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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
5d9ff70d-ff3a-4ee6-96b1-3fb14e1f0a9d | yolo2u-net-detection-guided-3d-instance | 2207.06215 | null | https://arxiv.org/abs/2207.06215v1 | https://arxiv.org/pdf/2207.06215v1.pdf | YOLO2U-Net: Detection-Guided 3D Instance Segmentation for Microscopy | Microscopy imaging techniques are instrumental for characterization and analysis of biological structures. As these techniques typically render 3D visualization of cells by stacking 2D projections, issues such as out-of-plane excitation and low resolution in the $z$-axis may pose challenges (even for human experts) to detect individual cells in 3D volumes as these non-overlapping cells may appear as overlapping. In this work, we introduce a comprehensive method for accurate 3D instance segmentation of cells in the brain tissue. The proposed method combines the 2D YOLO detection method with a multi-view fusion algorithm to construct a 3D localization of the cells. Next, the 3D bounding boxes along with the data volume are input to a 3D U-Net network that is designed to segment the primary cell in each 3D bounding box, and in turn, to carry out instance segmentation of cells in the entire volume. The promising performance of the proposed method is shown in comparison with some current deep learning-based 3D instance segmentation methods. | ['David Solecki', 'Abbas Shirinifard', 'Derek C. Ros', 'Amirkoushyar Ziabari'] | 2022-07-13 | null | null | null | null | ['3d-instance-segmentation-1'] | ['computer-vision'] | [ 9.30761024e-02 -6.45925552e-02 6.30174279e-01 -1.20238908e-01
-5.61487734e-01 -5.17644703e-01 1.90926671e-01 2.86080688e-01
-4.98693734e-01 7.74571240e-01 -4.86007035e-01 -9.49063525e-02
2.41146013e-01 -5.03395677e-01 -5.12732506e-01 -1.04643357e+00
-1.12286225e-01 7.49711871e-01 3.86562765e-01 1.94564372e-01
4.88411307e-01 1.04498732e+00 -1.23073113e+00 8.80074352e-02
4.97554004e-01 9.91804838e-01 3.42986405e-01 6.45094812e-01
-3.06434959e-01 1.30908817e-01 -6.36154413e-01 1.79255575e-01
3.02138209e-01 -2.05221534e-01 -5.50372005e-01 2.75027901e-01
3.88466507e-01 -4.13156927e-01 3.50260615e-01 9.79688644e-01
5.96505761e-01 -1.66170016e-01 1.05928695e+00 -7.54645348e-01
-4.06412855e-02 1.25214495e-02 -1.01145959e+00 4.30243641e-01
3.49877566e-01 -9.07781646e-02 4.85310316e-01 -1.09472275e+00
1.08491349e+00 1.08775485e+00 3.85542423e-01 6.42046452e-01
-1.40448534e+00 -5.58936238e-01 1.19601145e-01 -2.40535155e-01
-1.32516718e+00 -2.34793797e-01 1.06096160e+00 -8.20522130e-01
8.85376811e-01 1.01971082e-01 7.81873703e-01 4.79721129e-01
4.19074744e-01 6.53621018e-01 1.46163166e+00 -3.06158185e-01
2.38187104e-01 4.10628878e-02 4.21287209e-01 6.52998984e-01
5.70111983e-02 -2.85940260e-01 -1.71112508e-01 -1.66401610e-01
1.21375251e+00 1.82718739e-01 -3.20725828e-01 -4.82841522e-01
-1.10134959e+00 3.07064891e-01 2.73001969e-01 4.95119274e-01
-1.51578933e-01 -1.50850611e-02 2.92448580e-01 -3.20991665e-01
6.03429496e-01 2.17081055e-01 -1.53616771e-01 5.61294854e-01
-1.08833313e+00 3.56937438e-01 3.43137205e-01 6.28016353e-01
7.68686235e-01 -1.66681439e-01 1.18954137e-01 5.06668270e-01
3.37484419e-01 -3.69803011e-02 2.44346067e-01 -9.24507082e-01
1.40018523e-01 1.16509521e+00 1.53634362e-02 -6.98241949e-01
-7.53421009e-01 -1.53195798e-01 -9.75797951e-01 6.83140457e-01
7.74833024e-01 -1.92869410e-01 -9.50966716e-01 1.09486067e+00
8.34957242e-01 -4.51258086e-02 -8.48437920e-02 1.03118610e+00
8.62118185e-01 5.51250696e-01 -4.40575093e-01 -2.79031634e-01
1.44489706e+00 -6.06571317e-01 -5.28042972e-01 1.94700640e-02
6.23413503e-01 -5.16294956e-01 6.75691187e-01 1.58316046e-01
-1.15422463e+00 -3.17850500e-01 -8.51430357e-01 -1.18455574e-01
-4.36550409e-01 1.95155621e-01 2.03680247e-01 1.63912743e-01
-9.97535706e-01 2.54410148e-01 -8.35206807e-01 -2.99242884e-01
9.25496936e-01 5.15707731e-01 -6.73399985e-01 1.95583418e-01
-3.89332384e-01 6.87068999e-01 3.39149356e-01 2.73682207e-01
-1.01244462e+00 -7.14112580e-01 -5.60884893e-01 -7.69364461e-02
4.54694815e-02 -5.93868732e-01 6.21570528e-01 -2.56359577e-01
-1.10057306e+00 1.37498820e+00 -4.28321570e-01 -2.58808404e-01
6.12853408e-01 1.86086670e-01 3.58848125e-01 5.02446651e-01
2.62534395e-02 7.86657214e-01 2.61530399e-01 -1.67232287e+00
-6.11266911e-01 -1.15049434e+00 -1.50720567e-01 3.41801941e-01
1.84625462e-01 -1.49099678e-01 -2.36944258e-01 -1.77706271e-01
5.68384886e-01 -2.54257530e-01 -3.48944277e-01 8.43158290e-02
-7.24543452e-01 -2.66080443e-02 1.20480168e+00 -5.26880741e-01
5.65265656e-01 -1.97447991e+00 1.88146532e-01 -1.84264686e-02
6.49112284e-01 -7.85727575e-02 4.50827390e-01 8.42801034e-02
9.68478061e-03 -2.46241479e-03 -2.82055080e-01 -5.24142981e-01
-4.61682081e-01 -2.58610368e-01 8.04415196e-02 9.62945521e-01
-3.03238090e-02 5.86110473e-01 -4.81315970e-01 -8.01967263e-01
3.63661319e-01 6.21547818e-01 -2.26136476e-01 3.36430371e-01
-9.25177615e-03 9.47955072e-01 -3.91670078e-01 7.26555109e-01
7.36230969e-01 -2.24837258e-01 1.00498274e-01 -1.47981524e-01
-3.76349449e-01 -3.19248468e-01 -8.75263870e-01 1.28363335e+00
2.43214872e-02 5.10773361e-01 5.34655392e-01 -1.26873338e+00
9.92128551e-01 3.18507850e-01 6.92992151e-01 -2.33382329e-01
3.15475136e-01 2.71894932e-01 -2.48706341e-01 -3.43176901e-01
-1.83602437e-01 -2.14676425e-01 1.83025762e-01 4.87384677e-01
-3.37436050e-02 -2.79826641e-01 2.56030917e-01 -3.02400906e-02
6.09786153e-01 -9.15052369e-03 2.24391535e-01 -4.95726496e-01
7.59136558e-01 -1.25916973e-01 4.59512353e-01 2.12602675e-01
-2.74388313e-01 7.32294500e-01 8.33134472e-01 -7.49821126e-01
-1.18389881e+00 -8.41392457e-01 -4.25739110e-01 2.00635046e-01
3.55286449e-01 4.36529517e-01 -1.07480145e+00 -5.21224678e-01
-1.11189649e-01 1.49901636e-04 -6.42364740e-01 5.58079064e-01
-7.00791299e-01 -8.96103263e-01 1.24639601e-01 1.15042329e-01
4.81183380e-01 -7.32873440e-01 -1.36833537e+00 2.41584182e-01
2.00307667e-02 -1.21049929e+00 -1.06081970e-01 6.06890500e-01
-1.02485824e+00 -1.20755410e+00 -1.12405622e+00 -1.09522986e+00
1.21543908e+00 2.19814897e-01 5.72324038e-01 7.12345466e-02
-5.92558682e-01 -5.58686592e-02 -4.45300853e-03 -2.76349723e-01
2.08475385e-02 -3.23116750e-01 -2.58551314e-02 -2.28610635e-01
2.44446248e-01 -6.36767030e-01 -7.70911336e-01 2.63673216e-01
-6.67888701e-01 1.33870170e-01 6.01288155e-02 8.10773075e-01
1.18140495e+00 6.10576570e-02 2.81783640e-01 -9.68704939e-01
3.24309349e-01 -2.70805769e-02 -9.75984097e-01 6.48942590e-02
-6.16765060e-02 -4.72301006e-01 6.61180615e-01 -1.95676297e-01
-7.79195309e-01 4.38756764e-01 -7.38036856e-02 -5.20474136e-01
-6.64290071e-01 9.58333835e-02 -3.11165266e-02 -2.68538922e-01
3.63733530e-01 4.29378510e-01 1.67875260e-01 -3.33436906e-01
-1.66378006e-01 4.32757109e-01 1.89901680e-01 -2.03157634e-01
3.39423507e-01 1.09674394e+00 2.52458543e-01 -9.09777761e-01
-4.91051584e-01 -5.26523113e-01 -1.17144370e+00 -5.70194066e-01
1.09477413e+00 -6.76908612e-01 -8.37791800e-01 7.38477111e-01
-1.35816491e+00 -2.85285860e-01 9.67683643e-02 3.25186282e-01
-6.50686920e-01 4.04699385e-01 -8.33951056e-01 -1.02631903e+00
-4.41247314e-01 -1.56845176e+00 1.34557867e+00 4.59101915e-01
-1.49407655e-01 -9.77199376e-01 -4.10382487e-02 5.08621156e-01
-1.06406912e-01 5.13349354e-01 1.28979230e+00 -5.25210202e-01
-5.69789827e-01 -2.51231819e-01 -1.05660394e-01 -2.20079705e-01
-8.68293047e-02 1.34273708e-01 -1.19313061e+00 -1.42892629e-01
1.82343274e-01 -1.73128590e-01 6.87238514e-01 1.11831951e+00
1.09377933e+00 1.90346047e-01 -8.91717792e-01 6.95169568e-01
1.44063711e+00 5.52914023e-01 1.89222842e-01 4.88753840e-02
5.80756903e-01 9.86709654e-01 1.83455020e-01 3.04867834e-01
5.89955449e-02 5.07021964e-01 7.34263957e-01 -4.25019026e-01
1.23133950e-01 4.06295717e-01 -1.31739810e-01 4.69021738e-01
1.06587159e-02 -1.11429252e-01 -7.68788338e-01 5.56781530e-01
-1.28955448e+00 -7.13578820e-01 -1.99049085e-01 2.00992084e+00
4.52144653e-01 1.16967686e-01 2.60744989e-01 2.56010801e-01
9.48086441e-01 -1.32481948e-01 -7.19875574e-01 -2.61163980e-01
-2.11257115e-01 -1.70017794e-01 2.53831148e-01 5.78855634e-01
-1.11696744e+00 7.38973677e-01 6.43817091e+00 7.12299764e-01
-1.11288023e+00 -2.35253125e-01 1.27408576e+00 -1.02701262e-01
-1.87657446e-01 -3.36198151e-01 -1.26860237e+00 3.11196893e-01
3.42966169e-02 2.57140070e-01 7.94589221e-02 3.17544311e-01
3.82954359e-01 -5.51305473e-01 -1.18649912e+00 1.06466675e+00
-1.13209344e-01 -1.53799784e+00 -4.11142334e-02 4.62672889e-01
6.92066610e-01 -1.43549025e-01 -1.84652299e-01 -3.97382528e-01
-1.73918322e-01 -1.18484116e+00 4.94738758e-01 3.21782529e-01
6.59607172e-01 -7.91139424e-01 7.00662315e-01 8.61336291e-01
-1.09859693e+00 2.00168252e-01 -3.88968468e-01 7.18467757e-02
3.90686512e-01 9.27756906e-01 -8.85856271e-01 1.87177271e-01
6.94216311e-01 3.40006202e-01 -7.93239474e-02 1.06673872e+00
3.69708210e-01 -6.10344335e-02 -4.49267924e-01 -3.13620232e-02
1.18940130e-01 -5.79988718e-01 5.93340814e-01 1.05333364e+00
5.61795235e-01 3.92427832e-01 -6.49227649e-02 1.27624369e+00
1.95518974e-02 5.06357737e-02 -9.03045952e-01 1.37431100e-01
2.65158445e-01 1.54327095e+00 -1.59325325e+00 -3.57809007e-01
-1.96314320e-01 7.53730237e-01 5.33368766e-01 4.00753051e-01
-3.37531626e-01 -2.47788191e-01 3.86653483e-01 3.31416875e-01
1.66541010e-01 -2.88024426e-01 -4.99597758e-01 -7.53011048e-01
-7.80917630e-02 -1.59330159e-01 9.16283056e-02 -7.20174253e-01
-9.69037235e-01 5.83379090e-01 -2.20457837e-01 -9.12865639e-01
1.94002762e-01 -8.27958167e-01 -6.77538395e-01 1.02337801e+00
-1.15656722e+00 -8.25219691e-01 -5.43470345e-02 3.33878815e-01
4.53366965e-01 9.90560204e-02 7.76089728e-01 1.09314241e-01
-6.61114275e-01 -4.40883823e-02 1.44202739e-01 1.58546239e-01
1.68895021e-01 -1.27759683e+00 -4.43483107e-02 6.77260756e-01
-7.63407275e-02 5.63142657e-01 5.15592635e-01 -7.22217500e-01
-9.89672959e-01 -7.22256422e-01 6.41206980e-01 -2.80840248e-01
1.29871428e-01 -6.01577461e-01 -8.37370992e-01 4.58828151e-01
1.08720295e-01 3.35096687e-01 7.67687142e-01 -5.25675714e-01
2.13565499e-01 9.20342728e-02 -1.45859063e+00 5.93635857e-01
6.41918719e-01 -2.23045260e-01 -2.42047250e-01 3.10542196e-01
1.20435156e-01 -7.06336737e-01 -8.16048086e-01 2.38290995e-01
6.64055109e-01 -1.22448981e+00 9.47731674e-01 -2.53731251e-01
4.11616325e-01 -5.28661430e-01 1.04182422e-01 -1.10864615e+00
-2.56787967e-02 -8.05702880e-02 1.52031347e-01 8.66818964e-01
2.84384698e-01 -3.47275674e-01 1.26715696e+00 2.22056672e-01
-1.94793999e-01 -1.18857563e+00 -1.31024742e+00 -2.97908038e-01
1.37634367e-01 -1.72192454e-02 2.87667334e-01 6.77080750e-01
1.00328147e-01 3.13908815e-01 2.05732405e-01 1.97893232e-01
9.24880087e-01 4.99625862e-01 4.46547717e-01 -1.37347031e+00
3.08181554e-01 -5.66499293e-01 -4.92783487e-01 -1.08815384e+00
-1.50578842e-02 -7.94480383e-01 -1.88598409e-01 -1.65182936e+00
3.70504051e-01 -1.61660239e-01 2.23756395e-02 -2.92280558e-02
9.06368345e-02 2.70640284e-01 -1.03368424e-01 2.69031793e-01
-3.44963223e-01 2.26129353e-01 1.84272146e+00 -4.87895980e-02
-1.69423699e-01 -1.26464173e-01 -3.65937710e-01 9.16348338e-01
5.99173665e-01 -1.22101851e-01 -9.40632150e-02 -4.48437601e-01
-2.91571856e-01 3.33540022e-01 4.19880927e-01 -1.06340456e+00
3.62874597e-01 2.16268018e-01 1.00319517e+00 -1.31679082e+00
6.22710824e-01 -1.00447607e+00 6.59698918e-02 4.55967695e-01
-1.30960986e-01 -1.95793971e-01 1.25511186e-04 3.24584633e-01
-1.51955724e-01 -1.35414481e-01 1.19230676e+00 -7.35077620e-01
-1.35713249e-01 1.83859229e-01 -6.94231510e-01 -2.31143355e-01
1.55373180e+00 -9.11289334e-01 -2.95332670e-01 1.40439019e-01
-9.18782949e-01 2.97816664e-01 6.96653247e-01 -6.77432656e-01
9.11421299e-01 -9.89240110e-01 -2.12045416e-01 2.91250139e-01
-2.48513862e-01 7.44407296e-01 5.08597076e-01 1.01615715e+00
-9.68073845e-01 5.35784304e-01 -4.56018478e-01 -8.73387754e-01
-1.43318284e+00 2.59270847e-01 7.44264483e-01 -2.26797178e-01
-6.68414950e-01 1.20841849e+00 7.86003947e-01 -5.31978190e-01
1.96457520e-01 -4.25358295e-01 -7.13773549e-01 1.77230537e-01
4.50021416e-01 2.52545834e-01 -1.93381429e-01 -7.85359681e-01
-4.96948272e-01 1.01843464e+00 -1.19366489e-01 4.09105746e-03
1.41729081e+00 -3.41089815e-01 -4.24299210e-01 6.60184562e-01
1.17795813e+00 -3.13694566e-01 -1.46505868e+00 7.21819326e-02
-3.26122791e-01 -1.89492717e-01 -1.02191344e-01 -4.24175888e-01
-1.27697396e+00 1.39804888e+00 6.16056263e-01 1.63299412e-01
8.51592302e-01 1.13647334e-01 6.72494769e-01 2.70557366e-02
3.69319767e-01 -9.70530808e-01 -1.26134694e-01 4.05880719e-01
3.56672227e-01 -1.04025018e+00 -2.11492982e-02 -5.03420472e-01
-1.16832569e-01 1.40491343e+00 9.07024324e-01 -2.19398826e-01
6.14918172e-01 5.86487651e-01 1.89193055e-01 -7.04188347e-01
-5.23402214e-01 1.65821478e-01 -1.05604030e-01 6.26191735e-01
5.27838051e-01 -1.97081581e-01 -1.92583621e-01 3.05711627e-01
2.53574550e-01 -2.37896740e-01 4.88728195e-01 7.17316985e-01
-6.51494801e-01 -5.70996404e-01 -5.87750256e-01 5.56745410e-01
-5.83634555e-01 2.93254673e-01 -5.14749944e-01 6.70586586e-01
2.23623142e-01 3.51107299e-01 4.27000463e-01 -7.90466275e-03
7.18149841e-02 -5.63001707e-02 6.25785112e-01 -5.74029565e-01
-3.77407432e-01 7.71757424e-01 -4.26178932e-01 -2.11760536e-01
-3.84667784e-01 -5.59905171e-01 -1.80541909e+00 1.66331694e-01
-1.54253095e-01 -4.14281338e-03 4.85948622e-01 9.92637992e-01
8.46000388e-02 2.52789408e-01 4.82619017e-01 -1.28099501e+00
2.62478828e-01 -5.58444262e-01 -1.07454407e+00 -7.49157965e-02
5.06884515e-01 -6.17462397e-01 -4.17452097e-01 2.07164183e-01] | [14.443215370178223, -3.1395487785339355] |
f7e6ba7d-1ac5-428f-ab92-cef6b7a109b9 | move-as-you-like-image-animation-in-e | 2112.13647 | null | https://arxiv.org/abs/2112.13647v1 | https://arxiv.org/pdf/2112.13647v1.pdf | Move As You Like: Image Animation in E-Commerce Scenario | Creative image animations are attractive in e-commerce applications, where motion transfer is one of the import ways to generate animations from static images. However, existing methods rarely transfer motion to objects other than human body or human face, and even fewer apply motion transfer in practical scenarios. In this work, we apply motion transfer on the Taobao product images in real e-commerce scenario to generate creative animations, which are more attractive than static images and they will bring more benefits. We animate the Taobao products of dolls, copper running horses and toy dinosaurs based on motion transfer method for demonstration. | ['Lixin Duan', 'Wen Li', 'Yuning Jiang', 'Tiezheng Ge', 'Jiale Tao', 'Biao Wang', 'Borun Xu'] | 2021-12-19 | null | null | null | null | ['image-animation'] | ['computer-vision'] | [-4.24414217e-01 -2.71180630e-01 7.60039091e-02 6.42966032e-02
9.93225351e-02 -5.24678409e-01 4.04101998e-01 -7.84537733e-01
-4.13050316e-02 8.35412562e-01 -9.48991533e-03 -2.15795696e-01
3.68908167e-01 -8.49216580e-01 -5.14183700e-01 -5.02953947e-01
-1.22229263e-01 1.76409170e-01 3.94931585e-01 -4.93271112e-01
2.25412145e-01 3.11245769e-01 -1.02751279e+00 3.34466577e-01
4.58760649e-01 4.45064336e-01 2.41679609e-01 6.41199350e-01
-2.29923397e-01 6.29661083e-01 -6.54446840e-01 -4.47642714e-01
2.65789509e-01 -8.68395329e-01 -6.75639808e-01 2.95221478e-01
4.57164012e-02 -6.30746305e-01 -4.35043305e-01 9.77532566e-01
4.16125149e-01 6.44464344e-02 6.71591341e-01 -1.76003146e+00
-1.20323551e+00 4.24800485e-01 -8.78462732e-01 -3.26894373e-02
5.17431676e-01 4.53031331e-01 4.08766270e-01 -5.73430002e-01
1.13880301e+00 1.61743641e+00 5.52089572e-01 8.42565656e-01
-7.56725729e-01 -8.14463437e-01 -8.22142735e-02 3.40028286e-01
-1.15461409e+00 -2.34070271e-02 9.15570498e-01 -1.79820165e-01
2.90687144e-01 1.88865423e-01 1.18516600e+00 9.21965122e-01
2.95650512e-01 1.06750512e+00 1.09603691e+00 -1.38420194e-01
2.77891569e-02 2.55810559e-01 -5.52570522e-01 6.69672430e-01
-8.42311382e-02 5.71905188e-02 -3.39831412e-01 1.05999745e-01
1.48195696e+00 -1.18096195e-01 -3.61379415e-01 3.61015014e-02
-1.46730661e+00 8.10408890e-01 4.00001764e-01 2.54600823e-01
-4.59446162e-01 4.81727183e-01 4.43927974e-01 2.54327893e-01
5.39834462e-02 -6.47862777e-02 -7.52865076e-02 -3.93691361e-01
-4.79269326e-01 4.57627773e-01 5.77644050e-01 1.15447903e+00
4.33556378e-01 5.38514733e-01 1.37944952e-01 5.07108927e-01
6.19563758e-01 4.07774389e-01 5.93058050e-01 -9.75519717e-01
1.65002391e-01 2.10466340e-01 3.71866614e-01 -1.16417384e+00
-8.14984217e-02 3.73746991e-01 -6.80654585e-01 8.30404460e-01
1.48609191e-01 -4.30761307e-01 -8.66174161e-01 1.30293107e+00
6.71391606e-01 1.03346460e-01 8.24260339e-02 1.22998643e+00
1.22111189e+00 1.08496284e+00 3.05821627e-01 -1.27756685e-01
1.45604813e+00 -1.19438946e+00 -1.02404237e+00 1.60098553e-01
2.94215530e-01 -1.21057653e+00 1.24942684e+00 4.36585844e-01
-1.32673442e+00 -7.85299540e-01 -8.39891553e-01 2.63620410e-02
7.10149528e-03 -8.46476778e-02 8.95286202e-01 7.17967689e-01
-5.82037926e-01 7.92132318e-01 -5.60292542e-01 -3.52434576e-01
3.33319426e-01 2.60499597e-01 -3.00819755e-01 3.28358680e-01
-1.24644387e+00 7.56980658e-01 4.22028959e-01 2.36193687e-01
-8.36967170e-01 -3.42368722e-01 -5.92081428e-01 -8.22355524e-02
8.59076679e-02 -6.92934513e-01 1.28907347e+00 -1.27769864e+00
-2.13048363e+00 6.50950193e-01 3.47786307e-01 -5.51962890e-02
9.17159259e-01 -2.76359427e-03 -3.75405729e-01 2.12933704e-01
-2.09266305e-01 1.08155668e+00 7.98911750e-01 -1.29861510e+00
-6.44245684e-01 1.37480557e-01 6.13856949e-02 2.38598466e-01
9.49093327e-03 1.47787884e-01 -3.55320841e-01 -7.61654913e-01
-1.86920524e-01 -1.11534798e+00 -6.61883280e-02 4.30766016e-01
-2.30125442e-01 -2.03253403e-01 1.69587100e+00 -8.20668638e-01
8.32203805e-01 -2.03089476e+00 1.34660501e-03 -2.95797765e-01
-2.02953368e-01 5.17330110e-01 -1.27490476e-01 4.35546279e-01
1.81826010e-01 1.51377618e-01 4.50444147e-02 2.13470787e-01
-2.49542654e-01 1.38058797e-01 1.74039546e-02 1.42461777e-01
-2.19085291e-02 1.24367619e+00 -8.70633602e-01 -8.19386482e-01
1.82210758e-01 4.33188885e-01 -3.16753715e-01 -9.33521762e-02
-1.22905999e-01 7.92939425e-01 -6.68751657e-01 6.09357834e-01
8.47800732e-01 1.84461743e-01 6.00114614e-02 -9.38829258e-02
-1.58552527e-01 -2.98099250e-01 -1.19625449e+00 1.50508654e+00
-5.00574052e-01 6.70332372e-01 2.61779070e-01 -3.77068400e-01
8.98417890e-01 3.26378554e-01 3.88680607e-01 -5.74612617e-01
2.91831195e-01 -3.19697720e-04 2.17930853e-01 -1.00183046e+00
4.19231504e-01 -4.46577311e-01 4.70137484e-02 4.94674921e-01
-4.70249653e-01 -7.05316126e-01 3.01570930e-02 4.13588807e-02
4.86253440e-01 8.27917397e-01 2.17884988e-01 -1.42139211e-01
4.32470113e-01 3.69690150e-01 5.41808605e-01 -2.38063514e-01
-4.81000803e-02 8.34497809e-01 3.75813127e-01 -6.58769071e-01
-1.40362453e+00 -9.38724041e-01 3.55058044e-01 5.48285007e-01
6.41215146e-01 -4.85142879e-02 -6.83267593e-01 -5.74620426e-01
-6.48655817e-02 4.92082000e-01 -4.06485140e-01 -9.22318548e-02
-1.00053620e+00 -3.29739422e-01 1.86197340e-01 4.62518752e-01
1.13293886e+00 -1.64520895e+00 -8.97924066e-01 4.39089000e-01
-7.36038461e-02 -7.81408012e-01 -9.16323483e-01 -8.18250775e-01
-9.81504679e-01 -6.85740173e-01 -1.41156006e+00 -1.20445812e+00
5.02069175e-01 6.69827163e-01 7.16461301e-01 1.48409054e-01
-4.02476668e-01 5.23394533e-02 -4.48735148e-01 -4.79741782e-01
-7.38495469e-01 -3.28129679e-01 -1.07184201e-01 -8.82026479e-02
-9.19483826e-02 -6.00141048e-01 -9.11802709e-01 7.27766514e-01
-9.10005450e-01 5.05037904e-01 4.89493012e-01 7.04736292e-01
1.55596063e-01 -8.55759382e-02 6.14906251e-01 -8.59840214e-01
7.40060687e-01 -3.83985490e-01 -4.27139342e-01 1.62120774e-01
-7.59608671e-02 -3.46083164e-01 7.52635896e-01 -1.10349369e+00
-1.30396831e+00 1.21212445e-01 6.08276092e-02 -6.92253768e-01
2.50702798e-01 2.76894093e-01 -1.49680033e-01 -2.25001693e-01
3.61138314e-01 3.46515924e-02 6.12407997e-02 -3.41037065e-01
3.78893942e-01 7.57642150e-01 6.16819978e-01 -4.32242244e-01
7.60925353e-01 5.03584325e-01 -7.80957239e-03 -9.02593136e-01
4.84879285e-01 1.37519687e-01 -4.31508362e-01 -5.31583786e-01
9.80015695e-01 -4.18271095e-01 -1.33028913e+00 4.62320000e-01
-1.37394321e+00 -3.38738203e-01 -1.22748226e-01 6.18671179e-01
-6.24945045e-01 4.70216423e-01 -7.84308374e-01 -5.91478825e-01
-4.14542586e-01 -1.14945459e+00 8.86356413e-01 7.22368360e-01
-3.00157130e-01 -7.23584414e-01 -1.16750143e-01 7.60114640e-02
3.62599730e-01 6.08813167e-01 8.98955166e-01 3.43736410e-01
-7.82873333e-01 1.19453281e-01 -1.40881494e-01 -8.00691824e-03
3.08486223e-01 4.55959946e-01 -2.64295727e-01 1.56537071e-02
-2.28181913e-01 5.70736192e-02 2.03835100e-01 2.67542779e-01
6.87729359e-01 -4.87789422e-01 -4.54560220e-01 3.60320926e-01
1.23839486e+00 9.46139932e-01 1.07423556e+00 2.94125229e-01
6.30035520e-01 6.46581113e-01 8.94754708e-01 2.54954100e-01
2.00727075e-01 8.62465918e-01 2.70333648e-01 -2.76491284e-01
-4.09393489e-01 -4.85385001e-01 4.99271244e-01 8.72714162e-01
-6.41780674e-01 -1.70767769e-01 -5.04106820e-01 4.76854205e-01
-1.81236017e+00 -1.01992393e+00 -6.47655368e-01 1.93214524e+00
5.15331507e-01 -2.60034084e-01 4.42765653e-01 -3.00523013e-01
1.00909936e+00 -2.99300194e-01 -4.09499347e-01 -4.61505979e-01
1.59250125e-01 -4.13524806e-02 1.22757636e-01 2.11411536e-01
-7.00335205e-01 1.00716054e+00 6.98017406e+00 8.22547615e-01
-1.25109553e+00 4.46371973e-01 5.66723883e-01 1.18227020e-01
-3.46219331e-01 1.74963593e-01 -2.88857102e-01 8.87918651e-01
4.08395290e-01 -4.54079777e-01 1.50964603e-01 9.64241445e-01
3.67504865e-01 -5.15377708e-02 -8.22589695e-01 1.07707751e+00
-2.97793984e-01 -1.34894681e+00 9.91102681e-02 1.00496314e-01
8.58148575e-01 -1.00950444e+00 2.84185737e-01 2.29716718e-01
8.60260800e-02 -9.86883640e-01 9.16764796e-01 1.39914602e-01
6.23245180e-01 -7.88995206e-01 4.79643047e-01 2.50677556e-01
-1.14194739e+00 1.80666491e-01 -5.20791411e-01 1.24509141e-01
8.14193368e-01 -2.18886957e-01 -6.75173998e-01 2.41589338e-01
7.03494132e-01 3.02021712e-01 4.20168713e-02 1.07632446e+00
-2.24609375e-01 2.10485801e-01 -9.91578028e-02 -5.12529612e-01
3.51818055e-01 -7.27586806e-01 4.60275263e-01 1.04913592e+00
6.43011630e-01 4.68624622e-01 -2.44696453e-01 9.87482429e-01
1.20187057e-02 3.15289557e-01 -7.16689110e-01 7.69247338e-02
1.19460290e-02 1.43912303e+00 -9.34976637e-01 -4.06938672e-01
-3.30158383e-01 1.28662217e+00 -4.84870672e-01 9.68004018e-02
-1.31829047e+00 -6.94660246e-01 4.19062167e-01 3.21924508e-01
2.53700256e-01 -5.36250532e-01 2.21146166e-01 -9.74322915e-01
-3.73668015e-01 -7.64241159e-01 -7.43794814e-02 -1.37350535e+00
-8.02701950e-01 4.55254078e-01 2.01979205e-01 -1.77325416e+00
-1.46396592e-01 -5.04347086e-01 -1.25135553e+00 5.91336489e-01
-7.72715151e-01 -1.51562726e+00 -5.76589942e-01 2.98297256e-01
1.01513362e+00 1.55597879e-02 3.68160039e-01 4.49202001e-01
-4.51787673e-02 2.51057953e-01 8.46104026e-02 -1.01540461e-01
4.72053647e-01 -6.87056541e-01 7.20977068e-01 3.66534561e-01
-4.85008061e-02 4.85243917e-01 9.00949895e-01 -7.91729450e-01
-1.47363818e+00 -5.36568403e-01 7.57340565e-02 -5.73790818e-02
3.07227761e-01 -1.12089254e-01 -8.66875887e-01 7.01407194e-01
7.87461460e-01 -7.17895776e-02 -2.30339561e-02 -6.35574579e-01
4.59355026e-01 1.80661529e-01 -1.35413778e+00 8.42944980e-01
1.15594888e+00 2.08630890e-01 -3.60067278e-01 3.25489491e-01
4.95165408e-01 -4.54784393e-01 -7.92367399e-01 1.36495912e-02
9.73535538e-01 -7.82898784e-01 1.01074505e+00 -6.74533546e-01
5.38778007e-01 -5.90112150e-01 3.55958551e-01 -1.30480170e+00
-1.23714216e-01 -1.05378509e+00 2.58316010e-01 1.38784838e+00
5.61297201e-02 -4.45445895e-01 1.14050984e+00 3.62853497e-01
7.56387115e-02 -3.57298017e-01 -6.80235624e-01 -9.61759031e-01
1.55767635e-01 4.40405726e-01 7.44779229e-01 1.03514087e+00
-2.21793111e-02 3.16998929e-01 -8.99292946e-01 -4.17385310e-01
2.13883281e-01 1.40411705e-01 1.03644192e+00 -8.07692945e-01
-4.16432053e-01 -2.95413285e-01 -4.79804903e-01 -1.09200203e+00
-1.98304087e-01 -5.64414442e-01 2.21824273e-02 -1.55496836e+00
9.84537303e-02 -4.16109800e-01 5.25338471e-01 6.58474565e-02
1.48036489e-02 4.50971037e-01 6.84563816e-01 4.63563949e-01
-1.18373886e-01 6.00066066e-01 2.18554306e+00 -1.21534631e-01
-3.76467764e-01 -6.20353930e-02 -1.46686956e-01 7.50786364e-01
6.51205122e-01 -3.26203436e-01 -3.91078323e-01 -3.35259467e-01
-1.39612541e-01 6.11719906e-01 2.12974325e-01 -6.58199430e-01
8.37920681e-02 -4.76688117e-01 4.35016602e-01 -4.02130246e-01
3.11574101e-01 -7.52787650e-01 9.67374742e-01 1.06308877e+00
1.76311269e-01 4.40190822e-01 2.20826611e-01 2.31801495e-01
-1.48388907e-01 -4.81597304e-01 8.05359960e-01 -6.22878432e-01
-8.00584078e-01 1.75265640e-01 -5.98697424e-01 -5.06774247e-01
1.54549396e+00 -4.52345550e-01 -3.27852309e-01 -7.97649205e-01
-7.71970272e-01 -4.03630808e-02 3.62580597e-01 5.09629726e-01
7.43044615e-01 -1.80400050e+00 -7.67052352e-01 -8.83237273e-02
-4.91126865e-01 -1.55884147e-01 3.85259092e-01 6.73391640e-01
-1.46601367e+00 1.21728748e-01 -1.08378685e+00 -2.77189016e-01
-1.57779765e+00 7.19993830e-01 -2.42767651e-02 2.15616032e-01
-8.56292665e-01 4.32627469e-01 6.52431846e-01 -4.40897793e-02
-5.13839126e-01 -1.79841757e-01 -7.97622725e-02 -3.48821998e-01
2.92498738e-01 6.73595965e-01 -6.27756655e-01 -7.19273031e-01
-6.89657256e-02 9.03611124e-01 1.98960438e-01 -3.89356881e-01
1.05611873e+00 -1.64626971e-01 2.84158796e-01 -5.73170744e-02
1.03157949e+00 1.27858832e-01 -1.34727538e+00 5.99588215e-01
-4.24885929e-01 -9.22873795e-01 -4.94112641e-01 -3.04006338e-01
-1.31104910e+00 1.01674139e+00 5.24215758e-01 2.22703680e-01
8.83586586e-01 -1.05916746e-01 1.21301317e+00 -1.38929531e-01
8.64405811e-01 -7.39644825e-01 5.10576367e-01 -2.54549712e-01
1.06983554e+00 -9.35465515e-01 1.07820975e-02 -7.55803287e-01
-1.24405551e+00 1.34988904e+00 7.29376137e-01 -3.96897256e-01
2.01132074e-01 1.65775955e-01 2.57172793e-01 6.31892458e-02
-4.48270053e-01 2.34222323e-01 -1.41012713e-01 8.82545054e-01
2.30607584e-01 -1.33727238e-01 -8.88605177e-01 3.12140644e-01
-1.34706989e-01 3.52263868e-01 8.42502296e-01 1.08391368e+00
-3.17149431e-01 -1.06348681e+00 -6.77180946e-01 -1.34944946e-01
-3.74469131e-01 3.30458462e-01 -1.40789643e-01 1.50062311e+00
2.28228584e-01 5.46406686e-01 2.09366325e-02 -3.70631367e-01
2.36899644e-01 -3.35418075e-01 8.11420858e-01 -1.84517264e-01
-6.56131029e-01 2.20079973e-01 5.13535663e-02 -7.65912384e-02
-4.49608028e-01 -6.26864970e-01 -1.41779625e+00 -6.67933166e-01
-1.87267840e-01 1.83924600e-01 8.69308829e-01 2.56423295e-01
1.53257683e-01 2.96165347e-01 5.58456123e-01 -1.13598037e+00
-3.28637622e-02 -8.35562170e-01 -7.39740431e-01 6.59682393e-01
-2.16193214e-01 -5.94766259e-01 9.47936475e-02 5.06318212e-01] | [10.859989166259766, -0.7282785177230835] |
0da39448-565a-4ac4-8d15-a5ddabbed56a | hierarchical-dynamic-filtering-network-for | 2007.06227 | null | https://arxiv.org/abs/2007.06227v3 | https://arxiv.org/pdf/2007.06227v3.pdf | Hierarchical Dynamic Filtering Network for RGB-D Salient Object Detection | The main purpose of RGB-D salient object detection (SOD) is how to better integrate and utilize cross-modal fusion information. In this paper, we explore these issues from a new perspective. We integrate the features of different modalities through densely connected structures and use their mixed features to generate dynamic filters with receptive fields of different sizes. In the end, we implement a kind of more flexible and efficient multi-scale cross-modal feature processing, i.e. dynamic dilated pyramid module. In order to make the predictions have sharper edges and consistent saliency regions, we design a hybrid enhanced loss function to further optimize the results. This loss function is also validated to be effective in the single-modal RGB SOD task. In terms of six metrics, the proposed method outperforms the existing twelve methods on eight challenging benchmark datasets. A large number of experiments verify the effectiveness of the proposed module and loss function. Our code, model and results are available at \url{https://github.com/lartpang/HDFNet}. | ['Youwei Pang', 'Lihe Zhang', 'Huchuan Lu', 'Xiaoqi Zhao'] | 2020-07-13 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/4959_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123700239.pdf | eccv-2020-8 | ['rgb-d-salient-object-detection', 'thermal-image-segmentation'] | ['computer-vision', 'computer-vision'] | [-3.78330536e-02 -3.74732345e-01 -1.70235254e-03 -4.39927936e-01
-6.67730391e-01 -1.74836919e-01 3.85985345e-01 -9.97509658e-02
-3.12815249e-01 6.54860437e-01 4.31130469e-01 1.73173532e-01
-2.27607340e-02 -7.29346514e-01 -6.45040631e-01 -6.84855938e-01
4.00440633e-01 -4.03624207e-01 9.04671848e-01 -3.12651724e-01
4.22149301e-01 4.50888366e-01 -1.82945418e+00 3.02348018e-01
1.08198190e+00 1.22667706e+00 6.49401128e-01 1.18610084e-01
2.13370204e-01 5.50134003e-01 6.60483614e-02 -2.06132978e-01
4.03720438e-01 -2.00331345e-01 -5.53704560e-01 8.14716518e-02
1.33633465e-01 -2.50972092e-01 -2.98622608e-01 1.23447406e+00
8.13522220e-01 3.13134909e-01 2.65772551e-01 -1.13878715e+00
-6.10455334e-01 3.33188295e-01 -9.58421230e-01 5.04416287e-01
3.75674278e-01 2.40091383e-01 7.52326965e-01 -1.12527394e+00
4.24334824e-01 1.39689696e+00 4.83376622e-01 2.77816266e-01
-9.15073454e-01 -6.44858003e-01 2.25905761e-01 3.90788645e-01
-1.25237155e+00 -4.75619167e-01 1.10105038e+00 -5.11342548e-02
5.05374491e-01 2.52622843e-01 5.27726293e-01 7.33642876e-01
1.46619603e-01 1.04960775e+00 1.41691911e+00 -2.76315063e-01
-1.22345258e-02 2.91886538e-01 6.03239760e-02 9.45429742e-01
1.95413798e-01 -9.83811989e-02 -5.70430279e-01 2.28357166e-02
6.88372314e-01 3.57290804e-01 -4.81576800e-01 -4.42152560e-01
-1.31976855e+00 6.29798949e-01 9.70519006e-01 4.12783742e-01
-4.74410653e-01 -5.33382967e-02 1.65082812e-01 -1.73208445e-01
3.05225313e-01 3.09162587e-02 -2.80572087e-01 2.45342791e-01
-5.43489337e-01 1.34937003e-01 4.81941849e-02 6.69068217e-01
9.39850926e-01 -1.23992652e-01 -3.77974182e-01 8.46943140e-01
5.88152349e-01 4.40359831e-01 6.44261122e-01 -8.21849465e-01
4.45516080e-01 7.63333857e-01 1.53138101e-01 -1.12877059e+00
-5.09965956e-01 -4.58006591e-01 -8.11274409e-01 4.17991251e-01
1.04133479e-01 -6.81391917e-03 -9.34711218e-01 1.56200540e+00
6.74368978e-01 3.25783789e-01 -6.60008490e-02 1.43910706e+00
1.07524085e+00 7.06256211e-01 1.52480066e-01 -4.52351896e-03
1.39431787e+00 -1.11758149e+00 -5.93700826e-01 -2.58958280e-01
1.52773529e-01 -9.15168345e-01 1.19691038e+00 3.76881063e-02
-1.22107756e+00 -8.32038820e-01 -1.13023520e+00 -3.44313055e-01
-3.53169143e-01 4.46139038e-01 6.95321739e-01 1.88898087e-01
-9.21713531e-01 5.13285697e-01 -8.51743639e-01 -3.51022899e-01
5.16018927e-01 2.33967692e-01 -2.85820991e-01 -1.17203314e-03
-1.25014305e+00 8.48578393e-01 5.04775882e-01 2.28645667e-01
-5.38410187e-01 -4.15683985e-01 -8.31038713e-01 1.30688837e-02
2.40047961e-01 -7.28842080e-01 9.52076197e-01 -9.52758670e-01
-1.17806399e+00 6.91331327e-01 -1.54920399e-01 -1.62071049e-01
3.10157597e-01 -2.20359832e-01 -1.52915373e-01 1.74976990e-01
2.33474504e-02 6.15514934e-01 7.96901822e-01 -1.36208093e+00
-9.43576038e-01 -5.61295509e-01 8.51628631e-02 6.23854578e-01
-4.57938284e-01 7.57790059e-02 -4.28258389e-01 -5.73709309e-01
2.72746533e-01 -5.63746989e-01 -1.99874863e-01 8.81040022e-02
-3.43163371e-01 -1.71855554e-01 8.58126521e-01 -7.03063130e-01
1.10338819e+00 -2.29092526e+00 1.25248939e-01 -5.39750187e-03
2.23969996e-01 2.71646470e-01 9.20555368e-02 -1.02439232e-01
1.89225286e-01 -1.00935973e-01 -3.23179573e-01 -3.51585507e-01
-1.36296988e-01 -2.83761829e-01 -1.10143825e-01 4.13884491e-01
3.67080629e-01 9.73827720e-01 -7.70458162e-01 -6.93279862e-01
5.05829215e-01 5.79218507e-01 -3.26106817e-01 1.61876500e-01
1.48147479e-01 3.98418367e-01 -8.36018980e-01 9.01432991e-01
9.93915498e-01 -1.94282174e-01 -5.20407200e-01 -6.01555884e-01
-3.19748670e-01 -7.13562369e-02 -1.43979430e+00 1.84645760e+00
-2.99483120e-01 2.27906764e-01 2.25932058e-02 -7.96069384e-01
1.04658210e+00 -1.92167521e-01 3.20575148e-01 -7.80816197e-01
4.17803735e-01 3.20692539e-01 -2.77699858e-01 -5.30896187e-01
4.64183301e-01 7.65822753e-02 2.20457479e-01 7.02420697e-02
3.35697457e-02 9.39279124e-02 -3.85165364e-02 -2.50184275e-02
7.92889655e-01 2.02379197e-01 2.77367026e-01 -1.64477706e-01
9.04009998e-01 -1.92861244e-01 7.20822155e-01 4.33707386e-01
-4.53931987e-01 6.79982901e-01 6.95946142e-02 -2.29384854e-01
-7.44067729e-01 -1.05958271e+00 -1.17963485e-01 9.99981523e-01
8.80874336e-01 -4.97007146e-02 -3.07190865e-01 -6.09779775e-01
1.73843242e-02 4.00695235e-01 -5.73152423e-01 -2.56707758e-01
-3.85161698e-01 -7.96488643e-01 -4.11070883e-02 5.08211613e-01
1.00032425e+00 -1.18767154e+00 -8.95039260e-01 -3.49221425e-03
-3.29845577e-01 -9.01375592e-01 -3.21194708e-01 -2.47869641e-02
-9.48608100e-01 -7.65016079e-01 -8.74202669e-01 -1.05425858e+00
5.39412022e-01 6.34127915e-01 6.50888622e-01 2.59435084e-02
-2.28231505e-01 1.30263209e-01 -5.34509957e-01 -3.70396078e-01
1.93538740e-01 -3.92053137e-03 -2.28254899e-01 2.59623855e-01
2.14658558e-01 -5.83049476e-01 -1.08913231e+00 3.07877749e-01
-9.78686154e-01 2.80420959e-01 8.93809915e-01 7.28632152e-01
8.16222131e-01 -1.02693647e-01 4.44527090e-01 -2.23570064e-01
5.20522535e-01 -4.74760175e-01 -5.02303481e-01 3.04673284e-01
-2.99395502e-01 -2.95255035e-02 3.52993369e-01 -2.20070288e-01
-1.33376801e+00 2.49197990e-01 -2.27191091e-01 -3.98119718e-01
-1.18150711e-01 2.91872859e-01 -2.20666960e-01 -2.47875899e-01
3.53989154e-01 3.84445578e-01 -1.97987974e-01 -6.49796844e-01
2.26365015e-01 6.22682273e-01 3.40851068e-01 -2.48464525e-01
7.20366299e-01 6.69802487e-01 -9.65877995e-02 -4.29134607e-01
-8.11970770e-01 -4.03742462e-01 -3.44796240e-01 -2.50625908e-01
7.63472676e-01 -1.06839526e+00 -5.22703767e-01 6.12224102e-01
-1.00734687e+00 -4.01367098e-02 -1.05430879e-01 5.74407816e-01
-4.12564844e-01 2.98442423e-01 -4.26975816e-01 -6.95338190e-01
-4.92275029e-01 -1.21966147e+00 1.27979612e+00 8.70018721e-01
5.65505028e-01 -4.80347425e-01 -1.11388192e-01 1.90592960e-01
5.15914261e-01 3.26485395e-01 3.80551547e-01 -1.82063371e-01
-6.83668554e-01 1.27507597e-01 -6.63456619e-01 3.59437734e-01
9.77459401e-02 -8.21524709e-02 -1.05563045e+00 -1.56641871e-01
8.82396549e-02 -3.05840015e-01 1.19644713e+00 4.49784070e-01
1.27311563e+00 2.90222783e-02 -3.75922650e-01 6.91630542e-01
1.74546480e+00 2.54955292e-02 5.95751405e-01 4.48131919e-01
6.14723265e-01 5.21550357e-01 9.68859136e-01 5.43459296e-01
4.29049045e-01 6.18211925e-01 6.47702932e-01 -3.37735027e-01
-2.01630861e-01 -1.15406483e-01 1.99017152e-01 5.05443513e-01
-2.12483197e-01 3.47017869e-02 -7.14311123e-01 5.77959239e-01
-1.97847247e+00 -1.00485146e+00 -6.37566159e-03 1.91679740e+00
7.43817866e-01 1.41139776e-01 1.06213860e-01 5.45655005e-02
8.83684933e-01 1.51933849e-01 -5.24833024e-01 -8.80176425e-02
-3.48732889e-01 7.77771547e-02 3.51977736e-01 3.07787150e-01
-1.33251727e+00 8.12830925e-01 4.45707369e+00 9.48480666e-01
-1.24387169e+00 3.00127000e-01 7.64117241e-01 3.60517465e-02
-4.12895501e-01 1.37460455e-02 -7.78952122e-01 6.53631330e-01
2.67732143e-01 -7.96862692e-02 2.02050954e-01 6.75266325e-01
2.76255399e-01 -4.16902781e-01 -4.69089091e-01 1.03548443e+00
1.31899994e-02 -1.31493938e+00 -2.41028547e-01 -3.97004575e-01
7.18877733e-01 8.56927484e-02 1.80247247e-01 8.21410567e-02
-1.44696519e-01 -5.66815794e-01 9.12968755e-01 8.45615864e-01
2.79354066e-01 -6.17616475e-01 7.81390488e-01 1.56620309e-01
-1.44998789e+00 -3.83793801e-01 -4.77910668e-01 7.96878934e-02
1.85316756e-01 6.60182297e-01 -1.67236924e-01 7.59662926e-01
1.11801267e+00 8.97643447e-01 -8.77934635e-01 1.50094259e+00
-1.20339841e-01 1.36651650e-01 -3.43865365e-01 -1.20469518e-01
1.43917054e-01 2.61211190e-02 5.60452938e-01 1.06669402e+00
4.14077401e-01 2.00638399e-01 5.33717908e-02 9.51228619e-01
8.84971544e-02 3.03010702e-01 -3.62918556e-01 4.54396188e-01
4.78704989e-01 1.56926632e+00 -7.28766143e-01 -1.54430062e-01
-5.46991944e-01 9.82620597e-01 2.99526691e-01 2.72267044e-01
-1.11966658e+00 -5.48507988e-01 3.94814551e-01 2.93000098e-02
4.74458367e-01 -1.42073249e-02 -3.14936519e-01 -1.23292887e+00
3.18321705e-01 -5.14647961e-01 3.58612776e-01 -1.11242700e+00
-1.18911350e+00 6.54350996e-01 -1.41005054e-01 -1.41462779e+00
4.11975384e-01 -4.72195894e-01 -6.32970870e-01 9.66952026e-01
-1.88022566e+00 -1.37604880e+00 -7.13764966e-01 9.08715308e-01
4.04047072e-01 1.48338556e-01 2.00082064e-01 4.35558021e-01
-5.09146512e-01 3.39636773e-01 -6.61734194e-02 -1.50318280e-01
6.45557284e-01 -8.48199248e-01 -1.08342223e-01 1.05617154e+00
-2.88167298e-01 3.43001902e-01 5.82218230e-01 -4.46972460e-01
-1.25718558e+00 -1.11649668e+00 3.32951814e-01 -3.00925709e-02
5.05956590e-01 -1.81919247e-01 -8.34352434e-01 3.27026069e-01
2.25675166e-01 2.07969412e-01 7.13672563e-02 -4.62670028e-01
1.71414413e-03 -3.80943865e-01 -1.28640139e+00 3.90876621e-01
1.01321197e+00 -1.23408422e-01 -6.46174014e-01 5.67708649e-02
8.95101726e-01 -3.99918616e-01 -7.03400731e-01 8.08718622e-01
4.55338806e-01 -1.29748440e+00 1.12047589e+00 -8.96757245e-02
4.73080397e-01 -7.35654593e-01 -2.46523738e-01 -1.01548934e+00
-4.08361495e-01 -2.39352107e-01 -5.01108356e-02 1.33293092e+00
1.89719975e-01 -7.36672461e-01 3.78446400e-01 2.54284143e-01
-2.53655910e-01 -1.12961292e+00 -7.79718816e-01 -3.89416397e-01
-4.23442483e-01 -8.94737989e-02 5.55693507e-01 6.07671618e-01
-2.09535927e-01 5.06939441e-02 -2.25753069e-01 2.56609529e-01
6.27670169e-01 4.83205169e-01 4.28989053e-01 -9.36748743e-01
-4.49156128e-02 -4.79762852e-01 -6.26486897e-01 -9.00974154e-01
-2.05694169e-01 -6.62269175e-01 1.05245262e-02 -1.64476764e+00
4.56838906e-01 -5.08110166e-01 -7.86048114e-01 5.66007853e-01
-4.25560206e-01 5.34429967e-01 2.89103448e-01 1.15266666e-01
-8.10046494e-01 1.02125752e+00 1.44960976e+00 -3.97740677e-02
-2.99196094e-01 -8.40887055e-02 -9.05979753e-01 7.28252053e-01
1.09348631e+00 -1.27317548e-01 -2.14937195e-01 -4.79768544e-01
-2.02541858e-01 -2.36127824e-01 6.71405435e-01 -1.33198047e+00
2.02181235e-01 -2.60603100e-01 7.02808857e-01 -6.68122828e-01
5.13375998e-01 -7.62087941e-01 -1.29490569e-01 3.58492106e-01
-1.66509911e-01 7.85044879e-02 2.94389158e-01 4.23406988e-01
-3.26355636e-01 1.39051855e-01 1.03457510e+00 -1.56391207e-02
-1.15802085e+00 3.18764806e-01 2.73895860e-01 -1.15476839e-01
1.30286324e+00 -2.27870509e-01 -4.80050385e-01 -3.63989882e-02
-5.24309218e-01 3.47748786e-01 5.67780674e-01 5.39282382e-01
1.05439341e+00 -1.51025879e+00 -5.81663013e-01 2.08110541e-01
1.62798956e-01 -1.63793743e-01 5.81959724e-01 1.24920356e+00
-3.00742179e-01 1.26889914e-01 -7.02531397e-01 -5.43756008e-01
-1.12600434e+00 4.77798611e-01 4.07918453e-01 4.41339836e-02
-4.48702335e-01 9.63848889e-01 2.17369258e-01 -1.85195819e-01
1.33533686e-01 -3.01044494e-01 -4.45198715e-01 -1.26750067e-01
5.04681528e-01 3.11805367e-01 -8.08286518e-02 -7.88011789e-01
-5.77644289e-01 6.95640862e-01 1.01681173e-01 8.54137689e-02
1.49154401e+00 -5.09710312e-01 -1.43233582e-01 2.17930600e-01
1.11044133e+00 -1.20781930e-02 -1.35118580e+00 -3.98226768e-01
-3.26705426e-01 -8.14619303e-01 2.00634331e-01 -6.07122123e-01
-1.29112673e+00 7.24947453e-01 1.17422926e+00 7.13530928e-02
1.66424215e+00 1.59175083e-01 8.09588432e-01 -1.01529628e-01
2.45536268e-01 -8.42533708e-01 1.06023975e-01 1.53466716e-01
1.05626273e+00 -1.49513161e+00 9.58388001e-02 -4.49562967e-01
-7.54730344e-01 8.02154839e-01 9.23227072e-01 -3.23327899e-01
8.22652280e-01 -7.14708567e-02 -1.74338683e-01 -1.30958587e-01
-4.18484539e-01 -6.23053372e-01 3.47426087e-01 3.19779962e-01
2.51490593e-01 -1.04529761e-01 -6.07738972e-01 6.11486614e-01
1.49074435e-01 1.03553504e-01 2.83750862e-01 9.52323794e-01
-7.18332350e-01 -7.46415794e-01 -4.30940002e-01 4.14934516e-01
-2.81293780e-01 5.17865904e-02 6.16946407e-02 4.80760425e-01
4.52674985e-01 1.01240218e+00 -2.39931613e-01 -5.73343396e-01
3.99295449e-01 -3.51828963e-01 2.83433020e-01 -2.27648214e-01
-2.98707634e-01 6.22335821e-02 -3.30514163e-01 -7.84504592e-01
-8.44508350e-01 -6.64045036e-01 -1.28268600e+00 -5.29572517e-02
-4.26078647e-01 -1.69666320e-01 5.40708065e-01 6.44227803e-01
4.73507136e-01 6.00839019e-01 8.16790402e-01 -1.06598055e+00
-4.48211819e-01 -8.72120917e-01 -4.85955089e-01 4.64355111e-01
3.11351478e-01 -1.01937556e+00 -2.91485071e-01 -1.65214285e-01] | [9.697980880737305, -0.8338647484779358] |
97ad0800-0712-438a-a894-260742b4e14f | counterfactual-inference-in-duration-models | 1902.08502 | null | http://arxiv.org/abs/1902.08502v1 | http://arxiv.org/pdf/1902.08502v1.pdf | Counterfactual Inference in Duration Models with Random Censoring | We propose a counterfactual Kaplan-Meier estimator that incorporates
exogenous covariates and unobserved heterogeneity of unrestricted
dimensionality in duration models with random censoring. Under some regularity
conditions, we establish the joint weak convergence of the proposed
counterfactual estimator and the unconditional Kaplan-Meier (1958) estimator.
Applying the functional delta method, we make inference on the cumulative
hazard policy effect, that is, the change of duration dependence in response to
a counterfactual policy. We also evaluate the finite sample performance of the
proposed counterfactual estimation method in a Monte Carlo study. | [] | 2019-02-22 | null | null | null | null | ['counterfactual-inference'] | ['miscellaneous'] | [-1.83948740e-01 1.10603869e-01 -8.92396331e-01 -3.54592592e-01
-9.03982997e-01 -2.41538584e-01 4.05383140e-01 3.22674438e-02
-6.11079752e-01 1.71082985e+00 4.92459744e-01 -6.55806303e-01
-2.52241284e-01 -8.00235987e-01 -5.12431264e-01 -3.84814948e-01
-4.89279121e-01 1.47302315e-01 -4.08921838e-01 4.59540278e-01
1.56576574e-01 1.48639977e-01 -1.12417090e+00 -7.27105319e-01
1.27897465e+00 7.46723652e-01 -3.40961874e-01 5.32046616e-01
4.11831826e-01 4.90550548e-01 -2.49007404e-01 -3.72974277e-01
-2.30640188e-01 -3.51256877e-01 -4.07876104e-01 -4.55660047e-04
-9.07443985e-02 -6.93202198e-01 -2.73021609e-01 6.92879260e-01
4.20435131e-01 2.74164557e-01 1.19178617e+00 -1.28825355e+00
-6.68243945e-01 5.55461347e-01 -4.89669472e-01 1.61700547e-01
3.66109461e-01 -9.32596326e-02 4.66206729e-01 -5.52652538e-01
3.69467378e-01 1.16874933e+00 8.57024014e-01 4.54249889e-01
-1.31781852e+00 -6.05958998e-01 -2.80772775e-01 -3.13829333e-01
-1.38626528e+00 -2.10265294e-01 3.33791673e-01 -5.70291400e-01
2.77609050e-01 3.16793583e-02 5.67535639e-01 1.04067791e+00
1.00114882e+00 3.13703477e-01 1.24889994e+00 -5.14292657e-01
6.16276264e-01 7.46683637e-03 3.76612842e-01 3.22895348e-01
9.05274630e-01 8.78750622e-01 9.17486660e-03 -6.72961831e-01
9.80183363e-01 2.48858482e-01 -1.34792373e-01 -2.66589463e-01
-8.82521272e-01 1.07248414e+00 -3.19972396e-01 -5.13031185e-02
-8.12354147e-01 2.85669625e-01 2.77541786e-01 1.42436847e-01
5.48015118e-01 -4.89490688e-01 -4.82291818e-01 1.56799257e-02
-7.71311283e-01 5.97530246e-01 9.71092224e-01 6.10490859e-01
1.35223657e-01 2.00010136e-01 -5.57892621e-01 4.37733412e-01
2.52557397e-01 1.11969805e+00 2.93894231e-01 -1.19429779e+00
1.42596334e-01 -1.16656847e-01 8.89349222e-01 1.85249388e-01
-2.74651468e-01 -4.03913707e-01 -6.44678950e-01 1.61190361e-01
7.13238835e-01 -5.54556072e-01 -5.36757350e-01 2.16399336e+00
3.35620254e-01 6.68314248e-02 4.40075733e-02 1.84041157e-01
-2.04524934e-01 2.66040117e-01 5.35678327e-01 -1.20558977e+00
1.03701389e+00 -1.69293821e-01 -1.14542735e+00 3.08032244e-01
8.45595121e-01 -1.90469250e-01 7.31491208e-01 2.62187868e-01
-1.17145061e+00 -1.13150343e-01 -6.88744009e-01 4.70937073e-01
2.25714952e-01 -2.74147391e-01 4.92514223e-01 9.94838774e-01
-5.19284427e-01 8.78974080e-01 -6.33072317e-01 -2.07624555e-01
1.96849450e-01 -1.03943825e-01 -6.43220842e-02 -5.04458286e-02
-1.27967417e+00 7.11079657e-01 2.37721562e-01 -5.17325938e-01
-1.24849594e+00 -8.52634251e-01 -6.12175643e-01 6.96993992e-02
3.60976130e-01 -1.11620080e+00 1.38187492e+00 -6.02073550e-01
-1.42599034e+00 4.86168355e-01 -1.08629614e-01 -7.07126379e-01
7.39678264e-01 5.70659637e-02 -6.07656717e-01 -1.98368639e-01
5.77655792e-01 -2.36215860e-01 5.36921918e-01 -7.70288050e-01
-5.84672928e-01 -7.98276484e-01 -3.81066680e-01 4.14228030e-02
2.01381817e-01 -2.47029811e-01 4.27919298e-01 -8.58659267e-01
-2.84717232e-01 -6.75515711e-01 -3.38451803e-01 -1.61281198e-01
-1.69523984e-01 -3.55523139e-01 1.49143394e-02 -7.74612188e-01
1.38023174e+00 -2.10705590e+00 -7.06946015e-01 -1.28935531e-01
-3.17839205e-01 -5.66767514e-01 3.17175776e-01 4.21940893e-01
-1.53345495e-01 -3.56233725e-03 -5.24010062e-01 1.15013652e-01
7.83409923e-02 -1.40904114e-01 -3.12580436e-01 8.39148402e-01
-5.52917898e-01 4.44470793e-01 -6.31075323e-01 -4.97649819e-01
-3.96564566e-02 8.53807256e-02 -2.99069703e-01 1.66789204e-01
3.26858819e-01 2.42260709e-01 -3.27519089e-01 3.82986903e-01
7.99778759e-01 -9.81246829e-02 3.70765239e-01 6.33492053e-01
-5.45875490e-01 1.96571708e-01 -8.82770538e-01 1.30717134e+00
-5.31691253e-01 7.86250457e-02 -3.80223036e-01 -9.58503485e-01
5.49426675e-01 7.62347698e-01 2.05085695e-01 -4.64197218e-01
2.39864886e-01 3.62341315e-01 -3.66199404e-01 -4.80923444e-01
-3.20467390e-02 -1.20429766e+00 -5.59806466e-01 5.70671201e-01
-3.29329483e-02 2.64303952e-01 -2.31126353e-01 -9.94845107e-02
7.04782486e-01 -2.65615247e-02 1.21262312e+00 -6.04419649e-01
1.08082466e-01 -4.02243048e-01 5.96433401e-01 1.16780722e+00
-4.06110495e-01 1.11191515e-02 5.53739607e-01 -1.62774503e-01
-8.51448596e-01 -1.81184518e+00 -7.82511771e-01 3.73218417e-01
-3.62950712e-01 7.06764877e-01 -5.21162212e-01 -5.92041790e-01
4.55530554e-01 1.48573148e+00 -8.66623759e-01 -3.59814346e-01
1.51787442e-03 -9.71956730e-01 3.97324383e-01 6.94812596e-01
2.93087631e-01 -5.01201630e-01 -3.96277785e-01 4.73698348e-01
4.81269695e-02 -1.78574890e-01 -4.34047431e-01 -2.93442637e-01
-1.40565062e+00 -9.58587468e-01 -1.18842161e+00 -1.69689536e-01
1.20566180e-02 -2.25995883e-01 8.08849096e-01 -4.17477399e-01
2.19496086e-01 4.13173616e-01 3.46481234e-01 -3.63846719e-01
-2.68185079e-01 -5.53170919e-01 3.60640943e-01 -2.29521751e-01
1.59823194e-01 -5.47725797e-01 -1.13876498e+00 1.32427514e-02
-5.91305315e-01 -6.21630847e-01 1.15966976e-01 8.31192791e-01
1.13864601e-01 -2.33153090e-01 1.25845289e+00 -6.65968657e-01
5.28072894e-01 -8.92160952e-01 -7.92737722e-01 2.79836118e-01
-1.17939937e+00 5.68427481e-02 1.42807007e-01 -7.20806181e-01
-1.70174205e+00 -4.16610539e-01 1.50191829e-01 3.74770850e-01
-9.06970948e-02 6.47926092e-01 -4.88362551e-01 7.35204816e-01
3.95327747e-01 -1.03585392e-01 -7.98170492e-02 -6.30176723e-01
3.76609415e-01 7.40687549e-01 6.91505015e-01 -5.70355594e-01
9.75949597e-03 7.99293876e-01 8.65378305e-02 -3.85795474e-01
-7.59087801e-01 -1.27260020e-04 -1.65152818e-01 1.88296959e-02
7.98013508e-01 -9.28733826e-01 -1.23968053e+00 3.59963685e-01
-8.78855884e-01 -4.20149237e-01 -5.17813146e-01 1.36783302e+00
-1.26301467e+00 3.07918608e-01 -5.13512909e-01 -1.51699471e+00
2.30533257e-02 -3.23514491e-01 3.65360945e-01 2.76422232e-01
-3.52705389e-01 -1.54147172e+00 4.87629801e-01 -1.62700489e-01
-5.27551584e-02 7.20874071e-01 8.65526140e-01 -2.73926646e-01
-5.00910021e-02 -5.79578280e-01 8.71351175e-03 -5.85670173e-02
1.24288417e-01 -2.19103202e-01 -5.17188668e-01 -2.76907176e-01
3.29017371e-01 2.25470051e-01 5.97364068e-01 1.41182840e+00
7.79990554e-01 -7.81758845e-01 -6.42957091e-01 1.46125570e-01
1.69611192e+00 7.10530400e-01 7.20654607e-01 7.18831941e-02
-1.76315978e-01 6.75492764e-01 8.16109478e-01 9.93736923e-01
5.61207473e-01 3.21055233e-01 -1.57682732e-01 3.57250005e-01
5.51234841e-01 -5.79980195e-01 2.14491978e-01 9.08460692e-02
8.01063422e-03 -2.63132691e-01 -6.09015524e-01 1.07383251e+00
-1.56878662e+00 -1.20266974e+00 -2.68147528e-01 3.13315225e+00
8.74945164e-01 2.01924473e-01 5.75356424e-01 2.23042406e-02
9.77518439e-01 -4.08108145e-01 -5.80772817e-01 -4.48242009e-01
6.11448698e-02 4.03810441e-02 9.49458301e-01 6.30312383e-01
-5.98554552e-01 5.96136563e-02 7.83880186e+00 7.12417483e-01
-4.68978554e-01 4.26674187e-01 5.57803571e-01 1.57518551e-01
-4.44666684e-01 6.45668447e-01 -4.82398719e-01 7.69470990e-01
1.54100573e+00 -1.15759540e+00 1.99772734e-02 6.99778736e-01
7.22995281e-01 -6.06434405e-01 -9.59342837e-01 2.86334485e-01
-4.57650006e-01 -8.31066966e-01 -5.31640589e-01 4.64254260e-01
7.68486083e-01 -7.60009110e-01 6.33857846e-02 2.34744132e-01
7.73263752e-01 -6.05300248e-01 7.96370089e-01 1.04117429e+00
1.19045913e+00 -9.61087465e-01 6.95919931e-01 4.60698366e-01
-5.75012743e-01 -2.52758294e-01 -6.65537566e-02 -5.03821611e-01
7.61265337e-01 9.61593211e-01 -3.99479121e-01 2.41646260e-01
1.16045259e-01 7.51819611e-02 3.88075173e-01 1.21143389e+00
-2.17780441e-01 9.34358954e-01 9.52366740e-02 2.46958137e-01
-2.32298627e-01 -1.25891507e-01 3.40511978e-01 5.91169715e-01
7.59683549e-01 3.14420015e-01 -6.14578307e-01 7.62487829e-01
1.73236489e-01 -1.68617845e-01 -8.14646363e-01 2.52132744e-01
8.07851911e-01 2.44085416e-01 -4.35973257e-01 -5.72676063e-01
-6.31134927e-01 6.34099245e-01 -1.88673079e-01 5.89054585e-01
-7.12186038e-01 -3.77184361e-01 2.18202397e-01 2.62576252e-01
2.13648424e-01 2.61851817e-01 -5.59772253e-01 -1.07908916e+00
-2.98462123e-01 -2.22711623e-01 9.67189431e-01 -5.16599596e-01
-1.25253212e+00 -4.12094712e-01 4.65105742e-01 -8.27519894e-01
-5.29252887e-01 -6.97823539e-02 -8.83450329e-01 1.01459491e+00
-9.32624280e-01 -4.46258277e-01 6.12737536e-01 2.27271289e-01
1.39481440e-01 4.23429936e-01 4.42984968e-01 -1.37025407e-02
-4.99856442e-01 4.84168738e-01 5.92301667e-01 -4.37840015e-01
4.45834488e-01 -1.15366256e+00 1.58912197e-01 4.71710801e-01
-8.93816531e-01 7.41198957e-01 9.53972101e-01 -1.09900069e+00
-5.43652296e-01 -7.62705207e-01 1.24008477e+00 -1.87844798e-01
6.46032929e-01 9.42570493e-02 -6.87946081e-01 9.73208129e-01
-5.47279343e-02 -4.43904608e-01 4.99715477e-01 3.02467555e-01
1.05216866e-02 5.98558635e-02 -1.52075744e+00 3.87323320e-01
7.37566590e-01 -3.42873961e-01 -7.20819116e-01 8.87043923e-02
7.11273313e-01 3.42427075e-01 -1.12252915e+00 3.55252236e-01
1.20702922e+00 -8.89946818e-01 8.33781004e-01 -6.92066789e-01
7.61762410e-02 3.69114131e-01 -9.38043892e-02 -1.06873083e+00
-2.13693738e-01 -6.56790614e-01 1.24100521e-01 1.19591355e+00
2.07545340e-01 -1.00417233e+00 1.52707919e-01 7.05314636e-01
3.67083728e-01 -2.66091645e-01 -1.38846028e+00 -1.28425729e+00
5.75624943e-01 -4.38975662e-01 8.33980322e-01 7.61148930e-01
3.70851070e-01 7.93386102e-02 -6.67951405e-01 2.43752133e-02
1.17169797e+00 -3.46422531e-02 2.66539752e-01 -1.15644002e+00
-2.00337067e-01 1.96451381e-01 2.59398013e-01 -4.54188108e-01
2.70710289e-01 -2.07764149e-01 -1.24313861e-01 -1.16128588e+00
6.95954561e-01 3.92890722e-02 -4.05066580e-01 -2.13184059e-01
-3.62486094e-01 -5.53586900e-01 -4.19538796e-01 -1.26079723e-01
-4.85792086e-02 8.49016249e-01 8.23579073e-01 3.69526953e-01
-2.20019951e-01 5.75341463e-01 -4.30987179e-01 6.98484838e-01
6.17446899e-01 -9.36236918e-01 -2.80957848e-01 5.81016660e-01
-9.88945812e-02 1.36692393e+00 6.64690495e-01 -4.76748645e-01
-6.45499408e-01 -6.37751997e-01 1.31055564e-01 -6.22783184e-01
-2.57741243e-01 -3.18501383e-01 5.15642285e-01 9.05753851e-01
-3.60523880e-01 -2.16417998e-01 -7.71740079e-02 9.79240179e-01
3.90941262e-01 -3.31920952e-01 6.86922431e-01 1.62528723e-01
1.25786662e-01 1.73560455e-01 -7.61776030e-01 2.18747571e-01
8.98655832e-01 -1.11000307e-01 -1.25107870e-01 -7.14827180e-01
-9.85318303e-01 1.11931488e-01 6.50838077e-01 -3.32440346e-01
3.31109077e-01 -1.62175262e+00 -6.75131202e-01 -3.31750691e-01
-1.50426570e-02 -1.09648097e+00 8.01489770e-01 9.79898810e-01
1.36085242e-01 5.76620817e-01 -2.06972018e-01 8.99506733e-02
-4.11054224e-01 1.03860521e+00 3.02438796e-01 -2.71844357e-01
-3.36157978e-01 4.07659896e-02 4.67659235e-01 6.53468519e-02
-1.82011425e-01 3.86883914e-02 8.37593600e-02 -1.00506373e-01
5.93104541e-01 1.33625698e+00 -6.25834346e-01 1.18442485e-02
-3.45450610e-01 6.39316114e-03 4.43986565e-01 -9.10268545e-01
8.96401823e-01 -8.86262119e-01 2.11516887e-01 9.76999342e-01
1.15805411e+00 -1.12148233e-01 -1.58999896e+00 1.48626432e-01
2.75731981e-01 -4.95239913e-01 -1.17064804e-01 -6.49553478e-01
-1.23717628e-01 6.29126191e-01 9.74615335e-01 2.58200079e-01
9.39657152e-01 -2.04592068e-02 5.15368879e-01 -5.55340469e-01
3.60579371e-01 -1.05959857e+00 -6.76792264e-01 -2.83614188e-01
5.92995524e-01 -1.04876089e+00 1.04640022e-01 -1.17220730e-01
-2.43854582e-01 4.07987654e-01 -2.08104864e-01 -2.36127719e-01
9.46395695e-01 -4.23533656e-02 -2.63663858e-01 2.23947719e-01
-9.82180297e-01 -7.31518567e-02 8.57965052e-02 8.41676652e-01
6.07176423e-01 7.19923377e-01 -1.59670258e+00 8.04148853e-01
8.71672481e-02 7.53071010e-01 7.35359371e-01 6.02621078e-01
-1.83930457e-01 -6.85331464e-01 -3.90809506e-01 4.87988889e-01
-8.50787461e-01 9.10232738e-02 4.86546546e-01 1.12829459e+00
-2.56271541e-01 9.35710371e-01 2.96025455e-01 6.17902935e-01
2.29621187e-01 2.73880124e-01 4.85927343e-01 -2.38868341e-01
5.16978443e-01 1.62767857e-01 1.48686975e-01 -1.27942458e-01
-3.13296527e-01 -1.05235696e+00 -1.05989373e+00 -5.02268493e-01
-5.01022696e-01 2.53354490e-01 3.43138367e-01 1.17648911e+00
-9.94343087e-02 -8.14109109e-03 1.02813029e+00 -3.83484289e-02
-1.06995714e+00 -9.70798135e-01 -1.27143121e+00 -3.97190824e-02
4.53248978e-01 -7.68764973e-01 -7.85003543e-01 -2.06851527e-01] | [7.959960460662842, 5.197795391082764] |
20af043f-e54c-45cf-99c1-5a3239836049 | machine-translation-with-weakly-paired-1 | null | null | https://aclanthology.org/D19-1446 | https://aclanthology.org/D19-1446.pdf | Machine Translation With Weakly Paired Documents | Neural machine translation, which achieves near human-level performance in some languages, strongly relies on the large amounts of parallel sentences, which hinders its applicability to low-resource language pairs. Recent works explore the possibility of unsupervised machine translation with monolingual data only, leading to much lower accuracy compared with the supervised one. Observing that weakly paired bilingual documents are much easier to collect than bilingual sentences, e.g., from Wikipedia, news websites or books, in this paper, we investigate training translation models with weakly paired bilingual documents. Our approach contains two components. 1) We provide a simple approach to mine implicitly bilingual sentence pairs from document pairs which can then be used as supervised training signals. 2) We leverage the topic consistency of two weakly paired documents and learn the sentence translation model by constraining the word distribution-level alignments. We evaluate our method on weakly paired documents from Wikipedia on six tasks, the widely used WMT16 German$\leftrightarrow$English, WMT13 Spanish$\leftrightarrow$English and WMT16 Romanian$\leftrightarrow$English translation tasks. We obtain 24.1/30.3, 28.1/27.6 and 30.1/27.6 BLEU points separately, outperforming previous results by more than 5 BLEU points in each direction and reducing the gap between unsupervised translation and supervised translation up to 50{\%}. | ['Jian-Huang Lai', 'Tie-Yan Liu', 'Tao Qin', 'Jinhua Zhu', 'Fei Gao', 'Di He', 'Lijun Wu'] | 2019-11-01 | null | null | null | ijcnlp-2019-11 | ['unsupervised-machine-translation'] | ['natural-language-processing'] | [ 1.08087458e-01 -9.58122611e-02 -6.16531789e-01 -3.05194408e-01
-1.32337677e+00 -8.94794822e-01 9.69346881e-01 2.65473928e-02
-7.24266768e-01 1.26625681e+00 2.09769726e-01 -6.77318096e-01
1.71990737e-01 -6.50023639e-01 -1.00059676e+00 -6.47140622e-01
4.27837074e-01 8.05145025e-01 -1.61345720e-01 -4.85994071e-01
1.92298576e-01 -1.41362488e-01 -9.61500466e-01 2.91393220e-01
1.22396398e+00 3.81892622e-01 3.92525345e-01 1.80015087e-01
-3.82120579e-01 1.46219909e-01 -2.51613885e-01 -8.19265962e-01
4.14479584e-01 -7.10701227e-01 -8.12357962e-01 -2.56033510e-01
3.80842447e-01 1.02692433e-02 1.99818030e-01 1.14321554e+00
3.84368449e-01 -2.88205028e-01 7.20801055e-01 -6.77512884e-01
-6.34446323e-01 1.11296523e+00 -8.06852341e-01 3.89288031e-02
3.42916042e-01 -6.46846369e-02 1.14332867e+00 -1.26478219e+00
8.96455884e-01 9.46511030e-01 4.93501663e-01 4.04145271e-01
-1.29151022e+00 -8.78683746e-01 -1.86341405e-02 -1.37863979e-01
-1.37541533e+00 -5.03856957e-01 4.65270877e-01 -4.09967482e-01
1.14042377e+00 2.14337334e-01 2.59057522e-01 1.36716044e+00
2.12540239e-01 4.89311904e-01 1.55803978e+00 -8.39197159e-01
-2.43313536e-01 3.43286306e-01 4.01173420e-02 4.05956954e-01
3.45950663e-01 4.51866910e-02 -7.04269230e-01 -4.09643464e-02
3.91780287e-01 -2.79067785e-01 -2.27542162e-01 -6.89482391e-02
-1.78754425e+00 7.65492260e-01 8.56059268e-02 6.83979154e-01
-8.38376582e-02 -2.74477571e-01 4.56730247e-01 7.14827001e-01
6.90322995e-01 3.23611587e-01 -5.57596505e-01 -1.78797156e-01
-9.60730076e-01 4.56080250e-02 7.18401909e-01 1.36154878e+00
1.08590853e+00 -2.75621027e-01 5.27095646e-02 1.12172306e+00
1.31386548e-01 1.08243692e+00 5.84584057e-01 -4.46798235e-01
1.08380806e+00 2.61382163e-01 1.46105602e-01 -6.83844864e-01
7.28902034e-03 -4.63734776e-01 -9.76834178e-01 -5.61433375e-01
5.54329336e-01 -7.51245245e-02 -5.79856455e-01 1.98629522e+00
-6.37134165e-02 -5.38090527e-01 3.67531449e-01 7.35939384e-01
4.57330823e-01 7.27487743e-01 -2.27553889e-01 -5.57623863e-01
1.37852550e+00 -1.17547858e+00 -5.02429247e-01 -4.88031119e-01
8.35517704e-01 -1.44447482e+00 1.38119364e+00 2.08578050e-01
-1.12846768e+00 -4.39882517e-01 -8.29392731e-01 -2.42689922e-02
-3.44012916e-01 3.74538213e-01 3.60083103e-01 4.35353696e-01
-9.70770359e-01 4.53623205e-01 -7.18666613e-01 -6.12087846e-01
-8.41139071e-03 3.21552843e-01 -4.41125304e-01 -3.74330848e-01
-1.29219711e+00 1.08011806e+00 3.24872911e-01 -5.98794445e-02
-4.95092422e-01 -3.88066113e-01 -7.27758288e-01 -3.30946743e-01
3.07671994e-01 -5.90207160e-01 1.09359181e+00 -1.02564549e+00
-1.47156441e+00 1.16455841e+00 -4.13188547e-01 -3.29527080e-01
6.71194136e-01 -1.84969544e-01 -2.96056032e-01 -1.86190784e-01
4.90129709e-01 6.36903763e-01 5.20047963e-01 -9.72755194e-01
-5.01756847e-01 -3.99585158e-01 -3.11145425e-01 2.93233693e-01
-4.44730520e-01 2.14912936e-01 -5.51806271e-01 -8.55514765e-01
2.61615634e-01 -1.14628911e+00 -2.60140765e-02 -6.73823833e-01
-4.51114953e-01 -1.80028930e-01 1.98305830e-01 -8.62029910e-01
9.92670417e-01 -1.79413331e+00 3.28238547e-01 8.15377533e-02
-1.78850099e-01 -1.77994333e-02 -3.49724770e-01 6.94821596e-01
2.45661646e-01 1.19604550e-01 -5.12721598e-01 -3.52198988e-01
5.24968244e-02 3.51632595e-01 -3.34945351e-01 2.42534772e-01
8.88750777e-02 9.50653732e-01 -8.85030806e-01 -3.91100109e-01
-3.39696109e-01 8.83877724e-02 -2.62297183e-01 -5.21463193e-02
-7.09244907e-02 6.14375293e-01 -1.10526033e-01 5.49920380e-01
6.46463156e-01 6.49536774e-02 6.17408931e-01 9.65785161e-02
-2.73689598e-01 8.25870216e-01 -6.24942005e-01 2.01889038e+00
-7.73306906e-01 5.90547621e-01 -4.03239094e-02 -9.46415186e-01
1.07198775e+00 3.14335376e-01 1.93726301e-01 -9.24150050e-01
-4.09920104e-02 8.17013502e-01 1.20115146e-01 -2.67480046e-01
4.20927644e-01 -4.60577697e-01 -2.70794719e-01 7.54275560e-01
1.47427693e-01 -2.04194397e-01 6.54915214e-01 -4.84654531e-02
7.25231886e-01 2.03689918e-01 3.22668403e-01 -7.33617842e-01
4.70119357e-01 -2.70908177e-02 6.33846402e-01 5.84012628e-01
2.42457092e-01 5.85935950e-01 3.10592443e-01 -1.95770964e-01
-1.34272110e+00 -1.16269505e+00 -1.82507396e-01 1.03269196e+00
-1.40375905e-02 -5.48999786e-01 -7.94041455e-01 -7.77999103e-01
-3.19059461e-01 6.74426794e-01 -2.01978192e-01 8.06332678e-02
-9.29112136e-01 -9.86743331e-01 6.20962262e-01 3.71557921e-01
3.98450941e-01 -7.93168724e-01 6.40401021e-02 6.19505383e-02
-7.43842840e-01 -1.12334955e+00 -6.82781458e-01 2.53713042e-01
-1.01923358e+00 -6.61759615e-01 -9.23481524e-01 -9.60888088e-01
8.10061753e-01 2.25126818e-01 1.47937369e+00 -1.54581651e-01
3.03610265e-01 -2.03793988e-01 -3.13322693e-01 -1.73511103e-01
-5.92640162e-01 5.36690891e-01 4.06638235e-01 -3.79142344e-01
8.17820311e-01 -5.81482887e-01 -2.21562326e-01 4.88275528e-01
-6.50954902e-01 2.54603207e-01 9.91235912e-01 1.16803956e+00
6.43718898e-01 -5.27202606e-01 4.19757456e-01 -8.11775386e-01
4.78928566e-01 -3.94067615e-01 -4.57817137e-01 4.25918907e-01
-8.72551203e-01 2.68075436e-01 7.14705348e-01 -4.10454065e-01
-8.13196063e-01 -2.66407192e-01 -9.37009752e-02 1.32247999e-01
4.90040518e-02 5.94067991e-01 -1.32831544e-01 3.75802845e-01
7.79452145e-01 4.17178124e-01 -8.62682462e-02 -5.42697608e-01
2.34637544e-01 9.16034818e-01 2.76679903e-01 -8.34284961e-01
9.73195016e-01 9.06851217e-02 -4.73258972e-01 -5.83157003e-01
-5.43456614e-01 -3.00982654e-01 -7.92464674e-01 2.58554280e-01
6.01402223e-01 -9.72652614e-01 -5.88354357e-02 1.33598566e-01
-1.22464490e+00 -3.23620737e-01 3.91362794e-02 8.71424973e-01
-5.20930469e-01 3.01859170e-01 -7.60324419e-01 -3.05101424e-01
-5.04066229e-01 -1.19150233e+00 1.07889736e+00 -3.04616153e-01
-4.71956968e-01 -8.09855044e-01 1.89631715e-01 5.44411242e-01
3.02840769e-01 -2.29928657e-01 1.08894122e+00 -6.82218969e-01
-3.85285646e-01 8.77418146e-02 -1.37508795e-01 4.38872039e-01
2.48953596e-01 -3.23605984e-01 -5.21428168e-01 -4.91137803e-01
5.73281161e-02 -3.50572050e-01 7.62949467e-01 -9.61049199e-02
3.15817505e-01 -3.57849330e-01 -2.40504920e-01 2.85461038e-01
1.28473914e+00 -2.18477473e-02 4.31108743e-01 2.83876896e-01
4.36211228e-01 7.62591183e-01 6.06786549e-01 -5.09361997e-02
3.65533262e-01 8.19503367e-01 -2.29960755e-01 2.41880678e-02
-6.04079850e-02 -4.01162207e-01 7.95352399e-01 1.71081388e+00
-1.81053475e-01 -8.13307315e-02 -1.16290259e+00 6.78484380e-01
-1.66829574e+00 -6.03726566e-01 -1.68097094e-01 2.32462525e+00
1.35129237e+00 2.78248817e-01 -5.31399585e-02 -2.22502798e-01
6.78798318e-01 -1.20803460e-01 2.40257606e-02 -4.94496763e-01
-3.54854494e-01 3.92808199e-01 5.14183104e-01 6.51634932e-01
-7.02795088e-01 1.10577655e+00 5.28556108e+00 9.34724271e-01
-1.04596519e+00 3.61220181e-01 5.18950641e-01 -2.41695851e-01
-6.43239796e-01 1.89728484e-01 -8.63216698e-01 6.11049414e-01
1.00575864e+00 -1.60762534e-01 5.45984745e-01 3.78691882e-01
1.68152571e-01 -8.72660428e-02 -1.29502940e+00 9.04842973e-01
1.23635434e-01 -1.09168541e+00 1.24286242e-01 1.25273824e-01
1.11784101e+00 3.22172821e-01 1.04277648e-01 3.28282595e-01
3.24632496e-01 -8.73432279e-01 6.86610758e-01 1.19217321e-01
1.06415761e+00 -6.67733490e-01 8.53566647e-01 6.46184623e-01
-8.23700666e-01 4.27855104e-01 -4.31325406e-01 -7.77705237e-02
7.93437287e-02 7.94894636e-01 -7.44492054e-01 8.55607331e-01
6.86046064e-01 7.32393622e-01 -4.57834750e-01 1.83261767e-01
-5.53620160e-01 6.71538472e-01 -4.52477396e-01 -1.19258963e-01
3.20715994e-01 -7.21579611e-01 4.42161411e-01 1.33415270e+00
7.39895642e-01 -3.61879200e-01 -4.81204409e-03 7.30260372e-01
-4.10956621e-01 5.84772289e-01 -7.87419915e-01 -9.44656357e-02
3.41906875e-01 8.67211044e-01 -6.92534208e-01 -2.99637645e-01
-6.21803403e-01 1.21693087e+00 4.47618544e-01 3.86231273e-01
-7.25934207e-01 -2.91596353e-01 4.22166318e-01 -1.31386453e-02
6.25535622e-02 -4.46508408e-01 -4.05106544e-01 -1.47403491e+00
5.69255173e-01 -1.18565953e+00 -1.97017249e-02 -4.01022643e-01
-1.37683845e+00 9.09061790e-01 -4.99980785e-02 -1.35487211e+00
-6.25323355e-01 -4.64190125e-01 -2.69672543e-01 1.18547535e+00
-1.45065689e+00 -1.15073097e+00 2.31089190e-01 3.28124881e-01
6.12746477e-01 -3.69503647e-01 9.32809889e-01 4.77053881e-01
-3.82185251e-01 7.55649686e-01 4.90672290e-01 1.01889767e-01
1.25881553e+00 -1.10052598e+00 4.74682033e-01 9.11565304e-01
6.45473361e-01 1.04712915e+00 6.09059334e-01 -5.83585143e-01
-1.32843900e+00 -8.04471552e-01 1.87425518e+00 -6.49623454e-01
6.98944151e-01 -8.00523877e-01 -6.06183827e-01 6.45643532e-01
7.25808203e-01 -4.59647655e-01 7.18908072e-01 2.76623696e-01
-4.85314548e-01 -1.98358551e-01 -7.77450979e-01 9.85304296e-01
1.19660234e+00 -7.53684223e-01 -6.25641167e-01 4.81814206e-01
5.98004937e-01 -2.46027172e-01 -7.81529665e-01 4.27743316e-01
4.28871304e-01 -7.13393390e-01 6.92280948e-01 -4.47105944e-01
7.24064648e-01 -6.77022859e-02 -3.88516366e-01 -1.41013372e+00
3.87386531e-02 -6.37711823e-01 3.35081100e-01 1.30829430e+00
1.03312516e+00 -6.82657957e-01 4.98149365e-01 1.64215147e-01
-1.02304772e-01 -6.53913021e-01 -1.06098771e+00 -1.01826465e+00
7.90715277e-01 -2.73039579e-01 4.92716283e-01 1.08399832e+00
1.13836780e-01 6.46741688e-01 -3.31821233e-01 -3.69342715e-01
4.33465928e-01 4.55991238e-01 7.15748370e-01 -7.80734599e-01
-4.05909717e-01 -5.02811968e-01 2.54057407e-01 -1.28117108e+00
3.85862797e-01 -1.32814145e+00 1.20277002e-01 -1.25545430e+00
4.86983985e-01 -4.30827439e-01 -1.65681764e-02 4.29747760e-01
-6.92439079e-02 6.71154022e-01 -1.04375504e-01 4.41554397e-01
-1.22069284e-01 4.76157010e-01 1.04676366e+00 -3.15877020e-01
-1.62721910e-02 -1.94881946e-01 -6.73281133e-01 4.62038696e-01
9.58748460e-01 -5.91128111e-01 -1.31818101e-01 -9.28642631e-01
4.97422218e-01 -2.02749372e-01 -1.22589752e-01 -4.72884059e-01
2.58383676e-02 -1.42672047e-01 1.67687628e-02 -3.61065090e-01
-1.19268537e-01 -5.88178217e-01 -2.84495484e-02 3.77805799e-01
-3.44535828e-01 4.90949929e-01 6.06068224e-02 2.57316738e-01
-5.09276867e-01 -2.62518287e-01 5.37542939e-01 -1.77955419e-01
-2.00307548e-01 -2.16808263e-02 -2.91684777e-01 2.16055885e-01
6.89086080e-01 5.27604073e-02 -1.71588719e-01 -1.68144926e-01
-4.39990669e-01 -4.26081084e-02 6.06349409e-01 6.13500953e-01
2.04410493e-01 -1.53749216e+00 -1.05951011e+00 3.20940346e-01
3.32589477e-01 -2.07270101e-01 -2.49457672e-01 1.30011296e+00
-2.82731533e-01 6.76982343e-01 -9.32404548e-02 -7.34696627e-01
-1.02126372e+00 4.26181227e-01 -1.74177527e-01 -4.38506633e-01
-2.63379574e-01 5.34611225e-01 -3.75489779e-02 -8.46494138e-01
-1.43226460e-01 -2.16706991e-01 3.41412842e-01 -6.72961622e-02
1.21780902e-01 1.59447297e-01 3.60989451e-01 -7.76505888e-01
-2.90249854e-01 7.02344656e-01 -3.55086744e-01 -4.99501675e-01
1.09825516e+00 -2.96509176e-01 -5.92064559e-01 4.71715122e-01
1.32130802e+00 4.42934304e-01 -5.35507619e-01 -4.70396549e-01
3.56858552e-01 -3.92605960e-01 -5.83308697e-01 -8.49751890e-01
-5.85369647e-01 8.89202654e-01 1.91496268e-01 -2.35928565e-01
9.00614679e-01 1.38077691e-01 9.44259644e-01 5.52568495e-01
6.81628168e-01 -1.24346304e+00 -1.92175224e-01 8.00250649e-01
7.43844032e-01 -1.43517888e+00 -2.18197033e-01 -3.58525574e-01
-4.35589284e-01 9.04123068e-01 2.74149448e-01 2.11081102e-01
2.42394775e-01 2.56896973e-01 2.41531342e-01 2.44404852e-01
-6.34366214e-01 2.71298029e-02 3.61522913e-01 3.15496355e-01
8.93266499e-01 1.14708975e-01 -1.10453165e+00 5.81536770e-01
-4.89377111e-01 -3.61077368e-01 1.33379802e-01 7.77091205e-01
-1.01831399e-01 -1.86767232e+00 -3.12496096e-01 2.98897743e-01
-5.17095089e-01 -5.90736389e-01 -5.99326193e-01 7.66200483e-01
2.30786446e-02 9.67782736e-01 -9.99815613e-02 -2.49502376e-01
-2.67533818e-03 3.72347116e-01 6.18584692e-01 -5.87584853e-01
-6.05380416e-01 3.10718387e-01 2.04725087e-01 -2.33299315e-01
-4.70496863e-01 -7.62118280e-01 -9.61122513e-01 -4.12639380e-01
-2.05458581e-01 4.57013816e-01 6.28670216e-01 1.03353775e+00
2.92651236e-01 -1.31788895e-01 4.91920620e-01 -4.48559523e-01
-7.60522425e-01 -1.27552164e+00 -2.01652393e-01 3.60064149e-01
-7.34724924e-02 -3.03701222e-01 -2.42502198e-01 6.73905388e-02] | [11.561674118041992, 10.318110466003418] |
54c9751f-69e4-4124-9d26-a2f5f1e09673 | doad-decoupled-one-stage-action-detection | 2304.00254 | null | https://arxiv.org/abs/2304.00254v2 | https://arxiv.org/pdf/2304.00254v2.pdf | DOAD: Decoupled One Stage Action Detection Network | Localizing people and recognizing their actions from videos is a challenging task towards high-level video understanding. Existing methods are mostly two-stage based, with one stage for person bounding box generation and the other stage for action recognition. However, such two-stage methods are generally with low efficiency. We observe that directly unifying detection and action recognition normally suffers from (i) inferior learning due to different desired properties of context representation for detection and action recognition; (ii) optimization difficulty with insufficient training data. In this work, we present a decoupled one-stage network dubbed DOAD, to mitigate above issues and improve the efficiency for spatio-temporal action detection. To achieve it, we decouple detection and action recognition into two branches. Specifically, one branch focuses on detection representation for actor detection, and the other one for action recognition. For the action branch, we design a transformer-based module (TransPC) to model pairwise relationships between people and context. Different from commonly used vector-based dot product in self-attention, it is built upon a novel matrix-based key and value for Hadamard attention to model person-context information. It not only exploits relationships between person pairs but also takes into account context and relative position information. The results on AVA and UCF101-24 datasets show that our method is competitive with two-stage state-of-the-art methods with significant efficiency improvement. | ['Mike Zheng Show', 'Jiashi Feng', 'Fan Wang', 'Pichao Wang', 'Shuning Chang'] | 2023-04-01 | null | null | null | null | ['action-recognition-in-videos', 'video-understanding'] | ['computer-vision', 'computer-vision'] | [ 2.61747718e-01 -3.81769300e-01 -1.58312708e-01 -1.88118115e-01
-5.94881296e-01 -2.70226926e-01 9.14564192e-01 -7.57224709e-02
-6.07847393e-01 4.01744068e-01 6.15484655e-01 3.68087031e-02
1.76674098e-01 -5.41770816e-01 -4.58290726e-01 -7.90377975e-01
9.14664865e-02 2.55916059e-01 5.11053860e-01 -5.00722080e-02
1.04227163e-01 3.15772563e-01 -1.50510967e+00 5.91891289e-01
6.17240191e-01 9.59291875e-01 -8.71253312e-02 6.91499829e-01
-9.72774774e-02 1.27275681e+00 -5.58970571e-01 -3.94586593e-01
2.99303234e-01 -7.12670624e-01 -7.37163067e-01 3.04214597e-01
5.16410947e-01 -4.82311994e-01 -6.96941495e-01 7.85475075e-01
4.98788029e-01 2.96644241e-01 4.69135821e-01 -1.40891981e+00
-5.11881948e-01 3.37965369e-01 -8.01825345e-01 5.08066833e-01
6.15159094e-01 3.27111900e-01 1.03894532e+00 -1.00221527e+00
3.16500604e-01 1.44730282e+00 6.48196220e-01 7.74103224e-01
-8.74843657e-01 -4.02833253e-01 5.90271175e-01 5.54720342e-01
-1.53474844e+00 -2.93168485e-01 6.81819677e-01 -5.57131350e-01
1.05431235e+00 2.62741894e-01 1.01182151e+00 1.28544176e+00
-2.13623688e-01 1.26703036e+00 7.22417951e-01 -2.40541518e-01
-1.18232127e-02 3.35299447e-02 1.54102147e-01 6.90788746e-01
6.58764094e-02 -1.25213834e-02 -4.50017840e-01 -4.05168906e-02
9.38222408e-01 3.67538989e-01 -2.03111023e-01 -4.13005441e-01
-1.38429034e+00 7.30492473e-01 3.53679150e-01 3.68061811e-01
-3.99577916e-01 2.54546911e-01 4.18170840e-01 -1.10714398e-01
2.54142135e-01 -4.67334986e-02 -5.21590747e-02 -2.54201472e-01
-8.62194479e-01 4.03104782e-01 5.30058444e-01 6.75817132e-01
4.11481827e-01 -1.64211884e-01 -8.22700799e-01 7.19241619e-01
3.83289397e-01 3.87124151e-01 4.65509921e-01 -4.88371849e-01
6.93347871e-01 9.46616471e-01 2.13011399e-01 -1.23153448e+00
-3.14367652e-01 -1.67767927e-01 -8.24331522e-01 -1.51701763e-01
5.95206857e-01 -1.48481250e-01 -8.64149213e-01 1.72146106e+00
4.79984522e-01 6.35305882e-01 -9.35724154e-02 1.20823026e+00
8.75674903e-01 6.15468204e-01 2.29027987e-01 -8.22920725e-02
1.58197832e+00 -1.46044266e+00 -7.05548346e-01 -3.67269009e-01
6.73511863e-01 -3.34158450e-01 7.07470834e-01 1.21477358e-01
-9.34461236e-01 -7.41397917e-01 -8.39047968e-01 -2.56296098e-01
-3.89634252e-01 4.95659381e-01 6.06705844e-01 6.60613239e-01
-8.66532803e-01 2.93951809e-01 -7.36988485e-01 -5.92992961e-01
4.91006255e-01 1.86811149e-01 -5.38192511e-01 4.54543866e-02
-1.18461061e+00 6.84198260e-01 2.58861274e-01 2.21711099e-01
-8.30018222e-01 -3.50165665e-01 -9.59343374e-01 1.27324939e-01
7.17155278e-01 -6.30295277e-01 9.61356640e-01 -1.01666462e+00
-1.36949182e+00 8.17304671e-01 -2.48469874e-01 -5.08181572e-01
6.74949944e-01 -6.01366520e-01 -3.88231367e-01 2.99222976e-01
9.47643071e-02 7.07743883e-01 9.60789800e-01 -8.15268934e-01
-9.18036938e-01 -4.49419409e-01 1.61437318e-01 3.89102995e-01
-5.06592333e-01 3.32176238e-01 -9.39125836e-01 -8.29464495e-01
-1.44689038e-01 -7.58914769e-01 -1.24120437e-01 1.83641061e-01
-2.36339286e-01 -5.21383524e-01 9.25144851e-01 -7.32429028e-01
1.51827991e+00 -2.00173020e+00 2.83375174e-01 -3.50400247e-02
2.12057978e-01 6.93294942e-01 -1.48685634e-01 4.55551952e-01
-1.13637038e-01 -2.37043858e-01 -1.96755901e-02 -5.84537089e-01
-7.51494197e-03 8.57496858e-02 -4.05720584e-02 5.11077464e-01
5.30530155e-01 9.40840960e-01 -9.80048954e-01 -6.11796796e-01
4.42332983e-01 7.21880376e-01 -6.55359447e-01 2.45686010e-01
9.72420797e-02 5.32589078e-01 -4.46738899e-01 7.88764775e-01
4.58356529e-01 -2.88806081e-01 3.14736128e-01 -2.29749620e-01
-1.14917822e-01 5.87880611e-02 -1.58507657e+00 1.54942036e+00
-1.27220258e-01 4.52974707e-01 -3.12119555e-02 -1.16176832e+00
5.67845225e-01 4.25021380e-01 5.83406448e-01 -3.42731595e-01
1.36491761e-01 -1.76293269e-01 -3.21163535e-02 -7.05623388e-01
3.80044818e-01 2.42793709e-01 -4.59939055e-02 2.00428501e-01
3.78828607e-02 7.38068759e-01 3.66243273e-01 3.73975903e-01
1.12675190e+00 3.62116635e-01 5.24134815e-01 2.22708713e-02
