42134726e-01 -1.01088178e+00 4.04342085e-01 8.67287397e-01
7.62926340e-01 -5.10319591e-01 -3.41872983e-02 -5.66797137e-01
1.23479798e-01 8.70474160e-01 -3.84241343e-01 3.50076497e-01
-1.14922130e+00 -4.88414824e-01 4.55487192e-01 4.38699424e-01
9.84941199e-02 -1.02397668e+00 -1.04867423e+00 9.31518734e-01
-2.35551476e-01 -1.32829297e+00 -3.85626525e-01 6.14398956e-01
-2.51319110e-01 -1.47173667e+00 1.88750252e-02 -7.89070845e-01
5.43594658e-01 5.79800367e-01 1.00081718e+00 -2.42787860e-02
-1.42732084e-01 4.83799756e-01 -7.18124032e-01 -5.64395964e-01
-3.30577224e-01 -3.76668870e-02 2.97518652e-02 -1.10360479e+00
-1.40709549e-01 -2.78725266e-01 -6.71494529e-02 9.87778246e-01
-5.40297031e-01 -1.23762585e-01 2.18905270e-01 2.52524376e-01
6.45638466e-01 8.99875522e-01 8.27844024e-01 -6.55415416e-01
7.50006557e-01 -5.01925409e-01 -7.13427544e-01 5.53263903e-01
-5.59724092e-01 -3.22028041e-01 1.18473917e-01 2.47092828e-01
-9.42976177e-01 6.46100119e-02 2.58408576e-01 3.88574094e-01
-2.80597836e-01 8.77518892e-01 1.43715814e-01 -3.25299323e-01
3.56686503e-01 -2.98736095e-01 -2.51156539e-01 1.69478521e-01
1.97190583e-01 -1.56071499e-01 5.72316110e-01 -8.72992396e-01
8.12843859e-01 4.43974525e-01 6.00843489e-01 -7.96897858e-02
2.88083881e-01 -4.18439746e-01 -3.45192224e-01 -3.21348220e-01
4.24367070e-01 -4.14866537e-01 -9.16637599e-01 1.07873857e-01
-1.27427518e+00 -8.31367731e-01 -1.97091445e-01 2.11340517e-01
-1.17588770e+00 5.09563312e-02 -5.50544918e-01 -8.42126071e-01
2.15096787e-01 -9.91487265e-01 5.12729228e-01 1.02762699e-01
-3.21596414e-01 -5.75382829e-01 4.93662417e-01 7.35597759e-02
3.93076301e-01 8.10864270e-01 9.02444363e-01 -7.31233001e-01
-9.59875464e-01 5.50274365e-02 1.52611993e-02 -9.34023976e-01
-9.53132659e-02 -7.88636040e-03 2.50335783e-01 -5.17181695e-01
-7.82976389e-01 -2.10562065e-01 -1.51086986e-01 2.99848855e-01
1.06420137e-01 -2.76623994e-01 -1.09696937e+00 -3.56071949e-01
1.48235321e+00 9.26123619e-01 8.26586187e-01 1.08945072e+00
-2.67836452e-01 1.10475612e+00 1.67785215e+00 6.10981226e-01
1.04039919e+00 9.29802477e-01 6.55147493e-01 1.48075745e-01
5.12415171e-01 3.72414291e-01 1.31772980e-01 3.75404298e-01
-4.07000929e-01 -7.30197251e-01 -1.28588963e+00 8.67130518e-01
-2.48520350e+00 -1.00413156e+00 -3.80529642e-01 1.57117307e+00
3.84336323e-01 4.20885503e-01 4.74734604e-01 2.38941491e-01
5.37104964e-01 -1.46502659e-01 -1.89046919e-01 -8.21862996e-01
-1.08150646e-01 -6.78347871e-02 8.00552070e-01 6.38278961e-01
-7.25457072e-01 8.09448719e-01 7.26632929e+00 2.83368886e-01
-1.29201142e-02 1.01273969e-01 -1.90273896e-01 -2.06797466e-01
-4.16116387e-01 1.57150865e-01 -3.88754308e-01 -2.35040650e-01
1.02216542e+00 -4.81725335e-01 6.78406239e-01 5.56068897e-01
2.89141744e-01 -4.88969177e-01 -9.76382911e-01 1.90400183e-01
-4.54146981e-01 -1.77143002e+00 -7.61909604e-01 8.70312452e-02
9.03846323e-01 2.27675028e-02 -5.05713105e-01 7.08737969e-02
1.01051867e+00 -1.02549565e+00 9.47930336e-01 1.34865806e-01
5.94040975e-02 -1.58563805e+00 7.57804930e-01 3.00547987e-01
-1.36498821e+00 -3.06446850e-01 -9.41543803e-02 -1.02589972e-01
1.03767431e+00 7.74948224e-02 -1.15362716e+00 1.43043625e+00
1.00645399e+00 1.08881101e-01 3.83502096e-01 1.33548510e+00
1.11719146e-01 -3.70981216e-01 -5.09942830e-01 1.95346698e-01
8.33598733e-01 -3.83613408e-02 8.66972566e-01 1.03185034e+00
1.59652844e-01 6.20198369e-01 7.57473111e-01 2.20131010e-01
8.88395905e-01 -3.57490480e-01 -6.58070028e-01 1.63387179e-01
8.10731292e-01 9.59661424e-01 -1.44298971e+00 4.11619574e-01
-1.39656216e-01 5.15321791e-01 -1.05187051e-01 2.22270370e-01
-9.55865502e-01 -5.10864198e-01 7.17712581e-01 -8.21523592e-02
2.15144739e-01 -6.36494100e-01 -5.02841659e-02 7.70617276e-02
-1.97962299e-01 -8.88334155e-01 7.42178142e-01 -6.73597515e-01
-8.56774509e-01 6.74415290e-01 7.51040757e-01 -8.85851860e-01
-3.86831552e-01 -7.54731819e-02 -5.85525870e-01 4.10229266e-01
-1.92828131e+00 -1.04347956e+00 -4.86885086e-02 7.47025490e-01
5.65561771e-01 -1.39476135e-01 9.86939132e-01 2.06270993e-01
-2.91220844e-01 4.00879309e-02 -1.86868429e-01 -7.06029892e-01
4.90076482e-01 -1.03372717e+00 5.52110314e-01 8.27432632e-01
-6.50717139e-01 3.25920850e-01 1.02340126e+00 -1.00825393e+00
-1.75759089e+00 -7.91292250e-01 7.50322163e-01 -1.92704722e-01
4.84780133e-01 1.37135506e-01 -2.76723742e-01 1.00828004e+00
4.94464219e-01 -4.50994790e-01 3.99023026e-01 -1.40455112e-01
2.75634766e-01 8.72942582e-02 -1.38628280e+00 5.67040622e-01
1.45497513e+00 3.52558136e-01 -2.81364352e-01 7.06328630e-01
8.51513386e-01 -7.64112294e-01 -9.16663885e-01 4.24792886e-01
3.51414084e-01 -9.24438775e-01 8.22147787e-01 -7.37695992e-01
-1.45418450e-01 -7.64494181e-01 -3.02729547e-01 -1.43108547e+00
-7.22139299e-01 -1.22679496e+00 5.28187633e-01 9.88880873e-01
6.37921095e-01 -8.17763269e-01 7.97320485e-01 8.60486209e-01
-9.28968370e-01 -5.46872556e-01 -1.44023919e+00 -1.15414119e+00
-2.32421637e-01 -2.54371047e-01 1.24132121e+00 9.87770617e-01
6.85331225e-01 -3.61692816e-01 -2.43283838e-01 6.29127204e-01
6.31786823e-01 1.62658885e-01 6.85248613e-01 -1.09844398e+00
9.52862129e-02 -2.06705451e-01 -1.51307359e-01 -3.21220458e-01
2.80332211e-02 -4.86007184e-01 4.92447823e-01 -1.97463691e+00
-3.40259701e-01 -1.14130974e+00 2.40546260e-02 7.97042489e-01
7.99780726e-01 -3.45116228e-01 3.16534251e-01 2.13519521e-02
-9.52213168e-01 -3.32989357e-02 1.23981559e+00 -6.86534047e-02
-3.63762826e-01 -6.72527775e-02 -5.34695983e-01 2.28476629e-01
8.57589185e-01 -8.34555268e-01 -5.74669302e-01 -4.63278711e-01
7.21849084e-01 9.02855873e-01 -1.61430910e-01 -5.14475524e-01
8.63249421e-01 -1.35386658e+00 -5.27031898e-01 -9.71805096e-01
2.40072057e-01 -1.18440127e+00 9.05582249e-01 8.60312104e-01
-2.77126789e-01 9.35441434e-01 4.82280850e-01 6.74471378e-01
2.84232516e-02 -2.17341334e-01 1.00527108e-01 -2.61723071e-01
-1.30700529e+00 -7.60712177e-02 -8.25387955e-01 -2.60892302e-01
1.91186619e+00 -3.52361679e-01 -1.04642332e+00 -1.82348967e-01
-4.84998763e-01 1.08669484e+00 2.86228836e-01 2.49998897e-01
8.23574483e-01 -1.16169310e+00 -8.40405345e-01 -2.71585673e-01
1.18237272e-01 1.69200301e-01 2.64319688e-01 1.12279308e+00
-6.54677510e-01 5.90287447e-01 -8.05368483e-01 -1.93919674e-01
-1.21979809e+00 7.03013062e-01 4.21059988e-02 -5.50469458e-01
-6.99212015e-01 5.93172967e-01 -6.52660668e-01 -5.06081700e-01
1.34021640e-01 -1.93268910e-01 1.01935312e-01 -3.69238444e-02
6.59901857e-01 8.17623436e-01 1.06817931e-01 -2.64950246e-01
-1.19018865e+00 2.87076324e-01 -1.21457420e-01 -4.51178640e-01
1.79173541e+00 -4.67197359e-01 -2.95758069e-01 -1.30315840e-01
4.73307222e-02 -1.40959814e-01 -9.43963051e-01 1.77231893e-01
5.90000153e-01 -5.28654218e-01 -3.06532055e-01 -1.13803327e+00
-6.58721149e-01 -4.49153841e-01 -5.55487722e-02 5.70187747e-01
9.68670905e-01 7.67713487e-02 6.86813891e-01 5.82820594e-01
1.45198214e+00 -1.31682217e+00 9.76176858e-02 6.50260985e-01
1.08545685e+00 -5.66132784e-01 1.53135151e-01 -8.08472335e-01
-9.93008673e-01 1.18137026e+00 5.33229470e-01 -1.07446602e-02
1.49999619e-01 9.86554146e-01 -8.06656256e-02 -5.87621748e-01
-1.03563488e+00 -2.14869142e-01 -8.46881688e-01 1.13161099e+00
-6.84812486e-01 1.52521119e-01 -3.37595701e-01 3.61646384e-01
1.22135347e-02 -1.34668529e-01 1.19615591e+00 1.84942627e+00
-5.72579563e-01 -1.31855094e+00 -6.29465282e-01 -3.74890603e-02
3.49311829e-01 6.61225140e-01 -4.06071901e-01 1.18132186e+00
-1.23115376e-01 1.50936592e+00 4.01889682e-02 -2.89036274e-01
9.94478345e-01 -3.85467142e-01 7.01940894e-01 -3.77110481e-01
-7.48721123e-01 -6.41846135e-02 1.21563578e+00 -1.08143771e+00
-8.34550679e-01 -1.00267649e+00 -1.87102675e+00 -7.25931406e-01
-1.66473716e-01 4.10521805e-01 4.47277278e-01 8.71914208e-01
1.04227804e-01 5.85481107e-01 5.23550987e-01 -9.23455000e-01
-1.10728636e-01 -7.99374804e-02 -7.17681229e-01 -3.11313540e-01
-1.20834768e-01 -7.57368982e-01 1.71177372e-01 -5.98147094e-01] | [4.9651994705200195, 1.7822085618972778] |
76c71bd6-b015-4861-a5a1-f1d4678f5882 | dependable-intrusion-detection-system-for-iot | 2204.04837 | null | https://arxiv.org/abs/2204.04837v1 | https://arxiv.org/pdf/2204.04837v1.pdf | Dependable Intrusion Detection System for IoT: A Deep Transfer Learning-based Approach | Security concerns for IoT applications have been alarming because of their widespread use in different enterprise systems. The potential threats to these applications are constantly emerging and changing, and therefore, sophisticated and dependable defense solutions are necessary against such threats. With the rapid development of IoT networks and evolving threat types, the traditional machine learning-based IDS must update to cope with the security requirements of the current sustainable IoT environment. In recent years, deep learning, and deep transfer learning have progressed and experienced great success in different fields and have emerged as a potential solution for dependable network intrusion detection. However, new and emerging challenges have arisen related to the accuracy, efficiency, scalability, and dependability of the traditional IDS in a heterogeneous IoT setup. This manuscript proposes a deep transfer learning-based dependable IDS model that outperforms several existing approaches. The unique contributions include effective attribute selection, which is best suited to identify normal and attack scenarios for a small amount of labeled data, designing a dependable deep transfer learning-based ResNet model, and evaluating considering real-world data. To this end, a comprehensive experimental performance evaluation has been conducted. Extensive analysis and performance evaluation show that the proposed model is robust, more efficient, and has demonstrated better performance, ensuring dependability. | ['Rafiqul Islam', 'Kawsar Ahmed', 'Ziaur Rahman', 'Adnan Anwar', 'Sk. Tanzir Mehedi'] | 2022-04-11 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [-2.48997256e-01 -7.43616998e-01 1.21555757e-02 -3.54428768e-01
-4.05005328e-02 -4.69082534e-01 3.86489630e-01 1.19297951e-01
-4.94945019e-01 6.23874128e-01 -3.53924960e-01 -4.14750785e-01
-6.53339565e-01 -1.09160686e+00 -7.86458403e-02 -7.23172247e-01
-3.47039074e-01 6.64823949e-01 3.27355832e-01 -3.08539242e-01
5.80188148e-02 8.42611194e-01 -1.41569245e+00 -1.76754847e-01
6.08652771e-01 1.67820382e+00 -5.51393569e-01 3.17031920e-01
5.41552603e-02 5.48885047e-01 -8.25049281e-01 -2.65161335e-01
3.41349036e-01 6.72804862e-02 -7.01474011e-01 -5.74697077e-01
-2.40722477e-01 -4.82409120e-01 -5.53764820e-01 8.22736561e-01
7.63748348e-01 6.20695651e-02 3.40989083e-01 -1.74418378e+00
-4.06625420e-01 7.07849443e-01 -2.46437445e-01 4.36893284e-01
-2.61001829e-02 3.16281468e-02 4.78995711e-01 -2.43227199e-01
-1.34287208e-01 1.03939271e+00 6.53658092e-01 4.05432254e-01
-6.81566656e-01 -1.36559105e+00 1.26662865e-01 7.30403662e-01
-1.20670021e+00 -6.12189025e-02 8.04383516e-01 -3.70060913e-02
8.58248591e-01 9.08363461e-02 3.56515020e-01 1.56290781e+00
4.88117397e-01 1.70070320e-01 8.46890152e-01 -6.72059134e-02
4.14273828e-01 1.41621664e-01 2.52710611e-01 1.04893051e-01
5.69938242e-01 2.75795043e-01 2.00746045e-01 -1.73722968e-01
2.58729756e-01 5.23984969e-01 3.31998885e-01 -1.23433471e-01
-9.11928415e-01 7.61881471e-01 5.88479698e-01 5.29895723e-01
-6.34541452e-01 -8.92743915e-02 9.11383212e-01 4.32325184e-01
2.76620597e-01 1.70865372e-01 -8.55726182e-01 -1.90075010e-01
-2.50390410e-01 -2.72630483e-01 9.39515591e-01 6.16687596e-01
2.03498736e-01 4.29117531e-01 2.55145758e-01 2.81565070e-01
1.68466449e-01 7.83094645e-01 3.19549024e-01 -4.77584779e-01
2.57192105e-01 8.72718215e-01 -1.09978475e-01 -1.13441420e+00
-8.47679853e-01 -4.87441838e-01 -1.35891581e+00 1.51033640e-01
-1.78800941e-01 -2.37287134e-01 -7.56074607e-01 1.60719001e+00
5.06270885e-01 3.83808047e-01 1.03629299e-01 2.96996355e-01
4.45423216e-01 5.90692639e-01 7.07768500e-01 -1.22165725e-01
1.20441175e+00 -4.64817546e-02 -6.70802295e-01 1.38194486e-01
1.34353787e-01 -3.93183053e-01 6.70935035e-01 4.74213034e-01
-2.86779702e-01 -6.73194528e-01 -1.24407721e+00 5.73650181e-01
-8.39299679e-01 -7.05061674e-01 6.51278794e-01 1.15670836e+00
-6.85440242e-01 3.26818973e-01 -7.06350386e-01 -6.04899049e-01
4.53124076e-01 7.44651675e-01 -5.08822734e-03 5.61558865e-02
-1.72607291e+00 8.78585339e-01 7.96398699e-01 8.10874477e-02
-8.27539623e-01 -3.64169657e-01 -3.82536709e-01 2.08297089e-01
2.19260588e-01 -4.64337587e-01 8.28941226e-01 -3.81233156e-01
-1.37052119e+00 2.16727182e-01 9.05912220e-01 -5.58263421e-01
2.10299596e-01 -1.58126444e-01 -1.28366256e+00 -2.41934415e-02
-3.24789792e-01 -1.39951408e-01 8.43744338e-01 -1.04957998e+00
-8.01391602e-01 -4.77873921e-01 9.17218998e-02 -2.73329645e-01
-1.10076869e+00 4.42000866e-01 4.02567416e-01 -2.59794801e-01
-1.90142468e-01 -9.25221622e-01 -1.49229810e-01 -2.90050358e-01
-1.57069609e-01 -3.10201615e-01 1.58272696e+00 -2.87831098e-01
1.04074585e+00 -2.02822065e+00 -2.54988641e-01 4.37169909e-01
2.60702342e-01 8.67145121e-01 -3.51504348e-02 4.86807644e-01
-1.19730286e-01 1.94481790e-01 -6.55790120e-02 2.77300000e-01
-1.66086964e-02 2.15826243e-01 -3.07038248e-01 2.69746304e-01
-1.87421650e-01 4.65598881e-01 -6.55295491e-01 -3.59476686e-01
7.56512702e-01 8.07259500e-01 7.83170462e-02 3.65193963e-01
7.93567970e-02 7.39427924e-01 -9.41182733e-01 7.56415248e-01
8.34009290e-01 -2.25035831e-01 1.24560200e-01 -3.58777165e-01
7.71571137e-03 -8.53748471e-02 -1.11770988e+00 8.04935634e-01
-7.65032411e-01 1.48606494e-01 -3.51977795e-02 -1.32062280e+00
1.22335124e+00 5.40755630e-01 9.95444298e-01 -1.11627245e+00
7.79841304e-01 2.66481757e-01 1.59469664e-01 -4.96557474e-01
-1.47940814e-01 -4.39233817e-02 -4.00997311e-01 6.15215540e-01
-2.01451972e-01 2.35285565e-01 -2.40968272e-01 -4.14708816e-02
1.30740511e+00 -3.46387535e-01 3.03943455e-01 7.55522549e-02
9.22518909e-01 -4.91013378e-01 6.68176353e-01 5.03114820e-01
-7.24699557e-01 -3.31522018e-01 -9.67177525e-02 -1.02002525e+00
-5.49606442e-01 -9.60183263e-01 -3.80201846e-01 1.06037796e+00
3.49296570e-01 1.83486208e-01 -4.62408513e-01 -9.49070334e-01
-1.70821592e-01 7.34930515e-01 -3.97052765e-01 -6.53926611e-01
-8.58848274e-01 -9.97132540e-01 7.99196124e-01 5.02292216e-01
1.23576880e+00 -1.42353046e+00 -7.56119013e-01 4.37960327e-01
-1.63465947e-01 -1.45997250e+00 3.15114081e-01 3.58039200e-01
-5.76986611e-01 -1.22680426e+00 1.48372978e-01 -7.15725362e-01
1.18768252e-01 2.21282810e-01 8.87743294e-01 7.22103566e-02
-3.60969573e-01 4.94459867e-01 -6.06567025e-01 -7.40347624e-01
-5.47534883e-01 3.43439251e-01 5.28267026e-01 9.27023813e-02
6.41388893e-01 -8.18098783e-01 -6.62221134e-01 5.52968681e-01
-1.01719654e+00 -9.97476339e-01 7.58719325e-01 5.25439501e-01
1.55399188e-01 7.36648858e-01 1.18174183e+00 -4.89712179e-01
5.26663899e-01 -8.94538403e-01 -5.17333090e-01 3.34074646e-01
-1.05543339e+00 -3.08588356e-01 1.14809406e+00 -5.34810364e-01
-1.07997620e+00 -5.22166073e-01 -4.09331381e-01 -6.68981001e-02
-4.57574010e-01 1.24632709e-01 -4.52501446e-01 -3.71512830e-01
6.69734538e-01 -2.98260953e-02 -1.85328424e-01 -2.36562297e-01
1.16486186e-02 1.20653892e+00 1.37018621e-01 -4.86917347e-01
1.29419625e+00 4.69712883e-01 2.08288625e-01 -7.02132106e-01
-6.92911804e-01 -1.42825812e-01 -6.04831338e-01 -3.03829789e-01
7.06483483e-01 -3.92010778e-01 -1.16451001e+00 8.26954067e-01
-8.66112769e-01 1.97853297e-01 -1.96449608e-01 4.54549968e-01
-1.12103008e-01 3.17149520e-01 -6.97469234e-01 -6.87958598e-01
-1.08202815e+00 -1.01176322e+00 2.72042960e-01 3.46953958e-01
1.81543529e-02 -1.08822143e+00 9.51377004e-02 2.09745556e-01
1.03559768e+00 3.61483723e-01 9.98470843e-01 -1.39194357e+00
-3.31716448e-01 -7.24468470e-01 -3.59498560e-01 5.56322753e-01
4.19520557e-01 -1.31503999e-01 -8.09878588e-01 -5.89106858e-01
3.69570106e-01 -3.67588341e-01 2.11605594e-01 -7.68019781e-02
1.41923213e+00 -3.50002706e-01 -3.73714745e-01 7.38753378e-01
1.27521360e+00 8.46940339e-01 5.17947853e-01 8.22301686e-01
5.79609275e-01 4.46704388e-01 5.07850468e-01 6.28057241e-01
2.66906857e-01 2.39307016e-01 1.06391180e+00 2.70736981e-02
3.64928126e-01 1.48240656e-01 -6.26618564e-02 8.58514547e-01
-5.41253341e-03 -5.52123487e-01 -8.47342134e-01 -4.48537245e-02
-1.55494404e+00 -1.01275015e+00 3.10420334e-01 2.06393194e+00
1.78050935e-01 4.20705110e-01 1.59216315e-01 5.78238547e-01
7.77184665e-01 1.50093392e-01 -1.03487623e+00 -4.24316287e-01
9.88782421e-02 4.66335386e-01 3.44292670e-01 -3.35412353e-01
-1.52331257e+00 6.82012439e-01 6.02959633e+00 4.70358044e-01
-1.42756534e+00 1.69195503e-01 4.79421645e-01 4.80461419e-01
3.17794740e-01 -4.54566777e-01 -5.39266765e-01 6.22077107e-01
1.38627303e+00 -1.16798349e-01 8.79668891e-02 1.03142703e+00
-5.12327924e-02 7.62641370e-01 -6.04343712e-01 8.76828134e-01
-2.28333578e-01 -7.43048012e-01 8.94799083e-02 6.59225434e-02
2.65350819e-01 1.06876895e-01 3.98570597e-01 6.17037416e-01
4.51583743e-01 -7.90367603e-01 2.77828071e-02 1.51604526e-02
6.48382545e-01 -1.14762771e+00 1.27037024e+00 1.51471972e-01
-1.37278545e+00 -7.57272303e-01 -3.22759032e-01 -1.26600312e-02
-1.15646161e-01 2.98195601e-01 -5.15472054e-01 7.85391808e-01
9.94702995e-01 5.92435360e-01 -3.38316172e-01 9.34445798e-01
4.20209998e-03 7.19165266e-01 -2.84753263e-01 -2.51559258e-01
2.84233261e-02 2.91471556e-02 4.49972153e-01 8.10312450e-01
2.89825678e-01 5.37213683e-02 4.68101025e-01 2.57162184e-01
-1.38555269e-03 2.48298347e-02 -4.84035283e-01 3.88416499e-02
8.85571063e-01 1.50324261e+00 -9.18186128e-01 1.80329066e-02
-2.72317231e-01 6.34422183e-01 -1.19619854e-01 9.23246369e-02
-1.13620162e+00 -5.65393209e-01 9.04984713e-01 -1.15728654e-01
3.68337557e-02 -6.17260933e-02 -2.59834528e-01 -7.76139915e-01
-3.17321390e-01 -9.65172172e-01 9.33957517e-01 1.03293486e-01
-1.66864598e+00 1.09543133e+00 -6.00982979e-02 -1.53713727e+00
1.01948634e-01 -6.15708292e-01 -7.07701325e-01 2.23055750e-01
-1.52250135e+00 -1.41871524e+00 -4.26042885e-01 9.79022563e-01
4.10946071e-01 -4.34565097e-01 1.22476220e+00 6.38402164e-01
-9.59911168e-01 7.99265742e-01 5.61899506e-02 1.45920828e-01
5.12549996e-01 -5.57358146e-01 3.80020618e-01 8.61840844e-01
-3.16681027e-01 4.01383102e-01 4.59722459e-01 -4.53172714e-01
-1.38586760e+00 -1.21373045e+00 1.57467961e-01 -2.80991137e-01
7.75718331e-01 -2.04266250e-01 -9.07625318e-01 7.30075777e-01
-5.39235696e-02 1.04249619e-01 7.54805803e-01 2.70253103e-02
-6.54367864e-01 -8.44579220e-01 -1.63557804e+00 2.22511128e-01
7.04006732e-01 -1.67831510e-01 -3.58430475e-01 1.95240930e-01
7.84783721e-01 1.98394954e-01 -9.74212527e-01 7.60762751e-01
4.56577063e-01 -8.40606689e-01 1.15913689e+00 -5.76354563e-01
-6.84762537e-01 -1.08518340e-01 -6.72670230e-02 -1.04057467e+00
-6.02995336e-01 -4.44490254e-01 -1.68646321e-01 1.37941015e+00
-9.33207572e-02 -1.25580585e+00 5.81840992e-01 5.55366635e-01
1.38216585e-01 -3.47483009e-01 -1.13375640e+00 -9.06364799e-01
6.63471036e-03 -3.42093825e-01 1.09684968e+00 1.08447516e+00
-2.83707798e-01 3.12151641e-01 -4.52389657e-01 3.36074650e-01
8.65452111e-01 -1.89408764e-01 3.58512044e-01 -1.94793773e+00
3.55566084e-01 -2.28167236e-01 -8.62794757e-01 -1.88676983e-01
9.48058069e-02 -4.49028373e-01 -4.32707638e-01 -1.24857056e+00
-1.95493519e-01 -9.11286294e-01 -1.20727289e+00 7.06100225e-01
1.08252361e-01 4.09951024e-02 6.47719891e-04 1.66316956e-01
-6.87434375e-01 5.96719503e-01 5.59088588e-01 -3.53780597e-01
-5.70679754e-02 4.47348863e-01 -5.48473716e-01 7.44170725e-01
1.42365468e+00 -5.09010136e-01 -5.46228528e-01 -1.36016175e-01
4.45805043e-02 -8.67407918e-02 9.54902917e-02 -1.55615044e+00
2.54207432e-01 -5.45352064e-02 4.25787359e-01 -3.82707894e-01
1.66635096e-01 -1.53382814e+00 8.69862139e-02 9.73404825e-01
1.31315380e-01 2.32336953e-01 1.50206849e-01 4.64707971e-01
2.69290712e-02 3.05565685e-01 9.44143236e-01 2.47911364e-01
-9.58719850e-01 9.62174058e-01 -2.24757046e-01 -8.76182392e-02
1.49354315e+00 -2.13618368e-01 -3.39462638e-01 -1.78521574e-01
-2.84681410e-01 5.18819578e-02 4.78870757e-02 8.13113868e-01
8.09512436e-01 -1.02419269e+00 -5.24421155e-01 3.81818920e-01
2.51630656e-02 -4.17115539e-01 3.70538324e-01 2.41028428e-01
-3.98281276e-01 2.40714639e-01 -8.10043991e-01 -1.82684153e-01
-8.94327998e-01 1.06768119e+00 2.47163504e-01 -3.20693731e-01
-6.15482032e-01 4.14089501e-01 -2.95071691e-01 -5.78040779e-01
5.07142663e-01 2.95657337e-01 -4.71919328e-01 -1.11162566e-01
6.68502688e-01 8.49851906e-01 2.82835335e-01 -5.02529085e-01
-6.81108952e-01 4.37662780e-01 -1.21568896e-01 5.64025700e-01
1.33698988e+00 -8.72345343e-02 -1.90785870e-01 1.82340041e-01
9.23281193e-01 -4.69195276e-01 -6.10053778e-01 -1.29417464e-01
4.00032178e-02 -1.22902863e-01 -1.04126573e-01 -8.87194455e-01
-1.37519908e+00 7.92433321e-01 1.07962286e+00 8.57264221e-01
1.36597908e+00 -4.68488544e-01 1.41358948e+00 5.40264487e-01
8.51064265e-01 -8.18194926e-01 2.97136635e-01 6.10791028e-01
1.84024751e-01 -1.26312160e+00 -1.57351807e-01 -2.04478338e-01
-1.36768594e-01 1.18126321e+00 9.03742909e-01 1.38132453e-01
1.22579980e+00 4.41469133e-01 1.95943505e-01 -9.95054096e-02
-2.60271817e-01 1.98880598e-01 -3.76627982e-01 1.12188137e+00
-1.43645629e-01 2.52991803e-02 8.00605640e-02 3.30826104e-01
-1.17718270e-02 -3.11180353e-01 7.71364272e-02 8.70957851e-01
-5.03335297e-01 -1.31137109e+00 -3.33519608e-01 2.95172453e-01
-5.14487505e-01 3.82661790e-01 -8.64093751e-02 8.51026118e-01
5.94333969e-02 1.33825779e+00 -3.30117010e-02 -1.04012012e+00
3.41183484e-01 -1.25888497e-01 -1.22433268e-01 -9.90260541e-02
-6.41602755e-01 -7.36828983e-01 -2.70432502e-01 -3.77642214e-01
-1.69542015e-01 -7.22980872e-02 -1.22486305e+00 -6.68811262e-01
-1.30617812e-01 1.44162759e-01 9.06682312e-01 9.06720519e-01
5.77544510e-01 6.48421049e-01 1.30807507e+00 -4.47833240e-01
-1.02833891e+00 -8.11780572e-01 -4.64653254e-01 4.45751995e-01
-3.81471775e-02 -8.47674251e-01 -5.75413406e-01 -7.23221779e-01] | [5.277160167694092, 7.201507568359375] |
cc4836ee-e201-4b8f-89e8-dba34f83bff3 | how-does-information-bottleneck-help-deep | 2305.18887 | null | https://arxiv.org/abs/2305.18887v1 | https://arxiv.org/pdf/2305.18887v1.pdf | How Does Information Bottleneck Help Deep Learning? | Numerous deep learning algorithms have been inspired by and understood via the notion of information bottleneck, where unnecessary information is (often implicitly) minimized while task-relevant information is maximized. However, a rigorous argument for justifying why it is desirable to control information bottlenecks has been elusive. In this paper, we provide the first rigorous learning theory for justifying the benefit of information bottleneck in deep learning by mathematically relating information bottleneck to generalization errors. Our theory proves that controlling information bottleneck is one way to control generalization errors in deep learning, although it is not the only or necessary way. We investigate the merit of our new mathematical findings with experiments across a range of architectures and learning settings. In many cases, generalization errors are shown to correlate with the degree of information bottleneck: i.e., the amount of the unnecessary information at hidden layers. This paper provides a theoretical foundation for current and future methods through the lens of information bottleneck. Our new generalization bounds scale with the degree of information bottleneck, unlike the previous bounds that scale with the number of parameters, VC dimension, Rademacher complexity, stability or robustness. Our code is publicly available at: https://github.com/xu-ji/information-bottleneck | ['Jiaoyang Huang', 'Xu Ji', 'Zhun Deng', 'Kenji Kawaguchi'] | 2023-05-30 | null | null | null | null | ['generalization-bounds'] | ['methodology'] | [-2.28703991e-02 1.43318713e-01 -2.65989244e-01 -1.05847314e-01
-3.21737081e-01 -6.24851942e-01 3.85939568e-01 3.88988614e-01
-6.88303530e-01 6.69328034e-01 1.21999212e-01 -4.44866717e-01
-5.37661731e-01 -5.72370589e-01 -7.90410936e-01 -7.07236290e-01
-9.66767743e-02 6.76226392e-02 2.27202382e-02 -1.23280458e-01
5.63844740e-01 4.15174812e-01 -1.43116927e+00 -9.99177620e-02
7.55561769e-01 1.02849686e+00 1.61127031e-01 7.10216880e-01
-6.00825660e-02 6.96931303e-01 -3.92221808e-01 -4.28679258e-01
6.46191955e-01 -5.47016442e-01 -9.18044329e-01 -3.88980180e-01
4.41063374e-01 -1.41474739e-01 -4.19021904e-01 1.15914559e+00
5.46031773e-01 1.57161094e-02 5.74415147e-01 -1.31328452e+00
-6.17785931e-01 8.12890291e-01 -3.27605188e-01 7.15290964e-01
-3.72296840e-01 1.07849009e-01 1.38058388e+00 -6.53106511e-01
4.45743829e-01 1.22037137e+00 5.20865560e-01 7.29061723e-01
-1.15140319e+00 -6.36381269e-01 3.35995942e-01 2.13787302e-01
-1.17550421e+00 -4.75483179e-01 6.21912241e-01 -6.86314940e-01
6.33449256e-01 6.21717498e-02 5.17715454e-01 7.31507719e-01
3.96908134e-01 9.83517408e-01 7.05035627e-01 -5.35996616e-01
3.53244662e-01 4.90235358e-01 6.58799946e-01 8.19275379e-01
7.38039851e-01 2.49793768e-01 -6.34530306e-01 1.95015982e-01
8.07635069e-01 -1.17973439e-01 -3.22251618e-01 -5.06301284e-01
-7.66291082e-01 1.09506869e+00 6.37097716e-01 2.01577067e-01
5.60348295e-02 2.97373295e-01 4.36856836e-01 8.08908582e-01
3.34620833e-01 9.22840536e-01 -7.18720913e-01 -4.68949042e-02
-5.85202336e-01 1.15194112e-01 8.43915761e-01 9.03927565e-01
7.82070994e-01 -3.06993481e-02 -1.05937440e-02 5.93726873e-01
2.31680244e-01 7.12124631e-02 4.29314703e-01 -1.12045169e+00
4.50847894e-01 6.92429304e-01 -1.08454019e-01 -6.59401894e-01
-7.42833555e-01 -9.46176231e-01 -8.81670892e-01 1.50136366e-01
7.90317953e-01 -2.45584756e-01 -4.37169105e-01 2.25704288e+00
6.26976322e-03 -6.53022468e-01 -4.77249399e-02 9.29007649e-01
2.63570547e-01 2.34028324e-01 -2.20831051e-01 -1.80874512e-01
9.70679402e-01 -6.16750956e-01 -6.82758927e-01 -2.48895198e-01
9.67908800e-01 -3.02314639e-01 1.37874746e+00 2.06529111e-01
-1.14270246e+00 -2.66055882e-01 -1.21257854e+00 -1.34274498e-01
-4.04713303e-01 -2.60506094e-01 7.86436021e-01 8.47268105e-01
-1.15901911e+00 8.16903830e-01 -7.10227132e-01 -1.57663852e-01
6.71269476e-01 4.52433616e-01 -4.19350043e-02 9.59775969e-02
-1.19836521e+00 9.22770083e-01 4.63546664e-01 -1.13602698e-01
-7.04283774e-01 -1.02993262e+00 -5.71240783e-01 3.18663985e-01
2.77053833e-01 -8.59458625e-01 1.17905879e+00 -1.07086480e+00
-1.17355573e+00 4.96131420e-01 1.22052014e-01 -5.87918341e-01
7.64165580e-01 -3.60807389e-01 4.54307854e-01 -1.35776743e-01
-2.88557768e-01 6.30358696e-01 5.03479242e-01 -9.25338447e-01
-4.09323066e-01 -6.97437763e-01 2.73478270e-01 2.35849068e-01
-6.61379278e-01 -4.58737761e-01 -1.00821905e-01 -2.83921301e-01
8.32943618e-03 -8.18204701e-01 -2.34074607e-01 1.60821304e-01
-3.49307597e-01 -2.23573923e-01 3.95276338e-01 -1.29179940e-01
1.27367270e+00 -2.23250818e+00 1.59292892e-01 2.84814797e-02
7.89458930e-01 2.27561183e-02 -1.94017619e-01 3.63979191e-01
3.23806480e-02 6.96192801e-01 -1.63250864e-01 -6.59308881e-02
2.52310097e-01 -1.53063670e-01 -1.05579630e-01 7.46208429e-01
-6.14341674e-03 7.34492004e-01 -6.81600809e-01 -1.95704535e-01
-2.03429051e-02 4.69659835e-01 -9.60249722e-01 -1.30388200e-01
2.56974678e-02 1.86138064e-01 -4.11484212e-01 8.21938664e-02
5.45451462e-01 -5.60164690e-01 -4.71829996e-02 1.27701700e-01
-2.25974485e-01 4.49175775e-01 -1.11775100e+00 1.35112381e+00
-3.85585874e-01 1.17809105e+00 7.39287538e-03 -1.05231833e+00
4.84867692e-01 -5.93253188e-02 2.75570571e-01 -7.35074699e-01
2.89365351e-01 9.12816375e-02 3.55932891e-01 -1.69557154e-01
1.74842745e-01 5.94285876e-02 1.93807423e-01 6.11635685e-01
1.29170001e-01 2.56315202e-01 4.64609593e-01 3.90476197e-01
1.13862324e+00 -4.62940246e-01 2.68665552e-01 -7.26448357e-01
1.89448267e-01 -2.15627208e-01 4.17376697e-01 1.19158065e+00
-4.11351919e-01 2.79308766e-01 9.80763614e-01 -3.42306614e-01
-1.38745534e+00 -1.04513669e+00 -4.17973340e-01 1.30677533e+00
7.52239525e-02 -3.27772945e-01 -9.13448811e-01 -3.62160236e-01
1.12323530e-01 5.59569955e-01 -8.98059726e-01 -4.27613497e-01
-3.35366763e-02 -7.04619765e-01 5.00838161e-01 3.19208980e-01
6.90889239e-01 -5.69603324e-01 -7.22998381e-01 -1.78627566e-01
1.69032738e-01 -7.78019488e-01 -5.13528287e-01 3.42965662e-01
-1.19435990e+00 -1.14821684e+00 -5.46472967e-01 -3.54565322e-01
6.08977139e-01 2.11821020e-01 1.00939059e+00 3.94356251e-01
-3.75587225e-01 4.07423288e-01 -1.48215219e-01 -6.16663098e-01
-3.87767613e-01 5.08659661e-01 9.36029404e-02 -4.06769246e-01
1.74946070e-01 -4.35916305e-01 -9.04350221e-01 2.11294308e-01
-1.01466608e+00 1.71128750e-01 7.70264745e-01 7.37096786e-01
1.51129529e-01 -1.38312668e-01 5.49091101e-01 -9.28639948e-01
8.42514634e-01 -3.81242514e-01 -8.04723859e-01 7.88742900e-02
-9.58272994e-01 6.55903339e-01 5.79467535e-01 -4.31612730e-01
-6.88330352e-01 -1.54466555e-01 1.00643888e-01 -1.77449271e-01
1.58991367e-01 4.13917422e-01 1.66796148e-02 -2.83669960e-02
7.44583488e-01 1.27315402e-01 -8.54236409e-02 -3.74653548e-01
4.16018218e-01 2.30260760e-01 -1.92073762e-01 -3.20178241e-01
4.15286720e-01 2.64600575e-01 1.56810239e-01 -9.09251869e-01
-1.09623778e+00 -2.44158611e-01 -8.58384669e-01 -1.63735986e-01
5.42455018e-01 -6.39435410e-01 -1.05655682e+00 2.16131747e-01
-9.57833827e-01 -4.61804658e-01 -3.65278602e-01 5.69518685e-01
-5.80241144e-01 2.43385751e-02 -6.89881265e-01 -9.15632010e-01
-3.17634821e-01 -1.05422056e+00 2.62940139e-01 1.47010341e-01
-2.83844881e-02 -1.24352407e+00 -8.90585259e-02 1.20014451e-01
5.87823451e-01 -1.51258036e-01 1.07196331e+00 -7.13916898e-01
-7.61594772e-01 -8.59947354e-02 -3.75956416e-01 3.78773272e-01
-1.55870423e-01 -2.35614792e-01 -9.44300473e-01 -3.02389741e-01
4.61302772e-02 -2.21111298e-01 1.25371766e+00 6.66171312e-01
1.12432122e+00 -6.92527354e-01 2.17049289e-02 7.99157917e-01
1.59650135e+00 -6.01981506e-02 3.87591332e-01 3.77081245e-01
4.92307097e-01 5.96618533e-01 -1.63205862e-01 5.49629390e-01
1.13843672e-01 3.03005755e-01 2.81264573e-01 1.65061548e-01
-1.20725200e-01 -1.80189401e-01 2.45900616e-01 7.16940045e-01
1.97411135e-01 -1.94414154e-01 -1.03053737e+00 3.05811822e-01
-1.68292582e+00 -8.25787544e-01 1.44195497e-01 2.40641975e+00
9.24394011e-01 4.41830575e-01 1.04361899e-01 1.84253410e-01
5.53163767e-01 -1.70308948e-01 -1.05935800e+00 -3.67881685e-01
-5.53575382e-02 -4.65039730e-01 8.18078518e-01 8.92280161e-01
-7.91082740e-01 8.37043941e-01 6.65535641e+00 6.10206842e-01
-8.21700573e-01 1.68389186e-01 7.51947641e-01 -4.56442684e-01
-2.98793077e-01 -2.54806340e-01 -9.28025663e-01 3.23581576e-01
8.95637214e-01 -4.70360249e-01 5.34342945e-01 8.89924943e-01
5.38945608e-02 -4.68941517e-02 -1.46750867e+00 8.41277063e-01
-3.48642141e-01 -1.49222231e+00 -2.92502847e-02 2.93464303e-01
5.72318733e-01 2.54817873e-01 3.73593658e-01 1.03867888e-01
4.18089449e-01 -7.08265901e-01 6.16591394e-01 2.44195417e-01
5.11556029e-01 -7.69252360e-01 5.45352697e-01 4.58647668e-01
-7.75650918e-01 -4.01245594e-01 -4.75016207e-01 -2.53440529e-01
-5.73341966e-01 8.18741560e-01 -5.62087953e-01 -8.23643804e-02
4.97172028e-01 4.47828799e-01 -7.91162968e-01 1.11538994e+00
-3.20696272e-02 4.98589605e-01 -3.06714803e-01 -3.43299359e-01
5.21575958e-02 -6.00082101e-03 4.53146815e-01 1.27877116e+00
-9.08491686e-02 -1.48531169e-01 -3.15390050e-01 9.68414307e-01
-2.95673460e-01 9.07399729e-02 -7.53977597e-01 -1.82764083e-01
6.52891934e-01 7.94577718e-01 -7.19026327e-01 -2.15682257e-02
-2.62922555e-01 5.70082784e-01 6.41316116e-01 4.11440253e-01
-4.01739776e-01 -4.43058968e-01 8.01192939e-01 6.87326416e-02
-1.12492941e-01 -4.49594766e-01 -7.96985149e-01 -1.10242283e+00
-8.71252418e-02 -3.99492025e-01 4.00762767e-01 -2.23931652e-02
-1.12055004e+00 1.71866491e-01 -1.34724736e-01 -6.57504439e-01
2.58880872e-02 -8.47382963e-01 -2.14956716e-01 6.07753992e-01
-1.38109434e+00 -1.83800951e-01 -1.76466554e-02 4.49067891e-01
4.58460122e-01 -1.60242990e-01 4.16005403e-01 1.70311466e-01
-8.47346961e-01 9.20312941e-01 4.33052689e-01 2.20380336e-01
2.66882360e-01 -1.29499567e+00 2.88283110e-01 6.16237760e-01
-3.41756945e-03 8.23019505e-01 8.24067771e-01 -1.51980102e-01
-1.57702363e+00 -6.50406480e-01 4.61654365e-01 -5.56011617e-01
6.87811255e-01 -5.88120401e-01 -8.33005190e-01 5.06336570e-01
1.41265951e-02 -4.32414263e-01 8.79000425e-01 4.45143789e-01
-6.19415760e-01 -2.01341897e-01 -9.36612189e-01 6.91043377e-01
1.14032209e+00 -4.15394902e-01 -2.00598046e-01 3.53026479e-01
9.19836819e-01 2.99245259e-03 -6.55389965e-01 1.75473288e-01
5.95230818e-01 -1.08315575e+00 7.10731685e-01 -7.67541707e-01
3.53851378e-01 3.61279309e-01 -1.55970976e-01 -1.33068347e+00
-3.88886958e-01 -6.08172655e-01 -1.71869010e-01 7.11834550e-01
8.25152636e-01 -6.96354687e-01 7.79865026e-01 7.22042382e-01
-6.17443211e-02 -8.19351912e-01 -8.53972375e-01 -9.65070605e-01
7.07457602e-01 -4.06533509e-01 1.19150192e-01 8.23601842e-01
1.17811412e-01 3.68229955e-01 7.87293687e-02 -6.22122325e-02
8.26273203e-01 -3.45387816e-01 4.76804167e-01 -1.30946946e+00
-4.41619828e-02 -1.19304144e+00 -4.62637454e-01 -1.09458220e+00
-5.39491549e-02 -8.86510313e-01 -2.76228666e-01 -1.29310048e+00
4.34370905e-01 -3.37365150e-01 -5.64681649e-01 1.97534144e-01
-1.14209160e-01 -2.84137309e-01 4.88234043e-01 3.50199193e-01
-6.01100862e-01 4.90030646e-01 1.23591697e+00 3.47862691e-02
-3.16634268e-01 8.39307606e-02 -1.05650508e+00 6.88287139e-01
1.33056593e+00 -4.19621348e-01 -5.35554469e-01 -6.05194569e-01
7.09963262e-01 -3.39504987e-01 1.86016858e-01 -1.01187623e+00
4.60716069e-01 -4.15081568e-02 3.42572927e-01 -1.12783343e-01
1.28881305e-01 -5.81220448e-01 -6.26004577e-01 7.63119578e-01
-1.24651384e+00 1.03087589e-01 2.86170453e-01 6.02764487e-01
3.19960117e-01 -2.02429190e-01 1.13748729e+00 -1.49939626e-01
-4.85671699e-01 4.90069091e-01 -3.03980380e-01 5.85106075e-01
7.19843328e-01 1.30155995e-01 -5.13021231e-01 -4.25384581e-01
-5.66934168e-01 2.94033617e-01 1.77684292e-01 3.66674840e-01
6.04697168e-01 -9.85609591e-01 -6.64878249e-01 2.16171861e-01
-3.87884937e-02 -4.44072545e-01 9.88296196e-02 7.89952636e-01
-3.37227613e-01 6.78015947e-01 -2.16842264e-01 -4.49920833e-01
-9.35693383e-01 6.90835178e-01 6.56282544e-01 -1.76624022e-02
-4.72943187e-01 1.00803435e+00 4.17333424e-01 -5.26941121e-02
8.15021813e-01 -2.26507306e-01 1.75736725e-01 1.02110997e-01
7.64578223e-01 3.35179031e-01 -9.16934982e-02 2.02118233e-01
-2.07508817e-01 3.95533144e-01 -3.30512315e-01 -2.76041299e-01
1.00875747e+00 -3.34535956e-01 1.76551774e-01 7.03973651e-01
1.48279417e+00 -5.74344039e-01 -1.49143279e+00 -3.65273565e-01
1.87406778e-01 -3.65510702e-01 3.17662805e-01 -6.25164926e-01
-1.15816784e+00 1.32177126e+00 7.67108977e-01 5.11488557e-01
9.41495955e-01 1.94085196e-01 1.52243331e-01 8.52831602e-01
1.59969449e-01 -1.23803043e+00 1.63212806e-01 6.59031808e-01
8.86692166e-01 -1.36091566e+00 -8.50934461e-02 4.23200205e-02
-4.54934359e-01 9.78073776e-01 6.27338052e-01 -7.04402151e-03
1.03000128e+00 3.76803726e-01 -1.98862091e-01 -6.07327707e-02
-1.08269441e+00 -2.82707959e-01 1.39204383e-01 3.34353119e-01
6.38698161e-01 -9.86889079e-02 -3.03750187e-01 4.59383488e-01
-4.96188760e-01 -1.94563031e-01 5.03358364e-01 5.50849974e-01
-9.10089672e-01 -6.69040024e-01 -1.03840595e-02 4.67151701e-01
-6.57992840e-01 -3.46990287e-01 -5.24698675e-01 9.52500522e-01
-2.30222091e-01 9.04992044e-01 4.15966921e-02 -4.25186604e-01
1.00688161e-02 5.00773266e-02 4.91071671e-01 -4.03755844e-01
-2.27168575e-01 -4.16640908e-01 -4.92651343e-01 -4.81866330e-01
3.87062468e-02 -4.44455981e-01 -9.54617202e-01 -6.54168069e-01
-3.23279262e-01 1.47766143e-01 7.90166438e-01 9.13221896e-01
4.04298663e-01 4.70547497e-01 5.41076601e-01 -4.24964756e-01
-7.83858776e-01 -7.95535445e-01 -5.53308725e-01 2.18693376e-01
8.55231583e-01 -5.76195955e-01 -8.12397361e-01 -2.40404800e-01] | [8.05884838104248, 3.585968255996704] |
d17ec16b-68c0-4cc4-9958-3c15ce3b0dbf | noisy-multi-label-semi-supervised | 1902.07517 | null | http://arxiv.org/abs/1902.07517v1 | http://arxiv.org/pdf/1902.07517v1.pdf | Noisy multi-label semi-supervised dimensionality reduction | Noisy labeled data represent a rich source of information that often are
easily accessible and cheap to obtain, but label noise might also have many
negative consequences if not accounted for. How to fully utilize noisy labels
has been studied extensively within the framework of standard supervised
machine learning over a period of several decades. However, very little
research has been conducted on solving the challenge posed by noisy labels in
non-standard settings. This includes situations where only a fraction of the
samples are labeled (semi-supervised) and each high-dimensional sample is
associated with multiple labels. In this work, we present a novel
semi-supervised and multi-label dimensionality reduction method that
effectively utilizes information from both noisy multi-labels and unlabeled
data. With the proposed Noisy multi-label semi-supervised dimensionality
reduction (NMLSDR) method, the noisy multi-labels are denoised and unlabeled
data are labeled simultaneously via a specially designed label propagation
algorithm. NMLSDR then learns a projection matrix for reducing the
dimensionality by maximizing the dependence between the enlarged and denoised
multi-label space and the features in the projected space. Extensive
experiments on synthetic data, benchmark datasets, as well as a real-world case
study, demonstrate the effectiveness of the proposed algorithm and show that it
outperforms state-of-the-art multi-label feature extraction algorithms. | ['Cristina Soguero-Ruiz', 'Karl Øyvind Mikalsen', 'Filippo Maria Bianchi', 'Robert Jenssen'] | 2019-02-20 | null | null | null | null | ['supervised-dimensionality-reduction'] | ['computer-vision'] | [ 5.71271718e-01 -1.38800576e-01 -1.04483396e-01 -6.58194780e-01
-1.14914370e+00 -5.14419019e-01 5.34380734e-01 9.38718840e-02
-3.72393340e-01 6.98303103e-01 1.04994759e-01 3.32655013e-01
-2.77449548e-01 -4.93838340e-01 -3.36490929e-01 -1.20097744e+00
5.62653720e-01 5.43892801e-01 -4.36742097e-01 2.74115384e-01
-2.14816686e-02 2.52584249e-01 -1.87771153e+00 4.74713407e-02
4.81818438e-01 8.90414238e-01 4.73230816e-02 3.13815415e-01
-2.25902453e-01 5.86881101e-01 -5.57616055e-01 -6.23576231e-02
4.01019424e-01 -5.53549409e-01 -8.28964889e-01 7.51893222e-01
2.89449871e-01 2.82937646e-01 1.13675252e-01 1.44793940e+00
6.28937542e-01 2.46871427e-01 7.28626788e-01 -1.02925837e+00
-6.12672567e-01 5.29076099e-01 -8.25275540e-01 -6.94554374e-02
6.83567896e-02 -2.89350837e-01 8.23464274e-01 -1.08582652e+00
7.83999741e-01 1.53004658e+00 5.44297457e-01 6.72636032e-01
-1.47082412e+00 -6.03360593e-01 5.27609549e-02 -9.11452174e-02
-1.36782682e+00 -2.90811002e-01 8.89434218e-01 -4.14955735e-01
3.28040302e-01 2.30450645e-01 1.82217285e-01 9.46720123e-01
-6.58522844e-02 5.70705295e-01 1.61780095e+00 -6.81538343e-01
4.68691677e-01 4.19759035e-01 4.60723311e-01 4.25457627e-01
1.69036970e-01 6.32885098e-02 -4.27353770e-01 -4.42980111e-01
6.59407601e-02 2.48983577e-01 -8.97034928e-02 -4.92491335e-01
-1.16492879e+00 9.26413774e-01 -1.45625904e-01 3.76617759e-01
-3.94159645e-01 -1.24413989e-01 5.30012131e-01 4.68198508e-01
8.68892431e-01 1.83991194e-01 -4.30443525e-01 5.39524741e-02
-9.02113080e-01 1.49844084e-02 5.30786276e-01 8.30581844e-01
1.09075534e+00 -2.55181462e-01 8.87044296e-02 1.07271683e+00
4.39391285e-01 7.23391950e-01 7.48226345e-01 -1.02598190e+00
1.99987635e-01 5.83550394e-01 1.08931497e-01 -9.52210009e-01
-5.83191276e-01 -3.05896431e-01 -1.08450234e+00 -2.18846072e-02
3.03975970e-01 -1.46943331e-01 -8.60426784e-01 1.71313643e+00
9.70361292e-01 2.91060984e-01 1.98812529e-01 8.20045471e-01
7.00179279e-01 6.47438049e-01 9.08944607e-02 -7.75310576e-01
1.20982909e+00 -9.73717034e-01 -1.16169322e+00 -3.15590292e-01
1.01736319e+00 -9.31401253e-01 9.45241272e-01 5.16701639e-01
-5.80194771e-01 -4.39730227e-01 -8.61103237e-01 4.84742001e-02
-2.83620030e-01 2.16786265e-01 3.19753081e-01 8.95712674e-01
-6.51148379e-01 5.46316087e-01 -7.09218085e-01 -4.22125489e-01
3.82496953e-01 2.70326495e-01 -5.35392046e-01 -6.61060154e-01
-9.92234290e-01 7.00423837e-01 3.74128222e-01 1.90941960e-01
-7.47083426e-01 -3.32566917e-01 -8.00830543e-01 -3.56144369e-01
6.21641219e-01 -3.23812336e-01 9.94107664e-01 -6.83870971e-01
-1.24098730e+00 8.66852283e-01 -2.74601877e-01 1.86258927e-01
1.82400107e-01 -5.36184981e-02 -6.48769855e-01 -2.98019256e-02
3.56578082e-01 3.24995667e-01 9.96040821e-01 -1.56460917e+00
-8.32985520e-01 -7.11433649e-01 -4.57187742e-01 3.30283225e-01
-5.53794026e-01 1.27402335e-01 2.50740033e-02 -7.12921143e-01
6.51689768e-01 -1.24239993e+00 -2.11174279e-01 -3.56609613e-01
-4.63203996e-01 -3.22363228e-01 1.22211838e+00 -2.20575467e-01
1.10057247e+00 -2.29331422e+00 3.72132272e-01 2.17221245e-01
2.67124772e-01 2.63691515e-01 -1.81908846e-01 2.24628434e-01
-3.92175406e-01 2.49498785e-01 -4.07661647e-01 -7.67648101e-01
-1.23287417e-01 4.35283422e-01 4.88220826e-02 1.01138449e+00
-1.35592416e-01 4.31063294e-01 -1.14523697e+00 -5.19933939e-01
3.34949225e-01 4.66699958e-01 2.83477634e-01 2.91311648e-02
1.16397440e-01 6.43758714e-01 -4.49242890e-01 8.95495832e-01
8.49549472e-01 -3.34077686e-01 2.32635677e-01 -1.28135681e-01
2.31153011e-01 -3.24065506e-01 -1.80381966e+00 1.61129653e+00
-4.22758281e-01 6.94658980e-02 1.05609529e-01 -9.91173089e-01
9.05527353e-01 4.83135462e-01 6.35653377e-01 -2.67749757e-01
3.68630439e-01 2.50240058e-01 -7.37120092e-01 -4.91935700e-01
2.64205396e-01 -6.91363573e-01 -2.08386794e-01 9.72634435e-01
9.34595838e-02 -4.12041992e-02 3.51923257e-01 -1.11570721e-02
9.79095340e-01 -2.07051501e-01 5.66746473e-01 -2.02307608e-02
5.68361461e-01 -1.36132538e-01 8.09147835e-01 5.90248048e-01
-3.57227743e-01 6.16286457e-01 7.46532306e-02 -2.98208624e-01
-8.47473383e-01 -5.26074886e-01 -2.65809536e-01 1.07316899e+00
1.68721646e-01 -2.95504034e-01 -7.01022327e-01 -1.02447283e+00
-1.34613410e-01 5.40047526e-01 -6.32302582e-01 -1.91278234e-01
-2.72023290e-01 -1.12675834e+00 3.32109183e-01 4.01762687e-02
1.05228737e-01 -8.61456692e-01 1.37008550e-02 2.73048252e-01
-2.12445661e-01 -9.48184431e-01 -3.47690225e-01 5.17783046e-01
-7.87174344e-01 -1.02930844e+00 -3.51447523e-01 -8.75740349e-01
9.28011119e-01 6.37477398e-01 7.14572787e-01 -2.25635082e-01
-6.11093901e-02 2.15328231e-01 -5.36199868e-01 -1.38877705e-01
-7.28142679e-01 -2.37133317e-02 3.43301624e-01 4.37619120e-01
7.68219709e-01 -3.87061149e-01 4.70770933e-02 4.85938877e-01
-1.31983328e+00 -3.50462407e-01 5.30551851e-01 1.14272678e+00
1.14767301e+00 5.36735713e-01 8.69332969e-01 -1.37890303e+00
5.72627127e-01 -7.52228260e-01 -4.35549319e-01 3.03072095e-01
-8.99427414e-01 2.14273587e-01 6.39849424e-01 -6.02949977e-01
-1.00790989e+00 5.78591108e-01 1.85364991e-01 -3.37489456e-01
-6.17998242e-01 2.95933068e-01 -4.12812203e-01 -1.16490111e-01
7.68266559e-01 -6.00699447e-02 -5.97342439e-02 -7.24220872e-01
7.18984604e-01 1.18814540e+00 2.05989361e-01 -3.61298829e-01
7.90280879e-01 7.46917844e-01 2.96461970e-01 -5.89279115e-01
-1.30399549e+00 -1.03486383e+00 -8.40540230e-01 -8.15629438e-02
5.45862615e-01 -9.09740031e-01 -1.99490860e-01 5.48390567e-01
-7.10674167e-01 3.95865500e-01 -5.71286321e-01 5.45996845e-01
-3.52257103e-01 5.98715484e-01 -4.42988932e-01 -7.16807008e-01
-1.63800389e-01 -1.36126828e+00 1.24086273e+00 1.37090087e-01
-8.52031708e-02 -1.04768395e+00 2.67155349e-01 5.46943486e-01
3.48637328e-02 2.75922716e-01 8.02883446e-01 -8.51319313e-01
-2.90364325e-02 -3.95762116e-01 -7.26043358e-02 7.74472177e-01
4.86631721e-01 -5.60910165e-01 -1.10663056e+00 -4.41348761e-01
6.04859829e-01 -7.02581227e-01 8.24532688e-01 2.18775973e-01
6.41262114e-01 7.33425235e-03 -2.84478396e-01 2.37865925e-01
1.50363302e+00 -1.60706922e-01 -1.14513047e-01 1.82004452e-01
1.01939964e+00 6.26022995e-01 9.77902710e-01 5.61344564e-01
2.71583736e-01 3.53972673e-01 3.19489002e-01 1.19457260e-01
9.58024804e-03 2.34573364e-01 2.86192121e-03 1.21183956e+00
5.12150526e-01 -1.93152919e-01 -6.93058670e-01 3.76933753e-01
-1.89129961e+00 -5.61196506e-01 -3.14782560e-01 1.99417222e+00
9.81140137e-01 -3.33227694e-01 -1.75539806e-01 6.25060260e-01
1.16094923e+00 6.77776709e-02 -6.20741367e-01 -8.05204213e-02
-4.27973628e-01 -1.05541527e-01 5.47690809e-01 3.35275620e-01
-1.62321913e+00 6.47183001e-01 5.84438133e+00 1.03394949e+00
-7.75321364e-01 4.25674915e-01 5.40320933e-01 -2.17829749e-01
-7.32210055e-02 -6.37674555e-02 -9.84227180e-01 2.82675415e-01
1.09688592e+00 2.35456545e-02 4.63909984e-01 1.00338173e+00
1.94465160e-01 -2.62990981e-01 -1.12679207e+00 1.16536653e+00
2.75898129e-01 -7.30363965e-01 -2.78113782e-01 1.56930014e-01
1.30198038e+00 -2.53942490e-01 1.00261949e-01 1.03763796e-01
3.43805760e-01 -6.88142896e-01 4.86137718e-01 4.01533633e-01
7.80161917e-01 -8.98618281e-01 9.77649987e-01 6.74249232e-01
-1.01343775e+00 -3.53048265e-01 -5.32540917e-01 1.56352475e-01
2.15476021e-01 1.38467753e+00 -4.90709245e-01 4.95023251e-01
4.33992952e-01 8.37035894e-01 -6.03241205e-01 6.37370169e-01
-2.69032627e-01 5.50642014e-01 -3.69443595e-01 3.01928103e-01
2.83097208e-01 -3.25428486e-01 4.01672781e-01 9.66591120e-01
3.07146519e-01 8.56969059e-02 4.23254728e-01 2.93061972e-01
-2.69916028e-01 3.64372045e-01 -5.36285281e-01 -7.73273967e-03
5.32083750e-01 1.56105435e+00 -9.00177777e-01 -4.37338471e-01
-4.66986686e-01 7.35981047e-01 2.22509325e-01 2.06487477e-01
-3.90956461e-01 -5.78527935e-02 3.31511408e-01 -2.69810557e-01
1.06228560e-01 2.16287583e-01 -4.46212649e-01 -1.12561071e+00
-4.41485643e-03 -8.83244753e-01 3.74554932e-01 -3.04042518e-01
-1.50482547e+00 4.47942257e-01 -1.94317535e-01 -1.20904648e+00
-3.02494228e-01 -1.33320754e-02 3.04085016e-01 8.16632450e-01
-1.50655949e+00 -1.03056800e+00 -1.09628290e-01 3.68863374e-01
4.87464458e-01 -5.38784303e-02 1.09111273e+00 4.86001194e-01
-7.89858282e-01 3.63382131e-01 9.63756621e-01 -3.90945017e-01
9.64249194e-01 -1.26497424e+00 -1.66260451e-02 8.17710161e-01
1.28544480e-01 2.85925090e-01 6.20504379e-01 -6.94537282e-01
-1.13896370e+00 -1.58020794e+00 9.55761313e-01 -2.80635983e-01
4.18360263e-01 -3.19336116e-01 -9.16876018e-01 5.55405200e-01
-2.92445183e-01 5.60392082e-01 1.10249269e+00 -6.52897656e-02
-3.46327573e-01 -9.13838968e-02 -1.56459570e+00 2.52134458e-04
7.74245441e-01 -5.16268790e-01 -4.31210250e-01 8.39948237e-01
5.27779758e-01 -1.34652168e-01 -9.86605108e-01 1.99880064e-01
1.02651022e-01 -6.53373122e-01 7.41549671e-01 -2.03498900e-01
4.48764395e-03 -4.37862664e-01 -3.93380493e-01 -1.70231235e+00
-3.67917657e-01 -2.28577703e-01 -9.79465321e-02 1.44223773e+00
7.17006177e-02 -2.69805342e-01 8.08450699e-01 5.24710000e-01
1.45785049e-01 -6.68831229e-01 -9.55034614e-01 -6.54901981e-01
-2.16790780e-01 -3.52501810e-01 5.79119921e-01 1.15809596e+00
-2.24470958e-01 3.94817233e-01 -7.07323015e-01 1.57900512e-01
9.84047413e-01 1.38846897e-02 3.58368546e-01 -1.54859555e+00
7.42712915e-02 1.86865658e-01 -3.67416948e-01 -5.40353239e-01
4.71082479e-01 -8.84783149e-01 1.39959857e-01 -1.30465615e+00
2.82721579e-01 -5.27874827e-01 -3.15145522e-01 4.89272326e-01
-2.82996982e-01 4.28677261e-01 -7.60991797e-02 4.58314270e-01
-8.74191523e-01 4.80897188e-01 1.09033871e+00 -7.56445378e-02
-1.72276542e-01 2.85459794e-02 -8.24095249e-01 8.40293884e-01
6.58739090e-01 -1.24526739e+00 -4.52649266e-01 -1.00389317e-01
1.81119382e-01 -2.33899325e-01 -2.73398668e-01 -7.13801026e-01
1.72438741e-01 -2.02941656e-01 4.24488336e-02 -4.67524618e-01
2.63430923e-01 -1.02806664e+00 3.67023408e-01 -1.08250387e-01
-5.05612195e-01 -3.85529697e-01 -4.22836274e-01 1.04928267e+00
-2.82000691e-01 -5.62185526e-01 1.10722411e+00 -2.43231565e-01
-5.21164477e-01 2.78858431e-02 -3.45421493e-01 6.96095601e-02
1.25478888e+00 3.53835635e-02 -8.95574763e-02 4.64408994e-02
-1.08744609e+00 3.67270596e-02 4.02762473e-01 2.82443553e-01
4.12530363e-01 -1.56214559e+00 -5.37440181e-01 1.35503992e-01
2.59176522e-01 1.21079475e-01 4.02011544e-01 4.90287870e-01
3.48107261e-03 2.96761751e-01 2.86735624e-01 -6.00576282e-01
-1.37811577e+00 9.00568306e-01 -2.95935292e-02 -4.57864851e-01
-5.23837626e-01 7.13408351e-01 -1.58521861e-01 -8.58593106e-01
2.88813531e-01 -6.99738935e-02 -2.89039701e-01 4.62992221e-01
6.96308434e-01 7.99708724e-01 3.50864470e-01 -1.26530874e+00
-7.34611750e-02 8.51600945e-01 -8.38616937e-02 1.09717727e-01
1.36697853e+00 -7.90327787e-01 -4.21316445e-01 8.36943924e-01
1.55394328e+00 -4.19369459e-01 -8.88561785e-01 -8.62591445e-01
1.30289465e-01 -5.02302825e-01 2.72879750e-01 -5.50352752e-01
-1.16291869e+00 5.68906248e-01 9.16145325e-01 2.62535602e-01
1.25004709e+00 2.70758476e-02 7.34474003e-01 5.06710708e-01
3.03375930e-01 -1.43455410e+00 9.15184468e-02 2.01974377e-01
2.82214701e-01 -1.55925822e+00 1.21315941e-02 -6.55061841e-01
-6.38126373e-01 7.41257787e-01 1.75327167e-01 1.45154744e-01
1.00719726e+00 1.74479485e-01 4.09862727e-01 -2.87395507e-01
-4.62780207e-01 -3.69651802e-02 -1.23516023e-01 5.22964597e-01
-4.11343873e-02 1.68612584e-01 -3.41717035e-01 6.03240013e-01
2.80945301e-01 -8.19826126e-02 6.95972025e-01 1.29054379e+00
-6.65998340e-01 -1.37471402e+00 -1.02068722e+00 6.59242332e-01
-4.66919303e-01 2.27622092e-01 -1.14576131e-01 2.68435568e-01
5.43697894e-01 1.39679277e+00 -5.34606695e-01 -3.93214703e-01
2.69519299e-01 4.51673806e-01 -2.02321429e-02 -1.03263986e+00
-4.19912755e-01 4.27385002e-01 -2.15510488e-01 -3.29275161e-01
-9.72054720e-01 -9.42583561e-01 -1.21638429e+00 -1.03862047e-01
-8.06755960e-01 1.75874963e-01 9.60205853e-01 1.19310248e+00
2.71960825e-01 3.45696062e-01 1.02882719e+00 -7.85372555e-01
-9.24885452e-01 -1.10626197e+00 -1.27678108e+00 6.07380629e-01
3.12615007e-01 -8.54466736e-01 -8.45381916e-01 2.44182855e-01] | [9.453397750854492, 3.948620319366455] |
a4a54ee8-1731-498f-9ff0-95a2f68d1d23 | surge-routing-event-informed-multiagent | 2307.02637 | null | https://arxiv.org/abs/2307.02637v1 | https://arxiv.org/pdf/2307.02637v1.pdf | Surge Routing: Event-informed Multiagent Reinforcement Learning for Autonomous Rideshare | Large events such as conferences, concerts and sports games, often cause surges in demand for ride services that are not captured in average demand patterns, posing unique challenges for routing algorithms. We propose a learning framework for an autonomous fleet of taxis that scrapes event data from the internet to predict and adapt to surges in demand and generates cooperative routing and pickup policies that service a higher number of requests than other routing protocols. We achieve this through a combination of (i) an event processing framework that scrapes the internet for event information and generates dense vector representations that can be used as input features for a neural network that predicts demand; (ii) a two neural network system that predicts hourly demand over the entire map, using these dense vector representations; (iii) a probabilistic approach that leverages locale occupancy schedules to map publicly available demand data over sectors to discretized street intersections; and finally, (iv) a scalable model-based reinforcement learning framework that uses the predicted demand over intersections to anticipate surges and route taxis using one-agent-at-a-time rollout with limited sampling certainty equivalence. We learn routing and pickup policies using real NYC ride share data for 2022 and information for more than 2000 events across 300 unique venues in Manhattan. We test our approach with a fleet of 100 taxis on a map with 38 different sectors (2235 street intersections). Our experimental results demonstrate that our method obtains routing policies that service $6$ more requests on average per minute (around $360$ more requests per hour) than other model-based RL frameworks and other classical algorithms in operations research when dealing with surge demand conditions. | ['Stephanie Gil', 'Daniel Garces'] | 2023-07-05 | null | null | null | null | ['model-based-reinforcement-learning'] | ['reasoning'] | [-5.13909876e-01 2.37141788e-01 -5.08949876e-01 -5.57906568e-01
-1.00247943e+00 -4.25572634e-01 5.38591683e-01 3.29767853e-01
-4.01446015e-01 1.24620509e+00 6.20613813e-01 -3.73346120e-01
-5.90373695e-01 -1.53968191e+00 -8.27779412e-01 -3.40626329e-01
-7.68127978e-01 1.66263509e+00 2.38856182e-01 -6.45628393e-01
-5.54587282e-02 6.66549861e-01 -1.57694566e+00 9.65448155e-04
4.52380896e-01 1.00239444e+00 -1.68732762e-01 4.58590597e-01
-8.95880386e-02 8.38417113e-01 -4.61562365e-01 7.71855861e-02
7.36868858e-01 3.69560808e-01 -3.90638351e-01 -4.78166267e-02
-2.70806104e-01 -7.08341062e-01 -1.07094896e+00 9.65209976e-02
3.15591007e-01 4.92378116e-01 6.67432249e-01 -1.77972960e+00
2.64407992e-01 1.08618820e+00 -6.73925996e-01 5.39312541e-01
1.37292385e-01 5.37367761e-01 9.96703923e-01 -1.09831713e-01
3.85569513e-01 1.11710572e+00 3.66275847e-01 -7.09519312e-02
-1.18766665e+00 -9.09511924e-01 5.13848186e-01 2.97210783e-01
-1.45284033e+00 3.01295053e-02 3.61648589e-01 -7.89422393e-02
1.25696123e+00 8.49164501e-02 7.90798426e-01 1.02944636e+00
3.84632707e-01 8.50366592e-01 6.46564782e-01 3.27965975e-01
6.31652415e-01 -1.07543923e-01 -1.87272012e-01 -1.42236911e-02
9.80059952e-02 4.88031030e-01 -3.35229158e-01 -1.20814852e-01
5.35725176e-01 2.90778726e-01 4.17253375e-01 8.40613842e-02
-9.20001805e-01 1.08490062e+00 5.28726876e-01 -4.84388739e-01
-1.32610118e+00 3.11560631e-01 3.22398961e-01 2.76302725e-01
3.17454994e-01 3.62730622e-01 -5.70547462e-01 -3.19815129e-01
-9.56988931e-01 8.18824887e-01 1.13290298e+00 1.10323668e+00
9.87144947e-01 5.31981945e-01 -1.89325675e-01 3.75010163e-01
-6.61230162e-02 9.78466570e-01 -8.26526135e-02 -1.15102005e+00
8.04311454e-01 1.76637590e-01 5.65037251e-01 -1.01967013e+00
-1.15448654e+00 -2.01797470e-01 -4.42240834e-01 -1.06249355e-01
3.11762184e-01 -9.36197221e-01 -5.57703555e-01 1.79556322e+00
2.26489976e-01 4.25166160e-01 -1.77034140e-01 9.90798175e-01
-2.14924634e-01 1.20225978e+00 7.15568056e-03 1.74204987e-02
1.06185412e+00 -4.31274593e-01 -4.08025384e-01 -1.85394287e-01
4.10260290e-01 -1.84918121e-01 2.57814527e-01 4.25048113e-01
-1.30855429e+00 -1.15515850e-01 -5.37699938e-01 7.20756471e-01
-7.00022936e-01 -4.59567606e-01 8.25978935e-01 5.53758182e-02
-7.82539606e-01 1.15122490e-01 -7.27663875e-01 -2.55987078e-01
3.82444143e-01 5.86983621e-01 2.23755330e-01 -8.52359608e-02
-1.46818316e+00 9.61727142e-01 3.63999158e-01 -1.25319004e-01
-1.29194844e+00 -1.03140676e+00 -8.56635034e-01 4.62769002e-01
7.19133496e-01 -1.44144356e-01 1.19743121e+00 -2.63011187e-01
-1.11488760e+00 -2.51810223e-01 3.50229502e-01 -8.52257788e-01
3.18075269e-01 4.39813435e-01 -9.00620997e-01 -1.71431392e-01
3.04773927e-01 7.57476091e-01 1.60936072e-01 -1.15131450e+00
-1.56811047e+00 -2.00037643e-01 -2.63247043e-02 3.07011008e-01
3.04184675e-01 -4.91136521e-01 3.43244225e-02 -1.87675521e-01
-3.96185219e-01 -9.72114801e-01 -6.49320722e-01 -8.61039996e-01
-2.21808657e-01 -5.38580954e-01 6.34256184e-01 -4.24458385e-01
1.24530315e+00 -1.54902852e+00 -4.63390499e-01 1.05400252e+00
-8.96564350e-02 -3.54986191e-01 -3.17925692e-01 1.13083148e+00
3.75829905e-01 -2.86296725e-01 2.02977672e-01 -6.43938407e-02
5.10266960e-01 6.84655845e-01 -6.98601782e-01 3.95126343e-01
-3.82937468e-03 6.90917015e-01 -9.05381382e-01 2.06122287e-02
5.13484299e-01 5.72506860e-02 -6.45364225e-01 6.91253692e-02
-4.69500631e-01 1.40280500e-01 -5.30535042e-01 5.59116542e-01
6.66150689e-01 7.71702752e-02 3.40440303e-01 3.32326382e-01
-3.53721410e-01 4.82297838e-01 -1.53137434e+00 1.02862167e+00
-7.65951395e-01 4.87157732e-01 1.18782043e-01 -1.23905528e+00
9.27802026e-01 1.83532014e-01 1.10070646e+00 -1.15150046e+00
-1.01656474e-01 1.00744300e-01 -3.64584357e-01 -2.60631233e-01
6.87488198e-01 6.13998175e-02 -8.70148420e-01 9.58174169e-01
-2.66552597e-01 -1.40800849e-01 4.34723794e-01 3.06046695e-01
1.57705390e+00 -4.16672021e-01 -3.28558445e-01 -1.70021147e-01
-2.07115903e-01 4.01744962e-01 8.97574723e-01 7.43142843e-01
-1.02065116e-01 -1.39237791e-01 6.24002635e-01 -1.14911771e+00
-1.16290069e+00 -1.46673000e+00 1.08365692e-01 1.07686830e+00
1.03739478e-01 1.11242570e-01 -1.63299054e-01 -3.56196821e-01
7.09655702e-01 1.26091635e+00 -4.14408803e-01 1.58497125e-01
-6.99704647e-01 -6.98038697e-01 3.25559616e-01 5.50846279e-01
3.15653205e-01 -1.02303278e+00 -3.94741118e-01 8.45939040e-01
-2.49039918e-01 -1.08352923e+00 -4.63707447e-01 3.25117469e-01
-3.90683636e-02 -8.95025671e-01 -9.30675194e-02 -2.76669592e-01
4.20487106e-01 1.50147706e-01 1.21670055e+00 -5.55025280e-01
-1.46008164e-01 7.73308456e-01 1.83065295e-01 -7.11206853e-01
-3.42389219e-03 4.94315445e-01 4.54228759e-01 1.15019679e-01
7.00830698e-01 -8.14704835e-01 -7.77823627e-01 6.56296372e-01
-7.22256243e-01 -3.62453043e-01 4.21081007e-01 3.27420115e-01
5.09741843e-01 3.82933319e-01 1.41388309e+00 -5.28158426e-01
5.83187163e-01 -1.40204501e+00 -1.12316012e+00 -1.87436312e-01
-7.61565447e-01 -9.96278822e-02 6.46600008e-01 2.67912690e-02
-1.02452004e+00 -8.35929886e-02 6.96195588e-02 -5.47323599e-02
-2.76903421e-01 4.39820707e-01 3.34367990e-01 5.96034050e-01
5.33583283e-01 1.31199181e-01 8.92508104e-02 4.04828526e-02
2.87832499e-01 6.85967863e-01 4.61949408e-01 -7.82943308e-01
9.47718322e-01 4.56160873e-01 -1.05926566e-01 -5.58180809e-01
-2.67224699e-01 -4.06725019e-01 -3.54220439e-03 -3.53493273e-01
3.85115772e-01 -1.17671430e+00 -1.54331589e+00 -9.23751891e-02
-5.82854033e-01 -7.35197008e-01 -5.87344646e-01 7.77515590e-01
-1.06319821e+00 -6.66035891e-01 -7.19630241e-01 -9.29802239e-01
2.97760069e-01 -8.64407361e-01 6.15103304e-01 4.96838838e-01
-1.91442166e-02 -1.04359555e+00 3.81865472e-01 1.44848615e-01
9.04239595e-01 3.03321332e-01 6.64669454e-01 -7.96570957e-01
-9.60784912e-01 -3.37981880e-01 -1.90760523e-01 -2.80859739e-01
-7.13656992e-02 -2.05170050e-01 -4.34187740e-01 -3.06691229e-01
-1.08686316e+00 -3.59693706e-01 3.94051135e-01 7.64437675e-01
1.08442676e+00 -7.49772966e-01 -6.41038060e-01 1.18471824e-01
1.41362309e+00 2.56715953e-01 6.14007413e-01 3.73772651e-01
8.75514597e-02 7.54394352e-01 7.10401595e-01 1.19803965e+00
1.40329039e+00 6.46631598e-01 8.96151960e-01 -1.29513562e-01
5.36112607e-01 -7.49167025e-01 1.15496427e-01 1.22961521e-01
1.52585462e-01 -4.50934619e-01 -1.00991416e+00 9.29350913e-01
-1.97293985e+00 -1.28033996e+00 3.18711430e-01 2.36249614e+00
4.62566048e-01 3.73051971e-01 7.65864670e-01 -2.99350828e-01
4.13270831e-01 1.37482047e-01 -8.32729220e-01 -7.42691100e-01
9.68109965e-02 -6.10939711e-02 1.47534835e+00 5.05365908e-01
-6.04946911e-01 9.51672316e-01 6.31548834e+00 6.31147087e-01
-8.15441191e-01 -2.30964527e-01 7.56456673e-01 -5.89208663e-01
-4.88714665e-01 -2.27771759e-01 -1.05982924e+00 6.85327530e-01
1.75794005e+00 -3.42088073e-01 1.06528032e+00 7.65996218e-01
9.11227763e-01 -9.31663290e-02 -7.64203906e-01 6.43095255e-01
-4.13381577e-01 -1.70747960e+00 -1.68140128e-01 6.57232881e-01
9.16257262e-01 7.28653193e-01 6.70494586e-02 9.80123937e-01
1.41836333e+00 -9.79936898e-01 2.83460289e-01 7.64470518e-01
1.44083768e-01 -1.54994106e+00 6.07418001e-01 3.68824422e-01
-1.09796393e+00 -7.00836957e-01 -2.57275939e-01 -1.25717103e-01
8.51611674e-01 4.12679315e-01 -1.19067121e+00 3.10344100e-01
5.88593125e-01 4.11108911e-01 3.00262570e-01 1.04168439e+00
3.33309740e-01 7.51419842e-01 -9.70886409e-01 1.13792688e-01
8.85187149e-01 -3.68098319e-02 4.49620962e-01 8.41256380e-01
2.01920718e-01 1.21713668e-01 7.79658437e-01 6.52077019e-01
-1.58143669e-01 -4.11171526e-01 -9.23781574e-01 5.45940220e-01
1.10709739e+00 1.26622891e+00 -3.49518061e-01 -2.45999858e-01
-4.01448250e-01 1.35609940e-01 -1.80988133e-01 8.32794249e-01
-1.06112373e+00 -3.84543747e-01 1.18241465e+00 5.67519248e-01
2.90031731e-01 -4.03793246e-01 1.95494249e-01 -4.47700530e-01
-5.13635814e-01 -6.47375047e-01 4.94505018e-01 -5.32689571e-01
-1.41188073e+00 1.12904131e-01 2.92669117e-01 -1.16157436e+00
-7.78540194e-01 -2.22479597e-01 -8.60041559e-01 6.27963424e-01
-2.02340961e+00 -8.80706549e-01 1.75044149e-01 8.14384043e-01
7.66458333e-01 -2.29631171e-01 3.83780986e-01 3.80560905e-01
-6.53117061e-01 2.25069150e-01 1.86156765e-01 8.56571943e-02
3.43332618e-01 -1.17323875e+00 5.56013882e-01 1.54722542e-01
-1.80535749e-01 -2.86604464e-01 4.95731652e-01 -5.25654674e-01
-1.43144000e+00 -1.27788115e+00 6.48148477e-01 -2.29635000e-01
8.25944245e-01 -1.17568947e-01 -4.71010089e-01 1.16313648e+00
1.54660255e-01 -1.44903556e-01 6.14813387e-01 9.90342274e-02
2.40061179e-01 -7.25222826e-01 -1.19402421e+00 3.62439156e-01
5.51627874e-01 7.02624992e-02 -1.50663748e-01 4.77364779e-01
3.83324414e-01 -1.53621748e-01 -9.86277819e-01 -9.53813866e-02
4.20815617e-01 -5.07094443e-01 6.55060351e-01 -7.67752647e-01
-4.32197779e-01 -5.53843193e-02 -4.30482477e-01 -1.87082124e+00
-3.47588867e-01 -8.17546546e-01 3.39013398e-01 8.96006107e-01
8.01567376e-01 -9.22194123e-01 7.62043178e-01 1.18861091e+00
-2.42643923e-01 -4.85922307e-01 -1.33481586e+00 -5.43629944e-01
8.07444453e-02 -5.09602845e-01 1.29822648e+00 4.58692759e-01
-4.34155837e-02 -1.09317489e-01 -4.10611421e-01 5.36193430e-01
7.26034760e-01 -1.01777934e-01 8.29951227e-01 -1.18851233e+00
2.59552479e-01 -2.93756276e-01 -2.57053375e-01 -1.06759095e+00
3.70596856e-01 -6.38434410e-01 -3.32849920e-02 -1.55372250e+00
-5.04021823e-01 -9.55948591e-01 -3.61935377e-01 5.77115357e-01
9.73430097e-01 -2.91103780e-01 -1.29258499e-01 -4.99711744e-02
-8.25261235e-01 7.14320838e-01 5.75319767e-01 -1.67115510e-01
-5.07331789e-01 3.35808098e-01 -6.18936360e-01 3.41447115e-01
9.59972620e-01 -3.83300871e-01 -4.59589779e-01 -3.38406116e-01
4.72524375e-01 6.11433983e-01 1.09957397e-01 -1.01929438e+00
5.21268070e-01 -6.73134804e-01 3.86788964e-01 -1.10904086e+00
3.36768150e-01 -1.04819775e+00 9.55272242e-02 2.52701074e-01
-3.24741721e-01 5.31369627e-01 3.30018103e-01 8.51927757e-01
7.68579468e-02 6.45071924e-01 1.77832827e-01 -1.36703218e-03
-8.10800493e-01 7.71450222e-01 -1.22426784e+00 8.36693645e-02
1.22243822e+00 1.34718135e-01 -2.98641235e-01 -1.04584992e+00
-6.35631561e-01 1.56502867e+00 -1.17645673e-01 6.53735280e-01
2.94729173e-01 -1.41838706e+00 -8.80491555e-01 3.24557185e-01
-3.01535219e-01 -2.10954607e-01 4.42063451e-01 5.74513376e-01
-1.45602897e-01 6.30486727e-01 -3.87272388e-01 -3.02352697e-01
9.57449228e-02 4.54203963e-01 3.49839032e-01 -5.06512761e-01
-5.06621659e-01 -3.64485569e-02 -3.81532669e-01 -6.93172157e-01
1.05283760e-01 -2.78030455e-01 9.06509906e-02 5.07780015e-01
6.45813763e-01 7.45820701e-01 -1.55575290e-01 -4.34660673e-01
-1.32910460e-01 -3.18991505e-02 6.19771611e-03 -3.68013620e-01
1.57586777e+00 -4.78835434e-01 8.09242666e-01 2.88706332e-01
7.00591981e-01 -4.45090503e-01 -1.84797478e+00 -3.13343734e-01
-1.19118586e-01 -3.72511894e-01 3.46546054e-01 -9.55220699e-01
-1.38129973e+00 2.59966105e-01 2.42713973e-01 6.49191439e-01
7.50433326e-01 -1.53647080e-01 1.25488007e+00 3.90316248e-01
8.25850129e-01 -1.72079456e+00 -3.92953962e-01 2.36033440e-01
4.92552191e-01 -1.13607156e+00 -4.90514398e-01 5.64454198e-01
-1.02987266e+00 8.16859305e-01 4.98466223e-01 -5.28700590e-01
9.70322251e-01 3.55517685e-01 -1.96624443e-01 -2.42051527e-01
-1.37213564e+00 -5.09464204e-01 -5.69049418e-01 7.35227466e-01
-5.73395550e-01 6.53810263e-01 4.84438762e-02 6.10645831e-01
-2.18223333e-01 1.67473212e-01 8.77463281e-01 6.13956273e-01
-6.22144818e-01 -5.98975539e-01 -2.62408126e-02 7.82741904e-01
1.13034032e-01 4.27677035e-01 6.07996643e-01 1.11196601e+00
-1.99438512e-01 1.02529359e+00 9.54818666e-01 -2.39216343e-01
5.11231780e-01 -2.76434392e-01 -4.21661824e-01 -2.56307870e-01
-3.61877412e-01 -2.85084486e-01 2.21157357e-01 -1.05769897e+00
4.02542114e-01 -8.60137761e-01 -1.57098460e+00 -9.50931132e-01
3.81503373e-01 4.02524352e-01 8.19323063e-01 7.71198988e-01
7.42856860e-01 2.98883975e-01 1.54193008e+00 -1.05576384e+00
-7.26289153e-01 -6.93955421e-01 -1.14872718e+00 -6.57564923e-02
1.69289082e-01 -1.00162244e+00 -5.08626521e-01 -8.86274457e-01] | [5.443620681762695, 1.5459234714508057] |
5670c767-5864-4a0a-b98f-0838cf21b2d5 | unsupervised-heterophilous-network-embedding | 2203.10866 | null | https://arxiv.org/abs/2203.10866v3 | https://arxiv.org/pdf/2203.10866v3.pdf | Unsupervised Network Embedding Beyond Homophily | Network embedding (NE) approaches have emerged as a predominant technique to represent complex networks and have benefited numerous tasks. However, most NE approaches rely on a homophily assumption to learn embeddings with the guidance of supervisory signals, leaving the unsupervised heterophilous scenario relatively unexplored. This problem becomes especially relevant in fields where a scarcity of labels exists. Here, we formulate the unsupervised NE task as an r-ego network discrimination problem and develop the SELENE framework for learning on networks with homophily and heterophily. Specifically, we design a dual-channel feature embedding pipeline to discriminate r-ego networks using node attributes and structural information separately. We employ heterophily adapted self-supervised learning objective functions to optimise the framework to learn intrinsic node embeddings. We show that SELENE's components improve the quality of node embeddings, facilitating the discrimination of connected heterophilous nodes. Comprehensive empirical evaluations on both synthetic and real-world datasets with varying homophily ratios validate the effectiveness of SELENE in homophilous and heterophilous settings showing an up to 12.52% clustering accuracy gain. | ['Jun Pang', 'Daniele Grattarola', 'Guadalupe Gonzalez', 'Zhiqiang Zhong'] | 2022-03-21 | null | null | null | null | ['network-embedding'] | ['methodology'] | [ 1.17774121e-01 3.47318411e-01 -2.69063920e-01 -2.98746616e-01
7.00606853e-02 -6.55234218e-01 8.82787824e-01 2.92589039e-01
-2.18434617e-01 3.88777196e-01 1.44779682e-01 -1.58624604e-01
-6.12014890e-01 -1.09612823e+00 -3.66066962e-01 -7.60691941e-01
-3.99900168e-01 5.94890118e-01 -4.69489880e-02 -1.17647417e-01
-3.80584504e-03 5.99354267e-01 -1.30478215e+00 -2.30534628e-01
7.62300909e-01 6.99055731e-01 -2.71918207e-01 5.87736666e-01
6.17208295e-02 1.83828697e-01 -1.83568016e-01 -3.78228903e-01
4.87042427e-01 -3.35490853e-01 -6.64652467e-01 -3.75701450e-02
-1.52135044e-02 4.50630099e-01 -6.56507611e-01 8.55269611e-01
5.29781997e-01 -1.47567779e-01 9.55594957e-01 -1.58261180e+00
-4.81633872e-01 5.85667431e-01 -5.39366603e-01 2.36173704e-01
2.35692458e-03 -9.28655174e-03 1.62282073e+00 -6.67517424e-01
7.46254086e-01 1.24546742e+00 7.82293439e-01 3.14712822e-01
-1.83647883e+00 -5.41892171e-01 -1.17374808e-01 -6.19010627e-02
-1.40816593e+00 -4.37878817e-01 8.54176819e-01 -4.20119196e-01
5.27447045e-01 8.40173289e-02 6.79666996e-01 1.09924448e+00
-1.38160307e-02 4.29317772e-01 1.09431720e+00 -2.75514275e-01
2.05296189e-01 4.55802351e-01 2.94625498e-02 6.85064435e-01
5.47672331e-01 6.42566085e-02 -4.51459140e-01 -2.17643291e-01
6.43259048e-01 2.56922871e-01 -1.49907833e-02 -9.74197865e-01
-1.09521949e+00 1.00703275e+00 7.38153696e-01 2.34237164e-01
-3.24143589e-01 9.43832248e-02 5.84453642e-01 7.05939531e-01
4.57003117e-01 7.05961049e-01 -5.91396242e-02 4.30103764e-02
-5.96081018e-01 -4.65677410e-01 1.12899482e+00 5.68951905e-01
1.00244951e+00 -6.87948763e-02 3.25005263e-01 8.30600679e-01
2.61843473e-01 1.40559077e-01 1.55908495e-01 -5.92408180e-01
1.41636193e-01 1.03061497e+00 -3.13303620e-01 -1.44553006e+00
-4.11516041e-01 -6.25692010e-01 -9.46311057e-01 -1.18464619e-01
1.91209555e-01 -1.19084470e-01 -4.83648717e-01 1.71483254e+00
4.99199480e-01 3.33638519e-01 5.60888238e-02 6.20707810e-01
5.92769206e-01 4.16686416e-01 -1.12769842e-01 2.22162917e-01
1.09567642e+00 -8.34228516e-01 -3.88282448e-01 4.02544588e-02
8.56672049e-01 -2.52795964e-01 9.04905081e-01 -3.61697040e-02
-5.83109140e-01 -1.38778210e-01 -1.04507673e+00 4.18911248e-01
-6.32237971e-01 -5.73286861e-02 7.80758321e-01 9.39755738e-01
-1.01374316e+00 6.77616179e-01 -6.77759528e-01 -4.53485996e-01
5.26557267e-01 6.42822921e-01 -7.11295664e-01 -1.34479135e-01
-1.39312661e+00 5.44659555e-01 2.61595190e-01 9.04805511e-02
-7.77641416e-01 -6.11946344e-01 -8.73513460e-01 3.28138232e-01
2.46312827e-01 -5.18753111e-01 1.57576740e-01 -7.40416408e-01
-1.24693370e+00 8.96174252e-01 2.10541829e-01 -5.94998658e-01
4.52529639e-01 3.22545350e-01 -4.94598657e-01 4.31153327e-01
8.32784325e-02 4.93133128e-01 7.18736589e-01 -1.16552401e+00
-1.79215968e-01 -2.51189858e-01 5.00967354e-02 6.80633858e-02
-9.70509291e-01 -4.31723148e-01 -1.32015511e-01 -2.85326064e-01
9.75317806e-02 -9.93580520e-01 -3.82126532e-02 1.84410229e-01
-4.59028333e-01 -1.88842863e-01 9.94114578e-01 3.70084308e-02
1.07043898e+00 -2.16639042e+00 1.61450714e-01 9.35115039e-01
9.69525039e-01 1.13831908e-01 -1.74426049e-01 6.66120529e-01
-1.67069763e-01 1.90360546e-01 -6.08981848e-02 -1.84707716e-01
2.08814695e-01 3.07333261e-01 1.04987688e-01 6.97014272e-01
3.98455679e-01 9.23406482e-01 -1.01544523e+00 -4.64294225e-01
2.62817174e-01 6.99578404e-01 -5.77148557e-01 2.12498784e-01
3.47839952e-01 1.93868771e-01 -2.69255340e-01 4.83583778e-01
2.12974161e-01 -7.62107909e-01 6.87445045e-01 -1.46272004e-01
2.93980569e-01 7.74314627e-02 -1.12132347e+00 1.00461817e+00
-6.23628736e-01 9.34378922e-01 3.28996450e-01 -1.14224362e+00
1.34901798e+00 1.64814740e-01 6.68321133e-01 -1.67445004e-01
2.66449451e-01 2.58735836e-01 3.05851996e-01 -2.91321993e-01
9.94272009e-02 -2.41402730e-01 7.45191425e-02 7.02198088e-01
2.02501565e-01 2.87183970e-01 1.01244539e-01 4.97774750e-01
1.52700973e+00 -5.20060182e-01 1.46459177e-01 -5.13421237e-01
3.11728030e-01 -3.58606845e-01 5.17286301e-01 5.69750488e-01
-4.28417921e-01 4.07322794e-01 1.00243795e+00 -2.51398563e-01
-9.92967665e-01 -1.10619092e+00 -2.18819812e-01 1.00534582e+00
3.89048725e-01 -5.81250250e-01 -3.14374864e-01 -9.44363713e-01
3.23852748e-01 1.33607179e-01 -1.00640655e+00 -4.31368917e-01
-2.22366333e-01 -1.05923665e+00 6.18097067e-01 3.49385172e-01
2.26021662e-01 -6.72901571e-01 -1.78950876e-01 1.23606399e-02
2.90223867e-01 -1.05034733e+00 -9.45739597e-02 4.28263694e-01
-7.06665933e-01 -1.23405468e+00 -4.48765099e-01 -6.02204800e-01
9.02433515e-01 4.23672587e-01 9.78161514e-01 1.66318864e-01
-4.05471325e-01 6.24388456e-01 -3.33252430e-01 2.30659753e-01
-2.13962853e-01 4.40002978e-01 2.81809419e-01 3.48844260e-01
5.88929176e-01 -1.05946374e+00 -8.01380098e-01 6.74933255e-01
-7.85049379e-01 -3.39310586e-01 7.08939075e-01 1.10894942e+00
1.48162156e-01 2.09388345e-01 8.49885345e-01 -1.44649470e+00
6.26471758e-01 -9.90653038e-01 -3.50255370e-01 6.98731840e-02
-1.24941409e+00 5.71199283e-02 7.65737832e-01 -2.56484509e-01
-5.91815293e-01 -1.17152169e-01 2.54115164e-01 -3.83042395e-01
4.37040702e-02 3.96984875e-01 -3.27630311e-01 -3.01365107e-01
6.67008758e-01 -2.58772317e-02 2.21140012e-01 -1.40202433e-01
3.92370105e-01 7.79267251e-01 8.30588117e-02 -5.75423241e-01
9.72663164e-01 7.52987564e-01 2.73133010e-01 -9.69038486e-01
-2.17831403e-01 -6.83664739e-01 -8.16787720e-01 -1.42692477e-01
4.30293590e-01 -9.48430836e-01 -6.92643344e-01 -2.36974493e-01
-5.23898184e-01 -2.60662138e-01 -3.93008500e-01 3.47729266e-01
-2.37646297e-01 3.94250959e-01 -4.34660375e-01 -5.98155081e-01
-1.66446760e-01 -8.94545972e-01 7.10274041e-01 3.31981592e-02
-3.75932753e-01 -1.65229630e+00 1.82239786e-01 2.56941080e-01
3.97821397e-01 3.57446760e-01 8.44449341e-01 -1.09692574e+00
-3.64079237e-01 -4.62175488e-01 -5.02847075e-01 1.14590928e-01
2.23918781e-01 1.00300089e-02 -9.66105282e-01 -4.24333394e-01
-7.09083438e-01 -2.59189188e-01 9.75952268e-01 -1.34975970e-01
6.53684855e-01 -6.71290532e-02 -5.02914369e-01 6.91171646e-01
1.27584743e+00 -4.06909734e-01 4.04306769e-01 1.81678355e-01
6.83894992e-01 8.23466361e-01 1.65910095e-01 4.46809828e-01
2.76746869e-01 3.00964773e-01 5.27419269e-01 -1.88929066e-01
2.08815858e-02 -5.36214292e-01 8.17317367e-02 8.88685703e-01
3.38994950e-01 -3.02933156e-01 -1.06811047e+00 8.95206690e-01
-1.58006990e+00 -6.86900914e-01 1.34886932e-02 1.81233048e+00
5.52646816e-01 2.59395778e-01 2.62383789e-01 3.19322824e-01
1.08717573e+00 3.55075777e-01 -5.76732635e-01 -1.12583898e-01
-2.71035165e-01 -1.00435540e-01 4.23053533e-01 2.48746097e-01
-9.88139093e-01 8.68428290e-01 5.64730120e+00 4.86181021e-01
-9.64855671e-01 -4.48374301e-02 5.50403178e-01 1.88667700e-01
-5.76353312e-01 2.16171965e-01 -6.47342503e-01 4.07946467e-01
1.01517129e+00 -4.11824547e-02 8.66583735e-02 8.74754250e-01
3.35192420e-02 3.16309601e-01 -1.19342339e+00 9.38598216e-01
-1.16784297e-01 -1.27479339e+00 -2.03490868e-01 2.76477784e-01
9.67335403e-01 8.51230174e-02 1.04318537e-01 5.94670512e-02
4.96726096e-01 -1.15305579e+00 -2.12638780e-01 1.13952763e-01
9.20680761e-01 -5.49638391e-01 7.20825315e-01 -9.19262841e-02
-1.33267140e+00 -9.17169452e-02 -2.48153597e-01 4.72645536e-02
-1.25297084e-02 7.72633016e-01 -9.70938504e-01 4.51113641e-01
4.46733207e-01 1.24503553e+00 -5.91319859e-01 9.16530609e-01
-1.51253611e-01 7.52453625e-01 -5.26632547e-01 -4.87115420e-03
2.32158408e-01 -4.62534666e-01 5.72455108e-01 1.18593478e+00
-1.19813301e-01 -3.96804810e-01 -2.25071475e-01 8.70707393e-01
-4.52825785e-01 3.42889540e-02 -8.78017008e-01 -4.26599264e-01
8.44330907e-01 1.67246115e+00 -1.14406526e+00 -2.96352468e-02
-2.49188945e-01 8.26043427e-01 3.28482062e-01 3.02316129e-01
-4.33239102e-01 -8.95385623e-01 7.02291071e-01 3.84850144e-01
2.46420294e-01 -1.39112964e-01 -2.89302349e-01 -1.09187686e+00
-4.00527418e-01 -5.70991755e-01 4.11512464e-01 -1.13029070e-01
-1.54872561e+00 5.62359154e-01 -2.68852830e-01 -1.04183078e+00
2.53667794e-02 -6.55602038e-01 -9.28002656e-01 4.01798785e-01
-1.63624060e+00 -1.03203821e+00 -4.57279056e-01 2.77245045e-01
-1.13502607e-01 -3.68175715e-01 6.97349310e-01 4.58276629e-01
-8.67441654e-01 7.08494604e-01 4.79704410e-01 4.13923025e-01
7.16978550e-01 -1.43348062e+00 2.18131933e-02 5.65617383e-01
2.18337238e-01 9.40918505e-01 5.71419477e-01 -4.93633419e-01
-1.38886702e+00 -1.16953611e+00 7.25305855e-01 -2.67757595e-01
1.03639305e+00 -7.78465092e-01 -6.13884151e-01 5.72844803e-01
-1.03524134e-01 5.25457680e-01 1.08900106e+00 3.82527471e-01
-6.95994437e-01 -2.83514887e-01 -1.07754815e+00 5.79474986e-01
1.25799012e+00 -8.44219983e-01 -1.96377888e-01 1.92588031e-01
3.32500428e-01 4.59059596e-01 -1.31637585e+00 2.49864295e-01
5.59611857e-01 -8.64316106e-01 9.10723805e-01 -5.54587662e-01
3.12127292e-01 -1.12056814e-01 -3.26976255e-02 -1.37349403e+00
-2.20724463e-01 -7.94959784e-01 -1.77624166e-01 1.24786210e+00
4.20457006e-01 -1.10748506e+00 1.11184609e+00 4.01422471e-01
2.50118017e-01 -7.98247159e-01 -7.73748875e-01 -7.75233150e-01
-1.74425781e-01 1.92423090e-01 3.63307774e-01 1.28513777e+00
2.36500606e-01 5.97753882e-01 -1.81231827e-01 -9.75755230e-02
6.57532394e-01 2.31693998e-01 9.33714747e-01 -1.75101817e+00
-1.91787452e-01 -3.54638457e-01 -8.06100726e-01 -9.23575640e-01
6.08840406e-01 -1.32905293e+00 -2.71471620e-01 -1.14520562e+00
1.92010865e-01 -8.38427484e-01 -3.74943703e-01 2.60975361e-01
7.76185244e-02 4.23829794e-01 -5.37408702e-02 3.44231904e-01
-7.59756923e-01 8.20376813e-01 7.46169269e-01 -1.43675134e-01
-1.75112680e-01 -9.95991826e-02 -7.15883255e-01 4.39643621e-01
9.12504613e-01 -4.97056752e-01 -6.04655921e-01 2.89589502e-02
5.52738905e-01 -4.85239923e-01 4.81264114e-01 -7.63295472e-01
1.90533519e-01 2.66328424e-01 2.26669908e-01 4.19051349e-02
3.17854851e-01 -1.00193250e+00 4.28346135e-02 3.90601605e-01
-4.46978807e-01 -5.65354824e-02 -3.54959249e-01 1.27770400e+00
-1.93645939e-01 6.95929602e-02 8.19119394e-01 2.22704336e-01
-4.53559428e-01 3.45397562e-01 -1.10850751e-01 1.57654539e-01
1.09514236e+00 -4.60839391e-01 -3.08810025e-01 -4.27680731e-01
-6.53893828e-01 4.08734053e-01 5.09363055e-01 1.52454838e-01
6.00526094e-01 -1.21993113e+00 -4.18206215e-01 3.47904772e-01
2.49358892e-01 -3.53128105e-01 -1.02248471e-02 1.17380881e+00
-3.20585668e-01 2.15861827e-01 -3.65753502e-01 -8.02073956e-01
-1.19405162e+00 1.78619489e-01 2.57221907e-01 -1.86365664e-01
-5.95581174e-01 6.97731555e-01 7.64032528e-02 -7.44025588e-01
2.03698780e-02 4.98684496e-01 -2.83943951e-01 4.18580055e-01
-6.00400679e-02 7.10680008e-01 -1.93054885e-01 -6.59999311e-01
-2.81743079e-01 2.71496654e-01 -1.73425153e-01 1.73974484e-01
1.57574546e+00 -2.67334431e-01 -1.42071337e-01 3.31717193e-01
1.52193260e+00 -9.92841739e-03 -1.09160185e+00 -3.34841907e-01
2.86263347e-01 -5.48945606e-01 5.38212899e-03 -1.44384369e-01
-1.05117559e+00 1.01895809e+00 3.64190280e-01 5.20030499e-01
6.45285368e-01 1.17296398e-01 6.09440804e-01 4.25136179e-01
9.42187011e-02 -1.02833605e+00 3.57036054e-01 3.07815462e-01
2.62623549e-01 -1.28489971e+00 -3.62765603e-02 -5.27421415e-01
-4.45195913e-01 1.10795176e+00 5.12952149e-01 -3.26413155e-01
9.42835748e-01 -6.27244934e-02 -2.38548905e-01 -4.77534115e-01
-7.15062618e-01 -1.14459917e-01 -1.78729761e-02 6.59484684e-01
2.54422277e-01 4.80191596e-02 -8.21739063e-02 2.53812641e-01
7.72930980e-02 -6.42188728e-01 4.79354709e-01 5.88457584e-01
-2.13433057e-01 -1.05019283e+00 1.31217062e-01 8.25492799e-01
-1.42988056e-01 3.12549528e-03 -6.72118723e-01 8.13012719e-01
-2.62733310e-01 7.86210895e-01 7.96456113e-02 -6.51443481e-01
-6.18917979e-02 -1.40937809e-02 -7.22094029e-02 -7.24670708e-01
-5.64000189e-01 -1.28155515e-01 1.40565425e-01 -3.22329640e-01
-3.30086648e-01 -4.32939798e-01 -9.32578683e-01 -5.05850613e-01
-5.07592261e-01 4.13245022e-01 3.71187776e-01 5.20899415e-01
7.62829602e-01 2.69501328e-01 1.08168733e+00 -5.85287213e-01
-6.50245607e-01 -6.25369489e-01 -8.21784317e-01 6.16406381e-01
1.89850673e-01 -9.56957281e-01 -9.38556612e-01 -4.30310190e-01] | [7.02653169631958, 6.085538387298584] |
4f750c52-7d20-4f47-af00-c8f7f6a2d265 | improving-ecg-classification-interpretability | 2201.04070 | null | https://arxiv.org/abs/2201.04070v1 | https://arxiv.org/pdf/2201.04070v1.pdf | Improving ECG Classification Interpretability using Saliency Maps | Cardiovascular disease is a large worldwide healthcare issue; symptoms often present suddenly with minimal warning. The electrocardiogram (ECG) is a fast, simple and reliable method of evaluating the health of the heart, by measuring electrical activity recorded through electrodes placed on the skin. ECGs often need to be analyzed by a cardiologist, taking time which could be spent on improving patient care and outcomes. Because of this, automatic ECG classification systems using machine learning have been proposed, which can learn complex interactions between ECG features and use this to detect abnormalities. However, algorithms built for this purpose often fail to generalize well to unseen data, reporting initially impressive results which drop dramatically when applied to new environments. Additionally, machine learning algorithms suffer a "black-box" issue, in which it is difficult to determine how a decision has been made. This is vital for applications in healthcare, as clinicians need to be able to verify the process of evaluation in order to trust the algorithm. This paper proposes a method for visualizing model decisions across each class in the MIT-BIH arrhythmia dataset, using adapted saliency maps averaged across complete classes to determine what patterns are being learned. We do this by building two algorithms based on state-of-the-art models. This paper highlights how these maps can be used to find problems in the model which could be affecting generalizability and model performance. Comparing saliency maps across complete classes gives an overall impression of confounding variables or other biases in the model, unlike what would be highlighted when comparing saliency maps on an ECG-by-ECG basis. | ['Dr Jeff Dalton', 'Dr Fani Deligianni', 'Ms Yola Jones'] | 2022-01-10 | null | null | null | null | ['ecg-classification'] | ['medical'] | [ 5.20596743e-01 8.02751109e-02 2.66967565e-01 -4.52108920e-01
-4.76552755e-01 -5.91158330e-01 4.58530225e-02 7.41572559e-01
-2.77852476e-01 7.51541793e-01 5.78967715e-03 -3.34747106e-01
-2.46158555e-01 -4.48566049e-01 -3.72712880e-01 -6.69351995e-01
-4.11189020e-01 3.63803446e-01 2.26794511e-01 -1.45083433e-03
4.42397863e-01 4.58046287e-01 -1.39981592e+00 4.55109715e-01
6.83301270e-01 6.30581558e-01 1.46223336e-01 8.58837903e-01
2.68968552e-01 4.22316819e-01 -9.27152991e-01 -7.93141350e-02
-1.63334440e-02 -9.23605680e-01 -5.23668885e-01 -1.95088491e-01
1.79915845e-01 1.00771226e-01 3.31890672e-01 1.01149642e+00
9.58104253e-01 -3.91919196e-01 5.51350594e-01 -1.04762387e+00
-1.79728046e-02 2.82539457e-01 -3.09069157e-01 6.68271482e-01
4.24732596e-01 1.91335782e-01 6.12648845e-01 -5.58872521e-01
6.56691134e-01 8.19931686e-01 8.78304601e-01 3.50588858e-01
-1.42180395e+00 -4.42844868e-01 -1.43317163e-01 4.10470665e-01
-1.18778419e+00 -1.67629465e-01 8.36818993e-01 -5.62024355e-01
6.65429711e-01 8.46358299e-01 9.99699295e-01 8.19792628e-01
4.18662965e-01 2.92548805e-01 1.36113477e+00 -4.67690200e-01
4.48305041e-01 5.05534768e-01 -1.83221605e-02 3.07393432e-01
5.22747159e-01 -2.54886430e-02 -4.82091814e-01 -2.66448915e-01
7.07988143e-01 -2.81710196e-02 -5.06531179e-01 -4.76038128e-01
-1.33699620e+00 5.23362458e-01 5.04384816e-01 5.67492843e-01
-4.46840346e-01 -2.26011053e-01 4.41840440e-01 3.57570708e-01
2.57048726e-01 9.64814186e-01 -4.44200575e-01 -3.90563697e-01
-1.25750792e+00 1.83166072e-01 5.91759682e-01 1.28584251e-01
3.33390176e-01 -1.93759546e-01 -2.90068358e-01 6.29459977e-01
1.13317221e-01 3.03485543e-01 5.71471870e-01 -7.92596698e-01
-8.91218334e-02 7.14869976e-01 2.97200251e-02 -1.30113733e+00
-8.46733272e-01 -7.80315816e-01 -8.08397591e-01 6.96477234e-01
5.00564098e-01 -4.57962826e-02 -8.47602427e-01 1.42088842e+00
1.27991438e-01 9.37219113e-02 -1.32557809e-01 1.00951958e+00
6.46360397e-01 1.09929271e-01 2.45633587e-01 -2.13801086e-01
1.33583260e+00 -2.76674628e-01 -5.91718912e-01 -2.18805581e-01
8.07778716e-01 -4.83099371e-01 1.11741984e+00 6.55520618e-01
-9.73683178e-01 -4.88468915e-01 -1.30506241e+00 4.02473003e-01
-3.09142441e-01 1.14075625e-02 1.31567270e-01 8.46244037e-01
-9.80200350e-01 1.05183661e+00 -9.33663189e-01 -6.16986513e-01
6.05451107e-01 2.88711190e-01 -9.46090370e-02 2.83006042e-01
-1.14616358e+00 1.37309182e+00 5.28211035e-02 1.36043519e-01
-4.91752177e-01 -7.93571949e-01 -6.59945607e-01 5.33544421e-02
-5.74532039e-02 -6.33883178e-01 8.97810638e-01 -1.10689700e+00
-8.30753744e-01 1.02959776e+00 -1.97998896e-01 -4.83262390e-01
7.84566283e-01 2.57189311e-02 -5.17634094e-01 2.00545132e-01
2.16786727e-01 3.63795102e-01 6.86972380e-01 -1.17591703e+00
-4.39218462e-01 -5.95586777e-01 -2.52059162e-01 1.06987245e-01
8.10856000e-02 3.47071476e-02 1.35994866e-01 -6.08174920e-01
2.63448119e-01 -9.30784583e-01 -3.39813709e-01 7.63175562e-02
-2.51949608e-01 1.70568839e-01 6.83989108e-01 -8.86704445e-01
1.47141564e+00 -2.03928089e+00 -1.40008941e-01 4.39318538e-01
4.42606896e-01 2.87416697e-01 4.16090637e-01 1.04755156e-01
-2.48394042e-01 2.75146931e-01 -3.98985416e-01 2.32938126e-01
-3.79695415e-01 -1.01711594e-01 3.90326045e-02 4.35228288e-01
3.62505794e-01 7.72163987e-01 -8.49277794e-01 -4.16951299e-01
2.79164940e-01 5.30557692e-01 -1.19297832e-01 -3.31417918e-02
4.04393882e-01 7.15799749e-01 -1.46069676e-01 4.11439300e-01
2.67885834e-01 -3.99289221e-01 2.69447923e-01 -7.31549412e-02
1.65464938e-01 2.66950101e-01 -1.12491810e+00 1.54131150e+00
-3.80281471e-02 7.32952774e-01 -3.60302329e-01 -1.26483238e+00
1.02356839e+00 5.36362290e-01 3.56426239e-01 -7.09164202e-01
-1.76817738e-02 3.67054731e-01 5.85995674e-01 -6.10177398e-01
-1.93563759e-01 -3.94202560e-01 1.68430448e-01 2.18492270e-01
-3.59231561e-01 -2.08145622e-02 7.70187285e-03 1.46974223e-02
1.22305107e+00 -7.30961487e-02 3.56813222e-01 -4.65545446e-01
2.12249875e-01 1.86476842e-01 5.86995721e-01 7.18265235e-01
-3.58227879e-01 1.13290143e+00 7.04992175e-01 -7.50141919e-01
-8.88428807e-01 -1.01585543e+00 -4.67359453e-01 3.17495942e-01
1.25602037e-02 -2.79790878e-01 -8.28371882e-01 -7.29372084e-01
-6.79027438e-02 8.19708467e-01 -9.71016049e-01 -4.47373509e-01
-2.27148637e-01 -9.11433995e-01 4.83077705e-01 5.66429257e-01
9.86158252e-02 -1.30031800e+00 -1.83782935e+00 4.48978812e-01
-7.55649954e-02 -5.30630231e-01 1.64666265e-01 1.38264894e-01
-1.09897912e+00 -1.21560943e+00 -1.04214394e+00 -4.00050879e-01
6.09451294e-01 -9.24704596e-02 1.34249437e+00 2.76629776e-01
-8.41754794e-01 2.71333307e-01 -1.44496277e-01 -1.02344680e+00
-3.70974720e-01 -1.49947867e-01 -1.08653359e-01 -5.59062585e-02
4.46605206e-01 -5.33047855e-01 -1.01692331e+00 3.40579152e-01
-5.20954788e-01 4.35755551e-02 5.60121000e-01 7.17948914e-01
4.27517414e-01 -3.45200747e-01 8.87637615e-01 -1.09821665e+00
6.36163831e-01 -3.85013252e-01 -7.63004422e-02 2.66248018e-01
-9.00097370e-01 -1.81799665e-01 1.71557054e-01 -1.90589935e-01
-4.26060885e-01 -2.54549943e-02 1.77986383e-01 -2.99509972e-01
-2.56532252e-01 4.95636433e-01 1.63780153e-01 1.25440612e-01
1.11992037e+00 -4.57174703e-02 9.91213024e-02 -3.19021285e-01
-3.71566713e-01 5.49243271e-01 4.39658135e-01 5.07757105e-02
3.79627615e-01 3.25950414e-01 1.35815039e-01 -7.06615031e-01
-6.12272918e-01 -4.07792896e-01 -7.60493994e-01 -4.51890647e-01
8.36522937e-01 -4.74059075e-01 -2.52576202e-01 9.39996690e-02
-8.04770052e-01 -1.34566143e-01 -4.13501143e-01 3.64139169e-01
-3.23756933e-01 1.90410584e-01 -4.75127101e-02 -8.24403346e-01
-2.69292355e-01 -9.34841633e-01 8.08638930e-01 3.19959372e-01
-9.86815453e-01 -1.07418013e+00 2.01685458e-01 5.86711802e-02
6.36891186e-01 6.56994641e-01 9.35657620e-01 -6.21665299e-01
2.66322233e-02 -2.48288766e-01 -2.12795194e-03 1.64191559e-01
3.42410266e-01 -8.79601613e-02 -1.27122474e+00 -2.89658278e-01
1.27606764e-01 1.46743536e-01 5.30135453e-01 6.55922055e-01
1.11180305e+00 5.72610833e-02 -4.66508240e-01 1.99140772e-01
1.09598267e+00 4.24892515e-01 8.26188803e-01 4.32505190e-01
4.14230257e-01 6.69268310e-01 5.37631929e-01 6.42523542e-02
1.47692636e-01 6.24501526e-01 2.09114075e-01 -6.65255904e-01
-8.42647552e-02 1.01329997e-01 -1.55390073e-02 3.47031534e-01
-2.44171038e-01 4.18312550e-01 -1.21430659e+00 5.87944210e-01
-1.72368038e+00 -8.72399747e-01 -4.68053043e-01 2.68512774e+00
6.80385053e-01 4.05985892e-01 2.04683676e-01 6.65160179e-01
5.70690811e-01 -3.11857194e-01 -6.02263033e-01 -5.94709814e-01
-6.41664341e-02 3.42515409e-01 1.73708618e-01 1.63494840e-01
-8.64007413e-01 5.88913821e-02 6.43879652e+00 -4.65583615e-02
-1.71137750e+00 -1.13285724e-02 1.00949466e+00 -1.37977287e-01
-6.07937984e-02 1.46169262e-02 1.63192358e-02 5.57309270e-01
1.00563896e+00 -2.01705724e-01 -6.77980408e-02 5.11905551e-01
6.03113770e-01 -4.14397568e-01 -1.21234012e+00 1.17199826e+00
1.56257659e-01 -1.19095421e+00 -3.87427837e-01 -2.88725197e-01
3.21084350e-01 -2.43106276e-01 -6.76834285e-02 -1.50552973e-01
-4.98487175e-01 -1.23101401e+00 5.06432950e-01 7.92631030e-01
8.03252697e-01 -4.41974431e-01 8.61142099e-01 1.37084723e-01
-6.94705009e-01 4.26187254e-02 -9.16475058e-02 -2.93223977e-01
-3.74790542e-02 6.61538064e-01 -1.07537448e+00 1.73061058e-01
1.12211311e+00 5.80190063e-01 -1.07921290e+00 1.58170676e+00
-1.48493320e-01 8.24266076e-01 -1.92805141e-01 3.30796614e-02
-1.88265041e-01 1.97988316e-01 7.44724929e-01 1.24717855e+00
5.19347966e-01 -1.65977105e-01 -1.03255786e-01 9.15575981e-01
6.01447940e-01 2.68772721e-01 -7.54842401e-01 3.19976747e-01
7.96651989e-02 1.25193703e+00 -1.18624985e+00 -5.53507030e-01
-1.53632939e-01 9.66082871e-01 -2.54153997e-01 2.38955230e-01
-6.43537045e-01 -5.37890851e-01 2.38587007e-01 6.57342851e-01
-1.55100018e-01 3.95128429e-01 -9.10687268e-01 -7.74519801e-01
2.23646283e-01 -9.22271430e-01 4.28719789e-01 -1.00269723e+00
-8.66264224e-01 4.86387104e-01 -1.05543464e-01 -1.35692906e+00
-3.12701821e-01 -3.10102671e-01 -6.75716460e-01 1.07217109e+00
-1.09683752e+00 -4.41029876e-01 -3.88714999e-01 2.78623670e-01
2.21633270e-01 2.64619201e-01 1.09796238e+00 2.75087297e-01
-2.20187113e-01 4.26508248e-01 -3.01624596e-01 -2.22236104e-02
8.43765259e-01 -1.59632778e+00 2.10110664e-01 5.74462891e-01
4.55450684e-01 4.20877457e-01 9.58414495e-01 -5.96708238e-01
-6.12824798e-01 -6.81602001e-01 1.01530063e+00 -8.20153534e-01
1.99871864e-02 -1.05826095e-01 -1.32780635e+00 1.07799798e-01
-1.42152933e-02 7.99525827e-02 7.36132085e-01 1.96150720e-01
7.74767399e-02 -1.13737695e-01 -1.29526699e+00 5.13045073e-01
5.88279068e-01 -2.95048058e-01 -6.95384681e-01 2.38880352e-03
-2.41406560e-01 -2.58862764e-01 -7.37482309e-01 6.91627800e-01
8.18207681e-01 -1.23044062e+00 6.67741299e-01 -6.20332658e-01
1.92155153e-01 -4.48405385e-01 5.04504919e-01 -1.53940845e+00
-3.68062496e-01 -4.10366505e-01 2.89232403e-01 7.26173937e-01
8.41559410e-01 -6.95050538e-01 6.69706881e-01 6.62891388e-01
1.64630026e-01 -9.33316350e-01 -8.47673357e-01 -3.27227592e-01
-2.22211793e-01 -3.85564804e-01 2.34641150e-01 1.18992746e+00
3.52203578e-01 3.09487700e-01 -3.51468660e-02 1.50186315e-01
4.75315481e-01 -1.32502154e-01 3.70144874e-01 -1.65665889e+00
-1.82810217e-01 -4.83790129e-01 -1.00322044e+00 1.27651632e-01
-7.92383730e-01 -7.36130536e-01 -2.05543369e-01 -1.60051048e+00
2.74723530e-01 -5.30880034e-01 -8.04307997e-01 4.34783041e-01
-5.17683148e-01 5.59071362e-01 2.30212554e-01 1.41917214e-01
-2.55261838e-01 -2.26884112e-01 1.03966165e+00 -4.92315069e-02
-4.35286701e-01 1.42780155e-01 -7.98316061e-01 8.27616155e-01
1.02527893e+00 -7.42201269e-01 -4.49706435e-01 1.83940567e-02
2.69021124e-01 -1.25173792e-01 6.89733446e-01 -1.45216215e+00
6.10716753e-02 3.94290537e-01 9.48572695e-01 -2.13091537e-01
7.86775649e-02 -6.80274546e-01 4.37260211e-01 8.47932875e-01
-3.90244603e-01 4.40685183e-01 3.13002706e-01 3.29790950e-01
-1.78500816e-01 -1.49749845e-01 7.97031283e-01 -3.06713551e-01
-2.79912114e-01 -3.17308515e-01 -3.28210920e-01 1.36467203e-01
1.02352691e+00 -5.09846568e-01 1.68734968e-01 -4.30829376e-01
-1.27656806e+00 -4.95488830e-02 5.31876445e-01 3.86691749e-01
5.30305386e-01 -1.13451254e+00 -8.60053420e-01 3.11732352e-01
2.25666732e-01 -2.65388370e-01 3.48317951e-01 1.25008500e+00
-7.28415191e-01 4.51174155e-02 -3.37539345e-01 -1.03172362e+00
-1.50613308e+00 4.02170390e-01 5.51673770e-01 -1.99305072e-01
-8.28616977e-01 5.30339777e-01 -1.19611472e-02 1.36738092e-01
1.56640708e-01 -4.77202356e-01 -3.23236853e-01 2.89608091e-01
6.70945585e-01 3.05498481e-01 3.93736959e-01 -2.91374624e-01
-4.66808528e-01 6.26647174e-01 7.55562931e-02 -9.48686823e-02
1.07674837e+00 6.56846073e-03 1.23956032e-01 9.14745033e-01
7.22237706e-01 -2.47886702e-01 -9.53138888e-01 2.99960881e-01
2.28703469e-01 -4.56928045e-01 -2.52815541e-02 -1.34039092e+00
-8.23007822e-01 1.37412691e+00 1.34320116e+00 5.26815295e-01
1.21076834e+00 -1.73664600e-01 1.52312502e-01 -4.79919389e-02
2.24210054e-01 -1.18314397e+00 -2.81392962e-01 -3.65261555e-01
9.01442289e-01 -1.16122520e+00 -4.95572342e-03 -1.24005027e-01
-9.41635907e-01 1.02428877e+00 9.31182504e-02 -5.91240898e-02
7.57780254e-01 1.11265846e-01 6.41352296e-01 -3.83620471e-01
-5.27942955e-01 1.46231025e-01 4.02583539e-01 6.67164862e-01
5.39909601e-01 1.43599093e-01 -5.20896256e-01 4.68119502e-01
-8.49169716e-02 1.98201254e-01 4.75892484e-01 1.03808057e+00
-3.02006423e-01 -9.77684438e-01 -3.93332183e-01 7.76655495e-01
-7.39950120e-01 8.59175250e-02 -4.98384416e-01 5.61084151e-01
1.28377795e-01 7.38521278e-01 -1.35323033e-01 -2.57432252e-01
4.80614305e-01 4.54489201e-01 3.18300992e-01 -7.03515172e-01
-6.87133312e-01 -2.98403539e-02 5.89033775e-02 -4.31491375e-01
-2.86928624e-01 -8.01273823e-01 -1.18420553e+00 4.91569579e-01
-9.88774449e-02 1.92162052e-01 7.17325151e-01 7.53418624e-01
5.15782893e-01 8.94800901e-01 3.63115251e-01 -7.45487392e-01
-3.62388760e-01 -9.57134247e-01 -5.83109617e-01 6.84561193e-01
3.69646460e-01 -5.72593808e-01 -3.00441504e-01 2.04412952e-01] | [14.295592308044434, 3.2970781326293945] |
3befba5a-0058-442e-8ff6-8894f835f970 | towards-personalized-and-semantic-retrieval | 2006.02282 | null | https://arxiv.org/abs/2006.02282v3 | https://arxiv.org/pdf/2006.02282v3.pdf | Towards Personalized and Semantic Retrieval: An End-to-End Solution for E-commerce Search via Embedding Learning | Nowadays e-commerce search has become an integral part of many people's shopping routines. Two critical challenges stay in today's e-commerce search: how to retrieve items that are semantically relevant but not exact matching to query terms, and how to retrieve items that are more personalized to different users for the same search query. In this paper, we present a novel approach called DPSR, which stands for Deep Personalized and Semantic Retrieval, to tackle this problem. Explicitly, we share our design decisions on how to architect a retrieval system so as to serve industry-scale traffic efficiently and how to train a model so as to learn query and item semantics accurately. Based on offline evaluations and online A/B test with live traffics, we show that DPSR model outperforms existing models, and DPSR system can retrieve more personalized and semantically relevant items to significantly improve users' search experience by +1.29% conversion rate, especially for long tail queries by +10.03%. As a result, our DPSR system has been successfully deployed into JD.com's search production since 2019. | ['Wen-Yun Yang', 'Songlin Wang', 'Kang Zhang', 'Han Zhang', 'Yunjiang Jiang', 'Zhiling Tang', 'Yun Xiao', 'Weipeng Yan'] | 2020-06-03 | null | null | null | null | ['semantic-retrieval'] | ['natural-language-processing'] | [-3.40156376e-01 -8.18729520e-01 -4.65145856e-01 -4.30675417e-01
-1.01699221e+00 -6.97797596e-01 1.53072163e-01 9.53569189e-02
-4.11437273e-01 2.40850255e-01 7.40582496e-02 -1.47997096e-01
-4.75140005e-01 -1.00451076e+00 -4.47059780e-01 -8.53131637e-02
9.37135965e-02 1.09534872e+00 6.61706328e-01 -7.74081528e-01
3.71666998e-01 3.58890951e-01 -1.45490909e+00 3.24313849e-01
7.28542984e-01 1.52034736e+00 5.58948696e-01 4.05408055e-01
-4.68694270e-01 3.29651624e-01 -3.70737702e-01 -6.08914971e-01
3.95501107e-01 -9.04508773e-03 -1.04990768e+00 -3.65967631e-01
2.38827795e-01 -5.21501839e-01 -4.03334469e-01 9.05099392e-01
6.40922427e-01 3.49149555e-01 3.87572885e-01 -1.04392576e+00
-1.03692198e+00 6.06983781e-01 -3.77173841e-01 5.73004901e-01
3.62840384e-01 -1.24420002e-01 1.47663558e+00 -6.74245894e-01
5.05219221e-01 1.08048904e+00 2.18902901e-01 2.51817852e-01
-9.99416173e-01 -6.94295824e-01 8.50381255e-02 5.12058020e-01
-1.46680677e+00 -4.27588671e-01 5.73879838e-01 2.41224580e-02
9.32422101e-01 5.49759269e-01 3.37526888e-01 8.48478138e-01
4.44593169e-02 1.00549889e+00 3.40133935e-01 -2.12744012e-01
1.61836669e-01 2.83976376e-01 4.04683262e-01 1.67826384e-01
2.14668080e-01 -1.06237173e-01 -5.00877082e-01 -2.25684777e-01
5.65004826e-01 3.11712921e-01 3.45557958e-01 -2.62050867e-01
-8.60018611e-01 8.54031265e-01 4.57512587e-01 3.75733584e-01
-6.22637749e-01 2.74463803e-01 4.72660720e-01 4.60991055e-01
6.46330118e-02 5.90753019e-01 -5.74351728e-01 -1.68233663e-01
-5.35417855e-01 6.00518525e-01 9.71200824e-01 1.53734899e+00
6.02092564e-01 -4.79171842e-01 -4.39924955e-01 1.28614628e+00
5.25837898e-01 7.99099624e-01 6.99839354e-01 -9.78767395e-01
3.65047961e-01 4.43640471e-01 2.56189883e-01 -1.12646151e+00
6.05234541e-02 -6.90049827e-01 -4.77665573e-01 -8.61830413e-01
-9.69958771e-03 3.65671694e-01 -6.64905787e-01 1.36001587e+00
-1.00148864e-01 -3.13494384e-01 -2.00236633e-01 1.07593846e+00
5.70216894e-01 7.68102825e-01 4.40037251e-01 1.17535569e-01
1.67174852e+00 -1.07731080e+00 -3.76347959e-01 -2.91406453e-01
5.66099405e-01 -1.09966183e+00 1.29475653e+00 2.75254011e-01
-1.17216372e+00 -6.11794829e-01 -5.64950705e-01 -1.73642352e-01
-5.69177032e-01 -1.44982010e-01 7.66442955e-01 7.52663076e-01
-8.64774108e-01 2.77128130e-01 -2.12142646e-01 -6.28994942e-01
1.07690804e-01 4.99943078e-01 2.46226922e-01 -2.10001007e-01
-1.67338657e+00 4.81193185e-01 2.31535271e-01 -3.87996644e-01
-4.87491637e-01 -5.73010683e-01 -2.20186010e-01 5.96187413e-01
5.57316124e-01 -8.62353623e-01 1.67763150e+00 -4.30136532e-01
-1.17072082e+00 7.69453287e-01 -1.68889001e-01 -4.76423979e-01
9.17195156e-02 -3.09999883e-01 -1.01968825e+00 9.19367671e-02
3.38797092e-01 4.80976611e-01 3.69372368e-01 -1.07165980e+00
-1.22962058e+00 -5.34491181e-01 1.53504178e-01 3.82373363e-01
-3.88687849e-01 3.76536876e-01 -1.34002447e+00 -6.22093678e-01
1.22755365e-02 -7.64325261e-01 -1.09898567e-01 -2.91883528e-01
3.48405242e-02 -4.53614116e-01 6.53390169e-01 -6.69473588e-01
1.45616508e+00 -2.00099087e+00 -3.07719529e-01 6.32893026e-01
4.85735871e-02 3.28916460e-01 -2.14353964e-01 6.34615242e-01
5.67683101e-01 9.97076482e-02 5.65834343e-01 1.68654576e-01
6.72831774e-01 9.23020989e-02 -5.57933450e-01 -2.81914562e-01
-4.58447635e-01 1.35386002e+00 -8.09725285e-01 -5.02278924e-01
-1.03696689e-01 1.05744265e-01 -6.36944056e-01 6.40174374e-02
-3.82887810e-01 6.46857545e-04 -9.94514108e-01 9.67926323e-01
4.29758251e-01 -5.99912226e-01 6.32483214e-02 -2.54274011e-01
3.49329740e-01 3.39894742e-01 -8.50062728e-01 1.69414842e+00
-7.64104187e-01 1.18754983e-01 6.26845360e-02 -6.91854477e-01
9.15184915e-01 -6.55129403e-02 6.77895069e-01 -1.84498596e+00
6.10519089e-02 6.30173564e-01 -5.24180174e-01 -4.45505351e-01
9.95592892e-01 1.85402289e-01 -3.26921195e-01 4.61246401e-01
-2.72919983e-01 2.78700739e-01 3.46396804e-01 1.87587067e-01
9.54584599e-01 -3.26601982e-01 -1.58968687e-01 -2.67795026e-01
6.60613239e-01 8.46728757e-02 1.38685986e-01 1.01912892e+00
-1.00632645e-01 1.50960073e-01 -2.22964138e-01 -4.47963536e-01
-9.30572033e-01 -1.07406664e+00 -1.86323039e-02 1.82454216e+00
9.08396900e-01 -2.18722269e-01 -3.32760662e-01 -6.18869841e-01
3.17409068e-01 8.20899785e-01 2.29984358e-01 -3.98272812e-01
-6.75805151e-01 -2.55486906e-01 1.22463122e-01 4.32493120e-01
6.85047984e-01 -1.17586505e+00 -2.56889433e-01 4.37597901e-01
-5.19791722e-01 -9.79002833e-01 -1.08914685e+00 -1.26649484e-01
-6.01970255e-01 -7.07962155e-01 -8.96291792e-01 -1.06547761e+00
5.38836122e-02 8.70828986e-01 1.41217935e+00 1.19603254e-01
-3.24692369e-01 4.78738070e-01 -5.91631651e-01 -5.53236669e-03
6.04587719e-02 5.51201761e-01 -1.26236930e-01 -2.76967227e-01
9.51773763e-01 -3.57856005e-01 -1.20222175e+00 1.08768475e+00
-1.03956175e+00 -6.41099930e-01 7.37476230e-01 5.68444610e-01
6.40404344e-01 3.05812031e-01 9.23837602e-01 -8.29379022e-01
1.01410985e+00 -9.39256132e-01 -5.87787867e-01 6.54167950e-01
-1.25787556e+00 3.61378379e-02 4.46095914e-01 -2.53255427e-01
-8.10318112e-01 -5.68020880e-01 -4.92832005e-01 -2.72908539e-01
1.62737712e-01 3.84572476e-01 6.84114918e-02 4.39713486e-02