1.08160639e+00 -3.61469686e-01 8.16314697e-01 6.52161300e-01
-5.95445156e-01 4.85832185e-01 6.20384693e-01 -6.62217915e-01
-6.74886107e-01 -7.95372069e-01 3.87607753e-01 1.27224720e+00
2.02092037e-01 -7.24813223e-01 -7.55243838e-01 -1.06561887e+00
1.74804479e-02 2.50878036e-01 -9.10774708e-01 -5.67813292e-02
-8.20867479e-01 -6.43665433e-01 5.35903394e-01 1.00317430e+00
9.16462779e-01 -9.75761592e-01 -5.93611360e-01 1.17670037e-01
-3.89578104e-01 -1.25480652e+00 -8.48647058e-01 -4.52882022e-01
-4.20087457e-01 -1.30623567e+00 -1.04914546e+00 -5.98983049e-01
5.43539643e-01 5.46180964e-01 8.71789694e-01 1.28080249e-01
-2.88374245e-01 5.90373218e-01 -4.83581960e-01 -1.45878777e-01
1.04241304e-01 -2.22990647e-01 1.41249016e-01 7.15237260e-01
8.12007904e-01 -3.11499447e-01 -9.79904950e-01 5.90613306e-01
-6.44965231e-01 -8.53219070e-03 8.16424668e-01 7.75590003e-01
3.10827315e-01 -1.34937584e-01 2.21422061e-01 -4.16056871e-01
1.85858473e-01 -3.18796843e-01 -3.06745529e-01 4.49188769e-01
-1.16105989e-01 -2.22342223e-01 3.91224027e-01 -5.08880198e-01
-1.02412319e+00 4.03712243e-01 -2.95675592e-04 -5.61573863e-01
-2.14567065e-01 6.55177573e-04 -3.37441593e-01 9.98865664e-02
3.92957151e-01 4.10745263e-01 -1.16380751e-01 -4.86725271e-01
2.75096685e-01 5.73506415e-01 4.54272538e-01 -3.58262807e-01
6.37095690e-01 6.67850316e-01 -1.85553595e-01 -8.92205417e-01
-8.06152105e-01 -8.43207657e-01 -8.30437064e-01 -4.42383587e-01
1.37271762e+00 -1.11326635e+00 -8.47354591e-01 7.41180718e-01
-1.02179694e+00 -2.21238032e-01 -9.06148031e-02 5.03902853e-01
-3.11183333e-01 5.95231533e-01 -5.89414477e-01 -1.08142757e+00
-1.65673450e-01 -9.87909734e-01 1.35354435e+00 2.61391491e-01
4.40033525e-02 -8.22110355e-01 5.80151007e-02 6.32822812e-01
1.62213415e-01 2.66767174e-01 2.30499059e-01 -6.51055872e-01
-6.68510437e-01 -3.42438757e-01 -4.12778586e-01 2.31159270e-01
7.64085725e-02 -2.39085287e-01 -7.92767227e-01 -3.12803954e-01
-2.45093107e-01 -2.64704913e-01 1.07050037e+00 2.11572304e-01
1.04039490e+00 -2.88035840e-01 -5.39238036e-01 4.06759471e-01
1.03521001e+00 5.22141578e-03 7.89432108e-01 1.65788859e-01
1.05942893e+00 6.03215039e-01 6.60981655e-01 4.75456208e-01
4.14510965e-01 1.13289404e+00 3.56942602e-02 -7.75521547e-02
-2.24823773e-01 -3.18595320e-01 6.68853521e-01 2.96299487e-01
-5.06537139e-01 -2.49862239e-01 -8.79080832e-01 5.65395176e-01
-2.31258345e+00 -1.36632133e+00 -1.38885349e-01 2.11714458e+00
4.19690609e-01 2.76099537e-02 7.90878057e-01 1.09702304e-01
7.70664990e-01 3.96638662e-01 -3.86771172e-01 2.37961248e-01
1.21340370e-02 -3.02032679e-01 1.08315431e-01 2.38384396e-01
-1.65166247e+00 9.72085893e-01 5.43834209e+00 8.09152961e-01
-6.81790650e-01 1.27301604e-01 2.88330376e-01 -3.38601142e-01
4.48712617e-01 -1.92517668e-01 -1.02983129e+00 5.17009199e-01
4.52200979e-01 4.09231842e-01 6.34115040e-02 7.26239383e-01
2.50147283e-01 -5.90755865e-02 -1.24438584e+00 1.27544701e+00
3.57096493e-01 -1.04402757e+00 2.04456359e-01 -1.31090030e-01
4.36206609e-01 -4.61088270e-01 -3.05673689e-01 4.93037254e-01
-4.50836942e-02 -7.73467302e-01 8.11155558e-01 5.39019108e-01
4.22247052e-01 -6.13013506e-01 7.22785950e-01 1.65501669e-01
-1.72591221e+00 -4.65837687e-01 -9.14503783e-02 -2.79103547e-01
4.15125400e-01 1.47434667e-01 -2.05130786e-01 4.25046921e-01
7.19150543e-01 1.12236679e+00 -6.18768692e-01 9.26988959e-01
-4.60414082e-01 3.43454719e-01 -4.27668616e-02 -1.03864456e-02
4.39071327e-01 -2.20674485e-01 6.52302682e-01 1.39996910e+00
5.46902008e-02 2.89381534e-01 5.62955618e-01 5.67556143e-01
2.50528306e-01 2.95264143e-02 -4.85359430e-01 6.26900569e-02
2.55635321e-01 1.24206436e+00 -7.53617465e-01 -6.63744509e-01
-9.45537806e-01 1.24987578e+00 3.54190171e-01 2.53291905e-01
-1.22121346e+00 -2.72531003e-01 7.52629519e-01 1.42554313e-01
6.07223153e-01 -2.29304805e-01 1.80497572e-01 -1.41012907e+00
1.85155541e-01 -9.14447308e-01 7.35259175e-01 -4.09577757e-01
-1.10495961e+00 3.12595874e-01 1.31299093e-01 -1.30688179e+00
-1.15579283e-02 -7.89542019e-01 -6.91837907e-01 5.59786201e-01
-1.16742551e+00 -1.48947573e+00 -3.95064443e-01 8.56406629e-01
7.37411678e-01 -1.82123452e-01 4.89247710e-01 6.10650182e-01
-1.24334478e+00 7.68045127e-01 -2.75647044e-01 6.57728076e-01
6.15745604e-01 -1.29023814e+00 3.82981986e-01 1.16694558e+00
2.05170766e-01 5.51704347e-01 3.51786494e-01 -6.66828990e-01
-1.23831534e+00 -1.17523706e+00 9.13800895e-01 -6.10167623e-01
5.93726277e-01 -6.02967978e-01 -7.54887700e-01 7.33449101e-01
-5.33665232e-02 5.21802679e-02 7.26624548e-01 1.09718844e-01
-4.06564683e-01 -7.28842989e-02 -8.75051260e-01 7.81749666e-01
1.41924930e+00 -4.30850983e-01 -6.88491583e-01 4.20765340e-01
3.18763822e-01 -1.04745500e-01 -5.10790944e-01 2.18056604e-01
6.75768673e-01 -1.15224481e+00 1.23164666e+00 -7.84269273e-01
3.18416059e-01 -4.84027833e-01 -6.92945197e-02 -7.48983681e-01
-5.94971418e-01 -6.39424741e-01 -5.85339069e-01 1.26514518e+00
2.87099238e-02 -4.35681611e-01 8.36742342e-01 5.80667198e-01
1.18820392e-01 -7.86026597e-01 -7.55893290e-01 -8.76010180e-01
-3.14721495e-01 -1.23806693e-01 2.72870153e-01 7.41017759e-01
3.50603610e-02 5.79865575e-01 -8.32828879e-01 1.54215842e-01
4.68225032e-01 -1.48363609e-03 9.09475088e-01 -8.68978083e-01
-4.79836494e-01 -5.19980490e-01 -6.41512752e-01 -1.48486781e+00
-1.75012156e-01 -4.10500288e-01 -5.70939742e-02 -1.46851242e+00
5.91948211e-01 8.01365450e-02 -2.69010872e-01 5.40599287e-01
-5.40291309e-01 3.26285213e-01 2.54788160e-01 2.39246354e-01
-1.04284883e+00 6.18564010e-01 9.10152316e-01 -1.33298645e-02
-2.82941639e-01 3.33283655e-02 -4.61813897e-01 9.75802660e-01
4.72083002e-01 -1.18853286e-01 -3.81300539e-01 -4.26280469e-01
-1.24954626e-01 -1.43174544e-01 8.71851444e-01 -1.31952012e+00
2.55309701e-01 -1.10356510e-01 4.46202248e-01 -7.03732193e-01
5.29725611e-01 -6.66275203e-01 -3.07534933e-01 5.39911509e-01
-2.61005759e-01 9.95614100e-03 -1.88774049e-01 9.04820144e-01
-1.87077269e-01 1.68969214e-01 5.62482357e-01 -2.27337405e-01
-1.15909290e+00 4.16755110e-01 -4.47074324e-01 -4.94365096e-02
1.26623464e+00 -2.95334280e-01 -2.39516601e-01 -4.06810313e-01
-7.65775740e-01 4.15952504e-01 1.03768624e-01 5.20061314e-01
5.33760369e-01 -1.60053790e+00 -8.13329995e-01 1.24406584e-01
1.92441374e-01 -3.29799205e-01 5.26163638e-01 1.21312201e+00
-2.17887521e-01 5.27746201e-01 2.97862422e-02 -5.95542371e-01
-1.52390683e+00 7.55505621e-01 3.32506597e-01 -5.32732725e-01
-7.00928986e-01 1.07608902e+00 4.44270581e-01 -3.96081395e-02
4.94974583e-01 -1.79228246e-01 -5.43875456e-01 2.36456305e-01
1.10933697e+00 5.55364370e-01 -4.42926913e-01 -1.04145110e+00
-6.62188172e-01 6.98819220e-01 -9.97680798e-02 1.40555564e-03
9.80556130e-01 -1.12952784e-01 1.62137523e-01 2.29034692e-01
9.84908283e-01 -2.77191341e-01 -1.57699776e+00 -3.52358937e-01
6.73354650e-03 -7.22998798e-01 -1.42032474e-01 -5.78358054e-01
-1.01830697e+00 9.99943197e-01 5.67724407e-01 2.32267991e-01
1.09497011e+00 -9.11632851e-02 8.76874745e-01 3.09228718e-01
1.79048389e-01 -1.19872344e+00 4.70936000e-01 4.40333575e-01
9.33717072e-01 -1.32332873e+00 8.64319056e-02 -5.23009300e-01
-8.29615712e-01 8.54750335e-01 9.54310179e-01 -3.82277146e-02
4.53259557e-01 -1.50647849e-01 -8.29602256e-02 -2.02073321e-01
-5.36861062e-01 -5.60206234e-01 5.89165688e-01 5.13120234e-01
3.47541928e-01 -1.59421220e-01 -2.40059033e-01 6.75110698e-01
4.25210327e-01 -1.02407010e-02 -5.99365495e-02 9.59122956e-01
-4.30153191e-01 -1.04005945e+00 -4.77323949e-01 2.38571033e-01
-4.55816031e-01 4.75564785e-02 -5.13354421e-01 8.17126930e-01
3.86395007e-01 1.05583632e+00 1.75106972e-01 -4.74020511e-01
4.40492660e-01 1.62608176e-01 4.62939531e-01 -5.76618254e-01
-5.40836453e-01 6.02213368e-02 1.81006372e-01 -9.86566961e-01
-7.34971285e-01 -9.42475200e-01 -8.01615238e-01 -1.09813705e-01
-2.63785452e-01 -1.16048515e-01 -1.61192194e-02 1.05699611e+00
4.35197890e-01 4.74732816e-01 3.92937630e-01 -9.29079592e-01
-4.59366649e-01 -1.01815784e+00 -4.31229532e-01 5.94626606e-01
9.03056636e-02 -9.90835130e-01 -6.56165704e-02 5.84082492e-03] | [8.204155921936035, 0.5377448797225952] |
7e2d6e9c-9b43-4de6-b6c9-5bc6622ee90e | an-artificial-neural-network-functionalized | 2205.10118 | null | https://arxiv.org/abs/2205.10118v1 | https://arxiv.org/pdf/2205.10118v1.pdf | An Artificial Neural Network Functionalized by Evolution | The topology of artificial neural networks has a significant effect on their performance. Characterizing efficient topology is a field of promising research in Artificial Intelligence. However, it is not a trivial task and it is mainly experimented on through convolutional neural networks. We propose a hybrid model which combines the tensor calculus of feed-forward neural networks with Pseudo-Darwinian mechanisms. This allows for finding topologies that are well adapted for elaboration of strategies, control problems or pattern recognition tasks. In particular, the model can provide adapted topologies at early evolutionary stages, and 'structural convergence', which can found applications in robotics, big-data and artificial life. | ['Geoffroy Berthelot', 'Avner Bar-Hen', 'Fabien Furfaro'] | 2022-05-16 | null | null | null | null | ['artificial-life'] | ['miscellaneous'] | [-2.90429771e-01 -2.19560936e-01 -5.93706220e-02 -3.19125921e-01
1.06762934e+00 -2.09912956e-01 7.39541650e-01 9.98054892e-02
-5.56701779e-01 6.30428195e-01 -7.41864145e-02 -1.42840311e-01
-5.39934099e-01 -1.11254573e+00 -6.18482947e-01 -7.82872260e-01
-5.32280862e-01 3.97015959e-01 3.57328445e-01 -6.97166979e-01
4.04911757e-01 7.68536210e-01 -1.61075425e+00 -1.89225703e-01
8.43295574e-01 9.94289339e-01 3.19489241e-01 6.35201752e-01
-6.47175848e-01 6.14538252e-01 -3.78478944e-01 -4.13544595e-01
2.63778150e-01 -4.29324061e-01 -8.01648200e-01 -4.76028532e-01
-2.24881530e-01 6.56956136e-01 -8.55468437e-02 1.25687277e+00
2.31782302e-01 3.12076765e-04 6.60438597e-01 -1.06616151e+00
-7.98135161e-01 9.24196005e-01 9.63098276e-03 1.54665902e-01
-3.83954048e-01 7.40091354e-02 7.01931596e-01 -5.17792463e-01
6.86574817e-01 1.29725981e+00 7.29089439e-01 7.44866073e-01
-9.27044868e-01 -1.65213212e-01 -8.72599334e-02 4.92253423e-01
-1.12151384e+00 -1.25900075e-01 8.27825248e-01 -4.29943800e-01
3.24638933e-01 2.26708248e-01 1.02478898e+00 9.49005485e-01
6.24120653e-01 5.04620135e-01 8.34912419e-01 -3.48150015e-01
7.51276553e-01 2.90365554e-02 -1.57964543e-01 7.75365293e-01
5.23500085e-01 2.22853065e-01 -4.28418696e-01 3.88215303e-01
9.61499870e-01 -5.02371490e-02 -3.47503543e-01 -5.64862072e-01
-1.33722591e+00 6.33403361e-01 1.02234066e+00 8.67862821e-01
-5.22065699e-01 2.16347247e-01 9.97533575e-02 6.59743011e-01
-4.53735329e-02 1.04946685e+00 -6.68402493e-01 -1.17409363e-01
-2.98739880e-01 1.05389230e-01 9.41068888e-01 5.65441549e-01
6.76297843e-01 2.71639258e-01 4.96716768e-01 8.72713804e-01
7.79232308e-02 7.58942440e-02 8.60144198e-01 -8.48177135e-01
-1.65544644e-01 1.15889275e+00 -4.32572901e-01 -1.16998482e+00
-7.86296785e-01 -7.58688331e-01 -1.30844808e+00 3.63324195e-01
4.44323391e-01 -1.56320676e-01 -6.16857409e-01 1.60113835e+00
3.61306846e-01 5.53224683e-02 5.38429879e-02 9.14910138e-01
4.64612454e-01 5.13657391e-01 -1.70319781e-01 -8.28794986e-02
1.02804661e+00 -8.20181310e-01 -4.48488981e-01 2.21901294e-02
5.17684162e-01 -4.84323263e-01 7.47663498e-01 4.50577289e-01
-7.33160198e-01 -3.91880184e-01 -1.05139065e+00 3.55109304e-01
-8.16659629e-01 -3.30710895e-02 9.57974672e-01 5.61672091e-01
-1.04930711e+00 1.14670849e+00 -7.83207119e-01 -8.45102966e-01
3.67661059e-01 4.24728096e-01 -1.91148862e-01 3.22986037e-01
-1.00792933e+00 1.12265038e+00 9.50895488e-01 2.91537672e-01
-5.18603384e-01 -3.10103804e-01 -2.77266413e-01 -4.80848104e-02
1.31268933e-01 -8.72502506e-01 7.46393740e-01 -1.21109247e+00
-1.91568601e+00 2.34771296e-01 5.14540792e-01 -6.54129624e-01
4.29783702e-01 1.85795754e-01 -4.65169787e-01 -2.61943340e-01
-6.00943148e-01 6.54811025e-01 9.85600114e-01 -6.62368000e-01
-4.07116681e-01 -3.37980956e-01 2.92902552e-02 2.05704048e-02
-9.30384338e-01 -5.01897484e-02 1.42362073e-01 -6.75830603e-01
3.12816232e-01 -9.20188248e-01 -5.76192439e-01 2.31081545e-01
-6.03003316e-02 -3.38131160e-01 6.06731355e-01 -1.25113716e-02
9.12156940e-01 -1.80502439e+00 7.21391916e-01 1.98593229e-01
2.03035578e-01 4.41038430e-01 -4.64718379e-02 1.66160107e-01
2.24754661e-01 2.67443508e-01 -1.04434580e-01 4.59917098e-01
-3.07263553e-01 2.78186738e-01 2.22864121e-01 2.98109710e-01
3.92646223e-01 8.26800287e-01 -7.98186302e-01 -4.00799841e-01
2.34541625e-01 3.59823287e-01 -5.50277352e-01 2.28381500e-01
-3.69792014e-01 5.84469438e-01 -5.11113286e-01 4.64348525e-01
3.65592837e-01 -1.36991441e-01 -1.05822109e-01 -2.25384966e-01
-4.72400457e-01 -2.71118492e-01 -1.05730832e+00 1.66037583e+00
-4.42835480e-01 7.26783276e-01 9.08217579e-02 -1.51204562e+00
1.35660040e+00 -3.34636062e-01 3.69749606e-01 -6.19939744e-01
5.55563986e-01 4.13950562e-01 5.86042166e-01 -6.26316309e-01
4.00779158e-01 -1.39024124e-01 2.94720829e-01 2.31992707e-01
2.83337206e-01 -1.43167570e-01 3.87552917e-01 -2.29990616e-01
1.00204694e+00 9.51286331e-02 -4.63878067e-04 -6.65627480e-01
8.06494057e-01 9.30439010e-02 5.82709134e-01 3.35373878e-01
-5.50303869e-02 1.93459421e-01 2.25923926e-01 -1.03766203e+00
-1.28976405e+00 -7.16722548e-01 -2.56010026e-01 8.95044863e-01
2.54880041e-01 -5.43824509e-02 -7.40712166e-01 2.49429587e-02
-1.95888907e-01 2.93375999e-01 -8.02526176e-01 -5.17163098e-01
-6.62722826e-01 -6.59678459e-01 5.78502178e-01 1.66544944e-01
7.54381955e-01 -1.29080355e+00 -9.56882775e-01 2.95680076e-01
3.32335651e-01 -6.79437518e-01 2.01725692e-01 2.82871395e-01
-1.24676418e+00 -9.93158579e-01 -8.89545143e-01 -6.94914818e-01
5.90303838e-01 -2.26514623e-01 1.02650738e+00 4.07476574e-01
-5.41208327e-01 -2.41274863e-01 -5.41649282e-01 -5.06855071e-01
-6.23036027e-01 5.58288217e-01 4.17215049e-01 2.84067124e-01
-4.45179418e-02 -6.90602601e-01 -4.73093331e-01 5.74847579e-01
-8.21517110e-01 -1.80589687e-02 9.85579729e-01 8.98578227e-01
1.73579872e-01 3.45076174e-01 4.70837563e-01 -3.30706775e-01
7.94213951e-01 -4.18811172e-01 -5.45768142e-01 4.34642464e-01
-5.93708575e-01 6.68045104e-01 1.20354712e+00 -5.78591883e-01
-8.45086217e-01 -2.30817452e-01 -1.42635882e-01 -4.05565917e-01
-1.50593311e-01 3.79502147e-01 -6.59820018e-03 -5.36661088e-01
8.32957923e-01 1.77181624e-02 2.26235911e-01 -7.36829102e-01
3.14744174e-01 4.48070884e-01 5.23308396e-01 -6.74605429e-01
6.34730518e-01 8.42669681e-02 4.48158562e-01 -1.14462161e+00
-1.36346474e-01 1.25767753e-01 -7.58005083e-01 -3.86640728e-01
6.46188080e-01 8.33897442e-02 -7.94602334e-01 6.29325807e-01
-1.09248102e+00 -1.14048161e-01 -4.27521288e-01 6.15219831e-01
-2.83657551e-01 -1.93855949e-02 -4.16551977e-01 -5.82075477e-01
-3.69783282e-01 -8.84300709e-01 3.74212675e-02 7.26920545e-01
1.64044201e-01 -1.23850143e+00 2.26151630e-01 -5.85592687e-01
9.09793615e-01 1.66577816e-01 1.05878973e+00 -6.46632910e-01
-4.43699062e-01 9.90883559e-02 -5.04111201e-02 3.53876501e-01
-2.07498252e-01 3.79909486e-01 -4.47843850e-01 -8.69275630e-03
-1.33347034e-01 -1.99135020e-01 7.99569607e-01 2.94670910e-01
1.06223178e+00 -3.39560956e-01 -2.14373156e-01 8.12189043e-01
1.36592591e+00 1.56851128e-01 5.67243218e-01 4.72787499e-01
4.71662134e-01 9.94107723e-01 9.63905156e-02 2.21926779e-01
-2.43761973e-03 4.39180464e-01 7.62714505e-01 1.54443160e-01
-6.27188310e-02 -2.42294911e-02 5.34712896e-02 1.34283376e+00
-4.48841065e-01 1.94711596e-01 -1.03960967e+00 3.21167886e-01
-1.77056789e+00 -8.62885714e-01 -2.16656581e-01 2.09703541e+00
4.11653310e-01 2.00572789e-01 -8.00249353e-02 1.80394083e-01
8.22668910e-01 -3.23464632e-01 -6.39374733e-01 -8.07947934e-01
-4.30905074e-01 3.62018287e-01 4.27073747e-01 -5.66519983e-02
-7.50036657e-01 7.65706897e-01 6.32111692e+00 5.62372148e-01
-1.60126305e+00 -9.57954526e-02 3.81478548e-01 2.05548629e-01
-1.66738138e-01 1.11286035e-02 -3.42596114e-01 6.43598437e-01
8.72321784e-01 -1.38019383e-01 6.99295640e-01 9.48168159e-01
-2.75603347e-02 9.93126929e-02 -7.91973233e-01 8.47493470e-01
-3.54968011e-01 -1.59617794e+00 7.39349499e-02 1.07810266e-01
6.61545575e-01 1.54070750e-01 -1.42378911e-01 1.27151310e-01
1.78818733e-01 -1.05421638e+00 6.76331580e-01 7.39794731e-01
2.67441928e-01 -7.55785048e-01 7.18081713e-01 4.23828989e-01
-1.06592226e+00 -5.42207241e-01 -9.39555168e-01 -2.75973529e-01
-1.38363048e-01 6.91663206e-01 -5.89884996e-01 4.53714639e-01
9.25552905e-01 6.76424861e-01 -7.42365062e-01 1.66114688e+00
-4.04419489e-02 4.37491059e-01 -4.97083575e-01 -9.92128253e-01
1.48458391e-01 -3.75457108e-01 7.28949130e-01 8.32510948e-01
3.84086937e-01 -3.15015048e-01 -1.19894400e-01 1.15323091e+00
-5.94783910e-02 4.69733119e-01 -6.57287836e-01 -3.12090397e-01
4.57145125e-01 1.39728582e+00 -1.16163242e+00 1.67322680e-02
2.07132280e-01 4.74568874e-01 3.67310554e-01 -1.07917495e-01
-4.95627761e-01 -7.06916094e-01 5.59146643e-01 -2.13771641e-01
1.47697583e-01 -3.62182558e-01 -5.49166918e-01 -1.22330534e+00
-3.24571729e-01 -5.34631431e-01 -8.59467611e-02 -2.80244410e-01
-1.02203298e+00 6.98588312e-01 -3.19886684e-01 -1.12112141e+00
-1.89921126e-01 -9.61973548e-01 -9.71556365e-01 2.64911205e-01
-9.32973087e-01 -8.35965395e-01 -4.10082966e-01 5.96449256e-01
3.40752751e-01 -5.14655471e-01 7.29787827e-01 2.12140918e-01
-8.98454309e-01 3.15006763e-01 3.25795740e-01 -9.53150615e-02
8.44547376e-02 -1.12323952e+00 3.25147867e-01 8.23132932e-01
1.73930153e-01 6.07616901e-01 9.09371674e-01 -2.22728103e-01
-1.78664684e+00 -9.30681884e-01 4.95154738e-01 2.62182534e-01
7.32611001e-01 4.40137535e-02 -1.01266336e+00 -1.23466909e-01
6.39803112e-02 -9.69126299e-02 4.23073955e-02 1.75042287e-01
1.61806911e-01 -4.14647341e-01 -9.76163030e-01 7.51301706e-01
1.15151179e+00 2.92051613e-01 -4.34818506e-01 4.19664681e-02
7.06966102e-01 5.18620647e-02 -9.82975543e-01 4.85637814e-01
7.42305338e-01 -9.01002884e-01 7.77174711e-01 -7.44247317e-01
2.79496551e-01 -2.04333067e-01 2.74203151e-01 -1.54479527e+00
-4.61029619e-01 -7.50775337e-01 4.40018289e-02 1.11569643e+00
4.37694043e-01 -8.22140694e-01 7.68557727e-01 2.73589671e-01
-2.22441822e-01 -8.21545184e-01 -8.02582085e-01 -8.98135185e-01
2.88345367e-01 -4.39656526e-02 8.21872771e-01 9.67130363e-01
-2.24334270e-01 1.98190644e-01 -3.83745693e-02 -4.86663848e-01
5.14333785e-01 -1.52138323e-01 5.76253355e-01 -1.73954594e+00
-1.72524795e-01 -1.35206556e+00 -7.68876016e-01 -4.62950081e-01
-4.62389514e-02 -8.09804678e-01 -4.06120270e-02 -1.13616300e+00
-4.60969865e-01 -9.18756723e-01 -3.77355754e-01 2.32388273e-01
4.46734548e-01 1.03601115e-02 -7.04165176e-02 3.73138696e-01
-2.32272312e-01 6.87059522e-01 1.17451823e+00 -4.53366116e-02
-3.19084346e-01 3.14265266e-02 -3.01831871e-01 7.31584013e-01
1.23735738e+00 -2.67366678e-01 -2.03846872e-01 -5.17143309e-01
5.29675186e-01 -4.57171351e-01 1.69944763e-01 -1.36435926e+00
5.15731573e-01 -4.56598788e-01 2.17633158e-01 1.31366014e-01
-3.29031013e-02 -7.66007245e-01 4.01107192e-01 1.03123438e+00
-4.40170705e-01 3.58830541e-01 -3.57008666e-01 3.82017255e-01
-6.53709937e-03 -5.78096569e-01 9.58154142e-01 -4.08413857e-01
-7.63087153e-01 2.65023053e-01 -2.30839252e-01 -1.12337627e-01
1.12074053e+00 -2.69416600e-01 -2.66033202e-01 2.06269532e-01
-4.64799762e-01 -4.19555679e-02 6.48574948e-01 6.78246081e-01
5.36262572e-01 -1.24146712e+00 -5.24511993e-01 2.38761768e-01
-2.14015376e-02 -1.07485101e-01 -2.85681933e-01 8.53616357e-01
-1.04180670e+00 1.52525082e-01 -8.41876090e-01 -8.03848505e-01
-6.96524918e-01 5.42671561e-01 6.48804545e-01 2.91345119e-01
-4.22731310e-01 8.79743218e-01 -4.16209042e-01 -6.01498544e-01
2.27935791e-01 -3.10905039e-01 -4.80100334e-01 -3.17014575e-01
3.17464560e-01 5.08226395e-01 1.56373959e-02 -5.22900105e-01
-3.24309468e-01 6.43336713e-01 2.59864181e-01 1.48301199e-01
1.49100471e+00 1.68879256e-01 -6.10712767e-01 5.26472807e-01
1.06226420e+00 -6.46564543e-01 -6.58835232e-01 -4.60771332e-03
3.76845270e-01 -1.85124099e-01 1.73966978e-02 -5.28713286e-01
-1.19880247e+00 1.15250111e+00 6.33522809e-01 5.06898522e-01
1.01631439e+00 -2.13714808e-01 5.06989181e-01 8.07359755e-01
6.71484053e-01 -1.19157100e+00 2.65330553e-01 5.59704363e-01
9.64486778e-01 -8.02851617e-01 -2.44661346e-01 1.47502884e-01
-1.08264580e-01 1.68580592e+00 6.61598921e-01 -3.66618365e-01
9.47944403e-01 2.23226830e-01 -2.71639615e-01 3.63319740e-02
-8.59593928e-01 -1.30292460e-01 2.29780659e-01 5.41922987e-01
3.01731914e-01 1.65141135e-01 -7.24749804e-01 1.56879246e-01
-5.28895915e-01 -3.75640355e-02 4.63722974e-01 7.99350381e-01
-8.29657078e-01 -1.16609609e+00 -2.37801522e-01 4.14614081e-01
-2.61984646e-01 3.07616413e-01 -4.87510085e-01 4.71438736e-01
3.62492710e-01 4.63473797e-01 -3.43630128e-02 -8.52562547e-01
-1.02606546e-02 -3.21978003e-01 5.29170632e-01 -5.12682386e-02
-5.87173104e-01 -5.79879165e-01 -2.28991464e-01 -3.12941283e-01
-3.98314953e-01 -2.85260320e-01 -9.89586413e-01 -5.77547908e-01
-5.63212752e-01 4.22397643e-01 1.47879982e+00 8.36147666e-01
4.35266942e-01 4.18496996e-01 8.60565364e-01 -8.43152165e-01
-5.10187507e-01 -9.73781407e-01 -3.49157363e-01 1.93011090e-01
-1.79353043e-01 -9.42197680e-01 -1.37817070e-01 -1.78144172e-01] | [8.262261390686035, 3.159212112426758] |
d680a63d-1925-418c-bfba-5ee66721fd8c | morphological-disambiguation-from-stemming | 2011.05504 | null | https://arxiv.org/abs/2011.05504v1 | https://arxiv.org/pdf/2011.05504v1.pdf | Morphological Disambiguation from Stemming Data | Morphological analysis and disambiguation is an important task and a crucial preprocessing step in natural language processing of morphologically rich languages. Kinyarwanda, a morphologically rich language, currently lacks tools for automated morphological analysis. While linguistically curated finite state tools can be easily developed for morphological analysis, the morphological richness of the language allows many ambiguous analyses to be produced, requiring effective disambiguation. In this paper, we propose learning to morphologically disambiguate Kinyarwanda verbal forms from a new stemming dataset collected through crowd-sourcing. Using feature engineering and a feed-forward neural network based classifier, we achieve about 89% non-contextualized disambiguation accuracy. Our experiments reveal that inflectional properties of stems and morpheme association rules are the most discriminative features for disambiguation. | ['Antoine Nzeyimana'] | 2020-11-11 | null | https://aclanthology.org/2020.coling-main.409 | https://aclanthology.org/2020.coling-main.409.pdf | coling-2020-8 | ['morphological-disambiguation'] | ['natural-language-processing'] | [ 3.86421680e-02 -5.13724983e-01 -1.05902873e-01 -2.14451328e-01
-6.15541577e-01 -1.25583160e+00 2.00837493e-01 7.34320402e-01
-8.62336457e-01 6.29985809e-01 3.40583652e-01 -7.42461503e-01
-2.64335126e-01 -8.68802607e-01 3.82355670e-03 -4.45967406e-01
1.31574243e-01 3.92833829e-01 -1.52955219e-01 -4.81400609e-01
3.78613681e-01 7.92300344e-01 -1.12435722e+00 6.61971122e-02
1.36456561e+00 3.07119101e-01 3.83967906e-01 3.57073367e-01
-7.74475813e-01 3.02154988e-01 -7.81910956e-01 -5.95457435e-01
3.30082148e-01 -1.63656369e-01 -1.04687405e+00 -8.50511268e-02
3.55958819e-01 1.37776792e-01 1.71025932e-01 1.12870336e+00
4.33767319e-01 3.75662118e-01 7.43284881e-01 -4.00817543e-01
-1.08718002e+00 1.23159659e+00 -3.27866375e-01 5.48551619e-01
4.91299093e-01 2.01411575e-01 1.18520880e+00 -1.12928593e+00
7.26498067e-01 1.28678656e+00 4.66767669e-01 3.84084731e-01
-1.04164076e+00 -6.18790984e-01 2.32047498e-01 1.45879284e-01
-1.60224414e+00 -4.30564940e-01 8.55904460e-01 -3.69607478e-01
1.22672915e+00 2.29586475e-02 5.23448050e-01 4.04823869e-01
-9.59811807e-02 5.19342601e-01 1.11044335e+00 -8.42920184e-01
2.49193758e-01 -2.42673784e-01 1.91891715e-01 6.46689475e-01
4.45455432e-01 -2.82111257e-01 -4.52880263e-01 -1.16218530e-01
5.82726538e-01 -3.63523364e-01 7.85112530e-02 4.31581676e-01
-8.71765733e-01 8.82312059e-01 1.01934515e-01 9.29759800e-01
-1.89010471e-01 -4.50498968e-01 3.74169827e-01 2.61976153e-01
3.88891846e-02 8.55907261e-01 -5.89288116e-01 6.96640536e-02
-8.80332053e-01 -4.31655981e-02 5.94087005e-01 6.94928229e-01
8.87779117e-01 5.20369112e-01 3.76986891e-01 1.14645302e+00
1.13629408e-01 7.83894241e-01 8.81781399e-01 -6.06593311e-01
3.36186647e-01 9.06375170e-01 -2.44705826e-01 -1.03716433e+00
-4.34854507e-01 -1.32733593e-02 -3.13546151e-01 -1.21969571e-02
7.94911742e-01 -2.66290158e-01 -8.09459567e-01 1.60464609e+00
3.99668068e-01 -8.32851231e-01 1.78733408e-01 5.88566899e-01
7.52919316e-01 4.50140774e-01 5.52106142e-01 -3.52841169e-01
1.54693019e+00 -2.66975403e-01 -7.69710422e-01 -3.30222875e-01
5.84570050e-01 -8.31715107e-01 1.35917532e+00 4.43496495e-01
-1.20089638e+00 -2.23677441e-01 -1.09816682e+00 -2.60352075e-01
-8.74995112e-01 2.89196204e-02 7.57225692e-01 8.81588101e-01
-6.76228523e-01 3.29843879e-01 -6.52270257e-01 -3.98048460e-01
2.39986926e-01 4.06845093e-01 -7.49574602e-01 1.58535227e-01
-1.11619496e+00 1.07001781e+00 7.66569853e-01 1.15349917e-02
-3.83012667e-02 -3.48714411e-01 -1.17780781e+00 -2.59809107e-01
3.31140995e-01 -8.58819410e-02 1.07610381e+00 -9.09907639e-01
-1.18246484e+00 1.08072710e+00 -3.47302973e-01 -2.24262848e-01
-1.88180953e-01 -1.02755614e-01 -7.49408185e-01 2.69316465e-01
2.23884687e-01 3.59721035e-01 5.95939159e-01 -9.47690606e-01
-7.34473765e-01 -5.18681109e-01 -3.84269267e-01 1.35125369e-01
-4.61303800e-01 8.60966563e-01 1.89775333e-01 -1.07724047e+00
3.23715448e-01 -5.10073066e-01 -2.21838892e-01 -7.71322548e-01
-1.35739267e-01 -5.50091565e-01 4.49707091e-01 -1.00008237e+00
1.39954627e+00 -1.86233091e+00 -1.98521093e-01 2.68456668e-01
8.82256180e-02 4.85882223e-01 -4.97743636e-02 4.17361319e-01
6.54071644e-02 5.61618090e-01 -3.81734014e-01 2.05422208e-01
6.42492175e-02 2.94500053e-01 -1.89097986e-01 2.57824451e-01
5.59208095e-01 1.01253712e+00 -1.11316442e+00 -7.06717610e-01
2.55934894e-01 1.56214103e-01 -1.53198734e-01 -1.09771386e-01
5.09707108e-02 8.33358616e-02 -2.24170804e-01 1.02820277e+00
4.76406455e-01 4.16178167e-01 5.49508154e-01 2.35729650e-01
-3.04654032e-01 5.31581461e-01 -1.25240910e+00 1.49151754e+00
-6.55624926e-01 3.42672020e-01 -9.38210823e-03 -6.63977146e-01
1.04520535e+00 4.54878658e-02 -8.58157352e-02 -5.59924424e-01
5.18938303e-01 5.34524202e-01 3.25708359e-01 -3.72200370e-01
9.26013172e-01 -3.53220224e-01 -5.96744955e-01 5.44955969e-01
1.02492787e-01 -3.51779282e-01 6.15552306e-01 5.59388772e-02
9.06475842e-01 -1.81939691e-01 1.05028307e+00 -4.55380470e-01
5.65161824e-01 3.89264673e-01 9.83018041e-01 3.91809076e-01
-2.44181439e-01 2.49336734e-01 -2.09615335e-01 -4.65481907e-01
-5.77966869e-01 -1.26156735e+00 -2.28128210e-01 1.52755523e+00
-3.92815918e-01 -4.20114696e-01 -7.28918016e-01 -6.49262011e-01
-1.18883975e-01 7.56855786e-01 -2.08109424e-01 2.03951932e-02
-1.05312490e+00 -5.89437306e-01 8.89559031e-01 7.02207506e-01
7.75950924e-02 -1.38386667e+00 -4.86105412e-01 4.41564381e-01
-1.96361784e-02 -1.11233175e+00 -3.47906560e-01 4.89052951e-01
-6.07360959e-01 -9.81359959e-01 -6.80210814e-02 -1.24375296e+00
5.54010749e-01 1.26779631e-01 9.62861657e-01 2.14634061e-01
-3.95431280e-01 -3.92746106e-02 -4.18509990e-01 -5.83746076e-01
-6.46145701e-01 1.15996905e-01 3.92106086e-01 -4.73499894e-01
9.44216967e-01 -5.18587530e-01 6.10802993e-02 -1.24601014e-01
-1.02870357e+00 -6.33718908e-01 4.63773996e-01 4.40819710e-01
5.26925147e-01 2.06206262e-01 7.83126235e-01 -7.94710577e-01
6.94740415e-01 -1.70093372e-01 -5.68990111e-01 2.01108918e-01
-2.60336041e-01 1.56615913e-01 1.00842273e+00 -6.22540832e-01
-1.13497424e+00 4.16117281e-01 -2.04810709e-01 2.88655519e-01
-6.80741906e-01 6.69625163e-01 -7.13986039e-01 6.33349568e-02
9.01765108e-01 4.99191694e-02 -3.99701118e-01 -6.09869242e-01
5.71891963e-01 7.26140380e-01 8.67015302e-01 -8.30032170e-01
1.04194677e+00 2.68240511e-01 -3.13670367e-01 -1.19682753e+00
-4.46667135e-01 -4.00213361e-01 -1.14419281e+00 3.62123586e-02
8.30945849e-01 -6.24727011e-01 -4.14322972e-01 3.08976859e-01
-9.83416915e-01 -2.58223657e-02 -3.08503032e-01 2.98766792e-01
-9.15441364e-02 5.65792084e-01 -6.82067990e-01 -7.95487881e-01
-4.61731255e-01 -6.99699521e-01 5.79276145e-01 4.11563426e-01
-7.39985585e-01 -1.07895362e+00 4.87461165e-02 2.68026561e-01
1.90209672e-01 2.86011070e-01 1.47129762e+00 -1.17586935e+00
7.64050633e-02 7.55269360e-03 7.76382461e-02 2.28477456e-02
6.45084858e-01 2.51570135e-01 -7.25014031e-01 2.16702551e-01
-1.84017643e-01 -2.03173786e-01 4.91568059e-01 1.69011086e-01
3.17272902e-01 -3.30236435e-01 5.35880774e-02 3.97635043e-01
1.20817745e+00 3.34078312e-01 1.74595892e-01 4.22117054e-01
6.25915885e-01 8.70220900e-01 4.85061347e-01 2.28333101e-01
4.07118350e-01 9.72351208e-02 -3.45456600e-01 3.99462789e-01
-2.03277662e-01 -2.63586044e-01 5.00562429e-01 1.22285759e+00
2.92486161e-01 2.06959382e-01 -1.53351593e+00 8.69953334e-01
-1.31130219e+00 -6.13992095e-01 -1.94644690e-01 1.89515293e+00
1.32477605e+00 -2.14950517e-01 1.26302183e-01 4.96945620e-01
8.93583059e-01 -1.99689537e-01 -1.97198614e-01 -7.05434740e-01
-6.68378174e-01 8.45959008e-01 1.34545520e-01 6.13427997e-01
-1.08547318e+00 1.79497612e+00 6.72910976e+00 8.98082972e-01
-1.05971706e+00 -2.16007113e-01 2.55440306e-02 7.54682422e-02
-4.53825742e-01 -7.29407091e-03 -8.09587061e-01 1.24792330e-01
6.44749999e-01 -3.63089085e-01 4.21489000e-01 5.54077804e-01
2.60979772e-01 -2.62206048e-01 -8.05793107e-01 1.01026201e+00
1.16078593e-01 -1.01196802e+00 3.54356080e-01 -1.90407813e-01
4.30658966e-01 -3.30700070e-01 -7.76485726e-02 6.13679029e-02
8.68783951e-01 -1.25351536e+00 7.85352826e-01 -1.09750830e-01
7.63775468e-01 -1.13529241e+00 3.02852720e-01 2.21655384e-01
-1.41729665e+00 -9.01958272e-02 -4.90097523e-01 -4.06869113e-01
1.73332736e-01 6.08771861e-01 -8.01882446e-01 3.40066820e-01
4.13226485e-01 2.13944212e-01 -9.59935069e-01 7.62500346e-01
-7.80438781e-01 8.29104662e-01 -3.24523151e-01 -2.64660686e-01
1.48092046e-01 -2.84373969e-01 6.82656467e-01 1.59730446e+00
3.05688214e-02 1.89614728e-01 3.17357242e-01 5.41593969e-01
9.53835472e-02 9.11496103e-01 -6.33575559e-01 -2.49774083e-01
9.24222112e-01 1.44286025e+00 -1.29660165e+00 -1.37609884e-01
-1.11034378e-01 6.19256377e-01 6.71894550e-01 2.03519836e-01
-2.20042571e-01 -7.85036683e-01 7.33431697e-01 -1.02538720e-01
2.74914593e-01 -7.77216375e-01 -7.14919329e-01 -1.07611036e+00
-1.68515012e-01 -1.04277670e+00 7.33680189e-01 -3.09486449e-01
-1.46349382e+00 4.72692311e-01 -1.33533791e-01 -6.66420519e-01
-4.37243372e-01 -8.78078401e-01 -5.32073557e-01 8.69171143e-01
-1.16746235e+00 -1.16969442e+00 3.59278738e-01 4.86567080e-01
4.86715049e-01 -3.64551783e-01 9.84230459e-01 -9.25372541e-02
-5.91050804e-01 7.37400711e-01 -3.24436545e-01 7.19559968e-01
6.05556428e-01 -1.40444541e+00 3.08213532e-01 1.34967816e+00
6.27470493e-01 1.02251637e+00 5.68672776e-01 -9.08287108e-01
-1.41479170e+00 -1.01765788e+00 1.38965642e+00 -3.20168167e-01
1.05809152e+00 -3.45084995e-01 -1.03584468e+00 4.06879395e-01
2.69945920e-01 -3.67641151e-01 1.40240455e+00 2.16536164e-01
-5.97319484e-01 3.58344823e-01 -1.22249746e+00 1.01892292e+00
1.15415299e+00 -6.79541886e-01 -1.43418705e+00 -1.20618068e-01
5.05028784e-01 1.09373517e-01 -7.19315469e-01 -1.19081117e-01
2.39361688e-01 -3.28342885e-01 6.12694979e-01 -8.14152122e-01
4.42769416e-02 -5.19523561e-01 -2.50108391e-01 -1.42732286e+00
-5.30047596e-01 -9.92514253e-01 2.97107697e-01 1.56003356e+00
6.25296235e-01 -3.66739541e-01 6.48013711e-01 6.47609174e-01
2.94453028e-04 -1.74764562e-02 -7.55182445e-01 -9.83811498e-01
6.92673624e-01 -7.77899265e-01 7.69478738e-01 1.21924329e+00
5.59963167e-01 5.92738748e-01 4.41612035e-01 6.65122317e-03
3.44429165e-01 1.97783992e-01 1.85381070e-01 -1.28559864e+00
-5.59926741e-02 -8.10763776e-01 -4.38967019e-01 -2.64926940e-01
6.25755727e-01 -1.25827575e+00 -2.35839561e-02 -1.38689518e+00
-2.52536565e-01 -3.59156579e-01 6.68020099e-02 9.00689662e-01
-4.11234438e-01 1.59713790e-01 2.51941890e-01 -1.99539244e-01
-2.32191682e-01 7.50901550e-02 6.36797309e-01 -1.47874475e-01
-7.09584355e-01 -5.99475503e-01 -1.03581595e+00 1.15362298e+00
1.07816637e+00 -5.70290327e-01 8.93334821e-02 -7.15935946e-01
5.50180852e-01 -5.66825867e-01 -3.86592597e-01 -6.51530266e-01
3.08826476e-01 -6.46310508e-01 4.57072347e-01 -4.55902308e-01
-1.04802892e-01 -5.77937901e-01 -2.14702681e-01 3.03868264e-01
-2.93947384e-02 5.15632093e-01 3.70912164e-01 -4.96487552e-03
-4.16049846e-02 -6.53155744e-01 6.84903979e-01 -3.07105243e-01
-8.46819639e-01 -6.79539284e-04 -1.06103432e+00 5.15337288e-01
8.33939970e-01 -5.06728470e-01 -1.23018347e-01 1.47692695e-01
-5.00740767e-01 -7.71473497e-02 7.31132150e-01 3.08497250e-01
6.93798423e-01 -1.24320507e+00 -8.35597157e-01 1.71582550e-01
8.90439004e-02 -1.46504164e-01 -2.06567213e-01 1.22838974e-01
-5.36423922e-01 -6.90436969e-03 -2.90413529e-01 1.56829685e-01
-1.24154329e+00 6.07922912e-01 -1.45456299e-01 -9.87993181e-02
-3.08536589e-01 7.34893501e-01 -3.29549670e-01 -6.52721703e-01
-1.34721667e-01 -3.36435854e-01 -5.51137090e-01 3.94272238e-01
7.30826020e-01 4.52008963e-01 3.91014159e-01 -1.16705358e+00
-5.22688866e-01 4.57462668e-01 -3.52389812e-02 -4.81815219e-01
1.20131087e+00 -1.02149889e-01 -4.18146223e-01 3.56577724e-01
7.40372002e-01 9.58973467e-01 -2.87159175e-01 -2.04038650e-01
7.19027400e-01 -3.01096350e-01 -2.34875694e-01 -9.11281228e-01
-5.31681836e-01 8.42090607e-01 -9.74638164e-02 9.26987082e-02
9.66948271e-01 -4.95079644e-02 9.36808765e-01 7.38560557e-01
3.24106693e-01 -1.63406348e+00 -4.53400880e-01 1.26926768e+00
4.78858650e-01 -8.28536093e-01 -2.60080457e-01 -4.11285996e-01
-5.87116897e-01 1.05963778e+00 4.31698054e-01 -7.26976544e-02
5.65486431e-01 6.97362781e-01 4.53616738e-01 1.02720872e-01
-3.03088009e-01 -7.02682137e-01 1.61753476e-01 1.00750148e+00
6.38859868e-01 2.12824151e-01 -7.00623870e-01 1.12427437e+00
-9.96638775e-01 -8.11870337e-01 4.51175243e-01 1.09345746e+00
-7.00987101e-01 -1.23416245e+00 -7.52230108e-01 3.57691497e-01
-7.91824162e-01 -4.79319721e-01 -1.03496397e+00 6.79375112e-01
2.14459628e-01 1.38799870e+00 -7.01299915e-03 -1.81167647e-01
2.06010751e-02 4.97960627e-01 6.25441313e-01 -8.98836195e-01
-1.13796496e+00 -5.46109267e-02 2.18203798e-01 3.14517096e-02
-2.57531069e-02 -8.30970883e-01 -1.92080319e+00 -2.25031957e-01
-2.95765251e-01 2.31596097e-01 4.98785287e-01 1.20028031e+00
-6.96965382e-02 -3.55195440e-02 2.62173444e-01 -4.43658531e-01
-2.54361987e-01 -8.58677745e-01 -6.67360365e-01 4.55620825e-01
-2.96541005e-01 -4.36341971e-01 -2.57503152e-01 1.92430362e-01] | [10.410239219665527, 10.100655555725098] |
f93fa78d-74d6-4fc5-aacf-b75bf6abf8d2 | conceptual-cognitive-maps-formation-with | 2307.01577 | null | https://arxiv.org/abs/2307.01577v1 | https://arxiv.org/pdf/2307.01577v1.pdf | Conceptual Cognitive Maps Formation with Neural Successor Networks and Word Embeddings | The human brain possesses the extraordinary capability to contextualize the information it receives from our environment. The entorhinal-hippocampal plays a critical role in this function, as it is deeply engaged in memory processing and constructing cognitive maps using place and grid cells. Comprehending and leveraging this ability could significantly augment the field of artificial intelligence. The multi-scale successor representation serves as a good model for the functionality of place and grid cells and has already shown promise in this role. Here, we introduce a model that employs successor representations and neural networks, along with word embedding vectors, to construct a cognitive map of three separate concepts. The network adeptly learns two different scaled maps and situates new information in proximity to related pre-existing representations. The dispersion of information across the cognitive map varies according to its scale - either being heavily concentrated, resulting in the formation of the three concepts, or spread evenly throughout the map. We suggest that our model could potentially improve current AI models by providing multi-modal context information to any input, based on a similarity metric for the input and pre-existing knowledge representations. | ['Patrick Krauss', 'Andreas Maier', 'Achim Schilling', 'Paul Stoewer'] | 2023-07-04 | null | null | null | null | ['word-embeddings'] | ['methodology'] | [ 3.24151888e-02 1.18083939e-01 1.52868360e-01 -1.24584645e-01
-3.31537239e-02 -6.03347778e-01 1.06720614e+00 8.01305950e-01
-5.74299276e-01 3.36059988e-01 8.21853101e-01 -2.57230937e-01
-5.20902097e-01 -1.22878122e+00 -4.02930915e-01 -5.71035445e-01
-3.08983058e-01 5.64753413e-01 4.78131622e-01 -5.62116683e-01
8.15259039e-01 4.81641263e-01 -1.51587605e+00 3.84759694e-01
5.47159195e-01 5.54179668e-01 7.01822639e-01 3.12060118e-01
-3.83105636e-01 7.13126779e-01 -5.06825745e-01 1.88148543e-02
-2.49587782e-02 -3.74101996e-01 -8.80340993e-01 -3.66278172e-01
1.63780227e-02 9.50371921e-02 -2.16792092e-01 7.12619007e-01
1.71271980e-01 4.59170640e-01 7.89048016e-01 -7.44569957e-01
-1.10495615e+00 4.12365377e-01 -5.59712127e-02 7.33973026e-01
3.45868707e-01 -2.66337544e-01 9.97436881e-01 -1.02959859e+00
6.52351856e-01 1.11783171e+00 6.31433308e-01 2.74251014e-01
-1.41259706e+00 -3.56486768e-01 1.75567716e-01 4.01275992e-01
-1.48145878e+00 -1.79396808e-01 7.19237685e-01 -5.71958601e-01
1.23901284e+00 -1.73746124e-01 1.29404151e+00 7.16304123e-01
6.45878136e-01 3.69945496e-01 1.34557462e+00 -6.51708245e-01
5.05671322e-01 1.86020002e-01 6.04482507e-03 4.54997301e-01
8.95970836e-02 1.72071800e-01 -9.49518442e-01 -1.02574304e-01
1.05324841e+00 3.78902823e-01 -1.62772983e-01 -4.89944369e-01
-1.34419274e+00 6.80639327e-01 1.06420422e+00 9.01151836e-01
-5.48499763e-01 1.14410967e-02 1.14321016e-01 1.92989424e-01
1.35099694e-01 9.47860420e-01 -1.58530772e-01 2.16058135e-01
-8.57100964e-01 1.91627920e-01 3.04611892e-01 4.76599604e-01
9.38239753e-01 -2.08765998e-01 2.67127864e-02 7.98098743e-01
3.93143177e-01 4.56819013e-02 1.03482354e+00 -7.18848825e-01
3.12512070e-01 7.78937042e-01 -2.88089097e-01 -1.32540715e+00
-5.25446892e-01 -4.48825747e-01 -3.52732748e-01 2.48052344e-01
1.07485831e-01 4.79063570e-01 -7.46096492e-01 1.92676032e+00
-3.10420822e-02 4.98721488e-02 8.00816715e-02 6.47456884e-01
3.18182081e-01 6.53894365e-01 4.25823212e-01 3.37492645e-01
1.39440584e+00 -5.88212311e-01 -4.09560531e-01 -7.87924528e-01
3.92137200e-01 -2.22947299e-01 8.83861363e-01 5.12995524e-03
-1.06659627e+00 -5.34538805e-01 -1.47140801e+00 -2.90428400e-01
-1.26022589e+00 -6.03768289e-01 6.74036443e-01 2.44589880e-01
-1.72929776e+00 5.37215889e-01 -4.66703832e-01 -8.02518666e-01
5.26522338e-01 2.24387363e-01 -6.88494384e-01 -8.55510309e-02
-1.39559102e+00 1.47404516e+00 6.69301748e-01 -8.71301144e-02
-4.40828055e-01 -4.94788438e-01 -9.26966667e-01 3.36152971e-01
-2.11661622e-01 -8.13086987e-01 7.01113164e-01 -3.88833404e-01
-9.37349737e-01 9.31025267e-01 -1.28198862e-01 -2.14652568e-01
-4.98186320e-01 1.83476046e-01 -3.62623066e-01 2.19537869e-01
2.08242327e-01 1.01830697e+00 4.43875968e-01 -1.26942515e+00
-6.56918406e-01 -7.71160066e-01 1.16631307e-01 4.86550689e-01
-4.02567863e-01 -2.48584822e-01 2.27852035e-02 -7.81773567e-01
7.42628574e-01 -6.81115091e-01 -1.56881079e-01 -2.04163380e-02
3.76412719e-01 -2.18040541e-01 2.67301857e-01 -5.17002940e-01
1.17044234e+00 -2.25903988e+00 2.78600365e-01 4.63321686e-01
4.99618113e-01 -1.84956223e-01 -5.81775457e-02 7.32388079e-01
-9.22726654e-03 7.19031766e-02 -1.98757544e-01 2.78372228e-01
8.17481205e-02 2.65870661e-01 -4.27598089e-01 1.30112022e-01
1.25847399e-01 1.15358162e+00 -1.19615352e+00 -7.56367296e-02
-6.61279075e-03 5.22046745e-01 -4.72371936e-01 -1.22543715e-01
8.80239680e-02 6.10522032e-02 -1.29631251e-01 3.22133094e-01
3.32730502e-01 -7.81005993e-02 4.32999939e-01 1.74938023e-01
-2.51793653e-01 4.26000923e-01 -7.70539582e-01 1.85993302e+00
-4.69668299e-01 9.32559311e-01 -1.56325325e-01 -9.06258941e-01
1.15060699e+00 2.21248925e-01 -4.91481572e-02 -1.15646696e+00
6.05068542e-02 1.64951235e-02 2.09595993e-01 -1.38918355e-01
6.03902876e-01 -4.47824746e-01 -9.88711938e-02 8.38113844e-01
3.68832439e-01 -9.56064276e-03 4.18182798e-02 3.72923523e-01
1.08489871e+00 -1.13212757e-01 6.35349333e-01 -5.97587645e-01
1.59932449e-01 -2.34204978e-01 5.10044135e-02 4.59999919e-01
-2.21404895e-01 3.74828160e-01 3.21229815e-01 -5.25713265e-01
-8.76924992e-01 -1.59816456e+00 -1.47197455e-01 1.49368262e+00
1.38307020e-01 -4.76532817e-01 -3.90910715e-01 9.26290601e-02
7.14733005e-02 8.49787116e-01 -1.14545524e+00 -6.40821874e-01
-3.89612466e-01 -4.33054537e-01 3.43890548e-01 9.63120878e-01
3.26725811e-01 -1.45170307e+00 -9.46733654e-01 3.80325139e-01
-1.79520175e-02 -3.52966040e-01 9.27675068e-02 7.00714111e-01
-9.22096729e-01 -8.57694030e-01 -4.46471959e-01 -1.19894516e+00
7.85727143e-01 4.45909590e-01 1.15063095e+00 3.90322134e-02
-2.36493066e-01 4.45098042e-01 -1.95475891e-01 -4.05697674e-01
-3.22239473e-02 -2.90547628e-02 -2.69143078e-02 -4.61026728e-01
6.64411306e-01 -9.88814235e-01 -5.80708146e-01 1.03366099e-01
-1.19836152e+00 -8.42274129e-02 4.85717028e-01 7.07878530e-01
4.68820244e-01 -8.89383629e-02 8.84229183e-01 -6.60184145e-01
1.13429475e+00 -1.03064537e+00 9.68407914e-02 1.22315429e-01
-6.51435077e-01 2.54060417e-01 2.47450218e-01 -1.31403387e-01
-8.51529479e-01 -4.03755128e-01 1.26988336e-01 2.07879797e-01
-1.28063992e-01 8.95460129e-01 6.36504591e-02 2.26559509e-02
9.07562256e-01 5.01209557e-01 1.15923025e-02 -2.69770175e-01
6.18456364e-01 1.70454770e-01 6.07559323e-01 -7.78524697e-01
4.29590285e-01 3.88660610e-01 -2.00288415e-01 -6.87524140e-01
-6.02542639e-01 -2.81974196e-01 -1.04471374e+00 -5.23214266e-02
9.64677155e-01 -5.97175121e-01 -3.19386929e-01 4.12736423e-02
-9.52925265e-01 -3.03948075e-01 -4.29554075e-01 3.26937646e-01
-7.21949518e-01 -2.44349062e-01 -5.44624388e-01 -2.84671962e-01
1.53207541e-01 -6.16287351e-01 6.09686971e-01 1.62516624e-01
-6.92960680e-01 -1.38543594e+00 4.39758033e-01 -2.08393991e-01
6.98187470e-01 3.44454981e-02 1.43494856e+00 -7.15290487e-01
-4.86639053e-01 -8.80555660e-02 -1.61995247e-01 -3.69972855e-01
-6.68901857e-03 -8.00110161e-01 -9.27483678e-01 1.26929516e-02
6.63968250e-02 -2.86849380e-01 1.23171806e+00 -2.93160081e-02
8.50058615e-01 -1.99463591e-01 -4.43962187e-01 3.14600170e-01
1.40255845e+00 4.32287276e-01 7.47525454e-01 8.00356746e-01
2.02846751e-01 7.96236634e-01 7.12948069e-02 1.40289992e-01
7.43428588e-01 2.31418520e-01 9.00579169e-02 4.05190766e-01
9.25018415e-02 -4.25872177e-01 6.97759613e-02 6.36244357e-01
-5.28726494e-03 2.84545273e-01 -1.20529294e+00 8.76073778e-01
-1.68765557e+00 -1.21152031e+00 7.27901816e-01 1.89399028e+00
9.26404178e-01 1.01086266e-01 -3.35810542e-01 1.58995554e-01
3.94079119e-01 4.53689992e-01 -3.22255582e-01 -6.41536117e-01
-4.81746607e-02 3.77851456e-01 -1.48200095e-01 3.57634813e-01
-6.76479578e-01 9.79425371e-01 7.17001104e+00 3.34304184e-01
-9.11814094e-01 7.05752708e-03 1.73155621e-01 2.27706414e-02
-4.32785243e-01 -1.80568099e-01 -4.97455955e-01 1.59422889e-01
1.10855818e+00 -3.95790637e-01 7.53206670e-01 4.57744002e-01
-3.76017600e-01 -3.71724516e-01 -1.18282390e+00 7.20556200e-01
4.95903373e-01 -1.57088661e+00 2.08080366e-01 3.26662213e-02
4.68307972e-01 -5.29188551e-02 2.87330925e-01 4.60711807e-01
2.13750616e-01 -1.21926773e+00 7.82623887e-01 7.46072829e-01
5.39347112e-01 -5.51632702e-01 2.86142439e-01 3.53542030e-01
-1.26920128e+00 -3.11254680e-01 -5.27848065e-01 -4.35778975e-01
1.35737732e-02 -6.53404668e-02 -7.44460285e-01 -1.67489186e-01
6.06454372e-01 6.08629823e-01 -8.88135135e-01 9.26163971e-01
-2.10929871e-01 -3.76044996e-02 -1.00784503e-01 -1.12035543e-01
1.20619789e-01 4.64856513e-02 2.30112717e-01 1.02993584e+00
5.27568400e-01 4.21060979e-01 6.08087219e-02 1.00942135e+00
1.65445581e-01 7.76129737e-02 -9.52562928e-01 2.29734983e-02
9.03755307e-01 9.31587040e-01 -1.08804202e+00 -3.45064640e-01
-1.73181087e-01 8.13113093e-01 6.79406166e-01 4.45532411e-01
-1.77924484e-01 -3.59730691e-01 6.02794349e-01 2.89447606e-01
2.82552689e-01 -6.18808925e-01 -5.42729735e-01 -6.22004688e-01
-1.70061156e-01 -1.62240043e-01 2.40540341e-01 -9.91042376e-01
-1.14698327e+00 8.57364476e-01 -8.77298564e-02 -6.70020342e-01
-3.21178198e-01 -6.97128594e-01 -5.30750871e-01 1.17266250e+00
-1.17927098e+00 -1.12969184e+00 -1.86683387e-01 6.66405976e-01
2.25775957e-01 -3.42593014e-01 1.31364977e+00 -1.53817326e-01
1.69935316e-01 2.83019960e-01 9.11735520e-02 -3.99903208e-02
5.71326911e-01 -1.25349987e+00 4.46930438e-01 3.60185325e-01
4.79001194e-01 1.27612650e+00 3.16345602e-01 -6.19266689e-01
-8.78861725e-01 -8.98503780e-01 1.07734656e+00 -9.42943931e-01
7.76010931e-01 -5.00056744e-01 -1.05533683e+00 8.33378434e-01
2.19082326e-01 -1.93442613e-01 1.09864867e+00 3.60214293e-01
-7.58165121e-01 3.67126465e-02 -1.04034185e+00 7.06350386e-01
9.83899593e-01 -9.71549094e-01 -1.49141479e+00 -1.17866583e-01
6.21382296e-01 8.04140493e-02 -8.00566792e-01 -1.99463233e-01
6.01377130e-01 -1.03453064e+00 1.11619449e+00 -4.61809486e-01
4.57737684e-01 -4.16358322e-01 -4.42916662e-01 -1.81236506e+00
-1.11901236e+00 3.61771107e-01 -7.71738365e-02 8.78621697e-01
4.81302589e-01 -8.52311134e-01 4.94622588e-01 5.63174605e-01
-1.50429457e-01 -7.70185292e-01 -8.21514785e-01 -5.14508665e-01
3.11095566e-01 -2.13183016e-01 6.80787444e-01 1.03816676e+00
6.15774810e-01 4.40613776e-01 4.37102169e-01 -1.56730860e-01
2.10247636e-01 -9.33837891e-02 3.71132642e-02 -1.46673965e+00
5.17993495e-02 -6.73506379e-01 -7.54683852e-01 -6.46078348e-01
7.52754286e-02 -1.46585166e+00 -7.64590949e-02 -1.83583510e+00
2.08403856e-01 -4.24379766e-01 -7.98848629e-01 6.65172160e-01
1.47945434e-01 3.07027191e-01 4.50400859e-01 2.21053451e-01
-4.08641785e-01 3.96157056e-01 9.03972208e-01 3.09176464e-02
-1.52630329e-01 -6.52839839e-01 -1.33606291e+00 6.94305062e-01
9.23267663e-01 -2.14451313e-01 -5.89192152e-01 -5.24105608e-01
5.02491236e-01 -3.50868940e-01 3.95458728e-01 -1.40555012e+00
5.66501975e-01 5.26366867e-02 1.07453752e+00 -2.12012380e-01
4.49444741e-01 -9.17935133e-01 -4.09576064e-03 2.77119130e-01
-5.53196132e-01 4.99077588e-01 3.67571890e-01 6.67306185e-01
-3.39521676e-01 -7.42168427e-02 5.25851786e-01 -4.90377516e-01
-1.14941978e+00 -1.62019297e-01 -7.62927055e-01 7.19813779e-02
9.82267201e-01 -5.96978664e-01 -4.97097999e-01 -6.08185492e-02
-1.04107928e+00 4.68667550e-03 4.93028164e-01 6.91601813e-01
7.95412481e-01 -1.61412144e+00 -2.31018409e-01 5.91665328e-01
3.77257317e-01 -5.02359092e-01 6.03252538e-02 2.78402656e-01
-3.93179804e-01 5.86149752e-01 -8.58086169e-01 -2.16038287e-01
-4.15480226e-01 6.92242086e-01 7.58389384e-02 -2.22688280e-02
-4.93174314e-01 8.19764256e-01 3.33620608e-01 -3.83187503e-01
-1.48706034e-01 -2.20925301e-01 -7.69098639e-01 4.55481976e-01
8.87315214e-01 2.79059470e-01 4.23665792e-02 -6.58138990e-01
-3.58152032e-01 5.08058965e-01 -1.43208444e-01 -4.27671969e-01
1.42693019e+00 -1.65023088e-01 -5.34333646e-01 7.96861470e-01
9.54654276e-01 -1.73413083e-01 -8.66673052e-01 -2.92554915e-01
2.78575361e-01 -4.68099982e-01 -1.87651157e-01 -9.21902776e-01
-4.13450330e-01 1.02864194e+00 3.80129486e-01 9.73108709e-02
9.44227338e-01 3.15924972e-01 4.24167931e-01 4.96987075e-01
7.04911172e-01 -1.12575805e+00 3.98980439e-01 9.51658487e-01
9.99461710e-01 -6.31828845e-01 -1.36017054e-01 2.11741403e-01
-4.65332031e-01 1.05783284e+00 7.18793809e-01 -4.34928596e-01
8.68199646e-01 -1.12760533e-02 -9.88567919e-02 -4.08534706e-01
-7.61791408e-01 -8.11746120e-02 3.87606084e-01 8.58985007e-01
6.05612278e-01 1.37399763e-01 -1.07338101e-01 7.11523950e-01
-5.06150782e-01 -3.18984598e-01 2.90346950e-01 1.14119637e+00
-1.00198209e+00 -9.20241416e-01 -4.43463862e-01 5.23218751e-01
1.70911744e-01 -3.62953931e-01 -3.70082468e-01 5.94666481e-01
2.73507118e-01 5.03977835e-01 5.13103604e-01 -4.20297831e-01
9.00992900e-02 6.00022078e-01 5.60224354e-01 -8.58631313e-01
-4.00950640e-01 -3.72054189e-01 -3.78551394e-01 -5.02871871e-01
-1.82374880e-01 -6.53966248e-01 -1.59859383e+00 -1.68034911e-01
3.33111882e-01 9.97705534e-02 5.18090129e-01 9.28455651e-01
3.76004249e-01 4.78670418e-01 2.17205342e-02 -1.17387867e+00
-3.56664089e-03 -8.06987405e-01 -7.64626622e-01 2.35341266e-01
1.48205966e-01 -9.85121667e-01 2.21481896e-03 -8.67976174e-02] | [10.599699974060059, 2.381809949874878] |
a6ec2097-58ae-4b9e-b716-1555648beea1 | tuning-free-plug-and-play-hyperspectral-image | 2211.15307 | null | https://arxiv.org/abs/2211.15307v2 | https://arxiv.org/pdf/2211.15307v2.pdf | Tuning-free Plug-and-Play Hyperspectral Image Deconvolution with Deep Priors | Deconvolution is a widely used strategy to mitigate the blurring and noisy degradation of hyperspectral images~(HSI) generated by the acquisition devices. This issue is usually addressed by solving an ill-posed inverse problem. While investigating proper image priors can enhance the deconvolution performance, it is not trivial to handcraft a powerful regularizer and to set the regularization parameters. To address these issues, in this paper we introduce a tuning-free Plug-and-Play (PnP) algorithm for HSI deconvolution. Specifically, we use the alternating direction method of multipliers (ADMM) to decompose the optimization problem into two iterative sub-problems. A flexible blind 3D denoising network (B3DDN) is designed to learn deep priors and to solve the denoising sub-problem with different noise levels. A measure of 3D residual whiteness is then investigated to adjust the penalty parameters when solving the quadratic sub-problems, as well as a stopping criterion. Experimental results on both simulated and real-world data with ground-truth demonstrate the superiority of the proposed method. | ['Cédric Richard', 'Jie Chen', 'Xiuheng Wang'] | 2022-11-28 | null | null | null | null | ['image-deconvolution'] | ['computer-vision'] | [ 0.3860587 -0.49093235 0.39198965 -0.20951803 -0.6081874 -0.3630014
0.17625694 -0.65835184 -0.4316466 0.8157039 0.1338201 -0.24655177
-0.3628284 -0.4292173 -0.33002958 -1.2655188 0.36415634 -0.13636656
-0.16703553 -0.04611519 0.26118186 0.5003949 -1.2025503 -0.23954155
1.5907981 0.9046377 0.7609201 0.24414594 0.00803418 0.3947586
-0.30149376 0.09405237 0.52098083 -0.5345668 -0.5083351 0.7791456
-0.02575605 -0.49949223 -0.10849791 1.5852782 0.744915 0.4886622
0.4853845 -1.0162976 -0.5652912 0.12349863 -1.0502435 0.18160553
-0.157509 0.420859 0.5024181 -1.1029451 0.08461722 1.0564438
0.67049235 0.16172884 -1.3347994 -0.42252073 0.03081124 0.17094965
-1.3783343 -0.451279 0.9628916 -0.46163997 0.15617673 0.16335009
0.35860464 0.73060673 -0.14078279 0.4103134 1.4409643 -0.33272913
0.23036283 -0.04283301 0.16871816 0.45542124 0.5667987 -0.11651703
-0.22008248 -0.06348988 0.94246745 0.0253559 -1.059337 -0.06514162
-1.0692878 0.6196521 0.48282975 0.01369876 -0.7504469 -0.23204307
-0.05175822 0.07198853 0.5866004 0.36055863 -0.25865713 0.386512
-0.87342626 -0.05745874 0.46631643 0.5284822 0.8025409 0.2939001
-0.20605583 1.1503836 0.5931816 0.54361653 0.38415974 -1.117934
0.39419487 0.20617327 0.5950644 -0.884323 -0.11545131 -0.68173766
-1.1859312 0.27318838 0.3723541 -0.34937948 -0.94935966 1.3867812
0.44310653 0.38853487 -0.04185778 1.5792736 0.65506923 0.88411665
-0.18306369 -0.60268086 1.2460531 -0.9114662 -0.95810354 -0.46120682
0.04661645 -0.8826049 0.9241954 0.40850988 -1.0920576 -0.39791182
-1.1032463 -0.04892423 0.10457757 0.3817404 0.2501117 0.46751124
-0.73891073 0.70560724 -0.79329336 -0.03623621 0.30495864 0.2697436
-0.12315223 -0.2491835 -1.0804735 0.7873736 0.30767065 0.8137099
-0.96066517 -0.6675378 -0.5381873 -0.06338476 0.44825682 -0.72175914
0.810635 -0.99853873 -1.7549424 0.7654465 -0.17661537 0.12298558
0.44170558 -0.37792003 -0.30785277 0.14680074 0.15610027 -0.0445492
1.2646354 -1.499782 -0.27359688 -0.3051754 -0.07420754 0.44359976
-0.32341665 -0.02858789 -0.57666624 -0.747098 0.6046045 -0.68655705
-0.4016856 0.19356549 -0.5569048 0.44634873 0.6240221 -1.2109452
0.9017393 -2.2526646 0.2985825 0.16565953 -0.04211583 0.44496185
-0.14816996 0.03156538 -0.285229 -0.23659149 -0.7743411 -0.12132922
-0.10252117 0.10408023 -0.02567135 0.8920206 0.07346208 0.2636806
-0.78324884 -0.08058541 0.21170977 0.6702224 -0.3542656 0.60960704
0.04144417 0.839549 -0.5392842 0.4726111 1.300922 -0.31397298
-0.0440659 -0.70592976 -0.4844126 -0.13471349 -1.6349839 1.5720137
-0.19359036 0.2960946 1.0193137 -1.2238609 0.80323064 0.25831187
0.385607 -0.34980923 0.22990415 0.27280998 -0.3532943 -0.9821027
0.22466691 -0.54830736 0.68198806 0.21460073 -0.23662542 -0.18715815
-0.14655086 -0.3540812 0.6638908 0.16229312 0.05432042 -0.5053694
0.9716245 -0.05167791 0.8202948 0.618102 -0.13990085 0.82709444
0.23667422 -0.09235897 -0.65708953 -0.62388235 -0.13857721 0.70387787
0.4505228 0.369637 -0.70828754 -0.11345812 -0.17568189 0.7154993
-0.15546332 0.00879261 -0.3171571 -1.5215237 0.03736058 -0.01698132
0.985715 -0.592908 -0.2847801 0.21246007 -0.52462506 -0.97225106
-0.5229742 0.23942944 -0.8615567 -0.9654787 -1.0633081 -0.7042864
0.9430839 0.9129084 0.62025267 -0.0433873 -0.19614044 0.09377078
-0.15291765 -0.06261364 -0.14303185 -0.4372031 -0.05257855 0.40816873
-0.2282848 -0.6082837 -0.88593405 0.47846586 -1.245939 0.06448577
0.653112 0.97807544 0.5445391 0.5877041 0.22301732 -0.5411281
0.8286609 -0.35870865 -0.9536789 0.20347995 -0.5756313 0.04606663
0.44924912 -0.3662329 -1.5669373 0.01247683 0.06035846 -0.589218
0.04346244 0.8323395 -0.5277777 -0.3510925 0.45627934 0.45868495
0.23254927 -0.70751446 0.38978368 0.6575065 0.71855503 -0.5013251
1.144992 0.6038601 0.01185667 -1.1096243 -1.0243591 -0.65403354
-0.42624363 -0.30279815 0.8539806 -1.1022767 -0.70938194 0.7812969
-1.0551742 -0.23819715 0.15757415 0.47612834 -0.18372877 0.82947195
-0.47780696 -0.83209777 -0.41771856 -1.2663724 0.58435583 0.45292008
0.53802997 -0.9409955 -0.19383624 0.54961693 0.55114925 0.18298416
0.59737194 0.19852869 -0.577227 0.17525768 -0.6309134 0.7937616
0.3247603 -0.30758932 -1.0818672 -0.40774953 0.7881698 0.00573808
0.87117904 0.7592527 1.2757311 -0.30275238 0.04215754 1.1559546
1.7915794 -0.02314552 0.72095656 0.51577556 0.7029933 0.53044283
0.6002739 0.6054905 -0.02540313 0.39826837 0.65174377 -0.3000018
0.07136619 0.2523634 0.25542694 0.7913562 -0.3252793 -0.15818726
-0.61444277 0.35587555 -1.7155583 -0.71953106 -0.56062996 2.1658108
0.93512475 -0.2951448 -0.44940826 0.07462256 1.0103037 0.3045714
-0.62400025 0.45429942 -0.300041 -0.05338506 0.6585275 0.5895247
-1.1669285 0.47144347 5.4255657 0.7432824 -1.3212355 0.09507424
0.41807353 0.33054 -0.26645413 0.08427303 -0.37789723 0.5571027
0.29694656 0.13347776 0.9787314 0.24365935 1.0100209 -0.2268761
-0.37767404 1.2018535 -0.10142054 -0.8306037 -0.29626185 -0.11317202
0.73625815 -0.08501508 -0.0316575 -0.31258827 -0.01636732 -0.5762035
0.5855896 0.7561486 0.6552834 -0.37404934 0.6391039 0.30728775
-0.7834603 -0.11080054 -0.54556805 0.06177098 0.21146634 1.1888804
-0.24895927 0.6147307 0.57904583 0.71408194 -0.10421524 1.37166
-0.5230447 0.6803753 -0.09111746 0.5758178 0.14752047 -1.0815927
0.86597574 0.89537436 0.5436312 0.7324311 -0.07057254 1.2345091
0.02635702 -0.27539414 -0.06525761 0.03294921 0.17345951 1.4992667
-0.5496335 0.12290872 -0.41389337 1.1975346 -0.17953923 0.9728084
-0.74258375 -0.30400282 0.77961314 0.02362205 0.38384905 -0.23765366
-0.25644237 -1.4449133 0.06814793 -0.93175536 0.12397752 -1.1539291
-1.4398346 0.26587266 -0.26749712 -1.3270048 0.6716792 -0.6617713
-0.826434 1.3682383 -2.0655167 -0.65862876 -0.9091198 0.65662485
0.25493413 0.30194303 0.32595286 0.35329348 -1.1061157 -0.21434115
0.5939252 -0.25622934 0.7153359 -1.0599871 -0.28438637 1.2747719
-0.80364716 0.58287114 0.99786687 -0.53008777 -1.5163621 -1.061935
0.26124072 0.48449168 0.7360951 0.20926979 -1.2795522 0.26781708
0.25117093 0.12217176 0.4383791 -0.4440682 0.15208007 -0.3524471
-1.1528978 0.35407078 0.7677699 -0.35511306 -0.46109256 0.5295602
0.5713271 -0.4747758 -0.7633668 0.35371244 0.05102929 -0.87603486
1.1653701 -0.09810847 0.45939758 -0.87167394 0.01457075 -1.5827208
-0.500877 -0.8756131 0.2172148 1.3210423 0.07623304 -0.7750023
0.5385184 0.67028326 -0.3695491 -0.29316935 -0.52828425 -0.5749118
-0.47319132 -0.05699683 0.2850787 1.0898228 -0.4738782 0.09830704
-0.4930364 0.7800988 1.209437 0.08547673 0.46287712 -0.99666774
-0.20654045 -0.2506312 0.27887866 -1.2649473 -0.06020945 -0.61104697
0.5533968 -1.6174661 0.13084432 -0.3200293 -0.31043532 0.24505337
-0.5792496 -0.10916134 -0.1443208 0.5466428 -0.09689096 0.9081612
1.5364708 -0.2613929 -0.23074165 0.1346132 -0.80980575 0.63762134
0.6800429 -0.41945165 -0.4285268 -0.8103455 0.00920208 0.2271762
0.5279061 -0.7118786 0.21504022 -0.42061934 0.31293157 -0.4327694
0.24886921 -1.0018404 0.30491766 0.22523713 0.03489932 -0.52461964
0.06390318 0.5408851 -0.24605462 -0.588861 1.0574474 -0.24339217
-0.5438249 0.22667998 -0.18915197 -0.08238944 0.5917595 -0.22795674
-0.2975625 -0.26785564 -0.59394187 0.35362417 0.38553807 -0.25953072
0.6730732 -1.060814 -0.87921745 0.19810267 -0.15773763 0.14174233
0.37031534 1.1514708 -0.74788827 -0.09220511 -0.09300163 -0.47254625
-0.8798903 0.25979108 0.6841753 0.02611824 -0.4450578 0.9420224
0.08480126 -0.26310104 0.24743462 -0.14378783 -0.21432108 0.01724982
0.41388926 0.66678 0.02794337 -0.43623477 -0.13933326 0.47202906
0.39357048 -0.02711898 1.739704 -0.53612334 -0.6653462 -0.04893973
1.0044001 -0.26154536 -1.6537744 -0.29705834 -0.26992038 -0.560794
0.6980515 -0.57123595 -1.3464943 0.6234739 0.63272095 0.2744825
1.5458553 -0.6901784 0.67490935 0.12345029 -0.03461559 -0.9694483
-0.09647319 0.29843754 1.0357972 -1.2234743 0.34439763 -0.63353944
-0.48665312 1.1056746 0.35243925 0.08763167 0.6404449 -0.18153618
0.22375321 -0.19760104 0.07349344 -0.13003403 0.22813292 0.31072402
0.19422884 -0.16029774 -0.40413627 0.54459715 0.49965906 0.14869362
0.53351146 0.74981344 -0.5321794 -0.7362241 -0.98065114 0.21466133
-0.30379957 -0.16625613 0.16243227 0.22843498 0.14224625 1.2213321
-0.49664348 0.0173284 0.33342814 -0.20332967 0.18061025 -0.27445224
-0.25449052 0.4586984 -0.16798337 -0.33583608 -0.7994041 -0.52787435
-0.9303292 -0.06659514 -0.8511079 0.11332209 0.57297343 0.8890661
0.03748317 0.47457448 0.8867655 -1.1595738 -0.72226137 -0.8817738
-1.0451227 0.32644612 0.50320965 -0.58954513 -0.94634074 0.2588758 ] | [11.218779563903809, -2.4727227687835693] |
1d56f060-1379-4e88-b13b-4259ccf76a42 | clipter-looking-at-the-bigger-picture-in | 2301.07464 | null | https://arxiv.org/abs/2301.07464v1 | https://arxiv.org/pdf/2301.07464v1.pdf | CLIPTER: Looking at the Bigger Picture in Scene Text Recognition | Understanding the scene is often essential for reading text in real-world scenarios. However, current scene text recognizers operate on cropped text images, unaware of the bigger picture. In this work, we harness the representative power of recent vision-language models, such as CLIP, to provide the crop-based recognizer with scene, image-level information. Specifically, we obtain a rich representation of the entire image and fuse it with the recognizer word-level features via cross-attention. Moreover, a gated mechanism is introduced that gradually shifts to the context-enriched representation, enabling simply fine-tuning a pretrained recognizer. We implement our model-agnostic framework, named CLIPTER - CLIP Text Recognition, on several leading text recognizers and demonstrate consistent performance gains, achieving state-of-the-art results over multiple benchmarks. Furthermore, an in-depth analysis reveals improved robustness to out-of-vocabulary words and enhanced generalization in low-data regimes. | ['Ron Litman', 'Shai Mazor', 'Royee Tichauer', 'Oren Nuriel', 'Roy Ganz', 'Alona Golts', 'David Bensaïd', 'Aviad Aberdam'] | 2023-01-18 | null | null | null | null | ['scene-text-recognition'] | ['computer-vision'] | [ 7.95559108e-01 -5.94160020e-01 -1.30946651e-01 -2.51658618e-01
-1.02377939e+00 -9.02771473e-01 1.03997815e+00 1.23943113e-01
-4.41138208e-01 -3.66359204e-02 4.84210998e-01 -4.06180292e-01
4.14295912e-01 -4.39401239e-01 -8.62167537e-01 -6.35497034e-01
6.28692508e-01 3.33665103e-01 -3.94906811e-02 -2.42329329e-01
4.88471121e-01 2.40374357e-01 -1.58090937e+00 8.80410194e-01
7.60648906e-01 7.55728900e-01 4.15573925e-01 1.02801180e+00
-5.10990441e-01 7.50876784e-01 -3.96679848e-01 -4.26308572e-01
1.05171070e-01 -1.86476156e-01 -7.13363647e-01 6.49016559e-01
9.84840214e-01 -2.75829434e-01 -5.88862658e-01 7.87015975e-01
1.81956351e-01 -4.46108915e-03 6.42915487e-01 -5.54159462e-01
-1.21509290e+00 5.80059230e-01 -7.38479316e-01 1.37537807e-01
5.03914475e-01 3.91256869e-01 1.29509962e+00 -1.43944085e+00
6.03426933e-01 1.11060023e+00 2.86116511e-01 4.16058213e-01
-1.30517340e+00 -3.60202491e-01 7.38613188e-01 1.04124010e-01
-1.24307144e+00 -5.83992660e-01 6.00877166e-01 -3.15365106e-01
1.28930771e+00 3.14659297e-01 3.28030646e-01 1.55263126e+00
1.65422186e-01 1.26225853e+00 1.03639197e+00 -8.50206673e-01
-1.64453965e-02 -1.11042939e-01 3.17007333e-01 5.76084137e-01
2.31315285e-01 -2.64193863e-01 -8.88293684e-01 3.93351346e-01
2.59991497e-01 4.25055474e-01 -4.53029931e-01 -4.35892731e-01
-1.31802678e+00 5.98681390e-01 2.78843284e-01 3.78237158e-01
-2.13026300e-01 1.53604969e-01 4.36466485e-01 2.88003922e-01
5.71616650e-01 3.08492839e-01 -3.34887683e-01 -1.75055742e-01
-1.35058081e+00 1.70004612e-04 8.17505479e-01 1.06843305e+00
5.00981569e-01 1.55691639e-01 -5.14990926e-01 7.56036043e-01
1.67491719e-01 9.62130904e-01 5.37590921e-01 -6.73560351e-02
7.48867452e-01 5.81132829e-01 -2.36941516e-01 -6.59879565e-01
-1.89053491e-01 -5.79998493e-01 -7.59229779e-01 -3.57321580e-03
3.05627078e-01 3.56670678e-01 -1.49977398e+00 1.21305287e+00
-1.86065376e-01 2.21535042e-02 1.78393766e-01 7.67596662e-01
6.51511788e-01 7.13592291e-01 1.69992954e-01 2.90113747e-01
1.54827940e+00 -1.41398776e+00 -4.01798874e-01 -7.40066111e-01
5.18014312e-01 -8.72861743e-01 1.58534086e+00 5.41167200e-01
-6.99843884e-01 -5.36416411e-01 -1.01491988e+00 -3.85474145e-01
-8.19435120e-01 4.24901426e-01 2.05523312e-01 5.37803292e-01
-1.07747614e+00 2.01199517e-01 -6.58447266e-01 -7.71645665e-01
6.10375464e-01 -7.19174324e-03 -3.47528964e-01 -6.55447066e-01
-6.50529146e-01 6.93313897e-01 1.95577070e-01 -7.87503123e-02
-1.03609455e+00 -7.57025599e-01 -8.67064655e-01 1.76493540e-01
5.12349904e-01 -4.68305379e-01 1.13175809e+00 -1.28184545e+00
-1.39682162e+00 1.12594509e+00 -4.36816514e-01 -3.85797292e-01
4.76052105e-01 -5.57487965e-01 -3.63431126e-01 2.94205397e-01
-1.63213044e-01 6.68494403e-01 1.46255028e+00 -1.36377954e+00
-4.53845561e-01 -3.39853823e-01 -7.76892081e-02 3.51552844e-01
-5.47972918e-01 -7.57582411e-02 -7.88552761e-01 -9.32968616e-01
-1.28878206e-01 -6.27748132e-01 1.95919082e-01 3.93633842e-02
-5.22491038e-01 1.59371272e-01 1.00535929e+00 -7.59723246e-01
1.09444773e+00 -2.15786839e+00 2.09908098e-01 -1.16934426e-01
2.93359399e-01 4.00872976e-01 -5.92897534e-01 7.36195922e-01
-4.54214253e-02 -7.84266181e-03 -1.90335169e-01 -7.34867692e-01
1.19159624e-01 6.86048120e-02 -8.83103669e-01 3.20753306e-01
6.17674947e-01 1.39061892e+00 -6.45265996e-01 -1.78798676e-01
5.15865088e-01 4.42478210e-01 -4.14086252e-01 1.14210024e-02
-5.51052809e-01 -1.60347074e-02 -5.21261334e-01 8.00839424e-01
6.71226740e-01 -5.09313226e-01 2.66238637e-02 -1.28976628e-01
-1.34353012e-01 1.29536435e-01 -5.70354223e-01 2.09664583e+00
-5.97220421e-01 1.09354746e+00 6.04793765e-02 -1.07476461e+00
7.40875304e-01 -7.10923132e-03 -3.68170328e-02 -8.06643605e-01
1.27057269e-01 -2.43649244e-01 -4.08490777e-01 -5.18786669e-01
9.13568974e-01 2.51400590e-01 3.73679213e-02 5.62156379e-01
4.49868403e-02 -2.21305136e-02 8.46810713e-02 4.93011326e-01
1.15069032e+00 2.55567491e-01 1.35057732e-01 -2.41047055e-01
5.10148883e-01 2.27812245e-01 -4.60650682e-01 1.20541525e+00
1.97100583e-02 9.97647941e-01 1.48198262e-01 -3.18129390e-01
-1.10960484e+00 -9.49160516e-01 -2.29992811e-02 1.70148063e+00
2.93623861e-02 -4.25060421e-01 -6.97672784e-01 -7.31517732e-01
1.08335026e-01 7.86834896e-01 -8.32930088e-01 -1.20815776e-01
-3.79241079e-01 -5.31994700e-01 6.36545300e-01 7.10028827e-01
3.39284658e-01 -9.10051644e-01 -5.63854814e-01 6.30149394e-02
4.79288539e-03 -1.34904087e+00 -6.53764248e-01 4.13549453e-01
-4.91516590e-01 -7.28119373e-01 -8.47723663e-01 -8.22693467e-01
6.48624599e-01 7.86910713e-01 1.34011245e+00 2.33265892e-01
-5.25387883e-01 9.15379047e-01 -7.96711385e-01 -2.23473027e-01
-2.47914329e-01 1.85298488e-01 -4.05406028e-01 2.58713543e-01
6.28911078e-01 -5.05063199e-02 -3.14004421e-01 -1.05433233e-01
-1.10064411e+00 3.04175913e-01 8.58358979e-01 1.08579397e+00
3.84783268e-01 -4.58198547e-01 2.60752439e-01 -7.92299092e-01
4.90056068e-01 -1.06423751e-01 -3.16729844e-01 7.12562382e-01
-3.72090548e-01 7.23958388e-02 8.01008880e-01 -7.17083693e-01
-1.00421178e+00 1.52862564e-01 2.27856964e-01 -5.69578707e-01
-3.32951069e-01 4.74323928e-01 -1.79660991e-01 5.26842885e-02
5.82123220e-01 9.74202991e-01 -3.50900680e-01 -5.02928376e-01
7.89315104e-01 8.25117469e-01 6.64463878e-01 -6.52642727e-01
9.63615954e-01 7.14615345e-01 -4.49886024e-01 -1.05983949e+00
-1.02383685e+00 -6.29061520e-01 -8.71185362e-01 6.66896552e-02
9.01612163e-01 -1.20917451e+00 -3.86116058e-01 6.14536703e-01
-1.04844451e+00 -5.52177608e-01 -1.87260687e-01 7.21731260e-02
-3.36387485e-01 3.82730365e-01 -4.65706587e-01 -6.35757327e-01
-4.69709128e-01 -1.14189136e+00 1.72741508e+00 2.01892197e-01
2.55909503e-01 -9.69523609e-01 -1.40849292e-01 4.74584609e-01
5.20249248e-01 -2.99644291e-01 9.63459551e-01 -1.02029788e+00
-6.98745012e-01 -2.08412170e-01 -6.89208806e-01 1.39291301e-01
6.54815137e-02 -7.75262294e-03 -1.46298206e+00 -4.79498118e-01
-3.99856836e-01 -5.55486977e-01 1.42085731e+00 5.71720488e-02
1.19602168e+00 -2.58055460e-02 -3.63395065e-01 6.54624224e-01
1.40737724e+00 -3.40209156e-01 5.52254796e-01 1.03447512e-01
1.10003757e+00 3.54744583e-01 1.98970079e-01 3.02141935e-01
3.15816253e-01 4.65129763e-01 2.35532373e-01 -2.65053332e-01
-3.96197677e-01 -4.37546492e-01 4.29302871e-01 7.34157801e-01
4.74992365e-01 -8.73836219e-01 -1.10543704e+00 5.62295437e-01
-1.64758825e+00 -8.10979366e-01 1.33378208e-01 1.79428041e+00
7.35496581e-01 1.79764092e-01 -2.83763200e-01 -1.02295823e-01
5.65249681e-01 6.02568567e-01 -7.53797114e-01 -2.79189825e-01
-4.74937022e-01 3.20333600e-01 4.89412338e-01 3.70396942e-01
-1.27567840e+00 1.49821568e+00 6.60631466e+00 9.58635926e-01
-1.45524406e+00 -2.50739008e-01 5.94525814e-01 -1.94951117e-01
-2.25767776e-01 -3.05426102e-02 -7.88277507e-01 4.39144932e-02
7.47325957e-01 -1.25292718e-01 6.25951111e-01 6.65393829e-01
-2.13433104e-03 5.20621426e-02 -1.21480298e+00 1.06397021e+00
8.23396623e-01 -1.53215396e+00 5.50004542e-01 -1.25520512e-01
7.83068061e-01 2.60949999e-01 3.23735029e-01 4.00471240e-01
2.73534805e-01 -1.29219353e+00 7.55062342e-01 5.47605991e-01
1.03149760e+00 -2.20444694e-01 2.72489578e-01 1.80391356e-01
-1.27554679e+00 3.83170098e-02 -3.39289218e-01 -3.16794328e-02
-1.40684813e-01 4.18164700e-01 -6.24778867e-01 6.07325912e-01
5.02009034e-01 1.11103725e+00 -1.03257692e+00 4.94191021e-01
5.35556152e-02 7.59637356e-01 -1.23123676e-01 -1.55026555e-01
5.30786633e-01 4.95919064e-02 3.05930614e-01 1.74387622e+00
-3.21472138e-02 -1.34646133e-01 3.64728272e-01 7.98722863e-01
-3.99832815e-01 2.68147290e-01 -6.13768637e-01 -4.23091143e-01
8.83193612e-02 1.25897241e+00 -7.35625386e-01 -5.58369696e-01
-8.09585273e-01 1.31136799e+00 4.09925729e-01 6.91895604e-01
-5.14492095e-01 -2.47424260e-01 5.12271106e-01 -2.06452563e-01
8.93579304e-01 -2.38681540e-01 -4.65074271e-01 -1.57301223e+00
1.68784544e-01 -1.37088513e+00 2.23708585e-01 -1.09399152e+00
-1.21035302e+00 3.81941706e-01 -4.11987454e-01 -1.03013802e+00
9.83169004e-02 -1.18234241e+00 -6.74273729e-01 7.97412694e-01
-1.58048761e+00 -1.77910709e+00 -4.71639842e-01 5.76514423e-01
1.46324468e+00 -1.62974313e-01 7.45956719e-01 -9.74410698e-02
-6.22825444e-01 7.78140604e-01 5.60538769e-01 4.76206422e-01
8.68925393e-01 -1.18667638e+00 7.83942342e-01 1.10857129e+00
8.24310601e-01 4.81539428e-01 5.36010444e-01 -5.56391180e-01
-2.13297510e+00 -1.11758018e+00 4.76739228e-01 -8.29948008e-01
9.22889948e-01 -8.93560529e-01 -1.17828321e+00 7.00931013e-01
4.76623684e-01 9.21478961e-03 4.03857976e-01 1.49801774e-02
-1.13294101e+00 1.82860270e-01 -5.45094252e-01 7.91799724e-01
9.49906647e-01 -9.95788991e-01 -7.02433228e-01 3.16690892e-01
8.22304368e-01 -4.06865358e-01 -2.46644080e-01 -8.00997019e-02
7.41491616e-01 -5.89879036e-01 8.87501776e-01 -9.18997288e-01
6.15488589e-01 -2.02512927e-02 -4.39531446e-01 -1.13795078e+00
-1.94142103e-01 -4.33962166e-01 -3.17938291e-02 1.14756966e+00
4.57011700e-01 -2.20062256e-01 7.31045961e-01 2.07185432e-01
-1.53823420e-01 -4.88872439e-01 -6.59160852e-01 -5.65444112e-01
3.76166135e-01 -6.77044332e-01 2.13230059e-01 9.09617901e-01
1.21275559e-01 4.61362034e-01 -4.35427874e-01 3.39384675e-02
4.51498032e-01 2.95118243e-01 7.40425885e-01 -8.27803016e-01
-2.99346060e-01 -5.70383191e-01 -2.36445785e-01 -1.59085548e+00
3.94096375e-01 -8.84514928e-01 2.58747041e-01 -1.39037681e+00
4.71129954e-01 1.16139531e-01 -2.83913404e-01 4.98054594e-01
-4.11288291e-01 2.68719316e-01 4.46878850e-01 1.66980937e-01
-9.69337165e-01 5.43577611e-01 1.14068139e+00 -6.45414531e-01
1.42548800e-01 -4.88328427e-01 -7.03476489e-01 5.00351369e-01
5.10715246e-01 -2.82463104e-01 -2.53390700e-01 -1.00679743e+00
-3.55553813e-02 -4.82519120e-01 4.46833283e-01 -8.91062498e-01
4.49059665e-01 -1.55837461e-01 5.48707604e-01 -7.11456835e-01
3.25391084e-01 -7.34067202e-01 -6.31063581e-01 2.38291234e-01
-8.37617278e-01 -1.80508271e-02 4.52742934e-01 8.32727551e-01
9.40711796e-03 -5.41119054e-02 4.73671824e-01 1.86256960e-01
-8.43446076e-01 6.17996417e-02 -4.89751220e-01 2.81060159e-01
5.95902979e-01 -1.78611875e-01 -7.76703358e-01 -4.39659655e-01
-2.81213284e-01 1.07555427e-01 6.87164605e-01 7.21011043e-01
6.23058736e-01 -7.03402400e-01 -8.52215350e-01 2.95780212e-01
7.54604757e-01 -2.62115508e-01 2.48387516e-01 5.89342415e-01
-3.67416292e-01 7.44104028e-01 1.31065935e-01 -8.25605989e-01
-1.05395722e+00 8.80751312e-01 1.53370023e-01 -3.91140133e-01
-8.82289827e-01 6.13933384e-01 6.57502174e-01 -1.89313889e-01
2.70846397e-01 -5.15764892e-01 1.20077007e-01 -5.91662750e-02
8.54196370e-01 -2.57983357e-01 1.92139491e-01 -6.54682219e-01
-2.95302689e-01 9.17614520e-01 -6.07482910e-01 -1.40464276e-01
9.46231604e-01 -2.74006635e-01 3.60921592e-01 4.73838925e-01
9.97551918e-01 -1.32622356e-02 -1.46878505e+00 -7.36869991e-01
-6.55571818e-02 -3.87788564e-01 2.53740519e-01 -1.07387733e+00
-8.33685279e-01 1.24720502e+00 4.35576081e-01 2.58381944e-02
1.11511803e+00 -7.20935389e-02 4.69334513e-01 8.86608183e-01
-5.72816208e-02 -9.69382167e-01 3.30877423e-01 7.42770970e-01
7.95813441e-01 -1.59163916e+00 -1.04533359e-02 -1.74035072e-01
-8.61941040e-01 1.16465461e+00 3.60470027e-01 -7.95361474e-02
3.87472808e-01 4.40260440e-01 2.91128546e-01 -7.07128346e-02
-1.17293608e+00 -4.64013338e-01 5.78445911e-01 6.86371922e-01
3.39123547e-01 -7.66692758e-02 4.52082962e-01 3.89881611e-01
1.90304756e-01 -2.23155260e-01 3.91526312e-01 9.68760371e-01
-6.17336452e-01 -7.89275110e-01 -3.89892727e-01 5.17799079e-01
-2.56132066e-01 -7.35163450e-01 -6.97434962e-01 7.96904922e-01
-5.78748167e-01 9.08259809e-01 2.51879990e-01 -2.67298490e-01
1.77298471e-01 4.63423967e-01 3.39285195e-01 -6.32535219e-01
-6.03433311e-01 1.13137536e-01 -2.98636913e-01 -4.20595080e-01
-6.06384948e-02 -4.63861108e-01 -9.69783962e-01 -1.61235288e-01
-3.97329628e-01 -4.40570563e-01 7.21266389e-01 1.26778984e+00
5.08655787e-01 5.25698245e-01 3.39406550e-01 -9.00355220e-01
-5.58447897e-01 -8.80140007e-01 -2.84578472e-01 6.00389123e-01
5.57911515e-01 -1.88278586e-01 -2.79765427e-01 5.78635752e-01] | [11.794608116149902, 2.1817545890808105] |
9ec42af3-996b-474e-a763-d0acfab1f24a | 3d-skeleton-based-few-shot-action-recognition | 2112.12668 | null | https://arxiv.org/abs/2112.12668v1 | https://arxiv.org/pdf/2112.12668v1.pdf | 3D Skeleton-based Few-shot Action Recognition with JEANIE is not so Naïve | In this paper, we propose a Few-shot Learning pipeline for 3D skeleton-based action recognition by Joint tEmporal and cAmera viewpoiNt alIgnmEnt (JEANIE). To factor out misalignment between query and support sequences of 3D body joints, we propose an advanced variant of Dynamic Time Warping which jointly models each smooth path between the query and support frames to achieve simultaneously the best alignment in the temporal and simulated camera viewpoint spaces for end-to-end learning under the limited few-shot training data. Sequences are encoded with a temporal block encoder based on Simple Spectral Graph Convolution, a lightweight linear Graph Neural Network backbone (we also include a setting with a transformer). Finally, we propose a similarity-based loss which encourages the alignment of sequences of the same class while preventing the alignment of unrelated sequences. We demonstrate state-of-the-art results on NTU-60, NTU-120, Kinetics-skeleton and UWA3D Multiview Activity II. | ['Piotr Koniusz', 'Jun Liu', 'Lei Wang'] | 2021-12-23 | null | null | null | null | ['few-shot-action-recognition'] | ['computer-vision'] | [ 3.08124781e-01 -2.87883207e-02 -4.13442105e-01 -1.92351595e-01