4.70694900e-01 5.21828711e-01 -4.23800349e-01 -9.46186244e-01
4.76959437e-01 1.09137845e+00 3.74707997e-01 -8.04233193e-01
4.87023711e-01 6.85265884e-02 -4.64768559e-01 -3.71443599e-01
-5.10280490e-01 -1.32062972e+00 -1.02234751e-01 8.02494511e-02
4.11455870e-01 -7.39320636e-01 -9.85969365e-01 8.31077620e-02
-7.64390290e-01 6.44547418e-02 -2.75340766e-01 1.92679465e-01
-5.34330964e-01 3.86902303e-01 -5.77987671e-01 -4.71891522e-01
-8.80093157e-01 -1.02847385e+00 1.31241775e+00 2.95223951e-01
-3.50455008e-02 -6.29648983e-01 -1.11580081e-01 7.98940182e-01
9.66603398e-01 -9.13942933e-01 9.85072434e-01 -1.12610066e+00
-1.01899433e+00 -5.41557610e-01 -6.36284888e-01 1.39248148e-01
7.19344839e-02 -6.90221965e-01 -4.14312690e-01 -4.04714674e-01
-3.35288525e-01 6.92206947e-03 5.83558977e-01 -3.34382467e-02
1.26734090e+00 -3.29875916e-01 -5.26378572e-01 4.62187290e-01
1.37342894e+00 7.51686692e-01 6.05848968e-01 5.45801640e-01
1.54992834e-01 4.21467483e-01 8.58614683e-01 2.69786060e-01
5.71974695e-01 1.18593729e+00 1.93659753e-01 3.73832554e-01
-1.66840240e-01 -6.57163441e-01 -7.86453485e-02 8.49412620e-01
4.51807648e-01 -6.65436327e-01 -5.97956300e-01 5.81322670e-01
-1.87120259e+00 -8.32960248e-01 2.21184492e-01 2.14494014e+00
5.10687351e-01 8.03052709e-02 2.80351937e-01 -3.58378947e-01
6.65627420e-01 -2.12253645e-01 -7.21990108e-01 -1.98660493e-01
1.47069260e-01 3.47339213e-01 7.13560224e-01 5.96849956e-02
-7.33273625e-01 1.04113853e+00 6.38786888e+00 1.42561865e+00
-7.39460409e-01 9.48211029e-02 5.35864830e-01 1.94459319e-01
-5.30891836e-01 -4.05779220e-02 -1.08146954e+00 8.03619087e-01
8.71256173e-01 -5.65046132e-01 7.90209651e-01 1.08747339e+00
-1.25429422e-01 2.31888488e-01 -8.05691540e-01 1.23763311e+00
-4.18478847e-02 -1.20203376e+00 3.02162617e-01 1.38104185e-01
4.94849563e-01 -1.72014192e-01 2.44617015e-01 8.97291243e-01
3.98595184e-01 -7.38137662e-01 4.52155650e-01 5.86640418e-01
4.60927814e-01 -6.55277371e-01 7.15691566e-01 3.29626590e-01
-1.26448452e+00 -2.46860087e-01 -4.31047797e-01 6.02533638e-01
3.54708225e-01 2.41340116e-01 -5.87754250e-01 5.85076272e-01
1.06118870e+00 4.01705623e-01 -3.39892298e-01 1.32477760e+00
5.57116270e-01 2.73603022e-01 -3.95654142e-01 -3.57679695e-01
2.47562677e-01 -1.02062389e-01 3.09997350e-01 1.05119085e+00
6.23031795e-01 2.00584764e-03 1.33385941e-01 7.53623128e-01
-4.50198054e-01 5.04573047e-01 -1.91695973e-01 -1.26021653e-01
6.66964948e-01 1.15275824e+00 -5.11910021e-01 -2.65103817e-01
-3.01198751e-01 1.16633248e+00 -1.16211869e-01 4.13259864e-01
-8.18748057e-01 -6.11500084e-01 6.66488290e-01 3.87418121e-01
4.28837001e-01 -3.46474769e-03 2.94482350e-01 -8.67946684e-01
2.37365171e-01 -8.15988898e-01 7.04404652e-01 -7.66933203e-01
-1.68073785e+00 4.64382112e-01 -1.61035553e-01 -1.08745193e+00
-2.92714864e-01 -2.82465547e-01 -1.55789107e-01 1.05715215e+00
-1.66637218e+00 -9.85420823e-01 -1.41403198e-01 5.53895593e-01
7.95552492e-01 -4.10582364e-01 6.27261937e-01 1.09155774e+00
-6.17460310e-02 7.57760108e-01 4.29055989e-01 -2.38899976e-01
6.90401137e-01 -8.27926159e-01 5.48849106e-01 1.92049667e-01
2.37898245e-01 8.71581912e-01 3.86355221e-01 -4.81075734e-01
-1.73651922e+00 -9.44473863e-01 1.11983025e+00 -3.78972501e-01
6.51827812e-01 -3.04521304e-02 -6.82186663e-01 5.11857212e-01
-1.07004441e-01 -3.33689690e-01 6.37974918e-01 2.66967297e-01
-4.25993770e-01 -7.39334226e-01 -1.16163385e+00 4.85251427e-01
1.37151349e+00 -4.74939227e-01 -3.80432755e-01 5.10066032e-01
1.00699031e+00 -8.23819861e-02 -9.59854543e-01 3.85670811e-01
8.12503755e-01 -7.65149653e-01 1.50027275e+00 -6.30390286e-01
-3.00957590e-01 -7.39328042e-02 -5.18457174e-01 -8.12607348e-01
-5.43495119e-01 -5.33040643e-01 -2.48622298e-01 1.06093025e+00
3.35338891e-01 -7.46431351e-01 8.43869865e-01 6.26625657e-01
1.47667065e-01 -6.26165330e-01 -6.17407978e-01 -1.00263023e+00
-1.34111375e-01 -2.46620521e-01 1.12330127e+00 5.14961243e-01
-3.41751277e-01 3.54251862e-01 -1.84518948e-01 -8.57149214e-02
4.70051408e-01 4.42103148e-01 6.19638205e-01 -1.16486609e+00
-4.18332636e-01 -7.74462342e-01 -2.30045184e-01 -2.15456128e+00
-2.52259851e-01 -9.36912596e-01 -2.10935697e-01 -1.42659783e+00
3.76031965e-01 -9.00307536e-01 -7.03991473e-01 1.76537380e-01
2.65082210e-01 1.77882060e-01 8.68463963e-02 6.80863857e-01
-1.11586273e+00 4.23622489e-01 1.16442490e+00 -2.93586195e-01
-1.11456186e-01 4.90403295e-01 -1.25222719e+00 -2.12713569e-01
7.51577020e-01 -1.50221199e-01 -6.29518092e-01 -6.50866210e-01
5.29388011e-01 1.22949615e-01 1.24792442e-01 -5.61660051e-01
5.85354567e-01 -2.10502729e-01 -2.05049254e-02 -7.87372530e-01
1.78211927e-01 -1.21776628e+00 2.32968166e-01 2.14901492e-01
-5.09269595e-01 2.30890021e-01 -7.89044201e-02 6.04494810e-01
-2.90558606e-01 -3.26428115e-01 2.50610232e-01 -9.05629843e-02
-1.18230486e+00 6.04933977e-01 -3.32600577e-03 5.94963543e-02
1.02955151e+00 3.66061255e-02 -3.62910420e-01 -6.09509468e-01
-6.43921793e-01 7.69176662e-01 4.48497571e-02 9.06393707e-01
3.94739360e-01 -1.53978205e+00 -3.45267713e-01 1.07666500e-01
4.50651228e-01 -5.48492491e-01 4.21937764e-01 3.21808547e-01
-6.50993764e-01 1.16036379e+00 9.91861671e-02 -3.97653520e-01
-8.40999424e-01 7.29718745e-01 3.85369621e-02 -4.32531327e-01
-3.37640226e-01 7.09694386e-01 -1.25676561e-02 -2.89605469e-01
4.12301153e-01 -7.49853179e-02 -7.31671676e-02 -2.94617683e-01
4.50273722e-01 5.17615438e-01 2.21413255e-01 -5.11033177e-01
-2.19591230e-01 6.44928694e-01 -4.45191145e-01 1.34224400e-01
9.86622632e-01 -6.61755264e-01 1.54156491e-01 -1.30986914e-01
1.37177420e+00 -4.51932065e-05 -5.11462331e-01 -7.24616110e-01
1.87491611e-01 -8.03861022e-01 7.75065720e-02 -1.18540204e+00
-1.26459587e+00 4.22910362e-01 6.97574794e-01 5.99333048e-01
1.27708972e+00 2.67906576e-01 1.68708944e+00 5.84356785e-01
9.40092206e-01 -1.27819610e+00 -4.41059992e-02 3.76158744e-01
5.73966086e-01 -1.09295416e+00 -3.74761611e-01 -3.10194999e-01
-6.06705189e-01 8.15854728e-01 3.71725202e-01 2.05269083e-01
6.50732338e-01 -5.44355929e-01 3.04823909e-02 -3.99930179e-01
-6.43435419e-01 -4.19884831e-01 5.74897647e-01 2.45081484e-01
1.36946097e-01 -1.96318366e-02 -3.52670789e-01 7.10636854e-01
-6.75440058e-02 1.21334434e-01 -6.23103678e-01 8.03185701e-01
-7.72797704e-01 -1.49797130e+00 7.59623572e-02 6.44630253e-01
-3.67847264e-01 -8.90239328e-02 -1.10743325e-02 4.46702152e-01
-3.17011923e-01 1.06344008e+00 -2.90838890e-02 -6.20234728e-01
6.79947376e-01 1.41986936e-01 4.33269963e-02 -4.04366076e-01
-5.67810535e-01 5.41955270e-02 7.31836408e-02 -7.38807738e-01
1.40134826e-01 -2.59599417e-01 -1.08285642e+00 -6.07231617e-01
-4.03220028e-01 5.80302417e-01 7.99508750e-01 4.84886438e-01
8.36080730e-01 1.79943904e-01 8.00565839e-01 -1.10529751e-01
-9.24428642e-01 -5.92178226e-01 -7.51598716e-01 6.11649573e-01
-2.78553814e-01 -5.11401713e-01 7.30033591e-02 -2.43299678e-01] | [11.281376838684082, 7.3926100730896] |
033b5869-6929-4ab4-8360-a616951b3ca1 | crop-classification-under-varying-cloud-cover | 2012.02542 | null | https://arxiv.org/abs/2012.02542v2 | https://arxiv.org/pdf/2012.02542v2.pdf | Crop Classification under Varying Cloud Cover with Neural Ordinary Differential Equations | Optical satellite sensors cannot see the Earth's surface through clouds. Despite the periodic revisit cycle, image sequences acquired by Earth observation satellites are therefore irregularly sampled in time. State-of-the-art methods for crop classification (and other time series analysis tasks) rely on techniques that implicitly assume regular temporal spacing between observations, such as recurrent neural networks (RNNs). We propose to use neural ordinary differential equations (NODEs) in combination with RNNs to classify crop types in irregularly spaced image sequences. The resulting ODE-RNN models consist of two steps: an update step, where a recurrent unit assimilates new input data into the model's hidden state; and a prediction step, in which NODE propagates the hidden state until the next observation arrives. The prediction step is based on a continuous representation of the latent dynamics, which has several advantages. At the conceptual level, it is a more natural way to describe the mechanisms that govern the phenological cycle. From a practical point of view, it makes it possible to sample the system state at arbitrary points in time, such that one can integrate observations whenever they are available, and extrapolate beyond the last observation. Our experiments show that ODE-RNN indeed improves classification accuracy over common baselines such as LSTM, GRU, and temporal convolution. The gains are most prominent in the challenging scenario where only few observations are available (i.e., frequent cloud cover). Moreover, we show that the ability to extrapolate translates to better classification performance early in the season, which is important for forecasting. | ['Konrad Schindler', 'Jan Dirk Wegner', "Stefano D'Aronco", 'Mehmet Ozgur Turkoglu', 'Nando Metzger'] | 2020-12-04 | null | null | null | null | ['crop-classification'] | ['miscellaneous'] | [ 2.08016858e-01 -5.47796965e-01 -2.96621919e-01 2.19825599e-02
1.08750202e-01 -7.09296823e-01 6.27287388e-01 2.12486580e-01
-2.30463967e-01 6.66592538e-01 -2.24928617e-01 -6.36047959e-01
8.99402276e-02 -1.01722252e+00 -7.25755930e-01 -1.09092927e+00
-3.12853038e-01 2.16995925e-02 -1.01954348e-01 -3.28795224e-01
-2.94520080e-01 7.79295802e-01 -1.65388131e+00 -1.03095531e-01
1.01980376e+00 9.95883465e-01 6.00833535e-01 7.59052217e-01
-1.65664226e-01 8.93527150e-01 -2.09510893e-01 2.98265964e-01
4.95469086e-02 -4.22202170e-01 -3.91093969e-01 3.80112886e-01
-2.31357843e-01 -2.07217395e-01 -2.42981195e-01 9.45335388e-01
-2.10720628e-01 2.52942353e-01 5.56587815e-01 -9.61791873e-01
-5.57928324e-01 3.05546671e-01 -5.39153993e-01 1.08717799e-01
-2.70037800e-01 3.08111459e-01 8.03503156e-01 -5.40789187e-01
4.14782196e-01 8.55749726e-01 8.07919204e-01 2.35472098e-01
-1.38670206e+00 -4.31123555e-01 7.36342788e-01 -1.08444141e-02
-1.10123837e+00 -1.01844743e-01 6.42430902e-01 -6.59023106e-01
8.79755974e-01 2.24735782e-01 9.81561065e-01 9.69341338e-01
3.48087609e-01 7.97749281e-01 1.07277596e+00 -2.63026655e-01
2.43315116e-01 -1.48261994e-01 1.55010492e-01 2.45038748e-01
1.18119135e-01 3.26150149e-01 -9.35354307e-02 6.57296404e-02
9.73558605e-01 5.46403825e-01 -4.36608583e-01 -1.51892170e-01
-1.06055164e+00 8.68364096e-01 8.27794969e-01 4.14272487e-01
-1.04221213e+00 2.02411994e-01 2.54165500e-01 4.78273779e-01
9.24299181e-01 3.46191347e-01 -6.06331706e-01 2.50211716e-01
-1.35541618e+00 1.51045576e-01 6.33974135e-01 4.82033670e-01
7.70878255e-01 4.32410657e-01 1.45223975e-01 5.82079828e-01
1.54355779e-01 8.37066054e-01 5.80831230e-01 -7.41481602e-01
1.75581872e-02 3.68571937e-01 3.95940602e-01 -8.52650583e-01
-2.62366295e-01 -7.95142949e-01 -1.40621150e+00 7.55961463e-02
2.11544842e-01 -4.10983801e-01 -1.17508495e+00 1.78629756e+00
6.60979971e-02 5.74351788e-01 3.84009391e-01 7.54051387e-01
2.17491895e-01 1.29449153e+00 6.25659525e-02 -6.36207640e-01
1.24073899e+00 -7.21354246e-01 -8.23283613e-01 -3.93790185e-01
7.18297780e-01 -4.12761331e-01 5.62691331e-01 7.38935247e-02
-6.16147578e-01 -4.86293405e-01 -7.96500504e-01 3.72499913e-01
-7.43292689e-01 4.16229248e-01 8.16735506e-01 -4.30108123e-02
-1.12231898e+00 8.93432796e-01 -1.40062177e+00 -3.27000022e-01
-6.04475848e-02 5.51858023e-02 -3.70403081e-02 2.46894762e-01
-1.26171625e+00 7.32236564e-01 4.13281679e-01 8.78605306e-01
-8.25376809e-01 -6.39697254e-01 -9.15676951e-01 2.54739851e-01
2.77821362e-01 -3.78455222e-01 1.23035240e+00 -1.27807105e+00
-1.46659029e+00 4.33622688e-01 -6.44498825e-01 -8.92343760e-01
2.76483268e-01 -1.30781397e-01 -3.80589634e-01 -2.25842655e-01
-2.65551686e-01 4.37517613e-01 9.64229405e-01 -1.09002066e+00
-6.32501662e-01 -3.18311483e-01 -1.09025195e-01 9.60990489e-02
-1.72283798e-01 -3.41494292e-01 -7.67489150e-02 -6.27463698e-01
1.71546772e-01 -1.30716932e+00 -5.73633969e-01 -6.88592643e-02
-8.82517733e-03 -7.13013560e-02 9.24533010e-01 -8.28756213e-01
1.17119896e+00 -2.14597321e+00 3.03389341e-01 -1.02720216e-01
-3.95732634e-02 5.88718116e-01 -1.04187466e-01 5.53142607e-01
-3.19605678e-01 3.94141115e-02 -4.72804159e-01 -3.55936974e-01
-3.65148753e-01 5.87494195e-01 -7.73167551e-01 3.48661244e-01
4.40998614e-01 1.00901997e+00 -9.38722789e-01 2.18186483e-01
2.78648555e-01 5.06320179e-01 -1.64114069e-02 1.76125780e-01
-7.69106388e-01 5.36146998e-01 -2.94723839e-01 2.51808912e-01
6.40721440e-01 -5.72036445e-01 2.29786217e-01 2.67324805e-01
-7.14654148e-01 3.04291159e-01 -9.17491257e-01 1.33556521e+00
-6.50663197e-01 9.58803475e-01 -1.00935936e-01 -1.13554990e+00
7.55252361e-01 6.18154287e-01 2.88274199e-01 -2.79506624e-01
-2.19479218e-01 2.56187022e-01 -2.13538297e-02 -1.69113114e-01
6.08056068e-01 -9.19571053e-03 2.86269605e-01 2.71122366e-01
-2.70124584e-01 5.77891842e-02 2.37068102e-01 -3.29384476e-01
5.59049428e-01 3.97995859e-01 2.08267689e-01 -4.90279943e-02
3.25066924e-01 1.92858785e-01 7.25903690e-01 5.11345148e-01
2.69252300e-01 1.73331529e-01 5.26788771e-01 -6.82150722e-01
-1.03465462e+00 -6.39265001e-01 -1.72765013e-02 7.36479878e-01
-2.49860376e-01 1.14462517e-01 -3.47203791e-01 -1.06279537e-01
3.25809754e-02 6.58724070e-01 -8.14375639e-01 5.04063219e-02
-4.56831455e-01 -1.04040313e+00 1.14035830e-01 5.91856360e-01
6.17644906e-01 -1.17864203e+00 -8.57447803e-01 5.03422022e-01
-3.49839747e-01 -1.06503999e+00 1.74507834e-02 5.13393700e-01
-1.31144524e+00 -4.46524739e-01 -1.09265983e+00 -3.93721282e-01
5.25115788e-01 4.39610988e-01 9.67399001e-01 -5.72993495e-02
1.44856066e-01 -1.21305771e-01 -3.80638093e-01 -4.46560860e-01
-3.68059933e-01 3.00815463e-01 -5.14603639e-03 1.23043075e-01
2.50055760e-01 -7.02598810e-01 -4.39721316e-01 -4.02601063e-03
-1.03180003e+00 1.89566374e-01 4.86563265e-01 8.58741224e-01
5.69503188e-01 1.18191771e-01 2.36395434e-01 -6.26722097e-01
1.66857228e-01 -6.21676147e-01 -1.17829394e+00 3.48860443e-01
-4.10270005e-01 1.78268597e-01 6.21111572e-01 -5.68638027e-01
-9.60746467e-01 2.12425396e-01 8.82604718e-02 -6.50816739e-01
-1.38143763e-01 1.28497195e+00 5.25958478e-01 3.27000797e-01
3.59405071e-01 7.30436683e-01 9.53748450e-02 -4.47693437e-01
5.36633991e-02 5.20098209e-01 5.12097001e-01 -2.73725279e-02
7.95443654e-01 4.18144017e-01 -1.26807531e-02 -1.22663736e+00
-8.31563830e-01 -4.16266114e-01 -7.17351973e-01 -1.64918512e-01
7.32382178e-01 -1.29958081e+00 -6.61511064e-01 9.23593581e-01
-1.22302723e+00 -7.18545318e-01 -2.96375155e-01 7.01871693e-01
-2.20951051e-01 1.09365784e-01 -6.97484136e-01 -1.15984118e+00
-2.02167183e-01 -6.61262631e-01 8.74165416e-01 4.53367621e-01
1.54473141e-01 -1.29877722e+00 -6.94475546e-02 -6.07814848e-01
4.81914222e-01 4.72680837e-01 7.34828830e-01 -9.29848775e-02
-6.70890808e-01 -2.14319304e-01 -8.22311491e-02 4.97074962e-01
4.77831513e-01 3.69712293e-01 -1.03394878e+00 -3.67670357e-01
1.92478567e-01 1.70711339e-01 1.10115993e+00 8.74084353e-01
9.78439867e-01 -1.96899012e-01 -3.83765280e-01 7.07812846e-01
1.46053946e+00 2.67955601e-01 6.03165805e-01 1.32846832e-01
5.24942517e-01 4.44042921e-01 4.31328267e-01 2.39183173e-01
2.32959926e-01 3.00851315e-01 6.37030303e-01 -4.40573484e-01
3.43217194e-01 -6.03733547e-02 5.37032187e-01 1.01105535e+00
-3.25112700e-01 -3.51818204e-01 -9.74982381e-01 8.07938695e-01
-2.02175021e+00 -1.10416234e+00 -1.39746919e-01 2.48551702e+00
6.53845310e-01 -1.25312507e-01 -9.55191553e-02 5.37777916e-02
4.66854841e-01 4.81137931e-01 -8.40547681e-01 -1.75467655e-01
-2.97811538e-01 -4.36466001e-02 8.05632889e-01 4.41786289e-01
-1.15125561e+00 1.10245681e+00 5.60915327e+00 1.91214532e-01
-1.78520441e+00 -1.27260268e-01 4.63128746e-01 1.72179952e-01
3.29687856e-02 1.98980104e-02 -8.24244082e-01 4.25788909e-01
1.21440613e+00 6.86281845e-02 5.41770935e-01 4.17548388e-01
5.87824702e-01 -2.36570120e-01 -8.41859996e-01 4.92882073e-01
-4.93569165e-01 -1.30762804e+00 -1.10299841e-01 2.16445446e-01
7.54573762e-01 5.31404316e-01 1.44127950e-01 2.48949543e-01
5.77059925e-01 -8.53549659e-01 3.39982271e-01 7.69008338e-01
6.25461340e-01 -5.19318759e-01 6.54205263e-01 8.17978919e-01
-1.40092552e+00 -9.91800707e-03 -4.85795736e-01 -4.05422240e-01
5.34760989e-02 8.29529166e-01 -4.96100605e-01 7.77212679e-01
5.30092955e-01 1.22995889e+00 -1.52573749e-01 9.19373155e-01
-3.30796957e-01 8.93201888e-01 -6.44742548e-01 7.39572793e-02
7.14537501e-01 -5.10318041e-01 4.36217219e-01 9.67153907e-01
6.08751535e-01 1.31475732e-01 1.45264968e-01 8.88818026e-01
2.59427458e-01 -3.00594807e-01 -8.15784335e-01 -3.66763115e-01
2.13396803e-01 1.03022516e+00 -7.15298176e-01 -3.08945358e-01
-3.36731255e-01 1.05411375e+00 1.01403527e-01 8.52450490e-01
-5.27308285e-01 -6.68357462e-02 9.10623610e-01 -2.69030601e-01
7.21562564e-01 -6.57734156e-01 8.23123530e-02 -1.42442358e+00
1.05731405e-01 -7.01745272e-01 -4.04931866e-02 -9.06482875e-01
-9.59073901e-01 7.53989100e-01 -2.65342534e-01 -1.42886734e+00
-7.59306908e-01 -5.54868460e-01 -5.83750844e-01 1.28633869e+00
-2.00719595e+00 -1.09553814e+00 -2.74706274e-01 1.47376254e-01
5.20077884e-01 4.25422609e-01 1.13661385e+00 -7.18535557e-02
-6.17578506e-01 -1.82318702e-01 5.72930098e-01 1.33077592e-01
9.79920998e-02 -1.05695212e+00 7.96110988e-01 1.10992908e+00
1.00836486e-01 5.00574052e-01 8.38448942e-01 -6.70810580e-01
-1.11635864e+00 -1.39102983e+00 1.06306803e+00 6.88150078e-02
8.12342048e-01 -2.06806675e-01 -1.41305935e+00 9.79042292e-01
-7.83360470e-03 3.12225401e-01 3.31632078e-01 1.29912168e-01
-5.98201677e-02 -1.01559579e-01 -4.26869065e-01 4.59379286e-01
3.83908659e-01 -6.73008204e-01 -2.32266679e-01 3.65298629e-01
7.10679471e-01 -4.16017503e-01 -6.96421862e-01 4.61352110e-01
5.19774318e-01 -4.62923110e-01 6.04880989e-01 -6.34795547e-01
5.01446545e-01 -3.91389340e-01 -1.04864106e-01 -1.83414507e+00
-4.24041182e-01 -5.90814650e-01 -1.88310906e-01 8.85299444e-01
4.66663897e-01 -9.63044763e-01 5.15283585e-01 1.99935481e-01
2.50236601e-01 -6.56367123e-01 -5.44972003e-01 -1.06604087e+00
1.17608696e-01 -2.81615019e-01 6.02902949e-01 1.07439399e+00
-4.99957293e-01 5.57128154e-02 -5.66541493e-01 8.16495717e-01
3.23275089e-01 5.26555002e-01 4.09900725e-01 -1.65337873e+00
-1.62452489e-01 -2.83608317e-01 -1.37530163e-01 -1.36878920e+00
1.72783688e-01 -4.70875889e-01 9.75492522e-02 -1.27309370e+00
-2.25980699e-01 -4.03863937e-01 -3.67505163e-01 4.53276545e-01
-3.17863733e-01 -1.64062679e-01 2.61582136e-01 5.48367202e-01
2.21346870e-01 5.81202567e-01 1.04919970e+00 4.65670265e-02
-5.40065885e-01 4.52676415e-01 -7.11830556e-02 6.90348029e-01
8.11320126e-01 -3.78737420e-01 -1.75134480e-01 -6.05142355e-01
2.29963243e-01 3.87620896e-01 5.41336775e-01 -9.03948963e-01
1.11991659e-01 -2.69977063e-01 3.31359565e-01 -8.92495394e-01
3.47336978e-01 -8.31912518e-01 4.14331228e-01 7.17174113e-01
-1.26047865e-01 1.81147650e-01 5.07906854e-01 5.39667606e-01
-3.72784644e-01 -1.02032162e-01 5.40548563e-01 -2.75090337e-01
-6.18324578e-01 4.69606876e-01 -8.00277531e-01 -6.02169871e-01
6.69158101e-01 5.67111149e-02 -7.76419565e-02 -4.86680567e-01
-9.65335190e-01 5.07149220e-01 3.83142024e-01 4.24526691e-01
3.60337757e-02 -1.05216348e+00 -7.02143312e-01 2.72611469e-01
-1.42804429e-01 6.89434633e-02 1.86348483e-01 9.22655165e-01
-3.86762887e-01 6.86174929e-01 -1.84636321e-02 -9.72254813e-01
-9.04878139e-01 4.82910931e-01 5.87163687e-01 -6.35434449e-01
-5.74891925e-01 5.53822637e-01 1.95242867e-01 -3.50497931e-01
-3.41462158e-02 -9.35597837e-01 -2.73070633e-01 4.33503211e-01
4.51040804e-01 -1.38987601e-01 -3.38552357e-03 -4.98662382e-01
3.11410874e-02 4.55725968e-01 9.79188085e-03 -7.71128982e-02
1.59603083e+00 -1.55158877e-01 -1.37920111e-01 1.24734044e+00
8.16701651e-01 -6.80972278e-01 -1.56921232e+00 -7.16127217e-01
4.15592566e-02 -1.00648887e-01 4.28713471e-01 -5.72429895e-01
-1.21331716e+00 1.10044742e+00 6.59752369e-01 7.03629375e-01
1.26603103e+00 -5.45999169e-01 7.02560782e-01 5.18239439e-01
4.26235870e-02 -6.67224765e-01 -6.07268035e-01 7.74881482e-01
6.73630118e-01 -1.13036931e+00 -2.52083510e-01 -2.09644049e-01
-4.21112061e-01 1.07031655e+00 3.85839306e-02 -1.15616418e-01
7.44034350e-01 1.58314049e-01 1.69071570e-01 -5.71402116e-03
-1.16226876e+00 -5.24634659e-01 1.13715574e-01 2.80711144e-01
4.01836246e-01 3.91998947e-01 2.36673340e-01 6.61405027e-02
2.36935890e-03 3.61648053e-01 5.29570222e-01 9.34072077e-01
-2.91153312e-01 -7.14598060e-01 -3.28804493e-01 4.63581443e-01
-3.26996535e-01 -2.30053797e-01 8.93217996e-02 6.64423645e-01
-2.74135023e-01 6.57597184e-01 4.20626074e-01 7.44100139e-02
1.86589744e-03 2.60957003e-01 4.54035364e-02 -5.88531315e-01
-3.57194543e-01 2.24810615e-01 -4.31234568e-01 -3.27489525e-01
-5.85066080e-01 -1.07691514e+00 -9.53304172e-01 -3.11758429e-01
-6.26260936e-01 1.88418463e-01 8.83984208e-01 1.13288367e+00
2.71786213e-01 7.34950542e-01 8.00087214e-01 -1.17247200e+00
-4.52790558e-01 -1.02807593e+00 -7.74149716e-01 -2.16678292e-01
1.06741035e+00 -3.91201437e-01 -3.58957350e-01 3.68809342e-01] | [9.52322006225586, -1.5708585977554321] |
74896b9f-bc20-4b83-b799-9500a472c5e7 | conditional-gans-for-multi-illuminant-color | 1811.06604 | null | http://arxiv.org/abs/1811.06604v2 | http://arxiv.org/pdf/1811.06604v2.pdf | Conditional GANs for Multi-Illuminant Color Constancy: Revolution or Yet Another Approach? | Non-uniform and multi-illuminant color constancy are important tasks, the
solution of which will allow to discard information about lighting conditions
in the image. Non-uniform illumination and shadows distort colors of real-world
objects and mostly do not contain valuable information. Thus, many computer
vision and image processing techniques would benefit from automatic discarding
of this information at the pre-processing step. In this work we propose novel
view on this classical problem via generative end-to-end algorithm based on
image conditioned Generative Adversarial Network. We also demonstrate the
potential of the given approach for joint shadow detection and removal. Forced
by the lack of training data, we render the largest existing shadow removal
dataset and make it publicly available. It consists of approximately 6,000
pairs of wide field of view synthetic images with and without shadows. | ['Oleksii Sidorov'] | 2018-11-15 | null | null | null | null | ['shadow-removal', 'color-constancy', 'shadow-detection-and-removal', 'shadow-detection'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 7.13655114e-01 -1.52021751e-01 7.01533020e-01 -2.56497830e-01
-1.68887094e-01 -8.08635533e-01 5.53993940e-01 -6.47507250e-01
-3.13291252e-01 1.18940532e+00 -1.98416933e-01 -3.65206301e-01
3.60645115e-01 -6.49374485e-01 -7.45138645e-01 -1.00755477e+00
3.54029417e-01 2.09878325e-01 4.37488019e-01 -1.10492907e-01
2.32277419e-02 6.42829597e-01 -1.55819774e+00 6.23939373e-02
1.04898238e+00 7.22162783e-01 4.16937500e-01 8.71315897e-01
1.80034757e-01 5.26366472e-01 -7.99056172e-01 -4.13624346e-01
5.22459686e-01 -5.53441286e-01 -1.18061654e-01 2.59604305e-01
5.06120503e-01 -6.12897277e-01 -2.25796252e-01 1.07618976e+00
6.94338918e-01 1.09122731e-01 6.43185914e-01 -1.40425634e+00
-5.56791008e-01 -2.78397560e-01 -5.27574480e-01 -5.29826470e-02
1.63295865e-01 2.48881593e-01 4.21639800e-01 -8.28555882e-01
7.58301914e-01 9.90335464e-01 6.66970968e-01 3.30669492e-01
-1.22559845e+00 -5.20694077e-01 -1.20483674e-01 5.04595824e-02
-1.12834847e+00 -5.04216075e-01 1.15947759e+00 -1.12795353e-01
2.25653782e-01 6.03489637e-01 6.01790726e-01 1.33981848e+00
3.79047155e-01 4.29449081e-01 1.93497407e+00 -5.35738885e-01
1.83623835e-01 2.41036639e-01 -3.13841850e-01 6.79490685e-01
4.82194960e-01 3.66440535e-01 -3.59149426e-01 -2.30064437e-01
5.30077279e-01 5.73927946e-02 -5.60227871e-01 -5.84049702e-01
-9.27916110e-01 3.58794242e-01 3.53205889e-01 -1.24073565e-01
-1.03730358e-01 1.38017595e-01 -1.10255435e-01 1.67993084e-01
5.12515545e-01 1.62823841e-01 -3.44846427e-01 4.40005630e-01
-9.11240160e-01 3.00286829e-01 5.86843371e-01 8.68137896e-01
8.53664041e-01 3.45965058e-01 -2.42727399e-01 5.91455579e-01
2.01141581e-01 1.19685543e+00 8.61531198e-02 -9.81921792e-01
-1.46566004e-01 1.40816152e-01 5.99597394e-01 -9.41880107e-01
-2.01154396e-01 -2.58343071e-01 -8.65359902e-01 8.89795780e-01
4.72612143e-01 -3.01703036e-01 -1.41128063e+00 1.53248847e+00
3.91645074e-01 1.89112306e-01 2.08015591e-02 9.31388676e-01
7.28851020e-01 4.21923697e-01 -2.79576838e-01 -2.79574811e-01
1.07295167e+00 -8.21750641e-01 -7.99823225e-01 -2.76504904e-01
-3.86094689e-01 -1.27994943e+00 1.17325342e+00 5.55581927e-01
-7.79036403e-01 -5.13945699e-01 -1.11564732e+00 2.55611967e-02
-5.52580833e-01 -4.24517915e-02 6.85936272e-01 9.77703929e-01
-9.26279187e-01 4.34177935e-01 -5.35916924e-01 -1.74061611e-01
2.04294756e-01 4.31856923e-02 -2.28253752e-01 -5.23643017e-01
-8.37176919e-01 7.49901354e-01 9.31269974e-02 3.35798681e-01
-1.03316462e+00 -5.40514767e-01 -5.58935940e-01 -2.51467645e-01
5.02825439e-01 -5.79642534e-01 7.62561440e-01 -1.31141770e+00
-1.35646868e+00 6.20076060e-01 -1.96126938e-01 -7.48901516e-02
8.63324285e-01 -4.58598554e-01 -3.64859730e-01 4.97880541e-02
-1.03987083e-01 7.34543353e-02 1.47055709e+00 -2.09585333e+00
-1.15090692e-02 -1.74999759e-01 -1.68523341e-01 1.18072189e-01
9.39484388e-02 -2.01635629e-01 -4.55217600e-01 -7.09821224e-01
-8.40078220e-02 -1.01877189e+00 -1.22154474e-01 5.77693395e-02
-7.19157457e-01 7.56311238e-01 1.26016927e+00 -8.31160307e-01
4.85156000e-01 -1.88875985e+00 -2.85499096e-01 7.70959780e-02
-2.31463201e-02 2.54209697e-01 -6.67266473e-02 5.28638244e-01
-1.28883719e-02 -3.14720541e-01 -5.47001481e-01 -4.64294702e-01
-1.73734456e-01 3.94136071e-01 -4.99464661e-01 8.45519364e-01
-9.99687836e-02 8.15814257e-01 -8.76508057e-01 -5.21833718e-01
6.27676845e-01 8.02728772e-01 -2.68482447e-01 5.17157078e-01
-3.17913413e-01 7.08186626e-01 -2.34569415e-01 7.81555831e-01
1.24029517e+00 3.19334030e-01 6.63078204e-02 -2.61929095e-01
6.64631501e-02 -4.22416061e-01 -1.00596488e+00 1.47794247e+00
-4.47518766e-01 8.14716458e-01 2.93247670e-01 -4.65976983e-01
5.58730185e-01 4.30559106e-02 3.55012447e-01 -7.37380445e-01
1.03977457e-01 -5.71796335e-02 -2.22627088e-01 -2.50013828e-01
5.06226957e-01 -4.18742716e-01 2.57361829e-01 2.67933697e-01
-3.04091513e-01 -5.45020580e-01 -1.52657762e-01 1.33009121e-01
9.04892325e-01 5.30343175e-01 1.42702060e-02 -1.40145272e-01
3.88300389e-01 -2.75247633e-01 5.37194490e-01 7.84180403e-01
-5.30971363e-02 1.16085196e+00 1.28375888e-01 -6.13943897e-02
-1.12781751e+00 -1.40342879e+00 -5.00600860e-02 8.45902205e-01
2.88852632e-01 2.74895608e-01 -6.62239552e-01 -6.04123116e-01
8.60751723e-04 9.02569532e-01 -8.04164529e-01 -2.60423925e-02
-3.50930065e-01 -1.07233906e+00 2.81579912e-01 -2.09399275e-02
6.32147133e-01 -1.09383035e+00 -8.33291173e-01 -1.55908599e-01
-1.12371035e-01 -1.25562143e+00 -1.65613204e-01 2.42122054e-01
-4.83061701e-01 -1.29884338e+00 -9.23920214e-01 -1.76984951e-01
9.20564115e-01 6.51945472e-01 1.49465573e+00 7.78198941e-03
-6.66575015e-01 4.23097521e-01 -3.88635576e-01 -6.68158770e-01
-4.88199472e-01 -5.87175786e-01 -2.74739951e-01 3.84030104e-01
-3.72477114e-01 -7.82374799e-01 -9.96730089e-01 2.73455918e-01
-1.03368998e+00 2.86210239e-01 7.42295265e-01 7.19600201e-01
3.69128019e-01 8.17665681e-02 1.23179547e-01 -1.36360216e+00
3.55166465e-01 -7.86265284e-02 -7.56693006e-01 2.39425614e-01
-7.12470651e-01 -7.18114749e-02 8.25957179e-01 -8.76185670e-02
-1.63778365e+00 2.16485456e-01 1.69560552e-01 -5.63228250e-01
-4.18597728e-01 -3.98662120e-01 -5.20020068e-01 -3.21724564e-01
4.75686014e-01 4.49946463e-01 -3.96989554e-01 -3.21062148e-01
5.08198440e-01 1.84220001e-01 8.14759552e-01 -3.54796946e-01
1.44638157e+00 1.03028929e+00 3.47662389e-01 -1.00470376e+00
-8.98922801e-01 -3.34957719e-01 -5.52465141e-01 -3.90008479e-01
7.04978585e-01 -6.86805904e-01 -3.40870380e-01 7.07403898e-01
-9.75789368e-01 -5.03116906e-01 -1.12063885e-01 -3.30710076e-02
-5.04822850e-01 7.11164355e-01 -8.53591412e-02 -1.01934922e+00
-2.39014268e-01 -8.47576320e-01 1.10698247e+00 3.03690076e-01
3.58245492e-01 -9.18373108e-01 8.67621526e-02 3.74208927e-01
4.77448881e-01 7.96912432e-01 7.28435040e-01 1.58349827e-01
-7.15098858e-01 -4.90508042e-03 -3.63358468e-01 8.00390363e-01
3.49945515e-01 -1.75570678e-02 -1.54449975e+00 -4.25988138e-01
1.27751201e-01 -2.08671302e-01 1.09105015e+00 3.99059206e-01
1.03635824e+00 -1.80331804e-02 -9.50886756e-02 8.10101092e-01
1.83433008e+00 -6.94186091e-02 8.71169388e-01 3.47394198e-02
1.01008332e+00 4.90943402e-01 6.54970646e-01 4.13162470e-01
-2.32436970e-01 4.25672799e-01 7.78270066e-01 -6.90538287e-01
-6.10412598e-01 4.80445810e-02 1.43875092e-01 9.89436507e-02
-3.33750814e-01 -6.37719274e-01 -5.59641004e-01 4.68979686e-01
-1.46354461e+00 -9.37622249e-01 -2.95209378e-01 2.25147438e+00
6.36888862e-01 1.91046279e-02 -4.22127098e-01 8.90606344e-02
4.29200023e-01 3.72465700e-01 -5.60945630e-01 -5.63799329e-02
-5.29717863e-01 3.11909765e-01 8.68539333e-01 6.77843630e-01
-9.40317512e-01 9.71808672e-01 6.23593950e+00 6.81650460e-01
-1.05717969e+00 7.72718638e-02 4.94726539e-01 1.04885466e-01
-4.68616247e-01 -9.55492072e-03 -1.69624507e-01 5.17623067e-01
4.99828070e-01 3.73704255e-01 7.11266577e-01 5.35550356e-01
2.54240721e-01 -7.99904168e-01 -5.84382534e-01 8.73961687e-01
4.29259539e-01 -6.43459797e-01 -1.44706741e-01 -7.15640932e-02
9.73980844e-01 -4.71694693e-02 2.39762038e-01 -1.49542779e-01
4.38888639e-01 -9.98894632e-01 5.87986887e-01 7.93690443e-01
8.48549247e-01 -8.10077846e-01 5.45543492e-01 1.92408383e-01
-7.40814209e-01 2.94770211e-01 -4.30208653e-01 1.68617323e-01
1.80607095e-01 8.63714218e-01 -9.38601732e-01 7.77857363e-01
5.09716034e-01 1.90699443e-01 -7.46817827e-01 8.12068224e-01
-5.62669873e-01 5.75899005e-01 -3.30471188e-01 3.74268264e-01
-2.19109669e-01 -4.62316871e-01 6.20389342e-01 1.21176291e+00
1.93494871e-01 1.95535272e-02 -4.01402079e-02 7.93495178e-01
5.52229881e-02 -3.88227105e-01 -8.39312255e-01 3.54670137e-01
-8.40046778e-02 1.34971333e+00 -1.10709727e+00 -1.28068715e-01
-4.11279678e-01 1.69840801e+00 -2.23402902e-01 8.37797463e-01
-9.68314290e-01 -2.62607336e-01 5.86880088e-01 1.80559680e-01
2.79838294e-01 -3.32941771e-01 -4.29572701e-01 -1.05864751e+00
-3.88485491e-02 -7.37048447e-01 -1.91304803e-01 -1.27849221e+00
-1.05445516e+00 5.13251722e-01 -1.31322086e-01 -9.93473351e-01
1.56900547e-02 -6.54994309e-01 -6.78775132e-01 1.02684295e+00
-1.95111823e+00 -1.56390274e+00 -8.35968018e-01 9.16073382e-01
5.01352191e-01 2.10873168e-02 8.05084407e-01 1.11438930e-01
-1.48860067e-01 1.73367903e-01 6.41078770e-01 -1.08913444e-01
1.14802134e+00 -1.41536736e+00 2.63098896e-01 1.36562812e+00
-6.63909316e-02 2.72670597e-01 1.34299147e+00 -6.93499863e-01
-1.41004932e+00 -1.23404074e+00 2.56174922e-01 -6.21208370e-01
1.00492999e-01 -6.54881716e-01 -6.86636031e-01 3.49487066e-01
5.67001641e-01 1.81832194e-01 2.63782263e-01 -3.42935503e-01
-1.53938979e-01 -1.99919194e-01 -1.20719779e+00 6.34131312e-01
8.27489555e-01 -3.56690645e-01 -3.69316250e-01 4.51643139e-01
3.49390566e-01 -3.48261386e-01 -9.15223286e-02 3.51619095e-01
5.75765967e-01 -1.71576667e+00 1.27180600e+00 -2.53492482e-02
3.42130721e-01 -4.45550054e-01 -2.58027881e-01 -1.43486965e+00
6.03782292e-03 -7.36345053e-01 1.19987614e-01 1.18937147e+00
-6.75977999e-03 -6.50710404e-01 7.57845759e-01 3.18639815e-01
-2.96293855e-01 -2.29024976e-01 -5.68942428e-01 -6.78221107e-01
-2.70783186e-01 -4.28171396e-01 2.41029814e-01 6.27585948e-01
-1.17831206e+00 6.85287043e-02 -9.14455771e-01 4.47352827e-01
1.27391517e+00 6.36956632e-01 1.16792524e+00 -1.07001710e+00
-4.60310787e-01 1.72511831e-01 1.98957413e-01 -3.34689081e-01
1.26957282e-01 -3.10343415e-01 3.60520124e-01 -1.52130246e+00
3.30136448e-01 -3.59087884e-01 -1.71426073e-01 2.87254244e-01
-3.48690957e-01 8.58688056e-01 1.89709757e-02 3.28209600e-03
-3.54656845e-01 7.75051832e-01 1.55364263e+00 -5.06214090e-02
9.94994789e-02 2.40702368e-02 -2.92890191e-01 9.22688425e-01
6.67385101e-01 -4.10273105e-01 -6.38163686e-01 -1.95391625e-01
1.78460572e-02 -3.71750504e-01 8.49029183e-01 -1.00140035e+00
-3.26496214e-01 -4.60054457e-01 9.34931874e-01 -5.09352505e-01
7.82392979e-01 -1.08627999e+00 2.23740131e-01 3.84947926e-01
3.76163632e-01 -2.47474357e-01 1.73481286e-01 7.15754569e-01
1.01228349e-01 5.16620316e-02 1.06315851e+00 -2.97084510e-01
-6.85776949e-01 7.00899586e-03 -1.49364889e-01 3.15039903e-02
1.02302206e+00 -2.59208828e-01 -3.67595166e-01 -5.01884758e-01
-3.70454162e-01 -3.87265801e-01 9.76794720e-01 8.18716660e-02
5.82822323e-01 -8.40260983e-01 -6.54406965e-01 4.43587899e-01
-4.54672873e-02 -7.42133632e-02 3.66834015e-01 4.99946773e-01
-7.32643485e-01 9.65876579e-02 -5.43920934e-01 -2.02989712e-01
-1.47432089e+00 6.75252199e-01 2.01463148e-01 7.38891885e-02
-6.12348855e-01 6.12480521e-01 7.21171975e-01 -1.72051460e-01
-1.38228908e-01 -2.05136612e-01 2.78963774e-01 -3.48765194e-01
2.17383310e-01 2.93043375e-01 1.75558493e-01 -5.78898549e-01
-1.48042217e-01 3.88157487e-01 4.30852383e-01 -1.74921677e-01
1.37151039e+00 -3.32246542e-01 -3.29077579e-02 4.05274749e-01
8.51813555e-01 6.15097761e-01 -1.47937274e+00 1.50781289e-01
-6.71384633e-01 -9.36573386e-01 1.33442029e-01 -1.18791056e+00
-1.07819819e+00 7.22731888e-01 8.40937018e-01 8.87507796e-02
1.55515540e+00 -4.24244612e-01 6.52760029e-01 2.35589460e-01
3.20227563e-01 -9.62030113e-01 9.36263241e-03 2.66329229e-01
9.05638933e-01 -1.42781007e+00 3.21466535e-01 -6.10176444e-01
-4.78368044e-01 8.17605674e-01 3.27515543e-01 -2.79862255e-01
4.01607335e-01 5.19595325e-01 4.07797098e-01 5.19342236e-02
-3.34699601e-01 -3.67506921e-01 3.00495118e-01 1.07356930e+00
1.78476080e-01 -1.14797793e-01 -8.61237198e-02 1.58115834e-01
-6.92154688e-04 -3.45258743e-01 6.52855039e-01 9.21796978e-01
-2.30601877e-01 -1.04128957e+00 -8.87306571e-01 1.46448523e-01
-4.52731192e-01 -2.76360184e-01 -6.53032541e-01 7.59133458e-01
3.86707157e-01 9.14682865e-01 -4.45233941e-01 -5.37267849e-02
1.44927397e-01 -6.93421811e-02 6.35802507e-01 -2.54552126e-01
-1.29762650e-01 2.45012924e-01 1.45463571e-02 -6.91643238e-01
-4.63379592e-01 -6.25066876e-01 -6.90963507e-01 -1.98266298e-01
-2.46701151e-01 -3.11496675e-01 7.20961154e-01 7.07811475e-01
1.92716159e-02 8.22638214e-01 5.99965572e-01 -1.20183229e+00
-1.65072292e-01 -8.60248387e-01 -8.10620725e-01 6.30560815e-01
6.22824967e-01 -8.02682757e-01 -4.01135951e-01 4.09798950e-01] | [10.797304153442383, -4.039897918701172] |
bc7b89ae-8ccd-4a9d-b6ad-3cb8fff09c06 | harnessing-the-power-of-multi-task | 2212.04231 | null | https://arxiv.org/abs/2212.04231v2 | https://arxiv.org/pdf/2212.04231v2.pdf | Harnessing the Power of Multi-Task Pretraining for Ground-Truth Level Natural Language Explanations | Natural language explanations promise to offer intuitively understandable explanations of a neural network's decision process in complex vision-language tasks, as pursued in recent VL-NLE models. While current models offer impressive performance on task accuracy and explanation plausibility, they suffer from a range of issues: Some models feature a modular design where the explanation generation module is poorly integrated with a separate module for task-answer prediction, employ backbone models trained on limited sets of tasks, or incorporate ad hoc solutions to increase performance on single datasets. We propose to evade these limitations by applying recent advances in large-scale multi-task pretraining of generative Transformer models to the problem of VL-NLE tasks. Our approach outperforms recent models by a large margin, with human annotators preferring the generated explanations over the ground truth in two out of three evaluated datasets. As a novel challenge in VL-NLE research, we propose the problem of multi-task VL-NLE and show that jointly training on multiple tasks can increase the explanation quality. We discuss the ethical implications of high-quality NLE generation and other issues in recent VL-NLE research. | ['Stefan Wermter', 'Jae Hee Lee', 'Lukas Braach', 'Jakob Ambsdorf', 'Björn Plüster'] | 2022-12-08 | null | null | null | null | ['explanation-generation', 'visual-entailment'] | ['natural-language-processing', 'reasoning'] | [ 5.46044350e-01 1.09216774e+00 -1.16656363e-01 -7.18260527e-01
-9.83208299e-01 -3.03296804e-01 1.01311052e+00 -3.43729645e-01
-1.40631735e-01 9.49468732e-01 5.71388841e-01 -6.49858415e-01
-1.62209570e-01 -4.01329957e-02 -8.77453983e-01 -1.91577449e-01
4.46771502e-01 8.80392432e-01 3.66971008e-02 1.47721311e-02
2.40132868e-01 2.37684604e-02 -1.50974965e+00 7.23094881e-01
8.98148358e-01 8.03220212e-01 2.56571263e-01 8.53715599e-01
-1.31282210e-01 9.20262635e-01 -4.96526927e-01 -9.54672754e-01
-2.19793897e-02 -5.66729844e-01 -8.71233344e-01 2.38841474e-01
1.05961227e+00 -2.19103724e-01 3.79450135e-02 5.13835430e-01
3.62331897e-01 -2.46617824e-01 8.34406435e-01 -1.81785679e+00
-1.27940834e+00 5.69538057e-01 -8.65778029e-02 3.20195444e-02
2.24418372e-01 4.45894986e-01 1.34534132e+00 -1.15941453e+00
7.36469448e-01 1.40378523e+00 7.89903998e-01 1.17116570e+00
-1.57159507e+00 -4.21694517e-01 2.89320916e-01 4.15079653e-01
-7.10291326e-01 -7.24868059e-01 3.43953341e-01 -4.73356515e-01
1.42878354e+00 1.02200046e-01 5.38142383e-01 1.61755741e+00
5.70065022e-01 8.68229747e-01 1.09607649e+00 -1.72578365e-01
-1.02782324e-02 3.13408226e-01 5.40379174e-02 8.91489327e-01
4.64667618e-01 2.52160817e-01 -8.27867210e-01 -1.76958978e-01
6.19437575e-01 -3.56947124e-01 -4.28639412e-01 -4.34679717e-01
-1.30093598e+00 1.13248551e+00 4.73597527e-01 -1.81370392e-01
-4.70385045e-01 4.98238862e-01 1.50283724e-01 3.22104007e-01
4.55019981e-01 8.62653136e-01 -6.53136551e-01 3.10489833e-01
-8.03757489e-01 6.50668442e-01 7.25764155e-01 9.50617373e-01
5.67824483e-01 3.45260501e-01 -4.47982997e-01 4.96033996e-01
6.31892443e-01 2.46584713e-01 2.78916001e-01 -1.16210306e+00
5.17706573e-01 2.48251781e-01 -8.79079252e-02 -5.03867984e-01
-3.19955349e-01 -5.43837130e-01 -6.33293688e-01 3.72041464e-01
4.17951763e-01 -3.91225964e-02 -1.00786400e+00 1.96698964e+00
-1.45712540e-01 -8.15199614e-02 3.66930634e-01 8.25636864e-01
1.01525319e+00 2.81727344e-01 3.98668140e-01 -6.25486076e-02
1.25474107e+00 -1.48403561e+00 -6.68810904e-01 -1.01765788e+00
2.85980463e-01 -8.29069853e-01 1.12385654e+00 3.77458066e-01
-1.22975326e+00 -5.77829003e-01 -7.87604451e-01 -3.56263757e-01
3.68672609e-02 2.39924714e-01 1.00389731e+00 3.37082744e-01
-1.44552290e+00 1.75005734e-01 -4.16141421e-01 -2.38729224e-01
7.78577864e-01 3.17946166e-01 -2.87270188e-01 -1.53809682e-01
-8.61828387e-01 1.25369978e+00 2.43259251e-01 -1.30851373e-01
-1.16171372e+00 -6.22105598e-01 -9.93915975e-01 1.88866988e-01
3.28902513e-01 -1.39942682e+00 1.45843053e+00 -1.05869985e+00
-9.85952497e-01 9.55365777e-01 -5.62694073e-01 -7.54200161e-01
5.34527659e-01 -3.15271884e-01 -1.24846928e-01 -9.62615460e-02
5.02819836e-01 1.33759797e+00 8.74892533e-01 -1.34420550e+00
-3.95435661e-01 -1.00203887e-01 6.21912554e-02 1.19251525e-02
2.89654404e-01 -5.15883148e-01 -2.51127742e-02 -3.90499979e-01
1.34805560e-01 -1.07175720e+00 -3.99370074e-01 1.07730776e-01
-3.62363577e-01 -6.48628294e-01 5.68320632e-01 -3.84149522e-01
4.38219756e-01 -1.67364109e+00 -4.14755978e-02 -4.95777309e-01
6.80801868e-01 1.71540216e-01 -4.53309566e-01 -1.12563455e-02
-1.84251890e-01 3.70632112e-01 -6.20735949e-03 -8.36212158e-01
6.60874546e-02 4.30309594e-01 -7.41304755e-01 -4.12750617e-02
6.72399163e-01 1.41502798e+00 -8.06847751e-01 -5.72198093e-01
1.21916644e-01 3.03211927e-01 -7.30962574e-01 2.91930437e-01
-7.24016130e-01 4.05150205e-01 -4.05244172e-01 2.84773678e-01
2.13319018e-01 -7.00191140e-01 -2.13705093e-01 1.31405229e-02
2.10813969e-01 5.89998186e-01 -6.85355544e-01 1.77267218e+00
-4.25786614e-01 1.11718631e+00 -3.21720093e-01 -6.64417684e-01
4.84495014e-01 6.81011677e-01 -1.01992905e-01 -2.73488820e-01
-9.62461531e-02 3.92704368e-01 2.60458469e-01 -5.23915946e-01
4.56422091e-01 -3.24751973e-01 2.33577237e-01 5.57588279e-01
3.80096972e-01 -3.58669251e-01 -1.92286745e-01 4.14708614e-01
9.80140090e-01 3.52017254e-01 4.18321460e-01 -4.33196872e-02
9.10349786e-02 4.27702516e-01 3.06723416e-01 1.19433224e+00
-1.89296857e-01 8.55444074e-01 4.47958320e-01 -7.14214444e-01
-1.18242872e+00 -9.54464078e-01 4.43096366e-03 9.19836283e-01
-7.75613114e-02 -3.75674218e-01 -5.45796096e-01 -8.93368125e-01
-1.89069226e-01 1.44953358e+00 -7.25174606e-01 -1.05925851e-01
-1.85498610e-01 -4.27446991e-01 5.20061851e-01 5.90733349e-01
2.45423883e-01 -1.36608326e+00 -8.49498510e-01 1.75356343e-01
-4.32686836e-01 -1.45147693e+00 -9.40721706e-02 2.78734326e-01
-8.22916269e-01 -8.77110302e-01 -3.82132262e-01 -4.77324516e-01
5.37819564e-01 4.60133404e-01 1.72627723e+00 3.09202760e-01
-1.11407869e-01 4.52654243e-01 -3.67058702e-02 -8.39394391e-01
-6.31989956e-01 -6.01618998e-02 -8.26095194e-02 -4.50077653e-01
3.21057856e-01 -2.86768764e-01 -4.23012763e-01 2.61579394e-01
-5.55471659e-01 6.58344209e-01 8.25978577e-01 1.10802603e+00
5.72655916e-01 -7.21324503e-01 7.78972328e-01 -9.16692972e-01
7.56908357e-01 -4.96296197e-01 -8.03828016e-02 3.84465754e-01
-1.14475632e+00 5.96300662e-01 4.13150758e-01 -5.27781606e-01
-1.19182611e+00 -8.34802017e-02 -1.75402716e-01 -6.21390164e-01
-3.32800299e-01 8.72074962e-02 7.52350986e-02 6.83113113e-02
7.55916953e-01 2.17272714e-01 -1.17028803e-01 -1.36661157e-02
5.98150909e-01 4.51268740e-02 5.81605315e-01 -4.44654703e-01
7.14789331e-01 2.39063501e-01 -9.05670505e-03 -3.32420021e-01
-1.43351996e+00 1.34482279e-01 -1.92892373e-01 -6.65614661e-03
1.15863347e+00 -1.08479071e+00 -4.41788197e-01 -3.14739764e-01
-1.94510579e+00 -2.58728236e-01 -3.29821587e-01 4.34999019e-01
-8.43454659e-01 -1.31522715e-01 -1.63015947e-01 -7.03448415e-01
-2.40378544e-01 -1.24658954e+00 1.34338224e+00 1.17676809e-01
-6.90554023e-01 -1.10541141e+00 -1.02924787e-01 8.56863976e-01
4.07649904e-01 2.65885163e-02 1.06221128e+00 -9.34877098e-01
-1.11518669e+00 1.50405526e-01 -3.20508748e-01 -5.16918860e-02
-4.93404388e-01 -3.91597867e-01 -1.16558623e+00 9.55405384e-02
1.30491048e-01 -9.75756228e-01 1.15845811e+00 5.97650170e-01
1.13986015e+00 -3.94091457e-01 -2.83351302e-01 5.16756654e-01
1.21893573e+00 -3.35436314e-01 5.22318184e-01 2.08240241e-01
6.37457669e-01 7.70442486e-01 4.24289942e-01 -1.20633254e-02
6.34454072e-01 4.52017903e-01 8.49656343e-01 -2.19830826e-01
-5.27077734e-01 -4.50722814e-01 3.61592412e-01 9.92954522e-02
-8.66808072e-02 -7.67298818e-01 -7.22247362e-01 7.88983524e-01
-2.10680604e+00 -9.89128590e-01 -6.45742655e-01 1.64862144e+00
3.71729761e-01 2.86874622e-01 -3.62131089e-01 -4.89085197e-01
2.63018757e-01 1.95733532e-01 -6.86238706e-01 -6.42060161e-01
-2.79234886e-01 -6.68878034e-02 1.42747611e-01 6.87179565e-01
-5.93079805e-01 1.18811989e+00 7.08843660e+00 5.23699880e-01
-5.62612772e-01 4.92228508e-01 5.63396871e-01 -1.85644567e-01
-9.60764349e-01 2.36988828e-01 -7.87117183e-01 -1.81590274e-01
8.20468962e-01 7.12634474e-02 2.00065300e-01 9.14533198e-01
-2.05197129e-02 -8.98537859e-02 -1.62015247e+00 8.29139650e-01
4.68252748e-01 -1.64994240e+00 4.14905787e-01 2.87135392e-01
8.18607092e-01 2.76133031e-01 2.00298667e-01 2.46478930e-01
4.61630642e-01 -1.55327082e+00 1.02205348e+00 3.47793579e-01
6.82462633e-01 -1.46984667e-01 6.22453153e-01 4.90572989e-01
-9.25084412e-01 7.34571889e-02 -4.92084086e-01 -2.11150676e-01
3.69382173e-01 3.50364655e-01 -1.17359412e+00 2.08051562e-01
2.73096949e-01 3.93314898e-01 -7.22371876e-01 6.39181495e-01
-8.44783247e-01 5.26693642e-01 1.48067445e-01 6.88681528e-02
4.55322742e-01 3.49494398e-01 6.52542114e-01 9.92501080e-01
2.73044676e-01 1.93843633e-04 -9.76963863e-02 1.51145482e+00
5.31156249e-02 -4.10608143e-01 -9.50595617e-01 8.75979736e-02
8.32013860e-02 1.26908255e+00 -3.06252986e-01 -3.40607643e-01
-3.26651365e-01 1.08547854e+00 5.35751939e-01 3.34468812e-01
-9.33356225e-01 4.45687264e-01 6.12139702e-01 -1.41591569e-02
1.37714207e-01 -2.01086681e-02 -6.56074166e-01 -1.25726163e+00
-5.26483022e-02 -9.36779976e-01 1.08913720e-01 -1.33158362e+00
-1.16736984e+00 9.42474008e-01 -2.02833951e-01 -7.59604454e-01
-8.95432651e-01 -7.59879887e-01 -6.15916550e-01 7.92680085e-01
-1.67278278e+00 -1.57415962e+00 -2.58580238e-01 4.48498070e-01
1.27525473e+00 -5.01494944e-01 9.54816461e-01 -3.47896606e-01
-3.22032608e-02 4.56180900e-01 -6.00861967e-01 -5.08680642e-01
6.65296853e-01 -1.41668141e+00 8.18039179e-01 8.79226863e-01
7.63344288e-01 5.37065268e-01 1.10972190e+00 -6.57458961e-01
-9.39361453e-01 -9.23426032e-01 1.54256773e+00 -1.17398179e+00
3.33644450e-01 -3.77834201e-01 -8.70665193e-01 1.15430951e+00
4.69909877e-01 -1.02035575e-01 6.34207308e-01 4.50519472e-01
-4.09686267e-01 5.86310089e-01 -8.76043558e-01 6.99622869e-01
1.34602594e+00 -5.33393264e-01 -1.09166300e+00 6.06114268e-01
8.16214681e-01 -2.79169738e-01 -3.14893097e-01 1.97218612e-01
6.38276875e-01 -1.25949633e+00 9.65986788e-01 -9.53085065e-01
1.14115596e+00 -3.98694165e-02 -1.53473904e-02 -1.40882266e+00
-4.30531442e-01 -5.42074680e-01 -4.49118242e-02 9.17210042e-01
8.20287526e-01 -5.39666593e-01 6.24618351e-01 8.56634974e-01
-3.86145681e-01 -8.85309398e-01 -8.27386677e-01 -3.97531182e-01
-1.70145422e-01 -6.64833724e-01 2.79006273e-01 4.78807390e-01
-5.01285434e-01 9.66693401e-01 -5.73482096e-01 1.49229631e-01
1.04477227e+00 3.19305249e-02 9.67194796e-01 -1.40124869e+00
-4.45533991e-01 -3.35518241e-01 1.08773187e-01 -1.06759596e+00
6.86353087e-01 -9.91593182e-01 2.39797607e-01 -1.97826326e+00
4.76872653e-01 -1.90408714e-02 2.62551308e-01 6.75277293e-01
-3.41457605e-01 1.64708689e-01 3.31768364e-01 5.00009179e-01
-7.74658084e-01 5.83182633e-01 1.36431754e+00 9.33470428e-02
2.84496874e-01 -1.60895407e-01 -1.14318919e+00 1.01118922e+00
4.37914163e-01 -7.14401782e-01 -8.13711643e-01 -9.91881371e-01
2.83272207e-01 1.22606099e-01 1.00487208e+00 -7.27210820e-01
1.72228381e-01 -1.34869516e-01 5.80417395e-01 -3.04548025e-01
3.55351985e-01 -5.79659581e-01 2.49637797e-01 4.15537715e-01
-7.40964830e-01 1.16548568e-01 5.24427034e-02 7.77368248e-01
9.37902331e-02 -6.30104244e-02 4.67794627e-01 -3.02689284e-01
-6.52029812e-01 2.22582772e-01 -3.20099860e-01 1.92889705e-01
7.10204363e-01 -4.48962539e-01 -6.07922912e-01 -5.81928134e-01
-6.17236376e-01 3.74713719e-01 2.71033168e-01 5.09666681e-01
9.23507333e-01 -1.19830275e+00 -9.85264182e-01 -2.20790529e-03
3.23092252e-01 -1.26821890e-01 -4.72522303e-02 7.19141841e-01
1.60784259e-01 8.88146698e-01 -2.58333683e-01 -5.99227250e-01
-1.16551316e+00 1.64935827e-01 4.09466714e-01 -4.63008404e-01
-6.12449646e-01 1.18307066e+00 6.41270459e-01 -3.27151090e-01
1.75667228e-03 -5.91867603e-02 -1.03836551e-01 -2.37431064e-01
2.33739972e-01 -5.50875217e-02 -3.05738360e-01 -3.75184536e-01
-2.86417991e-01 3.19407314e-01 -9.21719298e-02 -3.63574892e-01
1.28499126e+00 -2.30573952e-01 3.17222476e-01 3.29258800e-01
5.53007901e-01 -3.49213511e-01 -1.56150210e+00 -9.95291695e-02
1.46600828e-02 -2.67979711e-01 -5.56791909e-02 -1.17927194e+00
-5.88833153e-01 1.05603921e+00 2.32312426e-01 -1.46493822e-01
7.08225548e-01 6.51991546e-01 6.01051211e-01 4.95756477e-01
2.84583122e-01 -6.14869535e-01 4.01925355e-01 4.10893142e-01
1.31009865e+00 -1.73308015e+00 -1.04020284e-02 -2.38573030e-01
-1.15707207e+00 9.53131557e-01 9.38718855e-01 9.53341543e-04
-4.26556952e-02 -1.12759799e-01 1.01835847e-01 -6.20963812e-01
-1.33639085e+00 -2.19749048e-01 7.33229578e-01 6.89781368e-01
6.02509856e-01 -5.26572093e-02 -9.04407352e-02 6.64781988e-01
-3.78474921e-01 -7.12622553e-02 4.56557363e-01 6.32298961e-02
-3.91825795e-01 -9.60647106e-01 -1.06231429e-01 5.01572490e-01
-2.37589508e-01 -4.24757928e-01 -7.28524923e-01 8.21589351e-01
3.23561162e-01 1.09285891e+00 -5.96216582e-02 -1.05233066e-01
-3.58197652e-02 3.89024943e-01 4.93193388e-01 -8.33106935e-01
-6.00992978e-01 -1.76460326e-01 3.82064432e-01 -7.17209220e-01
-3.91999125e-01 -6.42273724e-01 -1.11521471e+00 3.74600687e-03
-3.51685911e-01 -1.13663830e-01 6.06549919e-01 1.51212847e+00
4.67158079e-01 7.67099798e-01 -8.49051401e-02 -8.23201716e-01
-6.16872430e-01 -7.84068346e-01 6.15943410e-03 3.50412488e-01
5.40786922e-01 -6.34594917e-01 -4.65944707e-01 1.30813494e-01] | [10.836188316345215, 1.9552754163742065] |
8badde80-1f8f-4a21-bcaf-fc72fce2f7a3 | learning-from-partially-labeled-data-for | 2211.06894 | null | https://arxiv.org/abs/2211.06894v1 | https://arxiv.org/pdf/2211.06894v1.pdf | Learning from partially labeled data for multi-organ and tumor segmentation | Medical image benchmarks for the segmentation of organs and tumors suffer from the partially labeling issue due to its intensive cost of labor and expertise. Current mainstream approaches follow the practice of one network solving one task. With this pipeline, not only the performance is limited by the typically small dataset of a single task, but also the computation cost linearly increases with the number of tasks. To address this, we propose a Transformer based dynamic on-demand network (TransDoDNet) that learns to segment organs and tumors on multiple partially labeled datasets. Specifically, TransDoDNet has a hybrid backbone that is composed of the convolutional neural network and Transformer. A dynamic head enables the network to accomplish multiple segmentation tasks flexibly. Unlike existing approaches that fix kernels after training, the kernels in the dynamic head are generated adaptively by the Transformer, which employs the self-attention mechanism to model long-range organ-wise dependencies and decodes the organ embedding that can represent each organ. We create a large-scale partially labeled Multi-Organ and Tumor Segmentation benchmark, termed MOTS, and demonstrate the superior performance of our TransDoDNet over other competitors on seven organ and tumor segmentation tasks. This study also provides a general 3D medical image segmentation model, which has been pre-trained on the large-scale MOTS benchmark and has demonstrated advanced performance over BYOL, the current predominant self-supervised learning method. Code will be available at \url{https://git.io/DoDNet}. | ['Chunhua Shen', 'Yong Xia', 'Jianpeng Zhang', 'Yutong Xie'] | 2022-11-13 | null | null | null | null | ['tumor-segmentation'] | ['computer-vision'] | [-2.85816994e-02 3.71532679e-01 -3.01269889e-01 -3.47734541e-01
-8.22663963e-01 -4.86062676e-01 3.47858548e-01 1.45907653e-02
-4.90010947e-01 3.28557760e-01 7.99723789e-02 -4.21998650e-01
2.54243404e-01 -5.30499160e-01 -6.29436791e-01 -8.93408597e-01
2.29026690e-01 5.91861427e-01 4.75687623e-01 7.63736013e-03
-4.47907001e-01 4.02241528e-01 -7.60753095e-01 1.97757587e-01
9.27865386e-01 1.01595199e+00 3.80081028e-01 7.58175015e-01
-2.94243276e-01 9.21530545e-01 -2.56099701e-01 -2.24263102e-01
3.17378283e-01 -2.78597921e-01 -1.07745540e+00 8.35350752e-02
3.39050889e-01 -3.59652877e-01 -4.39136803e-01 8.70738745e-01
6.17067993e-01 -2.97900498e-01 5.39331436e-01 -1.21005917e+00
-6.80492580e-01 7.44064212e-01 -4.99407768e-01 2.60392606e-01
-1.84580147e-01 3.46625090e-01 7.66999424e-01 -6.06895804e-01
4.66179311e-01 9.37096536e-01 8.00284982e-01 8.20247829e-01
-1.16404891e+00 -5.70145845e-01 1.13313727e-01 -4.24161881e-01
-1.34088624e+00 -1.39019489e-01 4.08992320e-01 -5.17636120e-01
8.08132350e-01 1.65964335e-01 7.60848463e-01 8.92192304e-01
3.57819498e-01 9.95885015e-01 7.71595776e-01 2.82560894e-03
-1.02419943e-01 -1.82931405e-02 4.22932267e-01 1.01878726e+00
1.17555656e-01 -1.65019706e-01 6.45958930e-02 1.00915693e-01
8.42184961e-01 2.73182362e-01 -3.01177055e-01 -3.47404391e-01
-1.44788826e+00 7.27603734e-01 9.64813113e-01 3.46613824e-01
-3.49471182e-01 3.37734073e-01 5.59740067e-01 1.42761126e-01
4.48851228e-01 2.15476304e-01 -7.15661645e-01 3.96159351e-01
-7.87435889e-01 -1.16833955e-01 7.90760934e-01 8.29918087e-01
5.91118455e-01 -3.01769197e-01 -6.40815437e-01 6.94984257e-01
4.25977498e-01 1.90148249e-01 6.57523811e-01 -5.80478370e-01
1.12984352e-01 1.05624700e+00 -3.92105103e-01 -1.83955923e-01
-7.15459466e-01 -7.23000169e-01 -9.85447586e-01 -3.31956297e-02
6.28187656e-01 -1.86441034e-01 -1.50627255e+00 1.71826279e+00
6.63958430e-01 3.12832177e-01 3.02708112e-02 9.14704800e-01
1.35722125e+00 3.87779534e-01 3.91890407e-01 3.74503851e-01
1.56401265e+00 -1.65586746e+00 -4.01060134e-01 -1.81692332e-01
1.07140875e+00 -4.87427324e-01 1.07613993e+00 -1.93211213e-01
-8.82784009e-01 -2.34893546e-01 -7.96933889e-01 -4.77578372e-01
-4.49716598e-01 9.83148068e-02 8.22222471e-01 4.11512285e-01
-1.33814204e+00 1.46827862e-01 -1.20858657e+00 -4.00143832e-01
7.71087766e-01 6.19401932e-01 -3.46653789e-01 -1.57486439e-01
-9.18436468e-01 8.00279140e-01 3.89404416e-01 1.77728415e-01
-1.17940342e+00 -1.01625454e+00 -9.48155344e-01 1.38097957e-01
2.36443609e-01 -1.06499040e+00 1.30716681e+00 -8.67790401e-01
-1.38960230e+00 1.18963170e+00 1.40783101e-01 -3.85970920e-01
5.37560880e-01 1.33040279e-01 2.13157404e-02 2.68605232e-01
2.78083861e-01 1.14551306e+00 5.04233420e-01 -8.16830873e-01
-3.52575392e-01 -2.02749580e-01 5.82080856e-02 2.46377766e-01
-3.70885074e-01 -2.57385612e-01 -1.02589011e+00 -5.28302908e-01
-3.39445919e-02 -1.12375033e+00 -5.93968987e-01 1.50786966e-01
-7.42439866e-01 -3.17265689e-01 8.34083021e-01 -4.34951007e-01
9.65762913e-01 -2.09658813e+00 2.49764562e-01 -1.01329111e-01
5.28010428e-01 1.88925400e-01 -2.76521266e-01 -1.08650409e-01
-1.35651931e-01 6.77506849e-02 -3.14704388e-01 -5.05595326e-01
-1.34941563e-01 3.29990745e-01 2.81491458e-01 4.26723927e-01
5.88390157e-02 1.42006028e+00 -7.91022062e-01 -8.97078753e-01
9.67472941e-02 4.46872443e-01 -4.67189133e-01 2.63992399e-01
-1.45228431e-01 6.09908700e-01 -5.53688586e-01 1.03512716e+00
5.26371539e-01 -1.02582943e+00 -1.06244810e-01 -4.02240813e-01
1.20899275e-01 -4.77972478e-02 -5.55463970e-01 1.99858844e+00
-2.99832761e-01 1.45054027e-01 2.46552885e-01 -1.00522912e+00
3.92285138e-01 4.08921450e-01 9.48246062e-01 -4.89125580e-01
2.81159997e-01 2.32368007e-01 -3.42043722e-03 -4.15465593e-01
1.88232940e-02 2.25000549e-03 -2.49345854e-01 2.44226485e-01
2.96621144e-01 -5.21734729e-02 2.36919612e-01 3.41674924e-01
1.45167542e+00 2.23305635e-02 2.13310242e-01 -4.44700986e-01
5.52831948e-01 3.25341642e-01 6.64953113e-01 5.28164029e-01
-4.84288633e-01 5.12810886e-01 6.38583481e-01 -6.07773662e-01
-6.58965170e-01 -1.14060569e+00 -3.92124742e-01 1.09768772e+00
2.73354024e-01 -1.77364573e-01 -9.16984022e-01 -1.21667135e+00
1.42841950e-01 1.23101860e-01 -9.39671755e-01 7.69578516e-02
-4.25222486e-01 -1.02011931e+00 6.67545438e-01 5.83077371e-01
4.60421443e-01 -9.84283268e-01 -5.00724971e-01 1.05269395e-01
-1.38033301e-01 -1.26493800e+00 -9.13029909e-01 3.40083599e-01
-8.85739744e-01 -1.22685993e+00 -1.11009526e+00 -1.04535985e+00
1.03901637e+00 1.40402138e-01 1.20397007e+00 4.03916001e-01
-5.45339227e-01 2.67797321e-01 -7.35295117e-02 -3.47648352e-01
-2.54619896e-01 5.96035659e-01 -5.62784374e-01 -1.41015410e-01
1.99895412e-01 -2.10130394e-01 -8.25203240e-01 4.14944291e-01
-1.09632146e+00 4.72704053e-01 9.16748881e-01 1.07403457e+00
8.69584858e-01 -4.18377995e-01 4.35284942e-01 -1.14099467e+00
2.49606833e-01 -6.35152161e-01 -5.01704812e-01 3.56718153e-01
-2.97004730e-01 -1.04130596e-01 4.18313801e-01 -3.49517018e-01
-7.09827363e-01 5.04691422e-01 -2.39069209e-01 -5.25014341e-01
-9.78151187e-02 4.67714429e-01 7.82248676e-02 -2.71154195e-01
3.92470181e-01 1.53387859e-01 1.57312647e-01 -2.83871770e-01
7.99104050e-02 3.56292456e-01 3.61925036e-01 -3.27237725e-01
6.27818525e-01 4.51942354e-01 -6.55990914e-02 -5.02608538e-01
-1.13945627e+00 -4.95314419e-01 -6.23647988e-01 -9.00055617e-02
1.36788452e+00 -1.13263106e+00 -5.71058333e-01 6.43500030e-01
-8.86118054e-01 -9.38127458e-01 -4.02366012e-01 3.88090611e-01
-2.52504528e-01 5.21129146e-02 -1.20058477e+00 2.77325083e-02
-7.02817976e-01 -1.52099121e+00 1.30494857e+00 3.72999012e-01
5.64943179e-02 -1.29084516e+00 -7.40382671e-02 4.79332060e-01
6.26730859e-01 4.11304712e-01 8.64722908e-01 -7.40218818e-01
-8.02913189e-01 -1.21705726e-01 -4.74302828e-01 1.36252388e-01
2.53291458e-01 -2.73419887e-01 -7.84352601e-01 -5.23811996e-01
-3.29946458e-01 -5.13172507e-01 9.49220479e-01 6.33430660e-01
1.32536423e+00 1.65801812e-02 -6.55477285e-01 1.13109839e+00
1.44996703e+00 -2.02039152e-01 2.94056207e-01 8.70002210e-02
1.02421176e+00 2.92936265e-01 1.97213560e-01 -4.79920320e-02
7.46406853e-01 2.81004131e-01 5.94602942e-01 -8.65338445e-01
-3.24120551e-01 4.95904274e-02 -4.50411029e-02 8.59822989e-01
2.49410167e-01 -3.49208742e-01 -1.12738895e+00 7.03858674e-01
-1.69639468e+00 -3.50371867e-01 -3.68829555e-04 1.65084779e+00
8.45405281e-01 -5.87751828e-02 3.61972377e-02 -4.21200752e-01
4.77217406e-01 1.55139575e-02 -9.03693020e-01 -9.53257084e-03
2.11655840e-01 -5.55072539e-03 6.32594943e-01 3.66759062e-01
-1.30247521e+00 9.75869417e-01 6.39260435e+00 7.48904645e-01
-1.35736394e+00 2.58842766e-01 9.16079998e-01 -2.34158635e-02
-1.27832904e-01 -2.85336763e-01 -7.75203288e-01 4.46793318e-01
6.72207832e-01 1.24822855e-01 -3.54624353e-03 7.79377878e-01
-1.68184504e-01 2.75423247e-02 -1.10457826e+00 7.07998872e-01
-1.34971574e-01 -1.31847775e+00 -8.57523605e-02 2.40928948e-01
5.55121064e-01 6.17316604e-01 -2.67941225e-02 3.60476643e-01
5.30107737e-01 -1.06926453e+00 3.29405606e-01 2.54580319e-01
9.16796863e-01 -1.97937980e-01 8.50241601e-01 3.52576852e-01
-1.32200158e+00 7.54076913e-02 -9.15563256e-02 6.10275865e-01
1.82404250e-01 5.68603694e-01 -9.15479302e-01 3.42028052e-01
7.58991480e-01 8.12531173e-01 -6.64247513e-01 1.03342223e+00
-5.61894402e-02 5.80983877e-01 -4.52300936e-01 2.06159294e-01
3.96810621e-01 -9.75197926e-03 2.60205388e-01 1.34625006e+00
2.37155452e-01 -6.53745756e-02 5.26004136e-01 7.89179146e-01
-2.98695266e-01 5.64426221e-02 -2.30468243e-01 -8.30558594e-03
5.38246669e-02 1.71701860e+00 -1.05438030e+00 -5.01397014e-01
-4.97912437e-01 1.07757318e+00 3.69664967e-01 1.79137975e-01
-1.13532579e+00 -1.31903812e-01 4.38803703e-01 3.96831427e-03
3.36657256e-01 8.62348825e-02 -2.72001714e-01 -1.20174229e+00
-3.86798799e-01 -5.73084295e-01 7.12599933e-01 -4.71196175e-01
-1.32056952e+00 8.81380200e-01 -2.01158881e-01 -9.84647751e-01
2.81189710e-01 -6.58849239e-01 -6.29530728e-01 7.49638140e-01
-1.76126015e+00 -1.62743878e+00 -6.00561678e-01 8.02826107e-01
4.63173926e-01 1.72650754e-01 8.57061565e-01 5.88163674e-01
-1.02594316e+00 7.25390911e-01 -3.14369828e-01 5.43022037e-01
6.66139424e-01 -1.51619911e+00 2.51375884e-01 5.22062302e-01
-3.25148672e-01 3.49192142e-01 1.16010502e-01 -5.26573002e-01
-1.29817462e+00 -1.32095385e+00 6.68059766e-01 -4.93702859e-01
6.05469882e-01 -3.89172465e-01 -7.61977255e-01 8.79080534e-01
2.40240708e-01 7.85421073e-01 8.23418558e-01 -1.88586131e-01
-5.69477528e-02 -5.30126467e-02 -1.30437005e+00 3.81566465e-01
9.27638531e-01 -2.61442721e-01 -2.37412229e-01 6.87468350e-01
1.11029744e+00 -9.93663728e-01 -1.22020197e+00 4.01378781e-01
3.24977785e-01 -6.05947971e-01 9.98279810e-01 -3.44309777e-01
3.98515731e-01 -3.61102641e-01 1.34652868e-01 -1.09006500e+00
-4.10794944e-01 -3.43019813e-01 -6.07900806e-02 8.76710832e-01
5.95755219e-01 -8.02662373e-01 1.03889704e+00 6.50954187e-01
-5.83280861e-01 -1.23270571e+00 -7.53471911e-01 -3.38697165e-01
2.54684150e-01 6.98196888e-02 5.76006830e-01 9.96034563e-01
-3.66778105e-01 3.65973711e-01 8.64067227e-02 1.43871486e-01
6.99707627e-01 9.55011472e-02 5.30326009e-01 -9.85911250e-01
-3.62847954e-01 -3.76307338e-01 -1.98320374e-01 -1.14900208e+00
-1.15039218e-02 -1.34362864e+00 9.11411736e-03 -1.68059921e+00
3.82979959e-01 -6.18713498e-01 -6.39603853e-01 9.72623408e-01
-2.41457060e-01 4.16740566e-01 9.77202058e-02 2.97442257e-01
-8.22332621e-01 2.07243145e-01 1.78004718e+00 -2.93010354e-01
-1.96783364e-01 -6.22894652e-02 -8.01343262e-01 6.54638112e-01
6.59523189e-01 -4.87037241e-01 -3.71437073e-01 -6.98099375e-01
-2.36264750e-01 1.55864045e-01 4.10663754e-01 -9.33110774e-01
4.31452990e-01 6.87225312e-02 4.17873412e-01 -6.06639624e-01
8.08405653e-02 -8.62544537e-01 1.12171985e-01 7.04183936e-01
-2.69175768e-01 3.03321779e-02 2.24073231e-01 3.93971890e-01
-2.35962987e-01 1.74441993e-01 9.85634148e-01 -3.70798200e-01
-4.96718854e-01 1.03759253e+00 -1.15174055e-01 2.74776202e-02
1.33295166e+00 -1.04029641e-01 -3.26662809e-01 2.91196585e-01
-7.21671045e-01 8.16607237e-01 3.53289604e-01 1.79821894e-01
3.09040993e-01 -1.10886776e+00 -6.03971839e-01 1.59175750e-02
9.08577740e-02 7.13743925e-01 4.80260670e-01 1.39497185e+00
-7.86477983e-01 3.70344400e-01 -9.77278408e-03 -9.79308367e-01
-9.72007930e-01 5.12311161e-01 8.44686866e-01 -8.29282045e-01
-8.53955805e-01 1.10127711e+00 6.71738803e-01 -7.14109600e-01
2.96213388e-01 -6.30339026e-01 -9.29908156e-02 -2.47189388e-01
2.62968123e-01 -9.46623385e-02 -1.67907253e-02 -5.25958478e-01
-4.15463805e-01 4.88206983e-01 -3.01842779e-01 4.91591245e-01
1.26926672e+00 1.54586444e-02 -2.53399044e-01 -2.42528021e-02
1.37795460e+00 -4.40632731e-01 -1.48727000e+00 -3.59636426e-01
-1.46449760e-01 1.06631212e-01 2.19394118e-01 -9.43892241e-01
-1.65105832e+00 6.77988768e-01 6.37697577e-01 5.91925159e-02
1.27256048e+00 2.71810621e-01 1.20447326e+00 1.12816051e-01
-7.13747665e-02 -6.75701380e-01 1.16877044e-02 5.42133510e-01
4.95648861e-01 -1.31295371e+00 -2.19440103e-01 -5.74901700e-01
-6.02647305e-01 8.68027568e-01 9.14899468e-01 -1.54247642e-01
8.69761050e-01 7.31732190e-01 4.58620608e-01 -4.54482168e-01
-6.95130885e-01 -1.49781793e-01 2.71885574e-01 2.07212120e-01
5.52018642e-01 2.04184968e-02 5.18951053e-03 5.52528501e-01
1.10051177e-01 2.27115929e-01 7.09379315e-02 7.33378053e-01
-1.21551208e-01 -1.05502474e+00 -5.13292365e-02 6.14449441e-01
-7.07337737e-01 -6.73638657e-02 -8.08128715e-02 8.23581874e-01
2.18334779e-01 3.44817668e-01 -7.56993964e-02 -1.19187772e-01
1.69754073e-01 -1.42996907e-01 2.36863807e-01 -7.04695225e-01
-9.46905315e-01 1.14514366e-01 -3.24538618e-01 -6.62674844e-01
-4.04426664e-01 -3.07027072e-01 -1.51925635e+00 -3.89519706e-02
-1.65697455e-01 -1.60737224e-02 4.58707929e-01 8.06167424e-01
3.14921916e-01 1.08319914e+00 2.92470187e-01 -7.69279361e-01
-4.27490532e-01 -7.85219729e-01 -3.64060402e-01 3.78906399e-01
4.56978202e-01 -3.83928984e-01 -1.40083089e-01 -7.74265826e-02] | [14.675503730773926, -2.5243349075317383] |
7bdf0502-8a8f-49e7-bd40-32ad415558d7 | towards-high-fidelity-monocular-face | 2103.15432 | null | https://arxiv.org/abs/2103.15432v3 | https://arxiv.org/pdf/2103.15432v3.pdf | Towards High Fidelity Monocular Face Reconstruction with Rich Reflectance using Self-supervised Learning and Ray Tracing | Robust face reconstruction from monocular image in general lighting conditions is challenging. Methods combining deep neural network encoders with differentiable rendering have opened up the path for very fast monocular reconstruction of geometry, lighting and reflectance. They can also be trained in self-supervised manner for increased robustness and better generalization. However, their differentiable rasterization based image formation models, as well as underlying scene parameterization, limit them to Lambertian face reflectance and to poor shape details. More recently, ray tracing was introduced for monocular face reconstruction within a classic optimization-based framework and enables state-of-the art results. However optimization-based approaches are inherently slow and lack robustness. In this paper, we build our work on the aforementioned approaches and propose a new method that greatly improves reconstruction quality and robustness in general scenes. We achieve this by combining a CNN encoder with a differentiable ray tracer, which enables us to base the reconstruction on much more advanced personalized diffuse and specular albedos, a more sophisticated illumination model and a plausible representation of self-shadows. This enables to take a big leap forward in reconstruction quality of shape, appearance and lighting even in scenes with difficult illumination. With consistent face attributes reconstruction, our method leads to practical applications such as relighting and self-shadows removal. Compared to state-of-the-art methods, our results show improved accuracy and validity of the approach. | ['Louis Chevallier', 'Christian Theobalt', 'Philippe-Henri Gosselin', 'Junghyun Ahn', 'Cedric Thebault', 'Abdallah Dib'] | 2021-03-29 | null | http://openaccess.thecvf.com//content/ICCV2021/html/Dib_Towards_High_Fidelity_Monocular_Face_Reconstruction_With_Rich_Reflectance_Using_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Dib_Towards_High_Fidelity_Monocular_Face_Reconstruction_With_Rich_Reflectance_Using_ICCV_2021_paper.pdf | iccv-2021-1 | ['face-reconstruction'] | ['computer-vision'] | [ 1.80049777e-01 -1.40121967e-01 1.44743070e-01 -4.76059556e-01
-3.10839802e-01 -2.79735327e-01 5.96483588e-01 -5.86416841e-01
4.04103436e-02 7.17242718e-01 -3.35238711e-03 -3.03540509e-02
2.43084803e-01 -9.97634709e-01 -8.24224412e-01 -7.88623154e-01
4.01102662e-01 1.79917082e-01 -3.03382240e-02 -3.75485539e-01
-8.39093551e-02 1.01525486e+00 -1.97749162e+00 5.84774558e-03
8.26782405e-01 1.00235963e+00 6.80587888e-02 3.78389001e-01
-2.10878700e-01 6.33361816e-01 -1.98005632e-01 -5.89133859e-01
3.32009345e-01 -4.33596939e-01 -3.12353671e-01 2.98427016e-01
7.49410689e-01 -6.25539780e-01 -3.39821160e-01 8.13157558e-01
5.73783815e-01 -6.84050098e-02 5.29844403e-01 -8.20479035e-01
-6.96361423e-01 -2.00025588e-01 -6.94267333e-01 -5.12317777e-01
6.41799212e-01 2.34182194e-01 5.01066566e-01 -1.04552233e+00
5.77664852e-01 1.37662470e+00 9.04492855e-01 6.76624119e-01
-1.27753615e+00 -6.30315661e-01 7.86658749e-02 1.48754328e-01
-1.26339257e+00 -7.39726603e-01 1.02026343e+00 -1.14905424e-01
5.43506801e-01 4.99316752e-01 7.67304003e-01 1.08203912e+00
-1.37227684e-01 6.30208552e-01 1.31009912e+00 -4.37887132e-01
6.48499578e-02 3.63676161e-01 -4.35388833e-01 9.83221352e-01
5.44151887e-02 3.99009556e-01 -3.94011527e-01 1.48151860e-01
1.01340067e+00 3.40180516e-01 -6.29390419e-01 -4.11422521e-01
-7.50755072e-01 6.38255656e-01 6.86693251e-01 1.54891461e-01
-2.70977050e-01 1.70223027e-01 -9.60818604e-02 1.11441746e-01
7.66479075e-01 2.55697429e-01 -1.93632022e-01 3.26510221e-01
-1.08530962e+00 8.16633254e-02 7.48603404e-01 7.12616622e-01
1.08687389e+00 4.37262654e-01 8.12898949e-02 9.93614733e-01
3.77993762e-01 7.93149114e-01 -3.21568511e-02 -1.15803087e+00
2.17924081e-02 4.70696002e-01 1.25550941e-01 -9.62030292e-01
-4.84215081e-01 -4.87768918e-01 -8.70323181e-01 6.77659273e-01
1.86207786e-01 4.77039218e-02 -9.23830807e-01 1.57597756e+00
7.57464230e-01 1.86309412e-01 -8.79564583e-02 9.44429874e-01
9.15406108e-01 4.75088269e-01 -7.03952551e-01 -1.88304499e-01
1.25566328e+00 -8.59238505e-01 -7.33477831e-01 1.26116261e-01
1.17223248e-01 -9.38498139e-01 1.18365848e+00 6.06822550e-01
-1.15773356e+00 -3.51166159e-01 -8.17944109e-01 -2.89519787e-01
-2.04771813e-02 6.51061386e-02 7.91502655e-01 9.52122688e-01
-1.26840150e+00 7.63791978e-01 -7.11380482e-01 -4.37993616e-01
6.04009926e-01 2.03538820e-01 -3.62003148e-01 -3.97997558e-01
-6.48972809e-01 7.05269873e-01 -4.15508181e-01 1.49799094e-01
-7.71137118e-01 -8.11198831e-01 -9.65082049e-01 -1.17313452e-01
3.73570204e-01 -8.57164562e-01 8.51496637e-01 -1.13346040e+00
-2.31812692e+00 8.54030430e-01 -2.13176802e-01 -1.95807870e-03
6.78944468e-01 -2.34372959e-01 -1.00153811e-01 2.40219653e-01
-3.36235464e-01 4.06853765e-01 1.18824089e+00 -1.72283483e+00
-1.53778801e-02 -4.58142281e-01 9.23261642e-02 2.49065176e-01
-2.58225471e-01 5.68395630e-02 -5.67677617e-01 -4.01620269e-01
2.33640671e-01 -7.32338846e-01 1.98355056e-02 6.51513219e-01
-2.96437711e-01 3.90296966e-01 7.79186606e-01 -6.01874888e-01
5.59616387e-01 -2.01550603e+00 -1.42010888e-02 -5.19059300e-02
-4.78371680e-02 2.45527834e-01 -1.31835043e-01 4.41621721e-01
-2.09378134e-02 -1.20691277e-01 -3.03060293e-01 -9.64902520e-01
-1.14926584e-01 2.47657001e-01 -3.41466576e-01 8.72034311e-01
1.94135472e-01 8.77101064e-01 -6.15458310e-01 -1.69760406e-01
5.68001449e-01 1.26116598e+00 -6.68292761e-01 4.40545768e-01
-1.88375801e-01 9.00819302e-01 1.07138425e-01 8.19730878e-01
1.07706082e+00 -1.53998574e-02 -4.61414717e-02 -2.74434507e-01
-2.27739051e-01 3.89567874e-02 -1.14543462e+00 1.66696727e+00
-9.50087905e-01 5.60962498e-01 5.11049747e-01 -6.39412284e-01
9.44255471e-01 1.09209985e-01 5.02282262e-01 -8.61208975e-01
9.04477388e-02 1.51691630e-01 -6.05859339e-01 -3.77598733e-01
3.84998590e-01 -5.00124633e-01 7.69969463e-01 3.46838564e-01
-3.27716231e-01 -6.90417409e-01 -3.45573932e-01 -1.99658915e-01
6.09789789e-01 7.31453478e-01 1.34245520e-02 -1.97198287e-01
6.60946369e-01 -5.72235644e-01 4.84025061e-01 2.14239523e-01
4.35451031e-01 1.12069678e+00 2.89510209e-02 -5.39207339e-01
-8.53954434e-01 -9.80640411e-01 -4.48529720e-01 7.05164194e-01
1.91827700e-01 -1.60868376e-01 -8.31603408e-01 -3.24519455e-01
-2.84338091e-02 5.80860496e-01 -5.33437967e-01 1.02753155e-01
-7.60650873e-01 -7.62844145e-01 2.94452667e-01 1.64183080e-01
6.72863841e-01 -8.25017989e-01 -4.06808048e-01 -2.34800894e-02
2.38363706e-02 -1.29244184e+00 -9.81544554e-02 -2.95471132e-01
-8.01274657e-01 -9.94309366e-01 -8.85570705e-01 -3.19394171e-01
7.78187275e-01 4.51197386e-01 1.12650228e+00 3.60089928e-01
-5.53660572e-01 5.91026902e-01 -7.89580569e-02 -2.12435693e-01
-3.37366402e-01 -4.28970397e-01 7.62560219e-02 5.60979962e-01
-4.45487827e-01 -8.89150023e-01 -1.05491376e+00 5.08301973e-01
-1.00671840e+00 2.82283455e-01 2.06682324e-01 7.06201434e-01
4.30049002e-01 -2.45957717e-01 1.51218906e-01 -1.00205946e+00
-1.18969142e-01 -2.29010329e-01 -9.30600226e-01 -2.67469063e-02
-6.83355987e-01 -1.05792329e-01 8.44014823e-01 -1.20653979e-01
-1.51215553e+00 -6.08366653e-02 -4.44620401e-01 -5.23816407e-01
-1.49152845e-01 -6.71814680e-02 -2.11539686e-01 -5.10171950e-01
7.09901750e-01 4.08559203e-01 1.46375194e-01 -7.33769059e-01
4.30860221e-01 2.99336404e-01 4.28132623e-01 -4.30005938e-01
1.16974270e+00 1.23414636e+00 2.97962964e-01 -1.30382955e+00
-6.90553188e-01 -9.90185067e-02 -5.66315591e-01 -2.06649706e-01
5.28261006e-01 -9.41213846e-01 -8.73168647e-01 5.88690817e-01
-1.16861963e+00 -5.90602458e-01 -3.53930682e-01 2.21010894e-01
-5.24891913e-01 7.50157416e-01 -6.26230538e-01 -1.13748109e+00
-2.37807393e-01 -1.11626244e+00 1.31222999e+00 1.66824579e-01
4.75145102e-01 -9.40134406e-01 -2.67153047e-02 4.31812227e-01
5.93451500e-01 4.51330543e-01 5.19555569e-01 5.72230101e-01
-9.56930220e-01 2.01664761e-01 -4.96931195e-01 4.67620701e-01
3.32467109e-01 1.61553159e-01 -1.53099537e+00 -3.95415545e-01
2.71441013e-01 -1.39842302e-01 9.18878257e-01 2.46966705e-01
1.23929000e+00 -3.45904648e-01 -2.10444525e-01 1.46051002e+00
1.62378895e+00 -2.59398103e-01 8.54818702e-01 4.36885632e-04
1.08014226e+00 8.50079119e-01 3.72074097e-01 5.05718350e-01
3.97243291e-01 1.03419697e+00 7.61561334e-01 -5.84320724e-01
-5.77709079e-01 -8.19012448e-02 3.57064337e-01 5.17833292e-01
-3.37049007e-01 -1.04812942e-01 -5.17003357e-01 8.25119987e-02
-1.42871904e+00 -7.91688085e-01 -2.05794483e-01 2.37598300e+00
7.57184803e-01 -5.82612276e-01 -1.56372279e-01 1.75006747e-01
2.85809904e-01 1.72356486e-01 -4.51830059e-01 -2.58805066e-01
-3.80087435e-01 3.03844988e-01 3.56787741e-01 7.66708910e-01
-7.11229861e-01 1.01298988e+00 5.96099234e+00 5.70305347e-01
-1.54680347e+00 1.16500303e-01 5.93671918e-01 -1.10566445e-01
-7.94429004e-01 -1.30005151e-01 -7.63013780e-01 2.52423018e-01
5.08961558e-01 5.86663961e-01 8.78808558e-01 5.13241410e-01
1.97944999e-01 -1.37650266e-01 -9.96476650e-01 1.43865514e+00
4.25196022e-01 -1.26311457e+00 -2.04120632e-02 1.55950651e-01
9.46912706e-01 2.20293533e-02 1.46699056e-01 -1.06548533e-01
4.40931469e-02 -1.23039114e+00 6.10908270e-01 6.29008591e-01
1.08662963e+00 -6.67159438e-01 2.27934867e-01 1.70861617e-01
-9.51339900e-01 6.18723817e-02 -4.72312361e-01 -1.07233098e-03
1.65680632e-01 8.86487663e-01 -5.25502741e-01 8.28977108e-01
6.99630141e-01 7.55908132e-01 -2.50114590e-01 7.02771008e-01
-3.78078938e-01 3.67156774e-01 -5.25895000e-01 3.44952464e-01
-1.95680812e-01 -5.92946708e-01 4.98909891e-01 1.04285312e+00
2.38501698e-01 1.29466251e-01 -8.02358426e-03 1.12386107e+00
-1.75810307e-01 3.03228438e-01 -7.92472363e-01 4.57469612e-01
-1.72634274e-02 1.44721019e+00 -5.89950860e-01 -4.28434238e-02
-4.69688714e-01 1.10422432e+00 4.05852884e-01 6.15378261e-01
-7.57181764e-01 1.23165928e-01 7.72003055e-01 4.85321254e-01
2.51671284e-01 -3.03053349e-01 -3.14013153e-01 -1.49446833e+00
9.35795680e-02 -6.89847827e-01 -1.77275583e-01 -7.66765118e-01
-1.08714712e+00 7.16590285e-01 -2.51318187e-01 -9.49595749e-01
-4.30874452e-02 -6.58617616e-01 -6.17397189e-01 7.01183438e-01
-2.42338681e+00 -1.36867774e+00 -6.28171027e-01 7.79496491e-01
3.77668738e-01 2.52546240e-02 8.20464194e-01 4.09259826e-01
-5.23227751e-01 5.88405967e-01 2.42009461e-01 -3.52441400e-01
6.57787800e-01 -1.07273173e+00 1.79483458e-01 7.95987725e-01
7.99754187e-02 5.16342521e-01 6.17485106e-01 -3.38926017e-01
-1.87222564e+00 -9.64373887e-01 2.00942785e-01 -3.19213301e-01
1.10618606e-01 -5.10929823e-01 -9.08850789e-01 3.94595623e-01
1.31051183e-01 1.31341487e-01 3.94524723e-01 -1.00686103e-01
-4.47124243e-01 -4.88277972e-01 -1.29115641e+00 7.82524228e-01
1.20618832e+00 -4.76193160e-01 1.69075467e-02 5.24907351e-01
3.99915338e-01 -4.47236568e-01 -6.12373829e-01 4.05114263e-01
5.75064838e-01 -1.60324240e+00 1.14614129e+00 1.87209457e-01
3.76361281e-01 -2.57350266e-01 -2.22439155e-01 -1.12165713e+00
2.84236185e-02 -1.06366158e+00 -2.38428153e-02 1.23220241e+00
-5.46320062e-03 -9.93959904e-01 6.39493942e-01 4.55538094e-01
-2.76368111e-01 -8.70867968e-01 -8.42743516e-01 -6.16496444e-01
-2.00765416e-01 -4.77610230e-01 8.35511088e-01 8.14880490e-01
-6.75690949e-01 2.48429067e-02 -5.64545989e-01 1.98438779e-01
9.45009708e-01 5.00261724e-01 1.07566237e+00 -1.28015161e+00
-2.43678227e-01 -2.78475583e-01 -3.41485627e-02 -1.06801987e+00
3.28830063e-01 -7.91322410e-01 6.12695776e-02 -1.40245736e+00
-4.92804646e-02 -6.08684540e-01 2.96030581e-01 3.55999142e-01
3.40359882e-02 9.50822473e-01 7.27393106e-02 1.67224735e-01
4.53976681e-03 9.00250673e-01 1.41783524e+00 1.80904508e-01
-1.95664123e-01 -3.51569317e-02 -4.84631419e-01 8.68475080e-01
5.37977755e-01 -1.12999007e-01 -3.15630496e-01 -6.21564686e-01
3.53397369e-01 -1.04780369e-01 5.76028466e-01 -8.82078707e-01
-9.36186388e-02 -1.14110641e-01 5.35127699e-01 -3.01173329e-01
9.82065022e-01 -9.35258746e-01 3.29978853e-01 3.22113752e-01
3.08188587e-01 -4.54620898e-01 5.87996505e-02 4.22856271e-01
6.01918250e-02 4.57179733e-02 1.17262399e+00 -2.25159630e-01
-1.16759650e-01 5.60562015e-01 2.13378862e-01 -3.39970469e-01
6.49602532e-01 -5.43742597e-01 -6.73764870e-02 -6.26603186e-01
-3.44299525e-01 -3.21182102e-01 1.03075707e+00 5.73620088e-02
8.24534595e-01 -1.29566848e+00 -8.39503825e-01 6.81420624e-01
-1.38410121e-01 -1.75606995e-03 2.00060755e-01 8.33640635e-01
-8.55980337e-01 8.95313621e-02 -2.71113496e-02 -8.43262494e-01
-1.22344840e+00 3.70763183e-01 4.80029494e-01 2.34119594e-01
-9.62066352e-01 6.24526680e-01 6.60610914e-01 -6.32828772e-01
6.37543481e-03 -1.23083174e-01 -1.65339317e-02 -1.58462688e-01
5.21702707e-01 3.35499763e-01 1.23447731e-01 -7.06523776e-01
-1.50079906e-01 1.14739478e+00 4.37187135e-01 -2.23594140e-02
1.57797527e+00 -4.14579302e-01 -2.79170990e-01 2.15551659e-01
1.08864212e+00 4.13866669e-01 -1.68025923e+00 -1.93687961e-01
-7.28480697e-01 -9.55146492e-01 2.63553262e-01 -6.03312016e-01
-1.47472751e+00 1.07215989e+00 4.33190346e-01 -1.56075567e-01
1.21850300e+00 -1.56834215e-01 7.58275390e-01 8.53939652e-02
6.04690850e-01 -5.71758151e-01 1.77135080e-01 3.67186219e-01
1.06307948e+00 -1.29806578e+00 2.65192568e-01 -8.47595334e-01
-1.62481114e-01 1.28500569e+00 3.62695694e-01 -1.98464021e-02
5.80640793e-01 2.17597470e-01 2.20885158e-01 -1.62432343e-01
-2.86297143e-01 -3.16108525e-01 3.96564931e-01 6.76352799e-01
3.76674950e-01 -1.79514259e-01 1.39291719e-01 -1.36385024e-01
-2.45888948e-01 -2.45072514e-01 4.65576649e-01 4.31811571e-01
-1.24912344e-01 -1.03345144e+00 -6.56550109e-01 -4.69170921e-02
-3.23434174e-01 -1.14149868e-01 -3.13162096e-02 5.88679135e-01
-7.03021735e-02 8.22021604e-01 -1.08300067e-01 9.72784236e-02
1.97291896e-01 -1.78958669e-01 9.23631191e-01 -5.99740207e-01
-3.82390469e-01 1.80084020e-01 -1.14921056e-01 -8.50062847e-01
-3.78192157e-01 -4.82705116e-01 -1.00125873e+00 -4.27699655e-01
-4.42031801e-01 -4.01732922e-01 1.03890920e+00 7.15665877e-01
3.22173744e-01 2.34865606e-01 1.00722969e+00 -1.34058571e+00
-1.27588168e-01 -6.22770786e-01 -6.11998677e-01 3.76797080e-01
5.21778822e-01 -7.02621400e-01 -3.48839223e-01 -1.60999551e-01] | [9.907610893249512, -2.9204137325286865] |
e49d4253-fb0c-4f48-b125-93c45d8c03a4 | chinese-zero-pronoun-resolution-an | null | null | https://aclanthology.org/D14-1084 | https://aclanthology.org/D14-1084.pdf | Chinese Zero Pronoun Resolution: An Unsupervised Probabilistic Model Rivaling Supervised Resolvers | null | ['Vincent Ng', 'Chen Chen'] | 2014-10-01 | null | null | null | emnlp-2014-10 | ['chinese-zero-pronoun-resolution'] | ['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] | [-7.1648030281066895, 3.7107908725738525] |
72af4293-48e6-43fd-8dc0-168bf0b34bd0 | enhancing-opinion-role-labeling-with-semantic | null | null | https://aclanthology.org/N19-1066 | https://aclanthology.org/N19-1066.pdf | Enhancing Opinion Role Labeling with Semantic-Aware Word Representations from Semantic Role Labeling | Opinion role labeling (ORL) is an important task for fine-grained opinion mining, which identifies important opinion arguments such as holder and target for a given opinion trigger. The task is highly correlative with semantic role labeling (SRL), which identifies important semantic arguments such as agent and patient for a given predicate. As predicate agents and patients usually correspond to opinion holders and targets respectively, SRL could be valuable for ORL. In this work, we propose a simple and novel method to enhance ORL by utilizing SRL, presenting semantic-aware word representations which are learned from SRL. The representations are then fed into a baseline neural ORL model as basic inputs. We verify the proposed method on a benchmark MPQA corpus. Experimental results show that the proposed method is highly effective. In addition, we compare the method with two representative methods of SRL integration as well, finding that our method can outperform the two methods significantly, achieving 1.47{\%} higher F-scores than the better one. | ['Peili Liang', 'Meishan Zhang', 'Guohong Fu'] | 2019-06-01 | null | null | null | naacl-2019-6 | ['fine-grained-opinion-analysis'] | ['natural-language-processing'] | [ 5.01052439e-01 4.57227468e-01 -4.80815619e-01 -6.51476562e-01
-8.19360018e-01 -7.34582365e-01 6.78755760e-01 6.59915626e-01
-4.65926915e-01 1.06217456e+00 7.16416419e-01 -2.28038624e-01
-1.05028667e-01 -7.90669262e-01 -5.57768583e-01 -6.54019356e-01
4.78666395e-01 6.06454432e-01 3.49820822e-01 -5.90700150e-01
4.06722724e-01 1.79453924e-01 -1.66615081e+00 7.20575929e-01
9.45859611e-01 1.32423258e+00 -2.69438088e-01 1.11650482e-01
-3.53657037e-01 1.21269107e+00 -9.98931468e-01 -6.24633491e-01
-2.61586517e-01 -3.96365404e-01 -1.07629335e+00 -2.08834007e-01
-6.54935837e-03 3.62755626e-01 5.83542645e-01 1.07116008e+00
4.95419830e-01 2.61240870e-01 7.05326557e-01 -1.19214964e+00
-8.07201743e-01 8.77502501e-01 -5.76937854e-01 3.10075611e-01
4.04636979e-01 -6.18548512e-01 1.68866491e+00 -8.12972248e-01
5.38693190e-01 1.70630014e+00 2.44893610e-01 4.31910306e-01
-5.98753452e-01 -5.19280851e-01 7.65807807e-01 2.64054537e-01
-7.36410975e-01 -3.22509319e-01 8.68692935e-01 3.99758816e-02
8.80249918e-01 7.72938877e-02 4.70239580e-01 7.88225889e-01
4.92660748e-03 1.13213468e+00 1.14676154e+00 -4.98846650e-01
5.24773538e-01 2.78116405e-01 8.13760698e-01 4.84539419e-01
4.70468223e-01 -4.31821555e-01 -6.58853769e-01 -5.39079607e-01
1.75084531e-01 -1.52529627e-01 -2.77095526e-01 6.03147820e-02
-1.04200065e+00 1.20142436e+00 5.01340091e-01 5.43277264e-01
-5.80365598e-01 7.70571008e-02 4.21544760e-01 2.11765140e-01
8.45372379e-01 7.48002708e-01 -8.72594953e-01 1.31406486e-01
-3.79962713e-01 3.31264466e-01 8.13739240e-01 3.94393086e-01
4.46839064e-01 -3.94319445e-01 -5.13166726e-01 7.97357440e-01
5.12922585e-01 3.96358997e-01 5.78468263e-01 -6.71174347e-01
2.81160027e-01 1.21287608e+00 3.46706033e-01 -9.98071969e-01
-3.58049810e-01 -3.57030958e-01 -5.45457959e-01 -1.40946165e-01
-1.13523249e-02 -1.02116473e-01 -8.36779892e-01 1.63375127e+00
5.36370158e-01 2.47169118e-02 6.20479405e-01 7.77028382e-01
1.25519001e+00 8.68817687e-01 3.52001727e-01 -4.75294948e-01
1.84514821e+00 -1.05644429e+00 -9.96648192e-01 -3.32781106e-01
4.44909424e-01 -6.59610271e-01 9.92593646e-01 2.54827350e-01
-7.71515846e-01 -3.45645607e-01 -8.35178494e-01 2.56606694e-02
-5.06502569e-01 1.18618578e-01 8.71223569e-01 3.65986407e-01
-7.86484838e-01 3.43757004e-01 -3.44482362e-01 -2.27571264e-01
4.69544888e-01 3.75909984e-01 -1.31218746e-01 1.46041855e-01
-1.69419205e+00 7.39135146e-01 4.95688379e-01 -1.70370396e-02
-4.19246703e-01 -5.91465652e-01 -1.15687943e+00 1.26344822e-02
4.62885320e-01 -6.22961164e-01 1.43535709e+00 -1.16724944e+00
-1.28883410e+00 8.82067084e-01 -7.89526224e-01 -5.95670044e-01
-2.02467263e-01 -5.22379339e-01 -5.99638343e-01 2.25899175e-01
5.90093493e-01 5.23509562e-01 9.43395138e-01 -1.33705306e+00
-9.93498147e-01 -4.73005354e-01 7.70987093e-01 5.03399909e-01
-2.56761432e-01 2.90332466e-01 -1.09272584e-01 -7.63581038e-01
2.33593266e-02 -4.44230318e-01 -3.44434202e-01 -4.94810045e-01
-2.78209239e-01 -1.17125595e+00 7.17720151e-01 -3.05368096e-01
1.25341058e+00 -1.75729585e+00 -3.72321680e-02 2.16268003e-01
8.37749802e-03 2.19445959e-01 -8.36984068e-02 1.79973453e-01
-2.39630491e-01 1.98392883e-01 -3.80975544e-01 2.88790874e-02
2.28577331e-02 2.88338095e-01 -5.91545463e-01 -2.87355972e-03
4.40405071e-01 1.09096384e+00 -1.04696524e+00 -3.67295325e-01
-1.61977291e-01 4.82975423e-01 -2.40149170e-01 2.25001544e-01
-5.06403506e-01 2.92143941e-01 -7.30190277e-01 8.90238643e-01
3.03984821e-01 -4.90898073e-01 2.48213917e-01 -5.33148646e-01
2.67506510e-01 6.37036622e-01 -9.51405942e-01 1.07865703e+00
-4.64101911e-01 -2.87392586e-02 -4.49244305e-02 -1.04365647e+00
1.07167757e+00 3.80019277e-01 1.47719175e-01 -6.99888408e-01
3.42726260e-01 3.62619102e-01 -2.10597198e-02 -2.87696004e-01
3.92376482e-01 -4.46399450e-01 -2.63748139e-01 6.42869115e-01
-2.08950460e-01 -1.20007746e-01 4.31373328e-01 2.34354630e-01
8.04993570e-01 -1.03930168e-01 9.21942294e-01 -4.96285677e-01
1.16357505e+00 -3.86470445e-02 6.74860179e-01 5.10029197e-01
-6.03523813e-02 1.56206310e-01 8.04595828e-01 -3.96977246e-01
-2.84681141e-01 -6.03555143e-01 1.57279000e-01 1.30706251e+00
4.51882005e-01 -3.44000787e-01 -5.29021978e-01 -1.41955280e+00
-9.61774553e-04 8.62991393e-01 -8.96333396e-01 8.48651081e-02
-6.38564289e-01 -9.47909832e-01 1.59845859e-01 7.61357486e-01
3.30793023e-01 -1.46708131e+00 -3.96385372e-01 1.87578291e-01
-1.82687089e-01 -9.54668820e-01 -1.61905922e-02 2.72436291e-01
-5.30236363e-01 -1.28715372e+00 -4.60127801e-01 -8.98789704e-01
1.03995812e+00 1.91634402e-01 1.46912277e+00 1.54691234e-01
3.73254448e-01 2.68050525e-02 -8.96055102e-01 -6.23325169e-01
-2.21715093e-01 2.91621033e-02 2.45528016e-03 2.81738341e-01
5.83081782e-01 -1.67909890e-01 -7.00005054e-01 2.10666358e-01
-9.08594728e-01 -3.57222706e-01 6.24351859e-01 8.98032904e-01
8.25700998e-01 2.17925102e-01 9.74578440e-01 -1.52654684e+00
1.16959941e+00 -5.07369637e-01 -4.07450944e-01 3.84425908e-01
-6.82299912e-01 3.64961386e-01 5.99504530e-01 1.05122559e-01
-1.37038207e+00 -4.06815976e-01 -4.87523466e-01 3.25349629e-01
-1.46243006e-01 6.21927261e-01 -3.35748404e-01 3.72726589e-01
4.34704542e-01 -2.52944767e-01 -2.74222732e-01 -3.62182140e-01
4.64447886e-01 6.09438539e-01 7.08231702e-02 -6.14164591e-01
5.40880382e-01 7.15436697e-01 -3.39204997e-01 -3.88785511e-01
-1.74186039e+00 -5.84328771e-01 -2.25575075e-01 2.79542625e-01
7.23721087e-01 -8.39448571e-01 -6.72381222e-01 1.11134708e-01
-1.23185444e+00 4.03400779e-01 -4.34072882e-01 1.89015791e-01
-1.13177195e-01 1.99612007e-01 -3.61087710e-01 -8.55369925e-01
-7.89551198e-01 -1.02903509e+00 1.41073942e+00 4.94827896e-01
-4.28345978e-01 -1.12186277e+00 5.72064109e-02 6.89914823e-01
2.15416610e-01 1.64547697e-01 1.08138347e+00 -1.29201901e+00
1.03946850e-01 -1.48379356e-01 -3.41882586e-01 4.25303996e-01
4.38715696e-01 -3.58209044e-01 -8.80442798e-01 -1.43150166e-01
1.62822455e-01 -3.97536427e-01 1.14279962e+00 3.30473781e-01
1.03329515e+00 -4.65093434e-01 -3.22126508e-01 -1.52455345e-01
1.11638165e+00 4.26357657e-01 4.32472408e-01 3.21344286e-01
5.09259224e-01 8.77295554e-01 1.25449443e+00 1.71250269e-01
4.07097340e-01 3.98733795e-01 4.44686890e-01 -2.10977599e-01
1.38580978e-01 -1.11319348e-01 4.81449872e-01 5.84963560e-01
-2.41669461e-01 -4.63027507e-01 -4.63033497e-01 6.27837241e-01
-2.18122745e+00 -8.26592684e-01 -7.73243532e-02 1.56812429e+00
9.30691719e-01 2.45058492e-01 -1.71048075e-01 3.34928840e-01
6.70855343e-01 5.09661257e-01 -6.22361600e-01 -4.28060442e-01
-3.10843587e-01 7.71584809e-01 1.72699988e-02 6.81308806e-01
-1.34800231e+00 1.16794097e+00 5.55221987e+00 9.90200400e-01
-5.91620266e-01 2.58868337e-01 8.49143147e-01 4.11057055e-01
-7.84455359e-01 -1.48237661e-01 -1.11047769e+00 3.63019742e-02
6.42057955e-01 -1.92983612e-01 -1.89952686e-01 9.44986761e-01
1.75566897e-01 1.14910919e-02 -8.01435113e-01 6.44730151e-01
3.17348659e-01 -1.45907187e+00 4.82918322e-01 -4.22723979e-01
9.27639127e-01 -5.14180839e-01 -2.91414589e-01 3.53065372e-01
4.47177172e-01 -1.02033043e+00 5.14043391e-01 5.32063935e-03
2.38092884e-01 -9.11654592e-01 1.42873740e+00 -2.81974673e-02
-1.33066857e+00 -3.35523635e-02 -3.80474985e-01 -3.02822322e-01
3.21994990e-01 7.87855685e-01 -5.82008779e-01 6.19191170e-01
5.87727308e-01 1.05588925e+00 -4.02674884e-01 3.04803759e-01
-1.22506654e+00 7.35239565e-01 1.32134020e-01 -4.08225626e-01
3.51059943e-01 3.52010690e-02 3.37694079e-01 8.84143710e-01
7.18486905e-02 3.12799931e-01 9.93058980e-02 5.43239772e-01
-3.61224294e-01 4.57510293e-01 -4.80690986e-01 -2.69427210e-01
3.64480913e-01 1.48896611e+00 -9.29252207e-01 -5.51200330e-01
-2.64419824e-01 8.99135828e-01 1.18498541e-01 2.57695079e-01
-7.58791208e-01 -3.07033002e-01 8.02969337e-01 -1.49411440e-01
2.08237916e-01 5.64138234e-01 -4.01838645e-02 -1.01200962e+00
5.47565222e-02 -1.05944836e+00 7.91679263e-01 -6.30801916e-01
-1.63730776e+00 9.36964333e-01 -1.48382649e-01 -8.84407103e-01
-2.74701744e-01 -9.10634160e-01 -5.34051359e-01 6.26176000e-01
-2.07896876e+00 -1.39159179e+00 -5.09920791e-02 5.07247984e-01
8.37599277e-01 -1.59594595e-01 8.73079717e-01 -2.64705978e-02
-1.94343865e-01 2.59963036e-01 -5.41441560e-01 -1.06180184e-01
4.45181131e-01 -1.37492967e+00 2.34703451e-01 7.48861313e-01
1.34571642e-01 8.04154098e-01 4.96953785e-01 -6.75001025e-01
-7.87130415e-01 -1.11849177e+00 1.27743375e+00 -5.06023228e-01
7.07349300e-01 -4.55988683e-02 -6.72315359e-01 3.69434983e-01
4.02296454e-01 -2.98617214e-01 9.91092563e-01 2.13727400e-01
-3.93713623e-01 -9.01103765e-02 -1.23441935e+00 3.23396564e-01
9.01022613e-01 -3.09142470e-01 -1.22335565e+00 4.19089347e-01
1.44857740e+00 -6.79820105e-02 -5.97659051e-01 8.38065326e-01
2.97950476e-01 -8.35160673e-01 8.91890407e-01 -7.86914170e-01
5.24351776e-01 -7.82023907e-01 -2.20282063e-01 -1.27415705e+00
-6.20326363e-02 -1.60694346e-01 -1.82722926e-01 1.26524234e+00
8.25333416e-01 -7.60738671e-01 5.62575579e-01 1.68872520e-01
-4.37691696e-02 -1.03345835e+00 -6.92743480e-01 -2.73450464e-01
-1.88364074e-01 -2.82555282e-01 8.02941740e-01 8.03789496e-01
-1.36210725e-01 9.57571387e-01 -1.32189458e-02 1.95043951e-01
3.05708259e-01 5.99104762e-01 7.99349099e-02 -1.58878922e+00
-1.31497145e-01 -3.70396525e-01 -4.04509157e-03 -8.59704316e-01
6.13181829e-01 -8.16665888e-01 -6.29483834e-02 -2.04165721e+00
1.49284795e-01 -2.44496584e-01 -8.45385849e-01 7.14370191e-01
-6.69162452e-01 2.63501137e-01 -1.89585716e-01 -8.08579847e-02
-8.15613687e-01 5.44408023e-01 1.16989994e+00 -3.80005181e-01
-1.66370645e-01 2.75578409e-01 -1.58164120e+00 1.21628702e+00
8.91762316e-01 -2.61197805e-01 -6.99706793e-01 -1.94687113e-01
7.23776937e-01 -4.70574886e-01 -3.33436728e-02 -3.70148152e-01
-1.24195293e-01 -1.82105396e-02 5.25862351e-02 -5.52720487e-01
2.49592587e-01 -6.69376493e-01 -5.16932309e-01 3.58889788e-01
-6.24027550e-01 6.94246143e-02 -4.75867316e-02 5.55739462e-01
-7.11060822e-01 -3.54359031e-01 4.81535017e-01 -2.34146014e-01
-1.13753223e+00 -5.22374250e-02 5.24264611e-02 2.30273932e-01
9.14587259e-01 2.65926242e-01 -6.48636162e-01 -3.41024518e-01
-6.48467362e-01 3.57479692e-01 -4.28494066e-03 6.89656734e-01
7.69463778e-01 -1.18432724e+00 -7.14600801e-01 -1.73977446e-02
3.59141141e-01 1.46826714e-01 -1.26186192e-01 5.76412261e-01
-1.99966505e-01 4.46597457e-01 3.19091648e-01 -4.56611700e-02
-1.43335593e+00 3.63397598e-01 6.98276935e-03 -7.50873566e-01
-7.86785856e-02 1.26315093e+00 3.83808911e-01 -5.53817451e-01
1.33165956e-01 -3.56722534e-01 -1.15349507e+00 4.98969823e-01
7.50657976e-01 -4.88856956e-02 2.49814857e-02 -8.31069589e-01
-6.95074618e-01 6.35209560e-01 -1.57396436e-01 3.39478590e-02
1.37638581e+00 1.64927632e-01 -7.95843363e-01 3.69277507e-01
8.10091376e-01 3.22418749e-01 -4.41166997e-01 -1.66773483e-01
2.31354386e-01 -1.11003146e-02 -2.50903428e-01 -9.84699905e-01
-1.00115132e+00 5.83556235e-01 2.57940859e-01 4.11183596e-01
1.15876567e+00 5.86799085e-01 7.88766861e-01 3.15292031e-01
4.01123047e-01 -8.45671594e-01 1.72985137e-01 5.80000818e-01
1.07302749e+00 -1.26672733e+00 -9.22430754e-02 -7.36122608e-01
-8.71387124e-01 7.06160545e-01 6.46918893e-01 -3.80210206e-02
5.19536555e-01 -2.16623947e-01 3.18199664e-01 -6.59081578e-01
-8.17221582e-01 -6.18648052e-01 4.19791728e-01 3.87378305e-01
7.65020728e-01 8.86176080e-02 -8.90900254e-01 9.77919877e-01
-1.51565969e-01 -3.24526638e-01 2.23454069e-02 9.70062852e-01
-5.77025354e-01 -1.38364661e+00 -2.81208158e-01 4.25092995e-01
-7.45648324e-01 -3.46275300e-01 -7.36954749e-01 3.79287779e-01
3.37009668e-01 1.39035320e+00 -2.91489121e-02 -4.51110862e-02
3.97801876e-01 5.82954660e-02 2.13835724e-02 -9.17990386e-01
-8.15603316e-01 -1.59926251e-01 4.93650526e-01 -6.47281528e-01
-1.11334920e+00 -2.15084136e-01 -1.61915314e+00 4.03861016e-01
-3.80021006e-01 7.45752811e-01 2.96430737e-01 1.20742166e+00
3.87006909e-01 7.26194739e-01 5.16887844e-01 -3.17955285e-01
-2.79580355e-01 -8.03525805e-01 -3.83437425e-01 7.20266581e-01
1.32269561e-01 -7.81655729e-01 -4.07149196e-01 -1.64163619e-01] | [11.445561408996582, 6.709410190582275] |
6fc5452d-b15c-4fca-bb62-279073face32 | unsupervised-recurrent-neural-network | 1904.03746 | null | https://arxiv.org/abs/1904.03746v6 | https://arxiv.org/pdf/1904.03746v6.pdf | Unsupervised Recurrent Neural Network Grammars | Recurrent neural network grammars (RNNG) are generative models of language which jointly model syntax and surface structure by incrementally generating a syntax tree and sentence in a top-down, left-to-right order. Supervised RNNGs achieve strong language modeling and parsing performance, but require an annotated corpus of parse trees. In this work, we experiment with unsupervised learning of RNNGs. Since directly marginalizing over the space of latent trees is intractable, we instead apply amortized variational inference. To maximize the evidence lower bound, we develop an inference network parameterized as a neural CRF constituency parser. On language modeling, unsupervised RNNGs perform as well their supervised counterparts on benchmarks in English and Chinese. On constituency grammar induction, they are competitive with recent neural language models that induce tree structures from words through attention mechanisms. | ['Gábor Melis', 'Alexander M. Rush', 'Lei Yu', 'Chris Dyer', 'Yoon Kim', 'Adhiguna Kuncoro'] | 2019-04-07 | unsupervised-recurrent-neural-network-1 | https://aclanthology.org/N19-1114 | https://aclanthology.org/N19-1114.pdf | naacl-2019-6 | ['constituency-grammar-induction'] | ['natural-language-processing'] | [ 3.19380283e-01 8.53572249e-01 -1.84931502e-01 -7.77550876e-01
-1.07627141e+00 -6.46363556e-01 5.74572444e-01 -1.91992074e-01
-3.77716988e-01 6.77613854e-01 4.98537958e-01 -9.63416576e-01
5.97141623e-01 -1.10801566e+00 -9.80140984e-01 -3.59240174e-01
-1.64117347e-02 8.13521683e-01 -3.69576365e-01 4.38596262e-03
-1.32662490e-01 4.10876609e-02 -1.03280020e+00 2.01539621e-01
7.81269014e-01 2.06926957e-01 5.04463613e-01 9.33476150e-01
-6.21005237e-01 1.09567630e+00 -2.78622240e-01 -3.14908773e-01
-2.10421458e-01 -5.37641525e-01 -8.86475980e-01 3.88273410e-02
3.46868426e-01 -3.94835830e-01 -7.38388044e-04 1.06013024e+00
-1.80560760e-02 2.75183585e-04 6.72176182e-01 -6.60922587e-01
-8.78156662e-01 1.43679500e+00 -2.06786290e-01 9.78967473e-02
-1.75800249e-01 -6.39799088e-02 1.78070760e+00 -9.16784048e-01
7.23417997e-01 1.58953655e+00 8.75314251e-02 8.48065376e-01
-1.75602531e+00 -4.36913729e-01 7.38329589e-01 -5.95797479e-01
-9.34084773e-01 -3.60193700e-01 7.43930936e-01 -3.44619006e-01
1.53780603e+00 -1.60578743e-01 1.75312504e-01 1.26624775e+00
1.42562851e-01 1.01241851e+00 8.18487465e-01 -7.33289659e-01
3.26480091e-01 -2.03022033e-01 5.18803120e-01 9.08454478e-01
1.43995941e-01 -1.19156070e-01 -2.67592847e-01 -1.59508526e-01
7.89114952e-01 -3.94646645e-01 3.50370079e-01 -1.36524290e-01
-7.28347600e-01 1.38540363e+00 3.21209654e-02 8.44592154e-02
-3.66864204e-01 7.19135582e-01 2.05995277e-01 -8.15701634e-02
5.94910562e-01 1.12056375e-01 -5.70679665e-01 1.71295591e-02
-7.90227115e-01 2.96761960e-01 8.28432441e-01 1.18473589e+00
8.65837693e-01 2.20686957e-01 -4.90739569e-02 8.84092033e-01
6.95873082e-01 5.15871704e-01 1.72966644e-01 -9.96445894e-01
7.23012328e-01 3.29914153e-01 -2.70697922e-01 -2.66212821e-01
-2.84308374e-01 -2.61043966e-01 -6.01015568e-01 -5.92463836e-02
2.28012085e-01 -5.19711852e-01 -1.28819919e+00 2.37088609e+00
-1.84214581e-02 -1.27688155e-01 4.04143363e-01 3.51489246e-01
6.37392759e-01 1.11605036e+00 7.04490125e-01 -2.61229277e-01
1.32492232e+00 -7.28498995e-01 -6.74231470e-01 -7.42246926e-01
7.99279511e-01 -4.16719913e-01 1.19269741e+00 1.89502671e-01
-1.44382012e+00 -3.41268390e-01 -5.29932439e-01 -4.49325919e-01
-7.35418126e-02 2.70095497e-01 7.61020064e-01 6.39914691e-01
-1.23265910e+00 3.50602388e-01 -1.42765117e+00 5.18799238e-02
2.71269023e-01 1.86092585e-01 -5.09767197e-02 1.27270073e-01
-9.69746292e-01 5.54549873e-01 5.96461117e-01 1.39612824e-01
-1.10851872e+00 -3.62946361e-01 -1.30532062e+00 1.49290144e-01
2.36212283e-01 -6.18879020e-01 1.68295074e+00 -6.05831742e-01
-1.65704525e+00 9.61224318e-01 -5.25251806e-01 -6.65673733e-01
-2.38470301e-01 -3.31693739e-01 3.51613015e-02 -3.30191374e-01
2.08139747e-01 1.05479217e+00 4.80804771e-01 -1.11386800e+00
-5.72226286e-01 -1.65079698e-01 -2.38275863e-02 -4.93874699e-02
2.16175735e-01 1.60013691e-01 -3.24306875e-01 -5.85475147e-01
3.22186261e-01 -9.02237773e-01 -5.51706076e-01 -9.00841653e-01
-7.57905185e-01 -8.38817120e-01 4.42795426e-01 -7.44856119e-01
1.14776385e+00 -1.77750707e+00 2.34125838e-01 9.39652771e-02
5.62283881e-02 -2.04744071e-01 -2.10709617e-01 1.88400432e-01
2.74108462e-02 4.96552616e-01 -5.08329332e-01 -8.78336489e-01
3.23675752e-01 8.67597640e-01 -6.66881621e-01 -1.45902276e-01
5.88671863e-01 1.16936541e+00 -7.88422704e-01 -4.93715286e-01
-1.64478719e-01 3.01272780e-01 -1.09712934e+00 3.69758576e-01
-9.24268961e-01 9.53224376e-02 -3.47713411e-01 3.95596445e-01
4.15376484e-01 -4.22228009e-01 8.11439991e-01 2.62062758e-01
-3.18159200e-02 1.19417858e+00 -6.80425584e-01 1.68249440e+00
-8.94635856e-01 5.06727934e-01 -1.89328343e-02 -8.34728479e-01
8.71476769e-01 2.82515109e-01 -3.52350414e-01 -2.40417391e-01
1.35919936e-02 7.72718340e-02 8.99296626e-02 -1.67995170e-01
4.34109032e-01 -3.13986301e-01 -5.35068512e-01 6.04518533e-01
5.35180688e-01 -6.61817566e-02 3.54001373e-01 5.16547084e-01
9.27985370e-01 3.71779233e-01 8.81340355e-02 -2.15958744e-01
-1.04935402e-02 -1.62614629e-01 6.88605666e-01 9.54024732e-01
3.49832624e-01 3.01275402e-01 9.02260065e-01 -3.29890370e-01
-1.10034704e+00 -1.55247998e+00 -4.06731525e-03 1.67605734e+00
-5.95624983e-01 -7.12908924e-01 -9.32217181e-01 -8.18126380e-01
-4.26512301e-01 1.36562419e+00 -4.34253722e-01 2.88590699e-01
-1.18278015e+00 -8.69046032e-01 6.22846067e-01 8.86778533e-01
-8.53511244e-02 -1.62518072e+00 -4.73769218e-01 6.48906410e-01
-1.76337928e-01 -1.36364329e+00 -3.86301100e-01 7.60515392e-01
-1.08864093e+00 -5.73566198e-01 -1.31478131e-01 -1.06997931e+00
7.51729310e-01 -7.11157680e-01 1.69389009e+00 -1.95802286e-01
-7.07325935e-02 4.67087002e-03 6.33042678e-03 -4.30783302e-01
-9.08647180e-01 3.05953950e-01 -6.37560040e-02 -4.77922916e-01
4.95855719e-01 -7.85926163e-01 9.54019837e-03 -4.56556678e-01
-7.04903901e-01 3.01808327e-01 5.96594334e-01 8.71562839e-01
8.00370634e-01 -4.99709815e-01 3.67223442e-01 -1.34513271e+00
6.26901746e-01 -3.93268257e-01 -1.08069348e+00 2.03482077e-01
-4.63239431e-01 7.11071849e-01 6.56374514e-01 -2.86573991e-02
-1.41068268e+00 2.35561490e-01 -4.41623926e-01 3.30922455e-02
-4.74837929e-01 5.31714916e-01 -2.40303978e-01 9.39316690e-01
2.92844176e-01 2.42496565e-01 -3.90133113e-01 -6.00608766e-01
8.83827329e-01 1.37925535e-01 6.02517426e-01 -1.04808748e+00
5.48664272e-01 1.17977805e-01 -1.59154028e-01 -6.50634587e-01
-1.03494906e+00 5.31417951e-02 -6.27350867e-01 5.95281482e-01
1.23654628e+00 -9.36460316e-01 -4.20536250e-01 -6.36564419e-02
-1.74217868e+00 -6.40872180e-01 -2.30209351e-01 4.21907395e-01
-5.39838970e-01 4.00676206e-02 -9.89521146e-01 -1.04417920e+00
-4.20190185e-01 -9.72167611e-01 1.18311572e+00 9.99666527e-02
-3.80807072e-01 -1.14456153e+00 4.13763136e-01 -1.47980928e-01
1.01025395e-01 -4.07121964e-02 1.44034088e+00 -5.68191111e-01
-8.63534868e-01 3.14730436e-01 -1.86413471e-02 3.85161519e-01
4.71388325e-02 5.29225841e-02 -8.86385262e-01 9.93570015e-02
-3.95835131e-01 -5.27901769e-01 1.05374241e+00 7.90029943e-01
9.27259088e-01 -4.74223077e-01 -2.33944744e-01 6.17561638e-01
1.23361647e+00 9.89440754e-02 3.68486136e-01 -2.59407908e-01
7.62657046e-01 6.35817230e-01 -6.29453287e-02 1.05147697e-01
6.47799671e-01 9.44594666e-02 3.16274613e-01 1.41059337e-02
2.18082905e-01 -7.42069364e-01 6.42072201e-01 9.39194202e-01
6.30795881e-02 -4.69895601e-01 -9.07772243e-01 6.73079848e-01
-1.55981815e+00 -6.08608842e-01 -1.09446518e-01 1.66129506e+00
1.08664429e+00 4.09324497e-01 -1.72246441e-01 -5.18608987e-01
6.10199273e-01 3.04767847e-01 -2.16823071e-01 -9.00215387e-01
-1.46768689e-01 7.85356998e-01 3.48858684e-01 9.75389481e-01
-9.91087198e-01 1.71082163e+00 7.02208090e+00 4.26018268e-01
-7.99667060e-01 8.03837255e-02 8.65763783e-01 -5.91895059e-02
-8.13543618e-01 4.02064562e-01 -1.41957343e+00 2.54268609e-02
1.29771936e+00 2.88808256e-01 4.32956904e-01 1.01341844e+00
5.36771417e-02 1.35544300e-01 -1.33886695e+00 5.22251904e-01
-3.72031987e-01 -1.44832921e+00 1.82977363e-01 2.22700357e-01
6.69852436e-01 5.39478600e-01 -1.19621173e-01 6.55056000e-01
1.38262117e+00 -1.34386158e+00 6.26860976e-01 8.30911770e-02
5.63811183e-01 -6.09180868e-01 2.87913144e-01 6.16511345e-01
-1.12534189e+00 3.51452917e-01 -4.87738043e-01 -1.73282593e-01
7.15916216e-01 4.46302205e-01 -9.17754412e-01 -1.76666062e-02
3.73187214e-01 1.35388151e-01 -2.12299958e-01 -2.28171825e-01
-9.11241829e-01 1.40429711e+00 -5.09018719e-01 -1.61164373e-01
5.56358635e-01 -3.36328030e-01 3.59784245e-01 1.47874582e+00
1.86694916e-02 2.05181897e-01 3.88229370e-01 1.64400458e+00
-4.03986901e-01 3.11443470e-02 -6.28783107e-01 -3.97312403e-01
4.23905462e-01 1.20358682e+00 -7.72169650e-01 -6.10815883e-01
-3.15420538e-01 6.79909647e-01 8.47921014e-01 4.08486366e-01
-6.49383843e-01 -1.03941463e-01 4.45983052e-01 -1.73985183e-01
5.43005109e-01 -5.32124519e-01 -3.51717293e-01 -1.12146521e+00
-6.22112267e-02 -5.75989068e-01 3.51060480e-01 -5.11669397e-01
-1.06991625e+00 7.52474546e-01 1.37465045e-01 -1.29258111e-01
-9.68462527e-01 -7.85740733e-01 -8.88126016e-01 1.18940890e+00
-1.38243461e+00 -1.14953637e+00 5.39951921e-01 8.78731534e-02
8.15668881e-01 -4.74884398e-02 1.14367068e+00 -2.26008818e-01
-6.82779253e-01 3.44427407e-01 -3.04720610e-01 6.33879602e-01
-9.98208299e-02 -1.76954055e+00 1.29213929e+00 1.07533014e+00
7.59729624e-01 1.08669043e+00 5.70398688e-01 -7.36193240e-01
-1.03943932e+00 -1.23469162e+00 1.33238435e+00 -6.65698409e-01
5.74446917e-01 -9.68661904e-01 -1.04503942e+00 1.38873696e+00
4.90931809e-01 6.22091517e-02 5.90758801e-01 6.26675367e-01
-4.25594479e-01 5.60575902e-01 -4.49114412e-01 6.11880839e-01
1.04389489e+00 -6.34259760e-01 -7.88839757e-01 2.38872185e-01
9.73737001e-01 -3.15966338e-01 -4.58924711e-01 2.26300493e-01
2.61187077e-01 -5.92475593e-01 6.13466024e-01 -1.01352143e+00
6.73784733e-01 6.38277084e-02 -4.16885287e-01 -1.12971818e+00
-2.74130404e-01 -7.95626998e-01 -2.31652454e-01 1.33380699e+00
9.71581757e-01 -2.65400827e-01 1.07907403e+00 5.70614159e-01
-1.52470440e-01 -6.12864852e-01 -5.82650840e-01 -6.17940366e-01
7.11137533e-01 -7.48414278e-01 4.58330035e-01 5.03010809e-01
-1.95887908e-01 9.32570577e-01 -1.39020070e-01 2.74580300e-01
7.30105519e-01 1.23834953e-01 5.52176535e-01 -1.07651556e+00
-6.83459401e-01 -3.56288463e-01 3.99300754e-01 -1.43646455e+00
8.21619451e-01 -1.20124292e+00 5.25985599e-01 -1.56310833e+00
1.22029975e-01 -2.79934943e-01 -1.16957679e-01 6.07414842e-01
-3.43882710e-01 -2.22381547e-01 -9.98136327e-02 -2.69740999e-01
-4.88061875e-01 4.01969641e-01 6.82257950e-01 1.29889414e-01
-2.65725195e-01 -2.57093042e-01 -6.08614862e-01 1.03069592e+00
8.81583393e-01 -7.79801786e-01 -5.09973884e-01 -7.14615226e-01
4.34440851e-01 3.59450817e-01 1.51202694e-01 -2.01399609e-01
-3.28129940e-02 -2.32419267e-01 2.51924898e-02 -8.46295953e-01
4.80662771e-02 3.84666659e-02 -3.64153415e-01 2.16545016e-01
-6.31372333e-01 2.59750098e-01 1.67859402e-02 3.82595509e-01
-1.02588616e-01 -4.15921956e-01 6.12375379e-01 -4.73878682e-01
-4.10050452e-01 2.41974682e-01 -7.99591064e-01 4.98901606e-01
1.75912201e-01 3.76199216e-01 -1.26040913e-02 -3.33998621e-01
-1.04106975e+00 2.76393946e-02 -8.40158984e-02 3.23928595e-01
3.66069466e-01 -9.96413171e-01 -8.23649168e-01 2.73782015e-01
-1.87835395e-01 5.21153927e-01 -1.17741600e-01 9.56737846e-02
-2.83518612e-01 5.69195032e-01 3.66417825e-01 -5.77533424e-01
-8.47627401e-01 3.28802586e-01 1.67166680e-01 -8.24422956e-01
-6.17194831e-01 1.16947508e+00 7.96200991e-01 -9.18209612e-01
1.52339667e-01 -9.45196807e-01 4.92272638e-02 -5.30858457e-01
2.36244172e-01 -3.02549273e-01 -1.95139214e-01 -3.53357375e-01
-1.53898701e-01 1.98075071e-01 -3.17099929e-01 -5.00966668e-01
1.26744497e+00 9.21285227e-02 -2.49927312e-01 4.80929345e-01
1.00628233e+00 1.38457447e-01 -1.40142667e+00 -2.03594670e-01
5.71420789e-01 4.30809379e-01 2.81781629e-02 -5.04297972e-01
-6.08695269e-01 1.10049736e+00 2.89508775e-02 1.53460845e-01
5.80548644e-01 5.02671540e-01 8.18806708e-01 6.61408126e-01
1.16190195e-01 -1.06281257e+00 -2.22580433e-01 1.12584901e+00
4.88332152e-01 -1.26927924e+00 -5.47861576e-01 -3.79343837e-01
-4.18104559e-01 1.06362593e+00 5.63694358e-01 -4.61872727e-01
5.70595860e-01 7.65332162e-01 1.86216254e-02 -1.75214753e-01
-1.26021850e+00 -2.60448635e-01 1.23583093e-01 3.64536166e-01
1.13015497e+00 3.89759451e-01 -6.88336566e-02 9.14110899e-01
-5.26869535e-01 -2.67471999e-01 3.24329883e-01 7.64822066e-01
-6.07148349e-01 -1.55144882e+00 -1.05660766e-01 2.03336254e-01
-6.96579099e-01 -8.23290586e-01 -1.04779296e-01 7.03091979e-01
-1.49327308e-01 8.57796311e-01 4.10382330e-01 2.05797985e-01
-1.10510901e-01 4.33228076e-01 6.11195803e-01 -1.25604069e+00
-1.69727221e-01 4.22293574e-01 4.89095092e-01 -3.12881291e-01
-2.42996588e-02 -7.08823919e-01 -1.76055837e+00 1.64107546e-01
-1.63667381e-01 3.61796588e-01 6.43196881e-01 1.12581718e+00
1.32188186e-01 5.67022562e-01 3.02454025e-01 -5.28410614e-01
-6.23818278e-01 -1.08916891e+00 -3.89119595e-01 6.52427822e-02
6.42931834e-02 -5.12714051e-02 -7.11418167e-02 4.23119813e-01] | [10.394906044006348, 9.586811065673828] |
8ba58179-04c4-44e2-8958-b12d7eeb8c54 | improved-reinforcement-learning-with | 1903.12328 | null | https://arxiv.org/abs/1903.12328v2 | https://arxiv.org/pdf/1903.12328v2.pdf | Improved Reinforcement Learning with Curriculum | Humans tend to learn complex abstract concepts faster if examples are presented in a structured manner. For instance, when learning how to play a board game, usually one of the first concepts learned is how the game ends, i.e. the actions that lead to a terminal state (win, lose or draw). The advantage of learning end-games first is that once the actions which lead to a terminal state are understood, it becomes possible to incrementally learn the consequences of actions that are further away from a terminal state - we call this an end-game-first curriculum. Currently the state-of-the-art machine learning player for general board games, AlphaZero by Google DeepMind, does not employ a structured training curriculum; instead learning from the entire game at all times. By employing an end-game-first training curriculum to train an AlphaZero inspired player, we empirically show that the rate of learning of an artificial player can be improved during the early stages of training when compared to a player not using a training curriculum. | ['Cameron Browne', 'Simon Denman', 'Frederic Maire', 'Joseph West'] | 2019-03-29 | null | null | null | null | ['board-games'] | ['playing-games'] | [ 3.36093247e-01 4.44680452e-01 1.62082747e-01 -1.29121348e-01
-1.92412272e-01 -6.27124548e-01 5.17293692e-01 3.54810297e-01
-6.53026462e-01 7.75750577e-01 -2.10538521e-01 -4.77891356e-01
-8.98406282e-02 -1.19027376e+00 -9.27011132e-01 -4.97096330e-01
-1.79836705e-01 7.21282899e-01 5.88644028e-01 -5.16495168e-01
2.13863894e-01 1.23661637e-01 -1.70595062e+00 4.29369688e-01
7.30113864e-01 5.97097278e-01 2.14543968e-01 9.50149536e-01
-2.46642172e-01 1.14391387e+00 -7.65082657e-01 -4.72386688e-01
2.58872658e-01 -6.96268618e-01 -1.10018373e+00 -1.06130570e-01
-6.43418431e-02 -5.32959580e-01 -3.21685011e-03 1.04823947e+00
7.97037557e-02 3.81733447e-01 5.12236595e-01 -1.11241257e+00
1.08760618e-01 8.63137364e-01 -1.64657786e-01 3.74193676e-02
3.86072934e-01 3.28199953e-01 1.22157633e+00 -1.48836911e-01
5.08480191e-01 9.41708982e-01 5.18157065e-01 7.71980703e-01
-1.17991984e+00 -5.41727722e-01 4.88737911e-01 -8.81160572e-02
-7.76842892e-01 3.57397683e-02 5.03995001e-01 -4.00463581e-01
9.30137932e-01 -3.37687843e-02 1.14260983e+00 6.85987949e-01
5.54635152e-02 9.31317449e-01 1.03819501e+00 -3.93720597e-01
6.55413866e-01 -2.06652984e-01 -1.14654824e-01 5.22441804e-01
9.59184319e-02 6.15339160e-01 -3.36891502e-01 1.91607669e-01
7.90518939e-01 -1.46824062e-01 1.56816870e-01 -4.99804646e-01
-6.28199100e-01 8.18106830e-01 4.34517562e-01 2.27453217e-01
-4.51335222e-01 5.63282907e-01 6.31752253e-01 6.41257882e-01
5.02938665e-02 1.02301371e+00 -6.85784459e-01 -5.73187709e-01
-8.23672771e-01 6.39351189e-01 9.44104791e-01 5.50186157e-01
8.46491933e-01 -3.54175717e-02 8.25887397e-02 3.69122237e-01
7.33056441e-02 -1.94068074e-01 3.75286400e-01 -1.07113612e+00
1.37305722e-01 7.03521729e-01 5.68937883e-02 -2.92121589e-01
-2.06650361e-01 -5.41830242e-01 -1.41329736e-01 8.82598996e-01
7.19562113e-01 -6.55287445e-01 -8.86590004e-01 1.76668239e+00
3.63020003e-01 5.49707115e-01 3.49830925e-01 5.05617797e-01
5.99488676e-01 7.99001515e-01 3.52527142e-01 7.08152801e-02
1.10570848e+00 -7.91530907e-01 -5.77666685e-02 -6.08092666e-01
1.00046790e+00 -8.19681957e-02 1.03302872e+00 1.02385473e+00
-1.20262170e+00 -6.87546492e-01 -1.10812557e+00 2.69102961e-01
-3.24899554e-01 -6.66185856e-01 8.80712867e-01 5.76903045e-01
-8.14400375e-01 1.01538599e+00 -8.28144610e-01 -9.54979062e-02
4.42001134e-01 5.43006122e-01 -1.41131178e-01 -4.51165624e-02
-1.16210616e+00 9.60978925e-01 9.48476076e-01 -5.15033662e-01
-1.47779441e+00 -1.09833229e+00 -5.84541440e-01 3.96166265e-01
8.44725311e-01 -5.35747170e-01 1.89775014e+00 -1.57135952e+00
-1.82850838e+00 8.51857185e-01 5.54906428e-01 -5.78741133e-01
6.36083424e-01 -2.68497676e-01 2.36747190e-01 -5.12225972e-03
-1.06441461e-01 8.49942446e-01 5.99676847e-01 -1.19674826e+00
-1.03183889e+00 -5.63192461e-03 9.46241856e-01 4.74272728e-01
-1.43619990e-02 -2.84927320e-02 -3.12624693e-01 -1.70767322e-01
-4.15990561e-01 -8.59754860e-01 -4.48783070e-01 -1.05587117e-01
7.84382224e-02 -4.21014339e-01 3.92907351e-01 -2.32534677e-01
1.05265653e+00 -1.98799646e+00 2.15960428e-01 1.52895659e-01
3.29261482e-01 5.89165330e-01 -4.00726236e-02 4.06484634e-01
-3.61231863e-01 1.60386395e-02 -7.29688630e-02 -1.37616754e-01
2.86923703e-02 4.70561504e-01 1.01511694e-01 -1.70737624e-01
-1.68328676e-02 6.63854063e-01 -1.40025985e+00 9.74238850e-03
3.00717503e-01 -6.41523898e-02 -1.09374344e+00 2.72033989e-01
-7.28669524e-01 3.77056956e-01 -4.72965628e-01 -1.19872674e-01
1.77471265e-01 6.79572150e-02 1.80509821e-01 6.52028263e-01
-6.99410737e-02 5.13552547e-01 -1.22862697e+00 1.68543875e+00
-6.28192425e-01 4.91873026e-01 -1.87760398e-01 -1.23862708e+00
6.37779891e-01 3.47082078e-01 1.94477692e-01 -6.67076230e-01
1.00970514e-01 8.38358030e-02 5.39168060e-01 -1.89692467e-01
3.35109055e-01 -5.17100930e-01 -1.16424367e-01 4.76824254e-01
2.12860331e-01 -4.24077868e-01 4.25702333e-01 2.52724290e-01
1.23085201e+00 4.06599849e-01 3.66050512e-01 -1.28327176e-01
3.03414881e-01 4.90210019e-02 4.60268378e-01 1.06045175e+00
-2.90963724e-02 2.89613426e-01 1.20752180e+00 -5.79530120e-01
-9.94958937e-01 -9.10098255e-01 5.19062698e-01 1.44821477e+00
-1.87540084e-01 -6.97884202e-01 -9.65133667e-01 -7.98504770e-01
-1.52903900e-01 8.61814201e-01 -8.03922296e-01 -5.07214546e-01
-5.88758349e-01 -6.28880784e-02 2.34222010e-01 5.66197813e-01
5.67452252e-01 -1.53876209e+00 -1.20444071e+00 4.83505249e-01
4.15167063e-01 -5.83152890e-01 -7.60605857e-02 6.78639174e-01
-1.14128864e+00 -1.08514273e+00 -3.88102055e-01 -6.30772531e-01
5.93351960e-01 -2.94229835e-01 1.37730861e+00 6.17554724e-01
-9.18885246e-02 1.48343354e-01 -3.99864644e-01 -6.37575448e-01
-4.89225507e-01 9.96739343e-02 -2.71390021e-01 -4.77791280e-01
1.59005642e-01 -7.56577253e-01 -5.74782968e-01 -4.64743041e-02
-9.22659755e-01 4.32748854e-01 6.35927916e-01 7.46146798e-01
1.92979574e-01 4.58418638e-01 3.05145323e-01 -1.28111196e+00
7.14170456e-01 -2.74066657e-01 -6.57730162e-01 7.88732320e-02
-3.61668468e-01 3.81135345e-01 9.15904820e-01 -7.11936533e-01
-8.75945926e-01 1.08823374e-01 -2.77568430e-01 2.28774119e-02
-4.00420338e-01 8.38144720e-01 -8.24589133e-02 3.21989954e-01
8.09787571e-01 2.60082334e-01 -1.19108148e-01 -1.71816900e-01
3.40694308e-01 4.26360629e-02 4.53890711e-01 -1.14479399e+00
7.66155422e-01 -2.73404364e-02 -9.82277244e-02 -6.03477538e-01
-8.03995252e-01 -3.66717517e-01 -5.10638893e-01 -4.52239394e-01
6.56041682e-01 -7.15575218e-01 -1.17368317e+00 6.30770802e-01
-7.75585592e-01 -1.03453422e+00 -8.36864948e-01 1.49464831e-01
-6.44675434e-01 -1.27393872e-01 -2.56130606e-01 -7.21128643e-01
9.03320685e-02 -1.03145969e+00 4.22603428e-01 6.65730417e-01
-4.20711547e-01 -1.05891991e+00 1.44825011e-01 -1.39930695e-02
2.06243753e-01 2.05974281e-01 1.06757855e+00 -8.28713536e-01
-2.06591398e-01 -1.40861854e-01 3.83440435e-01 4.25701201e-01
1.56888459e-02 -1.33100808e-01 -7.15347290e-01 -2.47809276e-01
-2.52188742e-01 -6.38716877e-01 5.79938710e-01 2.58950084e-01
9.98272896e-01 -2.35685378e-01 -1.58354148e-01 4.44925904e-01
1.30252433e+00 5.39165556e-01 9.23944950e-01 6.21422768e-01
2.14779511e-01 4.91601557e-01 4.60591644e-01 2.91539878e-01
1.62002668e-01 3.68138194e-01 6.29126072e-01 -2.05561309e-03
2.95007348e-01 -4.61711198e-01 3.16284180e-01 1.18050240e-01
-1.62483320e-01 4.28971760e-02 -9.08362389e-01 5.97344041e-01
-1.78351748e+00 -1.04288721e+00 3.54239047e-01 2.33889151e+00
1.14715052e+00 9.77732718e-01 4.64653373e-01 3.51072699e-01
1.55110374e-01 2.31426060e-02 -6.77108765e-01 -7.28841066e-01
6.35096550e-01 9.21983361e-01 2.02776298e-01 6.74101055e-01
-8.24666023e-01 1.34450591e+00 6.18270206e+00 7.07795262e-01
-9.84793901e-01 -2.08103597e-01 4.83957529e-01 -8.74974206e-02
-1.17138073e-01 3.54411244e-01 -5.77434003e-01 1.73543707e-01
9.57920611e-01 -1.14127822e-01 5.11066973e-01 1.09147501e+00
7.44523993e-03 -3.38744134e-01 -1.42014766e+00 4.59866732e-01
-5.45819759e-01 -1.24770367e+00 -7.14373365e-02 8.75751674e-02
6.85433388e-01 -4.17271435e-01 -8.43032002e-02 9.26685929e-01
1.05545616e+00 -1.21074724e+00 8.04795384e-01 1.33162305e-01
3.77336621e-01 -1.16789424e+00 4.40807343e-01 7.08629847e-01
-9.69777584e-01 -2.10071757e-01 -2.43395776e-01 -6.88383698e-01
-4.06437755e-01 5.24550863e-02 -9.51343775e-01 3.28823894e-01
4.13391471e-01 4.11909968e-01 -3.73686790e-01 1.20620644e+00
-8.58149946e-01 9.09944415e-01 -1.87367991e-01 -4.59244043e-01
6.03742063e-01 -1.07396126e-01 2.53587693e-01 7.91292489e-01
-3.70774306e-02 6.05038106e-01 2.09094927e-01 7.86183476e-01
1.10977739e-01 -3.73716831e-01 -3.81486923e-01 -9.51756239e-02
1.60874397e-01 9.17496920e-01 -7.43826091e-01 -4.01938379e-01
-3.44447643e-01 9.34415162e-01 3.02601665e-01 1.47238269e-01
-5.34843028e-01 -4.66023266e-01 7.88913250e-01 2.32833788e-01
5.49251735e-01 -6.33953363e-02 -2.30194852e-01 -7.51463830e-01
-4.48123664e-01 -1.14791787e+00 3.82522523e-01 -8.87558877e-01
-7.44121313e-01 2.22999468e-01 1.75230190e-01 -9.04576898e-01
-3.71932685e-01 -7.38211274e-01 -1.31181145e+00 7.69359410e-01
-1.23437774e+00 -5.20999789e-01 -1.56153083e-01 5.69922209e-01
5.34539580e-01 -7.02745691e-02 8.35431278e-01 -2.97466338e-01
-1.21206522e-01 4.17896956e-01 -2.38645136e-01 3.64204407e-01
4.21913862e-02 -1.77265584e+00 4.51885521e-01 7.07477808e-01
1.38394862e-01 5.14664233e-01 9.59244072e-01 -5.14843285e-01
-8.17581773e-01 -6.23519003e-01 5.04094899e-01 -4.44840118e-02
5.70719182e-01 -4.21236426e-01 -8.17575336e-01 7.28147030e-01
9.59616676e-02 -2.56239295e-01 4.83900875e-01 3.87760431e-01
-1.51247412e-01 -1.99867904e-01 -8.63079846e-01 9.26199496e-01
8.23094785e-01 -4.17243898e-01 -1.00554001e+00 -5.44348219e-03
6.17275119e-01 -5.41335225e-01 -3.40591729e-01 -7.49185234e-02
5.58261395e-01 -9.46428418e-01 9.44334149e-01 -1.03252852e+00
8.47884536e-01 -1.58394337e-01 3.42923880e-01 -1.74340796e+00
-2.35927999e-01 -7.83440173e-01 -9.48495343e-02 7.28709459e-01
4.17495728e-01 -3.97556007e-01 1.29374743e+00 5.39717078e-01
-1.67943686e-01 -7.45004356e-01 -5.89838207e-01 -6.54217064e-01
5.63877881e-01 -5.16340673e-01 5.16473472e-01 6.44961834e-01
2.01026991e-01 3.51752460e-01 -1.56729162e-01 -5.95423281e-02
3.80306512e-01 -2.10850492e-01 9.54978645e-01 -1.56783080e+00
-7.08220243e-01 -5.22967398e-01 -3.46154064e-01 -1.32098103e+00
-8.23190883e-02 -6.24490082e-01 3.05175006e-01 -1.43185031e+00
1.49732996e-02 -5.69176018e-01 -3.77223849e-01 6.30423307e-01
-4.03557271e-01 -4.22658235e-01 3.65904629e-01 -2.91937143e-01
-6.83034897e-01 1.25253230e-01 1.38615572e+00 1.49270579e-01
-4.03829515e-01 2.90358394e-01 -8.21838975e-01 7.51975238e-01
7.42471159e-01 -6.63324535e-01 -7.34771729e-01 -1.13148995e-01
6.34843886e-01 2.92533964e-01 1.10736087e-01 -1.24905396e+00
4.14448589e-01 -4.06045169e-01 2.47757301e-01 -1.20293811e-01
1.94342643e-01 -6.75050795e-01 1.15028545e-01 9.78086352e-01
-6.31108105e-01 -8.99794996e-02 4.47313726e-01 2.89276749e-01
-1.28442878e-02 -8.34959745e-01 7.69214690e-01 -6.29830182e-01
-8.74274313e-01 1.02317907e-01 -8.88580322e-01 3.19377601e-01
1.20990872e+00 -5.15545547e-01 1.14697613e-01 -7.19107807e-01
-9.97558713e-01 3.63403827e-01 1.95379123e-01 3.47838551e-02
3.85908157e-01 -8.73467088e-01 -5.95154464e-01 -2.14488339e-02
-8.29832181e-02 4.02889788e-01 9.42196231e-03 3.26749325e-01
-7.55449593e-01 5.39040416e-02 -5.25491714e-01 -3.34184110e-01
-1.21220279e+00 3.92010868e-01 7.14586794e-01 -8.67079616e-01
-6.46347702e-01 1.14132249e+00 4.47718263e-01 -5.40153682e-01
3.85210276e-01 -4.90494311e-01 -2.70203978e-01 -1.45950783e-02
5.74691832e-01 -2.05967307e-01 1.24002464e-01 1.66198507e-01
5.85369393e-02 1.39067531e-01 -2.89347827e-01 -3.45986754e-01
1.60957074e+00 6.35861158e-01 3.04948539e-01 4.40479726e-01
6.62268996e-01 -4.57007498e-01 -1.74876356e+00 -1.73471630e-01
6.73650950e-02 -2.61771858e-01 -2.43954472e-02 -1.07124221e+00
-1.03556633e+00 9.66071486e-01 2.07674176e-01 2.48928517e-01
1.12647462e+00 -2.69423246e-01 3.49584222e-01 5.90978801e-01
5.68009079e-01 -1.20259976e+00 6.08757913e-01 8.81162822e-01
5.42489111e-01 -7.80200064e-01 -1.73398748e-01 -6.55059889e-02
-7.07435906e-01 1.30657160e+00 8.41646850e-01 -4.57975179e-01
4.34635967e-01 1.97387293e-01 -2.27025345e-01 -2.57097363e-01
-1.01824939e+00 -2.68297583e-01 -7.35515505e-02 7.47500896e-01
2.59499878e-01 -3.41829509e-02 5.13914116e-02 5.67183912e-01
-4.51518744e-01 1.41144097e-01 6.92952752e-01 1.36010766e+00
-7.30801404e-01 -1.28813326e+00 -2.91464198e-02 4.49222028e-01
-4.41890657e-01 -1.36376053e-01 -4.70484138e-01 1.04319155e+00
3.92885476e-01 6.73550367e-01 1.53843835e-01 -3.58712614e-01
4.42929000e-01 2.21924514e-01 8.60260010e-01 -1.19496214e+00
-1.06306565e+00 -3.62586856e-01 1.10164337e-01 -5.13788760e-01
-1.47321701e-01 -6.76534355e-01 -1.58127117e+00 -5.64671338e-01
-1.17286749e-01 4.97140497e-01 2.98942536e-01 1.29109764e+00
-3.30376297e-01 9.02851343e-01 2.97995836e-01 -8.12443376e-01
-4.39472824e-01 -6.01875603e-01 -4.11553890e-01 3.41261297e-01
3.43020499e-01 -5.82688987e-01 -2.89909154e-01 -8.82346928e-02] | [3.896543025970459, 1.4744749069213867] |
86280fa8-483b-4b28-8a5c-d1bd26557962 | leveraging-user-triggered-supervision-in | 2302.03784 | null | https://arxiv.org/abs/2302.03784v1 | https://arxiv.org/pdf/2302.03784v1.pdf | Leveraging User-Triggered Supervision in Contextual Bandits | We study contextual bandit (CB) problems, where the user can sometimes respond with the best action in a given context. Such an interaction arises, for example, in text prediction or autocompletion settings, where a poor suggestion is simply ignored and the user enters the desired text instead. Crucially, this extra feedback is user-triggered on only a subset of the contexts. We develop a new framework to leverage such signals, while being robust to their biased nature. We also augment standard CB algorithms to leverage the signal, and show improved regret guarantees for the resulting algorithms under a variety of conditions on the helpfulness of and bias inherent in this feedback. | ['Teodor V. Marinov', 'Claudio Gentile', 'Alekh Agarwal'] | 2023-02-07 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [ 4.11371768e-01 2.92425952e-03 -8.52829337e-01 -3.45946580e-01
-1.04692149e+00 -9.71959651e-01 4.82935846e-01 2.89168090e-01
-5.04375398e-01 9.64760900e-01 5.49790680e-01 -6.33885562e-01
-4.48064357e-01 -3.64130706e-01 -8.16047728e-01 -6.82136714e-01
2.58498579e-01 4.84937727e-01 -1.65841803e-01 -2.00133935e-01
3.19527805e-01 7.03593567e-02 -1.03579104e+00 4.11258489e-01
7.02763438e-01 1.06062412e+00 2.76949972e-01 1.05456889e+00
-7.96988904e-02 6.82524920e-01 -3.24491024e-01 -4.20784980e-01
3.84314060e-01 -6.01052701e-01 -7.09126294e-01 1.29005343e-01
4.24279273e-02 -4.58108634e-01 -3.17870170e-01 9.84121442e-01
2.58297116e-01 4.20283735e-01 4.48720127e-01 -1.01523280e+00
-4.18747216e-01 1.12244713e+00 -5.28122544e-01 2.26206884e-01
6.49931252e-01 -1.15381606e-01 1.57879138e+00 -4.43307757e-01
3.89995039e-01 1.01933324e+00 4.48551953e-01 6.88025355e-01
-1.46137667e+00 -5.45035541e-01 7.67020106e-01 -4.41378243e-02
-5.07791936e-01 -5.38588107e-01 8.76526058e-01 -3.10896724e-01
2.39545718e-01 7.27422476e-01 6.35256052e-01 1.32877755e+00
-4.88983780e-01 1.21038365e+00 1.00234711e+00 -6.51488900e-01
6.11668527e-01 3.31695169e-01 5.33322096e-01 1.40902340e-01
1.90738007e-01 7.50141591e-02 -7.82721341e-01 -6.97078407e-01
2.44886160e-01 5.37310779e-01 -5.67520738e-01 -4.00370389e-01
-7.92958438e-01 8.78574014e-01 2.23977849e-01 3.51676457e-02
-4.76598620e-01 2.32558429e-01 1.52555361e-01 3.76945764e-01
5.27272224e-01 4.40518379e-01 -6.38904333e-01 -4.52241600e-01
-7.02788293e-01 4.43277448e-01 9.84320641e-01 1.09484982e+00
6.18149519e-01 -5.01924455e-01 -6.76663637e-01 9.19776618e-01
-2.45945409e-01 3.74498129e-01 3.83094639e-01 -1.03309047e+00
7.27022886e-01 2.59173624e-02 1.01958680e+00 -6.62854075e-01
-9.21777412e-02 -3.27286988e-01 -4.40300852e-01 -4.29408193e-01
5.11336029e-01 -5.25742888e-01 -7.47785389e-01 1.87184465e+00
4.46652353e-01 -1.48493409e-01 -4.12339449e-01 8.78845394e-01
-2.61261407e-02 3.80359769e-01 -3.42927933e-01 -5.45717001e-01
8.00456882e-01 -7.67154753e-01 -7.29978383e-01 -5.03202140e-01
6.32410228e-01 -5.13151705e-01 1.26559973e+00 6.32746279e-01
-9.69142377e-01 2.65823990e-01 -7.52641320e-01 2.14979172e-01
1.66589156e-01 -1.62718877e-01 8.02120209e-01 8.41939569e-01
-5.29197037e-01 5.70446789e-01 -5.96379697e-01 -6.00671321e-02
3.62003714e-01 1.89399183e-01 7.61714429e-02 -3.49024087e-01
-9.46991563e-01 3.14814001e-01 -1.13705941e-01 2.15056896e-01
-6.61185026e-01 -7.74192750e-01 -2.48215809e-01 3.16728890e-01
9.51457083e-01 -5.87821662e-01 1.91817272e+00 -1.47189605e+00
-1.53911090e+00 3.62253010e-01 -5.51238000e-01 -4.58682060e-01
9.38229084e-01 -6.12723827e-01 1.69636726e-01 -1.28205299e-01
8.14220458e-02 4.08848226e-02 8.71984959e-01 -1.12931705e+00
-7.54048884e-01 -4.27035958e-01 4.95271623e-01 2.72734731e-01
-2.92842567e-01 -1.36746407e-01 -4.43871379e-01 -6.45932555e-01
1.92700922e-02 -1.15850890e+00 -3.69014353e-01 -3.50127190e-01
-4.63729680e-01 2.34575849e-02 4.64341879e-01 -3.34712148e-01
1.20110106e+00 -2.01142001e+00 -8.32519531e-02 5.00545204e-01
-5.58140390e-02 -3.44447553e-01 1.88420713e-02 6.69622242e-01
2.34433308e-01 3.40240300e-01 -5.57418279e-02 -3.82430971e-01
1.10704437e-01 1.68758303e-01 -7.08566606e-01 4.43689972e-01
-4.98066425e-01 5.25940657e-01 -9.95685637e-01 9.48420763e-02
-2.38450065e-01 -2.30648026e-01 -9.97498453e-01 1.91497549e-01
-5.06024122e-01 5.06346047e-01 -7.52160311e-01 3.53626996e-01
2.77492762e-01 -3.40108037e-01 4.40265417e-01 5.98113120e-01
3.72492403e-01 5.81974626e-01 -1.28317428e+00 1.32381129e+00
-6.82909787e-01 5.15181303e-01 5.68470955e-01 -9.58659708e-01
7.26273879e-02 1.99299514e-01 4.07100946e-01 -2.48877451e-01
-1.22005723e-01 -2.74365880e-02 -4.27543856e-02 -1.91145882e-01
5.24046481e-01 -3.95867415e-02 1.55993640e-01 8.74941885e-01
-4.03104901e-01 1.44400582e-01 -3.42204459e-02 5.35033822e-01
1.10450089e+00 -1.65422201e-01 3.99375856e-01 9.37540177e-03
-4.10608612e-02 -1.42359689e-01 6.15966082e-01 1.90543187e+00
-2.26737559e-01 4.50406283e-01 9.17819977e-01 -1.82300806e-02
-8.13906431e-01 -3.15983415e-01 1.81684062e-01 1.88225925e+00
4.91693877e-02 -3.78110856e-01 -5.03142774e-01 -9.05987144e-01
3.07637542e-01 8.00581932e-01 -7.92064726e-01 -3.47683430e-02
-2.93956339e-01 -6.21124685e-01 -9.13197249e-02 5.21158338e-01
4.33115549e-02 -5.52964747e-01 -4.97725636e-01 2.68489361e-01
-3.47905964e-01 -8.23369205e-01 -9.86017883e-01 3.45892102e-01
-1.11207044e+00 -9.95324373e-01 -4.62766290e-01 -4.81243357e-02
5.26162505e-01 6.68625832e-01 8.39092553e-01 3.73342298e-02
3.65414947e-01 6.22639656e-01 -5.02974868e-01 -2.65368193e-01
-8.07339028e-02 -3.58004496e-02 -1.28073484e-01 1.70040846e-01
1.16830263e-02 -3.83830756e-01 -8.32190990e-01 2.30368614e-01
-7.10042179e-01 8.76426995e-02 3.95704091e-01 1.41340172e+00
1.66830555e-01 -4.32080537e-01 6.85969412e-01 -1.45133603e+00
9.41031873e-01 -7.96048105e-01 -6.16450012e-01 1.48431912e-01
-6.38151169e-01 3.80406380e-02 5.59804261e-01 -7.66683102e-01
-1.09970236e+00 5.57632744e-03 1.48826346e-01 -2.83425629e-01
3.91527340e-02 6.21236622e-01 -4.90406007e-02 2.22736880e-01
7.95821130e-01 2.83077918e-02 -5.13475955e-01 -4.87729281e-01
5.14440119e-01 8.17248702e-01 3.02772701e-01 -1.04371381e+00
1.10514894e-01 3.49558651e-01 -1.94711357e-01 -5.03889978e-01
-1.05918896e+00 -7.28587687e-01 -1.83172803e-02 -1.41101941e-01
-1.49140760e-01 -5.99247754e-01 -1.10915220e+00 -1.83943525e-01
-1.03020644e+00 -5.30513346e-01 -1.24465771e-01 4.30966735e-01
-5.99937558e-01 2.88978040e-01 -5.23775280e-01 -1.50826538e+00
-3.72459516e-02 -1.00053096e+00 7.65325487e-01 -7.13413730e-02
-2.47840941e-01 -5.69266737e-01 -1.93491206e-01 4.21050459e-01
2.05579862e-01 -2.49511063e-01 7.91005969e-01 -1.06116176e+00
-5.57900906e-01 -4.77037489e-01 -1.61611941e-03 -1.83238983e-02
1.07182957e-01 -2.85175622e-01 -9.76380110e-01 -2.66782105e-01
-9.96595621e-02 -3.33907157e-01 1.03055286e+00 6.45771921e-01
1.38875997e+00 -8.61324310e-01 -3.83768469e-01 2.20622569e-01
8.87944639e-01 2.97119141e-01 -2.76446879e-01 2.30173260e-01
1.95511296e-01 4.57876742e-01 5.93567789e-01 8.76653314e-01
2.44011674e-02 6.94754899e-01 3.04923207e-01 3.57406437e-01
5.89096844e-01 -4.99815285e-01 2.09546074e-01 5.11422716e-02
1.83213234e-01 -2.60733247e-01 -5.43359995e-01 5.85163057e-01
-2.32886195e+00 -7.75866568e-01 3.26099455e-01 2.58427167e+00
1.22725046e+00 3.75351638e-01 3.26609373e-01 1.57657772e-01
6.98957264e-01 1.42371416e-01 -1.06544280e+00 -3.02275062e-01
-4.02022824e-02 -6.82031214e-02 8.81435752e-01 8.18539977e-01
-9.27513778e-01 6.58180892e-01 7.18519545e+00 7.24596024e-01
-8.41093719e-01 2.29465395e-01 1.06501317e+00 -8.78940642e-01
-4.90263760e-01 2.79849201e-01 -6.53964758e-01 6.80467665e-01
8.87020528e-01 -7.12368488e-02 9.50657070e-01 1.07174110e+00
3.27730030e-01 -5.86383939e-01 -1.46867692e+00 7.96638548e-01
-2.96091080e-01 -1.42586470e+00 -3.56884480e-01 -1.10953815e-01
8.79454374e-01 -1.88193738e-01 4.26681265e-02 2.69600987e-01
9.71574128e-01 -5.62624335e-01 7.52202809e-01 4.20045853e-01
4.81762201e-01 -7.76528239e-01 2.60802567e-01 7.34973907e-01
-3.19006145e-01 -7.39055753e-01 -9.14213359e-02 -4.21672493e-01
-1.15821935e-01 7.23582089e-01 -8.31291258e-01 2.64715731e-01
5.69645762e-01 2.82482117e-01 2.75107436e-02 1.01045811e+00
-3.19916070e-01 1.10367799e+00 -5.19667566e-01 -3.42652470e-01
2.40053788e-01 -2.49437869e-01 6.98356926e-01 1.08773363e+00
2.23462179e-01 3.09086144e-01 2.71981061e-01 4.72685277e-01
-4.43591028e-01 2.85616815e-01 -5.33167660e-01 5.38812093e-02
8.41539562e-01 8.62265170e-01 -3.75018656e-01 -3.95081460e-01
-3.30824614e-01 8.58507574e-01 3.78932655e-01 8.33873630e-01
-3.51628184e-01 -3.76786701e-02 6.59731686e-01 -5.76928630e-02
3.77816230e-01 1.73897967e-01 -7.33371556e-01 -1.06635642e+00
1.95591003e-01 -7.88809538e-01 7.64603019e-01 -7.57020772e-01
-1.14121306e+00 -2.06896737e-01 6.25555813e-02 -6.47054672e-01
-4.04947370e-01 -1.48152232e-01 -5.05794644e-01 7.62408435e-01
-1.13433611e+00 -5.93197465e-01 2.67152578e-01 4.81242001e-01
5.38033605e-01 3.02453071e-01 3.25775474e-01 -1.70877114e-01
-4.90936965e-01 6.04346216e-01 6.79941833e-01 -3.08292121e-01
8.55713904e-01 -1.48845637e+00 -2.18339086e-01 8.02487969e-01
1.91829741e-01 8.35980773e-01 1.07890534e+00 -5.64348400e-01
-1.57541800e+00 -7.61947513e-01 5.61622798e-01 -3.98684442e-01
6.24715447e-01 -5.79806924e-01 -4.99204487e-01 8.83976340e-01
3.00383978e-02 -1.34228587e-01 7.30327666e-01 7.93511152e-01
-4.07633126e-01 -3.19305003e-01 -1.05755043e+00 1.01839364e+00
9.94653344e-01 -4.63001817e-01 -4.14964676e-01 6.14451230e-01
8.81713450e-01 -5.46527147e-01 -1.36889070e-01 -6.41911551e-02
7.52796471e-01 -9.31117117e-01 5.43323040e-01 -1.13031006e+00
3.90644222e-01 1.73415184e-01 -3.31788301e-01 -1.51695442e+00
-9.11491513e-02 -1.33023167e+00 -4.69860464e-01 8.10516894e-01
4.91084278e-01 -6.33776665e-01 1.02330828e+00 1.18875706e+00
9.33040828e-02 -6.96232855e-01 -9.88022447e-01 -4.06891733e-01
-1.71072081e-01 -8.13515723e-01 7.59785235e-01 6.36755347e-01
4.48292166e-01 2.47332096e-01 -9.65084910e-01 -1.86053500e-01
2.67269343e-01 4.71102476e-01 8.31693351e-01 -9.37757492e-01
-7.54375517e-01 -5.30897260e-01 5.76651216e-01 -1.82035995e+00
-7.30984285e-02 -4.50526804e-01 2.22140074e-01 -9.07741427e-01
4.48000580e-01 -5.85611701e-01 -5.79444349e-01 4.18959439e-01
-4.82220113e-01 -2.57551908e-01 2.20693752e-01 1.58797354e-01
-6.00998938e-01 3.07440758e-01 9.65265691e-01 -3.54057401e-02
-6.83796108e-01 7.44064629e-01 -1.41324055e+00 4.38633025e-01
5.55916727e-01 -5.74510932e-01 -3.98817360e-01 2.57285442e-02
7.25342095e-01 6.85688794e-01 7.56482929e-02 -1.59541070e-01
4.57756937e-01 -6.78444684e-01 2.52836168e-01 -5.32955647e-01
3.10275644e-01 -8.36800754e-01 -1.61090389e-01 5.49159013e-03
-1.18608749e+00 -2.98557699e-01 -1.95044249e-01 1.41272366e+00
2.80327946e-01 -3.00917655e-01 5.14354527e-01 -4.30494808e-02
2.36451000e-01 1.94658816e-01 -3.70582879e-01 9.54217836e-02
6.29207969e-01 -1.31564111e-01 -2.25949943e-01 -1.24754095e+00
-6.37969375e-01 3.68293405e-01 1.75160438e-01 -6.31126985e-02
2.42672190e-01 -1.13446569e+00 -2.04968855e-01 -9.27632377e-02
1.06229097e-01 -1.53925374e-01 4.20102775e-02 1.07847261e+00
4.26935673e-01 5.48738897e-01 4.57328171e-01 -3.91585469e-01
-1.12055945e+00 7.09744215e-01 1.58593193e-01 -1.49178848e-01
-5.03434002e-01 6.70124173e-01 3.10008496e-01 9.56306532e-02
5.98332167e-01 -5.21163702e-01 2.27182060e-01 6.17629103e-02
4.33795720e-01 3.90704632e-01 1.54799938e-01 5.45182303e-02
1.07972734e-02 -2.98992902e-01 -4.06543583e-01 -7.23169327e-01
1.09999168e+00 -3.46153527e-01 2.06301779e-01 6.97679222e-01
5.10318160e-01 4.14863288e-01 -1.46220958e+00 -5.94710827e-01
2.28621379e-01 -9.29213345e-01 6.48257360e-02 -1.04684913e+00
-7.35435367e-01 3.67789656e-01 1.32189151e-02 5.72811127e-01
1.02174342e+00 -2.26442233e-01 5.43208838e-01 8.52733731e-01
2.88249731e-01 -1.35145032e+00 -2.17490092e-01 4.14576352e-01
7.77721941e-01 -1.26491892e+00 -5.95598929e-02 9.07311961e-02
-7.04513252e-01 8.11781883e-01 1.29345506e-01 9.95862335e-02
7.05285192e-01 -8.30943137e-02 -3.31458539e-01 5.40638924e-01
-1.32149649e+00 -5.09021990e-02 -4.65148501e-02 4.28385228e-01
3.66576195e-01 2.86182374e-01 -4.29111749e-01 1.08995819e+00
-8.35534111e-02 -1.63727209e-01 8.23909640e-01 1.05491745e+00
-4.44358319e-01 -9.05151784e-01 -5.23770154e-01 8.96130145e-01
-9.38678443e-01 -2.23852545e-01 -6.83486402e-01 2.39494428e-01
-7.15295374e-01 1.49859607e+00 -3.06148410e-01 -8.27397928e-02
1.33955300e-01 1.75688475e-01 2.08059579e-01 -5.04603624e-01
-6.49153948e-01 5.55439293e-01 2.42751792e-01 -6.46412194e-01
-1.82772651e-01 -7.12429643e-01 -5.88380754e-01 -2.65278131e-01
-5.03362298e-01 4.33633029e-01 5.11879265e-01 1.00092995e+00
3.44543815e-01 1.07064843e-01 9.81777012e-01 -6.82404935e-01
-9.45509374e-01 -1.09891522e+00 -5.89853525e-01 4.02735442e-01
9.69884217e-01 -5.81938446e-01 -4.88857239e-01 -1.94376111e-01] | [4.505217552185059, 3.2296671867370605] |
465821f8-cd15-49d1-827d-b56826d53ba2 | integrating-uncertainty-into-neural-network | 2305.08744 | null | https://arxiv.org/abs/2305.08744v1 | https://arxiv.org/pdf/2305.08744v1.pdf | Integrating Uncertainty into Neural Network-based Speech Enhancement | Supervised masking approaches in the time-frequency domain aim to employ deep neural networks to estimate a multiplicative mask to extract clean speech. This leads to a single estimate for each input without any guarantees or measures of reliability. In this paper, we study the benefits of modeling uncertainty in clean speech estimation. Prediction uncertainty is typically categorized into aleatoric uncertainty and epistemic uncertainty. The former refers to inherent randomness in data, while the latter describes uncertainty in the model parameters. In this work, we propose a framework to jointly model aleatoric and epistemic uncertainties in neural network-based speech enhancement. The proposed approach captures aleatoric uncertainty by estimating the statistical moments of the speech posterior distribution and explicitly incorporates the uncertainty estimate to further improve clean speech estimation. For epistemic uncertainty, we investigate two Bayesian deep learning approaches: Monte Carlo dropout and Deep ensembles to quantify the uncertainty of the neural network parameters. Our analyses show that the proposed framework promotes capturing practical and reliable uncertainty, while combining different sources of uncertainties yields more reliable predictive uncertainty estimates. Furthermore, we demonstrate the benefits of modeling uncertainty on speech enhancement performance by evaluating the framework on different datasets, exhibiting notable improvement over comparable models that fail to account for uncertainty. | ['Timo Gerkmann', 'Stefan Wermter', 'Dennis Becker', 'Huajian Fang'] | 2023-05-15 | null | null | null | null | ['speech-enhancement'] | ['speech'] | [-1.29975811e-01 3.76535542e-02 1.93569168e-01 -3.98108542e-01
-1.53782094e+00 -3.10344458e-01 6.98955059e-01 -2.38963068e-02
-1.99006990e-01 9.47248697e-01 5.36159217e-01 -2.99322367e-01
-4.50937152e-01 -6.51321948e-01 -7.80078650e-01 -9.23078239e-01
1.25029340e-01 3.61887589e-02 -2.54713982e-01 1.96713701e-01
-1.30479783e-01 2.27903187e-01 -1.50780511e+00 9.57126170e-02
1.17110646e+00 1.30677915e+00 3.36143039e-02 4.88294542e-01
-8.05132911e-02 7.52655387e-01 -1.14268565e+00 -3.66499901e-01
-6.50165156e-02 -1.12235308e-01 -1.12836175e-01 -3.39796335e-01
-4.89585847e-02 -4.77457970e-01 -2.53855675e-01 1.32903564e+00
8.13012779e-01 1.14008144e-01 9.91383374e-01 -8.53743732e-01
-3.08367133e-01 1.23825645e+00 -2.78081805e-01 1.95317283e-01
-2.14279085e-01 1.89647228e-01 7.85254419e-01 -7.01956987e-01
-1.79541901e-01 9.81968343e-01 8.24166417e-01 2.12052509e-01
-8.62952888e-01 -7.34522998e-01 -1.29678369e-01 1.10971525e-01
-1.44395208e+00 -8.25729907e-01 7.27879763e-01 -4.40893024e-01
6.55477464e-01 1.09795488e-01 -8.82012844e-02 1.28838706e+00
2.46848688e-01 5.23797512e-01 1.34738338e+00 -5.01401544e-01
5.54781258e-01 2.55080938e-01 1.81415692e-01 1.21177241e-01
1.67388633e-01 9.01048541e-01 -6.66688800e-01 -1.83423936e-01
1.46306977e-01 -5.28905272e-01 -3.21459979e-01 3.89124572e-01
-6.60386682e-01 4.85653967e-01 2.00190574e-01 8.66556242e-02
-3.59408319e-01 5.27758837e-01 1.78231433e-01 -1.82453290e-01
7.66115725e-01 8.76594186e-02 -3.51372659e-01 -2.78866678e-01
-1.30623257e+00 -3.51763628e-02 7.45737016e-01 5.58748662e-01
3.60468775e-01 6.00176394e-01 -5.60323238e-01 9.26419795e-01
5.79136431e-01 8.62349033e-01 -4.26505283e-02 -9.82403755e-01
4.62206811e-01 -4.31118995e-01 3.47707003e-01 -6.31098092e-01
-1.98139310e-01 -8.85274708e-01 -9.73021984e-01 2.61637330e-01
3.50182086e-01 -5.17471552e-01 -1.09790862e+00 1.93215299e+00
2.63037439e-02 3.42833608e-01 1.76688954e-01 6.29616737e-01
6.31685555e-01 7.33207047e-01 2.65266746e-01 -9.95894894e-02
1.32966757e+00 -5.66427529e-01 -1.20267284e+00 -1.25826687e-01
-4.85390760e-02 -7.03296721e-01 5.07409036e-01 5.81683218e-01
-1.08826041e+00 -4.42163318e-01 -1.32042038e+00 4.73956317e-01
-2.27302387e-01 4.03721213e-01 2.28774827e-02 1.32104528e+00
-7.89544642e-01 9.26589847e-01 -6.69583619e-01 6.01789296e-01
3.30666929e-01 1.18314564e-01 1.57231465e-01 2.63967007e-01
-1.67395067e+00 8.93304110e-01 4.68806624e-01 3.83451521e-01
-1.10028565e+00 -9.09089863e-01 -8.10309827e-01 3.84158313e-01
2.71619916e-01 -4.26657230e-01 1.37451398e+00 -3.85742962e-01
-1.59793901e+00 -4.20357287e-02 -6.49698153e-02 -8.67791951e-01
5.61200678e-01 -2.89414167e-01 -6.00664556e-01 -1.67354364e-02
-4.66679037e-01 2.29633659e-01 1.42458487e+00 -1.36943185e+00
-3.21873188e-01 -1.09526820e-01 -1.04634777e-01 -1.79524884e-01
-1.52623951e-01 -8.42122268e-03 -2.42875703e-02 -8.72446120e-01
3.11784446e-02 -5.22666752e-01 6.49600755e-03 -3.61950248e-01
-4.37706232e-01 1.82804510e-01 2.65961468e-01 -1.19277394e+00
1.34630787e+00 -2.24204874e+00 -1.91648111e-01 1.84717357e-01
1.79628253e-01 2.48356223e-01 3.34471762e-01 2.65304238e-01
1.92040786e-01 2.55299091e-01 -5.10230362e-01 -7.69667685e-01
4.99063313e-01 1.12028331e-01 -5.92516840e-01 2.40910396e-01
4.15528357e-01 6.03949666e-01 -5.22059441e-01 -3.14373314e-01
3.71217370e-01 9.86846328e-01 -1.36683986e-01 9.54653323e-02
-2.59961635e-01 3.52556616e-01 -8.83292332e-02 3.87645066e-01
1.04641104e+00 1.16673462e-01 -1.33575693e-01 -5.00222981e-01
2.25252826e-02 5.44021726e-01 -1.25319231e+00 1.13508058e+00
-8.77566695e-01 4.75023955e-01 2.94256330e-01 -5.74649870e-01
7.67546535e-01 4.99861568e-01 -1.09111788e-02 -1.43042609e-01
3.99985522e-01 3.39901268e-01 -3.35928611e-02 -1.31995529e-01
5.39267898e-01 -3.63821268e-01 -7.78877223e-03 4.35929269e-01
3.78019065e-01 -3.49749207e-01 -2.48604402e-01 -2.98969932e-02
6.76137209e-01 3.01471446e-02 6.55820221e-02 -2.27703571e-01
3.71152490e-01 -7.34687507e-01 3.43851656e-01 1.01284742e+00
-4.26212966e-01 6.74056470e-01 4.45181578e-01 3.82679045e-01
-8.89044642e-01 -1.42479515e+00 -4.41295773e-01 5.86983263e-01
-3.37745339e-01 -2.33870074e-02 -1.05018556e+00 -3.59920084e-01
-1.00271352e-01 1.22028816e+00 -4.77597505e-01 -3.52773637e-01
-7.72793405e-03 -9.46030557e-01 8.41209888e-01 6.47465706e-01
5.07678509e-01 -5.84568322e-01 -7.46207312e-02 5.02083637e-02
-2.62185663e-01 -1.07878745e+00 -2.59612083e-01 6.15948737e-01
-4.85342383e-01 -3.90352994e-01 -6.08947754e-01 1.41597331e-01
1.86718434e-01 -2.61636615e-01 8.08716178e-01 -4.39902455e-01
2.10972309e-01 3.84619325e-01 -1.02007121e-01 -7.08337367e-01
-7.53775597e-01 -3.47867757e-01 3.27415854e-01 -7.40542635e-02
1.04117639e-01 -6.36497796e-01 -3.41293454e-01 7.52827749e-02
-8.55005622e-01 -4.13163781e-01 4.49088216e-01 7.89630532e-01
3.67316723e-01 6.34082615e-01 9.22965527e-01 -3.93508792e-01
9.15994763e-01 -5.78097761e-01 -6.24122441e-01 2.12183937e-01
-6.52022958e-01 5.14710307e-01 2.72099227e-01 -1.46104604e-01
-1.79491389e+00 -3.54479641e-01 -5.32737911e-01 -3.88692498e-01
-1.58359677e-01 7.32735634e-01 -3.29794168e-01 1.44097686e-01
6.59234405e-01 4.86471578e-02 -3.29889029e-01 -4.98787493e-01
3.19293946e-01 1.01240432e+00 7.33424187e-01 -9.65647638e-01
5.67410529e-01 2.98476994e-01 -6.70142174e-02 -6.60728574e-01
-9.33241785e-01 -8.17700550e-02 -2.20652446e-01 -1.32927120e-01
5.98914504e-01 -1.12476277e+00 -4.48519558e-01 7.22111285e-01
-1.26191616e+00 1.65298879e-01 -3.71639550e-01 7.96183109e-01
-6.06930375e-01 4.70690817e-01 -5.89859784e-01 -1.67491269e+00
-4.99820292e-01 -1.17360389e+00 9.50538933e-01 7.98622593e-02
8.47318247e-02 -8.24130416e-01 -1.45746350e-01 2.86657363e-01
7.34749019e-01 -9.38900188e-03 7.71972775e-01 -4.94689018e-01
-4.19435889e-01 -1.57414570e-01 -2.26720259e-01 9.54409003e-01
2.73362044e-02 7.37014338e-02 -1.65410423e+00 2.01382741e-01
4.99668241e-01 5.43703325e-02 1.11727333e+00 9.87160802e-01
1.07374120e+00 -2.21745417e-01 1.80145741e-01 2.95740902e-01
1.24865484e+00 1.61317453e-01 7.58405685e-01 -1.16657987e-01
1.08818874e-01 5.81975460e-01 3.15529585e-01 8.15994978e-01
-2.59358361e-02 2.52792299e-01 3.99089903e-01 4.87809569e-01
-3.05690408e-01 -1.77438244e-01 3.26148540e-01 8.88089538e-01
4.17331941e-02 -4.16873455e-01 -7.10234225e-01 4.75382417e-01
-1.42836785e+00 -1.00879014e+00 2.49830529e-01 2.32887745e+00
1.11460090e+00 3.50006223e-01 -2.25236133e-01 2.66491026e-01
8.85803342e-01 2.18174979e-01 -3.72542679e-01 -2.88430095e-01
-2.21702054e-01 3.58525515e-01 5.49323976e-01 7.40991116e-01
-1.05186832e+00 4.04542297e-01 6.45676136e+00 1.28203988e+00
-7.91117549e-01 2.75474727e-01 8.43382001e-01 -1.06843218e-01
-5.55625677e-01 -2.10363179e-01 -7.33406901e-01 6.57501101e-01
1.41576838e+00 -1.73914973e-02 4.15572733e-01 4.57082391e-01
3.78696293e-01 -2.22479314e-01 -6.41193032e-01 6.23799145e-01
-2.26671904e-01 -1.20897722e+00 -3.02313477e-01 -6.34238729e-03
7.43143797e-01 -1.41046559e-02 4.86819953e-01 2.54827321e-01
4.12159353e-01 -1.08470666e+00 1.03561342e+00 1.08948267e+00
6.68222845e-01 -9.64247882e-01 8.91052246e-01 3.97864610e-01
-7.46275783e-01 -1.77207813e-01 -2.17882916e-01 1.27959356e-01
2.76849478e-01 1.22788250e+00 -7.41036177e-01 6.21819377e-01
7.13653564e-01 -3.07119335e-03 -6.95098750e-03 9.77661431e-01
-5.63556254e-01 9.21694815e-01 -4.76785988e-01 3.52984518e-02
7.49006402e-03 -8.98962319e-02 6.54179931e-01 1.15694213e+00
7.33681440e-01 -1.43459961e-01 -6.66018844e-01 1.32339704e+00
-8.42073336e-02 -4.76820439e-01 -1.25254989e-01 -1.83465108e-01
8.27903628e-01 8.30295086e-01 -1.46959037e-01 -9.09054205e-02
-1.12004150e-02 3.86232615e-01 6.12826757e-02 5.24764895e-01
-1.03603494e+00 -3.62257242e-01 5.63831806e-01 -3.16473722e-01
4.89194691e-01 -1.81416318e-01 -6.07976258e-01 -8.66590679e-01
9.40412953e-02 -8.44325602e-01 -5.50299417e-03 -8.82790983e-01
-1.33783948e+00 6.34911895e-01 2.61864305e-01 -9.55446899e-01
-5.29127955e-01 -6.28977656e-01 -6.60808980e-01 1.44819736e+00
-1.66146314e+00 -9.55224574e-01 1.44710988e-01 6.50997311e-02
1.12731636e-01 -4.36441265e-02 6.31771982e-01 1.87917650e-01
-5.37942708e-01 5.80006719e-01 5.55649161e-01 -2.51623034e-01
5.72251439e-01 -1.18101919e+00 3.47427547e-01 1.01932883e+00
-3.71485800e-02 6.53757095e-01 1.12496471e+00 -6.42236650e-01
-7.06309617e-01 -9.83718932e-01 6.52431846e-01 -4.85290408e-01
8.02102864e-01 -1.14244737e-01 -7.92319298e-01 1.56073108e-01
3.21734965e-01 -3.75045165e-02 5.43774426e-01 6.00680448e-02
-5.41858315e-01 -5.31056747e-02 -1.35545719e+00 2.47957885e-01
4.26398754e-01 -8.33078325e-01 -5.83809435e-01 -6.57179132e-02
1.02945161e+00 -3.59846681e-01 -9.32251096e-01 5.37348628e-01
5.12214065e-01 -1.10196829e+00 1.00348222e+00 -1.97458073e-01
4.22532320e-01 -4.19577003e-01 -6.13384068e-01 -1.63142371e+00
1.90715760e-01 -6.07060134e-01 -6.57328546e-01 1.67148018e+00
6.17625296e-01 -5.09221852e-01 3.25574487e-01 7.16709316e-01
-2.76228160e-01 -4.41410094e-01 -1.29733574e+00 -9.11252022e-01
3.17916483e-01 -1.24285483e+00 9.71598148e-01 2.13835165e-01
-2.61451215e-01 -3.33064049e-01 -6.32418275e-01 7.82440484e-01
9.96388555e-01 -4.39454228e-01 6.06673509e-02 -8.39868605e-01
-5.32901049e-01 -4.37573850e-01 -2.65268851e-02 -5.91899276e-01
1.77488938e-01 -3.46106082e-01 4.22173113e-01 -1.37461782e+00
-1.64954245e-01 -3.12125891e-01 -5.78112423e-01 -8.77683088e-02
-2.59540468e-01 -1.53659374e-01 2.96220370e-02 -2.17273459e-01
-1.40558496e-01 8.51375520e-01 5.45111418e-01 -4.13547993e-01
1.41650289e-01 5.11322260e-01 -6.86064482e-01 7.62784898e-01
9.54006970e-01 -6.10660970e-01 -3.36574852e-01 -3.49682897e-01
1.62580565e-01 4.93659861e-02 5.23683429e-01 -1.26128030e+00
1.70518160e-01 1.08151838e-01 2.68834859e-01 -7.33109891e-01
5.98302305e-01 -7.57350445e-01 2.15784267e-01 1.54085029e-02
-3.73418421e-01 -9.21080887e-01 5.44537723e-01 9.58343267e-01
-2.86175579e-01 -6.71088099e-01 8.09114814e-01 1.49293944e-01
-2.56676674e-01 -2.06095338e-01 -4.62217063e-01 -1.08784035e-01
3.68997514e-01 2.74696350e-01 -4.15011555e-01 -6.98767066e-01
-9.24611986e-01 -2.21571878e-01 -2.97059536e-01 5.67663983e-02
4.67851937e-01 -1.18746972e+00 -7.93262661e-01 -1.65726677e-01
-1.38492495e-01 -4.72029030e-01 7.96274900e-01 6.10576093e-01
1.60952825e-02 4.37131405e-01 3.95810366e-01 -5.97639859e-01
-6.69515014e-01 3.69838685e-01 8.02434385e-01 -2.23618612e-01
1.43330380e-01 9.49825048e-01 -1.21850550e-01 -5.98092437e-01
6.22640193e-01 -5.53504705e-01 -5.01524098e-02 -5.84507883e-02
5.33659875e-01 6.51288211e-01 3.75762373e-01 -4.61149335e-01
-1.51669830e-01 1.74785137e-01 4.25263017e-01 -6.00265265e-01
9.34730709e-01 -3.61772299e-01 -2.46757129e-03 4.18834388e-01
8.36078227e-01 2.40571350e-01 -1.55582225e+00 -3.83587182e-01
-5.05629964e-02 -1.70157164e-01 7.45490432e-01 -1.36147749e+00
-8.46192539e-01 1.23990166e+00 8.46602261e-01 1.05197139e-01
1.26653576e+00 -2.21098185e-01 5.65971136e-01 1.80914998e-01
3.28678131e-01 -1.21901357e+00 -3.81010294e-01 4.51763928e-01
8.73137534e-01 -1.21457744e+00 -4.53928187e-02 -1.10042877e-01
-6.66072369e-01 8.67554843e-01 2.07946718e-01 3.66825253e-01
1.13532543e+00 6.97849154e-01 -2.79719289e-02 2.42507070e-01
-5.52834630e-01 -1.84310794e-01 5.86294711e-01 7.36778438e-01
1.36102617e-01 2.67211825e-01 -1.04137650e-02 1.31626594e+00
-2.57402390e-01 -1.45072147e-01 3.40566367e-01 4.56242889e-01
-4.95749325e-01 -8.36930275e-01 -7.63594210e-01 4.54808146e-01
-6.68332696e-01 -5.58384955e-01 1.70120955e-01 1.86863095e-01
1.07946638e-02 1.42719483e+00 -8.81160945e-02 -3.40186626e-01
1.88980326e-01 2.02557161e-01 3.65683854e-01 -1.54869214e-01
-4.34468895e-01 1.97856113e-01 4.12530333e-01 7.26176798e-02
-2.16855064e-01 -6.53663456e-01 -9.61500525e-01 -1.72767833e-01
-6.67245686e-01 2.50661880e-01 1.25762224e+00 1.04392397e+00
3.39266151e-01 9.69411373e-01 5.67768037e-01 -7.13060975e-01
-1.33885920e+00 -1.18474483e+00 -7.97480583e-01 -3.32301557e-01
5.65013766e-01 -7.02212155e-01 -1.00166965e+00 -2.13056177e-01] | [7.2916412353515625, 3.6626293659210205] |
a75c6f22-2e4c-4840-9839-d991f3778669 | machine-learning-algorithms-for-depression | 2301.03222 | null | https://arxiv.org/abs/2301.03222v1 | https://arxiv.org/pdf/2301.03222v1.pdf | Machine Learning Algorithms for Depression Detection and Their Comparison | Textual emotional intelligence is playing a ubiquitously important role in leveraging human emotions on social media platforms. Social media platforms are privileged with emotional content and are leveraged for various purposes like opinion mining, emotion mining, and sentiment analysis. This data analysis is also levered for the prevention of online bullying, suicide prevention, and depression detection among social media users. In this article, we have designed an automatic depression detection of online social media users by analyzing their social media behavior. The designed depression detection classification can be effectively used to mine user's social media interactions and one can determine whether a social media user is suffering from depression or not. The underlying classifier is made using state-of-art technology in emotional artificial intelligence which includes LSTM (Long Short Term Memory) and other machine learning classifiers. The highest accuracy of the classifier is around 70% of LSTM and for SVM the highest accuracy is 81.79%. We trained the classifier on the datasets that are widely used in literature for emotion mining tasks. A confusion matrix of results is also given. | ['Mubashir Qayoom', 'Furqan Yaqub Khan', 'Danish Muzafar'] | 2023-01-09 | null | null | null | null | ['emotional-intelligence'] | ['natural-language-processing'] | [-2.40879074e-01 2.51497298e-01 -4.02327567e-01 -3.70641142e-01
-1.19451247e-01 6.03223394e-04 1.03778340e-01 5.96224844e-01
-5.24361670e-01 6.81208551e-01 1.87334672e-01 -2.45569423e-01
1.87617421e-01 -7.79794157e-01 1.41046226e-01 -4.19338167e-01
-2.38095626e-01 -2.02335730e-01 -3.92186642e-01 -4.33616400e-01
4.64886487e-01 3.02630007e-01 -1.47574389e+00 5.16068935e-01
7.43450999e-01 1.44991815e+00 -3.89188439e-01 7.28039682e-01
-2.73328543e-01 1.30889523e+00 -6.94176674e-01 -6.20372832e-01
-4.84607905e-01 -2.70372570e-01 -6.66076243e-01 -1.93628818e-01
-6.15809619e-01 -1.38808772e-01 6.03489801e-02 1.01268697e+00
7.02152669e-01 1.33512855e-01 7.68941224e-01 -1.27915871e+00
-5.06007075e-01 3.86024505e-01 -7.42171943e-01 5.52999139e-01
4.78158087e-01 -7.20293343e-01 4.15649772e-01 -7.05338120e-01
3.67128462e-01 9.51527894e-01 6.31130278e-01 5.29206932e-01
-2.33331189e-01 -8.76251519e-01 -2.60909855e-01 2.41610795e-01
-9.18775201e-01 -3.31671000e-01 9.24160421e-01 -6.74908578e-01
1.29789245e+00 -4.78054658e-02 9.55832243e-01 1.12461054e+00
6.83095396e-01 6.08763933e-01 9.00361538e-01 -4.52247232e-01
2.32350141e-01 7.47231007e-01 6.32800281e-01 8.71527612e-01
-3.09033394e-01 -3.81056458e-01 -7.51471162e-01 -4.44230467e-01
-1.73584342e-01 3.53905112e-01 5.50916314e-01 5.18091202e-01
-3.39781165e-01 1.26712465e+00 3.31619591e-01 5.36072254e-01
-7.12339044e-01 -4.06561494e-01 1.13252652e+00 5.93003929e-01
1.15839648e+00 3.14315170e-01 -4.55262423e-01 -3.95042539e-01
-3.74162108e-01 -2.90304959e-01 8.14568102e-01 3.68657075e-02
4.75680411e-01 1.42789990e-01 2.65664667e-01 1.34618342e+00
2.83413202e-01 2.99678653e-01 1.20960784e+00 -4.89711553e-01
-1.83795486e-02 1.13884258e+00 -4.53436464e-01 -1.50520742e+00
-6.05785966e-01 -1.57179222e-01 -8.81200552e-01 -8.01246092e-02
-4.65222985e-01 -9.57303941e-01 -1.37281343e-01 1.40840471e+00
4.51514244e-01 -7.29832649e-02 3.13390851e-01 2.18062177e-01
8.97676826e-01 6.38370752e-01 4.38747078e-01 -3.88847679e-01
1.49498749e+00 -5.08216083e-01 -9.92349863e-01 -1.56588316e-01
1.01501203e+00 -3.32324147e-01 5.92444181e-01 5.81687629e-01
-7.04384863e-01 -4.26516607e-02 -1.08185792e+00 2.40053579e-01
-1.06385195e+00 -9.45334211e-02 1.15845191e+00 9.27612066e-01
-7.04455554e-01 4.02284563e-01 -4.10851896e-01 -6.09661341e-01
7.87124038e-01 6.29663348e-01 -5.26203156e-01 7.14274824e-01
-1.41984296e+00 9.65341628e-01 7.70004988e-02 -1.65993333e-01
8.54661390e-02 9.52809770e-03 -8.82058084e-01 -2.29900017e-01
-2.38338590e-01 -6.70853630e-02 1.07695174e+00 -1.57337046e+00
-1.45283115e+00 1.23260808e+00 -3.93423392e-03 -4.43587184e-01
-1.01371467e-01 -1.35847032e-01 -1.00003624e+00 1.56230301e-01
-4.11020629e-02 1.18117839e-01 7.80041218e-01 -1.67651907e-01
-4.40167665e-01 -1.02484763e+00 -4.70808744e-01 1.79235041e-01
-1.26790285e+00 6.89904809e-01 4.50776756e-01 -3.28962326e-01
-1.76513851e-01 -7.52660930e-01 1.56821668e-01 -5.48130453e-01
-3.64962637e-01 -4.35915440e-01 1.34256828e+00 -7.07536578e-01
1.39617097e+00 -2.01056123e+00 -3.78204912e-01 2.81122684e-01
2.39671603e-01 4.17470366e-01 5.92728853e-01 4.98701006e-01
-1.27319962e-01 2.68499523e-01 3.44085664e-01 -3.00328165e-01
-9.24630612e-02 -2.04886004e-01 -1.90960944e-01 3.66879702e-01
1.30577773e-01 5.43485999e-01 -4.70409364e-01 -3.89210135e-01
1.29905492e-01 6.00757957e-01 -2.71019578e-01 3.23372334e-01
3.10610741e-01 -1.92533523e-01 -7.94875324e-01 6.45808458e-01
1.81176618e-01 -4.93082628e-02 1.47921089e-02 2.37998709e-01
-8.90790150e-02 -4.65756236e-03 -4.31310087e-01 1.00041437e+00
-5.69054067e-01 9.43950355e-01 -1.01521797e-01 -9.87712085e-01
1.23263490e+00 4.42950308e-01 5.39425373e-01 -6.05665982e-01
8.69617105e-01 -4.46847640e-02 -1.94128836e-03 -1.20797086e+00
4.18523222e-01 -3.57949257e-01 -2.98739552e-01 6.93232179e-01
9.12456214e-02 5.74551463e-01 -1.81771845e-01 1.95255011e-01
1.02790260e+00 -5.54950595e-01 6.33289158e-01 6.25614300e-02
4.87818062e-01 -3.78215641e-01 2.22597629e-01 8.66908282e-02
-5.59359193e-01 -3.09853137e-01 6.77678466e-01 -2.73648351e-01
-4.21221852e-01 -1.95814699e-01 -1.17986880e-01 1.57077575e+00
-4.38265920e-01 -2.58878589e-01 -7.70330250e-01 -5.20694375e-01
-9.86077860e-02 4.41127002e-01 -8.25425625e-01 -6.51892364e-01
3.53540927e-01 -9.91046011e-01 7.04505384e-01 3.59575421e-01
6.93284333e-01 -1.47310710e+00 -6.96353793e-01 1.02065064e-01
1.74695328e-01 -8.12899113e-01 3.39937657e-01 2.96602637e-01
-6.05614185e-01 -1.01804733e+00 -1.92796633e-01 -7.86618233e-01
2.33622074e-01 -6.15211613e-02 6.84036493e-01 -5.92752323e-02
-2.28281900e-01 1.92892775e-01 -5.19918323e-01 -1.04508471e+00
-2.27339581e-01 3.30076247e-01 3.15378875e-01 2.79187739e-01
1.08137155e+00 -7.05657065e-01 -3.23936254e-01 -3.23629916e-01
-7.08431304e-01 -2.12438896e-01 7.98983276e-02 3.48150998e-01
-3.18054974e-01 9.06536207e-02 1.54315877e+00 -1.15444374e+00
1.29626513e+00 -1.30012035e+00 3.07802737e-01 -2.20146194e-01
-4.98302072e-01 -5.60637653e-01 3.56909186e-01 -4.54396725e-01
-9.00785506e-01 -4.10496414e-01 -4.46441084e-01 1.53947666e-01
-2.63514221e-01 9.01535273e-01 2.83602536e-01 4.46816050e-02
5.86016178e-01 -3.68810087e-01 9.07732025e-02 -1.45096302e-01
-3.79138470e-01 1.69652176e+00 -2.11586729e-01 2.05584884e-01
-4.69932854e-01 2.18611524e-01 -4.76697713e-01 -1.23081112e+00
-9.64060485e-01 -5.31185091e-01 2.57525519e-02 -6.97999656e-01
1.02466941e+00 -5.36612034e-01 -1.25198567e+00 8.58743012e-01
-8.07630956e-01 2.41278470e-01 4.51593429e-01 3.91919971e-01
-7.74005875e-02 -1.19029023e-01 -1.04907823e+00 -1.41954994e+00
-9.55232620e-01 -4.83043820e-01 5.56791961e-01 4.07696396e-01
-6.59010828e-01 -1.28892052e+00 1.81058809e-01 4.02503371e-01
5.06027877e-01 6.39858961e-01 9.36763465e-01 -1.10427916e+00
1.07594419e+00 -8.01374435e-01 -4.71999161e-02 5.07979691e-01
7.56185427e-02 2.39726260e-01 -1.20368278e+00 2.70918101e-01
3.95997912e-01 -9.52858567e-01 5.93404353e-01 2.16107354e-01
1.05020821e+00 -2.57522881e-01 -1.77642703e-01 -1.00615248e-01
1.06958520e+00 6.09496057e-01 5.58743715e-01 2.98849463e-01
4.17262048e-01 8.89019370e-01 3.78156215e-01 8.49142730e-01
3.72470230e-01 -4.56966041e-03 3.16871077e-01 -5.89588769e-02
9.82658148e-01 2.41965358e-03 6.87942564e-01 7.97598064e-01
3.28385293e-01 -1.25695080e-01 -9.88501430e-01 1.24377221e-01
-1.89160359e+00 -1.04683638e+00 -9.83110666e-02 1.59252167e+00
5.45384347e-01 4.39985394e-02 2.94580936e-01 6.38818800e-01
8.19208920e-01 5.67730926e-02 -6.84475660e-01 -1.32170677e+00
1.02325588e-01 2.67572999e-01 -1.29477620e-01 2.60199755e-01
-1.20608282e+00 7.88065612e-01 5.44470406e+00 3.54390204e-01
-1.52827716e+00 1.69093058e-01 1.04028475e+00 -2.05266356e-01
5.41830540e-01 -7.39903092e-01 -5.21043658e-01 5.27078807e-01
1.37421238e+00 -9.55960527e-02 9.86732077e-03 1.09034336e+00
3.74287128e-01 -4.49561864e-01 -3.59572828e-01 1.04804826e+00
3.57424855e-01 -8.75967622e-01 -4.73754823e-01 4.07277122e-02
4.12696600e-01 5.27941249e-02 1.83564186e-01 5.44022977e-01
-2.60820866e-01 -9.31488454e-01 -9.73509178e-02 6.26365602e-01
4.36855614e-01 -1.25683153e+00 9.21970427e-01 5.06909728e-01
-4.72552985e-01 -3.34039509e-01 -5.38225211e-02 -6.65189803e-01
-5.45226969e-02 6.71654582e-01 -6.87810421e-01 -2.78572202e-01
7.08283663e-01 1.09437108e+00 -3.05042565e-01 2.27016285e-01
1.48217276e-01 5.28166294e-01 -2.01817751e-01 -5.90988398e-01
1.01176918e-01 -2.19480261e-01 6.00124225e-02 1.12364733e+00
1.38954982e-01 4.43394110e-02 -1.64975688e-01 6.53333217e-02
-1.82370231e-01 6.37617826e-01 -1.16381359e+00 -5.13808489e-01
2.18863133e-02 1.49953282e+00 -5.37314832e-01 -2.94632316e-01
-1.90215766e-01 9.61529911e-01 5.41276097e-01 -5.68860094e-04
-6.55065656e-01 -7.68249810e-01 4.83267367e-01 -3.93907502e-02
-4.69471306e-01 2.61637837e-01 -1.74011201e-01 -9.88672376e-01
-4.12263215e-01 -7.10546672e-01 5.77796519e-01 -8.47703516e-01
-1.41743195e+00 7.78360546e-01 -5.69367707e-01 -6.22586250e-01
-5.11549354e-01 -6.58771694e-01 -8.18944931e-01 3.98574263e-01
-1.01250207e+00 -7.11443782e-01 -1.78470805e-01 7.24132717e-01
2.83939421e-01 -5.93515337e-01 1.16967344e+00 3.47723663e-01
-1.16008437e+00 3.34729254e-01 -2.31334597e-01 2.54616052e-01
5.55630147e-01 -8.66544545e-01 -5.24479449e-01 1.96395874e-01
-5.62241614e-01 4.00352865e-01 3.77761871e-01 -6.29511654e-01
-1.27272582e+00 -8.49325478e-01 1.07523572e+00 -1.77543648e-02
7.87490368e-01 -2.15785176e-01 -4.84129310e-01 4.16412979e-01
3.91293526e-01 -4.23321307e-01 1.84057450e+00 2.71264017e-01
1.26128308e-02 5.17377779e-02 -1.61417508e+00 4.36132610e-01
2.90246725e-01 -7.18593299e-01 -2.81844229e-01 5.33474207e-01
1.53489664e-01 2.28926435e-01 -9.82322216e-01 1.30657151e-01
7.89049029e-01 -1.29818141e+00 3.53471577e-01 -1.14024162e+00
9.41425264e-01 6.65413976e-01 -2.04819828e-01 -1.16220164e+00
2.70215362e-01 -2.93117136e-01 -4.62310225e-01 1.31702423e+00
3.81651640e-01 -8.48129034e-01 7.02977240e-01 9.96652007e-01
2.74431854e-01 -1.20283389e+00 -4.05040979e-01 1.66897446e-01
-2.49199018e-01 -5.94862998e-01 -7.08392560e-02 1.34056580e+00
8.99400532e-01 9.39466774e-01 -4.12064791e-01 -4.72852409e-01
-2.50504911e-02 -4.28081721e-01 4.38728303e-01 -1.23867857e+00
3.22173014e-02 -2.96130925e-01 -6.16698623e-01 1.17595205e-02
6.24246657e-01 -7.74946213e-01 -8.29408944e-01 -1.20486009e+00
1.05025582e-01 3.73407602e-02 -3.88394207e-01 3.52853179e-01
3.17065805e-01 3.59208643e-01 -3.25794250e-01 -3.63880306e-01
-5.55366635e-01 5.29979527e-01 5.71645498e-01 1.83099821e-01
-5.33912599e-01 1.17679738e-01 -1.12752426e+00 1.19508469e+00
1.25885332e+00 -3.83053035e-01 -3.46022099e-01 2.33179420e-01
8.46257925e-01 2.13620380e-01 -2.32969165e-01 -7.66889989e-01
7.11511150e-02 1.96268216e-01 5.50415277e-01 -2.68935472e-01
5.57410240e-01 -5.61249077e-01 -3.80656540e-01 3.80564868e-01
-4.28797513e-01 1.69078991e-01 7.68701807e-02 2.65382111e-01
-3.17361623e-01 -2.92782634e-01 8.02735507e-01 1.51343476e-02
-4.32613790e-01 1.64581150e-01 -1.13640976e+00 -1.73805013e-01
1.24353004e+00 -1.12950504e-01 -1.78138018e-01 -7.48694122e-01
-8.58610213e-01 -2.48758271e-02 -1.45200416e-01 4.16800857e-01
6.63288057e-01 -1.06748998e+00 -2.73732305e-01 2.26247579e-01
9.77634862e-02 -9.63155329e-01 3.21930617e-01 7.84852087e-01
-1.22680798e-01 -4.43927683e-02 -3.16895634e-01 1.89978424e-02
-1.57120037e+00 2.72711754e-01 4.21485871e-01 8.74259858e-04
-1.00092389e-01 7.06471443e-01 -5.65683484e-01 -1.21906623e-01
2.47154132e-01 2.03995556e-01 -9.65904772e-01 8.12778890e-01
8.83235753e-01 6.52821958e-01 2.38258913e-01 -7.75793672e-01
-3.19241077e-01 9.40797478e-02 -2.75895059e-01 3.19149494e-02
1.51083577e+00 1.18067168e-01 -4.66453671e-01 9.60525751e-01
1.55870140e+00 -1.05223991e-01 -7.81241897e-03 2.61769354e-01
1.94109939e-02 9.76753235e-02 3.78267407e-01 -6.89710617e-01
-1.11562204e+00 1.03733349e+00 7.59622693e-01 8.53511870e-01
1.02759171e+00 -3.08120519e-01 9.72348869e-01 6.48448527e-01
4.50063637e-03 -1.56603611e+00 2.73364216e-01 7.20485091e-01
5.34779370e-01 -1.44186926e+00 -3.07372510e-01 1.41235694e-01
-1.03928554e+00 9.34890509e-01 6.49352610e-01 -3.61083448e-01
1.41116238e+00 3.97573203e-01 1.85768381e-01 -5.09290993e-01
-8.65475655e-01 8.04860815e-02 -1.47821203e-01 3.31187189e-01
7.69779384e-01 7.67576024e-02 -4.01957124e-01 1.27130818e+00
-2.82483548e-01 5.81809655e-02 3.66982102e-01 9.55474496e-01
-6.28537118e-01 -6.97172642e-01 -1.09687855e-03 1.09775960e+00
-1.31829762e+00 -9.46337078e-03 -9.46330070e-01 3.41980569e-02
2.54148580e-02 1.31741655e+00 -2.40720040e-03 -7.84044862e-01
-2.89705992e-02 5.12733161e-01 -1.84357166e-01 -5.61740756e-01
-1.12199831e+00 -4.62736130e-01 4.61121112e-01 -1.56806052e-01
-5.53069055e-01 -3.26060176e-01 -1.35098338e+00 -4.89511877e-01
-3.27513933e-01 2.50627875e-01 1.02817953e+00 1.02559078e+00
6.27141893e-01 3.06637943e-01 9.00903821e-01 -5.74595511e-01
2.01210946e-01 -1.18596756e+00 -7.98288345e-01 2.15700164e-01
1.23404764e-01 -6.88025773e-01 -3.13756466e-01 -3.48927706e-01] | [12.157289505004883, 6.397804260253906] |
46d194bf-3f7a-4471-8a07-da9368ca600f | refinement-of-direction-of-arrival-estimators | 2106.01011 | null | https://arxiv.org/abs/2106.01011v1 | https://arxiv.org/pdf/2106.01011v1.pdf | Refinement of Direction of Arrival Estimators by Majorization-Minimization Optimization on the Array Manifold | We propose a generalized formulation of direction of arrival estimation that includes many existing methods such as steered response power, subspace, coherent and incoherent, as well as speech sparsity-based methods. Unlike most conventional methods that rely exclusively on grid search, we introduce a continuous optimization algorithm to refine DOA estimates beyond the resolution of the initial grid. The algorithm is derived from the majorization-minimization (MM) technique. We derive two surrogate functions, one quadratic and one linear. Both lead to efficient iterative algorithms that do not require hyperparameters, such as step size, and ensure that the DOA estimates never leave the array manifold, without the need for a projection step. In numerical experiments, we show that the accuracy after a few iterations of the MM algorithm nearly removes dependency on the resolution of the initial grid used. We find that the quadratic surrogate function leads to very fast convergence, but the simplicity of the linear algorithm is very attractive, and the performance gap small. | ['Masahito Togami', 'Robin Scheibler'] | 2021-06-02 | null | null | null | null | ['direction-of-arrival-estimation'] | ['audio'] | [-7.38015026e-02 -1.15433209e-01 2.04544857e-01 -3.21085118e-02
-1.00709319e+00 -6.14640951e-01 3.94437939e-01 -1.44596219e-01
-2.99514383e-01 8.68930757e-01 6.44012332e-01 -2.82621831e-01
-4.05271918e-01 -4.30135220e-01 -4.29487526e-01 -1.00771487e+00
-1.26300797e-01 2.21650880e-02 1.07984081e-01 -6.92190528e-02
1.59024388e-01 3.62233162e-01 -9.75499392e-01 -3.26209068e-01
8.14603984e-01 1.02701807e+00 -1.06871307e-01 6.48642659e-01
7.61051029e-02 6.03613198e-01 -3.88828367e-01 2.51841336e-01
5.17869353e-01 -6.75825000e-01 -1.58746526e-01 1.71495050e-01
3.64936553e-02 -3.84821631e-02 -2.01682106e-01 1.00404918e+00
7.73193181e-01 3.54534090e-01 5.01277745e-01 -1.01088881e+00
6.97887465e-02 3.12316269e-01 -7.38547742e-01 2.42141157e-01
2.81098872e-01 -2.63352185e-01 4.21746045e-01 -1.41474056e+00
3.79301101e-01 8.71310711e-01 1.15755463e+00 -1.33816138e-01
-1.09345043e+00 -4.43004549e-01 -1.29818767e-01 -2.41894379e-01
-1.73949385e+00 -9.14363444e-01 8.63109887e-01 -3.84272188e-01
4.57388431e-01 4.51482773e-01 4.61656928e-01 4.42215472e-01
1.80181116e-02 2.61400819e-01 1.24987006e+00 -8.68631124e-01
6.33007824e-01 1.09013349e-01 2.24117011e-01 7.20614076e-01
4.31539237e-01 1.96408272e-01 -6.02100968e-01 -6.48580790e-01
9.44610298e-01 -3.44893575e-01 -6.75742149e-01 -4.64846373e-01
-1.37679601e+00 8.80345404e-01 7.56600797e-02 4.08300161e-01
-5.03944039e-01 2.33454585e-01 -9.11792368e-02 3.20350021e-01
5.47110617e-01 5.43423831e-01 -2.89939344e-02 -9.09414664e-02
-1.30737710e+00 1.59199938e-01 9.90166903e-01 7.90751159e-01
7.97194362e-01 4.31242257e-01 2.85380296e-02 8.22359085e-01
3.70131701e-01 8.90478849e-01 9.75231305e-02 -1.23187006e+00
2.97207087e-01 -2.44100481e-01 4.71703589e-01 -1.19268775e+00
-5.19712389e-01 -7.75818527e-01 -8.38770151e-01 -4.51165214e-02
5.94672322e-01 -9.02141392e-01 -6.22592568e-01 1.60843349e+00
5.86516619e-01 4.20080811e-01 4.42443369e-03 1.07745957e+00
3.08843851e-01 8.69958818e-01 -7.03783393e-01 -9.79361594e-01
7.63428390e-01 -8.59090030e-01 -1.00758600e+00 -7.08088726e-02
5.92045128e-01 -1.08964348e+00 5.96151352e-01 4.31248814e-01
-1.30521023e+00 -1.93798751e-01 -1.16899669e+00 5.08227348e-01
2.64956087e-01 -3.34259681e-02 5.68910003e-01 8.24983239e-01
-1.30019379e+00 1.13189936e-01 -8.27826023e-01 -2.01057822e-01
-2.90011525e-01 1.52524859e-01 -1.73663884e-01 8.94390270e-02
-7.13033974e-01 6.17646456e-01 -1.40129313e-01 1.45095810e-01
-3.56629580e-01 -7.23829687e-01 -6.59020901e-01 6.22615926e-02
1.88031614e-01 -6.74077928e-01 9.07446027e-01 -8.15452397e-01
-1.71293604e+00 2.09378481e-01 -6.04159772e-01 -2.83885837e-01
2.27799714e-01 -3.66018116e-01 -1.73539579e-01 2.23040342e-01
7.30364919e-02 9.69361514e-02 9.30138528e-01 -1.32005203e+00
-3.00191909e-01 -1.67892292e-01 -4.19924140e-01 4.14175302e-01
-3.26668143e-01 -5.31740002e-02 -3.93725902e-01 -5.96591592e-01
7.75927424e-01 -1.04613245e+00 -7.22872674e-01 -3.99219990e-02
-2.21650809e-01 4.96566921e-01 3.18505734e-01 -4.70323920e-01
1.44980109e+00 -2.26897407e+00 -1.37333330e-02 6.42596066e-01
9.69178453e-02 -2.66715735e-01 4.02445253e-03 6.73952639e-01
-1.31930754e-01 -1.99905559e-01 -2.00770855e-01 -7.75470734e-02
-3.16584885e-01 -9.53715369e-02 -4.03072804e-01 8.71881247e-01
-3.58064085e-01 3.97036552e-01 -7.27976799e-01 -9.39003602e-02
7.77166262e-02 3.38223279e-01 -8.28052521e-01 1.00442328e-01
3.39613497e-01 5.10470867e-01 -3.61115336e-01 2.83624917e-01
8.67397070e-01 -2.98443943e-01 1.05972819e-01 -3.91055614e-01
-6.71205878e-01 2.86127776e-01 -1.83385694e+00 1.49977684e+00
-5.30693650e-01 5.29616356e-01 8.33551168e-01 -1.13856113e+00
7.44842708e-01 5.13830721e-01 7.82125711e-01 -3.01182121e-01
-1.92347929e-01 5.40254712e-01 -1.29546881e-01 -7.74130896e-02
2.88840771e-01 -2.48755425e-01 1.05074935e-01 4.93929416e-01
-2.58280307e-01 -1.73197746e-01 1.10817134e-01 7.44285062e-02
1.06612146e+00 -3.18585098e-01 5.93298256e-01 -8.49936903e-01
3.99844140e-01 1.36595502e-01 7.07262099e-01 8.27140450e-01
1.22428343e-01 6.75813019e-01 1.82111204e-01 -3.28584969e-01
-8.50911796e-01 -9.05498862e-01 -2.93793380e-01 5.82362354e-01
2.33536899e-01 -3.73936862e-01 -4.98407990e-01 -8.11070055e-02
-2.46524975e-01 5.43545723e-01 -2.10360751e-01 2.48539776e-01
-8.57835174e-01 -1.14517701e+00 1.55754715e-01 3.15352142e-01
5.03565729e-01 -1.57683730e-01 -3.92357111e-01 3.90700817e-01
-2.27107704e-01 -1.10050333e+00 -5.71422935e-01 2.52457768e-01
-1.02661085e+00 -5.99019289e-01 -8.75021279e-01 -5.93263209e-01
7.91197181e-01 7.54017830e-01 7.45047331e-01 -2.15549707e-01
4.34075683e-01 7.63989151e-01 -1.41785458e-01 -1.22703843e-01
3.05636972e-02 -3.53731483e-01 3.95865649e-01 1.16817579e-01
-2.07525671e-01 -1.00395477e+00 -5.80482841e-01 3.46951127e-01
-2.97306806e-01 -7.72143528e-02 4.70644116e-01 9.41538811e-01
4.86957610e-01 -4.36435863e-02 4.77485448e-01 -5.71239948e-01
5.56770265e-01 -4.97863144e-01 -6.88885331e-01 -2.11115330e-01
-6.41424894e-01 8.15786645e-02 4.47699994e-01 -2.57550627e-01
-9.19717789e-01 1.87783778e-01 -1.87871382e-01 -3.61473978e-01
1.85337618e-01 8.63918662e-01 1.12425037e-01 -5.02648056e-01
8.47140253e-01 3.33509773e-01 -2.06324637e-01 -4.08128858e-01
3.90266150e-01 3.74782503e-01 4.84659374e-01 -4.32759643e-01
1.00853682e+00 5.93800485e-01 2.49834269e-01 -1.44723952e+00
-7.57761419e-01 -7.18977690e-01 -3.41423184e-01 -1.72644511e-01
4.08649504e-01 -9.77573156e-01 -1.40070349e-01 -4.54766788e-02
-1.04837263e+00 -1.75446451e-01 -3.77274662e-01 1.07267618e+00
-4.80940193e-01 4.79633361e-01 -3.18953007e-01 -9.55506623e-01
-2.34645084e-01 -7.04414546e-01 7.27949381e-01 1.62654612e-02
-1.51469022e-01 -1.00783515e+00 4.76452529e-01 -1.48992240e-01
6.52441978e-01 1.89177021e-01 5.84863305e-01 -2.62642503e-01
-6.37633562e-01 -1.05783314e-01 2.79135723e-02 -1.20071687e-01
5.98475710e-02 -3.83131206e-01 -5.62964559e-01 -4.73893940e-01
6.10653758e-01 1.20365478e-01 5.04378855e-01 9.27938044e-01
4.26824629e-01 -5.84033370e-01 -4.48194325e-01 1.07466066e+00
1.55414784e+00 1.68905661e-01 4.76353496e-01 1.11222334e-01
5.50671637e-01 1.17277920e-01 3.99121523e-01 5.60276508e-01
1.24146186e-01 6.16763651e-01 -2.14879939e-04 -3.21027309e-01
-1.11914590e-01 8.58245865e-02 2.49772489e-01 1.44156897e+00
-3.43108214e-02 -1.55390859e-01 -6.86076880e-01 6.34058356e-01
-1.83414578e+00 -8.47636759e-01 -3.16093832e-01 2.55187416e+00
6.42323911e-01 -1.53185055e-01 9.72831398e-02 1.76334590e-01
4.80688393e-01 3.06564540e-01 -1.10100515e-01 -3.27018708e-01
-3.71278124e-03 2.28354335e-01 7.91788280e-01 9.95144188e-01
-8.73177886e-01 5.30066371e-01 8.10207748e+00 6.64424658e-01
-1.03187621e+00 2.58847058e-01 2.03651428e-01 5.19338660e-02
-4.56569493e-01 1.69294223e-01 -7.43561864e-01 2.73320645e-01
8.35125089e-01 -2.91691422e-01 4.18687701e-01 7.37882197e-01
5.57451844e-01 -1.95013896e-01 -6.76997900e-01 1.10678065e+00
4.56543593e-03 -1.29846656e+00 -5.35129547e-01 9.23917145e-02
1.04416549e+00 -1.37628719e-01 -1.22027293e-01 -1.46681100e-01
1.11407086e-01 -7.05926120e-01 4.50066715e-01 2.91061312e-01
5.24663210e-01 -6.22283399e-01 4.23414886e-01 4.72299010e-01
-1.20534146e+00 -6.70441687e-02 -3.43618453e-01 -2.72227705e-01
5.45755029e-01 1.22672272e+00 -4.52212512e-01 4.00655508e-01
1.76265985e-01 3.10652107e-01 1.40207246e-01 1.26301622e+00
-6.94266558e-02 9.69930172e-01 -9.07935500e-01 1.12838950e-02
2.95851678e-01 -5.10623932e-01 1.23730338e+00 1.13331687e+00
8.84740293e-01 4.42129761e-01 4.79714811e-01 4.63469744e-01
5.37275732e-01 2.26424113e-01 -6.20062470e-01 3.50124031e-01
6.70002162e-01 1.03276026e+00 -6.25510335e-01 -1.27678722e-01
-5.95672369e-01 6.32913589e-01 -2.07916573e-01 7.60228574e-01
-5.10907590e-01 -2.27690771e-01 2.73855865e-01 2.86327571e-01
4.48362738e-01 -9.12940085e-01 -4.94152218e-01 -1.13960290e+00
6.88010156e-02 -9.33516383e-01 1.53871402e-01 -4.39541817e-01
-8.59697402e-01 5.07696271e-01 1.13155387e-01 -1.36213470e+00
-6.08266115e-01 -2.18924299e-01 -6.00779891e-01 8.70085716e-01
-1.16822040e+00 -6.54831290e-01 -1.41822278e-01 5.20083249e-01
3.88631016e-01 -3.96589488e-02 7.26046503e-01 3.93391997e-01
-2.67913908e-01 4.21383679e-01 3.67100477e-01 -2.84980446e-01
4.82791007e-01 -9.30840254e-01 -1.13231756e-01 1.22334683e+00
1.54774353e-01 8.02206397e-01 1.12768614e+00 -3.75154942e-01
-1.43661964e+00 -6.28586769e-01 8.17937136e-01 -1.58976078e-01
5.24147570e-01 -1.33409783e-01 -6.06573224e-01 6.08851075e-01
2.15154797e-01 8.87876824e-02 6.29357576e-01 2.43305817e-01
8.14625695e-02 -1.49991751e-01 -6.43160641e-01 4.55824345e-01
7.61265814e-01 -1.05683900e-01 -3.64593029e-01 5.73335409e-01
2.67885655e-01 -4.70174223e-01 -6.21138036e-01 5.30566037e-01
5.17268240e-01 -1.03655767e+00 1.02178419e+00 -1.73705921e-01
-2.28721410e-01 -5.46714485e-01 -4.89742100e-01 -1.37163115e+00
-7.63351679e-01 -1.11682260e+00 -1.55822992e-01 7.96463966e-01
5.02131820e-01 -7.67058372e-01 6.83619797e-01 2.99762756e-01
-3.57362717e-01 -6.14244998e-01 -1.11705947e+00 -9.05829668e-01
-1.59060732e-01 -2.24593118e-01 2.14497349e-03 1.04901040e+00
7.03296587e-02 5.28724492e-01 -5.69434524e-01 6.53385043e-01
8.63448322e-01 2.81867445e-01 8.58645082e-01 -8.83155525e-01
-6.77992105e-01 -6.66034669e-02 -8.49026314e-04 -1.66134346e+00
-5.07557750e-01 -5.00742733e-01 1.73481435e-01 -1.37252164e+00
-1.17113359e-01 -7.47882307e-01 -7.88491145e-02 8.14488754e-02
-1.42866164e-01 1.52101234e-01 -1.20057911e-01 3.00988913e-01
-2.86889881e-01 4.28135127e-01 9.02870238e-01 2.72752881e-01
-7.52260804e-01 2.15958908e-01 -5.38187325e-01 1.03149247e+00
4.52573299e-01 -5.49908340e-01 -4.77135748e-01 -5.23922443e-01
2.51194686e-01 3.90969306e-01 4.65519093e-02 -1.07732141e+00
4.54692900e-01 -1.06245384e-01 2.61850864e-01 -5.27170777e-01
4.11458015e-01 -5.18204570e-01 4.18054283e-01 4.01966512e-01
8.29883199e-03 6.74264655e-02 6.80257156e-02 4.94870514e-01
-2.16179937e-01 -3.08162481e-01 8.92724216e-01 2.94720605e-02
-3.52069706e-01 5.47346333e-03 -5.40441215e-01 7.02930465e-02
5.83543718e-01 -2.04299539e-01 2.13746578e-02 -1.02837873e+00
-7.35315561e-01 3.18778725e-03 1.96997777e-01 -3.47074807e-01
3.78645450e-01 -1.53551650e+00 -7.69532502e-01 1.12863280e-01
-4.77828622e-01 -4.32999372e-01 -4.36245315e-02 1.36355281e+00
-5.27505755e-01 4.03025240e-01 3.57461929e-01 -7.77982891e-01
-8.58869433e-01 5.03291249e-01 2.56191790e-01 -1.66981235e-01
-5.96613109e-01 8.38115573e-01 3.31879884e-01 -2.04738289e-01
3.05365790e-02 -6.69820756e-02 4.26670946e-02 -1.76076725e-01
5.91831088e-01 5.91844201e-01 8.67348537e-03 -5.09178698e-01
-5.10730505e-01 9.68338728e-01 5.25771320e-01 -6.66850567e-01
1.13687789e+00 -4.02714819e-01 -1.26736030e-01 2.87576854e-01
1.04723012e+00 9.25232708e-01 -9.49816644e-01 -1.50565073e-01
-3.06733400e-01 -5.37138700e-01 4.04308766e-01 -3.40508640e-01
-7.24430323e-01 5.01374424e-01 5.19485891e-01 2.93708563e-01
1.24572110e+00 -4.78386283e-01 6.43145859e-01 4.57207710e-01
4.77411360e-01 -8.00502181e-01 -3.92161384e-02 5.91165960e-01
8.81738901e-01 -7.51120567e-01 3.93906415e-01 -8.76954675e-01
2.09781691e-03 9.74555075e-01 1.47430584e-01 -2.28417680e-01
8.86426270e-01 6.63244009e-01 2.65392452e-01 -2.49958821e-02
-5.21170020e-01 -1.92736074e-01 3.10502201e-01 4.14199054e-01
3.49601239e-01 -3.08821350e-01 -7.66760886e-01 3.67053777e-01
-2.12750420e-01 -2.34452933e-01 5.20872474e-01 7.48382270e-01
-7.32963204e-01 -8.57930660e-01 -8.09774756e-01 1.49238065e-01
-3.21044058e-01 -3.22334141e-01 1.27571210e-01 6.48392439e-01
-2.99161017e-01 1.02873492e+00 -1.27184331e-01 -1.22402780e-01
2.27226436e-01 -6.79520564e-03 4.87517208e-01 -3.52999926e-01
-5.06668277e-02 9.11324084e-01 3.74299854e-01 -5.36194503e-01
-4.62740034e-01 -5.83048642e-01 -9.45745409e-01 -1.41928390e-01
-7.31785238e-01 6.79302216e-01 4.52071041e-01 9.39723253e-01
4.02716428e-01 8.73827636e-02 1.03405488e+00 -7.17571378e-01
-5.07790685e-01 -7.13667989e-01 -5.83716214e-01 -2.45113432e-01
3.91945004e-01 -4.39787447e-01 -7.04721630e-01 -3.01704928e-02] | [6.5186920166015625, 1.401112675666809] |
da9f28e7-906b-489f-b38b-4bed82a5c0c5 | levenshtein-ocr | 2209.03594 | null | https://arxiv.org/abs/2209.03594v2 | https://arxiv.org/pdf/2209.03594v2.pdf | Levenshtein OCR | A novel scene text recognizer based on Vision-Language Transformer (VLT) is presented. Inspired by Levenshtein Transformer in the area of NLP, the proposed method (named Levenshtein OCR, and LevOCR for short) explores an alternative way for automatically transcribing textual content from cropped natural images. Specifically, we cast the problem of scene text recognition as an iterative sequence refinement process. The initial prediction sequence produced by a pure vision model is encoded and fed into a cross-modal transformer to interact and fuse with the visual features, to progressively approximate the ground truth. The refinement process is accomplished via two basic character-level operations: deletion and insertion, which are learned with imitation learning and allow for parallel decoding, dynamic length change and good interpretability. The quantitative experiments clearly demonstrate that LevOCR achieves state-of-the-art performances on standard benchmarks and the qualitative analyses verify the effectiveness and advantage of the proposed LevOCR algorithm. Code is available at https://github.com/AlibabaResearch/AdvancedLiterateMachinery/tree/main/OCR/LevOCR. | ['Cong Yao', 'Peng Wang', 'Cheng Da'] | 2022-09-08 | null | null | null | null | ['scene-text-recognition'] | ['computer-vision'] | [ 6.20866716e-01 -2.25623056e-01 7.78733864e-02 -1.20605603e-01
-6.96466863e-01 -8.14862192e-01 8.87698233e-01 -1.57055393e-01
-3.11744481e-01 4.33426559e-01 2.70201147e-01 -3.27202320e-01
3.73363912e-01 -4.19130266e-01 -8.25424314e-01 -6.34091794e-01
7.28048682e-01 5.80113471e-01 1.17355518e-01 -9.28142763e-05
5.08945823e-01 2.85811961e-01 -1.42243648e+00 6.00127101e-01
1.04040825e+00 8.93221796e-01 5.80289900e-01 9.67799783e-01
-4.62700009e-01 1.05763781e+00 -3.17861408e-01 -6.71913862e-01
1.16686791e-01 -4.99875844e-01 -8.71320665e-01 5.26887536e-01
3.52156341e-01 -2.92257428e-01 -4.65974063e-01 1.27648056e+00
1.48620516e-01 -1.77661836e-01 9.08794820e-01 -1.11339831e+00
-1.04838729e+00 5.21603942e-01 -7.03544736e-01 -6.56825006e-02
7.29155838e-01 2.85582423e-01 1.07031715e+00 -1.35312831e+00
6.31497741e-01 1.30838382e+00 4.43756312e-01 3.18145245e-01
-1.02550530e+00 -3.11607748e-01 8.52877051e-02 5.32019615e-01
-1.35555923e+00 -4.25301462e-01 7.09389210e-01 -5.50411165e-01
9.71277833e-01 4.30849880e-01 6.37315691e-01 1.15147400e+00
3.30014944e-01 1.28484571e+00 9.85689223e-01 -6.62988961e-01
-1.79624036e-01 5.63311055e-02 1.31680056e-01 1.18137527e+00
-6.99030608e-02 -3.42140272e-02 -6.30117416e-01 3.72606367e-01
5.68663776e-01 -2.62592081e-02 -4.78479803e-01 -2.51220167e-01
-1.44376874e+00 6.43714428e-01 2.19253436e-01 1.98098496e-01
-1.04939990e-01 7.75182173e-02 5.82731426e-01 2.21544281e-01
3.01638722e-01 1.59640923e-01 -3.19048166e-02 -1.33604243e-01
-1.20704424e+00 -2.86011770e-02 6.66665137e-01 1.13694322e+00
5.89766681e-01 2.59388715e-01 -4.49844122e-01 8.62183690e-01
4.66114640e-01 7.66070902e-01 6.84768438e-01 -6.31437480e-01
6.04941607e-01 7.26374090e-01 -1.97157681e-01 -8.87580276e-01
-5.05847447e-02 -6.71287030e-02 -1.03284824e+00 8.23802948e-02
2.50677168e-01 2.49122605e-01 -1.12147307e+00 9.44121659e-01
1.90593079e-01 8.75900760e-02 1.20714240e-01 9.10596609e-01
8.66652489e-01 1.08868742e+00 -2.06508502e-01 -2.88829714e-01
1.39271343e+00 -1.43525100e+00 -7.12545514e-01 -2.07024544e-01
4.64968801e-01 -1.12004209e+00 1.32712638e+00 6.14672780e-01
-9.65868235e-01 -5.65752089e-01 -1.00905442e+00 -3.98382634e-01
-2.08359629e-01 5.46874821e-01 1.04238637e-01 3.36367279e-01
-9.44814801e-01 2.80558646e-01 -6.23022020e-01 -3.68983537e-01
3.86928260e-01 -2.18071193e-02 -2.65777290e-01 -6.23370968e-02
-9.03940022e-01 7.52987623e-01 5.01182437e-01 3.23179483e-01
-1.10490108e+00 -3.38694960e-01 -7.97091544e-01 -1.08357750e-01
3.47448915e-01 -5.88158727e-01 1.38982618e+00 -1.26224530e+00
-1.75519407e+00 1.07777154e+00 -2.18637452e-01 -5.35171151e-01
9.62809682e-01 -3.31445515e-01 -6.94915578e-02 2.01260671e-01
-1.04757577e-01 6.27252102e-01 1.30057669e+00 -1.17207754e+00
-6.13524318e-01 -2.30576128e-01 -3.90240818e-01 5.05560458e-01
-2.02016011e-01 9.04977769e-02 -7.29362905e-01 -8.41537952e-01
-1.86949968e-01 -7.40248203e-01 2.27246985e-01 1.30762473e-01
-5.75911224e-01 -2.36711994e-01 9.01076734e-01 -1.13764668e+00
1.12510800e+00 -1.95672739e+00 4.89589900e-01 -9.28812921e-02
2.86291510e-01 3.08777660e-01 -1.96122602e-01 5.19591928e-01
1.26656353e-01 -4.70446646e-02 -4.20957327e-01 -2.70993620e-01
8.26870650e-02 -1.71910316e-01 -5.69038391e-01 5.11874855e-01
5.40028233e-03 1.24190998e+00 -7.65919089e-01 -8.10503960e-01
5.98077953e-01 2.64693499e-01 -2.47981578e-01 2.09551170e-01
-6.38012588e-01 2.75140196e-01 -3.67137998e-01 7.61027455e-01
5.29604793e-01 -3.28038216e-01 3.31724398e-02 -2.05219701e-01
-1.52000532e-01 -9.94893759e-02 -8.00026357e-01 1.66186130e+00
-2.15068698e-01 9.51075256e-01 -1.93042994e-01 -1.08291829e+00
1.11674786e+00 8.23272541e-02 3.17991376e-02 -6.98707998e-01
3.24647814e-01 4.36256640e-03 -4.92487490e-01 -6.06842995e-01
7.03478456e-01 2.08504125e-01 -2.63552926e-02 2.20083654e-01
2.92400327e-02 -5.11134028e-01 2.36518502e-01 1.56151593e-01
8.48145723e-01 7.30068862e-01 5.19630313e-01 -1.53820841e-02
9.14860845e-01 3.40021670e-01 1.77443132e-01 6.65198445e-01
-1.03151836e-01 5.11589766e-01 2.81905532e-01 -3.10059160e-01
-1.45495069e+00 -1.04444969e+00 1.36066988e-01 1.02488112e+00
1.28130212e-01 -4.37300503e-01 -9.65109944e-01 -5.88198841e-01
-2.25783512e-01 9.22113299e-01 -3.35250914e-01 5.74626913e-03
-3.63529235e-01 -2.52039999e-01 7.35854447e-01 3.60697329e-01
8.63455474e-01 -1.19766235e+00 -5.44689119e-01 -4.51314710e-02
-4.57546324e-01 -1.16799104e+00 -7.55081117e-01 -8.03081393e-02
-6.35196805e-01 -7.36234903e-01 -7.65118420e-01 -1.09760380e+00
5.37218034e-01 1.64541274e-01 7.34405935e-01 -4.50358130e-02
-4.28428829e-01 4.19413269e-01 -5.97627997e-01 -1.60186797e-01
-7.82524824e-01 -1.73712239e-01 -2.46214688e-01 2.09024280e-01
2.61610836e-01 -3.16072106e-01 -4.03788894e-01 1.29907414e-01
-7.66385376e-01 8.38717282e-01 7.15045154e-01 1.04780269e+00
6.90109134e-01 -2.21990645e-01 -5.10572679e-02 -7.03304648e-01
6.14364564e-01 1.66545108e-01 -9.47237432e-01 6.33634627e-01
-3.81317198e-01 1.66733488e-01 9.05369341e-01 -4.65487361e-01
-1.18143475e+00 3.61126572e-01 1.60973053e-02 -7.46585071e-01
-1.09179012e-01 3.65624666e-01 9.59918648e-03 1.22065701e-01
5.67303479e-01 1.20930505e+00 -1.71517015e-01 -2.48835906e-01
6.17642343e-01 1.03784311e+00 7.10180402e-01 -2.88964570e-01
8.27099621e-01 3.77658933e-01 -4.59191293e-01 -1.12190461e+00
-6.81344509e-01 -2.82851338e-01 -7.16249406e-01 -4.69474196e-01
9.23286498e-01 -7.39481509e-01 -7.92922556e-01 9.27251995e-01
-1.33250940e+00 -5.24136126e-01 -8.96091983e-02 1.44415721e-01
-8.90790939e-01 8.18810344e-01 -6.91405773e-01 -7.23688781e-01
-7.49865532e-01 -1.09707439e+00 1.16851711e+00 8.55235904e-02
5.82776479e-02 -7.73347855e-01 -6.69121593e-02 5.65557599e-01
-7.30070397e-02 -1.66506991e-01 9.20427799e-01 -4.18479443e-01
-7.58044899e-01 5.62575050e-02 -4.04786229e-01 3.84377897e-01
-5.70751652e-02 3.07655454e-01 -8.13158929e-01 -1.73808724e-01
-9.53702480e-02 -3.59939098e-01 9.35320497e-01 1.69465140e-01
1.23206460e+00 -4.73430067e-01 -2.03187793e-01 7.93140471e-01
1.44888997e+00 1.12217151e-01 7.93290257e-01 3.18978906e-01
1.04471374e+00 2.52203166e-01 6.60487056e-01 4.78012800e-01
2.86975950e-01 4.54768151e-01 3.28380316e-01 1.58984050e-01
-2.74112761e-01 -6.06675625e-01 7.27683306e-01 1.14813685e+00
1.96813032e-01 -5.17923176e-01 -1.05426085e+00 2.82244503e-01
-2.01559377e+00 -8.75504196e-01 -1.20131850e-01 1.85801232e+00
8.02000344e-01 1.19630441e-01 -6.03377633e-02 1.44188985e-01
9.49821115e-01 2.65819788e-01 -5.18249154e-01 -5.95102727e-01
-1.70433804e-01 -2.25104749e-01 5.36561668e-01 6.18848979e-01
-9.45786715e-01 1.44858158e+00 5.43126583e+00 1.20458007e+00
-1.14579940e+00 -1.20829388e-01 5.87641954e-01 3.71712416e-01
-2.70437181e-01 -1.18185170e-01 -8.28715742e-01 4.13526207e-01
6.25494719e-01 -3.63848805e-01 8.12833428e-01 6.45433903e-01
3.60424012e-01 -3.83730722e-03 -1.02794349e+00 1.26627874e+00
4.23245102e-01 -1.60310793e+00 5.93918324e-01 -3.68618339e-01
5.90164423e-01 -5.22266440e-02 3.03648058e-02 1.86965615e-01
2.67050356e-01 -9.83213902e-01 1.23562515e+00 7.74476051e-01
1.08062577e+00 -3.69374692e-01 2.52210528e-01 5.31306386e-01
-1.30354190e+00 -6.02967031e-02 -4.18508768e-01 2.19463766e-01
-1.59993060e-02 3.24340791e-01 -9.98817027e-01 5.05890846e-01
4.92275417e-01 1.06566882e+00 -6.71238780e-01 8.14736664e-01
-4.76351142e-01 6.87090695e-01 -5.77659756e-02 -4.69786733e-01
3.05964708e-01 -3.82036567e-01 7.38668978e-01 1.55403030e+00
2.38593459e-01 -1.77104458e-01 1.33575544e-01 8.73457551e-01
-2.55129695e-01 2.62792796e-01 -5.88081002e-01 -2.94748843e-01
2.75808871e-01 1.22878981e+00 -7.65167892e-01 -5.07680476e-01
-4.04726833e-01 1.50250685e+00 2.60763496e-01 2.78367460e-01
-1.09004104e+00 -3.57214808e-01 -7.10403249e-02 -2.19784752e-01
5.50103545e-01 -1.72636569e-01 -3.84176940e-01 -1.41211605e+00
1.95656940e-01 -1.24563432e+00 1.89642340e-01 -1.26587844e+00
-9.18554008e-01 5.45157671e-01 -2.73484349e-01 -1.38608623e+00
-1.30294189e-01 -8.27656567e-01 -3.73906821e-01 5.14910877e-01
-1.27580166e+00 -1.41430056e+00 -4.58227307e-01 7.03325212e-01
1.39440477e+00 -2.23146453e-01 5.61141431e-01 -1.25934437e-01
-4.59348798e-01 5.87341309e-01 4.44109052e-01 2.92295426e-01
3.57757002e-01 -1.06925416e+00 5.03994644e-01 1.11746931e+00
4.41622615e-01 -8.06892216e-02 7.50373244e-01 -6.81396842e-01
-1.70300186e+00 -1.19250703e+00 6.65972650e-01 -3.31122369e-01
8.47767651e-01 -5.99668443e-01 -8.72526884e-01 7.37391889e-01
4.61448282e-01 -4.71370280e-01 -3.11844703e-02 -7.12588310e-01
-4.74305242e-01 -6.75856741e-03 -8.22327256e-01 9.20068204e-01
1.03165150e+00 -5.19662082e-01 -8.57003808e-01 4.49914813e-01
8.04709554e-01 -4.97609407e-01 -3.90267223e-01 3.70440260e-02
4.53650951e-01 -7.11685956e-01 6.08254492e-01 -1.69891059e-01
7.75831759e-01 -4.78989363e-01 -1.17926307e-01 -8.57231975e-01
-2.40888983e-01 -7.28041291e-01 -5.17818481e-02 1.24962556e+00
4.18398589e-01 -3.27409565e-01 5.85434377e-01 -1.44055307e-01
-1.70569748e-01 -4.56285298e-01 -8.27080607e-01 -5.99561989e-01
3.98260402e-03 -4.97184753e-01 1.82671934e-01 7.02433646e-01
5.62427528e-02 6.07924998e-01 -6.09851718e-01 -2.81869825e-02
6.82271898e-01 2.42397994e-01 6.32175028e-01 -7.77038991e-01
-2.29428902e-01 -4.66630518e-01 -4.24794286e-01 -1.55996048e+00
2.49544144e-01 -1.19750690e+00 3.86861086e-01 -1.62878132e+00
3.93186688e-01 2.12289780e-01 2.30595037e-01 5.08021593e-01
8.95227119e-02 1.61325663e-01 5.41965127e-01 3.05645466e-01
-7.50444591e-01 8.20582509e-01 1.36090052e+00 -6.03574812e-01
1.71056413e-03 -3.28816652e-01 -2.80854255e-01 7.47991920e-01
6.75417185e-01 -1.82728648e-01 -2.24535137e-01 -6.23497844e-01
1.19623244e-01 2.06747904e-01 4.23255324e-01 -8.09070230e-01
4.62197930e-01 -2.18165234e-01 4.55335468e-01 -8.58718455e-01
1.93140015e-01 -7.64903486e-01 -8.07393789e-02 5.98530233e-01
-5.12103617e-01 1.93949237e-01 1.00012720e-01 4.75692958e-01
-1.14548959e-01 -3.51299524e-01 7.77432919e-01 -9.84832528e-04
-9.96824920e-01 8.73824432e-02 -6.29116416e-01 2.55312435e-02
9.75738287e-01 -5.40677130e-01 -4.40448523e-01 -2.97099560e-01
-5.21506608e-01 2.04767704e-01 5.34775257e-01 5.21555424e-01
1.04850650e+00 -1.13459182e+00 -9.69780624e-01 1.40303001e-01
1.60016492e-01 -2.97693938e-01 6.31138962e-03 7.91557789e-01
-9.36583340e-01 5.07105112e-01 -1.14645272e-01 -9.56151426e-01
-1.51397669e+00 7.47960508e-01 4.24431175e-01 -3.09888303e-01
-8.29937458e-01 6.71307445e-01 3.43007565e-01 -3.26163083e-01
2.09857881e-01 -3.21880877e-01 -3.93030830e-02 -1.58874348e-01
3.73103470e-01 2.95954168e-01 -1.01347275e-01 -7.10079968e-01
-2.81604141e-01 6.63080990e-01 -2.53107488e-01 -2.54110426e-01
1.01885378e+00 -3.42689395e-01 -2.05031022e-01 5.09972453e-01
9.59580481e-01 -2.44981721e-01 -1.24046886e+00 -2.64935076e-01
-5.62863946e-02 -3.70468706e-01 -8.08484107e-02 -7.94498920e-01
-7.01411247e-01 9.95624185e-01 4.93503034e-01 -6.90699667e-02
1.08928430e+00 -1.79626331e-01 5.71162879e-01 5.99142849e-01
2.63654534e-02 -1.02826548e+00 3.65085959e-01 8.41494799e-01
1.18404210e+00 -9.05658484e-01 -1.85777962e-01 -2.21347794e-01
-9.27029192e-01 1.32930088e+00 4.34354872e-01 -1.46399319e-01
1.26244769e-01 3.93216610e-01 -1.01799890e-01 1.76126570e-01
-8.36719871e-01 -1.25631720e-01 2.86458790e-01 5.13565958e-01
2.50961751e-01 -4.37860377e-02 -1.92357168e-01 1.27999067e-01
-3.06246847e-01 6.26059547e-02 5.44618547e-01 5.92431247e-01
-6.12278044e-01 -6.86104953e-01 -4.72059697e-01 4.49753135e-01
-1.06218338e-01 -3.93614471e-01 -5.96217215e-01 4.85711396e-01
-2.71486968e-01 7.66504586e-01 -7.22885178e-03 -3.50225359e-01
1.56741828e-01 1.66449308e-01 5.94130039e-01 -3.99222791e-01
-3.42203259e-01 2.89016604e-01 -1.03978664e-01 -3.58447999e-01
-1.78733498e-01 -6.57554150e-01 -1.19081724e+00 -3.13954890e-01
-2.79354930e-01 -3.78695354e-02 4.74574327e-01 1.03844130e+00
-1.13523728e-03 4.32812542e-01 5.80754697e-01 -7.30002880e-01
-5.48944235e-01 -8.96887898e-01 -2.47514427e-01 2.13500991e-01
3.60098869e-01 -1.76839337e-01 -2.53881484e-01 7.65213251e-01] | [11.835088729858398, 2.147157669067383] |
045d5686-a8d0-4671-b2fd-568adb5f76c4 | dynamic-weights-in-multi-objective-deep | 1809.07803 | null | https://arxiv.org/abs/1809.07803v2 | https://arxiv.org/pdf/1809.07803v2.pdf | Dynamic Weights in Multi-Objective Deep Reinforcement Learning | Many real-world decision problems are characterized by multiple conflicting objectives which must be balanced based on their relative importance. In the dynamic weights setting the relative importance changes over time and specialized algorithms that deal with such change, such as a tabular Reinforcement Learning (RL) algorithm by Natarajan and Tadepalli (2005), are required. However, this earlier work is not feasible for RL settings that necessitate the use of function approximators. We generalize across weight changes and high-dimensional inputs by proposing a multi-objective Q-network whose outputs are conditioned on the relative importance of objectives and we introduce Diverse Experience Replay (DER) to counter the inherent non-stationarity of the Dynamic Weights setting. We perform an extensive experimental evaluation and compare our methods to adapted algorithms from Deep Multi-Task/Multi-Objective Reinforcement Learning and show that our proposed network in combination with DER dominates these adapted algorithms across weight change scenarios and problem domains. | ['Ann Nowé', 'Tom Lenaerts', 'Denis Steckelmacher', 'Axel Abels', 'Diederik M. Roijers'] | 2018-09-20 | null | null | null | null | ['multi-objective-reinforcement-learning'] | ['methodology'] | [ 2.05097154e-01 -2.18905255e-01 -2.76607871e-01 -1.43775508e-01
-7.42158055e-01 -5.63689053e-01 2.99108088e-01 3.04094970e-01
-9.18218732e-01 1.28545141e+00 3.91097404e-02 -1.13809161e-01
-7.85055935e-01 -5.82203090e-01 -6.03595853e-01 -7.96921253e-01
-1.00138754e-01 8.26807439e-01 1.70009196e-01 -5.05636930e-01
4.98869151e-01 6.92502689e-03 -1.52025592e+00 -1.20839924e-02
1.10606813e+00 8.76291871e-01 2.33431503e-01 8.14564407e-01
6.18735477e-02 4.78655308e-01 -8.20616543e-01 -2.90396452e-01
4.63347346e-01 -3.81958127e-01 -7.15686858e-01 -2.08394490e-02
3.31534892e-02 -2.69932956e-01 2.55022764e-01 7.96741545e-01
8.09039295e-01 6.07472658e-01 5.44904351e-01 -1.45158434e+00
-6.58666015e-01 6.82232976e-01 -7.12507963e-01 3.04907411e-01
1.93069935e-01 3.13149333e-01 1.10253966e+00 -4.71345007e-01
3.88092428e-01 1.38328505e+00 4.33427751e-01 4.62394387e-01
-1.33115470e+00 -3.58797520e-01 6.52464628e-01 5.26539385e-01
-7.53616810e-01 -1.43251196e-01 8.26139748e-01 -3.35146561e-02
1.09254634e+00 4.82632890e-02 6.71777487e-01 1.21267641e+00
3.20426911e-01 9.05947208e-01 1.32191527e+00 -4.93233085e-01
6.34350002e-01 -1.64476633e-01 -2.43238986e-01 3.31011146e-01
3.30356896e-01 2.76699275e-01 -4.33179200e-01 -7.83593804e-02
4.65811372e-01 -7.81487748e-02 1.36385468e-04 -8.21750104e-01
-1.22463405e+00 9.29315329e-01 2.20526889e-01 2.10449740e-01
-6.18338764e-01 4.18208480e-01 6.18254900e-01 9.66218412e-01
2.83085197e-01 8.70099545e-01 -7.14934528e-01 -2.91036904e-01
-7.33071268e-01 6.71283782e-01 5.93223989e-01 4.82234746e-01
6.38654947e-01 3.13601494e-01 -4.82080370e-01 8.43564630e-01
2.20345948e-02 3.08021516e-01 8.61012697e-01 -1.19981349e+00
5.72762907e-01 4.47380066e-01 6.69176221e-01 -7.25625932e-01
-6.34813964e-01 -6.84252262e-01 -4.64062899e-01 4.92892504e-01
5.60623467e-01 -4.54092979e-01 -8.48255217e-01 1.95280540e+00
3.87817353e-01 -2.09717318e-01 1.81911707e-01 9.64819670e-01
-1.10479183e-02 4.38891083e-01 1.43382043e-01 -4.53223616e-01
9.94769931e-01 -1.04089272e+00 -8.43772471e-01 -3.12838793e-01
3.85753274e-01 -4.86685991e-01 1.19705284e+00 5.98802626e-01
-1.14971268e+00 -4.63067561e-01 -1.14427888e+00 4.74032283e-01
-4.27047372e-01 -2.89663762e-01 6.06359661e-01 6.38159513e-01
-1.01680541e+00 7.87849128e-01 -5.63137889e-01 -1.62908182e-01
1.59840688e-01 6.23325169e-01 2.09394440e-01 1.26936302e-01
-1.56202710e+00 1.33080924e+00 6.07182741e-01 2.55997956e-01
-8.39433134e-01 -4.96285081e-01 -6.12582088e-01 9.67559740e-02
9.90282059e-01 -7.32685745e-01 1.50503898e+00 -1.35284948e+00
-1.94450939e+00 2.19031647e-01 4.08456087e-01 -3.95198375e-01
7.22990096e-01 -2.27822810e-01 -2.20409751e-01 -7.08522573e-02
-2.00204864e-01 5.42243838e-01 1.05924010e+00 -1.13933957e+00
-7.87326694e-01 -2.36622557e-01 5.11486351e-01 7.04492331e-01
-3.28483254e-01 -2.79088676e-01 1.98402002e-01 -5.26511431e-01
-5.60732126e-01 -7.64146328e-01 -5.36428392e-01 -1.61922231e-01
9.47829708e-02 -1.65104643e-01 4.09625173e-01 -1.74089760e-01
1.25716615e+00 -1.72462642e+00 6.69977069e-01 7.56818801e-02
-3.10883168e-02 1.26666367e-01 -5.77095330e-01 6.30735517e-01
1.01134501e-01 -1.97499633e-01 -2.80061096e-01 -1.90819710e-01
3.74500841e-01 4.29304093e-01 1.13335215e-01 3.82604569e-01
2.91638434e-01 8.23266149e-01 -1.17890024e+00 -2.72749931e-01
1.94719970e-01 -5.84617667e-02 -6.82840347e-01 2.80290172e-02
-5.56377828e-01 8.52012262e-02 -5.70803106e-01 5.32355011e-01
4.63985562e-01 -1.44864008e-01 2.90754229e-01 8.20864066e-02
-8.13586488e-02 -2.78364159e-02 -1.63326430e+00 1.60617018e+00
-6.30720735e-01 1.15098422e-02 7.38824755e-02 -1.27782309e+00
6.90326273e-01 2.26434320e-01 5.87790668e-01 -1.08373642e+00
7.91261941e-02 1.08324535e-01 3.54530364e-01 -3.00332367e-01
6.64787292e-01 -2.30995655e-01 -1.44105345e-01 5.64710259e-01
1.21135386e-02 -1.21457994e-01 6.66556239e-01 -2.19011530e-01
1.05100751e+00 4.60532427e-01 4.70688760e-01 -4.25067127e-01
5.11892438e-01 2.21953145e-03 6.73853338e-01 9.59452093e-01
-4.28496957e-01 1.87244073e-01 6.54396892e-01 -4.50876415e-01
-9.04791594e-01 -9.16913807e-01 -1.22762332e-02 1.58001125e+00
2.03982562e-01 1.59616604e-01 -2.95278668e-01 -6.86401427e-01
2.90260464e-01 7.55254805e-01 -7.45463669e-01 -3.55745256e-01
-4.99664634e-01 -1.12945223e+00 2.20182119e-03 3.56047630e-01
3.12869787e-01 -1.38865018e+00 -1.15371621e+00 7.12276757e-01
1.42912105e-01 -5.76511741e-01 -4.29749101e-01 6.74172163e-01
-6.87560856e-01 -1.00942564e+00 -8.18182766e-01 -3.53257835e-01
3.23998839e-01 -1.30983859e-01 1.17649579e+00 -3.35848480e-01
-4.78743762e-03 4.96786833e-01 -3.83089572e-01 -3.78463984e-01
-2.21111178e-01 2.24886477e-01 1.62901863e-01 -1.37607485e-01
-1.35602683e-01 -5.66856086e-01 -7.95085251e-01 3.90295357e-01
-1.14125323e+00 -1.76316157e-01 7.17181087e-01 1.15095782e+00
4.20244992e-01 -8.28608498e-02 1.12531483e+00 -9.37010109e-01
1.12000382e+00 -6.10693932e-01 -8.39129329e-01 5.65336943e-01
-9.30986881e-01 6.23359859e-01 8.67369711e-01 -8.66820097e-01
-1.03066838e+00 -3.12179774e-01 2.24589303e-01 -3.04638177e-01
2.84414083e-01 4.94109720e-01 2.19527915e-01 8.49371180e-02
4.88961548e-01 -1.31658018e-01 1.61872413e-02 -3.20161134e-02
3.44086945e-01 2.28428453e-01 -1.26146451e-02 -9.11549449e-01
6.68412864e-01 3.79111655e-02 3.13808694e-02 1.30382270e-01
-7.72107959e-01 -8.56409967e-02 -1.86364949e-01 -3.10206145e-01
4.20085222e-01 -5.78247309e-01 -9.56416249e-01 3.56085330e-01
-7.05684781e-01 -6.12877309e-01 -4.79716122e-01 3.88872802e-01
-9.86400068e-01 2.76259463e-02 -3.30494016e-01 -9.35903370e-01
-1.84205651e-01 -1.21412313e+00 6.77212656e-01 3.37078631e-01
-9.58752111e-02 -1.17571390e+00 4.01148587e-01 9.18242633e-02
7.09283352e-01 3.50579798e-01 1.02959311e+00 -4.99714464e-01
-1.34783700e-01 3.37908447e-01 1.88456103e-01 3.69461253e-02
2.16658399e-01 -1.00572526e-01 -5.07115483e-01 -5.47101855e-01
-2.35729843e-01 -8.84123862e-01 7.09654331e-01 4.31055337e-01
9.52399313e-01 -3.07022065e-01 2.37152457e-01 2.17057794e-01
1.55271876e+00 3.88618022e-01 2.13093162e-01 8.05385530e-01
1.69741899e-01 4.59514171e-01 8.18165243e-01 8.03998291e-01
4.55999881e-01 6.85824096e-01 7.69876480e-01 5.68546504e-02
4.25456643e-01 1.21237129e-01 5.10182917e-01 4.39915508e-01
-1.24152996e-01 -5.03235042e-01 -5.96734703e-01 5.46272218e-01
-2.26513982e+00 -8.92493784e-01 4.67114091e-01 2.22338676e+00
9.52950180e-01 4.17645991e-01 4.47647631e-01 1.59836009e-01
6.45350814e-01 3.22738856e-01 -1.11385763e+00 -9.44216907e-01
-8.76153335e-02 3.43423635e-01 5.68778336e-01 4.12218690e-01
-7.88201690e-01 6.47816896e-01 6.39161396e+00 7.25103915e-01
-8.54934752e-01 2.33848002e-02 5.19002378e-01 -5.19573629e-01
-3.54866117e-01 -3.05703670e-01 -2.75104553e-01 4.21455294e-01
1.03964305e+00 -3.85830969e-01 7.98551857e-01 4.89317000e-01
1.80372760e-01 -3.36279005e-01 -9.11115527e-01 6.75752819e-01
-1.51345447e-01 -7.82370865e-01 -2.56458312e-01 -2.66325057e-01
8.84635866e-01 -1.74695626e-01 3.60676974e-01 7.61128008e-01
8.62680078e-01 -7.57863402e-01 7.47496128e-01 3.67435426e-01
3.70877326e-01 -1.08964264e+00 7.03642190e-01 1.98110968e-01
-7.93593705e-01 -6.20374322e-01 -2.75693983e-01 -1.77384019e-01
1.49204154e-02 1.88436270e-01 -3.83821666e-01 6.73877656e-01
4.76858795e-01 3.08826774e-01 -2.75583446e-01 8.30736637e-01
4.72847521e-02 3.17287415e-01 -1.27617851e-01 -3.12317133e-01
6.34070456e-01 -2.66789556e-01 4.27721113e-01 6.36470735e-01
2.17730299e-01 -2.31236592e-01 2.03753620e-01 6.22906625e-01
6.92127198e-02 7.66986161e-02 -2.42455527e-01 2.89685559e-02
2.59503067e-01 1.16269970e+00 -6.72565281e-01 -3.89090851e-02
-3.82716775e-01 8.18193316e-01 5.20744622e-01 5.14218509e-01
-9.01800394e-01 -2.05810547e-01 8.37729871e-01 -3.85992467e-01
5.72466731e-01 -2.57822692e-01 9.10161585e-02 -9.27174270e-01
3.11274678e-02 -9.90886271e-01 6.80975974e-01 -4.30207580e-01
-1.39449394e+00 3.20462555e-01 1.33138627e-01 -1.12908995e+00
-4.15400267e-01 -4.81877685e-01 -3.29420298e-01 5.60455859e-01
-2.05862737e+00 -6.54879391e-01 2.49989182e-01 6.27254486e-01
7.82916129e-01 -9.23873708e-02 6.45997763e-01 1.17800280e-01
-6.44798517e-01 5.39359152e-01 5.33655167e-01 -5.94922006e-01
9.05790925e-01 -1.51025867e+00 -1.22265324e-01 3.77024651e-01
-2.99885392e-01 6.88285232e-02 8.90398622e-01 -2.59341002e-01
-1.52831852e+00 -7.41973817e-01 3.18691880e-01 -1.29744574e-01
8.70457053e-01 -1.92454830e-01 -6.82768703e-01 2.82699317e-01
4.29010183e-01 -8.90944451e-02 4.12287176e-01 1.72747597e-01
-5.37344217e-02 -3.71494144e-01 -1.24406087e+00 6.43791676e-01
9.90506947e-01 4.57258373e-02 -5.65680742e-01 1.92992479e-01
5.73099732e-01 -4.58766401e-01 -7.00720191e-01 3.69966775e-01
4.10959035e-01 -8.23543012e-01 8.32445323e-01 -8.91557097e-01
5.27753472e-01 -1.86290100e-01 4.23509115e-03 -1.94560623e+00
-4.39915061e-01 -6.07072532e-01 -2.62051612e-01 7.03396499e-01
3.98575157e-01 -7.07742453e-01 4.30551708e-01 5.21933436e-01
8.04079846e-02 -8.17027926e-01 -1.17662430e+00 -8.08625638e-01
1.49836227e-01 -7.50775561e-02 4.87159759e-01 7.35227346e-01
-1.90835997e-01 1.27674326e-01 -5.52084446e-01 -2.22727284e-01
5.64081788e-01 3.28772426e-01 2.42674202e-01 -9.45762575e-01
-6.39260471e-01 -6.52054608e-01 1.06900167e-02 -4.81973022e-01
1.62553310e-01 -5.35372257e-01 -6.36679605e-02 -1.40672243e+00
-7.19553083e-02 -4.29555506e-01 -9.65725243e-01 4.97960776e-01
-4.46370810e-01 -1.80068403e-01 3.06810051e-01 -2.59855986e-01
-1.20733428e+00 1.05614746e+00 1.44546461e+00 -2.47051194e-01
-4.26579118e-01 6.70937523e-02 -7.70668566e-01 2.96176881e-01
8.41093183e-01 -5.33429384e-01 -7.04086781e-01 -4.39125270e-01
7.37518787e-01 4.03737932e-01 -1.41841501e-01 -8.73543024e-01
2.67243031e-02 -7.67608821e-01 3.39868814e-01 -1.27088204e-02
1.51739895e-01 -1.00255549e+00 -5.55253699e-02 5.92084289e-01
-6.16742611e-01 7.18196929e-01 3.21319789e-01 6.85980380e-01
-2.34074518e-02 -2.36307651e-01 7.12184370e-01 -1.33312181e-01
-6.07415259e-01 1.45003304e-01 -4.67493594e-01 2.28460073e-01
1.15196574e+00 -2.35079318e-01 -2.39594147e-01 -4.10737455e-01