-6.97494507e-01 -1.67328522e-01 6.40521824e-01 -4.99917001e-01
-3.80794823e-01 1.98067650e-01 5.61939955e-01 3.32429379e-01
8.80493373e-02 -1.12892605e-01 -7.85890639e-01 -5.28447032e-01
-3.90818655e-01 4.10805613e-01 5.63294709e-01 -1.30906463e-01
4.62891906e-02 4.10501450e-01 -1.20598185e+00 2.84732342e-01
1.17949717e-01 8.28263760e-01 1.41905189e-01 9.69403625e-01
3.71883780e-01 1.26828718e+00 -2.54504919e-01 -3.03357810e-01
5.10670781e-01 -6.61939502e-01 -6.78143084e-01 7.75784373e-01
8.61931443e-01 -6.67207778e-01 -8.88888478e-01 6.04141533e-01
7.87480175e-01 5.74085951e-01 5.20467401e-01 -1.37723398e+00
-1.07468612e-01 1.21547855e-01 -7.98645556e-01 2.98366815e-01
7.49158442e-01 4.60876614e-01 1.06475306e+00 -7.77581334e-01
1.20077205e+00 1.20029950e+00 6.15782440e-01 5.92973471e-01
-1.05672419e+00 -2.84847289e-01 2.54200846e-01 7.21525252e-01
-9.64829087e-01 -5.22475243e-01 9.22526300e-01 -4.84275401e-01
1.44732630e+00 -2.16135025e-01 1.15907300e+00 1.77725065e+00
3.42751592e-01 9.68997478e-01 3.63873780e-01 -1.95674017e-01
1.37388915e-01 -1.11963689e+00 -1.04871206e-01 9.77449715e-01
-4.54343438e-01 5.06239869e-02 -1.11097586e+00 6.19371086e-02
9.32714283e-01 2.32168213e-01 5.95687255e-02 -1.16876376e+00
-1.26003611e+00 6.79997563e-01 8.21368769e-02 -3.73379923e-02
-3.11975539e-01 5.39800227e-01 6.78155243e-01 3.59469652e-01
2.46485442e-01 -2.07547545e-01 -2.95125574e-01 -5.86091816e-01
-7.29712665e-01 3.32937002e-01 5.21699727e-01 1.07737494e+00
3.45527411e-01 2.92786747e-01 -3.06740761e-01 6.10562563e-01
2.27566764e-01 2.99389005e-01 6.32889628e-01 -1.39310622e+00
6.31377816e-01 3.66320312e-01 -2.44171992e-01 -7.16795683e-01
-4.11741853e-01 -4.84436192e-02 -4.38534081e-01 3.41911793e-01
4.45620090e-01 2.23747313e-01 -1.03931463e+00 1.70545280e+00
5.42496800e-01 8.03109467e-01 -2.10335523e-01 9.84696507e-01
4.87255007e-01 1.35317311e-01 -2.29652956e-01 -1.85913086e-01
1.12263894e+00 -1.50324702e+00 -6.33857012e-01 -3.50342423e-01
6.15042329e-01 -4.36930180e-01 9.76546705e-01 5.31642288e-02
-1.33283091e+00 -6.77563906e-01 -1.18197060e+00 -3.14166844e-01
2.67183423e-01 -2.21912742e-01 3.52489471e-01 9.57029611e-02
-6.30358279e-01 9.85282123e-01 -1.56635594e+00 -4.84307438e-01
2.58666515e-01 5.82556939e-03 -6.40008450e-01 -3.65448967e-02
-9.60306108e-01 8.98497939e-01 -7.22244056e-03 -4.21687365e-01
-1.31950414e+00 -4.52043265e-01 -1.23406434e+00 -2.84746528e-01
6.82214260e-01 -9.92360353e-01 1.31574249e+00 -7.67378688e-01
-1.79734302e+00 9.63020027e-01 1.97205488e-02 -7.30372369e-01
8.16673100e-01 -5.62917829e-01 -1.37500167e-01 4.40090895e-01
1.70653671e-01 4.66210186e-01 1.21300447e+00 -3.10197949e-01
-3.71416122e-01 -5.88637114e-01 -8.41708109e-02 5.87830365e-01
-1.33652324e-02 -1.66536272e-02 -7.03983247e-01 -9.25000191e-01
1.24176845e-01 -1.06931674e+00 -1.94211915e-01 4.46076572e-01
-5.05763851e-02 -9.70113352e-02 8.59992802e-01 -7.68222809e-01
7.75882959e-01 -2.03479743e+00 8.26478481e-01 -1.46580458e-01
4.23861779e-02 3.45056038e-03 -3.44947189e-01 4.59161341e-01
-2.96723753e-01 -7.17098296e-01 -1.25820294e-01 -6.66672349e-01
-9.19991806e-02 5.04203856e-01 1.32667432e-02 8.56928825e-01
-2.11512558e-02 9.56200242e-01 -1.01816070e+00 -4.49440777e-01
5.26535332e-01 3.28601182e-01 -5.71100593e-01 4.17010367e-01
-2.57329315e-01 5.36510527e-01 -1.74344808e-01 6.88149571e-01
-2.61245091e-02 -3.63949597e-01 1.47474110e-01 -1.62466228e-01
2.54152715e-01 1.23409972e-01 -1.15745866e+00 2.76659822e+00
-7.85637423e-02 4.06300485e-01 -1.17985971e-01 -9.41680431e-01
6.16816878e-01 3.40805441e-01 1.01385868e+00 -6.83835387e-01
1.16487481e-01 -2.38307565e-01 -2.30560765e-01 -6.98490858e-01
1.70979485e-01 7.25285411e-02 4.94812056e-02 5.76532245e-01
5.91353178e-01 8.21109675e-03 1.55776843e-01 1.80940583e-01
1.48908889e+00 9.24633563e-01 4.62088406e-01 4.57838565e-01
1.61541641e-01 -3.60502362e-01 4.97695833e-01 4.90049571e-01
-5.61382651e-01 8.55425775e-01 5.35420120e-01 -4.82113123e-01
-1.32530940e+00 -1.17187393e+00 7.00361073e-01 1.16432238e+00
4.37325835e-02 -6.28112316e-01 -4.91377503e-01 -7.89310157e-01
-1.48673266e-01 4.02513295e-01 -5.99375248e-01 -2.85457581e-01
-8.77688169e-01 7.77747035e-02 5.36504865e-01 7.30585992e-01
3.40106875e-01 -8.15569878e-01 -1.14052451e+00 2.45653912e-01
-3.28622460e-01 -1.43014264e+00 -9.19388413e-01 1.63532406e-01
-1.06555128e+00 -1.24239469e+00 -8.71565044e-01 -4.62377846e-01
1.92093983e-01 3.88326615e-01 8.96678686e-01 -3.59946042e-01
-4.62974310e-01 6.41518772e-01 -5.70440471e-01 1.61527082e-01
-1.30677417e-01 -3.12812328e-01 2.27235720e-01 3.42247449e-02
-2.38407832e-02 -1.06307161e+00 -5.58763385e-01 4.85047936e-01
-4.84919339e-01 2.22925484e-01 1.74153432e-01 8.04976881e-01
7.21581101e-01 -8.29168260e-01 -2.38029987e-01 4.81310785e-02
-1.10845156e-01 -1.40246168e-01 -1.71337560e-01 2.05227107e-01
-9.70784426e-02 -4.89225425e-03 1.71232104e-01 -8.14331293e-01
-6.09238207e-01 3.81187975e-01 2.34692041e-02 -1.37933910e+00
7.19298422e-02 8.50571319e-02 -7.05659166e-02 8.24417770e-02
6.47599339e-01 1.70589447e-01 4.31822330e-01 -4.56088483e-01
5.64869702e-01 2.58575585e-02 8.16789925e-01 -3.61805826e-01
4.74182606e-01 6.85637414e-01 1.76291347e-01 -8.23172569e-01
-7.87308693e-01 -6.59947634e-01 -9.86729980e-01 -5.72781324e-01
1.18102252e+00 -1.18999541e+00 -4.72903907e-01 6.65449619e-01
-1.11424220e+00 -5.52212775e-01 -5.51291406e-01 7.87393510e-01
-1.48049843e+00 9.28084075e-01 -6.74695909e-01 -3.99280459e-01
-2.38311723e-01 -1.06218135e+00 1.42069483e+00 -3.02209616e-01
-6.30715489e-01 -5.46855509e-01 6.03309095e-01 4.95835513e-01
-3.59444767e-01 5.60187280e-01 4.70951855e-01 -5.34503102e-01
-3.97810966e-01 -1.13364622e-01 3.79299164e-01 3.27618927e-01
3.62669677e-02 -1.03848018e-01 -5.09598196e-01 -4.54135150e-01
9.84331742e-02 -5.50420940e-01 8.06595027e-01 5.28638840e-01
6.04232371e-01 -6.59904703e-02 7.83960894e-02 9.07654285e-01
8.18697453e-01 -4.61273305e-02 7.10376263e-01 2.66085327e-01
9.57578242e-01 4.32476521e-01 7.31549144e-01 7.75245965e-01
9.11285058e-02 1.17877293e+00 5.15961707e-01 4.10155475e-01
-4.17441458e-01 -3.28536570e-01 7.68116713e-01 7.28473186e-01
-4.62477148e-01 5.01917638e-02 -5.59468508e-01 3.75561684e-01
-2.20803666e+00 -1.44441843e+00 3.14851671e-01 2.01443863e+00
5.22239923e-01 2.47281909e-01 6.07268214e-01 -4.64387052e-02
5.50885141e-01 7.90796101e-01 -8.31424415e-01 5.15422300e-02
9.76843387e-03 2.44429022e-01 3.54194492e-01 4.29790378e-01
-1.11030221e+00 9.61871386e-01 5.92269945e+00 8.12925220e-01
-7.31084168e-01 2.69475162e-01 1.56199532e-02 -7.12236047e-01
2.89363295e-01 -1.03419490e-01 -3.88996750e-01 2.41979241e-01
7.31171906e-01 1.97927475e-01 2.68823832e-01 9.08564329e-01
3.12513828e-01 1.09628715e-01 -1.26523006e+00 1.15216303e+00
3.61482859e-01 -1.30057943e+00 -2.38568008e-01 3.86772193e-02
5.39474189e-01 3.55144858e-01 -3.22450876e-01 3.44722867e-02
1.85927570e-01 -5.55154502e-01 8.84789765e-01 5.23220778e-01
5.55238724e-01 -5.92609227e-01 -1.01410359e-01 2.83002496e-01
-1.24960291e+00 -4.37152339e-03 -1.97075471e-01 -1.95217758e-01
5.94557643e-01 -1.02628060e-02 -4.99227792e-01 6.04131758e-01
5.49102366e-01 1.38622200e+00 -2.72883952e-01 7.22475231e-01
-3.34152997e-01 1.47160932e-01 -1.32620245e-01 3.70384365e-01
3.08160841e-01 -1.88068643e-01 1.00170684e+00 7.33654082e-01
2.35305741e-01 5.67658059e-02 4.58810568e-01 2.77310759e-01
3.54681224e-01 -2.95279324e-01 -7.80174732e-01 2.81589907e-02
-1.45419776e-01 8.06007981e-01 -4.33833688e-01 -3.04280579e-01
-5.87310374e-01 1.55716002e+00 4.69063789e-01 1.95681214e-01
-1.01862907e+00 1.90866545e-01 9.63189900e-01 -4.80553787e-03
6.77519679e-01 -6.04122996e-01 2.10057110e-01 -1.34445477e+00
1.33360028e-01 -1.05851245e+00 6.25330687e-01 -8.70264769e-01
-1.03427041e+00 -9.45629478e-02 7.15827718e-02 -1.71807039e+00
-6.89058721e-01 -3.31079006e-01 -5.60180902e-01 2.07076788e-01
-7.61525035e-01 -1.22780621e+00 -4.71301526e-02 9.56610441e-01
1.13744235e+00 -2.94990748e-01 7.17038035e-01 9.18481946e-02
-2.57925153e-01 3.35896552e-01 -2.68891662e-01 1.25259340e-01
9.09075856e-01 -1.01094997e+00 9.50493336e-01 7.20748365e-01
4.76254672e-01 1.27689213e-01 6.72528088e-01 -8.72210920e-01
-1.45628452e+00 -8.89389217e-01 5.78888357e-01 -4.42775637e-01
8.45226526e-01 -2.49014214e-01 -5.71389496e-01 8.50408912e-01
1.25112226e-02 1.87505677e-01 5.52310765e-01 -2.05642894e-01
-5.60149193e-01 2.07419723e-01 -7.02046931e-01 8.60639453e-01
1.66139817e+00 -6.61275983e-01 -7.67926574e-01 4.83710527e-01
7.40627527e-01 -6.94412231e-01 -8.54602635e-01 1.38331622e-01
9.77359951e-01 -1.02437282e+00 1.28438044e+00 -1.09651375e+00
5.51575899e-01 -1.28137529e-01 -3.35407346e-01 -1.02756560e+00
-2.95274228e-01 -8.84885132e-01 -7.93877900e-01 3.19736630e-01
-2.87547141e-01 2.64121711e-01 1.20641863e+00 1.24735519e-01
-2.53279865e-01 -6.95637763e-01 -1.40621042e+00 -1.09665895e+00
-5.36628306e-01 -5.97938359e-01 -1.32419050e-01 7.50259399e-01
2.15513229e-01 4.70857143e-01 -1.09567630e+00 -2.52039701e-01
8.60931993e-01 -1.20209016e-01 1.09445512e+00 -6.40716076e-01
-8.35210741e-01 -2.32729465e-01 -9.85336125e-01 -1.22248232e+00
2.12464452e-01 -5.72849810e-01 -1.21988140e-01 -1.24295044e+00
2.04519004e-01 7.59588063e-01 -9.54300612e-02 4.69893992e-01
1.57773897e-01 2.08958648e-02 3.93958300e-01 1.10991105e-01
-8.01822960e-01 7.89820135e-01 1.48635805e+00 -1.53416887e-01
-8.86414573e-02 6.49880022e-02 3.89548391e-01 7.60383785e-01
2.48443872e-01 -3.95040929e-01 -6.07842326e-01 -3.46610487e-01
-8.68425891e-02 6.26515985e-01 5.69718480e-01 -1.14791083e+00
1.80961549e-01 -2.42734283e-01 2.87493825e-01 -7.79885888e-01
7.81010211e-01 -8.13659489e-01 2.35230044e-01 7.71719038e-01
-4.23418641e-01 1.42069995e-01 -3.67353350e-01 9.66517866e-01
1.25389099e-01 2.76971012e-02 7.62962162e-01 -3.11619163e-01
-9.80467677e-01 7.23896742e-01 -2.79546887e-01 1.97203934e-01
1.15809131e+00 -5.51046312e-01 6.92628250e-02 -5.32363474e-01
-1.19192314e+00 2.63634801e-01 5.73309898e-01 5.37703156e-01
8.03723872e-01 -1.59997678e+00 -5.33208847e-01 1.90282464e-01
2.99113035e-01 -3.41171980e-01 5.08258879e-01 1.13239944e+00
-3.80541503e-01 -5.34062684e-02 -6.36312008e-01 -7.58078277e-01
-1.58940589e+00 4.06323910e-01 4.16214287e-01 -4.61809456e-01
-1.18739688e+00 7.75547266e-01 -1.36275724e-01 -1.64926410e-01
6.81246519e-01 -1.02088995e-01 1.00917540e-01 1.30786851e-01
4.79683846e-01 8.07801068e-01 -7.47645274e-02 -6.37316227e-01
-3.89538348e-01 6.20903552e-01 6.75413013e-02 -3.84952009e-01
1.33452153e+00 -1.08080417e-01 4.26468819e-01 6.51525795e-01
1.40411222e+00 -6.37372911e-01 -1.86931849e+00 -2.94133276e-01
-2.44994536e-01 -5.95437050e-01 -4.25990760e-01 -2.21397579e-01
-8.71057093e-01 8.96402657e-01 6.46374762e-01 -5.05387187e-01
6.51633203e-01 1.19069722e-02 9.28547621e-01 5.07140517e-01
2.98559546e-01 -1.34358156e+00 1.00994992e+00 7.73759127e-01
8.67164552e-01 -9.70686853e-01 2.22437054e-01 -3.33700143e-02
-8.60927165e-01 1.07905221e+00 4.58632410e-01 -4.21227485e-01
3.57750654e-01 7.32834265e-02 -2.22107679e-01 -3.48181665e-01
-7.78737187e-01 -2.55065054e-01 2.11889073e-01 6.99627459e-01
1.00309722e-01 -2.55507886e-01 -1.41461164e-01 5.15224934e-02
2.05528721e-01 5.18200658e-02 3.19166780e-01 1.24102759e+00
-3.81314427e-01 -9.93622422e-01 4.25411724e-02 5.53949811e-02
-3.30230802e-01 3.36600363e-01 -3.63762528e-01 7.58698463e-01
-3.45109046e-01 4.85030651e-01 1.87567070e-01 -5.68434477e-01
5.51092625e-01 3.13838780e-01 1.17584133e+00 -6.23153508e-01
-2.70139724e-01 5.28518140e-01 2.88132608e-01 -1.28483236e+00
-7.47655153e-01 -9.56058204e-01 -1.10842121e+00 2.51567997e-02
1.40069693e-01 -6.88319564e-01 2.24760994e-01 1.05871058e+00
1.73971578e-01 4.60876107e-01 5.24291098e-01 -1.26783824e+00
-8.68842006e-01 -8.44384253e-01 -7.34886169e-01 5.82698226e-01
3.20543319e-01 -9.12169337e-01 -1.22539148e-01 2.30203077e-01] | [7.743653297424316, 0.14527355134487152] |
007c8436-f305-431e-b664-dd94b5100b9a | cdvae-co-embedding-deep-variational-auto | 1612.00132 | null | http://arxiv.org/abs/1612.00132v2 | http://arxiv.org/pdf/1612.00132v2.pdf | CDVAE: Co-embedding Deep Variational Auto Encoder for Conditional Variational Generation | Problems such as predicting a new shading field (Y) for an image (X) are
ambiguous: many very distinct solutions are good. Representing this ambiguity
requires building a conditional model P(Y|X) of the prediction, conditioned on
the image. Such a model is difficult to train, because we do not usually have
training data containing many different shadings for the same image. As a
result, we need different training examples to share data to produce good
models. This presents a danger we call "code space collapse" - the training
procedure produces a model that has a very good loss score, but which
represents the conditional distribution poorly. We demonstrate an improved
method for building conditional models by exploiting a metric constraint on
training data that prevents code space collapse. We demonstrate our model on
two example tasks using real data: image saturation adjustment, image
relighting. We describe quantitative metrics to evaluate ambiguous generation
results. Our results quantitatively and qualitatively outperform different
strong baselines. | ['Aditya Deshpande', 'Jiajun Lu', 'David Forsyth'] | 2016-12-01 | null | null | null | null | ['image-relighting'] | ['computer-vision'] | [ 7.12877631e-01 3.44672143e-01 4.10153657e-01 -9.59070325e-01
-1.19482422e+00 -5.53436339e-01 5.28304279e-01 -4.74500656e-02
2.19079889e-02 7.14421213e-01 2.26451442e-01 -1.26410127e-01
5.78154266e-01 -5.72442770e-01 -1.11079097e+00 -6.19313776e-01
1.17890246e-01 4.53556895e-01 2.49949917e-01 -1.69351041e-01
4.34058517e-01 2.04727486e-01 -1.67390108e+00 6.07293665e-01
1.01879513e+00 5.58628380e-01 6.32036865e-01 8.65398347e-01
-1.57831028e-01 5.84726155e-01 -9.27137852e-01 -5.70612311e-01
3.95873398e-01 -6.45880938e-01 -8.52411151e-01 4.61798608e-01
9.13528621e-01 -2.09281996e-01 3.25183898e-01 1.17097485e+00
2.29684696e-01 -8.25809762e-02 7.32474267e-01 -1.24129891e+00
-5.36588550e-01 3.13254058e-01 -7.86893725e-01 -4.75380450e-01
4.33239609e-01 1.86750174e-01 1.18005764e+00 -8.42753291e-01
7.41711795e-01 1.18723118e+00 7.59360313e-01 6.78943574e-01
-1.67856514e+00 -4.21143174e-01 2.82971691e-02 -3.01846802e-01
-1.17691290e+00 -1.63454056e-01 5.54648161e-01 -5.13931930e-01
4.92852598e-01 5.67021489e-01 4.26158011e-01 9.96225953e-01
6.62910938e-02 8.06851745e-01 1.36905932e+00 -4.32058305e-01
2.15293527e-01 1.62154853e-01 -7.80429021e-02 7.10927606e-01
-1.87582284e-01 1.33507624e-01 -6.46542013e-01 -1.64015919e-01
6.74376965e-01 -3.85670364e-01 -5.30848801e-01 -5.03961146e-01
-1.00043225e+00 6.71518683e-01 6.53232932e-01 3.29179354e-02
1.72860444e-01 4.33154941e-01 -2.85873562e-02 2.83440381e-01
1.45922825e-01 7.28457451e-01 -4.32178885e-01 1.45718411e-01
-9.42513227e-01 4.66204613e-01 8.84185731e-01 1.04323435e+00
1.26392603e+00 -1.76562205e-01 -2.03622222e-01 8.50160360e-01
1.63258836e-01 3.72888237e-01 2.19386935e-01 -1.45582974e+00
4.56052572e-01 1.48459837e-01 2.18354091e-01 -9.12596881e-01
-9.72083062e-02 -3.04676086e-01 -6.68283999e-01 7.50182152e-01
5.55068493e-01 -2.80617103e-02 -1.21268272e+00 1.97253239e+00
-1.03419803e-01 1.84393898e-01 -2.02946644e-02 7.83223271e-01
4.29029644e-01 9.03147876e-01 -1.87358916e-01 3.13117839e-02
7.86499798e-01 -1.11002481e+00 -3.39295179e-01 -5.66404581e-01
5.76407075e-01 -1.13112319e+00 1.60207260e+00 5.74354827e-01
-1.23246348e+00 -5.02364278e-01 -1.06760049e+00 -1.61601812e-01
-5.98728582e-02 -8.85501131e-03 6.94502234e-01 6.65826857e-01
-1.31101644e+00 7.70364642e-01 -5.69936633e-01 -1.64103270e-01
2.92367041e-01 1.77614644e-01 -2.76919127e-01 -2.18381226e-01
-5.81736147e-01 8.34235430e-01 -1.03457309e-02 -2.45533794e-01
-9.86851096e-01 -7.13604033e-01 -9.15811837e-01 7.37120509e-02
4.70593534e-02 -6.97887301e-01 1.34800184e+00 -1.45311463e+00
-1.13201857e+00 1.21538424e+00 -3.02864909e-01 -2.29596063e-01
5.60123622e-01 -3.69554669e-01 4.49213535e-02 -2.51475155e-01
2.57618427e-01 1.21860194e+00 9.59846258e-01 -2.32164955e+00
-4.97384876e-01 -1.42292500e-01 -3.45528387e-02 3.40249747e-01
7.20873028e-02 -1.47768646e-01 -5.45495272e-01 -7.37299979e-01
2.50337273e-01 -8.89604688e-01 -3.88996094e-01 2.03051493e-01
-7.66767979e-01 2.69848794e-01 4.66368377e-01 -5.99018037e-01
9.84165788e-01 -2.25854492e+00 4.67042476e-02 3.57978731e-01
8.33153874e-02 -2.08343580e-01 -2.04796568e-01 1.03947729e-01
-7.69847035e-02 2.79670268e-01 -9.07623947e-01 -7.40263641e-01
3.20255496e-02 6.41130745e-01 -7.65913427e-01 -2.81909686e-02
3.13214213e-01 5.02431989e-01 -8.89995277e-01 -4.49106127e-01
3.37792858e-02 3.69745642e-01 -8.99950266e-01 3.74692708e-01
-5.55983841e-01 4.09780890e-01 7.31157139e-02 4.62764621e-01
8.56012881e-01 -5.15712261e-01 1.26075521e-01 -4.67647761e-02
4.91580367e-02 2.25987807e-02 -1.14540148e+00 1.95737326e+00
-4.33115691e-01 7.52462268e-01 -1.41768128e-01 -6.62263691e-01
1.01796949e+00 -1.64451703e-01 1.08226381e-01 -3.72754276e-01
-2.36055911e-01 2.62336135e-01 -3.51423234e-01 -2.67285228e-01
6.22383416e-01 -1.87388271e-01 -3.44680510e-02 3.97093415e-01
-2.30384897e-02 -9.27457452e-01 -6.35697041e-04 3.63318950e-01
1.18497729e+00 4.20385838e-01 -2.29446679e-01 -3.58290285e-01
1.01554789e-01 -3.51646468e-02 6.59220755e-01 1.02369285e+00
4.12787125e-02 1.45356107e+00 5.89187086e-01 -3.89774442e-01
-1.23572969e+00 -1.23502874e+00 -1.20199300e-01 8.66979599e-01
4.85038131e-01 -4.78434443e-01 -9.31601048e-01 -5.78148127e-01
-1.66754961e-01 7.77655065e-01 -6.18274987e-01 2.65644968e-01
-5.69137633e-01 -7.73113847e-01 2.68734574e-01 1.90413207e-01
3.30286086e-01 -9.09116685e-01 -6.25340044e-01 -3.77364829e-02
-3.80591810e-01 -8.89037907e-01 -4.80430245e-01 4.35649723e-01
-6.68887436e-01 -8.96149993e-01 -5.74167132e-01 -1.00127816e+00
9.30115819e-01 2.27787837e-01 1.84908879e+00 5.89314699e-01
-4.46699798e-01 2.72412539e-01 -6.16638288e-02 -2.40843758e-01
-6.95899725e-01 -3.96796703e-01 -5.08230388e-01 -2.34314963e-01
-7.61218891e-02 -4.70727861e-01 -6.54448867e-01 3.46699864e-01
-1.06139052e+00 6.01514697e-01 5.03959119e-01 9.88877296e-01
6.60525620e-01 -1.87394604e-01 -5.88519610e-02 -1.03454256e+00
2.95815974e-01 -6.55089766e-02 -6.24354362e-01 5.84892452e-01
-3.83160651e-01 3.61069351e-01 5.04956007e-01 -9.04883817e-03
-1.33566523e+00 4.80597824e-01 -9.89603624e-02 -1.20012805e-01
-1.34936497e-01 2.04588741e-01 2.38974486e-02 4.69618812e-02
1.06638706e+00 -1.50046438e-01 -1.17852613e-01 -3.66335064e-01
4.68271226e-01 1.73132360e-01 8.63275766e-01 -8.68703604e-01
8.68391991e-01 5.53519607e-01 9.28383693e-02 -5.78601003e-01
-1.00994432e+00 -9.39963907e-02 -4.49108332e-01 -1.48959309e-01
9.34186220e-01 -7.81734049e-01 -8.96456838e-02 3.43409657e-01
-1.49284792e+00 -5.57175219e-01 -3.57179701e-01 5.53238355e-02
-7.40828454e-01 3.87134284e-01 -2.12010801e-01 -5.79217255e-01
2.25204423e-01 -1.20434439e+00 1.35002387e+00 1.14326105e-01
-2.66650081e-01 -8.52844417e-01 3.31317693e-01 1.86285943e-01
2.83604681e-01 5.24087667e-01 8.89143348e-01 5.56432307e-02
-8.83813918e-01 2.63208121e-01 -3.54892224e-01 5.44708788e-01
7.19377995e-02 2.37583324e-01 -1.24054217e+00 -1.41332760e-01
-2.29695700e-02 -7.84664810e-01 1.01857686e+00 2.61526883e-01
1.62401593e+00 -1.82776183e-01 -2.88729787e-01 6.82061911e-01
1.81235218e+00 -1.96446225e-01 1.02215672e+00 2.39092316e-02
7.00745881e-01 7.53821015e-01 6.19683087e-01 4.40970898e-01
2.48033941e-01 6.68508649e-01 5.32712340e-01 -4.64698315e-01
-4.38821316e-01 -2.68487602e-01 3.20025116e-01 3.98821652e-01
2.89870083e-01 -3.97861391e-01 -9.28356349e-01 5.02382576e-01
-1.81286716e+00 -7.83072174e-01 -2.86827445e-01 2.33301663e+00
1.20886111e+00 1.04486547e-01 -4.72819299e-01 -1.46053389e-01
6.82597458e-01 9.12190676e-02 -2.98551083e-01 -4.41299587e-01
-3.30782205e-01 4.03351605e-01 2.28258625e-01 1.11093318e+00
-9.59133446e-01 8.20252240e-01 7.23164034e+00 6.30606890e-01
-8.91797781e-01 -2.27105990e-01 1.17751336e+00 2.02368721e-01
-9.08443868e-01 4.92140532e-01 -5.48394322e-01 4.76893872e-01
3.39000732e-01 1.87847093e-01 4.45880622e-01 6.92653716e-01
-1.19565926e-01 -4.57276642e-01 -1.34344101e+00 1.16468155e+00
2.42661342e-01 -1.32720101e+00 1.12502955e-01 -1.30391046e-01
1.12901795e+00 4.71918657e-02 1.83514863e-01 -8.54018424e-03
8.50457549e-01 -1.23230457e+00 6.43243551e-01 5.22310674e-01
7.49854565e-01 -3.29597712e-01 2.75657624e-01 6.77255467e-02
-6.96341991e-01 1.92452073e-01 -4.76215273e-01 1.00338526e-01
2.10218802e-01 7.48658240e-01 -8.07311475e-01 2.26633176e-01
7.31229007e-01 5.00437200e-01 -9.41318810e-01 1.07187152e+00
-3.48799318e-01 4.69146818e-01 -3.47339481e-01 3.91484082e-01
5.37620373e-02 -1.96422502e-01 2.53683478e-01 1.20205772e+00
5.71192980e-01 -1.06448583e-01 1.58977732e-01 1.09724021e+00
-1.29574150e-01 -1.50607064e-01 -7.01783955e-01 5.36355019e-01
2.11918667e-01 1.05058026e+00 -5.66820621e-01 -2.18745768e-01
-4.62453783e-01 1.39775097e+00 4.18894410e-01 3.50615710e-01
-8.78396094e-01 3.65500338e-02 4.49679941e-01 -1.00242533e-01
5.12238871e-03 6.19339608e-02 -7.83354700e-01 -1.14085019e+00
1.28870741e-01 -7.65146792e-01 2.81048268e-01 -1.49980903e+00
-1.36697030e+00 6.02501690e-01 -2.02797055e-01 -1.37255263e+00
-3.54453415e-01 -4.76599067e-01 -6.36988223e-01 9.07694459e-01
-1.61594760e+00 -9.61098611e-01 -6.53862953e-01 4.18116450e-01
5.15506089e-01 6.21514469e-02 9.86880958e-01 1.96992382e-02
8.50231107e-03 4.33182389e-01 1.15176868e-02 -2.67975301e-01
1.06189919e+00 -1.85252810e+00 3.87161762e-01 9.90736127e-01
3.96083474e-01 3.66534114e-01 1.13726950e+00 -3.70480806e-01
-7.31696367e-01 -9.27332222e-01 9.79087412e-01 -7.00561702e-01
2.71025360e-01 -2.33867586e-01 -1.03831398e+00 6.48761272e-01
4.42436934e-01 -1.13583870e-01 6.35846019e-01 5.07766157e-02
-5.74168563e-01 7.26092830e-02 -1.04813480e+00 6.96492910e-01
1.04870152e+00 -3.46901894e-01 -2.96843499e-01 5.37127972e-01
6.92917347e-01 -4.83821124e-01 -4.35309291e-01 3.61110032e-01
3.19122434e-01 -1.64020085e+00 9.52753425e-01 -3.12797219e-01
9.65600848e-01 -4.53604937e-01 -3.63492370e-01 -1.41198838e+00
-1.05184242e-01 -6.82096004e-01 3.82539421e-01 1.01287413e+00
5.75172663e-01 -1.72106385e-01 8.95957947e-01 1.04331636e+00
-3.08296889e-01 -6.03599012e-01 -6.40034199e-01 -7.64252782e-01
1.98638692e-01 -3.58776987e-01 7.05633879e-01 8.77859771e-01
-3.28820735e-01 3.66603523e-01 -4.01951820e-01 1.77470669e-01
6.63389385e-01 3.53328496e-01 1.09979033e+00 -1.02303600e+00
-5.32034338e-01 -4.45430547e-01 -2.32964709e-01 -1.30156994e+00
4.86896699e-03 -7.03982949e-01 5.13626754e-01 -1.41111279e+00
4.16249901e-01 -8.15932930e-01 6.88156560e-02 5.65250218e-01
-4.22580272e-01 3.50854605e-01 2.76374221e-01 2.59797066e-01
-3.90155643e-01 3.41136813e-01 1.23675454e+00 -1.42676249e-01
-7.05651492e-02 -2.33840615e-01 -6.92067325e-01 7.80860186e-01
8.61016095e-01 -3.59783441e-01 -5.15342116e-01 -9.24770057e-01
4.49407279e-01 -1.16006322e-02 5.41805863e-01 -1.09878206e+00
-1.24268912e-01 -2.41353691e-01 4.39230531e-01 -4.61811274e-01
4.79110003e-01 -7.64163852e-01 2.98574775e-01 3.05665016e-01
-4.90971953e-01 7.40603637e-03 -6.84410334e-02 3.99738193e-01
-2.94002801e-01 -5.05449772e-01 9.12207067e-01 -3.71328145e-01
-7.17578709e-01 1.00321680e-01 1.14695586e-01 1.27514049e-01
6.53503478e-01 -2.28872299e-01 -4.07147050e-01 -5.65326989e-01
-8.09403360e-01 2.72644937e-01 1.08435297e+00 2.60945529e-01
7.64807224e-01 -1.28774095e+00 -7.50236094e-01 2.12932035e-01
2.79387832e-01 2.26154089e-01 -1.83143988e-01 1.10597983e-01
-7.61752248e-01 -3.71598035e-01 -1.24376245e-01 -8.56371224e-01
-1.29110479e+00 1.97142571e-01 3.28717619e-01 -2.55597532e-01
-4.14210558e-01 1.19468236e+00 7.61714041e-01 -5.39254487e-01
2.70731747e-01 -2.76492804e-01 2.12718397e-01 -4.26553071e-01
3.33670437e-01 -2.09999993e-01 -1.20605044e-01 -3.76961946e-01
4.13600579e-02 5.37835002e-01 9.95730385e-02 -6.55938685e-01
1.12270391e+00 -1.53290957e-01 -1.84135869e-01 3.42577130e-01
1.20469427e+00 -4.08765161e-03 -1.72247767e+00 -7.72122201e-03
-1.42177239e-01 -1.14770401e+00 -3.35890383e-01 -1.07086420e+00
-8.47215772e-01 9.52912688e-01 6.16398275e-01 2.28407994e-01
1.05917335e+00 -7.62445927e-02 5.81751108e-01 3.42134744e-01
3.63486379e-01 -1.18320799e+00 4.52238977e-01 4.58394319e-01
1.14576232e+00 -1.50062966e+00 -4.13751835e-03 -6.52148902e-01
-9.07323301e-01 1.04587746e+00 7.67377377e-01 -4.13039923e-02
5.93475163e-01 3.83321226e-01 3.46521109e-01 -2.96171933e-01
-6.01703584e-01 -9.70054865e-02 7.78038353e-02 7.44936585e-01
4.32569891e-01 3.31943557e-02 -2.32607909e-02 7.32145384e-02
-5.53857446e-01 -3.32477570e-01 7.69991219e-01 8.68327796e-01
-5.13477564e-01 -1.33432639e+00 -4.56527948e-01 3.72931898e-01
-1.56688437e-01 -2.10686445e-01 -5.54822803e-01 3.87568057e-01
8.91560167e-02 7.38484323e-01 8.08477923e-02 -3.13764125e-01
-7.87982568e-02 -4.96850051e-02 7.68090725e-01 -7.87919581e-01
-2.86449522e-01 -7.35346004e-02 -7.55939167e-03 -7.21210301e-01
-4.51985419e-01 -5.92529774e-01 -1.35762918e+00 -1.61037683e-01
-2.48960197e-01 -7.92584270e-02 7.41471231e-01 5.77614009e-01
1.89375773e-01 2.41313502e-01 9.41714704e-01 -8.98971558e-01
-1.63057446e-01 -4.21605617e-01 -4.70673263e-01 8.93711865e-01
3.82490188e-01 -3.50779772e-01 -5.54849982e-01 6.90507531e-01] | [9.910144805908203, -2.8825721740722656] |
ac6761ab-555a-4c1a-8cac-3baadfbcc9ec | reverberation-as-supervision-for-speech | 2211.08303 | null | https://arxiv.org/abs/2211.08303v1 | https://arxiv.org/pdf/2211.08303v1.pdf | Reverberation as Supervision for Speech Separation | This paper proposes reverberation as supervision (RAS), a novel unsupervised loss function for single-channel reverberant speech separation. Prior methods for unsupervised separation required the synthesis of mixtures of mixtures or assumed the existence of a teacher model, making them difficult to consider as potential methods explaining the emergence of separation abilities in an animal's auditory system. We assume the availability of two-channel mixtures at training time, and train a neural network to separate the sources given one of the channels as input such that the other channel may be predicted from the separated sources. As the relationship between the room impulse responses (RIRs) of each channel depends on the locations of the sources, which are unknown to the network, the network cannot rely on learning that relationship. Instead, our proposed loss function fits each of the separated sources to the mixture in the target channel via Wiener filtering, and compares the resulting mixture to the ground-truth one. We show that minimizing the scale-invariant signal-to-distortion ratio (SI-SDR) of the predicted right-channel mixture with respect to the ground truth implicitly guides the network towards separating the left-channel sources. On a semi-supervised reverberant speech separation task based on the WHAMR! dataset, using training data where just 5% (resp., 10%) of the mixtures are labeled with associated isolated sources, we achieve 70% (resp., 78%) of the SI-SDR improvement obtained when training with supervision on the full training set, while a model trained only on the labeled data obtains 43% (resp., 45%). | ['Jonathan Le Roux', 'Aswin Shanmugam Subramanian', 'Gordon Wichern', 'Christoph Boeddeker', 'Rohith Aralikatti'] | 2022-11-15 | null | null | null | null | ['speech-separation'] | ['speech'] | [ 3.12386960e-01 2.44694874e-01 4.10398692e-01 -1.58038855e-01
-9.98724997e-01 -6.30997837e-01 2.83582360e-01 -1.06510095e-01
-3.42191905e-01 6.34630263e-01 4.21490997e-01 -5.49315810e-01
-1.53222233e-01 -1.29173309e-01 -9.19574380e-01 -1.20151579e+00
-1.39199570e-01 1.72634318e-01 -9.90167186e-02 -4.53200266e-02
-4.58441138e-01 4.49763864e-01 -1.47545803e+00 3.37769121e-01
8.17782938e-01 1.00077522e+00 4.21511322e-01 1.13531625e+00
1.72440484e-01 7.03534305e-01 -8.76866162e-01 -7.84004927e-02
4.76565599e-01 -9.43149805e-01 -4.44694936e-01 8.99120495e-02
3.10894042e-01 -4.88107055e-02 -5.50850868e-01 9.54689443e-01
5.07604301e-01 1.71258211e-01 8.29216897e-01 -7.93165267e-01
-3.99146855e-01 9.27168012e-01 -2.23308563e-01 2.91184843e-01
-1.36397174e-02 -5.01720160e-02 9.76431251e-01 -7.77090371e-01
-1.46611845e-02 8.47146630e-01 4.83969182e-01 4.67940480e-01
-1.59668493e+00 -7.66197860e-01 2.15703994e-01 -1.72746226e-01
-1.27807629e+00 -1.08252215e+00 6.59976959e-01 -3.71745437e-01
6.68563783e-01 4.38178867e-01 3.52570206e-01 1.03837812e+00
-3.09910357e-01 5.22760987e-01 1.11586297e+00 -6.03768349e-01
2.67491519e-01 4.81893897e-01 2.41982624e-01 2.87224203e-01
-1.35309115e-01 3.30940634e-01 -5.83096087e-01 4.31627035e-02
6.06146038e-01 -5.46107769e-01 -7.35128462e-01 -7.65685812e-02
-1.02930248e+00 2.77738571e-01 6.24671400e-01 4.05847013e-01
-2.04913035e-01 -2.01831236e-01 -1.77981049e-01 3.90291214e-01
5.50118387e-01 5.42901278e-01 -5.12302339e-01 3.00538361e-01
-9.37018275e-01 -1.65189922e-01 8.97583306e-01 7.88902342e-01
6.19716167e-01 4.66087073e-01 1.78755164e-01 9.91403103e-01
3.57803762e-01 6.24929845e-01 4.50441152e-01 -8.39123189e-01
4.87398952e-01 -2.49402702e-01 1.92311868e-01 -5.85206509e-01
-9.27066654e-02 -1.07211804e+00 -8.27275455e-01 4.75220084e-02
7.70860434e-01 -4.18580234e-01 -1.03367329e+00 2.00855064e+00
1.06309503e-01 3.58685464e-01 4.48606700e-01 9.51345623e-01
4.93550688e-01 7.67653108e-01 -4.61448759e-01 -5.00378788e-01
8.13082337e-01 -9.25305486e-01 -5.62498569e-01 -5.75592816e-01
1.36247829e-01 -8.31060290e-01 7.67156303e-01 6.26127183e-01
-1.08355474e+00 -7.75022447e-01 -1.31525517e+00 4.89603847e-01
2.74458551e-04 1.09110191e-01 9.75636840e-02 6.95827663e-01
-9.37534153e-01 6.20999098e-01 -8.72965217e-01 2.06288338e-01
8.30045417e-02 6.75395578e-02 -2.24399269e-01 7.01272860e-02
-1.01950395e+00 6.09399319e-01 -9.85463858e-02 3.00925165e-01
-1.40457606e+00 -6.26095176e-01 -5.97443402e-01 2.77447939e-01
9.88403261e-02 -3.39481741e-01 1.15577900e+00 -1.39483535e+00
-1.44755864e+00 4.31147099e-01 -2.33639061e-01 -6.11736000e-01
4.41460252e-01 -3.26027632e-01 -7.11785436e-01 7.30448663e-02
-1.63568959e-01 3.00186604e-01 1.11561584e+00 -1.67438006e+00
-6.10170603e-01 -1.55235127e-01 -3.45889241e-01 2.43976519e-01
-8.04624259e-02 -9.88324434e-02 1.08935982e-01 -5.69898725e-01
3.35964411e-01 -8.12821448e-01 -4.54167500e-02 -2.95976460e-01
-5.84156811e-01 4.02808905e-01 2.16086328e-01 -1.27429664e+00
7.68984020e-01 -2.61401582e+00 1.85271531e-01 2.74322778e-01
1.29765049e-01 1.59801021e-01 -2.88168937e-01 8.98599699e-02
-5.26903212e-01 -2.01314569e-01 -4.29183394e-01 -7.70707250e-01
-1.05176397e-01 -1.27403244e-01 -6.06671393e-01 6.28966212e-01
-6.18952774e-02 -4.32610139e-02 -7.61912704e-01 1.68059900e-01
-6.37455285e-02 8.16411376e-01 -4.37141001e-01 6.55171990e-01
2.00390518e-01 9.12395835e-01 3.16576093e-01 -7.83645664e-04
5.37622452e-01 2.24467412e-01 7.44410902e-02 -7.74434879e-02
5.18423915e-02 5.94848990e-01 -1.30606401e+00 1.20938396e+00
-6.45219743e-01 8.06128561e-01 4.49646801e-01 -8.91495585e-01
7.70694315e-01 6.94854975e-01 2.24460393e-01 -4.46908116e-01
-8.24247524e-02 4.84420210e-01 6.21442199e-01 -9.36030969e-02
-4.19808440e-02 -3.91918868e-01 2.86154538e-01 2.75871128e-01
4.47846293e-01 6.07339107e-02 -3.17623734e-01 2.49141157e-02
9.39818680e-01 -1.40041277e-01 -1.61822751e-01 2.83688102e-02
2.20411599e-01 -6.10727072e-01 5.13089895e-01 7.11673081e-01
-7.04102069e-02 8.67714822e-01 1.49447665e-01 2.06238866e-01
-8.70121002e-01 -1.51424146e+00 -1.35740653e-01 8.57009888e-01
-1.39558807e-01 6.12990484e-02 -8.98648858e-01 -2.39582136e-01
-3.18361282e-01 1.01560080e+00 -2.13409513e-01 -2.33555079e-01
-5.14221847e-01 -6.19185328e-01 6.24340773e-01 3.87175858e-01
1.22668564e-01 -7.45573223e-01 -1.76046804e-01 2.04417557e-01
-3.43983382e-01 -1.10387361e+00 -4.59605396e-01 9.08286631e-01
-4.88560855e-01 -7.27773964e-01 -6.53752327e-01 -6.74491584e-01
7.71447301e-01 3.86195153e-01 6.87734485e-01 -2.44125396e-01
2.31594741e-02 1.74788058e-01 -1.05821997e-01 -2.45346680e-01
-6.82211101e-01 -4.30423349e-01 3.73734266e-01 4.31755573e-01
-4.36525971e-01 -9.17055309e-01 -4.40413028e-01 3.87859553e-01
-5.23208916e-01 -6.11090697e-02 3.12514991e-01 7.40252435e-01
4.04389799e-01 4.61759597e-01 6.79054797e-01 -4.24535781e-01
2.77484834e-01 -3.41344774e-01 -4.61786836e-01 4.80191642e-03
-2.36008584e-01 9.95688662e-02 8.01814139e-01 -7.39408076e-01
-1.21411288e+00 1.47631653e-02 -1.12879723e-01 -4.87963349e-01
-5.39016247e-01 1.76360458e-01 -4.70620424e-01 3.91038775e-01
8.50287497e-01 3.60913783e-01 -3.50286722e-01 -8.12153280e-01
5.04278898e-01 7.27915168e-01 8.84275496e-01 -3.69702637e-01
8.13611269e-01 1.46225438e-01 -4.77537036e-01 -1.02528214e+00
-8.40428293e-01 -5.06519735e-01 -4.65269089e-01 9.18430164e-02
7.93039978e-01 -1.02977490e+00 -3.46855283e-01 5.37931383e-01
-1.11500180e+00 -5.23499489e-01 -3.21078032e-01 8.78493249e-01
-3.88136894e-01 -6.24404512e-02 -5.59096098e-01 -1.36481953e+00
9.65759680e-02 -1.02573633e+00 4.74572361e-01 7.59718791e-02
-1.33570179e-01 -8.17053318e-01 -7.95787759e-03 5.01863301e-01
3.27628702e-01 -3.88855159e-01 1.04462981e+00 -1.07833099e+00
-2.88877845e-01 -2.77407348e-01 2.31974334e-01 1.05532777e+00
4.15772468e-01 -2.91103899e-01 -1.73953307e+00 -2.70966649e-01
6.16771936e-01 6.74829334e-02 9.36408937e-01 5.29578030e-01
8.86114419e-01 -4.81891692e-01 -1.24727495e-01 6.05428755e-01
9.79631484e-01 4.31918383e-01 4.98551816e-01 -2.69481450e-01
6.84799969e-01 7.25846410e-01 -1.83537304e-01 -4.00095880e-02
-3.54236960e-02 2.68439204e-01 2.11800724e-01 -3.31995130e-01
-4.85248685e-01 -3.23480636e-01 6.18328989e-01 1.15674150e+00
2.34012492e-02 -4.03529525e-01 -5.28186262e-01 6.22026622e-01
-1.29596889e+00 -7.00387895e-01 8.02686140e-02 2.74291348e+00
8.77677858e-01 3.57316047e-01 1.10869400e-01 4.83292192e-01
7.11684644e-01 -1.03794001e-02 -5.06765068e-01 -1.30420759e-01
-2.27589607e-01 3.82364124e-01 4.06967968e-01 9.93965447e-01
-9.81519282e-01 3.54005605e-01 5.93149233e+00 6.09503150e-01
-1.19308007e+00 5.30532412e-02 6.83558643e-01 -7.86276981e-02
-1.49433389e-01 5.66914119e-03 -5.52772462e-01 3.31746817e-01
1.31957793e+00 7.09892139e-02 9.78681862e-01 3.35359931e-01
1.83779284e-01 -4.16686870e-02 -1.54360795e+00 6.93030417e-01
4.50201146e-02 -4.88352954e-01 -4.57883507e-01 -2.95585138e-03
5.24112701e-01 -1.37765393e-01 3.52399439e-01 9.03614759e-02
4.31985199e-01 -1.08998227e+00 1.15745926e+00 5.06822169e-01
4.58414555e-01 -7.23213494e-01 4.46951926e-01 7.92004347e-01
-8.62193584e-01 -6.19595870e-02 -2.87955165e-01 -1.04159363e-01
-6.29962906e-02 7.59002030e-01 -1.14412618e+00 5.28517663e-01
4.99813139e-01 1.40147969e-01 -2.64273942e-01 1.07939351e+00
-5.09246826e-01 1.25724089e+00 -2.95767248e-01 5.43243051e-01
-2.89504439e-01 -8.59038234e-02 9.48746502e-01 1.13097906e+00
2.74160862e-01 -8.87793675e-02 3.58331725e-02 9.39082801e-01
-1.33135855e-01 -1.56542435e-01 -2.58880824e-01 -7.61689618e-02
4.08047915e-01 9.99522090e-01 -6.12919688e-01 -1.54851079e-01
-1.83123082e-01 9.17884886e-01 2.26660982e-01 8.86430264e-01
-7.68443763e-01 -4.77514923e-01 7.25940704e-01 1.09773509e-01
4.48925287e-01 -1.69743598e-01 -2.71556735e-01 -8.87304783e-01
-8.31423327e-02 -1.03195906e+00 -4.41226028e-02 -8.95656109e-01
-1.38481557e+00 9.72723186e-01 -1.76107511e-01 -1.21217155e+00
-1.42319381e-01 -3.70201439e-01 -4.75088149e-01 1.47877586e+00
-1.26561284e+00 -6.80724978e-01 2.51385123e-01 2.91525781e-01
4.15815949e-01 -1.22002386e-01 9.11629915e-01 3.32881987e-01
-5.14833510e-01 6.17012978e-01 2.14638129e-01 2.98292458e-01
8.22807074e-01 -1.33897424e+00 1.99881881e-01 1.07319033e+00
6.48697734e-01 5.81706405e-01 1.00249505e+00 -1.94979593e-01
-8.00807953e-01 -8.21249366e-01 7.13502586e-01 -3.12080115e-01
5.30570924e-01 -7.42279470e-01 -1.06668246e+00 6.58753633e-01
1.20653465e-01 7.57925585e-02 7.39481330e-01 -9.23663843e-03
-4.90465879e-01 -4.32970941e-01 -7.99093068e-01 5.29100418e-01
8.07943404e-01 -7.63009727e-01 -6.81890309e-01 5.53894527e-02
8.68422687e-01 -3.13477159e-01 -2.73132712e-01 -1.05606960e-02
3.74230713e-01 -8.83870900e-01 1.00961936e+00 -5.49212754e-01
3.05286467e-01 -3.54979426e-01 -4.97668594e-01 -1.93334496e+00
-1.67482078e-01 -5.06190658e-01 1.88596006e-02 1.41816652e+00
7.13630497e-01 -7.10737586e-01 1.83929741e-01 3.81902784e-01
-4.58298653e-01 -2.72359312e-01 -8.98535252e-01 -8.12899470e-01
1.95479766e-01 -5.59567332e-01 2.41547376e-01 7.18527555e-01
-2.68278778e-01 4.94895607e-01 -4.28436339e-01 7.81410098e-01
6.66457951e-01 -1.01661496e-01 4.97721434e-01 -8.88832688e-01
-9.50844109e-01 -2.80086786e-01 3.21527757e-02 -1.34079409e+00
2.19378307e-01 -8.89472425e-01 7.19768047e-01 -1.29396427e+00
-3.88135821e-01 -4.94153172e-01 -5.60054779e-01 1.71128392e-01
-2.23607019e-01 -1.41739547e-01 1.00148030e-01 1.88200176e-02
3.56069095e-02 4.82649535e-01 9.81382966e-01 -2.55146176e-01
-3.42545152e-01 5.38912654e-01 -7.59739399e-01 1.03756022e+00
4.57335353e-01 -6.20565414e-01 -4.39538777e-01 -5.05990088e-01
-2.12191671e-01 4.45511699e-01 3.47666442e-01 -1.25900221e+00
3.05230677e-01 2.49996111e-01 6.40470624e-01 -2.73461025e-02
5.67562461e-01 -9.93809640e-01 3.46212655e-01 1.94310531e-01
-7.45357990e-01 -6.93629205e-01 6.01132065e-02 6.63026690e-01
-1.09605663e-01 -1.14599571e-01 9.30340350e-01 7.94150829e-02
1.45817205e-01 -2.58909017e-01 -5.56120515e-01 8.34245607e-03
2.88665831e-01 -2.57625314e-03 -1.42810345e-01 -6.95350945e-01
-1.07482576e+00 -2.25521818e-01 -1.03509448e-01 6.89736903e-02
4.40806210e-01 -1.00115716e+00 -6.84506297e-01 4.49568778e-01
-3.92828494e-01 8.63715187e-02 1.59379750e-01 7.14136899e-01
1.76066726e-01 1.80749029e-01 1.41353965e-01 -4.09776181e-01
-1.11712885e+00 5.61931372e-01 9.38622415e-01 8.69009420e-02
-3.70516479e-01 1.24248064e+00 8.28574717e-01 -1.54802814e-01
5.50913632e-01 -5.23307741e-01 2.92228325e-03 -8.75853971e-02
4.57041651e-01 4.29041564e-01 1.88508421e-01 -7.74977446e-01
-1.88616306e-01 1.10993482e-01 1.93416029e-01 -5.71223974e-01
1.27485847e+00 -1.42470136e-01 8.02260414e-02 9.85387623e-01
1.25814593e+00 5.94611347e-01 -1.44085062e+00 -4.05976743e-01
-3.24769229e-01 -3.48323971e-01 3.66184711e-02 -1.09940994e+00
-1.00880814e+00 1.10152876e+00 6.53554857e-01 4.07706797e-01
1.18679118e+00 5.85510060e-02 4.58374292e-01 1.86367393e-01
1.49637043e-01 -5.70064306e-01 1.44547269e-01 3.73628676e-01
7.99490154e-01 -7.96070099e-01 -5.28074741e-01 -4.02061641e-01
-4.61902738e-01 8.27852666e-01 3.33381653e-01 -1.73461929e-01
8.22293699e-01 4.32679355e-01 3.40916425e-01 2.78548747e-01
-6.34067774e-01 -1.16227858e-01 4.77428973e-01 7.67894804e-01
4.43998456e-01 2.68975288e-01 6.23559296e-01 1.06509352e+00
-6.11171663e-01 -7.74180472e-01 3.30365688e-01 4.38751698e-01
-5.92298388e-01 -6.84914768e-01 -6.66326106e-01 2.26865709e-01
-4.10705417e-01 -3.64362478e-01 -2.87677079e-01 9.29669887e-02
2.83253431e-01 1.42329824e+00 4.27246578e-02 -3.09454501e-01
5.44477999e-01 3.57884467e-01 4.04344708e-01 -6.95390165e-01
-3.68072748e-01 7.98784316e-01 1.31195202e-01 1.42035872e-01
-2.07372352e-01 -4.51882213e-01 -1.21902835e+00 1.80495307e-01
-5.05322993e-01 3.69940758e-01 6.19721413e-01 1.14753151e+00
-1.68285728e-01 9.50872779e-01 9.05509889e-01 -8.61459911e-01
-5.85712016e-01 -1.01105309e+00 -9.00689423e-01 2.28406072e-01
1.02790117e+00 -3.27948630e-01 -1.07080340e+00 4.03914481e-01] | [15.174395561218262, 5.743254661560059] |
ed2c2d18-7248-444f-bf9b-41c5c3a44013 | accidental-learners-spoken-language | 2211.05103 | null | https://arxiv.org/abs/2211.05103v2 | https://arxiv.org/pdf/2211.05103v2.pdf | Accidental Learners: Spoken Language Identification in Multilingual Self-Supervised Models | In this paper, we extend previous self-supervised approaches for language identification by experimenting with Conformer based architecture in a multilingual pre-training paradigm. We find that pre-trained speech models optimally encode language discriminatory information in lower layers. Further, we demonstrate that the embeddings obtained from these layers are significantly robust to classify unseen languages and different acoustic environments without additional training. After fine-tuning a pre-trained Conformer model on the VoxLingua107 dataset, we achieve results similar to current state-of-the-art systems for language identification. More, our model accomplishes this with 5x less parameters. We open-source the model through the NVIDIA NeMo toolkit. | ['Boris Ginsburg', 'Samuel Kriman', 'Krishna C. Puvvada', 'Fei Jia', 'Travis M. Bartley'] | 2022-11-09 | null | null | null | null | ['spoken-language-identification'] | ['speech'] | [-2.59055138e-01 -3.31090838e-01 -1.74766779e-01 -7.35993028e-01
-1.05885804e+00 -9.56388474e-01 5.68195283e-01 -2.08924990e-02
-8.91565084e-01 1.21734485e-01 2.84321815e-01 -5.05942643e-01
4.62988198e-01 -2.42900372e-01 -5.33672810e-01 -2.45699435e-01
-1.76670969e-01 7.38564789e-01 -1.19847178e-01 -1.36192024e-01
-1.77172095e-01 3.68025839e-01 -1.41394937e+00 2.44764835e-01
6.66067183e-01 7.59227395e-01 -1.09645739e-01 1.02278614e+00
-4.80031557e-02 4.05294865e-01 -4.89630371e-01 -2.85693079e-01
2.76643395e-01 4.34562340e-02 -8.70075583e-01 -1.46980345e-01
9.72520232e-01 -4.91269559e-01 -5.12125731e-01 1.06317937e+00
7.46409357e-01 5.45005985e-02 5.26671648e-01 -7.56299734e-01
-1.01248372e+00 9.14925933e-01 1.30553037e-01 4.10850704e-01
3.65178615e-01 2.08607331e-01 9.96179521e-01 -1.06064582e+00
1.75550714e-01 1.76953816e+00 9.52062905e-01 8.84388387e-01
-1.35774982e+00 -8.76721382e-01 7.60562196e-02 -1.32092223e-01
-1.72013319e+00 -1.15710425e+00 5.45863032e-01 -4.21468586e-01
1.46589160e+00 1.93437710e-01 3.15212190e-01 1.60282302e+00
-2.98303843e-01 8.78462076e-01 9.96670902e-01 -5.04782677e-01
6.46033064e-02 5.10856569e-01 2.77290732e-01 7.81971216e-01
7.98180848e-02 1.25638351e-01 -5.76538444e-01 -3.85896146e-01
4.32660282e-01 -5.56935489e-01 1.35621965e-01 -1.83685705e-01
-1.12218297e+00 7.57885396e-01 6.58887029e-02 4.59277302e-01
1.73761979e-01 1.90577358e-01 7.68249750e-01 2.99350053e-01
5.16078591e-01 4.74884421e-01 -4.75870818e-01 -5.03112897e-02
-8.55325103e-01 -3.41356784e-01 8.38901520e-01 8.86409223e-01
6.46234751e-01 5.41420639e-01 3.21671575e-01 1.26675761e+00
5.36113143e-01 5.09251595e-01 9.94505227e-01 -7.06560910e-01
2.81955510e-01 1.12342529e-01 -4.12767023e-01 -1.17704555e-01
-2.86392927e-01 -4.81446922e-01 -3.84843290e-01 -1.13133028e-01
3.33127886e-01 -2.17063248e-01 -9.44288313e-01 1.73242831e+00
1.11266179e-02 2.41700083e-01 5.13570249e-01 5.60664654e-01
9.06433940e-01 4.37865555e-01 4.16751295e-01 4.95635331e-01
1.35282516e+00 -1.02006686e+00 -4.23932612e-01 -4.93332237e-01
7.96132803e-01 -9.17793393e-01 1.26267612e+00 3.14866781e-01
-6.41075313e-01 -7.88584828e-01 -1.27261436e+00 -2.52935499e-01
-5.55713832e-01 4.30229068e-01 8.12345266e-01 1.22253430e+00
-1.45879543e+00 2.17603371e-01 -8.57445955e-01 -8.70196640e-01
1.75753579e-01 6.00930035e-01 -5.36058962e-01 3.35650325e-01
-1.23740768e+00 7.09450424e-01 5.41953266e-01 -1.96608633e-01
-1.27853835e+00 -5.68108976e-01 -1.10055184e+00 -2.06670418e-01
-2.22655490e-01 -2.78903037e-01 1.36505508e+00 -9.45802510e-01
-1.74637616e+00 1.39737856e+00 -2.46232659e-01 -5.69480896e-01
6.49881363e-02 -9.73072201e-02 -8.48508656e-01 -7.11957440e-02
-5.44295870e-02 8.17698419e-01 6.04069054e-01 -1.08956671e+00
-3.83468688e-01 -1.56505480e-01 -1.72724053e-01 8.17456618e-02
-9.47010100e-01 5.73584914e-01 -3.58781606e-01 -5.96594274e-01
-2.35355705e-01 -8.84024501e-01 1.45871183e-02 -2.71609604e-01
-5.31528533e-01 -4.09471363e-01 5.87539911e-01 -6.51234388e-01
1.06425500e+00 -2.34489465e+00 -3.24461699e-01 1.34862423e-01
7.22205937e-02 7.04402447e-01 -4.98329908e-01 1.77635685e-01
1.41898496e-02 3.55039418e-01 -2.09461134e-02 -8.66290450e-01
3.52468252e-01 1.90455124e-01 -1.81191742e-01 4.89787251e-01
-1.85261350e-02 3.97739381e-01 -7.56950438e-01 -4.63060290e-01
2.81256557e-01 7.17352092e-01 -6.32627308e-01 3.85683507e-01
1.78491831e-01 2.33356938e-01 -1.10973544e-01 7.44058192e-01
7.15676248e-01 1.38313890e-01 4.40623462e-01 -3.02228797e-02
-1.16550818e-01 8.20593297e-01 -8.95999014e-01 1.85868466e+00
-7.78669477e-01 8.57365370e-01 3.66952091e-01 -8.34105492e-01
1.15708709e+00 2.48376548e-01 -1.77285366e-03 -4.70790863e-01
1.99014261e-01 6.31202936e-01 2.30277747e-01 -1.91210493e-01
5.21489322e-01 1.78493559e-01 -2.81141251e-01 4.23387945e-01
7.86628246e-01 -4.06897441e-02 -1.98945090e-01 1.35595701e-03
8.04603815e-01 -3.00275952e-01 1.39117569e-01 -5.92527270e-01
7.54727423e-01 -3.60158503e-01 3.12043965e-01 9.09020960e-01
-5.67545295e-01 5.31173110e-01 -2.79473722e-01 -4.29298460e-01
-1.03656864e+00 -1.42509758e+00 -5.50100088e-01 1.58359802e+00
-3.80195916e-01 -4.70677644e-01 -9.75512207e-01 -6.06798828e-01
6.34374097e-02 5.81139803e-01 -2.64112651e-01 -9.23535153e-02
-5.64607620e-01 -7.72709072e-01 1.49452615e+00 6.96283698e-01
3.96425873e-01 -6.66312873e-01 1.68285742e-01 1.29603162e-01
3.30784500e-01 -1.35715830e+00 -8.04260373e-01 5.05839586e-01
-3.56688380e-01 -5.55335820e-01 -3.22058141e-01 -1.44930601e+00
3.13838899e-01 -1.21127985e-01 1.19792926e+00 -9.54529271e-03
-2.23467201e-01 5.91358244e-01 4.20177821e-03 -1.62570149e-01
-8.96447539e-01 3.80253375e-01 8.06388617e-01 -1.25075519e-01
1.12025976e+00 -3.09440613e-01 -1.68872073e-01 3.01991757e-02
-5.82764268e-01 -5.11284649e-01 8.78972709e-02 7.67736614e-01
4.28245842e-01 -3.20319384e-01 3.74504417e-01 -4.53819335e-01
7.69464731e-01 -3.19380522e-01 -5.82234681e-01 3.80392194e-01
-2.88456887e-01 2.98368275e-01 8.58793557e-01 -7.49143839e-01
-7.13684618e-01 2.66589701e-01 -6.85023546e-01 -2.51654923e-01
-5.03961861e-01 -1.93496704e-01 -2.98530728e-01 -3.79929960e-01
6.11139536e-01 2.91774094e-01 -1.91196442e-01 -6.96574152e-01
4.20998752e-01 1.36932218e+00 7.19188094e-01 -8.20372880e-01
6.40728891e-01 3.31836700e-01 -7.64997065e-01 -1.10598302e+00
-4.47522432e-01 -3.94720644e-01 -7.78270423e-01 3.55104357e-01
9.08845365e-01 -1.34514582e+00 -7.43157685e-01 5.41188657e-01
-1.13115466e+00 -4.56578106e-01 5.12016527e-02 8.25267613e-01
-2.14259937e-01 3.61076087e-01 -8.56473804e-01 -1.02221918e+00
-4.32239920e-01 -1.31975162e+00 1.01874042e+00 8.83141756e-02
-4.33420807e-01 -1.47857881e+00 4.25692976e-01 1.05285510e-01
4.96814847e-01 -5.60036123e-01 5.83781421e-01 -1.49986029e+00
-2.67055668e-02 1.87693655e-01 5.72335208e-03 8.28866899e-01
1.38245299e-01 -3.77386026e-02 -1.58734548e+00 -5.06520152e-01
-2.21500769e-01 -7.36775160e-01 9.91847813e-01 -1.06905714e-01
9.71475720e-01 -5.37077785e-02 -1.09824672e-01 1.21486604e+00
1.02370834e+00 1.35433033e-01 1.05879018e-02 4.22872007e-01
9.76397216e-01 4.77484584e-01 -2.30439633e-01 -1.58178601e-02
7.85476267e-01 5.28215349e-01 -2.08206743e-01 -3.16387564e-01
-2.65092611e-01 -4.85809326e-01 6.79271162e-01 1.14746118e+00
5.94276726e-01 -1.90184399e-01 -1.25623679e+00 8.37063313e-01
-1.24314082e+00 -6.62431538e-01 3.30949485e-01 2.03856683e+00
1.06558204e+00 -1.31068572e-01 2.17879906e-01 -1.25197664e-01
5.50715506e-01 1.01032227e-01 -5.19891560e-01 -1.01955485e+00
-5.20925939e-01 4.64824736e-01 5.72863579e-01 1.00932097e+00
-1.46553385e+00 1.47336376e+00 8.04151535e+00 8.64382744e-01
-1.43591201e+00 1.99371979e-01 6.77563965e-01 4.93767485e-03
-4.47708786e-01 -1.59306347e-01 -1.25317383e+00 2.06415653e-01
1.46128142e+00 -1.02208480e-02 6.45712316e-01 1.06038356e+00
-3.31323177e-01 3.55176240e-01 -1.60636318e+00 1.10616744e+00
5.64113677e-01 -8.62369835e-01 -5.05064540e-02 -1.65812373e-02
4.22557980e-01 6.88934326e-01 4.34956074e-01 4.74491239e-01
7.21342444e-01 -1.26902032e+00 6.20578110e-01 -1.19478092e-01
9.03287649e-01 -4.56161827e-01 4.27905887e-01 5.88922855e-03
-1.15248263e+00 1.19704098e-01 -4.99011874e-01 1.61762878e-01
-1.24335170e-01 5.49056418e-02 -1.30526972e+00 1.03724590e-02
6.36762679e-01 3.56295735e-01 -7.82896101e-01 5.61712801e-01
3.27607132e-02 9.06134069e-01 -5.56586742e-01 -1.28727093e-01
2.17854425e-01 1.97108433e-01 4.13712710e-01 1.76484740e+00
1.25727639e-01 -5.90383291e-01 3.44546586e-01 7.35386193e-01
-2.67095923e-01 7.21129552e-02 -6.45194530e-01 -2.61153013e-01
7.43237019e-01 1.00124419e+00 -1.14065655e-01 -6.45704150e-01
-4.39369231e-01 1.06264591e+00 6.63172126e-01 4.23549652e-01
-3.66483063e-01 -2.22029030e-01 1.22327042e+00 -2.46016011e-01
9.79358703e-02 -5.28537214e-01 -1.10857725e-01 -1.34112608e+00
-4.14472729e-01 -1.17149770e+00 3.69034141e-01 -2.32293844e-01
-1.61075127e+00 1.06940627e+00 -2.77587742e-01 -7.18050480e-01
-5.36999226e-01 -1.18920779e+00 -4.73207504e-01 1.05036950e+00
-1.34524441e+00 -1.22452688e+00 2.28163496e-01 4.62853491e-01
4.57295060e-01 -8.77794325e-01 1.43345332e+00 4.46505398e-01
-6.47130549e-01 1.27534568e+00 3.02729487e-01 6.75271451e-01
9.01791632e-01 -1.25538576e+00 9.24645782e-01 8.04263115e-01
6.34174287e-01 1.10788488e+00 4.81276721e-01 -2.55475044e-01
-1.50665855e+00 -8.36371303e-01 9.85747099e-01 -6.35888994e-01
1.05996895e+00 -1.03610921e+00 -9.10536110e-01 9.15765166e-01
5.66724479e-01 1.73644722e-01 1.19957507e+00 4.79410678e-01
-1.04068923e+00 -2.25120619e-01 -1.14247513e+00 4.17340666e-01
1.04081953e+00 -1.47222412e+00 -5.30148149e-01 3.09655100e-01
7.52785683e-01 -1.21406250e-01 -8.10001075e-01 2.10010737e-01
7.51756370e-01 -7.52336323e-01 1.14140904e+00 -8.10417235e-01
-4.61704731e-01 1.63379859e-03 -5.17632484e-01 -1.37541175e+00
-1.76670402e-01 -6.36702716e-01 3.07294369e-01 1.54775465e+00
5.24368942e-01 -8.49373221e-01 5.08188665e-01 1.76672876e-01
-9.51621532e-02 -1.94777116e-01 -1.10509098e+00 -1.08072376e+00
6.08808458e-01 -5.98251104e-01 6.50053620e-01 1.09588027e+00
-1.37779964e-02 2.97906965e-01 -2.16275826e-01 3.68863016e-01
6.56031191e-01 -6.08643532e-01 6.28282368e-01 -8.37552965e-01
-4.38004732e-01 -3.85905266e-01 -4.58663821e-01 -1.10708237e+00
1.00933707e+00 -1.28261983e+00 1.19838072e-02 -8.24103296e-01
1.59681752e-01 -5.96301675e-01 -4.89065588e-01 4.08702999e-01
-8.14542770e-02 4.27077323e-01 -1.02417404e-03 -1.01078950e-01
-6.75158560e-01 3.17106038e-01 3.66195679e-01 -4.56716686e-01
-9.56284702e-02 -5.61550438e-01 -6.59022272e-01 8.32444370e-01
9.62169170e-01 -2.63612330e-01 -3.02235782e-01 -1.12860084e+00
-4.36812401e-01 -7.39104927e-01 6.52443096e-02 -1.14825785e+00
2.87271857e-01 3.85035932e-01 3.03702801e-01 -1.61430970e-01
5.46682298e-01 -3.51108879e-01 -4.01647866e-01 2.21871898e-01
-4.83290493e-01 1.62287936e-01 6.61252737e-01 -1.47382528e-01
-2.08007872e-01 -1.72668591e-01 6.48060918e-01 5.49639128e-02
-8.81058693e-01 6.87368587e-02 -6.68008089e-01 2.38077670e-01
3.48831922e-01 1.18958406e-01 -3.11816901e-01 -2.15440676e-01
-6.52727425e-01 1.42779469e-01 5.26827753e-01 6.87408209e-01
3.40059578e-01 -1.41193557e+00 -8.11906695e-01 6.73455417e-01
3.26150864e-01 -7.39781201e-01 -1.20012797e-01 2.77698100e-01
-5.13252676e-01 6.75023794e-01 7.36331269e-02 -7.67065048e-01
-1.22022653e+00 2.83057690e-01 6.67239308e-01 6.65197670e-02
4.66567511e-03 1.23285186e+00 2.72359371e-01 -1.27104056e+00
5.32720804e-01 -2.85888553e-01 6.32028952e-02 -2.40572751e-01
6.92120194e-01 -8.92020315e-02 3.05637233e-02 -1.02116561e+00
-9.30701673e-01 5.55122674e-01 -1.86875418e-01 -4.45143610e-01
8.47431421e-01 -1.20844357e-02 1.86352119e-01 6.03923619e-01
1.67777133e+00 5.07617116e-01 -7.98213899e-01 -2.48165905e-01
2.87006833e-02 -5.82418442e-02 1.28714681e-01 -6.02331102e-01
-6.33544087e-01 1.05327535e+00 1.01772809e+00 -1.29168266e-02
6.45757675e-01 1.17267326e-01 7.09679246e-01 5.33478200e-01
2.20817149e-01 -1.26979005e+00 -2.25728512e-01 7.62185514e-01
4.30727422e-01 -1.47296524e+00 -4.15114313e-01 -1.88676164e-01
-4.06481147e-01 1.10277188e+00 6.48200452e-01 -1.73478991e-01
6.54612660e-01 7.97483325e-01 7.69128919e-01 3.53536278e-01
-6.49976730e-01 -2.35645562e-01 4.02579308e-01 9.35911357e-01
1.04444671e+00 4.25872058e-01 1.24207266e-01 6.12492204e-01
-5.75887442e-01 -6.80195689e-01 1.78350862e-02 4.28218573e-01
-2.40493238e-01 -1.42171419e+00 -5.77607512e-01 -8.03356245e-02
-5.02205789e-01 -5.17487705e-01 -6.97591305e-01 5.52120745e-01
5.82765043e-02 1.03711474e+00 4.45224702e-01 -7.70625293e-01
3.31294276e-02 1.93307668e-01 3.93191963e-01 -7.57890821e-01
-8.77908051e-01 6.39861403e-03 5.07042229e-01 -3.61573994e-01
-9.15228482e-03 -5.90262949e-01 -1.17907333e+00 -2.49187276e-01
5.25623225e-02 1.31070346e-01 7.34322965e-01 6.77075267e-01
2.79105216e-01 2.58332074e-01 5.70503175e-01 -7.46899009e-01
-9.06562388e-01 -1.02262223e+00 -3.03239673e-01 2.53915220e-01
5.76022923e-01 -1.90869629e-01 -6.55704916e-01 -3.80512118e-01] | [14.192110061645508, 6.626827716827393] |
9689b8a2-6c1c-4110-894b-9e36554fa697 | uncertainty-aware-multiple-instance-learning-1 | 2302.03116 | null | https://arxiv.org/abs/2302.03116v1 | https://arxiv.org/pdf/2302.03116v1.pdf | Uncertainty-Aware Multiple-Instance Learning for Reliable Classification: Application to Optical Coherence Tomography | Deep learning classification models for medical image analysis often perform well on data from scanners that were used during training. However, when these models are applied to data from different vendors, their performance tends to drop substantially. Artifacts that only occur within scans from specific scanners are major causes of this poor generalizability. We aimed to improve the reliability of deep learning classification models by proposing Uncertainty-Based Instance eXclusion (UBIX). This technique, based on multiple-instance learning, reduces the effect of corrupted instances on the bag-classification by seamlessly integrating out-of-distribution (OOD) instance detection during inference. Although UBIX is generally applicable to different medical images and diverse classification tasks, we focused on staging of age-related macular degeneration in optical coherence tomography. After being trained using images from one vendor, UBIX showed a reliable behavior, with a slight decrease in performance (a decrease of the quadratic weighted kappa ($\kappa_w$) from 0.861 to 0.708), when applied to images from different vendors containing artifacts; while a state-of-the-art 3D neural network suffered from a significant detriment of performance ($\kappa_w$ from 0.852 to 0.084) on the same test set. We showed that instances with unseen artifacts can be identified with OOD detection and their contribution to the bag-level predictions can be reduced, improving reliability without the need for retraining on new data. This potentially increases the applicability of artificial intelligence models to data from other scanners than the ones for which they were developed. | ['Clara I. Sánchez', 'Caroline C. W. Klaver', 'Carel B. Hoyng', 'Bram van Ginneken', 'Coen de Vente'] | 2023-02-06 | null | null | null | null | ['multiple-instance-learning'] | ['methodology'] | [ 1.11428164e-01 9.93919745e-02 1.34122089e-01 -6.42157972e-01
-1.06306839e+00 -2.74348944e-01 1.12402029e-01 3.70253384e-01
-5.52469850e-01 7.46091783e-01 -2.66377449e-01 -4.19334143e-01
-4.80582803e-01 -7.32681811e-01 -9.56140220e-01 -7.29617655e-01
-1.26599580e-01 5.44485748e-01 1.83391571e-01 5.47242939e-01
-9.07262489e-02 5.69776475e-01 -1.50131655e+00 4.80477333e-01
8.76237273e-01 1.27261126e+00 1.31995216e-01 8.56474638e-02
-1.28648967e-01 5.87124825e-01 -7.76867986e-01 -2.86446273e-01
3.85662079e-01 -1.30839244e-01 -4.30173814e-01 2.16517180e-01
9.04864490e-01 -5.81903160e-01 1.60898396e-03 8.37617099e-01
7.36516953e-01 -3.31284374e-01 7.67014086e-01 -8.93539548e-01
-5.42657852e-01 3.40483248e-01 -5.99374354e-01 1.77012995e-01
-2.25790054e-01 4.79327500e-01 5.35565257e-01 -5.23713589e-01
5.17911494e-01 7.12273657e-01 9.76837814e-01 4.05195951e-01
-1.53618431e+00 -8.03229928e-01 -1.35313451e-01 1.56327382e-01
-1.16998255e+00 -8.23884159e-02 8.98838416e-02 -8.32902312e-01
9.13595021e-01 6.81065992e-02 5.37190497e-01 9.24589097e-01
3.26606333e-01 2.56237298e-01 1.20294130e+00 -3.18520397e-01
3.74025136e-01 3.13545942e-01 1.99280143e-01 5.88339508e-01
6.92341208e-01 1.26372417e-02 -1.87049359e-01 -1.62400663e-01
7.54088521e-01 -2.36708790e-01 -2.42084220e-01 -4.63033319e-01
-8.04079950e-01 6.91553175e-01 4.52805281e-01 3.36453170e-01
-4.76782471e-01 -9.11209434e-02 5.12838304e-01 1.44608200e-01
3.65876853e-01 5.48932195e-01 -5.50063372e-01 2.43032649e-01
-1.11180556e+00 1.21178357e-02 4.32122350e-01 5.99739611e-01
3.59420180e-01 -1.61711201e-01 -2.89815545e-01 1.05570102e+00
-6.07339479e-02 3.69178742e-01 6.50266409e-01 -7.32532382e-01
9.40965191e-02 7.15242863e-01 -5.66287823e-02 -6.04796231e-01
-7.74583340e-01 -8.82473707e-01 -7.95532763e-01 5.39382637e-01
7.70061910e-01 -1.95566818e-01 -1.36636055e+00 1.47147644e+00
3.80173065e-02 -2.42591456e-01 -3.45164180e-01 9.27752078e-01
9.22691464e-01 2.74011940e-02 2.15103030e-01 -1.36896193e-01
1.27170539e+00 -3.59696895e-01 -3.20658654e-01 -1.87075362e-01
7.91926324e-01 -3.64768565e-01 1.06217992e+00 7.46055365e-01
-8.28526318e-01 -5.35373569e-01 -1.01456928e+00 3.04999202e-01
-8.78077149e-02 1.43635407e-01 5.20330667e-01 8.17619622e-01
-9.69398141e-01 7.04996288e-01 -7.63522089e-01 -2.13314503e-01
9.46055591e-01 6.96234226e-01 -6.36034667e-01 -3.40484053e-01
-8.57720673e-01 1.07095110e+00 1.60051212e-01 -8.50920007e-02
-6.64912164e-01 -1.12396896e+00 -5.52416861e-01 -3.15712988e-02
3.02819371e-01 -6.62780166e-01 9.42638814e-01 -1.01980197e+00
-8.76307189e-01 1.03822029e+00 1.85278639e-01 -6.26635253e-01
6.35931253e-01 -7.52824023e-02 -4.55303133e-01 1.32537326e-02
3.08933526e-01 5.47490001e-01 8.89549911e-01 -1.27647960e+00
-5.22783816e-01 -7.34692395e-01 -2.40470231e-01 -4.61453766e-01
-6.41467841e-03 -2.04413280e-01 -2.91456521e-01 -4.14530009e-01
1.81363732e-01 -1.00921786e+00 -1.75678477e-01 1.47299781e-01
-2.70274490e-01 -6.63638785e-02 2.49543384e-01 -7.34470904e-01
9.17557299e-01 -1.96174622e+00 -3.55373293e-01 3.76709133e-01
4.13408637e-01 4.17304903e-01 -1.30487695e-01 -2.47636199e-01
-4.47243869e-01 2.08309054e-01 -3.65043521e-01 -7.55094178e-03
-3.84200662e-01 8.61398354e-02 3.59119922e-01 3.54301959e-01
4.61670071e-01 4.14449245e-01 -5.70422649e-01 -2.83473313e-01
3.83408427e-01 3.53867978e-01 -5.01467466e-01 -4.74506617e-02
1.20979808e-02 4.47871774e-01 -1.65147036e-02 5.48868537e-01
8.42795670e-01 -5.41989207e-01 1.59304023e-01 -2.90305912e-01
7.74167925e-02 9.12734587e-03 -1.03209519e+00 1.51052558e+00
-6.98661029e-01 5.90338171e-01 -2.52095371e-01 -9.81970310e-01
7.39168525e-01 1.43247917e-01 7.53715396e-01 -9.54031289e-01
3.84829678e-02 4.14797753e-01 6.48292243e-01 -7.27525234e-01
-2.32702494e-01 -4.25315768e-01 3.48054469e-01 1.55452192e-01
2.63709188e-01 7.54571185e-02 4.84345183e-02 -2.25338548e-01
1.29853630e+00 -1.37900814e-01 5.98423705e-02 -1.99977756e-01
2.18465120e-01 1.41113540e-02 4.74496901e-01 1.00842774e+00
-2.53289968e-01 9.33638036e-01 6.40630662e-01 -5.60445249e-01
-1.05048215e+00 -1.17012489e+00 -9.82338309e-01 3.89381647e-01
-3.49281698e-01 -1.05118573e-01 -6.49455428e-01 -9.08998668e-01
3.33255708e-01 7.86654532e-01 -6.49115324e-01 -1.19722813e-01
-1.72354445e-01 -1.17214215e+00 2.98189372e-01 4.58980411e-01
3.17249566e-01 -6.87855482e-01 -7.21344948e-01 2.71037310e-01
8.08617473e-02 -1.05402792e+00 3.15975606e-01 3.92658859e-01
-1.13314867e+00 -1.20372665e+00 -8.61965418e-01 -1.78738087e-01
4.97840047e-01 -2.48462632e-01 1.23903811e+00 -2.17665806e-02
-5.82490623e-01 3.79826576e-01 -1.08587041e-01 -6.55965745e-01
-4.81242627e-01 -4.50674109e-02 4.66160849e-02 -8.55754614e-02
5.15259981e-01 -3.52188319e-01 -7.05468893e-01 3.46297652e-01
-8.93123090e-01 -4.06603545e-01 7.88017631e-01 9.67997074e-01
6.39295816e-01 7.26240650e-02 6.26759052e-01 -9.99013066e-01
1.73463702e-01 -4.47968006e-01 -6.77989781e-01 4.61378172e-02
-9.60052133e-01 9.40457657e-02 2.61171848e-01 -4.24771070e-01
-8.06104660e-01 -7.38370838e-03 -6.25502900e-04 -4.71982419e-01
-3.43254834e-01 4.22693133e-01 2.33797476e-01 -1.67240039e-01
9.50724781e-01 -1.54506326e-01 3.00203443e-01 -4.97934699e-01
-1.98172674e-01 7.29295433e-01 1.30574703e-01 -2.73072720e-01
2.47113124e-01 3.14087600e-01 2.28135988e-01 -7.84069836e-01
-7.67461658e-01 -2.93232620e-01 -4.31274414e-01 -2.12332532e-01
6.79118693e-01 -8.84780228e-01 -4.60319102e-01 5.50285161e-01
-8.01292419e-01 -2.80913353e-01 -2.21175537e-01 7.24676490e-01
-2.00760543e-01 2.36528486e-01 -2.45762438e-01 -6.45422578e-01
-1.98316410e-01 -1.40932834e+00 8.35870624e-01 -2.00551841e-02
-3.31795871e-01 -6.79563284e-01 -2.25614712e-01 6.11811280e-01
3.70381743e-01 3.59011859e-01 1.30015826e+00 -8.33603203e-01
-3.78442198e-01 -3.76554787e-01 -4.48881924e-01 8.87907267e-01
1.93653718e-01 -6.30170926e-02 -1.19952357e+00 -1.11249581e-01
2.28923589e-01 -1.22514822e-01 9.01847363e-01 9.17249680e-01
1.61626768e+00 1.00127459e-01 -1.36294603e-01 2.90895522e-01
1.58811116e+00 2.70460963e-01 7.93226242e-01 5.67552149e-01
3.39920431e-01 6.70219898e-01 1.97018206e-01 2.84283102e-01
-1.85939953e-01 7.23920763e-01 6.20250285e-01 -1.79332644e-01
-2.84346968e-01 3.25118780e-01 3.85451131e-02 -8.46295208e-02
-1.21016711e-01 5.88974655e-02 -1.06582081e+00 6.00064516e-01
-1.49360406e+00 -5.20159066e-01 -5.95832944e-01 2.53389192e+00
5.39880574e-01 4.33196396e-01 5.60691692e-02 -2.16477048e-02
7.64147282e-01 -3.20365220e-01 -7.11855412e-01 -2.25693554e-01
-5.03816502e-03 6.32775128e-01 6.85887158e-01 -2.40133535e-02
-8.26118290e-01 2.43658885e-01 6.20441914e+00 7.07412183e-01
-1.38121772e+00 2.28506118e-01 7.27139890e-01 -4.65274751e-01
1.11734740e-01 -3.35818857e-01 -3.85268360e-01 7.88789928e-01
1.04605770e+00 3.99975657e-01 -5.34858592e-02 7.45869458e-01
2.96489954e-01 -5.85818350e-01 -1.14256263e+00 1.03062463e+00
-5.80678321e-03 -1.25263464e+00 -5.25529608e-02 2.25113764e-01
4.01561975e-01 1.59081429e-01 7.30885491e-02 2.29253039e-01
-3.18669349e-01 -1.25237072e+00 3.08506519e-01 5.29200852e-01
9.21594679e-01 -3.95408899e-01 1.18489504e+00 -2.77585592e-02
-1.69611782e-01 -2.63340995e-02 -2.84525335e-01 6.42184541e-02
-1.91657618e-01 1.26694667e+00 -1.14804506e+00 4.33572471e-01
1.01756477e+00 1.93256974e-01 -7.61668146e-01 1.19891417e+00
7.86947906e-02 6.28515482e-01 -3.46485883e-01 2.96837419e-01
2.67915279e-02 -2.05199756e-02 3.90188277e-01 8.96731079e-01
4.18394715e-01 -1.05121426e-01 -4.57772940e-01 1.01280844e+00
1.55813750e-02 1.69797745e-02 -4.65074509e-01 3.16452414e-01
7.53422752e-02 1.15233850e+00 -4.85883355e-01 -1.27472982e-01
-6.67030036e-01 6.22196138e-01 3.00022587e-02 6.62145466e-02
-9.39775705e-01 -1.02778137e-01 3.39076906e-01 8.06407571e-01
1.59729257e-01 2.72440672e-01 -4.68333453e-01 -7.89802134e-01
2.33605847e-01 -7.60447741e-01 4.55743253e-01 -9.20269430e-01
-1.55538833e+00 3.86688560e-01 -9.22847986e-02 -1.33740461e+00
3.33037786e-02 -9.90859330e-01 -1.25595778e-01 1.06806815e+00
-1.13125360e+00 -8.64376843e-01 -3.45641047e-01 2.90498048e-01
1.46810502e-01 -2.74950564e-01 7.45900869e-01 5.20030499e-01
-4.07514393e-01 6.00442588e-01 1.90122321e-01 1.37349233e-01
1.04118478e+00 -1.25631893e+00 -1.60873502e-01 5.49778521e-01
-1.52857676e-01 5.12646317e-01 4.24197912e-01 -5.22245407e-01
-6.95127189e-01 -1.05948997e+00 3.62043202e-01 -5.37329674e-01
3.78447026e-01 2.48958170e-01 -1.14839196e+00 4.82821286e-01
-2.45714307e-01 1.38894796e-01 9.30427551e-01 3.48084480e-01
-3.74297470e-01 -3.71962517e-01 -1.48811746e+00 2.01500267e-01
7.84271896e-01 -2.45631412e-01 -3.41921479e-01 3.45217079e-01
2.25904673e-01 -3.62507224e-01 -1.24350619e+00 6.47099972e-01
7.17831016e-01 -1.41928077e+00 7.40050495e-01 -6.24173880e-01
5.69200218e-01 -1.43544108e-01 -2.15625726e-02 -1.12951136e+00
-2.90983319e-01 2.37988830e-01 1.80213600e-01 9.04040098e-01
5.98779440e-01 -6.66982472e-01 7.15265036e-01 1.02423131e+00
-2.34641269e-01 -7.17566013e-01 -1.11974502e+00 -8.44951093e-01
2.87761629e-01 -6.19955182e-01 2.45332971e-01 7.94098675e-01
-4.69842702e-01 1.19086832e-01 5.22414036e-02 2.92406827e-01
7.31555879e-01 -3.17438900e-01 4.52283323e-01 -1.40493584e+00
-4.30656374e-01 -2.21472636e-01 -7.49501646e-01 -7.30720907e-02
-1.89321607e-01 -9.47998524e-01 -1.68677807e-01 -1.50516796e+00
3.04081917e-01 -6.88686907e-01 -4.83845264e-01 5.57334006e-01
5.85275926e-02 3.41488034e-01 1.05760150e-01 2.01073334e-01
3.16563286e-02 -2.32675951e-02 1.14850903e+00 -2.55176604e-01
-5.27581237e-02 2.30041996e-01 -6.39818966e-01 7.23462343e-01
7.06468165e-01 -6.97380304e-01 -1.96151167e-01 -5.16355693e-01
7.15313777e-02 -1.15877904e-01 6.76671922e-01 -1.44933438e+00
-2.93919683e-01 4.63959485e-01 8.96124780e-01 -3.35414380e-01
3.30662802e-02 -9.86807823e-01 4.41441298e-01 6.66031480e-01
-1.85474932e-01 -4.11833316e-01 5.66258490e-01 3.99215132e-01
8.94460827e-02 -5.40748119e-01 9.98414755e-01 -2.96590865e-01
-4.53556389e-01 1.62596568e-01 -3.48917574e-01 -1.10144660e-01
9.07454610e-01 -3.48624676e-01 -1.55482188e-01 -5.14177755e-02
-1.14229047e+00 -1.89972535e-01 3.53442907e-01 2.51062453e-01
2.44337037e-01 -9.14386392e-01 -6.73440456e-01 2.21302807e-01
3.42180818e-01 2.49237537e-01 6.38463616e-01 1.01169729e+00
-3.75172138e-01 3.25234175e-01 -2.95022428e-01 -1.16837418e+00
-1.14037502e+00 2.74011314e-01 7.72917926e-01 -2.36053273e-01
-5.42103708e-01 6.63327456e-01 1.53989434e-01 -2.35727191e-01
1.51487663e-01 -5.98883212e-01 1.14592090e-01 1.65739760e-01
3.46480012e-01 4.24116343e-01 9.13957119e-01 -1.08064041e-01
-5.21224320e-01 5.19211352e-01 -3.05685252e-01 7.78656453e-02
1.32705629e+00 3.22049648e-01 -6.47142753e-02 4.10882264e-01
1.01827800e+00 -2.17295662e-01 -1.01800084e+00 1.10408038e-01
-1.61067352e-01 -4.03720438e-01 4.09321547e-01 -1.32914841e+00
-1.25263727e+00 8.26118946e-01 1.44523561e+00 1.43981576e-01
1.02403772e+00 -2.65154224e-02 4.56167758e-01 5.55956503e-03
4.52669770e-01 -8.96915674e-01 -2.17515305e-01 1.11911118e-01
6.95817232e-01 -1.63211060e+00 1.98377296e-01 -2.36063570e-01
-6.01641476e-01 1.05758286e+00 5.71492970e-01 5.34142405e-02
6.05218112e-01 1.64922908e-01 7.62907639e-02 -2.35170290e-01
-2.54824221e-01 -9.16065052e-02 3.14421326e-01 9.87692475e-01
5.08505762e-01 1.01542681e-01 -4.26160455e-01 8.21403682e-01
2.44128332e-01 4.47526753e-01 6.89964831e-01 6.01676762e-01
-1.29423052e-01 -1.04151523e+00 -3.43539536e-01 1.22755015e+00
-5.35709620e-01 -6.62456453e-02 -1.27120271e-01 1.11914039e+00
6.11414790e-01 7.11404800e-01 3.84521753e-01 -2.31359452e-01
4.59186465e-01 2.17682734e-01 7.32758880e-01 -7.48899460e-01
-6.70217156e-01 -4.21322733e-02 1.76929116e-01 -6.53465509e-01
-5.17409980e-01 -6.64745212e-01 -1.03128862e+00 -2.36472003e-02
-3.33277196e-01 -2.83240467e-01 7.28117943e-01 7.88560569e-01
4.27375138e-01 7.73862123e-01 1.92764446e-01 -3.93933326e-01
-5.50967693e-01 -1.00075579e+00 -8.42577875e-01 4.29045439e-01
3.73887569e-01 -8.44196856e-01 -3.98680359e-01 -1.27295628e-01] | [15.017261505126953, -2.435399055480957] |
2aed6653-8b1c-46f3-860a-9a78f67e1652 | usd-unknown-sensitive-detector-empowered-by | 2306.02275 | null | https://arxiv.org/abs/2306.02275v1 | https://arxiv.org/pdf/2306.02275v1.pdf | USD: Unknown Sensitive Detector Empowered by Decoupled Objectness and Segment Anything Model | Open World Object Detection (OWOD) is a novel and challenging computer vision task that enables object detection with the ability to detect unknown objects. Existing methods typically estimate the object likelihood with an additional objectness branch, but ignore the conflict in learning objectness and classification boundaries, which oppose each other on the semantic manifold and training objective. To address this issue, we propose a simple yet effective learning strategy, namely Decoupled Objectness Learning (DOL), which divides the learning of these two boundaries into suitable decoder layers. Moreover, detecting unknown objects comprehensively requires a large amount of annotations, but labeling all unknown objects is both difficult and expensive. Therefore, we propose to take advantage of the recent Large Vision Model (LVM), specifically the Segment Anything Model (SAM), to enhance the detection of unknown objects. Nevertheless, the output results of SAM contain noise, including backgrounds and fragments, so we introduce an Auxiliary Supervision Framework (ASF) that uses a pseudo-labeling and a soft-weighting strategies to alleviate the negative impact of noise. Extensive experiments on popular benchmarks, including Pascal VOC and MS COCO, demonstrate the effectiveness of our approach. Our proposed Unknown Sensitive Detector (USD) outperforms the recent state-of-the-art methods in terms of Unknown Recall, achieving significant improvements of 14.3\%, 15.5\%, and 8.9\% on the M-OWODB, and 27.1\%, 29.1\%, and 25.1\% on the S-OWODB. | ['Siqi Wang', 'Yusong Tan', 'Wei Chen', 'Yulin He'] | 2023-06-04 | null | null | null | null | ['open-world-object-detection'] | ['computer-vision'] | [ 1.75511956e-01 1.19539171e-01 -1.18713170e-01 -2.84371585e-01
-8.12409759e-01 -3.38409215e-01 4.88273203e-01 -8.32384601e-02
-6.41528249e-01 5.15406072e-01 -3.07423115e-01 -4.32350636e-02
2.03631058e-01 -5.55938125e-01 -9.10897493e-01 -7.27678299e-01
3.32498014e-01 1.86557204e-01 9.58096862e-01 1.52078778e-01
1.27228722e-01 1.20419882e-01 -1.52054286e+00 4.19505276e-02
1.00833678e+00 1.35325944e+00 6.38384461e-01 3.87267381e-01
-2.00668558e-01 6.97185338e-01 -5.04812300e-01 -5.85128665e-01
3.17445099e-01 -9.45954919e-02 -3.79322499e-01 2.59414285e-01
7.40493774e-01 -5.94826281e-01 -2.92256743e-01 1.30638874e+00
5.14936566e-01 -4.81695421e-02 6.25583112e-01 -1.37394285e+00
-6.73056781e-01 5.01291811e-01 -9.75909352e-01 2.29387149e-01
-2.94363379e-01 3.49133313e-01 9.86505806e-01 -1.20814729e+00
3.35881442e-01 1.18996489e+00 5.93312562e-01 5.23695707e-01
-1.04189467e+00 -8.76803756e-01 4.15948004e-01 2.91078925e-01
-1.54818654e+00 -3.82603943e-01 5.05219460e-01 -4.97103363e-01
5.83088875e-01 1.24532327e-01 2.20276847e-01 9.62900519e-01
-1.26687482e-01 1.13322759e+00 9.07190323e-01 -2.96018958e-01
-9.34408046e-03 3.65978450e-01 3.43323171e-01 7.80610502e-01
6.74712956e-01 4.20494340e-02 -1.46833956e-01 2.07994059e-01
4.99436051e-01 5.92164285e-02 -2.93059856e-01 -4.53947127e-01
-1.09142601e+00 5.74840724e-01 6.05459392e-01 -9.14249495e-02
3.47634703e-02 1.59719456e-02 2.04274952e-01 -2.68982321e-01
3.08159828e-01 8.86696056e-02 -4.87560838e-01 2.44030640e-01
-6.70943022e-01 -4.80752485e-03 5.12052059e-01 1.18351483e+00
6.95968509e-01 -7.40913376e-02 -4.17493731e-01 1.02852976e+00
6.48613393e-01 7.29636848e-01 2.40600780e-01 -6.73644900e-01
5.36441207e-01 7.37555444e-01 3.10686082e-01 -8.73491585e-01
-2.89554477e-01 -7.89612889e-01 -6.49774432e-01 8.54484737e-02
4.21039313e-01 8.46907720e-02 -1.18120277e+00 1.65062737e+00
6.14258587e-01 4.40865904e-01 1.35860713e-02 1.17819226e+00
1.03896248e+00 5.67287326e-01 2.01922432e-01 -1.35850400e-01
1.37889910e+00 -1.24581671e+00 -7.43719459e-01 -5.59516907e-01
4.41836685e-01 -8.65484536e-01 1.03147531e+00 2.63581425e-01
-8.91862750e-01 -7.74534881e-01 -1.12749600e+00 -1.55250400e-01
-1.46679670e-01 6.17408454e-01 4.38190073e-01 5.65551341e-01
-4.70127851e-01 8.82278830e-02 -7.99100518e-01 -1.74882203e-01
7.93897092e-01 1.98297441e-01 -3.72019224e-02 -2.43017197e-01
-8.13692808e-01 7.81792223e-01 8.63465905e-01 2.29007542e-01
-1.29314435e+00 -4.64122027e-01 -8.15056622e-01 5.19929081e-02
1.00366378e+00 -4.25942361e-01 1.10522461e+00 -8.30284595e-01
-1.14938438e+00 7.02665210e-01 4.28448385e-03 -3.93119216e-01
6.48080707e-01 -5.54989159e-01 -3.82528186e-01 1.98698603e-02
2.26305187e-01 8.48719418e-01 8.80892098e-01 -1.40152466e+00
-1.14174974e+00 -4.48750436e-01 1.35212496e-01 1.14652023e-01
-4.72699553e-01 3.77341062e-02 -1.02225220e+00 -6.89749897e-01
2.71155149e-01 -8.61888170e-01 -1.96595162e-01 3.56618106e-01
-5.17020643e-01 -4.14815158e-01 9.49551463e-01 -5.96427917e-01
1.14979136e+00 -2.28315806e+00 -1.41005427e-01 -3.25061172e-01
2.97550380e-01 5.45389235e-01 -2.06275582e-01 -3.37293059e-01
3.79427224e-01 -1.48065373e-01 -3.81497741e-01 -4.34745759e-01
-7.66452216e-03 1.91444293e-01 -1.90778553e-01 4.88750845e-01
5.22761762e-01 7.59167671e-01 -8.17345023e-01 -7.27716506e-01
2.62464672e-01 3.05637121e-01 -4.48931545e-01 2.27299646e-01
-3.82677555e-01 4.94902804e-02 -4.49567050e-01 9.35523450e-01
9.59433436e-01 -3.16267669e-01 -1.07872896e-01 -2.64751852e-01
-1.70233026e-01 3.30920070e-02 -1.75014591e+00 1.36324418e+00
-7.36832544e-02 3.27152431e-01 1.39305875e-01 -8.24556947e-01
8.18507612e-01 3.14790644e-02 3.66539620e-02 -4.94626760e-01
2.84898967e-01 4.45245266e-01 1.05973765e-01 -6.28334045e-01
2.98266470e-01 8.84132013e-02 2.14513138e-01 -2.42143329e-02
2.46222064e-01 2.28673741e-01 2.06392661e-01 3.00936937e-01
8.70284975e-01 1.86556786e-01 2.45141745e-01 -1.97473541e-01
6.99214458e-01 -1.87833130e-01 9.81833398e-01 9.10148323e-01
-4.84015524e-01 6.13008440e-01 2.88741380e-01 -2.11298451e-01
-6.91288471e-01 -1.09709144e+00 -3.99049312e-01 9.82142568e-01
7.96156049e-01 -1.49937436e-01 -7.36808300e-01 -9.22783911e-01