-5.33593416e-01 9.71372724e-01 6.38340265e-02 4.85594064e-01
4.66704667e-01 -1.22305274e+00 -7.30545580e-01 -2.30353057e-01
1.03291571e-01 -2.64911175e-01 2.81538695e-01 6.92713857e-01
-1.94287710e-02 1.67435080e-01 -7.39652455e-01 -1.28477782e-01
-7.56841421e-01 6.40179932e-01 4.81780410e-01 -8.65788579e-01
-4.62519079e-02 3.58734936e-01 -2.31358334e-01 -4.44494069e-01
2.62494326e-01 -3.49318564e-01 -3.09671253e-01 4.80831176e-01
1.67957947e-01 4.34588790e-01 1.03416212e-01 1.77636474e-01
-2.09075600e-01 3.03349584e-01 -2.38244891e-01 -2.46588215e-01
1.59080601e+00 -1.67570874e-01 4.48203593e-01 6.02981448e-01
5.53175688e-01 -5.86681962e-01 -1.52340317e+00 -2.47341886e-01
1.19646773e-01 -4.20156300e-01 4.85263690e-02 -1.16083360e+00
-1.02242339e+00 3.70252579e-01 6.66977704e-01 1.51024580e-01
1.42844307e+00 -6.04201198e-01 3.81122261e-01 7.06224144e-01
5.12747049e-01 -1.67964506e+00 3.19835126e-01 5.42187274e-01
8.57698441e-01 -1.27472055e+00 -2.70593166e-02 5.35190642e-01
-8.87882411e-01 1.15215456e+00 8.49815130e-01 -1.78171948e-01
2.52704442e-01 1.70338839e-01 -3.57577242e-02 1.53419718e-01
-1.33827889e+00 -4.63057339e-01 -7.58407190e-02 4.95983452e-01
2.03215051e-02 -6.08514342e-03 -7.02568710e-01 1.51412070e-01
2.47519717e-01 -5.12390025e-02 4.12453979e-01 1.09360790e+00
-2.38563895e-01 -1.44802892e+00 -3.41843247e-01 4.38368559e-01
-2.93142378e-01 9.39569548e-02 -2.33405922e-02 9.57660317e-01
-2.78910585e-02 7.96278477e-01 1.46893729e-02 4.66544705e-04
4.12675112e-01 3.51371616e-02 7.12600887e-01 -2.50938177e-01
-1.10904980e+00 -8.11245292e-02 4.53248695e-02 -5.50229311e-01
-6.70578539e-01 -7.00834751e-01 -1.09376907e+00 -1.76614463e-01
-2.58020349e-02 6.53597638e-02 5.05712450e-01 9.56876278e-01
1.29238814e-01 8.32360029e-01 7.23836780e-01 -6.45393968e-01
-1.12772334e+00 -8.26404929e-01 -5.20231485e-01 4.18099016e-01
5.71336627e-01 -1.12671804e+00 -3.61552268e-01 -5.46463072e-01] | [4.172831058502197, 2.2688469886779785] |
b0604cda-a7bd-4212-ba7f-a039e5c0c1e6 | on-the-opportunities-and-challenges-of | 2304.06798 | null | https://arxiv.org/abs/2304.06798v1 | https://arxiv.org/pdf/2304.06798v1.pdf | On the Opportunities and Challenges of Foundation Models for Geospatial Artificial Intelligence | Large pre-trained models, also known as foundation models (FMs), are trained in a task-agnostic manner on large-scale data and can be adapted to a wide range of downstream tasks by fine-tuning, few-shot, or even zero-shot learning. Despite their successes in language and vision tasks, we have yet seen an attempt to develop foundation models for geospatial artificial intelligence (GeoAI). In this work, we explore the promises and challenges of developing multimodal foundation models for GeoAI. We first investigate the potential of many existing FMs by testing their performances on seven tasks across multiple geospatial subdomains including Geospatial Semantics, Health Geography, Urban Geography, and Remote Sensing. Our results indicate that on several geospatial tasks that only involve text modality such as toponym recognition, location description recognition, and US state-level/county-level dementia time series forecasting, these task-agnostic LLMs can outperform task-specific fully-supervised models in a zero-shot or few-shot learning setting. However, on other geospatial tasks, especially tasks that involve multiple data modalities (e.g., POI-based urban function classification, street view image-based urban noise intensity classification, and remote sensing image scene classification), existing foundation models still underperform task-specific models. Based on these observations, we propose that one of the major challenges of developing a FM for GeoAI is to address the multimodality nature of geospatial tasks. After discussing the distinct challenges of each geospatial data modality, we suggest the possibility of a multimodal foundation model which can reason over various types of geospatial data through geospatial alignments. We conclude this paper by discussing the unique risks and challenges to develop such a model for GeoAI. | ['Ni Lao', 'Rui Zhu', 'Ziyuan Li', 'Chris Cundy', 'Yingjie Hu', 'Gao Cong', 'Tianming Liu', 'Song Gao', 'Ninghao Liu', 'Deepak Mishra', 'Suhang Song', 'Jin Sun', 'Weiming Huang', 'Gengchen Mai'] | 2023-04-13 | null | null | null | null | ['scene-classification'] | ['computer-vision'] | [ 3.19004357e-01 -2.80015200e-01 -3.32631350e-01 -2.70197868e-01
-1.09155416e+00 -3.00791711e-01 9.61635828e-01 2.22363636e-01
-4.91367877e-01 5.97502351e-01 7.41088390e-01 -6.59997225e-01
-4.31902677e-01 -1.06294906e+00 -5.21811724e-01 -3.25986266e-01
-8.54204297e-02 2.50695556e-01 -7.23795267e-03 -4.09474134e-01
-4.63787876e-02 8.75685662e-02 -1.50516510e+00 2.82653421e-01
8.36347222e-01 6.53481543e-01 3.46390009e-01 7.06478596e-01
-1.95558965e-01 5.90350151e-01 -8.89489874e-02 -2.80573308e-01
-8.84520710e-02 1.25938565e-01 -6.45273149e-01 -6.89178631e-02
4.62814659e-01 -2.63074905e-01 -5.16275764e-01 8.30872595e-01
7.99982965e-01 2.50507325e-01 1.04747629e+00 -1.47721994e+00
-9.25419331e-01 5.22055328e-01 -5.06221354e-01 3.21274579e-01
1.53038502e-01 2.38673091e-01 1.21086848e+00 -1.24347234e+00
4.74924862e-01 1.22816288e+00 1.00712967e+00 3.69066328e-01
-1.11474228e+00 -4.30097610e-01 1.17479533e-01 -4.10172269e-02
-1.55095744e+00 -5.99200368e-01 2.74234831e-01 -7.24900603e-01
1.36401105e+00 3.27616423e-01 -1.22535728e-01 1.13392818e+00
-3.93687718e-04 7.35671520e-01 1.02651513e+00 -2.02597350e-01
2.36551329e-01 -9.42254588e-02 6.81297705e-02 4.33685422e-01
1.46656767e-01 -1.11707628e-01 -6.85605347e-01 -5.06368041e-01
5.35291612e-01 3.51083249e-01 1.03416361e-01 -4.20795120e-02
-1.59064388e+00 9.56725895e-01 4.65325564e-01 3.13803524e-01
-2.57986069e-01 3.03962588e-01 2.88297892e-01 -4.54457197e-03
7.49732614e-01 1.94967180e-01 -4.52184588e-01 -3.31053883e-02
-1.12939358e+00 -4.55349162e-02 6.54281676e-01 7.13645339e-01
9.44321036e-01 2.18593806e-01 -1.87411562e-01 1.18794453e+00
3.35914493e-01 9.94984031e-01 4.17737395e-01 -3.85954976e-01
8.87021780e-01 3.96857798e-01 -6.66487068e-02 -9.35276866e-01
-7.28023887e-01 -1.43864512e-01 -9.20728922e-01 -2.79840231e-01
4.02686223e-02 -5.02558112e-01 -1.39457870e+00 1.85085881e+00
7.17810318e-02 5.58665633e-01 2.48342723e-01 6.95229352e-01
1.26275647e+00 6.91241264e-01 7.70630300e-01 4.63277310e-01
1.47227478e+00 -5.63586354e-01 -3.50851625e-01 -5.33683598e-01
7.58107364e-01 -5.35963953e-01 1.17241085e+00 -5.06339788e-01
-4.69347298e-01 -4.34318721e-01 -7.47074604e-01 1.96844116e-01
-1.16495550e+00 1.10479876e-01 1.02546942e+00 6.02637053e-01
-1.00414515e+00 -2.34896783e-02 -7.84078002e-01 -1.22610343e+00
2.96653748e-01 -7.94916600e-02 -5.21010339e-01 -1.52796760e-01
-1.62114322e+00 1.08631861e+00 3.94543201e-01 -1.70442015e-01
-7.20200658e-01 -8.93391907e-01 -1.14330292e+00 2.72216529e-01
1.72934547e-01 -8.89202297e-01 6.94489777e-01 -3.41118366e-01
-7.61633813e-01 7.82490909e-01 -1.97494626e-01 -2.26650223e-01
-2.37941146e-02 1.51890084e-01 -9.18512940e-01 -8.31325874e-02
7.92475820e-01 8.04865658e-01 5.66167653e-01 -9.52232957e-01
-8.28126013e-01 -2.57873714e-01 -2.02575013e-01 2.51221299e-01
-6.97280407e-01 2.08841324e-01 -3.86628896e-01 -7.85529852e-01
-1.16442107e-01 -7.65093923e-01 -4.77429777e-01 -2.52654940e-01
-3.21462423e-01 2.40708645e-02 7.30718791e-01 -4.97544795e-01
1.36287415e+00 -2.29644632e+00 -1.15903474e-01 1.53498486e-01
3.69483302e-03 1.28630623e-01 -5.05454600e-01 8.29673588e-01
-2.92075262e-03 5.14564455e-01 -4.86959755e-01 -2.40816370e-01
1.25852317e-01 2.78884411e-01 -5.13400733e-01 1.89471945e-01
4.28682357e-01 1.27586102e+00 -1.01503277e+00 -2.63099432e-01
3.96375984e-01 2.29240671e-01 -2.73401111e-01 -2.56866425e-01
-4.94223461e-02 1.92145467e-01 -4.41703051e-01 1.07638526e+00
4.49907303e-01 -5.12269378e-01 -9.74085853e-02 -1.88713055e-02
-1.49668172e-01 -9.22966823e-02 -1.02442336e+00 1.74742758e+00
-4.44892466e-01 5.97797036e-01 -2.27499232e-01 -9.93096888e-01
4.95725065e-01 4.01728302e-01 5.44187367e-01 -5.96884906e-01
-5.05996346e-01 2.93137729e-01 -4.32683110e-01 -6.57244325e-01
6.34017050e-01 -3.32984924e-01 -6.22511089e-01 4.76768941e-01
1.98914409e-01 3.53243798e-02 -5.63209280e-02 2.46843219e-01
1.20107198e+00 -1.51629299e-01 3.23438257e-01 2.63452847e-02
8.40196833e-02 4.21347797e-01 1.67086855e-01 1.15338230e+00
-2.24319741e-01 8.93887401e-01 -2.85420399e-02 -5.00166655e-01
-9.58259940e-01 -1.39364314e+00 -4.16280657e-01 1.66349328e+00
-1.49530321e-01 -4.35507327e-01 -3.57571989e-02 -2.84765810e-01
2.15902850e-01 5.77695787e-01 -4.72647667e-01 3.97393927e-02
6.79973289e-02 -1.66621101e+00 1.16649711e+00 6.22812092e-01
6.41464591e-01 -4.03273463e-01 -2.96167165e-01 3.65247965e-01
-2.89895773e-01 -1.28674090e+00 2.72430535e-02 8.53206441e-02
-5.23199797e-01 -8.11473012e-01 -1.08468032e+00 -3.38211775e-01
2.06964955e-01 7.86953807e-01 8.94290268e-01 -4.78150159e-01
-1.35984197e-01 9.51431751e-01 -2.63572127e-01 -2.97816843e-01
1.37148023e-01 2.60323763e-01 2.29784429e-01 6.65092766e-02
6.33319914e-01 -7.02493429e-01 -4.37840134e-01 2.99411863e-01
-9.52118576e-01 4.77206185e-02 5.79232693e-01 7.11578131e-01
2.26902872e-01 -1.58168882e-01 9.91656125e-01 -6.67439938e-01
7.11010814e-01 -1.30929184e+00 9.78776589e-02 6.94780350e-01
-2.17626780e-01 -3.12513858e-01 7.05950260e-02 -3.72679681e-01
-1.17415953e+00 -2.83615943e-03 9.33337212e-03 -1.60982206e-01
-4.44191515e-01 1.33984792e+00 1.41306624e-01 6.33344725e-02
9.71360028e-01 2.79120833e-01 -3.37978125e-01 -3.70793670e-01
6.19998515e-01 8.73451233e-01 4.31441516e-01 -6.53214812e-01
9.37428832e-01 7.00022340e-01 -2.46885136e-01 -1.30844975e+00
-8.06304097e-01 -9.45787370e-01 -4.06423390e-01 -3.07057938e-03
1.31422901e+00 -1.67306173e+00 1.01105543e-03 5.37107587e-01
-8.68051291e-01 -3.86903942e-01 1.86649948e-01 6.28912330e-01
-3.39075506e-01 1.91895574e-01 -2.67045230e-01 -6.68195546e-01
-9.99030992e-02 -6.99832916e-01 1.33297515e+00 5.96810430e-02
-3.20020646e-01 -1.29713774e+00 3.35612863e-01 3.90353113e-01
8.12156677e-01 1.91867545e-01 1.05700481e+00 -6.89718068e-01
-2.62149870e-01 -5.34670986e-02 -7.08463430e-01 -1.52076662e-01
4.48603213e-01 -1.36833251e-01 -1.15141046e+00 4.67891172e-02
-6.48941755e-01 -4.38128173e-01 1.31853747e+00 3.74417514e-01
6.47101998e-01 4.79741767e-02 -3.98128092e-01 7.04261303e-01
1.45486987e+00 -3.20641220e-01 5.28936207e-01 4.12569702e-01
9.41261113e-01 4.15505439e-01 3.88405591e-01 4.32855368e-01
9.61452425e-01 3.39958757e-01 1.72545940e-01 -3.48833650e-01
-8.47789720e-02 -2.85964310e-01 3.07560891e-01 2.91869760e-01
-1.93787962e-01 -2.96197385e-01 -1.65461957e+00 9.71747100e-01
-2.16181946e+00 -1.17645645e+00 -7.59091601e-02 1.93721855e+00
3.73570234e-01 -3.13052654e-01 -2.38068774e-02 -4.78059351e-01
6.73288107e-01 6.03240311e-01 -4.16498929e-01 -1.18784569e-01
-4.94666964e-01 -1.13420956e-01 8.65189254e-01 1.20586373e-01
-1.72852194e+00 1.29635143e+00 6.39863586e+00 9.85399067e-01
-1.15474403e+00 3.69698107e-01 5.22179782e-01 1.42970942e-02
-4.99087363e-01 -7.32854083e-02 -7.88182318e-01 3.27756882e-01
1.14283657e+00 2.72339862e-02 1.59417033e-01 7.24875093e-01
2.33917922e-01 -2.12860301e-01 -7.10046351e-01 9.67656493e-01
-2.21286733e-02 -1.66512167e+00 1.98507264e-01 8.33941475e-02
1.04356372e+00 9.73730505e-01 3.22230041e-01 5.09233296e-01
7.01748371e-01 -1.16618848e+00 3.48772615e-01 4.72953707e-01
1.02470231e+00 -2.13266581e-01 4.65374529e-01 2.44687289e-01
-1.52680337e+00 -2.22525984e-01 -3.95094007e-01 -6.69366792e-02
3.16079587e-01 3.70740265e-01 -5.17763376e-01 7.44563699e-01
8.30731928e-01 9.49089944e-01 -6.67387426e-01 1.07526815e+00
1.27695814e-01 6.36792958e-01 -4.94589061e-01 3.26247722e-01
6.91813231e-01 1.92282006e-01 4.26618904e-01 1.43611491e+00
7.22057104e-01 2.94791967e-01 2.19912678e-01 6.36711657e-01
1.44455805e-01 -3.39171104e-02 -1.22753918e+00 -3.75508189e-01
4.64395344e-01 1.08354092e+00 -4.45593923e-01 -3.80290806e-01
-9.71286356e-01 7.23856747e-01 1.38132185e-01 8.63116562e-01
-8.18804920e-01 -7.09258690e-02 8.04857492e-01 -1.52544633e-01
5.66129908e-02 -3.65245819e-01 -5.00641942e-01 -1.58283830e+00
-5.10425746e-01 -5.29577732e-01 7.31341839e-01 -9.45866942e-01
-1.65854347e+00 1.54935688e-01 2.15075031e-01 -1.40478778e+00
-1.43870771e-01 -4.33982551e-01 -1.00686574e+00 9.10164475e-01
-1.72648025e+00 -1.98331356e+00 -9.33306962e-02 9.29728150e-01
3.42992872e-01 -4.36798960e-01 1.00862503e+00 3.60026807e-01
-5.56374490e-01 1.59431323e-01 2.68834323e-01 2.02074841e-01
8.75270486e-01 -9.33824480e-01 5.44098556e-01 9.17272747e-01
1.41105607e-01 6.58490241e-01 3.49575967e-01 -8.77204716e-01
-1.24659657e+00 -1.49970615e+00 1.14230800e+00 -4.54614729e-01
1.27644670e+00 -1.22919582e-01 -6.47532642e-01 8.26866388e-01
-2.48893782e-01 -1.55071262e-03 1.15091622e+00 4.78926212e-01
-4.87328023e-01 7.28256628e-02 -8.49306345e-01 6.93951786e-01
1.00847363e+00 -1.10021937e+00 -6.49279118e-01 5.24587333e-01
5.32343507e-01 2.11425498e-01 -9.43155944e-01 5.07409871e-01
2.71516562e-01 -2.74654031e-01 1.36248934e+00 -9.98981953e-01
5.75866044e-01 -2.13218480e-01 -7.84782827e-01 -1.34130847e+00
-4.89424676e-01 -4.84129488e-02 3.27156752e-01 1.07446742e+00
7.35099971e-01 -7.46995270e-01 3.83414805e-01 9.39518213e-01
-1.95458978e-01 -4.49611321e-02 -1.12551844e+00 -7.57852197e-01
1.16737396e-01 -9.70261753e-01 4.96442497e-01 1.30603719e+00
4.73341644e-02 3.36568832e-01 -6.67263150e-01 5.77503681e-01
5.66568792e-01 -1.27708003e-01 7.21449912e-01 -1.22085345e+00
7.28509994e-03 -3.98507416e-01 -4.40950066e-01 -4.15426582e-01
1.03011400e-01 -9.15342271e-01 -2.65040308e-01 -1.86768377e+00
3.80740672e-01 -5.69899499e-01 -4.50129956e-01 9.53576446e-01
-3.40660334e-01 3.02449435e-01 4.52886708e-02 3.53197664e-01
-5.81684053e-01 3.83085757e-01 5.37822723e-01 -4.96228606e-01
-1.47871748e-01 -7.02674985e-02 -8.06076467e-01 7.33351827e-01
7.27310538e-01 -2.34872177e-01 -4.67544585e-01 -7.11851060e-01
4.50431257e-01 2.87756249e-02 8.38275433e-01 -9.41862702e-01
4.68300611e-01 -6.06214166e-01 2.25093722e-01 -5.74290395e-01
3.63478363e-01 -5.79563916e-01 1.43827483e-01 3.56819145e-02
1.20901903e-02 -2.08284900e-01 3.05205971e-01 7.69303501e-01
-3.17163348e-01 1.96872339e-01 2.26175189e-01 -3.42519939e-01
-1.58906889e+00 5.20996928e-01 -8.16489935e-01 -7.07061123e-03
8.91784668e-01 -2.75220335e-01 -7.14181066e-01 -3.94521207e-01
-9.34080064e-01 3.90589327e-01 2.47107461e-01 5.91352940e-01
6.98761821e-01 -1.36346281e+00 -9.26174760e-01 1.65110335e-01
7.04949558e-01 -5.38099349e-01 5.36730468e-01 9.19538081e-01
7.21540861e-03 6.91644728e-01 -1.06546693e-01 -5.56300223e-01
-7.95194864e-01 1.96325690e-01 1.87669381e-01 -1.33931279e-01
-3.24960828e-01 7.27985024e-01 4.35327530e-01 -8.51240993e-01
-1.13123290e-01 -5.77033721e-02 -2.33132988e-01 4.73457575e-01
5.39582849e-01 2.53892094e-01 -1.39349267e-01 -8.48398805e-01
-5.27158380e-01 5.70489824e-01 5.16270578e-01 -2.99974740e-01
1.69148183e+00 -3.49074155e-01 1.19999023e-02 5.94223499e-01
1.03754842e+00 -3.58584106e-01 -9.55352426e-01 -4.99998301e-01
2.66431391e-01 -1.11191399e-01 5.10269441e-02 -9.36251760e-01
-6.67532027e-01 9.67008591e-01 5.04015207e-01 1.63155962e-02
8.08045506e-01 2.11495027e-01 6.93479180e-01 5.30393541e-01
3.38694602e-01 -1.19956446e+00 -2.02133879e-01 8.50781083e-01
4.62449729e-01 -1.83858228e+00 -2.20138222e-01 3.19933370e-02
-9.52644169e-01 7.99272537e-01 3.49235952e-01 4.79215056e-01
9.63241339e-01 -3.83461379e-02 -1.73874706e-01 -4.56950217e-01
-7.28436828e-01 -1.02377105e+00 5.37952900e-01 7.45464683e-01
2.44743973e-01 4.31186557e-01 1.84804380e-01 5.63110292e-01
3.55229825e-01 -4.08802330e-02 1.49970561e-01 8.76115561e-01
-7.10776448e-01 -7.16082096e-01 -5.10953486e-01 6.44227386e-01
-1.87362298e-01 -6.84231400e-01 -1.71363518e-01 6.09547913e-01
3.33916396e-02 1.14301789e+00 -6.41745627e-02 -4.40469742e-01
4.22345102e-02 3.35660160e-01 -3.21415067e-01 -9.36231673e-01
-2.69982874e-01 -7.92360902e-02 4.61056501e-01 -3.14395040e-01
-5.67911565e-01 -5.22734821e-01 -9.61416304e-01 -4.29996073e-01
2.00697809e-01 -3.87674987e-01 6.55847907e-01 1.08748257e+00
4.31389421e-01 1.45431265e-01 3.25761855e-01 -9.16722000e-01
-1.67436913e-01 -9.07623291e-01 -4.98374015e-01 3.21301550e-01
7.58376718e-02 -5.83264053e-01 -2.43492246e-01 -2.15349779e-01] | [9.387527465820312, -1.1977037191390991] |
f6ad6e95-8191-462b-bad8-856c18635a84 | focus-on-details-online-multi-object-tracking | 2302.14589 | null | https://arxiv.org/abs/2302.14589v3 | https://arxiv.org/pdf/2302.14589v3.pdf | Focus On Details: Online Multi-object Tracking with Diverse Fine-grained Representation | Discriminative representation is essential to keep a unique identifier for each target in Multiple object tracking (MOT). Some recent MOT methods extract features of the bounding box region or the center point as identity embeddings. However, when targets are occluded, these coarse-grained global representations become unreliable. To this end, we propose exploring diverse fine-grained representation, which describes appearance comprehensively from global and local perspectives. This fine-grained representation requires high feature resolution and precise semantic information. To effectively alleviate the semantic misalignment caused by indiscriminate contextual information aggregation, Flow Alignment FPN (FAFPN) is proposed for multi-scale feature alignment aggregation. It generates semantic flow among feature maps from different resolutions to transform their pixel positions. Furthermore, we present a Multi-head Part Mask Generator (MPMG) to extract fine-grained representation based on the aligned feature maps. Multiple parallel branches of MPMG allow it to focus on different parts of targets to generate local masks without label supervision. The diverse details in target masks facilitate fine-grained representation. Eventually, benefiting from a Shuffle-Group Sampling (SGS) training strategy with positive and negative samples balanced, we achieve state-of-the-art performance on MOT17 and MOT20 test sets. Even on DanceTrack, where the appearance of targets is extremely similar, our method significantly outperforms ByteTrack by 5.0% on HOTA and 5.6% on IDF1. Extensive experiments have proved that diverse fine-grained representation makes Re-ID great again in MOT. | ['Faquan Wang', 'Hongwei Wang', 'Ziwen Zhang', 'Huilin Ding', 'Shoudong Han', 'Hao Ren'] | 2023-02-28 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Ren_Focus_on_Details_Online_Multi-Object_Tracking_With_Diverse_Fine-Grained_Representation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Ren_Focus_on_Details_Online_Multi-Object_Tracking_With_Diverse_Fine-Grained_Representation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['multiple-object-tracking', 'online-multi-object-tracking'] | ['computer-vision', 'computer-vision'] | [ 4.23904508e-01 -3.52520019e-01 -3.68630320e-01 -3.39958489e-01
-8.77027690e-01 -6.30206704e-01 6.21116757e-01 -4.39981818e-01
-1.56293914e-01 8.70778143e-01 1.76462844e-01 4.32156086e-01
7.77069386e-03 -8.13976526e-01 -8.24364185e-01 -9.59135056e-01
2.22670153e-01 5.46787560e-01 5.64075589e-01 -1.62931308e-02
2.41444990e-01 5.15332937e-01 -1.87402272e+00 4.93047506e-01
8.90039563e-01 1.10167158e+00 3.28265280e-01 4.94038343e-01
-3.36830318e-01 3.89244229e-01 -1.04195905e+00 -1.83990687e-01
4.77818996e-01 -2.85232425e-01 -5.95465481e-01 2.24531487e-01
1.22063839e+00 -2.02114545e-02 -2.30101958e-01 1.12080109e+00
4.08122152e-01 1.72479555e-01 7.64197707e-01 -1.34512889e+00
-5.77033103e-01 1.74047351e-01 -1.01417720e+00 2.46964425e-01
7.07659647e-02 4.98421401e-01 9.10696864e-01 -7.18330204e-01
5.99601328e-01 1.49598634e+00 6.95841372e-01 6.37697041e-01
-1.31616855e+00 -9.44173276e-01 2.66965359e-01 1.80509403e-01
-1.47496259e+00 -2.89017648e-01 4.88791168e-01 -2.48548731e-01
5.02417266e-01 6.76085353e-01 5.63437343e-01 1.23917377e+00
3.82009268e-01 6.39141619e-01 1.21675503e+00 -1.25553861e-01
-3.04686099e-01 -2.81777438e-02 2.85059195e-02 7.52157569e-01
5.05690277e-01 2.17568308e-01 -7.66998947e-01 1.23813801e-01
9.82049167e-01 1.29296780e-01 -2.50721633e-01 -4.29143637e-01
-1.35448158e+00 7.15203702e-01 8.46084297e-01 2.83260643e-01
-2.38868296e-01 3.42509896e-01 1.84236497e-01 7.42901564e-02
1.80387542e-01 3.48264396e-01 -1.50491834e-01 2.17897296e-02
-8.69106412e-01 1.01822965e-01 2.73447007e-01 9.13585246e-01
1.12841296e+00 8.74933079e-02 -8.05284739e-01 7.08424687e-01
1.99746370e-01 8.83942723e-01 5.44594765e-01 -8.80531967e-01
4.71073300e-01 8.36425781e-01 7.81624317e-02 -1.20515847e+00
-2.35427395e-01 -7.30966389e-01 -8.27627063e-01 3.85746807e-01
5.70820689e-01 1.44247025e-01 -1.18195701e+00 1.73544240e+00
6.72140121e-01 2.81119734e-01 -2.61269271e-01 1.06629455e+00
8.25464189e-01 4.90844488e-01 1.95770711e-02 1.89577579e-01
1.44994986e+00 -1.19106674e+00 -5.17845929e-01 -3.53870124e-01
4.04620200e-01 -7.65400171e-01 1.02937496e+00 1.14461221e-01
-6.15908921e-01 -9.74071264e-01 -1.01548278e+00 1.66768264e-02
-3.96003187e-01 3.48313391e-01 6.29720747e-01 7.24295974e-01
-8.95216525e-01 4.06305432e-01 -6.34364247e-01 -2.42270306e-01
6.92494094e-01 6.21212542e-01 -5.28441370e-01 -1.76934719e-01
-6.84802353e-01 5.08326471e-01 2.42000505e-01 1.42952874e-01
-9.46165621e-01 -7.33124852e-01 -8.95967901e-01 -2.20290825e-01
2.33368263e-01 -8.20444882e-01 6.32484078e-01 -8.17329764e-01
-1.15694785e+00 7.93960810e-01 -3.93433452e-01 -1.47968128e-01
2.72655666e-01 -8.23420808e-02 -3.35805327e-01 -8.59600119e-03
7.06302524e-01 1.02864218e+00 1.29760158e+00 -1.26051712e+00
-9.86492872e-01 -3.89620245e-01 -1.53669000e-01 1.17275313e-01
-2.85526842e-01 -2.30157584e-01 -5.19019425e-01 -8.03153992e-01
-5.75826690e-02 -9.44832504e-01 -1.08016364e-01 1.87701449e-01
-4.73248065e-01 -1.47136569e-01 1.22392917e+00 -3.46834630e-01
1.12222052e+00 -2.18122840e+00 1.05436385e-01 -8.41549691e-03
3.22304010e-01 2.41136000e-01 -3.74056876e-01 -1.20853953e-01
2.67092943e-01 -1.29092321e-01 -2.26514861e-02 -2.52701640e-01
-1.46937743e-03 3.28974366e-01 -1.49751768e-01 5.34370780e-01
4.30686772e-01 1.23893392e+00 -8.51225555e-01 -6.82493150e-01
4.26605910e-01 4.50713426e-01 -4.40668046e-01 -3.96589600e-02
-1.55389562e-01 6.78643405e-01 -7.40565062e-01 1.02252030e+00
7.81044364e-01 -3.92825752e-01 -1.07006758e-01 -6.70948923e-01
-6.58033565e-02 -5.26815541e-02 -1.22039080e+00 1.68074667e+00
-3.29523116e-01 6.02890015e-01 -1.10577364e-02 -6.21005595e-01
1.16218972e+00 -3.14744294e-01 4.92723256e-01 -8.75614405e-01
1.49212003e-01 8.09821561e-02 -1.21859163e-01 -3.78635079e-02
7.18775690e-01 7.48741627e-02 -4.35892701e-01 2.40970656e-01
-1.43987685e-01 2.08922133e-01 1.61911622e-01 -1.82585511e-02
1.06400013e+00 3.41574371e-01 1.64191157e-01 -2.04280123e-01
4.87819225e-01 2.94722438e-01 7.81859696e-01 8.05699646e-01
-3.92913610e-01 8.42198730e-01 -1.52408123e-01 -4.62859094e-01
-5.37105620e-01 -1.12107015e+00 -1.86557233e-01 1.19697750e+00
6.07837558e-01 -4.75162894e-01 -4.87440169e-01 -1.02689409e+00
1.57177016e-01 1.07517377e-01 -9.27399635e-01 -1.04310021e-01
-8.53549838e-01 -7.29357243e-01 5.32345176e-01 5.86178482e-01
6.14959598e-01 -1.18960595e+00 -6.51275873e-01 9.44675580e-02
-2.91135967e-01 -1.02802253e+00 -7.10969985e-01 1.53900370e-01
-6.06650174e-01 -1.19043756e+00 -6.69121921e-01 -5.22185683e-01
6.79174423e-01 6.22364700e-01 9.88342941e-01 2.87304707e-02
-6.25257730e-01 5.84709793e-02 -4.20202911e-01 9.29697081e-02
1.05247609e-01 4.64946516e-02 2.21610107e-02 3.25671703e-01
3.26311618e-01 -2.92286485e-01 -6.81933284e-01 8.21309149e-01
-7.77334452e-01 -9.39452797e-02 7.57450461e-01 8.11010242e-01
9.26912546e-01 -4.39582840e-02 2.08442807e-01 -5.56475222e-01
1.42320031e-02 -1.88184083e-01 -3.89956206e-01 2.42297307e-01
-4.08054292e-02 9.82850119e-02 4.92611378e-01 -6.18114531e-01
-9.61616397e-01 1.37072146e-01 2.44237825e-01 -7.01154232e-01
-2.56680459e-01 -4.41484869e-01 -3.23609740e-01 -4.73599106e-01
7.17332482e-01 2.52402753e-01 -1.82887256e-01 -5.32363772e-01
4.53093618e-01 5.50098777e-01 6.93647921e-01 -6.18596494e-01
1.10173166e+00 6.94998264e-01 1.17191315e-01 -5.44371724e-01
-8.78403783e-01 -4.10644352e-01 -7.07648754e-01 -2.07233116e-01
9.99915421e-01 -8.55351985e-01 -6.27405405e-01 2.89912850e-01
-8.07478964e-01 -1.24461755e-01 -5.07145047e-01 2.22963452e-01
-3.24303120e-01 8.99774209e-02 -2.01220796e-01 -3.35665345e-01
-1.58355549e-01 -1.17664278e+00 1.60264444e+00 6.23073161e-01
-2.46317953e-01 -6.39938474e-01 -1.34678513e-01 4.49293792e-01
4.24946696e-01 4.90534902e-01 2.49541074e-01 -2.13522390e-01
-1.00056767e+00 3.34310941e-02 -4.75490242e-01 -8.00208822e-02
5.41741550e-01 -1.45945340e-01 -1.05595660e+00 -3.82134199e-01
-3.52296084e-01 -2.44514421e-01 1.00792837e+00 5.32023430e-01
1.10648930e+00 -4.13224809e-02 -8.71279001e-01 8.55831206e-01
1.19338381e+00 -8.80216733e-02 4.68538076e-01 5.21593869e-01
1.21981680e+00 4.82549727e-01 1.28155863e+00 3.38323206e-01
2.18287215e-01 1.08286822e+00 5.03949106e-01 1.03729732e-01
-7.74750590e-01 -2.29628459e-01 4.34801906e-01 1.53922051e-01
2.54439544e-02 -1.06378436e-01 -3.91717911e-01 4.73944902e-01
-1.74673021e+00 -9.13893223e-01 -2.44996548e-01 2.03088117e+00
6.45074010e-01 -4.29456532e-02 2.39252761e-01 -1.43839613e-01
1.00736320e+00 4.09708619e-01 -5.74374139e-01 1.36831984e-01
-2.17612639e-01 2.68584907e-01 7.03415692e-01 3.83132935e-01
-1.24623847e+00 1.19754934e+00 5.40505838e+00 1.53424788e+00
-1.17185640e+00 1.15372598e-01 3.18362176e-01 -2.64631659e-01
-1.53828293e-01 -3.08592767e-01 -1.52374399e+00 5.63396692e-01
4.90023732e-01 3.06765616e-01 1.69379398e-01 6.16518140e-01
-1.57076389e-01 9.75667760e-02 -6.16380036e-01 1.14982796e+00
3.61950323e-02 -1.46331096e+00 1.36677951e-01 1.99670956e-01
8.01370919e-01 -6.39640018e-02 1.20276414e-01 4.30839717e-01
2.10713550e-01 -1.06896234e+00 6.63455367e-01 4.08024669e-01
1.18102586e+00 -5.04472971e-01 4.31097776e-01 8.78515691e-02
-1.94799900e+00 1.23110984e-03 -5.56652963e-01 1.59668609e-01
1.08719401e-01 3.28798264e-01 -7.04988956e-01 8.44762504e-01
8.28110516e-01 8.86562109e-01 -9.19755697e-01 1.01053953e+00
-2.82729995e-02 1.98810846e-01 -4.18484658e-01 1.38209641e-01
2.33555153e-01 6.26938045e-02 6.02922082e-01 1.10323775e+00
2.45228976e-01 -3.80031407e-01 3.63755345e-01 7.37597227e-01
1.00346297e-01 -2.95939207e-01 -4.44817454e-01 7.70129338e-02
5.38532019e-01 1.60453248e+00 -7.50591457e-01 -2.59665132e-01
-1.31730720e-01 9.69531655e-01 2.44894877e-01 1.31919518e-01
-1.13994861e+00 -2.74838507e-01 9.75548148e-01 2.28789747e-01
8.25636148e-01 2.38074586e-02 -1.52578786e-01 -9.57116902e-01
8.56473744e-02 -8.67801905e-01 5.69322586e-01 -3.79810840e-01
-1.28205180e+00 8.16406369e-01 -3.82344276e-02 -1.74511492e+00
-2.37191543e-01 -5.22556067e-01 -4.42513347e-01 7.55080163e-01
-1.49891686e+00 -1.48527622e+00 -7.08360791e-01 8.11002493e-01
6.52883410e-01 8.10204148e-02 6.41937613e-01 5.41900337e-01
-6.15796506e-01 9.40922141e-01 -2.25517035e-01 1.34713039e-01
8.74618351e-01 -1.00213063e+00 3.21715981e-01 6.58023119e-01
4.31220829e-01 4.68939185e-01 3.87308329e-01 -8.25055003e-01
-1.30805206e+00 -1.61513186e+00 4.44733113e-01 -7.88534701e-01
3.96981061e-01 -2.99080133e-01 -7.46377826e-01 6.30870402e-01
-1.48741543e-01 5.90788424e-01 3.69398266e-01 -1.82061777e-01
-3.78765196e-01 -3.00549299e-01 -1.28321397e+00 4.02289122e-01
1.32570016e+00 -1.73544750e-01 -3.80625963e-01 1.56309471e-01
6.35715425e-01 -5.97805679e-01 -8.14994156e-01 5.57747781e-01
5.11660874e-01 -9.19646621e-01 1.12309480e+00 -2.12921292e-01
-2.44934522e-02 -9.22142506e-01 -3.89856011e-01 -1.03564334e+00
-5.97879648e-01 -5.72264194e-01 -1.37347504e-01 1.32621169e+00
-1.08758016e-02 -5.60430765e-01 1.01378751e+00 -5.25935888e-02
-2.67819434e-01 -6.62718952e-01 -1.11512887e+00 -8.19249511e-01
-3.87085676e-01 -1.49836630e-01 7.33889520e-01 7.42594123e-01
-7.57469594e-01 2.50567704e-01 -4.31086659e-01 2.79504657e-01
8.01413715e-01 6.61171556e-01 1.27189279e+00 -1.22023416e+00
-2.85447359e-01 -2.88548529e-01 -7.05108166e-01 -1.30614984e+00
-1.80555061e-02 -7.46858358e-01 9.41824466e-02 -1.34287488e+00
2.28856057e-01 -7.08650768e-01 -2.94613391e-01 6.11584604e-01
-2.81537712e-01 1.14202821e+00 2.27719098e-01 2.65318245e-01
-8.98906171e-01 7.09191144e-01 1.71941543e+00 -4.08951372e-01
-4.94984537e-02 -1.42118916e-01 -7.43200779e-01 4.55354750e-01
5.80880344e-01 -7.13993073e-01 -8.56566280e-02 -3.04741770e-01
-5.69652200e-01 -4.37020034e-01 6.56866848e-01 -1.28738391e+00
-2.15625465e-02 -3.80923063e-01 8.64981353e-01 -9.83707845e-01
7.03616202e-01 -7.64939487e-01 4.21250105e-01 4.14138705e-01
2.23041922e-01 -2.43525490e-01 3.40100050e-01 6.67188823e-01
-2.62833863e-01 4.75930348e-02 7.97948897e-01 -3.29166837e-02
-1.15781939e+00 5.16001284e-01 2.32444182e-01 6.07808195e-02
1.17825651e+00 -7.73839533e-01 -5.83090961e-01 2.34288663e-01
-5.03319144e-01 1.70704141e-01 6.73198879e-01 6.37934446e-01
4.06085134e-01 -1.67824721e+00 -5.56590140e-01 5.45008481e-01
1.81300953e-01 9.00286585e-02 4.86219853e-01 9.59296882e-01
-6.44136891e-02 4.73069310e-01 -5.35153568e-01 -1.07076931e+00
-1.45993066e+00 1.56485349e-01 1.86679184e-01 -3.60708356e-01
-7.54273236e-01 1.01113582e+00 7.98304141e-01 -2.77680278e-01
3.99763808e-02 -3.46645027e-01 -1.39880911e-01 2.99662706e-02
7.23195553e-01 2.34900177e-01 -3.01208615e-01 -1.04644442e+00
-6.08869433e-01 1.12798882e+00 -1.43719733e-01 2.62039095e-01
9.66009200e-01 -1.90167844e-01 2.26785630e-01 3.07471957e-02
1.02092969e+00 4.09156471e-01 -1.69126964e+00 -4.23336439e-02
-3.99470419e-01 -1.10546339e+00 -2.19788268e-01 -6.31146669e-01
-1.22363555e+00 6.86552644e-01 7.47201443e-01 -5.90423197e-02
1.08800304e+00 1.24683514e-01 8.12376797e-01 -7.92826042e-02
5.43419361e-01 -5.64189851e-01 3.73356819e-01 3.93845558e-01
7.11976230e-01 -1.21953368e+00 -1.15931161e-01 -5.74616253e-01
-5.87578952e-01 8.59115601e-01 1.07314539e+00 -1.31426468e-01
-1.38295576e-01 1.38541982e-01 1.64151013e-01 -3.06823850e-01
-3.21065784e-01 -5.05950272e-01 6.25996530e-01 1.06562114e+00
1.15524955e-01 1.67001337e-01 1.39822230e-01 3.40331584e-01
-3.25064451e-01 -4.07875627e-01 -3.66823897e-02 8.11297357e-01
-6.24740541e-01 -1.25642526e+00 -8.34581971e-01 4.16208446e-01
-2.15204656e-01 9.26712528e-02 -3.26678127e-01 9.95726109e-01
4.46226180e-01 8.51881504e-01 1.55792713e-01 -6.20860636e-01
1.63435072e-01 -4.75999087e-01 7.18992472e-01 -6.64597034e-01
-6.03585482e-01 2.08622012e-02 7.62593821e-02 -9.11577344e-01
-4.20895576e-01 -3.85208070e-01 -1.42981255e+00 -2.16868073e-01
-3.24423105e-01 -7.91803822e-02 3.53521317e-01 7.65679896e-01
4.55907404e-01 8.73786926e-01 4.26979542e-01 -1.23436201e+00
-2.95732707e-01 -9.26123798e-01 -5.05743384e-01 4.66375411e-01
5.08577168e-01 -1.26813042e+00 -2.07554147e-01 -1.89520806e-01] | [6.479081630706787, -2.1026859283447266] |
9a229652-9c84-4b9f-83fd-993eb31097bf | lbdmids-lstm-based-deep-learning-model-for | 2207.00424 | null | https://arxiv.org/abs/2207.00424v1 | https://arxiv.org/pdf/2207.00424v1.pdf | LBDMIDS: LSTM Based Deep Learning Model for Intrusion Detection Systems for IoT Networks | In the recent years, we have witnessed a huge growth in the number of Internet of Things (IoT) and edge devices being used in our everyday activities. This demands the security of these devices from cyber attacks to be improved to protect its users. For years, Machine Learning (ML) techniques have been used to develop Network Intrusion Detection Systems (NIDS) with the aim of increasing their reliability/robustness. Among the earlier ML techniques DT performed well. In the recent years, Deep Learning (DL) techniques have been used in an attempt to build more reliable systems. In this paper, a Deep Learning enabled Long Short Term Memory (LSTM) Autoencoder and a 13-feature Deep Neural Network (DNN) models were developed which performed a lot better in terms of accuracy on UNSW-NB15 and Bot-IoT datsets. Hence we proposed LBDMIDS, where we developed NIDS models based on variants of LSTMs namely, stacked LSTM and bidirectional LSTM and validated their performance on the UNSW\_NB15 and BoT\-IoT datasets. This paper concludes that these variants in LBDMIDS outperform classic ML techniques and perform similarly to the DNN models that have been suggested in the past. | ['Rahamatullah Khondoker', 'O. P. Vyas', 'Ranjana Vyas', 'Uphar Singh', 'P. Aditya Kumar', 'Saksham Sood', 'Kumar Saurabh'] | 2022-06-23 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [-2.79544383e-01 -2.31867164e-01 -4.04059924e-02 -3.07156593e-01
2.34437972e-01 -1.72191918e-01 8.63058150e-01 -6.83017001e-02
-5.28073013e-01 8.97432804e-01 -1.19414970e-01 -7.04646349e-01
-2.97620535e-01 -1.02923930e+00 -3.32902521e-01 -5.55336118e-01
-1.67582512e-01 5.41625500e-01 4.29962635e-01 -1.92652017e-01
1.40265360e-01 8.20557773e-01 -1.27646196e+00 2.91852027e-01
3.07362258e-01 1.52878284e+00 -1.87267587e-01 6.51812673e-01
-4.20645475e-01 1.15582991e+00 -7.33725250e-01 -4.67799067e-01
2.77776808e-01 1.43063992e-01 -7.96287954e-01 -4.61881757e-01
-9.08062533e-02 -1.77939236e-01 -5.22900224e-01 7.96368599e-01
6.97960317e-01 7.55446181e-02 3.43936533e-01 -1.39257491e+00
-6.41053915e-01 6.44880116e-01 -2.55604982e-01 4.99416739e-01
8.92144442e-02 3.39687392e-02 2.92501599e-01 -4.12629634e-01
1.16121709e-01 1.43278611e+00 8.41831446e-01 5.37987828e-01
-7.30368614e-01 -9.14743543e-01 -2.23804548e-01 6.38400912e-01
-1.05233586e+00 -1.74054310e-01 6.29772484e-01 -1.86993122e-01
1.51038933e+00 -2.71584272e-01 3.30282003e-01 1.58831799e+00
6.91201806e-01 5.39900839e-01 1.01919079e+00 -4.62941706e-01
2.75370538e-01 2.64352351e-01 3.63054842e-01 4.92022514e-01
4.81728733e-01 2.10453555e-01 -3.15959342e-02 4.25705872e-02
4.33481008e-01 4.34109896e-01 4.90855277e-01 3.19686323e-01
-7.70464897e-01 9.26007509e-01 6.70473278e-01 1.20908475e+00
-5.23197889e-01 2.32597440e-03 9.20267642e-01 4.62472826e-01
3.97382110e-01 2.12063584e-02 -8.19643557e-01 -2.57918358e-01
-3.87935519e-01 -3.25919777e-01 1.04417694e+00 5.09546638e-01
3.02679926e-01 6.21175766e-01 3.30382496e-01 7.04126120e-01
3.37000996e-01 4.08196509e-01 8.69814456e-01 -3.31618935e-01
3.50718498e-01 8.51569414e-01 -3.06344986e-01 -1.12596512e+00
-7.05442846e-01 -2.60224849e-01 -1.31823170e+00 1.62006274e-01
-3.69518399e-02 -1.67952076e-01 -1.08327746e+00 1.29555905e+00
4.84923832e-02 3.93141985e-01 2.04266623e-01 -1.48287835e-02
8.00478995e-01 7.76752830e-01 4.05797392e-01 3.20338309e-01
1.10553050e+00 -3.68149221e-01 -8.16959023e-01 3.63742933e-02
8.12643409e-01 -5.66545427e-01 4.43917900e-01 5.64098775e-01
-5.48245490e-01 -8.71650219e-01 -1.05997086e+00 2.81708568e-01
-1.30797708e+00 -3.59611630e-01 4.46662635e-01 1.18531191e+00
-1.14665401e+00 7.82401562e-01 -7.78756142e-01 -9.32042718e-01
5.87893188e-01 9.11761642e-01 -2.23065794e-01 2.36787513e-01
-1.36097300e+00 1.12947416e+00 8.38336349e-01 6.40115663e-02
-7.59692192e-01 6.42011017e-02 -3.92530352e-01 -3.12260509e-01
-6.22188337e-02 -5.02553463e-01 9.14489746e-01 -7.43624151e-01
-1.57222891e+00 5.90612888e-01 4.03619111e-01 -1.07282352e+00
-7.21331388e-02 -8.15596655e-02 -1.23966050e+00 -2.61174172e-01
-3.38704109e-01 6.06125295e-01 7.25092113e-01 -8.85659933e-01
-8.31820428e-01 -5.69737971e-01 6.78487495e-02 -6.24924719e-01
-6.93051815e-01 1.72098726e-01 3.46425056e-01 -3.60729873e-01
-2.33684406e-01 -1.02330947e+00 1.10492140e-01 -5.25836587e-01
-6.16612673e-01 -6.17053211e-01 1.86072755e+00 -5.28021097e-01
1.02424359e+00 -1.84990859e+00 -4.60330814e-01 1.62395224e-01
6.31683767e-02 1.34364140e+00 -1.52769657e-02 4.19835120e-01
-1.39353856e-01 4.29638952e-01 2.72373229e-01 -4.16169614e-01
-1.50349051e-01 9.53740895e-01 -1.92178413e-01 2.32984051e-01
-1.25904873e-01 8.30855310e-01 -3.84810984e-01 -2.45534390e-01
7.85305083e-01 7.39195824e-01 1.34906933e-01 1.74743950e-01
-1.65729165e-01 3.77520829e-01 -4.64182675e-01 6.39119565e-01
5.55828691e-01 1.40675232e-01 -2.70182103e-01 -2.98309743e-01
-2.13712305e-01 2.02434525e-01 -6.43605173e-01 1.11723828e+00
-5.84920645e-01 5.78304172e-01 -5.10955751e-01 -1.41627312e+00
1.15848815e+00 6.83491588e-01 5.23015618e-01 -1.22579598e+00
7.06964493e-01 2.43278071e-01 2.37859190e-01 -6.92469716e-01
-1.33995667e-01 -2.14453653e-01 1.01677455e-01 2.57056862e-01
1.58660039e-01 4.17486012e-01 -5.06474152e-02 -1.37909576e-01
1.16584873e+00 -2.67050594e-01 2.56701559e-01 -4.80443500e-02
8.70040834e-01 -4.22926396e-01 3.35519373e-01 7.08014369e-01
-3.39107871e-01 -1.98925167e-01 -5.61986677e-02 -1.11557007e+00
-9.10978913e-01 -8.61745179e-01 -7.07455054e-02 7.06786454e-01
-4.77912754e-01 2.46625207e-02 -4.55524564e-01 -7.82603562e-01
-2.69402623e-01 9.85616148e-01 -4.39048439e-01 -3.05973291e-01
-6.66075945e-01 -6.88022256e-01 1.32347822e+00 6.41423702e-01
1.27378702e+00 -1.57979321e+00 -4.69182462e-01 4.20051187e-01
2.71716446e-01 -1.54222608e+00 6.31650507e-01 4.44076359e-01
-1.10771477e+00 -9.18385923e-01 -1.58292115e-01 -6.34411156e-01
-3.57358716e-02 -1.52069569e-01 8.37616384e-01 -1.19493924e-01
-2.60890752e-01 8.80704746e-02 -5.91266334e-01 -8.86121035e-01
-4.07144606e-01 2.56574690e-01 4.37259853e-01 -1.00450709e-01
7.84588933e-01 -9.36934888e-01 -2.82892823e-01 1.94811255e-01
-1.12218988e+00 -5.76682985e-01 7.67079055e-01 4.50672328e-01
-5.37639223e-02 4.21427667e-01 8.31217468e-01 -5.83390772e-01
4.90558714e-01 -8.04737270e-01 -4.69466865e-01 -5.66825308e-02
-6.26418173e-01 6.75941333e-02 1.05847120e+00 -4.29727197e-01
-9.84573185e-01 -6.16516471e-01 -8.74122858e-01 -3.58704716e-01
-6.40883684e-01 3.08381796e-01 -1.21774465e-01 -2.95049965e-01
4.41328436e-01 -6.06962442e-02 -3.74527216e-01 -7.83228695e-01
-1.68711960e-01 1.20016861e+00 1.85231920e-02 -1.46895871e-01
5.41015327e-01 4.63575870e-01 9.93568376e-02 -1.16815257e+00
-6.51468158e-01 -9.58911628e-02 -6.34544551e-01 -3.61947976e-02
1.04994965e+00 -3.31020296e-01 -9.52614903e-01 8.91465724e-01
-1.24307728e+00 -1.58297092e-01 -1.84430093e-01 4.23220724e-01
1.28134340e-01 1.36350274e-01 -8.54448795e-01 -9.75128233e-01
-6.09029591e-01 -9.54610050e-01 5.66783488e-01 2.66982257e-01
-6.85945526e-02 -1.44612193e+00 7.38978460e-02 1.23085594e-02
9.51339245e-01 3.50218683e-01 1.09534180e+00 -1.19991839e+00
-4.90121208e-02 -5.43415964e-01 -4.54744399e-01 9.69662726e-01
3.46684396e-01 5.60841896e-03 -1.00879252e+00 -1.15310743e-01
4.25087392e-01 -6.78723864e-03 6.18514478e-01 3.44272435e-01
1.06804645e+00 -4.74636286e-01 -4.57391053e-01 4.79045063e-01
1.44827032e+00 1.07132924e+00 9.24595118e-01 6.99309051e-01
7.03702450e-01 1.21124059e-01 -1.01563387e-01 2.83497095e-01
1.82334214e-01 3.15090865e-01 6.77358627e-01 -2.93853576e-03
2.58704070e-02 9.51741412e-02 4.18202847e-01 8.69495213e-01
-1.17972806e-01 -7.01814175e-01 -8.82545233e-01 2.11024001e-01
-1.55485761e+00 -9.33913291e-01 -1.69505715e-01 1.87593389e+00
-3.48792411e-03 6.50513053e-01 2.59242684e-01 6.86175585e-01
5.47989070e-01 1.52564347e-01 -4.31592077e-01 -1.16754735e+00
-2.87123471e-02 6.75907850e-01 6.37169898e-01 9.00223032e-02
-1.26889157e+00 8.26906264e-01 5.65387249e+00 7.22373426e-01
-1.45259261e+00 4.33359325e-01 6.36300862e-01 3.51511717e-01
3.36662203e-01 -4.14828926e-01 -8.90922904e-01 6.53921485e-01
1.83883965e+00 4.83484268e-01 8.29090178e-02 8.92351151e-01
2.36907870e-01 6.17893785e-02 -5.82913339e-01 8.19939554e-01
-1.28812343e-01 -1.01192176e+00 1.68844566e-01 2.03690991e-01
4.53288168e-01 6.22354746e-01 1.40464947e-01 4.95729089e-01
5.06008208e-01 -1.02034223e+00 5.30999564e-02 4.11982268e-01
3.48207593e-01 -9.79832411e-01 1.28024423e+00 4.04361904e-01
-1.11152768e+00 -6.22227132e-01 -4.37823385e-01 -3.04104269e-01
8.93262103e-02 8.41223836e-01 -8.86085689e-01 5.97440541e-01
1.06614351e+00 6.80744171e-01 -5.25125921e-01 8.08334768e-01
3.14857364e-02 7.17882872e-01 -5.55382788e-01 -3.57938200e-01
6.57252371e-01 -1.49663314e-02 3.43624204e-01 1.12524390e+00
3.57441962e-01 -1.77803487e-01 5.67718111e-02 4.09986645e-01
-4.05579060e-02 -3.03618401e-01 -9.64470565e-01 -1.82525545e-01
2.50760376e-01 1.11321497e+00 -7.12044597e-01 -4.68557388e-01
-4.17634249e-01 6.93112731e-01 -1.78492725e-01 7.06513375e-02
-9.48164463e-01 -2.91500807e-01 6.14946365e-01 -9.09522250e-02
1.57436475e-01 -3.34214240e-01 -1.64878756e-01 -7.89095819e-01
-4.92611140e-01 -7.32106626e-01 6.22017384e-01 -4.69154626e-01
-1.58494520e+00 1.13693678e+00 -1.19787119e-01 -9.14505661e-01
-2.25318521e-01 -9.11389172e-01 -7.43993878e-01 4.21199828e-01
-1.29065192e+00 -1.27442384e+00 -7.45981336e-02 8.99591804e-01
5.15491068e-01 -3.96636456e-01 1.05091977e+00 7.50089109e-01
-9.44113612e-01 1.81578189e-01 1.17422372e-01 2.53443152e-01
1.68342873e-01 -7.86698997e-01 5.29419899e-01 8.06757450e-01
1.26819313e-01 5.41728437e-01 5.17930806e-01 -6.48111999e-01
-1.00928402e+00 -1.23988354e+00 7.98087001e-01 1.47723528e-02
7.29384243e-01 -1.60867900e-01 -5.74521720e-01 1.05454063e+00
3.93462151e-01 1.55359553e-02 5.22910357e-01 -2.20952049e-01
-2.62633711e-01 -5.69290936e-01 -1.58448613e+00 2.85007805e-01
7.55044580e-01 -4.75740194e-01 -5.81562459e-01 6.56160712e-02
5.29884160e-01 4.53540504e-01 -7.89786637e-01 6.80979967e-01
3.94362837e-01 -1.22002137e+00 1.06619334e+00 -4.70632762e-01
-5.18899202e-01 -1.58052012e-01 -2.07643807e-01 -9.42787290e-01
-1.38354525e-01 -2.62854934e-01 -4.00779575e-01 1.34945476e+00
1.22790985e-01 -1.23652101e+00 6.01418674e-01 4.36355233e-01
-1.08727448e-01 -5.43470562e-01 -1.10337400e+00 -1.00484335e+00
-1.43983871e-01 -7.67634749e-01 4.72153813e-01 6.30528569e-01
-4.52171773e-01 2.58660555e-01 -4.09913957e-01 -6.15284964e-02
4.08360183e-01 -6.75114989e-01 3.45054388e-01 -1.63345718e+00
1.69811055e-01 -3.01752925e-01 -9.34277415e-01 -5.48312783e-01
-4.61510606e-02 -4.59517628e-01 -6.47763908e-01 -1.54917455e+00
-3.67145330e-01 -4.74054217e-01 -8.01377058e-01 8.69311333e-01
7.07511187e-01 3.57412368e-01 -9.65578333e-02 -1.03350624e-01
-4.46778834e-01 2.07613423e-01 4.34932262e-01 -1.65965527e-01
-6.36942014e-02 3.35552037e-01 -2.58109182e-01 9.96247709e-01
1.33225846e+00 -5.58449149e-01 -2.55138427e-01 -2.86928356e-01
7.62502402e-02 -4.70347196e-01 3.37969154e-01 -1.68150485e+00
2.64129460e-01 5.97485185e-01 6.96797848e-01 -8.30703557e-01
4.62504715e-01 -1.20711505e+00 1.98672831e-01 1.00097358e+00
1.43247128e-01 4.42988276e-01 3.37259442e-01 1.49133399e-01
-1.13070883e-01 -5.73067479e-02 8.67349505e-01 -2.48766169e-01
-8.17459166e-01 4.12232906e-01 -6.79010570e-01 -6.21884167e-01
1.14456284e+00 -3.89981687e-01 -2.66779721e-01 -1.65987283e-01
-6.46927655e-01 -3.66152853e-01 -1.40839458e-01 6.23082995e-01
7.01828897e-01 -1.04697752e+00 -8.33290890e-02 2.45050669e-01
-3.88848305e-01 -4.02558893e-01 1.22509696e-01 8.09435427e-01
-4.45100665e-01 9.48243797e-01 -6.13033950e-01 -3.05841416e-01
-1.16983891e+00 8.51813853e-01 3.47261012e-01 -4.99350041e-01
-5.49419641e-01 7.38935709e-01 -5.68096936e-01 -4.00177211e-01
5.17555535e-01 -4.67451960e-01 -5.57817280e-01 -1.60554294e-02
5.57961345e-01 7.74576008e-01 3.65753740e-01 -6.35250509e-01
-5.35076380e-01 3.30971092e-01 -1.10876352e-01 2.74678797e-01
1.61048365e+00 -5.66363968e-02 -2.21633166e-01 5.61741710e-01
1.46587646e+00 -5.36731124e-01 -2.66179293e-01 2.27384293e-03
5.85236847e-01 7.69749358e-02 1.47237897e-01 -8.98864508e-01
-1.31885576e+00 1.01627421e+00 1.14340103e+00 8.02381277e-01
1.10010529e+00 -3.37393522e-01 1.47178340e+00 6.32309198e-01
5.15702486e-01 -9.46767688e-01 2.68713180e-02 7.88063228e-01
1.54815659e-01 -1.11672306e+00 -5.16130209e-01 3.21765363e-01
-1.14334457e-01 1.47456408e+00 3.78850043e-01 -2.71119833e-01
1.12441230e+00 6.68952465e-01 -1.34696424e-01 -2.59483814e-01
-4.70992565e-01 -2.97772199e-01 -9.79266763e-02 7.48733103e-01
1.77101254e-01 -7.96090290e-02 1.25690222e-01 1.64558545e-01
1.21526010e-01 2.88036495e-01 1.01047672e-01 9.19381440e-01
-4.32820439e-01 -1.54063392e+00 -2.93285131e-01 6.77417517e-01
-7.91497290e-01 1.52438328e-01 -3.28947812e-01 9.18615878e-01
6.93269074e-01 1.40285873e+00 -1.49122536e-01 -1.05137563e+00
2.37467691e-01 2.17186749e-01 1.40509978e-01 -1.18529700e-01
-7.38879561e-01 -6.24359429e-01 1.73023671e-01 -6.03239596e-01
-5.38773596e-01 -1.00617148e-01 -1.15296578e+00 -7.66879141e-01
-4.28183377e-01 -3.52098793e-02 1.02048934e+00 1.43099821e+00
3.91658396e-01 6.93929672e-01 5.08293867e-01 -9.01638031e-01
-1.16217874e-01 -1.24623346e+00 -4.73582596e-01 1.92008451e-01
7.48163313e-02 -6.59194052e-01 -2.60281205e-01 -4.60817695e-01] | [5.206057548522949, 7.178478717803955] |
42610569-6482-44f6-acfa-6b52e646072a | anode-unconditionally-accurate-memory | 1902.10298 | null | https://arxiv.org/abs/1902.10298v3 | https://arxiv.org/pdf/1902.10298v3.pdf | ANODE: Unconditionally Accurate Memory-Efficient Gradients for Neural ODEs | Residual neural networks can be viewed as the forward Euler discretization of an Ordinary Differential Equation (ODE) with a unit time step. This has recently motivated researchers to explore other discretization approaches and train ODE based networks. However, an important challenge of neural ODEs is their prohibitive memory cost during gradient backpropogation. Recently a method proposed in [8], claimed that this memory overhead can be reduced from O(LN_t), where N_t is the number of time steps, down to O(L) by solving forward ODE backwards in time, where L is the depth of the network. However, we will show that this approach may lead to several problems: (i) it may be numerically unstable for ReLU/non-ReLU activations and general convolution operators, and (ii) the proposed optimize-then-discretize approach may lead to divergent training due to inconsistent gradients for small time step sizes. We discuss the underlying problems, and to address them we propose ANODE, an Adjoint based Neural ODE framework which avoids the numerical instability related problems noted above, and provides unconditionally accurate gradients. ANODE has a memory footprint of O(L) + O(N_t), with the same computational cost as reversing ODE solve. We furthermore, discuss a memory efficient algorithm which can further reduce this footprint with a trade-off of additional computational cost. We show results on Cifar-10/100 datasets using ResNet and SqueezeNext neural networks. | ['George Biros', 'Kurt Keutzer', 'Amir Gholami'] | 2019-02-27 | null | null | null | null | ['multivariate-time-series-imputation', 'clustering-multivariate-time-series'] | ['time-series', 'time-series'] | [-3.08352083e-01 -8.49739835e-03 4.37149465e-01 -9.00514424e-02
-3.20704579e-01 -3.91880959e-01 2.42512703e-01 -2.48232663e-01
-8.50667357e-01 1.11200237e+00 -3.11811239e-01 -4.96195644e-01
-1.38172694e-02 -6.56345785e-01 -7.29809225e-01 -7.82098532e-01
-3.46674144e-01 1.44483492e-01 1.53996482e-01 -2.15700492e-01
2.00371623e-01 8.73843372e-01 -1.23897111e+00 -1.79229870e-01
1.06735730e+00 1.41694963e+00 -3.97991955e-01 8.05655122e-01
-2.45835662e-01 1.17533720e+00 -3.77595544e-01 -1.23588346e-01
5.00193477e-01 -5.97296357e-01 -9.66534257e-01 -3.37760448e-01
3.79595995e-01 -4.27504897e-01 -4.53659266e-01 1.13261890e+00
6.31968260e-01 6.99111521e-01 5.77809632e-01 -1.06166232e+00
-4.53891188e-01 3.97521198e-01 -3.42744380e-01 2.86920667e-01
-5.45883834e-01 1.33441091e-02 4.28440899e-01 -1.02126920e+00
5.73384702e-01 9.22613382e-01 1.38840973e+00 7.63123751e-01
-1.30172527e+00 -5.06227612e-01 1.15840010e-01 -1.17905691e-01
-1.43691194e+00 -2.22677231e-01 5.21500885e-01 -3.65109324e-01
1.16842318e+00 4.01754618e-01 8.07931364e-01 6.01940691e-01
2.54186958e-01 4.59579408e-01 1.21256435e+00 -1.61471173e-01
5.23170471e-01 1.01096876e-01 3.57026249e-01 9.05176640e-01
-1.87838711e-02 4.69956771e-02 -9.55444425e-02 -1.44948483e-01
1.09040654e+00 -1.40470669e-01 -4.52072144e-01 1.35077968e-01
-6.17020845e-01 1.09688568e+00 6.41771793e-01 3.43820870e-01
-3.51270646e-01 4.41308409e-01 6.79294527e-01 4.59507585e-01
7.73182333e-01 4.98385400e-01 -5.68559527e-01 -1.98142156e-01
-1.19882977e+00 4.22105581e-01 1.15240443e+00 5.57488322e-01
7.53359675e-01 6.35719180e-01 1.85313404e-01 8.73220026e-01
-4.10844088e-02 5.45298867e-02 6.26657426e-01 -1.21188438e+00
2.41130844e-01 2.86111623e-01 -1.05609514e-01 -7.71912336e-01
-5.07974327e-01 -4.80322897e-01 -1.49602842e+00 7.28472531e-01
5.56697607e-01 -7.29074776e-01 -9.23562288e-01 1.67869222e+00
2.62931049e-01 9.03218836e-02 -3.68915237e-02 9.16247368e-01
5.74315012e-01 8.11462879e-01 -1.94896564e-01 -2.25081429e-01
8.40557992e-01 -1.08487308e+00 -5.69891334e-01 -4.65014875e-02
1.12151289e+00 -3.82562995e-01 8.33607316e-01 2.15117797e-01
-1.37547243e+00 -2.78182358e-01 -9.78248596e-01 -3.21383923e-01
-4.71691757e-01 2.52341062e-01 7.17518270e-01 3.57539326e-01
-1.57528353e+00 1.49913168e+00 -1.03444457e+00 -3.16952239e-03
3.03142220e-01 6.09551787e-01 -9.45823640e-02 5.32754302e-01
-1.33626986e+00 9.48722780e-01 3.75063151e-01 6.08298659e-01
-4.38863158e-01 -1.05740142e+00 -6.69397712e-01 1.57177463e-01
4.41872142e-02 -5.42901754e-01 1.01390386e+00 -1.07257998e+00
-1.82403886e+00 4.86288548e-01 -1.62414610e-01 -8.57726932e-01
1.02691007e+00 -1.96951285e-01 -3.71289812e-02 -2.13932246e-01
-2.80642033e-01 5.40445805e-01 5.43597460e-01 -6.24952912e-01
-2.47754648e-01 -2.18448371e-01 -1.73131004e-02 2.23033771e-01
-4.26101625e-01 -1.59906358e-01 5.78930937e-02 -6.29539609e-01
1.97334528e-01 -1.16325247e+00 -6.44590020e-01 4.79646593e-01
-1.99215800e-01 -8.26749727e-02 6.06028080e-01 -7.52205253e-01
1.17095399e+00 -2.00508690e+00 3.03767193e-02 4.51900577e-03
2.92986453e-01 6.20230615e-01 2.43306458e-01 1.03733376e-01
-2.55871326e-01 2.03670278e-01 -5.89206517e-01 -3.80307972e-01
-2.57362217e-01 2.58279324e-01 -4.82430041e-01 5.43121159e-01
9.08546448e-02 7.25927591e-01 -4.06843156e-01 -2.74417788e-01
-1.25318766e-01 9.69799578e-01 -7.55927861e-01 -1.24385886e-01
6.22240156e-02 4.00281072e-01 -2.02371255e-01 -8.76439661e-02
8.46548498e-01 -2.17988029e-01 -1.56725436e-01 -2.13516191e-01
-6.97881639e-01 3.96366626e-01 -1.52742541e+00 1.26260233e+00
-5.28275907e-01 8.92089128e-01 3.65732938e-01 -1.38981915e+00
6.09835148e-01 2.89773792e-01 3.48745853e-01 -2.88880199e-01
3.14724892e-01 6.58490360e-01 -2.00036451e-01 -3.22003633e-01
2.79241323e-01 -3.43273163e-01 4.48044419e-01 2.06439704e-01
-1.39262840e-01 4.99058589e-02 2.09996834e-01 -3.91977988e-02
1.00340497e+00 3.90970223e-02 -9.14993212e-02 -5.87681293e-01
5.37711442e-01 2.50676036e-01 5.24001002e-01 6.02240086e-01
-1.39145449e-01 4.33655858e-01 8.69742036e-01 -6.63823724e-01
-1.01433754e+00 -4.61506069e-01 -3.59290868e-01 7.67572522e-01
-3.36249530e-01 4.56803804e-03 -1.02513409e+00 -7.30593726e-02
-1.67031273e-01 4.50449377e-01 -7.89406538e-01 -2.20768973e-02
-1.12596095e+00 -9.00632739e-01 9.49655294e-01 8.59669149e-01
9.35185134e-01 -1.00565147e+00 -7.57954061e-01 2.96203345e-01
2.61563897e-01 -7.88429141e-01 -3.63633633e-01 5.78139663e-01
-1.55770111e+00 -6.83504462e-01 -1.21267200e+00 -7.46155500e-01
7.86816359e-01 -3.87161523e-01 7.23356247e-01 1.46863371e-01
-1.66677281e-01 -2.16061696e-01 3.18361789e-01 1.39035732e-01
-2.22317263e-01 1.42113134e-01 2.55680889e-01 -2.24544734e-01
-2.78463840e-01 -7.05139697e-01 -5.39793372e-01 -1.01460680e-01
-8.66807699e-01 1.68246046e-01 5.84647618e-02 1.01176441e+00
4.80769366e-01 2.60376092e-02 2.48969808e-01 -8.76234710e-01
7.73868322e-01 -2.36264408e-01 -9.80880797e-01 -8.77050161e-02
-7.30935514e-01 4.45615232e-01 1.11369920e+00 -8.18907678e-01
-1.00374341e+00 -4.68065217e-02 -4.60983723e-01 -5.59645116e-01
3.52386534e-01 6.51661634e-01 5.86806357e-01 -7.15025246e-01
7.77665734e-01 9.68583003e-02 2.59402934e-02 -6.45635784e-01
1.48692131e-01 2.36617237e-01 4.04698581e-01 -4.68582600e-01
4.36545342e-01 4.60361511e-01 3.81777525e-01 -8.04907858e-01
-6.27038836e-01 1.48137078e-01 -4.33604240e-01 1.02243379e-01
6.61725342e-01 -6.82654262e-01 -9.48664546e-01 8.54203105e-01
-1.21362960e+00 -7.53088117e-01 -4.59221721e-01 5.00590265e-01
-3.47005606e-01 2.06075266e-01 -1.40476382e+00 -8.35352778e-01
-4.61370617e-01 -1.10514939e+00 6.32775277e-02 2.89681584e-01
-1.07759140e-01 -1.37056577e+00 -1.52053133e-01 -4.01253223e-01
8.77740800e-01 3.73543262e-01 8.11833203e-01 -3.30679446e-01
-2.91524321e-01 -3.75647545e-02 -3.57124835e-01 6.95257306e-01
-3.71399224e-01 2.37745624e-02 -8.19561124e-01 -2.66917676e-01
6.28343642e-01 -3.28971565e-01 8.98804724e-01 5.27813613e-01
1.20437455e+00 -7.39991367e-01 -7.93474391e-02 1.09266794e+00
1.61990225e+00 8.14338773e-02 4.28517282e-01 4.93365303e-02
8.27763438e-01 3.55359912e-01 -3.14873047e-02 3.07441920e-01
-1.10761918e-01 2.38112345e-01 1.47904530e-01 -4.50547278e-01
2.97932476e-02 2.42718264e-01 3.22147518e-01 1.03269136e+00
-4.32371765e-01 2.26026013e-01 -8.63615096e-01 3.39515656e-01
-1.79376447e+00 -6.02308691e-01 -5.01651347e-01 2.04126430e+00
1.07702482e+00 6.99788705e-02 -3.98482196e-03 3.75986129e-01
4.09217864e-01 -3.05884536e-02 -7.46988714e-01 -8.81523490e-01
-5.10834064e-03 4.68141019e-01 8.55238795e-01 7.74900317e-01
-7.47986615e-01 8.07406485e-01 6.42776251e+00 6.33013129e-01
-1.73839521e+00 3.10124457e-01 7.26207972e-01 -2.86040246e-01
1.90341055e-01 -1.10348850e-01 -7.99149930e-01 3.49900872e-01
1.01500940e+00 -9.69069377e-02 6.47151113e-01 1.01041651e+00
-5.17712794e-02 -5.48570566e-02 -9.60507333e-01 8.64392281e-01
-1.59567654e-01 -1.45773196e+00 -3.13612998e-01 1.96956322e-02
8.31053317e-01 3.98968965e-01 -1.27189934e-01 4.61187720e-01
1.10171691e-01 -1.07707095e+00 5.40132284e-01 1.73164919e-01
7.09809363e-01 -8.37847114e-01 4.01144505e-01 5.16557753e-01
-1.19468009e+00 5.56013845e-02 -5.60757339e-01 -5.39975405e-01
1.74563617e-01 8.16749573e-01 -2.09922820e-01 5.02834171e-02
7.83643126e-01 4.54114348e-01 -1.14604428e-01 9.20251608e-01
2.47015715e-01 3.30902189e-01 -6.47751868e-01 -2.36107409e-01
6.49151266e-01 -5.65801442e-01 3.82625937e-01 1.08700752e+00
4.75276262e-01 2.96714008e-01 -2.40951031e-01 1.14602256e+00
-5.71523011e-02 -6.76770359e-02 -2.62276292e-01 7.04287812e-02
4.24749292e-02 1.06810880e+00 -8.56489122e-01 -5.03411233e-01
-1.81362838e-01 1.21081746e+00 6.71793044e-01 7.44229138e-01
-1.00116372e+00 -7.16799617e-01 6.54235721e-01 7.58508816e-02
2.32232332e-01 -3.55694175e-01 -4.06865865e-01 -1.19415867e+00
1.96808070e-01 -3.86175781e-01 6.60908222e-02 -3.97194177e-01
-8.25202405e-01 8.71102571e-01 -2.01313764e-01 -8.25391650e-01
-2.07043752e-01 -9.82069254e-01 -4.53738481e-01 1.17932355e+00
-1.33853900e+00 -2.21223161e-01 -6.82604164e-02 3.70626748e-01
2.05173686e-01 3.77984524e-01 7.53008604e-01 6.88701749e-01
-7.20059335e-01 5.52130640e-01 3.94897640e-01 3.25166613e-01
9.95026529e-02 -1.01752532e+00 4.71452177e-01 6.71671927e-01
-4.70032841e-01 6.43122077e-01 7.14416802e-01 -4.18869525e-01
-1.05538428e+00 -1.05667877e+00 1.13488710e+00 2.67013401e-01
7.08788693e-01 -1.97027802e-01 -1.43082690e+00 6.97586536e-01
-1.70072868e-01 5.93623340e-01 1.06755532e-01 -2.44338155e-01
-2.09033266e-02 -4.58201766e-02 -1.14604056e+00 7.62678385e-01
1.04410803e+00 -5.22833347e-01 -5.91075569e-02 3.27697515e-01
2.96094656e-01 -9.39933181e-01 -9.80602562e-01 4.31350350e-01
4.75108206e-01 -1.21732497e+00 7.88725972e-01 -4.93721604e-01
2.98462480e-01 -9.59088504e-02 4.14987862e-01 -1.25562906e+00
-5.36200255e-02 -7.71127105e-01 -3.20281655e-01 8.95163476e-01
2.18594879e-01 -1.22094834e+00 5.98562360e-01 1.02704442e+00
-1.21954910e-01 -1.40324080e+00 -1.15331590e+00 -9.46861684e-01
6.83974743e-01 -3.83720309e-01 1.82718471e-01 9.73950267e-01
-1.90595508e-01 -1.16410442e-01 -2.58580774e-01 -1.49742007e-01
2.23590463e-01 -3.07881594e-01 2.19500646e-01 -1.19176352e+00
-2.54507869e-01 -7.14949012e-01 -5.09901345e-02 -1.28339946e+00
1.97063774e-01 -7.76634634e-01 7.72416666e-02 -1.30306995e+00
-3.74792606e-01 -7.79292762e-01 -1.19390063e-01 4.38076138e-01
1.48945570e-01 2.79523164e-01 -2.89089698e-02 2.49820188e-01
1.47656903e-01 4.24027294e-01 1.11578476e+00 2.52811551e-01
-4.73188728e-01 -1.99747846e-01 -1.71913385e-01 9.95797396e-01
7.50175953e-01 -6.08099282e-01 -3.43811572e-01 -6.19731486e-01
4.00837988e-01 2.48979315e-01 5.79651415e-01 -1.18275726e+00
4.02891845e-01 1.38586134e-01 3.42562854e-01 -3.21386337e-01
4.68870044e-01 -4.76989955e-01 2.31580660e-01 8.84035408e-01
-4.51807886e-01 2.17193931e-01 4.61457968e-01 -5.54867974e-03
-1.73290297e-01 -5.62888741e-01 1.03528380e+00 -2.30501518e-01
-1.88101500e-01 3.78139287e-01 -5.31103194e-01 2.28158697e-01
5.62574029e-01 -1.51603371e-01 5.03446050e-02 -2.08452389e-01
-8.00688267e-01 -6.35998249e-02 2.72983879e-01 -1.94043174e-01
2.14588791e-01 -1.33785117e+00 -2.56681979e-01 2.58186698e-01
-9.39981759e-01 2.76992559e-01 3.89580339e-01 1.15977550e+00
-1.03988004e+00 2.89699376e-01 1.89507063e-02 -3.05473387e-01
-8.03185880e-01 1.71159148e-01 9.92049396e-01 -4.25036639e-01
-8.59904051e-01 1.22348940e+00 -3.80695472e-03 -3.72177720e-01
4.38371032e-01 -5.90240002e-01 1.30637065e-01 -2.06621792e-02
3.30732822e-01 8.66002500e-01 1.27645090e-01 -3.41776490e-01
-2.91420907e-01 7.06866682e-01 1.50836021e-01 -3.78576547e-01
1.24717069e+00 2.40643352e-01 -3.37690443e-01 4.78298247e-01
1.64190102e+00 -4.37395275e-01 -1.45755625e+00 -4.32549790e-02
-1.89803004e-01 2.22662985e-01 2.85813659e-01 -3.42416286e-01
-1.42383635e+00 1.19436347e+00 6.68707311e-01 2.84585327e-01
1.06033885e+00 -7.22340047e-01 1.08855653e+00 6.37026668e-01
-1.26840159e-01 -1.28014970e+00 -4.98927414e-01 1.02747560e+00
8.41616869e-01 -8.78277421e-01 -6.53230101e-02 -2.99940228e-01
-3.34273517e-01 1.37269652e+00 6.13696396e-01 -6.04205787e-01
9.09610808e-01 5.69608867e-01 6.25540540e-02 2.19730753e-02
-3.99560153e-01 2.84039170e-01 1.21834666e-01 -1.52647585e-01
7.46725500e-01 -3.80215466e-01 -4.79383916e-01 2.78006524e-01
-1.96040809e-01 4.35960472e-01 2.56113350e-01 7.99679339e-01
-1.52729914e-01 -7.58074164e-01 -9.92558599e-02 3.67495149e-01
-6.40449226e-01 -2.85894811e-01 4.37769666e-02 8.60149324e-01
-5.64673655e-02 3.27130944e-01 1.98656455e-01 1.42816186e-01
1.06285907e-01 3.40781271e-01 4.45299327e-01 1.27802454e-02
-8.04261088e-01 -2.22094417e-01 -8.01690966e-02 -5.16416073e-01
-1.28153890e-01 -2.67225802e-01 -1.84577692e+00 -6.28729880e-01
-2.19775990e-01 2.19063342e-01 6.51912153e-01 1.01742542e+00
3.04018945e-01 6.15827262e-01 2.68550545e-01 -1.06251013e+00
-9.25076604e-01 -8.67556930e-01 -5.22026181e-01 1.41380847e-01
5.21784782e-01 -4.09929693e-01 -8.13240051e-01 -2.04984829e-01] | [6.865442276000977, 3.5000619888305664] |
b11a9732-96d3-4295-a72b-51cb609a6e4d | deep-deformable-3d-caricatures-with-learned | 2207.14593 | null | https://arxiv.org/abs/2207.14593v1 | https://arxiv.org/pdf/2207.14593v1.pdf | Deep Deformable 3D Caricatures with Learned Shape Control | A 3D caricature is an exaggerated 3D depiction of a human face. The goal of this paper is to model the variations of 3D caricatures in a compact parameter space so that we can provide a useful data-driven toolkit for handling 3D caricature deformations. To achieve the goal, we propose an MLP-based framework for building a deformable surface model, which takes a latent code and produces a 3D surface. In the framework, a SIREN MLP models a function that takes a 3D position on a fixed template surface and returns a 3D displacement vector for the input position. We create variations of 3D surfaces by learning a hypernetwork that takes a latent code and produces the parameters of the MLP. Once learned, our deformable model provides a nice editing space for 3D caricatures, supporting label-based semantic editing and point-handle-based deformation, both of which produce highly exaggerated and natural 3D caricature shapes. We also demonstrate other applications of our deformable model, such as automatic 3D caricature creation. | ['Seungyong Lee', 'Xin Tong', 'Jiaolong Yang', 'Soongjin Kim', 'Wonjong Jang', 'Yucheol Jung'] | 2022-07-29 | null | null | null | null | ['caricature'] | ['computer-vision'] | [ 2.43580133e-01 8.22051108e-01 1.97175965e-01 -6.83364689e-01
-3.91986638e-01 -5.45102596e-01 6.32581830e-01 -5.27111709e-01
2.41307765e-01 1.49704084e-01 7.21221194e-02 -6.90598711e-02
9.62510183e-02 -9.83541369e-01 -1.19728780e+00 -5.05210459e-01
3.98754254e-02 9.73372400e-01 -1.46405604e-02 -1.46999344e-01
5.87925352e-02 1.52384365e+00 -1.55314279e+00 1.85780108e-01
5.66536546e-01 6.65071070e-01 -1.11461490e-01 6.76232994e-01
-3.99932534e-01 -2.50830680e-01 -5.45825303e-01 -4.41036165e-01
4.07504439e-01 1.03014447e-01 -5.79990745e-01 3.70249391e-01
8.52862418e-01 -7.87042230e-02 1.51103616e-01 7.14070976e-01
3.52072567e-01 -2.02471744e-02 1.09789217e+00 -1.34404778e+00
-9.45402265e-01 4.03926075e-01 -3.75578761e-01 -9.37698901e-01
3.18328261e-01 -5.94562702e-02 5.23167849e-01 -1.09383881e+00
9.97789860e-01 1.95377636e+00 9.97780383e-01 1.03784740e+00
-1.37468803e+00 -2.09158346e-01 2.23859530e-02 -6.92805827e-01
-1.10302734e+00 -2.28922889e-01 1.21365166e+00 -6.09878540e-01
6.77079737e-01 5.54739416e-01 8.44236612e-01 9.72537816e-01
3.86302441e-01 8.03028584e-01 7.73026049e-01 -5.12969673e-01
1.81459382e-01 -8.26859381e-03 -2.86541998e-01 7.90553749e-01
-4.27941263e-01 -1.61567077e-04 1.77085847e-01 -3.81381959e-01
1.67879796e+00 3.23063940e-01 1.61570236e-01 -7.93206453e-01
-1.05369353e+00 7.26728797e-01 4.81668115e-01 1.78687468e-01
-1.75903991e-01 6.11743450e-01 -5.09747900e-02 1.24344662e-01
7.33791113e-01 4.73959178e-01 -3.56481850e-01 2.39289016e-01
-6.25953555e-01 6.96813583e-01 7.93618560e-01 1.01894701e+00
8.02045345e-01 1.82227269e-01 -2.56638825e-01 9.85683680e-01
4.85677451e-01 7.15029478e-01 -2.10028235e-02 -1.38346422e+00
-3.09094824e-02 7.98683405e-01 -6.27640262e-02 -9.04306054e-01
-3.85012031e-01 1.38775706e-01 -7.69758165e-01 1.14399421e+00
1.41194746e-01 -2.28815395e-02 -1.21336412e+00 1.60242355e+00
6.39979362e-01 1.22194603e-01 -1.58581004e-01 6.79362595e-01
8.20862353e-01 7.11433172e-01 -6.86316863e-02 4.42570031e-01
9.70085382e-01 -5.30667901e-01 -4.49911773e-01 2.14790657e-01
5.87329805e-01 -5.99432826e-01 1.28437102e+00 -1.57161385e-01
-1.65875554e+00 -5.36571205e-01 -8.27697456e-01 -3.07220727e-01
-4.17116016e-01 -1.11969784e-01 2.92077541e-01 3.42979044e-01
-1.47215986e+00 8.18082213e-01 -8.41222644e-01 -1.52669966e-01
5.54693222e-01 3.60797822e-01 -4.51203972e-01 4.74062681e-01
-7.33008564e-01 1.04976308e+00 7.01578101e-03 -3.25945579e-02
-4.96390700e-01 -9.80707407e-01 -1.04818749e+00 -1.34827629e-01
-1.81579608e-02 -8.57312918e-01 1.02436662e+00 -8.47576380e-01
-1.84650743e+00 1.61466765e+00 1.15444355e-01 2.28552544e-03
7.48578250e-01 1.00476012e-01 -1.98859081e-01 -3.26945484e-01
-3.23447466e-01 1.16887438e+00 1.25508702e+00 -1.56331861e+00
6.30172789e-02 -4.54084754e-01 1.61158204e-01 1.37376979e-01
3.11189115e-01 -2.55824506e-01 -5.84600151e-01 -1.08482134e+00
6.83875382e-02 -1.10863435e+00 -1.36062577e-01 9.46157694e-01
-6.28667176e-01 -4.59080011e-01 1.21996939e+00 -5.54682374e-01
4.98512030e-01 -2.15841913e+00 6.74205720e-01 5.30639529e-01
2.68752664e-01 1.59538716e-01 -2.94471234e-01 3.75277549e-02
-2.84870148e-01 5.08626282e-01 -7.80292749e-01 -6.71148419e-01
2.75167882e-01 3.04868519e-01 -1.83285609e-01 1.33984447e-01
4.98443604e-01 1.32364523e+00 -6.35388613e-01 -2.96291769e-01
2.84363121e-01 1.07734704e+00 -8.53762388e-01 3.33063364e-01
-7.69234419e-01 3.70094895e-01 -4.42363620e-01 6.55740619e-01
8.07112873e-01 1.60491392e-01 -3.45496029e-01 -2.58887053e-01
-1.57346055e-02 -4.40105945e-01 -1.21073604e+00 1.94123948e+00
-6.34756446e-01 4.13426429e-01 1.73770741e-01 -3.48847896e-01
1.41894424e+00 1.73749790e-01 3.29802930e-01 -7.17762634e-02
9.61086825e-02 2.06697434e-01 -7.17586935e-01 -3.37861270e-01
2.54904747e-01 -3.35244358e-01 -2.48355508e-01 5.69194973e-01
-2.89310068e-01 -1.04530835e+00 -7.66399682e-01 -2.41476834e-01
6.04379237e-01 7.35965252e-01 -2.08096027e-01 -3.00447851e-01
5.33097804e-01 -4.33921546e-01 3.71709354e-02 1.59650698e-01
3.25404763e-01 8.11203122e-01 5.24437904e-01 -9.08751428e-01
-1.48412156e+00 -1.23036790e+00 -3.56045663e-01 7.67675102e-01
-2.10103154e-01 3.83714378e-01 -1.24769211e+00 -6.48256719e-01
6.21037900e-01 7.56537080e-01 -9.79210258e-01 -7.68448263e-02
-9.67670143e-01 -6.26071915e-02 4.87831622e-01 4.38146889e-01
1.43616721e-01 -1.33422661e+00 -3.32368821e-01 2.21489947e-02
6.00795865e-01 -5.10179937e-01 -9.79052424e-01 -2.08707511e-01
-9.37230229e-01 -7.80196130e-01 -1.03888583e+00 -1.26200998e+00
1.00013113e+00 -3.80853862e-01 9.60420191e-01 1.69472054e-01
-3.82348210e-01 4.89018589e-01 2.34943047e-01 -4.87203002e-01
-1.19003475e+00 -1.09623469e-01 5.11429124e-02 1.03061952e-01
-1.50591776e-01 -8.35584104e-01 -1.68747857e-01 3.00202787e-01
-1.06612670e+00 7.42869452e-02 2.22829029e-01 2.60627866e-01
1.02799070e+00 -5.86349487e-01 4.04243678e-01 -1.07618201e+00
6.52050734e-01 -2.21805312e-02 -6.38747692e-01 1.96878701e-01
2.57713031e-02 5.67899644e-01 3.74749839e-01 -6.87981069e-01
-1.02363360e+00 2.33862087e-01 -4.03566509e-01 -8.10954332e-01
-3.01772863e-01 -1.64015651e-01 -4.05441761e-01 -2.39787400e-01
8.31613839e-01 -2.49745905e-01 5.56828678e-01 -8.02390754e-01
9.94914234e-01 5.58962166e-01 7.23506272e-01 -6.40995860e-01
1.22261953e+00 6.12215877e-01 1.76536053e-01 -6.76633060e-01
-2.96078295e-01 3.95157695e-01 -1.23346722e+00 -1.81223869e-01
1.01945615e+00 -4.63734657e-01 -7.82872796e-01 6.51671946e-01
-1.46443594e+00 -7.62987375e-01 -8.09211195e-01 -3.23814452e-01
-1.04048312e+00 1.00007346e-02 -3.41397047e-01 -3.37495238e-01
-4.01433259e-01 -1.13915753e+00 1.55466652e+00 -6.76512048e-02
-5.25790095e-01 -1.37112272e+00 2.77061194e-01 -1.42793387e-01
4.49584663e-01 1.12899542e+00 1.26234519e+00 -9.66720581e-02
-5.48829317e-01 -2.87857205e-01 2.15463370e-01 3.70716125e-01
1.71106026e-01 4.75442559e-01 -9.53002453e-01 9.72893909e-02
-4.37314421e-01 9.47035030e-02 2.62359887e-01 4.13217813e-01
1.59230959e+00 -4.20560926e-01 -4.56414461e-01 1.01133370e+00
9.44324732e-01 2.62701035e-01 5.68065405e-01 -1.48039475e-01
9.92881596e-01 7.34854817e-01 -1.37563393e-01 6.82647005e-02
1.82096437e-01 9.21866536e-01 6.30286038e-01 -3.22706491e-01
-2.94056833e-01 -4.59447861e-01 8.98711849e-03 4.27372396e-01
-2.48666629e-01 1.64975584e-01 -5.85118711e-01 2.09518537e-01
-1.37803054e+00 -6.34218633e-01 1.19391277e-01 2.03682113e+00
1.08377790e+00 -2.37356778e-02 1.04716033e-01 -4.46163453e-02
8.39884520e-01 1.35424808e-01 -8.62072706e-01 -1.01680434e+00
3.25695910e-02 2.92476743e-01 1.82677373e-01 9.34391499e-01
-8.41311395e-01 1.05550814e+00 6.38640690e+00 4.88486230e-01
-1.30145788e+00 -3.01852554e-01 4.08069253e-01 3.03278528e-02
-9.26153541e-01 -5.20497739e-01 -5.46697795e-01 2.05565751e-01
3.55342031e-01 3.32474448e-02 6.01925015e-01 9.49531674e-01
2.08125934e-01 7.78623700e-01 -1.36583173e+00 9.82707500e-01
-1.74563713e-02 -1.76099074e+00 5.55744708e-01 7.67663270e-02
8.30968797e-01 -4.51548249e-01 1.83553725e-01 7.04196393e-02
3.99278790e-01 -1.18757510e+00 8.71043682e-01 1.04750633e+00
1.44839382e+00 -6.63492799e-01 -5.87824993e-02 2.90756851e-01
-6.81655049e-01 5.15322268e-01 -2.75053948e-01 4.93111730e-01
1.82740182e-01 3.67249846e-01 -8.66567969e-01 -1.76276207e-01
4.87825483e-01 5.16224980e-01 -1.44733731e-02 7.96281874e-01
-1.15850471e-01 -1.33326814e-01 -2.62079924e-01 1.00319482e-01
2.34170817e-02 -2.41303042e-01 8.62961590e-01 1.13981676e+00
3.77812058e-01 -5.99460304e-02 -1.52819619e-01 1.44174361e+00
-5.44407487e-01 -8.52823108e-02 -9.18796122e-01 2.94134051e-01
5.82220018e-01 1.24347878e+00 -2.89884955e-01 -7.98701048e-02
2.51775801e-01 1.04811156e+00 3.91514182e-01 1.82900086e-01
-7.13551342e-01 -4.49438632e-01 8.75751972e-01 4.42447275e-01
1.54417641e-02 -2.15039268e-01 -2.92917013e-01 -7.60193288e-01
-1.08586840e-01 -4.36242282e-01 -2.43946239e-01 -1.17750132e+00
-1.13739145e+00 6.59818709e-01 9.57305133e-02 -8.72965038e-01
-2.41479576e-01 -5.64218760e-01 -7.87738204e-01 9.98851538e-01
-1.05527067e+00 -1.47737646e+00 -2.54580319e-01 4.20996904e-01
3.89598697e-01 -2.15369668e-02 1.01274288e+00 -5.46673387e-02
-1.17207654e-01 5.91685355e-01 -3.07797253e-01 -1.08228177e-01
5.07131279e-01 -1.35208476e+00 1.18502009e+00 2.84162164e-02
-4.57352474e-02 3.36718202e-01 3.89494032e-01 -6.43386662e-01
-1.37879860e+00 -1.37963128e+00 6.51058972e-01 -1.08735108e+00
8.81556496e-02 -4.85757858e-01 -1.21707952e+00 1.06464648e+00
-3.40834886e-01 1.45558134e-01 1.14879414e-01 -4.05285686e-01
-2.22862661e-01 1.05883464e-01 -1.66404057e+00 7.73104966e-01
1.54065740e+00 -5.05056322e-01 -7.59691715e-01 3.19975108e-01
9.89556551e-01 -1.00854731e+00 -1.23980594e+00 1.88231915e-01
6.12028122e-01 -4.75693226e-01 1.25635672e+00 -6.13107026e-01
1.93087846e-01 -1.94080740e-01 2.08396330e-01 -1.57686698e+00
-4.20208722e-01 -9.69166636e-01 -2.09844604e-01 8.27713370e-01
3.20147872e-01 -4.40873355e-01 8.84303868e-01 9.54879105e-01
-5.38033843e-01 -8.87241423e-01 -9.37220275e-01 -5.94884515e-01
5.20824552e-01 -4.25859183e-01 1.20691204e+00 9.31357801e-01
-5.84370315e-01 -3.87615532e-01 -2.13073254e-01 8.34718943e-02
5.64459860e-01 -1.85461994e-02 8.86391759e-01 -1.51062500e+00
1.65305585e-01 -5.76439202e-01 -3.33758950e-01 -1.04923296e+00
5.02922714e-01 -1.40051794e+00 -1.46600351e-01 -1.35732830e+00
-3.56248587e-01 -9.31077778e-01 5.78527093e-01 7.56826758e-01
2.54400790e-01 2.46958092e-01 2.40556046e-01 1.95127502e-01
2.24061072e-01 4.76081818e-01 1.93557358e+00 -1.16095863e-01
-7.51216888e-01 8.77933949e-02 -5.42517602e-01 1.03033185e+00
4.34964567e-01 -2.20200613e-01 -8.45784619e-02 -7.55711555e-01
-5.46601117e-02 5.24452031e-02 5.31331241e-01 -5.01149118e-01
-1.97429925e-01 -2.18950465e-01 3.74731898e-01 -5.51737070e-01
5.55203438e-01 -9.11768079e-01 6.02590919e-01 2.35152259e-01
-4.89769846e-01 -1.44365221e-01 1.88288093e-01 1.37457386e-01
2.97360092e-01 2.88462657e-02 9.19543624e-01 -1.93634346e-01
-1.87720686e-01 8.66496205e-01 1.61850318e-01 -1.34323910e-01
1.08491647e+00 -6.59073532e-01 1.87147349e-01 -1.56011090e-01
-1.21804440e+00 1.15593327e-02 9.60827529e-01 7.95481086e-01
7.11226583e-01 -1.82169151e+00 -7.46194184e-01 8.19449484e-01
-4.06093784e-02 5.36098599e-01 5.35512641e-02 -1.14051439e-01
-8.34511697e-01 1.12197297e-02 -3.40211809e-01 -6.90957189e-01
-1.08050740e+00 3.96909088e-01 9.99984741e-01 4.48473811e-01
-9.09027040e-01 8.20001841e-01 1.36947602e-01 -1.22082627e+00
3.32035631e-01 -8.67645621e-01 -1.21528670e-01 -3.21215868e-01
2.64756471e-01 2.25919083e-01 -8.59829709e-02 -6.69218481e-01
-9.72096771e-02 1.24764895e+00 4.37988043e-01 1.74092557e-02
1.54538202e+00 3.43978137e-01 -1.02859154e-01 3.53062689e-01
1.50867963e+00 1.12806261e-01 -1.81909454e+00 -9.87151116e-02
-3.70956272e-01 -3.38392258e-01 -2.55924284e-01 -6.97413027e-01
-1.08204269e+00 7.51996934e-01 2.79141933e-01 1.89182296e-01
4.60855186e-01 3.24171066e-01 6.97689712e-01 3.11593175e-01
2.29730338e-01 -7.44511485e-01 -9.05026868e-03 3.68007988e-01
1.83701980e+00 -6.40769124e-01 -3.89971644e-01 -4.64089990e-01
-4.54552621e-01 1.30352068e+00 1.67997628e-01 -4.43180114e-01
8.87740135e-01 4.25092548e-01 1.42581731e-01 -3.88529092e-01
-1.98107615e-01 5.04751980e-01 7.01854765e-01 9.79797184e-01
1.12352520e-01 3.04306716e-01 2.07911626e-01 1.76723942e-01
-4.80890661e-01 -9.93813425e-02 2.73195118e-01 6.88099980e-01
-3.75522703e-01 -1.26949584e+00 -5.93948543e-01 9.43315215e-03
9.17225331e-02 4.79484081e-01 -4.63025272e-01 8.57284606e-01
2.06064820e-01 -5.22803217e-02 4.65961367e-01 -1.10485181e-01
9.17091131e-01 3.55212092e-01 7.05331922e-01 -7.60996819e-01
-4.02585775e-01 -1.38862044e-01 -2.68926948e-01 -6.07881665e-01
-3.08059007e-01 -3.68635088e-01 -1.50214171e+00 -3.51818144e-01
2.50922918e-01 -3.91212642e-01 9.92483258e-01 5.35025954e-01
6.80533946e-01 2.12788463e-01 7.50784159e-01 -1.60037494e+00
-4.90963489e-01 -5.88114977e-01 -6.29300773e-01 7.08919227e-01
3.08501363e-01 -6.80857182e-01 -3.55689138e-01 3.23391616e-01] | [8.909466743469238, -3.629671573638916] |
aab7f104-9f2b-49ae-a06c-a161034e83a3 | srl4e-semantic-role-labeling-for-emotions-a | null | null | https://aclanthology.org/2022.acl-long.314 | https://aclanthology.org/2022.acl-long.314.pdf | SRL4E – Semantic Role Labeling for Emotions: A Unified Evaluation Framework | In the field of sentiment analysis, several studies have highlighted that a single sentence may express multiple, sometimes contrasting, sentiments and emotions, each with its own experiencer, target and/or cause. To this end, over the past few years researchers have started to collect and annotate data manually, in order to investigate the capabilities of automatic systems not only to distinguish between emotions, but also to capture their semantic constituents. However, currently available gold datasets are heterogeneous in size, domain, format, splits, emotion categories and role labels, making comparisons across different works difficult and hampering progress in the area. In this paper, we tackle this issue and present a unified evaluation framework focused on Semantic Role Labeling for Emotions (SRL4E), in which we unify several datasets tagged with emotions and semantic roles by using a common labeling scheme. We use SRL4E as a benchmark to evaluate how modern pretrained language models perform and analyze where we currently stand in this task, hoping to provide the tools to facilitate studies in this complex area. | ['Roberto Navigli', 'Simone Conia', 'Cesare Campagnano'] | null | null | null | null | acl-2022-5 | ['semantic-role-labeling'] | ['natural-language-processing'] | [ 2.30887130e-01 -1.78533167e-01 -1.03134654e-01 -7.25957096e-01
-2.45661303e-01 -1.02595317e+00 6.86872482e-01 7.69650042e-01
-5.25996864e-01 8.50521624e-01 4.71369863e-01 3.42315435e-01
1.26216441e-01 -4.43246245e-01 -7.15003386e-02 -6.79193735e-01
2.52449244e-01 3.69089454e-01 -1.06119625e-01 -4.76780862e-01
1.49180755e-01 2.25101754e-01 -1.99827731e+00 6.17588162e-01
4.86438721e-01 1.17743337e+00 -1.20635398e-01 2.15551913e-01
-4.73142326e-01 1.20101535e+00 -7.60748863e-01 -7.69188404e-01
-3.87831062e-01 -4.83493775e-01 -1.13313687e+00 1.08529508e-01
-2.94732340e-02 5.84035695e-01 5.83012521e-01 1.05498600e+00
4.14502025e-01 -1.98400710e-02 6.36417806e-01 -1.45347726e+00
-3.89251590e-01 6.96800590e-01 -1.38580576e-01 -2.32629687e-01
6.22124791e-01 -4.41671431e-01 1.37673032e+00 -5.06123781e-01
9.56212759e-01 1.09914434e+00 5.70938170e-01 7.51964390e-01
-1.02049601e+00 -5.30183494e-01 3.31845015e-01 7.89040998e-02
-9.29897606e-01 -2.85394609e-01 8.68514419e-01 -4.61391270e-01
6.88427925e-01 2.96112031e-01 5.70991635e-01 1.34399343e+00
-3.35982144e-01 8.23165238e-01 1.58438408e+00 -5.30726016e-01
2.71544009e-01 6.20255947e-01 4.32855725e-01 2.82103896e-01
-3.94566469e-02 -6.97805524e-01 -6.13221943e-01 -1.22217193e-01
4.56156246e-02 -2.93024302e-01 -2.38298714e-01 -3.14842641e-01
-1.05159748e+00 7.35979795e-01 1.19589180e-01 8.68610501e-01
-3.15366864e-01 -2.47735217e-01 7.99902797e-01 4.72302735e-01
7.85320878e-01 8.71576190e-01 -7.46795356e-01 -4.63773578e-01
-5.21262765e-01 1.52722895e-01 1.02266812e+00 3.13661098e-01
5.48750103e-01 -4.34524596e-01 6.06140867e-02 1.12148333e+00
1.27301544e-01 -4.09830362e-02 5.84703207e-01 -6.75307453e-01
7.75697976e-02 9.63048279e-01 2.83003896e-01 -1.07077587e+00
-7.04913795e-01 -1.78022400e-01 -7.20363438e-01 2.14878730e-02
3.49153012e-01 -1.85797110e-01 -3.45440209e-01 1.90837073e+00
3.30164105e-01 -1.71508327e-01 4.27313834e-01 8.49375129e-01
9.75317180e-01 4.33448285e-01 4.79960471e-01 -1.85671508e-01
1.67393684e+00 -7.85186410e-01 -8.32340002e-01 -2.76387423e-01
9.95800555e-01 -9.42445397e-01 1.35356641e+00 5.94719172e-01
-7.36268520e-01 -3.12132210e-01 -8.22254360e-01 6.17459863e-02
-9.39212322e-01 3.24701428e-01 7.41358221e-01 5.67027986e-01
-1.02874112e+00 3.99047077e-01 -4.27845031e-01 -4.90397692e-01
1.62130147e-01 6.11985363e-02 -5.87389231e-01 1.97794616e-01
-1.57710910e+00 1.07223582e+00 3.84037137e-01 -7.72055164e-02
-2.75973797e-01 -4.93179560e-01 -8.20323884e-01 -1.35275960e-01
3.23971003e-01 -3.21145386e-01 1.08001578e+00 -1.53189743e+00
-1.27201819e+00 1.50761795e+00 -1.87618181e-01 -2.26288810e-01
1.33333042e-01 -1.35637313e-01 -7.39772737e-01 -4.59942520e-02
3.25239450e-01 5.49586475e-01 4.27112788e-01 -1.40492857e+00
-8.07847500e-01 -4.72474903e-01 5.03691137e-01 2.12820977e-01
-5.38600683e-01 6.46711707e-01 -1.67361021e-01 -5.99227607e-01
-3.16271216e-01 -8.07490528e-01 -1.16320655e-01 -5.33612132e-01
-1.05374292e-01 -5.88532329e-01 5.61825454e-01 -1.12110443e-01
1.19379950e+00 -2.31607413e+00 2.76657432e-01 -9.44169685e-02
1.39075620e-02 3.19880322e-02 7.76968971e-02 4.64793026e-01
-2.97147572e-01 3.62500191e-01 -3.50577056e-01 -7.54408240e-01
2.17557535e-01 3.25850606e-01 -3.55295777e-01 2.00812444e-01
2.71250099e-01 5.57399869e-01 -1.02190232e+00 -6.08254850e-01
8.17745477e-02 5.65345883e-01 -1.38364822e-01 2.87148774e-01
-2.28782490e-01 5.86585939e-01 -5.08329749e-01 7.09711373e-01
2.45757803e-01 -1.61927685e-01 4.12892997e-01 -1.57555342e-01
-1.73613265e-01 3.76295239e-01 -1.12308609e+00 1.48934340e+00
-7.66215086e-01 4.72655952e-01 2.96660841e-01 -1.13335371e+00
9.82650518e-01 5.12981474e-01 7.06590652e-01 -5.21472871e-01
5.20282865e-01 3.55212003e-01 -2.56621093e-01 -4.65320826e-01
7.38952696e-01 -5.61459243e-01 -6.57380521e-01 5.03764927e-01
1.24089912e-01 -1.79943085e-01 4.53184336e-01 -1.47274926e-01
7.40015030e-01 -8.04418400e-02 4.00843889e-01 -1.91867366e-01
8.82975876e-01 5.19295000e-02 2.58992165e-01 1.32213846e-01
-2.64399618e-01 3.82268786e-01 9.77437556e-01 -3.96270186e-01
-5.06174803e-01 -4.63352382e-01 -2.33195633e-01 1.42811787e+00
2.76041150e-01 -4.38074410e-01 -6.52247787e-01 -8.31656218e-01
-3.82656753e-01 6.71337903e-01 -7.34562099e-01 8.92079547e-02
-1.14466675e-01 -9.47425902e-01 6.02216899e-01 2.89885044e-01
2.45770961e-01 -1.55173659e+00 -8.04893434e-01 2.27315381e-01
-4.31853384e-01 -1.35035765e+00 4.06741679e-01 5.12649536e-01
-3.65557641e-01 -1.21377826e+00 -3.28031838e-01 -7.20164359e-01
4.71482933e-01 -1.48243040e-01 1.82456458e+00 1.86728612e-02
1.09182537e-01 3.59481096e-01 -1.07038605e+00 -6.77100837e-01
-2.86949635e-01 1.25355229e-01 -2.43846923e-01 2.71070212e-01
5.47171533e-01 -2.54763484e-01 -1.80531397e-01 2.97907293e-01
-1.29693317e+00 -2.43354917e-01 2.14831874e-01 4.50771183e-01
4.19630319e-01 2.32986346e-01 5.24542451e-01 -1.08285522e+00
9.33131278e-01 -6.16362453e-01 -1.44276410e-01 3.64908725e-01
-3.67511779e-01 -1.53080039e-02 7.43880630e-01 -8.51316079e-02
-1.05759764e+00 -1.20061152e-01 -4.42640066e-01 1.68116033e-01
-6.12475932e-01 6.66989625e-01 -3.59288156e-01 2.09508151e-01
4.26396519e-01 2.06052270e-02 -2.54137427e-01 -3.35380197e-01
2.89409697e-01 6.81720257e-01 2.25577191e-01 -7.75300801e-01
1.65057749e-01 5.47163486e-01 -1.78654149e-01 -6.26531601e-01
-1.43115509e+00 -5.89056969e-01 -6.52054250e-01 -2.39107460e-01
1.00084949e+00 -8.87983859e-01 -3.82189542e-01 4.83682901e-01
-1.09836733e+00 -2.52866536e-01 -2.18756825e-01 2.17329741e-01
-3.55408996e-01 1.43455610e-01 -5.81370473e-01 -7.42507279e-01
-1.08341955e-01 -1.00220358e+00 1.13328171e+00 9.23269615e-03
-7.22916305e-01 -1.39090598e+00 9.20960978e-02 4.52936471e-01
2.44483739e-01 5.27988970e-01 8.64424109e-01 -8.39518189e-01
5.03778458e-01 -1.52643085e-01 -4.36617322e-02 6.33481264e-01
1.51098669e-01 -8.39743540e-02 -9.69630063e-01 1.07493803e-01
1.91172257e-01 -7.55853593e-01 6.92172825e-01 9.71651636e-03
1.08809209e+00 1.55891865e-01 -2.63463017e-02 5.61455265e-03
1.28660643e+00 1.18876033e-01 5.25953710e-01 5.42575419e-01
4.74559009e-01 1.31134748e+00 1.05826581e+00 3.67215544e-01
5.27854562e-01 6.38493538e-01 5.13876557e-01 -2.99811602e-01
3.04439694e-01 2.06585810e-01 3.74899477e-01 6.39052331e-01
-1.44390166e-01 -3.89220893e-01 -7.52664983e-01 5.36458552e-01
-1.82337463e+00 -6.71513736e-01 -3.04351419e-01 1.73172843e+00
1.03285110e+00 1.50041565e-01 1.41754895e-01 2.98516303e-01
6.62949979e-01 5.32521248e-01 -8.36264417e-02 -6.93308055e-01
-4.37675148e-01 2.00352460e-01 -1.32319584e-01 2.71343291e-01
-1.25602722e+00 1.13172019e+00 5.69859743e+00 6.28065705e-01
-1.39547825e+00 1.27312228e-01 7.28681147e-01 1.97291225e-01
-2.50328630e-01 -2.48591766e-01 -6.48190498e-01 4.13415283e-01
7.85063028e-01 1.92422107e-01 1.17298357e-01 9.76789653e-01
6.01180531e-02 3.53843309e-02 -9.36017215e-01 9.53896344e-01
3.97996068e-01 -7.50447929e-01 -2.09100664e-01 -4.30254221e-01
4.85303372e-01 -2.11887509e-01 -1.48316622e-01 3.19187194e-01
1.97178617e-01 -1.15897751e+00 8.00234795e-01 1.70004219e-01
4.47187483e-01 -6.95661962e-01 1.14726186e+00 -1.09223407e-02
-1.07043171e+00 1.58727661e-01 -4.99042347e-02 -4.82124031e-01
8.45428705e-02 5.26016176e-01 -4.17095363e-01 4.92615461e-01
8.85043681e-01 9.26448226e-01 -4.33486015e-01 3.62329274e-01
-5.14000595e-01 5.50847471e-01 6.77380105e-03 -2.72225231e-01
3.01353425e-01 -1.71832547e-01 1.76684588e-01 1.36114573e+00
2.55657077e-01 -1.35634109e-01 2.21332341e-01 4.68540370e-01
-1.94985151e-01 6.81170881e-01 -2.90435612e-01 -2.48076469e-01
3.27597708e-01 1.64563298e+00 -1.17320228e+00 -3.69318962e-01
-3.20007712e-01 9.39472497e-01 2.62177825e-01 -8.72468650e-02
-8.97985160e-01 -3.14841568e-01 9.71929491e-01 -5.99448495e-02
-2.58224785e-01 2.83317119e-02 -1.24475718e-01 -1.24607611e+00
-5.00553884e-02 -8.72846961e-01 5.36966622e-01 -9.35772777e-01
-1.48113060e+00 9.20639515e-01 -6.07570540e-03 -1.03364265e+00
-1.23180151e-01 -9.09397364e-01 -2.31863469e-01 4.93945301e-01
-1.77460504e+00 -1.18629301e+00 -2.67056584e-01 5.83713174e-01
4.36626315e-01 9.92329642e-02 1.07698190e+00 4.48401093e-01
-4.96043265e-01 2.16624841e-01 -3.62086117e-01 2.67235655e-02
1.06813061e+00 -1.44320822e+00 -2.57909715e-01 3.12284380e-01
2.14415863e-01 2.94299513e-01 9.27168965e-01 -1.45162612e-01
-7.68700123e-01 -8.64778817e-01 1.29227102e+00 -6.14102900e-01
1.08228076e+00 -4.69286203e-01 -9.17367935e-01 3.91281128e-01
2.93305963e-01 -9.64572579e-02 1.14095473e+00 5.83586097e-01
-4.77231175e-01 9.90988314e-02 -1.16253603e+00 3.32597315e-01
8.51604402e-01 -5.42983949e-01 -6.90750957e-01 1.41948268e-01
4.39762414e-01 -1.15128383e-01 -9.01484549e-01 4.38126683e-01
3.64789516e-01 -1.29225683e+00 6.04305565e-01 -6.04735315e-01
8.36423337e-01 -2.33387321e-01 -9.19468179e-02 -1.50772452e+00
4.67132144e-02 -8.93108323e-02 4.14451927e-01 1.70968354e+00
4.21893924e-01 -5.47338486e-01 5.55114448e-01 4.63884979e-01
-1.57196611e-01 -6.67177022e-01 -6.66807950e-01 -2.74604738e-01
6.94677746e-03 -8.01568925e-01 6.71593666e-01 1.31448412e+00
1.43387705e-01 7.42564738e-01 -3.51075083e-01 -2.09667206e-01
-8.61726925e-02 4.07853037e-01 5.10875642e-01 -1.62033057e+00
1.51166037e-01 -8.11885595e-01 -3.45986277e-01 -3.15824002e-01
7.38181114e-01 -7.64099538e-01 -6.64182305e-02 -1.49988329e+00
-3.48719023e-02 -5.47487497e-01 -5.95378935e-01 6.94248617e-01
-1.82567209e-01 6.45450056e-01 9.03414190e-02 7.75366649e-03
-1.07578754e+00 2.71656394e-01 9.08952534e-01 1.34100631e-01
-3.00288536e-02 -3.52182120e-01 -1.13543439e+00 1.23283541e+00
8.78103793e-01 -3.92247081e-01 -3.86583716e-01 -2.09468827e-01
9.03283477e-01 -4.84433621e-01 1.96113944e-01 -8.77292216e-01
-2.49972507e-01 -1.14108004e-01 1.12550735e-01 -1.09810926e-01
3.55079591e-01 -1.02975810e+00 -1.34834871e-01 5.97442687e-02
-4.94747102e-01 -1.92564800e-02 4.61198092e-02 -5.75703569e-03
-8.63935649e-01 -4.74434465e-01 5.28043449e-01 -1.80952370e-01
-1.13367105e+00 -1.68737993e-01 -4.06053603e-01 3.30168813e-01
1.11245513e+00 8.75223652e-02 -9.33771059e-02 -3.16098094e-01
-8.39665234e-01 2.48948410e-01 6.44631863e-01 7.02074349e-01
1.01453118e-01 -1.09840667e+00 -5.49247980e-01 -2.82964706e-01
5.54022968e-01 -2.61556298e-01 3.24430197e-01 7.05202639e-01
-1.48939565e-01 1.60280764e-01 -2.92270720e-01 -1.87173650e-01
-1.57757783e+00 4.35219139e-01 1.89611733e-01 -7.83477902e-01
1.62615269e-01 8.41110826e-01 -5.54470345e-02 -6.85394347e-01
1.81460343e-02 -2.80368835e-01 -1.00119174e+00 9.45137382e-01
3.36654007e-01 -5.21042719e-02 6.05655201e-02 -1.03618610e+00
-4.19661134e-01 6.27302349e-01 3.43647480e-01 1.82855055e-01
1.30888379e+00 -1.81134596e-01 -6.97451591e-01 8.72916043e-01
1.17913318e+00 2.11159602e-01 -5.55165112e-01 6.82816505e-02
4.85826641e-01 -6.58828318e-02 -1.28587514e-01 -1.01413453e+00
-8.37369561e-01 7.08516419e-01 1.86334804e-01 8.69042933e-01
1.36056185e+00 3.04640025e-01 5.56671441e-01 1.67022813e-02
3.04770708e-01 -1.29072785e+00 4.24392112e-02 8.04481566e-01
8.21741402e-01 -1.24270654e+00 -2.69392461e-01 -6.45970285e-01
-1.19387078e+00 1.00618732e+00 4.02036726e-01 1.28209874e-01
5.60901761e-01 2.04192489e-01 5.61879158e-01 -4.74304378e-01
-6.16602957e-01 -5.52863598e-01 9.01329517e-02 2.53887534e-01
1.11475539e+00 1.87353119e-01 -6.16214812e-01 8.53521109e-01
-5.64749539e-01 -1.20010749e-01 3.72938722e-01 7.95434535e-01
-1.39715776e-01 -1.70647311e+00 -3.16520393e-01 1.53578073e-01
-9.71994698e-01 6.83502555e-02 -1.00275087e+00 8.19349647e-01
6.06520653e-01 1.17773449e+00 2.86228731e-02 -2.53076941e-01
5.19467950e-01 2.82802314e-01 2.24348962e-01 -7.52613127e-01
-8.99312735e-01 -1.50769442e-01 5.51843047e-01 -3.25072020e-01
-1.30266881e+00 -6.26290560e-01 -1.20363355e+00 2.33146131e-01
1.68173984e-01 6.40166938e-01 7.13737428e-01 1.07922089e+00
1.94762930e-01 5.42571425e-01 5.58633626e-01 -6.29294395e-01
-5.77888414e-02 -9.24982309e-01 -7.17796683e-01 9.75758433e-01
-5.96788749e-02 -7.89161384e-01 -4.60018992e-01 6.91191629e-02] | [12.603468894958496, 6.326849937438965] |
65039fce-7ea5-488b-9697-ed89e7dff9e5 | obstacle-tower-a-generalization-challenge-in | 1902.01378 | null | https://arxiv.org/abs/1902.01378v2 | https://arxiv.org/pdf/1902.01378v2.pdf | Obstacle Tower: A Generalization Challenge in Vision, Control, and Planning | The rapid pace of recent research in AI has been driven in part by the presence of fast and challenging simulation environments. These environments often take the form of games; with tasks ranging from simple board games, to competitive video games. We propose a new benchmark - Obstacle Tower: a high fidelity, 3D, 3rd person, procedurally generated environment. An agent playing Obstacle Tower must learn to solve both low-level control and high-level planning problems in tandem while learning from pixels and a sparse reward signal. Unlike other benchmarks such as the Arcade Learning Environment, evaluation of agent performance in Obstacle Tower is based on an agent's ability to perform well on unseen instances of the environment. In this paper we outline the environment and provide a set of baseline results produced by current state-of-the-art Deep RL methods as well as human players. These algorithms fail to produce agents capable of performing near human level. | ['Jonathan Harper', 'Vincent-Pierre Berges', 'Ahmed Khalifa', 'Danny Lange', 'Adam Crespi', 'Julian Togelius', 'Ervin Teng', 'Arthur Juliani', 'Hunter Henry'] | 2019-02-04 | null | null | null | null | ['board-games'] | ['playing-games'] | [-9.95843261e-02 1.36984080e-01 2.44713947e-01 8.47600251e-02
-8.17573607e-01 -7.05152392e-01 8.24278712e-01 -1.21371247e-01
-7.58391380e-01 1.05134046e+00 1.02592163e-01 -1.26060173e-01
4.26386669e-02 -8.37841749e-01 -5.54232001e-01 -4.46160555e-01
-4.68310148e-01 1.03695476e+00 4.73505020e-01 -5.74517071e-01
1.35372043e-01 3.68796110e-01 -1.58945048e+00 1.80854306e-01
5.74412286e-01 5.34196138e-01 3.92947674e-01 1.05572140e+00
3.19681644e-01 1.34903789e+00 -7.49065280e-01 1.70409441e-01
7.31270015e-01 -6.66997790e-01 -6.05776787e-01 -7.14731812e-02
-1.47572488e-01 -4.68474865e-01 -4.20397103e-01 6.58108950e-01
7.54067302e-01 5.84158242e-01 4.35185730e-01 -1.25973833e+00
-6.14432283e-02 4.13402051e-01 -4.38416928e-01 1.01790532e-01
8.24263275e-01 8.09285998e-01 9.42187786e-01 -6.08470321e-01
9.32636857e-01 1.32493865e+00 4.97624606e-01 4.71645653e-01
-1.10988986e+00 -5.57347775e-01 1.98530212e-01 -3.59958261e-01
-1.08749115e+00 -3.33032101e-01 3.58057439e-01 -2.64068365e-01
1.39949715e+00 -9.28354338e-02 1.07854724e+00 1.19359851e+00
2.80309021e-01 1.00502026e+00 1.28664601e+00 -2.15666667e-01
6.65367901e-01 -4.51999038e-01 -7.53002465e-01 9.90554631e-01
1.07705042e-01 7.45574653e-01 -5.13393164e-01 -1.01796873e-01
1.30389893e+00 -6.60367727e-01 1.39247611e-01 -6.57703817e-01
-1.28599441e+00 7.78875113e-01 5.44468522e-01 1.19526714e-01
-6.85296834e-01 6.81825519e-01 1.56186491e-01 2.34153375e-01
-1.20822906e-01 1.09679675e+00 -1.98040172e-01 -6.45230353e-01
-5.26192248e-01 1.31435287e+00 9.53182220e-01 8.02777946e-01
3.93900901e-01 4.17710543e-01 -1.74911499e-01 4.30779725e-01
3.25773627e-01 2.46403947e-01 2.90172696e-01 -1.64570010e+00
3.59598607e-01 4.24457043e-01 6.56132817e-01 -8.02407324e-01
-7.44575560e-01 -4.74841535e-01 -2.21001446e-01 1.24520195e+00
5.99539757e-01 -5.42970240e-01 -1.03878212e+00 1.76164448e+00
3.99391651e-01 2.31460080e-01 4.15948093e-01 1.13729000e+00
7.95273423e-01 7.59506881e-01 4.67335507e-02 2.64947236e-01
9.26737428e-01 -1.30642128e+00 -7.55954310e-02 -7.88066089e-01
5.16897440e-01 -3.54251146e-01 1.01446390e+00 5.96756518e-01
-1.64753914e+00 -4.34407532e-01 -8.90914083e-01 1.51778430e-01
-1.02750473e-01 -3.98604095e-01 9.67449486e-01 2.64711231e-01
-1.23194051e+00 5.91842949e-01 -8.95954967e-01 -1.91498727e-01
5.33891380e-01 3.67310375e-01 -3.57026279e-01 -1.32969692e-01
-9.44463074e-01 1.25873876e+00 3.75489414e-01 -1.58593729e-01
-1.69305718e+00 -3.80991071e-01 -8.84639978e-01 -8.30546618e-02
6.51552558e-01 -8.49396586e-01 1.91840279e+00 -8.63888025e-01
-1.81461406e+00 9.81112540e-01 5.08166134e-01 -5.62035680e-01
1.04384732e+00 -3.17438453e-01 2.83195078e-01 -1.54340088e-01
3.30069214e-01 1.05138898e+00 3.94389600e-01 -1.46500754e+00
-8.82802486e-01 1.37855550e-02 4.92284894e-01 8.31329346e-01
8.39946628e-01 4.77272235e-02 -2.74645150e-01 -3.56031537e-01
-3.29371899e-01 -9.77657437e-01 -9.62900758e-01 -8.34467709e-02
-9.70114022e-02 -9.39337388e-02 2.69045532e-01 -9.60616246e-02
4.54190582e-01 -1.64971757e+00 3.60981464e-01 5.48017845e-02
1.54873818e-01 -9.98713449e-02 -3.46392959e-01 5.37907243e-01
2.91864514e-01 -8.99504870e-02 -1.52188256e-01 -3.22145849e-01
3.86284798e-01 3.10004056e-01 6.80729970e-02 3.45759094e-01
6.07301891e-02 1.09803450e+00 -1.37607932e+00 -3.49518239e-01
1.50122881e-01 1.37560427e-01 -7.51295865e-01 2.24602923e-01
-6.25211477e-01 7.15591013e-01 -7.41986990e-01 5.25820494e-01
-4.52844128e-02 -3.65334516e-03 3.69705036e-02 8.58787179e-01
-3.49200815e-01 3.53086084e-01 -1.10160136e+00 2.25786948e+00
-2.27312177e-01 3.86903554e-01 4.17932272e-01 -3.76653343e-01
6.75783217e-01 2.23280296e-01 4.06174064e-01 -1.15140998e+00
3.55856180e-01 2.49627918e-01 4.31934774e-01 -1.11040272e-01
7.00867176e-01 -1.15567505e-01 -4.39136446e-01 5.64494252e-01
-2.83382595e-01 -7.99191415e-01 3.27985495e-01 1.29769430e-01
1.59963644e+00 7.81313181e-01 3.43642861e-01 -2.43846938e-01
-1.57679528e-01 7.07926214e-01 4.60042894e-01 1.36573505e+00
-4.18394655e-01 5.63695312e-01 5.58579385e-01 -5.32577276e-01
-1.18992031e+00 -1.13673341e+00 5.42184174e-01 1.13273454e+00
2.41212755e-01 -3.72566342e-01 -7.54336774e-01 -3.33095819e-01
-1.22499801e-01 7.41679788e-01 -6.52640462e-01 5.73323332e-02
-8.65322351e-01 -5.81290722e-01 6.70961976e-01 5.75453579e-01
6.57988131e-01 -1.74960148e+00 -1.52499712e+00 5.75183868e-01
2.26808280e-01 -1.04561925e+00 2.92746853e-02 3.59000236e-01
-4.56294298e-01 -1.04415607e+00 -5.43111444e-01 -8.16003978e-01
2.60951251e-01 -1.05338311e-02 1.71845293e+00 1.26797706e-01
-5.30770302e-01 3.50880206e-01 -3.19388986e-01 -5.99542558e-01
-4.28654343e-01 1.99925117e-02 3.89314033e-02 -8.57531369e-01
-1.51223138e-01 -5.96522152e-01 -5.05645454e-01 1.72619700e-01
-4.73931789e-01 3.93159717e-01 7.24812746e-01 6.77563071e-01
4.76553351e-01 9.33318138e-02 2.73403615e-01 -7.03901649e-01
1.02792084e+00 -4.68984008e-01 -8.89994800e-01 -4.11943078e-01
-1.73110273e-02 -7.76057364e-03 5.00414789e-01 -2.21338302e-01
-8.19538951e-01 2.72172809e-01 -6.33926094e-02 -1.78076655e-01
-3.07063669e-01 3.56098711e-01 1.18453696e-01 -3.49762104e-02
1.04369795e+00 -3.88078950e-02 -9.49851274e-02 7.89590552e-02
3.10836494e-01 -1.99283332e-01 6.53037190e-01 -1.10699916e+00
8.22632134e-01 2.05315098e-01 -4.66757780e-03 -3.68403107e-01
-4.09586459e-01 9.89706349e-03 -2.30592147e-01 -3.84121180e-01
8.38322580e-01 -1.02709341e+00 -1.01376641e+00 5.65174222e-01
-8.74561548e-01 -1.40838659e+00 -6.85669124e-01 3.43152374e-01
-1.19957960e+00 -4.54653323e-01 -6.33430839e-01 -7.81305075e-01
2.28956789e-01 -1.38258016e+00 1.02660525e+00 3.98018003e-01
-4.00263369e-01 -6.40448570e-01 7.08785951e-01 1.20366037e-01
4.77341592e-01 7.77496815e-01 4.86488044e-01 -2.07741544e-01
-7.72527754e-01 6.05503321e-02 7.36859962e-02 -5.16385317e-01
-2.87359238e-01 -2.96708286e-01 -7.69776762e-01 -2.89597809e-01
-4.02468354e-01 -1.02647471e+00 4.39903677e-01 3.78341347e-01
4.82105196e-01 1.55728951e-01 -2.81466395e-01 3.73582721e-01
1.15711176e+00 4.67942834e-01 6.53676152e-01 8.11768889e-01
3.56536388e-01 4.61841285e-01 7.63867199e-01 6.10198259e-01
4.53475207e-01 6.74273551e-01 9.02246296e-01 -2.09572211e-01
9.02408436e-02 -4.17158723e-01 2.05657557e-01 3.04057971e-02
-4.25459683e-01 -3.84917319e-01 -1.22967625e+00 4.84809995e-01
-2.15608048e+00 -1.08878350e+00 3.21263403e-01 1.81590688e+00
6.00288272e-01 5.18244505e-01 4.82068479e-01 -1.37232587e-01
1.32283866e-01 2.93637335e-01 -7.66862214e-01 -5.07483125e-01
-5.80876274e-03 2.90266514e-01 3.02520275e-01 6.70080662e-01
-8.84126246e-01 1.54555750e+00 6.91861773e+00 5.20803511e-01
-5.65566301e-01 -1.23905517e-01 4.75166082e-01 -4.24400657e-01
-6.28674030e-02 -2.10475862e-01 -4.32305515e-01 -7.95813948e-02
6.24737561e-01 -1.14786565e-01 8.48414958e-01 9.55125928e-01
3.44547451e-01 -5.71292877e-01 -1.14069402e+00 8.41320992e-01
-1.62317201e-01 -1.26723766e+00 -5.31935096e-01 1.83884367e-01
9.46169138e-01 3.44383985e-01 2.34993875e-01 9.14127529e-01
1.60322118e+00 -1.76134336e+00 9.09259140e-01 1.57173306e-01
4.33615446e-01 -8.48789811e-01 4.32808846e-01 6.50780022e-01
-1.07042754e+00 -8.59203637e-02 -1.44430429e-01 -6.87621474e-01
1.42286867e-01 -4.40393746e-01 -9.43397403e-01 1.23598233e-01
7.29025960e-01 4.05779421e-01 -9.85854119e-02 1.20268285e+00
-4.51490194e-01 7.65383989e-02 -5.31174421e-01 -2.22411409e-01
1.12393022e+00 -1.12292998e-01 5.31163692e-01 8.42572153e-01
8.88215564e-03 5.83259165e-01 7.90754974e-01 8.53430748e-01
1.92963500e-02 -3.20145905e-01 -9.72552776e-01 1.92963500e-02
3.91955078e-01 9.50731397e-01 -8.34223449e-01 1.95276812e-02
-1.51856691e-01 8.33422422e-01 6.37350321e-01 3.47880900e-01
-8.79545987e-01 1.24976054e-01 9.27333713e-01 -4.19574510e-03
1.68215722e-01 -5.43718815e-01 -2.03234136e-01 -7.19219327e-01
-2.96062082e-01 -1.27624917e+00 1.32995933e-01 -1.03617060e+00
-7.33095706e-01 6.68356478e-01 -2.41260260e-01 -8.53527546e-01
-6.33007824e-01 -3.08189243e-01 -5.54319501e-01 6.90728426e-01
-1.30926907e+00 -7.71993458e-01 -4.29793656e-01 6.91070080e-01
7.17294812e-01 -4.44653600e-01 1.01730239e+00 -1.61923155e-01
-2.05798343e-01 1.44256018e-02 -1.38635457e-01 1.34193331e-01
2.62365609e-01 -1.37496769e+00 8.76645505e-01 5.43550849e-01
3.14095803e-02 -6.99329376e-02 9.22908008e-01 -6.90814853e-01
-1.27955425e+00 -6.62101805e-01 5.86836739e-03 -7.60075212e-01
4.51162398e-01 -4.40600216e-01 -3.06733519e-01 7.61525869e-01
2.74549842e-01 6.07548989e-02 1.97894294e-02 -9.41249803e-02
-2.25664303e-02 3.66021216e-01 -1.19029212e+00 1.01717591e+00
1.36445522e+00 -3.43660153e-02 -5.48022509e-01 1.64429024e-01
3.73697370e-01 -1.15728307e+00 -2.87945151e-01 1.27772585e-01
5.19833326e-01 -1.23459482e+00 1.08073413e+00 -9.01140332e-01
7.36915112e-01 -3.27558845e-01 -1.17131993e-01 -1.69158411e+00
-5.19278646e-01 -8.62598121e-01 1.40728265e-01 3.68769437e-01
3.65280569e-01 -1.73381582e-01 1.30359209e+00 6.15695894e-01
-2.24097639e-01 -7.73345888e-01 -7.91128039e-01 -5.32567084e-01
1.73785985e-01 -3.34763318e-01 3.91319573e-01 4.76045012e-01
-1.64128080e-01 3.58516991e-01 -3.55253905e-01 -2.34140046e-02
5.46855688e-01 3.89719978e-02 1.11795771e+00 -1.10570133e+00
-6.18208647e-01 -7.89797127e-01 -3.11071455e-01 -8.78144026e-01
1.77408025e-01 -5.30683577e-01 6.72886014e-01 -1.92230165e+00
-2.03791469e-01 -8.80958915e-01 5.37479110e-02 3.77714604e-01
1.50387302e-01 8.44380781e-02 4.68478024e-01 1.18856102e-01
-1.05973339e+00 5.67375124e-01 1.56925356e+00 8.07005763e-02
-4.93968993e-01 -3.13152403e-01 -5.20312071e-01 9.31391597e-01
9.42507148e-01 -3.55245739e-01 -6.00158393e-01 -6.74289823e-01
4.15726691e-01 4.44446623e-01 3.25627238e-01 -1.47304666e+00
3.32934171e-01 -5.44936419e-01 4.33585554e-01 -1.75278112e-02
8.05221796e-01 -6.72744155e-01 7.72853717e-02 7.02428877e-01
-3.19712192e-01 4.83376950e-01 4.07705128e-01 3.74920756e-01
3.13020609e-02 2.80640759e-02 7.62924254e-01 -9.50993657e-01
-1.04350376e+00 2.20968515e-01 -7.69921780e-01 6.25506520e-01
1.33885026e+00 -4.44875449e-01 -3.80937129e-01 -6.01598978e-01
-6.31706715e-01 6.43141091e-01 7.20642745e-01 3.28099459e-01
5.67059100e-01 -9.44209635e-01 -9.88901556e-01 -1.18192621e-01
-2.60525614e-01 4.42406714e-01 -6.35855794e-02 2.64446884e-01
-1.12682700e+00 2.00397611e-01 -7.79995501e-01 -2.26326600e-01
-8.15215349e-01 2.81022251e-01 7.31829822e-01 -6.98125660e-01
-8.42911839e-01 1.06951988e+00 2.82299250e-01 -6.63541317e-01
4.15949494e-01 -4.28491496e-02 -1.19642831e-01 -3.88805538e-01
4.63525325e-01 1.39115632e-01 -3.56840968e-01 -3.94514591e-01
-3.62009108e-01 2.09862694e-01 1.90956846e-01 -7.25414395e-01
1.45780706e+00 4.20308083e-01 5.90066373e-01 2.06722692e-01
2.56440252e-01 -2.19300017e-01 -1.82124424e+00 1.90754771e-01
-2.07230553e-01 -4.80161726e-01 -7.61935562e-02 -1.10178268e+00
-8.06694627e-01 4.99615222e-01 1.96693599e-01 -1.33834496e-01
5.69734931e-01 -1.51115075e-01 5.80253541e-01 6.22079670e-01
1.05202222e+00 -1.14324856e+00 4.97458875e-01 8.58735263e-01
9.12263036e-01 -1.21707654e+00 -1.02945134e-01 4.57705669e-02
-8.35945010e-01 6.11232638e-01 1.04504919e+00 -7.56826401e-01
-6.97746947e-02 7.73652732e-01 1.55671254e-01 -4.04298663e-01
-1.05280566e+00 -4.72400248e-01 -4.83446747e-01 9.75351572e-01
8.40655714e-02 -3.59067693e-02 3.24385136e-01 2.80499995e-01
-6.17631614e-01 6.14959560e-02 6.63307905e-01 1.38775468e+00
-6.68821037e-01 -7.88622200e-01 -3.12308490e-01 6.30015135e-02
-2.23662063e-01 2.63621807e-01 -5.96678734e-01 1.15867853e+00
-7.00582713e-02 7.58806705e-01 3.67279015e-02 -1.31311476e-01
5.05879343e-01 -4.64481831e-01 7.71470487e-01 -9.20100391e-01
-1.12380826e+00 -2.69623667e-01 4.73618448e-01 -9.86233771e-01
-1.43602878e-01 -6.85880303e-01 -1.68611515e+00 -4.63504374e-01
4.40578848e-01 1.87063262e-01 3.60586554e-01 8.47799778e-01
9.15071741e-03 7.37749696e-01 -8.59945640e-02 -1.48582733e+00
-1.96373478e-01 -4.32545781e-01 -4.60783213e-01 8.57179910e-02
1.84864938e-01 -8.25069189e-01 7.84106106e-02 -6.32459462e-01] | [3.9435184001922607, 1.3190914392471313] |
b517b3ae-36d2-4a6c-9535-8f441b1b349a | automatic-extraction-of-time-expressions | null | null | https://aclanthology.org/D15-1055 | https://aclanthology.org/D15-1055.pdf | Automatic Extraction of Time Expressions Accross Domains in French Narratives | null | ["Aur{\\'e}lie N{\\'e}v{\\'e}ol", 'Mike Donald Tapi Nzali', 'Xavier Tannier'] | 2015-09-01 | null | null | null | emnlp-2015-9 | ['temporal-information-extraction'] | ['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] | [-7.248049736022949, 3.7637124061584473] |
a93b4ac0-b972-4d7f-9462-7da7814b7688 | fine-tuning-siamese-networks-to-assess-sport | null | null | https://www.scitepress.org/PublicationsDetail.aspx?ID=qYIEOJLsu40=&t=1 | https://www.scitepress.org/PublicationsDetail.aspx?ID=qYIEOJLsu40=&t=1 | Fine-tuning Siamese Networks to Assess Sport Gestures Quality | This paper presents an Action Quality Assessment (AQA) approach that learns to automatically score action
realization from temporal sequences like videos. To manage the small size of most of databases capturing
actions or gestures, we propose to use Siamese Networks. In the literature, Siamese Networks are widely used
to rank action scores. Indeed, their purpose is not to regress scores but to predict a value that respects true
scores order so that it can be used to rank actions according to their quality. For AQA, we need to predict
real scores, as well as the difference between these scores and their range. Thus, we first introduce a new loss
function to train Siamese Networks in order to regress score gaps. Once the Siamese network is trained, a
branch of this network is extracted and fine-tuned for score prediction. We tested our approach on a public
database, the AQA-7 dataset, composed of videos from 7 sports. Our results outperform state of the art on
AQA task. Moreover, we show that the proposed method is also more efficient for action ranking. | ['Mégane Millan', 'Catherine Achard'] | 2020-02-27 | null | null | null | international-joint-conference-on-computer | ['action-quality-assessment'] | ['computer-vision'] | [ 3.45527261e-01 -2.07097337e-01 -2.98559576e-01 -4.77124661e-01
-1.08959448e+00 -4.05910790e-01 4.07216102e-01 -1.65244704e-03
-7.28402913e-01 6.45660102e-01 4.34423268e-01 4.10688370e-01
-2.59490788e-01 -7.82755911e-01 -7.29176521e-01 -4.97745812e-01
-3.59328151e-01 4.92973089e-01 6.76481307e-01 -2.50812173e-01
5.34061432e-01 4.41552699e-01 -1.61994374e+00 4.80218291e-01
7.68960238e-01 1.32134891e+00 -1.88117549e-01 7.11265266e-01
1.58386678e-01 1.25045729e+00 -8.13902557e-01 -4.69573259e-01
3.64955753e-01 -7.35773742e-01 -9.29736018e-01 -2.36545831e-01
7.06261754e-01 -5.80035627e-01 -3.08326542e-01 9.31674063e-01
2.30679929e-01 4.17805403e-01 6.63043380e-01 -1.23033309e+00
4.82184030e-02 7.65820801e-01 -1.23903729e-01 7.96367675e-02
5.87018192e-01 2.06063524e-01 1.49872303e+00 -5.45747101e-01
7.89165556e-01 1.00639617e+00 5.00661075e-01 6.45725429e-01
-7.53529966e-01 -5.79557478e-01 -4.34694849e-02 6.53384447e-01
-8.79984260e-01 -2.86038667e-01 7.61771858e-01 -3.81105751e-01
5.77612460e-01 1.60769761e-01 1.05243766e+00 1.07318962e+00
-1.34496614e-01 1.19259059e+00 8.22193146e-01 -6.08022697e-02
4.06497061e-01 -5.29129803e-01 -3.91262881e-02 6.15318835e-01
-1.62655264e-01 -7.49988109e-02 -9.97103333e-01 6.27513900e-02
5.63656211e-01 -3.37320268e-02 2.21450031e-02 -6.65353179e-01
-1.40520942e+00 5.07861018e-01 4.90313023e-01 3.12629402e-01
-6.87882841e-01 3.71763229e-01 5.32445371e-01 3.49340886e-01
1.43024161e-01 6.93203092e-01 -3.09249550e-01 -6.77904963e-01
-9.77419496e-01 6.17561102e-01 6.85725272e-01 1.96801439e-01
4.64870036e-01 -3.22446674e-01 -6.89621627e-01 9.02849376e-01
-5.24333026e-03 3.41750681e-01 6.84169054e-01 -1.43959928e+00
6.41974568e-01 8.07190359e-01 2.77242541e-01 -1.09463453e+00
-2.16111660e-01 4.26036939e-02 -3.86648059e-01 4.28455949e-01
1.07742882e+00 1.95687726e-01 -6.72115803e-01 1.78469622e+00
1.92222133e-01 3.87754112e-01 9.83769353e-03 1.19003105e+00
4.42473322e-01 4.08668131e-01 -1.36880249e-01 5.61752282e-02
9.54494715e-01 -9.94823813e-01 -3.86775345e-01 7.10655525e-02
7.65180707e-01 -3.84672642e-01 1.28721869e+00 8.42782795e-01
-1.23192465e+00 -5.64150035e-01 -1.05243003e+00 9.51226354e-02
-1.09268226e-01 2.78011620e-01 3.96324754e-01 2.09357306e-01
-8.67673635e-01 1.37766898e+00 -1.05916739e+00 -2.08560720e-01
5.93259811e-01 3.61279875e-01 -3.17793667e-01 1.01002000e-01
-1.20369077e+00 7.19967306e-01 5.18184423e-01 -3.11909840e-02
-1.10218644e+00 -3.74806106e-01 -6.44025564e-01 1.89121053e-01
6.28610373e-01 -2.00590640e-01 1.34646475e+00 -1.22918677e+00
-1.75671923e+00 6.83701754e-01 2.87614197e-01 -8.43503892e-01
8.51438165e-01 -4.75800008e-01 -1.95474386e-01 3.83958459e-01
1.23799115e-01 6.37158692e-01 6.20572448e-01 -5.94543576e-01
-9.10009027e-01 -3.56624216e-01 4.71288562e-01 1.58305541e-01
-5.66425860e-01 -1.43864408e-01 -4.76241559e-01 -4.33197647e-01
-1.49628576e-02 -9.40904856e-01 6.44202577e-04 1.35541841e-01
-2.47120723e-01 -5.98232865e-01 3.12371403e-01 -6.12289488e-01
1.37630284e+00 -2.03634453e+00 4.10904258e-01 3.23130548e-01
9.09752920e-02 4.86625493e-01 -4.49348986e-01 1.14092700e-01
1.46270469e-01 -3.29268157e-01 -2.32547536e-01 -2.75943521e-02
1.59789503e-01 1.16220959e-01 1.43375784e-01 2.67295092e-01
3.37826401e-01 7.00188875e-01 -1.01984966e+00 -7.12257862e-01
6.76894113e-02 1.90766200e-01 -8.48178923e-01 3.22108686e-01
-3.22056830e-01 2.93316156e-01 -5.97033978e-01 5.13351679e-01
2.31607139e-01 5.36137968e-02 1.95535526e-01 -2.27484062e-01
-2.62431260e-02 2.41691843e-01 -1.33504581e+00 1.93436790e+00
-2.73641884e-01 4.15434718e-01 -5.11884630e-01 -1.08202147e+00
9.08742785e-01 -1.71013328e-03 1.05650663e+00 -7.97108471e-01
9.21674892e-02 3.09767336e-01 7.70917833e-02 -6.37908101e-01
3.94734234e-01 1.94615319e-01 -1.67882293e-01 3.66157919e-01
9.27116573e-02 1.30602391e-02 8.53678286e-01 1.13729350e-01
1.38073659e+00 5.54107428e-01 1.59928322e-01 3.95963877e-01
8.75974178e-01 1.58542231e-01 6.47431970e-01 5.51198781e-01
-2.99192697e-01 5.84807575e-01 9.20093477e-01 -6.91840887e-01
-1.00083756e+00 -9.75546777e-01 5.09860396e-01 1.36887133e+00
1.67792495e-02 -3.94198388e-01 -9.10800815e-01 -1.00237012e+00
-1.70585319e-01 2.57615447e-01 -6.76423490e-01 -2.17127547e-01
-9.26960886e-01 -1.64436817e-01 5.92682719e-01 7.64716983e-01
5.52418351e-01 -1.51229167e+00 -8.04552853e-01 2.30750322e-01
-3.53019387e-01 -7.71999061e-01 -5.90811729e-01 -2.35241741e-01
-7.96703279e-01 -1.47990870e+00 -8.52531850e-01 -3.83384466e-01
1.62484169e-01 -1.28453434e-01 1.22579575e+00 -9.76731908e-03
1.55855075e-01 3.53415869e-02 -5.10129511e-01 -2.68611475e-03
-3.90573531e-01 1.53027177e-01 -1.06826723e-01 2.66268611e-01
3.15565765e-01 -5.17507136e-01 -7.60505259e-01 3.77228349e-01
-9.69201088e-01 -2.79922634e-01 8.58311415e-01 7.48785675e-01
5.99269986e-01 -2.65266448e-01 3.89358461e-01 -7.22183824e-01
5.97498536e-01 3.04349251e-02 -6.55789256e-01 2.04906568e-01
-4.12859112e-01 2.72737354e-01 6.97968960e-01 -4.80729133e-01
-7.44472325e-01 4.12226200e-01 -2.22660169e-01 -3.69052380e-01
-8.09074119e-02 4.50239152e-01 2.13289484e-02 2.56056607e-01
6.40368223e-01 4.12247963e-02 -4.80793007e-02 -3.33588451e-01
1.82883784e-01 3.05784702e-01 6.55266464e-01 -3.68177533e-01
5.71424663e-01 3.94204795e-01 2.04685017e-01 -1.78960696e-01
-1.04703677e+00 -4.11176234e-01 -8.53536367e-01 -6.32149756e-01
1.02970159e+00 -5.14891982e-01 -1.02867448e+00 4.86099094e-01
-8.92918169e-01 -3.63819629e-01 -4.10618275e-01 7.65438497e-01
-9.54787850e-01 1.85427800e-01 -6.19574904e-01 -6.87956393e-01
-1.11590602e-01 -1.02512109e+00 1.03662872e+00 1.44561037e-01
-3.40932816e-01 -5.37434518e-01 4.61610526e-01 6.48991227e-01
1.24596886e-01 4.86479521e-01 5.34525812e-01 -6.57811761e-01
-5.37817955e-01 -2.94297457e-01 -1.49075193e-02 5.28830111e-01
-6.09420203e-02 -4.28433977e-02 -5.72717190e-01 -4.67068553e-02
-5.13705134e-01 -8.28508258e-01 9.71056163e-01 3.57248485e-01
1.27450168e+00 -2.57936478e-01 2.64225304e-01 3.41652602e-01
1.09092653e+00 1.59615159e-01 7.96285152e-01 5.48848212e-01
5.66467404e-01 5.03005505e-01 1.11772811e+00 5.08416235e-01
1.65079236e-01 9.82579291e-01 5.65498471e-01 1.88189164e-01
4.88519110e-02 -4.07826841e-01 6.05121970e-01 6.91901088e-01
-5.80574036e-01 1.27880558e-01 -6.02858722e-01 3.93476337e-01
-1.93293965e+00 -1.23541546e+00 8.22736546e-02 2.30554152e+00
7.69971967e-01 4.79620516e-01 7.58502603e-01 3.23249459e-01
3.67258102e-01 3.59595865e-01 -7.50659466e-01 -1.57285795e-01
2.44691521e-01 2.94728458e-01 2.25358650e-01 2.37068266e-01
-1.13684726e+00 9.00213480e-01 5.54090071e+00 8.53701711e-01
-9.44006324e-01 -1.11826748e-01 3.02308917e-01 -4.13996786e-01
2.17027143e-01 -2.52654493e-01 -2.43530944e-01 4.50246423e-01
8.13838184e-01 2.78117284e-02 3.60051990e-01 9.36529458e-01
3.68670791e-01 -7.27903172e-02 -1.15867460e+00 8.72251093e-01
1.72795713e-01 -9.82165098e-01 5.36568649e-02 -1.85645342e-01
6.50954425e-01 -1.31512329e-01 -1.49446145e-01 5.00856817e-01
2.94194728e-01 -8.49820673e-01 7.43507206e-01 7.87710667e-01
4.61381167e-01 -8.75373662e-01 8.75645399e-01 1.02832422e-01
-1.06745923e+00 -2.17294425e-01 -3.63900214e-01 -1.20854311e-01
1.11776665e-01 2.52716482e-01 -6.14929795e-01 3.62506807e-01
6.65694535e-01 1.03434145e+00 -6.01992846e-01 1.23372304e+00
-2.70902693e-01 4.83653724e-01 -7.09978025e-03 -2.63674557e-01
3.44948143e-01 -1.83844388e-01 3.47809464e-01 7.82197773e-01
5.07831573e-01 -1.10960864e-01 1.70663536e-01 4.20374930e-01
-1.73274845e-01 2.99896806e-01 -3.39263022e-01 -2.39558503e-01
1.27544507e-01 1.00404644e+00 -6.15255117e-01 -4.66861129e-01
-2.02775896e-01 1.00744760e+00 3.80072951e-01 -1.64511412e-01
-9.46379006e-01 -4.65867728e-01 6.05446279e-01 -5.21732047e-02
2.44908139e-01 2.75311656e-02 1.51324123e-01 -1.16980267e+00
2.26098850e-01 -9.93184268e-01 6.05075836e-01 -7.35148132e-01
-9.77004349e-01 4.82504487e-01 -1.52770177e-01 -1.81544960e+00
-6.51752710e-01 -5.23840666e-01 -4.64846879e-01 1.56324089e-01
-1.22591329e+00 -6.64658785e-01 -4.92762625e-01 6.33362651e-01
6.79374218e-01 -5.69051467e-02 4.10196841e-01 5.16240060e-01
-3.05945605e-01 3.96035016e-01 -1.48230880e-01 4.75814372e-01
7.53164053e-01 -1.39020121e+00 5.95009439e-02 5.88639379e-01
3.46235484e-01 1.07783064e-01 6.23981714e-01 -4.20521796e-01
-1.00067019e+00 -7.63469279e-01 8.48794103e-01 -4.32916045e-01
7.37392843e-01 3.04148253e-02 -7.68313944e-01 2.14797571e-01
-2.67240286e-01 -6.74273670e-02 3.14832509e-01 4.71766368e-02
-2.44186044e-01 -5.89519918e-01 -8.92114282e-01 4.86056656e-01
1.26635242e+00 -2.38478094e-01 -5.50240099e-01 1.75517410e-01
3.34155619e-01 -2.83771962e-01 -9.10617292e-01 3.58654648e-01
1.10304832e+00 -1.42256200e+00 8.29712212e-01 -7.89971709e-01
8.84545326e-01 -2.12221652e-01 5.99495545e-02 -1.42479718e+00
6.38213707e-03 -1.30230904e-01 -9.90602449e-02 9.29993331e-01
3.17515820e-01 -1.58721179e-01 1.24229479e+00 2.43606746e-01
-3.50721218e-02 -8.11554790e-01 -9.23968077e-01 -8.02004278e-01
-2.09107757e-01 -3.33281666e-01 7.41829216e-01 6.52823925e-01
-1.54007077e-01 6.00957870e-02 -6.39567971e-01 -2.74299115e-01
5.44783950e-01 9.64095145e-02 9.61254478e-01 -1.36530900e+00
-3.69687110e-01 -7.12807119e-01 -6.99006557e-01 -8.77179384e-01
4.13746350e-02 -6.72364652e-01 2.07205549e-01 -1.35444105e+00
2.27528170e-01 5.74940536e-03 -8.24346483e-01 3.65044087e-01
-4.91778888e-02 4.90284830e-01 4.56594139e-01 2.14205384e-01
-1.20738471e+00 5.42388380e-01 1.47189176e+00 -2.53336668e-01
-3.46647620e-01 3.59147310e-01 -1.07712299e-01 8.08210075e-01
7.62933612e-01 -5.95102131e-01 -3.04302424e-01 -2.60346681e-01
3.53199095e-01 3.60489875e-01 3.39208066e-01 -1.57366049e+00
5.05454056e-02 -2.90062398e-01 2.46519998e-01 -7.15509653e-01
3.30585599e-01 -5.96936882e-01 -2.27064699e-01 5.45962214e-01
-8.14017713e-01 -3.16625275e-02 -6.46630168e-01 2.98890233e-01
-6.45912468e-01 -2.77178496e-01 6.81793928e-01 -8.26794058e-02
-8.07611644e-01 2.90072262e-01 -7.01906905e-02 1.33281603e-01
9.08332586e-01 -2.14834601e-01 -2.73387344e-03 -4.58843321e-01
-8.55073094e-01 2.05039337e-01 2.06259161e-01 3.80466729e-01
5.69919825e-01 -1.63899922e+00 -7.52839386e-01 -1.85486659e-01
3.23428690e-01 -3.53098303e-01 2.84054339e-01 8.68681610e-01
-6.04334295e-01 2.61692405e-01 -6.68277144e-01 -4.82683390e-01
-1.31808043e+00 2.09266171e-01 2.51717687e-01 -6.76267326e-01
-4.12350148e-01 5.56227505e-01 -3.83841574e-01 -2.50614792e-01
5.02037942e-01 -5.37183881e-01 -6.41657650e-01 2.43238404e-01
6.02177441e-01 7.02563107e-01 -1.70832083e-01 -5.71026504e-01
-2.95613289e-01 5.53281128e-01 2.70083040e-01 -2.31024459e-01
1.50914478e+00 3.25682521e-01 1.63404778e-01 4.21549201e-01
1.09913218e+00 -1.62370890e-01 -1.57010114e+00 -1.06473044e-01
3.19340646e-01 -6.05313897e-01 -4.34867531e-01 -6.74429536e-01
-1.34933484e+00 7.11566746e-01 7.83957839e-01 6.67925552e-02
1.29913509e+00 -3.40660848e-02 9.62756336e-01 7.43045032e-01
3.51078272e-01 -1.49362481e+00 8.13711345e-01 4.76297289e-01
7.64924347e-01 -1.18654585e+00 -2.77588785e-01 1.53959677e-01
-8.86336267e-01 9.86366570e-01 7.37729371e-01 -3.94524723e-01
1.08947866e-01 -3.08418721e-01 1.27522603e-01 -1.81058064e-01
-5.22562385e-01 -3.89798760e-01 4.17569041e-01 3.66695434e-01
2.44010076e-01 1.59205273e-02 -6.17714405e-01 3.31021994e-01
-3.29536200e-01 4.52108085e-01 3.32299918e-01 7.15727508e-01
-4.95721400e-01 -1.04871380e+00 -1.50181651e-01 6.26426458e-01
-5.67502737e-01 3.62115562e-01 -4.99237776e-01 5.20563900e-01
1.56943351e-01 7.73210704e-01 -2.10558578e-01 -6.27071738e-01
8.21540296e-01 1.45874605e-01 4.11621630e-01 -2.21445054e-01
-6.00136578e-01 -3.33779991e-01 1.71346247e-01 -1.33461261e+00
-7.68842041e-01 -8.20929646e-01 -1.14981663e+00 -5.04363840e-03
3.59738655e-02 1.75363868e-01 4.90202516e-01 8.88380766e-01
-1.90212220e-01 5.03234148e-01 6.87271476e-01 -6.61981940e-01
-7.59348869e-01 -1.03933609e+00 -6.52844548e-01 1.04173493e+00
6.96847886e-02 -8.19090664e-01 -1.85725704e-01 -5.03747724e-02] | [8.052640914916992, 0.6198624968528748] |
6a70853c-5658-443c-832f-57a9a558a057 | fast-interactive-video-object-segmentation | 2103.03821 | null | https://arxiv.org/abs/2103.03821v2 | https://arxiv.org/pdf/2103.03821v2.pdf | Fast Interactive Video Object Segmentation with Graph Neural Networks | Pixelwise annotation of image sequences can be very tedious for humans. Interactive video object segmentation aims to utilize automatic methods to speed up the process and reduce the workload of the annotators. Most contemporary approaches rely on deep convolutional networks to collect and process information from human annotations throughout the video. However, such networks contain millions of parameters and need huge amounts of labeled training data to avoid overfitting. Beyond that, label propagation is usually executed as a series of frame-by-frame inference steps, which is difficult to be parallelized and is thus time consuming. In this paper we present a graph neural network based approach for tackling the problem of interactive video object segmentation. Our network operates on superpixel-graphs which allow us to reduce the dimensionality of the problem by several magnitudes. We show, that our network possessing only a few thousand parameters is able to achieve state-of-the-art performance, while inference remains fast and can be trained quickly with very little data. | ['András Lőrincz', 'Viktor Varga'] | 2021-03-05 | null | null | null | null | ['interactive-video-object-segmentation'] | ['computer-vision'] | [ 3.48798990e-01 1.95193693e-01 -4.77845930e-02 -4.41507310e-01
-6.43390417e-01 -6.72224104e-01 2.49788657e-01 2.71781236e-01
-8.20870996e-01 5.68889141e-01 -5.97541988e-01 -3.71148348e-01
2.94459045e-01 -6.59617901e-01 -8.99026513e-01 -5.45783937e-01
5.39946370e-02 8.99079621e-01 6.62701666e-01 2.05410138e-01
1.82249881e-02 4.70325321e-01 -1.45671606e+00 4.74107526e-02
5.99248171e-01 1.09419286e+00 1.45437971e-01 9.25002813e-01
-2.29274750e-01 1.07893157e+00 -4.25949723e-01 -4.43095565e-01
2.20028743e-01 -2.79985785e-01 -1.10474098e+00 3.94076586e-01
7.48286605e-01 -4.94728565e-01 -1.81785852e-01 1.09239590e+00
2.18996361e-01 2.16402769e-01 4.29972440e-01 -1.14799511e+00
-7.75095224e-02 5.46855986e-01 -3.77794862e-01 1.74429044e-01
-9.99526754e-02 8.44860598e-02 8.78783405e-01 -6.06852889e-01
7.33239532e-01 7.68804431e-01 7.09747493e-01 5.86766124e-01
-1.29801488e+00 -2.37354040e-01 1.97799444e-01 1.63264737e-01
-1.24278533e+00 -2.12955400e-01 5.50143540e-01 -5.08192539e-01
7.32705414e-01 1.91719159e-01 9.54877198e-01 6.29132271e-01
-3.63808483e-01 7.02950597e-01 6.19051695e-01 -3.63286316e-01
3.06835562e-01 -1.85553700e-01 2.13759243e-01 1.11525536e+00
2.56978065e-01 -5.88568330e-01 -3.76912177e-01 1.69899300e-01
9.89573479e-01 -5.59564121e-02 -1.30362555e-01 -3.71901840e-01
-1.04877830e+00 7.93698192e-01 4.07433480e-01 2.10454047e-01
-2.62592554e-01 6.73658967e-01 6.53957427e-01 2.62486208e-02
5.78211308e-01 2.33150080e-01 -5.13431966e-01 -1.79530948e-01
-1.16828632e+00 1.05850354e-01 1.02044415e+00 9.11342561e-01
9.76691365e-01 -1.42827302e-01 3.77479853e-04 5.93403578e-01
2.78613549e-02 1.01708122e-01 -2.73859594e-02 -1.44024467e+00
3.09471726e-01 6.52669668e-01 1.59502149e-01 -9.46116567e-01
-4.05276835e-01 -2.78072827e-03 -9.40158904e-01 2.28700206e-01
1.01381791e+00 -3.02096128e-01 -1.08060920e+00 1.35724473e+00
4.47005689e-01 2.00219750e-01 -5.21636486e-01 1.05758309e+00
3.93610418e-01 5.38536310e-01 3.67781490e-01 -1.81116685e-01
1.35649753e+00 -1.25377917e+00 -5.53658664e-01 -3.09181571e-01
7.92709887e-01 -5.28765321e-01 9.61604357e-01 4.46178138e-01
-1.35515296e+00 -5.89601994e-01 -7.58746088e-01 -3.77576023e-01
-3.04337591e-01 1.33116096e-01 8.09863508e-01 5.26034832e-01
-1.24437678e+00 7.86519110e-01 -1.18328488e+00 -1.99334413e-01
8.55140150e-01 7.92149901e-01 -3.73671889e-01 -4.29503955e-02
-7.65421033e-01 5.89658439e-01 7.66393721e-01 3.93886715e-01
-7.94865668e-01 -5.45704782e-01 -7.61220276e-01 1.97552368e-01
8.59453559e-01 -7.51120567e-01 1.30606115e+00 -1.28244209e+00
-1.50021505e+00 9.27222788e-01 -2.02938586e-01 -5.96496522e-01
8.37382078e-01 -4.07510012e-01 4.08763289e-01 4.62155312e-01
-1.90738097e-01 9.04542089e-01 8.08384180e-01 -1.00506496e+00
-6.98612928e-01 -2.35863879e-01 1.86490864e-01 1.04276370e-02
-3.71813327e-01 7.17780367e-02 -1.04643118e+00 -3.75957638e-01
4.28216113e-03 -1.24302006e+00 -5.63121498e-01 2.59719342e-01
-3.16260636e-01 -3.94558251e-01 7.65224636e-01 -6.11879170e-01
1.03734314e+00 -1.90430820e+00 1.22714497e-01 1.06401622e-01
5.45116901e-01 5.71912348e-01 6.80693388e-02 -2.89635602e-02
2.47115597e-01 1.50198430e-01 -3.92924786e-01 -4.29933131e-01
-1.50940269e-01 4.54661340e-01 3.27668786e-01 4.25451577e-01
1.36768296e-01 9.76890385e-01 -8.38093996e-01 -8.67110610e-01
1.98602423e-01 4.66231287e-01 -5.86155474e-01 3.42292309e-01
-6.07388377e-01 4.48324710e-01 -3.63123745e-01 3.42305630e-01
4.54305530e-01 -7.39350200e-01 3.66589844e-01 -2.05941066e-01
2.58292574e-02 3.34878340e-02 -1.02841687e+00 1.84704304e+00
-2.31775835e-01 9.32022274e-01 2.36980990e-01 -1.34132135e+00
5.54970324e-01 2.55188912e-01 7.20661879e-01 -1.52640924e-01
4.38306779e-01 7.80709311e-02 -1.81095839e-01 -5.84567547e-01
3.73776108e-01 1.46018922e-01 1.37990206e-01 5.70562005e-01
1.40484855e-01 -1.34361954e-02 7.81201124e-01 6.14467636e-02
1.08559930e+00 1.83669895e-01 -5.15567251e-02 -5.25492132e-02
5.16163647e-01 2.17562795e-01 5.32618880e-01 6.89567566e-01
-8.46234709e-02 3.59507412e-01 7.39716887e-01 -9.14728105e-01
-1.37610233e+00 -6.13445580e-01 3.79020154e-01 1.27789652e+00
1.37650445e-01 -2.22552255e-01 -1.43708897e+00 -7.53709912e-01
-3.70342016e-01 3.03962678e-01 -4.41895634e-01 4.51377720e-01
-9.13384080e-01 -6.24287486e-01 4.58128482e-01 7.51174927e-01
5.76402962e-01 -1.03179622e+00 -7.92756677e-01 2.93896914e-01
-1.93409145e-01 -1.50500476e+00 -2.75460780e-01 1.41097695e-01
-1.00586557e+00 -1.14426982e+00 -6.55765831e-01 -8.48428607e-01
1.02397442e+00 1.39002018e-02 1.40324521e+00 4.66539800e-01
-4.35071051e-01 1.27518654e-01 -2.04258412e-01 -2.30241656e-01
-3.29316080e-01 3.63229483e-01 -3.33258390e-01 5.33106178e-02
3.46505761e-01 -4.43771601e-01 -5.97501576e-01 2.52175689e-01
-1.03465843e+00 3.06376368e-01 3.97641540e-01 6.56329155e-01
7.78684795e-01 -7.39644235e-03 1.05211139e-01 -1.33657527e+00
5.49923629e-02 -6.76141679e-02 -1.14219165e+00 1.82154447e-01
-3.09482217e-01 5.57052940e-02 6.30090594e-01 -4.28389519e-01
-8.72133374e-01 7.14904785e-01 -3.19941565e-02 -5.46049178e-01
-2.80289054e-01 3.33134204e-01 2.34579787e-01 -3.69441122e-01
3.69418293e-01 -2.08496138e-01 -7.05280975e-02 -3.76929134e-01
4.89713252e-01 2.50814289e-01 5.64742863e-01 -3.77763897e-01
6.08825326e-01 5.67331135e-01 2.45297953e-01 -7.03986883e-01
-1.10200083e+00 -5.50032198e-01 -1.07138968e+00 -5.70997775e-01
1.33997297e+00 -6.42495155e-01 -1.02131474e+00 4.15016323e-01
-1.36482918e+00 -8.74180853e-01 -2.15041876e-01 2.59830713e-01
-6.19011104e-01 4.23785627e-01 -1.00186241e+00 -6.05619788e-01
-2.77406096e-01 -1.06735265e+00 9.44147050e-01 1.93024337e-01
-1.02849126e-01 -1.03116405e+00 -2.83514768e-01 6.25368714e-01
1.78887933e-01 2.69059122e-01 6.64050043e-01 -6.00036144e-01
-1.00637925e+00 -3.68783444e-01 -5.47036409e-01 4.11975145e-01
-1.80260926e-01 1.08876891e-01 -8.55806649e-01 -6.78288043e-02
-1.58429593e-01 -2.86781579e-01 6.80376351e-01 3.54907960e-01
1.73260283e+00 -2.71467477e-01 -2.38002971e-01 5.52054286e-01
1.43784130e+00 -1.79477483e-01 4.92401898e-01 1.02686830e-01
1.28927696e+00 5.83387792e-01 4.70302820e-01 1.37168705e-01
2.27329612e-01 5.04171371e-01 4.33126211e-01 -2.92625576e-01
9.96701941e-02 1.66092277e-01 -2.87278086e-01 7.41223991e-01
-6.83168828e-01 -4.82159913e-01 -1.08855200e+00 4.77532178e-01
-2.08796453e+00 -7.21617877e-01 -3.98341477e-01 2.01261520e+00
7.59293795e-01 3.00507963e-01 1.62826106e-01 1.99134871e-01
7.62443960e-01 -7.22901747e-02 -3.33210856e-01 -2.22457498e-01
3.43747288e-01 1.36420116e-01 7.96081364e-01 4.40322578e-01
-1.32445610e+00 1.29801118e+00 6.52119732e+00 7.12331474e-01
-9.93025124e-01 1.08561069e-01 9.35806096e-01 -5.82837611e-02
2.90010601e-01 -7.15550631e-02 -5.59791386e-01 3.41867179e-01
9.92321849e-01 3.59363258e-01 4.46065277e-01 9.47526038e-01
1.41368344e-01 -4.11005229e-01 -1.23687279e+00 1.05321717e+00
-9.40721780e-02 -1.44230974e+00 -1.15697235e-01 -2.22886018e-02
8.10506880e-01 1.92498207e-01 -3.98740977e-01 -1.78436130e-01
3.63029182e-01 -9.45975780e-01 4.46427166e-01 4.06553298e-01
5.91256499e-01 -6.91775024e-01 8.46383393e-01 3.57661575e-01
-1.07099569e+00 9.81196165e-02 -4.06472117e-01 -8.94978419e-02
4.46441114e-01 6.49988413e-01 -6.29503310e-01 9.45188180e-02
7.09274113e-01 2.92935610e-01 -4.89037663e-01 1.00456858e+00
-2.01611325e-01 5.56796014e-01 -4.75560218e-01 -8.01000074e-02
4.90484595e-01 -3.11166435e-01 -6.69884437e-04 1.24332619e+00
8.00339878e-02 1.00403026e-01 5.07034898e-01 6.32685661e-01
-4.16034400e-01 -6.24550022e-02 -3.24485928e-01 -1.56563103e-01
1.42067671e-03 1.49035895e+00 -1.33855951e+00 -7.29195118e-01
-4.32009667e-01 1.21490574e+00 6.96401954e-01 2.93146282e-01
-9.46310163e-01 -2.22850844e-01 1.56032562e-01 2.45754674e-01
5.33031702e-01 -5.09576857e-01 -2.34238923e-01 -9.01149511e-01
-2.33480837e-02 -5.83897948e-01 2.27186352e-01 -5.96355975e-01
-8.86749446e-01 4.93181527e-01 -2.28145525e-01 -7.99497128e-01
-3.44289690e-01 -7.78612554e-01 -2.77999341e-01 4.78523046e-01
-1.28120601e+00 -9.92023885e-01 -6.73303664e-01 5.28750837e-01
6.53634608e-01 3.66398782e-01 6.77988648e-01 6.53557539e-01
-5.73546112e-01 2.00040966e-01 -2.47101575e-01 5.95549583e-01
3.19088191e-01 -1.32991493e+00 4.47103769e-01 7.79612422e-01
3.65572780e-01 2.16362387e-01 5.16725838e-01 -4.32047933e-01
-1.17877674e+00 -1.13027132e+00 9.26369727e-01 -2.97540098e-01
7.50934899e-01 -3.64521056e-01 -9.32921886e-01 7.14374483e-01
1.30979866e-01 4.54661846e-01 5.65011084e-01 -6.57623708e-02
-1.06929325e-01 6.18842170e-02 -6.66679561e-01 4.74379957e-01
9.89939153e-01 -4.61324573e-01 1.04186125e-02 6.48769140e-01
5.15977263e-01 -5.13819933e-01 -8.07013214e-01 4.72711436e-02
2.68365383e-01 -9.31685150e-01 7.53556848e-01 -5.23092210e-01
3.34533095e-01 -4.06240493e-01 3.80039126e-01 -6.01556361e-01
-3.48633081e-02 -6.81353986e-01 -1.23789221e-01 9.95543659e-01
4.28464621e-01 -2.42830783e-01 1.22356212e+00 1.00405490e+00
2.15782076e-02 -6.89771712e-01 -5.88331938e-01 -5.97200274e-01
-4.21346366e-01 -4.31062669e-01 6.47504404e-02 7.16234326e-01
-4.11061347e-01 2.73001194e-01 -2.04076916e-01 -1.29111800e-02
7.12505102e-01 -7.93559626e-02 8.46653223e-01 -1.38837910e+00
-2.25121826e-01 -3.32697958e-01 -3.94789517e-01 -1.09720075e+00
3.08996111e-01 -5.60311615e-01 3.89768869e-01 -1.54555011e+00
9.54190865e-02 -4.66433704e-01 9.45850462e-02 5.70612371e-01
-2.47128159e-01 8.50707948e-01 2.16206655e-01 1.82991609e-01
-1.06729639e+00 -1.51407361e-01 1.15851820e+00 -1.80839017e-01
-3.42460275e-02 -2.55195275e-02 4.31499332e-02 1.20220661e+00
7.89633930e-01 -6.14323378e-01 -3.24541062e-01 -7.61380732e-01
4.76143748e-01 1.49333635e-02 4.52505380e-01 -1.04668188e+00
4.61229354e-01 -3.83865088e-02 3.17739844e-01 -4.50165510e-01
4.50632185e-01 -9.65132892e-01 3.04098297e-02 3.66004229e-01
-3.00041437e-01 -1.02158047e-01 8.85160044e-02 5.77476203e-01
-2.83281714e-01 -6.23883247e-01 6.54536724e-01 -4.63377655e-01
-7.27361262e-01 5.96477330e-01 -4.43491518e-01 9.11131129e-02
1.11482906e+00 -1.15857244e-01 2.04565506e-02 -1.58009931e-01
-8.94420087e-01 1.61849543e-01 5.03983617e-01 -1.76195532e-01
1.39501497e-01 -9.06698704e-01 -3.27271879e-01 -2.84531385e-01
-3.46669257e-01 5.78807056e-01 2.65114039e-01 8.14168513e-01
-1.35243773e+00 2.32597306e-01 -5.72465919e-02 -8.83164048e-01
-1.43247890e+00 6.30049706e-01 2.29425982e-01 -4.77418661e-01
-6.53616428e-01 1.12674928e+00 -1.01003289e-01 2.00213771e-02
3.06484312e-01 -2.76645541e-01 -1.48862332e-01 2.14789823e-01
3.31667095e-01 4.59556311e-01 -4.85168956e-02 -4.41957831e-01
-1.39415801e-01 6.32982194e-01 -7.20698684e-02 -9.53956135e-03
1.29098797e+00 -3.52778323e-02 -3.85871023e-01 3.19167942e-01
1.22894883e+00 -5.34253895e-01 -1.75135708e+00 2.64449194e-02
1.65548444e-01 -3.42636198e-01 9.05152634e-02 -2.66507685e-01
-1.39052272e+00 1.03483617e+00 2.73615092e-01 5.19245863e-01
1.11895657e+00 -5.73324859e-02 1.00973248e+00 7.27845252e-01
2.14795217e-01 -1.54721630e+00 2.20548697e-02 3.53330582e-01
1.02116115e-01 -1.45349276e+00 3.34214829e-02 -7.62775183e-01
-3.23757797e-01 1.28764820e+00 4.59149897e-01 -2.71366477e-01
3.89311641e-01 2.51424491e-01 8.66759196e-02 -2.87989914e-01
-4.49364573e-01 -2.65753418e-01 2.46172547e-01 2.69344598e-01
4.11187112e-01 -1.15797468e-01 -1.80336803e-01 -2.01463774e-01
1.68567151e-01 1.53239116e-01 3.81003439e-01 8.58507633e-01
-3.95891309e-01 -1.11400318e+00 4.06036675e-02 4.28772867e-01
-7.65192866e-01 1.09029887e-02 -1.25300139e-01 7.46597826e-01
1.09434418e-01 7.39014983e-01 2.42136061e-01 1.72369465e-01
-1.00316122e-01 1.51109681e-01 5.11917353e-01 -4.41705197e-01
-5.16944468e-01 7.38224313e-02 1.24168076e-01 -7.79472589e-01
-8.41032982e-01 -2.38525465e-01 -1.37252617e+00 -3.36358637e-01
-3.31903368e-01 -2.96148397e-02 8.12366784e-01 1.07300472e+00
1.40190557e-01 6.80633068e-01 1.21204227e-01 -1.20485771e+00
3.37345712e-02 -7.14165330e-01 -2.65419990e-01 4.51767474e-01
1.02125295e-01 -2.67963320e-01 -1.32215191e-02 7.30498195e-01] | [9.31657886505127, 0.04311717674136162] |
1ccdcaa0-955e-43ab-888a-52b047bf14f3 | egohumans-an-egocentric-3d-multi-human | 2305.16487 | null | https://arxiv.org/abs/2305.16487v1 | https://arxiv.org/pdf/2305.16487v1.pdf | EgoHumans: An Egocentric 3D Multi-Human Benchmark | We present EgoHumans, a new multi-view multi-human video benchmark to advance the state-of-the-art of egocentric human 3D pose estimation and tracking. Existing egocentric benchmarks either capture single subject or indoor-only scenarios, which limit the generalization of computer vision algorithms for real-world applications. We propose a novel 3D capture setup to construct a comprehensive egocentric multi-human benchmark in the wild with annotations to support diverse tasks such as human detection, tracking, 2D/3D pose estimation, and mesh recovery. We leverage consumer-grade wearable camera-equipped glasses for the egocentric view, which enables us to capture dynamic activities like playing soccer, fencing, volleyball, etc. Furthermore, our multi-view setup generates accurate 3D ground truth even under severe or complete occlusion. The dataset consists of more than 125k egocentric images, spanning diverse scenes with a particular focus on challenging and unchoreographed multi-human activities and fast-moving egocentric views. We rigorously evaluate existing state-of-the-art methods and highlight their limitations in the egocentric scenario, specifically on multi-human tracking. To address such limitations, we propose EgoFormer, a novel approach with a multi-stream transformer architecture and explicit 3D spatial reasoning to estimate and track the human pose. EgoFormer significantly outperforms prior art by 13.6% IDF1 and 9.3 HOTA on the EgoHumans dataset. | ['Kris Kitani', 'Minh Vo', 'Richard Newcombe', 'Lingni Ma', 'Aayush Bansal', 'Rawal Khirodkar'] | 2023-05-25 | null | null | null | null | ['pose-estimation', '3d-pose-estimation', 'human-detection'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [-4.55750525e-01 -4.34825540e-01 7.75966123e-02 1.06430076e-01
-4.76347029e-01 -5.57630360e-01 4.07624871e-01 -4.95521218e-01
-3.01370740e-01 3.41270566e-01 5.48522413e-01 6.28908575e-01
2.10203543e-01 -4.67146963e-01 -8.25597823e-01 -4.35859710e-01
1.09888576e-02 8.40197325e-01 3.81823599e-01 -4.33295041e-01
-2.78796047e-01 3.96116674e-01 -1.76541698e+00 1.20867848e-01
1.59427509e-01 6.81971669e-01 -1.49901360e-01 9.14476514e-01
8.82698238e-01 4.75356698e-01 -3.35199356e-01 -3.82434636e-01
4.75831836e-01 1.25996605e-01 -4.28018540e-01 2.68418640e-01
1.09429967e+00 -6.52134359e-01 -7.90338457e-01 7.32614458e-01
1.00309277e+00 3.28264326e-01 3.63398910e-01 -1.32496870e+00
-8.52575973e-02 -8.47131461e-02 -7.84111738e-01 4.01104748e-01
1.11948156e+00 4.49376792e-01 4.70203966e-01 -1.02774751e+00
9.16124761e-01 1.40013933e+00 8.40103686e-01 5.77226460e-01
-7.22990930e-01 -6.80091858e-01 3.45719546e-01 2.21707717e-01
-1.49274933e+00 -5.64282537e-01 5.54732800e-01 -6.48211479e-01
1.01113999e+00 1.27547219e-01 1.22453308e+00 1.58408964e+00
3.45943004e-01 1.06448257e+00 6.64454162e-01 -1.06410041e-01
-1.30420655e-01 -3.70964289e-01 -5.13984524e-02 8.34258080e-01
4.01525378e-01 6.13915548e-02 -8.06991577e-01 1.31243309e-02
1.05379438e+00 4.01860446e-01 -2.83054173e-01 -1.14677727e+00
-1.65444469e+00 3.34982723e-01 3.87750864e-01 -2.71292597e-01
-4.53564584e-01 3.99227917e-01 6.01263046e-01 -9.95651633e-02
5.81710577e-01 2.42100190e-03 -3.75727624e-01 -1.67273313e-01
-6.45057499e-01 8.60369265e-01 3.66890550e-01 1.52376342e+00
2.25975037e-01 4.87318300e-02 -4.25033480e-01 3.76208901e-01
2.96102822e-01 1.05877233e+00 9.76289362e-02 -8.62060845e-01
5.97435236e-01 5.39510906e-01 3.66577566e-01 -1.00991082e+00
-7.56989121e-01 -5.80334187e-01 -5.72042346e-01 -2.85187274e-01
6.15850389e-01 -1.39589816e-01 -7.53013909e-01 1.62503994e+00
9.28051651e-01 3.34448278e-01 -5.02679467e-01 1.49140370e+00
9.55758333e-01 5.57760745e-02 -2.62795687e-02 2.08090842e-01
1.92028272e+00 -1.22558475e+00 -3.43244016e-01 -3.16020370e-01
5.23732066e-01 -5.38499117e-01 8.73626173e-01 3.51804405e-01
-1.14439917e+00 -6.40159547e-01 -7.31387556e-01 -9.96525213e-02
-1.15961418e-01 1.70747533e-01 6.55133307e-01 6.63675189e-01
-6.56391025e-01 1.65111095e-01 -1.00648201e+00 -7.43221700e-01
1.72392488e-01 1.76589295e-01 -7.79305696e-01 -2.17145309e-01
-1.06676638e+00 7.17366993e-01 9.59395543e-02 5.93734300e-03
-1.16872215e+00 -9.93004560e-01 -1.08046055e+00 -1.19167805e-01
7.56335139e-01 -1.59176099e+00 1.10702300e+00 -1.03956953e-01
-1.13078165e+00 1.25620687e+00 -4.98411059e-02 -3.09576184e-01
8.71528804e-01 -1.08994699e+00 -3.87912810e-01 2.19695225e-01
3.16718221e-01 2.78719097e-01 6.71871185e-01 -7.96330512e-01
-4.40955430e-01 -9.64555204e-01 1.13838106e-01 4.75322992e-01
1.42583400e-02 9.77864861e-02 -1.12385857e+00 -7.51103580e-01
8.09923559e-02 -1.28646886e+00 -6.72492608e-02 -2.69065402e-03
-4.84115660e-01 -1.09309647e-02 6.75440073e-01 -3.82796854e-01
9.57496524e-01 -1.85788751e+00 5.32133043e-01 -2.14835450e-01
5.39084375e-01 -4.58635055e-02 3.59416455e-01 9.59247202e-02
6.98432028e-02 -7.07986236e-01 4.58580822e-01 -6.55871332e-01
6.18589893e-02 -2.36670032e-01 6.19982332e-02 1.00120163e+00
-4.58235353e-01 9.86990035e-01 -1.14277565e+00 -5.98295271e-01
5.89680791e-01 6.02098227e-01 -8.71211827e-01 2.44520649e-01
2.39127368e-01 6.04964435e-01 -5.08418620e-01 1.12151957e+00
6.00782275e-01 -3.32821608e-01 -6.87252656e-02 -5.05374372e-01
9.19529200e-02 -2.50397950e-01 -1.40584564e+00 2.42433572e+00
-1.51231214e-01 2.61485040e-01 -1.84266933e-03 -3.88854742e-01
3.10903162e-01 3.69392365e-01 9.00155008e-01 -4.16711837e-01
4.35560018e-01 -1.98013514e-01 -5.74122190e-01 -4.34696198e-01
7.53245592e-01 1.80929065e-01 -2.96677858e-01 1.10366732e-01
1.45135775e-01 3.21232587e-01 1.97997332e-01 2.45507449e-01
1.16058576e+00 6.72692835e-01 3.28215063e-01 -2.04639882e-01
4.03823525e-01 1.73163170e-03 6.26496792e-01 6.00723147e-01
-5.94673336e-01 9.33407247e-01 -2.96679530e-02 -7.35927165e-01
-1.05301881e+00 -1.48821700e+00 2.50372171e-01 1.26107967e+00
4.37244058e-01 -6.60522461e-01 -7.37357616e-01 -5.72414815e-01
2.17896953e-01 2.24183723e-02 -5.55387437e-01 -7.30409985e-03
-8.56860518e-01 -6.93301499e-01 4.80289847e-01 7.29493439e-01
6.02065563e-01 -5.72740197e-01 -1.11804330e+00 1.17664658e-01
-5.09913921e-01 -1.60194635e+00 -8.47558975e-01 -5.25793016e-01
-5.99133193e-01 -1.58900166e+00 -1.21502173e+00 -2.70046592e-01
3.65411878e-01 6.20088696e-01 1.41474581e+00 -5.36026537e-01
-5.05383730e-01 1.12699997e+00 -1.01169750e-01 -4.02977429e-02
5.33291399e-01 -3.17785367e-02 8.35783601e-01 9.20769665e-03
5.13728023e-01 -4.00494456e-01 -1.19217324e+00 7.23491132e-01
1.32848412e-01 -6.75294036e-03 9.12156925e-02 5.03238261e-01
6.40599310e-01 -5.34954309e-01 -8.06504339e-02 -6.32392526e-01
-1.54160887e-01 -4.43099707e-01 -4.37190294e-01 1.28747493e-01
-5.76872788e-02 -5.26487708e-01 1.82361469e-01 -4.26997989e-01
-8.56755376e-01 3.38149667e-01 1.97515517e-01 -1.09009349e+00
-1.37496606e-01 -3.34087014e-01 -3.44586343e-01 -6.69605955e-02
7.95995474e-01 1.92173291e-02 -2.93112636e-01 -4.34498996e-01
3.35646003e-01 2.38825381e-01 1.11700213e+00 -6.19955778e-01
7.61618853e-01 8.44928205e-01 7.11166933e-02 -7.60492682e-01
-7.64224589e-01 -1.01568234e+00 -8.22285354e-01 -6.30852461e-01
1.06939352e+00 -1.80177546e+00 -1.21516597e+00 6.26857102e-01
-1.08030224e+00 -1.07550852e-01 -6.24773875e-02 7.53599524e-01
-9.06806707e-01 4.63843137e-01 -5.82972646e-01 -7.00211108e-01
-6.00813150e-01 -1.17291665e+00 1.77655709e+00 -6.74393773e-02
-3.99280101e-01 -6.60009086e-01 8.89671221e-02 6.34379447e-01
-1.06086925e-01 7.04277039e-01 -1.31677061e-01 -4.91634123e-02
-6.86206162e-01 -3.50793272e-01 1.06176659e-01 -4.48574901e-01
-3.01570445e-01 -6.15997076e-01 -9.07950997e-01 -6.37623310e-01
-3.44141930e-01 -1.65934145e-01 5.67139387e-01 5.91823578e-01
6.57805264e-01 1.81848571e-01 -6.59742355e-01 9.76249099e-01
8.98132741e-01 -5.07691026e-01 5.37231445e-01 3.82368296e-01
1.21422422e+00 3.52610469e-01 8.49401891e-01 8.14598322e-01
7.74500608e-01 1.25255978e+00 4.36544716e-01 1.94546863e-01
-1.90493315e-01 -3.87507826e-01 3.69673371e-01 4.95756984e-01
-5.80812395e-01 -1.62143007e-01 -8.71423602e-01 6.55642092e-01
-1.95144987e+00 -1.10351110e+00 -2.11474389e-01 2.16694283e+00
9.70071927e-02 5.97636402e-02 6.82421446e-01 -1.84377059e-01
7.59790778e-01 1.43801525e-01 -5.90019882e-01 5.91824114e-01
1.98641494e-01 -2.28276923e-01 6.63775086e-01 -3.53217125e-02
-1.57361639e+00 1.03825521e+00 5.84014320e+00 2.85793185e-01
-7.36742735e-01 3.48242193e-01 -9.90038663e-02 -7.91475713e-01
3.21473390e-01 -4.44664687e-01 -1.20964718e+00 3.75992507e-01
4.24242586e-01 1.04021206e-01 9.27280486e-02 1.22138071e+00
1.58763185e-01 -2.77581140e-02 -1.14725530e+00 1.68406224e+00
4.03084636e-01 -1.05118585e+00 -1.61185220e-01 7.71550834e-02
6.61365867e-01 1.82715654e-01 -5.37593588e-02 3.66365731e-01
1.74898833e-01 -5.55389106e-01 9.54030454e-01 6.91596985e-01
7.54185915e-01 -6.70399606e-01 4.22586679e-01 3.83658856e-01
-1.77274370e+00 1.20741196e-01 -1.90825105e-01 -7.78343389e-03
5.75506508e-01 3.17730665e-01 -3.28894824e-01 6.33878648e-01
1.26835024e+00 9.96954381e-01 -5.67067802e-01 1.07898200e+00
1.89096816e-02 -1.29047766e-01 -4.93606836e-01 3.89702797e-01
-1.42945126e-01 1.75665691e-01 9.49493706e-01 1.02202344e+00
3.29041004e-01 2.93913007e-01 5.15961647e-01 4.30063963e-01
1.03870183e-01 -1.84822336e-01 -7.22907662e-01 5.57603836e-01
1.92524865e-01 1.12185824e+00 -5.97603798e-01 -4.61130619e-01
-3.47793192e-01 1.28915393e+00 7.81311244e-02 2.13059679e-01
-1.14210403e+00 1.15802903e-02 1.05065405e+00 6.78007603e-01
2.55285263e-01 -4.50471759e-01 1.82810560e-01 -1.85911381e+00
1.03341885e-01 -8.04372907e-01 6.38109982e-01 -9.77485538e-01
-8.34431469e-01 4.36872900e-01 3.77742827e-01 -1.52859795e+00
-5.56251705e-01 -5.49531877e-01 -1.99689791e-01 5.00918567e-01
-8.53645325e-01 -1.56530571e+00 -1.01851964e+00 1.04386783e+00
6.93716049e-01 -3.42110187e-01 4.75874066e-01 6.76433980e-01
-5.05621195e-01 7.11687863e-01 -3.59368175e-01 1.78872302e-01
1.13187146e+00 -1.04026675e+00 8.05724919e-01 9.23186123e-01
-5.98898530e-02 8.20772886e-01 8.80471706e-01 -8.59963000e-01
-2.04230881e+00 -1.06694400e+00 4.01537806e-01 -1.37224889e+00
2.85927922e-01 -7.45933890e-01 -1.88025758e-01 9.46602464e-01
-4.46988195e-01 4.95426267e-01 2.87814915e-01 2.25797102e-01
-3.11188489e-01 9.11953822e-02 -1.00473595e+00 7.90166080e-01
1.90285468e+00 -3.04131418e-01 -4.21248913e-01 4.13341403e-01
5.05315423e-01 -1.15692544e+00 -9.04900551e-01 4.96696949e-01
1.01129723e+00 -1.06433868e+00 1.55422974e+00 -6.20205581e-01
1.41899243e-01 -5.52329957e-01 -3.18241060e-01 -1.02669203e+00
-4.88579392e-01 -6.28246248e-01 -7.41544068e-01 6.42605782e-01
-6.02006853e-01 -1.45653233e-01 1.28570807e+00 2.75437385e-01
-2.52055913e-01 -4.60003883e-01 -8.20196271e-01 -7.63531983e-01
-6.43672764e-01 -4.01842177e-01 4.99982685e-01 5.37831843e-01
-2.52475113e-01 1.51625052e-01 -9.14625764e-01 2.12575674e-01
1.17392385e+00 -2.99345446e-03 1.63624632e+00 -1.08554780e+00
-3.30769509e-01 1.75095141e-01 -9.52459753e-01 -1.43092728e+00
-3.80571336e-02 -5.17617106e-01 -2.77814299e-01 -1.04390025e+00
3.00081193e-01 1.01889521e-01 1.08965583e-01 8.02822004e-04
-9.63970423e-02 6.08584523e-01 2.73462623e-01 1.70223847e-01
-1.08556831e+00 6.28939033e-01 1.26082230e+00 -9.26351398e-02
3.00713480e-02 6.23195022e-02 -3.52516413e-01 9.07949507e-01
7.48907030e-02 -2.38594547e-01 -3.57243299e-01 -5.50298631e-01
7.03067631e-02 -2.15172339e-02 1.01761711e+00 -1.45807397e+00
5.53412855e-01 1.02762714e-01 7.01140821e-01 -9.74833608e-01
7.61173189e-01 -6.54260993e-01 4.35962617e-01 4.94508803e-01
3.65799934e-01 3.26442480e-01 6.77050874e-02 8.61438155e-01
2.40677759e-01 6.57575488e-01 5.97354352e-01 -5.21189272e-01
-1.07809746e+00 8.14238548e-01 1.85453117e-01 5.26592314e-01
1.21007478e+00 -3.81321549e-01 -1.34631872e-01 -2.57200390e-01
-9.16395783e-01 4.40751195e-01 7.51183271e-01 7.33506918e-01
6.21645987e-01 -1.50843704e+00 -6.74916506e-01 1.97414264e-01
5.40344059e-01 4.20558602e-02 6.71707809e-01 1.00668967e+00
-5.12115180e-01 5.83847463e-01 -4.27508742e-01 -1.09475470e+00
-1.44422674e+00 6.42656267e-01 4.54802632e-01 -1.36796251e-01
-1.13213551e+00 5.33776104e-01 6.45254314e-01 -2.87168413e-01
1.10445067e-01 -7.49114752e-02 -2.41728760e-02 -1.47148922e-01
6.33836150e-01 8.63041222e-01 -1.62554509e-03 -1.11338878e+00
-7.17994332e-01 1.03275025e+00 1.05907731e-01 1.33058921e-01
1.01203668e+00 -2.97646672e-01 5.59229255e-01 1.85445711e-01
8.62425327e-01 -1.08236991e-01 -1.47438741e+00 -2.15722874e-01
-6.41741693e-01 -7.55882800e-01 -2.07119778e-01 -2.09489033e-01
-1.03059256e+00 6.78761363e-01 7.94102609e-01 -6.46897018e-01
6.97149336e-01 7.87507519e-02 9.87977922e-01 3.09120476e-01
1.07044995e+00 -1.05510640e+00 3.04357797e-01 5.46678543e-01
7.93689013e-01 -1.22858799e+00 4.81877089e-01 -5.88546395e-01
-6.32754028e-01 6.72985792e-01 9.90031123e-01 -3.11618030e-01
5.60379803e-01 -3.13286260e-02 -2.49589995e-01 -6.56304002e-01
-4.00451183e-01 -2.56617576e-01 4.49852109e-01 6.73078656e-01
2.82676756e-01 4.81440239e-02 3.21153373e-01 5.96500814e-01
-2.35699430e-01 -4.53837924e-02 5.16946092e-02 8.62678528e-01
-3.43657136e-02 -3.36433172e-01 -7.62203038e-01 1.66674256e-02
-3.16472590e-01 3.99993062e-01 -1.44453831e-02 9.96879816e-01
1.18307546e-01 5.41153848e-01 6.61536306e-02 -4.40781415e-01
9.28039372e-01 -1.91137493e-01 9.17882621e-01 -5.40038884e-01
-6.28477514e-01 2.00971067e-01 1.35487914e-02 -1.16072607e+00
-4.40703213e-01 -9.28166747e-01 -7.32224822e-01 -6.43180549e-01
-1.62137970e-01 -4.10704166e-01 2.80502379e-01 8.18604529e-01
5.48211098e-01 5.72122395e-01 6.58166483e-02 -1.50273514e+00
-3.09764296e-01 -7.89471328e-01 -5.80397129e-01 6.93860471e-01
1.09812960e-01 -1.49620140e+00 8.43172073e-02 -6.73186183e-02] | [7.03298807144165, -0.8572015762329102] |
8a93cfde-560b-46e9-8e84-4fde90353f2e | click-carving-segmenting-objects-in-video | 1607.01115 | null | http://arxiv.org/abs/1607.01115v1 | http://arxiv.org/pdf/1607.01115v1.pdf | Click Carving: Segmenting Objects in Video with Point Clicks | We present a novel form of interactive video object segmentation where a few
clicks by the user helps the system produce a full spatio-temporal segmentation
of the object of interest. Whereas conventional interactive pipelines take the
user's initialization as a starting point, we show the value in the system
taking the lead even in initialization. In particular, for a given video frame,
the system precomputes a ranked list of thousands of possible segmentation
hypotheses (also referred to as object region proposals) using image and motion
cues. Then, the user looks at the top ranked proposals, and clicks on the
object boundary to carve away erroneous ones. This process iterates (typically
2-3 times), and each time the system revises the top ranked proposal set, until
the user is satisfied with a resulting segmentation mask. Finally, the mask is
propagated across the video to produce a spatio-temporal object tube. On three
challenging datasets, we provide extensive comparisons with both existing work
and simpler alternative methods. In all, the proposed Click Carving approach
strikes an excellent balance of accuracy and human effort. It outperforms all
similarly fast methods, and is competitive or better than those requiring 2 to
12 times the effort. | ['Kristen Grauman', 'Suyog Dutt Jain'] | 2016-07-05 | null | null | null | null | ['interactive-video-object-segmentation'] | ['computer-vision'] | [ 3.52737576e-01 -1.28252655e-01 -4.02648784e-02 -3.36760074e-01
-9.28156137e-01 -8.63578141e-01 2.73971617e-01 2.84446299e-01
-6.73455060e-01 4.13178116e-01 -2.99197465e-01 -1.90482900e-01
2.47036591e-01 -3.66687000e-01 -5.46397924e-01 -5.53423643e-01
-1.01827726e-01 7.48994231e-01 1.23021448e+00 1.27754271e-01
5.30345142e-01 3.93410742e-01 -1.45069039e+00 1.76461488e-01
8.63640785e-01 9.26548719e-01 4.86458212e-01 9.33311343e-01
-1.22416250e-01 3.77982497e-01 -3.65669429e-01 -3.14592332e-01
3.56523037e-01 -3.17137480e-01 -1.06689858e+00 5.22252858e-01
5.95908821e-01 -6.12708390e-01 3.48398387e-02 8.95136476e-01
1.40930548e-01 3.77297729e-01 2.59169698e-01 -1.07830119e+00
9.83669311e-02 2.96453208e-01 -7.79396474e-01 4.43287760e-01
5.06076217e-01 3.44884366e-01 1.07251477e+00 -1.30554521e+00
7.44928658e-01 9.45738852e-01 4.18168336e-01 3.39020312e-01
-1.20499253e+00 -3.52596313e-01 5.31477034e-01 -1.90756116e-02
-1.44831884e+00 -3.27980727e-01 2.50747561e-01 -5.25851727e-01
6.61007881e-01 4.76106495e-01 8.03611100e-01 3.58788669e-01
-7.58165643e-02 1.04974985e+00 5.67550838e-01 -2.07728520e-01
4.33299005e-01 4.82950173e-02 3.98547679e-01 7.30912089e-01
-6.47845939e-02 -3.21931243e-01 -5.30992508e-01 -3.06580335e-01
7.10244000e-01 -5.50020710e-02 -2.42633328e-01 -4.03366357e-01
-1.36683488e+00 3.82997483e-01 2.72874117e-01 7.20936805e-02
-5.01320183e-01 1.57839328e-01 1.60635486e-01 -1.32552430e-01
2.53098756e-01 2.05075160e-01 -3.84242356e-01 -2.07688257e-01
-1.70747066e+00 5.09704590e-01 6.01922929e-01 8.92995536e-01
8.15114379e-01 -4.16922897e-01 -2.69658923e-01 5.79630196e-01
2.15792239e-01 -1.82640210e-01 -8.93681198e-02 -1.19544625e+00
1.90176353e-01 4.26344812e-01 6.21900320e-01 -7.48611510e-01
-8.31898004e-02 -1.64472312e-01 -2.29375750e-01 3.34724277e-01
6.40857041e-01 -3.36144939e-02 -1.19149733e+00 1.12840998e+00
7.85943806e-01 4.08577651e-01 -4.98026729e-01 1.14019370e+00
5.50305665e-01 6.38164580e-01 1.68568119e-01 -3.21429193e-01
1.43679583e+00 -1.19771779e+00 -5.42336106e-01 -3.49191368e-01
1.15154609e-01 -9.38659906e-01 8.97464335e-01 5.55637836e-01
-1.37015438e+00 -6.49155200e-01 -7.56589532e-01 6.25578910e-02
-1.76396221e-01 2.68042609e-02 4.20986563e-01 2.67248482e-01
-1.04677474e+00 7.56260991e-01 -1.11016798e+00 -4.48857248e-01
3.88454586e-01 2.64922440e-01 9.72176194e-02 -3.00258584e-02
-7.25641191e-01 3.82240087e-01 4.76030141e-01 1.23923682e-01
-1.02322412e+00 -6.05929792e-01 -5.12440801e-01 -9.67941508e-02
8.08471680e-01 -6.35161102e-01 1.53974140e+00 -8.82731974e-01
-1.29366076e+00 5.02782464e-01 -6.00695193e-01 -3.61893058e-01
8.85359406e-01 -5.33009946e-01 1.36787325e-01 4.84989673e-01
1.90684125e-01 1.21581924e+00 9.33496833e-01 -1.28450048e+00
-1.15631175e+00 -2.80175183e-04 2.96644598e-01 2.77197152e-01
1.73861966e-01 4.66655165e-01 -1.57733285e+00 -5.16494572e-01
5.04628181e-01 -9.10531521e-01 -4.43239748e-01 1.56331182e-01
-4.90648866e-01 -3.47477406e-01 9.85673547e-01 -7.14131713e-01
1.45171714e+00 -2.17518449e+00 6.71563670e-02 2.48150364e-01
2.59499252e-01 1.49820641e-01 1.66622624e-01 2.40243211e-01
1.46601647e-01 2.13416293e-01 -4.11740214e-01 -5.33266604e-01
-2.58345246e-01 -3.21349874e-02 -4.18000259e-02 3.04472357e-01
2.17116460e-01 7.38900661e-01 -1.06141090e+00 -8.84783268e-01
2.61600703e-01 2.13427126e-01 -7.57049561e-01 9.20093358e-02
-4.51083332e-01 4.87768680e-01 -4.21753556e-01 6.17656052e-01
6.76666439e-01 -5.25407553e-01 -6.85926154e-02 -4.07050438e-02
-3.68911088e-01 2.48898700e-01 -1.53598464e+00 1.52998567e+00
2.44802967e-01 7.66025126e-01 6.01622611e-02 -5.50678432e-01
3.60513449e-01 2.81030864e-01 6.49263203e-01 -7.81552196e-02
-1.83742061e-01 -8.55912862e-04 -1.54810071e-01 -4.74454910e-01
6.94709957e-01 2.93469161e-01 3.16190362e-01 4.18157130e-01
-1.93324432e-01 3.56531069e-02 5.39368331e-01 4.93575424e-01
9.44766581e-01 5.04766703e-01 9.20688584e-02 -5.04024550e-02
4.08175647e-01 3.27427059e-01 6.29800320e-01 9.29707885e-01
-3.26685190e-01 9.61653471e-01 5.40928841e-01 -3.70984495e-01
-1.02870095e+00 -9.34533775e-01 6.24292791e-02 1.22975039e+00
5.20899355e-01 -6.15984619e-01 -9.96726453e-01 -7.84587622e-01
-2.07419485e-01 3.71273845e-01 -4.54229355e-01 6.16926014e-01
-7.82999575e-01 -2.21375510e-01 4.23144512e-02 4.87855941e-01
4.94359463e-01 -1.05281019e+00 -1.04517567e+00 1.95303693e-01
-3.36876750e-01 -9.49399829e-01 -7.81022429e-01 -1.41169757e-01
-9.75864768e-01 -1.02029443e+00 -8.37855160e-01 -7.92130709e-01
9.08208728e-01 4.47783351e-01 1.00214517e+00 3.46656531e-01
-4.36121166e-01 6.90768808e-02 -3.87930483e-01 1.66253522e-01
-1.08223356e-01 -1.42426223e-01 -2.65691191e-01 5.13287671e-02
5.08132689e-02 6.94727451e-02 -8.75554919e-01 5.98098695e-01
-7.87119210e-01 2.36005381e-01 1.99912548e-01 3.08715820e-01
8.98025095e-01 1.83617294e-01 1.11333743e-01 -8.63615692e-01
1.96772590e-01 -3.16726863e-01 -8.39333415e-01 8.30496326e-02
-1.05719298e-01 -2.35704049e-01 2.00974569e-01 -3.40453595e-01
-8.49294424e-01 5.17517567e-01 -2.80212350e-02 -5.65837979e-01
-3.17302346e-01 3.18626672e-01 1.01567924e-01 7.59077668e-02
4.46585298e-01 2.67174952e-02 -3.64061683e-01 -6.24228001e-01
4.09669101e-01 2.89663434e-01 6.61609888e-01 -1.30862355e-01
7.63938189e-01 4.36878651e-01 -6.05842471e-01 -7.01563716e-01
-5.65274477e-01 -8.41080964e-01 -9.76096570e-01 -5.40149450e-01
1.05711031e+00 -5.04384160e-01 -4.73557472e-01 2.07932398e-01
-1.27122927e+00 -2.79704064e-01 -5.90846278e-02 3.69358152e-01
-3.60713124e-01 4.64899689e-01 -6.99682474e-01 -9.15982127e-01
1.02025241e-01 -1.37860537e+00 1.07503104e+00 3.75408471e-01
-3.78908038e-01 -5.01566410e-01 -4.14491504e-01 2.56001204e-01
2.49289759e-02 1.77650467e-01 5.34483492e-01 -6.25201404e-01
-1.10306597e+00 -3.47506851e-01 -2.56469399e-01 -7.37872496e-02
3.41986422e-04 3.39485198e-01 -7.71223307e-01 -1.78415388e-01
-4.02185559e-01 1.44748554e-01 8.24840903e-01 4.88635927e-01
1.20116305e+00 -1.10496737e-01 -7.12537646e-01 1.60633340e-01
1.16148520e+00 4.70831990e-01 4.93279338e-01 2.38578573e-01
5.27122676e-01 4.89832610e-01 1.09084070e+00 4.66206789e-01
2.29782224e-01 7.21306801e-01 3.99693191e-01 -2.00585499e-01
8.53895992e-02 -8.15789625e-02 -2.17075665e-02 2.60870844e-01
-1.27061963e-01 -2.87684619e-01 -8.72088909e-01 5.87247491e-01
-1.92658246e+00 -8.27545524e-01 -1.80974320e-01 2.45745683e+00
6.42694831e-01 5.52578688e-01 5.62257886e-01 1.79836407e-01
8.84726346e-01 7.49112293e-02 -6.42112851e-01 5.18811382e-02
4.13097143e-01 -8.06414261e-02 4.20869499e-01 7.70347059e-01
-1.22436070e+00 1.29343116e+00 6.64979315e+00 6.04561031e-01
-8.94161344e-01 -3.87058444e-02 7.07797050e-01 -2.03055590e-01
-4.17834073e-02 3.37914109e-01 -8.98346603e-01 3.52477849e-01
4.64781046e-01 2.08695456e-01 3.96054685e-01 7.52835512e-01
5.78625977e-01 -7.06704855e-01 -1.24283826e+00 7.51845598e-01
-8.46230462e-02 -1.23505759e+00 -2.33387828e-01 -6.79850131e-02
5.21387935e-01 -9.33279544e-02 -2.50104100e-01 -9.65089872e-02
1.58104926e-01 -6.85231686e-01 1.04370701e+00 4.39415514e-01
4.41581964e-01 -6.22230411e-01 2.64588773e-01 4.98814106e-01
-1.34542310e+00 1.45291314e-01 1.06342286e-01 1.99425146e-01
5.75114489e-01 2.75280178e-01 -1.00449574e+00 1.52860686e-01
9.15793240e-01 2.72882432e-01 -6.15007043e-01 1.74331629e+00
-1.05352931e-01 6.74551785e-01 -5.81773937e-01 -4.99846006e-04
3.76386970e-01 -1.25174478e-01 6.47000611e-01 1.47363603e+00
-1.08036783e-03 3.75910223e-01 6.04748130e-01 7.45979130e-01
2.22659975e-01 5.50788902e-02 -8.61703828e-02 1.69194490e-01
5.42810678e-01 1.31304073e+00 -1.57215548e+00 -8.29716086e-01
-3.33998322e-01 1.40458512e+00 1.45305842e-01 6.04439914e-01
-1.04477739e+00 -4.34250206e-01 3.44042629e-01 3.08564365e-01
7.07650661e-01 -5.00866055e-01 -1.58818781e-01 -7.42869198e-01
1.25337347e-01 -7.40761161e-01 4.99849170e-01 -8.91335130e-01
-5.84536433e-01 5.20876288e-01 6.41662702e-02 -1.12368870e+00
-1.33359402e-01 -1.86798528e-01 -7.06310868e-01 7.06286848e-01
-1.14925849e+00 -4.73646939e-01 -3.26422811e-01 2.48548836e-01
1.28116250e+00 6.05424166e-01 1.67482778e-01 3.54112118e-01
-5.10332406e-01 1.80677727e-01 -3.27938795e-01 9.13379192e-02
5.13723493e-01 -1.24794233e+00 8.09887946e-01 1.00038254e+00
2.35750318e-01 6.79017246e-01 8.24936688e-01 -9.72373188e-01
-8.99498820e-01 -8.45802128e-01 8.52279723e-01 -3.15705895e-01
4.75916266e-01 -3.68268251e-01 -1.08611679e+00 5.49810588e-01
1.49895504e-01 1.11427985e-01 3.24131995e-01 -3.41406375e-01
2.20604569e-01 3.81394565e-01 -8.82351100e-01 7.87800193e-01
8.55031073e-01 -1.32889152e-01 -3.23985785e-01 4.99808282e-01
7.76040554e-01 -5.39475679e-01 -4.22176093e-01 1.49080753e-01
5.30984402e-01 -1.02868462e+00 8.05764556e-01 -6.75891101e-01
1.35819748e-01 -8.48243237e-01 1.63568944e-01 -6.95904076e-01
-3.23369026e-01 -1.07987463e+00 9.64286402e-02 9.85892713e-01
5.81655800e-01 5.51648252e-02 9.00137484e-01 8.94716203e-01
-2.79687122e-02 -9.45113778e-01 -6.91289067e-01 -3.87261629e-01
-5.81007481e-01 -6.74606860e-01 1.87848091e-01 5.45240462e-01
-1.08634889e-01 4.30807006e-03 -1.32212013e-01 4.96131867e-01
6.81131363e-01 1.78222910e-01 8.24728131e-01 -9.40960407e-01
-1.97587624e-01 -3.82081658e-01 1.20965326e-02 -1.71969569e+00
-5.32507777e-01 -3.92588764e-01 3.41414005e-01 -1.69104481e+00
9.84138399e-02 -5.39581716e-01 -1.19276211e-01 3.04739237e-01
-4.95410472e-01 5.61635971e-01 3.67573112e-01 3.61841410e-01
-1.09067559e+00 -1.49787202e-01 1.07027006e+00 3.03515941e-01
-6.64904773e-01 3.53454083e-01 -3.06392699e-01 1.09295106e+00
4.70401704e-01 -3.89252186e-01 -1.03869267e-01 -2.37875447e-01
-2.56508291e-02 3.31310123e-01 4.80881125e-01 -9.24949884e-01
4.93780494e-01 -3.61714542e-01 4.13040102e-01 -1.14781487e+00
3.54293674e-01 -6.67029202e-01 6.27861321e-02 3.91863912e-01
-2.86388695e-01 -3.12206466e-02 4.98259366e-02 5.71084499e-01
2.27952544e-02 -4.44447607e-01 8.39999735e-01 -2.95175433e-01
-7.54235029e-01 4.64320123e-01 -5.45333683e-01 -1.01964884e-01
1.15684259e+00 -7.59017587e-01 2.41622820e-01 -2.44073749e-01
-1.23110259e+00 4.97400105e-01 4.94490623e-01 3.19688052e-01
5.63594699e-01 -9.58843589e-01 -3.55016053e-01 3.10539883e-02
-1.92870229e-01 2.88257062e-01 -6.49734284e-04 9.61534441e-01
-6.99519634e-01 2.59731352e-01 4.00126666e-01 -1.04043603e+00
-1.40405750e+00 5.72487354e-01 2.30718568e-01 1.93547055e-01
-8.37422967e-01 1.07124269e+00 3.35393310e-01 3.50529492e-01
4.89305228e-01 -4.06028390e-01 -1.87015519e-01 1.59849077e-01
5.73723972e-01 4.34324861e-01 -6.96228370e-02 -4.95003998e-01
-4.66906428e-01 5.61802804e-01 -2.77674288e-01 -4.49396521e-01
9.84896541e-01 -3.83407980e-01 9.98278484e-02 4.67306137e-01
8.24388325e-01 4.73524928e-02 -1.75831616e+00 -1.71216264e-01
1.84890747e-01 -8.77742231e-01 -2.59160042e-01 -5.76291203e-01
-8.00179362e-01 5.70547998e-01 3.84541601e-01 4.77533102e-01
9.70608413e-01 2.57471532e-01 7.36855268e-01 1.17463022e-01
3.76880705e-01 -1.12670243e+00 2.60337323e-01 4.14997816e-01
5.25238037e-01 -1.06670308e+00 7.63049424e-02 -7.30310202e-01
-8.02329898e-01 9.69803214e-01 6.97433054e-01 -1.92850709e-01
4.44251835e-01 2.23100320e-01 -3.00938692e-02 -1.17823519e-01
-6.55348539e-01 -1.68401018e-01 2.49739543e-01 1.18641257e-01
3.91788065e-01 -3.10448974e-01 -3.58781427e-01 1.43357188e-01
-1.24670163e-01 -8.14999361e-03 3.55153650e-01 1.15002465e+00
-9.84461546e-01 -8.26523006e-01 -4.53894466e-01 3.54402214e-01
-6.34837151e-01 -1.11710116e-01 -2.81438112e-01 6.46663785e-01
1.46257252e-01 9.04166520e-01 3.39431375e-01 -3.66340987e-02
2.26977915e-01 9.93268937e-02 7.92185813e-02 -6.83426142e-01
-5.92605829e-01 7.48587847e-01 4.05025445e-02 -8.66627932e-01
-2.63097107e-01 -9.21179771e-01 -1.61685085e+00 1.40289530e-01
-6.26963139e-01 2.12370664e-01 5.40210307e-01 9.82053220e-01
2.41311938e-01 3.06482106e-01 4.31393921e-01 -1.15741396e+00
1.35337234e-01 -6.52227461e-01 2.40951125e-02 2.00857937e-01
3.58862251e-01 -4.97389227e-01 -1.74656466e-01 6.29629135e-01] | [9.244595527648926, -0.3002196252346039] |
46abb2ea-0692-4cb3-895f-4aad8ad7ff88 | mv-mr-multi-views-and-multi-representations | 2303.12130 | null | https://arxiv.org/abs/2303.12130v1 | https://arxiv.org/pdf/2303.12130v1.pdf | MV-MR: multi-views and multi-representations for self-supervised learning and knowledge distillation | We present a new method of self-supervised learning and knowledge distillation based on the multi-views and multi-representations (MV-MR). The MV-MR is based on the maximization of dependence between learnable embeddings from augmented and non-augmented views, jointly with the maximization of dependence between learnable embeddings from augmented view and multiple non-learnable representations from non-augmented view. We show that the proposed method can be used for efficient self-supervised classification and model-agnostic knowledge distillation. Unlike other self-supervised techniques, our approach does not use any contrastive learning, clustering, or stop gradients. MV-MR is a generic framework allowing the incorporation of constraints on the learnable embeddings via the usage of image multi-representations as regularizers. Along this line, knowledge distillation is considered a particular case of such a regularization. MV-MR provides the state-of-the-art performance on the STL10 and ImageNet-1K datasets among non-contrastive and clustering-free methods. We show that a lower complexity ResNet50 model pretrained using proposed knowledge distillation based on the CLIP ViT model achieves state-of-the-art performance on STL10 linear evaluation. The code is available at: https://github.com/vkinakh/mv-mr | ['Slava Voloshynovskiy', 'Mariia Drozdova', 'Vitaliy Kinakh'] | 2023-03-21 | null | null | null | null | ['unsupervised-image-classification'] | ['computer-vision'] | [-1.83619127e-01 1.50864556e-01 -3.25191945e-01 -3.43067616e-01
-7.98395991e-01 -5.00773430e-01 5.85876048e-01 -1.06380023e-01
-5.64190745e-01 6.26849651e-01 1.71833426e-01 1.05890013e-01
-3.18811834e-01 -6.63809597e-01 -9.28544402e-01 -8.46824110e-01
6.26237988e-02 4.68607575e-01 1.02426611e-01 -1.27357587e-01
-7.39137977e-02 3.77812803e-01 -1.52015710e+00 2.67684728e-01
7.45329499e-01 9.69898045e-01 2.67347157e-01 6.40366435e-01
5.75679541e-02 8.86744142e-01 1.83266290e-02 -2.56114125e-01
4.77838099e-01 -2.20825329e-01 -8.80712092e-01 1.49994642e-01
8.88280034e-01 -1.94339547e-02 -5.09467542e-01 9.86233592e-01
5.48852205e-01 4.11902547e-01 7.87250221e-01 -1.04844427e+00
-1.05602956e+00 4.59033132e-01 -5.39361835e-01 3.54169458e-01
-1.06470466e-01 -2.81702168e-02 1.05752563e+00 -1.25780487e+00
6.66669846e-01 1.03120553e+00 3.86035472e-01 4.71839577e-01
-1.32378161e+00 -5.45617461e-01 2.71038055e-01 6.28275633e-01
-1.59109104e+00 -2.20673785e-01 7.42196083e-01 -5.15158415e-01
8.99723768e-01 1.96198840e-02 5.77629745e-01 9.20478404e-01
-1.63653091e-01 8.79029095e-01 1.29125297e+00 -5.25558054e-01
1.78438604e-01 6.20600939e-01 2.93599606e-01 9.76068676e-01
1.53076470e-01 1.12322256e-01 -4.98553067e-01 -9.24390480e-02
7.80832767e-01 1.21781692e-01 -2.49735624e-01 -7.83724248e-01
-9.85313177e-01 8.26300025e-01 7.07285225e-01 4.17070806e-01
-1.10607110e-01 1.57255977e-01 3.84715259e-01 3.98598492e-01
7.31261969e-01 2.88847446e-01 -6.05062723e-01 2.57772684e-01
-1.08379555e+00 -2.53917843e-01 7.00346828e-01 9.63795006e-01
1.22599971e+00 4.21963751e-01 3.62152122e-02 9.28717375e-01
8.63232911e-02 4.91506010e-01 7.92342603e-01 -8.32396805e-01
2.66653508e-01 6.37140334e-01 -3.24367344e-01 -7.53782511e-01
-2.59810418e-01 -8.75614166e-01 -1.02989721e+00 2.73054689e-01
-8.53729025e-02 -1.98317748e-02 -1.21574867e+00 1.72432923e+00
1.56834006e-01 7.21229076e-01 4.08803761e-01 5.84114373e-01
9.41469193e-01 4.93757606e-01 -1.16738461e-01 -1.62016168e-01
1.09269953e+00 -1.32152724e+00 -5.12767375e-01 -2.24142186e-02
6.25422478e-01 -3.67308706e-01 8.90162051e-01 2.83800095e-01
-9.30974543e-01 -7.13110566e-01 -1.09240162e+00 -1.00991540e-01
-7.90047705e-01 3.34575832e-01 3.05927932e-01 4.43429947e-01
-1.23668659e+00 6.02082253e-01 -8.39203298e-01 -2.83989787e-01
4.96264428e-01 4.24142689e-01 -7.19558477e-01 -1.77627116e-01
-9.00497198e-01 1.16541421e+00 6.89902127e-01 -2.14022510e-02
-1.04689395e+00 -7.66585290e-01 -1.02187443e+00 -1.41110960e-02
5.42780876e-01 -7.29337513e-01 5.43685496e-01 -1.24206054e+00
-1.43689466e+00 1.06845450e+00 6.44431263e-02 -6.81099236e-01
3.84280920e-01 -4.97103810e-01 -3.71373057e-01 3.61694664e-01
9.68317036e-03 6.13593519e-01 1.09278953e+00 -1.50137281e+00
-3.19015682e-01 -3.39164525e-01 4.98628840e-02 5.91620684e-01
-4.64035034e-01 -4.26970810e-01 -5.69198906e-01 -7.03086376e-01
-6.27472401e-02 -1.02269053e+00 -1.57063887e-01 -1.83328524e-01
-3.34231973e-01 4.23974581e-02 9.35725331e-01 -5.72047114e-01
9.47323680e-01 -2.25709510e+00 4.82641518e-01 2.26592720e-01
2.03105211e-01 5.45892715e-01 -3.27028632e-01 1.82546407e-01
-3.70851666e-01 -1.10323727e-02 -4.55846280e-01 -4.83022332e-01
-1.98388085e-01 5.11956036e-01 -5.48562035e-02 6.35915041e-01
1.27920464e-01 8.28268766e-01 -9.34621274e-01 -3.76856208e-01
6.24103904e-01 8.35189044e-01 -4.99816775e-01 1.46139935e-01
-8.92543495e-02 4.01635170e-01 -8.95905346e-02 2.38629386e-01
7.43841767e-01 -4.40446883e-01 1.10911854e-01 -3.84204268e-01
5.93618155e-02 -2.35776544e-01 -1.39283001e+00 1.89224291e+00
-7.02721536e-01 4.67197418e-01 -1.62046641e-01 -1.21936262e+00
6.40361071e-01 2.83890724e-01 3.84717554e-01 -2.72318453e-01
-4.80195833e-03 1.02899343e-01 -3.78793836e-01 -3.65565687e-01
3.34103525e-01 -6.14310242e-03 3.41945767e-01 2.22961485e-01
7.64267325e-01 -1.60040800e-02 1.13326780e-01 4.28426921e-01
7.67200112e-01 2.36223325e-01 4.35500562e-01 -3.84771407e-01
7.30632722e-01 -2.78842747e-01 4.59317356e-01 5.69334149e-01
1.82145774e-01 6.04841590e-01 -2.69875601e-02 -2.31947914e-01
-8.08014691e-01 -1.20804667e+00 -1.25252649e-01 1.13210547e+00
-6.17652200e-03 -3.74424934e-01 -5.67916989e-01 -1.00323904e+00
-2.81278044e-03 6.57198071e-01 -8.96428287e-01 -2.44256392e-01
-3.35683405e-01 -6.45220637e-01 3.62412333e-01 4.63281035e-01
8.26311409e-01 -7.22938895e-01 -1.12100184e-01 -1.02818824e-01
-4.16949093e-02 -1.22313082e+00 -3.56155604e-01 3.49430144e-01
-8.46929550e-01 -1.19947922e+00 -8.54648411e-01 -9.59226072e-01
9.57922220e-01 3.29747587e-01 9.85741436e-01 -2.17938766e-01
-2.11809188e-01 9.58230078e-01 -4.09633636e-01 8.09433833e-02
-9.82968286e-02 1.72000989e-01 2.14280143e-01 3.50923657e-01
1.68262199e-01 -8.36694181e-01 -5.80052912e-01 1.61536321e-01
-1.00944936e+00 -1.58442464e-02 5.68089604e-01 9.00601387e-01
1.01972032e+00 -1.14819318e-01 4.94834632e-01 -1.09367859e+00
1.55100480e-01 -6.13073945e-01 -4.78151500e-01 3.84438902e-01
-7.85884082e-01 3.55558574e-01 7.57587790e-01 -4.98592973e-01
-1.05739045e+00 3.51476520e-01 1.06097152e-02 -1.02108252e+00
-1.72039643e-01 3.15249562e-01 -5.18856235e-02 -4.20442581e-01
7.61991024e-01 4.18757349e-01 -2.97933757e-01 -5.78046322e-01
1.09884107e+00 4.02090997e-01 5.73939025e-01 -3.53794843e-01
8.91327143e-01 7.03610837e-01 -1.87157661e-01 -9.60810542e-01
-1.08400762e+00 -7.83366024e-01 -8.63382936e-01 -1.44782951e-02
8.81298900e-01 -1.29568899e+00 -1.72258794e-01 2.77633280e-01
-6.60995245e-01 -3.10803294e-01 -7.25813508e-01 6.36552036e-01
-7.06099570e-01 5.07077634e-01 -3.45288426e-01 -5.05754471e-01
-3.39602023e-01 -9.54768777e-01 5.37890196e-01 8.89689848e-02
3.34746391e-01 -1.37438834e+00 2.40041122e-01 5.00185966e-01
3.11454445e-01 2.12664202e-01 5.97253799e-01 -1.02022672e+00
-4.82447684e-01 2.16667727e-02 -1.90833107e-01 1.07522595e+00
5.67632839e-02 -3.27564240e-01 -1.18007565e+00 -5.05842090e-01
1.23112220e-02 -6.90083086e-01 1.44951916e+00 3.70814711e-01
1.01560581e+00 -3.80852908e-01 -1.80164222e-02 9.99933660e-01
1.95042241e+00 -3.47504705e-01 6.27634048e-01 2.02818438e-01
1.10546041e+00 2.22700939e-01 1.05532348e-01 1.98286086e-01
6.54532015e-01 5.32842636e-01 4.43171233e-01 -1.04682840e-01
-3.22804540e-01 -9.19490531e-02 3.09156656e-01 1.28063810e+00
-2.95081496e-01 9.36047360e-02 -7.27039218e-01 8.29061568e-01
-1.96756375e+00 -9.13916647e-01 9.06282142e-02 2.21608782e+00
8.72577071e-01 -1.73618123e-01 2.09025536e-02 -9.57143754e-02
6.44543171e-01 3.50419253e-01 -5.89098632e-01 -1.84422806e-01
-3.88897181e-01 3.72365475e-01 7.73596525e-01 7.80691147e-01
-1.32475543e+00 1.10451424e+00 5.70746183e+00 1.04610300e+00
-9.51671302e-01 6.54673815e-01 3.99357051e-01 -1.28813028e-01
-3.61106582e-02 -7.45883435e-02 -8.03273618e-01 1.87963724e-01
7.44568110e-01 2.05556303e-03 3.35580766e-01 1.04558647e+00
-2.23648056e-01 -1.84091270e-01 -1.00018239e+00 1.22235501e+00
4.42191511e-01 -1.40015447e+00 2.02598885e-01 -1.46985531e-01
1.18753040e+00 3.93319845e-01 2.97224611e-01 3.73752207e-01
4.79022592e-01 -7.54414141e-01 3.25309098e-01 5.15241325e-01
8.30206633e-01 -6.66428387e-01 5.53266346e-01 1.36919245e-01
-1.24265981e+00 -9.36289951e-02 -3.77417475e-01 2.41771698e-01
-1.80847704e-01 6.56735539e-01 -3.82973433e-01 8.59214783e-01
5.48665583e-01 1.08080852e+00 -5.63866138e-01 7.74416208e-01
-4.20659274e-01 6.04040504e-01 -2.49944717e-01 5.30801952e-01
2.18222931e-01 -7.94932321e-02 4.81981903e-01 1.45072639e+00
-9.40030962e-02 3.72270751e-03 3.13415587e-01 5.47264099e-01
-1.95918962e-01 1.75668672e-01 -6.39611423e-01 3.39567244e-01
1.07081093e-01 1.28139281e+00 -4.97310698e-01 -6.93828881e-01
-5.69846153e-01 1.17410707e+00 7.55885005e-01 5.71796119e-01
-5.41356802e-01 -2.34399568e-02 3.89041185e-01 -6.83989003e-02
7.22702265e-01 -2.27883041e-01 -5.29526584e-02 -1.64480305e+00
-5.45309298e-02 -6.20315135e-01 6.44778907e-01 -5.95083058e-01
-1.43182445e+00 6.63912237e-01 2.45861545e-01 -1.20727360e+00
-9.76872891e-02 -5.14690220e-01 -3.21338266e-01 5.16272068e-01
-2.10116744e+00 -1.30262673e+00 -3.56746852e-01 1.03927541e+00
4.93203640e-01 -3.57342750e-01 9.11926389e-01 1.98848784e-01
-4.34575260e-01 6.58434987e-01 6.28160477e-01 1.16915248e-01
6.72144175e-01 -1.37817037e+00 -4.32638489e-02 6.65227413e-01
4.08284694e-01 2.25498661e-01 3.59707355e-01 -4.58181083e-01
-1.00182641e+00 -1.37322772e+00 3.49827588e-01 -3.53566825e-01
6.11288071e-01 -2.56418556e-01 -9.78028238e-01 9.93463933e-01
4.24281120e-01 5.14801264e-01 7.78240561e-01 -7.29064792e-02
-7.09010363e-01 -1.56105042e-01 -1.10575521e+00 2.13284090e-01
8.45756412e-01 -7.54706502e-01 -7.23651826e-01 5.57437539e-01
5.64500332e-01 -2.25339726e-01 -1.00145400e+00 3.80264670e-01
1.75010577e-01 -8.82497787e-01 1.06989467e+00 -4.24632847e-01
2.11574778e-01 -1.72784135e-01 -2.46305183e-01 -1.49699283e+00
-5.26641250e-01 -3.51092666e-01 -4.22447890e-01 8.65643144e-01
3.49285692e-01 -6.47288203e-01 7.63105214e-01 1.04458772e-01
-4.78468426e-02 -8.54152322e-01 -8.93459499e-01 -1.13122952e+00
1.17057756e-01 -2.34742209e-01 -1.51588157e-01 1.28730237e+00
-2.57386923e-01 3.79227579e-01 -6.09025896e-01 2.62055218e-01
9.08592761e-01 -1.41309842e-01 5.45709729e-01 -9.54191029e-01
-5.12900233e-01 5.06412704e-03 -6.88250244e-01 -9.08686936e-01
3.94901395e-01 -1.32303214e+00 -3.37853462e-01 -1.62772536e+00
3.24209124e-01 -2.33546272e-01 -6.42145336e-01 6.83936894e-01
2.83324849e-02 3.35121334e-01 4.10909891e-01 2.87815630e-01
-8.11095834e-01 7.50923276e-01 1.10751545e+00 -1.38178855e-01
-2.23590717e-01 -2.84298390e-01 -4.63518322e-01 8.88235390e-01
8.54381382e-01 -5.37922621e-01 -6.79994464e-01 -4.09893930e-01
1.30019011e-02 -3.81502777e-01 4.57886338e-01 -1.11438477e+00
4.08465564e-01 1.74315900e-01 2.64983147e-01 -4.17743921e-01
6.42921805e-01 -7.95995355e-01 1.97400898e-03 3.06729734e-01
-2.75158137e-01 -2.00004086e-01 2.21512511e-01 8.51317585e-01
-3.55986446e-01 -2.60638416e-01 1.00023997e+00 -5.29957652e-01
-9.71625268e-01 3.27854335e-01 -5.09691499e-02 3.37889850e-01
1.03692520e+00 -2.28950888e-01 -2.26268291e-01 -2.23269373e-01
-1.38574445e+00 2.05686435e-01 2.49154136e-01 2.55991548e-01
7.30275333e-01 -1.37006748e+00 -7.05467343e-01 1.56052291e-01
1.17983006e-01 -1.47459367e-02 4.95208740e-01 7.83482075e-01
-3.11516881e-01 2.04905286e-01 -2.74324477e-01 -5.39027870e-01
-1.17429125e+00 6.40173852e-01 4.66730982e-01 -5.85167348e-01
-6.55468345e-01 9.02236402e-01 4.13304597e-01 -6.06293619e-01
3.64264816e-01 -1.00524947e-01 -2.16388151e-01 -2.72291456e-03
3.67297143e-01 4.65010613e-01 -1.26019856e-02 -7.34297574e-01
-3.82122695e-01 8.25913131e-01 -3.49657208e-01 -8.26165229e-02
1.46592331e+00 -1.36632770e-01 8.68582278e-02 5.87461591e-01
1.55966115e+00 -3.28170002e-01 -1.34203374e+00 -6.57506943e-01
-2.64931291e-01 -1.92202032e-01 3.16215694e-01 -7.05619931e-01
-1.39186084e+00 8.83389652e-01 8.29078138e-01 -2.54600376e-01
1.12861252e+00 6.86520711e-02 3.20123434e-01 5.68363130e-01
1.36794180e-01 -1.23958969e+00 4.21721637e-01 5.81340909e-01
9.72264409e-01 -1.36076069e+00 1.90850019e-01 -3.52454543e-01
-8.28146935e-01 8.93649340e-01 6.61735117e-01 -5.42763174e-01
1.13992381e+00 -2.93226168e-02 -4.60214093e-02 -2.07313016e-01
-6.96527421e-01 -5.98859251e-01 5.16477227e-01 6.36322558e-01
3.31031010e-02 4.62786295e-02 9.34522301e-02 2.62385666e-01
1.37307674e-01 -2.49262135e-02 3.95366281e-01 8.60618889e-01
-3.16494852e-01 -8.27403307e-01 -8.42630416e-02 4.63749647e-01
-3.15992534e-01 -2.76554912e-01 -1.55979574e-01 7.62657166e-01
3.35949898e-01 5.76374531e-01 -1.13435067e-01 -5.39912045e-01
1.93111613e-01 9.63149518e-02 6.00098848e-01 -8.37234616e-01
-5.12171268e-01 -8.29381645e-02 -1.15775652e-01 -4.01512355e-01
-8.45409572e-01 -2.75441080e-01 -9.77607191e-01 -7.22436234e-02
-5.50868511e-01 9.56872851e-02 5.44253409e-01 9.09523666e-01
5.07683575e-01 3.78107816e-01 5.81582308e-01 -8.76144409e-01
-4.01695520e-01 -7.48289287e-01 -7.76510060e-01 4.81847137e-01
3.41996253e-01 -6.92634940e-01 -5.56622803e-01 2.91480422e-01] | [9.423757553100586, 2.6362264156341553] |
9cd5a928-51c5-476a-96f6-0d3c2834f214 | facenet-a-unified-embedding-for-face | 1503.03832 | null | http://arxiv.org/abs/1503.03832v3 | http://arxiv.org/pdf/1503.03832v3.pdf | FaceNet: A Unified Embedding for Face Recognition and Clustering | Despite significant recent advances in the field of face recognition,
implementing face verification and recognition efficiently at scale presents
serious challenges to current approaches. In this paper we present a system,
called FaceNet, that directly learns a mapping from face images to a compact
Euclidean space where distances directly correspond to a measure of face
similarity. Once this space has been produced, tasks such as face recognition,
verification and clustering can be easily implemented using standard techniques
with FaceNet embeddings as feature vectors.
Our method uses a deep convolutional network trained to directly optimize the
embedding itself, rather than an intermediate bottleneck layer as in previous
deep learning approaches. To train, we use triplets of roughly aligned matching
/ non-matching face patches generated using a novel online triplet mining
method. The benefit of our approach is much greater representational
efficiency: we achieve state-of-the-art face recognition performance using only
128-bytes per face.
On the widely used Labeled Faces in the Wild (LFW) dataset, our system
achieves a new record accuracy of 99.63%. On YouTube Faces DB it achieves
95.12%. Our system cuts the error rate in comparison to the best published
result by 30% on both datasets.
We also introduce the concept of harmonic embeddings, and a harmonic triplet
loss, which describe different versions of face embeddings (produced by
different networks) that are compatible to each other and allow for direct
comparison between each other. | ['Florian Schroff', 'Dmitry Kalenichenko', 'James Philbin'] | 2015-03-12 | facenet-a-unified-embedding-for-face-1 | http://openaccess.thecvf.com/content_cvpr_2015/html/Schroff_FaceNet_A_Unified_2015_CVPR_paper.html | http://openaccess.thecvf.com/content_cvpr_2015/papers/Schroff_FaceNet_A_Unified_2015_CVPR_paper.pdf | cvpr-2015-6 | ['disguised-face-verification'] | ['computer-vision'] | [ 8.59126542e-03 -2.10821643e-01 -1.31170571e-01 -8.95463228e-01
-6.75659597e-01 -4.43570793e-01 5.89034736e-01 -2.45140269e-01
-3.92405868e-01 2.69964188e-01 -9.85842198e-03 -2.21639901e-01
-1.17354795e-01 -7.96206057e-01 -7.89959788e-01 -4.85669792e-01
-4.63395029e-01 5.44867277e-01 -3.08578819e-01 -4.44513783e-02
1.47705495e-01 1.00809097e+00 -2.03344941e+00 4.44274753e-01
-1.81520525e-02 1.44628525e+00 -5.16856551e-01 6.54256880e-01
-6.89316317e-02 3.54011714e-01 -5.30905843e-01 -8.63870561e-01
9.41854894e-01 -1.69886369e-02 -7.83697546e-01 6.01570085e-02
1.22304988e+00 -6.25704944e-01 -5.88464499e-01 7.40832388e-01
7.81824768e-01 -1.85858503e-01 5.25304019e-01 -1.58806360e+00
-9.49011385e-01 1.78280905e-01 -5.37093759e-01 -3.42665911e-02
4.20390934e-01 6.01577237e-02 8.04896712e-01 -1.22096920e+00
5.21705210e-01 1.50651217e+00 9.89190102e-01 8.69251430e-01
-1.33038056e+00 -1.03962481e+00 -5.26924431e-01 3.67172301e-01
-1.80007350e+00 -1.13242280e+00 3.04540038e-01 -2.94311523e-01
1.15072429e+00 6.81998953e-02 3.98157269e-01 7.86683440e-01
-5.65050021e-02 2.19613492e-01 7.84379959e-01 -4.70793068e-01
-2.93647405e-02 6.60385713e-02 -1.97139874e-01 9.45171714e-01
8.46894458e-02 2.18300954e-01 -7.03927994e-01 -3.88417691e-01
6.26031637e-01 2.07307383e-01 -1.64808884e-01 -3.68594259e-01
-8.01298261e-01 8.77080798e-01 2.97311932e-01 8.71905535e-02
-9.31420773e-02 3.28769237e-01 4.79047000e-01 6.37761772e-01
3.04793656e-01 4.68275435e-02 -3.44813228e-01 1.95258744e-02
-1.08576834e+00 1.15242243e-01 1.00177753e+00 8.01214635e-01
9.45656776e-01 -8.79582390e-02 -4.40709703e-02 8.44814599e-01
4.19723839e-01 5.07380605e-01 3.51585418e-01 -1.08386898e+00
1.73695847e-01 4.15746838e-01 -2.79455543e-01 -1.12584627e+00
4.28131633e-02 2.13735715e-01 -6.84831738e-01 4.44860637e-01
3.54690194e-01 3.02997399e-02 -8.49210024e-01 1.72962296e+00
4.78674352e-01 4.30560708e-01 -1.45492211e-01 5.25507212e-01
6.45656884e-01 4.26654577e-01 -3.15846235e-01 1.24353841e-01
1.35020518e+00 -6.56907678e-01 -3.58144790e-01 2.38036573e-01
4.82976556e-01 -9.60972965e-01 7.23760903e-01 1.51462123e-01
-1.01042497e+00 -5.70998967e-01 -1.15277719e+00 -1.73428908e-01
-6.11226499e-01 -2.29684375e-02 6.14576340e-01 1.28785360e+00
-1.86710024e+00 7.30805993e-01 -5.60050726e-01 -4.86233562e-01
1.00925541e+00 1.07528949e+00 -9.47657824e-01 -3.00784737e-01
-5.99328995e-01 7.20624864e-01 -1.76658221e-02 -1.28424570e-01
-7.85072863e-01 -9.17798460e-01 -1.00661826e+00 1.72374874e-01
-6.93633184e-02 -1.17734127e-01 1.04905891e+00 -1.07487106e+00
-1.43803000e+00 1.13774550e+00 -3.73283803e-01 -4.21102226e-01
7.39080012e-02 1.80851314e-02 -6.83078349e-01 2.57941812e-01
-1.39200211e-01 1.01703274e+00 1.13516724e+00 -8.63043368e-01
-2.96417564e-01 -5.71901798e-01 -1.70616075e-01 -4.63430315e-01
-8.54885817e-01 4.52898115e-01 -5.06991625e-01 -3.77475232e-01
-3.26453239e-01 -7.87586153e-01 3.67756784e-01 7.66245961e-01
2.98406370e-03 -4.60750639e-01 1.15670180e+00 -4.52855766e-01
8.65175903e-01 -2.20442533e+00 -4.06675369e-01 4.57516462e-01
1.97184652e-01 6.03385031e-01 -7.21325278e-01 3.70305598e-01
-4.15597647e-01 1.59664184e-01 -2.07038552e-01 -6.92447364e-01
2.11443067e-01 1.64897442e-01 -2.17735425e-01 8.15566242e-01
6.00162208e-01 7.78014839e-01 -4.38911766e-01 -4.03026015e-01
-5.52060530e-02 9.53334451e-01 -8.49910021e-01 1.81456134e-01
1.87336534e-01 -2.93806940e-01 2.95517057e-01 9.15878296e-01
1.15626919e+00 -9.54518095e-03 3.73471111e-01 -3.20027530e-01
1.63641289e-01 9.25384983e-02 -1.21630812e+00 1.49355948e+00
-4.77297187e-01 8.08076560e-01 1.54881105e-01 -1.10269558e+00
1.09231400e+00 3.45363885e-01 5.38863838e-01 -6.07022166e-01
3.99935953e-02 2.12233663e-01 -5.55572957e-02 -1.92254320e-01
1.43093854e-01 1.93595812e-01 5.05079210e-01 8.89817953e-01
5.61025500e-01 4.01246339e-01 1.53474003e-01 -1.93494838e-02
1.17357779e+00 -1.25239015e-01 -3.43614779e-02 -3.21569830e-01
5.25180101e-01 -5.13367891e-01 2.50858456e-01 3.78889233e-01
-1.56209275e-01 6.71201706e-01 3.47690970e-01 -8.51051390e-01
-1.07089138e+00 -1.00749528e+00 -4.24291044e-01 1.09369469e+00
-4.23405617e-01 -8.32975328e-01 -7.91077375e-01 -7.74170399e-01
5.32000542e-01 -1.95584521e-01 -7.87533402e-01 2.15236340e-02
-6.39430165e-01 -5.49529195e-01 9.79508281e-01 5.64649045e-01
4.68695134e-01 -9.24562693e-01 -3.12726647e-01 -1.16257332e-01
3.99291307e-01 -9.70712125e-01 -7.08724737e-01 -2.20835969e-01
-6.16809130e-01 -1.36709976e+00 -5.43794215e-01 -9.65122521e-01
7.57767320e-01 2.12328240e-01 1.30030417e+00 6.27057493e-01
-9.16130364e-01 5.94561100e-01 -4.79679145e-02 -2.15968452e-02
-1.41853511e-01 -2.64910161e-01 4.16544646e-01 3.81665349e-01
9.38907146e-01 -4.71089333e-01 -6.20339274e-01 3.12345803e-01
-8.39757800e-01 -7.99716353e-01 2.52561331e-01 1.02486682e+00
3.61466885e-01 -2.94195324e-01 3.72113824e-01 -4.65884030e-01
3.67371500e-01 -4.05737698e-01 -5.84663689e-01 3.92116696e-01
-6.54681981e-01 5.29008918e-02 4.64250922e-01 -3.70566398e-01
-4.33391452e-01 3.13801408e-01 -2.16101035e-01 -6.88263953e-01
-1.18026495e-01 -1.56653062e-01 -8.72135758e-02 -6.79636300e-01
5.63076556e-01 1.30001679e-01 5.98995924e-01 -4.28043783e-01
3.26498479e-01 9.41883802e-01 4.47786748e-01 -6.71134531e-01
7.89780080e-01 5.79588890e-01 -2.30388045e-02 -7.17926562e-01
-2.30977401e-01 -2.37423539e-01 -6.67588234e-01 8.58118013e-02
3.98305088e-01 -8.25713873e-01 -1.35657215e+00 2.63421655e-01
-1.24307120e+00 -1.13423333e-01 -3.08299959e-01 1.81392878e-01
-3.51900101e-01 2.82745242e-01 -6.19209409e-01 -7.27083266e-01
-4.94953811e-01 -1.03775644e+00 1.26535738e+00 -5.88362515e-02
-1.14932852e-02 -7.11355925e-01 -3.06088757e-02 6.09194860e-02
6.20184779e-01 -1.60143264e-02 6.20737433e-01 -6.19121075e-01
-4.97403949e-01 -1.27021849e-01 -5.28981090e-01 5.69480062e-01
2.96398014e-01 3.23902130e-01 -1.27826858e+00 -6.73484921e-01
-2.67816514e-01 -6.57090664e-01 7.68430650e-01 -9.24199745e-02
1.45700765e+00 -5.02299011e-01 -2.24249318e-01 9.48823750e-01
1.47382724e+00 -2.64661238e-02 8.76874924e-01 2.53028776e-02
3.30760092e-01 6.44270957e-01 -4.46198620e-02 5.84211171e-01
2.30080426e-01 7.71575987e-01 2.24941283e-01 1.07118927e-01
-3.44119668e-01 -8.55538175e-02 5.28670847e-01 4.20060396e-01
2.12150007e-01 1.10451356e-01 -8.18181455e-01 4.79543656e-01
-1.36628091e+00 -1.22581875e+00 5.30105591e-01 2.30500722e+00
7.64858067e-01 -4.94490176e-01 1.31821185e-01 2.44992167e-01
6.68870866e-01 -1.14330519e-02 -3.72034907e-01 -6.84418738e-01
5.87163668e-04 1.10923016e+00 4.64655638e-01 4.62911367e-01
-1.08742750e+00 6.84689879e-01 7.06785059e+00 7.63242364e-01
-1.31928289e+00 2.18207598e-01 7.73971617e-01 -2.90925264e-01
1.10833049e-01 -3.55902255e-01 -1.03020966e+00 3.94230962e-01
1.16159332e+00 -7.97879174e-02 8.01462531e-01 8.21859121e-01
-4.52199161e-01 3.58573526e-01 -1.60458851e+00 1.62792373e+00
6.74499214e-01 -1.51122081e+00 2.57429957e-01 3.48708510e-01
6.12778544e-01 -3.66887189e-02 4.28127229e-01 1.88101918e-01
1.01540230e-01 -1.69650793e+00 2.75652647e-01 2.39466980e-01
1.49014640e+00 -9.40153956e-01 6.30305648e-01 -3.47214431e-01
-1.38977706e+00 -1.39109135e-01 -6.87672615e-01 9.46697071e-02
-3.77023280e-01 3.81138265e-01 -9.60132003e-01 1.56254813e-01
9.37625527e-01 6.51460946e-01 -6.84684217e-01 7.31206000e-01
3.63570929e-01 2.34275624e-01 -6.41815007e-01 3.28016609e-01
-7.78054297e-02 9.27483439e-02 -1.79674208e-01 1.18045294e+00
5.78807533e-01 -4.53544147e-02 -1.15074873e-01 6.43970490e-01
-7.35255122e-01 9.57420617e-02 -8.94813836e-01 -5.03819771e-02
7.32004523e-01 1.29584730e+00 -3.84087086e-01 -3.40996325e-01
-5.55194020e-01 1.00443697e+00 4.83606577e-01 -1.48141803e-02
-6.51602387e-01 -5.95935106e-01 1.26433289e+00 1.13274269e-01
6.18156850e-01 -1.79360941e-01 1.56077325e-01 -8.79059136e-01
2.20478445e-01 -1.06875157e+00 3.24186981e-01 -5.11145070e-02
-1.54836953e+00 8.33525300e-01 -3.35746169e-01 -8.71233165e-01
-3.68539929e-01 -1.04904711e+00 -4.83761847e-01 8.29986811e-01
-1.45030332e+00 -9.56126869e-01 -2.00855762e-01 1.00540841e+00
1.42015651e-01 -4.94776666e-01 1.21329880e+00 8.32345784e-01
-3.99659395e-01 1.39978755e+00 1.22285724e-01 5.15379488e-01
9.98639405e-01 -7.91926682e-01 7.06537485e-01 4.67814386e-01
5.56445301e-01 8.72140884e-01 -2.35393550e-02 -1.13601953e-01
-1.85451889e+00 -1.17811394e+00 9.76148665e-01 -4.72631216e-01
2.56134689e-01 -5.95464349e-01 -7.05987632e-01 6.26773596e-01
1.37866437e-01 7.19353914e-01 1.06119263e+00 1.46257192e-01
-1.09901059e+00 -6.10100627e-01 -1.65729129e+00 3.22233707e-01
1.33840966e+00 -9.77968037e-01 -4.20893952e-02 4.39265937e-01
3.10614347e-01 -3.50242667e-02 -9.92398143e-01 3.49293470e-01
9.83929038e-01 -1.11084259e+00 1.14217305e+00 -7.77099371e-01
1.39808476e-01 -1.81268826e-01 -4.83177513e-01 -8.51567209e-01
-2.34621286e-01 -6.61480129e-01 -3.00932735e-01 1.13684058e+00
1.32969603e-01 -7.57992089e-01 9.67423141e-01 5.83950877e-01
3.77771735e-01 -9.17491078e-01 -1.22106886e+00 -9.60454702e-01
-5.84485419e-02 -2.08023548e-01 1.18213058e+00 9.86198306e-01
-2.09041670e-01 -1.60867006e-01 -2.26174444e-01 8.29997659e-02
7.88080335e-01 5.16516306e-02 8.08668315e-01 -1.27277231e+00
-6.60269260e-02 -3.45038563e-01 -9.99124527e-01 -8.28684390e-01
5.35062432e-01 -1.12536454e+00 -2.84340113e-01 -8.44858110e-01
1.74557507e-01 -4.97687280e-01 -1.98583171e-01 8.16902280e-01
5.55014312e-01 1.07784510e+00 1.57214209e-01 -7.52135962e-02
-3.56759518e-01 5.70746660e-01 5.16811073e-01 -2.96020329e-01
2.63630807e-01 -6.07936203e-01 -5.78207016e-01 3.89497876e-01
7.35050023e-01 -3.53049755e-01 -1.04079075e-01 -7.41154134e-01
-3.72100294e-01 -3.74206901e-01 3.23478520e-01 -1.16586840e+00
4.19168770e-01 2.99988925e-01 7.92626798e-01 -1.53073058e-01
6.05253279e-01 -8.62076938e-01 1.10678956e-01 2.93579757e-01
-8.70066807e-02 4.79327828e-01 3.35995406e-01 1.15213528e-01
-1.59486979e-01 -2.49910611e-03 8.94580245e-01 3.03319514e-01
-5.49599528e-01 7.90294051e-01 1.52485088e-01 -1.38927013e-01
1.10175657e+00 -3.46099108e-01 -3.04918051e-01 -1.91887379e-01
-4.17633325e-01 -1.34902343e-01 5.47377229e-01 5.57346404e-01
8.92990530e-01 -1.73087990e+00 -8.41911316e-01 8.50672364e-01
2.95236744e-02 -6.97337747e-01 -5.23585752e-02 4.76524353e-01
-4.85509843e-01 3.15743834e-01 -4.63411748e-01 -5.26837349e-01
-1.66812336e+00 4.51976240e-01 4.05166686e-01 2.77268350e-01
-2.83038884e-01 9.71057951e-01 -2.94082791e-01 -5.43666780e-01
3.28561038e-01 3.20910335e-01 2.48325005e-01 -3.81402569e-05
1.20564413e+00 2.67263442e-01 4.49243993e-01 -9.14946973e-01
-6.78397298e-01 6.32091224e-01 -8.24302733e-02 4.33879485e-03
1.36948431e+00 4.16305989e-01 -5.03740370e-01 -2.42567837e-01
1.95038223e+00 -1.19017266e-01 -1.07178795e+00 -1.29491568e-01
-1.07420489e-01 -1.00971389e+00 -2.75463313e-01 -4.19057012e-01
-1.47535884e+00 9.29494619e-01 1.17603087e+00 -9.14218202e-02
1.00882554e+00 -4.06287750e-03 8.20603788e-01 6.68917775e-01
4.21568781e-01 -8.33571851e-01 2.02191472e-01 5.36486864e-01
7.93195128e-01 -1.28138483e+00 -7.98151195e-02 -1.95651799e-01
1.50613576e-01 1.39065957e+00 5.85603595e-01 -1.99843258e-01
1.04368246e+00 6.20521009e-01 -2.15188204e-03 -1.51102200e-01
-8.21304142e-01 -6.38421923e-02 1.26416922e-01 8.65182340e-01
6.05266392e-01 -1.90052465e-01 1.67439550e-01 1.08674221e-01
-2.26281583e-01 9.31631848e-02 -6.79782499e-03 8.97625446e-01
-1.91270232e-01 -1.48289835e+00 -4.94950354e-01 5.74928403e-01
-5.49312949e-01 -1.59083292e-01 -2.58732915e-01 4.87629324e-01
3.02392870e-01 9.54019010e-01 6.13374650e-01 -5.67877173e-01
7.70135224e-02 2.71542281e-01 8.09286118e-01 -5.66922665e-01
-5.34402847e-01 -7.17005849e-01 -2.90013283e-01 -1.02038443e+00
-3.63666862e-01 -4.52884465e-01 -9.33246434e-01 -9.88487363e-01
-8.98522884e-02 2.27305852e-02 8.44115317e-01 4.73969251e-01
8.42755198e-01 -2.76676387e-01 9.46546555e-01 -1.08626056e+00
-7.17337966e-01 -6.41195893e-01 -5.26930153e-01 4.51556206e-01
3.29783052e-01 -6.37190402e-01 -3.15166414e-01 1.64108485e-01] | [13.287086486816406, 0.7962698340415955] |
6cb35d66-2ecd-4b3b-a472-6de7864475b3 | sgd-qa-fast-schema-guided-dialogue-state | 2105.08049 | null | https://arxiv.org/abs/2105.08049v1 | https://arxiv.org/pdf/2105.08049v1.pdf | SGD-QA: Fast Schema-Guided Dialogue State Tracking for Unseen Services | Dialogue state tracking is an essential part of goal-oriented dialogue systems, while most of these state tracking models often fail to handle unseen services. In this paper, we propose SGD-QA, a simple and extensible model for schema-guided dialogue state tracking based on a question answering approach. The proposed multi-pass model shares a single encoder between the domain information and dialogue utterance. The domain's description represents the query and the dialogue utterance serves as the context. The model improves performance on unseen services by at least 1.6x compared to single-pass baseline models on the SGD dataset. SGD-QA shows competitive performance compared to state-of-the-art multi-pass models while being significantly more efficient in terms of memory consumption and training performance. We provide a thorough discussion on the model with ablation study and error analysis. | ['Boris Ginsburg', 'Evelina Bakhturina', 'Vahid Noroozi', 'Yang Zhang'] | 2021-05-17 | null | null | null | null | ['goal-oriented-dialogue-systems'] | ['natural-language-processing'] | [ 5.10560162e-02 6.41137898e-01 -1.15850613e-01 -5.22902966e-01
-1.04364014e+00 -8.36393356e-01 1.13321006e+00 2.33443573e-01
-4.35704291e-01 6.53288841e-01 6.60604835e-01 -4.51014072e-01
3.32992673e-01 -3.09884429e-01 -1.01269372e-02 -1.85945079e-01
1.19833477e-01 1.15831971e+00 8.08085144e-01 -1.13722539e+00
2.23763034e-01 -1.48953229e-01 -9.95184124e-01 4.25615013e-01
8.27974856e-01 7.31971323e-01 3.34929004e-02 9.04979825e-01
-5.76856434e-01 9.82067764e-01 -8.82036328e-01 -3.24274451e-01
-1.82331771e-01 -7.36646295e-01 -1.72518039e+00 1.09787405e-01
3.93811464e-01 -7.52164662e-01 -7.39951313e-01 7.74334967e-01
6.54570401e-01 2.88082808e-01 1.09971784e-01 -1.17954421e+00
-2.66071290e-01 4.23151910e-01 2.54708737e-01 8.33324492e-02
1.07012343e+00 2.87627429e-01 1.04087114e+00 -4.19607610e-01
6.87830150e-01 1.41325366e+00 1.68941364e-01 1.52905822e+00
-1.13065815e+00 4.56435606e-02 2.75530040e-01 -1.54082691e-02
-6.29978597e-01 -9.65275109e-01 5.51674247e-01 4.35710587e-02
1.35514081e+00 3.96828920e-01 3.19625229e-01 1.23680031e+00
-2.59639114e-01 1.27398837e+00 9.20365274e-01 -5.65695286e-01
3.05593461e-01 2.75818735e-01 6.75530970e-01 9.44303572e-01
-4.22133178e-01 -3.52890551e-01 -8.13376784e-01 -4.97919053e-01
4.23477471e-01 -6.40591383e-01 -2.34790623e-01 -3.86640579e-01
-9.59885240e-01 9.05557156e-01 -2.32531801e-01 8.05125833e-02
-7.30067939e-02 -3.09068084e-01 8.21944594e-01 6.32009745e-01
2.67552197e-01 5.80952644e-01 -5.22413611e-01 -8.88197303e-01
-4.87878561e-01 5.81730962e-01 1.59838974e+00 1.13582456e+00
2.07859740e-01 -7.50813410e-02 -6.37773991e-01 9.87344861e-01
3.53273541e-01 3.34529966e-01 3.74819785e-01 -1.12134933e+00
6.62367344e-01 9.79347408e-01 5.07509947e-01 -1.64981782e-01
-6.56992257e-01 5.56817986e-02 -3.76236677e-01 -2.52748549e-01
7.46068954e-01 -3.52247298e-01 -8.23892176e-01 1.78229725e+00
5.03936172e-01 -2.02328384e-01 6.75498366e-01 8.43415618e-01
1.20236290e+00 8.02891970e-01 1.71952456e-01 -2.85920501e-01
1.71258247e+00 -1.29348636e+00 -1.04951286e+00 -5.49747407e-01
8.92781198e-01 -3.68354350e-01 1.18819523e+00 1.26429200e-01
-1.39447474e+00 -1.19304568e-01 -7.91973412e-01 -3.43819767e-01
-1.27613500e-01 -1.56933144e-01 5.38550436e-01 7.47813165e-01
-1.15959048e+00 8.72634128e-02 -7.63172984e-01 -8.87986541e-01
-3.15131217e-01 1.47327557e-01 -8.52116197e-02 1.70077294e-01
-1.55204844e+00 1.09586680e+00 3.93061459e-01 -2.91618168e-01
-1.12841868e+00 -2.08262563e-01 -1.14550698e+00 3.20524089e-02
6.46224022e-01 -4.96749014e-01 2.24334121e+00 -9.22648013e-02
-2.16804338e+00 8.60158741e-01 -5.07479072e-01 -8.01922321e-01
4.48754847e-01 -3.54445577e-01 -4.99254853e-01 3.33949298e-01
-1.47560433e-01 3.59228432e-01 4.41104323e-01 -9.13711369e-01
-7.33572066e-01 -3.15872490e-01 6.31942749e-01 7.76309311e-01
-4.75525372e-02 1.64152980e-01 -8.68938744e-01 2.71329731e-01
1.58975378e-01 -8.45801950e-01 -1.71961471e-01 -4.81341481e-01
-4.82475281e-01 -6.13192320e-01 9.22819972e-01 -6.59331203e-01
1.55101347e+00 -1.82049215e+00 -1.53401736e-02 -3.47534806e-01
-2.82842107e-02 6.27035499e-01 -7.01518729e-02 9.56957221e-01
4.92768556e-01 -3.26658785e-01 -1.91458300e-01 -5.00849307e-01
4.88030612e-01 2.51649320e-01 -2.92700768e-01 2.22507529e-02
2.49838069e-01 8.53893220e-01 -1.15619826e+00 -4.16429818e-01
2.33422160e-01 -6.22912981e-02 -3.96131366e-01 7.62283087e-01
-8.45005691e-01 4.12271172e-01 -5.26573360e-01 4.01814073e-01
2.78629631e-01 -3.59696150e-01 5.11381745e-01 1.21234819e-01
7.58036897e-02 1.18284714e+00 -7.55313754e-01 2.31251836e+00
-2.25466982e-01 2.09468096e-01 3.04451078e-01 -4.34547395e-01
9.70093548e-01 7.11003423e-01 1.14944614e-01 -9.26384568e-01
-1.25710577e-01 7.69140944e-02 -7.74935558e-02 -5.33109605e-01
9.54635024e-01 -1.28861040e-01 -6.84273660e-01 6.40716016e-01
4.42412972e-01 -6.93522617e-02 2.74202377e-01 7.48144209e-01
1.22008085e+00 -1.22120371e-02 4.60855097e-01 -4.10014689e-02
9.29581940e-01 3.10821444e-01 2.25516051e-01 8.99666429e-01
-6.53322697e-01 5.49515039e-02 8.38147402e-01 -1.88773498e-01
-6.85461700e-01 -7.39607990e-01 1.29205182e-01 1.49809515e+00
3.20793688e-01 -6.56627417e-01 -1.01153338e+00 -1.14210689e+00
-2.54146874e-01 1.23030329e+00 -3.47802788e-01 -3.05216223e-01
-5.86348832e-01 -4.32615161e-01 9.29154575e-01 3.15610409e-01
9.64881659e-01 -9.75853562e-01 -3.81115317e-01 4.48711187e-01
-4.56047684e-01 -1.21405876e+00 -3.86010051e-01 4.65129279e-02
-7.00302362e-01 -8.60042512e-01 -4.98320401e-01 -7.38406360e-01
8.64878446e-02 1.35666430e-01 1.32438946e+00 -4.09765802e-02
4.07185316e-01 5.96288860e-01 -4.14355487e-01 -1.91016451e-01
-1.18673563e+00 5.44034600e-01 -1.59510583e-01 -2.55595684e-01
6.99122727e-01 1.01986460e-01 -5.36745906e-01 3.98948103e-01
-4.72386330e-01 7.42507279e-02 7.05795176e-03 1.05120623e+00
-1.13628976e-01 -6.94899797e-01 8.46360028e-01 -1.25014067e+00
1.15553999e+00 -3.49943489e-01 -4.19409961e-01 4.57018614e-01
-6.75661087e-01 3.14090431e-01 3.34023356e-01 2.55764604e-01
-1.67968786e+00 -6.33020997e-02 -5.37746847e-01 3.30538332e-01
-4.99650061e-01 1.82516620e-01 -2.01332927e-01 2.73565918e-01
7.67580867e-01 5.44538200e-01 2.84074843e-01 -7.93451846e-01
4.41222966e-01 9.49891806e-01 6.44510567e-01 -6.62640572e-01
3.96819144e-01 2.18027860e-01 -5.11418045e-01 -7.94287324e-01
-9.46782589e-01 -1.03115070e+00 -4.98134226e-01 -9.32512805e-02
8.02367747e-01 -7.02617168e-01 -9.73637700e-01 5.07205188e-01
-1.07092237e+00 -6.72200263e-01 -1.74188182e-01 -8.74001533e-02
-8.52002919e-01 6.52028918e-01 -9.05665338e-01 -1.24984097e+00
-6.91331387e-01 -8.05451334e-01 1.10942423e+00 4.27498907e-01
-3.20653945e-01 -1.14803290e+00 2.33354464e-01 6.42838955e-01
3.93900067e-01 -2.45746344e-01 8.51732433e-01 -1.52499342e+00
-2.06132844e-01 -1.46892905e-01 7.33208284e-02 4.06187698e-02
1.34862676e-01 -7.18274057e-01 -9.58373249e-01 -4.68718737e-01
9.39490423e-02 -8.49246919e-01 4.75766838e-01 -2.46356860e-01
2.99821079e-01 -4.09859270e-01 4.77679161e-04 -1.23464145e-01
1.01967406e+00 5.64318180e-01 4.59271014e-01 3.35112065e-01
5.55188097e-02 6.73709929e-01 9.70411122e-01 4.29890782e-01
7.94610739e-01 7.28444874e-01 3.13503712e-01 1.41440362e-01
7.38385972e-03 -4.02053535e-01 4.76002932e-01 7.65170395e-01
4.99681920e-01 -4.15439039e-01 -1.07042873e+00 6.97093308e-01
-2.09436750e+00 -8.94690573e-01 -1.06943853e-01 2.07181215e+00
1.04538584e+00 2.77432829e-01 5.31747341e-01 -1.11655697e-01
4.61249530e-01 5.22866845e-01 -6.86619580e-01 -7.83740759e-01
6.91029206e-02 -1.41525954e-01 -3.76004428e-02 8.30379009e-01
-1.02747226e+00 1.34079611e+00 6.80875111e+00 1.91457272e-01
-5.63424706e-01 1.86277255e-01 4.01985236e-02 1.21911742e-01
5.54571711e-02 -6.35510162e-02 -1.16010499e+00 1.41070291e-01
1.31745231e+00 -2.82060295e-01 2.04632029e-01 8.38852644e-01
-1.99073888e-02 -2.31601581e-01 -1.23868656e+00 4.12764966e-01
1.31474137e-01 -1.14055979e+00 -1.04404941e-01 -2.93601990e-01
3.25840712e-01 -6.24692626e-02 -3.74174088e-01 7.43490636e-01
6.83034062e-01 -4.91641253e-01 2.36205027e-01 1.68919027e-01
6.17510617e-01 -4.46343958e-01 7.23774850e-01 6.94403350e-01
-6.63380444e-01 -4.00682874e-02 -1.13602690e-01 1.22343414e-01
4.59077775e-01 -5.76616466e-01 -1.46274471e+00 5.44603348e-01
3.01008999e-01 9.44528282e-02 -3.59479040e-01 7.65522242e-01
-2.28190109e-01 8.59744549e-01 -1.64148226e-01 -2.85911858e-01
6.89826548e-01 2.67709809e-04 8.13990474e-01 1.29966617e+00
-3.79526943e-01 3.20907027e-01 2.72108793e-01 3.79636467e-01
3.91834565e-02 -1.15442187e-01 -3.81413609e-01 -1.02488235e-01
5.13277352e-01 9.43069279e-01 1.96735617e-02 -6.45677209e-01
-6.23127580e-01 1.28766608e+00 1.08815640e-01 2.56190300e-01
-4.35324162e-01 -3.52725029e-01 6.67708933e-01 -3.17891926e-01
-1.08690392e-02 -1.51531994e-01 1.65530145e-01 -1.23629248e+00
-1.64339259e-01 -1.25221777e+00 8.15185905e-01 -3.90675694e-01
-9.84501660e-01 7.28261054e-01 -1.28585756e-01 -8.57500613e-01
-9.02102649e-01 -2.80429423e-01 -3.32012653e-01 9.95049596e-01
-1.54463911e+00 -1.05767643e+00 -2.92396128e-01 8.05510700e-01
1.11045682e+00 -1.93545818e-01 1.47895801e+00 -1.22533798e-01
-5.57681024e-01 6.56944215e-01 1.24427723e-02 4.08649832e-01
8.34760845e-01 -1.71045113e+00 9.92702961e-01 5.99364460e-01
-2.11279795e-01 6.84480071e-01 8.32179427e-01 -5.85988462e-01
-1.56907761e+00 -7.12745905e-01 1.06287968e+00 -6.69183195e-01
6.05879903e-01 -5.52861810e-01 -1.25003397e+00 6.78591132e-01
6.49012685e-01 -6.84231520e-01 5.63733697e-01 3.95881385e-01
-1.54277146e-01 2.77917415e-01 -1.26498461e+00 6.15453601e-01
9.70362186e-01 -8.06749523e-01 -1.28556144e+00 3.70929241e-01
9.71514106e-01 -9.47995007e-01 -1.01589596e+00 -1.55612841e-01
1.75672621e-01 -8.03097069e-01 5.94070852e-01 -1.19939744e+00
-1.76433757e-01 -7.13952929e-02 1.31710079e-02 -1.00682056e+00
-6.18498251e-02 -1.26217985e+00 -8.15825820e-01 1.24129367e+00
4.99548078e-01 -4.64943022e-01 9.57323611e-01 1.20575345e+00
-4.08558011e-01 -4.08240825e-01 -1.02574742e+00 -6.69640362e-01
-1.24481671e-01 1.88446324e-03 5.81476092e-01 7.20510840e-01
7.39092886e-01 1.17231202e+00 -5.15632987e-01 1.10021472e-01
3.96283120e-01 1.38731405e-01 9.84234929e-01 -9.94284868e-01
-2.28756130e-01 -3.62325385e-02 1.03820607e-01 -1.76670551e+00
2.40529068e-02 -3.88309866e-01 2.00595379e-01 -1.83080757e+00
-1.57405123e-01 -3.57041061e-02 7.78288245e-02 5.45849800e-01
-2.47552648e-01 -5.22308409e-01 1.54724792e-01 8.62653330e-02
-1.19244778e+00 6.62789643e-01 1.18723130e+00 -3.99946235e-02
-4.77525562e-01 3.02403957e-01 -6.92015171e-01 4.49106425e-01
8.30276847e-01 -3.73262167e-02 -6.71661437e-01 -3.24498624e-01
-3.36015642e-01 8.08959007e-01 -1.38662979e-01 -7.21408725e-01
5.85921109e-01 -3.01610883e-02 -4.29364353e-01 -6.47141814e-01
8.61756921e-01 -4.11476403e-01 -7.55776763e-01 5.29034436e-01
-8.11617613e-01 -2.93836415e-01 2.81498343e-01 6.23247743e-01
-2.38270417e-01 -5.13922334e-01 6.48048460e-01 -3.07909220e-01
-1.12046874e+00 5.00192586e-03 -5.91601074e-01 3.89514953e-01
6.06578410e-01 9.56912246e-03 -7.55672216e-01 -9.36819136e-01
-7.94586241e-01 7.85807371e-01 3.76981199e-01 5.63581824e-01
4.43438768e-01 -8.11309099e-01 -6.31643355e-01 -1.43978894e-01
4.28336382e-01 -3.28188986e-02 2.98179418e-01 4.03391272e-01
-2.00456381e-01 1.01963913e+00 -1.15391309e-03 -5.69390476e-01
-1.32794082e+00 1.76157594e-01 5.66272020e-01 -6.27218068e-01
-4.93057519e-01 6.98886991e-01 -7.59966969e-02 -7.30131984e-01
5.11921406e-01 1.06927522e-01 -3.92976701e-01 -7.87833109e-02
7.30939627e-01 2.55313337e-01 1.55641750e-01 -4.39631552e-01
-2.76282310e-01 -2.56753385e-01 -6.15306139e-01 -5.07602096e-01
9.95238125e-01 -5.32960713e-01 2.02774927e-01 5.46626627e-01
8.30448210e-01 -4.74188358e-01 -1.26542354e+00 -8.05667043e-01
4.28431153e-01 -1.53792486e-01 -9.91988331e-02 -1.41112649e+00
-1.36112362e-01 7.55267680e-01 4.64206666e-01 7.47431517e-01
8.63429606e-01 1.83043793e-01 1.06471992e+00 8.43693376e-01
5.01266420e-01 -1.34784138e+00 2.48764277e-01 1.07651532e+00
7.43930936e-01 -1.37596047e+00 -4.47196007e-01 -1.86963618e-01
-1.12254322e+00 8.33186448e-01 9.86715376e-01 4.20346916e-01
-9.48371589e-02 -1.38105720e-01 3.60027492e-01 -5.21345139e-01
-1.29722250e+00 -3.25021356e-01 1.10137500e-01 5.94798505e-01
2.98512578e-01 -3.32848787e-01 -2.24077269e-01 5.14845252e-01
7.03308219e-03 -2.65924364e-01 4.57893640e-01 1.23209369e+00
-6.86617553e-01 -1.25168228e+00 6.77243248e-02 2.50626445e-01
-3.20286751e-01 -8.03283323e-03 -9.15890217e-01 7.93420911e-01
-9.45065737e-01 1.38499188e+00 -1.30193243e-02 -3.61025482e-01
8.42533171e-01 9.29319859e-01 2.31070593e-01 -9.43540931e-01
-1.08970368e+00 -1.35886103e-01 1.10472035e+00 -6.92723215e-01
-1.47771552e-01 -7.12043464e-01 -1.30481040e+00 -2.32484594e-01
-4.21759337e-01 6.54243588e-01 5.27741730e-01 7.62561321e-01
5.16235411e-01 8.99152085e-02 6.74967945e-01 5.11965267e-02
-1.14834976e+00 -1.13058019e+00 -2.14649677e-01 3.59413326e-01
4.12481993e-01 -3.09019953e-01 -5.99462017e-02 -2.61163324e-01] | [12.836320877075195, 7.927868843078613] |
dc2d5bcb-2a72-43b9-a914-a42db936834a | facial-feature-point-detection-a | 1410.1037 | null | http://arxiv.org/abs/1410.1037v1 | http://arxiv.org/pdf/1410.1037v1.pdf | Facial Feature Point Detection: A Comprehensive Survey | This paper presents a comprehensive survey of facial feature point detection
with the assistance of abundant manually labeled images. Facial feature point
detection favors many applications such as face recognition, animation,
tracking, hallucination, expression analysis and 3D face modeling. Existing
methods can be categorized into the following four groups: constrained local
model (CLM)-based, active appearance model (AAM)-based, regression-based, and
other methods. CLM-based methods consist of a shape model and a number of local
experts, each of which is utilized to detect a facial feature point. AAM-based
methods fit a shape model to an image by minimizing texture synthesis errors.
Regression-based methods directly learn a mapping function from facial image
appearance to facial feature points. Besides the above three major categories
of methods, there are also minor categories of methods which we classify into
other methods: graphical model-based methods, joint face alignment methods,
independent facial feature point detectors, and deep learning-based methods.