2.90075094e-02 5.60091555e-01 -3.00585240e-01 -2.10348949e-01
-6.28298700e-01 -9.78392005e-01 3.79195154e-01 7.30005562e-01
6.96161866e-01 -8.73611093e-01 -4.48506296e-01 6.43336251e-02
-3.48055005e-01 -1.57893395e+00 -5.13586819e-01 9.56443995e-02
-6.30919933e-01 -1.15414190e+00 -6.99007094e-01 -8.62252057e-01
6.82419240e-01 5.70588648e-01 7.76661396e-01 1.98693201e-02
-3.71467173e-01 8.64131674e-02 -2.69288749e-01 -7.12764204e-01
-1.36629686e-01 -4.92031686e-02 1.59376517e-01 4.45800245e-01
4.22187835e-01 4.28481326e-02 -7.41777420e-01 6.68469429e-01
-8.21713984e-01 5.44811301e-02 8.84707868e-01 8.88201237e-01
7.37711608e-01 -9.06876847e-02 6.22999430e-01 -6.11369908e-01
-1.26680434e-01 -3.86413783e-01 -8.94575238e-01 3.02386433e-01
-6.52305067e-01 -1.65058956e-01 4.09400672e-01 -6.56040013e-01
-1.22457707e+00 2.40604654e-01 6.57597706e-02 -6.19027972e-01
-9.37550962e-02 -8.36294889e-02 -6.38329864e-01 -1.05652943e-01
5.31200469e-01 1.87574908e-01 -3.40196341e-01 -5.59371769e-01
3.73137951e-01 9.41589177e-01 6.83933616e-01 -2.26506740e-01
8.86068761e-01 6.53249085e-01 -3.51230770e-01 -5.31331837e-01
-1.21621192e+00 -7.49376893e-01 -4.01994348e-01 -7.42409453e-02
8.58094811e-01 -1.22750843e+00 -3.99587721e-01 5.88056862e-01
-1.19089925e+00 4.45042066e-02 -1.80563956e-01 6.61368191e-01
-2.05634058e-01 3.93900394e-01 -4.81038421e-01 -9.41055417e-01
-2.81067431e-01 -1.23065746e+00 1.25297904e+00 3.64020050e-01
3.56518716e-01 -3.37592036e-01 -5.64430952e-01 6.93114340e-01
9.16612819e-02 -2.37991699e-04 4.63033527e-01 -6.17524862e-01
-1.02403426e+00 -1.42103583e-01 -8.00959229e-01 7.02405393e-01
-7.65511319e-02 -2.61817753e-01 -1.15873444e+00 -3.80318791e-01
3.59300524e-02 -3.35646689e-01 1.21248949e+00 1.71084359e-01
1.17996144e+00 -1.16206169e-01 -5.71703732e-01 5.51835835e-01
1.48504984e+00 1.38436854e-01 3.62438560e-01 1.99052602e-01
8.78970981e-01 5.19084811e-01 8.32351327e-01 4.49237734e-01
4.32699591e-01 6.87143087e-01 7.25688279e-01 -3.99080440e-02
-4.25855756e-01 -3.42853218e-02 5.15593886e-01 5.00388622e-01
1.13523014e-01 -3.27611059e-01 -7.48860776e-01 7.45232224e-01
-1.85900176e+00 -5.19186616e-01 -3.56928766e-01 2.04848313e+00
8.51006806e-01 6.05503798e-01 -5.11904210e-02 7.49119818e-02
9.07389104e-01 3.57343582e-03 -7.44571626e-01 4.42960322e-01
-1.19609430e-01 -2.53851175e-01 7.10098505e-01 2.00482368e-01
-1.38237321e+00 9.32167351e-01 4.77143383e+00 1.07938874e+00
-8.90580833e-01 3.37116748e-01 5.07620394e-01 -1.22748867e-01
2.36938626e-01 -4.60360572e-02 -1.41844642e+00 6.29553199e-01
4.15311098e-01 2.79204249e-01 1.51498858e-02 1.12845290e+00
1.26626147e-02 -1.88294306e-01 -1.03312147e+00 1.06886303e+00
2.84574986e-01 -8.74806046e-01 -3.72405956e-03 -2.52776772e-01
6.27265573e-01 1.47485152e-01 -6.84357658e-02 5.47461510e-01
1.55939832e-02 -5.42744637e-01 9.95676696e-01 2.66113609e-01
5.26575387e-01 -4.36765045e-01 8.92384350e-01 5.96815586e-01
-1.29987824e+00 -3.36563408e-01 -5.10489762e-01 1.54191777e-01
1.68824062e-01 6.63975716e-01 -5.51826119e-01 3.46464604e-01
9.60552692e-01 5.93475997e-01 -7.73945808e-01 1.21521604e+00
-4.12144244e-01 6.88816011e-01 -3.47121418e-01 5.50864190e-02
1.77087650e-01 1.83358882e-02 6.72588050e-01 1.11940515e+00
7.26421997e-02 1.26930729e-01 4.79280353e-01 1.00121737e+00
-2.00334787e-01 -9.71661657e-02 -4.74753454e-02 1.93403035e-01
2.90646881e-01 1.37122643e+00 -8.35761547e-01 -3.98872405e-01
-6.43938065e-01 9.39400613e-01 2.47290909e-01 2.28873238e-01
-1.28344846e+00 -4.61692780e-01 4.10145789e-01 4.39612679e-02
7.39218891e-01 -6.41744062e-02 -1.76586896e-01 -1.21982455e+00
4.93609130e-01 -7.01883197e-01 4.97827739e-01 -5.46898425e-01
-1.19654191e+00 3.66492689e-01 -2.53098994e-01 -1.32419825e+00
5.90925932e-01 -7.76152611e-01 -2.07845926e-01 5.19669592e-01
-1.96538126e+00 -1.17926776e+00 -5.23975968e-01 2.94702619e-01
8.06342840e-01 2.74035279e-02 8.82146582e-02 6.17423236e-01
-9.74945188e-01 6.84849799e-01 6.60434142e-02 1.81876943e-01
7.46968389e-01 -1.10481524e+00 1.20846748e-01 1.02921844e+00
2.27118015e-01 2.73220628e-01 3.77719760e-01 -5.11343539e-01
-1.32536602e+00 -1.53122890e+00 6.30945802e-01 -6.56721652e-01
4.52886432e-01 -5.35456777e-01 -1.05196702e+00 5.30666649e-01
-2.84771770e-01 6.27102196e-01 2.51164228e-01 -2.15840608e-01
-3.88579488e-01 -2.48434037e-01 -1.01216066e+00 5.99999249e-01
1.26622832e+00 -1.65900841e-01 -7.02163756e-01 3.55625302e-01
1.09684825e+00 -4.36603457e-01 -4.37375754e-01 7.99954295e-01
4.04568702e-01 -8.29834938e-01 9.96946275e-01 -3.10664505e-01
1.20573968e-01 -7.22093642e-01 -3.25139523e-01 -6.55818999e-01
-1.80456966e-01 -2.02296570e-01 -4.45171446e-01 1.49360299e+00
1.72788486e-01 -5.46164751e-01 4.85717773e-01 3.74750078e-01
-2.59092927e-01 -7.18823135e-01 -8.43640745e-01 -9.96333063e-01
-4.42257792e-01 -5.93888938e-01 2.15744704e-01 6.70128405e-01
-6.69402122e-01 4.71352905e-01 -2.92802423e-01 6.19391561e-01
7.56878078e-01 1.20291419e-01 7.67396390e-01 -1.23325205e+00
-3.07558507e-01 -3.63787740e-01 -4.27795351e-01 -1.38926971e+00
-1.67202085e-01 -7.12097764e-01 2.65764415e-01 -1.33692157e+00
5.10659039e-01 -4.64385211e-01 -6.63763642e-01 4.22116488e-01
-6.15933239e-01 4.41248327e-01 3.22168916e-01 3.36141884e-01
-1.14511704e+00 7.81834185e-01 1.19866228e+00 -2.70811409e-01
-2.25520004e-02 -1.70988627e-02 -6.07182801e-01 1.14058125e+00
4.27637964e-01 -6.83143437e-01 -1.61736697e-01 -6.19183719e-01
-1.90195054e-01 -4.19614464e-01 6.16719723e-01 -1.04730773e+00
1.47439614e-01 -4.97942930e-03 2.83679783e-01 -7.53001451e-01
2.69520044e-01 -8.58873069e-01 -4.18330401e-01 5.72788477e-01
1.94949768e-02 -6.73302054e-01 2.60994643e-01 9.69509184e-01
-1.37844726e-01 -2.69519717e-01 1.05490088e+00 2.32402161e-02
-1.10196614e+00 3.95359427e-01 2.54501671e-01 2.37260565e-01
1.36459017e+00 -2.18264461e-01 -3.48473579e-01 3.39302301e-01
-3.99658591e-01 6.29463077e-01 2.92037539e-02 6.86484098e-01
6.07326090e-01 -1.11836815e+00 -6.87531114e-01 2.00130433e-01
3.95727396e-01 3.77728283e-01 1.74971968e-01 9.00402129e-01
-2.08789304e-01 2.28158548e-01 1.93740681e-01 -8.44218791e-01
-1.15626323e+00 6.56235516e-01 3.19032192e-01 1.21752014e-02
-5.95682025e-01 1.03999221e+00 6.39566004e-01 -2.85440803e-01
6.11911833e-01 -3.87312561e-01 -2.29962975e-01 5.19784652e-02
5.97303391e-01 5.68454206e-01 -7.24892840e-02 -5.68148911e-01
-4.25980926e-01 5.48833728e-01 -2.61347681e-01 4.24075454e-01
1.06699264e+00 -2.83855528e-01 7.37282708e-02 3.24252486e-01
8.67075324e-01 -3.09691608e-01 -1.63881993e+00 -5.81511676e-01
1.45015106e-01 -4.80668128e-01 4.12751287e-02 -7.84567773e-01
-9.33350861e-01 8.78338516e-01 9.36172128e-01 1.06323406e-03
1.01575017e+00 2.25298584e-01 9.01196301e-01 4.04933333e-01
3.51994097e-01 -1.11077845e+00 1.85583502e-01 3.78191352e-01
7.15674818e-01 -1.75135553e+00 -5.21386042e-02 -7.26361096e-01
-4.77732241e-01 7.24895477e-01 1.22525251e+00 1.04782701e-01
4.66483861e-01 1.95782766e-01 -3.53641696e-02 7.09386170e-02
-4.98222142e-01 -4.48154986e-01 4.06358123e-01 2.82568604e-01
-8.21792632e-02 -5.48358299e-02 -3.53502750e-01 9.67678308e-01
5.18461466e-01 -8.05889145e-02 2.23950759e-01 7.56151676e-01
-7.05482304e-01 -6.92371964e-01 -5.24318039e-01 3.79929423e-01
-5.14216363e-01 -1.18882634e-01 -5.42598069e-02 6.86553776e-01
7.18091667e-01 9.36467469e-01 -7.95786083e-02 -1.24348886e-01
3.84819627e-01 -1.67269751e-01 1.17602155e-01 -6.58725023e-01
-1.18298329e-01 2.24628150e-01 -1.97713390e-01 -4.93919164e-01
-3.99063408e-01 -3.87066156e-01 -1.26233625e+00 4.10563886e-01
-1.17322373e+00 -2.22725302e-01 4.48723108e-01 8.85505021e-01
2.56461114e-01 6.28227413e-01 4.94335592e-01 -5.73388159e-01
-9.56351161e-01 -1.01057506e+00 -5.86627901e-01 3.53967339e-01
3.81807417e-01 -9.42864716e-01 -3.76562953e-01 1.57243818e-01] | [9.304661750793457, 1.2957957983016968] |
cf1309ae-a6f6-412b-b313-c19e6f44121d | duth-at-semeval-2017-task-4-a-voting | null | null | https://aclanthology.org/S17-2117 | https://aclanthology.org/S17-2117.pdf | DUTH at SemEval-2017 Task 4: A Voting Classification Approach for Twitter Sentiment Analysis | This report describes our participation to SemEval-2017 Task 4: Sentiment Analysis in Twitter, specifically in subtasks A, B, and C. The approach for text sentiment classification is based on a Majority Vote scheme and combined supervised machine learning methods with classical linguistic resources, including bag-of-words and sentiment lexicon features. | ['Symeon Symeonidis', 'John Kordonis', 'Dimitrios Effrosynidis', 'Avi Arampatzis'] | 2017-08-01 | null | null | null | semeval-2017-8 | ['twitter-sentiment-analysis'] | ['natural-language-processing'] | [-3.88724469e-02 1.31878699e-03 -4.28026199e-01 -9.21228349e-01
-7.09612072e-01 -7.42296934e-01 9.45391417e-01 8.66314054e-01
-8.08796108e-01 6.79063380e-01 6.86983705e-01 -2.59091377e-01
4.87059057e-01 -5.89184761e-01 7.68177509e-02 -3.89548451e-01
3.74471396e-01 4.16191906e-01 -2.29510441e-01 -9.91673589e-01
5.39975941e-01 9.53673199e-02 -1.29487622e+00 9.37426209e-01
2.75683105e-01 1.47055066e+00 -5.76040924e-01 6.63104594e-01
-7.33657598e-01 1.30917335e+00 -7.10099101e-01 -8.69290233e-01
-3.09405595e-01 -1.45023093e-01 -9.95484889e-01 -1.10431790e-01
2.43856311e-01 4.57971305e-01 3.70266080e-01 8.05158317e-01
4.15722370e-01 1.62180692e-01 9.03305709e-01 -8.56250703e-01
-3.26550484e-01 7.40990698e-01 -4.01842594e-01 2.09356606e-01
8.38653266e-01 -4.20873553e-01 1.53637874e+00 -1.14918089e+00
6.05980873e-01 1.16889274e+00 1.01570332e+00 4.01505649e-01
-8.32208633e-01 -5.39795041e-01 2.56275743e-01 -2.27339536e-01
-7.49609113e-01 -3.15985054e-01 5.61420918e-01 -8.20709586e-01
1.49249625e+00 1.88453525e-01 8.07617545e-01 9.87178981e-01
4.38853979e-01 8.98995757e-01 1.47125244e+00 -6.49523973e-01
3.64405245e-01 7.46622622e-01 6.93178535e-01 3.98456812e-01
1.78359568e-01 -5.49229264e-01 -1.12280858e+00 -2.93108255e-01
-6.29895210e-01 -2.18473688e-01 2.42319897e-01 1.89302191e-01
-6.74747050e-01 1.16983831e+00 1.87762216e-01 4.97093529e-01
-4.90512908e-01 -1.70464352e-01 1.18420541e+00 5.37604213e-01
1.50019336e+00 5.73851883e-01 -1.35595882e+00 -1.16238587e-01
-7.02233136e-01 3.46401274e-01 1.34521103e+00 5.89676797e-01
6.40011489e-01 -1.32620439e-01 -3.69402021e-03 8.34178627e-01
6.21841729e-01 6.33976638e-01 9.77507651e-01 4.02718559e-02
4.67292249e-01 9.19900835e-01 -7.75196999e-02 -1.14645052e+00
-7.72304475e-01 7.17806891e-02 -3.90851647e-01 -7.00997785e-02
-1.53802529e-01 -5.81195176e-01 -6.56063080e-01 9.47780013e-01
8.31987038e-02 -6.95371389e-01 2.55083740e-01 3.37569177e-01
1.63490343e+00 3.33157212e-01 4.62734997e-01 -1.99825794e-01
1.59511960e+00 -9.80547786e-01 -8.63382995e-01 -5.66416979e-01
1.27978706e+00 -1.15775609e+00 7.47979403e-01 4.05453175e-01
-7.38876104e-01 -1.80731699e-01 -8.24655712e-01 8.90781954e-02
-1.37556016e+00 3.35089825e-02 8.95653367e-01 1.11044574e+00
-1.23537338e+00 2.23336950e-01 -2.60764152e-01 -3.43828350e-01
5.90132535e-01 4.53262001e-01 -6.58292174e-01 4.22656715e-01
-1.18430972e+00 1.14834726e+00 1.85059890e-01 4.07566801e-02
5.78578562e-02 -4.15195763e-01 -1.40762675e+00 -5.17599285e-01
-2.25258783e-01 -1.42099500e-01 1.58757818e+00 -1.30315089e+00
-1.61084569e+00 1.59929883e+00 -6.47041917e-01 -3.03639650e-01
-9.84184965e-02 -2.92972624e-01 -6.19151950e-01 -2.58609653e-01
4.16154206e-01 1.83570206e-01 5.80502093e-01 -9.22419965e-01
-6.46073043e-01 -6.42670095e-01 -1.19076269e-02 1.40902370e-01
-5.86894333e-01 8.36020768e-01 1.92596555e-01 -4.47881609e-01
-1.93214700e-01 -9.04196560e-01 -4.77807969e-01 -1.03322065e+00
-4.80399758e-01 -6.44535065e-01 4.64038223e-01 -3.65050405e-01
1.13013899e+00 -1.88151467e+00 -4.35307801e-01 4.73412097e-01
-3.00030038e-02 3.73683311e-02 1.90559492e-01 6.14065349e-01
-3.76255244e-01 2.75536358e-01 2.78070599e-01 -9.02315855e-01
1.17437549e-01 -2.22144589e-01 -6.24571443e-01 3.93375069e-01
-7.56637156e-02 7.94418454e-01 -8.51922452e-01 -5.21727383e-01
1.71378419e-01 1.50892377e-01 -2.46238366e-01 -2.26761475e-01
-1.74977139e-01 2.27817729e-01 -6.96822643e-01 8.48554254e-01
3.50481987e-01 -1.50366098e-01 2.11980507e-01 -1.36185780e-01
-4.70960230e-01 9.38347459e-01 -4.59029585e-01 1.21652341e+00
-6.93686783e-01 7.64093220e-01 -8.92162696e-02 -8.79336894e-01
9.33102369e-01 5.46960533e-01 3.08928281e-01 -6.70917451e-01
7.61370838e-01 3.98800343e-01 -6.41253114e-01 -4.29920584e-01
7.92962372e-01 -4.37140942e-01 -8.93051565e-01 4.99967158e-01
1.95134863e-01 -7.80447245e-01 4.98003364e-01 2.44348586e-01
6.12307787e-01 -1.71577521e-02 7.36910939e-01 -6.31780267e-01
1.02963412e+00 3.23257446e-01 -7.41498619e-02 4.61104602e-01
-1.96974590e-01 2.44671330e-01 7.61811435e-01 -8.70270371e-01
-4.56116676e-01 -2.10149791e-02 -2.71785796e-01 1.85808253e+00
-3.84773135e-01 -1.00881851e+00 -4.86570388e-01 -1.06275809e+00
-7.27040544e-02 5.51505446e-01 -1.09744108e+00 2.64091074e-01
-2.66714156e-01 -1.00848627e+00 3.31449360e-01 3.68072003e-01
-6.83003590e-02 -1.56295431e+00 -9.54283699e-02 1.88103002e-02
2.02722289e-02 -1.19439828e+00 -1.86844423e-01 7.26505697e-01
-5.96250832e-01 -1.01408613e+00 -1.82598621e-01 -1.01587296e+00
5.31490743e-01 -8.23216215e-02 1.48962975e+00 -8.12600330e-02
9.88301635e-02 1.65351093e-01 -8.17924142e-01 -1.22197044e+00
-1.13671131e-01 4.28856462e-01 -6.14188164e-02 -3.19339335e-02
1.04675710e+00 8.80778730e-02 -2.41650537e-01 -1.52382359e-01
-4.88179564e-01 -4.46772218e-01 -1.13177793e-02 6.00533366e-01
3.89026105e-01 -4.28359568e-01 6.69664562e-01 -1.39323378e+00
8.59349906e-01 -5.40055394e-01 -1.17803499e-01 -2.53191262e-01
-4.53931719e-01 -5.45648932e-01 6.14401460e-01 2.28325978e-01
-9.40363050e-01 -1.01285186e-02 -7.77863204e-01 7.64021039e-01
3.89607623e-02 9.09733593e-01 2.77436614e-01 -1.60863489e-01
6.67545974e-01 -1.82913080e-01 -2.32727379e-01 -8.73370692e-02
2.06969008e-01 1.12258089e+00 -3.40078562e-01 -3.58810797e-02
1.39782250e-01 7.10440218e-01 -2.79822499e-01 -8.42101932e-01
-2.07904196e+00 -1.00182092e+00 -5.35468876e-01 -3.62350792e-01
1.05272043e+00 -1.31048238e+00 -6.54129624e-01 8.11561644e-01
-9.13565278e-01 -1.19253680e-01 -4.12812501e-01 4.20984089e-01
-2.79762655e-01 -2.24089444e-01 -6.93609536e-01 -7.84843445e-01
-9.19379532e-01 -7.59580910e-01 1.34397507e+00 9.12873298e-02
-9.28187728e-01 -1.57394779e+00 5.14065444e-01 5.19403458e-01
4.92506444e-01 1.93505466e-01 4.10260201e-01 -1.31607485e+00
9.26680982e-01 -8.71966481e-01 1.93615615e-01 5.77543616e-01
-1.02961116e-01 -3.35590616e-02 -1.24861395e+00 2.00234968e-02
-5.86112626e-02 -9.54766572e-01 1.15648007e+00 2.69746274e-01
9.34826434e-01 -5.00143617e-02 -2.08993480e-01 -3.07730921e-02
1.17575550e+00 -2.15648383e-01 1.08082309e-01 8.80294561e-01
5.10382652e-01 9.62285042e-01 7.11928487e-01 6.29539967e-01
8.06027114e-01 2.31451005e-01 2.04173952e-01 9.21278149e-02
5.55307746e-01 1.59895405e-01 4.82886076e-01 1.23622167e+00
2.86837369e-01 -3.00454438e-01 -1.01175117e+00 6.99754655e-01
-1.72556269e+00 -6.90222383e-01 -6.12980485e-01 1.37214530e+00
9.01098013e-01 4.74985778e-01 9.56228152e-02 3.07827443e-01
4.24193650e-01 6.70296848e-01 1.68909580e-01 -1.16550314e+00
-4.62144643e-01 6.89175367e-01 5.43679476e-01 3.32758427e-01
-1.48140121e+00 1.22735465e+00 7.45833397e+00 6.13496482e-01
-1.03287375e+00 3.51580828e-01 6.38786077e-01 -1.24316625e-02
-3.50591242e-01 -2.02581257e-01 -1.03197861e+00 1.61972523e-01
1.05596399e+00 1.83058813e-01 -4.14275944e-01 9.31452990e-01
-7.77411386e-02 -3.83861095e-01 -6.08159065e-01 3.97326678e-01
5.92900097e-01 -1.19943106e+00 -1.10177331e-01 -4.18370545e-01
1.06888342e+00 6.22981131e-01 5.43103591e-02 6.22932315e-01
3.69711727e-01 -9.75884616e-01 9.70603585e-01 1.53377965e-01
5.20422637e-01 -7.34003186e-01 1.41760898e+00 -3.53296548e-02
-9.58423197e-01 1.53904423e-01 -5.71616367e-02 -5.19707561e-01
2.70132482e-01 8.30675483e-01 -2.28376150e-01 8.14199150e-02
1.07815051e+00 1.34391940e+00 -4.76398498e-01 2.87511110e-01
-6.03311121e-01 8.05982769e-01 1.32736683e-01 -7.33941615e-01
4.81134742e-01 -2.67389584e-02 2.36677095e-01 1.95129371e+00
-2.66728580e-01 -1.96642205e-01 5.58835901e-02 -3.32400858e-01
-2.33565941e-01 9.26807761e-01 -5.29022098e-01 -1.74190998e-01
-1.04630396e-01 1.65061736e+00 -9.56133246e-01 -5.87592423e-01
-6.67611420e-01 4.21466500e-01 1.85181603e-01 -1.10131010e-01
-3.67142111e-02 -4.23256814e-01 4.90921080e-01 -2.32559785e-01
2.75255442e-01 2.22147346e-01 -8.08905184e-01 -1.21117377e+00
-2.66433150e-01 -8.31535816e-01 4.73226964e-01 -6.38257623e-01
-1.55997694e+00 1.01503181e+00 -5.31050324e-01 -7.84677327e-01
-1.21481337e-01 -1.04188406e+00 -8.54837060e-01 8.39283645e-01
-1.64114761e+00 -1.23610592e+00 -3.31106335e-02 6.53713048e-01
3.62132072e-01 -5.77020347e-01 1.33180249e+00 6.54821843e-02
-3.81221831e-01 3.50275397e-01 1.88510288e-02 4.95998673e-02
9.80294645e-01 -1.61946464e+00 5.75703740e-01 -1.47843882e-01
-3.10302645e-01 3.73745680e-01 8.50673437e-01 -5.14996886e-01
-8.70618820e-01 -9.36480820e-01 1.90960622e+00 -7.03465164e-01
1.36648226e+00 -5.13473868e-01 -1.25766799e-01 6.80601537e-01
4.24314380e-01 -5.15136644e-02 1.61581647e+00 6.55648768e-01
-5.11626303e-01 1.69909239e-01 -1.07284677e+00 2.56728560e-01
2.45950162e-01 -8.52107882e-01 -6.23523533e-01 8.14387202e-01
4.32672024e-01 -2.56008655e-01 -7.92553425e-01 4.19674575e-01
7.11356997e-01 -5.86510599e-01 5.18085659e-01 -8.73725295e-01
8.16409230e-01 1.70627132e-01 -2.30856404e-01 -1.47916698e+00
4.51145500e-01 -2.70067573e-01 3.20197552e-01 9.73869801e-01
9.92609978e-01 -7.44343400e-01 8.04768741e-01 1.48144960e-01
-7.09469691e-02 -7.53756821e-01 -5.30115724e-01 9.14229371e-04
3.10785383e-01 -8.37991714e-01 1.96378738e-01 1.05554020e+00
6.65377915e-01 8.64537835e-01 -8.09293017e-02 -6.87833786e-01
-1.22740887e-01 1.93716213e-01 6.00786686e-01 -1.40068328e+00
1.94134235e-01 -7.52542078e-01 -4.09907877e-01 -5.31505704e-01
7.80436814e-01 -8.97267163e-01 7.85463974e-02 -1.56547821e+00
-7.21098408e-02 -1.68210730e-01 -2.61090815e-01 4.06280845e-01
-1.06450863e-01 8.56574714e-01 -1.51306778e-01 5.30194268e-02
-1.22919381e+00 2.86121696e-01 8.76254320e-01 -7.03775287e-02
-1.01847919e-02 2.64049172e-01 -1.19499218e+00 1.35732758e+00
8.64934981e-01 -7.96134353e-01 3.32302392e-01 -3.40628140e-02
1.31436062e+00 -5.29994905e-01 -4.32235986e-01 -1.48218811e-01
5.99700585e-02 2.09995076e-01 1.26660481e-01 -7.76126146e-01
3.65514189e-01 -5.07723927e-01 -8.34635615e-01 1.71193048e-01
-4.57350880e-01 1.44043848e-01 4.10834700e-01 1.32166550e-01
-7.20488250e-01 -5.04376054e-01 4.11456943e-01 -2.09575906e-01
-4.46895152e-01 -1.04163438e-01 -1.04531729e+00 1.40939936e-01
8.64952326e-01 4.46657836e-03 -3.60828489e-01 -2.92528480e-01
-9.35030222e-01 3.38114828e-01 7.32010975e-02 4.28422123e-01
2.18462452e-01 -9.94355381e-01 -7.95506418e-01 2.27549635e-02
6.76899314e-01 -4.39550042e-01 -4.69307005e-02 9.38112080e-01
-4.83323842e-01 5.88951886e-01 3.02991509e-01 -3.92824374e-02
-1.42516375e+00 -6.02811687e-02 2.84997195e-01 -8.06820452e-01
2.07606807e-01 1.18144023e+00 -2.76221603e-01 -9.88825560e-01
3.56721878e-02 6.02550730e-02 -1.31227803e+00 1.01761138e+00
7.72202671e-01 4.95242812e-02 4.80406135e-01 -1.23375607e+00
-7.03010023e-01 6.56307459e-01 -3.00784081e-01 -1.58719242e-01
1.33134556e+00 1.52223150e-03 -8.12302768e-01 9.82358336e-01
1.31563532e+00 4.08256620e-01 1.09833635e-01 -2.38306344e-01
5.16537428e-01 -2.25746483e-02 2.41151467e-01 -1.06165946e+00
-6.70047879e-01 3.85272682e-01 -1.61241755e-01 8.56168389e-01
7.09449530e-01 8.26818421e-02 6.31699264e-01 5.96158266e-01
1.78238660e-01 -1.46243417e+00 -1.02632150e-01 1.31779182e+00
6.39964879e-01 -1.89303553e+00 1.35378137e-01 -2.25633100e-01
-9.95708644e-01 1.11023247e+00 2.69340575e-01 -3.41951460e-01
1.43647480e+00 3.29656392e-01 5.10524452e-01 -7.32307136e-01
-8.67693186e-01 -4.27210450e-01 6.21310532e-01 9.40625817e-02
1.22863865e+00 2.75912613e-01 -6.83075190e-01 9.82182503e-01
-7.65640438e-01 -2.54226834e-01 3.61893445e-01 1.15442419e+00
-4.39298391e-01 -1.07382643e+00 7.25433528e-02 8.27756643e-01
-1.34276664e+00 -4.53491449e-01 -1.00457966e+00 5.25238633e-01
-3.40674445e-02 1.53452027e+00 1.04502790e-01 -3.82554471e-01
5.48033774e-01 -6.22453839e-02 5.34039885e-02 -1.02918839e+00
-1.62649310e+00 -2.25238428e-01 7.59125233e-01 -4.77061480e-01
-1.17194223e+00 -9.26588953e-01 -9.76221323e-01 -3.82667959e-01
-5.48198819e-01 4.75222290e-01 1.33637381e+00 1.44107437e+00
-1.65813804e-01 4.01832879e-01 7.28031158e-01 -7.20374286e-01
-1.68735325e-01 -1.31782663e+00 -7.31553733e-01 2.70941138e-01
4.14728194e-01 -5.21818362e-02 -8.18296671e-01 -1.56244799e-01] | [11.196707725524902, 6.923301696777344] |
f19510e4-7ea0-4f97-a0aa-f21c5e9dd3f7 | on-the-convergence-of-policy-gradient-in | 2212.10439 | null | https://arxiv.org/abs/2212.10439v2 | https://arxiv.org/pdf/2212.10439v2.pdf | Policy Gradient in Robust MDPs with Global Convergence Guarantee | Robust Markov decision processes (RMDPs) provide a promising framework for computing reliable policies in the face of model errors. Many successful reinforcement learning algorithms build on variations of policy-gradient methods, but adapting these methods to RMDPs has been challenging. As a result, the applicability of RMDPs to large, practical domains remains limited. This paper proposes a new Double-Loop Robust Policy Gradient (DRPG), the first generic policy gradient method for RMDPs. In contrast with prior robust policy gradient algorithms, DRPG monotonically reduces approximation errors to guarantee convergence to a globally optimal policy in tabular RMDPs. We introduce a novel parametric transition kernel and solve the inner loop robust policy via a gradient-based method. Finally, our numerical results demonstrate the utility of our new algorithm and confirm its global convergence properties. | ['Marek Petrik', 'Chin Pang Ho', 'Qiuhao Wang'] | 2022-12-20 | null | null | null | null | ['policy-gradient-methods'] | ['methodology'] | [-2.39517599e-01 8.19378256e-05 -5.61324179e-01 3.94077115e-02
-1.12014163e+00 -5.12168646e-01 6.89340949e-01 -5.28927632e-02
-5.39381504e-01 1.47511113e+00 3.74162234e-02 -7.74387956e-01
-2.06330582e-01 -4.59519356e-01 -7.19762683e-01 -7.79196024e-01
-2.07702860e-01 5.60951948e-01 3.60267401e-01 1.81740150e-02
5.26478767e-01 6.86600327e-01 -1.13400280e+00 -3.15505117e-01
1.01923907e+00 9.09634054e-01 1.12322390e-01 5.83392143e-01
-1.05655221e-02 9.54551637e-01 -3.56061190e-01 1.06055886e-01
4.41083401e-01 -3.32466036e-01 -7.22414017e-01 8.84367749e-02
3.14760879e-02 -8.52275908e-01 -2.27523416e-01 1.07312727e+00
6.59856617e-01 6.60273612e-01 5.54956198e-01 -1.28107810e+00
-1.22916378e-01 5.15010238e-01 -6.50309503e-01 6.44203648e-02
3.78855914e-01 3.27679008e-01 6.31965160e-01 -4.59340066e-01
4.79778349e-01 1.79489815e+00 4.30803120e-01 6.87241673e-01
-1.23219001e+00 -4.30652738e-01 6.71837389e-01 4.21217736e-03
-8.01521599e-01 -2.22608790e-01 4.01006013e-01 -2.28352025e-01
8.63652825e-01 -1.47640824e-01 4.73183274e-01 1.12764347e+00
4.40632164e-01 1.14687717e+00 1.77803171e+00 -2.08419383e-01
7.98323572e-01 -1.46090806e-01 -3.49361181e-01 7.84449399e-01
2.02204540e-01 5.96308172e-01 -1.94912046e-01 -5.42973042e-01
1.04957092e+00 -2.01958358e-01 6.74255192e-02 -7.84955144e-01
-1.05003381e+00 9.94255006e-01 -3.65325660e-02 -2.35308483e-01
-6.19404137e-01 4.86354709e-01 3.13313186e-01 4.12633657e-01
5.09133458e-01 1.48189545e-01 -3.52558881e-01 -6.60567582e-01
-7.90776968e-01 8.31222653e-01 1.01095521e+00 8.68782282e-01
4.19187635e-01 4.60371137e-01 -5.25644541e-01 5.24772346e-01
5.41414738e-01 8.92946720e-01 2.57630348e-01 -1.53979790e+00
4.03070599e-01 -1.60435382e-02 9.80732143e-01 -5.01630187e-01
-2.99006790e-01 -1.24236457e-01 -4.52128053e-01 7.04350471e-01
5.52010238e-01 -5.67827344e-01 -6.96142614e-01 1.60910642e+00
7.65935719e-01 4.31049205e-02 1.63240865e-01 8.84007812e-01
-3.94210666e-01 8.17723155e-01 4.63716835e-02 -6.34487748e-01
5.08087456e-01 -8.10659468e-01 -6.80966020e-01 -3.56294811e-02
3.76257628e-01 -5.26884079e-01 1.01629412e+00 5.40887952e-01
-1.29092813e+00 -1.02902316e-01 -6.89507484e-01 4.79297668e-01
6.78999573e-02 -1.68894723e-01 6.00012004e-01 5.57430744e-01
-1.11062765e+00 1.02299273e+00 -1.12461424e+00 -1.04824044e-01
1.56434357e-01 3.29134554e-01 2.74244666e-01 3.06844339e-02
-8.14839959e-01 1.32651615e+00 4.73076552e-01 -3.70235369e-02
-1.46154404e+00 -3.55879903e-01 -6.12114608e-01 -2.47590527e-01
9.54037488e-01 -4.45343286e-01 2.07520056e+00 -6.50581419e-01
-2.60672283e+00 8.66546258e-02 -1.78423122e-01 -7.81304657e-01
1.06380808e+00 -4.11394358e-01 8.06518570e-02 1.98740616e-01
4.21551475e-03 2.43836567e-01 1.23089731e+00 -1.30415666e+00
-1.10083187e+00 -5.49707226e-02 3.81746702e-03 4.65010315e-01
2.20327646e-01 1.55905515e-01 1.54229283e-01 -2.54002988e-01
-1.65601999e-01 -9.96133387e-01 -7.75404513e-01 -2.89653391e-01
-2.72330660e-02 -4.96134996e-01 7.64203250e-01 -4.39752489e-01
1.03264022e+00 -1.66108608e+00 1.85955822e-01 4.05327380e-01
-1.83157071e-01 2.66110301e-01 -4.25767824e-02 5.47043324e-01
4.09394950e-01 -1.89547256e-01 -3.49729836e-01 -1.00977644e-01
4.96235132e-01 4.47870493e-01 -7.78732598e-01 6.94145799e-01
-1.14711545e-01 6.53705239e-01 -1.42845070e+00 -4.62096155e-01
3.56711715e-01 -1.42006487e-01 -4.25919324e-01 2.10376740e-01
-4.52378005e-01 4.64421451e-01 -7.57166505e-01 6.55258656e-01
4.52850699e-01 2.06136242e-01 3.56078923e-01 6.53326988e-01
-4.90283906e-01 1.85986891e-01 -1.35100436e+00 1.17420971e+00
-4.41154987e-01 9.77575593e-03 4.42378342e-01 -1.00678718e+00
7.38402069e-01 2.26933897e-01 8.17685902e-01 -5.11251926e-01
-3.85912210e-02 3.65032822e-01 -4.57429796e-01 -8.62790048e-02
6.94446146e-01 -2.13080555e-01 -4.28414345e-03 5.25657415e-01
-2.28032559e-01 -5.66993594e-01 3.98464143e-01 6.47948906e-02
8.71749520e-01 7.60736048e-01 3.58335793e-01 -4.02577937e-01
3.06009799e-01 2.38770922e-03 7.87695169e-01 1.20166385e+00
-5.68086386e-01 -1.02898397e-01 7.08376765e-01 -2.31739938e-01
-8.53867412e-01 -9.80045795e-01 9.58988592e-02 1.07358658e+00
-1.00992080e-02 1.65890753e-02 -6.60180330e-01 -9.13812399e-01
4.55167264e-01 9.13114429e-01 -2.63407588e-01 3.86843234e-02
-5.73818386e-01 -6.93005800e-01 2.03030705e-01 3.17135841e-01
5.88409901e-01 -9.24439192e-01 -6.82745755e-01 8.06331635e-01
1.61282480e-01 -9.66107309e-01 -2.57374644e-01 -6.66564405e-02
-1.13431978e+00 -1.00683892e+00 -8.47295642e-01 -2.89785564e-01
4.21525776e-01 1.09483339e-01 7.12498903e-01 -6.02117062e-01
4.30080384e-01 8.39522779e-01 -1.43005103e-02 -4.20586646e-01
-7.87685931e-01 -1.83108419e-01 5.23198664e-01 -2.31331006e-01
-1.19988494e-01 -2.46956140e-01 -4.75532860e-01 3.74951571e-01
-6.36005998e-01 -1.65125474e-01 4.21986610e-01 8.14507604e-01
8.15169454e-01 -7.93026686e-02 5.00456333e-01 -6.40695333e-01
1.37688708e+00 -4.07916725e-01 -1.42372644e+00 1.43890738e-01
-1.00787020e+00 3.68670821e-01 5.83466828e-01 -5.32518327e-01
-1.25798643e+00 1.93355102e-02 1.41196594e-01 -3.81773084e-01
3.50850113e-02 3.89294922e-01 4.16728765e-01 -3.32158431e-02
3.83629769e-01 1.86737955e-01 2.06805021e-01 -3.92192274e-01
4.15292442e-01 3.95213276e-01 2.52073318e-01 -1.32074893e+00
6.33364677e-01 4.20641869e-01 1.73031583e-01 -4.82301712e-01
-5.11840403e-01 -2.41819188e-01 -6.11851513e-02 -3.02054405e-01
4.09310251e-01 -6.60516858e-01 -1.07198274e+00 5.00801027e-01
-7.33688712e-01 -1.15673327e+00 -4.96744126e-01 7.43747830e-01
-1.25185406e+00 5.65098107e-01 -8.75298202e-01 -1.35697711e+00
-8.77761468e-02 -1.14342141e+00 7.61732519e-01 2.82325387e-01
2.37389758e-01 -9.20612991e-01 3.76146555e-01 -2.80508339e-01
3.25798273e-01 3.59239846e-01 5.23900270e-01 -1.32953942e-01
-3.39784414e-01 5.40484600e-02 1.85938086e-02 3.10239911e-01
4.64887684e-03 1.60341978e-01 -3.77776474e-01 -7.64360189e-01
4.95917685e-02 -3.28555226e-01 5.98091185e-01 6.26218438e-01
9.19502854e-01 -6.03362918e-01 -2.37157762e-01 2.02439815e-01
1.30290353e+00 4.40855473e-01 2.18438864e-01 6.91685677e-01
3.39022189e-01 1.93713978e-01 1.05198717e+00 9.04284656e-01
3.33875656e-01 3.68788153e-01 4.17396039e-01 2.39506245e-01
5.61351538e-01 -5.50065577e-01 1.02127314e+00 5.04632831e-01
-3.24976891e-01 2.26999253e-01 -7.69794583e-01 3.93654108e-01
-2.39158821e+00 -9.54699576e-01 2.62919337e-01 2.48420405e+00
1.04311287e+00 1.76655814e-01 4.63871121e-01 -2.50026464e-01
6.38453782e-01 -1.70651972e-01 -1.07558167e+00 -8.06953490e-01
2.70963699e-01 9.05611217e-02 9.28541601e-01 6.29256725e-01
-1.06822443e+00 1.28212714e+00 7.11878777e+00 9.23366606e-01
-8.01996291e-01 2.22466178e-02 2.69668996e-01 -2.48467624e-02
6.99140728e-02 1.28306448e-01 -1.00253880e+00 5.36701977e-01
1.07002234e+00 -2.96913773e-01 8.70971501e-01 1.23501730e+00
9.11104143e-01 -4.47500259e-01 -7.11886704e-01 6.72731876e-01
-8.16216171e-01 -9.80225086e-01 -3.40499818e-01 1.85082078e-01
1.13394654e+00 -3.00698988e-02 1.24036923e-01 7.27130651e-01
1.43797791e+00 -6.43244922e-01 6.94048107e-01 3.91106337e-01
4.82401490e-01 -1.20244873e+00 4.18012410e-01 3.54162604e-01
-8.34745646e-01 -4.95011479e-01 -7.01613903e-01 -8.85867402e-02
3.50878090e-01 4.56235051e-01 -8.92932832e-01 3.71741056e-01
4.88134325e-01 5.17598391e-01 1.23869650e-01 1.11383641e+00
-6.18817806e-01 7.26734519e-01 -4.41328138e-01 -2.27406651e-01
9.11235213e-01 -5.11994421e-01 8.53084922e-01 1.02054954e+00
2.33149230e-01 -2.45004207e-01 6.56711578e-01 5.87282002e-01
3.01806897e-01 2.21804772e-02 -5.89608252e-01 -1.57800838e-01
3.78061086e-01 9.19308782e-01 -6.07879817e-01 -4.33237731e-01
-1.76787123e-01 8.21717024e-01 4.66252178e-01 7.56755829e-01
-7.29587317e-01 -8.97607431e-02 9.02032614e-01 -2.87941486e-01
3.34355086e-01 -6.40994430e-01 1.77641496e-01 -1.04658377e+00
-8.12360421e-02 -1.17939961e+00 2.96301693e-01 -2.83175558e-01
-1.13674438e+00 -6.69018775e-02 2.95564473e-01 -1.18127477e+00
-9.65489745e-01 -4.68915462e-01 -2.30996922e-01 4.90963012e-01
-1.75567484e+00 -6.59202218e-01 5.16584635e-01 6.80698156e-01
5.11497378e-01 2.86629453e-04 6.22492731e-01 -3.56791288e-01
-5.34534872e-01 2.23755524e-01 1.04471684e+00 -3.89881849e-01
6.09843194e-01 -1.68381417e+00 1.22423686e-01 8.88304114e-01
-5.65275550e-01 4.65961307e-01 8.76016319e-01 -7.21246779e-01
-1.63121128e+00 -9.28880155e-01 -6.24747127e-02 1.42127713e-02
8.91255796e-01 1.34433821e-01 -6.47204459e-01 7.27615058e-01
-2.03578398e-02 -2.85313100e-01 -1.48284405e-01 -1.73663273e-01
1.00721635e-01 -3.05039119e-02 -1.30678391e+00 8.94482732e-01
7.35048831e-01 -2.06622273e-01 -3.96003813e-01 4.18641150e-01
4.48361188e-01 -6.78518832e-01 -9.91159499e-01 2.45938852e-01
3.48182619e-01 -7.25984752e-01 5.78117132e-01 -7.85988629e-01
-8.30452815e-02 -3.88268948e-01 8.29023942e-02 -1.78765428e+00
-5.62525652e-02 -1.48219419e+00 -7.34623373e-01 7.85093009e-01
1.92372948e-02 -9.90605891e-01 4.78231639e-01 7.35810220e-01
-1.45655066e-01 -6.82947576e-01 -1.18266916e+00 -1.26876771e+00
4.63757813e-01 -3.92276883e-01 5.33715963e-01 6.06092274e-01
2.00786769e-01 -2.77570456e-01 -4.81997371e-01 6.22106418e-02
9.46172476e-01 2.09298030e-01 7.46490180e-01 -6.54497802e-01
-3.97233516e-01 -6.00416780e-01 2.57817000e-01 -1.40054667e+00
4.03198153e-01 -3.36107045e-01 4.07873183e-01 -1.68625700e+00
-1.57580078e-01 -5.55345833e-01 -3.56728792e-01 3.20308357e-01
-2.82309055e-01 -6.45030618e-01 2.96606451e-01 1.61849800e-02
-8.03659558e-01 9.10763741e-01 1.39181149e+00 1.04982682e-01
-5.60480297e-01 4.53721136e-01 -2.45930478e-01 7.12861717e-01
1.36621392e+00 -6.03747606e-01 -4.95526344e-01 3.68549004e-02
-8.11847895e-02 3.41865391e-01 2.76940805e-03 -7.44940042e-01
-8.38005581e-05 -9.76786554e-01 6.90381415e-03 -4.57851350e-01
9.68995504e-03 -4.53862220e-01 -3.89575720e-01 7.59542108e-01
-3.05177331e-01 2.56926417e-01 2.45690897e-01 8.50413501e-01
9.89472643e-02 -2.72576660e-01 8.77913594e-01 -3.27673227e-01
-8.18619072e-01 3.70459199e-01 -1.00715399e+00 2.26939499e-01
9.69307899e-01 1.51736081e-01 4.75138202e-02 -5.72172403e-01
-5.22409022e-01 5.33758938e-01 3.42032075e-01 2.48414855e-02
6.71470225e-01 -1.13805342e+00 -5.47246039e-01 -4.35290903e-01
-5.58390617e-01 -9.61070955e-02 -3.49639595e-01 6.70540273e-01
-2.23188505e-01 4.60506052e-01 -7.83364624e-02 -2.78066516e-01
-7.72240281e-01 6.86528146e-01 4.94700074e-01 -6.05432451e-01
-5.12659132e-01 3.67582440e-01 -3.41858447e-01 -4.37464744e-01
3.95971417e-01 -5.45666695e-01 2.58231372e-01 -2.75637448e-01
5.72832763e-01 8.61312568e-01 -3.49067420e-01 -1.01513593e-02
5.42917550e-02 8.60976502e-02 -1.57227889e-02 -8.27798963e-01
1.14744937e+00 -2.05221981e-01 1.18708812e-01 3.84716928e-01
5.36379576e-01 -4.08593744e-01 -2.24753332e+00 -3.02682251e-01
4.09527451e-01 -4.89902884e-01 9.00430232e-02 -4.56064969e-01
-5.94755173e-01 4.81789261e-01 5.44166267e-01 2.69434415e-04
6.12517178e-01 -6.43283069e-01 6.26255393e-01 7.58638918e-01
8.37389827e-01 -1.85859668e+00 -1.09244019e-01 8.09914410e-01
7.21391261e-01 -1.08865392e+00 1.46637172e-01 2.62237310e-01
-9.17643130e-01 9.72745895e-01 4.34656203e-01 -2.39731342e-01
2.38816515e-01 1.18776239e-01 -8.94158408e-02 4.89723116e-01
-8.48935366e-01 -4.24192548e-01 -3.28375638e-01 7.74210513e-01
-1.47152647e-01 2.55252630e-01 -6.04758859e-01 -9.78232250e-02
2.02971727e-01 2.66360492e-01 4.80397582e-01 1.44341552e+00
-6.63672090e-01 -1.14945281e+00 -5.15447855e-01 1.77200094e-01
-5.64737380e-01 2.39956930e-01 9.93192047e-02 7.67797947e-01
-7.37476349e-01 1.10420632e+00 -3.72182846e-01 1.51626423e-01
2.55508780e-01 -7.14834314e-03 7.08340466e-01 -1.22809634e-01
-3.60641807e-01 2.17914969e-01 6.22424856e-02 -9.32284594e-01
-2.94735968e-01 -8.13928843e-01 -1.39371061e+00 -5.73588073e-01
5.32837138e-02 3.19457829e-01 7.58571923e-01 9.37743068e-01
1.64949283e-01 7.90175050e-02 8.28776121e-01 -9.93145227e-01
-1.68188024e+00 -7.35808432e-01 -6.47491753e-01 -1.83285344e-02
3.67559135e-01 -9.91859019e-01 -1.06221639e-01 -4.68669862e-01] | [4.156149864196777, 2.3501391410827637] |
855f4ff2-e04d-4a22-b9c1-975d0ef385c7 | hard-sample-guided-hybrid-contrast-learning | 2109.12333 | null | https://arxiv.org/abs/2109.12333v2 | https://arxiv.org/pdf/2109.12333v2.pdf | Hard-sample Guided Hybrid Contrast Learning for Unsupervised Person Re-Identification | Unsupervised person re-identification (Re-ID) is a promising and very challenging research problem in computer vision. Learning robust and discriminative features with unlabeled data is of central importance to Re-ID. Recently, more attention has been paid to unsupervised Re-ID algorithms based on clustered pseudo-label. However, the previous approaches did not fully exploit information of hard samples, simply using cluster centroid or all instances for contrastive learning. In this paper, we propose a Hard-sample Guided Hybrid Contrast Learning (HHCL) approach combining cluster-level loss with instance-level loss for unsupervised person Re-ID. Our approach applies cluster centroid contrastive loss to ensure that the network is updated in a more stable way. Meanwhile, introduction of a hard instance contrastive loss further mines the discriminative information. Extensive experiments on two popular large-scale Re-ID benchmarks demonstrate that our HHCL outperforms previous state-of-the-art methods and significantly improves the performance of unsupervised person Re-ID. The code of our work is available soon at https://github.com/bupt-ai-cz/HHCL-ReID. | ['Gang He', 'Chuang Zhu', 'Zheng Hu'] | 2021-09-25 | null | null | null | null | ['unsupervised-person-re-identification'] | ['computer-vision'] | [-2.27114990e-01 -2.86780626e-01 -3.35509151e-01 -6.43335521e-01
-5.64249694e-01 -3.28799099e-01 7.89510608e-01 2.04227164e-01
-7.77419567e-01 6.33857667e-01 1.89415082e-01 4.13616240e-01
-2.90524326e-02 -5.03056705e-01 -4.89191562e-01 -7.36423254e-01
2.22461417e-01 7.69352376e-01 1.15303788e-02 1.40871212e-01
2.43638158e-01 3.04815173e-01 -1.69182825e+00 -2.64579773e-01
1.02195776e+00 4.56516236e-01 1.71694458e-02 2.55966097e-01
2.10897028e-01 4.76642042e-01 -2.64023334e-01 -7.87460506e-01
4.55211252e-01 -4.85779077e-01 -7.84854352e-01 -7.10859075e-02
6.98655546e-01 -1.41093493e-01 -4.40505326e-01 1.37768531e+00
8.71894598e-01 5.09695351e-01 8.29661727e-01 -1.36688364e+00
-6.26637101e-01 6.36941910e-01 -1.00601757e+00 3.18386197e-01
2.40996093e-01 -5.84740452e-02 7.95573771e-01 -1.03823805e+00
4.10630077e-01 1.37215543e+00 7.88815796e-01 7.84280181e-01
-1.28833091e+00 -1.15201998e+00 2.50242203e-01 5.85120022e-01
-1.75357020e+00 -4.87988353e-01 8.14553857e-01 -4.29611862e-01
4.03682530e-01 2.18656361e-02 2.96539515e-01 9.18015718e-01
-4.64585274e-01 9.49590027e-01 1.06592548e+00 -5.02876937e-01
-8.75076428e-02 2.59015113e-01 6.51939750e-01 6.75861955e-01
4.66230869e-01 2.44407475e-01 -3.60681772e-01 6.39029965e-02
4.47171509e-01 4.24365759e-01 4.10292856e-02 -3.24099779e-01
-1.11984050e+00 4.76077646e-01 6.39235020e-01 1.79461032e-01
-3.13798562e-02 -7.30265805e-04 4.76023465e-01 2.07973197e-01
4.81467694e-01 -2.42901146e-02 3.68911289e-02 9.76515859e-02
-9.95806456e-01 4.78016317e-01 4.93647724e-01 9.57055748e-01
9.01538908e-01 -4.29523021e-01 -4.49608862e-01 1.11109006e+00
1.22503795e-01 5.42424679e-01 6.11490428e-01 -7.28364289e-01
3.53744030e-01 5.70298195e-01 -2.53360998e-02 -9.33510661e-01
-4.25535142e-01 -5.84256232e-01 -1.12299883e+00 -6.03290014e-02
5.59666514e-01 6.64277598e-02 -8.39290082e-01 1.75215340e+00
4.67384815e-01 7.21245050e-01 -2.57735729e-01 9.54472661e-01
8.67853582e-01 2.96561360e-01 2.38151908e-01 7.11604506e-02
1.28714216e+00 -1.17992330e+00 -5.22949815e-01 -1.25580519e-01
3.75257850e-01 -4.30970103e-01 6.80330873e-01 1.64177343e-01
-9.14282799e-01 -9.30716753e-01 -8.82837713e-01 -8.08415264e-02
-2.56325275e-01 3.24909449e-01 2.89446682e-01 7.45405197e-01
-8.33443224e-01 4.98608589e-01 -5.19902349e-01 -4.71241683e-01
5.36099911e-01 5.13245761e-01 -5.45770764e-01 -4.50239360e-01
-8.50863039e-01 5.22712767e-01 4.34183657e-01 7.20144138e-02
-6.06967866e-01 -6.21347725e-01 -7.40123928e-01 -6.89259768e-02
3.85632604e-01 -6.27332389e-01 1.08670783e+00 -8.10062110e-01
-1.19281566e+00 1.17870307e+00 -4.33609903e-01 -3.56344610e-01
8.16300631e-01 -2.28941858e-01 -4.32763159e-01 -2.22134385e-02
4.49959546e-01 8.10756385e-01 7.42455304e-01 -1.51026130e+00
-7.52055645e-01 -6.88756764e-01 -3.53272080e-01 2.98752695e-01
-4.74280626e-01 2.64319591e-02 -9.06819344e-01 -7.01043248e-01
-9.81072932e-02 -1.18077111e+00 -7.66655505e-02 -3.07073712e-01
-5.98415136e-01 -7.49301195e-01 3.96881700e-01 -6.27258956e-01
1.15020061e+00 -2.09077239e+00 -1.74378697e-02 3.63281816e-01
4.44068879e-01 3.39956045e-01 -8.02217722e-02 1.20498195e-01
-7.73223341e-02 -1.53113855e-03 -2.31113493e-01 -9.38594401e-01
1.31183147e-01 -2.74587393e-01 2.28192464e-01 7.83450544e-01
-2.45233744e-01 9.14427519e-01 -9.32429969e-01 -6.74459457e-01
2.23110154e-01 3.35765570e-01 -4.75441068e-01 1.82551116e-01
4.22988236e-01 6.73314095e-01 -2.24035591e-01 6.64620697e-01
8.80027235e-01 -1.65505379e-01 -7.40175396e-02 -3.56628984e-01
-4.43834998e-02 -3.23716402e-01 -1.18678093e+00 1.56313276e+00
5.54645099e-02 2.34905899e-01 -2.94077426e-01 -1.17853022e+00
8.04445148e-01 -1.34473726e-01 3.92371923e-01 -8.74060929e-01
1.01119980e-01 1.19449288e-01 -3.00840169e-01 -1.29635870e-01
5.41009367e-01 1.59704402e-01 -1.24566682e-01 4.27634835e-01
-6.36297166e-02 7.24574208e-01 3.33356589e-01 3.07688296e-01
5.64227402e-01 9.18137208e-02 1.54651463e-01 -3.33818763e-01
7.80363083e-01 -2.19499797e-01 8.85210752e-01 1.05279207e+00
-6.36858582e-01 6.79618061e-01 -9.93927568e-02 -1.61714494e-01
-1.05664694e+00 -1.05364370e+00 -2.92237639e-01 1.11170828e+00
6.43779457e-01 -2.95687586e-01 -9.04345751e-01 -9.16941047e-01
2.28270456e-01 3.76281679e-01 -6.08078837e-01 -4.96651866e-02
-5.73295176e-01 -8.92024457e-01 5.34040928e-01 4.90005225e-01
8.94852042e-01 -7.57820964e-01 3.74213159e-01 5.33678718e-02
-3.71324807e-01 -9.09437299e-01 -9.03366804e-01 -2.67582446e-01
-5.80064535e-01 -1.12125945e+00 -1.24358869e+00 -1.17843378e+00
1.14994705e+00 5.85585833e-01 8.91063988e-01 1.92171961e-01
-4.93262947e-01 4.50387865e-01 -3.75145525e-01 -2.11887047e-01
-1.93700586e-02 2.39473879e-01 4.91233200e-01 2.46835381e-01
9.58861828e-01 -4.11130339e-01 -7.94677615e-01 6.12557948e-01
-4.09843951e-01 -1.87947169e-01 4.41201419e-01 9.52442586e-01
7.55909681e-01 1.02259383e-01 7.03139424e-01 -9.97673690e-01
3.88498664e-01 -4.62125838e-01 -3.78222883e-01 2.97651708e-01
-9.54912186e-01 1.25324698e-02 3.81614745e-01 -4.07906681e-01
-1.21566784e+00 9.88507941e-02 8.75135697e-03 -2.75870055e-01
-2.90326804e-01 8.20194930e-02 -2.83118457e-01 -3.19582596e-02
4.55163926e-01 2.99220324e-01 -1.31422028e-01 -6.27976418e-01
3.08276474e-01 8.29701960e-01 9.37754631e-01 -6.43824220e-01
1.07206368e+00 5.94134688e-01 -2.65131205e-01 -4.87753421e-01
-8.61490965e-01 -9.98878658e-01 -7.91190326e-01 -2.39543125e-01
7.32590556e-01 -1.22575819e+00 -9.00815666e-01 8.48438859e-01
-5.88671148e-01 -2.37394214e-01 -1.00786984e-01 5.09860814e-01
-1.47610009e-01 6.44791007e-01 -4.75762069e-01 -6.02857769e-01
-4.33980316e-01 -9.82094705e-01 7.33825266e-01 6.95528090e-01
7.40281045e-02 -7.45171428e-01 2.66086042e-01 5.39341390e-01
1.80220604e-01 -3.37195359e-02 1.76059768e-01 -7.10056603e-01
-3.07568699e-01 -4.18594927e-01 -5.42582810e-01 2.94818223e-01
1.94584861e-01 -5.17386556e-01 -1.01450634e+00 -6.48605525e-01
-6.53333068e-01 -3.20842773e-01 1.24944913e+00 2.86063015e-01
1.32592964e+00 -2.66590744e-01 -6.89780354e-01 7.75048852e-01
1.42761946e+00 -2.36896977e-01 4.52011406e-01 4.60541159e-01
1.09975052e+00 6.20312989e-01 6.50597513e-01 4.81258839e-01
9.42338169e-01 8.26695979e-01 -8.69545043e-02 -1.06186844e-01
-2.06842750e-01 -3.99829417e-01 1.85930431e-01 5.18884718e-01
-4.10142899e-01 -1.46130845e-01 -7.82439113e-01 7.21123695e-01
-2.02500415e+00 -1.22974098e+00 -1.09575197e-01 2.32322931e+00
8.33212972e-01 -1.12613983e-01 5.70390284e-01 5.16029447e-02
1.46570277e+00 -2.03433886e-01 -7.30755806e-01 3.06768835e-01
-5.35533540e-02 -1.98897824e-01 6.44587338e-01 4.18026745e-01
-1.37461555e+00 9.71677065e-01 4.71875334e+00 9.25266802e-01
-5.54188132e-01 3.12895477e-01 6.22742414e-01 3.70291620e-02
1.20978415e-01 -2.66047180e-01 -1.31753159e+00 8.38718116e-01
5.71256697e-01 -3.24475825e-01 3.64228159e-01 7.70711005e-01
3.95470150e-02 -3.65141928e-02 -1.04204106e+00 1.51031685e+00
3.45881999e-01 -8.92362475e-01 -1.98707417e-01 1.62126105e-02
9.41223919e-01 -1.46212131e-01 1.11658864e-01 2.99157411e-01
4.14632946e-01 -7.62163460e-01 5.93801260e-01 7.59180367e-01
8.25213552e-01 -1.07481837e+00 9.30937111e-01 1.27708018e-01
-1.40770745e+00 -1.84843615e-01 -4.26836491e-01 2.00029820e-01
-8.96320678e-03 5.95703244e-01 -4.42166150e-01 5.75316310e-01
1.09532547e+00 1.09700358e+00 -9.03646410e-01 1.34826183e+00
-1.34442106e-01 6.18705869e-01 -2.09727108e-01 2.30532199e-01
-2.19903588e-01 -7.25134313e-02 3.96445811e-01 1.47668648e+00
-4.34093475e-02 3.16784121e-02 4.14619684e-01 6.14449501e-01
-2.99480885e-01 5.30561246e-03 -8.40043053e-02 3.74420077e-01
6.37938380e-01 1.24413919e+00 -6.00216091e-01 -3.82937968e-01
-2.80765712e-01 1.33696973e+00 5.24499238e-01 2.96576649e-01
-6.48240805e-01 -3.67137045e-01 5.02502620e-01 1.27729982e-01
1.25310242e-01 -6.31794706e-02 9.91909057e-02 -1.21537912e+00
-9.58629772e-02 -5.64412177e-01 7.12897301e-01 -1.95022523e-01
-2.03028011e+00 2.68759847e-01 1.28440723e-01 -1.32024014e+00
-1.51648685e-01 -1.67590126e-01 -4.88304883e-01 8.30150783e-01
-1.75887728e+00 -1.34929860e+00 -7.87285209e-01 9.17581379e-01
4.15438235e-01 -4.16003495e-01 3.44029337e-01 6.23201847e-01
-9.77000952e-01 1.42105722e+00 2.97736436e-01 5.81375599e-01
1.24614179e+00 -1.32277632e+00 3.34914029e-01 1.01020682e+00
-1.91432074e-01 6.46923900e-01 4.77999598e-01 -6.81125283e-01
-9.95703101e-01 -1.23953891e+00 8.64985883e-01 -4.06465948e-01
1.00235447e-01 -4.26759094e-01 -7.70573556e-01 7.32166648e-01
-6.98717311e-02 -4.37032618e-03 7.70824492e-01 1.12273835e-01
-4.49595451e-01 -5.16693056e-01 -1.36706805e+00 3.63499761e-01
1.29061532e+00 -3.59617740e-01 -4.01258200e-01 1.98061109e-01
2.96696454e-01 -9.70809087e-02 -7.77136981e-01 3.49155366e-01
4.60803539e-01 -7.94268966e-01 1.22072434e+00 -1.53282061e-01
-8.89674411e-04 -5.96293449e-01 2.79352963e-01 -1.07217252e+00
-7.20115840e-01 -3.07241619e-01 1.98223200e-02 1.75478792e+00
-6.90873414e-02 -7.51209617e-01 7.89709687e-01 6.55474484e-01
1.88799649e-01 -5.32853231e-02 -6.43762469e-01 -9.33862090e-01
4.57190014e-02 3.91516089e-02 5.06244063e-01 1.09544337e+00
-1.63498342e-01 8.34355354e-02 -5.47235906e-01 2.74015814e-01
1.41120684e+00 -7.63969496e-02 1.03050148e+00 -1.47170925e+00
-6.91588670e-02 -4.56111878e-01 -5.94145536e-01 -1.00530136e+00
4.66921777e-01 -1.31192636e+00 3.07609681e-02 -1.38595557e+00
9.36030507e-01 -7.59646058e-01 -4.55099910e-01 5.97096920e-01
-5.82620859e-01 5.39775848e-01 2.38683462e-01 7.29849398e-01
-9.61252570e-01 5.69299400e-01 8.09791744e-01 -3.63989800e-01
-2.03985438e-01 6.81346329e-03 -8.42567623e-01 4.14296657e-01
9.41043079e-01 -6.46577299e-01 -1.61364600e-01 -1.67547449e-01
-3.04960847e-01 -6.52740896e-01 6.04992926e-01 -1.20548952e+00
7.67911732e-01 2.47839093e-01 6.70703530e-01 -6.93239033e-01
5.50661497e-02 -4.50585157e-01 2.52129231e-02 2.56475598e-01
-3.85952562e-01 -1.65298488e-02 -1.72942266e-01 7.60469854e-01
-1.52479753e-01 -2.47410193e-01 9.02188957e-01 -1.89926773e-01
-1.00684786e+00 6.44721210e-01 2.04304516e-01 8.73500258e-02
9.05100524e-01 -2.98660278e-01 -2.04023749e-01 -5.97839430e-02
-5.31812012e-01 5.47567189e-01 6.72367930e-01 4.42877740e-01
4.75608587e-01 -1.30761957e+00 -1.05159986e+00 5.39857708e-02
3.60026270e-01 -6.04579002e-02 4.56708193e-01 7.18517601e-01
-1.35545105e-01 1.65215746e-01 -1.77678391e-01 -6.39941752e-01
-1.42579079e+00 6.43764675e-01 3.05090040e-01 -1.91777527e-01
-7.18074679e-01 8.98518562e-01 2.53716618e-01 -6.68711185e-01
4.68464166e-01 6.30106986e-01 -3.31427366e-01 1.67444255e-03
9.10147548e-01 7.79872894e-01 -2.64827669e-01 -8.72885168e-01
-4.97113526e-01 8.30974340e-01 -5.89205146e-01 3.67672741e-02
1.07990217e+00 -4.35748249e-01 -6.32574931e-02 1.00253515e-01
1.19919598e+00 -1.10736355e-01 -1.24634409e+00 -6.19334698e-01
3.73089686e-02 -4.82060015e-01 -1.63290218e-01 -6.51051998e-01
-1.05762339e+00 4.20413315e-01 9.81389940e-01 -3.67741257e-01
1.02060711e+00 1.11985326e-01 8.19963753e-01 4.17447507e-01
4.04503524e-01 -1.32225502e+00 1.70366108e-01 1.51243150e-01
3.99999976e-01 -1.69755602e+00 1.80017427e-01 -2.84004688e-01
-5.29635787e-01 5.99439442e-01 8.04750144e-01 -2.17322648e-01
6.92734718e-01 -1.51581973e-01 -1.37072384e-01 1.81585073e-01
1.62265658e-01 -5.08113563e-01 3.12949240e-01 7.04030395e-01
1.12716913e-01 1.87549710e-01 -4.09404010e-01 7.06553817e-01
-1.21995464e-01 2.07656715e-02 1.59351900e-01 6.52018607e-01
-2.17802599e-01 -1.24713719e+00 -3.74205947e-01 4.08526212e-01
-3.90107840e-01 1.77914463e-02 -2.39466816e-01 5.20751297e-01
3.58239532e-01 9.47985828e-01 1.83068905e-02 -4.03280199e-01
8.10065642e-02 -2.06640482e-01 5.18301964e-01 -4.18670923e-01
-4.63728815e-01 -2.56421804e-01 -2.61821926e-01 -3.54342371e-01
-7.45948017e-01 -8.84129167e-01 -1.23094618e+00 -5.79008579e-01
-2.88188308e-01 1.47819385e-01 4.41755563e-01 8.04023623e-01
3.39496464e-01 -1.74257495e-02 8.17031860e-01 -9.99222279e-01
-2.99471557e-01 -8.46253216e-01 -6.11429095e-01 7.57706702e-01
3.03721219e-01 -7.82948673e-01 -4.43237215e-01 1.26555458e-01] | [14.803421974182129, 1.049684762954712] |
44db4a75-e6fc-407d-bcf1-26023fe35d38 | learning-reward-machines-for-partially | null | null | http://papers.nips.cc/paper/9685-learning-reward-machines-for-partially-observable-reinforcement-learning | http://papers.nips.cc/paper/9685-learning-reward-machines-for-partially-observable-reinforcement-learning.pdf | Learning Reward Machines for Partially Observable Reinforcement Learning | Reward Machines (RMs), originally proposed for specifying problems in Reinforcement Learning (RL), provide a structured, automata-based representation of a reward function that allows an agent to decompose problems into subproblems that can be efficiently learned using off-policy learning. Here we show that RMs can be learned from experience, instead of being specified by the user, and that the resulting problem decomposition can be used to effectively solve partially observable RL problems. We pose the task of learning RMs as a discrete optimization problem where the objective is to find an RM that decomposes the problem into a set of subproblems such that the combination of their optimal memoryless policies is an optimal policy for the original problem. We show the effectiveness of this approach on three partially observable domains, where it significantly outperforms A3C, PPO, and ACER, and discuss its advantages, limitations, and broader potential. | ['Sheila McIlraith', 'Rodrigo Toro Icarte', 'Rick Valenzano', 'Ethan Waldie', 'Toryn Klassen', 'Margarita Castro'] | 2019-12-01 | null | null | null | neurips-2019-12 | ['problem-decomposition'] | ['miscellaneous'] | [ 5.08078523e-02 6.88476324e-01 -5.87325633e-01 1.78210080e-01
-9.73969221e-01 -8.02157283e-01 6.82098985e-01 -6.98535815e-02
-5.94264627e-01 1.25479555e+00 2.69576907e-01 -5.04249573e-01
-3.22878599e-01 -4.47246134e-01 -7.32434332e-01 -7.55623996e-01
-4.55501944e-01 7.50255644e-01 -5.05255647e-02 -2.51919866e-01
2.15150237e-01 4.47456956e-01 -1.48455977e+00 -3.01672071e-02
9.58646953e-01 8.50676894e-01 4.64802444e-01 7.97328472e-01
1.71182081e-01 1.20856333e+00 -6.54001892e-01 4.29077417e-01
3.33076775e-01 -4.37821120e-01 -1.31851053e+00 4.65806246e-01
-2.07280442e-01 -5.49530387e-01 -3.06651771e-01 9.27661180e-01
2.07455575e-01 4.35542941e-01 6.06532931e-01 -1.22435129e+00
-4.11008388e-01 6.81734085e-01 -2.47504413e-02 -1.56826541e-01
4.67922777e-01 5.37821054e-01 1.17843974e+00 -8.03681463e-02
5.14795303e-01 1.55542850e+00 -1.17256925e-01 8.97039533e-01
-1.61879838e+00 1.37868254e-02 4.37452495e-01 -5.27658612e-02
-6.38921678e-01 -3.19491595e-01 3.42572421e-01 -8.49451944e-02
1.36349440e+00 1.82975873e-01 7.42076218e-01 9.96815860e-01
2.01544687e-01 1.14397216e+00 1.55124593e+00 -4.44828898e-01
8.91165316e-01 1.06455740e-02 -2.17553839e-01 7.68010080e-01
2.10290793e-02 6.71366811e-01 -1.84374094e-01 -4.90210682e-01
9.70710695e-01 -7.99124092e-02 -1.52443245e-01 -6.50243938e-01
-1.18853617e+00 9.56451237e-01 6.77036420e-02 8.01025778e-02
-6.73535287e-01 5.65014005e-01 2.43583266e-02 7.15952277e-01
-9.36809578e-04 1.11530435e+00 -7.53100455e-01 -3.75962347e-01
-2.57993370e-01 6.84271514e-01 1.04884779e+00 7.30019331e-01
7.52233267e-01 4.82301652e-01 -2.68069327e-01 5.47632456e-01
2.24407241e-01 7.79988348e-01 4.22082782e-01 -1.85527742e+00
4.27075773e-01 2.74998426e-01 9.85179067e-01 -1.94795430e-01
-3.88537705e-01 -1.26188487e-01 -2.40846891e-02 5.41465342e-01
1.68850988e-01 -5.67639887e-01 -6.96254611e-01 1.89168143e+00
3.88259031e-02 2.19470039e-01 6.51386619e-01 8.18434656e-01
-4.41781469e-02 8.10519278e-01 -2.71165781e-02 -5.63058496e-01
7.59320021e-01 -1.00365424e+00 -6.18989170e-01 -5.62761068e-01
7.62725830e-01 3.08542140e-02 1.08011484e+00 5.68553984e-01
-1.33258879e+00 -8.05277377e-02 -1.04337716e+00 4.06991780e-01
1.14053704e-01 -3.92208248e-02 7.67425179e-01 2.26164490e-01
-1.31213725e+00 9.56397951e-01 -9.23706055e-01 -2.87591312e-02
5.94535321e-02 6.91022396e-01 2.00336158e-01 1.35172859e-01
-1.05368185e+00 1.16329598e+00 5.97512305e-01 -2.41678864e-01
-1.79132640e+00 -3.64396274e-02 -8.78231287e-01 9.31535810e-02
1.00451696e+00 -4.31774616e-01 2.09929585e+00 -1.11847305e+00
-2.18511438e+00 3.95684659e-01 1.67148598e-02 -7.11335540e-01
1.13779388e-01 -1.76985681e-01 -1.67104661e-01 2.73212075e-01
4.23005000e-02 6.72472298e-01 9.77869272e-01 -1.32782149e+00
-1.00623620e+00 2.18629055e-02 6.90137804e-01 6.40287161e-01
1.72401164e-02 -2.22631857e-01 3.10712587e-02 -1.73723429e-01
-3.71217579e-01 -1.20278084e+00 -8.56287420e-01 -6.45723939e-01
-5.84750213e-02 -4.56610054e-01 6.01438284e-01 -2.31831238e-01
8.34285080e-01 -1.85185230e+00 6.67683601e-01 5.46681248e-02
8.69806036e-02 1.58498138e-01 -6.32499635e-01 5.14052391e-01
2.53036857e-01 5.53633869e-02 2.19215825e-02 -1.44891798e-01
2.56758183e-01 7.73355126e-01 -6.03265285e-01 3.07694525e-01
1.74490303e-01 1.01890004e+00 -1.26224065e+00 -1.52548477e-01
6.76721260e-02 -4.13066655e-01 -4.93008733e-01 7.26469398e-01
-1.18889451e+00 6.34119153e-01 -8.96429360e-01 2.78170735e-01
-2.14581430e-01 -1.92191094e-01 6.81810379e-01 9.55626309e-01
2.72120629e-02 4.55034584e-01 -1.26914108e+00 1.44485438e+00
-4.41371709e-01 1.03294253e-01 2.65981942e-01 -1.08659112e+00
8.02576303e-01 6.77819133e-01 6.74799860e-01 -6.79169297e-01
-2.80986130e-02 1.52976438e-01 -1.45856857e-01 -4.93140817e-01
4.26042974e-01 -7.48595744e-02 -4.00310129e-01 8.88068378e-01
1.04854308e-01 -4.19222713e-01 1.59714907e-01 -9.66656767e-03
1.28683054e+00 3.55860710e-01 4.01169777e-01 -4.61618081e-02
3.62227708e-01 2.27055609e-01 6.15979493e-01 1.25847411e+00
-7.82088637e-02 -2.99783140e-01 8.63576710e-01 -2.61132061e-01
-9.18180287e-01 -9.72505748e-01 6.18683338e-01 1.20542192e+00
1.65288858e-02 -1.20617084e-01 -5.07675111e-01 -9.43074584e-01
8.14664587e-02 9.58438516e-01 -3.85942310e-01 -1.55904144e-01
-6.94607437e-01 -1.86972111e-01 1.38896585e-01 3.66577566e-01
2.94983387e-01 -1.61859167e+00 -1.01062965e+00 4.36256766e-01
-4.58430722e-02 -8.64460289e-01 -5.61890364e-01 4.51289743e-01
-1.20828199e+00 -1.13010311e+00 -4.92947608e-01 -6.40619874e-01
6.48257375e-01 3.26295316e-01 1.07050431e+00 2.53660735e-02
3.36739570e-01 1.08792043e+00 -1.62311345e-01 -2.00624034e-01
-6.55367792e-01 -1.18699312e-01 3.48813504e-01 -2.61852294e-01
-3.31559896e-01 -4.30600584e-01 -2.29052335e-01 7.43299797e-02
-8.33924890e-01 -8.90134946e-02 5.93716562e-01 8.95289063e-01
4.71673906e-01 9.31296647e-02 7.17333734e-01 -7.57658601e-01
1.13372731e+00 -4.20458496e-01 -1.02530289e+00 3.59559685e-01
-8.02353621e-01 8.34425151e-01 9.77353156e-01 -6.13667548e-01
-1.02144170e+00 2.25073844e-01 5.61882257e-01 -3.54265541e-01
-3.36774528e-01 4.01183486e-01 -8.02761167e-02 2.68675294e-02
5.40769100e-01 5.25120139e-01 1.85983226e-01 -3.71069700e-01
6.12281799e-01 4.62401956e-01 3.64660710e-01 -1.26188290e+00
4.80625510e-01 -5.51077612e-02 3.59861180e-02 -3.89714181e-01
-8.89189720e-01 -1.97099537e-01 -1.31498486e-01 -9.64156911e-02
5.16896963e-01 -6.68506086e-01 -1.13328564e+00 -1.41946167e-01
-8.95067871e-01 -1.00810814e+00 -7.73918748e-01 3.34702075e-01
-1.57048583e+00 1.61370218e-01 -5.33941269e-01 -1.16897106e+00
7.76069537e-02 -1.28593647e+00 7.38323867e-01 3.02082062e-01
-8.32492337e-02 -9.06668425e-01 3.62045288e-01 1.38013110e-01
1.52333602e-01 -2.60799658e-03 1.22912061e+00 -6.22655869e-01
-7.01859117e-01 4.20962721e-01 3.79029870e-01 2.81918913e-01
7.19252625e-04 -5.38009107e-01 -4.98828918e-01 -6.88028276e-01
-8.81457776e-02 -1.01772285e+00 4.43415344e-01 3.16970676e-01
9.99686241e-01 -9.65313315e-01 -9.22851562e-02 6.85436353e-02
1.31427753e+00 5.63856244e-01 3.02440941e-01 4.90299731e-01
-2.81847678e-02 3.00446928e-01 9.37638164e-01 6.00130260e-01
2.85758764e-01 5.21034777e-01 6.24120057e-01 3.87263536e-01
5.66397309e-01 -4.82171744e-01 8.72601449e-01 1.81656867e-01
-2.94970006e-01 4.87762094e-02 -5.91368318e-01 3.55119884e-01
-2.53543377e+00 -1.03086042e+00 7.02538013e-01 2.23417950e+00
9.59978104e-01 1.29244536e-01 5.02826333e-01 -2.00727224e-01
2.78117567e-01 1.99407578e-01 -1.08683217e+00 -8.87898624e-01
2.64348567e-01 3.48498791e-01 4.70708132e-01 7.39273071e-01
-8.45272005e-01 1.20188785e+00 7.68151808e+00 4.58526433e-01
-8.08963716e-01 -1.25168189e-01 3.02382499e-01 -7.19741511e-04
-2.99524337e-01 2.52600372e-01 -5.03173053e-01 6.71723438e-03
1.17287278e+00 -2.48473987e-01 1.36126220e+00 9.60968792e-01
5.38600206e-01 -2.47246265e-01 -1.47491062e+00 5.78291655e-01
-5.74119627e-01 -1.16790807e+00 -1.98229522e-01 2.62702584e-01
9.91447031e-01 -7.69173354e-02 2.54517645e-01 9.23819065e-01
1.36305749e+00 -1.20336199e+00 6.94967210e-01 2.52992511e-01
4.79612529e-01 -8.37330520e-01 2.80406833e-01 9.50532913e-01
-7.44674742e-01 -7.59998620e-01 -3.24387103e-01 -5.30974209e-01
-2.25188911e-01 -3.07605952e-01 -1.10925019e+00 2.65233636e-01
8.16148818e-02 5.55551708e-01 7.38538429e-02 7.03742802e-01
-6.87416494e-01 5.74220002e-01 3.13405367e-03 -4.16007012e-01
7.67139733e-01 -3.31938058e-01 6.04275465e-01 6.61849141e-01
-2.98770331e-02 3.61884356e-01 8.97133291e-01 8.87240887e-01
1.72728375e-01 -3.56127024e-01 -7.59387136e-01 -3.92393976e-01
3.91118050e-01 9.89707708e-01 -3.15994620e-01 -2.90421486e-01
-1.34909257e-01 8.10130954e-01 9.60615337e-01 6.78549349e-01
-6.54033065e-01 1.32416517e-01 7.87456572e-01 -4.50920790e-01
4.20933425e-01 -4.13979083e-01 3.36537808e-01 -1.15142477e+00
-5.00754774e-01 -1.45625019e+00 4.14798558e-01 -6.46577835e-01
-7.22251177e-01 2.31531978e-01 2.85037188e-03 -1.08252645e+00
-8.59600544e-01 -3.81956220e-01 -3.65296364e-01 5.14732420e-01
-1.34166789e+00 -4.43035901e-01 5.19003093e-01 5.73539317e-01
5.29625654e-01 -2.97371149e-01 1.13164353e+00 -6.79303229e-01
-4.35369253e-01 -7.27301016e-02 3.67274880e-01 -2.26926759e-01
-1.30533399e-02 -1.87285757e+00 -3.17371115e-02 4.50525016e-01
1.42859817e-01 1.16387740e-01 8.24518502e-01 -4.78035480e-01
-1.85412526e+00 -9.77307081e-01 2.84325570e-01 -1.63216770e-01
7.49611139e-01 6.31836057e-02 -3.09583396e-01 9.82029498e-01
2.22451434e-01 -1.78191051e-01 1.89329460e-02 1.26178935e-01
-3.49729843e-02 9.19593200e-02 -1.08171058e+00 8.69583666e-01
8.00228238e-01 -4.18897748e-01 -8.71946096e-01 3.37655663e-01
8.92962337e-01 -5.51692069e-01 -6.23576760e-01 -1.44134849e-01
1.71006899e-02 -4.70029950e-01 7.21487164e-01 -1.42280436e+00
3.32171351e-01 -8.86996016e-02 -5.67666404e-02 -1.84757638e+00
-3.56839687e-01 -1.31444561e+00 -7.04786181e-01 4.36461002e-01
3.54878366e-01 -8.46882224e-01 6.92231476e-01 6.59987688e-01
4.18896563e-02 -7.77787089e-01 -9.37535644e-01 -1.03137052e+00
2.46636383e-02 -1.18292861e-01 4.45823073e-01 4.19857323e-01
4.52646375e-01 3.76184016e-01 -5.69728673e-01 1.59135029e-01
5.82259715e-01 4.48742121e-01 5.31117380e-01 -8.03919017e-01
-9.94113445e-01 -2.55650669e-01 4.01147217e-01 -1.60627556e+00
7.06783235e-01 -7.08889663e-01 4.50565368e-01 -1.43317401e+00
3.41968685e-02 -6.03141606e-01 -3.81592274e-01 8.43641639e-01
2.41689444e-01 -8.10190082e-01 5.27831554e-01 2.13786647e-01
-1.22450125e+00 6.74315631e-01 1.67647064e+00 -1.82273805e-01
-6.02239192e-01 3.19809914e-01 -7.71653593e-01 6.42259240e-01
1.02694345e+00 -5.69321036e-01 -6.23219490e-01 -2.04373375e-01
1.03504010e-01 1.19656181e+00 8.62973481e-02 -5.76142251e-01
-8.58800486e-02 -1.06536233e+00 -1.49566829e-01 -3.09029464e-02
3.04650128e-01 -7.65125036e-01 -3.09220016e-01 8.31281245e-01
-8.82855713e-01 1.22502565e-01 -2.37846375e-02 7.20374465e-01
1.20809972e-01 -4.89441276e-01 4.98140097e-01 -3.98239225e-01
-6.11699283e-01 2.68347234e-01 -8.64489436e-01 2.92364448e-01
1.20211506e+00 3.54371876e-01 -2.02913985e-01 -7.99673676e-01
-9.02411044e-01 7.39413500e-01 1.64981946e-01 1.71208814e-01
6.88078880e-01 -1.04095566e+00 -3.65066826e-01 -2.47058392e-01
-2.50274479e-01 -1.57976732e-01 -4.45426434e-01 4.29430574e-01
1.43117115e-01 8.80091190e-01 -1.56960621e-01 -9.02721211e-02
-9.76393461e-01 6.47352755e-01 6.25264943e-01 -8.90594542e-01
-5.34458518e-01 9.42510441e-02 -1.53873429e-01 -5.73711276e-01
4.70367730e-01 -5.80586493e-01 -3.17815125e-01 -5.94376385e-01
3.43778640e-01 2.37308204e-01 -5.44248343e-01 1.68439656e-01
1.47292748e-01 -8.96995366e-02 -4.90759909e-02 -7.20508218e-01
1.52825260e+00 6.24678330e-03 1.22885117e-02 3.52074087e-01
5.70077181e-01 -4.26089585e-01 -1.77597582e+00 -4.33512986e-01
3.61677736e-01 -1.91919327e-01 -4.93959375e-02 -8.99281025e-01
-6.75146818e-01 3.37449223e-01 1.98042408e-01 4.17302251e-01
8.09981465e-01 1.09443218e-01 5.28923094e-01 9.58033502e-01
8.92776132e-01 -1.37748706e+00 5.71031094e-01 8.52684796e-01
8.49875987e-01 -8.16108108e-01 -1.36481375e-01 3.30817252e-01
-1.04014242e+00 1.09355092e+00 6.27238691e-01 -4.48934317e-01
1.40219569e-01 2.50490248e-01 -3.62763703e-01 8.00711736e-02
-1.41240931e+00 -5.95431924e-01 -7.49548450e-02 8.72239113e-01
-3.00981343e-01 3.42312574e-01 -1.56752914e-01 4.32982236e-01
7.47509897e-02 1.66108444e-01 6.79964900e-01 1.37791514e+00
-9.70719934e-01 -1.47485387e+00 -3.67191792e-01 4.04918313e-01
-2.07162455e-01 4.94182706e-01 -4.18540955e-01 4.71945345e-01
-2.77992666e-01 1.10918558e+00 -2.19455376e-01 -2.01207414e-01
2.05724761e-01 -8.77590850e-02 9.00649428e-01 -9.39268053e-01
-4.05080140e-01 1.94247381e-03 3.90185744e-01 -9.73848760e-01
-2.74476886e-01 -6.54229224e-01 -1.58374143e+00 -9.02769435e-03
2.08581343e-01 5.53239822e-01 1.21077806e-01 1.28959942e+00
2.10773662e-01 3.21029425e-01 1.10599804e+00 -6.94351912e-01
-1.62775135e+00 -5.56199849e-01 -6.92856848e-01 1.57711565e-01
6.34301901e-01 -6.90412045e-01 -2.76498795e-01 -3.58760893e-01] | [4.100226879119873, 1.7518945932388306] |
29a21968-c0a1-4a77-8418-5ebf2730871e | mobile-end-tone-mapping-based-on-integral | 2102.01289 | null | https://arxiv.org/abs/2102.01289v1 | https://arxiv.org/pdf/2102.01289v1.pdf | Mobile-end Tone Mapping based on Integral Image and Integral Histogram | Wide dynamic range (WDR) image tone mapping is in high demand in many applications like film production, security monitoring, and photography. It is especially crucial for mobile devices because most of the images taken today are from mobile phones, hence such technology is highly demanded in the consumer market of mobile devices and is essential for a good customer experience. However, high-quality and high-performance WDR image tone mapping implementations are rarely found in the mobile-end. In this paper, we introduce a high performance, mobile-end WDR image tone mapping implementation. It leverages the tone mapping results of multiple receptive fields and calculates a suitable value for each pixel. The utilization of integral image and integral histogram significantly reduce the required computation. Moreover, GPU parallel computation is used to increase the processing speed. The experimental results indicate that our implementation can process a high-resolution WDR image within a second on mobile devices and produce appealing image quality. | ['Orly Yadid-Pecht', 'Ulian Shahnovich', 'Ziyi Liu', 'Mengchen Lin', 'Jie Yang'] | 2021-02-02 | null | null | null | null | ['tone-mapping'] | ['computer-vision'] | [ 6.09686196e-01 -7.46083140e-01 -1.06618360e-01 -8.49372447e-02
-5.22774041e-01 -2.07626224e-01 2.02486575e-01 -1.58339813e-01
-5.37733555e-01 3.54279011e-01 -1.28872797e-01 -5.42125762e-01
-9.83994827e-02 -1.28845537e+00 -2.61848360e-01 -4.95289087e-01
5.36935806e-01 -1.51514769e-01 8.25292170e-01 -4.27707583e-01
3.53552580e-01 3.33605438e-01 -1.54660487e+00 2.94148624e-01
8.20633590e-01 1.39810312e+00 7.05175519e-01 6.68920279e-01
-4.28911112e-02 5.95279813e-01 -3.55010867e-01 -4.93740439e-01
3.95889968e-01 -2.77639776e-01 -2.22371608e-01 1.15504719e-01
2.04119131e-01 -9.05060768e-01 -3.69167775e-01 1.18247104e+00
6.59326911e-01 -3.91432978e-02 2.85417199e-01 -4.60307240e-01
-5.33700645e-01 1.41994342e-01 -1.16907978e+00 3.79920870e-01
1.19867422e-01 7.84488916e-02 5.01647830e-01 -6.68403864e-01
3.92394781e-01 6.98964953e-01 3.26178074e-01 1.29281610e-01
-9.03036594e-01 -6.48021400e-01 -4.06933397e-01 2.25546926e-01
-1.38303065e+00 -3.61725003e-01 6.97094500e-01 1.57543525e-01
7.36637294e-01 5.17832816e-01 5.29798508e-01 3.15270424e-01
4.20220703e-01 6.93474784e-02 1.26775861e+00 -3.79944742e-01
-2.28860583e-02 4.02816124e-02 -2.67931134e-01 4.13165510e-01
8.66554826e-02 -7.97894076e-02 -5.33150136e-01 3.19988728e-01
1.38355005e+00 1.92772835e-01 -3.65432292e-01 2.30728805e-01
-1.05916417e+00 3.14866602e-01 2.53691137e-01 3.13197941e-01
-3.27485859e-01 1.70736760e-01 2.18337536e-01 3.37954164e-01
3.53873491e-01 3.37052792e-02 2.54690588e-01 -4.72546548e-01
-8.75965893e-01 -1.28537223e-01 6.15027593e-03 5.65907896e-01
5.49977005e-01 4.16135304e-02 2.26215884e-01 1.07996011e+00
-3.51070389e-02 8.48349154e-01 5.68956435e-01 -9.74198759e-01
3.66361350e-01 4.27075118e-01 1.84544086e-01 -1.04442155e+00
-8.63867775e-02 -4.91451174e-02 -9.12967801e-01 3.28514010e-01
2.15098813e-01 2.54147708e-01 -5.15446305e-01 6.98720455e-01
2.86136717e-01 1.67094227e-02 -1.57914996e-01 1.17059422e+00
6.45407856e-01 1.08185327e+00 -1.43436953e-01 -2.00369060e-01
1.68582737e+00 -4.36592489e-01 -8.58325660e-01 -8.87978598e-02
-4.19508144e-02 -1.30328274e+00 1.50910580e+00 5.21767914e-01
-1.34605348e+00 -7.62378156e-01 -1.25210214e+00 -4.46825325e-01
8.30728412e-02 2.19471946e-01 5.17476082e-01 8.07223856e-01
-7.84200728e-01 2.73958117e-01 -4.59170610e-01 4.81572449e-02
2.18156397e-01 4.72207576e-01 1.52347842e-02 -1.73703790e-01
-9.13358510e-01 5.28526008e-01 -1.64589226e-01 -1.43319786e-01
6.33237734e-02 -7.06566095e-01 -3.07230800e-01 1.23980261e-01
2.18926415e-01 -2.77647972e-01 1.29844153e+00 -7.24814892e-01
-1.85298491e+00 8.16910207e-01 6.46617040e-02 -3.28929186e-01
4.28183228e-01 1.59898475e-01 -6.85391486e-01 4.93907660e-01
-3.06991726e-01 3.09625953e-01 1.12177575e+00 -6.60179913e-01
-7.86248386e-01 -1.79331362e-01 -1.41897053e-02 4.10681367e-01
-3.73910367e-01 2.76274621e-01 -6.89047456e-01 -4.79838163e-01
1.46035671e-01 -5.19922495e-01 6.46581426e-02 1.02195762e-01
9.12154987e-02 3.72716397e-01 1.16346931e+00 -6.07080996e-01
1.46707571e+00 -2.33803487e+00 -7.02496648e-01 2.94568151e-01
8.92430320e-02 4.27851856e-01 3.66084397e-01 3.80493179e-02
2.31467292e-01 -3.57409894e-01 1.03028521e-01 2.82963336e-01
-2.73885638e-01 -4.36723799e-01 -7.16480792e-01 3.58872592e-01
-2.27316990e-01 7.69942760e-01 -3.44577342e-01 -3.90081346e-01
6.36288524e-01 7.05839515e-01 -3.40634555e-01 -4.14239019e-02
4.93661314e-02 2.02470213e-01 -2.14583501e-01 6.75016522e-01
8.19845736e-01 -4.95179266e-01 1.49394646e-01 -5.52594066e-01
-6.06841743e-01 7.76617676e-02 -9.81364012e-01 1.26794767e+00
-7.37693846e-01 6.81220293e-01 -1.05584398e-01 -2.93791801e-01
1.16048419e+00 -3.87565903e-02 4.09880638e-01 -1.57075775e+00
3.87453258e-01 5.24072647e-01 5.31642176e-02 -2.73856908e-01
1.19200647e+00 -4.72800992e-02 -1.26632378e-01 6.01979852e-01
-7.72636116e-01 -2.42338583e-01 9.83213112e-02 -2.13173494e-01
6.49128079e-01 -1.26098245e-01 3.80413979e-01 -6.89168498e-02
5.57480812e-01 -1.43776417e-01 1.15275085e-01 2.70199835e-01
-1.33982562e-02 6.78710222e-01 1.00978687e-01 -4.90938514e-01
-1.26178920e+00 -1.10624039e+00 -2.99304992e-01 8.43779385e-01
8.06565464e-01 -2.62728751e-01 -7.52672851e-01 2.37172201e-01
-6.16999507e-01 2.92257339e-01 1.55959800e-01 2.46997327e-02
-7.92309999e-01 -4.98133451e-01 1.75458685e-01 3.12806189e-01
1.13352287e+00 -8.82588923e-01 -1.28180826e+00 2.81923860e-01
-1.70462564e-01 -1.23404813e+00 -7.78520465e-01 -3.88141394e-01
-8.79985869e-01 -6.78823709e-01 -8.36361051e-01 -9.26666021e-01
4.75496471e-01 8.23675692e-01 7.50523150e-01 -1.65080912e-02
-5.11679828e-01 -2.01137085e-02 -2.40138650e-01 -1.68790556e-02
-8.09205994e-02 -1.55626863e-01 -1.29751861e-01 5.53226471e-02
3.80711764e-01 -4.06103462e-01 -1.04850996e+00 6.78055286e-01
-1.11731410e+00 3.45653981e-01 6.13748431e-01 5.32664061e-01
1.01643395e+00 5.94045043e-01 4.11010981e-01 -6.96685076e-01
7.12336600e-01 1.93990692e-01 -1.04724646e+00 2.15617999e-01
-2.95347601e-01 -4.43674028e-01 7.73366809e-01 -6.64066434e-01
-1.37159765e+00 -3.52437526e-01 -2.75141478e-01 -1.94654297e-02
3.64333540e-01 -2.70145945e-02 6.01997972e-02 -2.36421585e-01
5.76168716e-01 3.33473474e-01 4.38327901e-02 -9.29495767e-02
-3.73208486e-02 1.10366046e+00 7.76133716e-01 -1.98814616e-01
5.96304178e-01 6.89632833e-01 -3.00970245e-02 -1.13226056e+00
-1.30378842e-01 -3.18153054e-01 2.93814451e-01 -4.27276284e-01
7.66469538e-01 -1.02186692e+00 -1.03966069e+00 5.25415301e-01
-6.40406549e-01 -3.90048414e-01 -1.09080940e-01 4.41599160e-01
-3.08820695e-01 1.17626935e-01 -6.64880872e-01 -5.73367417e-01
-6.89178228e-01 -1.34524500e+00 9.73408639e-01 6.94914997e-01
2.96576042e-02 -6.29916310e-01 -4.00930375e-01 1.93958849e-01
8.88298452e-01 -3.14146370e-01 7.60602832e-01 7.04273403e-01
-1.07776761e+00 -7.55612552e-02 -8.26035678e-01 -1.78708807e-01
3.78799677e-01 -1.57448933e-01 -9.23507929e-01 1.38614669e-01
6.37102798e-02 3.80637273e-02 4.89056438e-01 6.03435576e-01
1.40875769e+00 2.11550698e-01 -2.92578302e-02 5.24993777e-01
1.66161871e+00 5.65243661e-01 1.15776014e+00 1.65715277e-01
4.25480574e-01 2.01149866e-01 1.10009325e+00 3.66002202e-01
1.56791463e-01 1.09945643e+00 2.20723860e-02 -3.81135911e-01
-3.06316942e-01 -5.33575453e-02 2.85245240e-01 8.01670611e-01
-1.15232922e-01 -1.48467019e-01 -5.49504101e-01 1.40387058e-01
-1.28302252e+00 -1.05923355e+00 -1.61069229e-01 2.38457894e+00
7.66604483e-01 4.11732718e-02 6.72620162e-03 2.40567252e-01
8.77725482e-01 9.82874334e-02 -5.88089883e-01 -5.97396076e-01
-9.12690535e-03 6.56061769e-01 6.59144104e-01 4.06863809e-01
-7.17318714e-01 7.14281738e-01 5.60581160e+00 1.30576015e+00
-1.62693894e+00 1.51549116e-01 9.11383390e-01 -1.47760376e-01
-3.12479287e-01 -4.86617267e-01 -7.53770530e-01 5.81945896e-01
7.59788692e-01 -1.41898207e-02 4.99088496e-01 6.73098862e-01
4.23080057e-01 -7.02492535e-01 -2.10152239e-01 1.58161867e+00
-1.83225408e-01 -1.35788524e+00 -3.28689605e-01 3.18828553e-01
5.87332070e-01 -4.14925158e-01 7.20662892e-01 -4.00868148e-01
-4.52333033e-01 -7.02174366e-01 5.28201759e-01 2.04291046e-01
1.57658243e+00 -1.09642279e+00 3.43968749e-01 -3.61405984e-02
-1.32208145e+00 1.38805196e-01 -5.68721652e-01 1.19288173e-02
3.71000320e-01 7.29703546e-01 -5.12581110e-01 2.89608482e-02
7.91867077e-01 1.08742565e-01 -3.87247145e-01 7.06580639e-01
3.09503466e-01 2.52815783e-01 -2.81192571e-01 -1.13384105e-01
-1.33086771e-01 -4.57441121e-01 1.04899064e-01 7.30594337e-01
7.95937359e-01 1.96754739e-01 -3.55816334e-01 4.85109925e-01
-1.78892300e-01 1.13082789e-01 -3.75602067e-01 1.54668391e-01
3.54722947e-01 1.45023108e+00 -1.18818033e+00 -1.02151029e-01
-7.05293655e-01 1.23158109e+00 -3.48189652e-01 -8.82070512e-02
-1.12563658e+00 -8.82351696e-01 7.03995168e-01 5.92737436e-01
3.47541928e-01 -3.68000239e-01 -4.41913694e-01 -8.88574719e-01
1.17702171e-01 -7.85969615e-01 -2.31654480e-01 -7.82123268e-01
-7.72161961e-01 5.74798763e-01 -4.97073591e-01 -1.73825943e+00
-8.00514594e-02 -6.71543717e-01 -4.31748956e-01 9.41394508e-01
-1.75459623e+00 -6.22593462e-01 -5.04720688e-01 6.76677585e-01
5.12293041e-01 1.94722824e-02 5.25839686e-01 6.64866865e-01
-2.23330528e-01 5.22574186e-01 1.53932080e-01 -1.97821885e-01
7.00987637e-01 -6.60905480e-01 5.76204240e-01 8.57457995e-01
-1.20915867e-01 7.36113071e-01 2.68958509e-01 -5.26476920e-01
-1.45800543e+00 -8.63763213e-01 4.92741674e-01 2.61770725e-01
2.84885556e-01 -1.22237198e-01 -6.84451222e-01 -1.68384939e-01
1.81769520e-01 1.61569759e-01 7.25472689e-01 -6.15345597e-01
-6.72461763e-02 -6.54024720e-01 -1.26045918e+00 8.52152765e-01
7.78975785e-01 -7.39266574e-01 2.46248722e-01 -1.42937586e-01
4.35623318e-01 -8.21247160e-01 -8.07824850e-01 5.96930496e-02
7.43723631e-01 -1.29897892e+00 1.12730587e+00 8.22087169e-01
3.94073039e-01 -7.30630636e-01 -1.13108829e-01 -6.32728636e-01
6.79768175e-02 -5.87386668e-01 2.78790295e-02 1.08636856e+00
1.32402107e-01 -6.85259283e-01 5.71499646e-01 6.25410914e-01
2.42396578e-01 -6.44950688e-01 -7.16691852e-01 -4.33885515e-01
-7.33723998e-01 -4.21364188e-01 7.18924940e-01 4.30588961e-01
-1.47859514e-01 1.80151120e-01 -5.76482832e-01 -3.50812972e-02
6.22569263e-01 6.19851530e-01 5.33296406e-01 -6.37518466e-01
-5.53326190e-01 -2.48239174e-01 -4.77858245e-01 -1.34636378e+00
-6.89768434e-01 -2.28648335e-01 -2.54612148e-01 -1.13388753e+00
8.30551162e-02 -6.66186631e-01 2.24354833e-01 -1.36559829e-01
1.02864914e-02 1.20393336e+00 2.17168048e-01 2.83203840e-01
-3.93607855e-01 2.62462851e-02 1.41977167e+00 1.47450060e-01
-3.26768935e-01 3.69618274e-02 -6.77029192e-01 6.25659108e-01
9.34018910e-01 6.29291832e-02 -5.31185806e-01 -4.87564236e-01
4.67921764e-01 9.86369401e-02 1.87282860e-01 -9.63636935e-01
2.23455369e-01 -4.18966450e-02 6.72315121e-01 -4.64316994e-01
5.81343591e-01 -8.52041006e-01 3.44248891e-01 2.92890012e-01
1.72001794e-01 3.36249284e-02 -5.18787540e-02 6.34011626e-02
-2.36824691e-01 -8.71351957e-02 1.17223096e+00 4.95776683e-02
-9.47825313e-01 2.16995284e-01 -1.75752655e-01 -3.97473961e-01
1.02672195e+00 -6.69889450e-01 -4.85169023e-01 -4.61777449e-01
7.54833743e-02 -5.37715733e-01 6.82936549e-01 1.19194530e-01
1.04816079e+00 -1.28175080e+00 -8.61731991e-02 4.53403234e-01
-1.92953631e-01 -2.37468600e-01 6.61498070e-01 7.46307671e-01
-1.08992100e+00 1.62100255e-01 -3.94848675e-01 -5.08104026e-01
-1.33369708e+00 1.51688486e-01 2.20287628e-02 -1.02117598e-01
-8.25950563e-01 3.67157012e-01 1.41701475e-01 5.12444079e-01
-3.40932906e-01 -1.98961183e-01 -1.58413462e-02 -2.05574900e-01
1.22838211e+00 6.29049122e-01 1.61620125e-01 -6.12119555e-01
1.66520160e-02 1.17747188e+00 -3.00345778e-01 -3.90648931e-01