Though significant progress has been made, facial feature point detection is
limited in its success by wild and real-world conditions: variations across
poses, expressions, illuminations, and occlusions. A comparative illustration
and analysis of representative methods provide us a holistic understanding and
deep insight into facial feature point detection, which also motivates us to
explore promising future directions. | ['Xuelong. Li', 'DaCheng Tao', 'Nannan Wang', 'Xinbo Gao'] | 2014-10-04 | null | null | null | null | ['3d-face-modeling'] | ['computer-vision'] | [-1.40992165e-01 -2.06022248e-01 -3.56162637e-01 -6.21707499e-01
-4.76997107e-01 -2.14038983e-01 5.67315221e-01 -4.85633463e-01
-9.31941047e-02 1.61256149e-01 -3.51628102e-02 3.42803091e-01
1.18267328e-01 -2.47516260e-01 -2.90131629e-01 -9.16607320e-01
-1.30527288e-01 2.80110776e-01 -1.18275613e-01 -2.36817718e-01
2.20820755e-01 1.27938998e+00 -1.68027139e+00 -1.38403729e-01
3.00872982e-01 1.13255274e+00 -3.05454135e-01 1.46322533e-01
-1.65549055e-01 1.76500559e-01 -2.76220143e-01 -3.21127325e-01
2.85680294e-01 -2.28806019e-01 -3.17884237e-01 5.69846988e-01
3.90423089e-01 -4.03372198e-01 -3.85624245e-02 1.01910269e+00
5.88829815e-01 -2.83047706e-02 9.20236647e-01 -1.74581873e+00
-5.18961728e-01 -5.57275057e-01 -1.22864318e+00 -2.95743197e-01
6.03138030e-01 -1.62548199e-01 4.27575052e-01 -1.67191887e+00
6.85967565e-01 1.76121974e+00 8.27994466e-01 7.85251081e-01
-1.01956797e+00 -8.31076562e-01 6.63923174e-02 2.50653356e-01
-1.70902038e+00 -1.00338650e+00 1.13731265e+00 -3.75680655e-01
5.98048806e-01 2.43105263e-01 7.81850576e-01 7.62035668e-01
1.57242417e-01 7.53512681e-01 9.91922915e-01 -8.15652668e-01
6.26833290e-02 1.20055348e-01 -1.64559364e-01 1.42004144e+00
-2.55263120e-01 6.59077540e-02 -8.14802051e-01 -7.97821820e-01
1.25256085e+00 6.14087582e-02 -5.91572151e-02 -5.21004200e-01
-6.36231244e-01 8.36200953e-01 1.20475970e-01 5.77298068e-02
-3.94328415e-01 -1.01279572e-01 4.67592366e-02 2.48953760e-01
6.67221546e-01 -2.17832923e-01 -2.36918807e-01 2.59648710e-01
-1.10834932e+00 2.78539807e-01 6.10826194e-01 8.28031301e-01
9.68978107e-01 1.98857322e-01 2.10672200e-01 1.01749170e+00
1.19170547e+00 5.54192185e-01 3.80512476e-01 -9.54667747e-01
-1.62171513e-01 6.17250502e-01 8.17296878e-02 -1.47985423e+00
-4.58086014e-01 3.61305565e-01 -5.38505256e-01 7.76225567e-01
1.96021333e-01 3.85973416e-02 -7.78165460e-01 1.61155915e+00
7.44973481e-01 1.49623036e-01 -5.23673534e-01 9.68168497e-01
1.08997858e+00 3.24249536e-01 -7.75543675e-02 -5.14010608e-01
1.27757907e+00 -7.89870977e-01 -8.53956401e-01 2.49434233e-01
3.59523922e-01 -1.10350680e+00 5.00211954e-01 2.34825507e-01
-1.05813813e+00 -3.83167565e-01 -7.69764423e-01 1.53491378e-01
-1.84515268e-01 4.07137036e-01 5.39410114e-01 6.66072011e-01
-1.33292770e+00 1.97446689e-01 -9.26136017e-01 -5.24414420e-01
5.66870928e-01 7.25222886e-01 -7.98749030e-01 2.10265458e-01
-6.73079967e-01 9.79233563e-01 -6.48091793e-01 3.92428011e-01
-6.78608775e-01 -2.43235126e-01 -9.09617424e-01 -3.79883051e-01
-3.85321863e-03 -5.61513841e-01 1.12732732e+00 -1.37887764e+00
-1.89110351e+00 1.43731689e+00 -8.36203694e-01 4.05206025e-01
3.49603653e-01 -9.90934111e-03 -5.03498435e-01 1.91950619e-01
-2.21115470e-01 7.48301446e-01 1.37429941e+00 -1.25534213e+00
-1.45037726e-01 -7.25771308e-01 -5.25630474e-01 2.15173542e-01
-1.98545694e-01 9.74956155e-01 -5.12914956e-01 -3.19109261e-01
4.44619596e-01 -1.01904726e+00 -1.98404100e-02 9.55663502e-01
-2.54470438e-01 -5.99605739e-01 1.22883248e+00 -4.61124420e-01
7.28596985e-01 -2.09801221e+00 3.87306362e-02 3.99525940e-01
2.92466611e-01 2.48826340e-01 -3.66573960e-01 4.40116614e-01
-2.21133485e-01 -1.38347164e-01 3.13212425e-01 -6.18791997e-01
5.29841371e-02 -3.64786088e-02 -6.00918941e-02 9.59196627e-01
3.12947690e-01 1.04934382e+00 -3.98295939e-01 -7.49289572e-01
2.37751707e-01 7.72623122e-01 -2.73371279e-01 6.04382120e-02
3.49325538e-01 3.44855577e-01 -4.37853754e-01 1.35059428e+00
8.72314453e-01 1.11006588e-01 -1.10925607e-01 -3.30735326e-01
-1.03124753e-01 -4.90270615e-01 -1.26356339e+00 1.37959409e+00
-1.36863559e-01 6.22885048e-01 5.34887731e-01 -8.09558570e-01
1.28572404e+00 6.05711281e-01 7.41342187e-01 -2.95591980e-01
2.21436918e-01 3.47017646e-01 -3.96664888e-01 -5.99091470e-01
1.18716925e-01 -9.70214829e-02 4.83727694e-01 2.14152887e-01
7.56199434e-02 2.14701027e-01 -5.07577360e-01 -3.03764671e-01
4.86375272e-01 4.91867065e-01 5.71294308e-01 -1.61479101e-01
6.67296350e-01 -5.20608366e-01 5.21546900e-01 3.82398069e-03
-5.67895055e-01 5.12948751e-01 2.19581887e-01 -6.38743877e-01
-7.38345742e-01 -8.13216388e-01 -3.14898342e-01 1.01258051e+00
1.31227419e-01 -4.44891512e-01 -7.35518873e-01 -5.52906215e-01
1.29893348e-01 -2.27453113e-01 -7.74419367e-01 -1.93957891e-02
-5.38071811e-01 -6.04751587e-01 3.76818836e-01 4.37066495e-01
4.10631180e-01 -9.31683242e-01 -3.66359800e-01 -2.53649235e-01
1.59754306e-01 -8.28565717e-01 -4.02303427e-01 -4.94772345e-01
-7.84669816e-01 -1.04692483e+00 -8.87840033e-01 -9.08057809e-01
1.20793569e+00 2.37694696e-01 7.51139939e-01 3.59957427e-01
-4.90808219e-01 8.04436505e-01 -1.46641627e-01 -5.77751458e-01
-4.56486195e-02 -6.27761722e-01 5.52337468e-01 5.50723433e-01
6.81422651e-01 -4.26343530e-01 -4.78512108e-01 7.13438570e-01
-2.81949580e-01 -2.82875568e-01 4.42910761e-01 7.69708395e-01
6.39266908e-01 -5.56928813e-01 1.67032540e-01 -3.03654641e-01
4.22560275e-01 -2.45679840e-01 -4.45050597e-01 2.94241756e-01
-1.79747999e-01 -5.10786295e-01 6.50502294e-02 -4.84806716e-01
-8.79382074e-01 4.97272849e-01 -1.10364720e-01 -7.75322556e-01
-2.61190116e-01 2.01859921e-01 -4.00612414e-01 -7.77717650e-01
6.24426901e-01 2.48074844e-01 6.92792416e-01 -3.99793267e-01
8.42823163e-02 6.34139001e-01 1.45077765e-01 -3.69818449e-01
8.64624083e-01 6.24440312e-01 3.54245275e-01 -1.27098846e+00
-3.40740025e-01 -4.73325849e-01 -1.11243343e+00 -6.36596441e-01
4.38068360e-01 -6.13334119e-01 -8.07221353e-01 7.25304425e-01
-1.22760391e+00 3.96431163e-02 2.22341716e-01 3.34949493e-01
-6.18658781e-01 5.53849757e-01 -2.95378327e-01 -1.09502339e+00
-2.81722516e-01 -1.36789072e+00 1.42033708e+00 2.79580146e-01
-3.48881006e-01 -1.15112603e+00 -1.44158828e-03 1.49512112e-01
3.49073589e-01 5.63925207e-01 6.17520332e-01 -1.91873401e-01
-2.29926422e-01 -5.77623308e-01 -3.46306004e-02 7.14971498e-02
4.59874839e-01 6.40978217e-01 -9.80617523e-01 -3.13194662e-01
-6.85090944e-02 -4.63342786e-01 2.97297776e-01 4.98265535e-01
9.43373859e-01 -3.11709076e-01 -6.34691954e-01 6.92187905e-01
1.04384816e+00 1.60236046e-01 5.08159637e-01 1.20973319e-01
4.77373660e-01 6.98442578e-01 5.61326802e-01 4.72526819e-01
5.81783094e-02 1.17851257e+00 4.43701148e-01 -4.41223413e-01
-3.26989777e-02 2.78334375e-02 3.57399672e-01 4.89546359e-01
-5.54455221e-01 4.08602387e-01 -6.43891156e-01 -1.44961700e-01
-1.73270309e+00 -9.00737822e-01 1.31122991e-01 1.98033857e+00
5.77459514e-01 -5.93617857e-01 4.18613404e-01 1.30953550e-01
8.47198844e-01 -1.05067261e-01 -5.58345616e-01 -2.85572261e-01
-2.79024709e-02 6.19733594e-02 -1.90338016e-01 2.61913866e-01
-1.13438976e+00 8.82483125e-01 6.94796658e+00 8.00785244e-01
-1.48590946e+00 5.31499311e-02 5.16717970e-01 2.53091663e-01
9.29713920e-02 -1.22822419e-01 -1.01740015e+00 2.22091809e-01
2.58741409e-01 -6.57219579e-03 1.90342903e-01 1.02748573e+00
3.88690680e-01 3.85731645e-02 -1.01352811e+00 1.67200947e+00
4.59466010e-01 -1.07523012e+00 9.32025313e-02 2.32603967e-01
4.75120366e-01 -2.79190719e-01 1.47786126e-01 -8.36659446e-02
-3.94386262e-01 -1.09543991e+00 5.94400823e-01 8.23187709e-01
9.19024944e-01 -6.49061680e-01 2.92848051e-01 2.10626587e-01
-1.23042047e+00 1.93003386e-01 -4.86717284e-01 1.23215690e-02
-3.05325934e-03 2.16936350e-01 -5.32581508e-01 2.09533349e-01
4.82855350e-01 5.85733712e-01 -4.43200916e-01 1.07653260e+00
-8.27797409e-03 3.93009156e-01 -4.22550768e-01 2.63490397e-02
-8.90824422e-02 -2.94812530e-01 4.81146246e-01 9.73544419e-01
2.08777577e-01 1.10987872e-01 2.41138622e-01 8.55062306e-01
2.02164650e-01 5.18689215e-01 -5.69766164e-01 2.35454321e-01
4.19021726e-01 1.76300454e+00 -7.76924312e-01 1.09422527e-01
-6.88161731e-01 9.08313334e-01 2.15066984e-01 4.31927025e-01
-5.85337818e-01 -2.12309271e-01 6.50655091e-01 3.63463640e-01
-2.39206806e-01 -5.26091516e-01 2.55450219e-01 -1.04600859e+00
-2.10529447e-01 -7.66902566e-01 2.09518224e-02 -8.67237031e-01
-1.10128784e+00 5.57272136e-01 1.65660575e-01 -1.25762606e+00
-3.58754069e-01 -8.64968419e-01 -7.90319502e-01 9.67194259e-01
-1.41706955e+00 -1.45564914e+00 -5.21001637e-01 8.82181227e-01
4.00922447e-01 -6.17272317e-01 1.05160010e+00 1.11280583e-01
-5.96832097e-01 9.08691704e-01 -1.53099939e-01 3.46285045e-01
7.15940595e-01 -5.77926636e-01 1.55079260e-01 3.25338393e-01
1.57397956e-01 7.84047961e-01 3.80222380e-01 -3.12262088e-01
-1.66184556e+00 -7.56134570e-01 7.26333916e-01 -3.66643190e-01
2.17602998e-01 -3.37568372e-01 -7.12001681e-01 6.36507273e-01
-1.86249733e-01 6.00358486e-01 8.10741961e-01 -1.01147279e-01
-3.39658171e-01 -1.56645909e-01 -1.37454975e+00 5.05948663e-01
9.09221172e-01 -2.87740737e-01 -1.47193179e-01 4.47426289e-01
-2.64704645e-01 -4.34329838e-01 -6.97743297e-01 4.44989353e-01
9.54074860e-01 -9.72236454e-01 8.59686732e-01 -5.24858057e-01
-1.38321996e-01 -1.20382816e-01 -9.67769045e-03 -1.00441873e+00
-4.97713327e-01 -8.51921916e-01 -2.41205856e-01 1.14725065e+00
-2.80769076e-02 -7.31012464e-01 9.23311949e-01 8.09781194e-01
3.81436013e-02 -1.16956115e+00 -1.14767969e+00 -6.62563443e-01
-2.78705597e-01 2.19507422e-02 5.68243742e-01 1.05612385e+00
1.99394505e-02 -2.67892275e-02 -2.06028014e-01 -8.84481370e-02
6.49090707e-01 8.34572688e-02 1.05495965e+00 -1.35437655e+00
2.51933128e-01 -6.33191705e-01 -8.91353071e-01 -9.73965228e-01
5.93508661e-01 -6.25579000e-01 -3.03974003e-01 -1.14184773e+00
2.40869030e-01 -3.45212519e-01 1.32852629e-01 8.44451427e-01
2.83366382e-01 5.76518536e-01 -6.11779094e-02 4.57683712e-01
-3.24403554e-01 5.04734159e-01 1.26347208e+00 9.43276510e-02
-1.89188153e-01 1.67098090e-01 -2.98360199e-01 1.14434016e+00
6.15197420e-01 -1.31223142e-01 -1.38314068e-01 -6.81778118e-02
-1.45076402e-02 4.45864797e-02 5.15501976e-01 -4.98066097e-01
1.97697282e-01 -5.54970622e-01 8.16959620e-01 -4.11922723e-01
7.72581160e-01 -7.24634588e-01 -1.38034346e-02 2.99376637e-01
1.01484857e-01 1.52078167e-01 5.62123246e-02 1.77277729e-01
-2.59764373e-01 -1.64909616e-01 1.16732037e+00 -1.71413317e-01
-8.21124732e-01 6.95261836e-01 -3.13785940e-01 -5.39564371e-01
1.00227273e+00 -7.57414103e-01 2.01412395e-01 -5.94141960e-01
-8.62551451e-01 -2.60077357e-01 4.37259674e-01 5.10541916e-01
1.01701772e+00 -1.66268539e+00 -7.70582438e-01 7.60725439e-01
7.21214414e-02 -4.42679554e-01 -6.88109621e-02 1.15610754e+00
-4.63725775e-01 2.91393578e-01 -3.42296004e-01 -8.44647467e-01
-1.82350552e+00 2.52400249e-01 4.99663860e-01 6.19167387e-01
-2.34271392e-01 9.08299327e-01 2.49792963e-01 -3.42166632e-01
2.27163598e-01 2.15431511e-01 -3.91058534e-01 2.85261348e-02
5.75340986e-01 4.62986618e-01 9.86194462e-02 -1.51992738e+00
-6.16846621e-01 1.28834164e+00 1.72849640e-01 1.55635670e-01
1.01548398e+00 -5.24707101e-02 -3.60515088e-01 2.73091078e-01
1.36493719e+00 6.19131438e-02 -1.03024018e+00 -2.91582108e-01
-2.73894012e-01 -6.49938941e-01 -5.07271327e-02 -4.26434606e-01
-1.09909725e+00 7.81044900e-01 6.90775871e-01 -2.41431460e-01
1.25073802e+00 -1.43887708e-02 2.06699222e-01 2.46221900e-01
5.34212470e-01 -9.24364567e-01 4.08129930e-01 3.34719062e-01
1.29899049e+00 -1.20063710e+00 1.07585751e-01 -6.21904492e-01
-2.43224934e-01 1.45644498e+00 5.84076285e-01 -1.01833500e-01
1.25998998e+00 2.20621243e-01 3.36519301e-01 -2.72993118e-01
-5.19760370e-01 -3.28514352e-02 5.53701401e-01 7.42318809e-01
5.05767822e-01 -2.68042445e-01 -4.40420844e-02 2.47613639e-01
-9.66574699e-02 -3.44808660e-02 -1.89644635e-01 8.97800565e-01
-5.56400478e-01 -1.03573596e+00 -7.21758664e-01 1.86130509e-01
-1.92740604e-01 3.63262147e-01 -5.38288772e-01 9.66464102e-01
-9.25193541e-03 6.57307923e-01 1.22574195e-01 -2.39286155e-01
2.38127813e-01 2.95077503e-01 1.06136632e+00 -5.08549035e-01
-5.56280129e-02 4.65533733e-01 -4.07957852e-01 -5.21811962e-01
-7.20079660e-01 -7.96281040e-01 -1.15944958e+00 -2.08211064e-01
-7.23896384e-01 -9.46501195e-02 8.82918835e-01 8.12984109e-01
3.70253563e-01 -4.91037428e-01 9.90912795e-01 -1.45852280e+00
-3.51941019e-01 -8.69327486e-01 -7.76588738e-01 2.99739301e-01
2.48353601e-01 -1.20616400e+00 -3.69091898e-01 1.39240995e-01] | [13.29920768737793, 0.5050673484802246] |
3b3c922c-af8f-4648-852a-ced7176f84e3 | zero-shot-cross-lingual-opinion-target | 1904.09122 | null | http://arxiv.org/abs/1904.09122v1 | http://arxiv.org/pdf/1904.09122v1.pdf | Zero-Shot Cross-Lingual Opinion Target Extraction | Aspect-based sentiment analysis involves the recognition of so called opinion
target expressions (OTEs). To automatically extract OTEs, supervised learning
algorithms are usually employed which are trained on manually annotated
corpora. The creation of these corpora is labor-intensive and sufficiently
large datasets are therefore usually only available for a very narrow selection
of languages and domains. In this work, we address the lack of available
annotated data for specific languages by proposing a zero-shot cross-lingual
approach for the extraction of opinion target expressions. We leverage
multilingual word embeddings that share a common vector space across various
languages and incorporate these into a convolutional neural network
architecture for OTE extraction. Our experiments with 5 languages give
promising results: We can successfully train a model on annotated data of a
source language and perform accurate prediction on a target language without
ever using any annotated samples in that target language. Depending on the
source and target language pairs, we reach performances in a zero-shot regime
of up to 77% of a model trained on target language data. Furthermore, we can
increase this performance up to 87% of a baseline model trained on target
language data by performing cross-lingual learning from multiple source
languages. | ['Soufian Jebbara', 'Philipp Cimiano'] | 2019-04-19 | zero-shot-cross-lingual-opinion-target-1 | https://aclanthology.org/N19-1257 | https://aclanthology.org/N19-1257.pdf | naacl-2019-6 | ['multilingual-word-embeddings'] | ['methodology'] | [-2.44418606e-02 -4.33622524e-02 -3.51541907e-01 -4.78110850e-01
-1.18952847e+00 -7.09857464e-01 6.34891033e-01 4.57878590e-01
-8.02472353e-01 5.93428850e-01 8.01046863e-02 -2.91129291e-01
6.22309685e-01 -7.26976812e-01 -5.59257805e-01 -3.84449363e-01
3.86273772e-01 5.95650196e-01 -1.85058080e-02 -3.78577024e-01
-2.55096346e-01 2.95280755e-01 -1.23611283e+00 4.07814413e-01
7.24473953e-01 8.82389605e-01 -1.60781339e-01 5.94227493e-01
-6.54842257e-01 5.82904816e-01 -5.72130561e-01 -8.63835096e-01
2.76685178e-01 -1.18565731e-01 -5.09554386e-01 1.86398849e-01
2.57077903e-01 2.92041302e-01 3.58114660e-01 1.00697732e+00
2.92851806e-01 -3.67026299e-01 6.64031446e-01 -8.88580441e-01
-5.08017182e-01 5.81492484e-01 -4.99264002e-01 -1.49367392e-01
1.21789105e-01 -1.68147057e-01 1.44157588e+00 -1.11748469e+00
8.28309000e-01 7.84251690e-01 5.98874390e-01 2.79523104e-01
-1.25693977e+00 -5.53592384e-01 2.52950311e-01 -2.26464331e-01
-1.17614686e+00 -4.76577491e-01 8.72532070e-01 -4.82695878e-01
1.26172841e+00 -1.55025914e-01 7.30873346e-01 1.15046799e+00
1.26999676e-01 9.19643104e-01 1.11779451e+00 -7.83221602e-01
1.91097200e-01 7.22390592e-01 1.33574262e-01 5.91660857e-01
1.03562303e-01 -2.91128308e-01 -6.23839557e-01 -9.43765119e-02
5.33702858e-02 -3.87790710e-01 -8.80419090e-02 -4.92969185e-01
-1.16582274e+00 1.19653296e+00 1.11206330e-01 7.69371092e-01
-2.63806224e-01 -1.78654984e-01 7.75926411e-01 5.04974842e-01
9.84661102e-01 6.38431907e-01 -1.06816244e+00 -6.56150468e-03
-8.22631538e-01 -7.40029961e-02 1.13825858e+00 8.54490638e-01
9.17821705e-01 1.00288697e-01 2.36450747e-01 1.04994822e+00
9.61916819e-02 6.33098423e-01 6.07373357e-01 -1.95310086e-01
4.80975032e-01 9.44080114e-01 1.52245790e-01 -6.50972366e-01
-4.27633524e-01 -4.61449474e-01 -4.50577259e-01 2.89540701e-02
4.42547083e-01 -5.41844904e-01 -7.62368560e-01 1.56935549e+00
4.72190678e-01 -3.72467577e-01 5.35738707e-01 4.75758821e-01
5.20891905e-01 7.19712973e-01 5.80527075e-02 -1.07999556e-01
1.59021175e+00 -1.01038790e+00 -5.24058402e-01 -5.50495267e-01
1.13386571e+00 -1.10592949e+00 1.09330463e+00 5.27915776e-01
-5.53469896e-01 -2.70681381e-01 -1.08916390e+00 5.54054938e-02
-9.76291597e-01 5.26788473e-01 6.73771203e-01 7.05671966e-01
-8.63821924e-01 1.26984864e-01 -6.89072609e-01 -3.88446331e-01
2.02889889e-01 4.04735327e-01 -7.35000432e-01 -1.40637800e-01
-1.15236485e+00 1.00293648e+00 3.37655365e-01 -1.06067501e-01
-3.89257461e-01 -6.23058319e-01 -1.14872491e+00 -2.94932961e-01
2.54160076e-01 -2.55116194e-01 1.11127377e+00 -1.61389327e+00
-1.29699576e+00 1.23923087e+00 -2.29785085e-01 -3.27060491e-01
1.14253640e-01 -3.12349856e-01 -6.43708944e-01 -1.10283926e-01
3.46922696e-01 3.02712619e-01 7.98253119e-01 -1.06702673e+00
-7.26376951e-01 -4.14161980e-01 3.06156669e-02 -6.38581291e-02
-7.04282880e-01 3.57332289e-01 -5.46466768e-01 -5.49058080e-01
-3.26549172e-01 -1.06136799e+00 -2.62717426e-01 -2.77930826e-01
-1.47721753e-01 -3.39067876e-01 4.94481593e-01 -5.66371679e-01
8.78346741e-01 -2.01819110e+00 1.18745007e-01 1.31070480e-01
8.71671140e-02 4.25942987e-01 -4.00030196e-01 3.55966657e-01
-1.15884870e-01 -4.08385089e-03 -2.45532364e-01 -5.14750659e-01
1.31835759e-01 1.52815506e-01 -1.97644070e-01 3.75744432e-01
5.89585602e-01 8.29962969e-01 -9.47730958e-01 -3.26402307e-01
9.11472440e-02 6.51566982e-01 -5.67180932e-01 2.96553314e-01
-2.78358042e-01 1.02739453e-01 -4.52925354e-01 5.65137804e-01
4.01336789e-01 1.22255743e-01 4.36371982e-01 -1.77116677e-01
-1.85851336e-01 4.33695823e-01 -8.10580075e-01 1.45426226e+00
-1.30349028e+00 6.68917000e-01 -3.06474585e-02 -9.67554688e-01
1.19022357e+00 4.65750128e-01 4.56474543e-01 -4.44093287e-01
2.45074794e-01 7.01756775e-01 2.75828540e-02 -2.99513608e-01
3.74425679e-01 -5.23642421e-01 -5.57139933e-01 6.56044722e-01
5.82188547e-01 -2.61754543e-01 4.74271029e-01 -1.70815408e-01
8.80473316e-01 1.84100159e-02 5.10848701e-01 -3.95795822e-01
8.07416081e-01 1.24692008e-01 4.67501909e-01 1.26026213e-01
3.18110995e-02 3.12489837e-01 6.52899504e-01 -8.16001713e-01
-1.14448917e+00 -6.31514668e-01 -1.71421811e-01 1.37276578e+00
-4.78872240e-01 -7.34107316e-01 -5.81430674e-01 -1.00524223e+00
-2.04017803e-01 7.35992968e-01 -6.17079198e-01 1.25246629e-01
-4.58173573e-01 -8.76797438e-01 5.15551448e-01 3.35454315e-01
2.87060235e-02 -1.06191123e+00 -1.73068315e-01 3.58735889e-01
7.14161694e-02 -1.46543145e+00 -2.01498851e-01 5.48384368e-01
-4.59671289e-01 -9.39372241e-01 -5.51959991e-01 -8.86002064e-01
5.76191306e-01 -3.45277131e-01 1.38639557e+00 -3.39806348e-01
-8.72378610e-03 1.72517121e-01 -5.06637633e-01 -6.89668059e-01
-5.30528963e-01 5.70409656e-01 1.15285948e-01 3.96962166e-01
1.11875951e+00 -3.70536774e-01 -5.65730557e-02 -4.19521220e-02
-8.30075979e-01 -3.14708918e-01 6.08771980e-01 7.56975412e-01
4.68319356e-01 -3.91209483e-01 5.29471278e-01 -1.15120208e+00
5.13429582e-01 -3.49642366e-01 -7.79637456e-01 2.47974768e-01
-5.10743201e-01 3.55205089e-01 8.13318849e-01 -4.74703491e-01
-9.09848630e-01 1.72681540e-01 -4.33997124e-01 -8.12206045e-02
-2.16629222e-01 6.98084474e-01 -1.76268727e-01 1.01657666e-01
5.14753401e-01 5.60171343e-02 -4.51107323e-01 -3.88156801e-01
5.15562534e-01 9.02962625e-01 -4.78073545e-02 -5.10697663e-01
6.12924695e-01 2.83591151e-01 -5.17712295e-01 -9.90949690e-01
-1.22842658e+00 -5.71518302e-01 -8.84704649e-01 6.58334568e-02
9.39362884e-01 -1.13475645e+00 5.79538308e-02 3.89746964e-01
-1.19797814e+00 -8.49415883e-02 -2.56260812e-01 5.39036393e-01
-8.39238018e-02 -5.41223362e-02 -3.89051080e-01 -6.44485295e-01
-4.52809274e-01 -1.15567446e+00 1.33197868e+00 -1.52468771e-01
-4.31094587e-01 -1.37172711e+00 4.24065441e-01 2.20803276e-01
3.83895308e-01 6.36153147e-02 8.86863470e-01 -1.21076369e+00
-1.03535026e-01 -6.28013968e-01 -1.59727950e-02 5.32520652e-01
3.17942172e-01 -4.84005809e-02 -1.14095056e+00 -1.68043181e-01
-3.74008529e-03 -6.01518571e-01 7.83285558e-01 -1.18026935e-01
3.04730833e-01 -9.18684825e-02 -9.59832594e-02 4.64695662e-01
1.52727973e+00 -1.47297919e-01 1.06455371e-01 4.63705599e-01
7.22634077e-01 8.68116558e-01 4.02173668e-01 1.39653459e-01
4.50940073e-01 6.00000679e-01 -1.48807704e-01 -9.79006290e-02
2.79609829e-01 -1.26115724e-01 7.11681783e-01 1.29976475e+00
2.07500860e-01 -1.92209274e-01 -9.57303822e-01 1.16057265e+00
-1.42113972e+00 -2.49616057e-01 -1.27626676e-02 2.15167451e+00
1.08951128e+00 2.96290219e-01 -1.36840269e-01 -1.26732484e-01
2.36325040e-01 3.06619793e-01 -2.33517766e-01 -8.33733797e-01
-1.43490255e-01 5.64460635e-01 4.77438509e-01 5.55799484e-01
-1.29788947e+00 1.29752982e+00 5.36858368e+00 7.47505009e-01
-1.46442890e+00 3.50366205e-01 4.32853520e-01 -1.81237794e-02
-4.52248245e-01 -2.21819263e-02 -1.01459765e+00 -3.02762985e-02
1.23712540e+00 -1.29281864e-01 -4.33008224e-02 1.04539669e+00
-1.56889379e-01 2.39940807e-02 -9.80770469e-01 9.57595468e-01
2.94491261e-01 -9.26621139e-01 -1.35067791e-01 3.98472026e-02
9.40542996e-01 4.96354103e-01 -3.24554563e-01 5.73050022e-01
4.67541665e-01 -7.87316442e-01 3.73389453e-01 2.80064289e-02
7.43080258e-01 -1.01486635e+00 1.05445802e+00 1.77865654e-01
-1.06085086e+00 3.48131210e-01 -5.24239004e-01 1.17936723e-01
2.39388555e-01 9.02030587e-01 -7.97498524e-01 4.01016295e-01
3.78008366e-01 8.16321790e-01 -4.98062640e-01 3.56138557e-01
-5.41409969e-01 6.00808561e-01 -3.78967434e-01 -1.86533019e-01
4.51961219e-01 -3.06454718e-01 3.24941128e-01 1.39987707e+00
4.06299531e-01 -5.88908076e-01 2.60664791e-01 4.08185989e-01
-3.19787860e-01 8.74626338e-01 -9.68733907e-01 -4.25639600e-01
-2.34099865e-01 1.60053408e+00 -5.75781882e-01 -4.72085625e-01
-8.31704199e-01 9.94383454e-01 6.51787400e-01 1.31641477e-01
-5.08556485e-01 -4.67473984e-01 7.80411839e-01 -2.30606243e-01
6.41307473e-01 -2.17847154e-01 -1.25153050e-01 -1.66164100e+00
1.06892452e-01 -1.11507332e+00 2.14808434e-01 -3.62804234e-01
-1.45261872e+00 1.19117880e+00 -4.02297169e-01 -1.16883659e+00
-4.69463855e-01 -1.03984678e+00 -3.83130610e-01 9.41286862e-01
-1.66917276e+00 -1.44489706e+00 3.17732990e-01 3.86049747e-01
6.80914760e-01 -5.28211176e-01 1.26275945e+00 2.72573739e-01
-3.98987532e-01 6.43674433e-01 1.50614202e-01 4.67220277e-01
8.65669727e-01 -1.24714804e+00 3.42740536e-01 7.32040584e-01
7.68497884e-01 6.24616981e-01 5.48932195e-01 -3.23152632e-01
-1.29022729e+00 -1.14951789e+00 1.70661235e+00 -6.11934721e-01
1.23733449e+00 -7.46124327e-01 -8.24113786e-01 8.35224509e-01
4.35218066e-01 4.99871746e-02 1.20849097e+00 6.61585450e-01
-6.64836109e-01 -1.32489607e-01 -8.37542593e-01 5.36485076e-01
3.42795908e-01 -8.67543280e-01 -5.54525018e-01 3.96572739e-01
6.62101507e-01 -6.04547523e-02 -9.55574453e-01 1.41995892e-01
5.29253602e-01 -8.05296719e-01 4.76054072e-01 -7.17563093e-01
4.53497142e-01 -2.38656580e-01 -2.80043781e-01 -1.48630714e+00
4.20133054e-01 -1.14328831e-01 3.01221311e-01 1.35984623e+00
1.05882144e+00 -7.20905960e-01 4.96909857e-01 4.60672498e-01
2.23108649e-01 -8.02738607e-01 -7.70649314e-01 -5.48721015e-01
3.31318587e-01 -9.87011611e-01 3.42773944e-01 9.72963572e-01
5.99797294e-02 8.76627624e-01 -2.91278899e-01 1.24143511e-02
3.09460908e-01 3.80788833e-01 8.19766939e-01 -1.02083921e+00
-2.28029564e-01 -2.11289138e-01 -4.30757880e-01 -6.29803061e-01
7.42084861e-01 -1.04312313e+00 2.63874442e-03 -1.26731181e+00
-4.94635776e-02 -3.48074496e-01 -2.61494428e-01 4.81305599e-01
-3.68927382e-02 5.98730326e-01 -2.32325457e-02 -1.41780645e-01
-4.62022483e-01 5.72249651e-01 6.52325094e-01 -2.51132578e-01
-4.52631377e-02 -9.24047828e-02 -6.96567953e-01 1.11595857e+00
7.43667305e-01 -6.63193822e-01 -8.49082768e-02 -4.71837431e-01
4.98205394e-01 -5.07753730e-01 -2.88568616e-01 -6.10505581e-01
-1.86090454e-01 1.28836334e-01 1.73182756e-01 -2.08360016e-01
2.98491567e-01 -8.17552626e-01 -3.71486306e-01 1.13980047e-01
-1.91248700e-01 9.22391340e-02 3.74584138e-01 3.29817295e-01
-6.01220965e-01 -3.98110420e-01 5.68012714e-01 -9.16191936e-02
-7.34332502e-01 4.26529199e-02 -5.14673471e-01 1.03285745e-01
8.50276113e-01 3.28158081e-01 1.52572885e-01 -2.40296304e-01
-6.12208962e-01 -6.32594824e-02 5.81912518e-01 6.08588517e-01
2.23568439e-01 -1.34840858e+00 -6.70054197e-01 4.32860494e-01
6.84200525e-01 -3.38886023e-01 -3.13225657e-01 7.19072521e-01
-3.58599752e-01 5.16307712e-01 9.97078642e-02 -3.98015916e-01
-1.13710368e+00 4.48610872e-01 2.77660131e-01 -7.69906580e-01
-1.15060560e-01 7.11725295e-01 1.36647224e-01 -1.10965669e+00
-3.98876876e-01 -2.87969828e-01 -4.06583518e-01 4.97009993e-01
4.56839412e-01 -4.78310913e-01 3.09155285e-01 -1.08539748e+00
-3.98657411e-01 6.68312013e-01 -1.27367228e-01 -2.61763781e-01
1.54955745e+00 1.51654467e-01 -3.42044622e-01 7.62380242e-01
1.58442199e+00 5.71746945e-01 -6.68311477e-01 -2.91339964e-01
3.61524135e-01 -1.67546079e-01 2.26417016e-02 -3.92181575e-01
-1.09168708e+00 9.18328464e-01 2.93930173e-01 -4.68880348e-02
9.91640091e-01 1.53833002e-01 7.71844864e-01 4.52878326e-01
4.66547847e-01 -1.13311064e+00 -2.00068057e-01 7.59054065e-01
6.37427747e-01 -1.56064820e+00 -2.27992982e-01 -2.09456131e-01
-7.31314719e-01 1.16829312e+00 3.22133183e-01 -2.87933767e-01
7.97024071e-01 3.12735587e-01 6.60449386e-01 -2.90090561e-01
-8.76211166e-01 -4.75388557e-01 4.66002762e-01 4.81573582e-01
9.67647791e-01 1.02573685e-01 -3.96335334e-01 6.11654043e-01
-3.14928472e-01 -1.59159049e-01 4.84127164e-01 7.64946520e-01
-1.32481307e-01 -1.75684309e+00 -1.33820608e-01 3.57071787e-01
-9.43919003e-01 -4.42771763e-01 -5.34488559e-01 8.44197929e-01
1.03976585e-01 8.59754205e-01 -7.57481977e-02 -1.70577735e-01
3.31751555e-01 4.84901905e-01 2.18467966e-01 -8.97553146e-01
-6.41505897e-01 2.45222837e-01 4.84780729e-01 -3.85413259e-01
-6.89362705e-01 -6.50766611e-01 -7.65916049e-01 1.92520022e-01
-2.45226800e-01 3.26322734e-01 8.67763102e-01 1.30517626e+00
7.53304511e-02 2.18946815e-01 5.80711544e-01 -6.87460899e-01
-1.65004343e-01 -1.00311363e+00 -4.73294795e-01 4.02337879e-01
2.33769298e-01 -3.15838516e-01 -3.59863847e-01 1.23877369e-01] | [11.33719539642334, 7.0341477394104] |
fc7dcd8f-2d42-42b1-81ef-e9c82ec00cb9 | semantically-self-aligned-network-for-text-to | 2107.12666 | null | https://arxiv.org/abs/2107.12666v2 | https://arxiv.org/pdf/2107.12666v2.pdf | Semantically Self-Aligned Network for Text-to-Image Part-aware Person Re-identification | Text-to-image person re-identification (ReID) aims to search for images containing a person of interest using textual descriptions. However, due to the significant modality gap and the large intra-class variance in textual descriptions, text-to-image ReID remains a challenging problem. Accordingly, in this paper, we propose a Semantically Self-Aligned Network (SSAN) to handle the above problems. First, we propose a novel method that automatically extracts semantically aligned part-level features from the two modalities. Second, we design a multi-view non-local network that captures the relationships between body parts, thereby establishing better correspondences between body parts and noun phrases. Third, we introduce a Compound Ranking (CR) loss that makes use of textual descriptions for other images of the same identity to provide extra supervision, thereby effectively reducing the intra-class variance in textual features. Finally, to expedite future research in text-to-image ReID, we build a new database named ICFG-PEDES. Extensive experiments demonstrate that SSAN outperforms state-of-the-art approaches by significant margins. Both the new ICFG-PEDES database and the SSAN code are available at https://github.com/zifyloo/SSAN. | ['DaCheng Tao', 'Zhiyin Shao', 'Changxing Ding', 'Zefeng Ding'] | 2021-07-27 | null | null | null | null | ['nlp-based-person-retrival'] | ['computer-vision'] | [ 1.32360533e-01 -3.18475366e-01 -3.27645093e-01 -5.93689680e-01
-7.49473214e-01 -4.57395971e-01 6.43995941e-01 -6.60242885e-02
-5.28909147e-01 4.15228248e-01 5.24598598e-01 5.23903430e-01
-1.45865366e-01 -4.25161004e-01 -5.15721440e-01 -5.52592337e-01
3.87041509e-01 3.87793958e-01 -4.70042862e-02 7.99103230e-02
2.63238810e-02 4.71571922e-01 -1.58544230e+00 1.28232732e-01
4.81649131e-01 8.67437363e-01 7.48937810e-03 1.26044555e-02
8.56065452e-02 4.75522459e-01 -7.78198615e-02 -1.01038921e+00
4.97668445e-01 -4.38143522e-01 -7.54280746e-01 2.24278092e-01
1.15703607e+00 -4.20456350e-01 -6.77264094e-01 1.26838207e+00
7.30181515e-01 2.82102644e-01 5.69274545e-01 -1.43805790e+00
-6.42285526e-01 3.93100113e-01 -1.00449133e+00 1.26793608e-01
5.88794172e-01 -1.05955526e-01 8.65068257e-01 -1.06963921e+00
7.70481706e-01 1.48421299e+00 7.25937486e-01 7.82932103e-01
-1.07535994e+00 -9.94181633e-01 1.58470228e-01 4.44249839e-01
-1.83322418e+00 -7.36017108e-01 7.57214248e-01 -2.30584413e-01
3.44453990e-01 2.69179463e-01 3.89069855e-01 1.19488549e+00
-2.94879943e-01 9.31120753e-01 1.03689122e+00 -3.16997707e-01
-4.06058758e-01 1.24893352e-01 -1.05067968e-01 8.21380675e-01
2.83057451e-01 6.72935974e-04 -9.58856940e-01 -1.19380206e-01
7.60866940e-01 1.44254461e-01 -1.56373858e-01 -5.92875719e-01
-1.20304120e+00 4.52397615e-01 4.78550851e-01 1.78417057e-01
-1.10400639e-01 -3.48422192e-02 6.68150544e-01 -9.03810486e-02
3.82426322e-01 2.62793191e-02 8.00670590e-03 9.67499390e-02
-7.55601883e-01 2.80933827e-01 4.33938533e-01 1.11634958e+00
6.29452467e-01 -4.77575272e-01 -1.64384231e-01 1.13763356e+00
2.23174959e-01 5.64351261e-01 4.15018618e-01 -7.43122280e-01
6.63326561e-01 5.30892670e-01 -1.07234925e-01 -1.24574316e+00
-2.27796793e-01 -3.40191990e-01 -1.09800291e+00 -5.01793504e-01
3.62605602e-01 1.38827369e-01 -7.51816094e-01 1.76502359e+00
6.21281743e-01 -5.49217500e-02 -1.26295179e-01 1.10492790e+00
1.09718478e+00 1.20768212e-01 2.12608665e-01 1.24245040e-01
1.72819936e+00 -1.03802586e+00 -5.02132177e-01 -3.11277539e-01
2.39531115e-01 -6.20923221e-01 6.78707838e-01 -2.53581703e-01
-1.00727391e+00 -6.28514051e-01 -5.77587605e-01 -2.32986510e-01
-3.38734627e-01 4.58599508e-01 2.21888140e-01 3.56134921e-01
-9.66213346e-01 2.76196450e-01 -5.23842454e-01 -7.06731558e-01
3.64491552e-01 4.74253714e-01 -9.86034214e-01 -4.46764767e-01
-1.07436419e+00 7.17328608e-01 4.75031227e-01 2.90975839e-01
-4.80097443e-01 -4.51812804e-01 -1.17908907e+00 -1.25883818e-01
5.10336578e-01 -9.45366442e-01 1.02608716e+00 -8.09495986e-01
-9.86890614e-01 1.26311874e+00 -5.30564487e-01 -4.73769978e-02
7.01511979e-01 -7.72641897e-02 -4.37900841e-01 3.93783391e-01
5.96388519e-01 9.73188877e-01 8.05266559e-01 -1.39844716e+00
-7.13452876e-01 -7.55568683e-01 -1.41579211e-01 4.99518186e-01
-5.23173869e-01 2.91013926e-01 -1.08650458e+00 -7.86637843e-01
1.59398958e-01 -1.08813810e+00 1.58050179e-01 1.16985708e-01
-6.53584778e-01 -5.33810019e-01 4.32448983e-01 -8.46306205e-01
8.07806849e-01 -2.23916578e+00 1.58757895e-01 3.09385061e-01
4.57850486e-01 -1.05566811e-02 -3.68303657e-01 4.17687446e-01
-1.13814063e-01 -1.99077930e-02 4.50216345e-02 -8.24828207e-01
-8.69862512e-02 -1.40788451e-01 2.10902467e-01 7.47974455e-01
-1.45777747e-01 8.94057333e-01 -8.26949835e-01 -9.63207781e-01
4.24334295e-02 3.37200314e-01 -7.23787025e-02 8.27283189e-02
4.55511034e-01 5.79118311e-01 -4.97046888e-01 8.96634102e-01
8.16425979e-01 -2.24705160e-01 1.23221902e-02 -6.34400308e-01
-1.37083391e-02 -2.73098052e-01 -9.46926057e-01 1.75151205e+00
-1.41173437e-01 4.04518962e-01 6.78384257e-03 -7.58047998e-01
7.19519019e-01 1.62027463e-01 5.67423940e-01 -9.51662302e-01
1.89077646e-01 1.43127084e-01 -4.16630983e-01 -4.46096927e-01
5.20682752e-01 -3.29625010e-02 -5.89312315e-02 1.38715953e-01
1.88655555e-02 5.10898709e-01 4.82540876e-01 3.56219232e-01
5.92267752e-01 4.94537689e-02 1.92191660e-01 8.16693380e-02
8.48672330e-01 -2.28450805e-01 8.43940496e-01 7.40138769e-01
-4.29390252e-01 8.22188377e-01 1.28398523e-01 -2.66612023e-01
-1.05114245e+00 -9.81925309e-01 -1.67158306e-01 1.05294216e+00
7.01389909e-01 -5.58385313e-01 -6.73987627e-01 -7.46977329e-01
1.60734072e-01 1.48641557e-01 -7.08206534e-01 1.00988664e-01
-4.60374385e-01 -4.39835131e-01 7.39409924e-01 6.39303207e-01
1.04951012e+00 -6.95044935e-01 1.09861083e-01 -2.97672421e-01
-6.63045704e-01 -1.40953803e+00 -1.13067985e+00 -6.02931023e-01
-4.97584045e-01 -1.07326698e+00 -1.14024770e+00 -1.00096440e+00
9.95129347e-01 6.26570404e-01 8.22413445e-01 1.28023922e-01
-2.99586385e-01 6.87068760e-01 -3.25460285e-01 -6.89731613e-02
2.91888434e-02 1.51159003e-01 2.99509823e-01 2.87221879e-01
4.48725134e-01 -2.98624784e-01 -7.79758632e-01 6.03895128e-01
-6.60376906e-01 2.32440963e-01 6.05532944e-01 8.25227380e-01
8.38118315e-01 -3.76855209e-02 3.76094431e-01 -5.11370599e-01
4.00294095e-01 -2.24258959e-01 -3.87670636e-01 4.99153376e-01
-3.62970173e-01 -2.74100929e-01 2.78893977e-01 -3.55480909e-01
-1.19653404e+00 3.13843459e-01 7.72652626e-02 -5.43045640e-01
-1.89339146e-01 3.72358829e-01 -3.54695439e-01 -2.42449626e-01
1.24953747e-01 4.71117616e-01 9.08517912e-02 -5.23753762e-01
2.98282266e-01 7.47981727e-01 1.04029286e+00 -4.38244313e-01
1.14211237e+00 5.85209489e-01 7.48015754e-03 -5.67455709e-01
-1.05028534e+00 -9.80530024e-01 -8.92173588e-01 -3.06743562e-01
9.30592477e-01 -1.28329718e+00 -7.52428412e-01 6.17700815e-01
-9.22063112e-01 2.40807921e-01 1.09252669e-01 4.22837675e-01
-3.33833665e-01 7.30344653e-01 -5.54554999e-01 -5.21861255e-01
-5.60450852e-01 -8.19208384e-01 1.28498697e+00 5.66418409e-01
-8.73336270e-02 -6.42820537e-01 -1.04712144e-01 9.34462607e-01
1.54992426e-02 2.58136671e-02 3.70612979e-01 -6.72341406e-01
-3.44685763e-01 -3.84151250e-01 -7.43743300e-01 9.39649865e-02
2.78186947e-01 -4.78040963e-01 -7.28863180e-01 -5.45531511e-01
-3.11930895e-01 -2.64247686e-01 7.27834165e-01 1.08785845e-01
1.23496425e+00 -3.02824080e-01 -5.93183935e-01 6.65204704e-01
1.12272060e+00 -1.56160399e-01 4.12699133e-01 5.07055819e-01
1.13680804e+00 8.00246418e-01 6.65870786e-01 3.31978947e-01
8.38621616e-01 9.56759393e-01 5.84589876e-02 -1.09221049e-01
-3.48330557e-01 -5.71133375e-01 1.51113048e-01 5.75621545e-01
-1.00996576e-01 -1.97081745e-01 -8.04467440e-01 7.74819314e-01
-1.81535375e+00 -1.09152019e+00 3.04470304e-02 2.17924666e+00
7.09571660e-01 -3.99513006e-01 3.31962079e-01 -4.79747385e-01
1.23543298e+00 3.14668156e-02 -6.18105710e-01 4.29133654e-01
-1.75646722e-01 -4.69928354e-01 6.29462600e-01 2.80481856e-02
-1.43614101e+00 8.79986167e-01 4.83414412e+00 1.06605434e+00
-6.82098746e-01 1.10127002e-01 4.55617368e-01 -4.10652980e-02
6.30192235e-02 -2.89382339e-01 -1.14068007e+00 5.47753692e-01
4.43335921e-01 -3.80822271e-01 3.21482778e-01 6.34964108e-01
5.45850769e-02 -6.02378882e-02 -1.19828212e+00 1.55263066e+00
6.77762747e-01 -9.45519924e-01 1.88929394e-01 8.71290565e-02
5.66540122e-01 -1.41576990e-01 5.53412400e-02 -7.69713372e-02
-1.57618120e-01 -7.21826017e-01 7.60384738e-01 4.59354609e-01
9.95202005e-01 -8.45842898e-01 8.84323776e-01 1.12127043e-01
-1.60024226e+00 2.25425679e-02 -2.80506581e-01 5.49732268e-01
1.15662068e-01 2.00309455e-01 -5.53117335e-01 8.47880781e-01
8.54397655e-01 9.28666711e-01 -8.83165777e-01 1.25169253e+00
-7.83670321e-02 -5.39816208e-02 -1.46707594e-01 3.44706804e-01
-1.27928123e-01 -7.00121373e-02 5.48549652e-01 1.11835957e+00
1.58085316e-01 9.82108042e-02 2.94346809e-01 7.57578969e-01
-4.57581937e-01 2.50839949e-01 -4.52654362e-01 1.87487468e-01
5.44025540e-01 1.31112075e+00 -6.61187530e-01 -2.94111013e-01
-4.66941923e-01 1.34851635e+00 2.73706496e-01 2.26107046e-01
-6.44291997e-01 -2.62747258e-01 5.67616701e-01 5.87996244e-02
3.76304686e-02 4.25470099e-02 2.23427266e-01 -1.38034284e+00
2.82629639e-01 -8.63662958e-01 5.78970492e-01 -7.63051927e-01
-1.89672804e+00 4.43487704e-01 3.13183069e-01 -1.40126991e+00
-6.92460313e-02 -2.08697662e-01 -1.16773255e-01 8.36921334e-01
-1.41293693e+00 -1.67626595e+00 -6.67013049e-01 8.63306165e-01
6.05504811e-01 -2.57993639e-01 4.48049784e-01 5.99802494e-01
-8.41582060e-01 1.11257589e+00 1.38772309e-01 7.34980464e-01
1.12528479e+00 -8.49925578e-01 4.05060232e-01 7.87452340e-01
-4.88594659e-02 8.84048343e-01 3.19214880e-01 -8.95575523e-01
-1.39511740e+00 -1.17593825e+00 9.33442116e-01 -3.03887516e-01
3.96572620e-01 -3.66743863e-01 -7.69108891e-01 5.98276079e-01
4.08313833e-02 1.00878380e-01 7.09761798e-01 8.79365299e-03
-5.71264505e-01 -2.52281874e-01 -1.15425205e+00 6.60472989e-01
1.45992625e+00 -7.64164329e-01 -4.40190881e-01 3.29198211e-01
2.73936778e-01 -5.87362468e-01 -9.52422142e-01 3.94225746e-01
8.60814095e-01 -7.40776539e-01 1.16598570e+00 -2.11122528e-01
2.79715568e-01 -3.29458743e-01 7.41111264e-02 -1.00644743e+00
-2.70905793e-01 -1.81449026e-01 2.67385215e-01 1.73284698e+00
-5.57476990e-02 -6.67706370e-01 7.26565301e-01 9.39456820e-01
2.12746173e-01 -4.07097101e-01 -9.72058654e-01 -9.62760627e-01
-2.83527523e-01 9.78823528e-02 4.50873226e-01 9.13843632e-01
-1.01396210e-01 2.50423789e-01 -6.54259384e-01 2.01605827e-01
1.04886246e+00 1.16732880e-01 8.72793674e-01 -1.08862615e+00
1.58910215e-01 -2.29117215e-01 -6.04544282e-01 -1.00345612e+00
4.57839906e-01 -9.97146606e-01 2.50901058e-02 -1.34529328e+00
1.04808474e+00 -3.09060842e-01 -1.14048332e-01 5.92843473e-01
-2.44929314e-01 5.96899629e-01 3.68148148e-01 6.08301282e-01
-8.23032141e-01 7.10600197e-01 1.17176247e+00 -2.84859091e-01
1.32497847e-01 -1.82378054e-01 -6.52126729e-01 6.34108543e-01
7.25016296e-01 -4.94972795e-01 -1.47006735e-01 -4.03279752e-01
-1.53407261e-01 -1.25838548e-01 7.37511516e-01 -8.41383040e-01
5.49070239e-01 3.52355242e-02 7.25252748e-01 -9.16840553e-01
3.65188718e-01 -7.15558350e-01 1.67781219e-01 1.89068280e-02
-3.50461483e-01 2.34676465e-01 8.10199231e-02 7.15412199e-01
-3.61773252e-01 -1.87063500e-01 5.67329586e-01 -1.38380408e-01
-7.81371236e-01 6.03564203e-01 3.24969105e-02 -9.07622501e-02
7.97694564e-01 -3.45240176e-01 -3.85552168e-01 -4.62179899e-01
-4.60337251e-01 6.02543056e-01 6.20574415e-01 6.12858474e-01
7.10020602e-01 -1.51946044e+00 -7.35034764e-01 -1.95904784e-02
5.07558882e-01 -2.64326423e-01 6.58314526e-01 9.12524521e-01
-1.49822399e-01 4.39124912e-01 -2.35163182e-01 -4.39326257e-01
-1.90804994e+00 4.63868886e-01 4.27747846e-01 2.06895787e-02
-6.95728421e-01 8.43157232e-01 5.13168395e-01 -5.16017795e-01
2.76413471e-01 4.77709591e-01 -2.24172756e-01 2.17836231e-01
5.70777833e-01 3.77945155e-01 -2.39364773e-01 -1.34801555e+00
-5.88843882e-01 8.66848469e-01 -3.36103022e-01 -2.98328456e-02
1.09244335e+00 -6.08305275e-01 -3.86905819e-01 -6.28810823e-02
1.34006584e+00 -1.58859435e-02 -9.78120625e-01 -5.69945812e-01
-1.29497275e-01 -8.90398681e-01 -2.40548626e-01 -6.21802688e-01
-1.23067153e+00 4.10123080e-01 8.15531611e-01 -3.69487792e-01
1.13603377e+00 2.25340307e-01 8.90179694e-01 2.56628901e-01
3.94119710e-01 -9.90263462e-01 1.12868026e-01 2.04559714e-01
8.76206934e-01 -1.47844350e+00 2.68081099e-01 -6.38164103e-01
-6.46523297e-01 8.81622612e-01 8.96191776e-01 3.92653078e-01
1.68145657e-01 -3.14493358e-01 -1.79808326e-02 -1.87950030e-01
-2.29835548e-02 -4.17019486e-01 5.58167458e-01 5.78464866e-01
2.46155724e-01 -7.73685724e-02 -2.68813521e-01 3.26707006e-01
-1.49419025e-01 -5.72510436e-02 1.20076157e-01 6.87401652e-01
1.41390845e-01 -1.29669011e+00 -5.61777353e-01 3.98629487e-01
-4.11748558e-01 -4.92516793e-02 -6.22293234e-01 7.77532339e-01
1.93122655e-01 8.76440644e-01 -6.93266466e-02 -4.49812263e-01
2.82178938e-01 -1.33521125e-01 4.27098364e-01 -4.05943662e-01
-4.03204232e-01 8.06366652e-02 2.06357300e-01 -4.48097408e-01
-7.09331036e-01 -9.47298884e-01 -9.40069497e-01 -3.68676454e-01
-3.56491387e-01 -4.95757759e-02 4.04603392e-01 8.49837720e-01
4.68635589e-01 -2.38388847e-03 4.84342337e-01 -9.46259618e-01
-3.43173236e-01 -8.13952804e-01 -6.09749019e-01 6.75547063e-01
2.11508229e-01 -8.31408560e-01 -1.15014143e-01 1.14116430e-01] | [14.643595695495605, 0.8452453017234802] |
fde697cd-9bbe-4c6c-9bc4-8a9a15ac3ad8 | investigating-glyph-phonetic-information-for | 2212.04068 | null | https://arxiv.org/abs/2212.04068v3 | https://arxiv.org/pdf/2212.04068v3.pdf | Investigating Glyph Phonetic Information for Chinese Spell Checking: What Works and What's Next | While pre-trained Chinese language models have demonstrated impressive performance on a wide range of NLP tasks, the Chinese Spell Checking (CSC) task remains a challenge. Previous research has explored using information such as glyphs and phonetics to improve the ability to distinguish misspelled characters, with good results. However, the generalization ability of these models is not well understood: it is unclear whether they incorporate glyph-phonetic information and, if so, whether this information is fully utilized. In this paper, we aim to better understand the role of glyph-phonetic information in the CSC task and suggest directions for improvement. Additionally, we propose a new, more challenging, and practical setting for testing the generalizability of CSC models. All code is made publicly available. | ['Xipeng Qiu', 'Hang Yan', 'Yanjun Zheng', 'Xiaotian Zhang'] | 2022-12-08 | null | null | null | null | ['chinese-spell-checking'] | ['natural-language-processing'] | [ 2.36837089e-01 -4.11032468e-01 -3.31918329e-01 -3.54276866e-01
-1.02452409e+00 -8.26591611e-01 5.23798704e-01 2.22456425e-01
-5.37376046e-01 4.62578207e-01 5.07977903e-01 -7.68480241e-01
3.04452360e-01 -3.91391546e-01 -5.74015498e-01 -4.01912391e-01
3.24274510e-01 1.04320906e-01 1.21993661e-01 5.50450794e-02
5.20063996e-01 3.28138560e-01 -1.31861997e+00 5.77963471e-01
1.33048081e+00 5.54649234e-01 5.13761878e-01 5.28961599e-01
-2.02610120e-01 5.43431222e-01 -7.12336004e-01 -3.30691844e-01
7.36042038e-02 -4.63219374e-01 -7.00986803e-01 -1.66275620e-01
4.45852369e-01 -3.66525650e-01 1.58840030e-01 1.06976318e+00
3.04237694e-01 1.05942234e-01 5.39164960e-01 -7.45075643e-01
-1.02569401e+00 7.68091261e-01 -1.01127774e-01 2.65889376e-01
3.95437211e-01 7.38479719e-02 1.11716282e+00 -7.15113044e-01
4.37942505e-01 1.11773813e+00 5.54606736e-01 4.92047101e-01
-8.43313634e-01 -8.60205531e-01 3.67907882e-01 1.29429653e-01
-1.28173256e+00 -2.37123936e-01 4.24105853e-01 -3.21396530e-01
1.05037200e+00 2.76169121e-01 4.53076720e-01 1.14239693e+00
4.47018072e-02 1.10661948e+00 1.42592239e+00 -6.48167312e-01
3.09292451e-02 2.44104967e-01 3.17701012e-01 2.86292702e-01
2.80357331e-01 -1.27321556e-01 -4.66366976e-01 9.88212377e-02
5.26467979e-01 -9.58262607e-02 -2.60733902e-01 1.70400381e-01
-8.31941843e-01 7.93725729e-01 3.56960408e-02 5.94392836e-01
-4.64668460e-02 -1.45440742e-01 1.83190867e-01 8.90589729e-02
4.86358047e-01 1.00532401e+00 -6.74989343e-01 -7.52799332e-01
-1.07030833e+00 1.17288575e-01 8.65243971e-01 9.02887464e-01
5.35242200e-01 1.37955546e-01 -1.13497071e-01 8.14697504e-01
7.16985464e-02 4.06116158e-01 5.65351009e-01 -8.00016463e-01
6.62254155e-01 4.91640180e-01 1.13533512e-01 -8.28748345e-01
-3.22448850e-01 -3.16825569e-01 -5.50245456e-02 -2.61649311e-01
6.20916843e-01 -2.84705102e-01 -7.95830429e-01 1.59297645e+00
-3.53425086e-01 2.19764605e-01 -9.64157060e-02 9.19168830e-01
4.75768119e-01 5.99310040e-01 1.87585890e-01 1.23869263e-01
1.20708978e+00 -1.04128373e+00 -6.08075619e-01 -6.33769333e-01
8.80201757e-01 -9.29709971e-01 1.53890240e+00 4.24613416e-01
-1.01334047e+00 -5.90362549e-01 -8.79345417e-01 -2.95853406e-01
-4.58312988e-01 3.95388663e-01 6.55154586e-01 1.11000979e+00
-8.62294912e-01 4.22923952e-01 -1.03351462e+00 -3.72805774e-01
7.30967820e-02 1.25889331e-02 -1.61941454e-01 -3.15605700e-01
-1.19843388e+00 9.77104366e-01 2.78022975e-01 -8.13804716e-02
-4.88943189e-01 -6.98337138e-01 -1.06706643e+00 2.75542676e-01
2.91479260e-01 2.87225805e-02 1.33305311e+00 -8.87467802e-01
-1.39944756e+00 5.21370411e-01 -4.95940715e-01 -3.34324062e-01
2.38063306e-01 -3.68285626e-01 -4.91329789e-01 -4.64151753e-03
-8.44349619e-03 6.15811884e-01 3.59197408e-01 -9.92404819e-01
-7.91952729e-01 -4.56878513e-01 -6.64793849e-02 3.35780203e-01
-5.90370476e-01 3.96934390e-01 -6.44732475e-01 -8.10822427e-01
-1.11603811e-01 -1.14688790e+00 2.31960379e-02 -3.46876144e-01
-8.89684856e-02 -1.21091209e-01 5.77699900e-01 -1.06807673e+00
1.55718720e+00 -2.44184899e+00 -5.74563205e-01 -3.67950238e-02
-4.82678831e-01 6.54683411e-01 -2.49174356e-01 6.48623168e-01
4.09623891e-01 6.39896989e-01 -2.03898117e-01 -3.23430777e-01
-1.17313126e-02 1.55411130e-02 -3.70226204e-01 5.99283315e-02
3.60029757e-01 9.27979469e-01 -8.39082956e-01 -1.84716344e-01
9.19172317e-02 5.09447217e-01 -5.48127890e-01 -2.33525857e-01
-1.88465849e-01 2.44909093e-01 -3.58641326e-01 5.60678363e-01
7.28396952e-01 -1.17360651e-01 2.33601987e-01 4.36342299e-01
-4.49800253e-01 1.01186621e+00 -1.04461098e+00 1.33373797e+00
-3.04025888e-01 8.91029775e-01 -7.22468346e-02 -5.61227560e-01
8.96176636e-01 1.28598794e-01 -2.86808442e-02 -9.63969767e-01
-2.02621266e-01 2.40781680e-01 3.17645162e-01 -3.83959711e-01
8.74614239e-01 4.53386754e-02 -1.09702945e-01 3.34476024e-01
-6.58487439e-01 -9.61285681e-02 2.53666848e-01 -3.27340029e-02
8.50746036e-01 2.57399589e-01 1.64183930e-01 -2.00769052e-01
3.10410023e-01 3.72493416e-01 6.30433917e-01 9.85843539e-01
-3.59047174e-01 9.85199928e-01 3.45357358e-01 1.85269326e-01
-7.06792891e-01 -9.12350357e-01 -2.25060150e-01 1.02870882e+00
1.02131739e-01 -6.78472638e-01 -8.17117393e-01 -5.53795159e-01
6.11775331e-02 1.17503083e+00 -2.74674386e-01 -4.14896049e-02
-7.12376058e-01 -5.59823334e-01 8.60116422e-01 9.01136816e-01
2.98806608e-01 -9.88319695e-01 -4.00341094e-01 1.71843082e-01
-2.19649121e-01 -1.42226768e+00 -7.38627791e-01 -3.74247618e-02
-7.51233697e-01 -8.98227572e-01 -5.64891696e-01 -9.52599347e-01
3.54451835e-01 5.43845534e-01 7.15072691e-01 1.58777818e-01
1.51623145e-01 5.48195355e-02 -5.29801369e-01 -6.61891639e-01
-5.38560748e-01 2.86121875e-01 -3.82198244e-01 -4.05173272e-01
8.12496781e-01 1.90851361e-01 -3.56335968e-01 2.03117937e-01
-7.69469559e-01 1.40680671e-02 5.63531518e-01 5.05168498e-01
2.66538143e-01 -3.15125100e-02 3.81128401e-01 -9.72412705e-01
7.78028071e-01 -1.16852321e-01 -3.98386776e-01 1.08201616e-01
-6.29575193e-01 -8.97251889e-02 7.68318772e-01 -3.82481366e-01
-1.22319734e+00 -1.21690430e-01 -4.95869666e-01 2.07540706e-01
-4.20606762e-01 7.93203413e-01 -8.86496156e-02 1.28307313e-01
2.89961636e-01 2.09840328e-01 -1.70528471e-01 -7.84401238e-01
2.71846075e-04 8.20943117e-01 3.05390537e-01 -6.52794898e-01
3.24976891e-01 2.76207209e-01 -5.80364704e-01 -9.34356689e-01
-7.08887458e-01 -5.95967829e-01 -4.05708104e-01 2.69988656e-01
8.61434340e-01 -1.05704892e+00 -2.59696037e-01 3.73572469e-01
-8.53878796e-01 -6.50983930e-01 4.03100044e-01 7.60565937e-01
-7.14977756e-02 7.00791359e-01 -8.68575275e-01 -7.44021893e-01
-1.17231347e-01 -1.36083853e+00 8.93384099e-01 2.25511059e-01
-5.19184411e-01 -8.78086567e-01 -2.68176705e-01 6.19449794e-01
3.84659827e-01 -3.13639134e-01 1.11195123e+00 -8.42917800e-01
-5.02345741e-01 -2.23735228e-01 -1.90220445e-01 4.95929122e-01
1.49284080e-01 1.21643797e-01 -8.28988492e-01 -1.78276896e-01
-1.35645494e-01 -5.35199884e-03 9.31895912e-01 3.66747469e-01
1.18531096e+00 -1.99117333e-01 -2.97706366e-01 4.54386979e-01
1.11744606e+00 4.65148479e-01 6.59119248e-01 5.01749873e-01
5.06515622e-01 5.50914943e-01 6.92350745e-01 2.18458369e-01
7.50061631e-01 5.92839777e-01 -5.49404919e-02 2.20149696e-01
-1.90553173e-01 -3.98759693e-01 4.67383415e-01 7.87643433e-01
2.95047432e-01 -4.13483739e-01 -1.10935020e+00 6.54264331e-01
-1.48072410e+00 -6.76499546e-01 -1.73678964e-01 2.12767911e+00
8.40933979e-01 3.68019044e-01 -1.36038497e-01 1.00952260e-01
5.48959613e-01 1.44304588e-01 -1.09971754e-01 -6.41957223e-01
-2.96453953e-01 1.54696226e-01 3.33255291e-01 4.91215289e-01
-8.81098151e-01 1.26254046e+00 6.66714573e+00 5.69669604e-01
-1.43433821e+00 -1.43733114e-01 5.68217516e-01 1.98534802e-01
-4.70595628e-01 3.29360813e-01 -1.12136328e+00 5.92723429e-01
7.99212158e-01 1.25909507e-01 3.83131206e-01 7.67522395e-01
4.70148623e-01 -2.12732017e-01 -9.40639198e-01 4.43500221e-01
2.74220526e-01 -1.08357906e+00 -9.55858752e-02 -4.50579897e-02
8.66954684e-01 1.02508590e-02 1.87861860e-01 5.33068895e-01
2.26038769e-01 -8.99076998e-01 7.62414634e-01 8.04642513e-02
4.50908750e-01 -4.91443187e-01 5.00482738e-01 4.24171239e-01
-9.98195946e-01 -1.25292242e-01 -1.43139794e-01 -5.09383917e-01
7.46984705e-02 1.48109272e-01 -9.77317691e-01 2.69979030e-01
6.43482327e-01 6.52200699e-01 -1.10925031e+00 1.20204246e+00
-4.15272683e-01 1.18934858e+00 -3.95801701e-02 -2.74914771e-01
5.08989453e-01 -5.45787439e-02 1.59041986e-01 1.65037453e+00
4.89092112e-01 -1.20633662e-01 7.14795962e-02 7.50484645e-01
2.05548361e-01 2.49299690e-01 -3.21977943e-01 -5.67165494e-01
7.35926092e-01 7.96350300e-01 -8.01046252e-01 -1.08419061e-01
-8.57188284e-01 7.94414341e-01 3.01955193e-01 5.96257746e-01
-5.80673039e-01 -1.49211302e-01 8.38174701e-01 -3.51415039e-03
3.60790640e-01 -5.61274052e-01 -7.65920162e-01 -1.22438562e+00
3.36038284e-02 -1.08951604e+00 3.93913031e-01 -7.95433879e-01
-1.01701295e+00 1.45788699e-01 -2.38589853e-01 -8.90657306e-01
-2.11693227e-01 -8.22041035e-01 -5.70065558e-01 8.73735666e-01
-1.66805029e+00 -9.09764111e-01 -1.92809954e-01 1.79781348e-01
7.23744035e-01 1.73078418e-01 6.75029635e-01 9.64524448e-02
-6.48144722e-01 8.41670811e-01 3.08255702e-01 4.49227750e-01
1.01619744e+00 -1.16908193e+00 6.79545403e-01 1.15778196e+00
2.40674675e-01 9.92927134e-01 5.43260694e-01 -9.31531072e-01
-1.33856869e+00 -9.21448350e-01 1.23037159e+00 -4.79797065e-01
4.62472171e-01 -4.38822716e-01 -1.21320927e+00 7.11953640e-01
9.39509571e-02 -5.32627225e-01 6.37783170e-01 1.85404822e-01
-4.23161775e-01 1.90259293e-01 -6.83803141e-01 8.58073711e-01
6.52286232e-01 -7.28975296e-01 -4.84833926e-01 -1.09656997e-01
5.33321798e-01 -4.24260557e-01 -5.87344289e-01 3.77550930e-01
5.29929399e-01 -7.95312464e-01 6.74495578e-01 -3.80128801e-01
3.51185769e-01 -1.11012883e-01 -1.51892513e-01 -1.31541312e+00
-3.28673303e-01 -2.69785345e-01 3.84736478e-01 1.19755352e+00
7.51033902e-01 -5.77998936e-01 7.39656448e-01 8.40565503e-01
-5.88661253e-01 -6.08688056e-01 -4.62733537e-01 -8.47610891e-01
4.32398677e-01 -6.14872932e-01 6.22305751e-01 1.00874209e+00
2.49735028e-01 1.34399951e-01 -2.62942731e-01 2.55410403e-01
-8.77071768e-02 6.82929233e-02 4.91974652e-01 -9.14354205e-01
-2.45762661e-01 -7.71197736e-01 8.09530094e-02 -1.08377326e+00
3.26474965e-01 -9.29030597e-01 5.97842522e-02 -1.70147824e+00
5.28496839e-02 -4.06758517e-01 -9.53975990e-02 3.26300740e-01
-7.58641839e-01 8.81505199e-03 6.71679258e-01 5.25255986e-02
-4.99241799e-01 1.56048119e-01 1.11419976e+00 1.20072350e-01
-2.86169797e-01 1.22010477e-01 -1.03196442e+00 4.68550891e-01
9.75041509e-01 -2.51970768e-01 -1.68415576e-01 -8.64794016e-01
-3.89809012e-02 -2.15447262e-01 -4.49865796e-02 -8.08954477e-01
2.22341537e-01 -2.67756730e-01 4.26367253e-01 -6.74019158e-01
3.49125087e-01 -5.19754648e-01 -2.69676596e-01 2.82889962e-01
-4.47402567e-01 4.73656595e-01 5.70151806e-01 2.11462021e-01
-2.13547632e-01 -4.78024006e-01 4.88518894e-01 -1.27405554e-01
-8.06469560e-01 -2.04652622e-01 -7.11750209e-01 2.94310302e-01
6.93039358e-01 -3.74830395e-01 -3.89560699e-01 -4.55659240e-01
-3.85904349e-02 1.19639106e-01 8.03204000e-01 8.31426740e-01
3.30951929e-01 -8.86196613e-01 -5.85376918e-01 3.48244727e-01
3.54212672e-02 -4.96781379e-01 7.37994760e-02 6.16806924e-01
-5.19429207e-01 7.63575375e-01 4.31420915e-02 -2.77733654e-01
-1.15805423e+00 4.88955855e-01 8.49088579e-02 -6.11703210e-02
-4.47423488e-01 5.55825531e-01 3.99647355e-02 -3.56470317e-01
3.45729440e-01 -3.52354765e-01 -1.95312962e-01 -2.11474657e-01
6.11969888e-01 3.70718062e-01 8.14069584e-02 -5.26730597e-01
-4.52213198e-01 3.41843396e-01 -2.50011951e-01 -1.37740746e-01
8.69385719e-01 -1.73755288e-01 7.36112446e-02 4.04446930e-01
1.21514833e+00 3.73012781e-01 -1.16839170e+00 -7.29776025e-02
4.64315861e-01 -4.55634296e-01 -2.87118912e-01 -8.83641183e-01
-5.72721839e-01 9.68635142e-01 1.58715889e-01 -9.47656259e-02
9.21181500e-01 -2.65832365e-01 8.24371517e-01 2.22709313e-01
4.59793247e-02 -1.27828979e+00 -2.05427445e-02 9.95687842e-01
3.39653105e-01 -1.27306056e+00 -3.40780586e-01 -3.48791510e-01
-9.59562063e-01 9.81715083e-01 7.11911738e-01 1.15039267e-01
5.72387278e-01 4.81683195e-01 2.47693479e-01 3.92686903e-01
-6.18388712e-01 -3.06654602e-01 4.00470108e-01 4.63727176e-01
9.23516691e-01 1.55344039e-01 -5.31855047e-01 7.85972655e-01
-5.44079483e-01 -1.67591661e-01 5.93908429e-01 9.06351745e-01
-4.19522226e-01 -1.14864779e+00 -2.85098284e-01 5.19973934e-01
-8.00178051e-01 -5.76792717e-01 -5.78321338e-01 8.12275410e-01
-4.34935205e-02 1.29843605e+00 2.30703041e-01 -2.57877409e-01
7.40344971e-02 3.16691399e-01 1.56550989e-01 -1.00642419e+00
-7.49778271e-01 1.62383512e-01 2.30511248e-01 -2.59473950e-01
3.43445018e-02 -1.07850587e+00 -1.21464765e+00 -1.43383056e-01
-2.90550232e-01 6.31382614e-02 6.41284823e-01 1.10725808e+00
3.55812609e-01 1.33374438e-01 1.45885304e-01 -1.56221613e-01
-5.57788014e-01 -9.14345741e-01 -2.31109858e-01 4.34687674e-01
1.52661175e-01 -9.26287919e-02 -1.87444374e-01 -1.74413383e-01] | [10.923233032226562, 10.61236572265625] |
f79c4a28-8467-49f1-b1c7-637abaefd115 | deploying-offline-reinforcement-learning-with | 2303.07046 | null | https://arxiv.org/abs/2303.07046v1 | https://arxiv.org/pdf/2303.07046v1.pdf | Deploying Offline Reinforcement Learning with Human Feedback | Reinforcement learning (RL) has shown promise for decision-making tasks in real-world applications. One practical framework involves training parameterized policy models from an offline dataset and subsequently deploying them in an online environment. However, this approach can be risky since the offline training may not be perfect, leading to poor performance of the RL models that may take dangerous actions. To address this issue, we propose an alternative framework that involves a human supervising the RL models and providing additional feedback in the online deployment phase. We formalize this online deployment problem and develop two approaches. The first approach uses model selection and the upper confidence bound algorithm to adaptively select a model to deploy from a candidate set of trained offline RL models. The second approach involves fine-tuning the model in the online deployment phase when a supervision signal arrives. We demonstrate the effectiveness of these approaches for robot locomotion control and traffic light control tasks through empirical validation. | ['Peilin Zhao', 'Deheng Ye', 'Lanqing Li', 'Liu Liu', 'Ke Xu', 'Ziniu Li'] | 2023-03-13 | null | null | null | null | ['offline-rl'] | ['playing-games'] | [ 2.75343925e-01 2.43627146e-01 -3.96849781e-01 -2.77899563e-01
-7.08466470e-01 -3.97177428e-01 3.36076289e-01 9.57580805e-02
-7.22865224e-01 9.80630577e-01 -5.75562775e-01 -5.21705091e-01
-2.12782577e-01 -7.22477376e-01 -6.86699092e-01 -9.53935206e-01
-2.75859773e-01 8.23010564e-01 4.48263109e-01 -2.11289406e-01
4.02622193e-01 5.91331840e-01 -1.68195152e+00 -2.09156349e-01
9.34866965e-01 8.68841529e-01 6.50777638e-01 8.07529449e-01
2.23908812e-01 6.73276246e-01 -7.11200833e-01 2.81414002e-01
4.94666308e-01 -4.06571984e-01 -5.14007926e-01 2.66213268e-01
-1.97965771e-01 -4.78936702e-01 2.33424827e-01 7.52211750e-01
4.99496073e-01 5.94206572e-01 2.84465909e-01 -1.54668689e+00
2.19065979e-01 4.66506302e-01 -5.01620471e-01 8.80398154e-02
2.50764936e-01 5.96562266e-01 7.12479532e-01 -1.35133550e-01
3.89864147e-01 1.17580080e+00 3.41327488e-01 6.60043776e-01
-1.26180232e+00 -6.71396792e-01 3.95077199e-01 9.90240574e-02
-1.08711612e+00 -4.23882991e-01 5.86661637e-01 -2.97653109e-01
8.07080209e-01 -2.30928764e-01 7.13765979e-01 8.32917750e-01
-1.18840598e-02 6.93022907e-01 1.28719532e+00 -5.61880052e-01
7.60142326e-01 5.33756316e-02 -3.01404536e-01 8.57752383e-01
3.31805795e-01 5.93466520e-01 -3.18012744e-01 -2.84386754e-01
6.72294557e-01 -4.45267558e-01 1.76481321e-01 -4.41995829e-01
-7.46764958e-01 8.38615716e-01 2.28593916e-01 -1.12818912e-01
-7.57628202e-01 3.02112818e-01 3.10704887e-01 4.34763253e-01
8.09772611e-02 7.05212355e-01 -3.40048462e-01 -1.37145951e-01
-7.47511923e-01 4.57751989e-01 6.85451269e-01 6.49313450e-01
7.91313291e-01 1.24066778e-01 -6.82496503e-02 7.77704775e-01
2.47779071e-01 3.75250071e-01 3.17964375e-01 -1.22333944e+00
5.91454446e-01 2.89415717e-01 6.35703981e-01 -5.89469016e-01
-3.13926548e-01 8.14053193e-02 -4.72932570e-02 5.59666097e-01
4.82585341e-01 -5.76187730e-01 -7.30069160e-01 1.45753193e+00
7.39738524e-01 1.99606225e-01 2.11848076e-02 9.56623673e-01
-3.22861791e-01 6.27176225e-01 3.37211221e-01 -3.41973960e-01
7.24641383e-01 -9.55926120e-01 -5.05695760e-01 -3.50322038e-01
4.06090170e-01 -3.14761132e-01 1.10917139e+00 7.00188577e-01
-7.32617140e-01 -4.78520453e-01 -1.29924428e+00 6.68917537e-01
-1.29019439e-01 4.07533854e-01 4.67012405e-01 3.94908130e-01
-8.25449109e-01 8.50875616e-01 -1.02633703e+00 -3.90862793e-01
1.73303947e-01 6.14093721e-01 7.09509328e-02 -3.43316048e-02
-1.00657535e+00 9.52772617e-01 4.05580372e-01 1.98296532e-01
-1.37177384e+00 8.07901323e-02 -5.96624076e-01 -1.38223410e-01
7.71089494e-01 -2.73980618e-01 1.64192188e+00 -1.03883886e+00
-2.07371259e+00 2.75931835e-01 2.28449211e-01 -8.47661436e-01
6.21085525e-01 -2.34951213e-01 8.96362886e-02 2.07306951e-01
7.83873796e-02 5.88802040e-01 1.23541248e+00 -1.38068914e+00
-1.12033617e+00 -1.29888788e-01 5.37417352e-01 4.68853265e-01
-1.17349029e-01 -2.41787270e-01 -5.65233491e-02 -1.52558148e-01
-1.89024463e-01 -1.19206774e+00 -6.35417104e-01 -1.34047270e-01
-1.30163589e-02 -2.89695859e-01 6.78955257e-01 -2.12143138e-01
1.16172588e+00 -1.82050860e+00 -1.35346442e-01 2.75858611e-01
-4.68777984e-01 2.51653373e-01 -1.21147417e-01 5.11262178e-01
4.43798751e-01 -1.72000006e-01 -2.59754295e-03 -2.61358172e-01
3.10127120e-02 5.41776538e-01 -3.78825277e-01 3.40530306e-01
1.47609279e-01 1.73459709e-01 -1.16193724e+00 -4.18866009e-01
2.53818750e-01 -8.99738222e-02 -5.49958169e-01 6.53492987e-01
-4.31639880e-01 6.07891619e-01 -8.00589621e-01 7.36544609e-01
1.34162948e-01 4.24289145e-02 4.96512949e-01 3.65156442e-01
-1.88750550e-01 3.03688854e-01 -1.21853387e+00 9.95006263e-01
-7.70465434e-01 4.20418590e-01 2.42830127e-01 -1.22703958e+00
7.86145627e-01 1.93058282e-01 6.73397720e-01 -4.99021590e-01
1.38549134e-01 1.16073586e-01 7.75256008e-02 -7.81526983e-01
3.62758338e-01 -1.56816989e-01 -1.22531932e-02 4.72135991e-01
-1.78892672e-01 -2.71826416e-01 3.04905921e-01 -2.54645914e-01
1.15401709e+00 5.40785551e-01 3.69297206e-01 1.15381941e-01
3.00843507e-01 2.09026858e-01 6.33680761e-01 1.02955818e+00
-5.59690952e-01 -1.55010626e-01 4.38748747e-01 -3.79360259e-01
-8.32895160e-01 -6.13110304e-01 2.94234514e-01 1.31349385e+00
1.48195982e-01 -8.90208781e-02 -7.98403978e-01 -1.02676177e+00
1.49526432e-01 8.96659672e-01 -5.01893342e-01 -3.55709463e-01
-7.28179932e-01 -4.06234533e-01 1.73000455e-01 4.67244953e-01
3.89298141e-01 -1.16701579e+00 -1.48978674e+00 3.94590408e-01
-3.88294384e-02 -9.50942159e-01 -1.11243390e-01 5.16465962e-01
-7.51088142e-01 -1.22343624e+00 -1.51023641e-01 -1.94032624e-01
9.24704313e-01 2.61559963e-01 6.36955917e-01 2.40629271e-01
-7.39745721e-02 3.66221279e-01 -4.69808906e-01 -4.94810373e-01
-5.41483462e-01 1.42987594e-02 3.93730581e-01 -1.00607254e-01
-4.06569131e-02 -1.19537085e-01 -4.79544938e-01 4.69228059e-01
-5.37589431e-01 -2.60850012e-01 5.23230195e-01 1.04394937e+00
6.41561329e-01 4.04173642e-01 7.95298278e-01 -8.01336050e-01
7.72356629e-01 -4.12090540e-01 -1.24344051e+00 2.93132424e-01
-7.82159328e-01 1.64648205e-01 9.35498357e-01 -7.44274259e-01
-8.79918694e-01 4.30891901e-01 1.49322346e-01 -4.85181600e-01
-1.09009266e-01 3.37796748e-01 3.86210307e-02 -1.11060984e-01
5.71150362e-01 -2.75033742e-01 6.89560026e-02 -2.31402159e-01
-8.74084234e-03 5.58135390e-01 9.40068215e-02 -8.88377368e-01
7.78250039e-01 8.80725756e-02 -7.74287358e-02 -8.07896256e-01
-7.79822826e-01 -3.19070876e-01 -3.97485107e-01 -5.77097297e-01
5.51814497e-01 -5.46779096e-01 -8.58446658e-01 2.14030147e-02
-7.51569390e-01 -9.60370839e-01 -2.37148374e-01 3.85601908e-01
-9.74898100e-01 -8.19088742e-02 -5.33630513e-02 -1.30817378e+00
5.70009798e-02 -1.27138281e+00 1.03489614e+00 4.51692104e-01
-9.54148918e-03 -6.06491923e-01 3.81592400e-02 1.67598262e-01
2.34434202e-01 2.55683392e-01 7.23874688e-01 -4.88784194e-01
-4.32590306e-01 -2.72815257e-01 4.58977282e-01 1.53640449e-01
2.98151304e-03 9.93598774e-02 -8.44208241e-01 -6.25607371e-01
-1.80060655e-01 -7.17466116e-01 5.31698525e-01 2.67341733e-01
1.12206113e+00 -3.28172356e-01 -5.08003175e-01 1.19644269e-01
1.26301539e+00 7.61676371e-01 1.81391776e-01 6.79298460e-01
9.18152779e-02 6.40406728e-01 1.47393370e+00 6.92275286e-01
2.13935480e-01 7.73286641e-01 5.06735563e-01 2.13210970e-01
4.49251950e-01 -5.41019320e-01 5.85927784e-01 5.72180972e-02
-7.70418253e-03 -2.26276457e-01 -7.89174914e-01 1.91801086e-01
-2.10594797e+00 -8.58099282e-01 8.32211018e-01 2.50309968e+00
5.51303029e-01 4.05995727e-01 5.19470036e-01 9.60288644e-02
6.61443532e-01 -2.04053625e-01 -8.76531065e-01 -3.90390277e-01
5.43294489e-01 -3.77394725e-03 6.56069756e-01 4.38746065e-01
-1.13323021e+00 1.10409927e+00 6.81383657e+00 4.93297607e-01
-1.25593638e+00 -5.69702089e-02 4.60895300e-01 -3.59625109e-02
3.28362584e-01 3.24400902e-01 -9.57019925e-01 5.02284765e-01
1.07060337e+00 9.69195515e-02 7.56788015e-01 1.01964080e+00
7.29372025e-01 -7.80103683e-01 -9.96951580e-01 6.37800515e-01
-3.24525177e-01 -8.50618541e-01 -3.58320475e-01 -2.32645199e-02
4.36969668e-01 -2.54753292e-01 -2.17022136e-01 6.07917249e-01
6.92131400e-01 -6.85109198e-01 7.22363889e-01 4.14650887e-01
3.97241592e-01 -9.44846570e-01 4.57799613e-01 8.18853915e-01
-9.07850742e-01 -7.19854057e-01 -4.22934353e-01 -2.07786146e-03
1.63427398e-01 1.63955875e-02 -1.32899237e+00 2.78137207e-01
4.91057336e-01 8.41565803e-02 -4.02666777e-01 1.16692412e+00
-4.43237215e-01 7.97938287e-01 -4.70355183e-01 -2.88308859e-01
4.08953279e-01 -3.14111143e-01 5.10894477e-01 6.51905060e-01
7.35336691e-02 5.67762963e-02 7.36926794e-01 4.69296396e-01
5.81111073e-01 -2.14410082e-01 -7.58440197e-01 -2.18092352e-02
7.38754690e-01 1.08733225e+00 -8.62954795e-01 -1.53916612e-01
1.12469815e-01 5.67139685e-01 5.68130314e-01 4.00011152e-01
-1.01327932e+00 -2.83798426e-01 3.10252130e-01 1.69019729e-01
1.88598797e-01 -4.17971343e-01 1.79949552e-01 -6.26866043e-01
-1.56813428e-01 -9.04988766e-01 4.70131546e-01 -4.44279730e-01
-8.28838587e-01 4.32793587e-01 2.37817883e-01 -1.26368475e+00
-5.18214881e-01 -3.91700536e-01 -5.49619675e-01 1.53813481e-01
-1.38504386e+00 -6.02875173e-01 6.72949292e-03 1.61735699e-01
7.49664307e-01 -1.88233972e-01 3.96139920e-01 1.00208791e-02
-9.01144743e-01 4.44233298e-01 -9.30655226e-02 -4.43349540e-01
6.00884020e-01 -1.28699827e+00 -1.01229109e-01 6.07143521e-01
-2.83069760e-01 3.08403760e-01 8.84709239e-01 -5.70162177e-01
-1.36374879e+00 -9.75300968e-01 1.66702673e-01 6.81954175e-02
5.70249975e-01 -2.32775331e-01 -6.37579918e-01 5.03401220e-01
-1.23044550e-01 -2.23716110e-01 3.37960809e-01 -1.77317321e-01
3.87156934e-01 -2.39165217e-01 -1.19451392e+00 7.20790327e-01
5.91394305e-01 -3.76062514e-03 -2.82278895e-01 3.29319000e-01
4.19694215e-01 -4.32978660e-01 -4.32752967e-01 3.96735281e-01
5.47486246e-01 -5.88780701e-01 5.12974143e-01 -6.41158998e-01
-1.19166829e-01 -4.51219559e-01 -1.95323583e-03 -1.50401592e+00
5.92604205e-02 -9.09024239e-01 -5.46227843e-02 9.13347185e-01
4.06140506e-01 -6.17902219e-01 7.54959702e-01 6.47455871e-01
1.09560631e-01 -1.00696206e+00 -9.12752569e-01 -8.54619622e-01
-2.76197374e-01 -1.20935857e-01 4.15785491e-01 4.63341802e-01
3.27028930e-02 2.66263336e-01 -4.15704817e-01 2.87821054e-01
5.01935959e-01 1.28249153e-01 9.91261125e-01 -8.73108387e-01
-5.06030679e-01 -9.35372114e-02 6.52452707e-02 -1.06459820e+00
4.78414178e-01 -3.11766744e-01 8.25250864e-01 -1.36412323e+00
-4.89524096e-01 -9.93182898e-01 -2.26689532e-01 7.37339616e-01
-2.22278628e-02 -6.48050249e-01 1.56473279e-01 8.41038600e-02
-8.27885032e-01 5.51914752e-01 9.75873768e-01 1.27419643e-02
-5.06512642e-01 5.31205118e-01 -1.91280589e-01 6.74315929e-01
1.17855406e+00 -4.94967014e-01 -7.99745739e-01 -1.23017214e-01
-1.51015177e-01 4.07098234e-01 9.70653370e-02 -1.18217373e+00
2.91307598e-01 -8.37105274e-01 1.40778020e-01 -3.44201833e-01
2.02470452e-01 -9.74888802e-01 -2.15251833e-01 7.87568629e-01
-4.98116553e-01 1.99410439e-01 8.90500024e-02 8.64343941e-01
1.87981814e-01 -4.15847450e-01 1.06301486e+00 -3.53719480e-02
-6.45924747e-01 8.50261748e-02 -8.53343546e-01 -2.31984079e-01
1.34216237e+00 -1.81461483e-01 -2.52404343e-02 -3.11942369e-01
-6.10534728e-01 7.42033064e-01 3.55702132e-01 3.82765710e-01
5.05250514e-01 -7.93783903e-01 -9.78383273e-02 2.09956437e-01
-8.61145109e-02 -9.85343829e-02 -2.67649680e-01 7.08061159e-01
-3.89704227e-01 1.56711400e-01 -2.14786097e-01 -4.57633317e-01
-9.77277398e-01 6.90629125e-01 6.08548880e-01 -4.10776109e-01
-2.83902794e-01 3.75552893e-01 -2.83944935e-01 -2.97129869e-01
4.43907887e-01 -1.92895070e-01 -2.70913720e-01 -1.87608182e-01
2.00252831e-01 5.45764387e-01 -8.34725723e-02 -2.38854960e-01
-1.60848647e-01 1.13950945e-01 1.32266833e-02 -4.27978694e-01
1.06048763e+00 -1.95081159e-01 2.29248688e-01 4.48802352e-01
5.30456364e-01 -4.16200906e-01 -1.79576969e+00 8.04329887e-02
4.21536505e-01 -4.65106726e-01 -5.91601282e-02 -6.74882054e-01
-6.42726600e-01 5.58914244e-01 6.35602772e-01 1.90861911e-01
1.14847398e+00 -5.06431758e-01 5.05514920e-01 8.28779519e-01
9.14554954e-01 -1.79772031e+00 2.36403123e-01 4.24997509e-01
5.60361564e-01 -1.21550381e+00 -3.06945089e-02 1.86822042e-02
-9.37413871e-01 1.05293393e+00 1.19267726e+00 -3.16508204e-01
4.00162250e-01 2.36522034e-01 7.34962597e-02 1.76257696e-02
-1.20352459e+00 -3.03774804e-01 -3.35426211e-01 4.42578226e-01
-1.62527993e-01 2.60828793e-01 -3.07587832e-01 1.31080419e-01
3.85936797e-02 9.52631887e-03 4.23149616e-01 1.32764721e+00
-8.20528448e-01 -1.14016140e+00 -4.43673819e-01 4.26961094e-01
-1.54268771e-01 5.41176438e-01 -3.08298439e-01 7.34003305e-01
1.18860297e-01 1.20408452e+00 -1.66358203e-01 -4.23239291e-01
4.05200303e-01 -3.61693427e-02 5.13169765e-01 -6.89180255e-01
-5.35729468e-01 3.32436532e-01 2.92184472e-01 -6.84432924e-01
-2.95987368e-01 -6.33681476e-01 -1.31388068e+00 3.05306584e-01
-5.62404871e-01 3.73158634e-01 5.16359687e-01 9.49734211e-01
1.88695103e-01 3.82783145e-01 1.13699138e+00 -1.08920813e+00
-1.21007538e+00 -7.03350902e-01 -3.23479503e-01 -2.05693528e-01
4.09860134e-01 -1.28162289e+00 -1.71518788e-01 -1.06769867e-01] | [4.203892707824707, 2.2060322761535645] |
977c0a67-b1da-48ac-b13e-48bc158e0cde | abusive-language-detection-using-syntactic | null | null | https://aclanthology.org/2020.alw-1.6 | https://aclanthology.org/2020.alw-1.6.pdf | Abusive Language Detection using Syntactic Dependency Graphs | Automated detection of abusive language online has become imperative. Current sequential models (LSTM) do not work well for long and complex sentences while bi-transformer models (BERT) are not computationally efficient for the task. We show that classifiers based on syntactic structure of the text, dependency graphical convolutional networks (DepGCNs) can achieve state-of-the-art performance on abusive language datasets. The overall performance is at par with of strong baselines such as fine-tuned BERT. Further, our GCN-based approach is much more efficient than BERT at inference time making it suitable for real-time detection. | ['Chris Brew', 'Kanika Narang'] | null | null | null | null | emnlp-alw-2020-11 | ['abusive-language'] | ['natural-language-processing'] | [-3.38876247e-01 -9.39441249e-02 -3.87016118e-01 -5.01782298e-01
-7.82799244e-01 -5.99079311e-01 7.25596189e-01 3.51900250e-01
-7.01713622e-01 6.63640916e-01 2.93333918e-01 -7.58824646e-01
3.54105532e-01 -5.13783276e-01 -3.69063854e-01 -1.24880753e-01
-1.74232349e-01 6.37567163e-01 4.01366740e-01 -6.08255744e-01
2.68838704e-01 7.10058987e-01 -5.42594194e-01 4.89161015e-01
6.39307976e-01 6.58499122e-01 -5.12667358e-01 7.85439968e-01
-4.33133870e-01 1.46639907e+00 -9.39943850e-01 -9.47366893e-01
-3.24908316e-01 -2.29545370e-01 -9.92704749e-01 -4.90426600e-01
6.25101864e-01 -8.71194422e-01 -6.85997844e-01 9.82926250e-01
2.99733847e-01 -3.43814902e-02 4.67828453e-01 -9.86896336e-01
-4.98042285e-01 1.09098816e+00 -5.74909508e-01 1.13571215e+00
3.17928016e-01 7.30621666e-02 1.16318333e+00 -7.16746926e-01
5.35795927e-01 1.81955016e+00 7.91020751e-01 6.17046058e-01
-1.12812006e+00 -7.87011385e-01 4.65896547e-01 2.85299689e-01
-9.12986934e-01 -4.63955343e-01 7.20491827e-01 -3.39096844e-01
1.66150200e+00 7.07212910e-02 4.68202442e-01 1.89145815e+00
3.39500546e-01 1.06424117e+00 9.98834252e-01 -2.75289923e-01
-2.26032108e-01 -5.22857532e-02 8.28636527e-01 9.24587309e-01
4.16655660e-01 -1.39624089e-01 -6.38704658e-01 -5.07885039e-01
2.51128286e-01 -6.11536428e-02 1.86490253e-01 4.16185468e-01
-4.95747447e-01 1.52339160e+00 4.17355388e-01 8.89831483e-01
2.04825867e-02 4.05331552e-01 1.02072263e+00 4.27138627e-01
8.50554645e-01 4.95457739e-01 -2.12518111e-01 -4.03461516e-01
-1.19505048e+00 3.94191027e-01 9.96043026e-01 7.18960106e-01
7.67749473e-02 2.83017486e-01 -3.16153497e-01 8.28375578e-01
3.38394456e-02 1.63039699e-01 4.28880125e-01 -4.54083651e-01
8.06278229e-01 4.67583090e-01 -1.87551692e-01 -1.03546333e+00
-7.45119512e-01 -2.76014864e-01 -4.69148576e-01 -9.69751030e-02
5.32208145e-01 -1.38194039e-01 -8.17948103e-01 1.46456933e+00
-1.09295011e-01 -2.91076094e-01 -3.91527504e-01 4.67940122e-01
7.04140902e-01 6.18741512e-01 4.15494323e-01 -2.25494176e-01
1.10344768e+00 -8.17746699e-01 -9.71438944e-01 -7.90170848e-01
9.67100739e-01 -3.63143325e-01 8.62153888e-01 5.56843221e-01
-9.57146168e-01 1.29662782e-01 -1.06773078e+00 -5.15840590e-01
-5.48335910e-01 -4.56500322e-01 6.89193010e-01 9.48493242e-01
-9.14853990e-01 7.94104934e-01 -9.88622367e-01 -3.11913371e-01
7.91689217e-01 1.84260145e-01 -3.09391707e-01 2.07704633e-01
-1.38266599e+00 1.55503690e+00 2.52634883e-01 8.40200558e-02
-1.11039662e+00 -4.08345908e-01 -1.08710396e+00 -8.92346725e-02
2.33268976e-01 -1.18414216e-01 1.62589729e+00 -7.99645245e-01
-1.26839983e+00 1.22941518e+00 -1.20872281e-01 -8.90062511e-01
8.23579192e-01 -6.56915188e-01 -4.10859317e-01 2.56942838e-01
2.60742903e-02 1.13269702e-01 8.54730785e-01 -7.43817031e-01
-3.28465611e-01 -4.75906551e-01 1.63763881e-01 -1.83451876e-01
-6.77655160e-01 1.04218316e+00 2.54903644e-01 -5.45491993e-01
-4.85934049e-01 -5.82703412e-01 -7.56953843e-03 -2.87579715e-01
-8.55676889e-01 -5.77797532e-01 1.23284161e+00 -1.29395711e+00
1.68373311e+00 -1.87930441e+00 -8.53179619e-02 -1.27223581e-01
4.55833316e-01 7.35381126e-01 3.34308863e-01 5.79858899e-01
8.44124630e-02 4.52011287e-01 -1.72230586e-01 -8.91677976e-01
9.30751264e-02 3.79170358e-01 -4.49107945e-01 8.11272860e-01
2.26333827e-01 1.01063550e+00 -8.65861475e-01 -6.60610497e-01
1.39353529e-01 2.59339780e-01 -3.81941527e-01 3.92986953e-01
-1.34486794e-01 -1.58429015e-02 -2.68413007e-01 5.94059408e-01
2.35578299e-01 1.51422992e-02 2.61435568e-01 3.87243718e-01
-2.68819258e-02 9.59441185e-01 -2.66517013e-01 1.26923740e+00
-5.38394928e-01 1.09126520e+00 2.66677946e-01 -8.75217199e-01
7.07205713e-01 1.54623643e-01 -3.21336180e-01 -6.99311614e-01
4.47033644e-01 2.17216343e-01 2.38617599e-01 -5.85048318e-01
2.62298852e-01 -2.83350557e-01 -3.41969937e-01 5.19868433e-01
1.61173195e-01 1.21337451e-01 3.68627042e-01 6.17115021e-01
1.43047416e+00 -2.34549657e-01 4.90646362e-01 -3.09700131e-01
4.45841342e-01 -9.66501907e-02 3.80792260e-01 9.72310841e-01
-4.94135976e-01 1.79890513e-01 9.83437896e-01 -6.30761206e-01
-1.00447571e+00 -7.85199702e-01 -1.04487538e-02 1.53758168e+00
-4.26130623e-01 -4.61053938e-01 -9.66780424e-01 -1.14239943e+00
-2.36324128e-02 1.21959090e+00 -6.84104741e-01 -1.73722982e-01
-1.16158915e+00 -8.21964741e-01 1.07099009e+00 7.18119383e-01
4.80633140e-01 -1.25642276e+00 -5.99524617e-01 5.96089900e-01
-6.86410218e-02 -1.48430431e+00 -2.42001370e-01 2.11080492e-01
-7.25733042e-01 -9.11425889e-01 -9.35337171e-02 -2.24975407e-01
-7.12745488e-02 -1.60745919e-01 1.36042321e+00 3.26333821e-01
-3.74394864e-01 -3.04334909e-01 -3.15899909e-01 -4.74685043e-01
-7.57234991e-01 2.49915674e-01 -4.72481735e-02 -2.42744938e-01
6.93505764e-01 -6.65082872e-01 -3.07404667e-01 -3.14624369e-01
-6.54904723e-01 -5.88001251e-01 1.91843107e-01 7.11552680e-01
-5.24504721e-01 -4.84269917e-01 3.57754290e-01 -1.45081758e+00
8.70187640e-01 -5.40227473e-01 -1.84375435e-01 -1.52483791e-01
-3.64456117e-01 -2.59580225e-01 8.86047006e-01 -3.94247234e-01
-9.79161263e-01 -5.10412574e-01 -6.21352851e-01 -3.46881986e-01
-1.64197221e-01 2.82946527e-01 1.49454504e-01 2.62512088e-01
7.86717594e-01 -1.86888680e-01 -3.43858033e-01 -7.20720828e-01
3.23021231e-04 7.38210618e-01 2.54771262e-01 -1.63825437e-01
5.76283395e-01 4.32002187e-01 -3.11641455e-01 -8.50701213e-01
-1.29090881e+00 -3.45320135e-01 -9.00782168e-01 1.26123965e-01
8.57002735e-01 -6.61298037e-01 -4.51503187e-01 6.89132869e-01
-1.48224556e+00 -4.51946169e-01 5.32485843e-01 -1.19178578e-01
-1.41608655e-01 6.41460955e-01 -1.52489531e+00 -1.24981570e+00
-8.21093559e-01 -7.91575491e-01 9.65851903e-01 -3.20815623e-01
-6.12080634e-01 -1.21308053e+00 -7.49632576e-03 3.34032118e-01
4.39486265e-01 4.79502350e-01 1.20672190e+00 -1.33398926e+00
2.55794227e-01 -5.05517244e-01 -3.21483999e-01 2.99881399e-01
-4.12223786e-01 3.43664169e-01 -1.14378357e+00 -3.34496647e-01
-1.60255730e-01 -5.96626580e-01 1.03520596e+00 1.42239124e-01
9.45291162e-01 -6.55064404e-01 -5.06352603e-01 1.43662617e-01
9.60372448e-01 -6.05449826e-02 4.59573328e-01 3.73666525e-01
8.73407781e-01 4.56556827e-01 1.45258069e-01 2.89332926e-01
2.86741167e-01 6.13911986e-01 3.99195641e-01 1.06640309e-01
1.68661121e-02 -3.55313420e-01 8.08334947e-01 4.29382712e-01
5.09758413e-01 -4.15619999e-01 -1.26880443e+00 6.01604342e-01
-1.70686769e+00 -1.09643340e+00 -5.43157279e-01 1.46371746e+00
7.66114116e-01 8.51286054e-01 5.57914317e-01 2.52622992e-01
7.78710067e-01 3.98717821e-01 -2.79289365e-01 -1.39087152e+00
7.09738210e-02 2.88819581e-01 2.68902749e-01 7.66846001e-01
-1.31419265e+00 1.23566711e+00 7.58256721e+00 8.29402626e-01
-9.16522145e-01 6.66550875e-01 7.02895045e-01 -4.32125688e-01
8.40946808e-02 -2.19740927e-01 -1.15795434e+00 5.87376297e-01
1.56728423e+00 2.11190581e-01 1.17492758e-01 1.00859797e+00
1.87034860e-01 3.74297202e-02 -9.92472410e-01 8.08960974e-01
2.92137861e-01 -1.12153196e+00 -1.49910795e-02 1.70735512e-02
1.62570953e-01 3.94968301e-01 -1.51733845e-01 6.24832630e-01
8.78208518e-01 -1.27219450e+00 9.14646685e-01 -1.59293324e-01
5.01007080e-01 -9.06569481e-01 8.40401053e-01 6.88246489e-01
-3.61650050e-01 -4.33562934e-01 -3.39281946e-01 -3.60363990e-01
2.40297198e-01 6.60598576e-01 -6.67898655e-01 -2.73734536e-02
9.36532438e-01 7.59803712e-01 -8.82260978e-01 3.34491313e-01
-6.02200091e-01 1.09747612e+00 -3.90717328e-01 -3.95636737e-01
8.03499103e-01 2.74037391e-01 6.52053714e-01 1.93470538e+00
-3.68206322e-01 -1.43239051e-01 1.11737430e-01 7.58905649e-01
-1.45492896e-01 -1.62603171e-03 -8.61313581e-01 -3.91263455e-01
3.96397173e-01 1.23265886e+00 -4.59629834e-01 -4.36452240e-01
-3.50637525e-01 8.97558689e-01 1.16792607e+00 -1.80311546e-01
-6.70448542e-01 -2.04158396e-01 5.43840349e-01 1.29546687e-01
1.53084368e-01 -2.98853189e-01 -3.11181337e-01 -1.03714001e+00
-1.29418939e-01 -9.28070426e-01 1.03080332e+00 -3.31131458e-01
-1.62340903e+00 8.16925406e-01 1.02274090e-01 -4.34149235e-01
-7.16228545e-01 -9.86034930e-01 -9.23139513e-01 4.60891604e-01
-1.22022986e+00 -1.36528969e+00 2.67470211e-01 4.65118945e-01
6.75923884e-01 8.94559547e-02 8.41544032e-01 1.09610952e-01
-1.15543151e+00 6.88401818e-01 -2.02409387e-01 7.91500032e-01
4.56338793e-01 -1.32589555e+00 6.77725971e-01 1.16139317e+00
1.83428392e-01 6.51978374e-01 8.93190026e-01 -6.67299032e-01
-9.83202398e-01 -8.60240519e-01 1.20260024e+00 -8.83949518e-01
1.29161382e+00 -9.65091288e-01 -1.03177142e+00 1.10909832e+00
4.24544513e-01 -1.12392835e-01 5.85952520e-01 4.63427186e-01
-1.03405106e+00 3.66090119e-01 -1.13536847e+00 4.68308508e-01
1.23867989e+00 -6.31766796e-01 -7.53412783e-01 6.29880369e-01
5.22885382e-01 -3.37811291e-01 -3.28364849e-01 8.95017236e-02
2.46084049e-01 -1.24501133e+00 5.92275321e-01 -1.16397703e+00
8.38891625e-01 7.38418460e-01 1.43344045e-01 -1.07308602e+00
-4.23059732e-01 -6.30515099e-01 -6.00827813e-01 1.15503347e+00
4.75395262e-01 -6.95768058e-01 4.45082396e-01 4.76077557e-01
-3.56668867e-02 -5.42358637e-01 -1.28536570e+00 -8.33009899e-01
6.44601226e-01 -4.19399232e-01 -6.03488572e-02 1.01129031e+00
3.77739429e-01 6.43174529e-01 -2.86230505e-01 -1.52513608e-01
4.47268963e-01 -1.76122710e-02 3.89281183e-01 -1.06312609e+00
-1.54955238e-01 -8.02577674e-01 -3.86968106e-01 -5.33917248e-01
8.00038576e-01 -7.40982831e-01 -2.83471137e-01 -1.38287199e+00
4.51315612e-01 1.56377643e-01 -1.47386640e-02 6.96075559e-01
-2.41866782e-01 5.11563480e-01 1.25865147e-01 5.28260395e-02
-5.81954539e-01 1.99040204e-01 4.60641772e-01 -2.42089644e-01
1.06189549e-01 -4.03586507e-01 -6.08519971e-01 9.21979129e-01
8.50475848e-01 -7.59513378e-01 1.58754304e-01 -3.69092852e-01
1.97411016e-01 -1.17451213e-01 2.88390011e-01 -5.64139426e-01
-1.03768915e-01 1.34229437e-01 3.15440416e-01 -5.28187394e-01
3.40167522e-01 -2.00482488e-01 -6.53245449e-01 7.08001316e-01
-4.13443297e-01 3.60058367e-01 2.58011192e-01 4.53781724e-01
-4.75572422e-02 -6.37774646e-01 9.22389448e-01 -3.99664849e-01
-2.57523566e-01 1.76701307e-01 -8.99249375e-01 2.73757935e-01
7.03597128e-01 1.59234285e-01 -5.47623873e-01 -5.99983215e-01
-6.31799996e-01 -1.65385362e-02 4.90753353e-02 4.46721405e-01
2.41124764e-01 -6.76933765e-01 -8.51405561e-01 -3.98674607e-01
-1.32039502e-01 -5.12366295e-01 -1.46808773e-01 8.26439798e-01
-7.38137603e-01 4.85684842e-01 9.65589061e-02 -1.84999347e-01
-1.42419791e+00 6.67756140e-01 5.36941946e-01 -8.13503087e-01
-8.47466111e-01 1.09021914e+00 -1.52748689e-01 -2.79049933e-01
5.37440218e-02 1.28269985e-01 -3.54029894e-01 2.89184719e-01
8.87189031e-01 3.18507701e-01 3.04370284e-01 -6.76689208e-01
-5.26601970e-01 -4.28211153e-01 -7.67009020e-01 -1.06187865e-01
1.43867898e+00 1.58260673e-01 -3.03966641e-01 6.39023423e-01
1.23182452e+00 -1.09633252e-01 -7.76992917e-01 -1.99849173e-01
4.18461382e-01 -4.91973728e-01 2.83460677e-01 -7.63338029e-01
-7.81608641e-01 1.26161528e+00 -7.96936676e-02 4.21958536e-01
4.11716282e-01 -1.43817719e-02 9.97829139e-01 3.92444670e-01
2.07547337e-01 -1.09082162e+00 3.60171229e-01 7.97548592e-01
8.89328241e-01 -1.01735580e+00 4.49500307e-02 -4.45368022e-01
-4.65248108e-01 1.19943082e+00 7.02565134e-01 -5.17801821e-01
5.53888500e-01 5.01727283e-01 1.05432011e-01 -4.63693112e-01
-7.36567259e-01 -2.15737075e-02 -8.44824985e-02 3.51519823e-01
5.12751460e-01 -6.92344755e-02 -2.85279334e-01 3.91391695e-01
-4.50084895e-01 -7.81805634e-01 6.36345506e-01 9.30699289e-01
-3.45826417e-01 -8.25528681e-01 -1.33004412e-01 6.52457297e-01
-1.17773390e+00 -3.60117525e-01 -1.04399014e+00 1.01679003e+00
-4.77061510e-01 1.30119228e+00 9.07053947e-02 -2.09707931e-01
2.10624099e-01 3.65539193e-01 5.37604690e-01 -8.17728937e-01
-1.16024733e+00 -1.89398497e-01 8.67637575e-01 -8.24518919e-01
9.04555246e-02 -7.40274489e-01 -8.74340117e-01 -9.63920832e-01
-2.91969597e-01 -1.72650054e-01 3.22011739e-01 1.27225411e+00
-2.15349756e-02 1.72925405e-02 3.23502511e-01 -6.48325145e-01
-8.10026109e-01 -1.51934063e+00 -4.99607861e-01 5.25540948e-01
4.96803284e-01 -6.15242839e-01 -5.25731266e-01 -4.80632484e-01] | [8.800048828125, 10.530421257019043] |
040b109b-8977-4840-8d10-ffbe1a5f8d2d | learn-to-adapt-for-generalized-zero-shot-text | null | null | https://openreview.net/forum?id=2NOSFKxicWQ | https://openreview.net/pdf?id=2NOSFKxicWQ | Learn to Adapt for Generalized Zero-Shot Text Classification | Generalized zero-shot text classification aims to classify textual instances from both previously seen classes and incrementally emerging unseen classes. Most existing methods generalize poorly since the learned parameters are only optimal for seen classes rather than for both classes, and the parameters keep stationary in predicting procedures. To address these challenges, we propose a novel Learn to Adapt (LTA) network using a variant meta-learning framework. Specifically, LTA trains multiple meta-learners by using both seen classes and virtual unseen classes to simulate a generalized zero-shot learning (GZSL) scenario in accordance with the test time, and simultaneously learns to calibrate the class prototypes and sample representations to make the learned parameters adaptive to incoming unseen classes. We claim that the proposed model is capable of representing all prototypes and samples from both classes to a more consistent distribution in the global space. Extensive experiments on five text classification datasets show that our model outperforms several competitive previous approaches by large margins. The code and the whole datasets will be available after paper publication. | ['Anonymous'] | 2021-11-16 | null | null | null | acl-arr-november-2021-11 | ['generalized-zero-shot-learning', 'generalized-zero-shot-learning'] | ['computer-vision', 'methodology'] | [ 3.66693348e-01 2.27307510e-02 -4.67790902e-01 -6.03292763e-01
-8.04645240e-01 -3.35136771e-01 7.22571552e-01 1.94993436e-01
-3.25434774e-01 8.37849081e-01 -3.95188108e-03 1.19797103e-01
-1.32723719e-01 -1.01518536e+00 -4.51050550e-01 -7.61063278e-01
3.04510564e-01 1.05604994e+00 4.72143710e-01 -2.62724608e-01
1.41193390e-01 8.69547948e-02 -2.00144529e+00 6.92199409e-01
7.11966336e-01 8.70639801e-01 1.27356738e-01 7.37118244e-01
-6.01988673e-01 5.54194331e-01 -6.44114137e-01 -3.30582857e-01
1.51139900e-01 -6.13062024e-01 -7.65312552e-01 2.05317646e-01
3.32416207e-01 -1.87743545e-01 -1.73642561e-01 8.69201958e-01
5.91352165e-01 6.81271613e-01 9.22748089e-01 -1.45411837e+00
-7.76214063e-01 8.71896863e-01 -2.92419106e-01 6.06766008e-02
1.40030712e-01 -1.56376168e-01 9.92302597e-01 -1.16998458e+00
7.19877243e-01 1.20217276e+00 6.45905375e-01 9.59770381e-01
-1.01615298e+00 -5.98350763e-01 4.48904485e-01 5.41940928e-01
-1.17121029e+00 -5.11782408e-01 7.85504162e-01 -2.86597818e-01
6.47655189e-01 1.76898599e-01 5.31221330e-01 1.79743171e+00
6.65292963e-02 1.11373055e+00 6.35285437e-01 -6.21373355e-01
8.09889019e-01 6.39571130e-01 6.71619415e-01 2.51651585e-01
2.04855472e-01 -3.21903735e-01 -4.46680278e-01 -3.01565766e-01
-1.15864575e-01 4.50470150e-01 -1.34525582e-01 -8.21097374e-01
-9.08357024e-01 9.03753281e-01 1.37795821e-01 4.02256817e-01
-1.08691089e-01 -3.10233682e-01 4.60331202e-01 4.44161981e-01
6.36373222e-01 6.82368502e-02 -7.61394143e-01 3.37294228e-02
-7.10169375e-01 5.26405163e-02 8.31444919e-01 1.20463049e+00
7.50619411e-01 1.18086390e-01 -4.37691063e-01 1.21798801e+00
3.06720044e-02 3.87233019e-01 1.27466655e+00 -3.00978363e-01
5.48688471e-01 7.39095390e-01 2.00544344e-03 -4.66430306e-01
-2.75150836e-01 -5.48623562e-01 -7.34687567e-01 -7.94478059e-02
1.19444855e-01 -2.10675120e-01 -1.18744552e+00 1.51939845e+00
4.96206671e-01 4.44532931e-01 4.01916564e-01 3.45717490e-01
8.59386623e-01 1.06972420e+00 1.27927959e-01 -4.86871719e-01
8.96463394e-01 -9.08238113e-01 -7.41732359e-01 -4.70066190e-01
6.53382599e-01 -3.35879952e-01 1.15167105e+00 2.70391077e-01
-7.68442869e-01 -8.29281271e-01 -1.24546957e+00 1.84105620e-01
-8.89256418e-01 -1.01402245e-01 6.91318959e-02 6.96163893e-01
-6.75067008e-01 6.00949645e-01 -4.93098497e-01 -6.44580603e-01
4.65850890e-01 7.98140019e-02 7.31698349e-02 -6.87769279e-02
-1.32995045e+00 5.79661906e-01 1.03573906e+00 -1.92388430e-01
-9.46221650e-01 -5.66908062e-01 -7.77330339e-01 2.73432523e-01
3.26948553e-01 -5.17010748e-01 1.31326962e+00 -1.09257281e+00
-1.38232899e+00 6.45397127e-01 -4.55407314e-02 -4.04780954e-01
6.92209542e-01 9.43198353e-02 -4.99994636e-01 -1.17384858e-01
-8.59106332e-02 4.53971833e-01 1.19863892e+00 -1.44240499e+00
-8.87390435e-01 -4.86617863e-01 -4.36826199e-01 2.84221351e-01
-9.83980179e-01 -6.22470737e-01 -2.15177223e-01 -5.18038690e-01
1.13592081e-01 -5.99757552e-01 1.76837761e-02 -2.23461002e-01
-9.85548347e-02 -3.94330770e-01 1.36275589e+00 4.72410843e-02
1.22663176e+00 -2.01297259e+00 1.28084511e-01 5.42921238e-02
5.09386184e-03 3.07738900e-01 -3.22828114e-01 4.15092051e-01
-1.67683452e-01 -3.20115924e-01 -1.01211622e-01 -4.31413293e-01
-7.86533803e-02 4.47134137e-01 -5.57206452e-01 2.12739035e-01
-2.60153383e-01 9.12835300e-01 -1.20411563e+00 -4.60479885e-01
2.67274469e-01 2.80100614e-01 -1.35804653e-01 2.12834761e-01
-2.67104089e-01 -1.03882745e-01 -5.73079407e-01 5.55299580e-01
4.64945614e-01 -3.84477645e-01 3.48118991e-01 1.82050943e-01
3.98439884e-01 -4.86643195e-01 -1.30354536e+00 1.31579340e+00
-3.51695627e-01 3.71749133e-01 -4.92673099e-01 -1.42563462e+00
1.12326872e+00 3.68876725e-01 3.30994070e-01 -4.14978027e-01
3.78837079e-01 -1.96169559e-02 -2.87879318e-01 -5.90768814e-01
4.50596929e-01 -2.58271605e-01 -1.61424540e-02 4.19408798e-01
6.30046546e-01 5.04480600e-02 5.62805682e-04 1.45799801e-01
6.68336749e-01 -8.93739015e-02 6.14092767e-01 1.83445122e-02
4.53648120e-01 -1.10227466e-01 5.16784966e-01 1.28600657e+00
-3.60526145e-01 5.88576138e-01 1.95297912e-01 -7.45386958e-01
-1.10422623e+00 -1.17213643e+00 -3.22581232e-01 1.62908077e+00
1.62921667e-01 -1.66591018e-01 -5.03706753e-01 -9.55418408e-01
-1.41279921e-01 1.30553889e+00 -1.02434802e+00 -5.36107183e-01
-1.29686683e-01 -7.82834113e-01 -1.10513195e-01 5.84237278e-01
2.03318328e-01 -1.15766537e+00 -3.78251225e-01 4.08713490e-01
-5.86385317e-02 -5.87451398e-01 -1.14760704e-01 4.44437802e-01
-9.23480570e-01 -1.02897787e+00 -8.60516667e-01 -1.07513249e+00
5.90293825e-01 2.88371384e-01 7.01296031e-01 -1.68905988e-01
-3.96616787e-01 5.67048311e-01 -7.89528131e-01 -6.03469133e-01
-6.00691497e-01 2.56909043e-01 5.86024411e-02 5.10945439e-01
5.75927079e-01 -3.36754769e-01 -1.54608816e-01 2.28504270e-01
-9.77028608e-01 -4.85748537e-02 1.49554223e-01 1.27061617e+00
3.72452378e-01 1.87121749e-01 8.44419003e-01 -1.41040206e+00
7.06455886e-01 -9.39287663e-01 -2.59235948e-01 7.63029218e-01
-8.93750310e-01 -1.55603752e-01 9.94358480e-01 -9.50210512e-01
-1.28231192e+00 -3.21026728e-03 2.03544647e-01 -5.38255870e-01
-2.88682371e-01 1.91561326e-01 -1.57285437e-01 4.03818339e-01
9.17022586e-01 4.67196912e-01 -2.82744020e-01 -8.40772018e-02
3.97996157e-01 9.79264498e-01 2.87288874e-01 -5.69237769e-01
7.73271084e-01 4.51914787e-01 -4.87148613e-01 -8.30363512e-01
-1.22131479e+00 -7.25268126e-01 -8.72213840e-01 -3.18361521e-01
4.95541185e-01 -7.14590847e-01 -4.38854881e-02 4.77718681e-01
-8.27839255e-01 -2.99735725e-01 -7.19350398e-01 3.18830550e-01
-6.30484283e-01 1.52878210e-01 -4.14093882e-01 -9.90709603e-01
-3.73463929e-01 -8.90070021e-01 1.03651261e+00 3.29846114e-01
-9.54578370e-02 -1.27325642e+00 2.56514013e-01 -2.45598685e-02
2.51246929e-01 -8.39409232e-02 1.05515313e+00 -1.32578838e+00
-1.86182896e-03 -7.12468505e-01 2.58597702e-01 2.13515267e-01
1.31970942e-01 -9.42929462e-02 -1.24128926e+00 -6.64755702e-01
1.95178077e-01 -6.84222102e-01 9.73756969e-01 4.48282540e-01
1.33631527e+00 -1.92238182e-01 -5.69657564e-01 3.94030571e-01
1.29594493e+00 5.26788771e-01 5.17876089e-01 3.51445317e-01
3.63959402e-01 4.28923965e-01 6.37249589e-01 9.01882410e-01
4.84073348e-02 2.62751162e-01 1.23842843e-01 4.05510724e-01
1.71251878e-01 -3.45878303e-01 2.94614248e-02 9.18639421e-01
4.02129114e-01 -8.28603923e-01 -8.35209966e-01 4.76344049e-01
-2.11479068e+00 -1.24498928e+00 2.46758550e-01 2.16835070e+00
6.21455193e-01 2.41904214e-01 -3.04142889e-02 2.40599096e-01
1.10752237e+00 1.13046601e-01 -8.49633515e-01 -2.07248479e-01
-1.33551717e-01 2.72915304e-01 -8.25174269e-04 3.66503596e-01
-1.15230596e+00 8.16595316e-01 6.46414804e+00 9.15651739e-01
-7.82773435e-01 1.85105190e-01 6.44281685e-01 -1.69798091e-01
-1.59251317e-01 1.83614518e-03 -9.68825758e-01 3.37561041e-01
1.03193474e+00 -6.56897664e-01 6.42232522e-02 1.27720928e+00
-2.65326351e-01 4.96623784e-01 -1.11599970e+00 9.91484463e-01
4.35510099e-01 -1.31306624e+00 5.41540205e-01 -4.15352285e-01
1.01240289e+00 -2.00104818e-01 1.33328512e-01 1.04217219e+00
2.79975384e-01 -3.97539377e-01 5.21969914e-01 6.48415625e-01
5.54791570e-01 -7.36282289e-01 7.91301847e-01 7.34555900e-01
-9.99446690e-01 -7.04057038e-01 -7.35722661e-01 2.73044765e-01
-2.81839043e-01 5.18785156e-02 -9.77121055e-01 4.11860496e-01
5.97083807e-01 8.86057317e-01 -7.54704297e-01 9.72492993e-01
1.66026920e-01 5.84592223e-01 1.90340668e-01 -5.04181862e-01
2.01872110e-01 1.01342320e-01 3.97986025e-01 9.26653683e-01
4.95632499e-01 1.74715087e-01 3.88973922e-01 5.12167037e-01
-9.85631347e-02 3.66414487e-01 -5.03839731e-01 1.00555845e-01
5.92594624e-01 1.19551349e+00 -8.26224864e-01 -9.40937161e-01
-3.20228398e-01 1.10511911e+00 3.91804844e-01 5.23874044e-01
-6.13723874e-01 -7.05027461e-01 2.81080082e-02 -1.32434860e-01
3.52506459e-01 4.60311919e-01 -4.76354770e-02 -1.50280881e+00
-1.86812162e-01 -5.42090833e-01 9.50098574e-01 -7.84368277e-01
-1.72931385e+00 7.33841836e-01 1.19469784e-01 -1.66770554e+00
-3.59962434e-01 -5.64619422e-01 -6.79159820e-01 3.17869246e-01
-1.20108867e+00 -1.00232303e+00 -4.95218605e-01 6.77046478e-01
1.24707043e+00 -6.56991005e-01 9.90658224e-01 -2.42587149e-01
-4.93342638e-01 7.78234720e-01 7.64959991e-01 -2.69150771e-02
8.63422692e-01 -1.15505111e+00 2.22681254e-01 3.95230412e-01
1.09931268e-01 3.53368610e-01 6.57266438e-01 -6.34045422e-01
-1.07187533e+00 -1.22531593e+00 5.80238819e-01 -4.78674769e-01
7.48512685e-01 -4.39703226e-01 -1.32482910e+00 8.79220009e-01
6.66520447e-02 8.97085443e-02 7.77221799e-01 1.37358874e-01
-3.70485395e-01 -2.78507292e-01 -1.12216830e+00 4.72926825e-01
7.17667162e-01 -2.17983976e-01 -1.03994298e+00 5.86485088e-01
5.65530241e-01 -2.40792975e-01 -4.20009822e-01 3.00081760e-01
3.88881177e-01 -7.90382326e-01 7.45909333e-01 -1.11066914e+00
1.66695535e-01 9.00121853e-02 -2.40398273e-01 -1.49435079e+00
-4.80793685e-01 -1.60354063e-01 -6.05521739e-01 1.21521461e+00
3.33890587e-01 -8.19199800e-01 9.41819072e-01 3.01353753e-01
-1.08015068e-01 -5.51458120e-01 -9.95204866e-01 -8.58934224e-01
1.92919657e-01 -3.28637242e-01 5.15184522e-01 1.34920681e+00
1.55579567e-01 6.68447673e-01 -4.81987476e-01 -8.05696696e-02
8.37166250e-01 2.19361722e-01 5.90485752e-01 -1.69882691e+00
-2.32925937e-01 -2.95696825e-01 -4.33392256e-01 -6.99913383e-01
3.59398514e-01 -1.11634493e+00 1.80056185e-01 -1.26471031e+00
5.96844971e-01 -2.48564795e-01 -5.61754465e-01 4.56134856e-01
-3.88464570e-01 -2.68809348e-01 -7.35695884e-02 1.58807129e-01
-8.17854285e-01 8.13092470e-01 1.03132904e+00 -5.05717576e-01
-3.94433826e-01 3.51151109e-01 -4.96926904e-01 4.63642210e-01
7.38383949e-01 -4.41455454e-01 -9.06396985e-01 -1.13837883e-01
-1.58986062e-01 3.28400396e-02 6.56236112e-02 -1.06504393e+00
4.23465937e-01 -2.89526790e-01 9.57625091e-01 -8.57575357e-01
3.92888665e-01 -8.81884933e-01 -1.76621825e-01 3.21279079e-01
-8.74935269e-01 -4.83199894e-01 -1.95446387e-02 1.03884828e+00
1.14106283e-01 -7.39576817e-01 1.04200876e+00 -1.33210197e-01
-8.71305466e-01 5.54377198e-01 -1.91991046e-01 2.07756221e-01
1.37081683e+00 -4.65331078e-01 -4.62741643e-01 -4.51791018e-01
-9.71918166e-01 3.31053644e-01 2.09961459e-01 7.88786948e-01
8.10950577e-01 -1.42300785e+00 -6.21456981e-01 5.24300158e-01
6.58244073e-01 -2.80517370e-01 6.14309072e-01 7.88200721e-02
-2.23346073e-02 1.20155774e-01 -7.04954788e-02 -6.76826596e-01
-9.49803948e-01 1.01283026e+00 2.82232285e-01 7.27071390e-02
-7.01777041e-01 6.61416054e-01 9.44490358e-03 -7.99480081e-01
5.82009614e-01 8.64381567e-02 -2.75162280e-01 3.58408570e-01
9.13375795e-01 3.09390187e-01 -2.83330176e-02 -2.54615039e-01
1.25375941e-01 2.98345417e-01 -5.81156433e-01 2.01621607e-01
1.32461667e+00 -1.99822173e-01 4.84064162e-01 1.26695764e+00
1.34904134e+00 -5.39322972e-01 -1.10266316e+00 -6.15487099e-01
-3.67045291e-02 -2.87321836e-01 -3.12581658e-01 -5.55235386e-01
-6.60477102e-01 9.04869556e-01 7.93348968e-01 4.11469460e-01
8.74485791e-01 -2.07659099e-02 5.66917241e-01 7.88929343e-01
2.73027658e-01 -1.56121290e+00 4.58296120e-01 5.83941519e-01
4.43406969e-01 -1.23184156e+00 -1.24563001e-01 1.19853199e-01
-7.63899803e-01 1.51486135e+00 9.52224374e-01 1.59662724e-01
6.98349118e-01 -1.47008047e-01 3.98927592e-02 1.52451977e-01
-1.19720185e+00 4.99133132e-02 1.81506455e-01 6.56171560e-01
1.43526375e-01 -5.36966883e-02 1.64596885e-01 6.01974130e-01
1.58061385e-01 -2.52897263e-01 4.63767856e-01 1.14520693e+00
-9.78726149e-01 -9.51743245e-01 -3.07317436e-01 7.15256155e-01
2.77152389e-01 1.85295805e-01 -3.12641263e-01 8.27231288e-01
3.77291255e-02 7.82306790e-01 1.34571373e-01 -3.61495376e-01
3.30897629e-01 7.60579050e-01 2.18152598e-01 -9.68558252e-01
-8.27219337e-02 -2.85088625e-02 -4.72761124e-01 1.29789963e-01
-1.36577906e-02 -5.93706965e-01 -1.05253136e+00 1.27858058e-01
-5.23258507e-01 3.34974259e-01 2.88522452e-01 1.01581872e+00
4.27686460e-02 6.03999317e-01 9.96974766e-01 -8.11415613e-01
-9.96719480e-01 -1.03762555e+00 -7.53104329e-01 5.71191311e-01
1.87258348e-01 -7.03700960e-01 -7.02281356e-01 1.28959909e-01] | [10.126744270324707, 3.462315559387207] |
c4145b0e-5aca-4265-86e1-6606e63629bd | graph-level-anomaly-detection-via | 2307.00755 | null | https://arxiv.org/abs/2307.00755v1 | https://arxiv.org/pdf/2307.00755v1.pdf | Graph-level Anomaly Detection via Hierarchical Memory Networks | Graph-level anomaly detection aims to identify abnormal graphs that exhibit deviant structures and node attributes compared to the majority in a graph set. One primary challenge is to learn normal patterns manifested in both fine-grained and holistic views of graphs for identifying graphs that are abnormal in part or in whole. To tackle this challenge, we propose a novel approach called Hierarchical Memory Networks (HimNet), which learns hierarchical memory modules -- node and graph memory modules -- via a graph autoencoder network architecture. The node-level memory module is trained to model fine-grained, internal graph interactions among nodes for detecting locally abnormal graphs, while the graph-level memory module is dedicated to the learning of holistic normal patterns for detecting globally abnormal graphs. The two modules are jointly optimized to detect both locally- and globally-anomalous graphs. Extensive empirical results on 16 real-world graph datasets from various domains show that i) HimNet significantly outperforms the state-of-art methods and ii) it is robust to anomaly contamination. Codes are available at: https://github.com/Niuchx/HimNet. | ['Ling Chen', 'Guansong Pang', 'Chaoxi Niu'] | 2023-07-03 | null | null | null | null | ['anomaly-detection'] | ['methodology'] | [-1.25992224e-02 4.09104526e-01 4.51480411e-02 -9.95191932e-02
2.25139745e-02 -2.23179564e-01 4.48964179e-01 7.13734806e-01
5.02104223e-01 2.62052156e-02 -1.97842736e-02 -2.00465485e-01
-1.33467570e-01 -1.30581367e+00 -6.33647442e-01 -5.35495937e-01
-9.32323098e-01 4.73385185e-01 2.90892482e-01 -1.02057472e-01
1.16519295e-01 6.21468663e-01 -1.10556924e+00 2.15523526e-01
8.38899851e-01 8.61576259e-01 -6.02316618e-01 9.07578051e-01
1.84835389e-01 1.30789161e+00 -4.53710377e-01 -4.00485784e-01
2.39244342e-01 -4.74438459e-01 -8.24901402e-01 5.22618353e-01
5.76805770e-01 -3.18803340e-01 -8.79765570e-01 1.45493460e+00
2.27276519e-01 -1.02428362e-01 5.67489982e-01 -1.51671576e+00
-8.36158991e-01 5.42572200e-01 -6.74771428e-01 5.87379336e-01
2.34949216e-01 2.31152251e-01 1.30738819e+00 -5.13400555e-01
4.89181250e-01 1.24497640e+00 7.30478168e-01 3.40450406e-01
-1.09846401e+00 -1.04959898e-01 4.51630920e-01 3.50269020e-01
-1.39039969e+00 -1.15987979e-01 1.01819706e+00 -6.16701245e-01
1.22710085e+00 1.45232648e-01 6.17806435e-01 1.18823576e+00
6.29566133e-01 4.87275362e-01 3.57410371e-01 -5.12339622e-02
2.48310640e-02 -5.02921641e-01 4.54731077e-01 1.31390691e+00
7.22605348e-01 1.33078113e-01 -3.70381594e-01 -4.56835389e-01
6.68980360e-01 3.11555147e-01 -1.44718170e-01 -4.10906285e-01
-8.74746978e-01 1.00188231e+00 8.40173423e-01 6.95897043e-01
-7.59429157e-01 1.82485551e-01 6.98128164e-01 7.61711240e-01
6.16580188e-01 4.68732178e-01 -1.80702120e-01 3.66663814e-01
-3.84898663e-01 -2.27135643e-01 9.04997051e-01 6.76906645e-01
7.85686493e-01 5.08989394e-01 -3.22880261e-02 5.70426524e-01
3.98154914e-01 -4.59929518e-02 3.21743131e-01 -2.84413993e-01
1.48244753e-01 1.52467608e+00 -7.41079509e-01 -1.66039014e+00
-6.07556820e-01 -7.46043801e-01 -1.33622897e+00 -4.20735106e-02
1.84515297e-01 2.06566691e-01 -1.06497085e+00 1.45015109e+00
3.93049747e-01 4.51830417e-01 -2.73272336e-01 5.58898866e-01
1.06150973e+00 3.72399807e-01 3.95234637e-02 2.95200974e-01
1.06189334e+00 -7.88867235e-01 -4.56331819e-01 -4.11238343e-01
7.96044469e-01 -5.12076765e-02 7.25458443e-01 7.65292579e-03
-1.02446103e+00 -2.78580338e-01 -1.11361468e+00 2.24123403e-01
-6.71705842e-01 -3.43891948e-01 2.39668489e-01 3.63120586e-01
-1.36449528e+00 8.31067324e-01 -9.12862301e-01 -5.42350113e-01
3.41488421e-01 1.99816078e-01 -6.22835398e-01 -1.96536779e-02
-1.06194234e+00 3.36584806e-01 6.13721728e-01 3.33528779e-02
-1.21207142e+00 -3.37391615e-01 -1.33599341e+00 3.83854777e-01
3.05411816e-01 -5.16103268e-01 6.86609924e-01 -1.00546265e+00
-5.94038844e-01 9.86231208e-01 6.13567196e-02 -5.08281887e-01
-3.07453913e-03 2.40461707e-01 -9.06903744e-01 3.54489833e-01
1.29017264e-01 7.14124516e-02 1.17541695e+00 -1.17557859e+00
-2.29015902e-01 -8.41213226e-01 -2.38515720e-01 -1.68319643e-01
-5.58333695e-01 -3.01591426e-01 -3.08175147e-01 -8.52923989e-01
4.66644377e-01 -5.39712369e-01 -2.42582813e-01 -5.58399975e-01
-1.01792932e+00 -2.46614769e-01 1.26648664e+00 -8.51578176e-01
1.46001327e+00 -2.11119151e+00 1.29546821e-01 8.43730271e-01
1.15182686e+00 6.23344779e-02 -2.06861600e-01 6.86736107e-01
-4.60972100e-01 1.66932121e-01 -2.67406166e-01 -1.07701086e-01
-1.17148548e-01 2.74342328e-01 -1.07022852e-01 7.21479535e-01
4.17360067e-01 1.05148494e+00 -9.27892864e-01 -2.63446748e-01
-8.39080103e-03 1.95576921e-01 -2.83204764e-01 4.30435777e-01
-8.04023668e-02 3.89341652e-01 -2.82113224e-01 1.30257320e+00
5.05148172e-01 -8.11593175e-01 4.00880903e-01 9.99386236e-02
7.07602859e-01 -8.11492205e-02 -8.73372138e-01 9.21882510e-01
3.90249223e-01 3.70979786e-01 9.67849270e-02 -1.27950299e+00
9.57847953e-01 1.85194820e-01 2.41085708e-01 -9.68398094e-01
7.80515000e-02 1.23789936e-01 3.09069782e-01 -3.69562060e-01
1.31055743e-01 4.25633788e-01 -3.50141078e-02 4.11796212e-01
4.66099620e-01 6.07340336e-01 4.16568279e-01 6.31634355e-01
2.10598016e+00 -7.47475386e-01 5.00969410e-01 -2.72470951e-01
7.31159925e-01 -3.24403107e-01 6.28834903e-01 8.02146912e-01
-4.41552281e-01 2.46030912e-01 1.12233937e+00 -9.75005925e-01
-7.70549595e-01 -1.23321736e+00 4.05981570e-01 1.23945355e+00
8.47594440e-02 -6.65406287e-01 -7.09610701e-01 -1.19794536e+00
2.94610262e-01 3.68780047e-01 -8.58778238e-01 -7.50538290e-01
-5.20672977e-01 -5.79887927e-01 5.62506318e-01 6.33257151e-01
3.41833204e-01 -1.29507339e+00 1.11432359e-01 6.21727817e-02
2.82553852e-01 -1.08116448e+00 -3.82968903e-01 -7.23581463e-02
-1.01297987e+00 -1.57697523e+00 1.03015028e-01 -7.56458998e-01
1.10729146e+00 -8.54992568e-02 1.68555951e+00 8.56350005e-01
-6.16777301e-01 7.50850737e-01 -3.80990416e-01 -1.02763645e-01
-7.10688055e-01 7.71365464e-02 -2.43397057e-02 4.04448956e-01
4.35134530e-01 -9.76919591e-01 -4.13646817e-01 1.91917866e-01
-8.92837405e-01 -6.34341180e-01 6.88564658e-01 7.03604460e-01
5.08240163e-01 3.63844544e-01 6.26496851e-01 -1.24784160e+00
8.88987243e-01 -7.99919367e-01 -5.07087886e-01 1.04163699e-01
-7.88651943e-01 -2.01424614e-01 7.35275209e-01 6.38659224e-02
-3.01304191e-01 -4.17799205e-01 4.60239835e-02 -6.17720246e-01
-4.58057880e-01 4.58071232e-01 -2.97799319e-01 -2.35318780e-01
7.29399920e-01 4.64482248e-01 -1.04959048e-02 -7.29988664e-02
-1.48729356e-02 -5.05176932e-02 5.25347769e-01 -2.04228535e-01
9.44545805e-01 3.29570144e-01 2.00333551e-01 -9.35091972e-01
-4.13845658e-01 -5.78635693e-01 -7.71827161e-01 -4.03015196e-01
8.05842936e-01 -7.51912177e-01 -3.68779778e-01 7.21134305e-01
-6.84234500e-01 -4.83360589e-01 -2.62808502e-01 -2.55523443e-01
-2.66592324e-01 6.54596150e-01 -1.00723422e+00 -3.69431883e-01
-5.79062104e-01 -6.10229075e-01 1.01261079e+00 -8.55626445e-03
-1.41204938e-01 -1.69005358e+00 3.33823562e-01 -8.83853659e-02
3.53311539e-01 7.75722444e-01 1.20802498e+00 -1.21683073e+00
-6.08029008e-01 -5.01479089e-01 -3.01101059e-01 2.22786009e-01
1.41229659e-01 -3.43442410e-02 -7.11910307e-01 -7.83465743e-01
-2.62014955e-01 -1.51141137e-01 9.68640506e-01 1.78245455e-01
1.19020128e+00 -5.86857975e-01 -3.57848614e-01 5.63448668e-01
1.26502013e+00 -2.08740123e-02 4.80036825e-01 -3.67187709e-02
1.25884819e+00 3.76712203e-01 1.88892689e-02 3.29852462e-01
4.50742543e-01 9.64909047e-02 8.92517090e-01 -1.93330213e-01
-2.66770516e-02 -2.87973493e-01 3.67664397e-01 9.22459960e-01
1.12264864e-01 -3.58334452e-01 -1.26015437e+00 7.47489274e-01
-1.97807658e+00 -9.09355104e-01 -2.55531609e-01 1.77985072e+00
-1.14074975e-01 2.30338186e-01 3.63510877e-01 9.72452313e-02
9.91084278e-01 5.66642880e-01 -6.82723343e-01 -5.66893041e-01
-5.34427315e-02 -2.12284714e-01 4.34401259e-02 3.42467129e-01
-1.38876987e+00 6.81230843e-01 5.60672760e+00 3.84608120e-01
-8.15864682e-01 7.06179589e-02 7.34604537e-01 4.40885037e-01
-6.17975183e-02 -1.84867755e-01 -1.20184042e-01 3.18361849e-01
1.13405275e+00 1.29626198e-02 2.33408079e-01 9.43593383e-01
-2.50263572e-01 4.81825382e-01 -1.05427337e+00 7.32306480e-01
1.97331816e-01 -1.12776804e+00 2.01415315e-01 3.82039934e-01
6.03520334e-01 3.04405451e-01 -1.11255467e-01 4.21776086e-01
4.32933927e-01 -1.18110204e+00 -6.39871359e-02 6.63176954e-01
2.65889317e-01 -7.67724216e-01 8.50962341e-01 2.42839590e-01
-1.51644063e+00 -2.66611159e-01 -2.79719323e-01 1.27666280e-01
-3.50926012e-01 8.06459785e-01 -8.47276509e-01 6.96586668e-01
9.11162376e-01 1.15576971e+00 -1.08250499e+00 7.30535448e-01
-1.95401505e-01 7.44164407e-01 2.25394145e-01 5.17371178e-01
2.39134640e-01 -2.27962211e-01 9.23230290e-01 1.16681063e+00
1.06400199e-01 -3.05742145e-01 5.11529624e-01 8.94730330e-01
-3.26554596e-01 -9.72714275e-02 -1.13794637e+00 -5.52152991e-01
2.64402509e-01 1.50554991e+00 -8.11706543e-01 -2.25776538e-01
-5.38998663e-01 1.13218462e+00 7.85098493e-01 3.69327933e-01
-4.51248705e-01 -4.21443552e-01 6.41761720e-01 3.11339349e-01
1.50793031e-01 3.28510106e-02 8.75386670e-02 -1.19737172e+00
-2.06789235e-03 -1.06589937e+00 1.23598790e+00 -2.31279477e-01
-1.80673337e+00 8.66161287e-01 -6.12764597e-01 -1.10588777e+00
-4.06884670e-01 -7.30210781e-01 -1.42136848e+00 4.48449433e-01