9.25698876e-01 -5.86221457e-01 6.36684746e-02 1.07151866e-01
1.01004457e+00 4.49464977e-01 -1.25575507e+00 -6.11012094e-02
-5.18331945e-01 -1.01311707e+00 3.80132020e-01 -3.73928338e-01
-1.03922331e+00 9.49858189e-01 8.86819959e-01 1.93537459e-01
1.77692842e+00 -3.49824607e-01 1.36145914e+00 5.68944030e-02
7.08607256e-01 -1.30049205e+00 1.35764122e-01 6.86806738e-02
2.42104143e-01 -9.69430327e-01 1.91884026e-01 -6.88866317e-01
-6.77706897e-01 1.24183798e+00 4.65827614e-01 -6.87881634e-02
2.76509702e-01 5.90555847e-01 2.69815713e-01 1.80089265e-01
-1.86274096e-01 -2.41916582e-01 -1.43722482e-02 4.87819761e-01
3.66457164e-01 -8.61617327e-02 -4.46511447e-01 2.28393257e-01
-1.65181607e-01 4.48998669e-03 6.87115550e-01 6.50618851e-01
-5.03102601e-01 -1.13619709e+00 -3.33188981e-01 4.98894066e-01
-9.06892955e-01 -1.53807670e-01 4.13831204e-01 3.07472616e-01
-1.72912017e-01 8.74618053e-01 3.66309762e-01 -3.58003885e-01
1.93703726e-01 -6.35543346e-01 6.70086026e-01 -6.22446164e-02
-3.88388008e-01 3.12476665e-01 -1.95306882e-01 -5.72341800e-01
-1.95251510e-01 -8.74691382e-02 -1.34064436e+00 -6.38077199e-01
-1.34475291e-01 -4.57242817e-01 9.50388551e-01 3.57878268e-01
3.39812279e-01 4.42987770e-01 7.39342630e-01 -5.26998281e-01
-8.99524614e-02 -3.80459100e-01 -8.27179909e-01 4.19623926e-02
1.18268684e-01 -2.82700032e-01 2.56196946e-01 2.26306349e-01] | [10.882229804992676, -2.449770212173462] |
ec707aa0-93b2-4125-844a-b59fb9a64ab7 | moving-object-detection-by-detecting | 1109.0882 | null | http://arxiv.org/abs/1109.0882v2 | http://arxiv.org/pdf/1109.0882v2.pdf | Moving Object Detection by Detecting Contiguous Outliers in the Low-Rank Representation | Object detection is a fundamental step for automated video analysis in many
vision applications. Object detection in a video is usually performed by object
detectors or background subtraction techniques. Often, an object detector
requires manually labeled examples to train a binary classifier, while
background subtraction needs a training sequence that contains no objects to
build a background model. To automate the analysis, object detection without a
separate training phase becomes a critical task. People have tried to tackle
this task by using motion information. But existing motion-based methods are
usually limited when coping with complex scenarios such as nonrigid motion and
dynamic background. In this paper, we show that above challenges can be
addressed in a unified framework named DEtecting Contiguous Outliers in the
LOw-rank Representation (DECOLOR). This formulation integrates object detection
and background learning into a single process of optimization, which can be
solved by an alternating algorithm efficiently. We explain the relations
between DECOLOR and other sparsity-based methods. Experiments on both simulated
data and real sequences demonstrate that DECOLOR outperforms the
state-of-the-art approaches and it can work effectively on a wide range of
complex scenarios. | ['Can Yang', 'Xiaowei Zhou', 'Weichuan Yu'] | 2011-09-05 | null | null | null | null | ['moving-object-detection'] | ['computer-vision'] | [ 5.60330093e-01 -8.92546177e-01 -6.04828000e-02 -1.68555863e-02
-5.17227232e-01 -3.49082023e-01 4.32515293e-01 -8.40867087e-02
-6.06304049e-01 6.24099433e-01 -2.03207597e-01 -1.75742701e-01
3.52656096e-01 -3.80108535e-01 -4.99538571e-01 -1.14622176e+00
1.47532389e-01 6.03801571e-02 8.12524796e-01 1.87408596e-01
2.09523708e-01 4.85740602e-01 -1.51305389e+00 1.66500092e-01
7.52172112e-01 8.91694725e-01 4.13234562e-01 7.44309545e-01
-2.32377902e-01 8.82548809e-01 -6.30090773e-01 3.37851271e-02
5.22867024e-01 -5.12295306e-01 -7.47276023e-02 7.27948308e-01
6.67541206e-01 -3.79809171e-01 -3.23360205e-01 1.43514669e+00
4.53588545e-01 2.30336741e-01 4.55780029e-01 -1.21642840e+00
5.40253930e-02 -1.35109769e-02 -9.59998012e-01 6.95165098e-01
1.28913000e-01 5.40503338e-02 2.95337856e-01 -1.12351418e+00
4.73697215e-01 1.06459963e+00 5.00461698e-01 4.11076248e-01
-1.22055340e+00 -5.92475057e-01 3.93209249e-01 4.11007464e-01
-1.46050262e+00 -5.57383716e-01 7.88157046e-01 -6.94034457e-01
3.25887680e-01 5.05422235e-01 6.46585345e-01 7.63786733e-01
-8.44551250e-02 9.87147748e-01 9.58018184e-01 -3.84421676e-01
3.71174753e-01 3.62269022e-02 3.56125504e-01 6.13481641e-01
7.61381149e-01 -2.02770740e-01 -2.98380762e-01 -2.02572271e-01
7.70122349e-01 3.58210623e-01 -7.93972194e-01 -5.05626857e-01
-1.31455481e+00 6.36651993e-01 4.60658446e-02 3.30362320e-01
-2.93849409e-01 -3.16991583e-02 3.95175666e-01 1.18606433e-01
2.87006319e-01 -2.24151954e-01 -6.16590977e-02 1.23881422e-01
-1.23914313e+00 1.42696515e-01 8.25769126e-01 8.50074768e-01
4.74599987e-01 3.72888774e-01 -1.51971772e-01 6.11074626e-01
3.14682752e-01 6.45247161e-01 4.35461104e-01 -6.68748558e-01
3.83414805e-01 3.20618689e-01 4.41796780e-01 -1.25841546e+00
-1.09608583e-01 -3.80590022e-01 -1.08000183e+00 4.28201467e-01
6.11410260e-01 -7.97660723e-02 -9.74795640e-01 1.25562477e+00
6.60515904e-01 8.17944825e-01 -7.00794682e-02 1.18071520e+00
7.40422189e-01 8.01880538e-01 -2.07815155e-01 -7.69320488e-01
1.19109321e+00 -1.05434287e+00 -9.08019960e-01 -2.87203282e-01
4.06507015e-01 -9.22105789e-01 4.68095779e-01 7.43186653e-01
-7.41958737e-01 -5.25174320e-01 -1.10224783e+00 4.11740988e-01
8.84639025e-02 4.23607856e-01 3.47417831e-01 8.43752325e-01
-6.10710263e-01 3.37250322e-01 -9.42016900e-01 -3.98087710e-01
2.76101470e-01 3.84909600e-01 -4.08238351e-01 -2.92335182e-01
-5.53330183e-01 6.93980396e-01 5.55373728e-01 4.48147833e-01
-9.81716573e-01 9.58188400e-02 -7.12444425e-01 -2.41208866e-01
8.58326316e-01 -4.82982904e-01 9.23720896e-01 -1.18449700e+00
-1.26063514e+00 6.93348825e-01 -3.70517910e-01 -3.51704538e-01
6.90698922e-01 -5.65887570e-01 -4.18750048e-01 1.84651747e-01
-1.61994155e-02 -1.16181746e-01 1.43326163e+00 -1.21700752e+00
-8.19547534e-01 -1.85609490e-01 -3.58107060e-01 2.27652937e-01
-1.11835800e-01 3.91215265e-01 -9.13913429e-01 -8.35573137e-01
3.98888320e-01 -9.05160248e-01 -4.18747187e-01 3.63177881e-02
-2.66720444e-01 1.87817603e-01 1.28648436e+00 -7.59265721e-01
1.33145308e+00 -2.15976286e+00 1.06760927e-01 7.47866407e-02
1.34853289e-01 7.15432525e-01 2.04750299e-02 3.56057808e-02
5.08945398e-02 -3.88699591e-01 -2.96062857e-01 -2.20571682e-01
-4.73904878e-01 5.47597976e-03 -1.67712063e-01 9.56125140e-01
6.61052093e-02 2.61135191e-01 -8.78985703e-01 -7.39522815e-01
4.15344894e-01 3.34727138e-01 -4.77203071e-01 1.88251868e-01
1.53524829e-02 6.82630539e-01 -5.26111007e-01 8.21852624e-01
9.61997628e-01 -1.41342625e-01 7.07209334e-02 -2.21680090e-01
-2.22919807e-01 -4.51092452e-01 -1.79098427e+00 1.41943038e+00
1.28510609e-01 7.90062487e-01 6.86566472e-01 -1.42292857e+00
6.55680835e-01 1.59951106e-01 4.73637581e-01 -5.38988151e-02
1.47613809e-01 2.74608761e-01 1.33783624e-01 -7.07299769e-01
3.74065816e-01 -3.92385060e-03 4.34164941e-01 -7.60291442e-02
-1.41404897e-01 2.24819392e-01 4.01556939e-01 -1.09007105e-03
1.22632623e+00 7.02793226e-02 4.73238736e-01 3.21764722e-02
9.29734051e-01 -3.07538477e-03 1.10055292e+00 9.74669099e-01
-3.20553750e-01 5.40474892e-01 -7.28391409e-02 -3.28161895e-01
-6.93421662e-01 -8.00234735e-01 -2.25965772e-02 7.24711597e-01
4.09329742e-01 -3.27553481e-01 -5.64506888e-01 -6.45740092e-01
-1.91608086e-01 3.10041100e-01 -2.17837185e-01 1.85982659e-01
-9.01777804e-01 -1.04184175e+00 5.57323545e-03 2.32707620e-01
6.32670879e-01 -7.13762581e-01 -8.45203936e-01 4.13405001e-01
-2.05890104e-01 -1.33934200e+00 -4.12173301e-01 1.81687734e-04
-9.47371960e-01 -1.14672565e+00 -1.03874409e+00 -8.57639372e-01
7.66124427e-01 1.01509786e+00 8.05598736e-01 3.27175289e-01
-6.65926158e-01 3.55390698e-01 -4.00197983e-01 -2.45175436e-01
-2.61738509e-01 -5.39342701e-01 3.19030851e-01 4.99292046e-01
3.26488316e-01 -3.60557228e-01 -6.83611214e-01 5.65203547e-01
-1.05489194e+00 -7.18932897e-02 7.61064827e-01 8.85069668e-01
4.94270265e-01 2.86284029e-01 1.18879989e-01 -5.71717739e-01
9.50814635e-02 -4.14492875e-01 -1.01246333e+00 2.50715017e-01
6.67829439e-02 -3.51243973e-01 3.38212341e-01 -7.45088577e-01
-9.72619057e-01 5.67768037e-01 2.55196422e-01 -7.77920365e-01
-2.53780335e-01 2.14556128e-01 -4.43840444e-01 -3.53346109e-01
4.07666147e-01 6.09479308e-01 -1.49566576e-01 -5.12150466e-01
2.13846415e-01 6.75226748e-01 6.11708641e-01 -3.08782011e-01
1.05136788e+00 8.09889674e-01 9.26839113e-02 -1.43062675e+00
-6.13085628e-01 -1.09598124e+00 -7.25699842e-01 -4.04745638e-01
8.46336365e-01 -9.89575803e-01 -4.20296758e-01 5.40847778e-01
-1.12686074e+00 1.51618486e-02 6.69410303e-02 7.15982139e-01
-3.58853638e-01 8.86933506e-01 -1.95469350e-01 -1.15979195e+00
-2.88985875e-02 -1.15089297e+00 9.01432812e-01 2.22284049e-01
2.89393663e-01 -6.07584417e-01 -4.37018871e-02 1.91059127e-01
1.96001679e-01 2.68246055e-01 2.65347779e-01 -6.14525855e-01
-1.09760404e+00 -4.10770088e-01 -1.78881630e-01 4.81425762e-01
2.99782693e-01 -6.91961646e-02 -6.65772259e-01 -4.99565899e-01
4.95258540e-01 1.19824953e-01 1.07086957e+00 4.45781082e-01
8.93467426e-01 -3.19971830e-01 -4.75750446e-01 6.28306508e-01
1.46104538e+00 3.71856868e-01 5.50061643e-01 3.07051241e-01
7.59104073e-01 3.01401079e-01 8.71059299e-01 5.21070123e-01
-2.38648340e-01 8.33342135e-01 3.65839660e-01 7.09307566e-02
3.37332301e-02 3.66706371e-01 4.96797562e-01 7.88148761e-01
-1.57794565e-01 -1.61614314e-01 -8.47959638e-01 3.59471828e-01
-2.17061853e+00 -1.26261246e+00 -4.56454992e-01 2.38690877e+00
5.41038096e-01 9.29724798e-02 1.25099942e-01 2.68299013e-01
1.03455520e+00 1.10007174e-01 -4.15647209e-01 5.03439665e-01
-3.06529671e-01 -3.13224792e-01 5.23496449e-01 1.84670076e-01
-1.57151639e+00 8.89546275e-01 5.78464460e+00 8.24995577e-01
-1.11506855e+00 1.08724661e-01 3.17292154e-01 -2.55116709e-02
6.08930767e-01 6.42601624e-02 -9.77659881e-01 5.29343724e-01
3.10882479e-01 -1.52532551e-02 1.49723038e-01 8.54391098e-01
3.68229210e-01 -5.56452870e-01 -8.85149300e-01 1.54737663e+00
3.91956091e-01 -1.08114100e+00 -5.96163869e-02 -1.79986596e-01
7.26733625e-01 -2.08600625e-01 -2.69676864e-01 1.40974998e-01
-2.01359898e-01 -6.58125222e-01 5.11395037e-01 5.12128890e-01
2.66893357e-01 -2.29361027e-01 6.74987793e-01 5.51277161e-01
-1.27231598e+00 -1.41756639e-01 -5.79067230e-01 -1.34105813e-02
3.67422342e-01 7.06381798e-01 -4.13504928e-01 4.96233612e-01
6.84774280e-01 7.90694535e-01 -3.72757584e-01 1.81655800e+00
5.42274602e-02 6.92961037e-01 -4.71306175e-01 1.30717695e-01
3.25946398e-02 -6.27523303e-01 9.86563087e-01 1.31804848e+00
4.51152891e-01 4.60755289e-01 6.92106068e-01 4.16232169e-01
2.94872761e-01 3.02428663e-01 -5.22829354e-01 1.15154020e-01
7.18013495e-02 1.30357826e+00 -1.12320399e+00 -5.01486301e-01
-7.49319375e-01 1.24319351e+00 -2.47157186e-01 5.03826559e-01
-8.53512585e-01 -1.58353895e-01 4.77326572e-01 6.74283803e-02
6.96778953e-01 -5.37001014e-01 2.66226441e-01 -1.67711580e+00
1.94570467e-01 -1.12961185e+00 2.96890110e-01 -5.15817106e-01
-9.69496191e-01 3.31682444e-01 -4.12468202e-02 -1.73759580e+00
1.09311357e-01 -7.14215159e-01 -7.54176259e-01 5.88056624e-01
-1.32890093e+00 -6.58315241e-01 -7.90071130e-01 7.75733292e-01
8.20206523e-01 -2.25072503e-01 3.34451884e-01 5.63436568e-01
-9.24825191e-01 -1.13724871e-02 4.18404788e-01 3.00100595e-01
6.89460814e-01 -8.30564797e-01 -1.44071534e-01 1.48568618e+00
3.04423332e-01 3.84826601e-01 9.07785714e-01 -8.38798642e-01
-1.55669200e+00 -1.08669412e+00 1.82762697e-01 -8.63064546e-03
6.11432731e-01 -1.58563688e-01 -9.65820909e-01 4.86704677e-01
-5.37313819e-02 4.48782533e-01 5.89823544e-01 -4.52474654e-01
2.04500675e-01 -3.89968068e-03 -6.77260756e-01 5.42836607e-01
9.59566474e-01 1.05190285e-01 -5.30939758e-01 4.40619797e-01
1.34993926e-01 -3.84139627e-01 -1.94926500e-01 4.38736111e-01
2.32306868e-01 -9.26429570e-01 9.21671212e-01 -6.13864243e-01
-2.85980761e-01 -1.00895083e+00 -3.46594542e-01 -7.91002810e-01
-1.30577117e-01 -8.47049832e-01 -5.02367973e-01 1.05538952e+00
-3.19155753e-01 -3.25342685e-01 8.35440934e-01 2.89025158e-01
2.21107230e-01 -3.18481237e-01 -9.02468622e-01 -1.04020298e+00
-6.12615705e-01 -3.73123586e-01 -1.78789005e-01 8.66489828e-01
-4.89934385e-01 5.69958165e-02 -7.54265249e-01 4.77889061e-01
9.91919160e-01 9.89981666e-02 1.19261193e+00 -1.25606728e+00
-2.47382998e-01 -2.52247959e-01 -9.74684417e-01 -1.24611127e+00
-1.33214831e-01 -4.60573524e-01 2.06733912e-01 -1.44318831e+00
2.87924021e-01 -2.52347320e-01 -2.24701688e-01 6.55261055e-02
-4.64982212e-01 2.72832394e-01 4.10983503e-01 4.79043424e-01
-8.57297599e-01 5.28470576e-01 6.92458451e-01 -2.54556239e-01
-2.24565729e-01 2.63989151e-01 -4.75697443e-02 1.21555531e+00
5.45862138e-01 -5.30939519e-01 -1.24439821e-01 -2.12390393e-01
-2.56690174e-01 7.45901018e-02 4.60265845e-01 -1.36479115e+00
4.00953352e-01 -2.46024683e-01 5.35833597e-01 -8.68208468e-01
3.40662479e-01 -1.06318259e+00 8.97568986e-02 4.97437000e-01
3.54318887e-01 -8.49117115e-02 1.55907959e-01 1.00098372e+00
-4.02476549e-01 -3.45888704e-01 7.76506841e-01 -1.15452707e-01
-1.05522430e+00 2.42934182e-01 -5.84263146e-01 -7.27671385e-02
1.32089424e+00 -5.18414140e-01 -3.27528790e-02 -3.66276175e-01
-1.04108620e+00 7.42935091e-02 4.60067213e-01 1.94009677e-01
7.55300403e-01 -1.20395064e+00 -8.10730040e-01 2.17397228e-01
-1.56524107e-01 -7.17742890e-02 1.40463606e-01 1.26726329e+00
-7.12385058e-01 1.43483296e-01 -2.15469263e-02 -9.96400833e-01
-1.77355385e+00 7.07398832e-01 1.89878508e-01 -4.20534350e-02
-8.16023052e-01 7.25015223e-01 4.01026934e-01 4.61199284e-01
4.94481236e-01 -2.07910731e-01 -1.79528713e-01 2.82931770e-03
8.46375585e-01 5.86105049e-01 -1.71642765e-01 -9.15926278e-01
-3.64977211e-01 7.56245196e-01 -7.36337155e-03 -2.11036205e-02
1.11927378e+00 -1.46161109e-01 -9.85763073e-02 4.69192564e-01
8.68206620e-01 1.99841842e-01 -1.35301769e+00 -3.99908453e-01
1.65578380e-01 -9.62289512e-01 -1.68833490e-02 -1.52278677e-01
-1.14028895e+00 7.89525032e-01 7.86689639e-01 1.08940423e-01
1.37235284e+00 -4.29105610e-01 4.02151316e-01 6.91043198e-01
3.64473820e-01 -9.82874930e-01 2.16407344e-01 3.62581879e-01
7.75552332e-01 -1.42495024e+00 3.38063657e-01 -7.36911476e-01
-4.66954082e-01 1.16722214e+00 5.32307684e-01 -1.63484335e-01
6.64999425e-01 1.77080244e-01 2.05539376e-03 3.05145979e-01
-2.88121790e-01 -4.84046668e-01 3.63315165e-01 3.48659664e-01
1.47831649e-01 -4.25297737e-01 -2.74869889e-01 3.01313937e-01
5.01464725e-01 -9.24815016e-04 4.87324953e-01 1.28581917e+00
-8.02355647e-01 -9.47051466e-01 -9.80912864e-01 3.34859848e-01
-6.19711876e-01 2.17470359e-02 -4.66906242e-02 6.50133908e-01
1.76301554e-01 1.04621816e+00 -2.07381204e-01 -3.49056050e-02
7.95338303e-02 3.51461545e-02 4.29046333e-01 -6.12998545e-01
-1.15774170e-01 5.51362216e-01 -1.34297535e-01 -7.06938624e-01
-8.30862343e-01 -7.90538490e-01 -8.28168929e-01 2.48157948e-01
-8.11304986e-01 -2.79810857e-02 6.12323344e-01 6.99898422e-01
1.02945208e-03 3.86217684e-01 3.72020990e-01 -1.14802718e+00
-5.27854800e-01 -7.88311958e-01 -6.77965820e-01 4.99953359e-01
5.21759987e-01 -7.99057424e-01 -5.40978551e-01 4.32122856e-01] | [8.985371589660645, -0.7897384762763977] |
4f36f9e1-7821-4356-be66-43bb97d49e5f | neural-affine-grayscale-image-denoising | 1709.05672 | null | http://arxiv.org/abs/1709.05672v1 | http://arxiv.org/pdf/1709.05672v1.pdf | Neural Affine Grayscale Image Denoising | We propose a new grayscale image denoiser, dubbed as Neural Affine Image
Denoiser (Neural AIDE), which utilizes neural network in a novel way. Unlike
other neural network based image denoising methods, which typically apply
simple supervised learning to learn a mapping from a noisy patch to a clean
patch, we formulate to train a neural network to learn an \emph{affine} mapping
that gets applied to a noisy pixel, based on its context. Our formulation
enables both supervised training of the network from the labeled training
dataset and adaptive fine-tuning of the network parameters using the given
noisy image subject to denoising. The key tool for devising Neural AIDE is to
devise an estimated loss function of the MSE of the affine mapping, solely
based on the noisy data. As a result, our algorithm can outperform most of the
recent state-of-the-art methods in the standard benchmark datasets. Moreover,
our fine-tuning method can nicely overcome one of the drawbacks of the
patch-level supervised learning methods in image denoising; namely, a
supervised trained model with a mismatched noise variance can be mostly
corrected as long as we have the matched noise variance during the fine-tuning
step. | ['Sungmin Cha', 'Taesup Moon'] | 2017-09-17 | null | null | null | null | ['grayscale-image-denoising'] | ['computer-vision'] | [ 6.51201367e-01 -9.03035104e-02 2.46801645e-01 -6.72627747e-01
-1.05977118e+00 -2.81083107e-01 3.15714121e-01 -2.17827678e-01
-5.39082527e-01 5.85881174e-01 6.61919340e-02 3.79234105e-02
-3.55487108e-01 -9.46601331e-01 -1.01099086e+00 -1.22253871e+00
3.14853400e-01 3.73649597e-03 -1.30749360e-01 -3.36999655e-01
5.26911467e-02 2.48717859e-01 -1.28282678e+00 5.08780703e-02
1.11566722e+00 1.27709687e+00 2.32621342e-01 4.92290527e-01
1.84269231e-02 7.26045966e-01 -6.28574491e-01 -3.05084616e-01
4.92584407e-01 -7.72667050e-01 -3.31209004e-01 8.66962373e-02
6.14573300e-01 -1.27752095e-01 -3.98277909e-01 1.46845496e+00
6.72270954e-01 2.69774437e-01 6.71383321e-01 -6.02945268e-01
-7.55997002e-01 4.66071725e-01 -3.97873878e-01 5.89906201e-02
-2.19420061e-01 1.75254643e-01 5.78337848e-01 -7.44259238e-01
4.90166068e-01 1.04068816e+00 1.12290394e+00 5.73794484e-01
-1.41404366e+00 -4.10541207e-01 -3.98721956e-02 2.99535096e-01
-1.32171309e+00 -5.23353100e-01 1.13886750e+00 -1.62677690e-01
2.90470600e-01 1.25193536e-01 3.02866071e-01 1.10088682e+00
1.69480413e-01 3.30023110e-01 1.42497432e+00 -4.65429038e-01
4.38290805e-01 -2.32421666e-01 -2.06726134e-01 3.66779178e-01
-1.77673683e-01 1.42843843e-01 -2.84141153e-01 1.32649273e-01
8.60775411e-01 -1.62800491e-01 -5.66932738e-01 -2.90280670e-01
-9.24289703e-01 7.78403521e-01 8.08172405e-01 2.65525788e-01
-6.76522672e-01 3.02165002e-01 4.47488785e-01 4.99717683e-01
6.93072677e-01 3.51945341e-01 -2.93856412e-01 1.99126750e-01
-1.09800220e+00 2.65749414e-02 6.00626707e-01 3.57342422e-01
9.06654835e-01 3.77379566e-01 -1.86825290e-01 1.15080345e+00
1.31355524e-01 4.30998623e-01 6.30928516e-01 -1.14694500e+00
3.81914407e-01 2.20699638e-01 -1.84069742e-02 -1.01939535e+00
1.50791183e-02 -8.19521248e-01 -1.61957812e+00 6.37952626e-01
4.25814450e-01 -1.74480900e-01 -1.14591837e+00 1.81548393e+00
3.39799076e-01 3.74397337e-01 1.27311260e-01 8.93874049e-01
6.68652475e-01 7.35370696e-01 -1.97984710e-01 -4.90849853e-01
9.92581844e-01 -9.39291537e-01 -1.04383957e+00 -3.66171688e-01
4.93704565e-02 -7.71129251e-01 9.10295963e-01 6.57664955e-01
-1.15997469e+00 -8.49100292e-01 -1.28659117e+00 -6.54237717e-02
-2.40532517e-01 -2.93950401e-02 7.28238299e-02 6.37940168e-01
-1.04758787e+00 1.01567912e+00 -7.08674133e-01 -2.89182756e-02
5.14972627e-01 3.19897950e-01 -4.79884684e-01 -1.95401907e-01
-1.11239386e+00 8.54103804e-01 3.06933761e-01 7.47427404e-01
-9.80966389e-01 -6.73116922e-01 -7.93817699e-01 -1.04359379e-02
4.11970913e-01 -7.53970802e-01 9.13096726e-01 -1.49162424e+00
-1.70339441e+00 6.74741507e-01 -2.04201210e-02 -5.75045228e-01
7.35471010e-01 -2.13234395e-01 -3.27456564e-01 8.56110528e-02
7.94318616e-02 3.97307485e-01 1.61054349e+00 -1.52444446e+00
-3.02237809e-01 -2.76430339e-01 -3.04251313e-01 6.96488842e-02
-2.70836204e-01 -4.27603006e-01 -5.36899269e-01 -1.18322802e+00
3.12372535e-01 -3.30816060e-01 -4.26433682e-01 1.22187495e-01
-2.32338175e-01 2.11293429e-01 7.37464130e-01 -8.96960378e-01
1.04387283e+00 -2.27536058e+00 3.22435558e-01 3.95205051e-01
2.05773622e-01 3.84916812e-01 -3.10031086e-01 1.13239907e-01
-3.25770020e-01 -1.89788029e-01 -7.22821772e-01 -4.08404887e-01
-1.95726022e-01 3.36080521e-01 -2.21562773e-01 6.39942110e-01
1.65842190e-01 7.29875267e-01 -7.95392573e-01 -2.59509355e-01
3.05524290e-01 8.67459357e-01 -3.00274789e-01 4.76433486e-01
-1.09639624e-02 7.01792240e-01 -2.11021900e-01 1.79130957e-01
9.74645615e-01 9.76236612e-02 -1.00522533e-01 -6.56359136e-01
1.75430745e-01 -3.03041816e-01 -1.34272075e+00 1.68660104e+00
-6.62091494e-01 3.66902202e-01 4.47477102e-01 -1.49285960e+00
1.17052948e+00 2.56283909e-01 3.17019194e-01 -6.81261897e-01
9.90316644e-02 3.35955411e-01 -4.12625074e-01 -5.94174087e-01
-5.99035993e-02 -3.91703486e-01 2.53417730e-01 1.22914158e-01
2.22725227e-01 -1.48889944e-01 -1.45099238e-01 -2.70610511e-01
1.07392013e+00 2.48531759e-01 2.31857508e-01 -2.44872838e-01
7.46884346e-01 -3.66008073e-01 7.40441561e-01 1.03944695e+00
-7.75929242e-02 9.77989674e-01 3.28454167e-01 -5.63915849e-01
-1.18530595e+00 -7.79147625e-01 -2.14407191e-01 8.27938735e-01
1.91741120e-02 9.80780348e-02 -1.26536536e+00 -6.47488654e-01
-4.06997204e-01 2.94607818e-01 -7.76215553e-01 -2.82768756e-01
-8.57427776e-01 -9.58993495e-01 3.99729699e-01 3.23744029e-01
9.25406814e-01 -9.91075337e-01 -1.14717064e-02 2.63696879e-01
-1.42158002e-01 -9.36944485e-01 -5.41771591e-01 6.26237988e-01
-9.41755116e-01 -7.92037368e-01 -8.71218443e-01 -1.03174675e+00
8.65583181e-01 5.41822277e-02 1.06771433e+00 1.13879494e-01
2.12475881e-01 -2.49980316e-02 -1.25496283e-01 5.60515299e-02
-6.08529866e-01 1.02781253e-02 -1.43216625e-01 4.44967687e-01
-1.61850736e-01 -1.02281928e+00 -7.99331903e-01 3.50296021e-01
-1.26947832e+00 -1.52640998e-01 7.58938611e-01 1.29440689e+00
1.04640150e+00 4.89206851e-01 8.09743822e-01 -1.06723344e+00
6.45167053e-01 -3.11786205e-01 -6.16272211e-01 2.57166564e-01
-7.32365251e-01 2.99174994e-01 9.12574828e-01 -4.20725733e-01
-1.22358441e+00 2.17957586e-01 -5.96905589e-01 -4.66939449e-01
-2.08966553e-01 5.92947662e-01 -3.76625448e-01 -3.86216968e-01
9.96528685e-01 3.20967466e-01 9.40648615e-02 -7.57320464e-01
5.59245348e-01 4.82142329e-01 1.15919971e+00 -4.33685869e-01
1.08272970e+00 5.11755943e-01 1.83077291e-01 -4.06874269e-01
-9.92667794e-01 -2.78447121e-01 -6.19555295e-01 -1.48209110e-01
8.22004080e-01 -8.71453583e-01 -4.58371699e-01 8.73918831e-01
-9.94099617e-01 -5.26039422e-01 -5.14005899e-01 1.75906584e-01
-6.20919764e-01 4.08289373e-01 -5.30454278e-01 -5.32080948e-01
-4.47864145e-01 -1.10585070e+00 8.66790771e-01 1.28485337e-01
3.21255058e-01 -1.18976510e+00 6.04829304e-02 2.61279017e-01
7.13894188e-01 4.96778369e-01 7.91370332e-01 -3.13213736e-01
-2.96160936e-01 -2.88956702e-01 -1.24615580e-01 1.23757792e+00
2.14125812e-01 -4.99704510e-01 -1.02697420e+00 -3.52914065e-01
8.30670476e-01 -2.03939006e-01 1.16749454e+00 6.55556023e-01
1.41625559e+00 -4.54663426e-01 2.41067961e-01 1.08039129e+00
1.75436163e+00 -8.76600072e-02 8.97984028e-01 4.67726320e-01
8.50207448e-01 3.04150254e-01 2.24592701e-01 -7.01058134e-02
-1.15757085e-01 5.86375237e-01 6.38897538e-01 -3.08233291e-01
-2.99347967e-01 -8.03714022e-02 2.75719047e-01 9.38318908e-01
-1.42628655e-01 -3.87616344e-02 -4.10505235e-01 3.35786432e-01
-1.81074941e+00 -6.27722144e-01 5.02066538e-02 2.29050469e+00
1.23728645e+00 1.01102829e-01 -3.43680739e-01 3.30259234e-01
8.27653766e-01 3.55351985e-01 -5.76417148e-01 -1.32904425e-01
-2.86421150e-01 5.87650836e-01 5.76009512e-01 7.22408772e-01
-1.30273426e+00 6.45919621e-01 5.88493395e+00 1.27528739e+00
-1.26652539e+00 1.96057335e-01 8.93354118e-01 2.96863288e-01
-5.60302809e-02 -2.82691509e-01 -2.95286268e-01 5.08097649e-01
8.69931340e-01 1.73908278e-01 8.31455112e-01 7.49971628e-01
4.51339781e-01 1.29045218e-01 -9.01747644e-01 9.63973582e-01
1.37810573e-01 -1.26843524e+00 -7.49926716e-02 -2.55387813e-01
1.04847574e+00 -2.16637105e-01 1.45597503e-01 1.04644045e-01
1.57287046e-02 -1.15418267e+00 5.04196346e-01 9.15454984e-01
6.31357729e-01 -7.37988651e-01 1.03495157e+00 3.96381795e-01
-7.20568478e-01 1.12296008e-02 -4.07559365e-01 1.55886412e-01
1.28561392e-01 9.69457746e-01 -1.01068154e-01 5.66477895e-01
7.60572016e-01 7.19987988e-01 -4.48512614e-01 1.05779231e+00
-5.40680766e-01 7.51515985e-01 -1.63386583e-01 6.93185270e-01
8.52579400e-02 -5.47841668e-01 5.29644549e-01 9.04554427e-01
4.06402618e-01 -1.65208742e-01 -6.54716641e-02 7.76514113e-01
-3.78208250e-01 -9.66539904e-02 -2.34950691e-01 4.83478576e-01
1.40120462e-01 1.32985497e+00 -5.58197439e-01 -1.88521579e-01
-6.30281866e-02 1.17604923e+00 1.19315721e-01 6.64900541e-01
-4.57893163e-01 -4.79213625e-01 2.32016340e-01 -7.62150139e-02
6.32809997e-01 1.24777064e-01 -6.04934096e-01 -9.64977086e-01
2.91873515e-01 -1.10745013e+00 8.29591081e-02 -8.55978310e-01
-1.55592442e+00 8.87048244e-01 -3.99012059e-01 -1.25848138e+00
-2.97662318e-01 -4.70562190e-01 -7.91956186e-01 9.08443391e-01
-1.67935371e+00 -1.06451821e+00 -5.52939653e-01 5.95373273e-01
4.68617886e-01 -9.33107734e-02 7.18278050e-01 5.95460355e-01
-4.69656885e-01 5.88962138e-01 5.95771313e-01 1.62036791e-01
9.34943080e-01 -1.35146797e+00 1.84934735e-01 1.08978760e+00
-2.71996856e-02 4.20189768e-01 9.51839626e-01 -3.45876426e-01
-1.19117701e+00 -1.24842441e+00 4.06014919e-01 -1.60849065e-01
5.29301822e-01 -2.47013360e-01 -1.28248894e+00 4.85460520e-01
3.52850556e-01 5.13432264e-01 3.46406456e-03 -2.69504339e-01
-3.13879251e-01 -6.79873407e-01 -1.29638493e+00 3.25092733e-01
7.27038145e-01 -4.61983532e-01 -4.34405684e-01 2.87234426e-01
5.93549848e-01 -5.58646858e-01 -1.12179708e+00 4.36795592e-01
2.54100919e-01 -8.88561428e-01 1.14473152e+00 -4.08552319e-01
6.11380219e-01 -5.12674809e-01 1.98094305e-02 -1.74459195e+00
-3.27903003e-01 -7.39418983e-01 -7.86744282e-02 1.37656713e+00
7.04316571e-02 -5.53135395e-01 5.98844349e-01 5.08191660e-02
-2.42566213e-01 -8.06243896e-01 -1.10618734e+00 -6.16389334e-01
1.78436130e-01 -2.91171759e-01 4.35163379e-01 7.78189123e-01
-9.16552305e-01 2.08198458e-01 -6.76975429e-01 3.62404168e-01
1.09212327e+00 -3.60432923e-01 5.99184573e-01 -8.78502667e-01
-3.60260755e-01 -3.14681113e-01 -3.92852485e-01 -1.06255937e+00
1.38357118e-01 -4.46239948e-01 4.93704885e-01 -1.35379124e+00
-8.90018567e-02 -3.46323520e-01 -3.99087906e-01 3.81952614e-01
-3.81244004e-01 6.48476303e-01 -9.35861394e-02 2.13374645e-01
-3.73079628e-01 6.56434834e-01 1.39343905e+00 -2.59942710e-01
-4.17564176e-02 1.93855956e-01 -7.56454825e-01 8.11519980e-01
6.22309804e-01 -5.93993902e-01 -2.25111783e-01 -4.98094112e-01
2.32642949e-01 -6.89256191e-02 5.29039204e-01 -9.96411741e-01
4.23357010e-01 6.03541471e-02 4.28921640e-01 -1.78987965e-01
1.58563897e-01 -9.62915301e-01 2.68361628e-01 9.91953090e-02
-4.37603116e-01 -3.46099049e-01 -1.46535695e-01 7.33827472e-01
-5.51185250e-01 -5.03539622e-01 1.06574869e+00 -9.66593772e-02
-6.83394790e-01 1.81673497e-01 -1.74262494e-01 -2.69424226e-02
4.13894087e-01 -2.97983110e-01 -2.33972400e-01 -4.88281965e-01
-8.63551736e-01 -1.37666330e-01 2.40389019e-01 7.56287947e-02
4.47421998e-01 -1.49410784e+00 -8.22191536e-01 2.61646062e-01
-1.95054233e-01 2.80707479e-01 4.84098077e-01 8.13554764e-01
-4.50089753e-01 -3.49466145e-01 -1.86509669e-01 -5.34163237e-01
-8.44924808e-01 6.22799337e-01 7.24703789e-01 -2.74165928e-01
-6.44978702e-01 6.72061384e-01 1.27258658e-01 -4.94003892e-01
3.63518119e-01 -1.90717518e-01 -5.98288029e-02 -2.15606838e-01
4.93593335e-01 2.67891914e-01 4.27358478e-01 -6.55728340e-01
1.15634307e-01 7.34881997e-01 2.04361051e-01 4.77179214e-02
1.70692325e+00 -2.59914607e-01 -4.78275925e-01 1.50201246e-01
1.43130195e+00 -1.19579993e-01 -1.65687382e+00 -5.29079437e-01
-3.33408266e-01 -3.72376561e-01 5.93319952e-01 -7.96034455e-01
-1.56529522e+00 5.10728300e-01 1.12167108e+00 1.92329094e-01
1.70123553e+00 -4.01351482e-01 7.34207392e-01 4.60663140e-01
-1.83114737e-01 -1.16152894e+00 1.71712115e-02 2.17248902e-01
1.02704394e+00 -1.29549479e+00 -1.31417110e-04 -2.89829880e-01
-2.23514676e-01 1.14744878e+00 2.19148085e-01 -5.61975777e-01
7.25965798e-01 2.55921453e-01 4.78528112e-01 -6.49628416e-02
-1.98722646e-01 9.07487795e-02 3.74147862e-01 7.08655000e-01
-2.28278786e-02 -2.97067910e-01 -2.59371907e-01 5.36780775e-01
-8.43448564e-02 8.55046064e-02 2.26828679e-01 5.46000779e-01
-3.45678359e-01 -1.15102780e+00 -6.82131946e-01 2.89714009e-01
-6.10457957e-01 -8.32398459e-02 1.56388521e-01 3.84038866e-01
3.83991718e-01 7.13165998e-01 -2.37268612e-01 -8.82470310e-02
4.85683709e-01 -2.31349647e-01 3.13341618e-01 -2.75074214e-01
-7.11944520e-01 2.27609947e-01 -3.02207977e-01 -4.60510314e-01
-6.73130274e-01 -2.97727853e-01 -6.96082592e-01 -6.50074184e-02
-2.70139545e-01 4.43315729e-02 7.20625520e-01 1.19902742e+00
-6.88105775e-03 7.36274421e-01 8.25806260e-01 -1.17727256e+00
-8.41261625e-01 -9.65779662e-01 -6.08456194e-01 6.48520947e-01
5.98529458e-01 -3.36572558e-01 -5.38220882e-01 4.32597786e-01] | [11.550800323486328, -2.3362417221069336] |
b833dce8-4cbe-4a81-9beb-df8fb588ff2c | improving-generalizability-in-implicitly | null | null | https://openreview.net/forum?id=AfBuiCDtF_0 | https://openreview.net/pdf?id=AfBuiCDtF_0 | Improving Generalizability in Implicitly Abusive Language Detection with Concept Activation Vectors | Robustness of machine learning models on ever-changing real-world data is critical, especially for applications affecting human well-being such as content moderation. New kinds of abusive language continually emerge in online discussions in response to current events (e.g., COVID-19), and the deployed abuse detection systems should be updated regularly to remain accurate. In this paper, we show that general abusive language classifiers tend to be fairly reliable in detecting out-of-domain explicitly abusive utterances but fail to detect new types of more subtle, implicit abuse. Next, we propose an interpretability technique, based on the Testing Concept Activation Vector (TCAV) method from computer vision, to quantify the sensitivity of a trained model to the human-defined concepts of explicit and implicit abusive language, and use that to explain the generalizability of the model on new data, in this case, COVID-related anti-Asian hate speech. Extending this technique, we introduce a novel metric, Degree of Explicitness, for a single instance and show that the new metric is beneficial in suggesting out-of-domain unlabeled examples to effectively enrich the training data with informative, implicitly abusive texts. | ['Anonymous'] | 2021-11-16 | null | https://openreview.net/forum?id=eRRLIv7DrOX | https://openreview.net/pdf?id=eRRLIv7DrOX | acl-arr-september-2021-9 | ['abusive-language', 'abuse-detection'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.14016545e-02 2.48382818e-02 -4.22458351e-01 -4.73256648e-01
-4.36682254e-01 -7.49853015e-01 5.37315309e-01 4.39218074e-01
-4.10234839e-01 9.70649779e-01 3.32223296e-01 -4.51348513e-01
8.10517184e-03 -3.62596810e-01 -3.84075791e-01 -3.62331063e-01
-4.49509323e-02 4.16867733e-01 -8.44686478e-02 -4.51781660e-01
2.16405973e-01 6.13539517e-01 -8.96780372e-01 2.99617320e-01
1.06560123e+00 3.58949393e-01 -5.44281423e-01 5.24312615e-01
-1.47055313e-01 9.00123000e-01 -1.11087501e+00 -8.78570080e-01
-2.45143339e-01 -5.41941106e-01 -1.05590451e+00 1.48936629e-01
2.39101127e-01 -6.37633026e-01 -9.07711461e-02 1.20450890e+00
1.21918462e-01 3.97501923e-02 6.71451807e-01 -1.33127284e+00
-7.46166050e-01 9.16057706e-01 -7.17784166e-01 8.62900436e-01
5.54461777e-01 1.63858026e-01 8.29456270e-01 -7.26571441e-01
7.73674071e-01 1.52942991e+00 5.99717557e-01 8.44240665e-01
-1.03459668e+00 -1.12252641e+00 2.73121327e-01 2.08118558e-01
-9.14177716e-01 -4.14174676e-01 1.09326398e+00 -6.48577154e-01
5.11948586e-01 5.73218703e-01 5.04222929e-01 1.99021769e+00
-1.50584593e-01 6.58910394e-01 1.17566168e+00 -2.89950848e-01
2.70829171e-01 3.12091947e-01 8.15055430e-01 4.65737820e-01
5.94209135e-01 -3.34216028e-01 -3.78483742e-01 -7.77874529e-01
-4.90442626e-02 5.97204976e-02 -2.15271011e-01 2.56376415e-01
-3.57757747e-01 1.42422116e+00 2.29155242e-01 8.75143647e-01
-2.42787689e-01 -3.26603472e-01 7.88788736e-01 1.62602663e-01
1.04823327e+00 7.95096993e-01 -3.21836293e-01 -4.39341486e-01
-6.91214085e-01 2.77895421e-01 8.61999810e-01 3.07939261e-01
5.23245096e-01 -1.97442174e-02 1.45378008e-01 1.15315187e+00
4.98720296e-02 3.87656331e-01 7.09390700e-01 -6.36995792e-01
3.17965806e-01 5.97506642e-01 -2.29438052e-01 -1.09687793e+00
-4.60194945e-01 -3.95058364e-01 -6.64034545e-01 -4.44034547e-01
2.99979627e-01 -2.58042991e-01 -7.29094267e-01 1.93186915e+00
5.46935380e-01 1.34197280e-01 -3.04743737e-01 7.35755324e-01
3.46928239e-01 5.04191995e-01 4.58899438e-01 -6.90481007e-01
9.66185868e-01 -4.65290725e-01 -1.14318681e+00 -3.97756130e-01
9.62104559e-01 -2.90255070e-01 1.18526816e+00 3.64330769e-01
-6.46116376e-01 3.11038904e-02 -1.09936237e+00 1.13430247e-01
-6.70061707e-01 -6.89572394e-01 4.10753459e-01 1.05924702e+00
-4.50151294e-01 3.75315696e-01 -3.91619354e-01 -4.60671008e-01
4.13753122e-01 -1.32759929e-01 -3.84982586e-01 -1.88961804e-01
-1.49088621e+00 1.26962316e+00 3.23067963e-01 -1.15535825e-01
-8.42549622e-01 -5.35036445e-01 -8.71409237e-01 -2.65670538e-01
4.43727553e-01 -7.23488554e-02 1.25760615e+00 -1.34233952e+00
-1.10693026e+00 1.06007564e+00 4.81238253e-02 -3.91930759e-01
3.68756115e-01 -5.79138279e-01 -7.26211190e-01 1.68779925e-01
1.80731699e-01 -1.09451510e-01 1.09937179e+00 -1.33708453e+00
-7.78232515e-02 -7.03450799e-01 1.26991123e-01 -1.53609812e-01
-9.03210580e-01 5.73558331e-01 4.01759058e-01 -7.27184057e-01
-2.93258488e-01 -8.01313579e-01 1.28111690e-01 -2.54740655e-01
-6.60880983e-01 -2.63976455e-01 1.31576729e+00 -1.11318004e+00
1.65870202e+00 -2.17312598e+00 1.76920488e-01 4.00166988e-01
5.24231315e-01 6.40870988e-01 2.33356863e-01 2.14602262e-01
-6.44796714e-02 5.85696638e-01 -6.95396245e-01 -2.79088855e-01
-3.66153181e-01 6.33927643e-01 -4.14606661e-01 5.96691430e-01
3.81992877e-01 3.14854205e-01 -1.17679000e+00 -4.79856372e-01
-1.37374818e-01 3.30283828e-02 -3.70055169e-01 3.00751179e-01
-6.36562556e-02 5.27304113e-01 -4.40963119e-01 5.73418677e-01
5.47753811e-01 5.04831187e-02 2.85160899e-01 3.01550090e-01
2.29604393e-01 2.99574286e-01 -4.68950063e-01 9.73008692e-01
-2.78606206e-01 7.18551636e-01 2.48623818e-01 -1.06025219e+00
8.70369911e-01 5.79521954e-02 4.98043895e-02 -4.47239131e-01
3.45974296e-01 2.41184607e-01 3.65073085e-01 -7.76609600e-01
3.05127770e-01 -4.02795523e-01 -2.03356087e-01 5.96342981e-01
-8.26443825e-03 1.59576580e-01 7.71906227e-02 4.28453863e-01
1.16616404e+00 -6.00170493e-01 7.90207148e-01 -8.09223577e-02
5.49381614e-01 -2.78031975e-01 3.71658564e-01 6.22191548e-01
-6.83038235e-01 1.56496942e-01 6.74503863e-01 -2.89540976e-01
-1.02877903e+00 -7.39079893e-01 -3.64763826e-01 1.47049880e+00
-2.27727935e-01 -2.91330159e-01 -9.40051854e-01 -1.15136468e+00
2.14582849e-02 1.21449518e+00 -7.61350095e-01 -4.48458850e-01
-5.81585109e-01 -8.48587692e-01 8.42507780e-01 2.17956185e-01
2.79243082e-01 -1.00795412e+00 -4.36413348e-01 2.41295427e-01
-2.88286448e-01 -1.20164192e+00 -1.59417674e-01 3.35932486e-02
-6.43901587e-01 -1.18986309e+00 -3.18606347e-01 -1.05845854e-01
3.28671008e-01 2.04992648e-02 7.07724929e-01 6.29765570e-01
-9.91898626e-02 1.60355911e-01 -7.88562834e-01 -3.62028807e-01
-1.00849807e+00 8.33148416e-03 3.29170585e-01 2.68969983e-01
5.60006618e-01 -5.07433891e-01 5.17828912e-02 -2.78685130e-02
-1.16278207e+00 -3.81715328e-01 8.74088034e-02 8.95049691e-01
-2.98862606e-01 -5.94329059e-01 7.26941824e-01 -1.49166572e+00
1.10520720e+00 -1.19377530e+00 2.67421544e-01 -5.39243110e-02
-4.83386278e-01 -3.54308963e-01 5.70451200e-01 -9.78977621e-01
-9.61177468e-01 -6.17101848e-01 -3.41450512e-01 -4.04983133e-01
-2.19078943e-01 7.17944443e-01 4.60840464e-02 3.62498760e-01
1.10800588e+00 -9.81603563e-02 -8.05327669e-02 -5.61828732e-01
2.91237622e-01 1.10933137e+00 2.66366631e-01 -5.64991951e-01
8.63111138e-01 2.45878890e-01 -5.17245770e-01 -1.18593872e+00
-1.29254901e+00 -6.66619420e-01 -6.35207474e-01 -2.39436537e-01
6.54593706e-01 -3.07983190e-01 -1.57596841e-01 5.75325727e-01
-1.36232674e+00 -1.24093950e-01 2.23958164e-01 1.07779548e-01
-1.95606127e-02 6.65020645e-01 -8.96989167e-01 -1.35891426e+00
-4.44567442e-01 -6.82266831e-01 7.76377678e-01 -2.11939178e-02
-8.19420755e-01 -1.05704606e+00 4.39532906e-01 4.86246616e-01
2.74336845e-01 4.74947780e-01 1.25245643e+00 -1.63403904e+00
6.10772729e-01 -2.96028823e-01 7.32858777e-02 6.83594584e-01
2.77180165e-01 2.74395525e-01 -1.25407386e+00 -2.51722157e-01
3.36347163e-01 -5.74296176e-01 4.74155217e-01 -2.24539921e-01
1.31003809e+00 -9.64595914e-01 -2.08801374e-01 -6.38772827e-03
8.26602876e-01 2.16314584e-01 4.45015460e-01 1.67298004e-01
6.01208091e-01 7.01589644e-01 3.83209825e-01 6.27487481e-01
-2.42095187e-01 5.22192299e-01 2.61197358e-01 5.30867353e-02
3.26595008e-01 -2.32301161e-01 5.80378830e-01 7.75923669e-01
3.43677461e-01 -1.87389910e-01 -9.84910905e-01 4.90600944e-01
-1.53907716e+00 -1.02926481e+00 -1.51455954e-01 1.82008874e+00
1.27536869e+00 2.93250442e-01 3.64610076e-01 4.02381957e-01
8.60962272e-01 2.28607714e-01 -5.53066194e-01 -1.01037908e+00
-1.41056506e-02 3.45807463e-01 -1.41591251e-01 7.22507000e-01
-1.01787233e+00 7.53031790e-01 6.04963160e+00 6.99548542e-01
-1.06692171e+00 7.04072893e-01 9.36720550e-01 3.62293907e-02
-2.72612333e-01 -3.09731662e-01 -3.64547938e-01 7.80097604e-01
1.17727160e+00 -1.80331424e-01 1.84361741e-01 1.07222450e+00
2.93915216e-02 5.40898181e-02 -1.07824349e+00 7.30151176e-01
6.38943374e-01 -8.82916510e-01 2.09403038e-03 3.56190801e-02
3.79799515e-01 -3.32501888e-01 4.08123434e-02 5.43493748e-01
1.63779452e-01 -9.77187455e-01 3.91727000e-01 -4.97385934e-02
5.70076823e-01 -6.43461585e-01 7.73499250e-01 7.18100429e-01
-1.40470400e-01 -3.27430129e-01 -1.34478748e-01 -2.39804611e-01
-6.89986199e-02 6.23172462e-01 -1.03672159e+00 -1.19922832e-01
6.70014679e-01 5.06651282e-01 -7.27337778e-01 4.37983781e-01
-2.85821974e-01 1.09413207e+00 -2.98613608e-01 -2.07602322e-01
2.76632011e-01 6.56489953e-02 7.70286441e-01 1.49353600e+00
-9.83517915e-02 3.71259511e-01 -1.18421808e-01 7.84337342e-01
-2.32058316e-01 1.05297886e-01 -1.12683809e+00 -1.93290293e-01
5.56556582e-01 1.21176445e+00 -8.35255533e-02 -4.52397346e-01
-1.02305532e-01 9.75914180e-01 6.83471203e-01 2.05918923e-01
-7.20449507e-01 1.52146623e-01 7.53856719e-01 2.78310925e-01
-3.96410882e-01 -8.68842602e-02 -1.34724051e-01 -1.02258039e+00
-1.12156183e-01 -1.35587537e+00 7.03611851e-01 -5.83627284e-01
-1.82511520e+00 3.80694479e-01 2.84273624e-01 -5.91559231e-01
-5.02007484e-01 -6.24333203e-01 -6.57304645e-01 4.98905301e-01
-1.02182591e+00 -9.35986340e-01 -1.02828383e-01 4.86787826e-01
6.29813194e-01 5.13466038e-02 7.80268610e-01 2.58702576e-01
-8.58087480e-01 7.67785966e-01 -2.31176898e-01 5.52321017e-01
7.40281403e-01 -8.83591592e-01 2.18379255e-02 9.10715044e-01
-5.88613972e-02 7.89474189e-01 1.04941738e+00 -9.44923460e-01
-8.10153782e-01 -8.89642119e-01 6.72708333e-01 -7.73551762e-01
1.18013060e+00 -4.34775978e-01 -1.47163820e+00 9.71208155e-01
1.80402566e-02 -2.93105721e-01 1.03235495e+00 5.21352589e-01
-8.59194517e-01 2.85760611e-01 -1.49882424e+00 6.92293823e-01
1.06969166e+00 -5.75576544e-01 -9.64496791e-01 5.53754508e-01
8.46137762e-01 -1.72826096e-01 -3.48828346e-01 2.51462281e-01
2.28584841e-01 -7.46779919e-01 5.66041231e-01 -1.49891913e+00
6.16338313e-01 4.69464064e-01 -3.21218446e-02 -1.30511248e+00
-2.38112673e-01 -5.57602823e-01 -5.43948948e-01 1.25212765e+00
3.90782833e-01 -6.74027741e-01 3.04744244e-01 6.24674380e-01
2.96291392e-02 -5.09099066e-01 -1.26863635e+00 -5.67679584e-01
3.18750203e-01 -4.17906135e-01 1.61153793e-01 1.72081923e+00
4.84161675e-01 5.53577006e-01 -7.38697529e-01 4.22676541e-02
2.64385283e-01 -6.01614237e-01 7.46163607e-01 -1.13168716e+00
-2.01838851e-01 -3.47910583e-01 -4.08020258e-01 -3.87840837e-01
5.71311772e-01 -6.42755926e-01 -2.41704226e-01 -6.30682170e-01
5.38397610e-01 -2.35036358e-01 3.09377741e-02 3.87720913e-01
-5.92581272e-01 2.11388871e-01 1.44029289e-01 8.58995244e-02
-4.08870578e-01 4.28733766e-01 7.82981038e-01 -3.27914774e-01
-8.26440901e-02 -1.46401837e-01 -7.35933661e-01 9.99803305e-01
7.14359343e-01 -6.48581862e-01 -1.87416807e-01 3.54282595e-02
1.49301171e-01 -1.78708985e-01 2.64799207e-01 -5.49176574e-01
-5.25977433e-01 -3.47906440e-01 2.01246843e-01 -5.75084761e-02
4.43466425e-01 -7.14899004e-01 -3.79299372e-01 6.52443409e-01
-5.14156222e-01 1.73452213e-01 1.29638746e-01 5.60003936e-01
4.03237604e-02 -7.93093622e-01 8.77787650e-01 9.81837362e-02
-3.82020772e-01 1.04163744e-01 -5.78795910e-01 4.03158784e-01
1.07874358e+00 1.28036430e-02 -7.22914398e-01 -6.48964226e-01
-6.56857073e-01 -1.64183214e-01 1.75906599e-01 5.05908728e-01
6.11846089e-01 -9.98857260e-01 -9.73719060e-01 -1.42874733e-01
2.60520011e-01 -6.91101372e-01 9.23984200e-02 7.74606586e-01
-2.43122652e-01 -1.50587156e-01 2.24745292e-02 -3.82169008e-01
-1.39200664e+00 7.94784188e-01 3.47368151e-01 -3.44899416e-01
-3.18738312e-01 6.15136385e-01 1.48545057e-01 -2.17274129e-01
-4.25997414e-02 2.73511708e-02 -2.60975122e-01 1.04473658e-01
7.54642665e-01 3.04455221e-01 -6.91671446e-02 -7.95638263e-01
-3.71673048e-01 -2.45305523e-01 -4.99149173e-01 -6.08693734e-02
1.03559709e+00 9.46703851e-02 -2.87433654e-01 8.73339832e-01
1.27197874e+00 2.19787478e-01 -4.65811968e-01 -2.48462602e-01
3.61516714e-01 -6.11728191e-01 -4.25681382e-01 -8.54418576e-01
-5.62604368e-01 9.46605980e-01 4.09656823e-01 5.83793700e-01
7.31081605e-01 1.86977059e-01 7.46557951e-01 3.74754101e-01
-2.99996659e-02 -1.10007298e+00 5.32294393e-01 5.41684985e-01
1.06746256e+00 -1.14292741e+00 6.15096651e-02 -4.26756948e-01
-7.78964221e-01 9.01688874e-01 9.16654825e-01 1.53376684e-01
5.58044791e-01 5.10196500e-02 -5.31347608e-03 -2.94392228e-01
-3.85526776e-01 2.82401085e-01 -5.43720461e-03 7.39276588e-01
3.75453800e-01 2.58461922e-01 -6.46546006e-01 5.69061041e-01
-1.95559472e-01 -7.02209294e-01 7.37076700e-01 6.74113035e-01
-4.56534982e-01 -6.81218147e-01 -5.12885869e-01 5.95117986e-01
-7.49642015e-01 4.38535499e-04 -1.17562699e+00 9.20648634e-01
-3.91653329e-02 9.35458958e-01 3.85796241e-02 -3.67516547e-01
6.14434406e-02 5.25745869e-01 1.29783198e-01 -8.23704362e-01
-7.76205659e-01 -3.31666559e-01 6.26730323e-01 -1.80280596e-01
-3.31605077e-01 -5.79152942e-01 -8.57853174e-01 -6.21702254e-01
-3.27508420e-01 1.57214589e-02 4.81611878e-01 1.44983947e+00
-1.65351611e-02 -1.15852907e-01 6.96891606e-01 -3.37635607e-01
-8.75762820e-01 -1.54021358e+00 -3.77475679e-01 1.20630789e+00
5.83390474e-01 -7.71720648e-01 -6.93999350e-01 -3.02499056e-01] | [8.716962814331055, 10.48544979095459] |
17f17a91-9b4d-4aa9-8ccc-51937c45a4bc | self-supervised-co-training-for-video | 2010.09709 | null | https://arxiv.org/abs/2010.09709v2 | https://arxiv.org/pdf/2010.09709v2.pdf | Self-supervised Co-training for Video Representation Learning | The objective of this paper is visual-only self-supervised video representation learning. We make the following contributions: (i) we investigate the benefit of adding semantic-class positives to instance-based Info Noise Contrastive Estimation (InfoNCE) training, showing that this form of supervised contrastive learning leads to a clear improvement in performance; (ii) we propose a novel self-supervised co-training scheme to improve the popular infoNCE loss, exploiting the complementary information from different views, RGB streams and optical flow, of the same data source by using one view to obtain positive class samples for the other; (iii) we thoroughly evaluate the quality of the learnt representation on two different downstream tasks: action recognition and video retrieval. In both cases, the proposed approach demonstrates state-of-the-art or comparable performance with other self-supervised approaches, whilst being significantly more efficient to train, i.e. requiring far less training data to achieve similar performance. | ['Andrew Zisserman', 'Weidi Xie', 'Tengda Han'] | 2020-10-19 | null | http://proceedings.neurips.cc/paper/2020/hash/3def184ad8f4755ff269862ea77393dd-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/3def184ad8f4755ff269862ea77393dd-Paper.pdf | neurips-2020-12 | ['self-supervised-action-recognition'] | ['computer-vision'] | [ 6.63439929e-01 9.46843401e-02 -3.05542380e-01 -3.47710937e-01
-1.02901196e+00 -4.26005095e-01 8.54305029e-01 4.51030433e-02
-4.58865702e-01 8.21572423e-01 3.01307201e-01 2.46788993e-01
-8.54789242e-02 -4.72469240e-01 -9.40108955e-01 -7.29736030e-01
-6.39698282e-02 2.27624685e-01 2.64805138e-01 6.65888637e-02
4.04149473e-01 4.75924432e-01 -2.17740464e+00 5.62549829e-01
6.81314170e-01 1.42676246e+00 -3.61070745e-02 7.22319841e-01
6.66853413e-02 1.39035666e+00 -4.75404084e-01 -2.20342293e-01
3.01566631e-01 -7.20224440e-01 -9.58146751e-01 6.69281006e-01
8.70721042e-01 -2.89511681e-01 -2.87035316e-01 8.98258805e-01
3.20500761e-01 4.90618907e-02 6.70609355e-01 -1.31542408e+00
-1.36561960e-01 -1.39098898e-01 -3.67056906e-01 4.13841277e-01
5.02783358e-01 4.07210141e-01 9.46955502e-01 -8.62429857e-01
8.09927523e-01 1.09740925e+00 3.35054070e-01 4.55826104e-01
-1.21037221e+00 -5.31365931e-01 1.77063838e-01 4.30600524e-01
-9.95551586e-01 -9.43702996e-01 9.32348251e-01 -3.98714721e-01
8.68929803e-01 9.12271887e-02 7.34896839e-01 1.14041924e+00
-2.92680800e-01 1.23774290e+00 1.55545866e+00 -4.50779319e-01
5.28255463e-01 4.85845566e-01 -8.65024030e-02 8.86315644e-01
6.91569149e-02 3.50730926e-01 -9.50619519e-01 7.07380027e-02
5.70881426e-01 -1.30854905e-01 -3.57956409e-01 -7.30867028e-01
-1.16034532e+00 5.60967207e-01 3.95712137e-01 4.05839413e-01
-2.62549490e-01 2.60488868e-01 5.27026176e-01 4.69216526e-01
6.91831946e-01 2.75015473e-01 -4.10069644e-01 -3.83654743e-01
-1.16074371e+00 -1.31513104e-01 7.81293511e-01 6.95807755e-01
9.79648888e-01 3.49611223e-01 -2.56491452e-02 5.05371690e-01
2.30609298e-01 7.46099472e-01 5.16857505e-01 -1.25192118e+00
6.13796890e-01 5.01178205e-01 1.79292988e-02 -9.24325883e-01
-1.43595442e-01 -5.38717210e-01 -6.23464346e-01 5.66909075e-01
3.52149338e-01 5.08849733e-02 -8.49901438e-01 1.54095209e+00
2.29479261e-02 3.97915334e-01 4.08439010e-01 6.87708378e-01
7.71189570e-01 6.31220579e-01 -7.24026263e-02 -4.33253706e-01
8.13690782e-01 -1.05144465e+00 -6.59413695e-01 -2.15078205e-01
5.06260931e-01 -5.05531788e-01 6.60428286e-01 5.97877979e-01
-1.17183924e+00 -7.52398968e-01 -1.18066001e+00 2.75915444e-01
-3.34197074e-01 1.47244930e-01 6.72126710e-01 6.73667192e-01
-1.03931081e+00 9.57975566e-01 -8.92943621e-01 -2.33860880e-01
8.25356662e-01 9.99513790e-02 -7.04484880e-01 -3.21944594e-01
-7.58895040e-01 6.86489999e-01 2.74770588e-01 -2.81289607e-01
-1.29509783e+00 -6.93556428e-01 -1.02891195e+00 -2.99512707e-02
5.51257372e-01 -4.64747429e-01 6.24968171e-01 -1.72221160e+00
-1.49795580e+00 9.57559109e-01 -2.33274430e-01 -4.77524430e-01
7.85813928e-01 -4.28413004e-01 -2.23520964e-01 7.46647716e-01
2.12861085e-03 8.88336778e-01 1.29401708e+00 -1.59557498e+00
-5.58905482e-01 -5.02340496e-01 -5.01449220e-02 3.32555622e-01
-4.02203202e-01 -1.64662167e-01 -3.31987560e-01 -6.96127653e-01
-1.55318640e-02 -7.11944878e-01 3.17508802e-02 4.03961360e-01
-1.49871245e-01 1.12985738e-01 7.88925886e-01 -4.69893008e-01
6.52107596e-01 -2.12368965e+00 1.89024433e-01 4.15945016e-02
1.82691161e-02 6.64867401e-01 -1.84101865e-01 3.10392916e-01
-3.62366527e-01 -1.91480175e-01 -5.26526511e-01 -4.79541987e-01
-5.12089372e-01 3.97869200e-01 -1.59282833e-01 6.25710666e-01
7.47220337e-01 9.40375865e-01 -1.24322402e+00 -5.63066065e-01
7.44148552e-01 5.06018341e-01 -3.60719621e-01 4.46309060e-01
-4.97639589e-02 6.22132301e-01 -1.18009731e-01 6.39003992e-01
5.14765024e-01 -3.55848014e-01 -3.00166267e-03 -4.20883149e-01
3.35644856e-02 8.49632323e-02 -1.39666057e+00 1.77896595e+00
-3.44851881e-01 8.55772674e-01 -2.18308255e-01 -1.49852550e+00
5.46613097e-01 2.95567811e-01 7.27055907e-01 -8.28395844e-01
-3.56597379e-02 1.38068706e-01 -4.65884298e-01 -6.67034030e-01
-5.92921451e-02 -6.19895272e-02 5.19212961e-01 3.25853080e-01
7.02991486e-01 1.20778745e-02 2.08505318e-01 1.67382270e-01
1.11234903e+00 4.79646504e-01 4.36046153e-01 -7.07939863e-02
8.15270364e-01 -1.60218343e-01 3.81144136e-01 7.40488410e-01
-3.51754755e-01 6.91369057e-01 4.05086815e-01 -2.02484891e-01
-8.52013588e-01 -9.79296207e-01 3.87038030e-02 7.18168914e-01
9.20148939e-02 -2.60260999e-01 -5.59699059e-01 -1.03362191e+00
-5.97568788e-02 3.50392520e-01 -7.16962934e-01 1.63152050e-02
-5.87622464e-01 -5.88846505e-01 2.61936873e-01 7.15445101e-01
8.16078007e-01 -9.36927915e-01 -8.33840489e-01 -1.55293718e-01
-1.18834548e-01 -1.23354912e+00 1.54319584e-01 4.74704534e-01
-1.29134941e+00 -1.25813282e+00 -9.17623997e-01 -3.09421301e-01
6.95213497e-01 3.95692766e-01 1.17305243e+00 1.69955892e-03
-2.85055012e-01 1.09151030e+00 -4.69925314e-01 2.29201205e-02
-3.11320961e-01 -5.01786053e-01 -1.76195160e-01 4.35473293e-01
-4.23693918e-02 -4.97830421e-01 -5.90338886e-01 2.49774992e-01
-1.01535869e+00 -3.33923966e-01 6.06571853e-01 9.55957770e-01
5.00647783e-01 -1.21738628e-01 2.65635490e-01 -8.64236355e-01
-6.90071657e-02 -3.17816108e-01 -4.27477479e-01 1.57627538e-01
-8.37047994e-01 2.56278038e-01 4.85455006e-01 -7.82147646e-02
-1.12105799e+00 4.43025917e-01 4.72700521e-02 -5.71994245e-01
-4.81922954e-01 3.24705765e-02 -8.14418718e-02 -3.52521867e-01
6.02649927e-01 3.59274626e-01 3.82405907e-01 -3.20479721e-01
4.88143295e-01 3.84049892e-01 5.07485449e-01 -8.09917226e-02
7.58291185e-01 1.07025683e+00 2.19429478e-01 -8.88292015e-01
-8.86111200e-01 -9.37376201e-01 -8.28417420e-01 -5.50055623e-01
7.37865329e-01 -1.08481419e+00 -4.50963706e-01 5.66028476e-01
-8.03231359e-01 -1.48592070e-01 -6.44781768e-01 5.38433075e-01
-1.00709343e+00 6.94740117e-01 -2.53080159e-01 -1.03387296e+00
-8.69362243e-03 -9.19131696e-01 1.24712133e+00 8.02421849e-03
2.13703245e-01 -1.04863179e+00 4.46432419e-02 5.84202886e-01
2.32750982e-01 2.24164948e-01 3.04204285e-01 -6.41581059e-01
-7.79374182e-01 -1.23903148e-01 -4.12404835e-01 1.00360119e+00
1.77908376e-01 -2.26680428e-01 -1.50133491e+00 -2.69047141e-01
9.04647037e-02 -7.70232201e-01 1.32180715e+00 1.13793932e-01
9.51485097e-01 -1.84469253e-01 -5.18519208e-02 5.84998727e-01
1.66984963e+00 -6.29731566e-02 9.36103940e-01 3.16054672e-01
6.20408714e-01 8.32041264e-01 6.99089766e-01 4.20877963e-01
6.06273077e-02 6.79856896e-01 5.99471986e-01 -3.69139552e-01
-5.47986507e-01 -5.29179052e-02 6.52377427e-01 4.55263406e-01
-2.23402783e-01 1.43693062e-02 -4.19774950e-01 5.62777102e-01
-2.02004337e+00 -1.18471682e+00 1.20478511e-01 2.38634515e+00
6.32512331e-01 4.75877002e-02 2.03651547e-01 5.64082563e-01
3.73719037e-01 4.95399296e-01 -4.70660478e-01 2.57286251e-01
-3.51413101e-01 2.51461267e-01 5.27295768e-01 3.16502929e-01
-1.40220094e+00 7.02736497e-01 6.55042076e+00 6.43733144e-01
-1.05866838e+00 1.56900480e-01 7.36043036e-01 7.33550265e-03
-2.70404909e-02 -1.45529643e-01 -4.04256284e-01 3.88673455e-01
8.68294835e-01 4.11967397e-01 2.87236243e-01 6.94690943e-01
-1.06957043e-02 -5.03191888e-01 -1.22246945e+00 1.24900877e+00
6.92703485e-01 -1.30541372e+00 4.64385077e-02 -2.06284299e-01
7.01203406e-01 -5.51931858e-02 4.46301363e-02 -7.92439058e-02
-3.48816738e-02 -6.70422018e-01 5.31337202e-01 6.00306034e-01
5.60707927e-01 -4.77273762e-01 7.84048140e-01 7.35587329e-02
-8.13995659e-01 -2.12035105e-01 -1.01952299e-01 3.61089557e-02
9.18333977e-02 6.92961693e-01 -2.56811708e-01 7.79434562e-01
6.82276845e-01 1.31024230e+00 -8.18555713e-01 1.18434572e+00
-2.64736354e-01 7.19748855e-01 -6.75124153e-02 3.40688646e-01
2.42593214e-01 1.80017538e-02 6.88510418e-01 1.24456370e+00
-1.58011615e-01 -2.93566316e-01 7.32770376e-03 5.81144154e-01
1.59638360e-01 -1.73719563e-02 -6.60364270e-01 6.88898936e-02
-1.64459422e-01 1.04086637e+00 -7.08591223e-01 -6.27090156e-01
-5.08328855e-01 1.32824445e+00 1.76903725e-01 3.50862235e-01
-3.97352338e-01 -7.66207278e-02 4.49652761e-01 -3.35207768e-02
6.27982199e-01 9.78094786e-02 2.85032094e-02 -1.29850602e+00
1.45807758e-01 -6.32467926e-01 5.27738690e-01 -9.25023258e-01
-1.27188957e+00 4.41611737e-01 -1.69021133e-02 -1.71089506e+00
-5.66320419e-01 -7.79924750e-01 -1.97379917e-01 3.43551248e-01
-2.10443282e+00 -9.13485944e-01 -4.25912678e-01 7.15428829e-01
5.30219316e-01 -3.26295584e-01 7.73013651e-01 3.05879653e-01
-1.10504799e-01 4.43073750e-01 1.84093416e-01 4.55906279e-02
7.64672756e-01 -1.35812724e+00 -6.90407604e-02 7.52554655e-01
5.50472677e-01 2.28712745e-02 3.77512664e-01 -3.61304939e-01
-1.41993988e+00 -9.82120037e-01 4.39097911e-01 -4.96273458e-01
5.08433878e-01 -1.60188407e-01 -6.44216418e-01 2.81015992e-01
1.57468095e-01 4.72127140e-01 5.87897241e-01 -3.18901300e-01
-5.50046861e-01 -3.08484703e-01 -1.21726465e+00 5.23724966e-02
1.06251764e+00 -6.31879985e-01 -4.84853476e-01 4.86149788e-01
3.20888907e-01 -1.84679925e-02 -5.99361598e-01 4.19243991e-01
4.22778994e-01 -1.39767599e+00 1.02405226e+00 -6.67433858e-01
7.30566502e-01 -2.97712535e-01 -2.73977548e-01 -1.16256654e+00
1.79552153e-01 -5.33495188e-01 -4.39627022e-01 1.07372117e+00
2.11796194e-01 -5.52231550e-01 9.88923252e-01 1.68671861e-01
1.55725121e-01 -6.60878360e-01 -1.02914464e+00 -9.96633232e-01
-4.04762685e-01 -5.66847801e-01 -3.12957197e-01 8.49602580e-01
-1.51755363e-01 2.26364285e-01 -6.50731444e-01 -1.05526581e-01
8.82232189e-01 -8.54674801e-02 7.12535620e-01 -1.06244826e+00
-4.74154174e-01 4.70154593e-03 -9.70913410e-01 -1.08933270e+00
7.98166394e-02 -7.36745059e-01 2.70192642e-02 -1.27645695e+00
2.50433326e-01 -3.09475008e-02 -5.06720841e-01 3.67455333e-01
-1.00943901e-01 6.66542768e-01 3.62754554e-01 3.28690886e-01
-1.03478014e+00 6.46770954e-01 9.91234779e-01 -2.87851751e-01
1.16819456e-01 6.93067387e-02 -5.00379741e-01 6.88014984e-01
2.41771489e-01 -3.71715367e-01 -4.20398533e-01 -2.26267591e-01
-5.34256105e-04 1.36943376e-02 7.26611614e-01 -1.35312116e+00
9.95670632e-02 2.82641292e-01 7.28819251e-01 -4.32689130e-01
5.06137729e-01 -9.48998749e-01 -4.23245281e-01 4.71810818e-01
-4.55986649e-01 -2.56196022e-01 7.77808130e-02 9.61070716e-01
-4.47426587e-01 -4.26339358e-01 8.34199369e-01 -3.81333917e-01
-9.01156127e-01 -1.02216639e-02 -3.61626714e-01 2.18501762e-01
8.79080355e-01 -4.34941530e-01 -2.23452747e-01 -6.46110177e-01
-6.73969269e-01 -2.17456952e-01 3.50230038e-01 1.72484234e-01
7.65729725e-01 -1.05236852e+00 -5.65106213e-01 2.44751170e-01
3.86648655e-01 -3.80177766e-01 1.30971789e-01 7.57158875e-01
-3.30827147e-01 3.57671708e-01 -3.30796719e-01 -9.74701107e-01
-1.25346136e+00 5.30080259e-01 1.93144247e-01 -1.13354467e-01
-6.66205108e-01 8.21010470e-01 4.54865023e-02 -2.19685689e-01
5.21370292e-01 6.76507577e-02 -2.86159337e-01 2.49003693e-01
7.64749825e-01 6.59028828e-01 2.28024170e-01 -6.74255848e-01
-4.33438748e-01 6.47044122e-01 5.44554219e-02 -2.54069805e-01
1.40630281e+00 -5.84587753e-02 1.43996045e-01 5.96752942e-01
1.53484678e+00 -2.26734295e-01 -1.60651624e+00 -2.87319779e-01
-2.24537104e-02 -8.35492194e-01 2.65713990e-01 -8.67525995e-01
-1.35175014e+00 9.75855887e-01 1.11472404e+00 -6.70466349e-02
1.25635540e+00 4.18479070e-02 3.84122312e-01 4.77778405e-01
3.13151777e-01 -1.28445125e+00 5.74089885e-01 1.49487764e-01
6.34137094e-01 -1.61998320e+00 4.07528788e-01 -5.89354992e-01
-8.73223484e-01 1.03813648e+00 2.52002627e-01 -4.01057571e-01
6.54240787e-01 5.14642103e-03 5.37745468e-02 -2.93803513e-01
-6.97717786e-01 -6.63141310e-01 4.21449304e-01 9.30002689e-01
1.93749115e-01 -5.46070218e-01 1.06660381e-01 -2.59137601e-01
6.22654319e-01 1.95402443e-01 3.46312463e-01 1.09870684e+00
-1.33864135e-01 -8.47995758e-01 -1.27283990e-01 3.69666785e-01
-3.07010531e-01 -1.21933542e-01 -4.92610753e-01 6.43952370e-01
-1.91670328e-01 1.11485016e+00 7.59319663e-02 -1.88730314e-01
1.31109312e-01 5.06486446e-02 6.89689338e-01 -4.57166791e-01
-4.12266076e-01 -5.34616783e-02 1.13587014e-01 -1.11109602e+00
-1.31438482e+00 -7.25099504e-01 -7.86764562e-01 3.90975296e-01
-3.31254840e-01 -1.57654032e-01 6.95259273e-01 1.13731456e+00
4.05525118e-01 4.26325738e-01 9.72635627e-01 -1.12802529e+00
-4.12907958e-01 -6.34419262e-01 -4.18037415e-01 8.30739915e-01
4.98934954e-01 -1.03697956e+00 -7.65475690e-01 3.94232810e-01] | [8.815120697021484, 0.851734459400177] |
0f285daf-b882-4df2-a4dd-cc4ca687449a | on-the-generalisation-capabilities-of-fisher | 2103.01721 | null | https://arxiv.org/abs/2103.01721v1 | https://arxiv.org/pdf/2103.01721v1.pdf | On the Generalisation Capabilities of Fisher Vector based Face Presentation Attack Detection | In the last decades, the broad development experienced by biometric systems has unveiled several threats which may decrease their trustworthiness. Those are attack presentations which can be easily carried out by a non-authorised subject to gain access to the biometric system. In order to mitigate those security concerns, most face Presentation Attack Detection techniques have reported a good detection performance when they are evaluated on known Presentation Attack Instruments (PAI) and acquisition conditions, in contrast to more challenging scenarios where unknown attacks are included in the test set. For those more realistic scenarios, the existing algorithms face difficulties to detect unknown PAI species in many cases. In this work, we use a new feature space based on Fisher Vectors, computed from compact Binarised Statistical Image Features histograms, which allow discovering semantic feature subsets from known samples in order to enhance the detection of unknown attacks. This new representation, evaluated for challenging unknown attacks taken from freely available facial databases, shows promising results: a BPCER100 under 17% together with an AUC over 98% can be achieved in the presence of unknown attacks. In addition, by training a limited number of parameters, our method is able to achieve state-of-the-art deep learning-based approaches for cross-dataset scenarios. | ['Christoph Busch', 'Marta Gomez-Barrero', 'Lázaro J. González-Soler'] | 2021-03-02 | null | null | null | null | ['face-presentation-attack-detection'] | ['computer-vision'] | [ 4.25762028e-01 -2.21821472e-01 1.96446389e-01 -2.76552171e-01
-5.75868011e-01 -7.31969416e-01 8.60292494e-01 2.23566025e-01
-5.70693254e-01 5.95032334e-01 -3.82845253e-01 -2.57712603e-03
-2.89271414e-01 -6.66427612e-01 -4.45733786e-01 -8.31560910e-01
-3.05256724e-01 2.78552949e-01 1.62550788e-02 -1.87647447e-01
2.74101138e-01 1.02844465e+00 -1.96948063e+00 3.12926024e-01
4.29120272e-01 1.14983082e+00 -4.34353858e-01 3.39686930e-01
1.82713106e-01 -1.42666409e-02 -1.05527866e+00 -1.00306940e+00
3.82954627e-01 -3.52215208e-02 -2.79111445e-01 -3.20515662e-01
7.98428059e-01 -5.60996175e-01 9.49857831e-02 1.08911419e+00
7.73392081e-01 -3.18138629e-01 8.19373846e-01 -1.30933475e+00
-2.49121070e-01 3.82174820e-01 -5.73529780e-01 1.01608247e-01
6.43126428e-01 2.35539079e-01 6.34008706e-01 -7.04965889e-01
3.25354785e-01 1.23207843e+00 5.77298045e-01 5.96644759e-01
-1.27067864e+00 -1.12497556e+00 -3.68476927e-01 3.27344835e-01
-1.68575633e+00 -5.37654519e-01 8.54064345e-01 -3.51224780e-01
5.01994610e-01 2.60816246e-01 3.51819694e-01 1.58995056e+00
2.37167284e-01 3.45820069e-01 1.24858904e+00 -3.36051941e-01
1.03360102e-01 4.85343635e-01 -8.55572894e-02 3.35643858e-01
5.91083825e-01 3.21855545e-01 -4.01464462e-01 -4.99603450e-01
2.43607253e-01 -1.07356876e-01 -1.91220403e-01 -1.21404938e-01
-4.32209581e-01 8.74030590e-01 1.33680776e-01 5.91247678e-01
-3.48039746e-01 -5.14893293e-01 5.52405477e-01 1.24643266e-01
3.28696519e-01 3.07775468e-01 -1.96056679e-01 1.65823586e-02
-1.01593947e+00 1.98318228e-01 8.90658498e-01 1.34690255e-01
4.77085054e-01 2.25833684e-01 -3.44304964e-02 6.63657248e-01
1.84908330e-01 8.14284027e-01 3.01517278e-01 -1.01598263e-01
2.70128071e-01 4.80251610e-01 -9.32606161e-02 -1.48866057e+00
-3.96899551e-01 -4.84846652e-01 -9.11730886e-01 3.64770889e-01
6.95965052e-01 9.22593921e-02 -7.55691409e-01 1.65950286e+00
3.94003779e-01 2.24936187e-01 9.85264257e-02 7.06399322e-01
6.23426497e-01 3.77465993e-01 1.78550974e-01 -2.24445045e-01
1.42756855e+00 1.01497196e-01 -6.00542963e-01 2.34714165e-01
2.14191571e-01 -8.50772500e-01 9.02925789e-01 9.50323999e-01
-5.99005401e-01 -5.80738664e-01 -1.31085277e+00 6.96428537e-01
-7.83259988e-01 1.60032421e-01 2.84404576e-01 1.57404172e+00
-8.06882679e-01 6.14033103e-01 -3.58271629e-01 -3.59784216e-01
7.18845367e-01 6.59574926e-01 -8.14630210e-01 -7.47142583e-02
-1.29054117e+00 8.52864444e-01 5.82830869e-02 4.36034411e-01
-9.13645566e-01 -5.54285169e-01 -6.92013681e-01 1.41051352e-01
2.86137849e-01 1.82036996e-01 5.26860356e-01 -8.69235396e-01
-1.27365363e+00 9.42018926e-01 2.63299346e-01 -5.38134217e-01
6.18993878e-01 -2.39669681e-01 -6.37359619e-01 2.93094844e-01
-3.39685470e-01 1.29515603e-01 1.18752611e+00 -1.14946842e+00
-9.60534066e-02 -7.15848744e-01 -5.41877411e-02 -4.17901129e-01
-8.08230460e-01 3.95299584e-01 3.87274213e-02 -4.68470007e-01
-4.66074735e-01 -7.97228754e-01 3.23104560e-01 3.89430076e-02
-1.67705178e-01 -4.87207621e-02 1.00558102e+00 -6.72146142e-01
9.40988362e-01 -2.14917088e+00 -1.89032167e-01 5.29541194e-01
-3.06874048e-02 9.82255340e-01 -7.01520219e-02 3.56396824e-01
-1.96563974e-01 2.66220391e-01 -1.32125169e-01 -2.37727389e-01
2.27308087e-02 -1.78087339e-01 -3.69187206e-01 7.21027613e-01
2.94064105e-01 1.81822136e-01 -4.52320307e-01 -2.38948092e-01
3.55905235e-01 7.28268385e-01 -1.82317883e-01 3.04942131e-01
2.66035110e-01 3.27580005e-01 -1.78950772e-01 8.25474918e-01
1.27932858e+00 4.33856457e-01 8.78918394e-02 -2.17075229e-01
3.68332118e-01 -4.51275975e-01 -1.34282267e+00 1.00653493e+00
-2.47215852e-01 5.45836389e-01 -2.64102221e-03 -8.54840994e-01
1.16283250e+00 3.85495454e-01 3.95399511e-01 -5.14986873e-01
5.51898777e-01 2.50311077e-01 2.02237025e-01 -3.69675547e-01
1.95090175e-02 -1.14789113e-01 1.10732682e-01 2.37492442e-01
3.11575294e-01 1.61415145e-01 -5.43480320e-03 -1.21683449e-01
7.19583452e-01 -1.09367497e-01 2.52525479e-01 -3.86449508e-03
1.00602627e+00 -6.84318483e-01 2.86951482e-01 7.00882137e-01
-2.84525752e-01 4.29403067e-01 3.65631044e-01 -4.65909839e-01
-6.73816681e-01 -1.12831080e+00 -4.93206322e-01 5.73081255e-01
-1.25111371e-01 -1.97136045e-01 -9.57947791e-01 -8.80652726e-01
1.85753107e-02 3.57505172e-01 -6.95405066e-01 -2.90313154e-01
-2.62245744e-01 -8.07043254e-01 1.10855985e+00 1.99821785e-01
5.11361897e-01 -9.31681812e-01 -6.88180089e-01 -5.22138886e-02
2.66425103e-01 -1.14336658e+00 1.74799800e-01 -8.35096985e-02
-2.54482955e-01 -1.37143660e+00 -8.35336864e-01 -1.45358726e-01
3.75025898e-01 -1.86153814e-01 5.68687558e-01 1.04579248e-01
-8.54552984e-01 3.42280835e-01 -3.18662763e-01 -8.17488015e-01
-3.55597943e-01 -1.88814864e-01 3.49423677e-01 5.68759918e-01
7.17258513e-01 -3.82467508e-01 -4.14557576e-01 4.02635783e-01
-1.12042534e+00 -8.55095446e-01 5.15242338e-01 7.36645222e-01
5.61665185e-02 1.89987957e-01 4.88580108e-01 -6.38540149e-01
4.87897635e-01 -4.45221812e-01 -6.12654269e-01 3.70421052e-01
-2.10710004e-01 -2.15782627e-01 7.08263814e-01 -8.46194506e-01
-9.05340135e-01 -2.12290753e-02 -2.50862271e-01 -4.46668983e-01
-5.43277264e-01 1.69742033e-01 -5.14704287e-01 -5.21805048e-01
8.92880917e-01 1.32246047e-01 2.37516925e-01 -3.27292085e-01
-1.12312697e-01 7.63982594e-01 2.08534032e-01 -5.52181482e-01
1.13221467e+00 4.59650159e-01 4.17903692e-01 -1.12356210e+00
-3.64456922e-01 -2.33028904e-01 -6.94937170e-01 -5.26301682e-01
5.96516490e-01 -4.40436453e-01 -1.08429098e+00 9.29639995e-01
-1.00908458e+00 4.08505827e-01 3.27223569e-01 3.96672785e-01
-5.12960888e-02 7.70271659e-01 -1.87342823e-01 -1.36603081e+00
-3.25311869e-01 -1.30190539e+00 9.67915595e-01 3.38596553e-01
-1.29501477e-01 -8.45272303e-01 -1.89379290e-01 2.93069154e-01
5.67403913e-01 6.56181335e-01 6.18714213e-01 -1.19647563e+00
4.43256758e-02 -7.72656679e-01 -3.58800739e-02 7.04090059e-01
3.64121288e-01 1.44499779e-01 -1.48350811e+00 -5.15264392e-01
5.54555915e-02 -3.14295322e-01 5.00335991e-01 9.73797869e-03
1.08765733e+00 -2.32124299e-01 -2.03221202e-01 4.09050167e-01
1.31582248e+00 2.27624863e-01 6.95503056e-01 5.71706295e-02
2.09358886e-01 9.18340206e-01 3.68886441e-01 5.46392024e-01
-3.37308079e-01 7.84669280e-01 5.83220720e-01 -1.56893238e-01
2.80002594e-01 -3.25885639e-02 2.19935328e-01 -3.30670685e-01
-6.80512339e-02 -3.41846883e-01 -9.12525535e-01 1.52268454e-01
-1.22781146e+00 -1.01555467e+00 1.28976464e-01 2.45335937e+00
3.84322256e-01 2.40160897e-01 3.64968389e-01 7.25075364e-01
7.62313962e-01 1.09928712e-01 -2.79071599e-01 -5.23003757e-01
-3.11324269e-01 6.74849629e-01 3.30636710e-01 1.82534866e-02
-1.21636486e+00 7.61891186e-01 5.77499819e+00 9.20410514e-01
-1.54352713e+00 -1.12767659e-01 5.64839244e-01 5.05585410e-02
2.95992166e-01 -4.47184473e-01 -8.04702759e-01 5.58658600e-01
1.13906765e+00 1.92523390e-01 1.41960531e-01 6.69674754e-01
-3.08583695e-02 -6.58331066e-02 -7.68103778e-01 1.20278776e+00
6.18589759e-01 -8.31486464e-01 2.17901289e-01 4.35159296e-01
4.19007987e-02 -6.92460895e-01 6.00670457e-01 8.90162215e-02
-3.01197082e-01 -1.26825237e+00 2.40444213e-01 4.37239736e-01
7.30745852e-01 -1.07802808e+00 1.16447175e+00 1.53913885e-01
-7.98244953e-01 -2.84687102e-01 -4.14050251e-01 2.28780538e-01
-1.14645079e-01 3.04325402e-01 -1.02517235e+00 7.81446993e-01
6.45128489e-01 9.79138836e-02 -7.83454001e-01 1.07822943e+00
5.14860936e-02 6.18855596e-01 -4.69247162e-01 -4.90377620e-02
2.45957216e-03 8.37078989e-02 5.12175977e-01 1.08113873e+00
3.70112151e-01 -1.22395903e-01 -1.68592602e-01 5.72620153e-01
2.22535163e-01 4.61085379e-01 -8.38582993e-01 -9.59513411e-02
1.17620699e-01 1.40447903e+00 -5.83351135e-01 1.96422607e-01
-1.72992334e-01 7.57834673e-01 -8.70654806e-02 1.29299283e-01
-8.24140489e-01 -3.74175787e-01 5.71614146e-01 2.88996965e-01
1.39133543e-01 4.72290926e-02 1.86379433e-01 -9.27444220e-01
2.04218015e-01 -1.20356679e+00 5.59616804e-01 -3.03543031e-01
-1.37456882e+00 9.35842216e-01 9.83093157e-02 -1.25786293e+00
-3.51878136e-01 -1.09345889e+00 -3.57604414e-01 8.78261149e-01
-1.20173323e+00 -1.55501187e+00 -1.19539544e-01 6.28579795e-01
6.78632632e-02 -7.25049913e-01 1.03735912e+00 4.18666303e-01
-4.05543774e-01 1.17128658e+00 -1.21137746e-01 2.38369107e-01
7.61026442e-01 -7.31664360e-01 -1.89422697e-01 8.94387126e-01
3.07821035e-01 6.42963767e-01 6.33222759e-01 -4.60006475e-01
-1.21438181e+00 -5.53390026e-01 3.89478445e-01 -4.24177676e-01
3.93381864e-01 -6.95470512e-01 -1.00147748e+00 -4.04333994e-02
-1.15371525e-01 -2.78706476e-02 1.10732734e+00 8.78069177e-03
-7.40400851e-01 -3.23682278e-01 -1.64760733e+00 3.66793603e-01
4.57489550e-01 -4.59420472e-01 -3.52855116e-01 4.18195222e-03
-1.59759447e-01 1.72663108e-02 -7.53899932e-01 6.25196397e-01
9.79077518e-01 -1.02266145e+00 9.14882123e-01 -6.53472662e-01
-1.11226901e-01 -1.64379954e-01 -2.56208092e-01 -8.45245838e-01
2.67505944e-01 -4.29410875e-01 1.11731388e-01 1.48883832e+00
1.60182878e-01 -6.48470819e-01 7.92296588e-01 4.06734973e-01
6.98854983e-01 -4.87520218e-01 -1.14038360e+00 -8.35688293e-01
-1.04235999e-01 -3.60649377e-01 7.19499886e-01 6.64627433e-01
-3.48115295e-01 -2.47495368e-01 -4.99302149e-01 3.61593366e-01
9.85821843e-01 -4.61567163e-01 8.89281869e-01 -1.71840525e+00
-8.14181492e-02 -3.90288740e-01 -1.11285186e+00 4.76852767e-02
4.76141840e-01 -5.32076180e-01 -3.52446705e-01 -4.97970760e-01
1.65950745e-01 -9.36542526e-02 -4.87691432e-01 4.30646628e-01
3.34644988e-02 7.39607215e-01 2.97492117e-01 -1.00887187e-01
5.92440814e-02 2.62169003e-01 6.28169000e-01 -3.04695874e-01
1.17430925e-01 2.11013824e-01 -4.96831357e-01 6.05544686e-01
5.36082625e-01 -3.97133797e-01 -2.00220659e-01 1.72457039e-01
-2.14591727e-01 -2.50341505e-01 4.35151100e-01 -1.30774939e+00
-2.14425959e-02 6.80907592e-02 7.79481232e-01 -4.92542148e-01
6.16453350e-01 -1.05089796e+00 1.80149570e-01 4.98790920e-01
-1.84048731e-02 -2.26269543e-01 5.66901445e-01 4.18035090e-01
-1.55766591e-01 -3.44191521e-01 9.60443795e-01 2.39364803e-01
-2.31018648e-01 2.10955009e-01 -2.85715789e-01 -4.89492565e-01
1.10889494e+00 -5.07556438e-01 -2.07219854e-01 -4.17496830e-01
-5.80222905e-01 -4.74346876e-01 2.68632472e-01 3.92979592e-01
6.06844306e-01 -9.47137833e-01 -8.11585724e-01 5.04844129e-01
2.05404684e-01 -7.84101367e-01 5.11911869e-01 4.83647555e-01
-4.37191814e-01 2.52794385e-01 -6.88503087e-01 -6.72281086e-01
-1.86232746e+00 6.66357696e-01 2.65851170e-01 -1.61164925e-02
1.22072594e-02 6.30436063e-01 6.30578212e-03 -1.63712144e-01
2.42635518e-01 3.10734630e-01 -5.96297085e-01 4.31270957e-01
9.62977707e-01 1.82071373e-01 4.76306051e-01 -9.55932438e-01
-5.18688738e-01 4.96200889e-01 -2.52882004e-01 -5.38465902e-02
1.18579280e+00 5.05554914e-01 6.02958426e-02 4.93412204e-02
1.26687479e+00 2.26595595e-01 -7.51271129e-01 -7.28156567e-02
-2.92300954e-02 -8.18749487e-01 -2.20296323e-01 -9.04763103e-01
-1.10070491e+00 1.09682858e+00 1.22714102e+00 2.86168993e-01
1.17193210e+00 -3.62498432e-01 3.00693870e-01 2.96886683e-01
4.54448670e-01 -7.39297152e-01 1.93428658e-02 5.23437895e-02
8.52110684e-01 -1.17683280e+00 -3.47068273e-02 -3.27344358e-01
-4.37353641e-01 1.35016966e+00 4.51174825e-01 9.90978926e-02
7.57359684e-01 1.79934248e-01 1.50701985e-01 -9.86756608e-02
-1.40956700e-01 -3.67335565e-02 4.50700641e-01 9.32816803e-01
3.23574990e-01 -1.73689425e-01 -3.71123582e-01 6.65179193e-01
-2.00683862e-01 -3.42983007e-01 4.52875525e-01 6.17505193e-01
-1.54422866e-02 -1.24420762e+00 -7.28498578e-01 3.21308076e-01
-8.92178953e-01 3.21860522e-01 -6.70437574e-01 8.87654960e-01
2.11175308e-01 1.04263055e+00 -1.14540853e-01 -4.77846473e-01
3.01122159e-01 1.15053684e-01 5.36487579e-01 -2.90699124e-01
-8.96930456e-01 -8.32976028e-02 -1.27419859e-01 -4.42124903e-01
-3.57615858e-01 -7.01012075e-01 -4.24946100e-01 -3.03613842e-01
-5.78462303e-01 1.64000928e-01 9.95792091e-01 9.91702199e-01
8.79326016e-02 -2.18473990e-02 8.11829925e-01 -9.06866252e-01
-7.95982480e-01 -9.86186385e-01 -7.15992570e-01 5.77386200e-01
1.59721464e-01 -1.11068058e+00 -5.65340340e-01 -2.70865053e-01] | [13.068305969238281, 1.0729204416275024] |
e87be0b5-6fb5-47e1-96f9-ac0890998b0d | motion-compensated-three-dimensional | 2204.12882 | null | https://arxiv.org/abs/2204.12882v1 | https://arxiv.org/pdf/2204.12882v1.pdf | Motion Compensated Three-Dimensional Frequency Selective Extrapolation for Improved Error Concealment in Video Communication | During transmission of video data over error-prone channels the risk of getting severe image distortions due to transmission errors is ubiquitous. To deal with image distortions at decoder side, error concealment is applied. This article presents Motion Compensated Three-Dimensional Frequency Selective Extrapolation, a novel spatio-temporal error concealment algorithm. The algorithm uses fractional-pel motion estimation and compensation as initial step, being followed by the generation of a model of the distorted signal. The model generation is conducted by an enhanced version of Three-Dimensional Frequency Selective Extrapolation, an existing error concealment algorithm. Compared to this existent algorithm, the proposed one yields an improvement in concealment quality of up to 1.64 dB PSNR. Altogether, the incorporation of motion compensation and the improved model generation extends the already high extrapolation quality of the underlying Frequency Selective Extrapolation, resulting in a gain of more than 3 dB compared to other well-known error concealment algorithms. | ['André Kaup', 'Jürgen Seiler'] | 2022-04-27 | null | null | null | null | ['motion-compensation'] | ['computer-vision'] | [ 8.26890111e-01 -5.13902418e-02 2.02268854e-01 1.27816960e-01
-6.13403916e-01 -1.55623525e-01 5.08568227e-01 3.28394949e-01
-5.19847631e-01 8.99722219e-01 5.03673494e-01 -4.04897988e-01
-3.70252728e-02 -5.90339303e-01 -6.97838724e-01 -7.15858281e-01
-4.02060151e-01 -5.33679366e-01 2.05708608e-01 8.64819288e-02
5.22527397e-01 4.91786450e-01 -1.53562999e+00 4.27978933e-01
9.17772293e-01 1.19688225e+00 2.77668923e-01 1.09461987e+00
2.19349027e-01 7.89708138e-01 -4.34165120e-01 -2.39524782e-01
1.02142088e-01 -5.08831680e-01 -4.50748831e-01 2.67245948e-01
-5.37526198e-02 -7.56535053e-01 -3.82071286e-01 1.07136393e+00
3.30312848e-01 -2.18194127e-01 7.25575864e-01 -7.02325284e-01
-3.65734905e-01 4.06068601e-02 -3.79736423e-01 1.27230301e-01
7.38889039e-01 -9.74455103e-03 3.22850108e-01 -1.11985457e+00
4.23797309e-01 7.75861442e-01 9.54096556e-01 3.78172457e-01
-1.27126944e+00 -3.80505443e-01 -6.64744735e-01 5.78006387e-01
-1.58050632e+00 -6.58123374e-01 7.93485582e-01 -4.40127492e-01
6.47729158e-01 5.34436107e-01 7.74196982e-01 4.56099898e-01
6.77311718e-01 4.46129173e-01 1.03362465e+00 -5.67088127e-01
2.29787841e-01 1.17917046e-01 -5.13932705e-01 3.71236533e-01
4.11093026e-01 5.11755824e-01 -4.22885925e-01 -1.52878836e-01
6.25416398e-01 -4.62638661e-02 -9.30209517e-01 -2.27925807e-01
-9.41718876e-01 3.45635802e-01 2.36342043e-01 4.01586503e-01
-7.29421496e-01 2.23557919e-01 3.41913372e-01 2.11718872e-01
5.50794423e-01 -7.37784654e-02 8.22504163e-02 -2.24827081e-01
-1.56779063e+00 2.24230096e-01 5.34295082e-01 8.47803414e-01
5.81247747e-01 3.81021380e-01 -1.57359689e-01 3.50519389e-01
6.27569377e-01 2.21848890e-01 3.84949058e-01 -7.05904901e-01
4.39447522e-01 3.92334819e-01 3.28114063e-01 -1.17252564e+00
4.55333404e-02 -5.49651742e-01 -1.08533502e+00 7.29566574e-01