-9.53401268e-01 -1.28291321e+00 -8.38141918e-01 1.09741688e+00
7.74178803e-02 -6.07186556e-01 8.75839233e-01 1.87238678e-01
-8.58948886e-01 5.51317275e-01 -2.73399383e-01 7.50686765e-01
2.39018068e-01 -1.79446375e+00 8.98225605e-01 1.37797236e+00
2.28577599e-01 4.00648147e-01 2.96144098e-01 -1.02031410e+00
-1.23525822e+00 -1.64787221e+00 5.96922755e-01 -3.76186490e-01
8.46334875e-01 -4.14768100e-01 -1.43041539e+00 1.15829659e+00
1.09659463e-01 5.86309016e-01 5.99859834e-01 2.04373464e-01
-5.26402295e-01 5.36732636e-02 -1.21225023e+00 3.70961964e-01
1.12936068e+00 -7.65548110e-01 -1.49857715e-01 3.81827444e-01
5.21726727e-01 -2.90145308e-01 -1.26221693e+00 6.19721174e-01
-1.31549522e-01 -1.28724647e+00 7.49399245e-01 -8.16293418e-01
9.64747146e-02 -2.46463016e-01 3.25806737e-02 -1.50696707e+00
-8.45643938e-01 -4.39905196e-01 -1.18016589e+00 8.40231240e-01
1.99450746e-01 -1.06882823e+00 8.46738636e-01 -1.02179609e-02
-3.77764493e-01 -7.94149697e-01 -7.51161098e-01 -6.96713746e-01
-3.16448539e-01 -6.06644992e-03 5.23795545e-01 1.25753546e+00
-1.23180419e-01 3.68057042e-01 -1.27345785e-01 7.12112963e-01
7.52753735e-01 8.65900964e-02 6.78351820e-01 -1.49581611e+00
-2.33648628e-01 -6.13874197e-01 -1.24033046e+00 -1.35479778e-01
2.73595512e-01 -1.11554086e+00 -5.02325773e-01 -1.48944008e+00
1.37815893e-01 3.41563046e-01 -4.40473586e-01 5.70855200e-01
-3.61841738e-01 1.37364820e-01 -3.13824505e-01 2.30707079e-02
-9.82870996e-01 5.12074530e-01 8.62415373e-01 -2.82683671e-01
-8.68755654e-02 -2.13623699e-02 -7.63237238e-01 8.78922582e-01
9.85400200e-01 -1.88883290e-01 -3.68288219e-01 -9.97653231e-02
1.05420142e-01 -2.11169869e-01 7.11040854e-01 -1.21360111e+00
2.87126094e-01 4.03628409e-01 6.75113678e-01 -5.05033672e-01
-1.98711768e-01 -6.27197921e-01 -2.13606402e-01 7.91969299e-01
2.72723362e-02 6.62696660e-01 4.28886153e-02 9.92272735e-01
-4.56101209e-01 2.78540522e-01 7.29203701e-01 -2.54221082e-01
-9.08915341e-01 9.22656655e-01 -3.12094718e-01 1.97796017e-01
1.10608351e+00 -3.89283672e-02 -6.36313498e-01 -6.64965630e-01
-1.08455920e+00 3.31335127e-01 3.96877229e-01 5.41086435e-01
8.53330433e-01 -1.43325186e+00 -8.06068063e-01 6.37501240e-01
4.81286228e-01 -1.80778921e-01 5.09673178e-01 9.61323917e-01
-4.74855423e-01 4.04923335e-02 -1.79914370e-01 -7.72556901e-01
-1.15003574e+00 7.20834434e-01 7.74704516e-01 -7.26557016e-01
-8.77356708e-01 6.90184891e-01 2.86335856e-01 -7.77996421e-01
1.37449965e-01 2.17671450e-02 -1.22561872e-01 -2.43194550e-01
3.71652037e-01 6.03737950e-01 2.00283825e-01 -7.87992239e-01
-3.55119050e-01 9.22287330e-02 -1.86539233e-01 9.06545937e-01
1.10395098e+00 3.58039737e-02 -6.81412816e-01 2.21971005e-01
1.01096010e+00 -1.76528618e-01 -8.41450214e-01 -3.41692001e-01
4.73279238e-01 -3.11760306e-01 -1.24595985e-01 -4.29266274e-01
-1.41514802e+00 7.32728660e-01 4.22121584e-01 9.55018878e-01
1.35289347e+00 1.37106195e-01 6.80478275e-01 3.92486364e-01
-9.80105177e-02 -6.83804810e-01 5.73321700e-01 5.61459184e-01
9.25197601e-01 -1.28080058e+00 -3.16576183e-01 -3.23218673e-01
-3.76906812e-01 1.11522913e+00 1.19319963e+00 -6.34530067e-01
7.91831732e-01 -4.12598923e-02 -1.35969460e-01 -1.08592856e+00
-7.94092059e-01 -8.32142085e-02 6.60349667e-01 6.73388362e-01
2.05380186e-01 1.96467802e-01 5.05415380e-01 3.06668699e-01
9.23841894e-02 -9.33022618e-01 4.16850746e-01 6.78437889e-01
-4.38909709e-01 -7.55686641e-01 -1.65379405e-01 1.04205263e+00
-4.40550774e-01 1.16135851e-01 -1.10265028e+00 7.50203967e-01
-1.50970563e-01 9.06999707e-01 3.18119258e-01 -5.92642784e-01
3.30193341e-01 7.05534667e-02 2.51465701e-02 -6.17791295e-01
-5.66857994e-01 -1.59815341e-01 1.09361283e-01 -1.14226341e+00
2.95561016e-01 -2.92091966e-01 -9.91379440e-01 -6.28604650e-01
-1.27098663e-02 -5.42829596e-02 -2.03097746e-01 6.79342091e-01
8.33655894e-01 9.32700455e-01 5.16754031e-01 -6.03004277e-01
-2.27367863e-01 -1.02313304e+00 -1.09278035e+00 7.50611842e-01
6.03195846e-01 -4.13447261e-01 -4.63180989e-01 -4.75909799e-01] | [6.654242515563965, 5.835501670837402] |
f4ce8ab6-efa7-484d-8319-cee9b39696a1 | modeling-sequential-sentence-relation-to | 2302.01626 | null | https://arxiv.org/abs/2302.01626v1 | https://arxiv.org/pdf/2302.01626v1.pdf | Modeling Sequential Sentence Relation to Improve Cross-lingual Dense Retrieval | Recently multi-lingual pre-trained language models (PLM) such as mBERT and XLM-R have achieved impressive strides in cross-lingual dense retrieval. Despite its successes, they are general-purpose PLM while the multilingual PLM tailored for cross-lingual retrieval is still unexplored. Motivated by an observation that the sentences in parallel documents are approximately in the same order, which is universal across languages, we propose to model this sequential sentence relation to facilitate cross-lingual representation learning. Specifically, we propose a multilingual PLM called masked sentence model (MSM), which consists of a sentence encoder to generate the sentence representations, and a document encoder applied to a sequence of sentence vectors from a document. The document encoder is shared for all languages to model the universal sequential sentence relation across languages. To train the model, we propose a masked sentence prediction task, which masks and predicts the sentence vector via a hierarchical contrastive loss with sampled negatives. Comprehensive experiments on four cross-lingual retrieval tasks show MSM significantly outperforms existing advanced pre-training models, demonstrating the effectiveness and stronger cross-lingual retrieval capabilities of our approach. Code and model will be available. | ['Nan Duan', 'Daxin Jiang', 'Ming Gong', 'Yaobo Liang', 'Shunyu Zhang'] | 2023-02-03 | null | null | null | null | ['xlm-r'] | ['natural-language-processing'] | [-4.47831824e-02 -3.64272386e-01 -6.15518808e-01 -5.63466907e-01
-1.56464148e+00 -4.89776075e-01 8.81711006e-01 1.09672405e-01
-5.11586487e-01 5.73048651e-01 4.70620692e-01 -3.31365943e-01
2.07528472e-01 -4.49960947e-01 -8.62335086e-01 -3.17061365e-01
2.66633332e-02 5.01374543e-01 -8.66677612e-02 -5.32738984e-01
1.28185257e-01 3.01393047e-02 -1.20238101e+00 7.24556863e-01
8.88741314e-01 7.02831507e-01 6.03727639e-01 4.65671688e-01
-4.30886388e-01 7.20918894e-01 -4.58454698e-01 -5.60980737e-01
1.23680398e-01 -2.69924432e-01 -8.20673943e-01 -1.11976862e-01
6.79613531e-01 -3.19055051e-01 -3.87132019e-01 9.69581664e-01
6.06471658e-01 -2.01976094e-02 7.78338253e-01 -6.41004384e-01
-1.10996616e+00 9.04784262e-01 -6.60591483e-01 1.18758902e-01
3.41518074e-01 -2.15154514e-01 1.68050051e+00 -1.22904086e+00
7.69432187e-01 1.62016261e+00 4.70644474e-01 4.44136381e-01
-1.04971552e+00 -4.83200759e-01 2.22113743e-01 7.26532713e-02
-1.79270744e+00 -5.02006352e-01 5.70853353e-01 -1.22764759e-01
1.49534416e+00 1.44216880e-01 2.00626388e-01 1.12479305e+00
6.50477767e-01 1.34760153e+00 6.61660552e-01 -6.90913975e-01
-2.98247159e-01 5.20395041e-01 5.69551811e-02 6.55076921e-01
-2.31776521e-01 -2.90221542e-01 -6.64418936e-01 -8.46023709e-02
1.38618141e-01 2.30854787e-02 -2.03288928e-01 -1.40279740e-01
-8.96915793e-01 1.04857409e+00 4.91133690e-01 4.68074858e-01
-2.04990000e-01 -1.36219636e-01 8.86851072e-01 5.44658899e-01
8.58387828e-01 2.20679894e-01 -4.48299825e-01 3.60350937e-01
-1.14651954e+00 1.58587009e-01 5.50970614e-01 1.06831670e+00
7.27752447e-01 -2.39182025e-01 -3.60506594e-01 1.36174119e+00
4.92279798e-01 7.52093017e-01 9.32879031e-01 -1.71089709e-01
8.03799987e-01 5.29072821e-01 -4.56163615e-01 -8.17718744e-01
-1.76659087e-03 -5.41493654e-01 -9.27767277e-01 -7.50507116e-01
-5.23265183e-01 2.62634426e-01 -6.17213666e-01 1.74189973e+00
-1.51180089e-01 -2.99184233e-01 4.49839443e-01 8.35069418e-01
7.68745720e-01 1.03994823e+00 4.38937508e-02 -1.51790366e-01
1.24143291e+00 -1.18567896e+00 -6.12348437e-01 -3.86372834e-01
1.17232358e+00 -1.09333730e+00 1.27570236e+00 -4.81657349e-02
-1.21257436e+00 -5.56723654e-01 -1.01507390e+00 -6.28333211e-01
-5.57808101e-01 4.76952136e-01 5.38181484e-01 2.01573640e-01
-1.14439464e+00 1.94728509e-01 -6.75394058e-01 -4.45158273e-01
-3.32844034e-02 9.62887108e-02 -3.08233440e-01 -4.48152989e-01
-1.62902200e+00 1.02399790e+00 4.20065790e-01 1.33710325e-01
-8.17593157e-01 -5.54995954e-01 -1.17588055e+00 9.13757831e-02
-9.12928581e-02 -6.74644411e-01 1.07084906e+00 -7.26357937e-01
-1.21197927e+00 1.15368557e+00 -4.36151236e-01 -6.48020029e-01
1.23859145e-01 -2.98407465e-01 -4.88135427e-01 8.63016993e-02
5.16104758e-01 7.84035385e-01 7.05240250e-01 -1.01794851e+00
-4.04651672e-01 -2.20007122e-01 -5.96778654e-02 5.55758774e-01
-4.90716994e-01 3.05429727e-01 -8.94386113e-01 -6.50395215e-01
-1.94864333e-01 -7.41030991e-01 -7.65688792e-02 -5.51479578e-01
-5.44129074e-01 -6.15352035e-01 5.15796900e-01 -6.68633819e-01
1.45896494e+00 -2.05493736e+00 1.42287657e-01 -1.24617241e-01
-3.78323227e-01 1.12485752e-01 -6.60069704e-01 9.26287949e-01
1.17861949e-01 -8.87747630e-02 -1.60598248e-01 -1.05251694e+00
1.26077265e-01 2.88565040e-01 -8.76635373e-01 2.73357034e-01
3.80468100e-01 1.17728353e+00 -8.00931096e-01 -7.40737975e-01
3.09635811e-02 5.95332026e-01 -5.07000983e-01 3.22053939e-01
-1.78620696e-01 1.70540586e-02 -4.75507677e-01 4.99980807e-01
5.14507353e-01 -2.78624356e-01 4.17563230e-01 -1.60946593e-01
-4.44513466e-03 9.85879600e-01 -4.93353873e-01 2.11093068e+00
-8.99371505e-01 3.73603761e-01 -2.31353473e-02 -8.59175861e-01
8.44625115e-01 3.42115223e-01 1.65048063e-01 -1.08576703e+00
-1.66064143e-01 4.77632880e-01 -4.77956206e-01 -3.49965751e-01
9.89778996e-01 -2.54441351e-01 -4.78080809e-01 5.04931211e-01
3.36862415e-01 -2.04650253e-01 3.43690455e-01 7.22468436e-01
7.90584505e-01 -1.23046653e-03 2.38271222e-01 -3.66796732e-01
8.54876101e-01 -1.98234752e-01 2.19191194e-01 8.65093648e-01
2.84915656e-01 5.29015541e-01 8.96888822e-02 -1.70958161e-01
-9.64070499e-01 -9.74915683e-01 -4.16732341e-01 1.33024633e+00
8.21327697e-03 -8.44568551e-01 -2.81585068e-01 -6.60462677e-01
8.06684121e-02 5.88090539e-01 -3.04217994e-01 -3.07459652e-01
-7.44032323e-01 -7.16461837e-01 5.19760311e-01 2.33986616e-01
1.72249451e-01 -1.13985860e+00 3.05605352e-01 4.78739776e-02
-2.78906614e-01 -1.23672903e+00 -1.08829165e+00 1.79093167e-01
-5.61682761e-01 -5.86050987e-01 -6.59411788e-01 -1.17328680e+00
5.42686343e-01 4.42061573e-01 1.52495539e+00 -5.75326979e-02
-2.79803216e-01 3.92975181e-01 -3.70926708e-01 -5.96290231e-02
-4.35963005e-01 5.91972709e-01 2.16525689e-01 -8.52071866e-02
5.62651157e-01 -2.11872846e-01 -5.00498474e-01 -2.32841149e-01
-1.03697968e+00 -7.30115101e-02 6.98939443e-01 1.00289476e+00
8.33155394e-01 -2.96121091e-01 6.61621332e-01 -9.59436417e-01
9.61445510e-01 -5.53580344e-01 -3.84292036e-01 7.15112448e-01
-5.02434433e-01 2.38199428e-01 7.20492661e-01 -1.25441939e-01
-8.62941325e-01 -3.18205446e-01 -3.47331762e-01 -3.83043408e-01
4.20038223e-01 8.57590556e-01 -6.63536564e-02 3.32615852e-01
1.92197546e-01 7.22172618e-01 -3.03209662e-01 -7.40895092e-01
5.27280629e-01 8.87127638e-01 1.19490236e-01 -6.11519158e-01
4.45020437e-01 4.39833254e-02 -4.46784258e-01 -8.69524539e-01
-1.29628921e+00 -5.68473041e-01 -5.66189468e-01 2.48586044e-01
6.52097344e-01 -1.38710940e+00 -2.37924308e-01 -3.84555794e-02
-1.39681637e+00 -2.35445313e-02 -7.71322027e-02 4.51650858e-01
-1.49940386e-01 4.21919256e-01 -1.08031046e+00 -6.76409543e-01
-8.76417041e-01 -1.17014360e+00 1.76413667e+00 -3.38669509e-01
-1.08567081e-01 -1.28940558e+00 2.88416356e-01 2.76344806e-01
5.41940331e-01 -7.02600360e-01 1.07662499e+00 -6.11141503e-01
-5.04085660e-01 -4.03459907e-01 -8.37224126e-02 6.87072515e-01
1.22597747e-01 -3.00675511e-01 -7.49489069e-01 -6.93690777e-01
-4.54366067e-03 -8.37153018e-01 1.21416628e+00 2.14693770e-01
1.00077748e+00 -3.36220831e-01 -3.42421293e-01 4.46914136e-01
1.40402889e+00 -3.31215084e-01 2.89516896e-01 1.07174754e-01
5.89676976e-01 5.51633358e-01 3.51632833e-01 7.58502353e-03
7.37386227e-01 8.17521989e-01 -9.23029557e-02 -3.85879464e-02
-2.07231015e-01 -4.78444368e-01 7.38945484e-01 1.91651070e+00
7.22512722e-01 -2.45244637e-01 -5.45845926e-01 6.86598182e-01
-1.65163648e+00 -6.96611762e-01 2.19604582e-01 2.04490209e+00
1.22539532e+00 8.44674185e-02 -4.13827568e-01 -5.46291590e-01
2.61926562e-01 5.21816790e-01 -3.31909031e-01 -5.96349537e-01
-6.47049785e-01 2.78428167e-01 2.44554713e-01 8.70724201e-01
-1.09371424e+00 1.42235923e+00 6.20764112e+00 1.24192667e+00
-1.24994981e+00 2.13967219e-01 3.91866684e-01 -2.18666986e-01
-6.47583902e-01 -4.43481579e-02 -1.33732653e+00 2.26857066e-01
9.08935249e-01 -3.18390161e-01 1.25439823e-01 7.65415013e-01
-2.24193670e-02 2.81224787e-01 -1.09179008e+00 8.05477977e-01
5.52171230e-01 -1.22462249e+00 6.65546596e-01 -3.24227959e-01
7.81577170e-01 5.91042995e-01 1.55708909e-01 8.97022367e-01
1.40561685e-01 -7.84338236e-01 4.20677215e-01 4.49337125e-01
8.44013095e-01 -6.51687443e-01 7.28847265e-01 4.76580650e-01
-1.32056189e+00 4.43756521e-01 -6.58703923e-01 3.28953356e-01
2.97667652e-01 4.95718926e-01 -7.64865875e-01 7.55094111e-01
4.18701917e-01 1.21277881e+00 -7.19968796e-01 3.78535807e-01
-1.90675661e-01 3.64717245e-01 -1.47692710e-01 5.78789786e-02
5.09421051e-01 -1.78572848e-01 3.06386888e-01 1.73172331e+00
2.33963072e-01 -6.95607483e-01 3.66540819e-01 6.40517473e-01
-4.21956152e-01 6.79123402e-01 -8.77819419e-01 -1.49231538e-01
3.11708361e-01 1.18676078e+00 8.90300721e-02 -5.15972555e-01
-7.09647655e-01 1.26715147e+00 7.74513721e-01 4.25616682e-01
-2.71593630e-01 -2.11464405e-01 5.98601937e-01 -3.07131201e-01
2.03199863e-01 -1.94484279e-01 1.15911603e-01 -1.63986981e+00
3.20406049e-01 -8.83600831e-01 4.76953328e-01 -5.81532538e-01
-1.87859976e+00 9.91217434e-01 -3.37774843e-01 -1.34010506e+00
-4.68725443e-01 -4.24769729e-01 -1.48611784e-01 1.25337040e+00
-1.96277010e+00 -1.50485945e+00 6.71337128e-01 6.10938370e-01
9.13804233e-01 -3.89335841e-01 1.10916996e+00 6.19755626e-01
-5.81419289e-01 8.54838312e-01 3.86907876e-01 9.25728232e-02
1.05158353e+00 -1.03542721e+00 3.32473993e-01 5.78922868e-01
4.11860347e-01 1.19744706e+00 1.28649443e-01 -5.80479026e-01
-1.72127759e+00 -1.32773137e+00 1.61797237e+00 -4.40352201e-01
9.16806936e-01 -8.06450129e-01 -1.04136109e+00 8.57972085e-01
6.73675478e-01 -1.64288655e-01 7.93688238e-01 2.36851200e-01
-5.08174896e-01 -2.16731787e-01 -4.13751870e-01 7.04767704e-01
5.06972730e-01 -1.19755960e+00 -6.70610607e-01 7.51900554e-01
1.00902295e+00 -2.49802202e-01 -8.15129101e-01 4.44383025e-01
4.29831505e-01 -5.04253864e-01 1.02061856e+00 -6.94744825e-01
6.36535168e-01 3.89760621e-02 -5.10867476e-01 -1.25465786e+00
-1.19800001e-01 -1.90481707e-01 -6.07013293e-02 1.31146467e+00
5.91840208e-01 -4.83305275e-01 2.92601168e-01 -3.93840037e-02
-3.13095093e-01 -9.71778393e-01 -9.67715859e-01 -7.58090675e-01
5.46466112e-01 -5.05786538e-01 2.63170689e-01 8.50806952e-01
8.35776031e-02 1.09019625e+00 -5.56726456e-01 7.54134282e-02
4.50648129e-01 3.94003898e-01 4.50521350e-01 -5.89886725e-01
-4.55896020e-01 -4.42028582e-01 7.21774921e-02 -1.77059507e+00
8.31293106e-01 -1.60750878e+00 3.63620035e-02 -1.37390292e+00
6.60118937e-01 -2.72321433e-01 -3.86806190e-01 1.86234429e-01
-3.70099574e-01 2.43795455e-01 -2.25096103e-02 4.62208927e-01
-1.07571101e+00 1.13811147e+00 1.17709315e+00 -4.15630639e-01
1.46599516e-01 -3.11992764e-01 -5.97391784e-01 2.61782438e-01
3.24824184e-01 -4.16109473e-01 -3.34127545e-01 -9.79365885e-01
3.00080597e-01 7.50512406e-02 -1.69432685e-01 -4.51144189e-01
3.43108177e-01 3.37010086e-01 7.96350092e-02 -8.73475075e-01
5.64000785e-01 -5.78377724e-01 -4.82626736e-01 2.46505141e-01
-7.63576150e-01 1.99477255e-01 1.64429605e-01 3.38561773e-01
-8.05287540e-01 -2.25415573e-01 2.93335825e-01 -1.43332437e-01
-4.30018693e-01 3.94324839e-01 -1.81519464e-01 8.19879025e-02
3.17219019e-01 6.37082458e-01 -3.03278685e-01 -3.45802099e-01
-2.83401310e-01 5.23742497e-01 5.75804859e-02 8.39460313e-01
5.53404689e-01 -1.54088032e+00 -1.01034451e+00 3.53963077e-01
5.42585850e-01 -2.84885883e-01 3.15929383e-01 7.09427357e-01
-1.00335628e-02 1.14674115e+00 5.40128708e-01 -5.93895793e-01
-1.10739470e+00 4.90209460e-01 2.23385692e-01 -7.74819851e-01
-4.96754199e-01 9.15808022e-01 4.36686218e-01 -8.29940975e-01
1.71661317e-01 -2.77139135e-02 -6.87883496e-02 -8.70449767e-02
6.05911970e-01 -4.25480247e-01 2.83815354e-01 -8.37406158e-01
-3.88445556e-01 6.38177335e-01 -7.81366050e-01 5.48877530e-02
1.30711257e+00 -4.27156061e-01 -6.59582794e-01 7.48488665e-01
1.94037569e+00 2.88376100e-02 -5.46996593e-01 -8.27500701e-01
1.46182314e-01 -7.31641203e-02 -8.06799345e-03 -5.26342750e-01
-7.68493712e-01 1.02927196e+00 1.59688979e-01 3.19867842e-02
1.08581769e+00 3.15795332e-01 1.12972891e+00 5.43975353e-01
4.40063089e-01 -9.89712775e-01 -2.39294935e-02 9.85863805e-01
1.10174870e+00 -1.54089904e+00 -1.07405409e-01 -1.24792792e-01
-5.67301393e-01 8.20736110e-01 4.17360365e-01 -2.17462257e-01
6.27381206e-01 1.73838302e-01 9.90963057e-02 -1.10493556e-01
-1.20221245e+00 -6.03373088e-02 7.59079397e-01 -3.34105454e-02
1.02225387e+00 6.32421672e-02 -5.39577007e-01 4.55698133e-01
-3.31100613e-01 -3.88824642e-01 -2.18116179e-01 7.76908040e-01
-1.66976050e-01 -1.51171422e+00 1.12336621e-01 2.97750175e-01
-5.93006313e-01 -7.99394548e-01 -3.19729477e-01 3.97780955e-01
-3.99670303e-01 6.30775273e-01 7.20419362e-02 -2.55358249e-01
1.91958964e-01 2.99768180e-01 3.58016312e-01 -8.85995567e-01
-5.94863415e-01 1.83829874e-01 2.22974531e-02 -4.97200161e-01
-1.10626765e-01 -4.65643287e-01 -7.47140706e-01 6.26249164e-02
-1.79563180e-01 4.01283294e-01 7.00480402e-01 1.03680003e+00
5.35348594e-01 3.67741078e-01 8.07477355e-01 -6.12756729e-01
-7.32361734e-01 -1.32962894e+00 -6.27635956e-01 3.86183113e-01
4.36553389e-01 -7.76647106e-02 -3.41279089e-01 -8.20645019e-02] | [11.300262451171875, 9.80214786529541] |
70e6bbd3-c6d1-45da-ace4-c2bc9cf27672 | constraints-on-parameter-choices-for | 2206.02575 | null | https://arxiv.org/abs/2206.02575v1 | https://arxiv.org/pdf/2206.02575v1.pdf | Constraints on parameter choices for successful reservoir computing | Echo-state networks are simple models of discrete dynamical systems driven by a time series. By selecting network parameters such that the dynamics of the network is contractive, characterized by a negative maximal Lyapunov exponent, the network may synchronize with the driving signal. Exploiting this synchronization, the echo-state network may be trained to autonomously reproduce the input dynamics, enabling time-series prediction. However, while synchronization is a necessary condition for prediction, it is not sufficient. Here, we study what other conditions are necessary for successful time-series prediction. We identify two key parameters for prediction performance, and conduct a parameter sweep to find regions where prediction is successful. These regions differ significantly depending on whether full or partial phase space information about the input is provided to the network during training. We explain how these regions emerge. | ['B. Mehlig', 'K. Gustavsson', 'L. Storm'] | 2022-06-03 | null | null | null | null | ['time-series-prediction'] | ['time-series'] | [ 2.33502775e-01 1.67553514e-01 -2.25675359e-01 1.21591790e-02
2.59461194e-01 -8.09085608e-01 6.82656825e-01 -1.72875628e-01
-2.49552444e-01 7.67349958e-01 -2.61082381e-01 -3.13315362e-01
-2.04528674e-01 -5.45069754e-01 -5.18361211e-01 -1.05919099e+00
-6.50535166e-01 2.83842146e-01 3.25507700e-01 -4.77222085e-01
1.12940453e-01 7.40793347e-01 -1.27508152e+00 -2.24750146e-01
5.94200730e-01 6.11863315e-01 6.05884977e-02 9.84297693e-01
3.12270969e-01 5.55310488e-01 -5.29799759e-01 3.23311687e-01
5.08427799e-01 -8.59148264e-01 -4.65686649e-01 -2.36926913e-01
-3.73721808e-01 4.73395884e-02 -7.76123047e-01 9.84669864e-01
1.94982052e-01 7.35695958e-02 6.78313732e-01 -1.27646708e+00
4.45952788e-02 5.74515581e-01 6.30339980e-02 3.11624587e-01
6.61338717e-02 4.03155744e-01 6.86759233e-01 -3.44275981e-01
9.44632888e-01 5.64100206e-01 7.47946799e-01 6.69014513e-01
-1.61825466e+00 -6.98772073e-01 -4.76948246e-02 -3.02896947e-01
-1.30811965e+00 -6.34123862e-01 1.17394698e+00 -2.56650776e-01
8.26715112e-01 1.95600897e-01 1.21723664e+00 1.00437415e+00
9.31147635e-01 7.12480098e-02 1.13837171e+00 -3.54347378e-01
4.51245546e-01 -3.74634005e-02 -6.27943724e-02 5.96162140e-01
8.34017992e-02 6.09502316e-01 -3.84094864e-01 -3.01785797e-01
9.58755374e-01 -8.54821205e-02 -3.74557436e-01 -4.90310103e-01
-1.29633260e+00 5.12105107e-01 2.58615673e-01 6.28484070e-01
-6.24007821e-01 3.68421227e-01 1.75772280e-01 9.97401893e-01
1.69559851e-01 8.10685039e-01 -2.31983587e-01 -2.08603188e-01
-1.10400546e+00 1.97380438e-01 1.18991113e+00 4.26818788e-01
6.52804136e-01 4.62471992e-01 3.71535510e-01 1.70114920e-01
9.39058065e-02 6.97984636e-01 4.05657917e-01 -1.05124390e+00
-9.89442989e-02 3.17180097e-01 1.18700460e-01 -9.97236013e-01
-5.78866363e-01 -6.16417229e-01 -1.10857213e+00 5.80234006e-02
6.05660558e-01 -5.50552547e-01 -6.59126401e-01 1.95979679e+00
2.69794818e-02 5.21199703e-01 3.89542758e-01 8.32346439e-01
4.22477163e-02 8.83324742e-01 -4.51453924e-01 -8.82293820e-01
5.79869926e-01 -2.36531824e-01 -6.15158379e-01 -6.14095479e-02
6.43781066e-01 -3.86297405e-01 4.23751920e-01 8.88759550e-03
-1.14564383e+00 -1.81302890e-01 -1.15607452e+00 9.60813344e-01
-6.42277971e-02 -3.00833464e-01 1.67178199e-01 1.71257704e-01
-1.14473045e+00 1.17314494e+00 -1.14634073e+00 -3.13293099e-01
-5.71700811e-01 6.74487531e-01 -1.89890295e-01 7.23108470e-01
-1.50355101e+00 7.62861967e-01 3.21697980e-01 3.65763426e-01
-9.48410869e-01 -2.86886871e-01 -4.48439479e-01 5.75315543e-02
-1.90748453e-01 -4.90407646e-01 1.04569948e+00 -1.06149745e+00
-1.60294330e+00 5.43793619e-01 -2.80762970e-01 -8.20599437e-01
6.09811366e-01 6.94891453e-01 -5.44395328e-01 2.79059827e-01
-2.56641746e-01 4.71029371e-01 9.91759241e-01 -1.11446404e+00
1.55114075e-02 2.49400884e-01 -2.44619921e-01 4.49156910e-02
-2.07640007e-01 -2.26049036e-01 1.42877296e-01 -1.94946215e-01
4.07949299e-01 -1.52405262e+00 -5.28022468e-01 -2.68583894e-01
-3.78155798e-01 -2.36316379e-02 8.64262283e-01 1.67394541e-02
1.15272498e+00 -2.09800506e+00 -1.16064502e-02 8.06245565e-01
3.30609620e-01 8.22869763e-02 -7.95377344e-02 1.00580883e+00
-1.50223821e-01 1.76239178e-01 -4.77488860e-02 3.58561762e-02
-2.39628285e-01 3.03987026e-01 -7.85603225e-01 7.37582743e-01
1.63030729e-01 8.41426611e-01 -6.99516177e-01 -1.79229304e-01
-1.00662395e-01 4.33486551e-01 -5.08361340e-01 2.53210843e-01
-1.20800674e-01 9.67816591e-01 -4.71799314e-01 -3.40163298e-02
1.24193437e-01 -5.58174670e-01 4.42443371e-01 1.55335993e-01
-4.84353691e-01 2.10045561e-01 -1.01073587e+00 7.00886905e-01
-2.50644267e-01 9.27608013e-01 8.45702440e-02 -1.23686409e+00
1.09328377e+00 6.10293627e-01 8.60048354e-01 -5.16008079e-01
3.47179562e-01 3.96286339e-01 6.19558990e-01 -2.81176925e-01
2.20016807e-01 -2.92788118e-01 -7.32208341e-02 7.71000147e-01
-2.52115309e-01 -8.79782438e-02 2.20127150e-01 1.07753344e-01
1.38534117e+00 -2.26543665e-01 -1.41353635e-02 -4.70513463e-01
2.78435141e-01 1.90339014e-01 5.75812101e-01 8.46002102e-01
-2.33417884e-01 1.41387790e-01 8.33799958e-01 -5.09976387e-01
-1.58463597e+00 -7.20100522e-01 -9.22243372e-02 4.63604003e-01
3.74326319e-01 -3.49692583e-01 -5.01917481e-01 1.04206530e-02
-1.36674091e-01 9.35138986e-02 -7.69447327e-01 -5.39935827e-01
-9.26364720e-01 -3.97973657e-01 5.12957513e-01 1.56487942e-01
2.96199501e-01 -1.22364461e+00 -7.98594475e-01 4.74864930e-01
1.48768991e-01 -9.64429438e-01 -2.98005939e-01 5.07387698e-01
-1.13807166e+00 -7.72352993e-01 -6.13945305e-01 -7.17336953e-01
8.31001520e-01 1.87372733e-02 7.81680524e-01 4.05138552e-01
1.78810328e-01 2.95789897e-01 8.58689547e-02 1.77998945e-01
-9.83097494e-01 1.99483186e-01 5.16257644e-01 -2.77991802e-01
-3.29025030e-01 -1.07396257e+00 -5.88707864e-01 4.72598970e-01
-7.63124883e-01 2.15051711e-01 3.04620057e-01 8.87467623e-01
4.85211521e-01 1.04079030e-01 5.76340675e-01 -4.03560787e-01
8.19690168e-01 -3.36640865e-01 -6.87539101e-01 6.24236763e-02
-5.71273267e-01 3.26166391e-01 1.19587660e+00 -8.93175840e-01
-3.97312939e-01 2.77703583e-01 1.20545169e-02 -6.25809491e-01
9.47979391e-02 6.59805179e-01 7.31129110e-01 -3.64076823e-01
6.42461836e-01 7.43596673e-01 5.07189214e-01 1.79988563e-01
-1.26075044e-01 1.30259678e-01 2.34027401e-01 -3.81225973e-01
1.05127895e+00 3.65955472e-01 4.04405475e-01 -8.70665550e-01
-1.38895497e-01 6.40518591e-02 -7.97686458e-01 -5.70241153e-01
3.56223583e-01 -4.36122358e-01 -1.09955513e+00 3.00950885e-01
-9.55562651e-01 -5.34531415e-01 -3.47472250e-01 4.29264843e-01
-7.13707209e-01 -1.76830709e-01 -7.18465030e-01 -9.95946050e-01
-6.73549175e-02 -7.85115778e-01 1.34508833e-01 2.33926415e-01
-3.01257282e-01 -1.17388678e+00 4.96824533e-01 -8.67054939e-01
7.29781508e-01 3.01720649e-01 6.08046055e-01 -7.31147826e-01
-5.09028971e-01 -4.50412154e-01 4.09786612e-01 -9.22103748e-02
-2.08784238e-01 4.79644507e-01 -4.12350893e-01 -4.98528451e-01
5.01465127e-02 1.31243140e-01 5.67513227e-01 6.30696893e-01
4.13135141e-01 -4.56550777e-01 -6.41399562e-01 5.07725656e-01
1.18829846e+00 3.94279331e-01 3.13297302e-01 7.69199505e-02
1.66989237e-01 5.71434557e-01 -2.32319571e-02 2.66757995e-01
5.68564497e-02 2.81756967e-01 2.72156268e-01 1.39212906e-01
3.92855495e-01 -3.17376643e-01 5.35231590e-01 1.25639427e+00
-9.08458829e-02 6.94998074e-03 -1.14863908e+00 4.29744661e-01
-1.68934417e+00 -1.14371586e+00 -2.39850208e-02 2.10757613e+00
8.13464284e-01 5.96671700e-01 2.30122715e-01 1.77357376e-01
7.18962014e-01 1.68740287e-01 -7.65126586e-01 -3.40638250e-01
-2.48081163e-01 -5.58265261e-02 5.38781941e-01 5.61564505e-01
-5.99161506e-01 7.50850081e-01 7.74873877e+00 2.86488831e-02
-1.78092122e+00 -2.78713614e-01 3.03584009e-01 6.09549545e-02
-3.57986063e-01 2.65786171e-01 -5.89494765e-01 5.51366389e-01
1.41344213e+00 -6.07776880e-01 4.36818033e-01 4.15404469e-01
5.13600349e-01 6.89990595e-02 -7.94203520e-01 3.31555754e-01
-6.37730300e-01 -1.43728673e+00 -4.58578765e-01 2.17465401e-01
7.45956123e-01 1.38177484e-01 -6.07750146e-03 1.29515886e-01
1.90723285e-01 -7.36612260e-01 3.91582668e-01 7.23015904e-01
6.18508458e-01 -5.72648585e-01 4.39827710e-01 7.84336209e-01
-1.22301388e+00 -4.55882363e-02 -2.08721519e-01 -3.63560557e-01
3.34534943e-01 5.33557653e-01 -6.14171386e-01 -1.65688246e-01
7.10947067e-02 7.73988187e-01 -1.45089090e-01 9.31282461e-01
1.10373735e-01 1.12466037e+00 -7.35217869e-01 -3.73424679e-01
1.91077083e-01 -3.09058756e-01 1.00600410e+00 7.79341280e-01
3.15123707e-01 2.42442355e-01 9.47005004e-02 9.30662751e-01
3.13637048e-01 -3.78813148e-01 -1.02006149e+00 -4.50329274e-01
5.93421519e-01 9.26115334e-01 -1.03105521e+00 -2.31273115e-01
1.50391132e-01 5.28792202e-01 3.73378918e-02 6.92231417e-01
-7.18281150e-01 -4.50406224e-01 4.73796904e-01 1.63104549e-01
1.33781508e-01 -7.91526675e-01 -1.25001092e-02 -1.18351197e+00
-1.34515360e-01 -3.94000649e-01 -2.65153974e-01 -5.80158651e-01
-7.58429289e-01 7.20569611e-01 -1.88631833e-01 -1.55132890e+00
-9.09441411e-01 -7.55301192e-02 -9.13635790e-01 6.27308249e-01
-9.97895300e-01 -6.00773335e-01 1.20643876e-01 5.27211845e-01
-3.45644623e-01 -7.38649741e-02 8.24466825e-01 -1.83782890e-01
-6.27801836e-01 2.48465836e-01 2.28087634e-01 6.93557635e-02
2.35201597e-01 -9.28014398e-01 3.90941024e-01 7.33422399e-01
3.28928120e-02 8.51805151e-01 1.43122447e+00 -6.60029113e-01
-1.56841683e+00 -5.69877386e-01 8.86221111e-01 6.83321953e-02
1.06625605e+00 -3.84218097e-01 -1.12820542e+00 5.75288355e-01
1.97445333e-01 2.43812889e-01 1.84541658e-01 -2.30206117e-01
3.70463133e-02 -4.29459102e-02 -6.98556244e-01 7.92963505e-01
7.77749956e-01 -5.08696973e-01 -2.06082821e-01 1.36443332e-01
5.31637251e-01 -2.99540997e-01 -7.97206759e-01 3.65287513e-01
7.19391227e-01 -8.05597425e-01 5.32847047e-01 -6.00809455e-01
2.42504984e-01 -1.65679529e-01 2.21246958e-01 -1.36540997e+00
-3.80792707e-01 -1.16443205e+00 -3.16764206e-01 5.87050438e-01
5.26754141e-01 -1.19579268e+00 8.59221101e-01 4.91757065e-01
1.31440595e-01 -8.20479155e-01 -1.05140102e+00 -9.96547401e-01
3.30483735e-01 -9.42777283e-03 1.22334860e-01 9.29629624e-01
4.19042677e-01 3.06302290e-02 -3.62490863e-01 1.82029530e-01
2.65323013e-01 2.84878314e-01 4.96062666e-01 -1.24824762e+00
-1.51798874e-01 -6.26357377e-01 -4.10441309e-01 -1.11949015e+00
3.98597449e-01 -7.02908635e-01 9.64400843e-02 -1.08424532e+00
-3.03673178e-01 -6.47536755e-01 -2.65036851e-01 2.35697314e-01
4.19273227e-01 1.01456121e-01 2.84080654e-01 9.18422282e-01
-2.54846811e-01 4.28779513e-01 1.15384173e+00 4.73841578e-01
-6.24770105e-01 3.83554071e-01 1.64700393e-02 3.54875147e-01
1.08716905e+00 -6.20958686e-01 -2.81313032e-01 2.31789052e-01
5.88430285e-01 7.63012826e-01 4.37111169e-01 -1.15171075e+00
6.67650521e-01 -2.69000769e-01 1.78327650e-01 -4.80463982e-01
4.59659874e-01 -7.73357749e-01 5.75963497e-01 1.11038065e+00
-5.94075799e-01 3.03647041e-01 -7.07829138e-03 6.00562274e-01
-2.10696638e-01 -3.86611745e-02 9.31982994e-01 3.11783880e-01
-1.97937742e-01 1.93072602e-01 -1.01091862e+00 -1.45182863e-01
9.47261572e-01 -2.01942727e-01 -2.14262724e-01 -7.46680379e-01
-9.30740654e-01 2.63983816e-01 5.39228737e-01 -3.20672207e-02
3.27238858e-01 -1.23803353e+00 -3.86049688e-01 5.23721635e-01
-3.00388664e-01 -6.18732452e-01 -3.17782932e-03 1.14871216e+00
-2.26465330e-01 5.30171335e-01 -2.69982159e-01 -8.24599206e-01
-9.61869240e-01 2.62725413e-01 8.35432649e-01 -1.73264176e-01
-7.38329172e-01 2.01000050e-01 -4.67057854e-01 -1.69835806e-01
-1.55149803e-01 -2.58428782e-01 -3.98918279e-02 -2.23537043e-01
6.83930963e-02 -2.29954705e-01 -3.19553107e-01 -5.61815619e-01
-2.11011454e-01 5.64020932e-01 2.49325782e-01 -4.06554461e-01
1.27676892e+00 -6.02045804e-02 -9.74437594e-02 7.51841187e-01
1.07450891e+00 -1.95624396e-01 -1.43513143e+00 -2.81551063e-01
-2.17506573e-01 1.82356462e-01 -2.55361438e-01 -1.90732121e-01
-9.75392461e-01 6.71456814e-01 3.00951630e-01 1.09021711e+00
9.39386666e-01 -1.74661547e-01 6.22972190e-01 4.97955412e-01
2.96018004e-01 -8.78472626e-01 -4.36087251e-02 9.68381822e-01
6.57609642e-01 -7.21099555e-01 -2.79751420e-01 6.76280260e-02
-2.82303333e-01 1.38976943e+00 2.97979742e-01 -8.89986396e-01
1.11911726e+00 5.19344509e-01 -1.21355161e-01 -4.38902937e-02
-1.34853888e+00 1.70140862e-01 8.53068829e-02 2.25639090e-01
3.11848611e-01 -4.41388078e-02 -3.04987758e-01 9.27203745e-02
-5.48347056e-01 -1.52481109e-01 7.40056813e-01 9.78717744e-01
-6.22467935e-01 -7.89313555e-01 -6.87658489e-02 3.43372375e-01
-1.47216052e-01 2.39796877e-01 -4.31257367e-01 8.53762567e-01
-7.32971847e-01 5.50043702e-01 3.63141060e-01 -5.66559851e-01
9.66620892e-02 1.17382824e-01 2.36422956e-01 -1.05181135e-01
-5.94708800e-01 7.23323077e-02 -1.69727236e-01 -3.48831326e-01
-3.49623561e-01 -5.64914346e-01 -1.40729702e+00 -7.12163806e-01
-4.26955789e-01 3.97549778e-01 5.84478676e-01 1.01566899e+00
3.78528476e-01 3.95328909e-01 1.08605337e+00 -8.58473301e-01
-5.36836684e-01 -7.29409456e-01 -4.49199557e-01 -7.86070675e-02
8.09079766e-01 -3.81557077e-01 -9.62433219e-01 2.10995562e-02] | [6.61785364151001, 3.5565025806427] |
376c9528-5fd1-4f82-8440-3d16af4244d8 | a-maintenance-planning-framework-using-online | 2208.00808 | null | https://arxiv.org/abs/2208.00808v2 | https://arxiv.org/pdf/2208.00808v2.pdf | A Maintenance Planning Framework using Online and Offline Deep Reinforcement Learning | Cost-effective asset management is an area of interest across several industries. Specifically, this paper develops a deep reinforcement learning (DRL) solution to automatically determine an optimal rehabilitation policy for continuously deteriorating water pipes. We approach the problem of rehabilitation planning in an online and offline DRL setting. In online DRL, the agent interacts with a simulated environment of multiple pipes with distinct lengths, materials, and failure rate characteristics. We train the agent using deep Q-learning (DQN) to learn an optimal policy with minimal average costs and reduced failure probability. In offline learning, the agent uses static data, e.g., DQN replay data, to learn an optimal policy via a conservative Q-learning algorithm without further interactions with the environment. We demonstrate that DRL-based policies improve over standard preventive, corrective, and greedy planning alternatives. Additionally, learning from the fixed DQN replay dataset in an offline setting further improves the performance. The results warrant that the existing deterioration profiles of water pipes consisting of large and diverse states and action trajectories provide a valuable avenue to learn rehabilitation policies in the offline setting, which can be further fine-tuned using the simulator. | ['Hajo Molegraaf', 'Nils Jansen', 'Zaharah A. Bukhsh'] | 2022-08-01 | null | null | null | null | ['dqn-replay-dataset', 'dqn-replay-dataset'] | ['miscellaneous', 'playing-games'] | [-1.35761276e-01 1.63660347e-01 -2.69350465e-02 1.07023440e-01
-7.11981177e-01 -5.69330513e-01 -1.12567656e-01 3.79030049e-01
-2.99275577e-01 9.43945885e-01 -3.35281454e-02 -5.20355821e-01
-7.91438878e-01 -1.04309881e+00 -8.29976916e-01 -1.11028051e+00
-5.59474289e-01 7.56172121e-01 1.27476171e-01 -5.10820448e-01
2.66779870e-01 5.13050914e-01 -1.14346993e+00 -3.08904260e-01
9.38481033e-01 7.60582507e-01 5.99465549e-01 9.41154063e-01
4.94598836e-01 6.85599387e-01 -6.75805390e-01 1.28338575e-01
6.51907384e-01 -1.86375052e-01 -8.73749554e-01 5.09981513e-01
-5.49152255e-01 -1.01867080e+00 -4.55843031e-01 5.34775794e-01
8.43903720e-01 3.19751889e-01 3.67730618e-01 -1.40012813e+00
-3.17345440e-01 6.43372595e-01 -3.12193930e-01 3.10534723e-02
7.67057613e-02 8.16030145e-01 9.07733083e-01 -3.90467467e-03
-9.91276279e-02 1.22034407e+00 4.68574673e-01 4.65728909e-01
-9.94905114e-01 -2.17204001e-02 3.21923882e-01 2.63377964e-01
-6.05987191e-01 3.96216005e-01 3.46100360e-01 -1.68755293e-01
1.11034155e+00 -1.78850457e-01 1.08497202e+00 8.44003558e-01
4.81739342e-01 9.17127788e-01 7.87831128e-01 -2.64160395e-01
9.04450834e-01 -4.80939656e-01 -3.53173047e-01 2.54669160e-01
2.74232447e-01 2.76082426e-01 3.88235450e-02 -1.64957881e-01
9.55604494e-01 1.98066995e-01 -1.27957180e-01 -4.32809561e-01
-7.95118868e-01 7.49597847e-01 3.27039629e-01 -4.18554842e-01
-8.60225141e-01 3.32968503e-01 4.74670917e-01 5.87661803e-01
-9.19382274e-02 7.74509966e-01 -8.67392004e-01 -2.16371104e-01
-2.02886701e-01 5.90049863e-01 1.15061522e+00 9.38554347e-01
4.68975902e-01 3.28379750e-01 -3.03025752e-01 5.79799414e-01
2.13744059e-01 6.70003355e-01 3.56109329e-02 -1.80274880e+00
6.79120660e-01 1.38004065e-01 9.64791179e-01 -1.84653953e-01
-7.34992862e-01 -3.25914830e-01 -5.03508449e-01 4.43232775e-01
3.17732662e-01 -7.76465178e-01 -7.74034739e-01 1.37981379e+00
3.62970114e-01 9.26155150e-02 3.10824990e-01 7.74103463e-01
-4.92020369e-01 7.44353414e-01 -5.81618957e-02 -3.60660911e-01
9.42806125e-01 -8.29767406e-01 -6.27822518e-01 5.27922623e-03
7.51622140e-01 -5.86780682e-02 1.02355707e+00 7.29435325e-01
-1.29199529e+00 6.07389845e-02 -8.40738416e-01 8.25489283e-01
3.07408944e-02 -2.64363348e-01 2.53023118e-01 5.50878346e-01
-1.07809365e+00 1.00254607e+00 -1.15004468e+00 -2.22653359e-01
5.49036622e-01 4.90984321e-01 4.08184469e-01 -3.22300941e-01
-1.16870713e+00 1.08981383e+00 2.78268278e-01 4.52906400e-01
-1.79093349e+00 -7.84157157e-01 -6.89876020e-01 2.06647381e-01
8.16688359e-01 -5.45157671e-01 1.95770979e+00 -1.80377543e-01
-1.79375577e+00 -3.22618008e-01 6.80349410e-01 -5.61373711e-01
6.75164461e-01 -4.29679811e-01 1.43806696e-01 4.72698092e-01
-1.41724974e-01 1.88127179e-02 4.65791494e-01 -1.48752832e+00
-9.11557496e-01 -2.96470094e-02 5.92008948e-01 4.28265512e-01
-4.22258168e-01 -5.01387477e-01 1.87846795e-01 -2.38391250e-01
-3.12178314e-01 -1.01412845e+00 -8.83260906e-01 -2.02605993e-01
-2.81381935e-01 -3.17493349e-01 5.41013837e-01 -7.05549896e-01
8.01461101e-01 -1.53093326e+00 1.43224627e-01 2.57103145e-01
-4.41307098e-01 6.36864826e-02 -3.02764356e-01 1.11243117e+00
2.55809754e-01 -1.98820412e-01 -4.91781563e-01 -8.37915987e-02
2.05145374e-01 7.56849229e-01 -9.65447128e-02 4.84416455e-01
3.08220714e-01 4.88045871e-01 -1.38250697e+00 -8.80944431e-02
1.85318202e-01 -2.16242492e-01 -5.84939301e-01 6.01356864e-01
-4.50435370e-01 4.66253728e-01 -7.42694557e-01 7.72745728e-01
5.48371136e-01 3.23776063e-03 3.78185004e-01 2.02496320e-01
-9.77573022e-02 -3.16250622e-01 -1.18049598e+00 1.41498435e+00
-8.91468883e-01 9.13718622e-03 3.77721101e-01 -1.57347369e+00
7.85434484e-01 4.44492251e-01 1.05029070e+00 -7.68434286e-01
-4.42625955e-02 1.64568096e-01 -1.85381472e-01 -1.20031583e+00
5.06253123e-01 -2.17855230e-01 -2.47800753e-01 6.19431257e-01
-1.61207944e-01 -4.04179841e-01 1.63233072e-01 -1.34801865e-01
1.68308330e+00 1.23490922e-01 -2.24301189e-01 -1.17687635e-01
1.20232858e-01 5.45680858e-02 7.20750272e-01 8.84357214e-01
-2.32118443e-01 -5.65265529e-02 7.23144710e-01 -1.22274138e-01
-1.23187232e+00 -1.14848840e+00 1.28582060e-01 7.29514837e-01
4.88540322e-01 4.14338112e-01 -6.55467629e-01 -4.34902906e-01
3.90274763e-01 8.26863825e-01 -3.12607646e-01 -2.96807528e-01
-1.03878295e+00 -9.22523320e-01 2.44964566e-02 8.07378590e-01
3.38122845e-01 -1.16918194e+00 -1.05802572e+00 8.58517945e-01
1.97229564e-01 -7.45464861e-01 -2.31183559e-01 4.20334965e-01
-1.06468976e+00 -1.23952758e+00 -8.68258834e-01 -7.39965796e-01
7.04077244e-01 5.93328662e-02 8.89209330e-01 1.26525059e-01
-1.23727091e-01 8.17078590e-01 -3.79432023e-01 -2.78479546e-01
-4.90220547e-01 6.99529871e-02 9.24441963e-03 -3.68136525e-01
-3.91399562e-01 -4.13554132e-01 -1.12579238e+00 5.30606389e-01
-9.22359288e-01 -5.30695736e-01 4.96293455e-01 9.14140761e-01
3.81276011e-01 6.34348810e-01 1.18661952e+00 -5.70780456e-01
9.09772158e-01 -7.15182722e-01 -7.31851935e-01 4.56276238e-01
-6.21974826e-01 1.20208368e-01 9.03463602e-01 -3.55716825e-01
-1.08183193e+00 2.68563982e-02 -6.98864982e-02 -1.31187186e-01
7.90498331e-02 3.90984505e-01 -2.44527400e-01 3.18449885e-01
2.87610978e-01 4.14109975e-02 1.42623037e-01 -5.24116099e-01
-2.56629474e-02 6.34122908e-01 2.95138568e-01 -1.10274768e+00
7.33991027e-01 1.16578192e-01 1.69551373e-01 -1.62380457e-01
-7.45287538e-01 -3.03370357e-01 -3.39200914e-01 -4.42257196e-01
4.20055121e-01 -8.23236704e-01 -1.32157075e+00 7.95211196e-01
-7.79357672e-01 -1.20565343e+00 -7.11099982e-01 1.66658357e-01
-1.20604527e+00 2.71467119e-01 -8.67019057e-01 -1.40823781e+00
-2.02416793e-01 -9.80636775e-01 7.85881698e-01 4.45703626e-01
5.35724998e-01 -1.01776886e+00 2.33341545e-01 5.63759394e-02
3.69600564e-01 5.05569041e-01 7.92769313e-01 -1.73542544e-01
-5.82331002e-01 -3.56893614e-02 3.30983073e-01 4.71633077e-01
3.26359779e-01 -3.41255724e-01 -3.58011365e-01 -8.91913712e-01
-1.09892271e-01 -6.44589067e-01 2.45171458e-01 4.77678776e-01
1.36280382e+00 -6.19773507e-01 -2.43380100e-01 -1.16343416e-01
1.54862773e+00 6.61212087e-01 4.50502485e-01 9.41475153e-01
2.70147890e-01 6.45418167e-01 1.21153009e+00 1.16548693e+00
7.14093566e-01 2.36194566e-01 1.19597077e+00 1.77034423e-01
6.96329713e-01 -1.04657315e-01 4.80577648e-01 2.64393419e-01
-1.34232631e-02 -4.52215463e-01 -1.05232120e+00 8.64573956e-01
-2.09810400e+00 -9.41947103e-01 4.58919942e-01 2.24524522e+00
9.17542934e-01 2.90347278e-01 4.97675568e-01 2.72691488e-01
5.88973999e-01 -5.03711581e-01 -1.39520657e+00 -4.25143421e-01
1.99995890e-01 1.37075633e-02 9.59545195e-01 2.38564089e-01
-7.32635498e-01 1.75202966e-01 6.28268623e+00 2.16473594e-01
-6.32817030e-01 -3.14603776e-01 6.07511640e-01 -1.19957395e-01
-1.92199320e-01 -9.42324996e-02 -5.53263426e-01 4.94237721e-01
1.42861605e+00 -1.04523428e-01 7.52148569e-01 6.16011322e-01
1.02437735e+00 -1.43074125e-01 -9.92273152e-01 2.32229501e-01
-7.17312396e-01 -1.02879381e+00 -5.45096278e-01 7.27838129e-02
7.83157051e-01 -6.75403923e-02 -2.92116135e-01 5.33843875e-01
9.87911463e-01 -6.45431757e-01 7.09900498e-01 6.39542520e-01
1.33750677e-01 -1.07885194e+00 1.04959929e+00 5.06764770e-01
-1.06117308e+00 -1.25135076e+00 -2.02536181e-01 -1.46762118e-01
8.32410336e-01 2.29009345e-01 -8.12178254e-01 6.50552332e-01
8.23537946e-01 5.90053499e-01 6.38049282e-03 1.45480359e+00
-6.91163242e-02 7.36235142e-01 -2.38659829e-01 -2.27968261e-01
2.93902189e-01 -2.03037798e-01 2.35548094e-01 6.46236122e-01
3.61337215e-01 6.41396120e-02 8.02316129e-01 4.09757614e-01
3.64533424e-01 -5.35297096e-01 -1.36884466e-01 2.11045146e-01
6.23728156e-01 8.20830643e-01 -5.23281753e-01 7.81180486e-02
-1.87043294e-01 4.98549640e-01 3.38152610e-02 4.24451023e-01
-7.61645675e-01 -4.61554438e-01 7.02544868e-01 -1.62211433e-02
4.20065314e-01 -4.03563023e-01 -1.05064183e-01 -4.36091423e-01
-1.28535479e-01 -6.56573474e-01 3.42690647e-01 -6.38173401e-01
-1.59089577e+00 -7.15532601e-02 2.18478411e-01 -1.48386478e+00
-4.18417394e-01 -5.72125077e-01 -7.56240070e-01 6.20347679e-01
-1.93998373e+00 -5.06304145e-01 -1.08737528e-01 4.01487648e-01
8.07342529e-01 9.78921205e-02 4.79933470e-01 4.27240133e-02
-8.31013858e-01 1.98966026e-01 6.84349954e-01 -1.37021706e-01
1.79088354e-01 -1.76759827e+00 1.26666039e-01 4.15017486e-01
-1.08581543e+00 -1.90428257e-01 1.06061053e+00 -6.21705115e-01
-1.82752645e+00 -1.14139402e+00 -3.76580447e-01 2.78459266e-02
8.72466385e-01 1.28075734e-01 -1.01943111e+00 2.67666757e-01
2.77055830e-01 -1.90035567e-01 1.22168496e-01 -5.40415704e-01
7.00715661e-01 -3.21747333e-01 -1.35387349e+00 5.14027059e-01
8.24859679e-01 1.02030821e-01 -1.10849388e-01 6.09852195e-01
1.00603330e+00 -6.48301125e-01 -1.28789830e+00 1.91952422e-01
2.19107777e-01 -3.20612580e-01 7.32073665e-01 -7.41787493e-01
2.82538831e-01 4.57340963e-02 7.45451450e-02 -1.81347716e+00
-8.69733095e-02 -9.03716505e-01 -4.11739409e-01 9.46608245e-01
3.02596271e-01 -6.50872171e-01 8.25057089e-01 7.62614667e-01
-4.88944650e-01 -1.17515659e+00 -1.01235795e+00 -9.52480078e-01
4.30640638e-01 -4.08688113e-02 8.31293941e-01 2.83872396e-01
-3.44822071e-02 -3.07742804e-01 -3.31688970e-01 7.05284774e-01
8.39850187e-01 1.61841258e-01 4.26940084e-01 -7.15736508e-01
-5.36045313e-01 1.89552918e-01 2.65529513e-01 -1.03753448e+00
4.32284363e-02 -9.17395279e-02 6.94500268e-01 -1.90356195e+00
-1.39189586e-01 -7.77986825e-01 -4.44964617e-01 6.66601717e-01
9.67187360e-02 -8.06609094e-01 1.63380131e-01 3.17483842e-02
-5.34561932e-01 1.02097964e+00 1.70035887e+00 -3.16467226e-01
-4.19705033e-01 6.55881703e-01 -3.77787441e-01 1.28811479e-01
1.19799554e+00 -4.30347264e-01 -6.30549252e-01 -5.33064365e-01
2.42788374e-01 8.67932498e-01 2.75045216e-01 -6.39807999e-01
2.53780577e-02 -7.96675026e-01 -1.87711164e-01 -5.64986050e-01
7.89430961e-02 -9.40706849e-01 -2.42829233e-01 1.13285565e+00
-3.26045364e-01 3.64308178e-01 1.14034951e-01 1.14602065e+00
2.01229647e-01 -5.78986406e-01 5.33833444e-01 -3.37289751e-01
-5.20703137e-01 5.26723087e-01 -6.03481770e-01 3.76188084e-02
1.12967992e+00 -1.51313692e-01 -3.81165981e-01 -4.71760422e-01
-7.06744671e-01 1.24989915e+00 1.13275856e-01 2.33214468e-01
5.72503984e-01 -8.63188386e-01 -7.14698613e-01 -3.66091043e-01
-4.17741328e-01 4.63825911e-01 4.41782176e-01 5.22706509e-01
-4.75518912e-01 -1.33939356e-01 -3.00456703e-01 -5.08393049e-01
-3.21027666e-01 4.20044214e-01 8.22081208e-01 -5.21954060e-01
-8.35914552e-01 1.66904882e-01 -4.16231513e-01 -4.05300915e-01
2.22834408e-01 -3.56826484e-01 -3.08270361e-02 -4.77257110e-02
2.33137488e-01 8.05423379e-01 2.14709178e-01 3.07852834e-01
5.36955111e-02 1.22962825e-01 1.40509695e-01 -1.65302351e-01
2.10719466e+00 -2.37990856e-01 4.12078112e-01 1.95590988e-01
6.29544258e-01 -9.25054908e-01 -2.31957698e+00 -3.77654396e-02
2.07365692e-01 -3.60194534e-01 3.33338859e-04 -7.15068161e-01
-1.26589382e+00 3.98582935e-01 6.63896918e-01 7.08928645e-01
1.29080772e+00 -3.23170334e-01 9.37006295e-01 6.01368308e-01
5.76726019e-01 -1.56226611e+00 6.09963059e-01 4.69177425e-01
8.82470250e-01 -9.02619064e-01 -2.07324922e-01 3.93065542e-01
-7.84956634e-01 1.15211558e+00 8.94055605e-01 -3.95699620e-01
7.09773719e-01 4.86934036e-01 -1.56183183e-01 1.84921790e-02
-9.48573232e-01 1.51600922e-03 -9.47372139e-01 7.36172199e-01
-4.11846191e-01 1.00741468e-01 8.41598138e-02 3.81166488e-01
1.86227530e-01 -7.54346773e-02 1.11387849e+00 1.51266801e+00
-5.50149739e-01 -1.11282325e+00 -4.16238457e-01 4.94737744e-01
9.13468227e-02 6.46550119e-01 5.71455836e-01 7.06417441e-01
-3.31051797e-01 1.00933456e+00 4.05322425e-02 2.13172451e-01
9.21010792e-01 -4.95986372e-01 3.28421265e-01 -5.52897751e-01
-6.51535749e-01 -2.50391871e-01 1.02027848e-01 -4.89270210e-01
-3.21128458e-01 -9.32227552e-01 -1.69918549e+00 -1.15401722e-01
-4.01652813e-01 1.03210002e-01 4.54727620e-01 1.14293528e+00
7.20026568e-02 8.33199024e-01 1.43799603e+00 -1.04434502e+00
-1.69745386e+00 -7.00878680e-01 -9.47222888e-01 4.00074795e-02
5.42571366e-01 -7.66177297e-01 -3.17744672e-01 -2.90548325e-01] | [4.438408851623535, 2.251450300216675] |
d349acfe-96af-49d8-9555-c1dcf59a8333 | d3vo-deep-depth-deep-pose-and-deep | 2003.01060 | null | https://arxiv.org/abs/2003.01060v2 | https://arxiv.org/pdf/2003.01060v2.pdf | D3VO: Deep Depth, Deep Pose and Deep Uncertainty for Monocular Visual Odometry | We propose D3VO as a novel framework for monocular visual odometry that exploits deep networks on three levels -- deep depth, pose and uncertainty estimation. We first propose a novel self-supervised monocular depth estimation network trained on stereo videos without any external supervision. In particular, it aligns the training image pairs into similar lighting condition with predictive brightness transformation parameters. Besides, we model the photometric uncertainties of pixels on the input images, which improves the depth estimation accuracy and provides a learned weighting function for the photometric residuals in direct (feature-less) visual odometry. Evaluation results show that the proposed network outperforms state-of-the-art self-supervised depth estimation networks. D3VO tightly incorporates the predicted depth, pose and uncertainty into a direct visual odometry method to boost both the front-end tracking as well as the back-end non-linear optimization. We evaluate D3VO in terms of monocular visual odometry on both the KITTI odometry benchmark and the EuRoC MAV dataset.The results show that D3VO outperforms state-of-the-art traditional monocular VO methods by a large margin. It also achieves comparable results to state-of-the-art stereo/LiDAR odometry on KITTI and to the state-of-the-art visual-inertial odometry on EuRoC MAV, while using only a single camera. | ['Rui Wang', 'Nan Yang', 'Lukas von Stumberg', 'Daniel Cremers'] | 2020-03-02 | d3vo-deep-depth-deep-pose-and-deep-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Yang_D3VO_Deep_Depth_Deep_Pose_and_Deep_Uncertainty_for_Monocular_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Yang_D3VO_Deep_Depth_Deep_Pose_and_Deep_Uncertainty_for_Monocular_CVPR_2020_paper.pdf | cvpr-2020-6 | ['monocular-visual-odometry'] | ['robots'] | [-5.35462499e-01 2.65934691e-02 -4.13113624e-01 -5.00565171e-01
-5.57921410e-01 -2.73999125e-01 5.93530536e-01 -4.98251498e-01
-6.10428572e-01 6.34375095e-01 2.10805312e-01 1.65277630e-01
4.22746807e-01 -5.78993261e-01 -1.20921075e+00 -5.21926880e-01
4.28861141e-01 8.92591238e-01 3.12066734e-01 5.27529269e-02
1.11306265e-01 4.57364172e-01 -1.67050755e+00 -4.55804676e-01
7.76593924e-01 1.33187604e+00 2.77555585e-01 6.94699347e-01
2.45384410e-01 9.15843785e-01 -4.17117700e-02 -1.40299112e-01
5.82008123e-01 1.25644013e-01 -4.26331013e-01 9.54168066e-02
1.46723628e+00 -7.62128413e-01 -9.48203981e-01 1.16045713e+00
6.85461938e-01 -6.91483393e-02 5.64299464e-01 -1.16145730e+00
-5.22016585e-01 -2.26870775e-01 -4.40626472e-01 -7.81197026e-02
5.14271855e-01 5.54202676e-01 7.25460231e-01 -1.13607061e+00
1.11755180e+00 1.46197021e+00 7.54607379e-01 4.53118861e-01
-1.22765708e+00 -5.76035857e-01 -5.65760210e-02 2.42838711e-01
-1.44995415e+00 -3.57473254e-01 5.54233432e-01 -7.32301533e-01
1.33941591e+00 -5.18357158e-01 8.70403230e-01 8.61287832e-01
5.66978455e-01 7.99691200e-01 8.34897995e-01 3.24990340e-02
1.27825558e-01 -9.21568424e-02 -1.55891806e-01 9.74212110e-01
2.34942943e-01 7.66073942e-01 -8.22749496e-01 3.45180690e-01
8.76470089e-01 3.15641090e-02 -3.68111789e-01 -1.28391540e+00
-1.04798186e+00 6.80663168e-01 7.95952201e-01 -5.41449726e-01
-7.60989562e-02 5.65299809e-01 3.11982602e-01 1.75092757e-01
6.01353109e-01 5.89104779e-02 -5.60878277e-01 -2.12732896e-01
-8.01813722e-01 1.95855767e-01 4.89931494e-01 1.17412758e+00
1.47722328e+00 4.49917495e-01 2.61666119e-01 4.74637717e-01
8.29341352e-01 1.14948642e+00 4.30320650e-01 -1.31400371e+00
6.57936811e-01 6.18921161e-01 3.36796381e-02 -6.08084142e-01
-5.01300871e-01 -3.06944072e-01 -6.83739364e-01 7.88784385e-01
1.92266643e-01 -7.90362954e-02 -1.25020790e+00 1.52312732e+00
5.11781335e-01 2.08565354e-01 1.77196786e-01 1.24074674e+00
1.07242489e+00 2.81862348e-01 -6.32348955e-01 2.85848916e-01
8.33173871e-01 -1.17709565e+00 -6.36742890e-01 -8.88985097e-01
5.42558908e-01 -5.86825311e-01 6.63892984e-01 3.71227592e-01
-9.73225892e-01 -8.26482475e-01 -1.55751896e+00 -7.54163504e-01
-7.20712766e-02 1.19181924e-01 3.71319979e-01 4.67858136e-01
-1.33525336e+00 5.75036526e-01 -9.12117720e-01 -2.57156968e-01
6.28432259e-02 3.93063903e-01 -5.44394076e-01 -3.22822422e-01
-9.86450315e-01 1.03369892e+00 2.86098868e-01 3.70909348e-02
-1.44593918e+00 -8.16754520e-01 -1.74395251e+00 -4.13191348e-01
1.05033182e-01 -1.32664204e+00 1.17745173e+00 -5.53891301e-01
-1.81716800e+00 1.27278316e+00 -3.74099225e-01 -6.31659865e-01
7.85044789e-01 -6.97820127e-01 7.37270564e-02 5.51397465e-02
2.77391553e-01 1.33386779e+00 7.73238122e-01 -1.21912682e+00
-7.74811208e-01 -7.57836282e-01 -9.99741852e-02 6.34049177e-01
4.67766702e-01 -9.10911977e-01 -8.46369982e-01 1.43247306e-01
6.16791785e-01 -1.18963945e+00 -3.58210988e-02 4.20074940e-01
-3.11017722e-01 2.73566186e-01 7.71212876e-01 -3.11085165e-01
4.44995284e-01 -1.91632271e+00 3.37896377e-01 -2.77472228e-01
3.54870081e-01 -3.06559280e-02 2.29563326e-01 -1.91084191e-01
1.33341506e-01 -7.43027449e-01 2.20754161e-01 -1.13573527e+00
1.49308860e-01 4.49024498e-01 -1.63092747e-01 1.07561827e+00
-1.90918148e-01 1.06879807e+00 -9.27753031e-01 -3.31231326e-01
9.57333148e-01 6.81970179e-01 -5.66638112e-01 3.25034499e-01
-2.04619527e-01 6.61246061e-01 2.99611300e-01 8.21456492e-01
1.00586891e+00 -6.09655678e-02 -1.39286295e-01 -2.99827546e-01
-1.62546396e-01 5.43751717e-01 -1.26257241e+00 2.30691004e+00
-5.04537046e-01 9.94553506e-01 -1.09132230e-01 -2.07536399e-01
8.76057446e-01 -2.01250851e-01 2.37910047e-01 -9.08859074e-01
2.41403580e-01 4.24954116e-01 -5.35468698e-01 -1.54005155e-01
8.54635954e-01 -5.58823533e-03 1.07784502e-01 -2.25808591e-01
3.91400665e-01 -9.31746602e-01 -9.36917588e-02 1.33190513e-01
6.05119050e-01 8.58303070e-01 3.43670040e-01 -1.00106619e-01
6.55847907e-01 -8.41125250e-02 7.28385091e-01 2.59415805e-01
-3.22613239e-01 9.23971713e-01 -2.92792264e-02 -4.56343412e-01
-1.17262197e+00 -1.32327676e+00 -2.04797119e-01 3.05633307e-01
7.37004638e-01 -1.01736955e-01 -3.73586625e-01 -2.99130589e-01
5.86110890e-01 1.01800054e-01 -5.98335683e-01 2.25419030e-02
-2.67240733e-01 -1.48101315e-01 4.85022068e-01 8.13058436e-01
7.22152770e-01 -5.96503139e-01 -4.27196145e-01 -7.21818283e-02
-1.90271307e-02 -1.53975523e+00 -5.80871701e-01 3.89144719e-01
-1.07645941e+00 -1.01545060e+00 -5.65160751e-01 -6.48635268e-01
2.96943218e-01 4.20898765e-01 1.22130692e+00 -5.96817195e-01
4.81528975e-02 4.54765737e-01 3.24292332e-01 -1.89386815e-01
-2.70559173e-02 1.55320596e-02 3.73816073e-01 -4.66571569e-01
4.66622770e-01 -4.95449454e-01 -8.61775160e-01 3.99504244e-01
-4.88169670e-01 -6.17610887e-02 2.32930958e-01 8.03320348e-01
8.60158801e-01 -8.56653810e-01 -4.69474047e-01 -2.79712975e-01
-4.04438823e-01 -1.30466715e-01 -1.26730001e+00 -5.39966106e-01
-9.22987759e-01 3.95465195e-01 2.25415736e-01 -1.07215932e-02
-8.94237936e-01 4.78295922e-01 -1.24905244e-01 -1.07494128e+00
8.97753537e-02 4.04458381e-02 -1.59627020e-01 -4.63062823e-01
7.92806685e-01 1.34169698e-01 1.37661129e-01 -1.10403217e-01
3.35636765e-01 4.21352446e-01 9.19248939e-01 -5.29188663e-02
1.04005277e+00 1.13225305e+00 4.21165973e-01 -6.43513262e-01
-9.11103606e-01 -8.77834141e-01 -9.47247803e-01 -9.25649032e-02
1.01284647e+00 -1.90824544e+00 -8.34700823e-01 8.87697041e-01
-1.24193513e+00 -5.69819093e-01 -2.05126747e-01 6.98548853e-01
-8.32253814e-01 6.29865468e-01 -6.87826991e-01 -4.89733130e-01
-3.33270878e-01 -1.43839896e+00 1.65420842e+00 1.45566180e-01
1.13845788e-01 -1.16726220e+00 6.07540667e-01 3.27171743e-01
-2.49520078e-01 1.04465581e-01 -9.09698978e-02 4.49860364e-01
-1.25434732e+00 5.39341196e-02 -3.32031101e-01 5.14315486e-01
-1.92123905e-01 -2.33353525e-01 -1.35295975e+00 -4.56542104e-01
-1.45037964e-01 -6.62858963e-01 1.10968196e+00 6.13876641e-01
2.99962461e-01 5.29362559e-01 -1.74646094e-01 1.61823809e+00
1.73163664e+00 -5.36654890e-02 8.55022907e-01 7.90193200e-01
1.37748814e+00 1.46751568e-01 8.52769613e-01 3.06354135e-01
1.05540586e+00 9.75151718e-01 1.15682626e+00 -8.54664221e-02
-1.71889216e-01 -5.53121328e-01 6.00114942e-01 5.54300189e-01
2.30943456e-01 1.54330522e-01 -7.76857436e-01 6.53975427e-01
-1.96221471e+00 -4.30769831e-01 -2.57209152e-01 2.34523559e+00
5.38089037e-01 1.15790084e-01 -2.43510351e-01 -6.38438016e-02
2.35547915e-01 5.66025496e-01 -8.22549701e-01 -2.35727191e-01
-4.00369287e-01 -6.92400113e-02 1.15013385e+00 1.01491714e+00
-1.18918419e+00 1.25226617e+00 5.49319077e+00 2.27090150e-01
-1.34519005e+00 1.66783497e-01 1.60835221e-01 -2.78193504e-01
-5.86546883e-02 -3.18155587e-02 -1.37569940e+00 1.02078423e-01
5.75337052e-01 2.19744250e-01 3.50920081e-01 1.30169535e+00
8.56069624e-02 -4.24970388e-01 -1.39001489e+00 1.55587244e+00
3.42719942e-01 -1.36100936e+00 -2.68073261e-01 3.46161544e-01
1.24919868e+00 9.86442387e-01 -1.32269524e-02 2.12375984e-01
5.40042341e-01 -8.44739079e-01 1.00151241e+00 2.75881112e-01
1.04838109e+00 -5.28308451e-01 8.80896032e-01 3.01679015e-01
-1.30166638e+00 7.24555850e-02 -7.62501717e-01 -3.95100474e-01
4.63885695e-01 6.24411464e-01 -6.17303073e-01 7.27432191e-01
8.58507395e-01 1.36372447e+00 -4.23964679e-01 9.04042780e-01
-8.09209585e-01 -3.59165192e-01 -3.08484405e-01 6.21145546e-01
4.33788598e-01 -1.72798350e-01 3.96215349e-01 7.09138036e-01
3.11363995e-01 -5.03123343e-01 1.04633778e-01 6.03667378e-01
-1.60332203e-01 -3.25117111e-01 -9.08207655e-01 7.50870466e-01
2.59843588e-01 8.45406115e-01 2.99520463e-01 -4.32549536e-01
-3.62211138e-01 1.17150509e+00 5.23583055e-01 1.26044109e-01
-5.74946582e-01 1.39279410e-01 1.37572920e+00 9.03916955e-02
3.93839568e-01 -4.29453880e-01 -4.20296550e-01 -1.56911004e+00
4.76000644e-02 -4.36925918e-01 5.49319796e-02 -1.32422721e+00
-9.51388299e-01 4.47254837e-01 -3.78289163e-01 -1.57059550e+00
-7.00111389e-01 -9.69520569e-01 -7.91729167e-02 9.20640528e-01
-2.22208738e+00 -9.00509894e-01 -1.06098735e+00 5.77351391e-01
2.85254151e-01 5.80612533e-02 2.97889441e-01 3.50928485e-01
-1.31264731e-01 2.80146688e-01 2.11478293e-01 -2.12778449e-02
1.13407946e+00 -1.48179758e+00 9.15104866e-01 9.13874209e-01
9.75663662e-02 1.78237692e-01 6.81604922e-01 -6.44874990e-01
-1.58157682e+00 -1.08872736e+00 7.80782461e-01 -8.07317138e-01
4.89823312e-01 -2.82108605e-01 -5.07022500e-01 9.95674729e-01
1.89629477e-02 5.02321959e-01 -2.75605619e-01 -3.80781293e-01
-3.06082070e-01 -3.60747755e-01 -1.06345224e+00 4.63639975e-01
1.41909361e+00 -9.69039083e-01 -3.84315938e-01 8.48219469e-02
8.70537400e-01 -1.37439001e+00 -6.94207251e-01 4.61965263e-01
7.63678551e-01 -1.45482171e+00 1.24026906e+00 1.20025173e-01
4.81626570e-01 -5.58497727e-01 -3.38356376e-01 -1.33326209e+00
-2.34915670e-02 -4.39801514e-01 -3.14263463e-01 6.01145566e-01
-1.07716657e-01 -7.00527549e-01 1.26303089e+00 1.78382397e-01
-3.88002872e-01 -3.66248190e-01 -1.23064756e+00 -9.31081295e-01
-1.13575593e-01 -7.30852365e-01 3.14002365e-01 5.36070228e-01
-3.84833574e-01 3.79459351e-01 -7.81538665e-01 4.16989893e-01
1.03243077e+00 -2.15177000e-01 1.57453847e+00 -1.15008855e+00
-2.35752791e-01 1.03287481e-01 -1.15684950e+00 -1.99684441e+00
3.72306854e-01 -5.93778193e-01 3.04987043e-01 -1.44219387e+00
-2.70055473e-01 1.79792598e-01 2.84875929e-01 -1.11005105e-01
-2.23589353e-02 5.14505923e-01 8.72144029e-02 1.91837087e-01
-3.72680008e-01 9.24527943e-01 1.32522833e+00 -1.66615531e-01
-2.54103869e-01 -3.61531645e-01 1.45343199e-01 8.68200004e-01
1.77142456e-01 -4.06137675e-01 -4.72784728e-01 -7.12777853e-01
3.91974688e-01 1.21265411e-01 5.57595015e-01 -1.37677181e+00
2.97831327e-01 8.74806568e-02 5.21350145e-01 -1.14646006e+00
7.91966915e-01 -7.65088797e-01 1.03163436e-01 6.51513338e-01
4.34253246e-01 -1.77981183e-02 1.38438016e-01 4.75729138e-01
-3.65230203e-01 1.65157109e-01 1.06667542e+00 -1.59467477e-02
-1.23032820e+00 6.43483758e-01 1.91864505e-01 2.01592818e-01
6.44677162e-01 -6.06429994e-01 -6.06596589e-01 -5.76169312e-01
-3.25743705e-01 3.75351816e-01 1.13063800e+00 4.13187712e-01
7.52289116e-01 -1.42844987e+00 -1.34232655e-01 4.67599630e-01
5.66637278e-01 6.60887897e-01 2.37053022e-01 8.75056028e-01
-1.02166879e+00 6.98608100e-01 -2.28087321e-01 -1.38898993e+00
-1.07305431e+00 4.22375381e-01 7.68468916e-01 -1.61417544e-01
-6.29830241e-01 7.31040061e-01 5.99288046e-01 -1.00734115e+00
3.27016115e-01 -7.54731894e-01 2.31417239e-01 -2.42346987e-01
2.02242404e-01 4.38510299e-01 6.20197831e-03 -8.77650917e-01
-4.92263854e-01 1.21484983e+00 3.67313325e-01 -1.61361665e-01
1.01277220e+00 -6.46947920e-01 2.52842903e-01 5.14788270e-01
1.36544979e+00 -2.49608979e-01 -2.09809351e+00 -4.71364468e-01
-4.47404921e-01 -7.74267852e-01 3.87173504e-01 -3.53281528e-01
-1.14380920e+00 1.13059080e+00 8.52359653e-01 -8.32530916e-01
6.27527893e-01 -2.33995050e-01 7.37896621e-01 4.42551017e-01
7.04764783e-01 -1.01542711e+00 2.28761941e-01 1.09757972e+00
6.02785468e-01 -1.82208681e+00 3.43770593e-01 -3.21436763e-01
-6.01920664e-01 9.06200230e-01 9.19728756e-01 -3.82338494e-01
4.78497922e-01 1.57189220e-01 6.67324543e-01 -1.32997051e-01
-5.01892388e-01 -4.02532101e-01 4.60286111e-01 7.56643951e-01
-3.41729335e-02 -2.95266926e-01 3.73106837e-01 -4.67918754e-01
-1.99375317e-01 1.43825546e-01 4.71668661e-01 7.91877031e-01
-3.19032818e-01 -7.80557871e-01 -4.33217049e-01 -2.77039945e-01
1.24770470e-01 -1.78188175e-01 -1.89086944e-01 9.78208303e-01
2.17524409e-01 4.27926540e-01 1.67145967e-01 -5.20967126e-01
3.53009760e-01 -2.32576579e-01 6.85101271e-01 -5.56872904e-01
-2.21476391e-01 -8.27521309e-02 3.14367074e-03 -1.08974636e+00
-3.65865916e-01 -5.97074449e-01 -1.22703016e+00 -5.26273489e-01
-3.84668517e-03 -6.57791972e-01 8.99143994e-01 1.02148652e+00
2.58602202e-01 2.25323483e-01 4.34907168e-01 -1.63986135e+00
-4.17937785e-01 -9.06508327e-01 -6.25883102e-01 1.85036883e-01
7.78319001e-01 -9.96914566e-01 -6.00132585e-01 -2.79583812e-01] | [8.142760276794434, -2.2437946796417236] |
5499acf3-5dc1-4165-a9b6-101453404d73 | adequacy-of-the-gradient-descent-method-for | 1704.01704 | null | http://arxiv.org/abs/1704.01704v2 | http://arxiv.org/pdf/1704.01704v2.pdf | Adequacy of the Gradient-Descent Method for Classifier Evasion Attacks | Despite the wide use of machine learning in adversarial settings including
computer security, recent studies have demonstrated vulnerabilities to evasion
attacks---carefully crafted adversarial samples that closely resemble
legitimate instances, but cause misclassification. In this paper, we examine
the adequacy of the leading approach to generating adversarial samples---the
gradient descent approach. In particular (1) we perform extensive experiments
on three datasets, MNIST, USPS and Spambase, in order to analyse the
effectiveness of the gradient-descent method against non-linear support vector
machines, and conclude that carefully reduced kernel smoothness can
significantly increase robustness to the attack; (2) we demonstrate that
separated inter-class support vectors lead to more secure models, and propose a
quantity similar to margin that can efficiently predict potential
susceptibility to gradient-descent attacks, before the attack is launched; and
(3) we design a new adversarial sample construction algorithm based on
optimising the multiplicative ratio of class decision functions. | ['Benjamin I. P. Rubinstein', 'Yi Han'] | 2017-04-06 | null | null | null | null | ['computer-security'] | ['miscellaneous'] | [ 3.72911394e-01 2.31287464e-01 -1.71963692e-01 -3.74892712e-01
-6.62680447e-01 -1.08806503e+00 9.60438371e-01 1.19824652e-02
-3.83706331e-01 7.60569334e-01 -1.54075503e-01 -1.01007211e+00
-2.29330540e-01 -9.90508735e-01 -6.40626132e-01 -4.64424938e-01
-3.66247684e-01 3.31239194e-01 3.24889988e-01 -5.37843704e-01
5.64027369e-01 8.33920836e-01 -9.38117266e-01 3.13128561e-01
7.70630240e-01 6.65429235e-01 -9.34105992e-01 9.62990522e-01
3.58943641e-01 7.45135128e-01 -9.35618997e-01 -9.40663278e-01
7.48179197e-01 -5.70822895e-01 -8.51478398e-01 -1.66649923e-01
4.10793722e-01 -3.27019662e-01 -3.84732753e-01 1.11370540e+00
3.80839080e-01 -1.27138002e-02 8.65925729e-01 -1.51663768e+00
-7.38798320e-01 5.71549058e-01 -1.37521088e-01 2.46416360e-01
1.98684558e-01 4.23522711e-01 7.58028984e-01 -2.99133331e-01
4.88199919e-01 1.37888145e+00 9.14339840e-01 9.02965307e-01
-1.38662887e+00 -8.26574087e-01 -9.70878825e-02 -1.29249230e-01
-8.89294863e-01 -4.04908687e-01 8.33317637e-01 -3.84395063e-01
3.94623607e-01 7.01792777e-01 2.07513109e-01 1.45213556e+00
2.06170797e-01 4.96349096e-01 1.36832905e+00 -4.57609296e-01
4.43329334e-01 5.91671109e-01 7.09065199e-02 6.22498035e-01
4.50919181e-01 4.67139840e-01 1.14738367e-01 -1.05031943e+00
4.85549748e-01 -1.31415918e-01 -1.51490897e-01 -4.11468267e-01
-6.38317525e-01 1.34514797e+00 2.45961651e-01 2.77360290e-01
-6.27121404e-02 -2.24807058e-02 5.91780484e-01 6.90484941e-01
5.31782269e-01 8.28570187e-01 -4.51181471e-01 3.30387950e-01
-6.43109977e-01 4.41638380e-01 1.15843260e+00 3.32935393e-01
3.75129670e-01 4.02967274e-01 -1.93722285e-02 5.62492549e-01
7.71883577e-02 4.95736718e-01 4.20696735e-01 -5.68757713e-01
5.89439511e-01 1.54741749e-01 1.84618220e-01 -1.12293971e+00
2.67956108e-02 -3.20206136e-01 -6.10880971e-01 8.08034956e-01
7.23591506e-01 -3.87244523e-01 -7.55759895e-01 1.57523835e+00
3.87568831e-01 2.69701064e-01 2.40810111e-01 5.30813396e-01
4.38536443e-02 1.69733807e-01 1.25691965e-01 1.35746464e-01
9.04534221e-01 -4.38109159e-01 -1.45610601e-01 -1.44240156e-01
7.00854242e-01 -5.11542618e-01 1.15225101e+00 4.27194238e-01
-8.37249517e-01 -1.30364284e-01 -1.31944132e+00 7.30796218e-01
-5.85309505e-01 -4.73820418e-01 8.22515249e-01 1.51031709e+00
-6.64520264e-01 1.08234775e+00 -5.31388819e-01 4.46022376e-02
5.62054694e-01 4.09255803e-01 -5.04428089e-01 2.30818033e-01
-1.46605301e+00 9.34121609e-01 2.37944975e-01 -2.84749568e-01
-9.49855268e-01 -7.32054412e-01 -7.50349641e-01 -8.89348090e-02
1.29710361e-01 -4.05803531e-01 8.94081056e-01 -1.15423310e+00
-1.27666712e+00 8.04863274e-01 2.95511127e-01 -8.64421487e-01
9.21856046e-01 3.64270508e-02 -6.00608647e-01 2.80104727e-02
-1.90143436e-01 5.43092825e-02 1.41820550e+00 -1.41931713e+00
-3.16362053e-01 -3.47579479e-01 2.50214577e-01 -3.47052395e-01
-6.17010534e-01 2.03219071e-01 5.27750552e-01 -9.88803744e-01
-3.73571724e-01 -1.14829171e+00 -5.95537782e-01 -2.64797002e-01
-7.81305909e-01 7.70201981e-02 9.42299962e-01 -4.92937684e-01
1.06745780e+00 -2.02730107e+00 -1.90236613e-01 7.96656549e-01
1.52591825e-01 8.54797363e-01 -4.19914983e-02 3.40343177e-01
-3.70442152e-01 5.13690352e-01 -4.49908495e-01 -1.59953646e-02
8.75577554e-02 6.86429814e-02 -9.68876898e-01 7.21975565e-01
3.58464241e-01 7.03349471e-01 -7.84997463e-01 -2.94951182e-02
9.37976763e-02 2.40989015e-01 -6.12952113e-01 8.49392563e-02
-3.64916585e-02 1.80276006e-01 -6.47157133e-01 6.14108145e-01
6.73771441e-01 -3.19477692e-02 1.68629028e-02 1.84510961e-01
5.98358214e-01 7.08152056e-02 -1.12579310e+00 6.20839059e-01
-2.57124126e-01 5.57541490e-01 9.23360363e-02 -1.10112405e+00
8.33418667e-01 -1.58131346e-02 6.03923313e-02 -1.64142758e-01
2.39551932e-01 2.58272737e-01 -2.07897779e-02 -1.47416934e-01
1.21918410e-01 -1.49136052e-01 -2.63172090e-01 5.89829683e-01
-2.29724407e-01 -8.02525356e-02 -1.85174853e-01 2.95002818e-01
1.24180853e+00 -3.75641495e-01 1.03870966e-01 -2.52446294e-01
8.35845470e-01 3.21131274e-02 2.69124240e-01 1.11765194e+00
-2.94598132e-01 1.91680253e-01 7.82755613e-01 -7.23556221e-01
-1.09002721e+00 -1.16384828e+00 -1.58474565e-01 1.09347391e+00
-5.49289286e-02 -4.86604050e-02 -9.92186546e-01 -1.49043465e+00
5.82724035e-01 9.66136694e-01 -8.59806895e-01 -5.00258803e-01
-6.09874308e-01 -1.13208258e+00 1.15688288e+00 3.19399893e-01
1.93186983e-01 -9.20384109e-01 -1.91135053e-02 -1.75565518e-02
4.88098174e-01 -7.26699889e-01 -3.85529429e-01 2.18115270e-01
-6.99489534e-01 -1.43390107e+00 -2.57502317e-01 -3.90272141e-01
7.84965992e-01 -1.27686188e-01 1.00424755e+00 4.27482963e-01
-4.60634172e-01 3.82817388e-01 -3.86565864e-01 -6.27617240e-01
-1.19294190e+00 2.30595060e-02 3.54891926e-01 1.11183807e-01
3.09130132e-01 -5.83934903e-01 -3.34206045e-01 4.15106446e-01
-1.04278815e+00 -8.83399546e-01 3.92690331e-01 8.84046614e-01
-1.93330422e-02 1.22663163e-01 8.62495065e-01 -1.48578560e+00
1.03638041e+00 -5.71761131e-01 -5.34867704e-01 2.33435526e-01
-8.86261761e-01 6.67029321e-02 1.14050651e+00 -8.56677115e-01
-5.57401776e-01 -2.00274363e-01 -2.61100084e-01 -2.84950644e-01
-2.47653261e-01 -1.83731928e-01 -1.32628933e-01 -8.30884159e-01
1.34659183e+00 2.06214443e-01 2.19793037e-01 -1.62681699e-01
4.44156945e-01 5.42722344e-01 3.17467421e-01 -7.72497416e-01
1.51275492e+00 5.60455799e-01 1.24971412e-01 -7.77647853e-01
-6.33789420e-01 2.51866542e-02 -2.69117296e-01 1.22786447e-01
3.67704511e-01 -4.01553869e-01 -6.21895671e-01 4.46316779e-01
-7.64154434e-01 -4.44204956e-01 -1.97860599e-01 2.03915477e-01
-4.68502760e-01 6.89663947e-01 -7.62609363e-01 -9.43401814e-01
-3.31700236e-01 -9.91805553e-01 4.71519768e-01 -7.80166015e-02
-2.79470384e-01 -1.38138020e+00 1.52394071e-01 3.53612244e-01
5.06319165e-01 6.53093278e-01 8.70286524e-01 -1.52500832e+00
-1.11313298e-01 -7.67570436e-01 1.24995768e-01 7.53398299e-01
2.68602315e-02 7.22373500e-02 -1.06148291e+00 -6.12994730e-01
1.38144031e-01 -4.80842710e-01 6.48384750e-01 -8.41433853e-02
1.12497020e+00 -7.67267048e-01 -1.65285408e-01 7.53251135e-01
1.27909029e+00 -1.16097823e-01 7.53714621e-01 6.53008103e-01
4.91752386e-01 6.82906628e-01 3.55816513e-01 3.46725613e-01
-3.41382712e-01 3.59150916e-01 5.54137766e-01 -1.69129759e-01
3.20909292e-01 -2.18557134e-01 4.00812566e-01 -1.54436365e-01
1.77107975e-01 5.41715091e-03 -4.86990690e-01 1.73667282e-01
-1.39162135e+00 -1.28745234e+00 -1.31905871e-03 2.25628877e+00
9.09865022e-01 6.75236583e-01 4.94990945e-01 3.41628015e-01
5.53114295e-01 1.37228131e-01 -3.83208066e-01 -7.54455268e-01
-3.93401599e-03 5.62153757e-01 9.43590164e-01 6.93006933e-01
-1.37297344e+00 9.23649490e-01 7.24042416e+00 1.11254108e+00
-9.38528597e-01 -1.44307464e-01 9.43037868e-01 1.17509700e-01
-2.47410730e-01 5.96691594e-02 -7.83299625e-01 6.24894798e-01
9.71288085e-01 -2.07649156e-01 4.46633190e-01 1.19617474e+00
-1.59230202e-01 5.29651880e-01 -9.87711966e-01 1.01934500e-01
9.33778360e-02 -1.36595130e+00 4.06965613e-02 2.29890853e-01
6.14116490e-01 -4.09529388e-01 3.84678960e-01 4.03307945e-01
8.38106573e-01 -1.13290870e+00 4.60929692e-01 1.08063355e-01
4.39475745e-01 -9.96147215e-01 6.18490815e-01 3.85126561e-01
-5.76141059e-01 -2.44244263e-01 -3.86683106e-01 3.07066947e-01
-1.98767170e-01 4.93158668e-01 -9.91826892e-01 2.74398893e-01
3.02500695e-01 1.29328417e-02 -7.46489346e-01 5.80246985e-01
-2.03689307e-01 9.98008311e-01 -1.72093675e-01 -1.45686835e-01
3.45917821e-01 -1.41799748e-01 6.42773092e-01 1.29813123e+00
-1.80347428e-01 -2.16049895e-01 2.01873869e-01 7.70334780e-01
1.39918372e-01 -3.10506914e-02 -8.58859897e-01 2.97366064e-02
4.88627762e-01 1.02481091e+00 -5.80893397e-01 -1.30196318e-01
-3.28253210e-02 1.03085649e+00 2.48959690e-01 3.06581169e-01
-8.31940532e-01 -8.16468060e-01 8.48356247e-01 1.71745062e-01
2.20191300e-01 9.08633415e-03 -3.64531308e-01 -9.77329016e-01
-6.38060048e-02 -1.35272586e+00 4.69949394e-01 -1.64423604e-02
-1.73951483e+00 5.94015956e-01 -2.45206371e-01 -1.07950020e+00
-5.42204738e-01 -7.84359992e-01 -9.82337117e-01 1.05000579e+00
-1.10878789e+00 -1.11761308e+00 4.76622462e-01 7.70612836e-01
1.41448706e-01 -6.60048306e-01 8.68076861e-01 -4.33556885e-02
-3.91343862e-01 1.17367566e+00 1.71016887e-01 5.58091581e-01
5.76629996e-01 -1.36205828e+00 7.88435876e-01 8.81633401e-01
1.77272648e-01 8.76448095e-01 9.28541541e-01 -8.04015934e-01
-1.11399591e+00 -1.15607238e+00 4.23644155e-01 -9.92640853e-01
9.98330891e-01 -4.18416381e-01 -9.57457662e-01 5.19163251e-01
-3.32242280e-01 1.37242094e-01 8.44867826e-01 1.52296945e-02
-7.86844254e-01 1.31533980e-01 -1.89749932e+00 7.50162482e-01
6.13323092e-01 -6.05726421e-01 -4.82867122e-01 5.87299407e-01
6.76106036e-01 -4.55858074e-02 -7.35184431e-01 2.12423205e-01
4.44802076e-01 -1.07425356e+00 1.45890939e+00 -1.46427214e+00
1.06256202e-01 3.94564644e-02 1.81914195e-02 -1.21744955e+00
-3.35873991e-01 -1.04653811e+00 -2.57579148e-01 1.04292035e+00
5.30653894e-01 -1.03024900e+00 1.22440600e+00 6.01404727e-01
3.81775916e-01 -7.86453784e-01 -8.74392152e-01 -9.92338359e-01
4.48716134e-01 -2.31552139e-01 4.47487652e-01 1.22784257e+00
-1.14432737e-01 -2.41082594e-01 -5.44453442e-01 3.86302352e-01
9.05248821e-01 -4.32968110e-01 1.02777004e+00 -1.10402393e+00
-6.69948220e-01 -4.57987130e-01 -6.48989499e-01 -3.61706108e-01
4.85635430e-01 -7.64430225e-01 -3.93372476e-01 -4.77861524e-01
-3.29083443e-01 -7.80940652e-01 -1.96782455e-01 4.60741460e-01
-5.43020010e-01 3.77702147e-01 -2.99413893e-02 1.99796528e-01
-1.60848185e-01 2.83464082e-02 7.90273070e-01 -5.39543070e-02
-7.70765841e-02 5.14335811e-01 -9.93449569e-01 8.06578219e-01
8.38555038e-01 -6.77835405e-01 -3.74069721e-01 3.10047686e-01
2.42167599e-02 -2.56907910e-01 6.51876628e-01 -8.17882538e-01
-2.57584780e-01 -3.54576737e-01 5.38530707e-01 1.37691617e-01
-2.74011698e-02 -7.45085597e-01 -2.21143708e-01 8.65898013e-01
-6.35434985e-01 -1.02271043e-01 5.83860129e-02 8.97321522e-01
3.09731394e-01 -5.76386213e-01 1.02736330e+00 -3.62261869e-02
-1.54638320e-01 3.59154612e-01 -2.82093018e-01 2.79860795e-01
1.10927474e+00 -4.46500145e-02 -3.86052936e-01 -3.95648748e-01
-5.36029935e-01 -1.34178072e-01 6.44764662e-01 2.66465813e-01
4.16718304e-01 -1.22666728e+00 -9.04915154e-01 5.62601864e-01
-1.86657518e-01 -8.24024975e-01 -1.80114314e-01 2.87823588e-01
-5.88765860e-01 8.18042755e-02 -1.80795789e-02 -2.29809824e-02
-1.44582713e+00 8.15487444e-01 4.01264220e-01 -3.55766863e-01
-4.51005042e-01 9.14536059e-01 -2.88950086e-01 -5.17892659e-01
1.72588080e-01 4.01748359e-01 5.34964204e-02 -4.06678885e-01
6.24999762e-01 4.41013753e-01 1.40735051e-02 -4.98105496e-01
-2.72459865e-01 -8.24162811e-02 -4.29267466e-01 1.02948174e-02
1.07445943e+00 5.67173839e-01 1.92992501e-02 -2.12809563e-01
1.25473440e+00 4.59627271e-01 -1.05796778e+00 1.37686767e-02
1.14716798e-01 -8.40882957e-01 -4.86659288e-01 -5.61722636e-01
-6.53272212e-01 4.46738183e-01 4.76171345e-01 9.59153712e-01
7.03229427e-01 -3.29253912e-01 7.09486008e-01 5.92613280e-01
2.80519277e-01 -8.86329472e-01 8.86895433e-02 2.15706453e-01
5.35440803e-01 -1.20130646e+00 2.44393885e-01 -3.96362990e-01
-6.51537359e-01 1.04967964e+00 3.34550977e-01 -6.72386825e-01
5.46265066e-01 1.93656489e-01 -8.56082886e-03 8.32281262e-02
-3.94417703e-01 5.34382880e-01 3.01232547e-01 9.42696214e-01
-3.83361466e-02 1.14708841e-02 -8.67959931e-02 4.59572792e-01
-3.70303571e-01 -4.14947420e-01 6.41347766e-01 9.61941481e-01
-3.67939293e-01 -1.42863011e+00 -7.59539425e-01 6.36183560e-01
-8.87994111e-01 -4.76377755e-02 -8.20920169e-01 8.01136553e-01
-1.86405703e-01 1.02870512e+00 -4.79745388e-01 -6.37591302e-01
2.53121346e-01 1.08792685e-01 3.19402039e-01 -3.48514467e-01
-1.04722714e+00 -6.15170956e-01 2.84398586e-01 -4.23835248e-01
2.16357753e-01 -5.46424448e-01 -5.85183680e-01 -7.62210906e-01
-5.34566760e-01 4.34214473e-01 5.58498919e-01 8.48100245e-01
1.30297452e-01 6.26259595e-02 1.25296080e+00 -6.55193090e-01
-1.62444794e+00 -6.40402317e-01 -6.43933237e-01 8.77710462e-01
3.83485436e-01 -4.10179019e-01 -1.22008920e+00 -1.55206054e-01] | [5.7113823890686035, 7.682979106903076] |
ac32c6dc-60b0-4b84-8d1f-ed7338bb0ba7 | few-shots-is-all-you-need-a-progressive-few | 2107.10064 | null | https://arxiv.org/abs/2107.10064v3 | https://arxiv.org/pdf/2107.10064v3.pdf | Few Shots Are All You Need: A Progressive Few Shot Learning Approach for Low Resource Handwritten Text Recognition | Handwritten text recognition in low resource scenarios, such as manuscripts with rare alphabets, is a challenging problem. The main difficulty comes from the very few annotated data and the limited linguistic information (e.g. dictionaries and language models). Thus, we propose a few-shot learning-based handwriting recognition approach that significantly reduces the human labor annotation process, requiring only few images of each alphabet symbol. The method consists in detecting all the symbols of a given alphabet in a textline image and decoding the obtained similarity scores to the final sequence of transcribed symbols. Our model is first pretrained on synthetic line images generated from any alphabet, even though different from the target domain. A second training step is then applied to diminish the gap between the source and target data. Since this retraining would require annotation of thousands of handwritten symbols together with their bounding boxes, we propose to avoid such human effort through an unsupervised progressive learning approach that automatically assigns pseudo-labels to the non-annotated data. The evaluation on different manuscript datasets show that our model can lead to competitive results with a significant reduction in human effort. The code will be publicly available in this repository: \url{https://github.com/dali92002/HTRbyMatching} | ['Beáta Megyesi', 'Yousri Kessentini', 'Alicia Fornés', 'Mohamed Ali Souibgui'] | 2021-07-21 | null | null | null | null | ['handwriting-recognition'] | ['computer-vision'] | [ 4.56664383e-01 -1.66715249e-01 -5.30316867e-02 -3.00763160e-01
-9.16630447e-01 -6.81852341e-01 4.45346504e-01 7.03309402e-02
-4.27029431e-01 8.94331098e-01 -1.10050708e-01 -9.80190337e-02
1.81565389e-01 -5.43742120e-01 -6.77588761e-01 -6.33811891e-01
6.17283285e-01 7.79906988e-01 4.10044461e-01 -1.05640925e-01
6.17278636e-01 6.38466477e-01 -1.43293440e+00 2.44213328e-01
9.90596712e-01 7.73748994e-01 6.52851641e-01 8.14072490e-01
-4.98319805e-01 7.17783630e-01 -7.64320374e-01 -5.09255767e-01
2.53652453e-01 -5.77688754e-01 -5.13373673e-01 5.83435118e-01
4.50659484e-01 -3.07434201e-01 -4.11346793e-01 1.08305538e+00
6.32644296e-01 2.12635204e-01 6.10062838e-01 -5.62126338e-01
-5.45420229e-01 4.62265521e-01 -5.16229093e-01 1.12785343e-02
1.06645718e-01 -1.62878148e-02 7.04627991e-01 -1.28666973e+00
9.64874148e-01 7.41890490e-01 4.65449542e-01 4.63381082e-01
-7.94805825e-01 -4.77221310e-01 -2.88873702e-01 3.47476721e-01
-1.51049376e+00 -6.47020280e-01 6.95465922e-01 -3.90831113e-01
5.17205298e-01 1.38945147e-01 1.43571898e-01 1.14776337e+00
-2.09201545e-01 6.10097587e-01 8.98926735e-01 -1.03383970e+00
3.34226042e-01 2.85735250e-01 4.95162904e-01 9.74956751e-01
1.63996324e-01 -5.54422259e-01 -7.30155945e-01 7.24500865e-02
7.63332963e-01 -7.17924014e-02 -1.25066951e-01 -1.82865053e-01
-1.02664852e+00 5.35981238e-01 -1.79411545e-01 4.10435349e-01
-1.06577985e-01 -2.75729567e-01 5.70922017e-01 1.66411877e-01
1.73390880e-01 2.86754221e-01 -3.92043926e-02 -3.63433182e-01
-1.26564312e+00 -2.12569445e-01 8.74255300e-01 1.30806947e+00
7.45009422e-01 -2.22919900e-02 -6.87298775e-02 1.27086627e+00
-1.14546575e-01 5.09662151e-01 5.44126868e-01 -5.50234795e-01
6.02212250e-01 4.52872574e-01 2.56576389e-01 -9.77271736e-01
8.75220075e-02 -3.32371853e-02 -7.79373646e-01 -2.49418640e-03
7.31677771e-01 -4.40063551e-02 -1.17162824e+00 9.52640533e-01
-2.87038833e-02 -4.79700305e-02 -2.68922038e-02 8.13548267e-01
5.05211294e-01 7.96754718e-01 -3.59252155e-01 -2.11038038e-01
1.39045703e+00 -1.19086468e+00 -7.43004382e-01 -3.49997222e-01
4.76980686e-01 -1.09134603e+00 1.25740445e+00 5.51441133e-01
-7.41356194e-01 -6.05480015e-01 -1.21359730e+00 -1.49732426e-01
-4.09860849e-01 7.99290717e-01 2.52692401e-01 5.87673187e-01
-5.25426090e-01 6.55804157e-01 -8.07418585e-01 -5.61767757e-01
5.31866610e-01 7.74282217e-02 -2.82601655e-01 -2.80400813e-01
-8.11002254e-01 7.44837940e-01 5.80040157e-01 2.67384470e-01
-6.70200765e-01 -1.96133286e-01 -6.87846482e-01 2.26191469e-02
6.58932209e-01 2.81010717e-01 1.03115821e+00 -1.05720842e+00
-1.53358734e+00 7.77883232e-01 -2.82454789e-01 -2.51102298e-01
7.73696303e-01 -1.90975219e-01 -2.50012904e-01 3.21949363e-01
-1.11409187e-01 3.36479962e-01 1.11212015e+00 -1.13155293e+00
-4.29445028e-01 -1.63396135e-01 -6.75135553e-01 1.01765648e-01
-6.59679234e-01 1.96663618e-01 -1.04247248e+00 -8.77218664e-01
-4.51535843e-02 -8.56258631e-01 -3.53416055e-02 1.94930583e-01
-3.84305865e-01 -9.58503410e-03 7.94010699e-01 -1.02331138e+00
1.11479414e+00 -2.19631648e+00 -8.13374370e-02 2.63817966e-01
-2.77105629e-01 4.59235013e-01 -2.15259731e-01 4.92472708e-01
4.32872742e-01 -2.25652054e-01 -4.77151811e-01 -2.64690906e-01
-1.85737371e-01 2.63586760e-01 -6.28264427e-01 4.04906571e-01
7.04290122e-02 5.62902570e-01 -8.33779037e-01 -7.37303495e-01
1.04353614e-01 1.68710575e-01 -1.99631439e-03 1.86900020e-01
-4.13181573e-01 3.54638427e-01 -2.28433862e-01 7.63254046e-01
6.42523348e-01 -2.31157914e-01 3.49657923e-01 -3.49934474e-02
-2.28332430e-01 -2.13658541e-01 -1.30126143e+00 1.81096768e+00
-2.93435514e-01 8.14128876e-01 -4.65920597e-01 -8.66491973e-01
1.36982834e+00 1.16417788e-01 9.80063602e-02 -5.58507562e-01
1.17202677e-01 5.42930126e-01 -1.17495716e-01 -6.05515301e-01
5.86388409e-01 2.38280043e-01 -1.62666515e-01 4.68394250e-01
3.03383261e-01 -1.31501988e-01 6.05384409e-01 8.90783146e-02
8.35329056e-01 3.29702020e-01 2.45208547e-01 1.11579634e-01
5.32952011e-01 2.08234191e-01 5.58092177e-01 8.97125185e-01
1.60752498e-02 7.29641199e-01 3.97223055e-01 -5.03016889e-01
-1.63672936e+00 -5.83551586e-01 -4.90981936e-02 9.32383835e-01
1.91967469e-02 -2.56273389e-01 -7.47130394e-01 -5.46915591e-01
-1.93191737e-01 6.17950082e-01 -3.49512130e-01 1.69086486e-01
-6.55178010e-01 -4.78130281e-01 8.22198272e-01 3.51658016e-01
4.57899570e-01 -1.23046708e+00 -6.25318766e-01 6.00558668e-02
5.25410753e-03 -1.12175572e+00 -5.23914099e-01 2.81931847e-01
-6.54168248e-01 -7.32560754e-01 -1.08586192e+00 -1.13535976e+00
1.00138390e+00 -1.98384393e-02 4.73139524e-01 -6.96916580e-02
-6.34881556e-01 -2.45688837e-02 -5.41070342e-01 -3.79955560e-01
-3.93742532e-01 6.51465878e-02 -2.10109036e-02 2.29218259e-01
3.93317401e-01 -8.50666985e-02 -1.53781146e-01 3.51185292e-01
-7.21706927e-01 8.41569901e-02 6.73751354e-01 1.03896832e+00
6.77855968e-01 -1.65729135e-01 2.89175957e-01 -9.27396417e-01
4.63044733e-01 -4.69627194e-02 -8.89532864e-01 6.45387053e-01
-3.07842344e-01 2.28651296e-02 8.48860264e-01 -5.81408799e-01
-1.14574659e+00 3.94461393e-01 1.83625653e-01 -4.87609774e-01
-1.71962827e-01 4.31279808e-01 -1.49546023e-02 -1.00829192e-01
6.84875727e-01 6.75280333e-01 1.33270491e-02 -6.88532531e-01
1.93941072e-01 1.19598389e+00 7.80444562e-01 -4.43800420e-01
6.29757404e-01 2.83031136e-01 -2.55926132e-01 -1.31008673e+00
-7.56387770e-01 -3.46136242e-01 -1.00557578e+00 -2.63129890e-01
4.05872375e-01 -5.92512608e-01 -3.31834823e-01 6.86976492e-01
-1.19808316e+00 -3.43258649e-01 -2.86349088e-01 3.98934722e-01
-5.00823736e-01 9.24453795e-01 -7.08711684e-01 -7.58234918e-01
-3.45309407e-01 -8.77455473e-01 8.69316757e-01 4.13237870e-01
-2.42301941e-01 -5.21717131e-01 4.01707888e-02 3.98851216e-01
4.86894250e-02 -1.30575910e-01 8.88457477e-01 -9.23079133e-01
-5.61124325e-01 -5.33050656e-01 -2.38594562e-01 4.39353675e-01
2.36168221e-01 1.23338282e-01 -7.98866928e-01 -1.58243597e-01
-1.79926261e-01 -6.16517901e-01 6.55095398e-01 -1.36808902e-01
1.20149577e+00 -1.95305213e-01 -5.37513830e-02 3.16513956e-01
1.25410140e+00 3.48182350e-01 6.90919876e-01 2.27096483e-01
6.94773972e-01 4.85196948e-01 8.89022708e-01 8.18759799e-01
-3.13421249e-01 5.71415007e-01 -3.41742516e-01 1.93857267e-01
-2.31937855e-01 -3.05610180e-01 2.35110462e-01 1.09351504e+00
-3.13696824e-02 -4.91030484e-01 -1.26287317e+00 4.82912719e-01
-1.82247376e+00 -8.71676981e-01 9.43261757e-02 2.38678050e+00
1.07979858e+00 5.93819208e-02 -9.44931135e-02 3.71751785e-02
9.61902738e-01 -2.23833043e-02 -5.97766340e-01 -4.30086017e-01
-3.65228415e-01 2.55694538e-01 6.17308021e-01 4.74534601e-01
-1.07908893e+00 1.31824172e+00 5.50225925e+00 1.14881647e+00
-1.08456719e+00 -3.51034328e-02 4.72073972e-01 -1.10018766e-02
2.47426108e-01 1.79273095e-02 -1.02785540e+00 7.78451979e-01
8.90514493e-01 -2.37336487e-01 4.40754205e-01 7.08942652e-01
2.01731756e-01 -1.71010181e-01 -8.52119207e-01 1.12143564e+00
6.16378546e-01 -1.42301130e+00 3.00126016e-01 -1.96901545e-01
6.04144454e-01 -3.17245811e-01 -3.56485188e-01 3.65373045e-02
1.53146330e-02 -8.19316268e-01 7.80153155e-01 9.06658292e-01
1.07253385e+00 -6.28779590e-01 6.59372151e-01 4.19331461e-01
-7.50574291e-01 1.29504010e-01 -8.15105438e-01 2.01087981e-01
1.33135691e-02 4.84348208e-01 -9.60706651e-01 4.11695600e-01
3.02941769e-01 6.72829688e-01 -6.59655929e-01 1.09956038e+00
-3.26452255e-01 6.04262948e-01 -1.94196597e-01 -3.10008258e-01
3.09265852e-02 -4.41096693e-01 3.16393256e-01 1.42762959e+00
4.99548793e-01 -2.99072266e-02 2.17394814e-01 5.33499420e-01
-3.18754166e-01 3.93244833e-01 -6.32935762e-01 -4.13311958e-01
5.85648656e-01 1.12812865e+00 -1.09356248e+00 -6.55673087e-01
-3.96717101e-01 1.60893846e+00 3.31822366e-01 1.57635957e-01
-7.41551518e-01 -1.02634799e+00 -1.53395280e-01 -1.23974681e-01
6.05569184e-01 -3.93659681e-01 -3.53418261e-01 -1.11285186e+00
3.01004231e-01 -1.00952089e+00 2.24590987e-01 -7.76137173e-01
-1.01426053e+00 5.47788382e-01 -5.15783548e-01 -1.46928132e+00
2.51870416e-02 -6.54788911e-01 -4.01623875e-01 8.48018587e-01
-9.67869580e-01 -1.02926779e+00 -4.91815209e-01 2.67440617e-01
1.08446872e+00 -5.58979392e-01 8.08531106e-01 3.89407724e-01
-7.24134505e-01 7.13892758e-01 5.75928271e-01 4.32535976e-01
1.05032754e+00 -8.96022141e-01 2.25282699e-01 1.02970707e+00
2.25048169e-01 5.44697940e-01 4.97151822e-01 -8.90628219e-01
-1.33252192e+00 -9.38926399e-01 1.01979876e+00 -2.37251103e-01
7.47911692e-01 -6.57643259e-01 -1.08651257e+00 5.01430452e-01
5.28353751e-02 -2.25315597e-02 7.40762651e-01 -4.72237706e-01
-2.41690382e-01 4.36873175e-02 -8.02837133e-01 6.79759920e-01
7.64564335e-01 -4.58491117e-01 -5.10378242e-01 6.19530559e-01
1.20654762e-01 -3.75195175e-01 -6.97907686e-01 -1.60016492e-01
6.29634798e-01 -2.78403640e-01 4.66276288e-01 -4.62543696e-01
6.02114677e-01 -3.71706754e-01 -3.89282480e-02 -7.49730349e-01
-7.33233709e-03 -6.06932521e-01 -7.21248463e-02 1.36306739e+00
4.36587781e-01 -2.37395033e-01 8.47116411e-01 4.35068786e-01
-1.28686521e-02 -4.63332623e-01 -8.31339300e-01 -1.13758481e+00
-2.21635938e-01 3.75487320e-02 7.79055208e-02 8.75302553e-01
1.92521542e-01 2.49087021e-01 -8.53889763e-01 -6.56389520e-02
7.13666975e-01 3.05651248e-01 8.49399447e-01 -1.02821863e+00
-2.01381505e-01 -7.23544657e-02 -4.21254128e-01 -1.04286742e+00
2.82085780e-03 -9.43752229e-01 4.12418216e-01 -1.17911339e+00
3.60426039e-01 -3.40701371e-01 1.03895208e-02 7.01329589e-01
6.38770759e-02 2.74059087e-01 1.93873838e-01 5.29394329e-01
-5.88037312e-01 3.79653007e-01 9.49666083e-01 -2.74725348e-01
-2.33741656e-01 -3.18727493e-01 -2.18827382e-01 6.68294966e-01
8.70803297e-01 -5.31924307e-01 -8.87434334e-02 -4.27122295e-01
-2.40130022e-01 -1.25216898e-02 -5.07655218e-02 -1.02659702e+00
5.84628582e-01 -6.36622533e-02 5.16373873e-01 -5.95642924e-01
1.68471202e-01 -5.34874499e-01 -1.12422675e-01 5.30719280e-01
-3.18546355e-01 -2.75121838e-01 1.51942760e-01 5.43498337e-01
-1.15916699e-01 -8.69648635e-01 8.57806265e-01 -2.04819202e-01
-9.86242712e-01 -1.12751955e-02 -4.69693124e-01 -2.20182076e-01
1.05125916e+00 -4.76033241e-01 -2.63478786e-01 -6.49339855e-02
-7.36440897e-01 -9.27199721e-02 6.64632618e-01 3.91518176e-01
5.87959468e-01 -9.71960723e-01 -6.28202915e-01 1.94935799e-01
2.87322223e-01 -2.83036947e-01 1.68528676e-01 5.93088448e-01
-1.08585191e+00 4.63084728e-01 -4.94192988e-01 -3.39081734e-01
-1.48920643e+00 5.51152885e-01 1.60564389e-02 -1.19154520e-01
-8.41167986e-01 6.51700497e-01 -4.82075661e-01 -1.72898024e-01
5.44115663e-01 1.40369415e-01 -7.90636167e-02 2.32384324e-01
5.27711034e-01 5.14075160e-01 1.58413798e-01 -4.84076560e-01
-2.62520581e-01 6.75265312e-01 -5.32038689e-01 -1.67995974e-01
1.33466840e+00 1.01652808e-01 7.49565363e-02 5.58620811e-01
8.93044055e-01 2.24659294e-01 -1.28334200e+00 -3.78999978e-01
4.04314309e-01 -6.01453781e-01 -4.47604269e-01 -6.80381835e-01
-5.99241614e-01 8.19714129e-01 5.80381513e-01 -4.21153933e-01
8.46460044e-01 -1.88037604e-01 6.73699319e-01 9.66195047e-01
4.10667658e-01 -1.57230771e+00 1.79402426e-01 6.22850418e-01
9.14598823e-01 -1.20158792e+00 1.62954628e-01 -3.47524613e-01
-8.80241334e-01 1.51658404e+00 4.06209767e-01 -8.71597528e-02
1.31830588e-01 3.03984523e-01 1.39064565e-01 1.04815230e-01
-3.36448729e-01 1.23994492e-01 -7.52172172e-02 2.35763714e-01
2.58014649e-01 -2.05755070e-01 -5.35327196e-01 5.77217519e-01
-1.41499389e-03 2.75874883e-01 7.92332947e-01 1.34634304e+00
-5.70126414e-01 -1.27350163e+00 -4.36899364e-01 5.51657438e-01
-2.88102984e-01 -9.96925756e-02 -7.03828216e-01 3.35571468e-01
-7.89114088e-02 5.92535913e-01 1.97166085e-01 2.11330857e-02
2.47122809e-01 3.96242648e-01 4.67581272e-01 -7.75387168e-01
-1.83486596e-01 2.15618432e-01 -5.40961102e-02 1.55942384e-02
1.82746332e-02 -6.61435127e-01 -1.14586329e+00 -7.58711025e-02
-3.78401250e-01 8.73259008e-02 5.88596463e-01 5.64057529e-01
3.97483975e-01 1.96341544e-01 5.58507562e-01 -7.39165902e-01
-6.54946148e-01 -1.01811540e+00 -6.03872955e-01 3.88408810e-01
-6.74833134e-02 -3.09200704e-01 -7.18172267e-02 6.41591728e-01] | [11.816390991210938, 2.4538464546203613] |
9756e96f-6b7a-4917-a099-6a9628593544 | drugehrqa-a-question-answering-dataset-on-1 | 2205.01290 | null | https://arxiv.org/abs/2205.01290v1 | https://arxiv.org/pdf/2205.01290v1.pdf | DrugEHRQA: A Question Answering Dataset on Structured and Unstructured Electronic Health Records For Medicine Related Queries | This paper develops the first question answering dataset (DrugEHRQA) containing question-answer pairs from both structured tables and unstructured notes from a publicly available Electronic Health Record (EHR). EHRs contain patient records, stored in structured tables and unstructured clinical notes. The information in structured and unstructured EHRs is not strictly disjoint: information may be duplicated, contradictory, or provide additional context between these sources. Our dataset has medication-related queries, containing over 70,000 question-answer pairs. To provide a baseline model and help analyze the dataset, we have used a simple model (MultimodalEHRQA) which uses the predictions of a modality selection network to choose between EHR tables and clinical notes to answer the questions. This is used to direct the questions to the table-based or text-based state-of-the-art QA model. In order to address the problem arising from complex, nested queries, this is the first time Relation-Aware Schema Encoding and Linking for Text-to-SQL Parsers (RAT-SQL) has been used to test the structure of query templates in EHR data. Our goal is to provide a benchmark dataset for multi-modal QA systems, and to open up new avenues of research in improving question answering over EHR structured data by using context from unstructured clinical data. | ['Daisy Zhe Wang', 'Kirk Roberts', 'Anthony Colas', 'Jayetri Bardhan'] | 2022-05-03 | null | https://aclanthology.org/2022.lrec-1.117 | https://aclanthology.org/2022.lrec-1.117.pdf | lrec-2022-6 | ['text-to-sql'] | ['computer-code'] | [ 0.05530621 0.58343136 -0.11705893 -0.6199553 -1.4335594 -0.82259655
0.06517588 1.0712956 -0.07276095 0.7982391 0.70154124 -0.7803476
-0.6316555 -1.1245301 -0.5284573 0.18857652 0.09967741 1.1590321
0.39734632 -0.5691644 -0.24624126 0.07260022 -1.0751652 1.1471002
1.053648 0.9909971 -0.25114498 0.64401466 -0.48306373 1.4420781
-0.44312996 -0.8164081 0.03422433 -0.6303452 -1.3723894 -0.57666945
0.4728958 0.05959406 -0.05290731 0.6808219 0.7413009 -0.461854
0.21315546 -1.0142643 -0.64582425 0.6959065 0.48545557 0.06798696
1.1825594 0.24412996 1.0768832 -0.2828083 1.2601284 1.193985
0.7408474 0.5531872 -1.1223371 -0.19666465 -0.6335964 0.19908899
-1.1866815 -0.5033498 0.09078373 -0.34865442 1.2815758 0.79544264
0.31020102 0.67339814 0.32386637 0.3775767 0.98812366 -0.35271734
0.28171206 0.2971231 0.47426638 0.7800799 0.3555032 -0.22369751
-0.30995852 -0.81812555 0.13893992 -0.24311739 -0.2638671 -0.12191665
-1.1384623 0.69987935 0.23741905 0.03730328 -0.6022027 -0.46451843
0.54179895 0.67947274 -0.39578483 0.9529188 -1.1134198 -0.01068049
-0.38732487 0.7822644 1.5844012 1.1985863 0.45822376 -1.0300616
-0.6632121 0.5712089 0.3232355 0.4134047 0.28702098 -1.0795772
1.0344439 1.1124738 0.22207879 -0.6968482 -0.74562615 0.22405131
-0.40828487 -0.7231399 0.66673964 -0.07501984 -0.7594395 1.4003516
0.41457313 -0.4814661 0.5874383 0.47163182 1.6955123 0.25416073
0.57930034 -0.19104849 2.185999 -0.34527752 -1.1952584 0.31509888
1.3134415 -0.7347099 0.84815353 0.11168343 -1.2227498 -0.26309475
-0.43956718 -0.43957546 -0.76135 -0.42705715 0.13144875 0.67625284
-0.9039271 0.172861 -0.66231996 -0.6575151 0.24013671 0.18157925
-0.4515789 -0.54165006 -1.6344781 1.0485528 0.4685644 -0.4912238
-0.17705362 -1.264382 -1.069046 0.04020005 0.71942914 -1.4254768
1.2547433 0.01749383 -0.8409512 0.734511 -0.19404007 -0.49521372
0.13027178 0.16343631 -0.7499228 0.24067591 0.36148113 0.45806724
-0.26285216 -0.93575066 -0.3991119 -0.8844443 0.10241815 -0.03104704
0.44726294 0.15447351 -0.2756944 -0.34553257 -0.04136898 -0.5422655
-0.154761 -0.32015845 -0.74516034 -0.34828198 0.21507122 -1.0983381
1.6535081 -1.6723567 -0.31202367 0.18948849 0.2688658 -0.03162561
-0.1206941 1.075177 -0.17150226 0.34821376 -0.40744582 0.23190694
-0.041139 0.5676223 -0.35355732 -0.3744679 0.5085889 1.2866416
-0.75643784 -1.0538343 -0.44179648 0.10181939 -0.76352614 0.5584371
-0.70034313 0.35944307 -0.55784714 1.0447049 0.44096532 -0.47504085
0.30920887 -0.54514605 0.2533556 0.8715564 -1.1348467 1.5321068
0.01477582 -0.35736835 -0.16386248 -0.55577284 0.5846207 0.7132927
0.91236985 -1.055703 -0.24583255 0.31893632 -0.09916565 -1.4000074
0.35060722 -0.03996221 -0.3071777 0.10936976 0.0162273 -0.17027654
0.34689662 0.3441766 1.6918571 -0.11031341 0.594216 -0.14382751
0.6794261 0.76292354 0.42473584 0.8170557 0.03261313 0.4608756
0.5702914 -0.52736664 -0.62129545 -1.0069109 -0.7208941 0.62074506
-0.33688673 -0.86679447 -0.5742954 -0.8257287 0.18814549 0.6793416
-0.5816519 0.15314662 -0.64699346 -0.5180736 0.8675953 0.4225929
0.14454323 -1.1459508 -0.68428177 0.76429397 -0.66242117 -1.2947109
-0.1729141 0.36698171 -0.9898178 -1.8732822 -0.18299706 -0.4883114
0.0835815 -0.6701154 1.9420363 0.08698042 -0.22417209 0.714305
-0.4033794 -0.5625981 -0.7647001 0.20981203 -0.6754477 -0.63579565
0.75925183 0.09089264 -0.62451166 0.3138921 -1.4985036 -0.47606373
0.13242096 0.6877729 0.84392905 -0.4164265 0.85774225 -1.3987638
0.9411787 -0.78031254 -0.45497003 0.90613556 -0.656779 0.41736782
0.6958947 0.32597905 -0.70962954 0.1789657 -0.8384887 0.27578914
-0.47673672 1.0389271 -0.34644225 0.48630497 0.87970793 -0.20155768
-0.03967185 -0.65882754 0.61141884 0.65593785 0.5589826 -0.6703866
0.14343621 0.06514463 -0.03116237 -0.18588652 -0.6829816 -0.79268914
-0.4115312 0.54308003 1.1969961 -0.7605947 -0.9451262 -0.3760749
-1.0526083 -0.04254183 -0.8099833 0.09663762 -0.5228473 0.24124019
-0.80321896 -0.46890616 -0.4334101 -1.086543 1.2370816 -0.2200162
-0.4906677 -0.864677 0.5455921 0.72945404 0.25853905 0.2548258
1.6764792 -1.2471458 -0.5690417 -0.01098976 -0.11903089 -0.45910296
0.35677978 -0.39143288 -0.6743236 0.12707996 0.00836584 -0.5622307
0.3946766 0.02845009 1.049001 -0.50412256 -0.2694419 0.3048561
1.5724005 0.39962798 0.92750293 0.3541385 0.38930196 0.9509374
0.4276679 0.33702 1.1004388 0.56966174 0.2770116 0.05985606
0.22327757 -0.3229713 -0.23204903 0.70776063 0.5009218 -0.25838807
-1.2927856 0.49767312 -1.5893718 -0.6832657 -0.518614 2.002312
1.4482716 -0.3366181 0.13112478 -0.0610279 -0.07676447 -0.5321218
-0.49041227 -0.3134788 0.02079062 0.5991928 0.3333364 0.374368
-0.8524215 0.3868752 6.4055295 0.27483702 -0.38768488 -0.05569353
0.4267686 0.42307615 -0.6658309 -0.19067286 -0.90933883 0.33859336
1.5700078 0.14523691 0.09970563 0.40979025 -0.02174886 -0.06282481
-1.5595871 0.693096 -0.23524725 -1.6865269 0.2263733 0.03013883
0.19911578 0.05248244 -0.35353982 0.4702481 0.52272373 -1.0379702
0.01929766 0.9013601 0.72784925 -0.07911783 1.1147463 0.2959719
-1.1461765 -0.12663186 -0.0953784 0.6320399 0.11516218 0.30483967
-1.0628306 1.3269203 0.85207385 0.26593116 -0.88666266 1.1096166
0.42392403 0.34230012 -0.37026983 0.41463012 -0.14153829 0.13639264
0.10365614 1.2434894 0.17282258 0.6124199 0.11225803 0.73574775
-0.2699278 0.5000146 -0.84836614 -0.07857809 0.4424913 0.8595313
-0.22358187 -0.6678003 -0.5467033 0.3329184 0.06322728 0.2024399
-0.48942038 -0.24770221 0.33391467 0.41724786 0.02087855 0.50780463
-0.356193 -1.1961204 0.04573562 -1.6154205 1.4913318 -0.7217096
-1.6899494 0.7872292 0.16003716 -0.9667427 -0.62880737 -0.42502934
0.14992872 0.99242353 -1.4919866 -0.7820974 -0.2028828 1.0280776
0.04472487 -0.02215041 1.3982377 0.7148187 -0.11053044 0.54213995
-0.27678385 0.37923783 0.978755 -1.4552009 0.17377646 -0.02096634
-0.10528625 0.9847043 0.42630368 -0.87885135 -1.7676997 -1.115459
1.5588139 -1.3234575 0.35346904 -0.07711692 -1.4902822 0.5952695
0.45591682 0.00781213 1.3292925 -0.01818188 -0.40287444 -0.07323109
-1.5790236 0.09483087 0.68432987 -0.713017 -1.1738857 0.38267943
1.0359337 -0.5351865 -1.8214381 0.76170295 0.23592156 -0.63894033
1.1258682 -1.3041635 0.39684212 -0.39162 -0.42194346 -0.7632691
0.15309772 -0.3620433 -0.32001927 1.0257863 0.7817649 -0.7574057
0.62970865 1.2448282 0.00969252 -0.9904959 -0.91776 -0.34092164
0.09457494 -0.20009947 1.0509592 1.1810528 0.22336413 0.49060127
0.53072625 0.32681218 0.16135097 0.17368984 0.49890253 -1.3033929
-0.3303587 0.11049024 -0.09847465 -0.7057386 -0.69904375 -0.95543164
-0.18956293 -2.0702443 0.11152451 -0.7083457 -0.02423107 0.78200734
-0.25526595 -0.23671536 0.08161364 0.1453993 -0.7536937 0.07244142
1.113666 -0.21458595 -0.23165393 -0.22948429 -0.8348443 0.17147443
0.43159935 -0.85157377 -0.4430217 -0.23690349 0.70167816 1.021952
0.15483463 -0.5930967 0.41144308 0.05116303 0.02674462 -0.8689603
-0.06542777 -1.1694152 0.29106998 0.63415575 -0.7560545 0.6287585
0.21843497 0.2496299 -0.5641619 -0.19913247 0.19542764 -0.4768039
-0.04933745 0.14422022 -0.21278617 0.845818 0.54779744 0.17351982
-0.6242611 -0.14128424 -1.0022974 0.693265 0.08824498 0.27069548
0.54215765 -1.0338918 -0.76110667 0.08860963 0.68005574 0.1003525
0.14389965 0.7088646 -0.5737324 1.0172745 -0.00785604 -0.531687
-1.1552643 0.8894289 0.4467333 -0.8931845 -0.38188723 0.18718235
-0.31534028 -1.014016 0.17846029 -0.808782 -0.42141986 0.0460074
0.5518979 -0.01195089 0.6398681 -0.21195967 -0.4925829 0.120148
0.03425271 0.05015112 1.1378812 0.03427376 -0.39265126 0.13816111
1.1180727 -0.0113913 -0.11960982 -0.23693943 0.59241456 -0.111141
-0.56635654 -1.4458474 -0.53007495 0.48347837 0.62959886 0.6765384
1.1215066 0.2821523 0.8477729 0.77380604 0.24281913 -0.68948895
-0.3189977 0.49779114 0.89491135 -1.319393 -0.2458285 -0.49616522
-0.6148872 1.0462259 0.51078063 0.72641397 0.6122568 0.3878515
0.48415715 -0.89830303 -1.0301176 -0.43860653 0.44225633 0.6224921
0.7221681 -0.2325445 -0.318685 0.8166476 -0.13398275 0.60451925
0.04431144 1.1939592 0.05279475 -1.692897 -0.5309256 0.8854072
-0.8820062 -0.3984736 -0.38405934 0.77113193 0.2581572 1.1530824
-0.27813682 -0.0673127 1.0484159 0.5817067 0.31451926 -0.7481783
-1.4042307 -0.39180568 0.5756962 -0.7375916 -0.3196027 -0.48406476
-1.4405571 0.17409179 0.14824574 0.49243924 0.2743314 0.7665012
0.7176051 0.53655195 -0.04663041 0.7832573 -0.60644215 -0.6300647
0.09317149 0.6556632 0.1589301 -0.02599826 0.4915136 0.26952678] | [8.776321411132812, 8.52126693725586] |
39efe4a9-f9ae-4286-a255-9ca85c16ac21 | unsupervised-continuous-object-representation | 2007.15627 | null | https://arxiv.org/abs/2007.15627v2 | https://arxiv.org/pdf/2007.15627v2.pdf | Continuous Object Representation Networks: Novel View Synthesis without Target View Supervision | Novel View Synthesis (NVS) is concerned with synthesizing views under camera viewpoint transformations from one or multiple input images. NVS requires explicit reasoning about 3D object structure and unseen parts of the scene to synthesize convincing results. As a result, current approaches typically rely on supervised training with either ground truth 3D models or multiple target images. We propose Continuous Object Representation Networks (CORN), a conditional architecture that encodes an input image's geometry and appearance that map to a 3D consistent scene representation. We can train CORN with only two source images per object by combining our model with a neural renderer. A key feature of CORN is that it requires no ground truth 3D models or target view supervision. Regardless, CORN performs well on challenging tasks such as novel view synthesis and single-view 3D reconstruction and achieves performance comparable to state-of-the-art approaches that use direct supervision. For up-to-date information, data, and code, please see our project page: https://nicolaihaeni.github.io/corn/. | ['Jun-Jee Chao', 'Nicolai Häni', 'Volkan Isler', 'Selim Engin'] | 2020-07-30 | null | http://proceedings.neurips.cc/paper/2020/hash/43a7c24e2d1fe375ce60d84ac901819f-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/43a7c24e2d1fe375ce60d84ac901819f-Paper.pdf | neurips-2020-12 | ['single-view-3d-reconstruction'] | ['computer-vision'] | [ 3.38852793e-01 2.19200507e-01 -1.24751702e-01 -5.10670662e-01
-7.85919070e-01 -6.87484145e-01 8.15684438e-01 -5.22286117e-01
2.97424316e-01 3.50311279e-01 1.56442001e-01 -2.58747160e-01
4.97998774e-01 -7.37153530e-01 -1.29729152e+00 -2.09787875e-01
4.88671690e-01 5.99320471e-01 2.45556518e-01 -6.67884126e-02
2.25952521e-01 5.74823260e-01 -1.61068320e+00 5.34321666e-01
4.80618030e-01 9.27597582e-01 2.80331552e-01 8.53067756e-01
2.81616859e-02 7.77155936e-01 -2.83975065e-01 -3.26583475e-01
5.85480690e-01 -5.07092178e-01 -7.61124671e-01 5.51655889e-01
9.15747583e-01 -9.07151997e-01 -3.78438085e-01 1.04095078e+00
2.08346710e-01 -1.06837284e-02 5.70245683e-01 -1.27152181e+00
-9.09069300e-01 -1.92525722e-02 -4.79758352e-01 -1.74967632e-01
6.77595317e-01 3.90282154e-01 8.24453115e-01 -1.25137854e+00
9.71162260e-01 1.21334946e+00 1.95654988e-01 6.59919858e-01
-1.45296693e+00 -3.93123150e-01 3.22710127e-01 -2.16752872e-01
-1.19537997e+00 -8.18551362e-01 8.97736132e-01 -5.46802640e-01
9.83861446e-01 2.18614951e-01 6.94289386e-01 1.34881842e+00
1.37740523e-01 6.89544082e-01 1.14918613e+00 -3.69835258e-01
1.73812717e-01 2.00008750e-01 -4.18068826e-01 7.41935968e-01
1.30210817e-01 2.91003138e-01 -6.56692624e-01 1.30633697e-01
1.28838491e+00 1.28488511e-01 -5.66549063e-01 -8.69004607e-01
-1.34759009e+00 4.74358261e-01 5.43505251e-01 -2.69323885e-01
-2.89430380e-01 2.82960653e-01 -8.36585835e-02 2.05056667e-01
5.97751677e-01 4.16880935e-01 -3.10177356e-01 9.74678025e-02
-6.37107074e-01 4.26753879e-01 5.72576880e-01 1.27875781e+00
7.84583449e-01 3.01253617e-01 3.89687806e-01 3.90754253e-01
3.51252794e-01 7.92771578e-01 -2.19202824e-02 -1.49193919e+00
4.73448634e-01 6.28591299e-01 1.87748492e-01 -6.72337592e-01
1.24716312e-01 -1.99685156e-01 -5.90248048e-01 7.25282252e-01
1.07726060e-01 1.85355932e-01 -1.08832145e+00 1.50501561e+00
4.02255863e-01 1.30332544e-01 1.36576116e-01 1.02236426e+00
1.14908397e+00 7.94171393e-01 -6.75649405e-01 1.10678650e-01
8.82056594e-01 -1.04586387e+00 -3.26222926e-01 -4.70385790e-01
1.02856480e-01 -8.85741770e-01 1.15132868e+00 4.58359241e-01
-1.54836142e+00 -5.66300988e-01 -1.06780529e+00 -4.25605714e-01
-1.34341896e-01 -7.09753782e-02 4.55238402e-01 1.65847704e-01
-1.24525964e+00 3.79844159e-01 -9.11826551e-01 -2.22852439e-01
4.30709630e-01 1.23809092e-01 -5.74877441e-01 -5.23648143e-01
-5.31121850e-01 8.89514208e-01 6.25585616e-02 3.96652333e-03
-1.28355944e+00 -7.14078188e-01 -1.02021313e+00 -2.01634303e-01
5.01547277e-01 -1.22280991e+00 1.55218184e+00 -1.09996057e+00
-1.64724815e+00 1.15815771e+00 -2.93787092e-01 1.49316117e-02
5.49232721e-01 -3.02198172e-01 -1.57004863e-01 3.69371206e-01
1.17141910e-01 8.27426672e-01 9.47497308e-01 -1.90708089e+00
-1.73056513e-01 -2.75798559e-01 5.13469100e-01 4.06956464e-01
4.76032853e-01 -1.38099134e-01 -4.48625386e-01 -4.04952645e-01
5.49621165e-01 -7.73536623e-01 -1.44462481e-01 4.11519796e-01
-5.15941143e-01 2.67869085e-01 7.88664818e-01 -5.63585281e-01
2.82889873e-01 -2.10999894e+00 3.50143045e-01 -1.75682247e-01
2.86834329e-01 -5.83516359e-02 -1.52038932e-01 3.79205048e-01
-1.92218199e-01 2.58123279e-02 -1.27374902e-02 -6.15527868e-01
-9.68450308e-02 1.20341070e-01 -5.66182256e-01 4.06818032e-01
2.82441139e-01 1.01092649e+00 -7.67880797e-01 -1.37079552e-01
6.75700724e-01 6.72817051e-01 -6.53948545e-01 6.38483047e-01
-6.77694142e-01 9.11074400e-01 -1.94233909e-01 6.78995132e-01
7.27116942e-01 -7.65214682e-01 3.71484458e-02 -3.08594644e-01
-4.62104566e-02 5.48979104e-01 -1.07380128e+00 2.07263160e+00
-5.00015199e-01 5.13381004e-01 1.10367976e-01 -7.00439930e-01
8.57038677e-01 2.68955201e-01 1.92403629e-01 -5.34884214e-01
8.70950595e-02 1.42461613e-01 -4.57425654e-01 -2.84098536e-01
3.57673943e-01 -2.19112828e-01 2.48005494e-01 6.07165158e-01
6.51588067e-02 -7.94564486e-01 -2.72951126e-01 3.45504701e-01
7.40193427e-01 6.52832568e-01 3.65702033e-01 2.47943457e-02
2.01072872e-01 -9.77139696e-02 4.60435331e-01 4.63148475e-01
2.75922656e-01 1.25527310e+00 3.08649331e-01 -6.43319368e-01
-1.39045465e+00 -1.54295540e+00 1.03153393e-01 4.40199554e-01
3.96784008e-01 -2.86318362e-01 -4.35620338e-01 -4.44471359e-01
-2.61874616e-01 8.48097384e-01 -5.29890716e-01 2.52898812e-01
-5.18290401e-01 1.60165370e-01 -1.12060048e-01 5.39645135e-01
4.91223156e-01 -9.10052121e-01 -8.42131972e-01 -1.96265414e-01
-3.10511798e-01 -1.34116900e+00 -3.92482996e-01 -1.00703597e-01
-1.03367949e+00 -1.11651731e+00 -6.12180412e-01 -5.76649308e-01
1.01107812e+00 7.35504329e-01 1.47565246e+00 6.42234311e-02