2.40585908e-01 -1.84616193e-01 -6.53297603e-01 1.15520823e+00
1.62026912e-01 2.02054903e-01 3.19415867e-01 7.77151823e-01
2.69524664e-01 1.08980954e+00 -5.71235642e-02 -6.95398748e-01
1.09082973e+00 -5.32540143e-01 -1.20678997e+00 1.35949999e-01
1.16561726e-01 -1.28595841e+00 1.27480358e-01 6.23457313e-01
-1.28805614e+00 -5.41698217e-01 -1.50751197e+00 4.94118147e-02
1.23168238e-01 7.53146112e-02 9.42059457e-02 9.73366499e-01
-1.06519365e+00 7.26236045e-01 -4.10075009e-01 3.87442172e-01
3.21239084e-01 2.61249781e-01 -4.19229031e-01 -2.63835937e-01
-1.15817785e+00 7.36440063e-01 2.40813017e-01 2.21292108e-01
-5.38658202e-01 -8.47453177e-01 -8.61599147e-01 5.05137555e-02
-2.40715723e-02 -5.23479342e-01 7.63161302e-01 -1.20759404e+00
-1.34422243e+00 4.12121266e-01 -5.08442044e-01 -7.65470445e-01
9.92173791e-01 -1.81207523e-01 -5.80438733e-01 5.62683880e-01
-4.61720407e-01 7.89826289e-02 1.25421965e+00 -1.35485077e+00
-4.70752865e-01 -1.97987497e-01 -3.05336565e-01 1.67316616e-01
5.96303307e-02 -1.96065485e-01 -8.00933242e-02 -9.97220695e-01
1.03719577e-01 -2.89564371e-01 -2.02103138e-01 4.66890723e-01
4.45802882e-02 7.67800331e-01 7.60464251e-01 -1.35701454e+00
1.59934080e+00 -2.32182789e+00 8.92988890e-02 -1.54365832e-02
6.63068220e-02 4.43814576e-01 4.03879225e-01 6.40772820e-01
-1.54126197e-01 -1.03221811e-01 -8.01886141e-01 -3.25179249e-01
-4.97081608e-01 -2.71549731e-01 4.84125279e-02 8.92658472e-01
1.05860764e-02 2.65052438e-01 -6.41622066e-01 -1.56218946e-01
3.88202697e-01 1.03157997e+00 -4.26048160e-01 1.98163375e-01
2.51910895e-01 3.54669690e-01 3.60220224e-02 4.75470334e-01
1.27149105e+00 3.42582136e-01 1.39679266e-02 -1.40866220e-01
-4.03132498e-01 -3.62022333e-02 -1.42095184e+00 1.28261054e+00
-3.87374789e-01 8.55603695e-01 2.40369916e-01 -5.95848262e-01
9.45056677e-01 9.90571856e-01 4.82855320e-01 -4.95428562e-01
1.22059293e-01 6.27013743e-01 -3.24342489e-01 -6.03559852e-01
9.68340814e-01 -2.78935701e-01 4.95509952e-01 -6.47396073e-02
-3.67944419e-01 2.56904274e-01 -2.46757820e-01 -1.74301669e-01
7.32358277e-01 1.36167333e-01 6.08060777e-01 -3.19010377e-01
9.62028384e-01 -3.71023566e-01 3.66947442e-01 1.23262666e-01
-1.58623859e-01 8.19183707e-01 -9.22196079e-03 -2.85885543e-01
-1.43643117e+00 -6.79626703e-01 -3.71600479e-01 -1.89427808e-01
2.80954480e-01 -2.72637576e-01 -8.20901394e-01 -9.99769270e-02
-1.96150541e-01 5.05144119e-01 -3.13837796e-01 -2.29206145e-01
-6.22406304e-01 -5.73418498e-01 4.70030099e-01 -1.77757461e-02
9.27098632e-01 -4.87060726e-01 -7.26611853e-01 6.23936534e-01
-2.64743716e-01 -9.63348448e-01 -4.60086048e-01 -3.30149055e-01
-1.14475560e+00 -9.07850504e-01 -1.38944161e+00 -5.76993108e-01
4.91794974e-01 2.36413822e-01 7.34354377e-01 3.12809676e-01
-2.69238651e-01 8.58083293e-02 -4.44987684e-01 -2.24132314e-02
-1.04245865e+00 -6.94583356e-01 -2.44655728e-01 5.77504694e-01
7.46168047e-02 -5.23391247e-01 -1.06143975e+00 5.02157882e-02
-1.06086898e+00 2.95233391e-02 5.58046699e-01 6.70279324e-01
1.80041626e-01 3.79290760e-01 4.22583759e-01 -4.85170275e-01
3.38872492e-01 -4.86075848e-01 -6.29252553e-01 -1.71152875e-01
-8.38647962e-01 -7.40282089e-02 7.02973545e-01 -2.14757904e-01
-1.19666290e+00 1.53474296e-02 -5.56980789e-01 -4.70107347e-02
-5.88470213e-02 2.33639881e-01 -1.25200897e-01 -5.34861445e-01
5.47711074e-01 7.01770663e-01 9.24769640e-02 -5.81985891e-01
-8.34431779e-03 8.98321211e-01 4.40177917e-01 2.58049518e-01
6.05581999e-01 3.66115928e-01 2.46230260e-01 -9.98230815e-01
5.01166403e-01 -3.06069404e-01 -9.54735354e-02 -5.06878674e-01
6.89381421e-01 -9.73965228e-01 -4.95716006e-01 9.77764010e-01
-1.43855178e+00 8.82189274e-02 -1.74798995e-01 7.36846924e-01
-6.23851955e-01 9.47405815e-01 -6.26922667e-01 -1.15900636e+00
-2.16866687e-01 -1.23981154e+00 5.12739956e-01 -4.32002321e-02
1.61415085e-01 -8.98363531e-01 -2.03249156e-01 5.61190769e-02
7.80204475e-01 3.63552779e-01 4.86378491e-01 3.25446486e-01
-8.72337520e-01 -2.00858414e-01 -1.44830689e-01 5.14080226e-01
1.59627929e-01 -3.11266482e-01 -8.60494196e-01 -3.21105659e-01
4.26711082e-01 5.88065743e-01 8.35445821e-01 3.97430509e-01
6.61672533e-01 -6.54707134e-01 -6.59680888e-02 7.37585604e-01
2.06713676e+00 5.28994560e-01 1.32537532e+00 2.98088878e-01
1.27537310e-01 3.74907464e-01 4.38719213e-01 7.59724557e-01
9.13129002e-02 8.01462889e-01 5.12232661e-01 -6.54412061e-02
-3.99918020e-01 -4.92181629e-02 2.29430392e-01 5.27449012e-01
-3.58748317e-01 -3.00695539e-01 -5.85135102e-01 8.00704122e-01
-1.23556530e+00 -1.10545123e+00 -4.66486007e-01 2.59940505e+00
8.34045231e-01 -6.82501942e-02 -1.41324401e-01 1.03597021e+00
7.97603965e-01 1.87314644e-01 -3.83946821e-02 -6.24715447e-01
-1.29874781e-01 -1.38115942e-01 9.77178276e-01 1.00691450e+00
-7.71076202e-01 9.62279588e-02 6.05240440e+00 1.03694308e+00
-1.12299991e+00 -2.69131307e-02 2.71528870e-01 5.11528552e-01
-3.60687912e-01 3.58703732e-02 -3.76566142e-01 6.53978288e-01
1.12698138e+00 -2.95320988e-01 2.98517168e-01 2.56907821e-01
4.89903212e-01 -5.70209086e-01 -5.95311522e-01 1.02423966e+00
1.29503906e-01 -1.29998147e+00 7.22190514e-02 3.03433806e-01
4.66717124e-01 -9.80652213e-01 -9.26522091e-02 -6.34543002e-01
-6.72706962e-01 -8.81839395e-01 9.09477353e-01 6.26603007e-01
9.49553609e-01 -9.15147305e-01 7.93772459e-01 3.02491486e-01
-1.25266242e+00 2.18680408e-02 -2.19343320e-01 -9.69663635e-02
7.22354770e-01 8.62725198e-01 -5.80668449e-01 9.04578805e-01
3.62647116e-01 5.00054836e-01 7.60343596e-02 1.49199462e+00
-3.25869881e-02 6.22097969e-01 3.56042795e-02 3.42898607e-01
1.20602481e-01 -1.71538219e-02 1.08189487e+00 1.41901362e+00
7.54431129e-01 1.35422856e-01 -7.78946161e-01 5.32602906e-01
2.20806718e-01 1.89426243e-01 -6.46409094e-01 4.13874358e-01
2.54411757e-01 4.07608777e-01 -8.48226398e-02 -4.00314897e-01
-3.16722661e-01 1.47229600e+00 -7.02225745e-01 3.45997989e-01
-6.10934854e-01 -6.24921381e-01 4.88764912e-01 4.77661461e-01
5.12560308e-01 -4.61965770e-01 -3.70252699e-01 -8.93783152e-01
1.51182726e-01 -7.55043745e-01 -2.75641263e-01 -5.79817176e-01
-4.88901794e-01 7.27750540e-01 -2.41400868e-01 -1.76815653e+00
-3.16968381e-01 -1.91531375e-01 -8.81118551e-02 1.23941028e+00
-1.85874224e+00 -5.80311894e-01 -1.12771206e-01 4.85171974e-01
7.17326701e-01 -2.67527755e-02 7.89478421e-01 8.28732848e-01
1.36432737e-01 3.89792442e-01 3.71364892e-01 -3.39274794e-01
2.95512587e-01 -7.75617301e-01 3.99280563e-02 1.22106516e+00
-5.45647860e-01 2.42489576e-01 1.19329000e+00 -6.98031068e-01
-1.26551294e+00 -9.61517692e-01 1.31942272e+00 5.66721678e-01
4.06455435e-02 1.23863794e-01 -1.10128236e+00 1.73569262e-01
5.11353254e-01 -2.26313889e-01 4.72486347e-01 -1.07891047e+00
1.30362902e-03 -1.10902525e-02 -1.77077091e+00 4.14325684e-01
5.34690142e-01 -3.88448775e-01 -3.55318755e-01 -3.04418534e-01
4.42280293e-01 -4.46855605e-01 -9.89674807e-01 3.18896413e-01
6.10292375e-01 -1.42763650e+00 1.07744133e+00 4.73439485e-01
6.32877231e-01 -5.38335025e-01 -3.91559869e-01 -1.01240289e+00
-3.57467651e-01 -9.79794085e-01 -1.37338817e-01 1.09685278e+00
2.47647449e-01 -4.81344491e-01 7.61668384e-01 3.32249433e-01
6.23105839e-02 -5.11928022e-01 -1.23247921e+00 -6.02690995e-01
-3.54656607e-01 -3.62733424e-01 4.36751485e-01 5.10991871e-01
-8.39312077e-02 -4.32338685e-01 -9.24466372e-01 3.44031155e-01
1.12608683e+00 -4.74927753e-01 2.74108112e-01 -7.15745270e-01
-3.82771701e-01 -3.71629670e-02 -8.45217228e-01 -1.24710691e+00
-6.21958196e-01 -3.56506765e-01 -1.18214481e-01 -1.44998741e+00
-3.14288408e-01 -2.66462982e-01 1.78826362e-01 -4.24337447e-01
-1.75408304e-01 7.21255720e-01 2.71953255e-01 5.25206588e-02
3.63187969e-01 5.88247478e-01 1.08361137e+00 1.65944919e-02
-9.73145217e-02 2.73945957e-01 -2.51821846e-01 5.75843155e-01
7.37508893e-01 -4.26652640e-01 -2.95622945e-01 -2.38376349e-01
1.70973599e-01 5.37470877e-01 5.24168670e-01 -1.35404301e+00
2.59815991e-01 2.79126972e-01 3.82258326e-01 -5.28245270e-01
4.73320305e-01 -1.41442442e+00 8.48347127e-01 8.39749217e-01
-5.57986870e-02 -8.13228078e-03 3.44341367e-01 7.58136511e-01
-5.55085719e-01 -5.59392750e-01 9.78827715e-01 -4.38917056e-02
-8.35925102e-01 -1.37702242e-01 -1.04534125e+00 -6.12580299e-01
1.09722733e+00 -9.68922317e-01 8.59984234e-02 -4.35451776e-01
-8.71874154e-01 -6.47189796e-01 5.47896862e-01 -1.15943819e-01
1.06810808e+00 -1.17138350e+00 -8.49952519e-01 3.93588454e-01
-2.98880041e-01 -4.55268830e-01 7.06502974e-01 8.39101851e-01
-1.07566631e+00 2.33290836e-01 -2.64951676e-01 -3.48777443e-01
-1.52183759e+00 6.49287462e-01 3.89097214e-01 -1.39830336e-02
-8.35502982e-01 3.62790078e-01 -4.08501923e-01 7.36696243e-01
-4.92080785e-02 -2.47377604e-02 -2.78395861e-01 -2.70922691e-01
9.50642347e-01 7.17145681e-01 1.91765323e-01 -8.08913946e-01
-2.99988776e-01 8.29362214e-01 4.44300711e-01 -2.63500929e-01
1.03772867e+00 -6.51831329e-01 -2.13248320e-02 1.05091467e-01
1.33632386e+00 6.15114868e-01 -1.47726405e+00 1.03048161e-01
-1.71608090e-01 -9.80901599e-01 2.99724787e-01 -6.28238261e-01
-8.32692742e-01 6.73240304e-01 8.21943045e-01 1.25470117e-01
1.66132677e+00 -8.06999564e-01 8.58294904e-01 -6.61244392e-01
4.81166571e-01 -5.26160181e-01 -5.45594335e-01 -4.47489284e-02
9.95367587e-01 -7.22676694e-01 3.06279540e-01 -7.05981135e-01
-2.47876883e-01 1.36256731e+00 -4.83487606e-01 -2.08122373e-01
6.36255443e-01 4.11674708e-01 -2.35760018e-01 5.76699555e-01
-4.58261281e-01 1.20504461e-01 2.78502673e-01 6.54264271e-01
4.85425591e-01 -1.73200965e-01 -1.12522900e+00 9.59880501e-02
2.96021640e-01 7.12109566e-01 8.39994907e-01 9.35763001e-01
-5.42911947e-01 -1.11574626e+00 -8.08791399e-01 -2.00831309e-01
-8.57018292e-01 -2.42042094e-01 4.29601073e-01 3.87715191e-01
2.46204391e-01 1.14456522e+00 -2.86608875e-01 -1.73910707e-01
2.65663773e-01 -2.24841163e-01 4.71755445e-01 2.74483919e-01
-6.44648492e-01 4.51454520e-02 2.66434938e-01 -4.52717632e-01
-5.80304205e-01 -4.77553546e-01 -1.02956843e+00 -5.18860459e-01
-9.45590883e-02 1.68698579e-01 1.08033931e+00 6.05837524e-01
3.39655131e-01 6.59222603e-01 8.25791538e-01 -9.42553520e-01
-1.27956167e-01 -5.59764326e-01 -6.22443020e-01 1.73215464e-01
1.19724131e+00 -3.00381154e-01 -6.20648265e-01 4.35272872e-01] | [11.32832145690918, -2.2104036808013916] |
cea292fa-e394-40a1-b91b-0b1a3110c312 | a-stacked-deep-convolutional-neural-network | 2111.12689 | null | https://arxiv.org/abs/2111.12689v1 | https://arxiv.org/pdf/2111.12689v1.pdf | A stacked deep convolutional neural network to predict the remaining useful life of a turbofan engine | This paper presents the data-driven techniques and methodologies used to predict the remaining useful life (RUL) of a fleet of aircraft engines that can suffer failures of diverse nature. The solution presented is based on two Deep Convolutional Neural Networks (DCNN) stacked in two levels. The first DCNN is used to extract a low-dimensional feature vector using the normalized raw data as input. The second DCNN ingests a list of vectors taken from the former DCNN and estimates the RUL. Model selection was carried out by means of Bayesian optimization using a repeated random subsampling validation approach. The proposed methodology was ranked in the third place of the 2021 PHM Conference Data Challenge. | ['Joaquin Borrego-Diaz', 'Juan Galan-Paez', 'David Solis-Martin'] | 2021-11-24 | null | null | null | null | ['remaining-useful-lifetime-estimation'] | ['time-series'] | [-8.29327293e-03 -2.27412969e-01 1.64289534e-01 -4.74752992e-01
-2.11850017e-01 -6.75808564e-02 4.19323146e-01 3.33499573e-02
-3.84992421e-01 9.19803977e-01 -6.56506270e-02 -4.14897174e-01
-1.10538208e+00 -8.09131324e-01 -4.02012259e-01 -8.90564322e-01
-4.04321164e-01 6.17849648e-01 4.75473814e-02 -1.39671803e-01
2.27764815e-01 1.26184571e+00 -1.84302878e+00 2.80532122e-01
1.60184771e-01 1.66716576e+00 -2.49263123e-01 7.42914975e-01
4.19749826e-01 8.83147657e-01 -8.56481552e-01 2.16442168e-01
5.22119582e-01 6.32651821e-02 -5.23357034e-01 -2.41695374e-01
-1.07083045e-01 -2.45997444e-01 -3.52968544e-01 4.80871350e-01
4.64975119e-01 5.68294883e-01 8.51422071e-01 -1.49045110e+00
3.09375882e-01 3.69328916e-01 2.58470565e-01 4.71920192e-01
-2.86248684e-01 2.85934564e-02 5.53043902e-01 -9.67335999e-01
2.75073409e-01 1.14054644e+00 6.87824726e-01 1.94291130e-01
-9.17042136e-01 -4.57605958e-01 -5.22698760e-01 2.18632400e-01
-1.31664264e+00 -2.71473765e-01 8.40816855e-01 -7.08797574e-01
1.33227265e+00 1.73734888e-01 7.01418221e-01 9.35837030e-01
1.02060425e+00 1.05120435e-01 9.31526542e-01 -2.61504620e-01
6.71441019e-01 -1.65291533e-01 3.12728822e-01 5.27447581e-01
6.29415512e-01 9.35041547e-01 -5.05997062e-01 -2.20863581e-01
3.96084189e-01 9.29051265e-03 3.33198369e-01 -3.30256343e-01
-7.51459420e-01 8.32802176e-01 2.22467750e-01 2.60825813e-01
-8.12317908e-01 3.37364376e-01 6.46413267e-01 3.36544484e-01
2.09965110e-01 7.48893082e-01 -9.33393419e-01 1.53918087e-01
-1.19742703e+00 3.62207472e-01 8.63860548e-01 5.47312200e-01
3.27119738e-01 4.47720945e-01 -3.54368985e-01 2.69948274e-01
5.43321967e-01 1.40516341e-01 1.98725075e-01 -9.68629241e-01
3.54570225e-02 4.76915210e-01 2.67497033e-01 -8.32544744e-01
-7.31226623e-01 -7.14783490e-01 -8.38471055e-01 7.31000006e-01
-5.89166582e-02 -5.62603116e-01 -1.17091727e+00 8.94754827e-01
-1.32518023e-01 -5.41190445e-01 -7.23320395e-02 5.74857831e-01
5.38860559e-01 6.62234902e-01 -1.26809791e-01 -3.09014291e-01
7.26666451e-01 -4.73343015e-01 -6.21087670e-01 2.37861469e-01
4.52226289e-02 -2.45093465e-01 1.04033306e-01 1.16295338e+00
-9.11039174e-01 -8.02554071e-01 -1.75216782e+00 6.05516315e-01
-6.62350237e-01 3.39583635e-01 1.48248911e-01 6.13807559e-01
-9.84177053e-01 1.10269856e+00 -7.64830470e-01 1.26267061e-01
4.04487789e-01 7.67320991e-01 -2.37665787e-01 -9.42252129e-02
-1.27443552e+00 1.05630279e+00 8.43732774e-01 4.48548734e-01
-1.70389700e+00 -3.59706372e-01 -5.66027641e-01 1.87884104e-02
2.08878443e-01 -4.30033714e-01 8.36773992e-01 -5.12739897e-01
-1.33100057e+00 4.44984324e-02 4.63160783e-01 -6.45575523e-01
5.01375496e-01 -2.61617184e-01 -7.64639676e-01 -1.84270650e-01
-3.68901253e-01 2.67794073e-01 8.29430640e-01 -1.18391514e+00
-7.79961526e-01 -1.68068439e-01 -2.69555897e-01 -4.93398041e-01
-1.19307213e-01 -1.00157268e-01 8.02924633e-02 -4.20951456e-01
-3.86493802e-02 -5.96318245e-01 -2.38315254e-01 -7.78014600e-01
-5.57467580e-01 -3.53227854e-01 6.50690258e-01 -6.84955955e-01
1.43402541e+00 -1.70348322e+00 3.06055486e-01 5.36783755e-01
5.03040403e-02 3.27496789e-02 3.41691792e-01 8.40825200e-01
-3.44655603e-01 3.56101096e-02 -9.06198770e-02 -7.47241685e-03
-6.60669878e-02 3.57168056e-02 2.96354257e-02 5.31664789e-01
4.65135396e-01 1.30450010e-01 -4.76762295e-01 -2.18344748e-01
3.47582042e-01 2.94662535e-01 -1.80222327e-03 4.18674916e-01
-1.26203790e-01 9.63700488e-02 -3.44053447e-01 9.05269444e-01
5.64066768e-01 4.55810487e-01 -3.71143520e-02 -5.08079886e-01
-2.84990937e-01 -2.92274684e-01 -1.03830266e+00 1.02817106e+00
-3.87576848e-01 7.00649142e-01 -3.05122048e-01 -9.47848320e-01
1.37201071e+00 3.58497500e-01 7.81582952e-01 -3.23478103e-01
6.89534068e-01 2.08179921e-01 1.31437391e-01 -5.76783061e-01
4.16110516e-01 -3.11439961e-01 -3.63091290e-01 -1.09905258e-01
4.58395571e-01 8.71295780e-02 3.48777503e-01 -2.74311155e-01
1.40891635e+00 2.52371311e-01 -9.75224376e-03 -4.86869156e-01
5.44150412e-01 2.63051242e-02 9.53717411e-01 6.50674105e-01
-1.93607897e-01 2.40353301e-01 7.98929811e-01 -1.15070236e+00
-1.08054960e+00 -1.04076552e+00 -1.40796751e-01 4.60707694e-01
-5.39350092e-01 8.20161309e-03 -3.56447905e-01 -7.02384949e-01
2.06146896e-01 9.34223533e-01 -8.51301551e-01 -4.82991099e-01
-2.99875438e-01 -7.21965849e-01 4.40046579e-01 6.20313883e-01
9.10665691e-02 -1.08911622e+00 -1.03226483e+00 2.35459253e-01
7.31652558e-01 -5.33491910e-01 7.12469101e-01 8.39830935e-01
-8.06859195e-01 -1.17866039e+00 -1.55605793e-01 -3.21106553e-01
4.47581291e-01 -6.60725236e-01 1.25964367e+00 6.34349091e-03
-6.39722347e-01 -2.34350506e-02 -2.28655025e-01 -9.29338098e-01
-4.42268431e-01 9.62980539e-02 4.92308915e-01 -9.63631123e-02
3.98193032e-01 -2.06951857e-01 -3.59196216e-01 1.94982678e-01
-8.16324770e-01 -6.04036510e-01 8.58832300e-01 9.17826533e-01
5.05860865e-01 1.01868153e+00 6.88552976e-01 -4.16731268e-01
8.65820527e-01 -7.17077971e-01 -8.78333688e-01 1.21684335e-01
-8.85308743e-01 5.00183851e-02 7.70769119e-01 -1.39901843e-02
-7.07876086e-01 1.81411654e-01 -2.08501536e-02 -7.25033820e-01
-3.50055844e-01 6.90881968e-01 -1.84704497e-01 3.24465483e-01
4.22172278e-01 -3.60729426e-01 2.19105836e-02 -5.64563036e-01
-2.28502527e-01 5.66714525e-01 3.73054415e-01 -4.01754797e-01
5.80978751e-01 -9.68127847e-02 6.92901254e-01 -5.84025085e-01
-4.24679637e-01 -2.51819380e-02 -8.35994899e-01 -1.01895404e+00
7.16314554e-01 -4.26793069e-01 -7.18887091e-01 3.79992515e-01
-8.33479047e-01 3.03732697e-02 -4.70409125e-01 6.49588108e-01
-5.72183013e-01 -4.16447282e-01 -2.17045844e-01 -1.18346989e+00
-3.38611782e-01 -1.11982799e+00 4.53220874e-01 1.91238314e-01
6.15638308e-02 -4.83379483e-01 1.10884748e-01 -5.62422648e-02
4.51349765e-01 7.15259850e-01 1.12701356e+00 -1.18536329e+00
-1.36140361e-01 -1.12998998e+00 3.36104572e-01 1.13366139e+00
-9.15347189e-02 3.88859212e-01 -8.80529344e-01 -6.68655813e-01
-1.60727929e-02 -3.16580951e-01 7.56037593e-01 6.98354661e-01
1.30173123e+00 1.95254236e-01 -1.07980445e-01 1.99919060e-01
1.63512039e+00 5.92758775e-01 4.60048199e-01 4.03806150e-01
2.47774318e-01 7.93990254e-01 8.65614414e-01 5.73494613e-01
-6.15897417e-01 1.39097542e-01 1.01877582e+00 2.69231230e-01
3.21261585e-01 3.79365832e-01 2.04212248e-01 6.17107987e-01
-1.57046765e-01 -4.44303215e-01 -1.26844382e+00 5.91921806e-01
-1.57539940e+00 -6.74056411e-01 7.44078085e-02 2.19298506e+00
-2.26656385e-02 8.18211079e-01 1.22511230e-01 5.33816218e-01
3.41391921e-01 -5.29192574e-02 -5.47168851e-01 -6.18490696e-01
1.49367124e-01 3.03745598e-01 8.11126351e-01 1.12889856e-01
-1.16346490e+00 1.00534119e-01 6.75596046e+00 6.74083292e-01
-5.41845798e-01 -2.00946644e-01 7.22764194e-01 -5.35233855e-01
1.75877199e-01 -7.22340196e-02 -8.26619804e-01 3.93422991e-01
1.67506289e+00 1.08239107e-01 2.67340153e-01 6.45445645e-01
2.22619325e-01 -2.64469385e-01 -1.07330430e+00 3.84012192e-01
7.93609023e-03 -1.24787474e+00 -1.10251166e-01 3.00275624e-01
6.81113422e-01 4.33958396e-02 -3.01408637e-02 3.92341048e-01
2.29430899e-01 -1.33442843e+00 6.45752490e-01 1.54256511e+00
4.99477983e-01 -1.61762190e+00 1.58609700e+00 1.00643590e-01
-6.51991785e-01 -8.31954420e-01 -4.11050946e-01 1.18695289e-01
5.80521375e-02 9.78168309e-01 -7.12017417e-01 9.49057996e-01
1.05682278e+00 4.05371666e-01 -5.36183536e-01 1.11280024e+00
4.64764744e-01 5.73710024e-01 -1.17301993e-01 -1.00903556e-01
1.01086564e-01 1.10605434e-01 6.84665620e-01 7.86627054e-01
5.69426537e-01 -6.66385353e-01 3.28366319e-03 6.15291536e-01
3.18807960e-02 -4.64551181e-01 -6.11068189e-01 -1.28202721e-01
6.07312500e-01 1.20967686e+00 -6.15667760e-01 -1.28790185e-01
8.49770755e-02 1.06444426e-01 -2.70653486e-01 1.42705783e-01
-6.01017833e-01 -7.49491155e-01 4.11287218e-01 3.96568030e-02
3.02175671e-01 -2.82776445e-01 -1.01935595e-01 -8.90095234e-02
-2.65360057e-01 -5.66470504e-01 1.92292020e-01 -6.33028805e-01
-1.35523856e+00 1.01943219e+00 4.94585335e-01 -1.39957047e+00
-5.08084059e-01 -1.33240485e+00 -3.77247363e-01 1.03815937e+00
-8.72187912e-01 -6.87083066e-01 -8.70695338e-02 2.62200832e-03
8.19856524e-01 -1.14605689e+00 5.83690941e-01 1.23848349e-01
-8.63468468e-01 -2.51361039e-02 4.35962737e-01 -1.96105540e-01
2.55238682e-01 -1.12015986e+00 5.29877171e-02 7.95345604e-01
-8.02481055e-01 3.31693411e-01 9.67900157e-01 -9.25841093e-01
-1.30884957e+00 -1.26192415e+00 6.46664143e-01 -4.26112175e-01
4.49922442e-01 1.64579879e-02 -3.76112908e-01 2.33659551e-01
1.85962990e-01 -2.44910438e-02 6.02571368e-01 -2.29542062e-01
5.35734951e-01 -5.15065372e-01 -1.31302774e+00 -1.59748569e-01
3.87174487e-02 -4.43235338e-02 -3.07122082e-01 9.35956463e-02
4.32131767e-01 7.50874728e-02 -1.40342319e+00 8.28334093e-01
7.19122350e-01 -1.04150534e+00 9.76427257e-01 -9.68381107e-01
3.55939537e-01 -2.35064134e-01 -4.65522975e-01 -1.19386995e+00
-4.31783617e-01 -1.80283457e-01 -4.52469975e-01 9.92162526e-01
2.29706451e-01 -2.67500699e-01 4.53462213e-01 3.48348588e-01
-2.26655930e-01 -1.30075943e+00 -1.30339026e+00 -1.04681480e+00
-6.63037598e-02 -3.72405231e-01 7.41351962e-01 1.17933549e-01
-9.92283940e-01 -1.78540275e-01 -3.85810554e-01 2.54237264e-01
6.98409319e-01 -6.32134080e-01 6.53396100e-02 -1.84832585e+00
2.31300429e-01 -3.48160237e-01 -5.66338718e-01 3.69999886e-01
-1.22415368e-02 -3.58111799e-01 4.34735507e-01 -1.34327316e+00
-3.34308058e-01 -1.32216364e-01 -1.20232427e+00 -6.91482872e-02
6.21847272e-01 -1.93627238e-01 -1.47108167e-01 -6.66257069e-02
-1.47242457e-01 4.73515064e-01 2.85339177e-01 -1.81646571e-01
1.03419028e-01 3.89117151e-01 -5.55195333e-03 6.10462010e-01
1.10890388e+00 -6.70616508e-01 -4.70945388e-01 7.45154321e-02
4.70362127e-01 1.76935881e-01 5.52357018e-01 -1.69867516e+00
5.95346205e-02 -1.53880447e-01 1.01642334e+00 -1.28628075e+00
2.53193080e-01 -1.10053575e+00 6.68825805e-01 6.81332350e-01
-2.32679367e-01 5.05774379e-01 3.62966716e-01 6.74871683e-01
-1.42805591e-01 -5.66750705e-01 5.67945957e-01 -3.58301885e-02
-5.33335268e-01 4.21993360e-02 -7.22886741e-01 -7.52352178e-01
9.59527969e-01 -2.76642442e-01 1.54555991e-01 2.39407554e-01
-9.40309286e-01 1.65816158e-01 7.30703473e-02 4.75883216e-01
8.67879450e-01 -1.41834211e+00 -6.74632728e-01 2.71336198e-01
-1.35907263e-01 -2.95896769e-01 2.96550632e-01 5.82017064e-01
-6.71365917e-01 7.20940530e-01 -7.56042659e-01 -1.65167704e-01
-7.05703735e-01 5.64541101e-01 7.67968178e-01 -4.58670318e-01
-9.23023373e-02 9.05580640e-01 -8.28498900e-01 -2.73010463e-01
2.83265203e-01 -1.08620010e-01 -5.55343986e-01 3.95865142e-01
1.15958698e-01 1.07881236e+00 7.62145758e-01 -4.86151934e-01
-5.18752635e-01 6.76184818e-02 3.03886116e-01 -2.07511991e-01
1.61060143e+00 3.57464284e-01 -2.82216460e-01 8.44884992e-01
1.19304192e+00 -7.02800155e-01 -1.19437540e+00 4.87889707e-01
2.60812223e-01 -1.41818151e-01 7.58192778e-01 -1.22704244e+00
-1.08815253e+00 7.07641363e-01 1.05369413e+00 4.10678595e-01
1.37844217e+00 -5.79169869e-01 2.84273237e-01 5.29254436e-01
3.11287582e-01 -1.53772950e+00 -2.09335595e-01 3.65493178e-01
1.16261292e+00 -9.00168478e-01 3.69831204e-01 5.49082220e-01
-3.22490513e-01 1.66255522e+00 5.38661897e-01 -3.37905914e-01
1.02603054e+00 2.06512749e-01 -4.33244526e-01 -6.57092690e-01
-1.09890211e+00 2.39144534e-01 4.34585869e-01 4.06841338e-01
1.46456450e-01 1.03772216e-01 -2.82887250e-01 8.43653023e-01
1.56871185e-01 2.31535316e-01 3.56876969e-01 1.23579419e+00
-4.27745491e-01 -8.15658689e-01 -4.74852473e-01 1.05846334e+00
-3.69868785e-01 3.42665792e-01 -2.83686459e-01 8.24848831e-01
5.90481162e-01 1.11416197e+00 4.23612595e-02 -1.04001975e+00
5.27882814e-01 2.23596036e-01 -2.03719120e-02 -1.60072073e-01
-6.60579324e-01 -2.89277047e-01 3.37615579e-01 -5.92701316e-01
-3.18215936e-02 -7.63158143e-01 -6.74399316e-01 6.01181760e-02
-1.41207546e-01 5.69738671e-02 1.13195753e+00 7.69230664e-01
9.84079242e-02 1.38736844e+00 9.44778025e-01 -1.19444942e+00
-6.43874645e-01 -1.17101097e+00 -8.55306268e-01 -3.78273070e-01
6.04781508e-01 -1.12615180e+00 -5.25591195e-01 -2.76896000e-01] | [6.726168632507324, 2.4764046669006348] |
75bd02a6-2204-415e-88d8-ba6041d4ced7 | where-are-the-facts-searching-for-fact | 2010.03159 | null | https://arxiv.org/abs/2010.03159v1 | https://arxiv.org/pdf/2010.03159v1.pdf | Where Are the Facts? Searching for Fact-checked Information to Alleviate the Spread of Fake News | Although many fact-checking systems have been developed in academia and industry, fake news is still proliferating on social media. These systems mostly focus on fact-checking but usually neglect online users who are the main drivers of the spread of misinformation. How can we use fact-checked information to improve users' consciousness of fake news to which they are exposed? How can we stop users from spreading fake news? To tackle these questions, we propose a novel framework to search for fact-checking articles, which address the content of an original tweet (that may contain misinformation) posted by online users. The search can directly warn fake news posters and online users (e.g. the posters' followers) about misinformation, discourage them from spreading fake news, and scale up verified content on social media. Our framework uses both text and images to search for fact-checking articles, and achieves promising results on real-world datasets. Our code and datasets are released at https://github.com/nguyenvo09/EMNLP2020. | ['Kyumin Lee', 'Nguyen Vo'] | 2020-10-07 | null | https://aclanthology.org/2020.emnlp-main.621 | https://aclanthology.org/2020.emnlp-main.621.pdf | emnlp-2020-11 | ['image-similarity-search', 'ad-hoc-information-retrieval'] | ['computer-vision', 'natural-language-processing'] | [-2.62456447e-01 2.22346276e-01 -6.19247019e-01 2.13050917e-01
-4.90333736e-01 -8.81631196e-01 8.68848205e-01 3.38571757e-01
-1.58721313e-01 6.93120420e-01 3.11868578e-01 -3.76812786e-01
5.29954791e-01 -1.16789150e+00 -6.85977519e-01 -1.04135402e-01
3.76739442e-01 1.89622551e-01 8.33110809e-01 -6.30891323e-01
7.48664021e-01 -1.11418806e-01 -1.01032436e+00 6.08514667e-01
9.52571929e-01 5.72285652e-01 -7.60581642e-02 1.27497595e-02
-3.09146821e-01 1.02392638e+00 -8.65108669e-01 -8.30221653e-01
1.00374706e-01 -5.64476848e-01 -5.38605928e-01 -3.55376080e-02
6.37622893e-01 -5.69751561e-01 -7.37587631e-01 1.82999527e+00
-6.57500990e-04 -5.45239806e-01 7.82048982e-03 -1.29611361e+00
-1.09401214e+00 9.27748978e-01 -7.53967166e-01 5.21678448e-01
4.21909273e-01 -2.60976572e-02 6.21811986e-01 -7.48551011e-01
1.08302641e+00 1.34108901e+00 6.18561625e-01 4.06064987e-01
-5.63004434e-01 -8.67826879e-01 -2.36500978e-01 3.50285806e-02
-1.31944633e+00 -3.03775042e-01 7.47144163e-01 -5.97381473e-01
-2.03428324e-02 4.78520811e-01 7.43856311e-01 1.46232605e+00
3.51758271e-01 9.29507196e-01 1.29229796e+00 5.47053218e-02
-4.93414793e-03 5.99585414e-01 1.16654158e-01 7.21971035e-01
7.78837740e-01 -4.17434312e-02 -7.54088819e-01 -5.89255035e-01
5.74962914e-01 2.52444237e-01 -4.42739457e-01 2.49333113e-01
-1.26808202e+00 1.25177109e+00 6.91708982e-01 5.61110675e-01
-2.21363932e-01 -1.94396183e-01 6.12282455e-01 4.40525562e-01
1.12847114e+00 4.56233829e-01 7.11119622e-02 2.43145619e-02
-7.59379566e-01 4.15364593e-01 7.75353789e-01 7.43604958e-01
5.86661994e-01 -5.19646943e-01 -1.17168359e-01 6.26961708e-01
9.19620469e-02 1.01660669e+00 4.73454624e-01 -3.89933407e-01
6.08838737e-01 8.59399378e-01 6.43976569e-01 -2.18071389e+00
1.93966657e-01 -4.77085859e-01 -5.29743314e-01 -3.40799481e-01
4.37636375e-01 -1.55865803e-01 -3.84205043e-01 9.12330329e-01
4.75534111e-01 1.38825446e-01 -6.32320225e-01 1.26022840e+00
7.43899643e-01 7.30746627e-01 -4.24899668e-01 -1.78394064e-01
1.51148021e+00 -1.01699996e+00 -1.13416874e+00 -3.24613094e-01
7.23319590e-01 -1.04509163e+00 8.28952014e-01 4.07964766e-01
-8.12701643e-01 6.72537684e-02 -9.18050408e-01 1.05458135e-02
-9.30024981e-01 -1.52509421e-01 3.03784460e-01 6.08392894e-01
-4.72646087e-01 6.92822337e-01 -3.84267390e-01 -3.16322863e-01
8.57011497e-01 -3.95558864e-01 3.70717309e-02 -1.94766864e-01
-1.61744797e+00 7.23448575e-01 2.25564405e-01 -7.62385279e-02
-5.98627269e-01 -2.13790894e-01 -1.67524144e-01 -3.78370196e-01
9.40626562e-01 -3.15080106e-01 1.04655290e+00 -1.18487990e+00
-9.71363187e-01 8.89540553e-01 1.23330422e-01 -4.21002716e-01
1.08199918e+00 -3.01397532e-01 -8.34224045e-01 4.27302510e-01
7.01380610e-01 -1.14972293e-01 1.23093343e+00 -1.31801891e+00
-6.62489951e-01 -2.22890377e-01 -1.26826406e-01 -3.54739726e-01
-4.55528319e-01 1.87376767e-01 -1.84444085e-01 -8.61038208e-01
4.35352594e-01 -9.90532875e-01 2.15109140e-01 8.16194341e-02
-9.08859551e-01 1.92105528e-02 1.49109185e+00 -7.70878613e-01
1.38178062e+00 -2.01293540e+00 -6.93308175e-01 1.68552533e-01
7.43634522e-01 7.18374848e-01 1.36949807e-01 7.92865634e-01
6.42657399e-01 7.22341120e-01 2.15644315e-01 1.98951527e-01
-3.85500431e-01 -2.01479912e-01 -8.07175934e-01 9.31514204e-01
-3.43519270e-01 9.62776721e-01 -1.50006461e+00 -3.63230169e-01
-1.63965628e-01 2.49461278e-01 -2.70534456e-01 -2.99065858e-01
-4.04313117e-01 5.01690686e-01 -1.01126158e+00 5.33234954e-01
1.01930475e+00 -8.00788343e-01 3.24750282e-02 1.08356915e-01
-2.93964565e-01 7.86561549e-01 -5.13221264e-01 6.12235904e-01
-1.57061256e-02 8.73637736e-01 1.53695002e-01 -2.65646756e-01
5.66396415e-01 1.53321698e-02 1.60738349e-01 -6.02784276e-01
4.84002531e-01 4.36277300e-01 -5.45229733e-01 -6.04685843e-01
7.15614796e-01 1.39497563e-01 -7.47899339e-02 9.34564769e-01
-6.57454371e-01 1.23167410e-01 -2.59369519e-02 8.14214051e-01
1.12811089e+00 -4.88908082e-01 9.91389304e-02 -6.94160461e-02
2.47785479e-01 6.08806729e-01 1.85602587e-02 1.12406898e+00
-4.04532194e-01 6.21247664e-02 4.87920433e-01 -4.90972668e-01
-9.70722079e-01 -6.30775571e-01 6.26232922e-02 8.88644338e-01
8.78520548e-01 -5.67959845e-01 -6.69339657e-01 -1.10063457e+00
2.34285012e-01 5.74000835e-01 -5.55627346e-01 -7.16568753e-02
-3.73710960e-01 -4.61831242e-01 5.04978538e-01 -6.24109387e-01
1.03725946e+00 -8.77670586e-01 -7.58680105e-02 2.57115394e-01
-9.42334116e-01 -1.10756874e+00 -5.61399579e-01 -9.73554611e-01
-5.66078186e-01 -1.23040831e+00 -5.76128960e-01 -2.83249378e-01
8.02126050e-01 1.18014789e+00 6.45095408e-01 8.80940437e-01
2.57869929e-01 -1.76350757e-01 -7.26500392e-01 -4.59714174e-01
-8.34616482e-01 -4.31182161e-02 -5.58387376e-02 2.41030529e-01
3.00987929e-01 -1.28051057e-01 -7.53996849e-01 7.90963769e-01
-1.10950744e+00 -1.86851025e-02 7.04032406e-02 5.13079226e-01
-8.38111788e-02 2.57077396e-01 4.14396852e-01 -1.37456381e+00
8.46256435e-01 -1.05892754e+00 -6.20248318e-01 -1.15652636e-01
-5.45261741e-01 -5.79737067e-01 3.73124778e-01 -7.12949216e-01
-6.40130341e-01 -6.01404607e-01 2.40783930e-01 -1.82784926e-02
3.33848268e-01 5.30082524e-01 5.89814544e-01 -3.51794392e-01
9.62082505e-01 2.72263229e-01 -7.49242678e-02 -4.53963757e-01
1.51187554e-01 9.05796111e-01 -1.34992534e-02 1.44888639e-01
1.25203216e+00 1.18004298e+00 -7.11734593e-01 -8.76484513e-01
-1.41031384e+00 -6.97134197e-01 8.67610350e-02 -4.56385195e-01
2.43554115e-01 -8.14918399e-01 -7.41748631e-01 6.33524477e-01
-1.48885822e+00 2.29848549e-01 4.31037754e-01 -3.31527740e-02
3.23058605e-01 6.68035448e-01 -8.52864921e-01 -7.01809347e-01
-1.00038134e-01 -5.23466647e-01 6.87791348e-01 -1.10864416e-01
-9.17207375e-02 -8.23704362e-01 1.57157276e-02 7.51170039e-01
7.48826385e-01 1.45829543e-01 1.05674945e-01 -8.73429120e-01
-1.05535972e+00 -8.18778813e-01 -6.92746699e-01 -2.03192830e-02
1.91975757e-01 -1.37697026e-01 -7.16910779e-01 -2.34250858e-01
1.52658254e-01 -1.48864910e-01 7.05821455e-01 -1.07637353e-01
7.27756381e-01 -1.35123801e+00 -8.74022126e-01 -2.40419909e-01
1.18906617e+00 -4.26191717e-01 6.21526182e-01 5.31393230e-01
6.08527601e-01 6.11024201e-01 7.03390300e-01 5.08945942e-01
3.29697102e-01 4.89085615e-01 6.27218664e-01 2.28464171e-01
1.16621353e-01 -9.33798313e-01 4.16171342e-01 9.49069083e-01
1.31197974e-01 -5.60359359e-01 -6.96726680e-01 6.32102787e-01
-1.82987750e+00 -1.30632913e+00 -8.51932228e-01 1.85584533e+00
6.39836729e-01 4.09919739e-01 1.72919497e-01 -9.63959023e-02
1.24664700e+00 5.49684107e-01 5.77121554e-03 1.92919746e-01
-1.02215432e-01 -6.38634026e-01 9.60251749e-01 6.71539962e-01
-1.05940115e+00 1.18565309e+00 5.27206421e+00 1.00516486e+00
-1.27771187e+00 7.12736011e-01 4.16325271e-01 1.66142628e-01
-4.54197437e-01 1.85561135e-01 -8.51090550e-01 1.04852498e+00
6.05004966e-01 -2.05994397e-01 4.42574978e-01 8.05873454e-01
6.25686646e-01 -3.53716820e-01 -1.29813507e-01 6.90732539e-01
1.68228880e-01 -1.92456162e+00 2.52477676e-02 2.62390196e-01
9.56196785e-01 3.07985574e-01 3.06714308e-02 4.89111319e-02
2.73446739e-01 -4.27612334e-01 9.29360092e-01 2.52294123e-01
1.90389633e-01 -1.54877529e-01 8.03524315e-01 7.72259295e-01
-3.53553832e-01 1.34502813e-01 -2.78721899e-01 -2.07703754e-01
4.62005377e-01 1.34840798e+00 -7.97798455e-01 2.28245668e-02
5.50142705e-01 8.91922772e-01 -7.65874684e-01 9.60929215e-01
-5.46547353e-01 7.17050612e-01 -2.03766793e-01 -6.09181285e-01
3.23333770e-01 -1.92578509e-01 1.01065564e+00 1.02431381e+00
2.80173600e-01 4.15964797e-02 -3.45915630e-02 1.01573586e+00
-4.38104242e-01 -5.62297888e-02 -8.76430690e-01 -6.83328748e-01
6.21847034e-01 1.00082183e+00 -9.66432035e-01 -6.75620198e-01
-1.09289885e-01 1.17640710e+00 1.46029800e-01 1.45864651e-01
-9.71846461e-01 -2.08428130e-01 7.22741261e-02 9.98010516e-01
7.50771165e-02 -1.14685811e-01 -8.19342807e-02 -1.62668777e+00
8.30630437e-02 -1.02263713e+00 1.08691320e-01 -7.83603787e-01
-1.66360104e+00 4.04808015e-01 -5.01803458e-01 -1.24319243e+00
5.98944366e-01 -1.29527524e-01 -4.34825152e-01 2.58630663e-01
-1.48126757e+00 -7.92173088e-01 -1.80807710e-01 4.63886410e-01
1.54025719e-01 2.38744646e-01 -6.95854872e-02 4.63311046e-01
-2.31947079e-01 1.50402755e-01 7.51707405e-02 3.98386091e-01
7.11315334e-01 -3.34088236e-01 4.77067620e-01 6.30163848e-01
1.29182428e-01 9.46921766e-01 9.87369955e-01 -1.42713702e+00
-1.18247092e+00 -1.07757580e+00 1.40419412e+00 -8.75988960e-01
1.46028626e+00 -3.90642256e-01 -9.18474615e-01 5.76184332e-01
4.09838334e-02 -7.62260705e-02 1.33220866e-01 -3.07029873e-01
-6.25438154e-01 5.56306243e-01 -1.36223912e+00 8.03503692e-01
1.06846440e+00 -6.71560109e-01 -4.46956307e-01 1.13292205e+00
5.46277404e-01 -3.77456844e-01 -7.94107467e-02 -3.65778238e-01
3.91933113e-01 -1.08378887e+00 6.37209117e-01 -4.57445204e-01
8.44206512e-01 -2.32967094e-01 5.01446605e-01 -1.08826518e+00
-6.00639060e-02 -8.66506577e-01 -2.19996106e-02 8.94970953e-01
3.53059232e-01 -9.75659907e-01 5.84876955e-01 2.28948258e-02
3.79891396e-01 -1.63529545e-01 -7.20689476e-01 -7.11167216e-01
-3.08770090e-01 -4.57644276e-02 2.18511418e-01 1.63198340e+00
2.90208340e-01 1.31757841e-01 -7.50291824e-01 3.21036667e-01
6.78121567e-01 5.74380253e-03 8.04982603e-01 -1.02365673e+00
7.01566860e-02 -2.29352564e-01 -7.13950535e-03 -1.20222604e+00
-5.39493740e-01 -6.51361108e-01 -1.08258203e-01 -1.12568104e+00
3.01124841e-01 -2.00056344e-01 4.50848639e-01 3.40575993e-01
9.23630595e-02 6.17078245e-01 -5.57723641e-02 8.86875570e-01
-8.08119595e-01 1.83483824e-01 1.81532168e+00 -1.99036047e-01
-5.67498570e-03 -3.37713100e-02 -6.39751434e-01 8.92948985e-01
7.46092439e-01 -1.14944911e+00 1.88907072e-01 -4.88795899e-02
9.56073105e-01 -3.22237536e-02 8.80084574e-01 -4.62224722e-01
2.72959173e-01 -3.69490653e-01 -2.77784109e-01 -5.66201687e-01
-4.49746065e-02 -6.64420784e-01 -1.21979915e-01 7.75028110e-01
-1.65355578e-01 -9.09258798e-02 -2.53782779e-01 9.51768219e-01
-2.40462825e-01 -1.51394323e-01 7.53552020e-01 -5.04068375e-01
-3.93162370e-02 1.98747128e-01 -7.95933127e-01 2.05012158e-01
7.94904053e-01 1.12614408e-01 -1.32597625e+00 -8.52353036e-01
-4.07181054e-01 -1.51836621e-02 5.84099352e-01 4.62089002e-01
4.57492411e-01 -1.24143815e+00 -8.03802133e-01 -2.84830451e-01
1.87975273e-01 -7.02847421e-01 8.39313269e-02 1.01233947e+00
-6.36902809e-01 3.01347405e-01 4.93743718e-02 -8.26232806e-02
-9.68414605e-01 6.67695165e-01 3.09475843e-04 5.84124625e-02
-6.03718877e-01 6.76823199e-01 -3.48391742e-01 -2.67772883e-01
-3.14388275e-01 9.24473554e-02 1.72921628e-01 7.54674003e-02
8.92036021e-01 5.34403801e-01 -1.96126223e-01 -8.39661062e-01
-3.10258180e-01 -1.92142621e-01 -2.18232557e-01 7.72001222e-02
1.02545822e+00 -4.81086463e-01 -4.41511422e-01 2.70855516e-01
1.17428136e+00 7.32500076e-01 -4.87290561e-01 -3.73594105e-01
7.95607939e-02 -1.17678440e+00 1.87052518e-01 -8.32178652e-01
-1.07535076e+00 4.08442676e-01 -1.62971944e-01 1.10693729e+00
1.10413499e-01 2.06946969e-01 1.41129148e+00 2.23201618e-01
6.69053674e-01 -1.01755679e+00 4.77092892e-01 4.27477032e-01
8.66509140e-01 -1.56466925e+00 3.73380125e-01 -9.72739756e-01
-5.14551222e-01 8.53794217e-01 1.76152080e-01 -3.94736469e-01
6.64280713e-01 -3.65733922e-01 1.80736989e-01 -8.21290135e-01
-2.21654207e-01 2.58840352e-01 4.61418107e-02 1.03095211e-01
7.88122192e-02 1.96541264e-03 -7.87711024e-01 2.89193243e-01
-1.61734208e-01 7.51286075e-02 1.09875083e+00 8.77060473e-01
-8.86061668e-01 -7.45117843e-01 -5.77848971e-01 4.47977006e-01
-8.37459147e-01 -1.17392287e-01 -9.32785571e-01 7.01291740e-01
1.34310231e-01 1.09422505e+00 -4.67008054e-01 -4.22842175e-01
-1.30745143e-01 -4.81238633e-01 -1.18151926e-01 -5.88132441e-01
-5.99405289e-01 -1.12432785e-01 3.95212650e-01 -5.68163991e-01
-3.52639645e-01 -3.75031233e-01 -6.59608066e-01 -1.00162947e+00
-8.08411658e-01 2.87720293e-01 7.07096934e-01 9.22844708e-01
4.48090017e-01 -3.78718287e-01 7.04264998e-01 -2.05813512e-01
-4.02157158e-01 -8.54207873e-01 -4.37672883e-01 5.32341719e-01
6.11481249e-01 -5.70400238e-01 -9.70301092e-01 -1.67122126e-01] | [8.136796951293945, 10.247846603393555] |
548759a8-9444-4e10-8948-cd8db3d38c2e | on-robust-face-recognition-via-sparse | 1303.01624 | null | http://arxiv.org/abs/1303.1624v1 | http://arxiv.org/pdf/1303.1624v1.pdf | On Robust Face Recognition via Sparse Encoding: the Good, the Bad, and the Ugly | In the field of face recognition, Sparse Representation (SR) has received
considerable attention during the past few years. Most of the relevant
literature focuses on holistic descriptors in closed-set identification
applications. The underlying assumption in SR-based methods is that each class
in the gallery has sufficient samples and the query lies on the subspace
spanned by the gallery of the same class. Unfortunately, such assumption is
easily violated in the more challenging face verification scenario, where an
algorithm is required to determine if two faces (where one or both have not
been seen before) belong to the same person. In this paper, we first discuss
why previous attempts with SR might not be applicable to verification problems.
We then propose an alternative approach to face verification via SR.
Specifically, we propose to use explicit SR encoding on local image patches
rather than the entire face. The obtained sparse signals are pooled via
averaging to form multiple region descriptors, which are then concatenated to
form an overall face descriptor. Due to the deliberate loss spatial relations
within each region (caused by averaging), the resulting descriptor is robust to
misalignment & various image deformations. Within the proposed framework, we
evaluate several SR encoding techniques: l1-minimisation, Sparse Autoencoder
Neural Network (SANN), and an implicit probabilistic technique based on
Gaussian Mixture Models. Thorough experiments on AR, FERET, exYaleB, BANCA and
ChokePoint datasets show that the proposed local SR approach obtains
considerably better and more robust performance than several previous
state-of-the-art holistic SR methods, in both verification and closed-set
identification problems. The experiments also show that l1-minimisation based
encoding has a considerably higher computational than the other techniques, but
leads to higher recognition rates. | ['Conrad Sanderson', 'Yongkang Wong', 'Mehrtash T. Harandi'] | 2013-03-07 | null | null | null | null | ['robust-face-recognition'] | ['computer-vision'] | [ 6.06864870e-01 -2.47448027e-01 1.21062417e-02 -3.33043724e-01
-7.86109686e-01 -3.16817552e-01 5.97630858e-01 -2.36381948e-01
1.14309601e-01 5.91574132e-01 -3.99469733e-02 2.68049985e-01
-5.17289162e-01 -6.09531760e-01 -3.59156698e-01 -1.13109720e+00
9.11651701e-02 1.62308916e-01 -2.67019391e-01 -7.32211173e-02
3.38250905e-01 9.87161577e-01 -1.98108709e+00 5.60363121e-02
4.24231827e-01 1.27019858e+00 -2.39012301e-01 1.64515674e-01
7.09227100e-02 5.04465222e-01 -6.30938530e-01 -3.01451713e-01
3.92067879e-01 -5.48286498e-01 -5.39623857e-01 2.66838074e-01
6.96383417e-01 7.37530738e-02 -2.30951563e-01 1.36634600e+00
7.14641392e-01 2.59251088e-01 8.40596318e-01 -1.17006779e+00
-6.61595583e-01 2.95452923e-01 -8.35983217e-01 7.50047937e-02
6.22941554e-01 -2.96821743e-01 5.55800676e-01 -1.06849933e+00
4.83169466e-01 1.37944925e+00 8.18201721e-01 5.03628433e-01
-1.34156084e+00 -8.76806557e-01 -2.29576990e-01 2.44736373e-01
-2.05667281e+00 -9.46492076e-01 9.92939472e-01 -2.50993043e-01
4.56556648e-01 4.00403976e-01 2.22887397e-01 9.18741643e-01
-1.49194077e-01 3.75343502e-01 1.32417786e+00 -6.07482612e-01
8.86945352e-02 2.38969371e-01 1.54861629e-01 5.96332848e-01
3.06389153e-01 2.70720512e-01 -4.60335433e-01 -3.39283168e-01
7.00070918e-01 6.70830309e-02 -5.17316461e-01 -4.08641517e-01
-8.89162421e-01 8.42140615e-01 1.28172383e-01 7.03752398e-01
-4.36073959e-01 -3.05942655e-01 1.21951945e-01 2.34191090e-01
1.32328764e-01 -4.97456789e-02 5.91557212e-02 3.20594579e-01
-1.42384028e+00 1.75921544e-01 7.77712941e-01 6.38312101e-01
7.81104028e-01 2.73964852e-01 -6.94809109e-02 1.08420861e+00
4.71424043e-01 5.96420884e-01 5.02880573e-01 -6.82418466e-01
1.13324061e-01 4.86173749e-01 -2.35109299e-01 -1.49424005e+00
5.23751192e-02 -4.07859594e-01 -1.19404972e+00 2.22543791e-01
2.41425455e-01 2.30977133e-01 -7.22416580e-01 1.64540040e+00
4.58680689e-01 4.67963785e-01 3.55878681e-01 7.38631487e-01
1.02845073e+00 5.13574660e-01 -2.29766056e-01 -4.09336984e-01
1.31616616e+00 -3.66127014e-01 -8.25547934e-01 2.63491631e-01
3.92987840e-02 -1.13090825e+00 9.93860736e-02 5.16234636e-01
-7.22226083e-01 -8.70145798e-01 -1.03868866e+00 3.64088029e-01
-2.58852482e-01 5.29621601e-01 3.29477459e-01 9.82556880e-01
-1.01456559e+00 5.56492150e-01 -3.51622999e-01 -4.91401941e-01
4.31655020e-01 6.21478200e-01 -8.40193570e-01 -2.45151326e-01
-8.52155268e-01 6.89229667e-01 1.81550637e-01 4.45231229e-01
-7.50227392e-01 -3.40925068e-01 -1.13047135e+00 -2.63324827e-02
1.26502767e-01 -2.75614917e-01 5.55453300e-01 -9.84431148e-01
-1.44004476e+00 8.93626511e-01 -4.73722607e-01 -1.03154518e-01
-1.28124291e-02 1.94892004e-01 -7.75043011e-01 1.66861534e-01
-2.88880076e-02 3.48976433e-01 1.38659072e+00 -1.39645958e+00
-1.53581873e-01 -7.08966315e-01 -3.51118743e-01 -6.34045377e-02
-2.99108103e-02 4.00004178e-01 -1.62684098e-01 -7.73037970e-01
3.91217858e-01 -8.82406175e-01 2.51895756e-01 -9.19101834e-02
-3.70236307e-01 -2.95040846e-01 1.24222815e+00 -7.35721350e-01
1.14313269e+00 -2.48372936e+00 2.42948040e-01 5.33922136e-01
-2.83164203e-01 4.92437661e-01 -8.09710249e-02 3.23196530e-01
-5.05421281e-01 -2.52524227e-01 -5.20048499e-01 -3.68351609e-01
-2.37022072e-01 1.91479385e-01 -1.02470376e-01 9.54091966e-01
1.46348268e-01 3.35591823e-01 -3.10709149e-01 -7.00275719e-01
1.86420858e-01 9.27072406e-01 -1.56447574e-01 2.03917548e-02
6.00105584e-01 4.57047433e-01 -3.39990050e-01 1.06528795e+00
1.11677825e+00 1.11630797e-01 5.31198978e-02 -4.78001356e-01
6.14881366e-02 -4.19606835e-01 -1.93720698e+00 1.44084942e+00
-1.34721085e-01 4.20339733e-01 3.04231316e-01 -1.36913729e+00
1.27254462e+00 5.63689768e-01 4.62887824e-01 -2.46983856e-01
2.14483187e-01 1.74144372e-01 -1.32891521e-01 -2.78779685e-01
2.04216957e-01 -3.27330858e-01 2.86046267e-01 3.57265174e-01
3.40185463e-01 1.47159174e-01 -4.16223891e-02 -2.32105583e-01
5.65290272e-01 -3.59870791e-02 5.89149356e-01 -2.34709337e-01
1.34535694e+00 -5.67582548e-01 5.90794384e-01 3.88942093e-01
-2.21798703e-01 7.31300592e-01 8.24213773e-02 -1.93266556e-01
-6.47653401e-01 -8.29929531e-01 -5.93416631e-01 4.13040429e-01
6.59850165e-02 -1.00582398e-01 -7.56114125e-01 -5.96029103e-01
2.00598672e-01 2.75919199e-01 -6.16124570e-01 -1.37599275e-01
-4.81756806e-01 -6.62424505e-01 6.67737544e-01 1.97238460e-01
5.84641576e-01 -9.71605480e-01 -1.31259114e-01 -1.61483232e-02
7.73817450e-02 -9.31066215e-01 -4.07856584e-01 -3.77294272e-01
-8.00542176e-01 -1.16644347e+00 -8.70079517e-01 -8.37676466e-01
7.79505432e-01 3.74473184e-01 4.62384939e-01 8.78036171e-02
-4.08972561e-01 5.77838063e-01 -2.52853781e-01 -9.51679125e-02
-3.62075686e-01 -5.06736696e-01 3.68326724e-01 8.34289610e-01
6.39095902e-01 -4.72084403e-01 -2.97600955e-01 4.64451730e-01
-7.61948228e-01 -8.46387565e-01 4.76187557e-01 9.62186098e-01
6.23931170e-01 3.73026639e-01 5.70466757e-01 -5.21468878e-01
3.59095126e-01 -3.75550330e-01 -5.37954390e-01 2.95145661e-01
-4.06430930e-01 -8.33144635e-02 3.26761127e-01 -4.95553404e-01
-1.11435509e+00 2.08777532e-01 -1.32109031e-01 -7.28919923e-01
-4.63907689e-01 2.92827547e-01 -3.47553968e-01 -7.61819184e-01
4.90423352e-01 6.83833241e-01 4.93736058e-01 -5.32548845e-01
-4.56408188e-02 9.26285207e-01 4.42808151e-01 -4.86561507e-01
9.99618292e-01 5.57786047e-01 2.08705068e-01 -1.19256091e+00
-4.29745615e-01 -7.20378399e-01 -6.02802753e-01 -1.74208194e-01
5.95544457e-01 -9.79258835e-01 -8.63430321e-01 5.12977302e-01
-9.66648400e-01 6.10805452e-01 -1.82347983e-01 5.19828379e-01
-4.46554661e-01 8.17330241e-01 -1.59493491e-01 -1.19421649e+00
-4.35068309e-01 -1.35036433e+00 1.20339036e+00 3.59703541e-01
-6.06870651e-02 -6.27681971e-01 -5.05442172e-02 4.29901987e-01
4.12475765e-01 2.97522575e-01 5.00996590e-01 -7.41624594e-01
-2.46428266e-01 -5.94044209e-01 -1.09009594e-01 6.76712871e-01
4.12754595e-01 -2.28805915e-02 -1.15482891e+00 -4.09493655e-01
3.54873002e-01 -7.76604265e-02 6.32366836e-01 3.66601437e-01
9.82692182e-01 -2.07667366e-01 -2.38379017e-01 5.06433964e-01
1.57284355e+00 3.16101521e-01 7.38085568e-01 -1.47662655e-01
5.54258347e-01 7.65748441e-01 4.89141136e-01 3.86772543e-01
-1.01599306e-01 9.48302150e-01 5.38138598e-02 1.61000296e-01
-1.26693845e-01 1.22420527e-01 5.22926211e-01 5.63583136e-01
-1.91027522e-01 1.61827028e-01 -4.02258664e-01 4.92014587e-01
-1.52589631e+00 -1.43533111e+00 1.23444214e-01 2.48342848e+00
4.32477325e-01 -5.65868795e-01 -1.36387686e-03 6.89781249e-01
9.65934694e-01 1.79433152e-01 -8.83380771e-02 -2.90828377e-01
-4.05093700e-01 5.42035460e-01 9.46135819e-02 3.15491170e-01
-1.19670057e+00 5.64728856e-01 5.80823088e+00 1.11516833e+00
-1.10132575e+00 1.04291141e-01 4.73584652e-01 3.95263970e-01
2.17869878e-01 -8.52679312e-02 -1.06380916e+00 3.61446351e-01
7.06145704e-01 5.31233102e-02 4.40590739e-01 7.19702303e-01
-1.16333142e-02 -1.16735786e-01 -1.05358648e+00 1.43775773e+00
7.52316177e-01 -9.33198571e-01 5.47006018e-02 2.13642493e-01
7.10755646e-01 -6.00440800e-01 1.76839814e-01 4.42022309e-02
-3.62083137e-01 -1.30414891e+00 3.72841805e-01 7.37567306e-01
8.85998964e-01 -9.30191636e-01 9.67823684e-01 2.41391212e-01
-1.41601872e+00 5.54069644e-03 -5.24530530e-01 3.48734230e-01
-1.61255032e-01 5.23398161e-01 -4.30076301e-01 8.94107342e-01
3.92152548e-01 6.67710066e-01 -3.27635705e-01 9.03153598e-01
8.11691508e-02 2.75621027e-01 -3.84513170e-01 3.38761538e-01
-1.49521992e-01 -3.94219756e-01 7.58807302e-01 9.24275577e-01
6.65049255e-01 2.35714301e-01 -2.64622830e-02 9.28333759e-01
7.60050341e-02 2.73027152e-01 -8.34177971e-01 8.32747817e-02
4.06898052e-01 1.30266321e+00 -3.50665301e-01 -1.97120398e-01
-4.15125132e-01 8.82200420e-01 -1.36093199e-01 2.58941799e-01
-4.58594710e-01 -2.77170092e-01 6.98451579e-01 -1.37878463e-01
6.00859463e-01 7.90261626e-02 7.67403767e-02 -1.03937376e+00
1.00296522e-02 -1.26889753e+00 4.65492457e-01 -3.70582610e-01
-1.26815236e+00 7.42087483e-01 3.00085900e-04 -1.32829189e+00
-2.90589631e-01 -4.93431658e-01 -4.49928582e-01 1.15037596e+00
-1.45669973e+00 -1.26890957e+00 -2.01430768e-01 9.63262856e-01
2.60398775e-01 -6.66995287e-01 1.04055607e+00 4.64172661e-01
-6.32768035e-01 8.39447796e-01 4.56233136e-02 1.75196573e-01
6.22269571e-01 -6.86249554e-01 -4.77569699e-01 9.04392421e-01
4.03254658e-01 9.27586555e-01 5.05999207e-01 -4.05938238e-01
-1.47513843e+00 -8.40098441e-01 9.63627577e-01 -1.46531701e-01
1.30866095e-01 2.87700519e-02 -9.50678706e-01 3.42368543e-01
-1.20019205e-02 3.54424953e-01 8.44487667e-01 -9.80079621e-02
-4.39796925e-01 -3.29654872e-01 -1.62696433e+00 -1.49450004e-02
6.48668408e-01 -7.69998372e-01 -6.73100591e-01 9.53021869e-02
-1.32110164e-01 6.14226051e-03 -1.16295981e+00 5.25892973e-01
6.46467090e-01 -1.03722954e+00 1.20736432e+00 -1.08258754e-01
-7.01451898e-02 -6.69423223e-01 -5.15568912e-01 -8.38744402e-01
-2.25305736e-01 -4.59491700e-01 -1.47248849e-01 1.50143766e+00
-9.11045969e-02 -6.58348799e-01 6.48223817e-01 1.18817665e-01
3.37498516e-01 -5.04491568e-01 -1.28062725e+00 -8.73895764e-01
-3.91757071e-01 -2.80012507e-02 7.13120043e-01 1.05642915e+00
-2.49819949e-01 4.58234735e-02 -4.29727346e-01 3.93192530e-01
9.82705712e-01 2.82525152e-01 7.93343961e-01 -1.41846347e+00
-1.76293492e-01 -3.54507804e-01 -9.08764303e-01 -5.84665298e-01
4.68706191e-01 -7.56347895e-01 -1.03331603e-01 -8.98231268e-01
2.94906974e-01 -4.55052912e-01 -3.47035855e-01 3.52534086e-01
4.14912216e-03 5.40757775e-01 1.50091887e-01 2.47391552e-01
-3.91862914e-02 5.79672337e-01 8.69803190e-01 -1.47814542e-01
2.10900437e-02 9.67006236e-02 -6.23728275e-01 4.65443522e-01
4.86045301e-01 -3.78819704e-01 -1.24615654e-01 1.66886553e-01
-5.37532151e-01 1.03661433e-01 5.09027123e-01 -1.30398178e+00
3.15313190e-01 1.96912467e-01 5.67646980e-01 -4.61545467e-01
5.75424612e-01 -1.12762749e+00 6.50799453e-01 1.60855263e-01
1.10167325e-01 -1.81336895e-01 1.43783823e-01 5.44220448e-01
-6.97360516e-01 -4.91765559e-01 1.13108921e+00 2.62700357e-02
-5.21073937e-01 3.25803578e-01 -2.98275780e-02 -5.88976145e-01
1.12799144e+00 -7.59887695e-01 2.91452497e-01 -2.02623010e-01
-7.78481543e-01 -4.34971869e-01 3.07482958e-01 1.78416491e-01
8.61064374e-01 -1.46238160e+00 -8.78382206e-01 6.91871226e-01
9.18945447e-02 -3.15179318e-01 4.95148838e-01 9.36581135e-01
-1.35917395e-01 4.58111107e-01 -1.24251731e-01 -6.83834553e-01
-1.68552160e+00 6.31933868e-01 2.89292544e-01 3.98979932e-02
-4.26075429e-01 8.36855233e-01 2.22693682e-02 -3.34666967e-01
1.48640156e-01 2.77031243e-01 -6.16074800e-01 2.68000215e-01
6.87756181e-01 3.59775931e-01 1.30517900e-01 -1.60225320e+00
-6.21531069e-01 1.13484514e+00 1.23599231e-01 1.61874086e-01
1.25153387e+00 9.91902035e-03 -3.59894156e-01 5.32605574e-02
1.45490599e+00 2.02990964e-01 -5.83984017e-01 -3.53557378e-01
-9.91056859e-02 -7.08140135e-01 1.18139878e-01 -3.48409235e-01
-1.23337662e+00 5.91148376e-01 9.30096686e-01 -1.22019956e-02
1.36208129e+00 -1.10674195e-01 4.15000081e-01 -1.45113319e-02
4.89932567e-01 -7.22949266e-01 -3.48252863e-01 9.83972922e-02
1.10712540e+00 -1.26292431e+00 2.12672114e-01 -6.01025999e-01
-3.13779831e-01 1.28020585e+00 1.45401523e-01 -2.10378289e-01
8.85692120e-01 -2.91689541e-02 -2.74192363e-01 -1.42168090e-01
-1.86986893e-01 -1.21138155e-01 5.63820481e-01 6.66858017e-01
3.03196281e-01 -1.26036003e-01 -1.86770871e-01 3.73316824e-01
-2.49696942e-03 -2.33127568e-02 1.39344186e-01 7.95223236e-01
-1.84262127e-01 -1.10552227e+00 -1.09420443e+00 2.85871446e-01
-5.90085626e-01 1.49549633e-01 -2.59830058e-01 6.79571688e-01
2.95301288e-01 9.67429519e-01 -1.60684600e-01 -2.26709574e-01
2.17376709e-01 1.90690205e-01 7.25116253e-01 -4.54241127e-01
-4.46640223e-01 2.02035472e-01 -1.54030040e-01 -4.79045898e-01
-1.00770295e+00 -1.06764734e+00 -8.12158883e-01 -2.89962679e-01
-6.35218620e-01 2.04304367e-01 7.05273271e-01 8.58184099e-01
2.88924605e-01 3.43031958e-02 8.09765100e-01 -8.81194353e-01
-6.04269266e-01 -1.04412115e+00 -8.88079107e-01 3.18690181e-01
3.81201118e-01 -9.85583305e-01 -6.21530354e-01 1.14014402e-01] | [12.873790740966797, 0.540584146976471] |
11536c1d-4f13-402e-8fe4-1a1f8c61f46c | learning-spatially-variant-map-models-for-non | null | null | http://openaccess.thecvf.com//content/CVPR2021/html/Dong_Learning_Spatially-Variant_MAP_Models_for_Non-Blind_Image_Deblurring_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Dong_Learning_Spatially-Variant_MAP_Models_for_Non-Blind_Image_Deblurring_CVPR_2021_paper.pdf | Learning Spatially-Variant MAP Models for Non-Blind Image Deblurring | The classical maximum a-posteriori (MAP) framework for non-blind image deblurring requires defining suitable data and regularization terms, whose interplay yields the desired clear image through optimization. The vast majority of prior work focuses on advancing one of these two crucial ingredients, while keeping the other one standard. Considering the indispensable roles and interplay of both data and regularization terms, we propose a simple and effective approach to jointly learn these two terms, embedding deep neural networks within the constraints of the MAP framework, trained in an end-to-end manner. The neural networks not only yield suitable image-adaptive features for both terms, but actually predict per-pixel spatially-variant features instead of the commonly used spatially-uniform ones. The resulting spatially-variant data and regularization terms particularly improve the restoration of fine-scale structures and detail. Quantitative and qualitative results underline the effectiveness of our approach, substantially outperforming the current state of the art. | ['Bernt Schiele', 'Stefan Roth', 'Jiangxin Dong'] | 2021-06-19 | null | null | null | cvpr-2021-1 | ['blind-image-deblurring'] | ['computer-vision'] | [ 3.89799178e-01 -2.87111253e-01 -1.14907816e-01 -2.78670251e-01
-7.45893061e-01 -3.94158900e-01 7.01088488e-01 -2.40691289e-01
-4.72190946e-01 6.65422320e-01 4.88933980e-01 2.07036138e-02
-4.04867023e-01 -4.61206734e-01 -4.36822683e-01 -1.20370400e+00
-2.11082511e-02 -6.38187900e-02 4.66327146e-02 -9.27326381e-02
3.01971674e-01 6.61005020e-01 -1.66993248e+00 -1.01049215e-01
1.07091570e+00 1.09863162e+00 3.08532804e-01 4.88889515e-01
2.06466854e-01 5.77447295e-01 -7.24130571e-02 -8.68075341e-02
4.34290081e-01 -3.84424269e-01 -4.51910347e-01 4.39226866e-01
6.28683686e-01 -3.48602206e-01 -4.17050719e-01 1.22498047e+00
4.36647773e-01 2.29961187e-01 6.86755657e-01 -4.54572737e-01
-9.11455274e-01 6.73097596e-02 -5.87395310e-01 9.36101675e-02
2.22349316e-02 3.98755014e-01 9.74667370e-01 -7.97249913e-01
4.29671466e-01 8.29922974e-01 7.51313686e-01 3.67363662e-01
-1.40769315e+00 -3.01048636e-01 -7.79323205e-02 1.53496087e-01
-1.37786603e+00 -8.10104430e-01 9.03528631e-01 -6.39924288e-01
5.63481390e-01 3.47113729e-01 3.99084985e-01 7.61591971e-01
7.07523599e-02 4.28508103e-01 1.44056129e+00 -5.67717254e-01
1.92903310e-01 8.13358203e-02 4.66427999e-03 4.98771995e-01
1.67910382e-02 4.60663050e-01 -4.86743659e-01 5.55165112e-02
8.96961212e-01 -7.48869255e-02 -8.93484294e-01 -5.56261182e-01
-1.31090999e+00 4.78797883e-01 6.10159338e-01 5.19158006e-01
-6.48210227e-01 7.35454410e-02 -1.33307144e-01 4.45554107e-02
6.74255610e-01 5.22772789e-01 -3.60696822e-01 1.88995406e-01
-1.36826754e+00 1.22375518e-01 3.06118488e-01 3.39183897e-01
9.54485774e-01 -7.38338893e-03 -3.38178247e-01 9.00994837e-01
4.98162925e-01 5.02048731e-01 2.96389014e-01 -8.20515394e-01
9.64446515e-02 2.46271849e-01 4.77884114e-01 -1.11972129e+00
-2.96141326e-01 -8.27203751e-01 -1.09953225e+00 4.98003751e-01
4.14372921e-01 1.27572224e-01 -9.59149480e-01 1.80128586e+00
3.41602594e-01 2.74026752e-01 -1.28094509e-01 1.29358518e+00
4.93528694e-01 5.70952952e-01 -1.99299287e-02 -3.88444364e-01
1.18114412e+00 -1.00158703e+00 -8.74930084e-01 -4.20944005e-01
1.15494922e-01 -8.45701993e-01 9.46409345e-01 1.74551889e-01
-1.12840748e+00 -2.99702048e-01 -1.13480401e+00 -1.90146506e-01
-6.71664923e-02 3.38599354e-01 5.32453001e-01 5.23280084e-01
-1.31436825e+00 7.31324136e-01 -6.94144607e-01 -3.78673635e-02
4.64555860e-01 1.61055788e-01 -4.22139496e-01 -2.04305455e-01
-9.54366624e-01 7.90156603e-01 -3.00641991e-02 6.34953618e-01
-7.26514697e-01 -7.48823702e-01 -6.25999808e-01 9.55596119e-02
1.44202426e-01 -7.60985494e-01 7.15864003e-01 -9.07552719e-01
-1.74913943e+00 9.03566718e-01 -1.60979047e-01 -1.39735997e-01
7.48539984e-01 -3.24847281e-01 -1.78204939e-01 3.25917482e-01
-1.91919114e-02 3.72036010e-01 1.34092510e+00 -1.52977133e+00
-1.96249828e-01 -3.49085987e-01 -1.71522051e-01 1.85027793e-01
-5.66530347e-01 -3.78711931e-02 -4.77717847e-01 -7.85478354e-01
2.05063090e-01 -6.11532569e-01 -2.77792513e-01 4.06393737e-01
-4.22600627e-01 3.31585824e-01 4.37992573e-01 -9.15685475e-01
1.21142721e+00 -2.08014107e+00 5.77465057e-01 -1.98699348e-02
5.06999493e-01 2.99217224e-01 -2.09257498e-01 2.60595620e-01
-1.84461147e-01 -1.84537306e-01 -7.60642231e-01 -5.16141236e-01
-8.41294155e-02 -1.39659718e-01 -1.78678811e-01 1.01579356e+00
1.73581347e-01 9.22454298e-01 -8.13623846e-01 -9.56202894e-02
5.76549709e-01 8.60212684e-01 -4.12084281e-01 3.79792601e-01
-1.75812125e-01 9.01856303e-01 -4.09195513e-01 3.29546154e-01
9.22766626e-01 -3.43880355e-01 5.45635931e-02 -4.27276611e-01
-4.11765009e-01 -3.64204831e-02 -1.07916272e+00 1.63357866e+00
-5.92820048e-01 7.13060677e-01 5.80349803e-01 -1.03620863e+00
7.06730425e-01 2.68756092e-01 4.75616366e-01 -8.11488211e-01
7.88698047e-02 4.12641078e-01 -3.89902830e-01 -6.44096434e-01
4.09023881e-01 -4.24329638e-01 4.27004427e-01 2.96942770e-01
-4.40306738e-02 3.95692587e-02 -2.27589890e-01 -1.17344551e-01
8.21467817e-01 1.08671710e-01 3.39429051e-01 -5.17004371e-01
7.98777282e-01 -3.82741988e-01 2.66499847e-01 6.88999891e-01
-2.14820787e-01 9.99558210e-01 1.33096054e-01 -2.48870015e-01
-1.07707095e+00 -7.72505820e-01 -3.53644103e-01 7.47778356e-01
3.37239712e-01 6.38585351e-03 -7.20520318e-01 -5.37250996e-01
-2.40283221e-01 3.80310625e-01 -6.55384481e-01 6.96368366e-02
-5.14422715e-01 -8.60784113e-01 1.53916907e-02 -2.51311641e-02
5.02782643e-01 -7.64863253e-01 -3.84435356e-01 9.59032029e-02
-2.44568840e-01 -1.11686397e+00 -5.18958986e-01 3.04384846e-02
-7.74931908e-01 -9.48508501e-01 -1.22666132e+00 -5.53778350e-01
7.84979105e-01 5.61816275e-01 1.04976368e+00 2.22176239e-01
-2.31404811e-01 2.76633978e-01 -2.43604511e-01 3.89401466e-01
-2.76332825e-01 -1.79597437e-01 -1.51983961e-01 5.18439353e-01
-1.98386654e-01 -1.03670776e+00 -9.28924382e-01 2.82841474e-01
-1.15472710e+00 2.01912850e-01 9.08016145e-01 1.06973934e+00
3.54114920e-01 -4.21536565e-02 3.17463785e-01 -4.52731550e-01
4.41422999e-01 -1.92969784e-01 -6.83265150e-01 3.16213042e-01
-7.72379816e-01 1.72834724e-01 4.64931428e-01 -2.55514473e-01
-1.06726396e+00 5.07237986e-02 -1.48338705e-01 -2.71982551e-01
-2.27023527e-01 5.01467526e-01 -2.84879148e-01 -4.36852992e-01
7.13569462e-01 4.82265085e-01 5.45404181e-02 -5.89336634e-01
4.91636962e-01 5.75786233e-01 8.17476451e-01 -3.67303431e-01
9.81441677e-01 6.99314833e-01 -2.54370179e-03 -8.72837126e-01
-7.90024638e-01 -5.54743826e-01 -7.15710580e-01 -2.03852415e-01
8.08728635e-01 -8.39666784e-01 -4.66451019e-01 8.91086102e-01
-1.03489220e+00 -4.09464926e-01 -2.81427622e-01 3.96998316e-01
-5.30539632e-01 7.12530375e-01 -4.92283970e-01 -7.79621124e-01
-1.75199717e-01 -1.27396965e+00 1.13677430e+00 1.68933064e-01
2.98647612e-01 -1.04578352e+00 1.63721427e-01 4.91231531e-01
8.86849463e-01 5.92716895e-02 8.99768949e-01 -1.44775407e-02
-8.51531327e-01 -2.75324821e-01 -7.02240467e-01 4.34380591e-01
1.28036574e-01 -4.11745667e-01 -1.19444251e+00 -3.19775850e-01
1.90490797e-01 -1.25967432e-02 1.13012183e+00 7.21488774e-01
8.52005541e-01 -3.57409090e-01 -3.70932221e-02 9.17317629e-01
1.66759920e+00 -5.26963413e-01 8.50437701e-01 3.45828354e-01
7.48203993e-01 7.00177968e-01 2.03426227e-01 3.37810427e-01
3.58978510e-01 1.06343782e+00 5.74438095e-01 -3.09965998e-01
-5.44654667e-01 1.71116501e-01 1.93745017e-01 7.39031732e-01
-1.91455618e-01 -8.37859437e-02 -6.60504699e-01 6.52624071e-01
-1.88419366e+00 -8.20585728e-01 -2.14168042e-01 2.36621189e+00
1.07422948e+00 -3.96585494e-01 -3.88530672e-01 4.20083068e-02
6.48864329e-01 6.36800706e-01 -3.37278217e-01 2.35458046e-01
-3.73417944e-01 1.60771802e-01 4.48177874e-01 8.49512637e-01
-1.29834294e+00 7.63132393e-01 6.18517685e+00 9.51020360e-01
-1.40179157e+00 1.33520141e-01 5.19369543e-01 2.11310107e-02
-5.80199003e-01 7.00925514e-02 -2.09654346e-01 4.31552470e-01
6.00437164e-01 3.12740535e-01 9.22074974e-01 2.89855868e-01
5.25753796e-01 -4.70246971e-02 -7.78652847e-01 1.01881027e+00
-4.93479818e-02 -1.37258637e+00 -3.09158087e-01 2.66040117e-01
9.32440877e-01 -1.72766950e-02 2.10838199e-01 -2.84888744e-01
-1.17436342e-01 -1.10041404e+00 8.82948875e-01 8.82952154e-01
8.59780788e-01 -2.69241214e-01 6.07284844e-01 2.34735683e-01
-9.18731511e-01 -3.20493728e-02 -1.09632518e-02 6.57733604e-02
3.63777280e-01 1.14051700e+00 9.97217372e-03 6.94104791e-01
5.55304885e-01 9.74994659e-01 -3.58523697e-01 1.14156711e+00
-4.93477643e-01 5.33252358e-01 -2.17659295e-01 5.89279890e-01
9.12046283e-02 -5.00767767e-01 9.33993220e-01 1.15732229e+00
2.29223490e-01 -4.96567935e-02 -1.04454562e-01 1.06883752e+00
6.22092106e-04 -1.26845047e-01 -2.45707512e-01 7.46498257e-02
1.31810337e-01 1.37166941e+00 -5.07003844e-01 -7.49426112e-02
-2.69524664e-01 1.12551105e+00 2.83950776e-01 6.55321360e-01
-4.72756267e-01 -9.13791135e-02 7.62234628e-01 7.61139616e-02
4.91006762e-01 -3.66894752e-01 -5.59030294e-01 -1.38596690e+00
4.78789866e-01 -8.31181109e-01 -1.96376875e-01 -5.96997201e-01
-1.44119442e+00 6.04394913e-01 -3.42795730e-01 -1.30984533e+00
4.33132518e-03 -6.21606350e-01 -5.72428167e-01 1.08501291e+00
-2.20974565e+00 -1.40661538e+00 -4.22623247e-01 4.11979944e-01
1.35542154e-01 1.68680489e-01 7.69101799e-01 5.56101382e-01
-6.73009217e-01 3.23175907e-01 5.38532495e-01 -3.35031897e-01
7.17981219e-01 -1.15293097e+00 2.14457959e-02 1.30591273e+00
-7.88091123e-02 7.21023977e-01 9.15050209e-01 -3.36828709e-01
-1.37306106e+00 -7.76623607e-01 7.73760498e-01 -1.63435817e-01
6.01547480e-01 -1.11727253e-01 -1.03335512e+00 1.24899261e-01
2.13483438e-01 3.94191206e-01 3.04056972e-01 -1.09001964e-01
-4.65957612e-01 -2.01208144e-02 -1.03111458e+00 4.66891140e-01
7.49043465e-01 -7.09101439e-01 -2.81627685e-01 2.15366229e-01
3.85928005e-01 -2.17240795e-01 -7.01037765e-01 4.36242491e-01
6.60326540e-01 -1.37511575e+00 1.25341940e+00 -2.74665356e-01
6.77932739e-01 -4.68486249e-01 -1.82666317e-01 -1.34308076e+00
-4.10190403e-01 -7.44329214e-01 -2.19211504e-01 1.10974407e+00
3.18161994e-01 -7.23330319e-01 6.10723317e-01 5.17088234e-01
-1.96229160e-01 -7.03643501e-01 -1.02724433e+00 -4.70939159e-01
-1.03500243e-02 -2.95219570e-01 3.01728189e-01 1.07260549e+00
-3.59712601e-01 2.73192115e-02 -8.08771849e-01 4.63814050e-01
9.36163187e-01 2.12203816e-01 4.19041544e-01 -1.15250850e+00
-3.33176017e-01 -7.45859027e-01 -3.26015890e-01 -1.38338494e+00
9.71770585e-02 -6.57657921e-01 4.25021052e-01 -1.51443815e+00
2.06693202e-01 -5.23716211e-01 -3.38194698e-01 2.45733902e-01
-4.85098392e-01 5.25889218e-01 -9.07617584e-02 6.09959006e-01
-3.43091458e-01 8.75715137e-01 1.31104851e+00 -9.12925974e-02
-1.86807141e-01 -1.13715760e-01 -7.20426679e-01 4.64389056e-01
4.11073387e-01 -2.45512962e-01 -1.35506541e-01 -6.94332242e-01
1.13898642e-01 -1.28665879e-01 7.31767595e-01 -8.38180780e-01
2.82853335e-01 -2.36975983e-01 3.14384490e-01 -3.68172191e-02
3.19945633e-01 -9.76026416e-01 -7.98254162e-02 -2.81996652e-02
-3.57025474e-01 -5.96299112e-01 1.79049857e-02 7.96058953e-01
-3.39399040e-01 -9.01350304e-02 1.11469734e+00 1.06089354e-01
-5.68271697e-01 2.42818728e-01 -1.27683714e-01 -2.93872297e-01
6.67170525e-01 -2.33408958e-01 -1.98917851e-01 -4.59479481e-01
-5.53869724e-01 -1.74020797e-01 5.51949620e-01 2.44823903e-01
4.64972168e-01 -1.13966072e+00 -8.40152979e-01 3.06270391e-01
3.52176540e-02 -1.95639208e-01 6.70347929e-01 1.36911893e+00
-4.08573091e-01 1.71055064e-01 -1.70894295e-01 -5.77450156e-01
-8.40444922e-01 3.99355501e-01 5.85736513e-01 -3.43656927e-01
-6.96774304e-01 8.51576567e-01 4.53127414e-01 -4.40161824e-01
5.65214902e-02 4.01374139e-03 -2.39420414e-01 -1.16215661e-01
6.12612486e-01 2.92910188e-01 1.09877519e-01 -7.93753982e-01
-1.58797160e-01 7.76772618e-01 9.53915492e-02 -1.80428699e-01
1.70767117e+00 -6.04823530e-01 -4.70100403e-01 -4.13655750e-02
1.17580318e+00 3.06585282e-01 -1.63633704e+00 -4.35910314e-01
-9.61316153e-02 -7.84181297e-01 6.81039810e-01 -8.05610716e-01
-1.26251101e+00 8.89842927e-01 8.49839926e-01 2.19095379e-01
1.37274754e+00 -2.08992973e-01 6.57896519e-01 -1.97487041e-01
1.54718578e-01 -5.89131415e-01 -1.20998926e-01 1.66411266e-01
9.86953020e-01 -1.26106727e+00 2.56633967e-01 -4.57621813e-01
-2.25490302e-01 1.16685236e+00 1.22913104e-02 -8.26545879e-02
5.86915493e-01 -4.74865660e-02 1.74471661e-01 -2.35486180e-01
-2.16167524e-01 -2.90770471e-01 7.62243986e-01 5.98481596e-01
3.99593443e-01 -1.07543595e-01 -2.83710629e-01 3.27457339e-01
2.45483354e-01 8.01944211e-02 1.51608586e-01 5.08665800e-01
-3.59193027e-01 -1.04316795e+00 -4.26415622e-01 1.42436653e-01
-4.39464986e-01 -3.11189711e-01 -6.11268617e-02 4.28210706e-01
-2.96929572e-02 9.19063449e-01 -2.76509047e-01 -1.33379400e-01
9.33089703e-02 -3.12563211e-01 3.40753138e-01 -2.02423513e-01
-4.72639531e-01 2.58068591e-01 -1.02521814e-01 -5.63219488e-01
-6.83151662e-01 -4.92442340e-01 -5.62970579e-01 -1.60499051e-01
-4.18326110e-01 -1.70149803e-01 6.86321497e-01 1.10060370e+00
4.42334443e-01 3.29516023e-01 8.49166214e-01 -1.49629688e+00
-6.14280522e-01 -9.57220674e-01 -5.78018188e-01 2.81648219e-01
9.85445142e-01 -5.75699449e-01 -6.20229959e-01 1.99086487e-01] | [11.562132835388184, -2.5125339031219482] |
49e31c4e-8ec5-4c93-8ac3-11bcf85ac1a6 | hierarchical-and-progressive-image-matting | 2210.06906 | null | https://arxiv.org/abs/2210.06906v1 | https://arxiv.org/pdf/2210.06906v1.pdf | Hierarchical and Progressive Image Matting | Most matting researches resort to advanced semantics to achieve high-quality alpha mattes, and direct low-level features combination is usually explored to complement alpha details. However, we argue that appearance-agnostic integration can only provide biased foreground details and alpha mattes require different-level feature aggregation for better pixel-wise opacity perception. In this paper, we propose an end-to-end Hierarchical and Progressive Attention Matting Network (HAttMatting++), which can better predict the opacity of the foreground from single RGB images without additional input. Specifically, we utilize channel-wise attention to distill pyramidal features and employ spatial attention at different levels to filter appearance cues. This progressive attention mechanism can estimate alpha mattes from adaptive semantics and semantics-indicated boundaries. We also introduce a hybrid loss function fusing Structural SIMilarity (SSIM), Mean Square Error (MSE), Adversarial loss, and sentry supervision to guide the network to further improve the overall foreground structure. Besides, we construct a large-scale and challenging image matting dataset comprised of 59, 600 training images and 1000 test images (a total of 646 distinct foreground alpha mattes), which can further improve the robustness of our hierarchical and progressive aggregation model. Extensive experiments demonstrate that the proposed HAttMatting++ can capture sophisticated foreground structures and achieve state-of-the-art performance with single RGB images as input. | ['Xin Yang', 'Guofeng Zhang', 'Qiang Cai', 'Yuxin Wang', 'Ziqi Wei', 'Yuhao Liu', 'Yu Qiao'] | 2022-10-13 | null | null | null | null | ['image-matting'] | ['computer-vision'] | [ 4.81903464e-01 -1.42429560e-01 1.27197787e-01 -4.01615441e-01
-5.08784592e-01 -2.14704096e-01 2.78863043e-01 -4.45236176e-01
-1.55612528e-01 5.27559936e-01 -9.40825045e-02 -2.06178233e-01
3.00039649e-01 -8.78687561e-01 -1.02325332e+00 -9.23054934e-01
3.03500503e-01 3.05295885e-01 7.67265260e-01 -4.55783755e-02
6.98660538e-02 2.26544648e-01 -1.28493547e+00 7.51110494e-01
1.54076076e+00 1.52208447e+00 3.41175824e-01 7.52818406e-01
-3.79638791e-01 1.09766936e+00 -6.68066680e-01 -7.16827452e-01
4.34170216e-01 -4.45302963e-01 -4.86918628e-01 4.35134143e-01
7.46345401e-01 -5.45897722e-01 -4.01138633e-01 9.95248675e-01
1.37874663e-01 -8.02196339e-02 5.34650326e-01 -1.23116481e+00
-9.31852639e-01 5.25093436e-01 -1.07370639e+00 3.75222296e-01
2.96205152e-02 8.03567290e-01 8.42635930e-01 -9.65224326e-01
1.65284984e-02 1.40077436e+00 4.85178858e-01 1.53181240e-01
-1.15384066e+00 -7.30699301e-01 6.13464773e-01 3.30762029e-01
-1.14269125e+00 5.37293069e-02 9.77211416e-01 -1.09707505e-01
4.64776516e-01 3.32631856e-01 9.64303136e-01 1.03333724e+00
2.54602164e-01 1.09938228e+00 1.43006480e+00 -3.51868756e-03
-1.36358580e-02 -2.47709770e-02 -1.84455246e-01 1.06726849e+00
3.58153254e-01 -1.83727235e-01 -6.21842980e-01 2.08594963e-01
1.25041091e+00 8.12774226e-02 -4.61646676e-01 -2.30349004e-01
-1.46300364e+00 4.80485886e-01 1.10670817e+00 -4.78644110e-02
-3.58552426e-01 2.67868459e-01 3.71736772e-02 2.99058072e-02
4.57462847e-01 1.79952398e-01 -3.89253378e-01 1.01307720e-01
-9.39688146e-01 -1.29396021e-01 2.47375920e-01 9.20758188e-01
1.07033300e+00 4.78837609e-01 -4.02801901e-01 5.46652973e-01
4.41459835e-01 9.24424231e-01 1.03211626e-01 -1.26483822e+00
6.64959073e-01 9.32172835e-01 -1.28334373e-01 -1.00103259e+00
5.33606485e-02 -4.95795965e-01 -1.01626015e+00 3.52304399e-01
5.24360538e-01 1.60285980e-01 -1.58690929e+00 1.28504002e+00
3.76577854e-01 5.07060468e-01 -2.29293987e-01 1.20305085e+00
7.17708349e-01 9.57216084e-01 -1.13407150e-01 -1.37053311e-01
1.32666934e+00 -1.45301557e+00 -5.70165753e-01 -4.70138818e-01
-8.65955129e-02 -7.20217645e-01 1.47645473e+00 6.20946527e-01
-1.58351588e+00 -7.63354480e-01 -1.14004993e+00 -4.11890626e-01
-6.59060404e-02 -4.24664058e-02 7.34990239e-01 5.53884804e-01
-8.39244008e-01 3.88760865e-01 -9.99830186e-01 2.79358596e-01
7.87356913e-01 3.28810513e-01 2.16135666e-01 -5.81865348e-02
-8.93102586e-01 4.05639052e-01 3.15502942e-01 4.33300793e-01
-1.11004591e+00 -9.37345803e-01 -6.82114720e-01 -5.56178056e-02
5.66510558e-01 -1.08575070e+00 7.38302588e-01 -1.39298904e+00
-1.62564492e+00 6.54835105e-01 -1.41896717e-02 -2.48837337e-01
7.12711930e-01 -3.92504245e-01 -1.61250785e-01 5.10254622e-01
4.45091799e-02 7.74003029e-01 1.27900672e+00 -1.76685643e+00
-6.57081664e-01 -1.24477737e-01 1.49570659e-01 2.45510489e-01
-2.67434537e-01 -2.04416245e-01 -6.90774441e-01 -1.05664945e+00
2.86916345e-01 -5.71396112e-01 -1.04905352e-01 3.71047199e-01
-4.94981080e-01 4.46244389e-01 8.06150675e-01 -8.44836950e-01
1.04457176e+00 -1.95436502e+00 5.01997113e-01 2.52522737e-01
4.74780917e-01 2.68163621e-01 -4.11528861e-04 -5.36931396e-01
3.94072831e-01 -6.59698620e-02 -6.30605698e-01 -3.51098299e-01
-2.11801857e-01 4.09855485e-01 -5.05666614e-01 1.81748316e-01
5.31420469e-01 1.22282100e+00 -7.93355525e-01 -7.72286415e-01
5.49880326e-01 6.84147775e-01 -5.57909250e-01 4.60070789e-01
-5.28045893e-01 4.07719553e-01 -3.83022100e-01 1.13100564e+00
8.47347558e-01 -6.58880115e-01 -3.43962133e-01 -6.39724791e-01
2.70404041e-01 -1.28325358e-01 -9.29773569e-01 1.54219925e+00
-4.69695508e-01 4.53877449e-01 2.05123052e-01 -3.48508865e-01
6.80111110e-01 -3.82001549e-01 2.13084906e-01 -5.63805878e-01
3.60471129e-01 1.51456282e-01 -9.80602764e-03 -1.10944837e-01
2.59969652e-01 1.36989370e-01 2.56316483e-01 1.50121838e-01
-3.31063509e-01 -1.73849463e-01 -2.28243321e-01 2.38137707e-01
9.79611278e-01 1.48930326e-01 -2.85971075e-01 -5.22066094e-02
5.73937118e-01 -2.15557829e-01 6.40343606e-01 6.62259340e-01
-1.70072183e-01 1.07969022e+00 2.83614278e-01 -4.87191826e-01
-1.08837521e+00 -1.47669220e+00 5.58743775e-02 1.16865981e+00
7.66445100e-01 -5.32249585e-02 -8.21559846e-01 -7.20811248e-01
5.38521260e-02 4.68620807e-01 -9.99612391e-01 -1.19270325e-01
-8.19689274e-01 -7.44704545e-01 3.75578314e-01 7.59099483e-01
1.21371853e+00 -1.00252509e+00 -5.03501415e-01 -1.34020254e-01
-3.57446820e-01 -1.06713128e+00 -7.08995223e-01 1.98233798e-01
-9.53322232e-01 -8.33757460e-01 -8.77959549e-01 -4.74495023e-01
7.50818014e-01 4.10722673e-01 1.16531467e+00 3.50822836e-01
-7.33229890e-02 7.85581172e-02 -1.97544008e-01 -6.59645647e-02
-2.06540689e-01 -7.06903040e-02 -3.71364594e-01 4.10266876e-01
-1.53760597e-01 -7.23029077e-01 -1.04513836e+00 4.01414007e-01
-1.07623363e+00 4.27629083e-01 1.12096691e+00 8.01549315e-01
5.85753679e-01 -3.88398528e-01 -4.56576310e-02 -8.89654994e-01
-8.38332176e-02 -1.90680817e-01 -3.38904560e-01 4.23929006e-01
-5.31916440e-01 -4.84491102e-02 6.78153038e-01 -5.78575969e-01
-1.34976017e+00 -3.62160116e-01 1.07839093e-01 -1.01114500e+00
9.29712430e-02 -6.11534528e-02 -5.70338786e-01 -4.38111663e-01
3.92836034e-01 4.81075019e-01 -1.70873046e-01 -2.32156292e-01
4.93644834e-01 1.33374155e-01 8.06149423e-01 -7.07839191e-01
1.11741877e+00 7.43275225e-01 -8.67802948e-02 -3.48915845e-01
-1.17764437e+00 -3.59268347e-03 -4.84337300e-01 -2.50883043e-01
1.01070082e+00 -1.02657366e+00 -5.81802309e-01 6.95361555e-01
-9.62363482e-01 -7.40276039e-01 -1.74046680e-01 -2.85350569e-02
-4.68349785e-01 5.17291069e-01 -1.07159400e+00 -5.33223569e-01
-5.62866151e-01 -1.32157874e+00 1.37562740e+00 3.18879247e-01
5.14398813e-01 -6.07791424e-01 -6.01508796e-01 8.99823308e-01
4.31040764e-01 2.62262911e-01 7.52214730e-01 2.60665983e-01
-1.36028576e+00 5.83064139e-01 -8.41147304e-01 4.52698976e-01
1.29243895e-01 2.53118783e-01 -1.12071860e+00 -1.96888432e-01
-5.76826669e-02 -1.85963407e-01 1.42557013e+00 4.17205811e-01
1.55281401e+00 -4.26045567e-01 2.08781101e-02 1.32832158e+00
1.15527081e+00 -2.02574451e-02 9.42755640e-01 4.72440690e-01
1.52666509e+00 1.76275760e-01 7.59482503e-01 3.76667202e-01
3.64542097e-01 3.92539591e-01 7.17208564e-01 -6.08564615e-01
-5.02489924e-01 -1.32757062e-02 4.78052139e-01 8.13891232e-01