-1.33216217e-01 5.45284867e-01 -2.14728519e-01 -2.29860157e-01
-5.70790410e-01 -3.17800611e-01 -2.69986391e-02 -1.51230276e-01
-2.94861108e-01 -8.69940460e-01 -6.09244764e-01 2.83471733e-01
-8.59583676e-01 9.61384177e-01 4.38910097e-01 6.13886833e-01
7.57973075e-01 -4.26694512e-01 5.49981892e-02 -9.26211596e-01
-2.25719661e-01 -1.30038545e-01 -9.19135213e-01 1.62353620e-01
-3.86548638e-01 -9.84413624e-02 5.33430398e-01 -1.12059735e-01
-1.22006762e+00 6.65892512e-02 1.28425464e-01 -9.29758430e-01
-3.97650361e-01 8.02256763e-02 -3.05108666e-01 2.13038877e-01
7.05259144e-01 2.15951145e-01 -4.39764075e-02 -4.52465415e-01
5.36450148e-01 2.63140231e-01 6.28683329e-01 -5.94350696e-01
9.47317243e-01 7.78302372e-01 -1.03434846e-01 -4.37895030e-01
-1.05685520e+00 -1.17861167e-01 -8.88863087e-01 -2.55544990e-01
9.20095325e-01 -1.24231565e+00 -4.72816765e-01 3.58619034e-01
-1.39757550e+00 -6.39433324e-01 -3.64088744e-01 4.84827250e-01
-9.44225907e-01 6.49585249e-03 -4.14148748e-01 -5.80537617e-01
-1.06310837e-01 -1.17352009e+00 1.41806042e+00 1.17667675e-01
-5.67935854e-02 -7.70303011e-01 -2.75089949e-01 6.03888869e-01
1.84233919e-01 5.75627327e-01 7.41356075e-01 2.02803656e-01
-1.48804605e+00 -1.86885409e-02 -2.57011592e-01 4.32225764e-01
2.40062609e-01 7.43364021e-02 -1.18752146e+00 -4.02647018e-01
8.91758502e-02 -5.79181671e-01 5.04521549e-01 3.92698258e-01
1.16805136e+00 -3.45788389e-01 -1.30627677e-01 9.90566432e-01
1.52493775e+00 7.03138113e-02 3.98454696e-01 7.27378651e-02
9.33511198e-01 4.53318000e-01 3.41950357e-01 3.21648151e-01
5.11918664e-01 6.01218820e-01 8.23540986e-01 -2.58628488e-01
-4.11618114e-01 -5.94084322e-01 3.36608887e-01 7.92598784e-01
-2.16028661e-01 -3.76750827e-01 -7.69009829e-01 3.57458115e-01
-1.62215960e+00 -9.14765537e-01 2.60828175e-02 2.14784575e+00
4.03431714e-01 1.61069423e-01 -2.37279311e-01 -2.26108149e-01
4.70766038e-01 4.08083141e-01 -8.84393930e-01 -3.34344059e-01
-1.01143055e-01 -8.62283483e-02 2.58079469e-01 6.62711024e-01
-7.04606950e-01 8.61358047e-01 5.91882706e+00 7.58830979e-02
-1.21386874e+00 -4.84243855e-02 6.42151654e-01 -3.67921263e-01
-8.05769205e-01 2.90318489e-01 -5.06382108e-01 1.32436603e-02
4.71302986e-01 -1.33201107e-01 8.28905702e-01 7.51426876e-01
1.48843169e-01 -6.15759157e-02 -1.46156466e+00 1.23957956e+00
4.96138573e-01 -1.74425113e+00 2.85364509e-01 1.08112812e-01
1.17100596e+00 2.66019315e-01 1.42478570e-01 -1.41946405e-01
6.84066296e-01 -1.06098783e+00 1.05907214e+00 6.14009380e-01
1.20786083e+00 -3.11604559e-01 2.05996931e-01 4.99819428e-01
-9.57672894e-01 3.16389531e-01 -1.73250929e-01 -3.55325341e-02
4.41584378e-01 2.32170731e-01 -6.67023838e-01 7.83406258e-01
7.73246825e-01 1.01793540e+00 -4.27214622e-01 6.50513411e-01
-5.16896963e-01 2.13635296e-01 -2.84642845e-01 5.69593608e-01
6.17641136e-02 -1.41505659e-01 5.59658229e-01 4.42521423e-01
3.54378194e-01 3.54093581e-01 2.31511250e-01 1.31118357e+00
-2.02302068e-01 -3.90833616e-01 -1.06366336e+00 1.55087262e-01
2.21227139e-01 9.55467522e-01 -6.03883743e-01 -4.34493065e-01
-5.99237621e-01 1.04009974e+00 2.92404205e-01 6.17574632e-01
-6.97608352e-01 1.90376937e-01 5.26736498e-01 4.35679585e-01
3.84417534e-01 -3.59530479e-01 -4.31671768e-01 -1.55647087e+00
1.87580019e-01 -9.06588018e-01 -6.15825132e-02 -1.57860804e+00
-1.13886428e+00 7.26717830e-01 2.19689980e-01 -1.57887161e+00
-3.62307936e-01 -7.23257780e-01 -5.76082647e-01 8.66677821e-01
-1.38403749e+00 -1.46621120e+00 -6.06445432e-01 5.74813068e-01
8.17117155e-01 1.03033409e-02 8.73748779e-01 -4.47435938e-02
-8.32913220e-02 1.34885639e-01 -2.49718159e-01 -9.70192030e-02
6.44920111e-01 -1.10164940e+00 9.64810967e-01 8.99925709e-01
2.55389005e-01 5.14374971e-01 5.79876184e-01 -4.96872425e-01
-1.62241733e+00 -9.35228109e-01 5.29250681e-01 -9.63218331e-01
8.82288069e-02 -5.85966706e-01 -7.45426595e-01 1.20285034e+00
3.88037562e-01 3.86298746e-01 3.33750695e-01 -1.31608337e-01
-5.95879078e-01 9.72521007e-02 -8.72522414e-01 6.28144264e-01
1.39879954e+00 -5.91995597e-01 -4.24405873e-01 2.78402716e-01
8.42403591e-01 -9.95760739e-01 -6.87532008e-01 4.09328520e-01
5.65261900e-01 -1.32349408e+00 1.03789937e+00 -4.93904352e-01
8.65120173e-01 -5.54935515e-01 -5.97950697e-01 -1.15831244e+00
-1.22581139e-01 -4.43956733e-01 -2.43095770e-01 6.95090711e-01
3.32808763e-01 -6.07734799e-01 7.31364787e-01 7.07175612e-01
-4.19158429e-01 -6.81483626e-01 -5.64621925e-01 -7.14767158e-01
-1.81462839e-02 -2.79038399e-01 7.15414345e-01 9.00568247e-01
-4.95231450e-01 4.80451733e-01 -3.65457416e-01 3.88322920e-01
7.75481701e-01 7.32498109e-01 1.30476391e+00 -9.99029458e-01
-5.02867222e-01 -1.04019836e-01 -2.04894572e-01 -1.37432301e+00
6.53424039e-02 -9.57969427e-01 5.14713526e-02 -1.93316126e+00
1.70416847e-01 -2.60202467e-01 3.50969791e-01 3.52470785e-01
1.45296037e-01 3.67338836e-01 3.44504714e-01 3.13162476e-01
-6.13016725e-01 6.98635340e-01 1.63297713e+00 1.41765520e-01
-7.59481192e-02 -7.24099651e-02 -7.19324052e-01 8.48310947e-01
7.36596167e-01 -2.19240531e-01 -7.10130394e-01 -1.02870560e+00
2.22369730e-01 3.75418454e-01 8.48443270e-01 -6.94555163e-01
3.24389637e-02 -4.23169225e-01 7.19515681e-01 -8.66305947e-01
8.58965993e-01 -7.24062681e-01 5.71605861e-01 1.42556295e-01
-1.31887555e-01 1.75481394e-01 1.15903780e-01 4.14363265e-01
-8.07251185e-02 6.92056790e-02 6.89963520e-01 -6.38760865e-01
-6.51143372e-01 5.10626972e-01 1.88506637e-02 4.92344350e-02
8.00845206e-01 -4.66543734e-01 -3.87238413e-01 -5.88650107e-01
-6.33793414e-01 -4.43437397e-02 1.22282779e+00 5.44146717e-01
9.54568565e-01 -1.38201296e+00 -7.27021575e-01 5.62773764e-01
1.70817479e-01 7.72010922e-01 3.29736561e-01 2.60616869e-01
-7.43083298e-01 2.41885230e-01 -2.35015586e-01 -8.93660069e-01
-1.09000432e+00 5.81945300e-01 3.92408073e-01 2.81661391e-01
-9.77684498e-01 7.07697690e-01 8.50225270e-01 -7.14415550e-01
-1.33270949e-01 -4.30388570e-01 5.29805601e-01 -7.63641477e-01
3.74255389e-01 3.97002921e-02 -1.70493186e-01 -6.45750582e-01
-1.88527271e-01 6.85718715e-01 3.89597528e-02 -3.89792949e-01
1.33668458e+00 -1.99625492e-01 8.94365683e-02 7.10118830e-01
1.19566572e+00 -1.93410084e-01 -1.83675277e+00 -3.54796678e-01
-7.10977793e-01 -1.00696695e+00 6.12595081e-02 -7.11351633e-01
-9.83931720e-01 9.93588507e-01 2.79179007e-01 -3.01475704e-01
1.02318752e+00 3.09869856e-01 4.66627300e-01 3.91658098e-01
5.24826586e-01 -4.87404436e-01 3.74096870e-01 3.83947521e-01
1.36650372e+00 -1.49583530e+00 2.32676402e-01 -5.52320719e-01
-5.37344873e-01 9.87590790e-01 9.42857981e-01 -2.95717001e-01
5.89451551e-01 2.06249043e-01 2.04075500e-01 -4.05741274e-01
-9.92850721e-01 1.01255491e-01 3.70628268e-01 6.33332193e-01
2.18772560e-01 -3.91085446e-02 7.16817021e-01 -6.19402677e-02
-4.12726730e-01 -2.70418320e-02 6.59247339e-01 8.43148470e-01
-5.35656773e-02 -9.66029882e-01 -2.31377304e-01 3.27383041e-01
-2.88776774e-02 -3.27813067e-02 -3.54211628e-01 6.48323417e-01
-8.57721046e-02 7.16412008e-01 1.45204201e-01 -1.28013164e-01
3.73280376e-01 -1.44767985e-01 9.25427318e-01 -1.06075442e+00
-3.36688571e-02 1.28698438e-01 -2.39558816e-02 -8.14984858e-01
-5.18607378e-01 -6.91276193e-01 -1.01512575e+00 -5.29438078e-01
-1.30746856e-01 -4.07370627e-01 6.21306837e-01 6.51512206e-01
4.24828023e-01 3.74927282e-01 7.48876333e-01 -1.33371902e+00
-2.22949073e-01 -5.51337123e-01 -3.48591447e-01 3.80779982e-01
6.40101731e-01 -6.92979515e-01 -2.82651752e-01 5.08615136e-01] | [9.058724403381348, -3.060664653778076] |
9ffa7a8a-1bb7-4628-9497-94c94d5348d7 | distributional-reinforcement-learning-for-5 | 2203.00636 | null | https://arxiv.org/abs/2203.00636v2 | https://arxiv.org/pdf/2203.00636v2.pdf | Distributional Reinforcement Learning for Scheduling of Chemical Production Processes | Reinforcement Learning (RL) has recently received significant attention from the process systems engineering and control communities. Recent works have investigated the application of RL to identify optimal scheduling decision in the presence of uncertainty. In this work, we present a RL methodology tailored to efficiently address production scheduling problems in the presence of uncertainty. We consider commonly imposed restrictions on these problems such as precedence and disjunctive constraints which are not naturally considered by RL in other contexts. Additionally, this work naturally enables the optimization of risk-sensitive formulations such as the conditional value-at-risk (CVaR), which are essential in realistic scheduling processes. The proposed strategy is investigated thoroughly in a parallel batch production environment, and benchmarked against mixed integer linear programming (MILP) strategies. We show that the policy identified by our approach is able to account for plant uncertainties in online decision-making, with expected performance comparable to existing MILP methods. Additionally, the framework gains the benefits of optimizing for risk-sensitive measures, and identifies online decisions orders of magnitude faster than the most efficient optimization approaches. This promises to mitigate practical issues and ease in handling realizations of process uncertainty in the paradigm of online production scheduling. | ['Ehecatl Antonio del Rio Chanona', 'Dongda Zhang', 'Max Mowbray'] | 2022-03-01 | null | null | null | null | ['distributional-reinforcement-learning'] | ['methodology'] | [ 3.26210856e-01 3.75580102e-01 -1.40090257e-01 -4.65808213e-02
-5.80635369e-01 -4.77113754e-01 4.36090976e-01 7.96828032e-01
-2.18047515e-01 1.08839226e+00 -3.31033170e-01 -3.28790128e-01
-9.88823593e-01 -9.95183051e-01 -6.85130596e-01 -7.69490123e-01
-4.58963186e-01 7.76682198e-01 -3.25581402e-01 -1.22561380e-01
3.12279910e-01 8.14363658e-01 -1.55829012e+00 -1.14169523e-01
9.75029111e-01 1.28333497e+00 4.20042008e-01 3.70925516e-01
-8.56676474e-02 8.12320292e-01 -4.88539070e-01 -8.66077468e-02
4.38696772e-01 6.84411600e-02 -7.07107723e-01 3.94213974e-01
-3.40106815e-01 -1.61846623e-01 4.64503169e-01 8.80862355e-01
2.51064450e-01 4.68313992e-01 5.57422519e-01 -1.49719095e+00
1.30717577e-02 7.37329602e-01 -4.93830144e-01 2.57766154e-02
2.94635415e-01 3.32252771e-01 1.04518306e+00 -3.51138830e-01
3.29618573e-01 1.36833310e+00 1.24990150e-01 -1.18664660e-01
-1.30717146e+00 8.55759811e-03 5.33037305e-01 -3.81960310e-02
-1.03261316e+00 1.43249482e-01 4.76372302e-01 -4.72179651e-01
8.21745276e-01 1.84492260e-01 4.78141993e-01 6.92965627e-01
7.06427276e-01 8.76309752e-01 1.33444631e+00 -5.90230703e-01
7.44051099e-01 -9.91318673e-02 -1.76810026e-01 -5.23448922e-02
3.46208632e-01 4.05413061e-01 1.61063839e-02 -3.95782897e-03
5.38123369e-01 1.39347138e-02 3.08440086e-02 -5.67675591e-01
-1.00037682e+00 9.14925992e-01 -2.75564566e-02 1.15631521e-02
-8.69451642e-01 1.60171375e-01 4.42312986e-01 4.52552259e-01
5.05246639e-01 9.29541886e-01 -5.41883528e-01 1.48254126e-01
-6.21106148e-01 4.39526290e-01 1.15684080e+00 1.13356745e+00
2.46371225e-01 2.89436430e-01 -6.08767390e-01 3.31879765e-01
3.05455536e-01 3.44469190e-01 -3.90090019e-01 -1.19042552e+00
6.19598269e-01 6.97220340e-02 7.50362813e-01 -6.03100479e-01
-5.15472472e-01 -3.60584676e-01 -5.77764928e-01 4.87995237e-01
4.37291652e-01 -3.50658715e-01 -5.64206779e-01 1.33037925e+00
1.66222960e-01 -1.45785108e-01 1.89360946e-01 6.19140148e-01
-4.45722640e-01 7.33205557e-01 1.96185082e-01 -9.76957142e-01
1.18019378e+00 -5.59221327e-01 -1.10946214e+00 8.23036209e-02
1.67161286e-01 -7.47384906e-01 5.08996546e-01 9.44248676e-01
-1.17941034e+00 -3.07566017e-01 -8.50345075e-01 8.86549652e-01
-3.10714602e-01 7.57638505e-03 5.72041988e-01 6.65607989e-01
-6.68789566e-01 1.04789293e+00 -6.07661963e-01 -3.95358764e-02
5.38387857e-02 4.92468625e-01 2.18458101e-01 1.76831614e-02
-1.35752165e+00 1.26570404e+00 7.67235756e-01 6.29512846e-01
-1.06272614e+00 -7.67069995e-01 -9.93365943e-01 3.33460808e-01
1.37999737e+00 -1.83297127e-01 1.53971636e+00 -5.81124067e-01
-1.95238924e+00 -8.06999579e-02 5.17386794e-01 -6.15729451e-01
9.27159250e-01 -2.99872816e-01 -3.08038682e-01 1.16703667e-01
-3.70678544e-01 1.66969374e-02 8.62103164e-01 -1.24608397e+00
-7.11992681e-01 -1.52671948e-01 5.38018465e-01 1.30267754e-01
3.30118775e-01 -6.22154698e-02 4.16204274e-01 -4.01371479e-01
-3.40656549e-01 -8.74577045e-01 -9.30472553e-01 -6.33583724e-01
-4.41082537e-01 -3.08487322e-02 2.91586399e-01 -5.25269508e-01
1.03252220e+00 -1.52926219e+00 3.71346831e-01 6.18230820e-01
-5.18066943e-01 -7.48392986e-03 1.88315794e-01 1.04396272e+00
-1.57656334e-02 5.14891632e-02 -3.97508800e-01 -8.47111046e-02
6.20998442e-01 5.72120130e-01 -3.94917667e-01 4.20643330e-01
6.70018256e-01 4.95336831e-01 -9.70265090e-01 -1.47387639e-01
5.36299407e-01 -2.49100715e-01 -1.11537822e-01 2.32367486e-01
-8.68024409e-01 4.28429693e-01 -7.24874318e-01 7.76464105e-01
6.81990921e-01 3.69689941e-01 5.65463901e-01 2.25610688e-01
-5.13439834e-01 -4.22711313e-01 -1.64683449e+00 1.11217022e+00
-1.00244033e+00 -2.17416123e-01 5.49806654e-01 -1.20379400e+00
8.69538546e-01 2.69581079e-01 7.38142073e-01 -3.29991549e-01
1.83793157e-01 1.27742821e-02 -5.71992956e-02 -2.96104848e-01
5.14194071e-01 -2.37285718e-01 -2.14734837e-01 2.68812209e-01
-6.02329038e-02 -6.25934780e-01 6.19601250e-01 -3.86698335e-01
7.81439781e-01 4.55630004e-01 4.87367183e-01 -6.76386535e-01
6.94744706e-01 -1.69039324e-01 7.18743443e-01 6.54255092e-01
4.91860621e-02 -8.50663930e-02 1.13480544e+00 -3.68613042e-02
-7.85840750e-01 -7.85830200e-01 -1.42937914e-01 7.13332474e-01
2.68629454e-02 1.68870211e-01 -3.27609062e-01 -5.55951774e-01
5.90080976e-01 1.32320797e+00 -3.45570952e-01 1.91579819e-01
-2.44992062e-01 -8.43772411e-01 -2.81660348e-01 3.59982550e-01
-1.87187061e-01 -9.20995355e-01 -6.63295329e-01 7.78192222e-01
5.97696304e-01 -1.05970860e+00 1.01907261e-01 5.62803984e-01
-6.20913148e-01 -1.08985257e+00 -5.65448880e-01 -9.33582485e-02
5.31729102e-01 -5.31318903e-01 1.11128879e+00 -6.30507350e-01
-4.39701617e-01 6.26648366e-01 -2.06144065e-01 -7.31331348e-01
-7.48495460e-01 -1.30436763e-01 1.42336711e-01 1.46955386e-01
-2.18316808e-01 -7.75855780e-02 -1.18875550e-03 2.93278664e-01
-1.06731904e+00 -4.65955675e-01 6.01855218e-01 8.31521213e-01
8.06614161e-01 7.24794030e-01 1.11454892e+00 -9.96990740e-01
8.44579518e-01 -5.05818546e-01 -1.48177731e+00 6.72888756e-01
-9.54962492e-01 3.54540259e-01 8.24956238e-01 -1.62559927e-01
-1.37417448e+00 2.03400344e-01 2.73220778e-01 -3.63045722e-01
-3.33494931e-01 7.40126193e-01 -3.84267598e-01 9.60237905e-02
-2.18177214e-01 -1.75194249e-01 6.11830316e-02 -1.49347231e-01
1.16027914e-01 1.07256047e-01 -8.11702386e-02 -1.09102523e+00
6.21697962e-01 -1.36099592e-01 7.35773742e-01 -4.30727154e-01
-5.91851890e-01 -3.01722914e-01 -5.84625781e-01 -3.94001126e-01
6.45590305e-01 -5.20966053e-01 -1.22256601e+00 1.49382085e-01
-7.97544062e-01 -2.13808581e-01 -7.65731275e-01 5.40857971e-01
-1.26161993e+00 2.17954159e-01 -5.74640274e-01 -1.48687303e+00
1.32443458e-01 -1.48607016e+00 7.76601911e-01 1.43418357e-01
9.51562375e-02 -1.08849013e+00 -2.01573879e-01 2.93368623e-02
3.52166682e-01 7.57365227e-01 9.27595079e-01 -5.72214901e-01
-6.62372530e-01 -1.28927514e-01 2.71190077e-01 4.71551239e-01
1.75862357e-01 1.66058764e-01 -6.52679861e-01 -3.63386989e-01
2.07171161e-02 -2.45032892e-01 3.54248792e-01 5.60821831e-01
1.06150115e+00 -1.97834596e-01 3.43472287e-02 -1.44452214e-01
1.66453600e+00 4.67189640e-01 3.31260562e-01 3.43276918e-01
1.21117666e-01 1.27417111e+00 1.45276034e+00 1.05926692e+00
-5.27858064e-02 6.39352977e-01 8.70597601e-01 3.49849701e-01
9.41492975e-01 7.21046180e-02 2.43980139e-01 1.46707535e-01
-2.14485303e-01 -4.17765886e-01 -8.13927174e-01 2.70387799e-01
-2.09845352e+00 -8.70467365e-01 3.89085487e-02 2.36544132e+00
4.83865350e-01 2.71812350e-01 -4.49834056e-02 1.19554415e-01
7.46059895e-01 -2.17226949e-02 -4.73441958e-01 -1.17273283e+00
1.20190486e-01 1.64860711e-01 1.02499986e+00 5.71741879e-01
-1.26698244e+00 3.35794508e-01 6.25555897e+00 7.29172170e-01
-4.47961777e-01 -4.33403432e-01 8.14825177e-01 1.26317012e-04
-1.75316319e-01 6.92246109e-02 -7.81722903e-01 3.41959238e-01
1.13225949e+00 -4.51058626e-01 6.10348880e-01 9.24801826e-01
7.06816733e-01 -4.66595411e-01 -1.34323359e+00 2.42812127e-01
-5.66542268e-01 -9.95238841e-01 -4.35553670e-01 2.42089823e-01
9.61371243e-01 -6.81626439e-01 9.19800699e-02 3.98722202e-01
5.83983719e-01 -1.13137317e+00 6.21854186e-01 7.66250610e-01
1.71603128e-01 -1.50187719e+00 1.12865222e+00 3.19881648e-01
-1.03625047e+00 -6.49223626e-01 -2.32133746e-01 -6.66113943e-02
5.35799623e-01 9.29996431e-01 -8.53912294e-01 1.27234280e+00
1.16509333e-01 3.18332046e-01 2.08392605e-01 9.07571197e-01
-1.52231351e-01 1.30264923e-01 -3.23874712e-01 -6.47523329e-02
5.38652539e-01 -6.03439867e-01 3.78607392e-01 9.69167769e-01
3.19370925e-01 -3.06215346e-01 7.39686310e-01 9.98162091e-01
4.52536911e-01 1.92390963e-01 -5.63426375e-01 -2.52874792e-01
2.23638088e-01 1.07822609e+00 -7.78309643e-01 2.34265268e-01
-1.39119178e-01 3.78119528e-01 -1.51433304e-01 2.08452865e-01
-7.78688610e-01 -2.48881042e-01 5.36193073e-01 -1.36831626e-01
3.41567278e-01 -2.54851371e-01 -5.06961755e-02 -4.05431956e-01
4.61360142e-02 -6.95013702e-01 3.26037467e-01 -2.27484196e-01
-1.55053425e+00 -3.02561224e-02 4.82116699e-01 -1.10201275e+00
-7.24657118e-01 -9.99687016e-01 -5.78230679e-01 1.01696897e+00
-1.88610125e+00 -7.51929820e-01 4.44323212e-01 6.05887324e-02
5.94559968e-01 -2.40394603e-02 5.39192259e-01 -2.12632135e-01
-7.86817849e-01 -1.59418032e-01 5.43477297e-01 -7.67157257e-01
3.51255834e-01 -1.72096550e+00 -2.31765375e-01 7.23979592e-01
-6.95348680e-01 2.27374911e-01 1.01543236e+00 -6.98349297e-01
-1.80265141e+00 -1.14902508e+00 4.20895368e-01 -3.76944505e-02
1.03585589e+00 -8.13854635e-02 -5.68408012e-01 3.43965679e-01
2.32666016e-01 -2.29861706e-01 2.71522522e-01 2.45235418e-03
4.29392427e-01 -2.27580205e-01 -1.49314415e+00 2.61482507e-01
4.51732278e-01 -2.54540816e-02 -3.07468414e-01 5.89784086e-01
6.82089388e-01 -3.86127442e-01 -1.42632222e+00 4.76972282e-01
-3.96032296e-02 -4.45900887e-01 8.38426113e-01 -6.61033392e-01
2.52972186e-01 -8.15908015e-02 -6.58217221e-02 -1.67386115e+00
-1.78160861e-01 -9.99579132e-01 -1.89947188e-01 1.28081679e+00
3.52541953e-01 -6.24440014e-01 2.21999645e-01 9.69149232e-01
-2.33003303e-01 -7.96490133e-01 -1.11989975e+00 -1.15945256e+00
5.51188476e-02 -2.39757836e-01 6.97029054e-01 7.04542398e-01
5.43616489e-02 -4.54470932e-01 -3.32482159e-01 5.12286723e-01
9.08181906e-01 3.06147188e-01 1.54609099e-01 -1.07782102e+00
-4.63381320e-01 -3.62253368e-01 -1.39280140e-01 -1.06408939e-01
4.90933746e-01 -4.77426440e-01 3.24505657e-01 -1.48546028e+00
-5.93359768e-01 -5.71019471e-01 -6.60157919e-01 1.67525142e-01
8.63835737e-02 -9.16115582e-01 5.05417049e-01 -4.80143428e-01
-3.88209343e-01 7.16623485e-01 1.35920131e+00 -1.24244429e-01
-2.53623903e-01 5.24283230e-01 -3.26649666e-01 5.65157115e-01
1.00849700e+00 -2.26531237e-01 -6.23967111e-01 2.44810179e-01
3.63223791e-01 6.71939552e-01 -3.62451412e-02 -6.67923033e-01
-2.29243329e-03 -9.69325960e-01 7.37857148e-02 -5.68112552e-01
1.74750954e-01 -1.40397179e+00 5.11241257e-01 5.79441965e-01
-5.39673150e-01 1.82687789e-01 3.33811104e-01 8.97973657e-01
-2.34962583e-01 -7.82619059e-01 4.60116059e-01 -2.39067405e-01
-7.36622095e-01 1.28489420e-01 -5.57175279e-01 -2.73093909e-01
1.59706783e+00 2.64338672e-01 6.73905835e-02 -2.15335265e-01
-1.02757597e+00 9.26295042e-01 -5.87920547e-02 2.11844400e-01
2.00945944e-01 -8.29979241e-01 -7.51886368e-01 -2.65280575e-01
-1.58423841e-01 7.36594014e-03 2.83758014e-01 8.36015642e-01
-4.73309726e-01 5.61265886e-01 -2.67945170e-01 -3.32763612e-01
-4.99053150e-01 9.89177346e-01 2.22812787e-01 -1.08155024e+00
-1.97992563e-01 1.87554494e-01 -1.54420391e-01 -1.90483585e-01
1.43698364e-01 -7.64592648e-01 -1.32046312e-01 5.07893562e-01
2.87697136e-01 7.01484025e-01 2.12209687e-01 2.58254656e-03
-1.64437920e-01 2.39077546e-02 2.18946472e-01 -1.48472592e-01
1.37583947e+00 -1.44196570e-01 6.23475835e-02 5.61464071e-01
3.11116606e-01 -1.75351888e-01 -1.68106556e+00 4.23401333e-02
7.31048703e-01 -3.09049278e-01 1.28268734e-01 -1.00340152e+00
-9.48177755e-01 5.23863494e-01 2.59488106e-01 5.65887451e-01
1.05797398e+00 -6.42670751e-01 1.26258329e-01 4.85852242e-01
7.91184008e-01 -1.63737988e+00 -3.60913128e-01 5.18562734e-01
1.15988934e+00 -1.01951778e+00 9.24412161e-02 -6.05930567e-01
-9.23577607e-01 1.26326442e+00 3.55156004e-01 -1.58845335e-01
5.39629161e-01 6.69260323e-01 -5.20599365e-01 3.72374088e-01
-7.38147974e-01 -3.83173794e-01 -8.79764333e-02 6.22532368e-01
7.50779063e-02 5.12638986e-01 -3.45365465e-01 5.24795711e-01
4.72476721e-01 5.31174382e-03 7.44659305e-01 1.42216504e+00
-2.41902158e-01 -1.29649878e+00 -6.68105841e-01 3.61506462e-01
-4.70838875e-01 2.54881471e-01 1.89271107e-01 9.88881350e-01
-1.36872027e-02 1.04108310e+00 -1.11720324e-01 3.46475661e-01
7.93441772e-01 1.54904842e-01 4.73999828e-01 -6.92183197e-01
-7.90243149e-01 1.55935571e-01 4.40996021e-01 -5.61865985e-01
-3.01512748e-01 -8.77224803e-01 -1.20684707e+00 7.51213208e-02
-4.09177691e-01 1.93899363e-01 8.15509737e-01 9.27144349e-01
7.57493824e-03 9.71146226e-01 9.46685016e-01 -9.67184007e-01
-1.40826046e+00 -4.95916247e-01 -1.05458462e+00 -1.36089832e-01
-3.49902511e-02 -9.30983722e-01 -2.67324746e-01 -3.54541034e-01] | [4.704955101013184, 2.4353713989257812] |
0d511a01-194d-4b3d-8dc2-896adc5af925 | autonomous-asteroid-characterization-through | 2210.05518 | null | https://arxiv.org/abs/2210.05518v1 | https://arxiv.org/pdf/2210.05518v1.pdf | Autonomous Asteroid Characterization Through Nanosatellite Swarming | This paper first defines a class of estimation problem called simultaneous navigation and characterization (SNAC), which is a superset of simultaneous localization and mapping (SLAM). A SNAC framework is then developed for the Autonomous Nanosatellite Swarming (ANS) mission concept to autonomously navigate about and characterize an asteroid including the asteroid gravity field, rotational motion, and 3D shape. The ANS SNAC framework consists of three modules: 1) multi-agent optical landmark tracking and 3D point reconstruction using stereovision, 2) state estimation through a computationally efficient and robust unscented Kalman filter, and 3) reconstruction of an asteroid spherical harmonic shape model by leveraging a priori knowledge of the shape properties of celestial bodies. Despite significant interest in asteroids, there are several limitations to current asteroid rendezvous mission concepts. First, completed missions heavily rely on human oversight and Earth-based resources. Second, proposed solutions to increase autonomy make oversimplifying assumptions about state knowledge and information processing. Third, asteroid mission concepts often opt for high size, weight, power, and cost (SWaP-C) avionics for environmental measurements. Finally, such missions often utilize a single spacecraft, neglecting the benefits of distributed space systems. In contrast, ANS is composed of multiple autonomous nanosatellites equipped with low SWaP-C avionics. The ANS SNAC framework is validated through a numerical simulation of three spacecraft orbiting asteroid 433 Eros. The simulation results demonstrate that the proposed architecture provides autonomous and accurate SNAC in a safe manner without an a priori shape model and using only low SWaP-C avionics. | ["Simone D'Amico", 'Nathan Stacey', 'Kaitlin Dennison'] | 2022-10-11 | null | null | null | null | ['simultaneous-localization-and-mapping', 'landmark-tracking'] | ['computer-vision', 'computer-vision'] | [-4.08565909e-01 -2.30033040e-01 3.04794729e-01 7.88599849e-02
-8.12040269e-02 -1.05746865e+00 8.87154162e-01 -2.91612476e-01
-6.40087247e-01 9.61129904e-01 -5.26571989e-01 -2.47291997e-01
-5.73321581e-01 -5.72441399e-01 -5.47721267e-01 -6.43444121e-01
-4.76668447e-01 9.50544238e-01 2.05574736e-01 -6.35308266e-01
1.94395900e-01 1.07055831e+00 -1.34004867e+00 -9.73313570e-01
1.27566946e+00 9.95363057e-01 4.62633491e-01 6.62673056e-01
4.52328205e-01 2.77804047e-01 -4.54692692e-01 5.48105001e-01
6.22064710e-01 -1.70967668e-01 -4.08013850e-01 1.06057078e-01
-2.90049836e-02 -1.69412866e-01 5.09892264e-03 1.06057453e+00
1.18032306e-01 4.60004240e-01 4.56955016e-01 -1.43433928e+00
1.42994821e-01 -2.36283317e-01 -4.34719808e-02 -5.86967915e-02
1.74333677e-01 2.85076588e-01 5.01518369e-01 -8.72560143e-01
3.68645668e-01 8.22079659e-01 9.21257794e-01 -1.74297050e-01
-6.89296007e-01 -3.23040217e-01 -2.10688412e-01 -1.13620631e-01
-1.61669576e+00 -2.68754989e-01 -3.01299058e-02 -6.27030253e-01
7.51922905e-01 4.79890674e-01 9.90959585e-01 2.12369084e-01
6.54933333e-01 9.21550989e-02 1.03261542e+00 -1.57733232e-01
5.61541677e-01 -1.54047132e-01 1.22747002e-02 8.04948211e-01
1.16733873e+00 3.07277948e-01 -4.37067360e-01 -4.46098387e-01
6.62406564e-01 -8.77282247e-02 -6.11848295e-01 -7.16458738e-01
-1.48915708e+00 7.07796931e-01 1.96900591e-01 -2.98371576e-02
-7.24005401e-01 -5.70109561e-02 -1.48944240e-02 3.09782863e-01
-2.33326599e-01 7.91984975e-01 -1.70056641e-01 -8.62129331e-02
-7.94691682e-01 2.81266242e-01 1.15550637e+00 9.81178105e-01
8.23026061e-01 9.51407075e-01 6.21345043e-01 3.11336666e-02
6.83591247e-01 1.51513064e+00 4.68487412e-01 -8.49931836e-01
3.88834812e-03 4.84248787e-01 1.00312328e+00 -7.36061275e-01
-7.13728487e-01 -6.14753187e-01 -4.75785971e-01 6.33175492e-01
-2.41122730e-02 -5.13741255e-01 -5.73074579e-01 1.05294967e+00
6.53453588e-01 -9.44332927e-02 6.37294054e-01 1.26080751e+00
4.20767069e-02 5.27123928e-01 -5.12901127e-01 2.55171885e-03
1.47253549e+00 -5.49650788e-01 -5.59284329e-01 -7.11822689e-01
6.16283178e-01 -4.84504461e-01 3.54513973e-01 3.33392859e-01
-5.26327908e-01 -1.91954583e-01 -1.51960051e+00 5.46445429e-01
-1.88523516e-01 5.58714390e-01 5.72451591e-01 5.24797499e-01
-1.11228418e+00 1.55939579e-01 -1.20040846e+00 -5.66505671e-01
-5.10954082e-01 5.61119258e-01 -2.74559319e-01 3.76106828e-01
-1.13922787e+00 1.44870090e+00 5.19877493e-01 2.30536610e-01
-1.33240938e+00 -5.27454138e-01 -1.00387430e+00 -2.09666699e-01
1.86711982e-01 -8.11851144e-01 9.68398035e-01 -4.95135874e-01
-1.52022016e+00 1.09759286e-01 6.36473671e-02 -9.71418083e-01
2.42914110e-01 -5.19743204e-01 -5.61339200e-01 8.91069025e-02
2.13501468e-01 6.60029277e-02 6.27507567e-01 -1.19856501e+00
-9.08063412e-01 -4.26141798e-01 -2.18157515e-01 7.91883051e-01
2.02302292e-01 -4.50700819e-01 1.92339018e-01 -2.42980689e-01
8.12803030e-01 -1.64615405e+00 -3.11073899e-01 2.67646890e-02
3.67820300e-02 3.78772557e-01 1.06984532e+00 -4.60370183e-01
3.98942351e-01 -1.86131442e+00 2.13120356e-01 3.51132631e-01
-1.82606220e-01 -2.72444449e-02 4.47914600e-01 5.90941310e-01
4.24602419e-01 -4.80385393e-01 -7.88106993e-02 2.14616768e-02
9.68981013e-02 3.33859742e-01 -5.30606270e-01 1.20771456e+00
-6.69091046e-01 3.89429867e-01 -7.19842494e-01 2.58068025e-01
2.12734997e-01 2.63440162e-01 -1.40151083e-01 -1.97972521e-01
-9.59437415e-02 6.15967035e-01 -6.79442525e-01 8.67439806e-01
7.57990003e-01 -5.39076701e-02 1.82895079e-01 -2.74017572e-01
-1.10062289e+00 -1.23203866e-01 -1.43213499e+00 1.57217669e+00
-2.90177107e-01 3.71683359e-01 1.01578116e+00 -5.42895138e-01
1.14062953e+00 3.27376127e-02 5.50926089e-01 -1.87756017e-01
2.63923675e-01 6.47147834e-01 -7.12257624e-02 -1.79227009e-01
1.08998346e+00 -1.72084555e-01 -2.39223510e-01 3.07177901e-01
-9.83111635e-02 -6.54865921e-01 -1.08064070e-01 1.88499972e-01
7.60264277e-01 1.24513246e-01 4.83098090e-01 -9.57581997e-01
7.09783852e-01 6.20265245e-01 8.50810468e-01 4.96489495e-01
4.71606851e-02 -1.47578433e-01 -3.18317145e-01 -3.24832767e-01
-7.61963248e-01 -8.11399400e-01 -6.96080998e-02 2.44511113e-01
9.27886844e-01 -1.42003428e-02 5.18027656e-02 -9.51959714e-02
4.82666641e-01 7.03507423e-01 -1.48263156e-01 -1.60370767e-02
-2.49947965e-01 -7.63688684e-01 6.41447902e-01 -6.47141039e-02
6.14124238e-01 8.68588965e-03 -1.24634576e+00 5.90925291e-02
-5.48107401e-02 -8.72644842e-01 1.10322237e-01 -6.39904439e-02
-8.81602526e-01 -1.08693552e+00 -3.37733924e-01 -1.59196675e-01
6.13803983e-01 7.29041696e-01 3.40443760e-01 -6.07307851e-02
3.61825936e-02 9.57953632e-01 -4.23516184e-01 -5.57907939e-01
-6.44716620e-02 -5.53598285e-01 1.11575043e+00 6.94508851e-02
-2.51338005e-01 -2.39625424e-01 -4.36014503e-01 8.05412531e-01
-3.96336883e-01 -2.86717832e-01 6.88852668e-01 4.82146859e-01
5.31964839e-01 1.56217545e-01 1.18556894e-01 8.58665928e-02
1.32304102e-01 -5.88798702e-01 -1.42280138e+00 5.11786975e-02
-6.60942495e-01 -6.23450987e-02 4.21607465e-01 5.44668511e-02
-1.08414197e+00 2.23464698e-01 6.28127515e-01 -2.30314404e-01
2.26155639e-01 7.94657588e-01 2.92202443e-01 -9.84336019e-01
6.40307248e-01 7.27927864e-01 8.33755732e-01 -2.07440540e-01
-4.83377390e-02 6.13432348e-01 7.01346457e-01 -4.16705251e-01
1.22205150e+00 1.21986115e+00 4.47588533e-01 -1.24342811e+00
-4.11012888e-01 -5.42449653e-01 -4.89696056e-01 -4.89873737e-01
7.33587146e-01 -1.34095192e+00 -7.98799574e-01 3.08472723e-01
-6.57147944e-01 -1.74826175e-01 -1.15924001e-01 1.33715594e+00
-5.34977913e-01 7.51677275e-01 6.80823550e-02 -1.04516435e+00
-5.23420513e-01 -1.04048491e+00 7.56173909e-01 3.24443460e-01
-1.29888281e-01 -8.26865435e-01 4.09085035e-01 1.74311146e-01
5.94089508e-01 1.95050582e-01 -1.95976287e-01 -4.23622787e-01
-1.05022621e+00 -3.83540750e-01 2.97629923e-01 -2.36215100e-01
1.32409245e-01 -4.58666950e-01 -8.79244655e-02 -1.21948791e+00
4.35501188e-01 -1.71582431e-01 1.70530558e-01 1.03769667e-01
-2.26028442e-01 -1.70969531e-01 -6.05306864e-01 8.49775910e-01
1.57996356e+00 4.08495873e-01 1.05123557e-01 9.45377648e-01
2.35300288e-01 1.72601745e-01 1.39496017e+00 8.76581669e-01
3.93405259e-01 4.27107006e-01 1.06791413e+00 3.71844292e-01
4.28110212e-01 2.76938736e-01 5.60622871e-01 6.83018386e-01
-3.32727939e-01 -1.92848723e-02 -1.11559749e+00 6.22230709e-01
-1.58951461e+00 -5.16349077e-01 -4.55689728e-01 2.42022395e+00
-1.37865722e-01 -4.41389143e-01 -5.95363140e-01 -4.86493587e-01
5.24255812e-01 1.42268660e-02 -6.85406804e-01 2.76721388e-01
-1.50852010e-01 -7.23475993e-01 1.40382826e+00 6.64523125e-01
-7.56201267e-01 6.61197901e-01 5.20721245e+00 8.08897167e-02
-1.16678059e+00 -5.37773222e-02 -1.00322926e+00 3.54663670e-01
-3.08249176e-01 7.77640641e-01 -1.19726133e+00 1.78858161e-01
9.51529264e-01 -4.94502604e-01 3.45654070e-01 9.24961030e-01
2.21628085e-01 -5.06005287e-01 -3.80372614e-01 7.68335938e-01
1.89181820e-01 -1.40500987e+00 -4.21957206e-03 2.91275829e-01
5.45500875e-01 6.12035334e-01 -3.05127531e-01 5.93732148e-02
5.25264502e-01 -1.85632020e-01 1.12500858e+00 8.78065646e-01
4.94903028e-01 -5.53675234e-01 6.61351025e-01 4.98519748e-01
-1.31503892e+00 -3.88230145e-01 -4.51789349e-01 -2.33157516e-01
6.99225843e-01 3.71496320e-01 -1.05464554e+00 1.28678632e+00
8.28816295e-01 3.28055650e-01 -1.59130186e-01 1.29265797e+00
2.30847131e-02 1.15078859e-01 -8.68900061e-01 -2.27500036e-01
6.21322513e-01 -8.17353845e-01 1.41694283e+00 1.76517293e-01
1.14211333e+00 3.80022794e-01 3.98683935e-01 5.49359143e-01
7.25605309e-01 -2.46847481e-01 -5.31351388e-01 7.27098212e-02
7.55859911e-01 1.33206117e+00 -4.61991966e-01 -2.69994587e-01
-2.81165034e-01 4.61880356e-01 -4.33379859e-01 2.50536114e-01
-8.13247204e-01 -4.29981053e-01 7.65832961e-01 -5.49175870e-03
2.27974325e-01 -1.17998052e+00 -2.19575204e-02 -1.01579750e+00
-4.37037140e-01 -5.95071793e-01 1.98227733e-01 -1.02379799e+00
-4.29649323e-01 6.75937772e-01 -8.00621212e-02 -2.05677104e+00
-1.86483875e-01 -4.37030971e-01 -4.95445579e-01 8.86773109e-01
-1.27659631e+00 -1.16508853e+00 -6.86537504e-01 4.65197980e-01
1.35942131e-01 -7.56393373e-01 6.31796122e-01 -4.08315122e-01
2.35869903e-02 -5.36023855e-01 5.58238328e-01 -5.77970147e-01
5.97035527e-01 -9.69083071e-01 6.76961392e-02 1.11649227e+00
-3.50462228e-01 6.55259013e-01 1.22031057e+00 -1.40083909e+00
-2.37110448e+00 -1.06106055e+00 3.33500832e-01 -7.87841678e-02
9.52148080e-01 2.56206859e-02 -4.70685482e-01 1.18576109e+00
2.62970999e-02 -2.95221508e-01 1.81052536e-01 -1.68103456e-01
5.92814922e-01 -1.36098906e-01 -9.67439294e-01 6.04571760e-01
1.04872793e-01 -1.08336657e-01 -6.40748322e-01 7.01811135e-01
2.96294212e-01 -4.99844164e-01 -1.00372124e+00 7.68002331e-01
4.08717155e-01 -5.17067611e-01 7.13678181e-01 -1.51281655e-02
-1.26261389e+00 -1.31811678e+00 -2.88450360e-01 -1.33608842e+00
-2.37923071e-01 -8.45376730e-01 2.46524006e-01 4.54158098e-01
5.35036735e-02 -1.07221353e+00 8.00989270e-01 3.31076324e-01
-7.53667891e-01 2.72393048e-01 -1.20493269e+00 -1.43003762e+00
-5.52409232e-01 1.42421886e-01 2.78433144e-01 9.56721127e-01
-1.56194627e-01 -5.03039081e-03 -3.74366701e-01 1.33389449e+00
1.05654812e+00 3.67565066e-01 1.09673822e+00 -1.66753900e+00
4.03602831e-02 4.68707234e-01 -3.41934800e-01 -9.78883028e-01
2.26267781e-02 -6.06722653e-01 2.11070552e-01 -1.32120764e+00
-7.58484066e-01 -6.58374727e-01 5.74265242e-01 1.60352692e-01
6.65389717e-01 -2.10840702e-01 -6.23809397e-02 7.84661651e-01
-4.62939322e-01 1.15013802e+00 7.66460240e-01 4.64131162e-02
-1.05080575e-01 1.68712512e-01 1.89278752e-01 8.56317401e-01
5.98123431e-01 -2.64981270e-01 -5.18115342e-01 -5.54729879e-01
3.39665473e-01 4.78140742e-01 6.31795108e-01 -1.58036077e+00
1.07966983e+00 -1.83149487e-01 2.26777084e-02 -1.07978749e+00
7.92738318e-01 -1.01029706e+00 8.58119667e-01 9.63770866e-01
8.88998449e-01 2.01324329e-01 2.72267312e-01 8.94867122e-01
-4.40381646e-01 -3.71397525e-01 8.65889013e-01 -1.77927971e-01
-1.19368684e+00 1.18564181e-01 -9.26327169e-01 -5.80559850e-01
1.40263915e+00 -4.48374331e-01 -3.43028754e-01 -4.09507334e-01
-6.68313384e-01 6.02649570e-01 9.01000619e-01 -4.17781770e-02
5.52100599e-01 -9.04756546e-01 -4.96453077e-01 4.30437028e-01
-1.04261801e-01 1.40305301e-02 4.55618948e-01 1.13925683e+00
-1.20204866e+00 7.31493950e-01 -5.49177587e-01 -8.04915428e-01
-8.79348278e-01 9.50559229e-02 5.47554910e-01 5.07863283e-01
-5.06936610e-01 4.14844066e-01 -1.22193083e-01 -7.19553649e-01
-6.83813393e-01 1.26620978e-01 3.84659623e-03 -1.02764063e-01
3.73777241e-01 5.25027215e-01 9.35230404e-02 -9.86297905e-01
-5.36759198e-01 6.24876320e-01 2.99230158e-01 -4.32226658e-01
9.80472326e-01 -6.97971702e-01 -2.84344554e-01 1.43807992e-01
1.74756825e-01 2.51399159e-01 -1.17368686e+00 9.60118994e-02
8.29334185e-02 -3.15263450e-01 3.53174746e-01 -3.95313352e-01
-4.37083334e-01 -9.45461541e-02 3.99119496e-01 -5.56367002e-02
5.74242413e-01 -4.53806341e-01 3.70233536e-01 9.91142511e-01
1.09967518e+00 -1.08039749e+00 -4.91813093e-01 1.10112691e+00
1.07006526e+00 -7.08677232e-01 5.02838314e-01 -1.05771340e-01
-7.91422188e-01 1.07974195e+00 4.92245376e-01 -3.38947833e-01
6.33937895e-01 1.87287524e-01 6.63688257e-02 -3.60854656e-01
-4.91354734e-01 -1.38468921e-01 -1.52815729e-02 2.90400326e-01
-5.07667363e-01 -1.69852808e-01 -3.56638193e-01 3.38108152e-01
-3.32008421e-01 -4.47596461e-01 9.30040002e-01 1.48796666e+00
-1.10022461e+00 -4.44990903e-01 -1.11057770e+00 5.46536110e-02
1.22608654e-01 2.31070980e-01 9.36630517e-02 1.13605928e+00
-3.54859740e-01 5.78989863e-01 1.35116354e-01 -1.78089619e-01
3.24093461e-01 -4.96528834e-01 8.71216208e-02 -1.78488851e-01
-6.34309128e-02 -1.14141867e-01 2.73491293e-01 -5.64972401e-01
-1.68299317e-01 -8.80005896e-01 -1.47588909e+00 -4.19455133e-02
-4.38268393e-01 1.04263318e+00 1.30030382e+00 9.60163474e-01
5.38380504e-01 1.83108337e-02 4.84906763e-01 -1.03803742e+00
-9.88565385e-01 -9.09559846e-01 -1.29651582e+00 -5.51529408e-01
3.18219841e-01 -1.19617319e+00 -6.55668914e-01 -4.06414866e-01] | [7.307408332824707, -1.8262383937835693] |
333eb8a1-f866-4815-b693-0811e4e6bad6 | a-boundary-aware-neural-model-for-nested | null | null | https://aclanthology.org/D19-1034 | https://aclanthology.org/D19-1034.pdf | A Boundary-aware Neural Model for Nested Named Entity Recognition | In natural language processing, it is common that many entities contain other entities inside them. Most existing works on named entity recognition (NER) only deal with flat entities but ignore nested ones. We propose a boundary-aware neural model for nested NER which leverages entity boundaries to predict entity categorical labels. Our model can locate entities precisely by detecting boundaries using sequence labeling models. Based on the detected boundaries, our model utilizes the boundary-relevant regions to predict entity categorical labels, which can decrease computation cost and relieve error propagation problem in layered sequence labeling model. We introduce multitask learning to capture the dependencies of entity boundaries and their categorical labels, which helps to improve the performance of identifying entities. We conduct our experiments on GENIA dataset and the experimental results demonstrate that our model outperforms other state-of-the-art methods. | ['ong', 'Ho-fung Leung', 'Jingyun Xu', 'Yi Cai', 'Gu Xu', 'Changmeng Zheng'] | 2019-11-01 | null | null | null | ijcnlp-2019-11 | ['nested-named-entity-recognition'] | ['natural-language-processing'] | [-3.90005052e-01 1.23253353e-01 -4.07184750e-01 -5.78094542e-01
-5.55375040e-01 -9.05865610e-01 9.45481360e-02 4.34287369e-01
-6.34235084e-01 7.24599361e-01 3.59399676e-01 -2.29444936e-01
2.44014740e-01 -9.69548583e-01 -6.71928048e-01 -1.56309813e-01
-3.24866056e-01 3.35929930e-01 4.02392298e-01 1.76200226e-01
1.67163819e-01 7.76928216e-02 -8.55032384e-01 5.58494568e-01
1.19918251e+00 6.10442936e-01 4.41213325e-02 1.99587792e-01
-7.96482205e-01 1.05878305e+00 -5.99611938e-01 -5.04009783e-01
-4.77077402e-02 9.56395939e-02 -1.22644508e+00 -5.21680355e-01
1.01758830e-01 -7.37324506e-02 -2.06727207e-01 9.07825589e-01
4.50116724e-01 5.60000837e-02 7.97553003e-01 -8.69530439e-01
-1.10472429e+00 9.09467399e-01 -5.24113595e-01 3.08097780e-01
1.10277094e-01 -4.76244539e-01 1.39117050e+00 -1.10254860e+00
6.28252804e-01 9.50125456e-01 1.00256252e+00 3.56647640e-01
-8.21998954e-01 -7.64177322e-01 5.70128679e-01 1.09815426e-01
-1.74323213e+00 -1.80821106e-01 3.25546741e-01 -4.46326762e-01
1.21739531e+00 -1.13865897e-01 -6.92155585e-02 5.11778057e-01
-1.44989103e-01 8.25155318e-01 6.00098789e-01 -4.39126730e-01
4.05387729e-02 -1.78649768e-01 6.01279259e-01 8.81994784e-01
3.77929330e-01 -4.37930882e-01 -6.78476095e-02 -1.45237803e-01
5.11736333e-01 -3.38747017e-02 9.27191377e-02 -1.35894164e-01
-1.12988508e+00 5.84153116e-01 5.99216223e-01 6.12419605e-01
-4.63477820e-01 -7.29936780e-03 6.20797098e-01 -8.90480205e-02
5.44697225e-01 3.29137921e-01 -9.75711346e-01 2.38753945e-01
-7.16162086e-01 -2.42146641e-01 9.78087366e-01 1.23911202e+00
9.32470500e-01 -3.33450496e-01 -4.22455281e-01 1.09857416e+00
4.05619532e-01 1.28386185e-01 3.86350691e-01 -3.40075761e-01
7.25732565e-01 1.09893692e+00 2.34517246e-01 -8.84420455e-01
-8.48097265e-01 -3.83582026e-01 -6.67120516e-01 -6.74095213e-01
2.07191929e-01 -6.71275020e-01 -1.07243931e+00 1.76274443e+00
5.69160640e-01 5.63292444e-01 1.24379709e-01 6.23946786e-01
1.29458773e+00 7.91521907e-01 7.24754393e-01 7.35880733e-02
1.65776885e+00 -1.07427669e+00 -7.18435645e-01 -1.97523743e-01
1.32941043e+00 -3.73081625e-01 6.15320683e-01 -2.68869728e-01
-4.17871118e-01 -4.77143437e-01 -5.87061405e-01 -2.71810979e-01
-6.93728685e-01 6.78544283e-01 5.95709860e-01 6.27255380e-01
-7.69492865e-01 2.67103672e-01 -8.56969416e-01 -1.54378504e-01
3.07887226e-01 3.34085524e-01 -2.98994362e-01 2.17081919e-01
-1.72171426e+00 8.03658009e-01 9.64847982e-01 3.39562118e-01
-5.14717579e-01 -7.45028019e-01 -1.25114346e+00 4.43406165e-01
3.99578363e-01 -4.02412176e-01 1.17267895e+00 -3.97912174e-01
-8.00829411e-01 9.41364884e-01 -5.81776202e-01 -2.91207701e-01
5.24929166e-02 -2.02482149e-01 -6.91031814e-01 -2.39542305e-01
3.34728479e-01 6.83443427e-01 -8.38098153e-02 -1.03157902e+00
-9.02534366e-01 -7.47607872e-02 -4.50976044e-02 2.17187792e-01
-4.30960774e-01 5.05347140e-02 -4.61576104e-01 -5.45913041e-01
2.24514324e-02 -6.52301550e-01 -3.61273378e-01 -7.60619640e-01
-7.20522285e-01 -8.78150105e-01 4.37410295e-01 -8.24256599e-01
1.71960306e+00 -1.98652005e+00 -4.23076391e-01 2.38000378e-02
3.93155843e-01 2.68051177e-01 -9.00604725e-02 5.86405754e-01
-1.19794481e-01 5.81875563e-01 -5.95148690e-02 -1.53111905e-01
1.17943548e-01 -7.31656328e-02 -3.37904960e-01 1.93595752e-01
3.21095556e-01 1.18575191e+00 -7.69887626e-01 -7.71361589e-01
-3.00409406e-01 2.96080053e-01 -7.51978680e-02 1.87571555e-01
-9.46060643e-02 1.27519965e-01 -7.72260547e-01 5.88518798e-01
7.80221581e-01 -6.42169893e-01 3.91724557e-01 -1.57254398e-01
-2.73017436e-01 6.31123126e-01 -1.20918405e+00 1.46234000e+00
-3.96030277e-01 1.75132036e-01 -2.90373504e-01 -7.80939519e-01
9.76014078e-01 5.66203356e-01 1.74592644e-01 -4.04204220e-01
-2.25249320e-01 1.95132703e-01 -2.47117862e-01 -6.45982742e-01
4.95907605e-01 7.58576989e-02 -4.60239738e-01 2.03609928e-01
-4.39716615e-02 9.49591935e-01 2.84994990e-01 1.10252865e-01
1.14668560e+00 1.24134123e-01 4.95498478e-01 -9.97885019e-02
6.62542164e-01 8.15203339e-02 1.20905077e+00 6.10113919e-01
-2.76471525e-01 5.69163971e-02 3.93410981e-01 -5.95591366e-01
-9.17291045e-01 -7.99528539e-01 -6.26655817e-02 1.48651874e+00
3.59585136e-01 -3.02606404e-01 -7.45208204e-01 -1.26723683e+00
-1.04781665e-01 6.30705118e-01 -5.92175961e-01 2.34465018e-01
-8.88381243e-01 -6.95538223e-01 1.04613400e+00 8.06099594e-01
5.62477767e-01 -1.24039745e+00 9.21366364e-02 2.82506377e-01
-4.71223295e-01 -1.21221018e+00 -6.34864688e-01 1.97231978e-01
-6.12388670e-01 -8.56179655e-01 -7.13271558e-01 -1.53028703e+00
6.97432935e-01 -7.52438381e-02 1.07784486e+00 1.10733390e-01
1.11646689e-01 -2.15304449e-01 -3.74429196e-01 -1.87215641e-01
2.15150714e-02 7.76196778e-01 -2.19895646e-01 -2.89595306e-01
8.79570484e-01 -3.31684560e-01 -4.17569727e-01 4.41119373e-01
-7.01660275e-01 -5.68620712e-02 6.99781120e-01 7.56078720e-01
4.20731157e-01 -1.02911862e-02 1.08755839e+00 -1.45323443e+00
3.36448401e-01 -9.10168588e-01 -4.85476017e-01 8.27457845e-01
-5.05695820e-01 3.08615446e-01 8.37846696e-01 -4.08748209e-01
-1.53526664e+00 7.69893304e-02 -1.85760513e-01 5.65918442e-03
-5.07322848e-01 9.44693863e-01 -2.47717157e-01 2.84330636e-01
3.53509605e-01 1.75981998e-01 -1.16659653e+00 -6.92894876e-01
4.25330669e-01 7.76894748e-01 2.99713910e-01 -6.39753222e-01
4.17976141e-01 1.72285214e-01 -4.28037882e-01 -4.30733651e-01
-1.19990361e+00 -8.26050282e-01 -8.92685771e-01 1.38491869e-01
9.47777629e-01 -1.19545889e+00 -8.39634359e-01 3.15775275e-01
-1.34881926e+00 -1.24907069e-01 2.83595175e-01 3.97105306e-01
7.36963972e-02 4.63019311e-01 -1.26849067e+00 -6.85424447e-01
-4.00767237e-01 -5.00734150e-01 9.42126811e-01 6.15234435e-01
6.74178377e-02 -1.19168997e+00 4.08829778e-01 6.38801381e-02
9.70401987e-02 -8.77503678e-02 1.07450533e+00 -1.36810923e+00
-4.81375486e-01 -2.42878329e-02 -4.77091134e-01 -3.91287744e-01
8.94968733e-02 -2.63642639e-01 -6.54516220e-01 8.37071911e-02
-6.16387606e-01 -2.81444699e-01 9.77203012e-01 2.03858390e-01
8.08234692e-01 -2.75393277e-01 -9.09733772e-01 5.12968004e-01
1.30111408e+00 2.29411617e-01 3.84483308e-01 4.67101306e-01
1.01816177e+00 7.33871341e-01 6.55515492e-01 2.13086367e-01
1.05682611e+00 1.59466252e-01 -2.09667534e-02 -3.11353832e-01
3.22150201e-01 -6.42437220e-01 -4.72511537e-02 9.24980164e-01
3.45843256e-01 -4.25516516e-01 -1.29367459e+00 9.74750757e-01
-1.92032027e+00 -7.19782472e-01 -2.01856390e-01 1.68721151e+00
1.03959310e+00 -7.64707401e-02 -1.26334473e-01 -6.90359831e-01
1.53124762e+00 1.01855591e-01 -6.82317555e-01 -1.58778816e-01
6.28534588e-04 -1.82021260e-01 6.02989733e-01 1.98259234e-01
-1.68773234e+00 1.36660707e+00 6.21635866e+00 7.78614104e-01
-6.83567286e-01 1.13393143e-01 7.27264941e-01 5.74090600e-01
-2.68161267e-01 1.60373058e-02 -1.42843866e+00 5.16587555e-01
9.07673299e-01 -1.77758813e-01 -1.90261856e-01 1.06433380e+00
-5.56799136e-02 2.88037211e-01 -8.70422482e-01 5.67165494e-01
-1.54567510e-01 -1.23540545e+00 -1.40677989e-01 -1.31740831e-02
8.36115718e-01 2.38416836e-01 -4.24119085e-01 7.95500994e-01
9.15395796e-01 -9.36074138e-01 4.38653797e-01 4.81431574e-01
7.49551058e-01 -7.73094058e-01 9.67541456e-01 5.97839653e-01
-1.76170254e+00 -4.02415879e-02 -4.56953257e-01 8.06477740e-02
4.31955993e-01 6.42544746e-01 -9.45395947e-01 6.93748295e-01
5.96025050e-01 5.71296453e-01 -3.66111159e-01 1.21976089e+00
-5.38702130e-01 8.78481925e-01 -3.27090710e-01 -1.54899031e-01
2.37576932e-01 -8.23459029e-02 3.46719213e-02 1.79895937e+00
1.86324343e-01 2.54053682e-01 4.97697175e-01 7.41343975e-01
-7.41107881e-01 4.69074756e-01 -2.91798979e-01 -1.22031100e-01
1.03732777e+00 1.43223226e+00 -9.95251596e-01 -4.29371655e-01
-7.49353111e-01 8.46869946e-01 1.02187729e+00 3.49809557e-01
-9.56676066e-01 -9.33375001e-01 3.43394488e-01 -5.49631834e-01
6.73057556e-01 -2.81809211e-01 -5.12210369e-01 -1.36666238e+00
8.40467811e-02 -1.56044215e-01 9.28330779e-01 -4.38993245e-01
-1.52630174e+00 5.28951347e-01 -3.61429155e-01 -8.20215404e-01
3.70615907e-02 -2.28445813e-01 -6.26217246e-01 7.63959110e-01
-1.62209463e+00 -1.24164331e+00 3.56313922e-02 1.92039665e-02
3.49954784e-01 2.77838379e-01 7.37135470e-01 5.69675684e-01
-9.57197249e-01 7.57281661e-01 1.23583958e-01 1.19497406e+00
8.68132651e-01 -1.30849993e+00 9.76128757e-01 9.08476412e-01
2.63249010e-01 1.03112519e+00 1.22403009e-02 -1.04115093e+00
-5.11839807e-01 -1.28061497e+00 1.61678064e+00 -4.10401583e-01
5.40295959e-01 -4.54328150e-01 -1.09086144e+00 1.08418858e+00
1.59291252e-01 -1.00350203e-02 1.07417798e+00 6.20893002e-01
-7.00034261e-01 3.45641196e-01 -8.35718930e-01 3.75185639e-01
1.21815658e+00 -4.68313456e-01 -9.82187033e-01 7.10278228e-02
1.08867645e+00 -3.03679556e-01 -9.86974239e-01 5.13378203e-01
2.50880420e-01 -3.20225120e-01 8.35845232e-01 -8.04218590e-01
1.58378646e-01 -5.31357765e-01 4.71850783e-01 -1.10183823e+00
-5.24052262e-01 3.64427567e-02 -2.19507650e-01 1.83309424e+00
1.02511477e+00 -5.59861004e-01 7.54727721e-01 8.90368342e-01
-1.21885411e-01 -5.81334591e-01 -6.91516995e-01 -7.31945455e-01
5.12827113e-02 2.13526525e-02 5.98615944e-01 1.53703451e+00
1.93932861e-01 7.51932144e-01 -1.68053642e-01 5.77123225e-01
1.62715957e-01 3.58111531e-01 1.43376440e-01 -1.27395797e+00
2.01617092e-01 -5.25269285e-02 -5.84773570e-02 -1.35225618e+00
5.68739116e-01 -9.57305253e-01 2.50001669e-01 -1.83100796e+00
3.23166698e-01 -9.10213768e-01 -5.70832491e-01 1.01647401e+00
-6.21986747e-01 -9.14790109e-02 -1.53188154e-01 3.32562178e-01
-1.29433417e+00 3.37547481e-01 5.69677353e-01 1.14385262e-01
-3.55049193e-01 -3.39294106e-01 -6.38786614e-01 7.31327057e-01
5.94608605e-01 -8.41519654e-01 1.07117251e-01 -6.92452252e-01
5.29635251e-01 -2.04668254e-01 -4.31096375e-01 -7.00205684e-01
6.75466597e-01 -6.27110973e-02 5.73188961e-01 -7.44292915e-01
-4.02882040e-01 -5.14781535e-01 -1.03357613e-01 1.45974129e-01
-6.33671522e-01 -2.15111058e-02 -4.28783819e-02 6.50074065e-01
-2.73136079e-01 -4.62504745e-01 2.91015238e-01 -9.58582088e-02
-1.19684052e+00 4.38001484e-01 -3.82446885e-01 4.58538115e-01
9.93880689e-01 1.73655942e-01 -4.81569946e-01 9.84608904e-02
-7.75585413e-01 8.74505401e-01 1.86494231e-01 3.67206931e-01
1.81744248e-01 -1.27984929e+00 -6.48055851e-01 -2.84368813e-01
1.69221878e-01 2.38621742e-01 4.22706783e-01 4.79269385e-01
-5.51967502e-01 6.15088284e-01 2.34437093e-01 -2.97699898e-01
-1.16708422e+00 8.20619226e-01 2.94975042e-01 -6.72706485e-01
-3.04910153e-01 1.03391206e+00 5.32588184e-01 -1.15128398e+00
2.46962011e-01 -1.28973678e-01 -8.44724715e-01 7.02012405e-02
5.00628233e-01 2.26298273e-01 -9.31104198e-02 -5.67773342e-01
-6.26975119e-01 3.66078615e-01 -5.29021382e-01 2.01292112e-01
1.21771002e+00 -2.94292390e-01 -2.92658567e-01 4.33427274e-01
1.00428832e+00 2.35465914e-01 -8.68289351e-01 -5.64012945e-01
8.62167895e-01 9.49270651e-02 -3.17533463e-01 -7.09920704e-01
-7.04512000e-01 7.11807728e-01 7.95504674e-02 2.12905496e-01
8.49492729e-01 3.72471921e-02 1.05461586e+00 7.03687906e-01
4.12249178e-01 -9.81256247e-01 -5.32422245e-01 9.75540698e-01
-8.96053165e-02 -1.06714332e+00 -3.91156703e-01 -8.24407816e-01
-7.51722038e-01 9.32137907e-01 9.42543149e-01 1.87637135e-02
5.72563529e-01 3.36367458e-01 1.72640994e-01 1.24560436e-02
-7.21174240e-01 -5.36406577e-01 1.83841377e-01 3.61004621e-01
8.19091618e-01 -1.03775291e-02 -6.93278790e-01 1.00175810e+00
3.95375162e-01 -1.06118858e-01 1.35781482e-01 8.42316985e-01
-6.45736098e-01 -1.25907421e+00 -2.29627024e-02 3.26972663e-01
-8.81800294e-01 -6.16900325e-01 -4.41775322e-01 4.67156976e-01
3.00997823e-01 9.11626041e-01 1.24427356e-01 -1.63754836e-01
3.74036223e-01 4.64741766e-01 -2.27171198e-01 -8.50719213e-01
-7.49412954e-01 -1.85844004e-01 4.65384752e-01 -3.08061223e-02
-3.45981032e-01 -3.28659028e-01 -1.94924259e+00 -1.41164195e-02
-7.03769386e-01 6.82165802e-01 2.84677416e-01 1.06575763e+00
8.47852290e-01 3.82171869e-01 5.61899960e-01 -6.01947010e-02
-1.88154399e-01 -1.01622581e+00 -4.51811790e-01 3.73757213e-01
6.15033992e-02 -5.86546838e-01 5.76466024e-02 9.61874574e-02] | [9.534306526184082, 9.4483060836792] |
0996155a-ff03-4edc-8ab9-0a355feb24f5 | simulating-user-satisfaction-for-the | 2105.03748 | null | https://arxiv.org/abs/2105.03748v1 | https://arxiv.org/pdf/2105.03748v1.pdf | Simulating User Satisfaction for the Evaluation of Task-oriented Dialogue Systems | Evaluation is crucial in the development process of task-oriented dialogue systems. As an evaluation method, user simulation allows us to tackle issues such as scalability and cost-efficiency, making it a viable choice for large-scale automatic evaluation. To help build a human-like user simulator that can measure the quality of a dialogue, we propose the following task: simulating user satisfaction for the evaluation of task-oriented dialogue systems. The purpose of the task is to increase the evaluation power of user simulations and to make the simulation more human-like. To overcome a lack of annotated data, we propose a user satisfaction annotation dataset, USS, that includes 6,800 dialogues sampled from multiple domains, spanning real-world e-commerce dialogues, task-oriented dialogues constructed through Wizard-of-Oz experiments, and movie recommendation dialogues. All user utterances in those dialogues, as well as the dialogues themselves, have been labeled based on a 5-level satisfaction scale. We also share three baseline methods for user satisfaction prediction and action prediction tasks. Experiments conducted on the USS dataset suggest that distributed representations outperform feature-based methods. A model based on hierarchical GRUs achieves the best performance in in-domain user satisfaction prediction, while a BERT-based model has better cross-domain generalization ability. | ['Maarten de Rijke', 'Zhumin Chen', 'Pengjie Ren', 'Zhaochun Ren', 'Krisztian Balog', 'Shuo Zhang', 'Weiwei Sun'] | 2021-05-08 | null | null | null | null | ['movie-recommendation', 'user-simulation'] | ['miscellaneous', 'natural-language-processing'] | [-1.9235784e-01 3.4448105e-01 1.1937593e-01 -8.5154754e-01
-7.2868431e-01 -5.0301355e-01 5.3111231e-01 7.4643729e-04
-4.3380353e-01 6.2935805e-01 5.7515764e-01 -2.2486727e-01
9.7677276e-02 -5.6633246e-01 1.3431673e-01 3.6640245e-02
2.0842475e-01 1.0091921e+00 1.8090160e-01 -1.1531436e+00
2.6652503e-01 -4.9565665e-02 -1.2899814e+00 6.8456805e-01
9.7127026e-01 9.0771645e-01 3.2534635e-01 8.6685008e-01
-7.0506476e-02 7.7478266e-01 -8.7129021e-01 -5.4593980e-01
-1.7453666e-01 -4.7978774e-01 -1.3867432e+00 3.3428696e-01
-3.4826803e-01 -5.5448049e-01 -8.3183810e-02 4.7220948e-01
7.2965443e-01 6.4290154e-01 7.1197164e-01 -1.3054824e+00
-1.7110419e-01 3.8395381e-01 3.5109770e-01 -2.0189497e-01
1.0335673e+00 2.3115091e-01 1.2650013e+00 -6.3769215e-01
6.8132949e-01 1.5624669e+00 3.3249500e-01 8.6580384e-01
-1.1007133e+00 -1.3523485e-01 -2.1607612e-01 -1.2889136e-01
-6.9129848e-01 -6.4953703e-01 4.8908952e-01 -3.4926626e-01
1.1431237e+00 3.8369039e-01 3.6600175e-01 1.1961753e+00
-2.2282983e-01 1.1409882e+00 9.2134374e-01 -4.1195562e-01
3.0786842e-01 7.0026636e-01 3.5088167e-01 6.8484920e-01
-6.6887635e-01 -4.1436836e-01 -4.9084064e-01 -4.5039284e-01
6.0860616e-01 -4.0892163e-01 -3.1377235e-01 -9.5980376e-02
-8.4343183e-01 1.2638103e+00 -4.7705028e-02 4.9856377e-01
-2.9012907e-01 -5.8708268e-01 7.5972414e-01 8.5064000e-01
6.0769492e-01 1.1334153e+00 -1.1230110e+00 -7.6622409e-01
-1.8397565e-01 6.9220018e-01 1.6403995e+00 8.7267780e-01
2.9844284e-01 -2.1125679e-01 -4.9312380e-01 1.5125989e+00
1.9777691e-01 1.1671667e-02 8.3723563e-01 -1.2084217e+00
3.2362342e-01 9.1683590e-01 5.8224326e-01 -6.8938190e-01
-7.2641277e-01 1.7029037e-01 -4.8445764e-01 -2.1258490e-01
5.9504604e-01 -6.7287564e-01 -2.4405605e-01 1.6638331e+00
7.1400791e-02 -3.2536304e-01 4.8193124e-01 8.2551104e-01
1.3189135e+00 7.1558762e-01 2.2557173e-02 -3.9598548e-01
1.4067558e+00 -1.1193846e+00 -7.8099501e-01 7.9894066e-02
1.1242523e+00 -4.8863918e-01 1.4882134e+00 3.8917527e-01
-1.2002534e+00 -5.8731461e-01 -5.5467099e-01 5.8644105e-02
-3.0038756e-01 -9.5216967e-02 5.6761009e-01 5.8204931e-01
-9.9077553e-01 6.2872428e-01 -3.1151786e-01 -4.5145419e-01
-3.4627068e-01 3.3383024e-01 -2.4989940e-01 3.0479601e-01
-1.6424340e+00 1.2647210e+00 4.2010788e-02 -4.4578052e-01
-3.5319236e-01 -2.6915440e-01 -1.0173337e+00 1.7554520e-01
2.4633023e-01 -4.6375921e-01 2.0164108e+00 -8.7855214e-01
-1.9507886e+00 6.7565930e-01 -1.1152618e-03 -2.0369838e-01
3.3231413e-01 2.4667511e-02 -4.0899447e-01 -3.5087252e-01
6.8552978e-02 4.5267755e-01 2.3736589e-01 -1.0611175e+00
-7.4649894e-01 -2.4066317e-01 3.8343391e-01 6.3172203e-01
-4.8856351e-01 2.7068388e-01 -4.7159863e-01 3.0776873e-02
-2.4605249e-01 -8.8258320e-01 -5.3840280e-01 -7.5667906e-01
-1.2410242e-01 -8.2007879e-01 3.3938515e-01 -6.5090245e-01
1.2613916e+00 -1.9206395e+00 3.2914734e-01 -4.6919949e-02
1.2768863e-01 3.5154173e-01 -3.5232416e-01 5.7054061e-01
3.5844699e-01 -2.7444469e-02 2.7197042e-01 -5.0132424e-01
2.2134475e-01 1.3482486e-01 2.0182216e-01 -1.8322818e-01
1.4335583e-01 6.9887036e-01 -1.0405812e+00 -3.8491413e-01
1.2149887e-01 -1.6528739e-01 -7.9741889e-01 1.0491116e+00
-5.1720661e-01 6.6147763e-01 -6.1462438e-01 2.5367031e-01
-8.5118882e-02 -3.5296729e-01 2.9316205e-01 -1.4907999e-01
1.5197678e-01 7.2575271e-01 -8.9559776e-01 1.7560368e+00
-9.3236709e-01 2.4870788e-01 -2.5761036e-02 -6.9786590e-01
1.1964893e+00 6.2140310e-01 4.7623053e-01 -1.0031806e+00
3.0261630e-01 6.8576910e-02 2.8044221e-01 -6.5014356e-01
1.0240859e+00 1.2924676e-01 -5.6865394e-01 6.8068147e-01
3.7050417e-01 -5.6374234e-01 2.9901922e-01 3.9754570e-01
1.0883248e+00 -1.6070384e-01 3.3041504e-01 -8.2805522e-02
5.8365035e-01 6.0295798e-02 3.1075615e-01 6.3545305e-01
-2.8745350e-01 2.4209234e-01 7.1684408e-01 -2.1875167e-01
-7.4333376e-01 -3.5917571e-01 2.1291021e-02 1.7942290e+00
-1.2870087e-01 -7.2196943e-01 -8.8192761e-01 -9.0875745e-01
-2.3516053e-01 1.0228481e+00 -2.7901986e-01 -1.4965276e-01
-2.9714066e-01 -3.2672775e-01 4.5377827e-01 2.5010991e-01
4.9103901e-01 -1.2766781e+00 -4.7673449e-01 5.0131482e-01
-7.7590191e-01 -1.1319935e+00 -2.5791067e-01 8.5866518e-02
-4.7651514e-01 -8.9815176e-01 -5.7899553e-01 -7.3794341e-01
1.4618973e-01 1.1310901e-03 1.4608849e+00 9.7861536e-02
2.3775920e-01 6.1696327e-01 -7.7419192e-01 -2.3851576e-01
-9.7738558e-01 1.0389076e-01 2.6127526e-01 -3.4734866e-01
4.1059989e-01 -1.4208587e-01 -2.5516376e-01 9.2066020e-01
-4.8484272e-01 1.5054013e-01 -5.6174066e-02 1.2869961e+00
-2.9519272e-01 3.6091022e-02 8.9079571e-01 -1.0836748e+00
1.5919995e+00 -5.2007502e-01 -1.5802543e-01 2.3743488e-01
-4.9081451e-01 -7.7905774e-02 6.6895616e-01 -2.8549600e-01
-1.2480633e+00 -2.3841093e-01 -5.0046074e-01 2.9806817e-01
-3.0093995e-01 6.5535808e-01 -6.3576102e-02 3.2013205e-01
8.8517410e-01 -5.2030269e-02 2.4016717e-01 -4.3567237e-01
2.2567551e-01 1.2525071e+00 1.4207202e-01 -6.5767050e-01
1.3710304e-02 -6.5208387e-01 -7.6340628e-01 -1.1146381e+00
-6.5915483e-01 -8.5378027e-01 -5.1153970e-01 -2.8949445e-01
5.4017174e-01 -7.0009083e-01 -9.8949736e-01 3.0209234e-01
-1.2746911e+00 -6.7006212e-01 1.2633694e-02 9.3750663e-02
-8.7057769e-01 2.6772451e-01 -9.3132675e-01 -1.1077749e+00
-3.1647784e-01 -1.3714200e+00 9.7190034e-01 1.4976299e-01
-9.3992269e-01 -1.1758506e+00 1.5205671e-01 8.2023096e-01
4.5775607e-01 -2.7471101e-01 9.4384098e-01 -1.4667091e+00
3.9774963e-01 -1.7494889e-01 -7.1301539e-03 5.1259291e-01
9.0651967e-02 -3.0510950e-01 -8.9848620e-01 -1.8330486e-01
1.3272470e-01 -1.1263697e+00 6.6471189e-02 -3.7330673e-03
7.9167241e-01 -3.6983910e-01 2.0006308e-01 -2.6334342e-01
4.7695917e-01 3.9304417e-01 5.7262379e-01 9.6931927e-02
1.3180491e-01 1.0905789e+00 1.2181033e+00 8.3892047e-01
6.8390322e-01 1.0719643e+00 -1.1080471e-01 -1.9610472e-01
4.9831152e-01 6.9076789e-04 3.3388978e-01 9.6748435e-01
-8.5311532e-02 -5.1216692e-01 -7.6625502e-01 1.0519366e-01
-2.0484521e+00 -6.2171209e-01 -8.3631121e-02 1.9951844e+00
1.1732744e+00 1.2522094e-01 6.2260628e-01 -1.4468825e-02
3.7065867e-01 2.0238588e-02 -2.2939631e-01 -8.2034796e-01
4.1512290e-01 -3.0977001e-02 -1.9570129e-01 5.9218878e-01
-9.0662766e-01 1.2379133e+00 5.8542295e+00 5.2692795e-01
-6.2858737e-01 4.0235359e-02 8.5562778e-01 4.6203607e-01
-1.5952523e-01 -3.9143705e-01 -6.1772817e-01 3.3710599e-01
1.2618493e+00 -2.3587663e-01 2.6363829e-01 1.0454692e+00
5.1291341e-01 -2.2008845e-01 -1.3492724e+00 7.1365666e-01
-1.7577176e-01 -8.0424446e-01 -3.2472888e-01 -2.3196463e-01
4.3192405e-01 -2.2386473e-01 -3.0680239e-01 1.0329537e+00
5.9623843e-01 -9.9780446e-01 3.8332272e-02 2.0688221e-01
7.6456910e-01 -4.6507713e-01 1.1572791e+00 7.9790789e-01
-6.0983914e-01 2.1471059e-01 -2.4175157e-01 -3.2368401e-01
3.2133344e-01 -1.6065942e-01 -1.5467813e+00 1.8582901e-01
3.1305695e-01 3.6326513e-01 -4.0468577e-01 4.4323519e-01
3.4944624e-01 4.9614891e-01 -5.4798286e-02 -6.1121041e-01
2.8372622e-01 -2.8744456e-01 2.3120241e-01 1.4648155e+00
-1.8807103e-01 4.9207866e-01 5.1139998e-01 5.8100492e-01
-4.2927057e-02 4.4822595e-01 -6.9446445e-01 -3.1627804e-01
3.5922074e-01 1.3502929e+00 -2.3448889e-01 -3.1829625e-01
-4.8933095e-01 1.1492606e+00 1.3964853e-01 1.5028767e-01
-4.8642451e-01 -2.8303176e-01 6.0918528e-01 -9.3387878e-03
-4.1716662e-01 -6.7179608e-05 -1.8393432e-01 -1.0206815e+00
-4.7421893e-01 -1.5246783e+00 1.9608730e-01 -4.7243416e-01
-1.1041027e+00 7.7880192e-01 -1.5887716e-01 -1.0057807e+00
-9.2552555e-01 -5.7067770e-01 -4.6535188e-01 9.4060355e-01
-8.5117155e-01 -7.7538592e-01 -3.7346044e-01 6.5764874e-01
1.1617513e+00 -5.7970834e-01 1.4559654e+00 1.4843237e-01
-3.9388844e-01 5.0757158e-01 -1.9874430e-01 9.4729804e-02
8.1156629e-01 -1.3680761e+00 4.4825235e-01 -1.4988227e-01
-2.6564249e-01 5.0164735e-01 9.7342920e-01 -5.1413804e-01
-1.1751332e+00 -6.9211280e-01 1.0469205e+00 -5.3672534e-01
6.4409757e-01 -2.0048694e-01 -9.1816896e-01 3.0267203e-01
1.3125685e-01 -8.0194342e-01 8.0396950e-01 6.6804284e-01
1.9495489e-01 4.4818801e-01 -1.2839967e+00 5.3715187e-01
6.9977337e-01 -4.3942475e-01 -7.2884911e-01 7.3013538e-01
6.8506253e-01 -6.8201983e-01 -1.1284533e+00 4.1613993e-01
4.5082012e-01 -9.4578028e-01 7.4224854e-01 -1.0603603e+00
6.4022499e-01 5.4563564e-01 -2.4101658e-01 -1.7125187e+00
-2.1928436e-01 -6.6122013e-01 1.7439908e-01 1.2504019e+00
6.4073068e-01 -3.6739779e-01 7.0218194e-01 1.3136874e+00
-1.5503125e-01 -7.6723677e-01 -4.6754283e-01 -3.3683068e-01
-1.0129051e-01 -3.2610109e-01 3.9113590e-01 9.2970997e-01
7.6705503e-01 1.1426305e+00 -4.8784128e-01 -4.4466051e-01
-1.8678334e-01 -2.3858353e-01 1.1806594e+00 -1.4692183e+00
-4.1955033e-01 -4.7999206e-01 -2.8245425e-02 -1.3860319e+00
3.3200869e-01 -4.6524253e-01 3.1077421e-01 -1.4400691e+00
-8.8857941e-02 -5.0191408e-01 1.0263237e-01 3.0172649e-01
-5.4499507e-02 -3.3115456e-01 -8.2411632e-02 -9.8885000e-02
-1.0409225e+00 6.2914723e-01 1.3510770e+00 6.8101346e-02
-6.9120628e-01 5.9826314e-01 -5.3861988e-01 8.1044191e-01
7.4277937e-01 1.2530825e-01 -5.8956850e-01 6.5969720e-02
-1.7884266e-01 1.0022478e+00 -2.6130664e-01 -4.9957246e-01
1.1400260e-01 -3.3187991e-01 -1.5952489e-01 -1.4067301e-01
6.8371171e-01 -5.5113375e-01 -5.5375355e-01 1.2270676e-01
-8.5716563e-01 -8.3432883e-02 -1.0970766e-01 4.5583099e-02
-3.0132779e-01 -5.4039693e-01 7.1849859e-01 -2.7644050e-01
-7.0789528e-01 -2.0902514e-02 -6.9433784e-01 2.3871490e-01
6.5964293e-01 1.3719800e-01 -1.7804907e-01 -1.1363498e+00
-6.8131125e-01 5.5994302e-01 2.0812313e-01 6.3235050e-01
5.5263740e-01 -8.9775115e-01 -6.7077416e-01 8.3552778e-02
4.0622079e-01 -2.7183065e-01 7.4733995e-02 3.8915291e-01
-1.1111558e-01 5.4590321e-01 -2.0854777e-01 -4.4537723e-01
-1.5138083e+00 -2.6376233e-01 3.3194283e-01 -5.0946146e-01
-1.9309431e-01 9.3371785e-01 2.1566017e-02 -1.1756139e+00
5.4918361e-01 4.7703158e-02 -8.3471197e-01 1.6394022e-01
4.3242490e-01 4.2053652e-01 1.1372304e-01 -5.9582990e-01
5.9792917e-02 -3.9939883e-01 -8.0703810e-02 -4.0806684e-01
1.3109529e+00 -1.2770171e-01 1.7763530e-01 7.3131931e-01
1.0343581e+00 -3.7080592e-01 -6.9944793e-01 -3.6967111e-01
1.5726174e-01 -4.7156522e-01 -9.7048022e-02 -1.1832354e+00
-4.4107816e-01 5.5398029e-01 2.7293086e-01 7.7647495e-01
7.0139217e-01 -1.0544750e-01 8.4787840e-01 9.7491270e-01
5.9076029e-01 -1.4978938e+00 5.5325973e-01 1.0982466e+00
1.1099167e+00 -1.7022253e+00 -5.3209555e-01 -1.8238954e-01
-1.5816256e+00 1.0244995e+00 1.1077296e+00 3.9078170e-01
2.7266800e-01 -7.3406979e-02 2.2409274e-01 -1.9494152e-01
-1.1435727e+00 -1.6335174e-01 2.7888757e-01 4.1288731e-01
1.0550164e+00 1.7515354e-01 -4.5274502e-01 1.1068050e+00
-3.5850662e-01 -1.3896922e-02 5.1439238e-01 7.2049367e-01
-5.5812055e-01 -1.2829236e+00 1.4850403e-01 7.2782642e-01
-8.9924753e-02 5.6349218e-02 -6.8455511e-01 5.1964998e-01
-5.6453675e-01 1.5727571e+00 -2.7905232e-01 -8.0703408e-01
1.0069877e+00 3.1640995e-01 1.2706362e-02 -1.0062561e+00
-1.0318495e+00 -1.2970372e-01 1.1804599e+00 -5.7107389e-01
-1.4078501e-01 -3.7002537e-01 -1.1137034e+00 -4.2124307e-01
-6.3973480e-01 6.8423933e-01 6.6069818e-01 1.0162983e+00
3.1787645e-02 4.5359266e-01 1.1240509e+00 -7.2241968e-01
-1.0351875e+00 -1.4868357e+00 -7.3322982e-01 9.3324256e-01
-2.0517169e-01 -6.2843239e-01 -1.6033769e-01 -3.3530754e-01] | [12.87171459197998, 8.05517864227295] |
b6fd360d-f762-40d1-8595-10b2f75e91f1 | deep-sky-modeling-for-single-image-outdoor | 1905.03897 | null | https://arxiv.org/abs/1905.03897v1 | https://arxiv.org/pdf/1905.03897v1.pdf | Deep Sky Modeling for Single Image Outdoor Lighting Estimation | We propose a data-driven learned sky model, which we use for outdoor lighting estimation from a single image. As no large-scale dataset of images and their corresponding ground truth illumination is readily available, we use complementary datasets to train our approach, combining the vast diversity of illumination conditions of SUN360 with the radiometrically calibrated and physically accurate Laval HDR sky database. Our key contribution is to provide a holistic view of both lighting modeling and estimation, solving both problems end-to-end. From a test image, our method can directly estimate an HDR environment map of the lighting without relying on analytical lighting models. We demonstrate the versatility and expressivity of our learned sky model and show that it can be used to recover plausible illumination, leading to visually pleasant virtual object insertions. To further evaluate our method, we capture a dataset of HDR 360{\deg} panoramas and show through extensive validation that we significantly outperform previous state-of-the-art. | ['Jean-François Lalonde', 'Yannick Hold-Geoffroy', 'Akshaya Athawale'] | 2019-05-10 | deep-sky-modeling-for-single-image-outdoor-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Hold-Geoffroy_Deep_Sky_Modeling_for_Single_Image_Outdoor_Lighting_Estimation_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Hold-Geoffroy_Deep_Sky_Modeling_for_Single_Image_Outdoor_Lighting_Estimation_CVPR_2019_paper.pdf | cvpr-2019-6 | ['lighting-estimation'] | ['computer-vision'] | [ 2.43298888e-01 -2.98052192e-01 3.25801879e-01 -5.31280816e-01
-1.01334310e+00 -1.05480146e+00 4.77887064e-01 -5.45985758e-01
8.84236544e-02 6.50460124e-01 1.66353181e-01 -2.24114463e-01
2.55368471e-01 -7.83072531e-01 -9.75324392e-01 -5.92225492e-01
2.14598089e-01 2.89293945e-01 9.21634883e-02 -2.51908869e-01
-1.16336323e-01 6.30374849e-01 -1.89214861e+00 -1.19008757e-01
8.17233145e-01 1.09374988e+00 4.14017290e-01 1.09975684e+00
1.91943705e-01 9.18923259e-01 -4.02753025e-01 -1.88814938e-01
7.75648177e-01 -4.04326469e-01 -4.32968676e-01 4.97390300e-01
1.05661738e+00 -7.47618794e-01 -4.71830130e-01 7.14714646e-01
5.58130145e-01 1.10010825e-01 4.41486925e-01 -9.97220993e-01
-3.14260602e-01 -3.74668211e-01 -5.00800312e-01 -3.03600162e-01
4.34046835e-01 3.96006078e-01 9.49523330e-01 -7.45581985e-01
7.38432169e-01 8.03731382e-01 6.86120450e-01 5.32072149e-02
-1.30954146e+00 -4.11991149e-01 1.11215897e-01 2.49964055e-02
-1.42009699e+00 -5.88004351e-01 8.65284562e-01 -1.79646343e-01
5.52058637e-01 4.35557395e-01 9.20992732e-01 9.69959259e-01
-2.40458965e-01 4.12160575e-01 1.73180556e+00 -4.93338466e-01
2.94755280e-01 1.43073440e-01 -4.10215646e-01 5.91016769e-01
-1.30771175e-01 1.91583857e-01 -6.87375188e-01 7.56979361e-02
7.00830638e-01 -1.21314906e-01 -5.11015654e-01 -6.68292761e-01
-1.00323272e+00 5.04714787e-01 7.06522048e-01 -5.77920616e-01
-2.96703428e-01 3.02846789e-01 -3.50815088e-01 7.79013336e-02
4.08801109e-01 1.79044917e-01 -4.43018198e-01 2.78411731e-02
-9.53755796e-01 3.23746175e-01 7.06810772e-01 9.34237957e-01
1.05316341e+00 5.43081984e-02 3.27531427e-01 6.80814683e-01
3.61849368e-01 1.28563452e+00 -2.88931072e-01 -1.51017284e+00
9.16778073e-02 2.13612065e-01 3.90366167e-01 -7.62838423e-01
-2.77064204e-01 -3.32240969e-01 -2.91943401e-01 5.77486932e-01
3.39294970e-01 8.24763328e-02 -9.83476818e-01 1.60565674e+00
6.29542172e-01 2.14138776e-01 5.09254355e-03 9.17663097e-01
5.96681833e-01 6.53964520e-01 -5.33371925e-01 -3.42412945e-03
1.02220774e+00 -8.87735903e-01 -4.02432859e-01 -4.83778000e-01
1.57355189e-01 -9.32381392e-01 1.43553150e+00 7.49586105e-01
-9.31008279e-01 -2.64040768e-01 -1.11651754e+00 -4.68947321e-01
-2.54524618e-01 5.53897470e-02 7.78192878e-01 7.68555403e-01
-1.26442063e+00 3.01999271e-01 -5.64712763e-01 -5.12707472e-01
1.53337151e-01 -2.39672542e-01 -1.31856397e-01 -3.67064059e-01
-7.78802276e-01 8.93079400e-01 -8.79995897e-02 1.09308369e-01
-1.39800107e+00 -8.82435024e-01 -7.94570625e-01 -2.26967856e-01
2.32023865e-01 -6.58358455e-01 1.24372935e+00 -9.76993561e-01
-1.67152166e+00 9.71228361e-01 -1.43685877e-01 -2.43831113e-01
7.04249799e-01 -2.77817726e-01 -3.05314660e-01 2.51494408e-01
-1.56342059e-01 5.90672910e-01 6.48969114e-01 -1.96445012e+00
-3.09770346e-01 -2.09051967e-01 1.34500697e-01 5.52961767e-01
-1.07347257e-01 -2.53499836e-01 -8.19495976e-01 -2.56028086e-01
3.31530683e-02 -1.07222378e+00 -9.12831724e-02 4.58259314e-01
-3.96468818e-01 9.56081450e-01 6.52012706e-01 -9.19239342e-01
7.67913043e-01 -1.91651976e+00 -1.73847765e-01 3.53634268e-01
-5.68988994e-02 -2.50636250e-01 3.36888172e-02 3.56638908e-01
2.34860763e-01 -4.29438740e-01 -3.75547856e-01 -5.79147518e-01
8.27182755e-02 3.68523747e-01 -6.91054106e-01 8.58994007e-01
-1.90825537e-01 6.52310014e-01 -6.47045910e-01 -4.40057218e-01
8.25655520e-01 9.59893167e-01 -3.59750360e-01 5.78765392e-01
-4.55486804e-01 6.58876359e-01 6.73991218e-02 7.65861750e-01
1.00166118e+00 -2.71948036e-02 2.44906470e-01 -2.81369507e-01
-1.43902719e-01 2.52918571e-01 -1.29057467e+00 2.05311823e+00
-1.00260472e+00 9.30192232e-01 1.94926053e-01 -1.00002326e-01
1.00499821e+00 -2.06125602e-01 2.49227256e-01 -1.02118957e+00
2.75942795e-02 1.48649886e-01 -8.08935225e-01 -4.04342145e-01
5.66759229e-01 -1.31808326e-01 1.40237793e-01 4.03444856e-01
-2.70102620e-01 -8.78540635e-01 -3.59813839e-01 1.81079328e-01
8.18825126e-01 7.11184382e-01 2.28499882e-02 -2.50381857e-01
2.11093381e-01 1.06683336e-02 3.40877533e-01 5.87723255e-01
6.57901838e-02 1.23498821e+00 -1.06801875e-01 -4.10974741e-01
-1.32921052e+00 -1.42121983e+00 -2.69405752e-01 9.43210125e-01
3.01273286e-01 -1.25732422e-01 -7.49426603e-01 -2.84944534e-01
-4.73814048e-02 8.94059598e-01 -7.32120574e-01 3.98673236e-01
-2.61118501e-01 -7.08144903e-01 2.62116998e-01 2.72578984e-01
5.35763144e-01 -5.28829098e-01 -7.62638867e-01 -3.69429410e-01
-3.21220577e-01 -1.32328677e+00 -5.29882535e-02 1.28336800e-02
-4.72620547e-01 -1.02284908e+00 -4.32622105e-01 -2.20696360e-01
4.14987862e-01 7.03907192e-01 1.71723104e+00 -2.32220031e-02
-6.27459109e-01 7.64710069e-01 -2.03703150e-01 -4.87888634e-01
-3.36314321e-01 -4.90604311e-01 -1.68215409e-01 -4.88717370e-02
-1.95142403e-01 -7.60595262e-01 -1.08705854e+00 4.96035218e-01
-8.77708733e-01 4.31388259e-01 1.91963613e-01 6.33859411e-02
8.23366702e-01 -2.09203988e-01 -4.10881527e-02 -6.23844743e-01
-2.70736396e-01 -2.89326966e-01 -1.07477725e+00 2.04745948e-01
-4.83142167e-01 -2.45650336e-01 4.51273769e-01 1.62721172e-01
-1.50982380e+00 3.50826085e-01 2.33430392e-03 -4.86096025e-01
-2.09068522e-01 -2.39633903e-01 -4.35396641e-01 -4.50150132e-01
7.89837420e-01 2.02233270e-01 -3.80045354e-01 -3.75545263e-01
9.25939620e-01 6.00346804e-01 1.07650983e+00 -6.84050441e-01
1.11588025e+00 1.24584782e+00 2.41664276e-01 -1.03997898e+00
-1.07142985e+00 -3.94853860e-01 -4.34282809e-01 -4.86784369e-01
7.07430482e-01 -1.37528384e+00 -5.18132806e-01 4.56900716e-01
-7.16227889e-01 -8.96370411e-01 -2.48342633e-01 1.30164012e-01
-8.79312873e-01 3.20045233e-01 -2.22231567e-01 -1.02578616e+00
-1.69640496e-01 -8.70100498e-01 1.58744419e+00 1.58153802e-01
2.14111373e-01 -1.04427171e+00 4.81095046e-01 5.40883541e-01
3.43975157e-01 6.79344714e-01 3.28341067e-01 5.92131436e-01
-1.02848959e+00 5.91720976e-02 -3.60492170e-01 3.33521485e-01
5.81561886e-02 3.31391722e-01 -1.70067835e+00 -1.86898515e-01
-5.22174835e-02 -6.23956382e-01 8.29040766e-01 3.39286923e-01
1.17608702e+00 9.44253504e-02 2.40904853e-01 1.26985800e+00
1.98231328e+00 -5.24008989e-01 9.11487401e-01 6.03265822e-01
9.82235432e-01 5.92670500e-01 6.73615873e-01 5.23700774e-01
8.29980016e-01 8.20620179e-01 8.41825008e-01 -5.48614681e-01
-5.06586432e-01 -3.30711097e-01 1.87156111e-01 2.48043671e-01
-1.09762117e-01 -3.05451959e-01 -8.27442646e-01 5.27887702e-01
-1.43127775e+00 -6.65205538e-01 -1.70348227e-01 2.69448090e+00
6.78270102e-01 -3.16291988e-01 -2.67222431e-02 -1.28230020e-01
2.53727525e-01 4.20897871e-01 -6.03225350e-01 -1.54086858e-01
-5.87026000e-01 1.90271065e-02 9.36438441e-01 6.33835196e-01
-8.70598614e-01 8.34633350e-01 6.95432997e+00 3.56098413e-01
-1.10509062e+00 -1.64659899e-02 6.85509920e-01 -3.78431082e-01
-7.65381098e-01 1.40096173e-01 -4.56806272e-01 1.46510199e-01
8.72752845e-01 3.73813719e-01 1.00949717e+00 8.66764784e-01
3.70455742e-01 -3.63046497e-01 -8.58235598e-01 1.16503048e+00
3.80470306e-01 -1.13946462e+00 -4.50854748e-01 1.58952922e-01
1.19021130e+00 4.05318975e-01 9.19556692e-02 -1.87726825e-01
6.87565923e-01 -9.87390876e-01 9.58998024e-01 5.53460836e-01
9.50911403e-01 -5.54230928e-01 7.12847114e-02 1.44250179e-02
-1.15940893e+00 1.71614110e-01 -3.44274282e-01 7.16205239e-02
3.46528977e-01 6.20344877e-01 -7.27510989e-01 7.12616265e-01
8.68933260e-01 5.47223330e-01 -7.92458475e-01 8.80878270e-01
-5.83200276e-01 3.89248669e-01 -6.99699342e-01 5.84194005e-01
-1.68863788e-01 -4.06087488e-01 1.97040096e-01 1.09480190e+00
2.32239336e-01 -1.01385482e-01 3.42425816e-02 7.77429938e-01
-1.88511416e-01 -6.21909723e-02 -6.20136321e-01 5.61873913e-01
2.78396517e-01 1.58096254e+00 -6.54495478e-01 -1.05118945e-01
-2.37784028e-01 1.18151677e+00 2.92167157e-01 5.12713790e-01
-9.90236819e-01 8.53098109e-02 7.93180943e-01 4.35866565e-02
3.08977604e-01 -3.07686210e-01 -5.06502509e-01 -1.29409564e+00
1.20044880e-01 -8.79839957e-01 -4.77555282e-02 -1.47394669e+00
-1.02958262e+00 3.97835076e-01 -1.50947154e-01 -1.28852141e+00
-6.71053231e-02 -6.07700467e-01 -6.22356057e-01 8.17429662e-01
-1.91087437e+00 -1.56734073e+00 -1.12156689e+00 5.47351480e-01
3.18177462e-01 5.72229743e-01 8.79607677e-01 1.44001544e-01
-2.33406410e-01 1.48031384e-01 6.67576969e-01 -3.56239319e-01
8.67819965e-01 -1.61685884e+00 4.70219612e-01 1.13403702e+00
3.12251091e-01 1.99393570e-01 1.17218876e+00 -2.70716071e-01
-1.92027652e+00 -1.16401088e+00 3.48214209e-02 -9.00533378e-01
4.71496046e-01 -6.98677599e-01 -6.06759012e-01 6.58288121e-01
1.17144048e-01 3.18603009e-01 4.70553517e-01 -9.17634070e-02
-6.83691740e-01 -4.21323329e-01 -1.12626636e+00 5.99051714e-01
1.06172717e+00 -6.80827022e-01 -8.39745030e-02 5.64500809e-01
6.00941658e-01 -7.76268303e-01 -7.72513330e-01 3.61604989e-01
8.86153579e-01 -1.53278625e+00 1.22965026e+00 1.46691144e-01
4.76023674e-01 -6.60892963e-01 -7.41616726e-01 -1.28578174e+00
2.71397084e-01 -5.66361666e-01 -1.70532420e-01 1.25839412e+00
2.49472350e-01 -4.29213345e-01 6.45510793e-01 7.31927335e-01
-7.10961595e-03 -3.63998950e-01 -4.44784552e-01 -7.11235523e-01
-1.81084856e-01 -7.84174681e-01 7.26395428e-01 7.56676912e-01
-8.19631517e-01 -1.39395863e-01 -6.68303967e-01 6.33995414e-01
1.10137069e+00 7.35893548e-01 1.46819746e+00 -8.60160649e-01
-7.45794833e-01 1.16493471e-01 -1.45949170e-01 -9.16685164e-01
-7.83050880e-02 -2.83682644e-01 4.06068206e-01 -1.72563946e+00
2.32610211e-01 -5.66280067e-01 2.12710593e-02 1.45475775e-01
-6.23495206e-02 7.94401228e-01 2.47891396e-02 8.01308453e-02
-6.18739188e-01 6.42818987e-01 8.72218013e-01 2.93865502e-01
-1.27328530e-01 -5.31020582e-01 -4.56517220e-01 7.89066613e-01
5.90906858e-01 -6.12643585e-02 -5.30999422e-01 -7.54763246e-01
5.36518157e-01 -2.88365424e-01 5.42408764e-01 -1.03427958e+00
-3.20729613e-01 -3.11475843e-01 7.34488964e-01 -5.42712033e-01
7.24652886e-01 -8.31682682e-01 4.87514347e-01 -1.74218982e-01
-5.12026511e-02 -3.61204386e-01 1.52112335e-01 5.40917218e-01
3.42844933e-01 2.84359813e-01 8.32927227e-01 3.30017693e-02
-8.61875296e-01 2.33592391e-01 3.11407626e-01 8.05502683e-02
8.85317326e-01 -4.41027544e-02 -6.09897554e-01 -6.87679768e-01
3.97449499e-03 1.52302772e-01 1.44634795e+00 2.54677534e-01
4.15087432e-01 -1.12743866e+00 -6.13287508e-01 1.41019538e-01
5.30867934e-01 2.76108950e-01 4.86352533e-01 4.10403103e-01
-1.09896302e+00 -2.42227137e-01 4.51867022e-02 -7.21918344e-01
-1.14671147e+00 4.57067877e-01 5.16561449e-01 2.96217561e-01
-7.96838522e-01 5.14164150e-01 5.39660394e-01 -7.16653287e-01
1.05918080e-01 -7.66529217e-02 4.73773956e-01 -4.72134620e-01
6.14045978e-01 2.70554155e-01 2.54269302e-01 -7.58819580e-01
-2.05445170e-01 8.56902778e-01 6.57514632e-01 -3.80455524e-01
1.44913661e+00 -6.93603814e-01 1.54928921e-03 4.79720026e-01
1.14915144e+00 3.91689271e-01 -1.98338842e+00 -1.17018968e-01
-8.67714167e-01 -1.07290602e+00 4.34845388e-01 -9.39689040e-01
-1.02478075e+00 7.99081564e-01 9.05658364e-01 -1.19361646e-01
1.45845914e+00 -5.07183112e-02 6.06759548e-01 4.62920278e-01
5.29354036e-01 -1.15872550e+00 -1.66465685e-01 1.19382903e-01
7.25249529e-01 -1.44422424e+00 2.74691761e-01 -4.49013829e-01
-6.84557319e-01 1.03764200e+00 2.64451206e-01 -3.90949696e-02
2.25426450e-01 3.49221140e-01 6.11697674e-01 -1.03746198e-01
-3.91917795e-01 -3.10061812e-01 1.22162610e-01 6.82838976e-01
-9.49478801e-03 6.95765913e-02 6.90154016e-01 -2.32628673e-01
-5.68091035e-01 -1.33424357e-01 7.01673508e-01 8.26911211e-01
-4.90378410e-01 -6.18771315e-01 -7.25795031e-01 -7.24990070e-02
-5.17333597e-02 -1.05668209e-01 -2.06371248e-01 5.03334939e-01
-5.41288294e-02 1.01342249e+00 -1.35978892e-01 -1.14192761e-01
1.57470912e-01 -3.49656969e-01 7.93013334e-01 -1.18618034e-01
5.39406091e-02 9.60419513e-03 2.09701881e-01 -9.14493263e-01
-3.63504320e-01 -6.17592037e-01 -7.57695794e-01 -3.50075334e-01
-1.77247435e-01 -4.61857766e-01 1.29514194e+00 6.56924009e-01
2.12955669e-01 2.74028659e-01 1.13995385e+00 -1.37846649e+00
-9.76382047e-02 -4.95724916e-01 -6.98040783e-01 5.35475791e-01
6.00314677e-01 -4.81586188e-01 -5.50550401e-01 2.46492952e-01] | [9.772184371948242, -2.9588284492492676] |
1a1f79da-0ca1-4571-a48b-d7aab58a5fcc | rethinking-backdoor-data-poisoning-attacks-in | 2212.02582 | null | https://arxiv.org/abs/2212.02582v1 | https://arxiv.org/pdf/2212.02582v1.pdf | Rethinking Backdoor Data Poisoning Attacks in the Context of Semi-Supervised Learning | Semi-supervised learning methods can train high-accuracy machine learning models with a fraction of the labeled training samples required for traditional supervised learning. Such methods do not typically involve close review of the unlabeled training samples, making them tempting targets for data poisoning attacks. In this paper we investigate the vulnerabilities of semi-supervised learning methods to backdoor data poisoning attacks on the unlabeled samples. We show that simple poisoning attacks that influence the distribution of the poisoned samples' predicted labels are highly effective - achieving an average attack success rate as high as 96.9%. We introduce a generalized attack framework targeting semi-supervised learning methods to better understand and exploit their limitations and to motivate future defense strategies. | ['Vincent Emanuele', 'Marissa Connor'] | 2022-12-05 | null | null | null | null | ['data-poisoning'] | ['adversarial'] | [ 8.06545168e-02 1.09310828e-01 -7.37322271e-01 -2.90289789e-01
-8.51047099e-01 -1.19717538e+00 5.65813005e-01 4.78514373e-01
-5.67461193e-01 9.04066384e-01 -2.87574381e-01 -5.81974030e-01
3.20961505e-01 -8.90332162e-01 -7.24319160e-01 -8.48919749e-01
-1.92790464e-01 5.98415017e-01 1.57918811e-01 1.07294559e-01
1.54485643e-01 8.24534059e-01 -6.03862584e-01 4.26298946e-01
3.63708615e-01 4.98457432e-01 -9.10793304e-01 4.47416544e-01
1.86085403e-01 1.08590710e+00 -1.07419240e+00 -5.63670576e-01
5.89119673e-01 -2.62445748e-01 -7.00731635e-01 -1.56509392e-02
2.30539829e-01 -7.14314520e-01 -6.11903429e-01 1.25116885e+00
2.90194333e-01 -4.11831915e-01 8.61429334e-01 -1.64203227e+00
-3.34354490e-01 9.55286682e-01 -7.56803155e-01 2.69937426e-01
1.00979924e-01 4.50601578e-01 6.68131828e-01 -4.27285016e-01
1.33954555e-01 9.07189071e-01 4.95871663e-01 1.09454274e+00
-1.36115491e+00 -1.33434558e+00 -2.54371971e-01 -9.65707526e-02
-1.33164132e+00 -4.79516655e-01 7.83803821e-01 -1.32242754e-01
3.94273043e-01 1.88691840e-01 3.66437621e-02 1.50000513e+00
-3.77925746e-02 7.28202105e-01 1.38864076e+00 -2.51507074e-01
5.83259404e-01 6.52688026e-01 5.79973757e-01 6.02016628e-01
8.95754695e-01 5.99004626e-01 -4.86086905e-01 -1.09345388e+00
2.78930157e-01 3.10384631e-01 -4.13577743e-02 -2.86060214e-01
-4.20931727e-01 1.34337664e+00 2.81237423e-01 -7.60178566e-02
-4.04365174e-02 1.40236557e-01 5.03306985e-01 2.59220600e-01
3.94502461e-01 6.27003014e-01 -6.62344158e-01 3.92313987e-01
-5.81784725e-01 -5.63625619e-03 9.26756918e-01 6.44871652e-01
7.50450015e-01 2.46754214e-01 1.85035616e-01 1.94475845e-01
2.48592272e-01 8.48850429e-01 2.68197268e-01 -4.74601567e-01
4.19326276e-01 3.32402647e-01 3.73860478e-01 -5.31165242e-01
-1.23885654e-01 -1.64039254e-01 -3.90197188e-01 2.19826981e-01
8.21532965e-01 -7.32049406e-01 -5.75706124e-01 1.48567629e+00
4.62167412e-01 -2.79572338e-01 1.42985523e-01 3.13155860e-01
8.31352100e-02 4.73149091e-01 4.88382906e-01 -4.52747494e-01
8.65722418e-01 -6.46562219e-01 -4.44428533e-01 1.97020639e-02
1.20734680e+00 -9.75394472e-02 9.74030554e-01 6.73478842e-01
-7.09104717e-01 1.54128253e-01 -1.13844371e+00 6.97652936e-01
-6.63593769e-01 -3.83551896e-01 8.48168015e-01 1.57425451e+00
-1.68830186e-01 7.48795569e-01 -9.27803218e-01 -1.83286909e-02
1.17838216e+00 7.60709167e-01 -2.86615372e-01 -3.71933505e-02
-1.06322396e+00 6.22943163e-01 3.51051271e-01 -5.26682734e-01
-1.62182200e+00 -6.75656855e-01 -4.94067639e-01 -6.94283843e-02
3.15555722e-01 2.56605685e-01 1.28989112e+00 -2.23780602e-01
-9.66855586e-01 6.51940525e-01 2.39652067e-01 -9.31801796e-01
4.13183212e-01 -2.53662854e-01 -4.09381777e-01 5.64252138e-01
-1.21627092e-01 2.79893994e-01 1.10203254e+00 -1.20083880e+00
-4.03068423e-01 -6.55494928e-01 -1.71533912e-01 -3.34599495e-01
-1.16096663e+00 3.24836403e-01 4.97404128e-01 -4.84625727e-01
-5.03908098e-01 -8.45945835e-01 -5.12998641e-01 -1.79727376e-01
-9.83417332e-01 1.57510098e-02 1.41589379e+00 -2.86776349e-02
8.69794846e-01 -2.04162741e+00 -7.07066357e-01 4.38648701e-01
3.29213858e-01 8.03190589e-01 8.01261440e-02 6.67833984e-01
-1.96653381e-01 4.06771988e-01 -1.47200927e-01 -1.82870269e-01
-1.70513615e-01 -1.15392748e-02 -1.13384044e+00 1.01227319e+00
-2.12662946e-02 7.66895652e-01 -8.68372858e-01 -2.28421837e-01
4.53563146e-02 2.13903829e-01 -2.48032987e-01 3.54996443e-01
-3.74340653e-01 4.34103459e-01 -7.91565061e-01 7.68369615e-01
5.00277162e-01 -9.62720662e-02 1.97956860e-02 9.59504470e-02
5.77954054e-01 3.52974474e-01 -5.23878634e-01 6.79378271e-01
1.96931332e-01 2.01858327e-01 -3.14720541e-01 -6.55976951e-01
9.11443472e-01 3.09264809e-01 4.02452111e-01 -1.55522466e-01
1.60668164e-01 -2.09434237e-02 -1.47498041e-01 -9.74500775e-02
-3.76951814e-01 -3.72207254e-01 -2.52151877e-01 1.15974104e+00
-9.28314403e-02 2.65241042e-02 -2.42766902e-01 5.33579588e-01
1.12843287e+00 -5.13394594e-01 1.90507844e-01 -1.79643691e-01
1.94261566e-01 4.99248356e-01 3.16915214e-01 1.02444696e+00
-3.69264960e-01 -8.73277932e-02 6.19877994e-01 -6.96075559e-01
-1.02830696e+00 -1.12543249e+00 -1.43705353e-01 7.59979606e-01
-9.05716866e-02 -6.24839723e-01 -1.14326096e+00 -1.75915802e+00
1.73199639e-01 7.82053471e-01 -6.24340534e-01 -6.38654888e-01
-3.14942151e-02 -1.02163672e+00 1.36735904e+00 4.48691964e-01
5.14491796e-01 -8.03035319e-01 -2.78454512e-01 -1.76970780e-01
4.74954516e-01 -9.73295271e-01 -2.70887911e-01 7.53376007e-01
-9.72592354e-01 -1.57774663e+00 -7.17349211e-03 -2.65108734e-01
1.08772206e+00 2.07796216e-01 4.23100591e-01 1.91650301e-01
-2.70710170e-01 1.64445132e-01 -3.35527956e-01 -8.13677430e-01
-7.22021282e-01 7.55128043e-04 6.51906490e-01 3.13730799e-02
9.22103524e-01 -5.33810496e-01 2.97540391e-04 3.44463706e-01
-1.01587653e+00 -8.37009251e-01 2.29366988e-01 6.71273291e-01
2.93650299e-01 4.18425739e-01 8.46058309e-01 -1.67475462e+00
4.47907895e-01 -5.25742352e-01 -6.99544311e-01 1.54693186e-01
-5.74086666e-01 1.10514812e-01 1.20776975e+00 -1.11397374e+00
-5.70146322e-01 4.01415914e-01 1.05271474e-01 -5.48977852e-01
-4.60706204e-01 -1.01828039e-01 -2.84215242e-01 -5.01249254e-01
1.37105465e+00 1.27112672e-01 -1.82376444e-01 -4.93612707e-01
2.64801919e-01 8.19335818e-01 1.54383242e-01 -5.34666836e-01
1.64148498e+00 6.27163887e-01 8.97028223e-02 -7.76622295e-01
-1.24117756e+00 -1.78168118e-01 -4.39259052e-01 2.59773761e-01
3.92064929e-01 -6.51638329e-01 -6.12614453e-01 5.54347515e-01
-7.99663424e-01 -5.95318198e-01 -3.52791518e-01 3.47975820e-01
-2.27998808e-01 5.82849383e-01 -9.42650437e-01 -1.02807963e+00
-5.02451003e-01 -7.86171377e-01 4.72413868e-01 1.09174522e-02
-2.73671091e-01 -1.15093231e+00 3.44582722e-02 4.16263998e-01
1.05855614e-01 3.76593798e-01 9.03732419e-01 -1.89656055e+00
-9.99609157e-02 -8.20057929e-01 2.38232449e-01 3.79949063e-01
2.93958694e-01 -2.73538649e-01 -1.24168861e+00 -6.12480998e-01
4.85474288e-01 -1.06462109e+00 6.67987943e-01 -2.00301215e-01
1.30978286e+00 -6.82362795e-01 -4.67490107e-01 3.84810179e-01
1.13853371e+00 2.01922178e-01 4.90890503e-01 -4.56029065e-02
7.37935483e-01 5.73960364e-01 5.75573087e-01 6.38604939e-01
-5.43179214e-01 7.42404312e-02 4.48704392e-01 8.26680958e-02
5.97365320e-01 -8.08029413e-01 5.44545412e-01 -2.29645818e-01
8.04973006e-01 8.68921354e-02 -7.58643329e-01 4.21862490e-02
-1.22896004e+00 -9.55914259e-01 -1.69585750e-01 2.46855330e+00
1.22670388e+00 4.73042488e-01 6.31350875e-01 4.72119004e-01
6.66900098e-01 9.61702410e-03 -8.16219687e-01 -2.75026381e-01
1.03072412e-01 5.38982451e-01 1.12024713e+00 4.65755105e-01
-1.51116455e+00 1.22742283e+00 7.30348396e+00 1.02492714e+00
-7.95042217e-01 1.71333283e-01 9.29541290e-01 -3.00316334e-01
8.41501057e-02 9.99433845e-02 -1.17812622e+00 3.27932090e-01
1.35260630e+00 2.62899902e-02 3.56491625e-01 1.10113597e+00
-1.05193540e-01 1.17823817e-01 -1.21691608e+00 6.94573998e-01
-7.56300092e-02 -1.13504815e+00 1.96949884e-01 4.97290850e-01
5.47522664e-01 -1.13333419e-01 2.91588902e-01 1.46165401e-01
9.63801861e-01 -1.16725099e+00 1.73145961e-02 -3.47225219e-01
7.51884520e-01 -1.25630236e+00 2.82781869e-01 8.21730554e-01
-3.62185746e-01 -5.19180059e-01 -5.28289258e-01 -4.58666217e-03
-3.00339431e-01 5.30507147e-01 -8.48598480e-01 -2.05736607e-01
3.77497137e-01 3.73324990e-01 -7.11444318e-01 6.19894207e-01
-6.07099712e-01 1.45226586e+00 -5.71045935e-01 -7.72375837e-02
2.93636233e-01 2.32126966e-01 2.36482218e-01 9.36336398e-01
-4.95810211e-01 2.20513895e-01 3.58361930e-01 8.03572297e-01
-2.15260416e-01 -1.44087210e-01 -1.04043007e+00 -6.27702415e-01
6.59564734e-01 9.04579222e-01 -6.05924189e-01 -2.67963320e-01
-5.61200008e-02 6.31694555e-01 3.01079571e-01 2.46915281e-01
-7.04858243e-01 -2.99279004e-01 4.70865697e-01 9.36018825e-02
-1.65198565e-01 -1.63585007e-01 -5.38592100e-01 -1.11809838e+00
-6.43660367e-01 -1.19923067e+00 6.10098183e-01 -3.27697545e-02
-1.56462371e+00 3.27462703e-01 3.92553657e-02 -9.59736228e-01
-1.13864765e-01 -6.13252640e-01 -6.78762615e-01 4.65138614e-01
-1.05262721e+00 -9.76523638e-01 5.23317158e-01 9.55736578e-01
5.05747609e-02 -6.22911215e-01 1.08485925e+00 -2.79556900e-01
-8.73399496e-01 9.71382320e-01 2.92304121e-02 6.37298882e-01
5.93782067e-01 -9.04511273e-01 3.39130580e-01 8.11391592e-01
6.14278913e-01 7.67437220e-01 6.84944093e-01 -9.91955221e-01
-1.47152293e+00 -1.11963773e+00 4.31429267e-01 -8.24320972e-01
7.03435063e-01 -5.49884081e-01 -8.26228917e-01 7.25326538e-01
-2.83337474e-01 3.00482035e-01 1.32757366e+00 -1.77119255e-01
-1.18282294e+00 1.25258267e-01 -1.82085001e+00 5.63012302e-01
4.62065876e-01 -7.95856416e-01 -2.24561423e-01 9.71995413e-01
7.22638249e-01 3.09906393e-01 -5.37651896e-01 1.09578639e-01
2.30548368e-03 -5.43819070e-01 8.93637240e-01 -1.33739293e+00
4.03653830e-02 -1.53398104e-02 -4.60031033e-02 -8.67127061e-01
1.62320182e-01 -8.45831037e-01 -6.69832885e-01 1.07484710e+00
5.24563789e-01 -7.56798804e-01 1.45796716e+00 7.88584948e-01
8.51263344e-01 -4.03726220e-01 -4.23327386e-01 -8.69707346e-01
4.12620604e-01 -1.31991297e-01 2.21911848e-01 1.24159408e+00
6.82507873e-01 1.79383367e-01 -5.76514244e-01 5.03799260e-01
1.41699183e+00 -4.02502745e-01 7.60451972e-01 -1.05118573e+00
-4.62333888e-01 3.14842939e-01 -1.84578300e-01 -4.75632578e-01
3.07558239e-01 -6.16183817e-01 -5.92770994e-01 -3.29995841e-01
3.46227229e-01 -3.84423018e-01 -2.59180754e-01 1.07344437e+00
-1.73929483e-02 7.55924940e-01 -9.63352844e-02 4.05055642e-01
-5.20499170e-01 1.91959247e-01 3.84899169e-01 -2.36476153e-01
-2.49581531e-01 3.52425963e-01 -9.39113915e-01 7.33577132e-01
1.35868299e+00 -1.29722464e+00 -6.31666064e-01 1.60600916e-01
-2.64514834e-01 -3.18926543e-01 1.25617534e-01 -8.89381886e-01
2.00613305e-01 -4.42556024e-01 5.60876727e-01 -3.27874810e-01
1.19478609e-02 -9.45882320e-01 -3.08849692e-01 1.01838958e+00
-8.90600085e-01 -4.90841925e-01 2.21090531e-03 7.51485229e-01
3.05612296e-01 -5.76543450e-01 1.14408624e+00 -1.90353140e-01
2.86382049e-01 6.13018095e-01 -6.28172338e-01 1.36643827e-01
1.44977713e+00 1.05278045e-01 -5.79352379e-01 -4.33199883e-01
-5.89467943e-01 3.53371240e-02 6.24674141e-01 -2.09079519e-01
5.76910019e-01 -9.27652717e-01 -3.28958035e-01 1.86820999e-01
6.83356375e-02 -4.47661489e-01 -4.62605655e-01 1.62192538e-01
-2.21115276e-01 3.31221700e-01 -1.59242749e-01 -1.11700911e-02
-1.22334254e+00 1.35182631e+00 2.60725021e-01 -2.78472960e-01
-2.58843303e-01 5.39365232e-01 -1.48218144e-02 -2.78691858e-01
6.00417852e-01 6.20356917e-01 3.99810076e-02 -3.59561741e-01
1.04239535e+00 2.95908660e-01 -3.33380878e-01 -2.41806775e-01
-1.76654562e-01 -2.78892279e-01 -5.63202739e-01 -1.14458092e-01
1.11546755e+00 5.84872246e-01 1.10878125e-01 3.39092277e-02
1.26380873e+00 2.21122488e-01 -1.21195734e+00 -4.26115721e-01
2.10330307e-01 -4.92831320e-01 -1.66126862e-01 -8.53719115e-01
-1.02441919e+00 1.05776215e+00 3.01640362e-01 4.11774725e-01
6.59032404e-01 -1.87928118e-02 8.09095979e-01 1.00991261e+00
7.68040240e-01 -6.07679188e-01 3.73403877e-01 -6.74598515e-02
2.07382422e-02 -9.90775466e-01 1.41296595e-01 -6.30492389e-01
-6.86530113e-01 9.76950347e-01 5.79071164e-01 -4.32967812e-01
5.91614962e-01 7.09096789e-01 1.08058780e-01 -1.33679941e-01
-6.97267950e-01 2.94103801e-01 -3.84936661e-01 1.10210717e+00
-1.99386925e-01 -1.28385752e-01 2.56045144e-02 6.33309901e-01
1.99408934e-01 -4.88295525e-01 7.18539119e-01 1.03850174e+00
-6.19307101e-01 -1.53765249e+00 -6.78279042e-01 5.56890011e-01
-9.63939428e-01 -4.43423167e-02 -1.10685217e+00 3.46949041e-01
-5.18663824e-01 1.29892397e+00 -4.59319085e-01 -7.25751221e-01
-5.82361221e-02 2.87644267e-01 1.92621857e-01 -9.58368778e-01
-8.49150538e-01 -1.89595252e-01 9.68656316e-02 -2.74059117e-01
-3.11353318e-02 -5.88023543e-01 -1.20437908e+00 -4.95293528e-01
-7.13807881e-01 6.20571554e-01 4.26740140e-01 9.73722816e-01
-7.40437955e-02 -5.69982648e-01 1.40159011e+00 -4.75090355e-01
-1.57331049e+00 -6.52598083e-01 -9.10110891e-01 4.05064702e-01
1.79530978e-01 -2.89940745e-01 -7.38407671e-01 -3.05016711e-02] | [5.831053733825684, 7.503674030303955] |
c229ad15-5abb-40fd-9635-0d205bbb01cd | 3d-face-reconstruction-from-single-image-with | null | null | https://doi.org/10.1016/j.jksuci.2022.11.014 | https://reader.elsevier.com/reader/sd/pii/S131915782200413X?token=585B515A57883E1FC38B2C59EA999ADE93A1CD1FA4472234FB6507500E717B7477F6BE947D46B1573BF0B3D67BD6EAAD&originRegion=eu-west-1&originCreation=20221224162725 | 3D face reconstruction from single image with generative adversarial networks | Traditional reconstruction techniques extract information from the object’s geometry or one or more 2D images. On the other hand, the limit of the existing methods is that they generate less precise objects. Thus the lack of robustness towards several face reconstruction problems, such as the position of the head, occlusion, noise, and lighting variation. Therefore, generative neural networks and graphical convolution networks have marked a significant evolution in the field of 3D reconstruction. This paper proposes a model for 3D face reconstruction from a single 2D image. Our model is composed of a generator and a discriminator based on convolutional graphic layers. Indeed, in order to generate a face mesh with expression, our idea is to use the landmarks associated with this image as input to the generator to reconstruct a face geometry with expression and improve the convergence rate. As a result, our model offers an accurate reconstruction of facial geometry with expression; thus, our model outperforms state-of-the-art methods through qualitative and quantitative comparison. | ['Mehdi Malah;Mounir Hemam;Fayçal Abbas'] | 2022-12-01 | null | null | null | journal-of-king-saud-university-computer-and | ['3d-face-reconstruction', 'face-reconstruction'] | ['computer-vision', 'computer-vision'] | [-1.03235794e-02 3.27767998e-01 3.48806947e-01 -4.34049636e-01
-2.99235702e-01 -3.06863248e-01 6.58795059e-01 -3.17023814e-01
3.14894388e-03 5.25356650e-01 -9.92709398e-02 1.04932211e-01
2.33786687e-01 -1.27116799e+00 -6.56350911e-01 -5.53382158e-01
3.55335027e-01 6.01047456e-01 -2.01895609e-01 -3.25934440e-01
9.66466963e-02 1.14397609e+00 -1.77261984e+00 -1.07890107e-02
5.40773988e-01 1.19130266e+00 -1.03269920e-01 1.32825240e-01
-4.34862912e-01 1.96392313e-01 -6.07252538e-01 -5.08945048e-01
1.84137091e-01 -6.92937434e-01 -3.59366745e-01 3.18910658e-01
3.40927958e-01 -5.58615506e-01 -9.66553390e-02 9.82758164e-01
6.39257252e-01 -1.45683691e-01 8.09208155e-01 -1.00714874e+00
-6.57381773e-01 2.04210758e-01 -5.79166770e-01 -5.89618802e-01
5.76975584e-01 -5.20978961e-03 3.25209737e-01 -1.07649505e+00
8.56058300e-01 1.42462957e+00 7.06096530e-01 7.41715610e-01
-1.32688868e+00 -4.82769370e-01 -1.56810924e-01 -3.24466556e-01
-1.40485311e+00 -4.81405199e-01 1.16330290e+00 -5.87126732e-01
2.29934201e-01 2.21695840e-01 9.49475110e-01 1.08771682e+00
-8.30940008e-02 3.18181843e-01 1.09881806e+00 -4.51022267e-01
2.01852113e-01 3.08416694e-01 -6.78899109e-01 8.91272128e-01
-1.19637009e-02 2.03520909e-01 -2.60265678e-01 -1.83683917e-01
1.24472249e+00 -9.99024361e-02 -2.82433063e-01 -4.71705765e-01
-6.48791969e-01 6.96885467e-01 4.44398433e-01 3.79291862e-01
-4.17589664e-01 9.81687456e-02 -1.90088049e-01 -3.04344222e-02
6.17518306e-01 3.46692264e-01 1.06717803e-01 7.11337551e-02
-1.15324998e+00 4.40057635e-01 7.32748687e-01 5.91102064e-01
8.31841648e-01 3.11657727e-01 -4.63111065e-02 7.38853574e-01
5.00318646e-01 4.50006902e-01 2.16768846e-01 -8.91808212e-01
-1.30220847e-02 8.30829799e-01 -1.28529340e-01 -1.31512594e+00
-4.02856797e-01 -4.21027064e-01 -8.55312943e-01 7.62918472e-01
3.85484636e-01 1.01199225e-02 -8.65939677e-01 1.71932173e+00
6.04218483e-01 -2.24266127e-02 -2.19380543e-01 1.03278327e+00
1.02656484e+00 4.43381280e-01 -1.46998346e-01 -1.10647969e-01
1.00575757e+00 -3.39995086e-01 -6.41566634e-01 4.46943343e-02
2.42197827e-01 -7.19408095e-01 7.01760709e-01 3.52638364e-01
-1.41187215e+00 -7.10969985e-01 -7.97927380e-01 -1.03143252e-01
-2.81383872e-01 4.09612983e-01 3.61352593e-01 5.34713507e-01
-1.23007619e+00 6.58218384e-01 -4.73224163e-01 -9.37951654e-02
5.88921070e-01 3.28780562e-01 -5.62786222e-01 1.26903534e-01
-9.45911884e-01 9.19352472e-01 3.80200781e-02 4.02190924e-01
-5.53444743e-01 -7.04183400e-01 -9.02386487e-01 4.09317911e-02
-7.60150999e-02 -8.65533710e-01 9.57134843e-01 -1.24065435e+00
-1.93903017e+00 1.19149709e+00 4.11527865e-02 1.12686738e-01
9.34322357e-01 1.13221616e-01 -6.02601096e-02 7.38517568e-02
-1.47212490e-01 7.20982671e-01 9.34688687e-01 -1.54131472e+00
1.34083470e-02 -6.15370810e-01 -3.82380225e-02 -2.48793233e-02
3.86896655e-02 -1.56906530e-01 -5.00383377e-01 -5.78926504e-01
3.26113790e-01 -6.64608777e-01 1.83242839e-02 5.06321430e-01
-2.88252145e-01 -9.48232263e-02 9.32761014e-01 -6.68532729e-01
8.19086671e-01 -2.29644847e+00 2.51172245e-01 3.87107819e-01
2.00652853e-01 2.00606987e-01 -6.35752082e-02 1.65930137e-01
-2.66278118e-01 1.82951808e-01 -2.47216031e-01 -6.76399112e-01
-6.88266680e-02 -1.88613981e-02 2.06665266e-02 5.47421157e-01
3.72098953e-01 7.80499935e-01 -7.71659791e-01 -4.92364854e-01
3.35913658e-01 1.04335225e+00 -4.86476272e-01 3.74240279e-01
-2.98051834e-01 8.91197622e-01 -4.95476276e-01 5.04009545e-01
8.91842008e-01 8.63474905e-02 2.87376225e-01 -2.72722781e-01
-1.83827266e-01 -4.43764590e-02 -1.13663566e+00 1.71910942e+00
-5.27075887e-01 4.63020355e-01 2.08246723e-01 -8.11965346e-01
1.37514198e+00 6.19528174e-01 6.25565767e-01 -7.50834405e-01
4.67764020e-01 3.40885609e-01 -3.32926840e-01 -4.39815342e-01
1.49285853e-01 -4.35085207e-01 3.22599828e-01 3.33407909e-01
-7.87312537e-02 -5.16976058e-01 -1.94669291e-01 -1.46752030e-01
4.22408640e-01 5.56855798e-01 9.97427627e-02 -1.00017814e-02
6.17876172e-01 -5.53748965e-01 3.66013080e-01 2.04860629e-03
4.58860785e-01 1.04236472e+00 7.88107097e-01 -5.92176437e-01
-1.09215462e+00 -9.23507631e-01 -1.90803930e-01 2.56700456e-01
-1.52034476e-01 -2.35491216e-01 -1.01623952e+00 -3.83519530e-01
-2.50554346e-02 5.63176095e-01 -8.50433171e-01 -3.39134745e-02
-7.03623235e-01 -5.35578430e-01 4.57202852e-01 2.82871544e-01
6.02009118e-01 -1.05887544e+00 -6.58473969e-01 9.81863514e-02
2.15350371e-02 -8.90496850e-01 1.26878796e-02 -4.23028529e-01
-9.20057118e-01 -9.47600842e-01 -8.48555744e-01 -6.31781161e-01
1.00981569e+00 -3.47275138e-01 1.10704756e+00 3.27408224e-01
-3.31871450e-01 1.25890121e-01 5.09189479e-02 -2.70896435e-01
-7.27486551e-01 -2.78122157e-01 -1.98301345e-01 3.73561442e-01
-2.76832283e-01 -8.52374494e-01 -6.18887603e-01 2.22736716e-01
-9.66996074e-01 3.58332217e-01 3.14342856e-01 4.71641779e-01
7.07581341e-01 -6.26170263e-02 3.10081393e-01 -8.20621908e-01
5.51246345e-01 -2.86133647e-01 -7.91667819e-01 1.49872918e-02
-3.03513527e-01 1.34323701e-01 5.91592252e-01 -2.31622383e-01
-1.30974972e+00 3.25155646e-01 -4.75679457e-01 -4.71862942e-01
-3.85212988e-01 1.92229375e-01 -3.73085678e-01 -8.83895755e-02
7.15295911e-01 1.58382684e-01 4.74996865e-01 -7.01200485e-01
2.13055477e-01 4.72227395e-01 4.00736392e-01 -5.17401636e-01
8.19247305e-01 5.67127705e-01 4.57561940e-01 -8.27964127e-01
-4.14490521e-01 2.45318979e-01 -8.46197903e-01 -4.71950114e-01
6.59609556e-01 -6.56283140e-01 -7.31312573e-01 6.08051717e-01
-1.44686723e+00 -2.31099337e-01 -1.55100286e-01 1.97729334e-01
-5.84042966e-01 1.74964055e-01 -3.72601300e-01 -8.78659487e-01
-2.81372845e-01 -1.28691769e+00 1.27501798e+00 3.85161370e-01
-9.55234692e-02 -7.60445893e-01 -9.31912884e-02 7.55273327e-02
4.83179688e-01 8.92464936e-01 9.13407266e-01 1.66274205e-01
-5.43318570e-01 -2.93259352e-01 -7.31461588e-03 3.58925909e-01
1.26912981e-01 2.97362566e-01 -1.02690327e+00 2.17311587e-02
-2.97065862e-02 -1.14298180e-01 2.42665738e-01 2.47502968e-01
1.34494770e+00 -3.59738648e-01 -1.81278706e-01 8.66607070e-01
1.37513769e+00 1.17798425e-01 9.07248974e-01 -6.88254461e-02
5.52533805e-01 8.34254563e-01 9.91985574e-02 4.70562488e-01
4.04549129e-02 1.00699270e+00 5.94783783e-01 -2.67970383e-01
-5.42872369e-01 -4.63120610e-01 -3.46541754e-03 3.37803096e-01
-5.44508338e-01 5.68702631e-02 -6.87472403e-01 6.80892617e-02
-1.38380289e+00 -7.72529483e-01 -6.95110783e-02 2.18460703e+00
6.76596045e-01 -2.00948268e-01 7.03391060e-02 1.70244709e-01
5.15792489e-01 -1.47513360e-01 -3.28614116e-01 -3.28100920e-01
-1.69425294e-01 3.88866723e-01 -1.31289870e-01 2.78483719e-01
-5.42056978e-01 7.25521147e-01 5.97348356e+00 5.15840113e-01
-1.59267557e+00 -2.26419896e-01 8.33652258e-01 -4.00685053e-03
-4.57882434e-01 -3.17714721e-01 -5.32648206e-01 3.84394437e-01
4.45306987e-01 1.90200806e-01 3.23691130e-01 8.93071592e-01
2.71512151e-01 -2.52080318e-02 -9.42731798e-01 1.22345543e+00
3.24890763e-01 -1.25972235e+00 1.95560932e-01 2.30384603e-01
5.82997084e-01 -5.82201540e-01 9.92510170e-02 -1.21511728e-01
-2.79085428e-01 -1.29130733e+00 1.02021086e+00 7.97991574e-01
1.20417023e+00 -7.17242539e-01 5.47542691e-01 3.04817915e-01
-7.35899806e-01 4.57486570e-01 -1.09003365e-01 2.00243428e-01
2.19280839e-01 5.88116705e-01 -7.89483070e-01 5.08606076e-01
4.62708771e-01 2.65033543e-01 -3.99981618e-01 7.66001046e-01
-4.61891472e-01 2.39192367e-01 -3.30602974e-01 1.49463832e-01
6.49303105e-03 -5.21692455e-01 3.60202074e-01 8.85792792e-01
5.64487696e-01 1.71106622e-01 -2.16991380e-01 1.42105532e+00
-1.97448313e-01 2.68271774e-01 -7.69395053e-01 1.75128654e-01
1.58409789e-01 1.32934189e+00 -6.67272210e-01 -5.06349578e-02
-1.30774438e-01 8.47316325e-01 3.38209152e-01 3.67905289e-01
-7.77285933e-01 -7.96887130e-02 4.91628677e-01 6.13290727e-01
7.05972165e-02 -2.59113431e-01 -3.27933908e-01 -8.09430897e-01
3.80747393e-02 -6.64635539e-01 -2.49680176e-01 -9.30531085e-01
-8.81916046e-01 8.60906720e-01 -1.64846912e-01 -9.29239988e-01
-4.93065059e-01 -5.27204633e-01 -3.77905011e-01 1.04390740e+00
-1.25401604e+00 -1.17289186e+00 -5.78791797e-01 4.49310333e-01
1.14446029e-01 4.72901724e-02 9.97407258e-01 3.90759408e-01
-2.52106071e-01 4.64618862e-01 -4.27098304e-01 3.24373782e-01
2.41077140e-01 -9.20356333e-01 2.56902695e-01 3.75461489e-01
7.07014799e-02 3.82067412e-01 5.56714058e-01 -4.52113092e-01
-1.27887583e+00 -6.58460200e-01 7.65083671e-01 -2.07005545e-01
-8.14915597e-02 -4.13978755e-01 -7.90061355e-01 3.90505761e-01
-2.88853377e-01 1.06443301e-01 3.27419430e-01 -2.62678891e-01
-2.71495134e-01 -1.67590365e-01 -1.40949738e+00 6.60012186e-01
1.04921770e+00 -3.04176420e-01 -2.27962136e-01 -1.01037956e-04
-2.76696905e-02 -6.81497574e-01 -7.40216911e-01 3.51384729e-01
7.57871568e-01 -1.37800694e+00 8.99111271e-01 -1.55957431e-01
5.65894425e-01 -2.51798034e-01 1.68323204e-01 -1.32659209e+00
-1.62936836e-01 -4.95734215e-01 6.90409318e-02 1.05690765e+00
2.58389711e-01 -5.25069356e-01 8.08135390e-01 6.03694379e-01
4.72926861e-03 -8.55524659e-01 -9.43981826e-01 -3.84963065e-01
5.10019213e-02 -1.24685310e-01 9.56056237e-01 7.52115548e-01
-4.04441953e-01 1.84327736e-01 -1.75360605e-01 -1.15638733e-01
5.41175008e-01 3.94786417e-01 9.18203533e-01 -1.48916030e+00
-5.11416197e-02 -6.05696261e-01 -4.75631714e-01 -9.12949622e-01
2.46370241e-01 -7.80283868e-01 -1.17111899e-01 -1.48307419e+00
-2.35775173e-01 -5.86915314e-01 4.13708210e-01 3.84617269e-01
2.56788731e-01 5.35285532e-01 1.25792801e-01 7.63624161e-03
2.22885475e-01 7.24438667e-01 1.65172791e+00 1.92807227e-01
-1.90876529e-01 -1.39833745e-02 -5.02933383e-01 8.78056109e-01
7.72286594e-01 -2.50456542e-01 -1.11222796e-01 -4.43965733e-01
2.90456265e-01 2.26769954e-01 5.07593572e-01 -9.29429352e-01
-1.50550306e-01 1.03093520e-01 8.20345461e-01 -3.63269389e-01
6.06503308e-01 -8.46510410e-01 7.80664206e-01 3.15432072e-01
1.20816473e-02 -7.34219551e-02 8.80692005e-02 6.70148134e-02
-1.55537307e-01 -1.38211623e-01 1.03405726e+00 -3.49787354e-01
-1.24739505e-01 4.27763849e-01 9.41675995e-03 -3.59969258e-01
7.45116651e-01 -5.20265341e-01 1.53721780e-01 -5.20875275e-01
-7.18735039e-01 -4.21352327e-01 6.69591069e-01 2.64081001e-01
7.82643557e-01 -1.38709939e+00 -7.93082833e-01 6.71215713e-01
-2.72631615e-01 3.40543956e-01 1.28092468e-01 5.89453280e-01
-8.09889436e-01 1.57122493e-01 -4.58495378e-01 -4.84548569e-01
-1.24086297e+00 2.51895040e-01 7.25549161e-01 1.79936513e-01
-6.08718574e-01 5.20945966e-01 1.96829826e-01 -4.51013774e-01
-1.96970664e-02 -7.02208802e-02 -1.87145084e-01 -1.13341045e-02
3.95324826e-01 1.57789841e-01 1.15594700e-01 -8.66205871e-01
-6.18636161e-02 8.84029388e-01 5.87156296e-01 -2.22285748e-01
1.37146389e+00 2.83937812e-01 -2.72703439e-01 1.91551074e-01
1.16339219e+00 1.16985455e-01 -1.24915838e+00 8.75431076e-02
-4.63106811e-01 -6.46237016e-01 -7.95826763e-02 -7.19830990e-01
-1.52126682e+00 8.40103626e-01 3.74147505e-01 2.58712955e-02
1.11033428e+00 1.25534739e-02 4.91825521e-01 -2.36950919e-01
4.10865426e-01 -8.77331316e-01 1.20070599e-01 1.51240915e-01
1.28598523e+00 -9.44821596e-01 -1.74384803e-01 -6.00181103e-01
-4.62533347e-02 1.50213575e+00 5.27471542e-01 1.36876013e-02
7.55933583e-01 3.08279157e-01 1.41134307e-01 -3.99923205e-01
-1.45211443e-02 3.20132449e-02 3.61770213e-01 5.27311802e-01
6.36230171e-01 2.17705444e-02 -3.44525367e-01 2.06898719e-01
-5.28680265e-01 1.89022273e-01 1.33019522e-01 5.33510387e-01
-5.05260304e-02 -1.14877045e+00 -4.76638556e-01 -7.76920374e-03
-3.78096491e-01 3.47434610e-01 -4.70267743e-01 7.72273242e-01
4.01501000e-01 5.30880213e-01 1.75455198e-01 -4.88545932e-02
5.28346300e-01 1.12104155e-01 8.45126212e-01 -5.26272058e-01
-4.62144375e-01 1.01944648e-01 -2.50579745e-01 -5.68479478e-01
-3.33069921e-01 -3.91967595e-01 -1.18074226e+00 -3.20839375e-01
-4.23174798e-02 -9.45947543e-02 1.12768006e+00 5.71504414e-01
4.26272750e-01 3.50223571e-01 5.80768883e-01 -1.07636774e+00
-2.67881244e-01 -8.93224180e-01 -7.08043694e-01 6.81095600e-01
5.23924530e-02 -8.66711318e-01 -1.91510782e-01 3.41198519e-02] | [12.733809471130371, -0.2692641615867615] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.