-2.81218022e-01 -1.28914416e-01 -4.63293344e-01 1.59100667e-01
-1.65794635e+00 -6.16113484e-01 -1.16971694e-01 1.79679000e+00
1.10414279e+00 4.70598996e-01 1.38146043e-01 -8.29400346e-02
5.80487490e-01 3.35157126e-01 -8.65278304e-01 5.26384544e-03
-4.93959755e-01 1.08626172e-01 5.40362239e-01 5.52593946e-01
-1.02818418e+00 1.03113818e+00 5.50403833e+00 1.03400922e+00
-1.19309747e+00 3.88686061e-02 1.20609319e+00 -7.91765973e-02
-6.83934927e-01 -1.25335127e-01 -3.80999476e-01 9.85146523e-01
3.60278904e-01 3.33645016e-01 5.23067534e-01 6.47496998e-01
-5.44409500e-03 -9.23302546e-02 -6.13000691e-01 8.47800493e-01
1.97193637e-01 -1.26767588e+00 4.70365077e-01 -5.73095009e-02
1.02231419e+00 -2.32673779e-01 5.03057837e-01 4.23041098e-02
2.30164081e-01 -9.71163571e-01 9.47287560e-01 5.44872582e-01
8.25587749e-01 -6.86805785e-01 6.55034959e-01 -2.41195709e-01
-1.50863314e+00 -1.58772334e-01 -3.58823448e-01 1.47627577e-01
2.53906220e-01 7.18221426e-01 -4.75291103e-01 5.94465494e-01
8.35764647e-01 5.78537524e-01 -9.27124321e-01 8.35582733e-01
-2.38231376e-01 7.00912952e-01 -3.41958672e-01 1.52263924e-01
3.51332396e-01 -3.28443050e-01 3.60472649e-01 1.11196673e+00
4.32017446e-02 1.40821308e-01 3.34963232e-01 1.15314662e+00
1.41964294e-02 -2.15190873e-01 8.98028985e-02 2.35059455e-01
3.25905502e-01 1.22257495e+00 -9.63672757e-01 -6.52171016e-01
-3.03145945e-01 1.48630357e+00 2.64687032e-01 6.28521740e-01
-1.37502575e+00 -1.98831588e-01 7.78213263e-01 2.95976818e-01
5.92648864e-01 3.42278136e-03 -7.06644893e-01 -1.34786999e+00
2.38382984e-02 -9.58766580e-01 2.28894502e-01 -1.17504096e+00
-1.27765512e+00 5.79474747e-01 -9.40990895e-02 -1.04886389e+00
4.26429749e-01 -8.12037766e-01 -6.86810434e-01 6.46739483e-01
-1.54457653e+00 -1.69681823e+00 -9.30354714e-01 6.18994176e-01
6.20459735e-01 1.51315197e-01 -1.32634193e-01 3.17889571e-01
-6.78238153e-01 6.56450748e-01 -2.84089118e-01 2.50777788e-02
6.87429011e-01 -1.55391145e+00 3.60943586e-01 1.00984907e+00
2.43864022e-02 5.15527904e-01 5.63075006e-01 -8.42627048e-01
-1.36762977e+00 -1.40586793e+00 -1.36535913e-01 -5.36620796e-01
6.33032024e-01 -2.79241681e-01 -1.16665018e+00 5.69364429e-01
2.98875302e-01 1.76295787e-01 4.50437441e-02 -4.00907844e-01
-4.86166209e-01 -4.59598631e-01 -9.67553496e-01 8.08260739e-01
1.16047895e+00 -2.95709282e-01 -2.85357654e-01 8.61343741e-02
1.18842900e+00 -6.46599710e-01 -7.76240110e-01 6.96871817e-01
3.19551378e-01 -1.30640829e+00 1.25379515e+00 -8.30807760e-02
7.01173782e-01 -8.12779546e-01 1.36377336e-02 -8.43792498e-01
-3.39166522e-01 -5.66916883e-01 -5.47562540e-01 1.15949190e+00
2.44189128e-01 -4.01051998e-01 1.02275968e+00 4.23987567e-01
-4.59410489e-01 -1.24318612e+00 -7.06813514e-01 -3.58853042e-01
-1.18521228e-01 -3.12679082e-01 6.55193746e-01 6.45720720e-01
-7.75529504e-01 1.74959660e-01 -5.66074312e-01 3.49494547e-01
8.66940558e-01 4.18327242e-01 7.86558747e-01 -6.82701766e-01
-4.83726591e-01 -5.16533792e-01 -3.56306642e-01 -1.39828587e+00
-2.79296309e-01 -3.69424224e-01 1.04545958e-01 -1.33117306e+00
4.17870164e-01 -4.55019921e-01 -3.97689939e-01 4.15525377e-01
-8.49431276e-01 6.75091445e-01 2.71038502e-01 1.30547673e-01
-8.18783343e-01 8.76764476e-01 1.74640310e+00 -4.47075397e-01
5.86271062e-02 -2.45071203e-01 -5.45683920e-01 7.65401006e-01
4.88865793e-01 -8.87090247e-03 -3.66039157e-01 -6.45500839e-01
1.19150551e-02 -3.52365285e-01 6.05493486e-01 -1.05417597e+00
-8.52786750e-03 -5.19611359e-01 9.56070721e-01 -5.93755603e-01
6.40276551e-01 -7.42004693e-01 5.04233316e-02 3.32757622e-01
-9.37369764e-02 2.90727913e-02 7.42306635e-02 7.53119648e-01
6.18392825e-02 2.39570960e-01 7.29509473e-01 -2.99419373e-01
-6.49253845e-01 6.17314994e-01 8.17147195e-02 1.12691700e-01
9.24535275e-01 -5.54338217e-01 -4.85087126e-01 -1.05208866e-01
-3.52359980e-01 2.24098802e-01 9.86971140e-01 8.11953619e-02
9.42154527e-01 -1.30743217e+00 -5.38684249e-01 2.69885898e-01
-1.25031635e-01 5.32589257e-01 3.06078792e-01 9.45234001e-01
-9.95972991e-01 -2.85943151e-01 -3.40067327e-01 -9.39507961e-01
-1.00086224e+00 5.80716491e-01 2.09089175e-01 -6.96119890e-02
-6.96514249e-01 1.33381438e+00 9.33401465e-01 1.21800609e-01
2.59304851e-01 -7.06404865e-01 5.11500180e-01 -3.32688302e-01
5.91469646e-01 2.58885264e-01 -2.91437209e-01 -4.15364981e-01
-2.00958624e-01 5.66645324e-01 -2.79080480e-01 6.25871122e-02
9.57869768e-01 -3.89712811e-01 -1.19862981e-01 4.71574426e-01
7.24650085e-01 -7.77385682e-02 -1.96617603e+00 -2.05969647e-01
-5.76818824e-01 -1.10627615e+00 6.05062395e-02 -7.57954359e-01
-1.58716440e+00 1.13775742e+00 5.76604307e-01 -5.96761666e-02
1.41258395e+00 -6.72004670e-02 1.22150433e+00 1.09335221e-01
1.59818649e-01 -7.28751302e-01 8.69583309e-01 2.03150678e-02
7.44534552e-01 -1.39879847e+00 1.63016289e-01 -6.56978250e-01
-8.83003831e-01 8.56268764e-01 1.29254758e+00 -1.35818809e-01
1.36917844e-01 4.15488541e-01 3.04728240e-01 6.16519805e-03
-6.68677568e-01 -2.74081409e-01 4.69449431e-01 5.01192212e-01
5.00056073e-02 -1.76462874e-01 4.63723809e-01 4.32397693e-01
1.13003671e-01 -4.94886279e-01 2.64244437e-01 6.51103318e-01
-6.79313779e-01 -5.23900628e-01 -5.83532155e-01 6.36823475e-01
-2.82501727e-01 -3.36963892e-01 -4.35619056e-01 4.40443873e-01
3.21489185e-01 6.38419926e-01 2.40591802e-02 -4.65247065e-01
1.89208239e-02 -5.08656263e-01 7.49745667e-01 -3.23053837e-01
-6.17020428e-01 2.62591332e-01 -3.13085288e-01 -7.31186271e-01
-2.83894002e-01 -3.28104496e-01 -1.28714108e+00 -6.06975675e-01
-4.43867862e-01 -2.64474124e-01 3.38002257e-02 8.88050377e-01
1.26506746e-01 9.44071889e-01 6.48952901e-01 -1.08768880e+00
-1.12720810e-01 -7.55678713e-01 -2.77820081e-01 5.07936180e-01
4.32873547e-01 -7.24190354e-01 -3.96783769e-01 2.71760732e-01] | [10.622556686401367, -0.9487459063529968] |
c5f3f26c-962e-4136-a821-0a7b40e05663 | boosting-multitask-learning-on-graphs-through | 2306.14009 | null | https://arxiv.org/abs/2306.14009v1 | https://arxiv.org/pdf/2306.14009v1.pdf | Boosting Multitask Learning on Graphs through Higher-Order Task Affinities | Predicting node labels on a given graph is a widely studied problem with many applications, including community detection and molecular graph prediction. This paper considers predicting multiple node labeling functions on graphs simultaneously and revisits this problem from a multitask learning perspective. For a concrete example, consider overlapping community detection: each community membership is a binary node classification task. Due to complex overlapping patterns, we find that negative transfer is prevalent when we apply naive multitask learning to multiple community detection, as task relationships are highly nonlinear across different node labeling. To address the challenge, we develop an algorithm to cluster tasks into groups based on a higher-order task affinity measure. We then fit a multitask model on each task group, resulting in a boosting procedure on top of the baseline model. We estimate the higher-order task affinity measure between two tasks as the prediction loss of one task in the presence of another task and a random subset of other tasks. Then, we use spectral clustering on the affinity score matrix to identify task grouping. We design several speedup techniques to compute the higher-order affinity scores efficiently and show that they can predict negative transfers more accurately than pairwise task affinities. We validate our procedure using various community detection and molecular graph prediction data sets, showing favorable results compared with existing methods. Lastly, we provide a theoretical analysis to show that under a planted block model of tasks on graphs, our affinity scores can provably separate tasks into groups. | ['Hongyang R. Zhang', 'Aneesh Sharma', 'Haotian Ju', 'Dongyue Li'] | 2023-06-24 | null | null | null | null | ['node-classification', 'community-detection'] | ['graphs', 'graphs'] | [ 5.19554794e-01 -6.17736131e-02 -3.30738842e-01 -1.26329795e-01
-5.85140824e-01 -7.12772727e-01 3.43964458e-01 4.96747673e-01
-3.28648686e-01 5.55945814e-01 -9.38323140e-02 -4.72792357e-01
-3.86495799e-01 -4.54501510e-01 -8.59129429e-01 -8.34680915e-01
-5.40211260e-01 9.18935239e-01 4.95719254e-01 1.94130793e-01
2.33323932e-01 9.27799568e-02 -1.00808620e+00 6.15470588e-01
1.05924022e+00 4.52944458e-01 3.25139195e-01 8.66411328e-01
1.50142416e-01 6.31129146e-01 -2.75943637e-01 -3.72185409e-01
1.22627422e-01 7.35666137e-03 -1.12535274e+00 1.77338988e-01
3.32959205e-01 3.19572479e-01 1.00156575e-01 9.56977308e-01
3.56988221e-01 -2.28688270e-01 1.03390241e+00 -1.73302603e+00
-1.88540548e-01 5.63664615e-01 -1.05443442e+00 2.66131014e-01
8.91800970e-02 -4.77070212e-01 1.46122622e+00 -8.05181563e-01
5.72823048e-01 1.06262994e+00 8.47722411e-01 1.70438856e-01
-1.91595507e+00 -7.90675223e-01 3.86222661e-01 1.56034946e-01
-1.39200151e+00 -1.11873411e-01 6.71562195e-01 -1.02463579e+00
8.40624034e-01 1.59443617e-01 4.14964110e-01 8.05548728e-01
5.05548775e-01 6.85694516e-01 1.04025781e+00 -2.42356196e-01
-4.49510291e-02 -1.40086412e-02 5.65978169e-01 8.51523817e-01
2.77372926e-01 -4.38898027e-01 -6.38470054e-01 -6.50472820e-01
2.90337503e-01 4.37960923e-02 -2.70945579e-01 -8.69754314e-01
-1.23447883e+00 8.53156865e-01 5.60702801e-01 2.38370240e-01
-1.80397719e-01 2.67527640e-01 4.39525485e-01 5.93784988e-01
8.23802233e-01 6.74553514e-02 -4.59604144e-01 6.20436728e-01
-6.19685531e-01 -2.50829272e-02 8.47658694e-01 8.23152184e-01
1.18196225e+00 -7.17129707e-01 -1.66316018e-01 7.91244984e-01
2.06781670e-01 1.93622723e-01 -9.58218127e-02 -6.29912794e-01
4.12191421e-01 4.07587647e-01 -2.42106929e-01 -9.55180705e-01
-7.37640798e-01 -6.51435614e-01 -1.08421493e+00 -2.73345470e-01
5.61715245e-01 -9.13274810e-02 -5.40358484e-01 1.99349964e+00
3.18333417e-01 4.04117823e-01 -3.66023988e-01 5.07309020e-01
1.30552381e-01 4.65854049e-01 1.89063147e-01 -5.49099386e-01
1.36247790e+00 -9.03249323e-01 -4.55007136e-01 -3.30786228e-01
1.20946646e+00 -6.99265063e-01 7.41420209e-01 3.51758718e-01
-6.24309659e-01 -2.47874588e-01 -7.68695831e-01 2.36684114e-01
-7.09782839e-02 -1.96462944e-01 8.47805500e-01 6.22663915e-01
-1.51274371e+00 4.57883269e-01 -6.21869326e-01 -4.76199597e-01
2.66403049e-01 6.78097904e-01 -1.91018462e-01 -2.12750942e-01
-9.74764228e-01 6.53350413e-01 1.98783875e-01 -3.61762077e-01
-7.53073931e-01 -6.70557082e-01 -5.09204149e-01 6.68210909e-02
5.82580209e-01 -8.74161303e-01 6.98067904e-01 -9.09047544e-01
-5.24874330e-01 9.26413655e-01 -3.35774124e-01 -1.11119993e-01
1.83652833e-01 3.79895836e-01 6.83489218e-02 -6.29469603e-02
4.91924256e-01 4.83433992e-01 7.42367566e-01 -1.25737154e+00
-6.25731528e-01 -5.35522997e-01 -1.22787125e-01 3.03094953e-01
-7.54976034e-01 -1.65400598e-02 -3.51088196e-01 -4.28421795e-01
1.76623359e-01 -1.32633746e+00 -3.34491611e-01 -2.06152216e-01
-5.39942265e-01 -5.72989047e-01 4.42087650e-01 -2.98382580e-01
1.15652096e+00 -1.91985810e+00 6.37557983e-01 6.78370535e-01
1.06281102e+00 -3.90249312e-01 -3.26765925e-01 3.05460274e-01
-1.87635273e-01 3.63653183e-01 -1.78183734e-01 -4.48758245e-01
-3.18839401e-01 2.81751202e-03 1.55136392e-01 6.31924093e-01
1.78128183e-02 7.69349694e-01 -8.75122666e-01 -6.03364944e-01
-5.09092093e-01 3.37495953e-02 -6.52110815e-01 -7.90292099e-02
5.67361005e-02 1.88890189e-01 -4.78979856e-01 5.04654408e-01
5.74350893e-01 -1.09112751e+00 8.88548315e-01 -7.02351183e-02
4.28845584e-01 1.68249086e-02 -9.12516713e-01 1.38766551e+00
-2.07996547e-01 4.91056114e-01 3.57114285e-01 -1.32519317e+00
4.79642361e-01 2.87678987e-01 8.48359466e-01 -1.66754145e-02
-2.55714506e-01 3.00342381e-01 4.24478859e-01 2.63820272e-02
5.94230928e-02 -1.98115304e-01 -1.04684770e-01 7.37649977e-01
2.71686260e-02 4.06476974e-01 1.12693839e-01 8.06803644e-01
1.63357747e+00 -5.20560503e-01 1.65365085e-01 -6.42599463e-01
3.09864819e-01 -2.22745799e-02 3.70295316e-01 8.34302545e-01
-3.52445900e-01 8.71158857e-03 8.13354135e-01 -1.79742128e-01
-8.02379191e-01 -9.59860921e-01 -3.31694377e-03 1.78467166e+00
6.13956191e-02 -4.35514748e-01 -5.74523389e-01 -8.53579700e-01
3.11206639e-01 -2.48498961e-01 -6.49714112e-01 -1.64342746e-01
-3.47667068e-01 -1.19238329e+00 3.59199941e-01 2.60639399e-01
-2.60344356e-01 -6.38935328e-01 3.04863095e-01 8.07415396e-02
-4.83522058e-01 -1.04788351e+00 -9.21921492e-01 7.14901447e-01
-8.28267395e-01 -1.20670593e+00 -5.94575405e-01 -1.29997993e+00
6.29000366e-01 7.67996669e-01 1.19064832e+00 4.51714993e-01
-8.56569931e-02 3.79828036e-01 -4.05547544e-02 2.71873251e-02
-3.85242075e-01 6.32154286e-01 2.63733983e-01 1.82143599e-01
1.88023120e-01 -7.38052309e-01 -2.79105693e-01 4.11622018e-01
-5.34301937e-01 2.51884311e-01 6.15912974e-01 8.66725087e-01
3.48354846e-01 -1.17495716e-01 8.32764983e-01 -1.16109717e+00
7.41218328e-01 -7.89050639e-01 -3.73484164e-01 6.84014976e-01
-8.20689142e-01 1.26260027e-01 3.31724077e-01 -6.86114609e-01
-4.99025583e-01 3.17334563e-01 5.88704884e-01 -3.21834087e-01
4.94846106e-01 7.90192068e-01 1.47228278e-02 -2.99427778e-01
6.40294552e-01 -2.05723122e-02 5.73445633e-02 -7.81399086e-02
1.97532088e-01 4.74777371e-01 6.72699660e-02 -8.86761606e-01
7.87466586e-01 3.01219195e-01 2.68348813e-01 -7.23686159e-01
-7.72633374e-01 -9.13830996e-01 -7.54347146e-01 -4.86139774e-01
8.45504820e-01 -1.01987898e+00 -1.22976983e+00 2.14124709e-01
-1.32224047e+00 -5.86417556e-01 5.88850975e-01 2.69960642e-01
-2.90010810e-01 6.33240998e-01 -9.29605365e-01 -7.82276750e-01
7.06983637e-03 -1.07220221e+00 1.04465461e+00 -5.14530778e-01
-1.70993549e-03 -1.47118032e+00 2.29024142e-01 4.51844454e-01
-4.20619771e-02 -1.36522681e-01 1.37818110e+00 -8.76025736e-01
-7.36205757e-01 1.22236297e-01 -4.19038147e-01 -2.71740943e-01
-1.98703445e-02 -3.69387269e-01 -6.89915955e-01 -7.91880310e-01
-2.12164760e-01 -4.92259979e-01 1.24546134e+00 5.44815481e-01
9.66436148e-01 -7.41387382e-02 -1.08943748e+00 2.99595416e-01
1.22920513e+00 -2.73756921e-01 7.29357153e-02 -5.49356118e-02
1.13435078e+00 8.37214828e-01 3.43999386e-01 1.42334983e-01
5.58399320e-01 7.05837429e-01 1.44950122e-01 -1.85768232e-01
1.52318269e-01 -8.04920308e-03 4.44234133e-01 1.08310795e+00
-2.58351192e-02 -3.80129963e-01 -1.27964807e+00 4.02654171e-01
-2.13895202e+00 -7.07406163e-01 -6.81705415e-01 2.15882850e+00
9.06102896e-01 1.72401682e-01 5.08180201e-01 -4.94369157e-02
1.18349671e+00 -9.79867950e-02 -4.54723746e-01 8.14189464e-02
-5.30399680e-02 -6.84599280e-02 7.25070834e-01 7.27720976e-01
-1.20276022e+00 8.62421274e-01 6.62436152e+00 1.05131495e+00
-7.44767487e-01 4.56690639e-01 7.84976363e-01 1.53890893e-01
-3.20903897e-01 1.52368963e-01 -6.18594229e-01 1.34783670e-01
8.18997562e-01 -3.35408777e-01 5.29430985e-01 5.51986933e-01
-1.73138753e-01 1.12258293e-01 -1.39882576e+00 5.94071090e-01
6.00824803e-02 -1.09930253e+00 -4.54455838e-02 6.40487254e-01
7.63200641e-01 1.44328073e-01 -9.01174173e-02 2.09036842e-01
7.42510140e-01 -8.43188167e-01 2.65805483e-01 -1.87428459e-03
6.26379073e-01 -4.58630532e-01 2.95648366e-01 6.38148427e-01
-1.52980065e+00 -1.17244653e-01 -2.97140002e-01 -2.43312642e-01
-1.07314207e-01 7.02682555e-01 -9.98392522e-01 4.23408419e-01
4.55014795e-01 8.32555413e-01 -6.81327224e-01 8.25942338e-01
3.48723888e-01 4.12702233e-01 -1.71085045e-01 -2.21442636e-02
-2.09547028e-01 -3.00467610e-01 3.84891868e-01 1.03419507e+00
1.53829036e-02 -2.43839309e-01 9.32066917e-01 4.88334030e-01
-3.30624700e-01 3.62928629e-01 -7.66408205e-01 1.73326761e-01
3.29540431e-01 1.45853877e+00 -1.16747367e+00 -2.81556159e-01
-4.97475386e-01 1.07764256e+00 9.01494801e-01 3.90724599e-01
-7.58909464e-01 8.91460776e-02 4.09250945e-01 1.54017672e-01
-3.47200967e-02 -1.13964386e-01 -3.58212113e-01 -1.19388759e+00
-1.18209921e-01 -5.26832104e-01 7.11562216e-01 -3.54981631e-01
-1.76057649e+00 1.93739995e-01 -1.48498997e-01 -8.15543890e-01
9.61183086e-02 -5.25747836e-01 -4.46523368e-01 6.73735619e-01
-1.20739079e+00 -1.21359253e+00 -5.68221584e-02 6.44270957e-01
2.61026174e-01 6.68960391e-03 6.08494341e-01 4.50642914e-01
-5.99172771e-01 5.89612365e-01 6.22947440e-02 -4.25360836e-02
1.00457549e+00 -1.35749388e+00 1.70278370e-01 6.59386873e-01
1.94083661e-01 4.39536691e-01 3.37026745e-01 -9.78128552e-01
-1.16509998e+00 -1.10743821e+00 1.02235770e+00 -5.74744046e-01
1.08870828e+00 -8.71582806e-01 -9.90722954e-01 8.74821544e-01
-1.52099341e-01 1.95275340e-02 9.50755119e-01 8.62742722e-01
-5.39750099e-01 4.25723754e-02 -5.76313615e-01 2.65683621e-01
1.36335337e+00 -7.62987614e-01 8.28417316e-02 9.23807800e-01
6.87423348e-01 2.61149317e-01 -9.79157567e-01 3.86725008e-01
4.53799069e-01 -7.09832549e-01 8.79834414e-01 -6.86934114e-01
2.65992492e-01 -1.45810530e-01 3.17087024e-02 -1.27746236e+00
-8.14126015e-01 -4.70863849e-01 1.08879432e-01 8.19967270e-01
7.50636339e-01 -6.52770996e-01 9.60677922e-01 2.35694483e-01
2.61134934e-03 -5.38072109e-01 -8.70952368e-01 -7.69995034e-01
2.18446374e-01 -5.91622889e-02 -2.35965714e-01 1.35632539e+00
2.18451411e-01 1.23475730e+00 -5.41928649e-01 2.24580809e-01
9.40892577e-01 1.74785793e-01 6.45243227e-01 -1.90911460e+00
-5.72014272e-01 -4.28533763e-01 -1.52452469e-01 -1.02617276e+00
6.41143501e-01 -1.63630629e+00 7.04408586e-02 -1.35829794e+00
1.11771655e+00 -7.59224057e-01 -3.36455911e-01 6.16953075e-01
-5.15245080e-01 3.43982190e-01 -9.51091200e-03 6.98684216e-01
-1.05692863e+00 2.24998351e-02 1.10052431e+00 -3.01604301e-01
-1.39009520e-01 5.69264479e-02 -6.13475740e-01 4.76061910e-01
5.50443709e-01 -7.87756026e-01 -4.50271219e-01 -1.93720549e-01
5.60222149e-01 2.80136645e-01 2.08885118e-01 -6.26279891e-01
5.94484746e-01 -9.13585201e-02 1.49205476e-01 -3.11718374e-01
1.72086626e-01 -6.11323297e-01 2.24985674e-01 8.17744911e-01
-4.32049245e-01 5.34895435e-02 -3.22058707e-01 1.13876867e+00
3.70349258e-01 2.19228610e-01 5.71570933e-01 2.82934904e-01
-2.15388194e-01 4.72803712e-01 -5.58919191e-01 -8.01072177e-03
1.07768106e+00 -2.09602788e-02 -4.20624167e-01 -2.62846291e-01
-1.00255513e+00 6.77820086e-01 4.78336215e-01 1.37889773e-01
2.95050830e-01 -1.25117290e+00 -8.34733248e-01 -9.04570892e-02
2.88837075e-01 -3.81376743e-01 1.33276999e-01 1.29577374e+00
7.49873891e-02 5.49867786e-02 6.94736317e-02 -1.04564500e+00
-1.74795222e+00 7.83044279e-01 1.86339632e-01 -8.06747139e-01
4.44248952e-02 6.91342890e-01 8.37588847e-01 -4.18263912e-01
1.85506374e-01 1.70103893e-01 -1.23154834e-01 2.67480522e-01
-3.36825438e-02 2.60002792e-01 3.79166082e-02 -4.44298923e-01
-5.93482316e-01 5.32490313e-01 -3.11911732e-01 8.25927556e-02
1.11235821e+00 -8.86629447e-02 -6.09921575e-01 4.83270973e-01
1.26210046e+00 -1.42707884e-01 -7.29438484e-01 -4.42951590e-01
5.48142493e-01 -1.84051737e-01 -2.70861864e-01 -3.86122644e-01
-8.67000639e-01 7.09852695e-01 2.89886922e-01 6.12230718e-01
9.46686208e-01 4.59830076e-01 2.03331470e-01 5.09315789e-01
3.33146602e-01 -8.19476128e-01 6.17429733e-01 5.07028699e-01
3.89482468e-01 -1.17942691e+00 7.94910491e-02 -1.10114110e+00
-4.35100347e-01 6.82105064e-01 5.44194520e-01 9.07617956e-02
9.64134991e-01 2.09658653e-01 -4.54872638e-01 -4.97547507e-01
-1.09321737e+00 -3.45462531e-01 2.75106937e-01 6.98830962e-01
6.00953937e-01 3.64279389e-01 -2.19618037e-01 4.33052331e-01
3.52081776e-01 -3.67398649e-01 4.86667782e-01 4.87211108e-01
-5.56772947e-01 -1.35713196e+00 -2.07932442e-01 9.15537417e-01
-3.52332771e-01 -2.91874856e-01 -6.62585497e-01 2.25588113e-01
-2.19235852e-01 9.26884890e-01 5.01427762e-02 -6.46862626e-01
-1.84300572e-01 2.06394896e-01 4.34029996e-01 -8.51263106e-01
-5.72328508e-01 3.42324764e-01 1.54589951e-01 -1.05624497e-01
-4.82424766e-01 -6.37280047e-01 -8.47119927e-01 -5.21138728e-01
-6.76162481e-01 2.53773272e-01 2.73120224e-01 8.44383240e-01
3.17449778e-01 4.40128684e-01 7.94653296e-01 -7.77477682e-01
-5.05587101e-01 -8.45019102e-01 -7.54022419e-01 3.55218351e-01
1.18034527e-01 -7.29156196e-01 -4.59077328e-01 2.34899763e-02] | [7.180486679077148, 5.435795307159424] |
8915797a-3044-4dce-ad94-87eed0cbad9a | blood-oxygen-saturation-estimation-from | 2212.07116 | null | https://arxiv.org/abs/2212.07116v2 | https://arxiv.org/pdf/2212.07116v2.pdf | Blood Oxygen Saturation Estimation from Facial Video via DC and AC components of Spatio-temporal Map | Peripheral blood oxygen saturation (SpO2), an indicator of oxygen levels in the blood, is one of the most important physiological parameters. Although SpO2 is usually measured using a pulse oximeter, non-contact SpO2 estimation methods from facial or hand videos have been attracting attention in recent years. In this paper, we propose an SpO2 estimation method from facial videos based on convolutional neural networks (CNN). Our method constructs CNN models that consider the direct current (DC) and alternating current (AC) components extracted from the RGB signals of facial videos, which are important in the principle of SpO2 estimation. Specifically, we extract the DC and AC components from the spatio-temporal map using filtering processes and train CNN models to predict SpO2 from these components. We also propose an end-to-end model that predicts SpO2 directly from the spatio-temporal map by extracting the DC and AC components via convolutional layers. Experiments using facial videos and SpO2 data from 50 subjects demonstrate that the proposed method achieves a better estimation performance than current state-of-the-art SpO2 estimation methods. | ['Hitoshi Imaoka', 'Yoshifumi Onishi', 'Yusuke Akamatsu'] | 2022-12-14 | null | null | null | null | ['spo2-estimation'] | ['medical'] | [-1.20710187e-01 -2.66435534e-01 1.65079862e-01 -5.56492269e-01
9.82001275e-02 1.27740040e-01 -1.06166549e-01 -3.16430479e-01
-4.29562390e-01 6.69865668e-01 2.19548434e-01 1.89988166e-02
1.85344413e-01 -5.39489865e-01 -1.30673856e-01 -8.74003112e-01
-1.76906288e-02 -6.44798458e-01 1.34822473e-01 1.28702790e-01
2.18449011e-02 5.56302607e-01 -1.51779449e+00 -2.82506365e-02
8.83079469e-01 1.85028052e+00 -4.30296749e-01 6.46832764e-01
-8.99519771e-02 1.09819114e+00 -6.98742390e-01 1.02603309e-01
2.10192546e-01 -9.38450754e-01 -2.69588947e-01 -2.22110882e-01
4.67862636e-01 -5.79343915e-01 -6.97423398e-01 9.16962922e-01
8.67160857e-01 1.48911968e-01 1.87825978e-01 -1.35717356e+00
-6.48320392e-02 1.56894118e-01 -3.97028029e-01 4.95843530e-01
3.81978266e-02 6.13601863e-01 1.33737832e-01 -5.24596334e-01
-1.44144952e-01 1.05150771e+00 8.31716418e-01 5.18847823e-01
-8.80566478e-01 -7.45661139e-01 -4.21518952e-01 7.30974972e-01
-1.42580307e+00 -8.20231140e-01 1.02928638e+00 -2.51107961e-01
5.53349257e-01 1.49755016e-01 9.50892389e-01 4.83169407e-01
3.08412731e-01 4.72517967e-01 1.31507671e+00 -1.34403035e-01
-4.84543443e-02 2.41174683e-01 -4.21898544e-01 8.60308290e-01
-2.27787063e-01 2.33186278e-02 -6.10861599e-01 2.46521577e-01
8.47648025e-01 1.10317864e-01 -6.88290477e-01 1.00011751e-01
-8.66475224e-01 3.83693308e-01 6.47202730e-01 2.73205847e-01
-7.50390768e-01 2.81476796e-01 2.86995798e-01 -2.84481775e-02
3.33857298e-01 4.86807339e-02 -1.16152473e-01 -4.44822937e-01
-9.86590207e-01 -4.07529652e-01 9.08695579e-01 5.98203063e-01
5.19617558e-01 2.09472831e-02 -3.93369794e-01 7.10815966e-01
2.26951092e-01 6.73476577e-01 5.35791576e-01 -1.13709021e+00
-4.44503464e-02 3.01539093e-01 2.85101891e-01 -1.11717498e+00
-5.14865041e-01 -8.94702375e-02 -8.13546002e-01 2.00103652e-02
4.17834669e-01 -3.10552150e-01 -4.14345801e-01 1.43383241e+00
4.53107476e-01 3.11582267e-01 -1.99085191e-01 1.30574417e+00
1.17427373e+00 2.99935669e-01 7.86413252e-02 -6.55287743e-01
1.18057096e+00 -9.38608289e-01 -1.26494646e+00 1.39267907e-01
-2.43792266e-01 -3.48218083e-01 4.84993905e-01 2.63664454e-01
-1.20061219e+00 -9.84627128e-01 -9.79602933e-01 -1.26338914e-01
-1.96989655e-01 1.51459754e-01 2.61680424e-01 6.08041406e-01
-8.20811808e-01 9.17853117e-01 -1.04430389e+00 -5.13976514e-02
5.79657793e-01 8.46116915e-02 -2.33297884e-01 1.98720489e-02
-1.35825074e+00 7.63538778e-01 -2.39143334e-03 6.24154329e-01
-7.23610282e-01 -7.60970712e-01 -8.11954618e-01 8.48875195e-02
1.39102414e-01 -4.26506847e-01 8.27679574e-01 -9.91623938e-01
-2.06749344e+00 2.90706217e-01 -6.59143329e-01 -5.60283214e-02
6.11862183e-01 -2.38625616e-01 -6.07650399e-01 8.43274295e-01
-4.50208843e-01 4.40691262e-01 9.69351649e-01 -5.59813321e-01
-4.65924859e-01 -1.93918735e-01 -2.01846018e-01 3.76899424e-03
-4.50431526e-01 1.57970309e-01 -2.65728176e-01 2.30552256e-01
1.50273904e-01 -5.17128229e-01 2.48078987e-01 7.10020721e-01
-2.81174928e-01 -3.35410625e-01 5.92927516e-01 -8.63439381e-01
1.08663619e+00 -2.10261559e+00 -3.61684144e-01 -3.19211856e-02
4.73629504e-01 5.37291229e-01 1.44497395e-01 -2.36673713e-01
2.61446182e-02 -5.69355898e-02 -4.95583899e-02 -1.44664466e-01
-3.20680797e-01 -3.67600441e-01 2.47327819e-01 9.21941996e-01
3.81196946e-01 7.72405267e-01 -1.10303676e+00 -7.10413337e-01
7.27268875e-01 1.06690848e+00 -7.93180019e-02 6.97210312e-01
4.22508210e-01 8.38637590e-01 -1.08274788e-01 5.57406902e-01
8.45236361e-01 3.34885381e-02 -6.20133169e-02 -8.61501694e-01
-3.75926793e-01 3.57430369e-01 -7.74780095e-01 1.50201082e+00
-3.96643758e-01 1.11619735e+00 6.69322833e-02 -1.02758467e+00
1.15672469e+00 5.91533482e-01 9.52260852e-01 -9.47899044e-01
5.50982237e-01 2.25996062e-01 2.84925755e-02 -1.28851354e+00
-2.79666364e-01 -5.44694066e-01 7.21477389e-01 6.38205558e-02
4.57955189e-02 -6.50742054e-02 1.47191733e-02 -5.56477845e-01
7.09380627e-01 9.47001353e-02 1.58387855e-01 -2.35527322e-01
9.71466541e-01 -5.78037024e-01 6.62704945e-01 2.91431069e-01
-1.15839922e+00 5.40472925e-01 5.04628778e-01 -7.67178953e-01
-5.89118123e-01 -6.27005160e-01 -4.01480496e-01 3.02882135e-01
3.44708204e-01 -1.61227986e-01 -8.17193568e-01 -3.54269445e-01
-1.22220576e-01 2.24770471e-01 -7.46782601e-01 -2.34811425e-01
-6.94249570e-01 -2.63073832e-01 4.36942875e-01 6.41603351e-01
1.05197942e+00 -1.23075402e+00 -1.25054371e+00 2.54454523e-01
-7.46786058e-01 -1.39507532e+00 -2.57552803e-01 -3.05013414e-02
-6.67080998e-01 -1.21784198e+00 -6.02514744e-01 -2.04215392e-01
1.48115769e-01 2.42563158e-01 8.39834929e-01 -4.81807142e-02
-6.49022281e-01 1.29712541e-02 -4.51483913e-02 -7.37497330e-01
5.02473116e-02 -3.47979933e-01 2.92392131e-02 4.61281955e-01
8.53330612e-01 -7.00805128e-01 -1.32803535e+00 3.03304136e-01
-5.50416708e-01 1.94153618e-02 1.19947284e-01 1.61737934e-01
4.47779626e-01 -5.26252277e-02 3.23837638e-01 -1.65035665e-01
3.21407646e-01 -3.62405360e-01 -3.07383478e-01 -9.17161629e-02
-3.55076522e-01 -3.36904854e-01 7.60746002e-01 -3.35422099e-01
-7.78820932e-01 -2.02535942e-01 -1.00630745e-01 -7.70762861e-01
-4.65639144e-01 1.59042105e-01 -1.48080969e-02 2.81364210e-02
6.64812028e-01 2.63911843e-01 4.33027506e-01 -8.48141238e-02
-9.02691558e-02 9.45434630e-01 8.05528879e-01 4.64355722e-02
3.71785790e-01 8.31776500e-01 3.93395275e-01 -1.11735952e+00
-7.15071440e-01 -6.08966231e-01 -8.25604677e-01 -6.94486916e-01
9.61994827e-01 -9.50069070e-01 -1.29554999e+00 8.89404297e-01
-1.07661533e+00 -2.73618698e-01 -1.56458154e-01 8.08363140e-01
-4.35895175e-01 3.01981181e-01 -5.28755426e-01 -9.54678655e-01
-6.11697376e-01 -1.08319604e+00 5.12817979e-01 7.77950108e-01
6.68417290e-02 -9.73428190e-01 -1.31218389e-01 1.91716641e-01
1.14067972e+00 6.98713899e-01 2.71529526e-01 -2.79686619e-02
-2.52184011e-02 -3.04168910e-01 -3.55306774e-01 6.74947619e-01
6.04740977e-01 -5.22517376e-02 -1.43846941e+00 1.03524113e-02
2.23570004e-01 -3.13129693e-01 6.88630164e-01 7.42360532e-01
1.57894993e+00 -9.12783146e-02 -7.31145591e-02 1.16626573e+00
1.36345100e+00 2.84710497e-01 9.74917293e-01 -5.18833473e-02
6.98097467e-01 6.79523289e-01 1.41391486e-01 6.28517270e-01
2.41405815e-01 3.76549214e-01 5.42150319e-01 -2.37095132e-01
-4.64144081e-01 -6.74051940e-02 3.40900600e-01 5.02164245e-01
-3.41566205e-01 6.64895028e-02 -2.89279789e-01 3.23100448e-01
-1.26576066e+00 -8.08746517e-01 -2.23431826e-01 2.07506156e+00
1.06917453e+00 -4.90048081e-01 3.60548973e-01 3.10480744e-01
9.01415646e-01 2.81986762e-02 -9.21341836e-01 -8.14510211e-02
1.65927157e-01 4.88074213e-01 2.76670247e-01 2.35551760e-01
-1.09493124e+00 3.30255270e-01 5.81642103e+00 -3.40726376e-02
-1.75186861e+00 -5.15563414e-02 6.96133137e-01 -3.75921875e-01
2.06917882e-01 -4.36215609e-01 -4.16857898e-01 8.29319358e-01
1.29389751e+00 1.93397596e-01 5.37131608e-01 6.47685170e-01
4.91014510e-01 -1.59427896e-01 -9.88963366e-01 1.35758126e+00
2.77162105e-01 -7.76560068e-01 -6.88870549e-01 -1.87471241e-01
1.79650977e-01 -1.40349507e-01 -2.93419123e-01 -2.34667465e-01
-5.84211111e-01 -1.24797547e+00 5.87062120e-01 8.28562438e-01
1.45133221e+00 -4.78989810e-01 9.38849807e-01 -4.18497026e-02
-1.07668769e+00 -2.30633467e-01 -4.65221137e-01 1.45809874e-01
1.60566181e-01 7.54434824e-01 -3.63594770e-01 1.62717655e-01
7.81045258e-01 1.03190649e+00 -3.13398272e-01 1.32977533e+00
-4.87144858e-01 5.66295564e-01 -3.43710721e-01 -1.75596714e-01
-8.09558481e-02 -1.63475513e-01 2.13037118e-01 1.09727156e+00
2.89132297e-01 5.24451375e-01 -3.99391413e-01 1.32776856e+00
-2.99676359e-02 -1.61482096e-01 -1.81378081e-01 1.21166922e-01
4.74777967e-01 1.39675581e+00 -2.75184840e-01 -3.42446387e-01
-3.97797674e-01 7.37442911e-01 -1.35951161e-01 3.29821944e-01
-9.06033337e-01 -1.11177135e+00 6.52473509e-01 2.57720500e-01
-3.11887972e-02 1.05892710e-01 5.68857826e-02 -9.17131543e-01
-3.74743566e-02 -4.39281732e-01 -2.17287224e-02 -8.26889813e-01
-9.30864036e-01 6.53730273e-01 -2.76703954e-01 -1.00963187e+00
1.24193594e-01 -6.23128474e-01 -7.66018391e-01 1.36286128e+00
-2.40117049e+00 -3.57086182e-01 -1.58615708e+00 7.72178710e-01
2.24992231e-01 3.40730131e-01 5.50596476e-01 3.53264868e-01
-7.80813336e-01 5.56003094e-01 -3.50475818e-01 3.64015609e-01
7.71328449e-01 -8.99054527e-01 -8.15593675e-02 7.40919173e-01
-8.18478823e-01 4.79834139e-01 4.18101370e-01 -1.16536565e-01
-1.32592750e+00 -1.01836610e+00 7.50266731e-01 5.80159612e-02
2.05543861e-01 1.57115579e-01 -7.05604911e-01 9.82104540e-02
2.47727334e-01 9.76282477e-01 5.68236947e-01 -7.72781312e-01
-1.17284022e-01 -6.61586940e-01 -1.46094990e+00 -1.80737656e-02
6.10551536e-01 -5.81459522e-01 -4.81442139e-02 9.70731899e-02
4.24567789e-01 -4.17636842e-01 -1.13640821e+00 4.41056311e-01
9.07364249e-01 -1.30627084e+00 7.21597135e-01 -3.51633608e-01
5.46599507e-01 -4.33127463e-01 2.03306913e-01 -1.40170109e+00
-1.55766085e-01 -7.79650807e-01 -4.27889615e-01 7.95029044e-01
-3.00896704e-01 -7.52344728e-01 4.76134688e-01 7.30506599e-01
-8.97865817e-02 -4.93125051e-01 -7.81070471e-01 -4.26833063e-01
-9.15255621e-02 -1.63239971e-01 7.15341926e-01 6.14542663e-01
3.39330167e-01 3.72234099e-02 -3.91835153e-01 -6.45039529e-02
5.77186167e-01 -2.38122120e-01 5.75145423e-01 -1.16444731e+00
2.15497985e-01 -3.60007852e-01 -4.96597767e-01 -9.95683789e-01
1.98219009e-02 -2.03023925e-01 5.69960415e-01 -1.51099300e+00
2.67144721e-02 -2.49434188e-01 -6.97415233e-01 3.98221791e-01
-4.13821608e-01 3.88838619e-01 -5.68973087e-03 -3.92394923e-02
-2.00497687e-01 4.99341398e-01 1.36066091e+00 -6.68358356e-02
-3.65539551e-01 -2.94329762e-01 -2.85253167e-01 2.94098645e-01
8.18865478e-01 -3.30645926e-02 -1.40819907e-01 -1.98097050e-01
-3.81050140e-01 3.43463600e-01 2.79287189e-01 -1.49763691e+00
2.75240511e-01 -2.06673831e-01 8.08591127e-01 -1.60290018e-01
3.23860615e-01 -9.29603934e-01 -5.12621813e-02 6.10105097e-01
-4.45143551e-01 -2.15044007e-01 8.09855536e-02 -1.28419045e-02
-2.91003913e-01 1.57201812e-01 1.26608789e+00 -3.91258955e-01
-2.85172403e-01 6.38423324e-01 2.74889506e-02 -1.42804489e-01
7.46852934e-01 -2.15360254e-01 -5.93783140e-01 -4.97819871e-01
-4.55582500e-01 -1.75662667e-01 3.98173667e-02 1.80193424e-01
9.04709637e-01 -1.20689178e+00 -7.08838582e-01 5.79164326e-01
8.48799795e-02 -6.61984310e-02 2.76218385e-01 1.55413437e+00
-6.92893684e-01 3.25783491e-01 -4.96726573e-01 -6.35251820e-01
-9.51748908e-01 4.41617668e-01 1.14941382e+00 5.90874016e-01
-6.12705588e-01 8.73140693e-01 -2.52979666e-01 3.88137072e-01
4.76366282e-01 -2.31300220e-01 -4.94627774e-01 -1.69712633e-01
1.10488892e+00 7.79384971e-01 -1.15211539e-01 -7.10288703e-01
-4.17142361e-01 6.35267258e-01 7.80166686e-01 2.56844431e-01
1.26236832e+00 -3.92702878e-01 -3.56485367e-01 4.28380698e-01
1.60507369e+00 -3.12052310e-01 -1.66498685e+00 -1.02406465e-01
-6.85300827e-01 -7.16158807e-01 2.50581980e-01 -7.74076700e-01
-1.66319811e+00 1.57125592e+00 1.09438336e+00 2.18918264e-01
1.49568927e+00 -4.54690278e-01 1.07354403e+00 -2.29206756e-01
-1.21897981e-01 -8.12783659e-01 1.45447537e-01 -1.08697675e-01
4.67075586e-01 -1.11122251e+00 -1.87278643e-01 -3.85604382e-01
-2.65949398e-01 1.62654161e+00 7.12327480e-01 2.07781553e-01
8.49643409e-01 4.92721461e-02 3.75859529e-01 2.06281953e-02
-2.96752512e-01 -1.72237381e-01 5.93057796e-02 7.49996841e-01
6.20804846e-01 -4.14871424e-02 7.30580464e-02 6.26594573e-02
9.03251991e-02 4.35017407e-01 4.57146347e-01 5.30966938e-01
-4.70496416e-01 -1.94734365e-01 -1.12350248e-02 2.85785645e-01
-6.73304081e-01 1.61294099e-02 2.52987653e-01 3.61380249e-01
3.85923773e-01 1.53162038e+00 2.66536623e-01 -4.88427609e-01
4.63597506e-01 2.61117846e-01 5.52188933e-01 1.47737324e-01
-2.50942647e-01 -2.65867673e-02 -2.58321851e-01 -9.18982387e-01
-8.39369416e-01 -3.40678424e-01 -1.06586301e+00 -4.93850768e-01
-1.99849010e-01 1.73659801e-01 1.06983209e+00 8.83552670e-01
2.48185068e-01 7.04243541e-01 1.04223144e+00 -8.43182027e-01
-1.84913561e-01 -1.39342380e+00 -7.08090365e-01 6.33118272e-01
9.75524068e-01 -5.75046778e-01 -5.85214198e-01 1.79146022e-01] | [13.899896621704102, 2.766494035720825] |
5211d304-4503-45c9-810f-a2f447f92137 | pnlp-mixer-an-efficient-all-mlp-architecture | 2202.04350 | null | https://arxiv.org/abs/2202.04350v2 | https://arxiv.org/pdf/2202.04350v2.pdf | pNLP-Mixer: an Efficient all-MLP Architecture for Language | Large pre-trained language models based on transformer architecture have drastically changed the natural language processing (NLP) landscape. However, deploying those models for on-device applications in constrained devices such as smart watches is completely impractical due to their size and inference cost. As an alternative to transformer-based architectures, recent work on efficient NLP has shown that weight-efficient models can attain competitive performance for simple tasks, such as slot filling and intent classification, with model sizes in the order of the megabyte. This work introduces the pNLP-Mixer architecture, an embedding-free MLP-Mixer model for on-device NLP that achieves high weight-efficiency thanks to a novel projection layer. We evaluate a pNLP-Mixer model of only one megabyte in size on two multi-lingual semantic parsing datasets, MTOP and multiATIS. Our quantized model achieves 99.4% and 97.8% the performance of mBERT on MTOP and multi-ATIS, while using 170x fewer parameters. Our model consistently beats the state-of-the-art of tiny models (pQRNN), which is twice as large, by a margin up to 7.8% on MTOP. | ['Diego Antognini', 'Peter Staar', 'Damian Pascual', 'Francesco Fusco'] | 2022-02-09 | null | null | null | null | ['slot-filling'] | ['natural-language-processing'] | [ 2.90638059e-01 5.52915990e-01 -5.96311271e-01 -3.64015669e-01
-1.45379889e+00 -5.06347120e-01 2.30821788e-01 6.09784126e-02
-4.77150142e-01 3.52392137e-01 3.88954401e-01 -7.46115923e-01
5.71532190e-01 -6.67641699e-01 -8.39148462e-01 -1.71694741e-01
4.49218839e-01 7.69996166e-01 2.77359307e-01 2.51433942e-02
-2.84952343e-01 -1.91215217e-01 -1.13518667e+00 8.23050141e-01
6.99261367e-01 1.38464403e+00 3.52717817e-01 8.32822621e-01
-6.29053175e-01 9.18476403e-01 -3.95147085e-01 -6.35148466e-01
6.07600920e-02 1.16180353e-01 -7.61004031e-01 -5.99313915e-01
7.01061368e-01 -4.50169504e-01 -1.96561515e-01 4.50845838e-01
7.11798787e-01 -2.66204596e-01 1.58452287e-01 -9.93290067e-01
-4.45239395e-01 9.95964885e-01 -3.60463172e-01 4.38714363e-02
1.34640276e-01 1.01254496e-03 1.51624095e+00 -7.28899717e-01
3.01353097e-01 1.45358109e+00 9.37044322e-01 5.57612658e-01
-1.17998648e+00 -7.30133057e-01 1.06811941e-01 -8.79628360e-02
-9.23376143e-01 -8.02323639e-01 4.04094756e-01 2.56741524e-01
2.04195452e+00 7.55075589e-02 2.26578206e-01 1.32685137e+00
4.82720524e-01 1.24168348e+00 1.03483415e+00 -4.28873003e-01
1.84090570e-01 -5.80769219e-02 2.88562030e-01 9.50722337e-01
8.76938701e-02 -5.20539880e-01 -8.52725148e-01 -2.66183674e-01
3.72725397e-01 -2.36379370e-01 4.25157875e-01 1.60953060e-01
-9.44778919e-01 8.22935343e-01 1.59991115e-01 3.01994085e-01
-2.47225031e-01 5.73280334e-01 7.24732935e-01 3.97288270e-04
6.63652599e-01 2.04012662e-01 -1.14433813e+00 -8.00264359e-01
-9.59590554e-01 -1.27534689e-02 1.03170717e+00 1.14481652e+00
3.51149648e-01 2.69305650e-02 -1.65989578e-01 1.14274526e+00
3.93840790e-01 7.97008574e-01 6.46091223e-01 -9.20476139e-01
1.28122950e+00 4.86306369e-01 -2.04898462e-01 -3.51265460e-01
-4.99758214e-01 -2.68434137e-01 -5.95606208e-01 -4.07543421e-01
2.59325147e-01 -1.02527082e-01 -1.09084475e+00 1.64598572e+00
2.40648583e-01 -8.05459097e-02 3.52420146e-04 2.28890046e-01
9.12013292e-01 1.04596829e+00 6.17778063e-01 1.62629828e-01
2.00874996e+00 -1.26457477e+00 -6.66250467e-01 -9.11485732e-01
9.32728946e-01 -7.10925400e-01 1.61000431e+00 3.63420784e-01
-1.21973825e+00 -4.72147793e-01 -1.02997458e+00 -8.82426202e-01
-2.49240041e-01 2.64362156e-01 9.17038679e-01 9.78752136e-01
-1.07063055e+00 6.23100519e-01 -1.24162817e+00 -3.40991378e-01
7.19257236e-01 5.36909938e-01 -9.73466784e-02 -9.89850238e-02
-9.60541368e-01 6.23685658e-01 2.69552857e-01 -1.55262724e-01
-5.92074990e-01 -1.16955125e+00 -8.58552456e-01 4.15050745e-01
2.40082741e-01 -7.24640965e-01 1.62334800e+00 -3.26017022e-01
-1.98079515e+00 7.91363657e-01 -5.74884951e-01 -6.90579057e-01
3.59053002e-03 -6.35647178e-01 -4.42107409e-01 -2.03963066e-03
2.29848132e-01 8.61098111e-01 6.90378249e-01 -4.25422370e-01
-7.97598422e-01 -3.61032188e-01 1.15365081e-01 1.53228238e-01
-6.47247374e-01 1.59767777e-01 -6.23061419e-01 -3.25910062e-01
6.33145869e-02 -8.29713345e-01 -1.24786600e-01 -1.67014211e-01
-3.81172419e-01 -4.44081753e-01 7.23453462e-01 -9.86520350e-01
1.25885797e+00 -2.04709363e+00 -3.47238839e-01 -4.57238853e-01
1.82769895e-01 2.32103288e-01 -1.64271101e-01 2.59928077e-01
4.74526942e-01 3.06842059e-01 -6.20051064e-02 -9.19744551e-01
4.43738312e-01 4.20831978e-01 -4.20223117e-01 -1.34452954e-01
2.00757235e-01 1.22988474e+00 -6.41173184e-01 -6.63872361e-01
1.51317269e-01 5.21173477e-01 -7.33600855e-01 -2.24991180e-02
-4.83011574e-01 -1.55067086e-01 -5.11373639e-01 9.77102518e-01
4.19481844e-01 -5.22613108e-01 6.17690921e-01 -4.09829646e-01
1.34583801e-01 1.31352031e+00 -5.53513587e-01 2.15001726e+00
-1.17724192e+00 5.85344970e-01 5.60149141e-02 -7.20509827e-01
5.09595335e-01 5.06894350e-01 4.37535942e-01 -1.26965380e+00
9.09777358e-02 5.17396986e-01 -3.72714669e-01 -7.59445950e-02
7.91635513e-01 -1.74017727e-01 -5.34572959e-01 4.25733149e-01
4.06096637e-01 -3.64755876e-02 4.34557945e-02 3.21269929e-02
1.44555461e+00 1.40040115e-01 8.13860446e-02 -1.85588554e-01
-2.81995293e-02 -2.24234700e-01 6.01201594e-01 6.87900186e-01
-1.90545395e-01 2.89098382e-01 4.58680958e-01 -4.13022161e-01
-1.09575999e+00 -1.17719233e+00 -2.25256309e-01 1.47165978e+00
-3.79492193e-01 -8.52951705e-01 -8.37145627e-01 -7.56969571e-01
-3.26472938e-01 6.61315918e-01 1.60832908e-02 2.50020474e-01
-9.91481423e-01 -1.09435523e+00 9.12041903e-01 8.55698764e-01
8.11375916e-01 -9.21706438e-01 -6.61689878e-01 6.13291442e-01
-4.62662280e-01 -1.78468740e+00 -3.04071218e-01 4.94921863e-01
-1.24210942e+00 -5.14011919e-01 -3.15049767e-01 -8.89794528e-01
1.29569083e-01 -3.64383131e-01 1.59676969e+00 -4.98505622e-01
4.92412560e-02 -2.58923359e-02 -1.40422434e-01 -3.11546326e-01
-3.28564942e-01 7.08898246e-01 -1.48984373e-01 -5.05144656e-01
5.26448548e-01 -7.17198789e-01 -4.68260050e-01 1.67595185e-02
-3.95728022e-01 1.97594851e-01 8.39907885e-01 6.98483646e-01
7.78210461e-01 -1.19870096e-01 3.08103979e-01 -9.83518183e-01
3.28663379e-01 -2.92092294e-01 -3.76299769e-01 2.10896954e-01
-6.22829676e-01 4.12927091e-01 8.62367809e-01 -5.07171929e-01
-1.08675587e+00 6.70156926e-02 -7.21479595e-01 1.95731577e-02
3.82842898e-01 1.88684985e-01 -3.82673621e-01 2.62125254e-01
4.28265601e-01 -1.93339214e-02 -5.98916233e-01 -6.91956818e-01
5.58264434e-01 6.92668080e-01 2.81761676e-01 -7.95782506e-01
2.82966882e-01 3.43045384e-01 -2.54422605e-01 -7.35851705e-01
-1.00030959e+00 -3.78164560e-01 -1.85637191e-01 6.87594712e-01
9.04472232e-01 -1.18735003e+00 -7.79096484e-01 1.98978066e-01
-1.20292962e+00 -7.99543142e-01 -3.91653061e-01 1.95224971e-01
-5.20319283e-01 1.99311405e-01 -1.14762926e+00 -5.49055219e-01
-1.06333756e+00 -9.01159346e-01 1.64924967e+00 3.23443413e-02
-3.42565835e-01 -9.10786808e-01 -1.45678237e-01 7.91152358e-01
6.21109784e-01 -2.83853233e-01 1.21106362e+00 -4.48447436e-01
-5.68349659e-01 1.38016809e-02 -2.99750924e-01 3.66599202e-01
-2.01911122e-01 -7.22986758e-01 -1.38406146e+00 2.13662162e-02
8.86419043e-02 -5.51733553e-01 7.11822152e-01 5.29629767e-01
1.16530323e+00 -3.84280473e-01 -5.80304801e-01 6.15276098e-01
1.27692926e+00 -5.32362238e-02 7.17161596e-01 8.74548256e-02
8.24657321e-01 7.93334618e-02 3.00175011e-01 6.21677525e-02
7.07004964e-01 8.39435041e-01 1.20949008e-01 1.01432554e-01
-3.34775686e-01 -7.70366192e-01 5.86015105e-01 1.35307062e+00
3.42945963e-01 -5.43666244e-01 -8.96709263e-01 5.32555759e-01
-1.58761454e+00 -3.57004613e-01 1.21103689e-01 1.77778780e+00
8.86909783e-01 6.26768470e-01 -2.83478826e-01 9.55251139e-03
6.12096786e-02 5.23113191e-01 -5.36114097e-01 -1.01727867e+00
2.21971851e-02 8.15867662e-01 7.79462576e-01 4.81676280e-01
-1.11966312e+00 1.26318407e+00 6.02970982e+00 1.27681136e+00
-9.69914675e-01 8.03047180e-01 9.43808913e-01 -2.69334793e-01
-2.24507630e-01 -5.65322712e-02 -1.44730103e+00 4.84881669e-01
1.79250574e+00 3.76470506e-01 3.69136035e-01 1.13623321e+00
-1.94182083e-01 1.44871682e-01 -1.21269763e+00 1.19971001e+00
-9.85839814e-02 -1.29728949e+00 -8.28424543e-02 1.33165643e-01
4.63047594e-01 6.16917789e-01 2.27205604e-01 9.17417288e-01
2.63940990e-01 -9.31773126e-01 6.20919645e-01 -4.09836441e-01
9.43739831e-01 -4.12425190e-01 5.16042471e-01 2.52590060e-01
-1.49998224e+00 -1.66352764e-01 -4.42918956e-01 3.01484801e-02
5.04903316e-01 7.89136827e-01 -9.06543791e-01 8.16765353e-02
8.90701532e-01 4.24960375e-01 -2.11215168e-01 1.34196982e-01
-2.12245405e-01 1.07822621e+00 -8.34987521e-01 -1.11882024e-01
2.88341731e-01 3.97025257e-01 3.34694050e-02 1.18723130e+00
3.38214487e-01 -5.52326627e-02 -1.10209323e-01 6.25558734e-01
-6.37720883e-01 9.05201733e-02 -1.36059940e-01 -3.31447363e-01
5.57137370e-01 1.22598588e+00 -6.52050197e-01 -5.75293422e-01
-4.44361627e-01 1.21089268e+00 4.11069691e-01 -1.63279101e-01
-1.01141834e+00 -1.16196983e-01 6.21485829e-01 1.03937350e-01
6.03634775e-01 -2.05591515e-01 -5.23600817e-01 -1.23160088e+00
5.60743868e-01 -9.51953173e-01 2.71440208e-01 -4.77203548e-01
-1.01522171e+00 7.57784069e-01 -3.06138307e-01 -7.76222706e-01
-3.07139307e-01 -9.54544902e-01 -2.49883622e-01 6.60528481e-01
-1.67215705e+00 -1.51816630e+00 2.34694630e-01 5.46646751e-02
8.67901444e-01 1.64881364e-01 1.25441515e+00 7.79909849e-01
-4.61087972e-01 8.83283019e-01 1.21820271e-01 -1.25445938e-02
2.26954311e-01 -1.19384325e+00 1.14169395e+00 5.15910566e-01
3.87457192e-01 4.73643214e-01 2.24455625e-01 -3.01756889e-01
-1.67677581e+00 -9.79923248e-01 1.59567547e+00 -9.06241536e-01
6.41650319e-01 -1.15442932e+00 -4.30801511e-01 9.18224633e-01
5.72550483e-02 2.10253835e-01 6.01767123e-01 6.48532689e-01
-5.25665402e-01 -3.10251117e-01 -1.19025004e+00 6.16380036e-01
1.26338506e+00 -8.67145121e-01 -3.81238520e-01 5.97116649e-01
1.24349439e+00 -6.21534407e-01 -1.09424651e+00 3.03667217e-01
8.45995665e-01 -5.45872748e-01 1.09679019e+00 -2.80858427e-01
4.97212410e-01 3.00590962e-01 -4.73285317e-01 -6.92712367e-01
-3.30751389e-02 -6.76265061e-01 -5.49196422e-01 1.23789644e+00
7.89431930e-01 -7.09760249e-01 1.25578678e+00 6.62685990e-01
-2.13452026e-01 -9.90706146e-01 -1.28410482e+00 -7.44028151e-01
1.87312409e-01 -9.58460391e-01 6.35733068e-01 3.25590879e-01
2.69630533e-02 9.00546908e-01 -2.40920812e-01 -3.62992771e-02
3.44090134e-01 -3.54497246e-02 4.88977462e-01 -8.44839752e-01
-7.86375821e-01 -1.69823393e-01 -7.45791495e-02 -1.69686019e+00
2.35581487e-01 -7.93892801e-01 -1.26762986e-02 -1.62618351e+00
8.04025903e-02 -8.40720594e-01 -3.15367550e-01 8.37450325e-01
1.80786714e-01 3.14289212e-01 2.95461476e-01 -1.01032719e-01
-7.85890877e-01 4.19555306e-01 6.20864391e-01 -3.85561377e-01
-1.57839671e-01 -3.13121974e-01 -7.39938080e-01 7.44157672e-01
8.21133375e-01 -6.82809234e-01 -4.85395491e-01 -1.08132839e+00
4.12897319e-01 -2.57428661e-02 -1.14372142e-01 -1.06836033e+00
-4.51291800e-02 5.02246559e-01 7.51742572e-02 -5.33833444e-01
8.68749380e-01 -7.09310949e-01 -3.53449494e-01 3.30146194e-01
-2.08132863e-01 1.03141397e-01 6.25257909e-01 2.11197153e-01
-8.72447621e-03 -4.99523617e-02 2.97417790e-01 -9.53052044e-02
-5.64215362e-01 1.92412034e-01 -2.31487706e-01 3.95666748e-01
2.68209517e-01 7.52948271e-03 -4.88245249e-01 -4.84020077e-03
-3.00845981e-01 -9.85081941e-02 -2.11202309e-01 4.76786345e-01
1.83452815e-01 -9.79524553e-01 -3.15057397e-01 1.28553599e-01
-3.69863771e-02 1.85851790e-02 3.03006321e-01 6.07736647e-01
-4.57464457e-01 1.00354111e+00 1.55979529e-01 -6.12061203e-01
-1.02952981e+00 7.93211684e-02 7.10875764e-02 -1.18834758e+00
-7.81515479e-01 9.81257498e-01 2.03797370e-01 -7.71076918e-01
6.57721385e-02 -1.19091260e+00 3.33782256e-01 -3.96187752e-01
4.98650938e-01 2.16901079e-01 3.74234915e-01 -1.60921767e-01
-4.38541234e-01 4.86787617e-01 -1.15419827e-01 -1.22965574e-01
1.26334214e+00 7.52764344e-02 1.22430034e-01 3.62675279e-01
1.45046103e+00 4.24939208e-02 -1.09049439e+00 -2.49822155e-01
1.95005722e-02 1.19016379e-01 1.38606712e-01 -1.08234227e+00
-7.05217540e-01 1.08055270e+00 6.24536097e-01 -2.73672998e-01
1.02340949e+00 1.32802784e-01 1.68307161e+00 6.18715227e-01
6.49728179e-01 -1.30380917e+00 7.56824017e-03 6.53161108e-01
3.50265294e-01 -1.28269696e+00 -1.59426853e-01 -5.00355661e-01
-3.62507969e-01 7.50043452e-01 2.70967901e-01 7.89411888e-02
6.08937621e-01 6.25429571e-01 -2.12013394e-01 -6.04225285e-02
-1.00634527e+00 2.64084607e-01 5.94960600e-02 4.71546918e-01
6.33438826e-01 4.17862654e-01 2.41057780e-02 8.99510205e-01
-4.37244415e-01 6.27677888e-02 -1.49112493e-01 1.00321376e+00
-2.41001159e-01 -1.39763844e+00 1.81325465e-01 6.53727055e-01
-8.95835519e-01 -7.58927524e-01 2.39152372e-01 5.24797142e-01
3.66381332e-02 9.93436635e-01 1.55834258e-01 -2.25850508e-01
2.44322956e-01 3.93686980e-01 4.26010996e-01 -7.14209259e-01
-7.55172253e-01 -1.33508936e-01 6.56882107e-01 -1.05264521e+00
-4.66846339e-02 -3.36471319e-01 -1.35497344e+00 -3.70764732e-01
-4.33551520e-02 -2.15668425e-01 1.08397949e+00 8.49366307e-01
7.61526704e-01 6.24557555e-01 -4.20000255e-02 -4.77294981e-01
-6.22568250e-01 -9.67890620e-01 -2.75050610e-01 -2.05480322e-01
-9.61135998e-02 -9.94690284e-02 1.87534634e-02 -1.12208910e-01] | [8.774942398071289, 3.72241473197937] |
9167449a-ae86-472e-b6f9-3fb7d5214676 | approximate-spectral-clustering-with | 2302.11297 | null | https://arxiv.org/abs/2302.11297v1 | https://arxiv.org/pdf/2302.11297v1.pdf | Approximate spectral clustering with eigenvector selection and self-tuned $k$ | The recently emerged spectral clustering surpasses conventional clustering methods by detecting clusters of any shape without the convexity assumption. Unfortunately, with a computational complexity of $O(n^3)$, it was infeasible for multiple real applications, where $n$ could be large. This stimulates researchers to propose the approximate spectral clustering (ASC). However, most of ASC methods assumed that the number of clusters $k$ was known. In practice, manual setting of $k$ could be subjective or time consuming. The proposed algorithm has two relevance metrics for estimating $k$ in two vital steps of ASC. One for selecting the eigenvectors spanning the embedding space, and the other to discover the number of clusters in that space. The algorithm used a growing neural gas (GNG) approximation, GNG is superior in preserving input data topology. The experimental setup demonstrates the efficiency of the proposed algorithm and its ability to compete with similar methods where $k$ was set manually. | ['Masahiro Takatsuka', 'Mashaan Alshammari'] | 2023-02-22 | null | null | null | null | ['graph-clustering', 'spectral-graph-clustering', 'graph-partitioning'] | ['graphs', 'graphs', 'graphs'] | [-8.12590420e-02 -1.33755267e-01 2.12050945e-01 6.25537485e-02
-4.04012114e-01 -6.55570924e-01 4.52843681e-02 1.02687150e-01
-3.44338477e-01 4.61797118e-01 -2.75573462e-01 -2.60273933e-01
-5.59670687e-01 -7.66834199e-01 -3.34524632e-01 -1.08643401e+00
-3.39134544e-01 4.95007008e-01 7.17553049e-02 1.49739712e-01
6.08302653e-01 3.92237306e-01 -1.73207235e+00 -1.60133794e-01
1.00601327e+00 9.37920272e-01 3.63376826e-01 3.43044907e-01
-3.77348274e-01 2.04009846e-01 -4.01432544e-01 -1.97116539e-01
7.12543130e-01 -5.76395273e-01 -5.83963513e-01 3.08342069e-01
-3.02815825e-01 4.56549615e-01 -5.41570112e-02 1.34502637e+00
3.55795234e-01 4.08569187e-01 6.75448239e-01 -1.51328170e+00
-4.94439900e-01 8.13079596e-01 -1.00562072e+00 6.49057776e-02
8.05520511e-04 -8.29152763e-02 7.89610088e-01 -8.31474304e-01
4.86057192e-01 9.21542764e-01 8.30660939e-01 2.29291067e-01
-9.48957801e-01 -1.01335919e+00 3.05504464e-02 1.48297444e-01
-2.20485520e+00 -1.12623893e-01 1.13601553e+00 -5.04930317e-01
6.64127767e-01 1.57872885e-01 7.14589834e-01 1.83431208e-01
-4.26039070e-01 3.34519923e-01 1.25912023e+00 -7.16346562e-01
5.02856910e-01 4.69071656e-01 -6.00377023e-02 7.22193778e-01
4.24769759e-01 -2.70000547e-01 -2.72324234e-01 -3.14144939e-01
6.55357659e-01 -9.96375754e-02 -1.66097388e-01 -6.40863061e-01
-9.21914339e-01 9.43367243e-01 3.31223756e-01 3.82062316e-01
-1.50959089e-01 4.74161953e-02 3.06537300e-01 1.04877882e-01
6.18678518e-03 5.17488539e-01 -2.60205269e-01 -1.19383052e-01
-1.09874845e+00 -2.06009716e-01 3.83555889e-01 1.00624120e+00
7.30282664e-01 1.56626925e-01 7.50769675e-01 6.98469520e-01
1.26394495e-01 1.45754218e-01 6.35017455e-01 -8.32045853e-01
4.70581532e-01 1.13688576e+00 2.04511866e-01 -1.44744563e+00
-3.48452181e-01 -2.49107033e-01 -1.16915452e+00 7.99960420e-02
5.44128895e-01 -2.02685341e-01 -6.33944690e-01 1.62134087e+00
3.77894908e-01 2.42772087e-01 -1.01680934e-01 7.94569731e-01
2.44669944e-01 5.93684375e-01 -1.58601403e-01 -5.97582877e-01
1.11133540e+00 -6.30275488e-01 -6.64734364e-01 4.51451764e-02
4.40961510e-01 -8.27580392e-01 1.09803140e+00 4.72522497e-01
-9.21819866e-01 -4.89398092e-01 -1.16269135e+00 6.62745655e-01
-5.26803017e-01 4.33944583e-01 8.29671144e-01 1.02339602e+00
-1.32625663e+00 4.60444659e-01 -7.87467897e-01 -5.94798148e-01
1.33959144e-01 7.16018498e-01 -1.84086949e-01 2.59182334e-01
-9.53415394e-01 3.77277166e-01 7.50646353e-01 3.84221196e-01
-1.96070641e-01 -4.54406021e-03 -4.42372322e-01 2.43268143e-02
3.64262462e-01 -3.67189467e-01 3.67802113e-01 -9.76708651e-01
-1.28729987e+00 6.33477569e-01 -3.41930836e-02 -2.98160583e-01
4.32096273e-01 3.12710077e-01 -2.69940764e-01 3.38353574e-01
-1.66196585e-01 3.66781920e-01 8.61119747e-01 -1.38237143e+00
-5.36262512e-01 -4.07527596e-01 -4.79898483e-01 3.90706003e-01
-5.67851067e-01 -2.04365432e-01 -5.19006968e-01 -3.39459836e-01
7.55109191e-01 -8.74044776e-01 -4.39586431e-01 -3.20456028e-01
-4.05629098e-01 -3.11359376e-01 1.02579927e+00 -5.45968056e-01
1.74693751e+00 -2.32152462e+00 7.75568783e-02 8.83307099e-01
2.18830496e-01 2.10483223e-01 3.26330841e-01 6.60645068e-01
-2.85239488e-01 3.38289112e-01 -4.57699090e-01 4.53281887e-02
-1.03697628e-01 -2.80054539e-01 -1.81202416e-03 6.98518097e-01
-3.06015044e-01 2.06923887e-01 -7.36662030e-01 -7.62567103e-01
2.49705151e-01 2.83850461e-01 -3.73270333e-01 3.23905386e-02
2.04898879e-01 -8.72708708e-02 -3.32986414e-01 6.78855956e-01
7.64219940e-01 -5.07111907e-01 5.21372914e-01 -2.36232460e-01
-1.48506492e-01 -5.60821950e-01 -2.02018309e+00 1.28752065e+00
5.59191145e-02 3.51156026e-01 3.54017973e-01 -1.42438316e+00
1.18740320e+00 3.08791280e-01 9.70075965e-01 1.16324350e-02
3.24746937e-01 3.47070903e-01 1.57045677e-01 -5.00726163e-01
3.02589208e-01 -1.36116013e-01 1.18727878e-01 5.87817252e-01
-3.43744814e-01 9.60562527e-02 1.66166201e-01 1.87611416e-01
9.85396385e-01 -1.79426864e-01 3.90564173e-01 -5.65874696e-01
5.97900093e-01 1.06541224e-01 5.27959645e-01 4.84319776e-01
-4.78295743e-01 4.20038760e-01 2.56752849e-01 -2.98533142e-01
-9.86075044e-01 -9.24279451e-01 1.11195259e-01 5.69972217e-01
5.47034085e-01 -2.46697277e-01 -1.12405360e+00 -4.66181010e-01
-1.94255397e-01 3.53311270e-01 -5.63409865e-01 -1.15337260e-01
-4.56547529e-01 -1.05455959e+00 4.26331669e-01 2.56624371e-01
6.19889140e-01 -1.07847238e+00 -6.99615955e-01 8.90029520e-02
-2.06577688e-01 -6.65933013e-01 -4.19862717e-01 3.27413410e-01
-9.87519085e-01 -1.43192136e+00 -3.07554960e-01 -1.01853418e+00
1.19104302e+00 4.26481903e-01 6.83975220e-01 2.36648992e-01
-4.17344600e-01 -1.01484160e-03 -5.13391256e-01 -1.23448804e-01
-5.22432514e-02 1.18755534e-01 2.70054609e-01 9.83091071e-02
7.75657177e-01 -7.35237360e-01 -8.03065479e-01 4.37880278e-01
-7.59443402e-01 -1.91484034e-01 6.38189256e-01 6.48717284e-01
4.64563817e-01 1.08989656e+00 5.91491461e-01 -7.00018644e-01
8.49224925e-01 -3.49900156e-01 -5.66627145e-01 2.32395053e-01
-9.41308260e-01 -1.75883006e-02 1.06723607e+00 -3.25909615e-01
-6.31708145e-01 4.58763838e-01 2.63427198e-01 -5.19089401e-01
-3.44304405e-02 3.84873539e-01 -1.04321361e-01 7.61594847e-02
5.84528923e-01 3.33571672e-01 -3.60206515e-02 -2.82172590e-01
2.54839122e-01 7.20520794e-01 2.70558625e-01 -4.66259986e-01
9.22812223e-01 4.77126747e-01 -4.11583558e-02 -6.97584808e-01
-3.62954801e-03 -8.26289833e-01 -8.02329123e-01 -8.27742442e-02
6.70753956e-01 -5.39473236e-01 -9.43184853e-01 1.62974462e-01
-5.28024554e-01 1.83565497e-01 -5.43794446e-02 4.48450476e-01
-5.16207099e-01 6.12914741e-01 -2.61087269e-01 -1.23121583e+00
-2.39305809e-01 -1.11073852e+00 3.96305919e-01 1.73714727e-01
-1.06195835e-02 -6.96836352e-01 -2.05402017e-01 2.40442008e-01
1.42937213e-01 4.55541164e-01 1.06087518e+00 -4.51377153e-01
-2.72387564e-01 -4.16986853e-01 -2.53264844e-01 -1.90186277e-01
3.67195994e-01 2.76925325e-01 -6.08590484e-01 -5.98662257e-01
-4.08691876e-02 -2.56319996e-02 3.77007961e-01 5.88756025e-01
1.29939139e+00 -2.67190963e-01 -6.07534170e-01 5.80617666e-01
1.85725653e+00 9.04750943e-01 4.46238160e-01 2.60472983e-01
4.04871464e-01 4.56807703e-01 5.38578987e-01 5.22748113e-01
-2.65579987e-02 3.95842999e-01 3.37382495e-01 -1.78840131e-01
2.83492148e-01 -6.79520518e-02 4.22953069e-02 1.16154373e+00
-2.96876609e-01 -1.19318463e-01 -1.14117122e+00 7.33979344e-01
-1.85807848e+00 -1.02235520e+00 6.36973679e-02 2.29604173e+00
6.49130464e-01 1.69858560e-01 3.55410993e-01 6.33885980e-01
1.23409736e+00 -3.27172548e-01 -6.44275069e-01 -3.74391377e-01
7.76352957e-02 1.98149607e-01 5.65887272e-01 4.02165055e-01
-9.79369938e-01 7.95259595e-01 5.73152447e+00 9.57362831e-01
-9.81533110e-01 -2.30911866e-01 6.02380216e-01 8.61302577e-03
-2.43439451e-02 7.04936013e-02 -5.33670425e-01 5.89217603e-01
6.53302014e-01 -1.95523798e-01 6.83427811e-01 9.09544647e-01
3.37133080e-01 -4.93496470e-02 -6.88066125e-01 1.22941399e+00
-1.61304712e-01 -1.02155733e+00 -8.90303329e-02 -6.50100969e-03
6.94109261e-01 -4.21348095e-01 -4.78532128e-02 -7.73513317e-02
4.95278776e-01 -1.21680272e+00 3.50342274e-01 7.65284300e-02
8.32785130e-01 -1.18893862e+00 7.78117716e-01 5.23148298e-01
-1.63188088e+00 -2.31135577e-01 -4.72913355e-01 1.30625680e-01
-1.50938600e-01 3.96132231e-01 -9.99507010e-01 4.42479193e-01
9.03412342e-01 1.47072360e-01 -5.22494733e-01 9.63075280e-01
4.74903017e-01 4.84832138e-01 -6.57849789e-01 -2.23776281e-01
4.52784359e-01 -7.89717853e-01 2.13875756e-01 1.07819021e+00
7.89846897e-01 3.07728469e-01 1.64421409e-01 5.70894718e-01
1.81925297e-01 5.02040625e-01 -6.08487606e-01 -4.46110398e-01
1.10440063e+00 1.08285511e+00 -1.60392332e+00 -1.30642392e-02
-1.30356997e-01 7.99685001e-01 1.98996991e-01 2.52475500e-01
-7.30218053e-01 -7.38729239e-01 1.07155852e-01 3.24946880e-01
3.54347706e-01 -3.16298723e-01 -3.92374486e-01 -6.93753421e-01
-4.83480431e-02 -7.25970089e-01 7.05456614e-01 -4.50875640e-01
-1.02709949e+00 7.36651003e-01 -6.82676956e-02 -1.46107924e+00
8.40528458e-02 -3.53860438e-01 -3.61483455e-01 6.30536079e-01
-6.12848043e-01 -8.41261744e-01 -2.71827936e-01 6.75740600e-01
4.00670707e-01 -2.55837739e-01 6.47929847e-01 1.92028359e-01
-7.78366268e-01 6.36156380e-01 4.73044127e-01 1.24167256e-01
4.55274552e-01 -1.23210156e+00 -2.25383833e-01 1.02218723e+00
-2.00857744e-02 9.70524609e-01 7.53316820e-01 -5.17158687e-01
-1.19659269e+00 -7.20725298e-01 5.28827727e-01 8.61335918e-02
5.42270124e-01 -3.27159613e-01 -6.25468552e-01 3.30459207e-01
1.15020104e-01 -2.15451658e-01 8.56603801e-01 -1.33715913e-01
-4.86373231e-02 -1.87748224e-01 -1.35485995e+00 5.53222358e-01
9.33864176e-01 -2.16052771e-01 -4.85798065e-03 2.87128873e-02
3.15429151e-01 -1.84325483e-02 -7.22415507e-01 2.55532265e-01
2.94087380e-01 -1.09836102e+00 9.83225882e-01 -1.91972509e-01
-2.85731698e-03 -8.16835523e-01 -1.65596992e-01 -1.03734887e+00
-3.77722919e-01 -5.70039749e-01 9.89351273e-02 1.13206673e+00
3.67677361e-01 -4.92015064e-01 1.17212045e+00 3.19731861e-01
1.97004989e-01 -8.07663023e-01 -9.74467099e-01 -5.69299042e-01
-8.75024721e-02 -2.96019353e-02 6.30423427e-01 1.40431595e+00
2.86654949e-01 -2.72182636e-02 -1.47209853e-01 3.95928442e-01
7.74470687e-01 2.45756865e-01 4.57190752e-01 -1.36107206e+00
-1.39502943e-01 -6.42677486e-01 -3.91486734e-01 -4.77211654e-01
-2.09986418e-01 -6.94705129e-01 -1.31624505e-01 -1.31508934e+00
1.26999557e-01 -9.29190099e-01 -4.13281709e-01 1.32563874e-01
-2.32395440e-01 1.36826053e-01 -2.53670085e-02 5.31212866e-01
-3.55062664e-01 2.06352994e-01 8.71969879e-01 8.97162501e-03
-6.52520180e-01 -1.13963626e-01 -7.96270847e-01 9.66846764e-01
1.02257872e+00 -2.41464168e-01 -9.20308113e-01 -1.03501705e-02
1.82136163e-01 8.66784900e-03 -2.63349444e-01 -1.09505892e+00
4.10328150e-01 -2.86919057e-01 2.25526318e-01 -6.89912856e-01
1.13470271e-01 -1.29141319e+00 6.61069095e-01 4.22454596e-01
-4.26558070e-02 4.50950354e-01 1.13549894e-02 7.33724058e-01
-2.31573761e-01 -2.62544960e-01 8.85713398e-01 -3.33248496e-01
-6.35795534e-01 1.51306868e-01 -2.47044116e-02 -1.95816234e-01
1.47936857e+00 -8.89093637e-01 1.20505556e-01 -2.32875898e-01
-6.55559480e-01 1.88489497e-01 6.25315726e-01 -5.10278866e-02
6.35214090e-01 -1.22894716e+00 -2.38047957e-01 1.37510195e-01
-1.48603335e-01 2.23043144e-01 -4.41483855e-02 7.71576881e-01
-1.01294172e+00 1.57801524e-01 -1.63927954e-03 -5.69726765e-01
-1.28945816e+00 9.15112853e-01 2.74023533e-01 -1.43348739e-01
-4.72735882e-01 9.63945389e-01 -2.72180289e-01 -2.96091795e-01
4.06254768e-01 7.18770698e-02 -2.30214417e-01 6.49280176e-02
7.88917616e-02 5.55001378e-01 -1.47262648e-01 -4.77443904e-01
-4.08397049e-01 8.12929928e-01 1.56167969e-01 -7.68056735e-02
1.20847595e+00 -3.76445264e-01 -2.65230030e-01 2.58285046e-01
9.90794361e-01 -4.93408889e-02 -9.99711156e-01 -3.74882333e-02
1.37000024e-01 -3.87188911e-01 -2.73518980e-01 -3.77233088e-01
-1.15422440e+00 5.88554084e-01 7.45898068e-01 5.20180941e-01
1.58022153e+00 -2.37449512e-01 6.08392358e-01 3.67604285e-01
5.28621197e-01 -1.61607814e+00 -3.45545113e-02 -2.79465616e-01
3.83128792e-01 -1.05816901e+00 6.02968633e-02 -6.35463297e-01
-4.97190565e-01 1.18383384e+00 6.54718220e-01 -1.42434940e-01
8.79695356e-01 2.65324473e-01 1.28618330e-01 -4.68538821e-01
-2.52087861e-01 1.71135858e-01 -5.39337881e-02 5.77576637e-01
3.07240963e-01 7.24676773e-02 -6.43211842e-01 3.62904966e-01
-5.12135088e-01 -2.19549865e-01 4.10594970e-01 8.65509510e-01
-6.67194963e-01 -8.97796333e-01 -6.43410146e-01 4.00004119e-01
-3.08327973e-01 -2.02549621e-03 -4.81566280e-01 9.58989620e-01
5.88770807e-01 1.06365669e+00 -3.32236029e-02 -5.44048727e-01
-8.75436515e-02 1.24468103e-01 7.38238916e-02 -1.58067539e-01
-3.89608592e-01 1.85066149e-01 -4.71597344e-01 -2.96485454e-01
-5.80757916e-01 -6.78955495e-01 -1.38738537e+00 -4.37633395e-01
-6.94802821e-01 8.49788129e-01 5.40334463e-01 5.65414369e-01
1.82926893e-01 1.13036707e-02 8.69665563e-01 -2.70815432e-01
-2.43458584e-01 -7.83980548e-01 -8.70850563e-01 4.82290357e-01
-2.43584901e-01 -7.20417202e-01 -6.36384606e-01 4.61941719e-01] | [7.541796684265137, 4.687739849090576] |
1acdae83-85c6-4832-9c0c-d8e590f890ed | object-detection-based-inspection-of-power | 2212.11017 | null | https://arxiv.org/abs/2212.11017v1 | https://arxiv.org/pdf/2212.11017v1.pdf | Object detection-based inspection of power line insulators: Incipient fault detection in the low data-regime | Deep learning-based object detection is a powerful approach for detecting faulty insulators in power lines. This involves training an object detection model from scratch, or fine tuning a model that is pre-trained on benchmark computer vision datasets. This approach works well with a large number of insulator images, but can result in unreliable models in the low data regime. The current literature mainly focuses on detecting the presence or absence of insulator caps, which is a relatively easy detection task, and does not consider detection of finer faults such as flashed and broken disks. In this article, we formulate three object detection tasks for insulator and asset inspection from aerial images, focusing on incipient faults in disks. We curate a large reference dataset of insulator images that can be used to learn robust features for detecting healthy and faulty insulators. We study the advantage of using this dataset in the low target data regime by pre-training on the reference dataset followed by fine-tuning on the target dataset. The results suggest that object detection models can be used to detect faults in insulators at a much incipient stage, and that transfer learning adds value depending on the type of object detection model. We identify key factors that dictate performance in the low data-regime and outline potential approaches to improve the state-of-the-art. | ['Giovanni Sansavini', 'Etienne Auger', 'Blazhe Gjorgiev', 'Mohammad Hossein Saadat', 'Laya Das'] | 2022-12-21 | null | null | null | null | ['fault-detection'] | ['miscellaneous'] | [ 2.55640268e-01 -3.42397869e-01 2.99693704e-01 -1.19824700e-01
-8.04526269e-01 -7.56388128e-01 2.45577008e-01 1.16599716e-01
1.94754049e-01 2.34973729e-01 -5.77383280e-01 -4.56230164e-01
-2.07459539e-01 -7.98399031e-01 -6.77163661e-01 -8.44740450e-01
-1.44395307e-01 3.81441742e-01 4.94427711e-01 -1.47674382e-01
2.60128498e-01 7.44353294e-01 -1.42247331e+00 5.08671761e-01
9.97780681e-01 1.02833533e+00 4.52457845e-01 8.20802748e-01
6.42078757e-01 3.76257300e-01 -1.36554754e+00 1.61227897e-01
2.47960985e-01 -1.53316155e-01 -6.14309609e-01 5.18372953e-01
6.85893536e-01 -5.38053036e-01 -1.89745754e-01 8.55189681e-01
6.04460716e-01 -4.13026392e-01 9.69945252e-01 -1.12327099e+00
-8.30336928e-01 1.08864464e-01 -6.80525124e-01 7.85383701e-01
1.14433751e-01 4.03473377e-01 8.45720470e-01 -6.24906063e-01
-2.95194127e-02 8.88541579e-01 1.00415134e+00 -7.02009052e-02
-1.03453994e+00 -3.30779731e-01 -1.29253671e-01 3.34957331e-01
-1.33993256e+00 -5.74293733e-02 7.86074936e-01 -7.62003720e-01
9.95114207e-01 3.53625923e-01 5.24699211e-01 6.70293272e-01
4.83737022e-01 8.62509608e-01 7.40486026e-01 -5.39994895e-01
1.63792953e-01 8.96742418e-02 4.06622648e-01 5.96876919e-01
5.37859738e-01 8.92369598e-02 2.38276385e-02 5.90610690e-03
7.35599756e-01 -2.22529009e-01 -3.72725308e-01 3.93142290e-02
-8.23633790e-01 8.32853138e-01 5.65579236e-01 5.24503887e-01
-2.23255396e-01 -2.71215916e-01 4.25601333e-01 5.22293627e-01
5.40850520e-01 9.22919214e-01 -6.06262445e-01 2.66327113e-01
-1.19573975e+00 -5.19446097e-03 5.76284707e-01 3.57714295e-01
4.30620581e-01 2.53423333e-01 -4.30697739e-01 8.27846825e-01
6.08730465e-02 4.84995574e-01 -7.86546245e-02 -4.25131649e-01
2.71052390e-01 5.53042650e-01 3.25121820e-01 -7.48333931e-01
-5.95669627e-01 -7.37977505e-01 -4.80441719e-01 6.90830350e-01
3.48347694e-01 -2.47703344e-01 -1.13430667e+00 8.11956286e-01
4.06794548e-02 2.67027449e-02 -1.99357897e-01 8.86441112e-01
6.08036637e-01 8.24120283e-01 -4.65952933e-01 2.28219032e-01
1.20513558e+00 -9.13845897e-01 -3.47272426e-01 -4.44258332e-01
6.05707407e-01 -5.65508306e-01 1.11939144e+00 8.76982808e-01
-6.54963613e-01 -8.71240079e-01 -1.50876796e+00 2.15234324e-01
-4.60167974e-01 8.20658684e-01 4.03438807e-01 6.14549220e-01
-1.04339504e+00 6.86512709e-01 -8.33630979e-01 -4.28812504e-01
6.71958089e-01 3.73461246e-01 -9.14989263e-02 -1.53454050e-01
-7.78697371e-01 1.01503396e+00 3.71862590e-01 2.88883239e-01
-1.36373341e+00 -6.84748650e-01 -6.21174991e-01 3.63211751e-01
2.17488632e-01 -7.47946501e-02 8.91124249e-01 -9.28995192e-01
-7.24155486e-01 7.75133789e-01 5.24916589e-01 -4.97321606e-01
4.72394437e-01 -1.86161220e-01 -5.39332628e-01 2.65799701e-01
1.20369427e-01 2.21213385e-01 1.24486244e+00 -1.28926384e+00
-8.04531455e-01 -1.65276602e-01 3.01550090e-01 -3.94809365e-01
-4.46334541e-01 6.22332469e-02 -1.44835010e-01 -5.03772497e-01
-2.91974217e-01 -6.21521950e-01 2.73256361e-01 1.23422503e-01
-4.99841273e-01 -3.34408283e-01 1.33882546e+00 -7.83175528e-01
7.94694006e-01 -2.20413637e+00 -4.23418164e-01 -5.20760752e-02
-3.73550057e-02 7.36856043e-01 -3.40309024e-01 3.29717249e-01
-9.03683603e-02 -2.41762727e-01 -2.09710047e-01 5.17535955e-03
-1.08122699e-01 -8.33156332e-02 -2.95905948e-01 7.80907393e-01
7.96938479e-01 7.15003669e-01 -7.59910107e-01 -9.89591181e-02
4.95027423e-01 3.45633149e-01 -9.66933891e-02 3.16452056e-01
8.96399170e-02 2.30183899e-01 -1.39063090e-01 1.04934216e+00
9.18507576e-01 -3.83848935e-01 -2.95568734e-01 -6.36014044e-01
-4.58826385e-02 -1.16807684e-01 -9.18815255e-01 9.95199740e-01
-1.75338134e-01 1.12764740e+00 1.14450671e-01 -1.35546041e+00
1.03166592e+00 1.67809054e-01 2.24049076e-01 -5.98588288e-01
1.09468568e-02 4.29044813e-02 2.33822107e-01 -8.14605355e-01
1.20492816e-01 5.63074499e-02 1.33545890e-01 4.32656407e-02
8.98492932e-02 -4.88549083e-01 3.69635582e-01 -1.57400891e-01
1.55033982e+00 -3.18799049e-01 -2.47541249e-01 -5.79024255e-01
2.78820097e-01 4.38485779e-02 2.93424129e-01 1.12262106e+00
-4.49808016e-02 7.39198625e-01 6.02440238e-01 -5.87456882e-01
-9.79867637e-01 -1.12070119e+00 -6.87032878e-01 8.44321370e-01
1.59417957e-01 1.12354666e-01 -6.94038689e-01 -8.49770546e-01
3.97528172e-01 5.61541259e-01 -7.32799411e-01 -3.88361067e-01
-1.68350667e-01 -1.28826702e+00 4.21982437e-01 8.55000257e-01
3.27108979e-01 -1.16220093e+00 -6.92852616e-01 1.23342037e-01
3.49612594e-01 -1.06261420e+00 6.27247915e-02 6.79015279e-01
-6.51856065e-01 -1.26295698e+00 -6.96727693e-01 -9.97709930e-01
1.00363278e+00 4.15958643e-01 1.31151807e+00 5.09388208e-01
-8.93783808e-01 2.40917251e-01 -3.52870643e-01 -7.82588243e-01
-3.70011657e-01 -1.40774444e-01 -6.95319427e-03 1.01499045e-02
1.73330218e-01 -1.24373205e-01 -5.01295745e-01 4.08351958e-01
-7.92226017e-01 -3.92407686e-01 6.17413759e-01 8.90636921e-01
5.31726107e-02 6.83284342e-01 7.64648914e-01 -6.17786169e-01
5.70455909e-01 -2.86143959e-01 -8.21216285e-01 2.44296908e-01
-3.50227863e-01 -4.35781896e-01 6.27338707e-01 -2.80702680e-01
-6.04815483e-01 -1.16183534e-01 -1.95486709e-01 -4.51419711e-01
-2.08967730e-01 5.23947299e-01 -1.69291366e-02 -3.52218270e-01
7.76849926e-01 -1.27422258e-01 -2.16351137e-01 -5.80217898e-01
-3.53862882e-01 8.75275731e-01 5.95474362e-01 -2.80652255e-01
8.72347057e-01 5.74137390e-01 -5.83161190e-02 -1.06804681e+00
-1.06111813e+00 -4.31660712e-01 -8.48403573e-01 -2.39027128e-01
4.55927283e-01 -8.62637699e-01 -3.56878638e-01 7.65076637e-01
-8.92625272e-01 -5.89364707e-01 -2.49219462e-01 4.75754812e-02
-2.17666581e-01 3.62683952e-01 -7.51915038e-01 -8.40712786e-01
-3.82827967e-01 -9.70304549e-01 1.47383785e+00 2.10072830e-01
1.62248299e-01 -8.66650283e-01 -1.45437494e-01 2.71181285e-01
3.26393545e-01 2.79304832e-01 1.12823606e+00 -5.77765822e-01
-3.12587976e-01 -7.13683546e-01 -4.28212851e-01 8.95042300e-01
3.80258679e-01 3.27953041e-01 -1.09544170e+00 -8.07257771e-01
-8.45162347e-02 -3.57830524e-01 1.11221266e+00 5.57883382e-01
1.17232978e+00 1.76725790e-01 -4.67201531e-01 3.07315558e-01
1.47534478e+00 2.74334937e-01 7.35524595e-01 3.43860894e-01
5.25299191e-01 2.98140496e-01 7.58555651e-01 3.26107204e-01
-5.28671592e-02 4.34700817e-01 6.08879805e-01 -6.65950716e-01
-8.42506662e-02 3.93527538e-01 1.89382300e-01 3.40687633e-01
2.11738974e-01 -4.80144680e-01 -1.06530595e+00 9.11642432e-01
-1.52163839e+00 -6.70313954e-01 -9.40386057e-02 2.02709174e+00
3.76552075e-01 6.53007865e-01 1.51652366e-01 6.67102933e-01
6.61725938e-01 -2.77483046e-01 -4.89562303e-01 -1.44763857e-01
-2.80199170e-01 -3.18180397e-02 2.45732278e-01 8.40313956e-02
-1.51064205e+00 4.86394256e-01 7.27839851e+00 6.98040009e-01
-1.37615120e+00 8.14742744e-02 6.45170152e-01 9.99162570e-02
3.77007544e-01 -2.73701489e-01 -6.93808675e-01 7.03791618e-01
7.01896489e-01 3.87923449e-01 2.23582168e-03 7.28240967e-01
2.35232756e-01 -3.07060659e-01 -1.23504221e+00 7.07807660e-01
3.34994048e-01 -1.21669447e+00 -3.13695431e-01 -1.18906341e-01
8.05798829e-01 1.93322703e-01 1.59968585e-02 3.36544603e-01
-2.41392747e-01 -9.34334934e-01 5.36945105e-01 4.13039446e-01
4.84962970e-01 -5.28867900e-01 1.04004729e+00 1.58132717e-01
-9.01496291e-01 -5.55301309e-01 -6.54422343e-01 -9.68207023e-04
-6.66779093e-03 9.32082891e-01 -9.00388181e-01 3.69561553e-01
1.10111380e+00 8.55666876e-01 -1.00609004e+00 1.52403975e+00
-9.82045084e-02 1.02516520e+00 -2.76412219e-01 2.14560792e-01
1.84103936e-01 -3.65505069e-02 3.11616540e-01 1.35102999e+00
3.77076358e-01 -5.38918018e-01 3.87014627e-01 9.23530340e-01
2.66148657e-01 -6.48352861e-01 -6.13689423e-01 9.73433852e-02
1.78791434e-01 1.54159737e+00 -9.69465613e-01 -2.67087251e-01
-4.62154150e-01 7.92910695e-01 8.90814066e-02 2.01626927e-01
-1.05099213e+00 -6.59347773e-01 1.87728539e-01 4.12027270e-01
8.37912798e-01 -9.30848122e-02 -2.19275847e-01 -7.43301988e-01
5.78391775e-02 -8.31041098e-01 3.96643072e-01 -1.12627685e+00
-1.40572548e+00 3.76097351e-01 -2.13197246e-01 -1.33940172e+00
2.52551615e-01 -1.13696492e+00 -1.24164236e+00 5.76012731e-01
-1.47203851e+00 -1.41548145e+00 -4.63121295e-01 -8.93022213e-03
7.54193962e-01 -8.64265545e-04 3.26934844e-01 3.49184901e-01
-7.84863234e-01 4.72868890e-01 4.30379808e-01 3.19930047e-01
4.64151323e-01 -1.41722906e+00 5.05667090e-01 9.28038120e-01
1.23635769e-01 -1.39517495e-02 8.06458533e-01 -5.81554413e-01
-1.39677155e+00 -1.28402269e+00 6.61701187e-02 -7.19470382e-01
5.29007196e-01 -3.94125968e-01 -1.29823387e+00 4.88185883e-01
5.22092767e-02 1.81339607e-01 2.69191504e-01 1.49553016e-01
5.23185506e-02 -3.14510435e-01 -1.26763248e+00 9.55893770e-02
3.82688016e-01 -5.12254059e-01 -4.63440865e-01 8.06652069e-01
1.84455156e-01 -1.83914244e-01 -9.39881325e-01 6.14338696e-01
2.02334523e-01 -7.50531971e-01 9.70499456e-01 -2.86945105e-01
8.94053355e-02 -5.62340736e-01 1.59341007e-01 -1.35944593e+00
-7.92504191e-01 -1.56162977e-01 -2.39397302e-01 1.25493991e+00
3.53635281e-01 -3.60968500e-01 5.47133148e-01 -2.15579346e-02
-5.85711658e-01 -7.17042804e-01 -7.74840593e-01 -1.09203577e+00
-1.93666872e-02 -8.78222659e-02 4.14947033e-01 6.99638247e-01
-2.67430246e-01 2.17787459e-01 -3.67111750e-02 7.72220254e-01
4.45389032e-01 2.49361217e-01 4.34399277e-01 -1.53653169e+00
-1.04911238e-01 -1.59904703e-01 -7.14628756e-01 -5.58739662e-01
-2.66676825e-02 -6.96410894e-01 3.55203658e-01 -1.70067608e+00
3.65108997e-02 -3.29935312e-01 -5.81255376e-01 6.76301539e-01
-3.21341962e-01 6.52039111e-01 -9.77473035e-02 6.70899544e-03
-6.21076524e-01 2.18858227e-01 1.02550852e+00 -7.94946134e-01
8.65249187e-02 -1.07968394e-02 -5.48054755e-01 7.15433061e-01
7.63109267e-01 -2.65539706e-01 -8.47362727e-02 -5.99605560e-01
-8.31793994e-03 -3.75831306e-01 8.42511296e-01 -1.61502469e+00
-1.08259358e-02 3.69464636e-01 1.02310157e+00 -7.95300424e-01
1.95430100e-01 -7.02300787e-01 -4.74390507e-01 6.09111845e-01
2.18469903e-01 7.88225979e-03 6.95426106e-01 4.17675465e-01
1.10013178e-02 -6.26024425e-01 9.19692934e-01 5.46228513e-02
-8.43271554e-01 -1.92810297e-01 -7.17484891e-01 -2.14056253e-01
9.68498886e-01 -4.80909467e-01 -4.73926783e-01 8.38987678e-02
-3.52879196e-01 1.74557820e-01 3.41666192e-01 7.06613004e-01
3.80141735e-01 -9.47398901e-01 -9.87768590e-01 5.97817659e-01
1.49938136e-01 -2.10544378e-01 1.32299483e-01 7.77377725e-01
-5.63817739e-01 4.05117720e-01 -3.80543739e-01 -1.12007546e+00
-1.13864577e+00 6.27789259e-01 5.31698108e-01 -5.39189242e-02
-6.72063470e-01 7.62374282e-01 2.53897697e-01 8.23641103e-03
5.02162613e-02 -5.48860967e-01 -3.25078309e-01 5.57160862e-02
4.63853538e-01 4.40635294e-01 6.92780495e-01 -1.62264735e-01
-3.60838801e-01 6.43156946e-01 -2.51386642e-01 6.84338093e-01
1.26849115e+00 2.83358037e-01 -1.12241106e-02 3.59175473e-01
8.88406456e-01 -5.07989943e-01 -1.55059361e+00 2.36750945e-01
-8.52133557e-02 -4.61526185e-01 3.70738983e-01 -1.05790687e+00
-1.25200212e+00 1.00111187e+00 1.03342617e+00 7.49994814e-01
1.28513432e+00 1.17772795e-01 3.06014031e-01 4.14208025e-01
2.93425173e-01 -1.18281817e+00 2.52493769e-01 2.94991583e-01
1.01300716e+00 -1.37108660e+00 1.85959250e-01 -2.74854898e-01
-2.17980891e-01 1.14940393e+00 8.25506389e-01 -4.19979215e-01
4.67746407e-01 6.77565575e-01 -6.58763275e-02 -5.44904292e-01
-4.90039200e-01 -9.42808390e-02 4.04010236e-01 8.84196401e-01
1.66265145e-01 -1.79044134e-03 2.79133648e-01 4.24058169e-01
3.58798355e-01 -1.13438621e-01 5.37876129e-01 1.17032433e+00
-7.64062822e-01 -5.60792387e-01 -7.21751392e-01 1.18694472e+00
-4.24171120e-01 1.03601120e-01 -3.78089219e-01 6.21456742e-01
3.81263912e-01 1.20249426e+00 3.49189162e-01 -2.23669007e-01
5.47325611e-01 -4.90032196e-01 5.73235452e-01 -5.55429399e-01
-6.54248774e-01 -1.65214583e-01 -4.53118198e-02 -3.10017448e-02
1.97300538e-01 -5.64935744e-01 -7.49438286e-01 3.59970093e-01
-1.04436958e+00 1.22118562e-01 4.06500608e-01 1.01193213e+00
3.06318879e-01 9.85824466e-01 6.27010107e-01 -1.21437466e+00
-6.65190279e-01 -1.10365713e+00 -9.85511303e-01 2.70930558e-01
6.88302577e-01 -7.53970265e-01 -5.59352875e-01 9.02528167e-02] | [7.420025825500488, 1.9078953266143799] |
8a7a165c-ed14-4e0d-9461-079478cb6b89 | can-rumour-stance-alone-predict-veracity | null | null | https://aclanthology.org/C18-1284 | https://aclanthology.org/C18-1284.pdf | Can Rumour Stance Alone Predict Veracity? | Prior manual studies of rumours suggested that crowd stance can give insights into the actual rumour veracity. Even though numerous studies of automatic veracity classification of social media rumours have been carried out, none explored the effectiveness of leveraging crowd stance to determine veracity. We use stance as an additional feature to those commonly used in earlier studies. We also model the veracity of a rumour using variants of Hidden Markov Models (HMM) and the collective stance information. This paper demonstrates that HMMs that use stance and tweets{'} times as the only features for modelling true and false rumours achieve F1 scores in the range of 80{\%}, outperforming those approaches where stance is used jointly with content and user based features. | ['Sebastian Dungs', 'Kalina Bontcheva', 'Ahmet Aker', 'Norbert Fuhr'] | 2018-08-01 | can-rumour-stance-alone-predict-veracity-1 | https://aclanthology.org/C18-1284 | https://aclanthology.org/C18-1284.pdf | coling-2018-8 | ['rumour-detection'] | ['natural-language-processing'] | [-4.77171749e-01 3.59026939e-01 -4.35600877e-01 -2.33220756e-01
-2.34630153e-01 -1.17989138e-01 1.23860705e+00 4.66751009e-01
-3.61140490e-01 8.40692163e-01 6.27814770e-01 -4.94486719e-01
6.03188038e-01 -6.54812038e-01 -1.19240144e-02 -3.57350111e-01
-2.01532498e-01 4.61537510e-01 2.74916112e-01 -4.91797894e-01
8.27118039e-01 1.99921094e-02 -1.47152233e+00 3.68932933e-01
5.77596188e-01 7.05792606e-01 -6.04123592e-01 9.05980289e-01
-2.94556826e-01 2.00152993e+00 -1.39186954e+00 -7.14452744e-01
-1.01985656e-01 -6.91730022e-01 -8.20686162e-01 3.03480536e-01
-1.10400036e-01 -5.08116245e-01 -4.72567707e-01 6.83116734e-01
3.09263289e-01 -3.85904834e-02 6.41158104e-01 -1.39525390e+00
-7.91309953e-01 8.19924414e-01 -5.49565256e-01 8.42749774e-01
8.20717037e-01 2.10938603e-01 8.56755495e-01 -5.35799742e-01
5.42497933e-01 9.66680706e-01 1.02533400e+00 1.33443400e-01
-9.91463721e-01 -5.97878993e-01 -3.16723138e-01 2.37382635e-01
-1.12045538e+00 -6.62137330e-01 5.39175928e-01 -8.03321958e-01
1.05626082e+00 4.87682641e-01 7.33572364e-01 1.24672103e+00
3.55520397e-01 5.95384121e-01 1.52399600e+00 -2.40836188e-01
2.16917470e-01 7.03546882e-01 2.73426473e-01 4.69970703e-01
3.47139508e-01 -3.09008747e-01 -6.30071938e-01 -1.02547109e+00
3.77345800e-01 -1.41918212e-01 -1.03672236e-01 7.33768702e-01
-9.49827969e-01 1.44831121e+00 -1.26038371e-02 3.23687941e-01
-7.31243610e-01 -2.49243379e-01 6.49245918e-01 6.22959375e-01
1.06105340e+00 5.91794491e-01 1.62239313e-01 -5.44069588e-01
-1.41739428e+00 3.69241506e-01 1.65849555e+00 6.17630720e-01
5.07173717e-01 2.78005600e-01 -4.71150875e-02 7.50007451e-01
1.98432177e-01 2.35872462e-01 6.80486798e-01 -9.69216168e-01
2.10641503e-01 3.73928130e-01 5.50933957e-01 -1.42585397e+00
-4.42573756e-01 -2.68800735e-01 -4.23768133e-01 -2.22304970e-01
7.40193725e-02 -3.47885847e-01 -5.43637931e-01 8.54316235e-01
1.39576405e-01 7.63582811e-02 -6.91107288e-02 6.40991032e-01
1.00629008e+00 4.90757883e-01 1.01034284e-01 -8.27747464e-01
1.15724659e+00 -5.94220340e-01 -9.51890647e-01 -1.62998274e-01
6.91533327e-01 -1.05531335e+00 4.54067796e-01 3.43162924e-01
-1.10842359e+00 1.86844394e-01 -8.49254787e-01 3.26016665e-01
-6.86402544e-02 -8.90745580e-01 6.12457573e-01 1.06188560e+00
-9.89957392e-01 5.71365476e-01 -4.43124682e-01 -2.32760280e-01
3.06040019e-01 -1.72182262e-01 1.92284122e-01 4.08264279e-01
-1.44219625e+00 1.29393888e+00 -6.77128695e-03 -3.71242523e-01
-6.79332674e-01 -7.43985698e-02 -4.94946033e-01 -4.33844119e-01
1.12351775e-02 -4.12878454e-01 1.47937548e+00 -9.33403611e-01
-1.47523201e+00 7.85486817e-01 -1.86462030e-01 -8.36031139e-01
8.98355901e-01 1.76046655e-01 -6.72491372e-01 1.42783865e-01
3.25058788e-01 -2.46153325e-01 8.67457926e-01 -1.00159228e+00
-3.92251819e-01 1.58200208e-02 -1.49360299e-01 -9.88552421e-02
-8.32491890e-02 6.07563734e-01 4.21644866e-01 -2.80437380e-01
-8.16031471e-02 -7.64275014e-01 -1.31679270e-02 -9.34244633e-01
-5.81793308e-01 -2.50120819e-01 5.73673427e-01 -9.51824188e-01
1.65120077e+00 -1.43679142e+00 -6.41066730e-01 9.60335508e-02
6.60871625e-01 1.46089911e-01 5.47408938e-01 9.77010429e-01
4.19800222e-01 7.53572583e-01 4.25341547e-01 -4.83081728e-01
5.00699179e-03 1.04226731e-01 -2.82834142e-01 1.02926028e+00
-2.63597250e-01 5.53748786e-01 -7.38188922e-01 -6.39660001e-01
-2.22082257e-01 1.51180178e-01 -3.02786767e-01 9.22678858e-02
-4.04522307e-02 1.22473203e-01 -2.23558128e-01 3.58970940e-01
3.32586139e-01 -5.27864456e-01 1.22174390e-01 7.26854384e-01
-4.34108645e-01 7.74394214e-01 -4.43891436e-01 1.22080311e-01
-1.07599273e-01 9.62084293e-01 -1.67163968e-01 -3.44260603e-01
1.13360310e+00 7.65547216e-01 3.10171902e-01 -5.33356011e-01
3.55847716e-01 2.00501323e-01 -4.35730591e-02 -6.62304223e-01
8.67972851e-01 -7.12576628e-01 -1.29828095e-01 9.99585330e-01
-4.27829236e-01 -7.42046468e-05 -2.02283859e-01 3.46106589e-01
1.13430262e+00 -6.38691068e-01 7.37825453e-01 5.70424311e-02
9.82541069e-02 3.22708279e-01 3.40294778e-01 1.07566071e+00
-5.64986467e-01 3.83052886e-01 9.21047390e-01 -2.33117059e-01
-1.33003616e+00 -3.24622661e-01 -8.04391131e-02 1.18132818e+00
-1.87120631e-01 -3.05763096e-01 -4.79273707e-01 -2.94887632e-01
4.36227806e-02 1.13056695e+00 -9.03602302e-01 4.27370787e-01
-1.90753579e-01 -1.07227492e+00 1.02389932e+00 -8.84795859e-02
3.71586591e-01 -9.09590781e-01 -4.37772602e-01 3.63022864e-01
-6.25235319e-01 -1.09187555e+00 -3.34929787e-02 -2.58890361e-01
-6.30814075e-01 -9.46572125e-01 -4.47373629e-01 -8.31824318e-02
-1.00521490e-01 3.66176754e-01 1.02048385e+00 7.06179678e-01
4.06216681e-01 1.06437178e-02 -5.58340549e-01 -5.55284142e-01
-1.01064658e+00 -2.02804446e-01 7.77391642e-02 -2.28603885e-01
5.78694165e-01 -6.33218288e-01 -2.53905594e-01 3.59358221e-01
-5.98814070e-01 -3.76832373e-02 1.98168635e-01 6.44930601e-01
-5.33786535e-01 -2.31709227e-01 1.09116340e+00 -1.21441531e+00
1.13341987e+00 -1.47061932e+00 2.76446611e-01 -5.83989024e-01
-7.86903799e-01 -6.72173142e-01 2.15736508e-01 -2.81095088e-01
-7.55915821e-01 -9.92446482e-01 -1.48765817e-01 -1.21439602e-02
-1.66150004e-01 7.09660590e-01 8.36257756e-01 1.53939366e-01
9.60704088e-01 2.64737248e-01 4.87648666e-01 -2.02171639e-01
-1.31335825e-01 1.23476195e+00 1.77975018e-02 2.46012285e-01
2.92376757e-01 5.61667264e-01 -5.60912371e-01 -1.23716617e+00
-1.02028561e+00 -9.60767031e-01 -2.30252221e-01 -3.21688890e-01
4.45652515e-01 -8.49408984e-01 -1.02677763e+00 3.97941053e-01
-1.16306198e+00 -1.31685689e-01 2.58338541e-01 3.16920131e-01
-5.88559985e-01 6.77264154e-01 -1.30282986e+00 -1.36300242e+00
-3.02071363e-01 -4.84110981e-01 4.44489606e-02 -1.39636561e-01
-8.45603824e-01 -1.32745516e+00 2.95918703e-01 8.18015277e-01
7.37991631e-01 4.73680854e-01 3.60195607e-01 -1.46311522e+00
1.57925323e-01 -5.81630528e-01 -1.54837027e-01 1.91716701e-02
-1.99977830e-01 2.79288050e-02 -8.50934565e-01 -2.13147216e-02
4.30174589e-01 -2.57945061e-01 6.43782556e-01 -1.91399418e-02
6.68031350e-02 -1.39511526e+00 1.69167947e-03 -3.07440311e-01
1.05169392e+00 -3.58169705e-01 8.32613707e-01 9.43272114e-01
2.65020669e-01 5.37627757e-01 1.76928133e-01 1.29549408e+00
8.26217532e-01 3.46489817e-01 5.22963107e-01 3.58722121e-01
4.06295419e-01 -1.05815046e-01 5.88383794e-01 9.49806809e-01
-2.81946778e-01 -3.77929449e-01 -1.08260190e+00 6.60353839e-01
-1.54048705e+00 -1.71985245e+00 -8.81406963e-01 1.79516244e+00
8.40162456e-01 4.10806358e-01 7.24587739e-01 2.83814520e-01
8.01496089e-01 6.20323420e-01 2.37777814e-01 -1.00682735e+00
-1.14512391e-01 -5.19600093e-01 6.58917904e-01 7.31990635e-01
-7.03744173e-01 5.27328789e-01 7.66021729e+00 4.90080193e-03
-9.97200787e-01 4.12993520e-01 3.49596888e-01 -2.43099019e-01
-4.08201307e-01 1.05708659e-01 -5.68340719e-01 6.80203080e-01
1.50792348e+00 -2.47175023e-01 1.50693148e-01 6.72681272e-01
7.95023441e-01 -5.31005383e-01 -5.29722393e-01 3.91004801e-01
5.22093534e-01 -1.30075705e+00 -3.36738944e-01 4.48303789e-01
9.53033864e-01 3.91898751e-01 -7.95535296e-02 4.63425577e-01
5.95799387e-01 -1.07278931e+00 9.75099444e-01 6.39327109e-01
6.88667521e-02 -6.14795268e-01 1.14172971e+00 1.05679536e+00
5.75692393e-02 -7.00338408e-02 -2.04248652e-01 -8.64154220e-01
4.13915068e-01 6.92743123e-01 -1.92577326e+00 -8.26061368e-02
2.55622566e-01 4.06053811e-01 -4.82623816e-01 9.52789485e-01
5.58306836e-03 1.33140242e+00 -1.36244744e-01 -4.42189783e-01
5.31449080e-01 4.23689187e-01 8.57722223e-01 1.58654249e+00
-9.45401564e-02 1.18294902e-01 2.28939906e-01 5.17336607e-01
-4.14652424e-03 1.96916327e-01 -3.69266033e-01 -3.32903802e-01
6.14051044e-01 7.31444001e-01 -4.14093316e-01 -6.12031400e-01
-4.11688358e-01 7.05804706e-01 9.90514085e-02 8.10548663e-02
-7.93909550e-01 1.23714544e-01 3.75713468e-01 7.35064864e-01
1.78670675e-01 -2.10252538e-01 -5.68576992e-01 -9.74010885e-01
-4.58459646e-01 -1.00376475e+00 2.32010841e-01 -4.72219110e-01
-1.45956135e+00 5.31339407e-01 -2.50622272e-01 -8.80910873e-01
-4.79909658e-01 2.39707097e-01 -1.03212416e+00 9.74639297e-01
-1.41702497e+00 -8.05316567e-01 -2.83397108e-01 1.56786904e-01
2.87130058e-01 -4.65316046e-03 6.52520716e-01 -1.52070120e-01
-5.05392790e-01 2.18659282e-01 -2.70305462e-02 2.35567093e-01
5.62564075e-01 -9.77099180e-01 4.54866797e-01 3.95618796e-01
-3.67182225e-01 6.29015863e-01 1.62676859e+00 -1.08064139e+00
-7.52624273e-01 -7.01206744e-01 1.44044781e+00 -7.58859634e-01
1.18923974e+00 2.30697960e-01 -1.14460456e+00 8.47091854e-01
4.24834013e-01 -3.86004090e-01 1.07528567e+00 3.60855609e-01
-3.74445200e-01 1.05656278e+00 -1.09558070e+00 2.30098218e-01
2.77153760e-01 -5.58220327e-01 -8.33350956e-01 4.05969083e-01
2.30339125e-01 -5.46553433e-02 -1.11498034e+00 -3.16566169e-01
4.70417410e-01 -1.53330529e+00 2.95541167e-01 -7.34323263e-01
6.36611521e-01 2.21243411e-01 -2.57652756e-02 -1.09721792e+00
-4.23647135e-01 -7.69283295e-01 -2.87253022e-01 9.93138134e-01
3.84033263e-01 -8.29502106e-01 6.89753592e-01 4.22345817e-01
1.79726228e-01 -5.11318624e-01 -7.13258684e-01 -5.80417633e-01
1.13531537e-01 -3.66034597e-01 4.08060640e-01 1.44228721e+00
7.13922024e-01 2.32140765e-01 -8.88907969e-01 2.85189250e-03
5.19780517e-01 -1.88646853e-01 8.36825132e-01 -1.34129941e+00
-2.35345557e-01 -6.48687065e-01 -5.89372158e-01 -5.95689416e-01
-5.39238844e-03 -4.38100427e-01 -2.52708718e-02 -1.55108631e+00
5.21384716e-01 -4.14798528e-01 3.18600953e-01 1.14171267e-01
-1.00190528e-01 4.96903479e-01 8.92421156e-02 9.95242655e-01
-7.20624804e-01 1.86499596e-01 8.60884011e-01 3.14662665e-01
-2.19234064e-01 2.52121687e-01 -7.35119879e-01 9.35188890e-01
9.68072414e-01 -7.02780426e-01 2.46266693e-01 2.74325669e-01
6.23351753e-01 6.29627645e-01 5.38453996e-01 -5.28248250e-01
3.79633099e-01 -4.73896414e-01 7.59410784e-02 -4.30076003e-01
4.23220068e-01 1.35465518e-01 2.59726141e-02 1.79051578e-01
-4.24227774e-01 1.66086182e-01 -3.08549970e-01 8.10433328e-01
-1.67254820e-01 -4.00607318e-01 5.94458640e-01 -5.64525008e-01
-3.09713662e-01 -1.45453930e-01 -1.11791217e+00 2.09323853e-01
8.74051034e-01 -3.82475287e-01 -6.40284717e-01 -1.31901956e+00
-8.06592941e-01 -1.33150131e-01 8.72387707e-01 9.51388180e-02
3.71340990e-01 -7.15416372e-01 -1.20835698e+00 -4.47346926e-01
-3.02705541e-02 -1.11705732e+00 5.43620102e-02 1.58898926e+00
-4.52761799e-01 3.22225511e-01 -7.68229738e-02 -1.99512511e-01
-1.14492702e+00 3.83181214e-01 1.78815126e-01 -3.96661796e-02
-6.02232456e-01 5.61677277e-01 -7.57731080e-01 4.96844985e-02
-2.29021847e-01 6.33876622e-01 -5.27521968e-01 4.59963053e-01
8.48509967e-01 8.61621261e-01 6.57319427e-02 -1.30237055e+00
-7.32131228e-02 -8.35977614e-01 -6.17178231e-02 -3.17234486e-01
1.19525456e+00 -6.40565753e-01 -2.72489995e-01 1.12001455e+00
1.06774998e+00 3.85675430e-01 -5.64488173e-01 -2.28501275e-01
3.06364000e-01 -5.89440823e-01 1.29516587e-01 -5.39727151e-01
-2.95964479e-01 4.60880637e-01 -4.48046029e-01 1.22243845e+00
4.06767474e-03 -1.09223545e-01 7.42945850e-01 -9.27881524e-02
3.92209440e-01 -1.04401970e+00 8.02514404e-02 7.66226709e-01
8.49700749e-01 -1.24244785e+00 4.14669335e-01 -2.24500656e-01
-1.27030945e+00 8.60732734e-01 4.98916805e-02 -1.35897279e-01
5.52575171e-01 2.86462083e-02 2.58043677e-01 -4.29274917e-01
-1.16277409e+00 6.92128092e-02 -2.83426195e-01 3.12634319e-01
7.75321782e-01 3.91086906e-01 -6.44895077e-01 5.59963048e-01
-5.93614221e-01 4.32501361e-03 1.56999052e+00 9.44392979e-01
-1.16047108e+00 -3.52749676e-01 -6.17294431e-01 8.91551197e-01
-1.18399286e+00 -1.03199549e-01 -5.11734486e-01 7.06518054e-01
-2.25007832e-01 1.50810421e+00 -3.56595702e-02 -6.72405720e-01
-2.02959791e-01 2.45770097e-01 -2.02648357e-01 -8.19242954e-01
-9.79696572e-01 -3.21364969e-01 9.66808319e-01 -8.93136188e-02
-3.18214178e-01 -9.75957751e-01 -8.87965381e-01 -1.38703346e+00
-6.13647759e-01 5.35120904e-01 5.11883497e-01 1.18127036e+00
-6.54046312e-02 -1.65205345e-01 8.84869218e-01 -6.42246842e-01
-9.15186405e-01 -1.39741778e+00 -6.73414230e-01 3.78945857e-01
5.64322591e-01 -4.83708829e-01 -8.68040681e-01 -1.50486872e-01] | [8.263701438903809, 10.117100715637207] |
71789437-961d-4d05-8ce4-75c828f1cee7 | vehicle-interfacing-neural-network-verifiers | 2202.05207 | null | https://arxiv.org/abs/2202.05207v1 | https://arxiv.org/pdf/2202.05207v1.pdf | Vehicle: Interfacing Neural Network Verifiers with Interactive Theorem Provers | Verification of neural networks is currently a hot topic in automated theorem proving. Progress has been rapid and there are now a wide range of tools available that can verify properties of networks with hundreds of thousands of nodes. In theory this opens the door to the verification of larger control systems that make use of neural network components. However, although work has managed to incorporate the results of these verifiers to prove larger properties of individual systems, there is currently no general methodology for bridging the gap between verifiers and interactive theorem provers (ITPs). In this paper we present Vehicle, our solution to this problem. Vehicle is equipped with an expressive domain specific language for stating neural network specifications which can be compiled to both verifiers and ITPs. It overcomes previous issues with maintainability and scalability in similar ITP formalisations by using a standard ONNX file as the single canonical representation of the network. We demonstrate its utility by using it to connect the neural network verifier Marabou to Agda and then formally verifying that a car steered by a neural network never leaves the road, even in the face of an unpredictable cross wind and imperfect sensors. The network has over 20,000 nodes, and therefore this proof represents an improvement of 3 orders of magnitude over prior proofs about neural network enhanced systems in ITPs. | ['Ekaterina Komendantskya', 'Luca Arnaboldi', 'Robert Atkey', 'Wen Kokke', 'Matthew L. Daggitt'] | 2022-02-10 | null | null | null | null | ['automated-theorem-proving', 'automated-theorem-proving'] | ['miscellaneous', 'reasoning'] | [ 3.14689666e-01 7.56681085e-01 -3.95869836e-02 -1.67333037e-01
5.84406639e-03 -9.37615573e-01 5.67729354e-01 -3.87518466e-01
-4.09427620e-02 9.60586786e-01 -8.42010498e-01 -1.34731233e+00
-3.02696675e-01 -9.72782195e-01 -1.15216815e+00 -3.22810471e-01
-6.91925049e-01 2.10693240e-01 6.78220034e-01 -3.59040260e-01
-3.83539528e-01 5.18878341e-01 -1.90051556e+00 -5.31154461e-02
2.84562588e-01 8.29212010e-01 -2.23802835e-01 1.29915810e+00
4.84373868e-01 1.05723369e+00 -4.09205943e-01 9.56665352e-02
3.83953899e-01 -7.17802867e-02 -7.97445595e-01 -6.45092249e-01
4.85115856e-01 -5.33649623e-01 -3.53921175e-01 1.07998681e+00
4.45135646e-02 -4.32435721e-01 -3.06571946e-02 -2.49735451e+00
6.53676689e-02 9.65165794e-01 1.07063100e-01 -2.85876781e-01
6.95617422e-02 2.47749105e-01 1.00860751e+00 1.50515765e-01
5.54318428e-01 1.28098381e+00 9.54885244e-01 6.64700747e-01
-1.28530455e+00 -9.69080687e-01 -1.55566916e-01 -5.68273552e-02
-1.51584601e+00 -6.01519704e-01 2.79905379e-01 -4.40385818e-01
1.46133065e+00 4.63514000e-01 5.12080431e-01 7.22063601e-01
3.88069004e-01 3.31230491e-01 7.02417433e-01 -3.41930926e-01
2.58237630e-01 5.16044736e-01 2.26021841e-01 9.64484513e-01
7.45466709e-01 7.36505508e-01 5.53963818e-02 -3.37551802e-01
7.97816396e-01 -5.84585965e-01 -5.67690097e-02 -3.62368137e-01
-1.16918564e+00 5.39782703e-01 4.87326197e-02 5.36053926e-02
1.72343865e-01 9.24357116e-01 8.89697373e-01 5.96504450e-01
-3.63923490e-01 3.17001641e-01 -8.54458928e-01 -8.58224183e-02
-5.96811414e-01 7.02011466e-01 1.41395652e+00 1.06897378e+00
5.79201341e-01 4.40314263e-01 6.79952800e-01 -2.82507122e-01
5.83894730e-01 7.55026042e-01 -1.67246342e-01 -1.41314876e+00
2.16710150e-01 4.32391286e-01 5.96034266e-02 -1.06747615e+00
-3.00831586e-01 -6.65225908e-02 -6.37887776e-01 9.76007581e-01
2.46259853e-01 -7.51183331e-01 -4.77106124e-01 1.76246929e+00
3.78723711e-01 2.06564322e-01 3.79708201e-01 4.33471411e-01
2.74349838e-01 6.86228693e-01 -5.23890734e-01 1.04524754e-01
9.25667882e-01 -4.44591194e-01 -4.69844043e-01 1.76245347e-02
8.86574447e-01 7.58022517e-02 2.10278496e-01 6.54593825e-01
-1.03100574e+00 -2.34156594e-01 -1.53668725e+00 4.57758456e-01
-6.56875253e-01 -2.94915617e-01 8.66026819e-01 9.00839746e-01
-1.53740478e+00 6.26247466e-01 -1.16167676e+00 -1.76738888e-01
1.11128382e-01 1.05006909e+00 -5.50841033e-01 3.02184790e-01
-1.46777642e+00 1.18796813e+00 7.48100340e-01 3.94620061e-01
-1.11588180e+00 -6.41436577e-01 -1.16183269e+00 -1.75089747e-01
3.88166428e-01 -4.80727822e-01 1.61694682e+00 -8.58627915e-01
-1.43451476e+00 3.90424989e-02 2.73501694e-01 -9.55222070e-01
4.14010346e-01 6.12454891e-01 -6.86036527e-01 -8.22844878e-02
-2.20432639e-01 6.67465806e-01 3.71507883e-01 -8.79819155e-01
-8.33901107e-01 1.12284251e-01 7.30907977e-01 -7.35403955e-01
2.72859663e-01 1.11303829e-01 3.70820731e-01 4.41568404e-01
-4.80636150e-01 -9.57188845e-01 -6.03868850e-02 3.39270145e-01
-5.58163643e-01 -1.79741815e-01 1.39587700e+00 -6.78069443e-02
7.19749689e-01 -1.85848498e+00 -3.94174755e-01 7.26802647e-01
1.91735640e-01 5.53729594e-01 -3.84606160e-02 5.42314649e-01
-9.56922919e-02 4.00190204e-01 1.45266458e-01 4.51035976e-01
6.52097702e-01 6.37401879e-01 -4.89029855e-01 6.69839025e-01
2.66797602e-01 7.35579073e-01 -6.89152598e-01 -4.57128912e-01
2.85612226e-01 3.56451392e-01 -3.33267003e-01 -3.52459252e-01
-4.89920706e-01 -3.38689268e-01 -3.97251844e-01 4.28156614e-01
5.20432413e-01 -2.05341130e-02 4.27002400e-01 2.82478720e-01
-3.96849185e-01 3.75181973e-01 -1.63104701e+00 8.87253404e-01
-2.41824538e-01 1.10762799e+00 8.16129982e-01 -8.21200252e-01
6.56624675e-01 7.88261831e-01 -2.25071967e-01 -2.86493868e-01
5.87673426e-01 2.33622536e-01 4.18276399e-01 -3.38866681e-01
2.26151198e-01 -1.47535935e-01 -7.83614665e-02 5.49303889e-01
-1.34704754e-01 -4.98859584e-01 3.84547889e-01 1.44238174e-01
1.42699194e+00 2.79568106e-01 1.42709807e-01 -2.30601281e-01
5.23734808e-01 1.95533246e-01 4.61704999e-01 7.88894892e-01
-1.68851599e-01 -3.21820289e-01 8.77105117e-01 -3.59555572e-01
-1.25337398e+00 -7.16983795e-01 -3.14336382e-02 2.43920311e-01
-1.07186548e-01 -4.70542669e-01 -8.02089095e-01 -3.54848772e-01
2.78060317e-01 7.82341957e-01 -4.57009941e-01 -8.01286399e-02
-2.03984752e-01 1.84628129e-01 1.58534157e+00 8.28883410e-01
4.71439242e-01 -6.96631312e-01 -7.58897483e-01 1.05590066e-02
5.47708929e-01 -1.04884160e+00 3.91606450e-01 3.90697807e-01
-6.23883843e-01 -1.34813309e+00 5.21103740e-01 -7.19383478e-01
5.94886482e-01 -1.11369237e-01 6.27606750e-01 4.79930252e-01
-1.51487231e-01 1.27561793e-01 2.19801798e-01 -8.55978072e-01
-1.03850174e+00 -2.32057258e-01 3.84490162e-01 -8.73200059e-01
-1.96774110e-01 -6.17657423e-01 4.23862278e-01 3.49433899e-01
-9.05092180e-01 1.71592936e-01 1.93008736e-01 5.65614760e-01
-4.91969474e-02 8.19190621e-01 4.46096092e-01 -7.35630155e-01
5.75400114e-01 -1.46971971e-01 -1.62565243e+00 1.25775307e-01
-7.74208188e-01 1.24224931e-01 8.06673527e-01 -2.71297544e-01
-3.20536226e-01 3.20696414e-01 1.52193069e-01 -1.86530411e-01
-2.59221256e-01 5.16581178e-01 -1.70290664e-01 -5.05547345e-01
6.05649769e-01 -2.94648349e-01 6.25755847e-01 4.52028483e-01
2.49197096e-01 4.85703886e-01 6.31531417e-01 -6.60497963e-01
1.40257895e+00 2.02458277e-01 5.94687343e-01 -5.77900589e-01
5.13297133e-02 2.71595001e-01 -2.27126136e-01 -2.76723474e-01
2.28470325e-01 -6.35930419e-01 -1.77227378e+00 3.82397443e-01
-1.33756888e+00 -7.90522397e-01 -2.73474813e-01 2.37086236e-01
-4.10124809e-01 -3.22490521e-02 -4.37809318e-01 -1.17437518e+00
-3.71803753e-02 -1.11979926e+00 6.36592984e-01 -6.34213015e-02
-5.99067628e-01 -9.70053852e-01 1.41166002e-01 -4.07878429e-01
4.47896302e-01 5.22249043e-01 7.64989495e-01 -6.27555966e-01
-6.55454934e-01 -7.30844319e-01 -3.23672771e-01 7.50572979e-01
-2.08447948e-01 7.23802924e-01 -1.03116095e+00 -2.76952773e-01
-4.45056736e-01 -2.72646666e-01 -1.80257693e-01 1.34934830e-02
5.09806752e-01 -6.38203502e-01 -4.76775378e-01 -2.00483724e-02
1.44484413e+00 9.31887403e-02 6.51042819e-01 2.94576794e-01
2.31966496e-01 4.83853400e-01 3.50456595e-01 -1.53770596e-01
3.00000787e-01 2.32233956e-01 8.07286680e-01 5.16742244e-02
4.08758581e-01 -7.85869211e-02 6.25043929e-01 -4.83352749e-04
-6.83927312e-02 4.24739942e-02 -1.05263960e+00 4.82043654e-01
-1.64506662e+00 -1.20617378e+00 -3.99087131e-01 1.90460312e+00
6.88851833e-01 4.67065156e-01 -1.27775492e-02 7.80537426e-01
5.29347301e-01 -4.21764314e-01 -3.38162780e-01 -8.02128673e-01
3.77444446e-01 9.70085114e-02 9.39049661e-01 8.75757456e-01
-7.73609102e-01 5.03365457e-01 7.04925346e+00 2.92210162e-01
-1.02386284e+00 -3.04657638e-01 -2.79239237e-01 3.25594544e-01
2.76057981e-02 3.82137835e-01 -8.70589554e-01 -6.98374882e-02
1.56555080e+00 -4.62025255e-01 6.13104522e-01 1.01189327e+00
4.43735957e-01 -2.69569665e-01 -1.45728493e+00 8.94322246e-02
-2.55386919e-01 -1.36080182e+00 -6.45968676e-01 9.29067358e-02
4.29854989e-01 1.96256131e-01 -3.34583610e-01 4.12381202e-01
9.19133246e-01 -1.04839885e+00 8.84136140e-01 1.45259693e-01
6.61690176e-01 -9.33126628e-01 1.03014636e+00 6.63963974e-01
-9.60194051e-01 -5.48217408e-02 9.09665525e-02 -7.00243115e-01
-1.15900569e-01 1.67956427e-01 -1.42776060e+00 4.63768512e-01
4.88137662e-01 2.98521847e-01 -1.30002633e-01 8.53758574e-01
-2.89150864e-01 5.44043064e-01 -7.72746861e-01 -3.19830358e-01
1.82426155e-01 2.71492809e-01 4.34149057e-01 9.12666917e-01
6.29317090e-02 -2.24241331e-01 -2.63495237e-01 9.28496301e-01
4.69430774e-01 -1.02878213e+00 -1.10340393e+00 -2.33561039e-01
2.94866532e-01 1.17355418e+00 -4.87536639e-01 -4.02161956e-01
-1.77626088e-01 1.43235158e-02 -2.12291107e-01 2.41261899e-01
-1.15348017e+00 -8.75944436e-01 7.34192073e-01 -4.57451865e-03
3.41714144e-01 -4.68792617e-01 1.18341699e-01 -5.07519424e-01
-3.28805856e-02 -1.03706610e+00 -1.93920657e-01 -1.01260912e+00
-5.14687836e-01 3.11422795e-01 3.25987816e-01 -8.91869247e-01
-7.04946756e-01 -8.79855812e-01 -4.68585283e-01 7.06705809e-01
-1.20731771e+00 -1.16779268e+00 1.17602944e-01 3.01547706e-01
-4.45243061e-01 2.41218850e-01 1.04443264e+00 2.54172087e-01
-5.47338605e-01 3.38654518e-01 -2.62918621e-01 1.77360684e-01
6.13440312e-02 -1.06748331e+00 5.00592589e-01 1.03637123e+00
-3.53532374e-01 7.82805979e-01 1.23714471e+00 -6.03997111e-01
-2.02465320e+00 -1.25377011e+00 9.38843191e-01 -4.59390610e-01
1.33010483e+00 -6.79285645e-01 -5.08474588e-01 1.34783590e+00
1.73514307e-01 1.06348552e-01 -1.30393341e-01 -1.84742007e-02
-4.82522309e-01 -3.21611017e-01 -1.15599930e+00 4.83473808e-01
5.58710575e-01 -7.27385998e-01 -2.69269586e-01 3.62108499e-01
8.45928490e-01 -6.08202577e-01 -7.54444659e-01 3.61376494e-01
7.78392196e-01 -6.32932723e-01 5.83938241e-01 -5.59320867e-01
-8.89596567e-02 -1.13508677e+00 -4.39821482e-02 -7.62282848e-01
3.02077383e-01 -9.70709026e-01 -1.32332474e-01 1.21354401e+00
7.22442508e-01 -1.16752088e+00 6.06852293e-01 9.30188417e-01
-8.93501714e-02 -3.07628810e-01 -8.65844429e-01 -1.04130673e+00
-8.83995593e-02 -1.04570663e+00 5.93081474e-01 7.11890876e-01
6.20259941e-01 2.05390453e-01 3.68543863e-02 6.15620255e-01
2.73299247e-01 -3.72629881e-01 9.60164785e-01 -1.24156570e+00
-3.80660623e-01 -3.81363869e-01 -9.18392360e-01 -2.87636250e-01
4.17404175e-01 -7.98832238e-01 3.85394514e-01 -1.39576828e+00
-4.46372896e-01 -3.97518873e-01 2.44082555e-01 1.14628828e+00
1.03044760e+00 -4.28724699e-02 -1.26535028e-01 -4.92693722e-01
-5.29493153e-01 -3.02757889e-01 7.88651884e-01 -2.71870345e-01
1.16484754e-01 1.88905030e-01 -3.90642762e-01 8.42683375e-01
7.61676729e-01 -4.32658553e-01 -6.12213016e-01 -2.50766221e-02
7.29323208e-01 8.80374685e-02 1.09291291e+00 -1.33779836e+00
7.08129883e-01 -2.21650943e-01 -2.12909445e-01 -4.10733461e-01
3.24450061e-02 -1.36391008e+00 7.27144301e-01 6.88189626e-01
-2.72650003e-01 1.68734565e-01 7.79907763e-01 3.24405342e-01
-3.45248096e-02 -1.64820284e-01 3.43405038e-01 3.26424748e-01
-6.33389235e-01 -7.33462870e-02 -8.17355573e-01 -4.83019710e-01
1.21661985e+00 -9.04152244e-02 -7.48435080e-01 -3.27030778e-01
-3.97419363e-01 6.29327238e-01 5.05126715e-01 6.13390952e-02
3.49143982e-01 -7.86454141e-01 -2.10690543e-01 3.89282793e-01
-3.48182231e-01 8.20873603e-02 -2.09963232e-01 7.58018494e-01
-6.29897356e-01 6.32482111e-01 -1.77514061e-01 -4.05622512e-01
-1.53871787e+00 4.16763574e-01 7.06582844e-01 2.56097943e-01
-4.43284303e-01 3.23395967e-01 -4.37115490e-01 -8.38247240e-01
3.91404212e-01 -9.23225164e-01 3.22544605e-01 -6.77109241e-01
7.47189522e-01 3.34561527e-01 1.24947436e-01 -1.71242341e-01
-4.52552289e-01 8.75527114e-02 2.92556226e-01 -2.59033769e-01
1.14320910e+00 3.04121375e-01 -5.29110670e-01 3.97611290e-01
8.27754080e-01 -2.92773962e-01 -9.16257143e-01 4.73278701e-01
-3.38601649e-01 3.13661963e-01 7.63940588e-02 -7.18393564e-01
-7.97687590e-01 6.50222123e-01 2.04385921e-01 7.25516617e-01
7.63672531e-01 -2.24167585e-01 4.62367266e-01 1.14642227e+00
6.12659395e-01 -8.73025298e-01 -9.73504841e-01 6.58827066e-01
4.59641695e-01 -6.63931310e-01 1.22299023e-01 -4.33992743e-01
1.31678116e-02 1.19697773e+00 5.00515163e-01 -3.70569646e-01
5.03295660e-01 1.40205407e+00 -1.70848787e-01 -1.60163313e-01
-1.12010324e+00 3.32153767e-01 -4.55256581e-01 8.18235874e-01
-9.61227641e-02 1.84158757e-01 3.63402009e-01 3.37510914e-01
-3.88223469e-01 5.69717884e-01 9.06410575e-01 1.12395692e+00
-4.31177169e-01 -1.16766524e+00 -2.84140736e-01 2.44995207e-01
-1.62926286e-01 2.33292609e-01 -2.54469752e-01 1.80311012e+00
1.46395847e-01 1.11821032e+00 -8.43873620e-02 -7.10986018e-01
3.54690433e-01 -1.26991078e-01 5.19712806e-01 -2.80284911e-01
-5.01013458e-01 -7.18438566e-01 8.50403368e-01 -6.66407108e-01
-4.04143155e-01 -7.15102375e-01 -1.73115647e+00 -7.74777830e-01
-7.19607890e-01 2.37497389e-01 8.95497441e-01 9.92145836e-01
1.21353261e-01 6.03401363e-01 2.09749341e-01 -7.65238643e-01
-5.81215858e-01 -2.91548997e-01 -7.48534918e-01 -7.48165786e-01
7.37951934e-01 -3.45637530e-01 -4.83832389e-01 7.55637139e-02] | [6.412506103515625, 7.4565253257751465] |
28ab2c40-29b1-4770-962f-56c33af019db | toplight-lightweight-neural-networks-with | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Yu_TOPLight_Lightweight_Neural_Networks_With_Task-Oriented_Pretraining_for_Visible-Infrared_Recognition_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Yu_TOPLight_Lightweight_Neural_Networks_With_Task-Oriented_Pretraining_for_Visible-Infrared_Recognition_CVPR_2023_paper.pdf | TOPLight: Lightweight Neural Networks With Task-Oriented Pretraining for Visible-Infrared Recognition | Visible-infrared recognition (VI recognition) is a challenging task due to the enormous visual difference across heterogeneous images. Most existing works achieve promising results by transfer learning, such as pretraining on the ImageNet, based on advanced neural architectures like ResNet and ViT. However, such methods ignore the negative influence of the pretrained colour prior knowledge, as well as their heavy computational burden makes them hard to deploy in actual scenarios with limited resources. In this paper, we propose a novel task-oriented pretrained lightweight neural network (TOPLight) for VI recognition. Specifically, the TOPLight method simulates the domain conflict and sample variations with the proposed fake domain loss in the pretraining stage, which guides the network to learn how to handle those difficulties, such that a more general modality-shared feature representation is learned for the heterogeneous images. Moreover, an effective fine-grained dependency reconstruction module (FDR) is developed to discover substantial pattern dependencies shared in two modalities. Extensive experiments on VI person re-identification and VI face recognition datasets demonstrate the superiority of the proposed TOPLight, which significantly outperforms the current state of the arts while demanding fewer computational resources. | ['Wei Peng', 'Xu Cheng', 'Hao Yu'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['person-re-identification', 'face-recognition'] | ['computer-vision', 'computer-vision'] | [ 4.60231006e-01 -3.45720172e-01 -2.12571118e-02 -3.76079768e-01
-1.30549893e-01 -1.92528903e-01 6.43743277e-01 -7.54023850e-01
-4.61375237e-01 5.77287674e-01 9.37718451e-02 -5.16306907e-02
-1.86960429e-01 -5.34620583e-01 -6.05303586e-01 -8.84957612e-01
3.92833740e-01 9.77118090e-02 -1.87232673e-01 -2.02842161e-01
4.55532484e-02 3.70882064e-01 -1.71257710e+00 2.48541355e-01
8.89553130e-01 1.36788237e+00 1.70679569e-01 1.78044140e-01
-1.44519061e-01 8.21177721e-01 -4.27451909e-01 -6.17555916e-01
3.64011496e-01 -4.89215404e-01 -3.88706744e-01 5.97430300e-03
7.33920634e-01 -3.32341999e-01 -6.21871769e-01 1.24900365e+00
8.05952609e-01 1.43606603e-01 5.27268410e-01 -1.36574674e+00
-8.69981706e-01 2.35537753e-01 -6.37110054e-01 2.35877067e-01
5.55074625e-02 2.51471192e-01 5.03922343e-01 -9.93820667e-01
3.15714955e-01 1.26158988e+00 6.96066976e-01 8.25876355e-01
-9.43034530e-01 -9.14160252e-01 4.39946830e-01 7.88517296e-01
-1.56607997e+00 -6.20929062e-01 1.05267477e+00 -1.88637137e-01
7.34483659e-01 2.84238815e-01 2.88275778e-01 1.54484034e+00
-2.07846999e-01 6.62870407e-01 1.29344857e+00 -3.28049362e-01
2.84497510e-03 3.28703642e-01 -4.83787246e-02 5.35080969e-01
1.38464347e-01 3.52756023e-01 -5.48115313e-01 2.11290434e-01
6.10658169e-01 3.42588991e-01 -3.60426545e-01 -2.97103375e-01
-8.95600557e-01 3.11407566e-01 6.66944981e-01 2.34714448e-01
-3.31698477e-01 -2.35008672e-01 5.15350282e-01 4.25602853e-01
4.77240354e-01 -3.18190962e-01 -3.81853938e-01 3.13743204e-01
-6.28339171e-01 -2.89345205e-01 4.15367424e-01 8.69436502e-01
6.45835459e-01 2.90005565e-01 -3.17547619e-01 1.06858444e+00
3.71876627e-01 5.24893880e-01 4.17905182e-01 -2.63911068e-01
6.50202870e-01 5.69117308e-01 -1.43239960e-01 -1.14288461e+00
-3.41678530e-01 -7.93049455e-01 -1.42982566e+00 2.49170899e-01
2.45970696e-01 -6.27609249e-03 -1.09036458e+00 1.81612647e+00
3.35721105e-01 5.57948649e-01 2.00253919e-01 1.20014060e+00
1.16153252e+00 3.65289927e-01 3.18027377e-01 -1.37533024e-01
1.40474093e+00 -1.03761411e+00 -4.82492417e-01 -5.01816809e-01
2.04909429e-01 -5.48891425e-01 8.67855966e-01 1.71417177e-01
-6.71437204e-01 -9.56786692e-01 -9.72357631e-01 1.14971459e-01
-1.56407341e-01 3.51303518e-01 6.62563860e-01 8.72387528e-01
-1.05434370e+00 3.98188740e-01 -2.30659932e-01 -3.94901425e-01
6.83977783e-01 3.94744664e-01 -4.90171015e-01 -6.15183890e-01
-1.22860074e+00 6.59780264e-01 4.80231285e-01 8.35398197e-01
-8.41121674e-01 -7.35689878e-01 -7.57995784e-01 8.86990130e-02
4.87587661e-01 -7.71083951e-01 5.78918457e-01 -1.53954566e+00
-1.71316278e+00 8.02642167e-01 -2.04802714e-02 1.10630557e-01
6.42142296e-01 5.67522347e-02 -6.13981426e-01 -4.51132935e-03
-3.18491876e-01 4.01670754e-01 1.31959093e+00 -1.28126264e+00
-5.50606310e-01 -6.07518435e-01 7.79489279e-02 4.44316328e-01
-8.19387317e-01 1.09166302e-01 -6.73106253e-01 -5.67249775e-01
-1.12048842e-01 -8.25983167e-01 1.31141812e-01 -3.90359387e-02
-2.40584999e-01 -4.14577127e-01 6.06660247e-01 -8.79724562e-01
6.32803619e-01 -2.32481337e+00 2.58282781e-01 3.03367525e-01
1.14603631e-01 4.60066110e-01 -4.48195547e-01 -3.66071388e-02
-2.95100957e-01 -2.70988226e-01 5.96738700e-03 -3.91524613e-01
1.57008767e-02 2.44631954e-02 -7.93718025e-02 4.78131711e-01
1.20840140e-01 9.24541056e-01 -6.08977258e-01 -3.19180280e-01
3.04505080e-01 7.67584503e-01 -2.41042003e-01 3.96060437e-01
1.77859038e-01 7.91340590e-01 -2.77781397e-01 6.98325932e-01
1.21877933e+00 -2.25020573e-01 3.46585453e-01 -6.38931632e-01
8.64405632e-02 -1.23396523e-01 -1.13957679e+00 1.69639862e+00
-4.45970446e-01 1.91604093e-01 9.84364003e-02 -1.46509850e+00
9.55449283e-01 1.56416968e-01 2.59006590e-01 -1.29972827e+00
2.73321509e-01 -2.25055739e-02 -7.08357468e-02 -5.20669580e-01
4.26741503e-02 -7.91199431e-02 2.77558982e-01 1.32028803e-01
1.96665265e-02 8.32205117e-01 -4.08768743e-01 -1.15976043e-01
7.43325770e-01 3.31492245e-01 -2.18473792e-01 -6.62141759e-03
9.08392847e-01 -3.95117551e-01 8.36857557e-01 7.35759914e-01
-3.47360522e-01 4.43547696e-01 -4.51436173e-03 -5.54077327e-01
-7.26243198e-01 -7.21329331e-01 -1.18174799e-01 1.33021855e+00
6.02162540e-01 4.78568440e-03 -5.60021996e-01 -7.68351376e-01
-1.61970034e-01 1.62042394e-01 -7.35057890e-01 -4.20996994e-01
-4.45986718e-01 -7.85825193e-01 5.04958749e-01 5.72239637e-01
1.12630606e+00 -1.17269778e+00 -1.29520953e-01 -1.55714214e-01
-2.32094929e-01 -1.22474921e+00 -2.94989377e-01 -3.48812938e-01
-5.94948649e-01 -9.75681901e-01 -9.27634656e-01 -1.06709158e+00
6.78374827e-01 6.12628758e-01 8.29106808e-01 2.85885245e-01
-3.80633473e-01 3.78778577e-01 -4.20874536e-01 -1.02032468e-01
9.67018828e-02 -1.17331862e-01 1.30605340e-01 5.96071839e-01
6.36301696e-01 -4.42602009e-01 -8.72757554e-01 4.00028437e-01
-7.36661971e-01 1.34388581e-01 8.32745552e-01 1.31843066e+00
3.40109617e-01 1.73368484e-01 5.20073831e-01 -6.88040674e-01
1.37892187e-01 -2.95850307e-01 -4.16434824e-01 7.12808311e-01
-6.11636758e-01 -1.87934771e-01 6.96711957e-01 -6.09280944e-01
-1.66979897e+00 -4.57981341e-02 -8.72312933e-02 -6.85684681e-01
-3.64509851e-01 3.87411475e-01 -5.73945463e-01 -5.20722032e-01
4.32732522e-01 4.76224363e-01 -1.71592347e-02 -5.37088037e-01
1.59968928e-01 6.48270488e-01 5.72280407e-01 -4.65092212e-01
1.08368397e+00 5.02233267e-01 -1.92760348e-01 -6.75793827e-01
-7.55463839e-01 -2.70457745e-01 -5.34986377e-01 -1.83872998e-01
7.33863413e-01 -1.29404497e+00 -8.46649051e-01 8.40715587e-01
-9.57164288e-01 -1.52682841e-01 7.80244395e-02 5.57596684e-01
-1.19860165e-01 5.56723833e-01 -4.47222292e-01 -7.24706233e-01
-4.28953320e-01 -9.13520157e-01 6.72412217e-01 5.25314748e-01
5.66468120e-01 -8.41999650e-01 -2.32129991e-01 6.73471212e-01
7.31417954e-01 -8.19249302e-02 7.49939740e-01 -2.98627853e-01
-4.38596129e-01 1.14430748e-01 -8.14451098e-01 5.15901506e-01
7.65049085e-02 -4.70566750e-01 -1.38612640e+00 -5.98983586e-01
-9.80459452e-02 -4.96154517e-01 1.03217208e+00 9.88971367e-02
1.52070260e+00 -1.91484854e-01 -2.77168185e-01 1.11787605e+00
1.34378624e+00 1.67125762e-02 9.40299451e-01 4.30820346e-01
1.22480845e+00 8.33114743e-01 4.13673878e-01 3.34084541e-01
4.91434574e-01 6.98178768e-01 4.45140809e-01 -3.97098392e-01
-3.71993721e-01 -1.46130994e-01 2.76318789e-01 5.74180245e-01
-4.85165626e-01 -1.72963068e-02 -5.66475093e-01 1.41793087e-01
-1.83349109e+00 -9.87241566e-01 3.02209437e-01 2.12858200e+00
5.05761623e-01 -2.95203865e-01 -8.26129019e-02 -9.83505249e-02
9.01246428e-01 1.05554104e-01 -8.17707002e-01 1.83808491e-01
-3.59103203e-01 2.54452169e-01 3.67469490e-01 -4.19616438e-02
-1.06791496e+00 1.01143122e+00 5.03580523e+00 9.38812733e-01
-1.25541556e+00 2.73133636e-01 5.82474113e-01 6.85224235e-02
-1.10075317e-01 -2.30187222e-01 -6.86531425e-01 4.37009066e-01
6.23128533e-01 2.18749583e-01 6.42707884e-01 7.23090351e-01
-1.87516510e-01 3.62674743e-01 -8.88401151e-01 1.54580462e+00
5.21167397e-01 -8.71483803e-01 6.10463768e-02 -3.41929607e-02
5.75079739e-01 -7.02725500e-02 4.03363377e-01 5.43585777e-01
-1.00155808e-01 -1.13683641e+00 5.29307961e-01 6.51916742e-01
1.06356823e+00 -7.84208596e-01 8.38666379e-01 1.27588287e-01
-1.29829931e+00 -3.81706476e-01 -9.24094081e-01 5.95451221e-02
-3.04214001e-01 3.32562476e-01 -3.19700330e-01 9.14446652e-01
1.00800717e+00 8.47197175e-01 -7.11803496e-01 8.02363694e-01
-8.36746916e-02 3.94337058e-01 -1.34317249e-01 2.75639355e-01
-3.32281560e-01 -2.47429714e-01 2.10385457e-01 8.86898816e-01
1.84665188e-01 1.94093771e-02 2.14741126e-01 6.79396808e-01
-2.96976328e-01 1.60855409e-02 -4.31474686e-01 2.05977634e-01
2.86332011e-01 1.39458275e+00 -2.92658329e-01 -2.48319656e-01
-7.85384536e-01 1.40844202e+00 4.95433539e-01 7.47794271e-01
-8.59265625e-01 -1.53381288e-01 6.35211349e-01 -3.50701392e-01
4.20991510e-01 1.78459197e-01 -1.00296840e-01 -1.53360009e+00
2.53813326e-01 -1.07107663e+00 6.54060721e-01 -5.52632213e-01
-1.77059090e+00 8.42558086e-01 -3.30338389e-01 -1.27312100e+00
1.15012981e-01 -7.42322564e-01 -4.95802641e-01 1.01351881e+00
-2.06208825e+00 -1.45393848e+00 -8.07495892e-01 1.15829027e+00
3.01395863e-01 -4.89604473e-01 5.99059224e-01 6.94250762e-01
-1.04261923e+00 1.26615393e+00 4.40366007e-02 1.55268833e-01
9.33890760e-01 -6.01282537e-01 1.34643633e-02 8.51577938e-01
-1.04919039e-01 4.71771300e-01 2.27636620e-01 -5.01310468e-01
-1.68766642e+00 -1.14523828e+00 3.33982050e-01 -1.27009749e-01
2.54181474e-01 -4.17921752e-01 -1.01807952e+00 4.90227252e-01
3.10424250e-02 4.77939248e-01 7.00134039e-01 3.40014607e-01
-8.63354325e-01 -5.89662492e-01 -1.07661903e+00 3.37858468e-01
1.56902301e+00 -7.26837933e-01 -3.12028319e-01 2.70546585e-01
2.99750149e-01 -1.56029299e-01 -6.92856669e-01 5.87242603e-01
7.94199407e-01 -9.71219838e-01 1.20682573e+00 -5.71207643e-01
1.53893113e-01 -3.23606610e-01 -7.17464164e-02 -1.18191028e+00
-5.88415921e-01 -8.54765326e-02 3.13004740e-02 1.46040082e+00
-8.55409876e-02 -9.56773698e-01 7.37158597e-01 7.73675621e-01
1.26650527e-01 -2.09880099e-01 -1.02443564e+00 -7.43707240e-01
-2.89069384e-01 -1.88478276e-01 6.68409884e-01 1.16828263e+00
-5.60813189e-01 1.88500077e-01 -8.26457024e-01 4.10286248e-01
9.45229352e-01 1.30040646e-01 7.17255712e-01 -1.24469614e+00
-3.97604316e-01 -1.54892102e-01 -5.87689281e-01 -8.83972585e-01
3.70324194e-01 -8.54250848e-01 -3.01928092e-02 -1.12138462e+00
4.63801444e-01 -6.51174128e-01 -8.75184953e-01 5.24190426e-01
-4.20992821e-01 6.39834166e-01 2.69967139e-01 4.22387511e-01
-7.57714629e-01 9.82131779e-01 1.08763778e+00 -5.47713518e-01
-5.68771288e-02 -2.86570400e-01 -9.06304419e-01 5.71381152e-01
7.24339247e-01 -1.58611119e-01 -5.84532320e-01 -7.32582390e-01
-2.30690837e-02 -2.58702189e-01 7.08419204e-01 -1.00562942e+00
3.40787917e-01 -1.56185016e-01 9.43711102e-01 -1.86671734e-01
2.47371390e-01 -9.59389687e-01 1.77829042e-01 2.56727844e-01
-1.21772535e-01 -3.15613389e-01 2.18700975e-01 6.63714647e-01
-1.66576996e-01 1.70589462e-01 6.82408631e-01 8.58940557e-02
-1.16292131e+00 7.07142413e-01 3.22856218e-01 -2.20196232e-01
9.35712814e-01 -2.65762895e-01 -5.45921803e-01 -4.77143787e-02
-4.72709060e-01 2.43979879e-02 2.75999486e-01 7.47509301e-01
8.23978543e-01 -1.46288073e+00 -7.85991073e-01 4.66877669e-01
4.39379007e-01 -2.43087038e-01 1.17220163e+00 8.20605695e-01
-3.44609916e-02 7.33721405e-02 -5.32701015e-01 -6.38464808e-01
-1.27018058e+00 7.83934355e-01 4.53328341e-01 -5.98304681e-02
-8.23355258e-01 9.41531360e-01 6.86968505e-01 -4.52658057e-01
3.75469327e-01 5.42676926e-01 -5.09772897e-01 -1.18603610e-01
7.99633980e-01 7.41845816e-02 5.76121435e-02 -7.46544123e-01
-6.86066091e-01 6.73144341e-01 -3.27075213e-01 3.57983947e-01
1.19640458e+00 -3.50269496e-01 -6.48583248e-02 -1.18151218e-01
1.02284873e+00 -4.79180604e-01 -1.47109890e+00 -6.07832015e-01
-4.55947012e-01 -7.07032204e-01 1.30486593e-01 -8.47603261e-01
-1.34759426e+00 9.18936729e-01 1.17480075e+00 -3.45579803e-01
1.50458896e+00 -2.55569488e-01 5.95745325e-01 4.14964557e-01
5.30761123e-01 -1.05432248e+00 7.44874626e-02 2.40138263e-01
7.16026664e-01 -1.39688325e+00 -1.23187706e-01 -3.98681521e-01
-6.43272579e-01 8.85687113e-01 9.68410730e-01 1.00939706e-01
4.64633226e-01 -3.07535797e-01 1.98147699e-01 6.54897690e-02
-3.83386850e-01 -3.27832282e-01 3.88433158e-01 9.54771996e-01
3.85887688e-03 -8.87304991e-02 -3.29979099e-02 5.75332880e-01
2.87202805e-01 5.77924103e-02 -6.15259036e-02 5.90698659e-01
1.07909791e-01 -1.00794339e+00 -3.91536295e-01 2.75456607e-01
-2.04991058e-01 -1.13131933e-01 -1.74074724e-01 4.65181231e-01
4.06233490e-01 9.03738201e-01 -3.42067927e-02 -7.96780646e-01
3.05930406e-01 -1.26344815e-01 5.82916796e-01 -1.67457476e-01
-4.60081458e-01 -2.24641412e-01 -1.91992730e-01 -5.82171977e-01
-7.10796773e-01 -2.91560709e-01 -5.84925711e-01 -4.76868242e-01
-9.95404273e-02 -1.97512493e-01 6.31442130e-01 1.05246496e+00
5.86382985e-01 6.05558455e-01 8.70685279e-01 -8.86438549e-01
-6.13219261e-01 -9.68200684e-01 -5.65885603e-01 8.30826521e-01
2.16657251e-01 -8.80187750e-01 -2.92003751e-01 -1.88941628e-01] | [14.656471252441406, 0.9393826723098755] |
d0763971-cc7a-4737-a56f-547d3ebd6c18 | learning-programs-by-combining-programs | 2206.01614 | null | https://arxiv.org/abs/2206.01614v2 | https://arxiv.org/pdf/2206.01614v2.pdf | Learning programs by combining programs | The goal of inductive logic programming is to induce a logic program (a set of logical rules) that generalises training examples. Inducing programs with many rules and literals is a major challenge. To tackle this challenge, we introduce an approach where we learn small non-separable programs and combine them. We implement our approach in a constraint-driven ILP system. Our approach can learn optimal and recursive programs and perform predicate invention. Our experiments on multiple domains, including game playing and program synthesis, show that our approach can drastically outperform existing approaches in terms of predictive accuracies and learning times, sometimes reducing learning times from over an hour to a few seconds. | ['Céline Hocquette', 'Andrew Cropper'] | 2022-06-01 | null | null | null | null | ['program-synthesis', 'inductive-logic-programming'] | ['computer-code', 'methodology'] | [ 4.91474658e-01 4.41451609e-01 -5.64840555e-01 -5.32268226e-01
-8.10805202e-01 -8.90680015e-01 3.67438793e-01 9.08019114e-03
2.22923551e-02 9.57639039e-01 -3.87823910e-01 -8.36132228e-01
-3.27115357e-01 -1.31803954e+00 -1.05157721e+00 -3.77839953e-02
-1.99607238e-01 8.37881923e-01 5.48542619e-01 -1.80677578e-01
1.03380844e-01 2.82721758e-01 -1.78999662e+00 7.98222542e-01
1.01883185e+00 8.61721635e-01 -2.97078907e-01 8.70858073e-01
-2.99949706e-01 1.39583135e+00 -2.84969956e-01 -2.66732067e-01
1.41100049e-01 -3.09736520e-01 -1.15493000e+00 -2.63861865e-01
2.21677154e-01 7.28000933e-03 4.73789684e-02 1.12416339e+00
-6.70557544e-02 8.39202702e-02 2.51688927e-01 -1.39779055e+00
-3.52454394e-01 9.50148284e-01 2.09311116e-02 -2.36998886e-01
7.18517363e-01 1.41305298e-01 1.53005433e+00 -3.58819962e-01
5.20900905e-01 1.22013676e+00 5.91388822e-01 6.40174508e-01
-1.72355878e+00 -5.38681030e-01 1.02008834e-01 4.68933545e-02
-1.38328242e+00 -3.13985795e-01 5.01528084e-01 -3.73897344e-01
1.28238893e+00 4.36026901e-01 5.65222144e-01 3.76323611e-01
-6.04459681e-02 1.03847337e+00 1.18030250e+00 -7.01633453e-01
4.74955082e-01 5.16889870e-01 3.22234213e-01 1.01127613e+00
2.05496341e-01 2.55840153e-01 -3.23307753e-01 -3.63178849e-01
5.88027120e-01 -4.67114508e-01 4.65403572e-02 -3.17125738e-01
-9.23267484e-01 1.10471201e+00 -5.89929335e-02 -1.49892382e-02
2.16270331e-02 3.57876599e-01 5.21314442e-01 6.42859459e-01
1.33775324e-01 1.08865130e+00 -8.21067989e-01 -1.69060513e-01
-8.08431149e-01 7.88934171e-01 1.48591363e+00 1.23386335e+00
7.93485284e-01 -1.13641948e-01 -2.55863797e-02 4.66761887e-01
-7.65695944e-02 4.08324301e-01 1.56911425e-02 -9.61886466e-01
3.97979558e-01 7.91566908e-01 -9.74620283e-02 -7.22286344e-01
-1.45723194e-01 2.73890316e-01 -2.87287444e-01 2.61105269e-01
2.73829341e-01 -4.32331055e-01 -7.54583299e-01 1.59031463e+00
2.91307494e-02 2.02125683e-01 1.93042934e-01 2.52370834e-01
7.65360236e-01 8.23931396e-01 6.60698265e-02 -2.80377448e-01
1.11263227e+00 -7.56785989e-01 -2.05047920e-01 -5.67771256e-01
1.01188505e+00 -2.93074436e-02 1.17033994e+00 7.05338478e-01
-1.19751334e+00 -2.08679929e-01 -9.62283134e-01 1.83901668e-01
-2.84444958e-01 -2.20483750e-01 1.47951281e+00 6.15151703e-01
-8.73479247e-01 5.90553582e-01 -4.42854732e-01 9.61445048e-02
5.55346787e-01 9.86924827e-01 -3.24229121e-01 -2.16807649e-01
-1.03564668e+00 6.62727773e-01 1.26299953e+00 -5.33529878e-01
-5.96796572e-01 -1.03777933e+00 -1.20097685e+00 2.09516846e-02
7.90504932e-01 -6.18296206e-01 1.36958647e+00 -9.66469824e-01
-1.80865657e+00 8.31624568e-01 -1.33337408e-01 -6.03363395e-01
2.14859433e-02 7.54905716e-02 -2.34853968e-01 -2.07600653e-01
-2.53837198e-01 5.03293037e-01 3.10166031e-01 -9.93025482e-01
-9.46056306e-01 4.11351062e-02 8.52029502e-01 -3.33852470e-02
9.39252302e-02 3.43048513e-01 -3.41161489e-01 -1.00022063e-01
2.02001538e-02 -1.04651332e+00 -3.52908403e-01 -5.13321102e-01
-4.18399811e-01 -5.79870105e-01 5.09779632e-01 -1.83412284e-01
1.19030392e+00 -1.79289424e+00 3.72749031e-01 5.13819218e-01
9.76331905e-02 1.13855228e-01 1.62441418e-01 -9.96155217e-02
-8.82287323e-02 2.83324957e-01 -3.14069211e-01 4.04712111e-01
3.73135567e-01 6.43740594e-01 -5.01735568e-01 -1.54914990e-01
7.08189845e-01 1.21105921e+00 -1.01247585e+00 -6.77494645e-01
-2.99847480e-02 -2.35841393e-01 -1.29062998e+00 2.87676543e-01
-1.22491586e+00 1.19814761e-01 -4.09988403e-01 8.04124773e-01
2.87218601e-01 -2.33628452e-01 5.79403222e-01 3.76920700e-01
1.81272402e-01 7.28505373e-01 -1.24860740e+00 1.46415818e+00
-6.31415308e-01 6.01451933e-01 -3.63940567e-01 -1.32790983e+00
9.74487484e-01 1.35540962e-01 1.31827787e-01 -3.46193790e-01
1.47673609e-02 2.82572985e-01 -4.64266054e-02 -6.39867187e-01
6.16062395e-02 -5.11616588e-01 -5.87797821e-01 4.83838797e-01
1.52289823e-01 -8.32682490e-01 5.39040029e-01 -1.04534082e-01
1.30168855e+00 1.98388636e-01 5.17798364e-01 -3.86825681e-01
7.11260080e-01 5.75959980e-01 8.08529913e-01 9.38814819e-01
5.24772108e-01 -7.82418698e-02 1.26817107e+00 -8.39012980e-01
-7.67246842e-01 -6.97746277e-01 6.80276006e-02 1.47954214e+00
-3.22337270e-01 -6.53970838e-01 -3.73094618e-01 -9.11479354e-01
1.20549038e-01 9.10240471e-01 -2.56177098e-01 -7.07490817e-02
-9.32446301e-01 -6.39420748e-01 7.68633544e-01 6.84219718e-01
1.98132202e-01 -1.31768525e+00 -4.17614549e-01 1.08365893e-01
1.30178779e-01 -1.09111476e+00 1.86423317e-01 6.70374036e-01
-8.97963524e-01 -1.27410865e+00 4.48426843e-01 -1.05984330e+00
7.89299190e-01 -6.69355333e-01 1.54102278e+00 9.25673321e-02
-3.25764753e-02 2.56863125e-02 -1.36103317e-01 -6.18190169e-01
-6.58617079e-01 2.22466707e-01 -5.77619262e-02 -6.01308227e-01
6.43556297e-01 -6.25667572e-01 5.40349066e-01 -3.70609164e-01
-8.95862401e-01 1.58038765e-01 5.19398928e-01 9.77310300e-01
6.77973270e-01 6.88977659e-01 2.55201906e-01 -1.74631417e+00
7.70249724e-01 -3.29403549e-01 -1.21592045e+00 5.21334410e-01
-7.42968619e-01 5.59567392e-01 9.51993704e-01 -4.38900024e-01
-7.17934847e-01 5.05743921e-01 8.10714364e-02 -2.25375462e-02
-2.74167955e-01 7.64318824e-01 -3.64349484e-01 -3.08039725e-01
8.70312035e-01 2.90960997e-01 -2.67573416e-01 1.04162537e-01
4.39162225e-01 2.57962763e-01 6.02704287e-01 -1.48986375e+00
9.48859453e-01 -1.42590195e-01 2.15801090e-01 -2.19590679e-01
-9.65504229e-01 7.56203085e-02 -6.01675510e-01 2.98541009e-01
2.67720342e-01 -6.96530700e-01 -9.22935903e-01 -2.10369200e-01
-1.07703125e+00 -8.01506996e-01 -4.79221821e-01 9.19263735e-02
-7.91342497e-01 -2.12740287e-01 -3.41879398e-01 -7.16440082e-01
-1.99255526e-01 -1.13710225e+00 6.97476864e-01 1.70003608e-01
-5.01756489e-01 -1.16800416e+00 1.96367368e-01 -9.90588292e-02
6.94532916e-02 3.47922802e-01 1.45034611e+00 -9.72452879e-01
-6.54836476e-01 -2.63931334e-01 -1.52583137e-01 5.14798939e-01
2.60649528e-02 2.50271000e-02 -7.19261229e-01 1.62313744e-01
-3.40009212e-01 -7.44508922e-01 4.55839336e-01 1.24043882e-01
1.48686373e+00 -6.42996609e-01 -2.34002739e-01 6.94506645e-01
1.52404833e+00 3.09172809e-01 6.71827257e-01 3.25718373e-01
3.51331860e-01 4.08489645e-01 7.40616143e-01 1.80827886e-01
2.65730888e-01 4.59178835e-01 2.17932463e-02 3.12264830e-01
4.19807553e-01 -2.99287856e-01 2.94845432e-01 6.93965033e-02
-1.70140922e-01 3.64019006e-01 -1.31244147e+00 5.76494396e-01
-1.86052394e+00 -1.02400684e+00 2.90521920e-01 1.83870423e+00
1.55605495e+00 5.52199125e-01 1.94340572e-01 3.74905527e-01
2.18877584e-01 -1.63019300e-01 -3.52007627e-01 -9.14517462e-01
1.35924667e-01 1.15375423e+00 3.86940807e-01 8.29014182e-01
-1.30147171e+00 1.29147136e+00 7.89126682e+00 5.70060670e-01
-1.16204786e+00 -3.55258405e-01 3.03816289e-01 -1.37596160e-01
-4.56822008e-01 2.69992977e-01 -8.23079765e-01 1.21035770e-01
1.01027071e+00 -5.92393041e-01 9.14133489e-01 1.39263988e+00
-4.13856536e-01 2.11932813e-03 -1.69721413e+00 6.87020600e-01
-1.19643666e-01 -1.55342841e+00 -9.18862224e-02 -2.80296743e-01
8.65598679e-01 -3.98008794e-01 -1.33146361e-01 9.75388646e-01
9.46931541e-01 -1.47563672e+00 4.02831078e-01 1.66555330e-01
7.60615766e-01 -1.03445399e+00 6.43191397e-01 4.06671673e-01
-9.42172587e-01 -2.74890006e-01 -1.64377615e-01 -6.86850071e-01
-4.16243404e-01 6.29407585e-01 -1.00316405e+00 2.15707302e-01
3.29240352e-01 6.37043595e-01 -3.94556224e-01 6.41982317e-01
-7.66155124e-01 6.01169586e-01 -4.38838214e-01 -2.80664384e-01
2.64385752e-02 1.17259264e-01 1.67100966e-01 1.37455249e+00
-7.31815994e-02 5.49801528e-01 3.09287518e-01 1.27194071e+00
-2.66244002e-02 -2.67778724e-01 -8.12235713e-01 -2.26057246e-01
3.52335632e-01 9.53167260e-01 -3.06011826e-01 -5.64282477e-01
-3.73645216e-01 4.10265893e-01 3.74945909e-01 8.58281702e-02
-1.04983544e+00 -6.37531459e-01 5.80252469e-01 -1.44672915e-01
2.75284410e-01 2.51648091e-02 -6.14698350e-01 -1.04243183e+00
2.19876155e-01 -1.33266044e+00 4.46272492e-01 -2.74214119e-01
-8.06065261e-01 3.84793907e-01 3.48693907e-01 -6.75044298e-01
-5.98745465e-01 -1.04132581e+00 -5.65468490e-01 7.33486474e-01
-1.40336680e+00 -9.12497699e-01 7.08132237e-03 4.92567927e-01
9.84226987e-02 -2.00177953e-01 1.16733468e+00 6.86871335e-02
-4.22494859e-01 6.17160618e-01 -6.85776174e-01 1.08711153e-01
2.00765476e-01 -1.49746764e+00 1.19245261e-01 7.16534317e-01
6.71817139e-02 7.09961057e-01 6.98808610e-01 -2.59719104e-01
-1.82348216e+00 -1.21055102e+00 9.96675432e-01 -6.14658594e-01
7.74124384e-01 -1.70998633e-01 -6.55443728e-01 1.19605088e+00
-2.81257510e-01 2.68085092e-01 6.98117793e-01 6.29322767e-01
-6.25845969e-01 -1.52635857e-01 -1.08466780e+00 4.55031276e-01
9.94436145e-01 -6.94727540e-01 -8.90482962e-01 5.18177748e-01
7.09181905e-01 -7.99286604e-01 -9.21060443e-01 6.71956182e-01
3.96468192e-01 -6.73792660e-01 8.20613682e-01 -1.24975860e+00
8.59258592e-01 -3.37020874e-01 -9.02753472e-02 -1.10201192e+00
-1.88698664e-01 -7.16546953e-01 -5.38061202e-01 7.60060489e-01
6.89430714e-01 -6.18644953e-01 9.74567950e-01 1.07271528e+00
7.90983886e-02 -9.77226913e-01 -4.90040213e-01 -8.06414425e-01
3.04696113e-01 -8.52182508e-01 7.80721426e-01 7.59502828e-01
7.20932007e-01 4.45981741e-01 -6.62329420e-02 2.26269960e-01
2.82812119e-01 6.42574072e-01 8.45720708e-01 -1.34089041e+00
-7.88731933e-01 -5.77485383e-01 -4.24170583e-01 -8.13183784e-01
8.33398163e-01 -1.19809866e+00 2.46920049e-01 -1.15592492e+00
1.87561929e-01 -8.01685452e-01 -3.57722491e-02 1.20882332e+00
-5.48150530e-03 -1.71047911e-01 -1.28074273e-01 -3.92160028e-01
-8.25807810e-01 -2.03850955e-01 7.95538366e-01 -4.73986745e-01
-5.00688493e-01 1.28242180e-01 -1.17085838e+00 8.01493287e-01
8.39769900e-01 -4.55366790e-01 -5.63161254e-01 -3.78668666e-01
8.02356005e-01 -7.13668913e-02 2.31803805e-01 -1.08151805e+00
3.55831236e-01 -8.31195891e-01 -2.30224028e-01 -8.13042838e-03
-8.89131054e-02 -5.86656451e-01 1.28659755e-01 3.53753656e-01
-5.61040163e-01 -3.45090121e-01 6.49300218e-01 -5.28383180e-02
-3.21755856e-01 -4.88468558e-01 5.65476894e-01 -2.79919893e-01
-9.02082264e-01 1.54729486e-02 -1.28158256e-01 2.77778387e-01
1.23898041e+00 1.33729219e-01 8.22268520e-03 -2.40570363e-02
-6.12161517e-01 4.54078376e-01 3.18045825e-01 8.78955498e-02
5.31531274e-01 -1.10647726e+00 -5.65163255e-01 3.60589892e-01
2.65667737e-01 4.90333378e-01 -6.74760520e-01 5.06554186e-01
-6.86180651e-01 9.20316458e-01 -4.43460383e-02 -3.99904490e-01
-1.37070537e+00 4.82331336e-01 4.19841558e-01 -7.16265023e-01
-4.07616347e-01 9.91879761e-01 -1.20031826e-01 -8.62304568e-01
2.05368251e-01 -6.87146544e-01 1.48852104e-02 -5.35636425e-01
6.64623260e-01 -1.73615664e-01 -1.08010143e-01 2.48631716e-01
-4.20678288e-01 2.37247542e-01 3.79567854e-02 2.03433454e-01
1.63028371e+00 9.46933568e-01 -6.51636481e-01 2.77757436e-01
8.47535551e-01 -6.63000494e-02 -6.82971299e-01 -4.00366813e-01
4.17803824e-01 -2.58838266e-01 -3.78840044e-02 -7.06388474e-01
-7.08049417e-01 3.67130190e-01 -2.63235986e-01 1.82064787e-01
1.33482027e+00 2.23372951e-01 3.54949087e-01 1.03040850e+00
7.02178121e-01 -9.22428429e-01 -2.26893112e-01 1.09888601e+00
4.73002583e-01 -1.23714566e+00 1.77950516e-01 -6.33069098e-01
-4.59406793e-01 1.18309176e+00 6.08747184e-01 -2.72651315e-01
2.60499507e-01 8.24807465e-01 -6.05563939e-01 -2.40619138e-01
-1.12237775e+00 -2.67034590e-01 2.35220730e-01 5.66161036e-01
4.23321903e-01 3.02723199e-01 -5.16994111e-02 9.23950970e-01
-5.34681261e-01 5.82321405e-01 2.35694036e-01 1.27003348e+00
-4.26543117e-01 -1.56822050e+00 -2.03914002e-01 6.06069088e-01
-3.70653600e-01 -2.74996519e-01 -2.80476570e-01 8.10123980e-01
2.53549993e-01 6.18984103e-01 -2.49800086e-01 -6.64201200e-01
3.75741720e-01 4.93489385e-01 9.29503024e-01 -1.28995943e+00
-3.34088713e-01 -7.31226802e-01 6.51899517e-01 -5.64787745e-01
-3.41371834e-01 -4.37959343e-01 -1.39338326e+00 -4.41347480e-01
-1.60075918e-01 2.41918460e-01 1.65530726e-01 1.02594972e+00
-2.32584495e-02 5.40614724e-01 4.84092385e-01 -2.69253224e-01
-5.23661375e-01 -3.57482046e-01 -4.72202957e-01 4.36790697e-02
4.67364639e-02 -2.77956069e-01 -1.98174685e-01 2.34402016e-01] | [8.717412948608398, 7.151403427124023] |
e9929964-1972-4d8c-9b75-5f5bd5488ba4 | recognizing-scenes-from-novel-viewpoints | 2112.01520 | null | https://arxiv.org/abs/2112.01520v1 | https://arxiv.org/pdf/2112.01520v1.pdf | Recognizing Scenes from Novel Viewpoints | Humans can perceive scenes in 3D from a handful of 2D views. For AI agents, the ability to recognize a scene from any viewpoint given only a few images enables them to efficiently interact with the scene and its objects. In this work, we attempt to endow machines with this ability. We propose a model which takes as input a few RGB images of a new scene and recognizes the scene from novel viewpoints by segmenting it into semantic categories. All this without access to the RGB images from those views. We pair 2D scene recognition with an implicit 3D representation and learn from multi-view 2D annotations of hundreds of scenes without any 3D supervision beyond camera poses. We experiment on challenging datasets and demonstrate our model's ability to jointly capture semantics and geometry of novel scenes with diverse layouts, object types and shapes. | ['Georgia Gkioxari', 'David F. Fouhey', 'Justin Johnson', 'Devendra Singh Chaplot', 'Nikhila Ravi', 'Alexander Kirillov', 'Shengyi Qian'] | 2021-12-02 | null | null | null | null | ['scene-recognition'] | ['computer-vision'] | [ 2.46050254e-01 8.43147561e-02 3.37861001e-01 -7.15822756e-01
-1.94505721e-01 -1.22474754e+00 5.61333954e-01 -1.49362355e-01
-1.15003526e-01 5.96484775e-03 -1.94495097e-01 -7.74173364e-02
3.28531891e-01 -6.96623147e-01 -8.80606949e-01 -1.74783304e-01
2.23798633e-01 7.93066621e-01 3.93627584e-01 -1.30266905e-01
2.17559591e-01 1.04718828e+00 -1.62864733e+00 3.79139543e-01
1.68561742e-01 8.65045607e-01 6.47768080e-01 9.34579134e-01
-1.33477166e-01 7.11296916e-01 -1.17308468e-01 -9.04271975e-02
7.32712507e-01 -6.93591386e-02 -7.05800772e-01 9.80035007e-01
7.83535719e-01 -6.06179357e-01 -4.08269316e-01 7.52126276e-01
-1.80674046e-02 2.84933746e-01 4.73389119e-01 -1.05665827e+00
-6.12627923e-01 -2.11950794e-01 -4.88084823e-01 9.00919959e-02
9.72345412e-01 4.11427915e-01 8.78405988e-01 -1.08424425e+00
8.47318649e-01 1.21540713e+00 3.12947631e-01 4.30264711e-01
-1.29678106e+00 6.25738278e-02 5.65265656e-01 1.10165365e-01
-1.23553324e+00 -4.43724364e-01 1.06289470e+00 -5.47068417e-01
1.33995461e+00 1.54322973e-02 8.59727144e-01 8.51716876e-01
-3.16023201e-01 6.37373626e-01 1.11740279e+00 -3.08464974e-01
4.31858391e-01 4.83820498e-01 -1.82208680e-02 6.97524905e-01
1.02180853e-01 -4.82939929e-02 -5.33926725e-01 3.29017133e-01
1.04122651e+00 5.94997406e-01 4.40669805e-02 -1.10927463e+00
-1.20685875e+00 2.53248781e-01 6.83395863e-01 7.32326657e-02
-3.12886745e-01 2.06060093e-02 -1.10190779e-01 1.52621374e-01
2.83687949e-01 6.43565536e-01 -7.46813238e-01 1.70439221e-02
-2.02710941e-01 -1.22922853e-01 6.10575318e-01 1.34712136e+00
9.31640983e-01 -1.18944168e-01 7.65953124e-01 2.75145948e-01
1.15954250e-01 9.20320511e-01 1.85191594e-02 -1.21772921e+00
3.35720748e-01 1.05252111e+00 1.06710374e-01 -1.10562027e+00
-5.13852000e-01 1.08870141e-01 -5.47890723e-01 3.55448842e-01
2.65746176e-01 4.48394865e-01 -1.09899473e+00 1.33790362e+00
5.25785804e-01 -1.53484672e-01 3.03395055e-02 1.01912010e+00
7.35541165e-01 4.41312164e-01 -5.63599467e-01 3.76124978e-01
1.15952313e+00 -8.57305229e-01 1.26071880e-02 -7.46969163e-01
3.11134934e-01 -5.36945760e-01 1.14590561e+00 5.86879551e-01
-1.20420516e+00 -9.53807056e-01 -8.09754252e-01 -4.54453468e-01
-6.10033154e-01 9.17928070e-02 6.49533987e-01 4.68541384e-01
-1.15479839e+00 1.69589564e-01 -1.01558244e+00 -7.51004636e-01
2.96472281e-01 2.52706617e-01 -8.61233473e-01 -4.47292984e-01
-1.94513351e-01 1.01435137e+00 3.36681783e-01 8.63661021e-02
-1.14459300e+00 -2.97733635e-01 -1.10031581e+00 -2.56872326e-01
5.02422810e-01 -9.60770845e-01 1.06502640e+00 -9.70779777e-01
-1.42088640e+00 1.20173919e+00 -2.49472573e-01 2.26327688e-01
2.17092752e-01 -3.43781620e-01 4.31229174e-02 5.11565983e-01
-4.62690048e-04 6.12194419e-01 7.76813388e-01 -1.78905857e+00
-5.27350187e-01 -9.79862094e-01 6.00641847e-01 6.03201807e-01
1.08653538e-01 -4.68382746e-01 -4.10495162e-01 -7.90894218e-03
8.37139070e-01 -9.70757365e-01 -4.52951819e-01 2.21272305e-01
-4.97287631e-01 2.54629493e-01 8.10638249e-01 -1.88344792e-01
-2.02358037e-01 -2.15689945e+00 5.50053775e-01 1.95230067e-01
3.29972118e-01 -3.17914635e-01 -2.09176376e-01 3.05156469e-01
8.76074433e-02 -1.39526904e-01 1.32945523e-01 -6.08443797e-01
-6.26287088e-02 4.46076870e-01 -4.04214621e-01 5.24607837e-01
1.64485410e-01 8.07729602e-01 -9.70671892e-01 6.52497783e-02
7.49081552e-01 3.86527389e-01 -7.69590378e-01 6.35039449e-01
-3.49630535e-01 7.65532434e-01 -5.18500090e-01 5.80542982e-01
8.44636202e-01 -5.57888508e-01 1.59561500e-01 -1.98120743e-01
-1.93908326e-02 2.49526843e-01 -1.18338966e+00 2.27680874e+00
-4.92023140e-01 3.42486531e-01 -1.67326227e-01 -9.88444686e-01
9.53459263e-01 -1.15088359e-01 2.29857624e-01 -4.81009573e-01
-2.76283883e-02 -9.32955742e-02 -4.58217770e-01 -7.28913009e-01
4.05209869e-01 -1.75949022e-01 -2.95792520e-01 5.23831427e-01
1.44179717e-01 -8.98314595e-01 -3.12766612e-01 3.43646973e-01
1.02991784e+00 2.40706965e-01 3.43709797e-01 3.38973284e-01
1.07896000e-01 1.71036169e-01 7.15986416e-02 8.43180001e-01
1.37949005e-01 7.08181858e-01 -9.24986973e-02 -9.90638316e-01
-1.20818341e+00 -1.69365323e+00 3.19343835e-01 9.76939023e-01
5.89750588e-01 -1.09633133e-01 -4.19777334e-02 -6.28818035e-01
3.99345681e-02 5.41455626e-01 -6.08197808e-01 8.28319788e-02
-3.21766049e-01 1.32218912e-01 -5.41417599e-02 7.12236941e-01
4.59249288e-01 -7.56032944e-01 -1.33898950e+00 -4.32699114e-01
3.56184654e-02 -1.49705803e+00 -1.54199392e-01 3.16752791e-01
-9.03587818e-01 -1.06019866e+00 -1.73476323e-01 -7.33170033e-01
1.15645909e+00 1.04587817e+00 1.20839310e+00 -2.47484416e-01
-5.49703777e-01 1.18148637e+00 -5.24388790e-01 -1.61951482e-01
-8.77167210e-02 -1.84638456e-01 2.54147828e-01 -1.54548623e-02
3.20600569e-01 -8.57897341e-01 -3.94151598e-01 1.59894437e-01
-7.64927804e-01 5.64468622e-01 6.11338854e-01 3.27097446e-01
6.76794469e-01 -2.06406817e-01 -1.70620620e-01 -9.66393113e-01
-4.88292545e-01 -2.44957924e-01 -5.87579012e-01 2.62545109e-01
1.90478191e-01 -1.83996037e-01 7.33875453e-01 -2.02262402e-01
-1.06801379e+00 8.14043581e-01 5.79952240e-01 -6.96965992e-01
-9.34338272e-01 1.14111774e-01 -4.42549944e-01 7.09321648e-02
6.40114784e-01 2.98681080e-01 -3.07375401e-01 -5.57648897e-01
9.30028617e-01 2.38745824e-01 6.30413651e-01 -4.84042108e-01
1.09155798e+00 1.00416958e+00 -7.91167989e-02 -8.37066472e-01
-1.08769095e+00 -7.65071869e-01 -1.72663784e+00 -9.27462205e-02
9.46619093e-01 -1.07905936e+00 -6.47957087e-01 3.23255718e-01
-1.33603966e+00 -3.72438878e-01 -4.81800377e-01 3.82293284e-01
-7.95803010e-01 1.72369301e-01 -1.22609451e-01 -7.69972861e-01
4.12372798e-01 -8.97152781e-01 1.40601671e+00 2.33302504e-01
-3.63391340e-02 -9.14804697e-01 -7.22862929e-02 5.22349238e-01
-7.83432275e-02 4.01395440e-01 6.04330420e-01 -4.53421772e-01
-1.10701835e+00 -7.95209184e-02 -2.20283419e-01 1.84968933e-01
4.65920568e-01 -3.02995861e-01 -1.35411489e+00 -1.67396948e-01
2.89785296e-01 -4.60932255e-01 5.89081705e-01 1.14751063e-01
1.12298787e+00 -2.59200949e-02 -2.94263750e-01 9.28482711e-01
1.53088129e+00 3.07947487e-01 1.04728699e-01 1.04217350e-01
1.12328160e+00 7.02842593e-01 2.53629118e-01 4.31515396e-01
6.44480348e-01 5.27695060e-01 6.41090035e-01 -1.42630860e-01
6.46530688e-02 -2.98708707e-01 1.43023595e-01 8.64139378e-01
-1.11757606e-01 -7.38535747e-02 -1.10789323e+00 3.81992042e-01
-1.44942856e+00 -7.39983559e-01 2.69735605e-01 1.96065617e+00
3.18764448e-01 1.03351675e-01 -1.44294798e-01 -2.47529700e-01
2.94469565e-01 1.16260899e-02 -1.19643676e+00 -3.74632388e-01
-2.36796498e-01 -9.40474421e-02 3.40058327e-01 3.64827991e-01
-9.25731063e-01 9.79067206e-01 6.48335075e+00 -2.86373258e-01
-8.13197017e-01 -2.36550644e-01 4.42221910e-01 -9.75435227e-02
-4.87695903e-01 2.75498122e-01 -4.62930262e-01 -3.87572706e-01
2.41975531e-01 1.87329218e-01 8.12641501e-01 9.78159249e-01
-2.62829751e-01 -2.99134165e-01 -1.70866120e+00 1.29262602e+00
6.23383939e-01 -1.01761460e+00 1.73550516e-01 -8.12385045e-03
8.55826855e-01 1.69891104e-01 1.99460000e-01 -1.81738243e-01
6.40994430e-01 -1.04017961e+00 8.15841973e-01 7.03685641e-01
7.43147790e-01 -1.31066948e-01 2.15876579e-01 6.95065975e-01
-1.05446696e+00 -1.32798195e-01 -3.81887853e-01 -5.30150294e-01
1.88903674e-01 2.35660851e-01 -1.21430254e+00 4.75673914e-01
7.90390730e-01 1.23142314e+00 -9.78374600e-01 6.63273275e-01
-1.92822337e-01 -2.09874958e-01 -5.39916515e-01 2.01164156e-01
-6.96837381e-02 -2.06990898e-01 3.43487501e-01 6.29859626e-01
3.49115521e-01 5.43787181e-01 4.08116400e-01 9.32046831e-01
8.12049583e-02 -4.91810411e-01 -1.23436475e+00 2.11310983e-01
4.05870974e-01 1.06906748e+00 -9.29950237e-01 -3.65597785e-01
-4.81381387e-01 1.37706530e+00 3.74073178e-01 4.90981549e-01
-3.68585914e-01 -2.98226681e-02 4.24279600e-01 2.28533484e-02
3.58864754e-01 -8.04427803e-01 -3.61936480e-01 -1.48718750e+00
9.47121978e-02 -5.77597737e-01 1.69626981e-01 -1.60238552e+00
-1.38374913e+00 5.78679979e-01 1.75682858e-01 -1.22386217e+00
-1.03011601e-01 -9.88044977e-01 -6.59011826e-02 6.01999044e-01
-1.32961237e+00 -1.41853118e+00 -6.96211338e-01 7.62857318e-01
7.51921415e-01 -7.58913681e-02 8.72971594e-01 -3.40391815e-01
1.18621826e-01 -1.48946092e-01 -7.66530335e-02 9.79071669e-03
4.63510603e-01 -1.40976334e+00 7.06199944e-01 5.59908688e-01
6.50514960e-01 5.26629984e-01 4.72231001e-01 -3.28453124e-01
-1.91202593e+00 -8.43315959e-01 4.76496398e-01 -1.42241597e+00
1.55309021e-01 -8.11178505e-01 -5.54290295e-01 1.08364618e+00
-3.59756872e-02 2.89623052e-01 5.67646444e-01 1.47175327e-01
-8.88268292e-01 -1.06717013e-01 -9.60587919e-01 4.19593871e-01
1.48933196e+00 -9.23606038e-01 -7.59299219e-01 3.63869518e-01
7.72104681e-01 -6.23865962e-01 -5.73728144e-01 2.49681130e-01
5.12868524e-01 -1.14659846e+00 1.35204554e+00 -7.25312114e-01
2.15447560e-01 -5.48381805e-01 -7.29669988e-01 -1.35849810e+00
-1.61557660e-01 -7.11504221e-02 2.49613568e-01 6.03878319e-01
8.25830083e-03 -5.33205569e-01 6.38029397e-01 8.15065265e-01
-6.65696859e-02 -3.60593870e-02 -7.36455619e-01 -5.92821598e-01
-2.47238249e-01 -5.64043105e-01 5.48271656e-01 8.16976786e-01
-2.01067865e-01 4.63787049e-01 -8.67895409e-02 5.58257401e-01
5.54306209e-01 7.79045045e-01 1.40249467e+00 -1.27234757e+00
-4.41161036e-01 -1.00463321e-02 -8.29678237e-01 -1.31527126e+00
8.00001025e-02 -8.53462160e-01 1.44231424e-01 -1.48230004e+00
5.30550063e-01 -3.60774130e-01 7.41348416e-02 5.11596262e-01
2.30655491e-01 2.72096217e-01 4.56331074e-01 1.99258864e-01
-1.11903203e+00 4.99756634e-01 1.32002485e+00 -1.55972227e-01
-2.05382392e-01 -3.05210054e-01 -4.94216055e-01 1.01511061e+00
5.57824016e-01 -1.89697847e-01 -6.27374291e-01 -1.09616196e+00
3.93811524e-01 -4.78391573e-02 7.44580626e-01 -9.79818106e-01
3.11612457e-01 -4.56414044e-01 9.43201005e-01 -7.09042430e-01
8.81133318e-01 -1.18141568e+00 2.50812858e-01 4.35587354e-02
-3.50471586e-01 2.19290867e-01 2.42300496e-01 8.16175103e-01
2.06724226e-01 2.36772940e-01 2.67149150e-01 -7.00338602e-01
-1.10813498e+00 2.01960504e-01 -1.69309825e-01 -1.00682385e-01
1.03392041e+00 -4.99430060e-01 -2.66213149e-01 -4.11936104e-01
-1.10174668e+00 1.83593035e-02 1.25362480e+00 4.45029080e-01
1.05476904e+00 -1.19120467e+00 -1.46625593e-01 7.30932474e-01
4.52611357e-01 5.80439031e-01 4.42085773e-01 1.75574064e-01
-6.81515753e-01 4.05897409e-01 -3.98349434e-01 -1.05404496e+00
-1.04084289e+00 9.04307902e-01 2.93782264e-01 3.81883889e-01
-7.36140847e-01 8.59458268e-01 8.26982319e-01 -1.06132400e+00
4.41776067e-02 -6.47034883e-01 2.14208156e-01 -5.59813261e-01
3.67642730e-01 -2.60550063e-02 -1.90443888e-01 -8.92465353e-01
-4.57169473e-01 9.80042934e-01 1.73972249e-01 -2.31699899e-01
1.45584214e+00 -5.52390754e-01 -1.27253160e-01 1.08388019e+00
1.30549157e+00 -1.43523067e-01 -1.56551778e+00 -4.37114120e-01
-1.99211076e-01 -1.03586757e+00 -2.97262281e-01 -7.72623837e-01
-7.53484428e-01 1.25529003e+00 4.33255881e-01 -2.37837434e-02
1.14913142e+00 5.38292527e-01 4.37971115e-01 9.84164715e-01
8.90074253e-01 -7.88725734e-01 7.98343480e-01 6.27584696e-01
6.80826604e-01 -1.41527700e+00 -8.38529691e-03 -3.69170338e-01
-7.01762080e-01 1.21812201e+00 7.07230628e-01 -2.89594084e-01
6.77394152e-01 -3.09768710e-02 7.79669508e-02 -5.91565371e-01
-7.46483326e-01 -3.16297919e-01 2.87637055e-01 8.78637552e-01
-2.70686626e-01 1.66909263e-01 1.22333264e+00 1.47738382e-01
-2.22237602e-01 -5.04706204e-01 6.42982781e-01 8.96080434e-01
-5.62874675e-01 -7.09448099e-01 -2.12196976e-01 6.11761063e-02
3.92421246e-01 1.92196816e-01 -7.54115999e-01 5.57176471e-01
2.25442857e-01 8.97072792e-01 2.93689758e-01 -5.09386301e-01
5.73756039e-01 -2.90753040e-02 9.75919724e-01 -1.25915861e+00
7.54131703e-03 -8.42519253e-02 -3.04414392e-01 -8.69304419e-01
-6.74658358e-01 -8.51479352e-01 -1.16517544e+00 9.92466435e-02
3.64646614e-02 -3.93895268e-01 7.95971632e-01 9.16522145e-01
3.73814106e-01 2.13382453e-01 1.04279613e+00 -1.24742568e+00
-1.09132707e-01 -3.96273911e-01 -7.75591493e-01 6.45128667e-01
6.72366261e-01 -5.80068529e-01 -3.77657980e-01 6.16647422e-01] | [8.365771293640137, -2.971015453338623] |
0e688b28-9894-4c6d-b1a2-ef1908bfb464 | an-implicit-alignment-for-video-super | 2305.00163 | null | https://arxiv.org/abs/2305.00163v1 | https://arxiv.org/pdf/2305.00163v1.pdf | An Implicit Alignment for Video Super-Resolution | Video super-resolution commonly uses a frame-wise alignment to support the propagation of information over time. The role of alignment is well-studied for low-level enhancement in video, but existing works have overlooked one critical step -- re-sampling. Most works, regardless of how they compensate for motion between frames, be it flow-based warping or deformable convolution/attention, use the default choice of bilinear interpolation for re-sampling. However, bilinear interpolation acts effectively as a low-pass filter and thus hinders the aim of recovering high-frequency content for super-resolution. This paper studies the impact of re-sampling on alignment for video super-resolution. Extensive experiments reveal that for alignment to be effective, the re-sampling should preserve the original sharpness of the features and prevent distortions. From these observations, we propose an implicit alignment method that re-samples through a window-based cross-attention with sampling positions encoded by sinusoidal positional encoding. The re-sampling is implicitly computed by learned network weights. Experiments show that the proposed implicit alignment enhances the performance of state-of-the-art frameworks with minimal impact on both synthetic and real-world datasets. | ['Angela Yao', 'Michael Bi Mi', 'Xin Wang', 'Ziwei Yu', 'Kai Xu'] | 2023-04-29 | null | null | null | null | ['video-super-resolution'] | ['computer-vision'] | [ 4.19197649e-01 -2.80172914e-01 -2.52558351e-01 -3.43025327e-01
-5.21523178e-01 -1.96572408e-01 5.78238606e-01 -1.56492114e-01
-3.15542459e-01 7.81710565e-01 6.97045863e-01 2.20968515e-01
-4.73808311e-02 -9.30186033e-01 -8.06686044e-01 -7.78909147e-01
-1.23105623e-01 -5.79726160e-01 5.15185893e-01 -3.23015988e-01
4.25976455e-01 3.98128480e-01 -1.60199106e+00 5.99442899e-01
9.50183153e-01 7.30008543e-01 4.19285119e-01 5.00136018e-01
1.01955198e-01 7.49440312e-01 -2.47620806e-01 -2.10388377e-01
5.89130282e-01 -5.42222977e-01 -6.74953163e-01 -3.31458412e-02
7.54448533e-01 -7.03127563e-01 -5.99537075e-01 1.13269508e+00
5.82155406e-01 3.37182909e-01 2.26345256e-01 -5.84024906e-01
-7.23639369e-01 4.51918781e-01 -7.38451123e-01 7.28418171e-01
2.47990102e-01 1.13509528e-01 8.68046284e-01 -9.97170568e-01
7.36756265e-01 1.11346531e+00 7.63509929e-01 4.73711908e-01
-1.31948388e+00 -6.84156895e-01 -5.58047257e-02 4.75295573e-01
-1.33220327e+00 -5.94006181e-01 9.45022285e-01 -2.97179848e-01
5.46187699e-01 3.86363268e-01 5.85851610e-01 8.21238995e-01
3.66263390e-01 1.43652171e-01 1.05360734e+00 -3.98724586e-01
2.59611066e-02 -1.71849951e-01 -2.18363419e-01 3.15242857e-01
1.40493214e-01 2.27330327e-01 -9.07025158e-01 8.97736624e-02
1.52104795e+00 -1.10042654e-01 -6.84318364e-01 -1.39649361e-01
-1.19765341e+00 5.35561204e-01 4.53698307e-01 4.84147370e-01
-5.42928755e-01 2.60897521e-02 3.91473413e-01 1.16753235e-01
8.05223227e-01 4.18636292e-01 -2.05458894e-01 1.27729863e-01
-1.31671882e+00 2.54171044e-01 -6.61816522e-02 5.65022528e-01
6.52160347e-01 2.46978030e-01 -5.33919871e-01 8.65121067e-01
-1.51279628e-01 2.95224134e-02 5.97100019e-01 -1.19610667e+00
4.71713036e-01 1.13103531e-01 4.52843696e-01 -1.20929480e+00
-7.27385134e-02 -5.76859295e-01 -1.09364414e+00 3.69808346e-01
4.31782752e-01 7.91672170e-02 -5.43455124e-01 1.70773864e+00
2.92563647e-01 7.98893213e-01 -2.27137268e-01 1.34782398e+00
5.04969358e-01 6.46296561e-01 -5.73869096e-03 -4.52356279e-01
1.29652011e+00 -1.03342676e+00 -1.07011509e+00 3.26044969e-02
-7.32185319e-02 -8.75367939e-01 9.14483845e-01 2.58326024e-01
-1.44157374e+00 -9.83238995e-01 -1.18583739e+00 -2.72520363e-01
1.88046530e-01 -1.36085153e-01 2.42028862e-01 3.71978402e-01
-1.04757917e+00 1.04928553e+00 -6.76898599e-01 -2.72513721e-02
3.79529655e-01 3.64116393e-02 -3.72387350e-01 -9.27433670e-02
-1.39425504e+00 8.73164475e-01 1.04242466e-01 8.68563801e-02
-5.40712833e-01 -9.56186354e-01 -6.20192409e-01 8.47075507e-02
1.38744280e-01 -7.30946124e-01 7.55019665e-01 -1.31615794e+00
-1.51176310e+00 5.07481217e-01 -3.91543567e-01 -6.13078475e-01
5.95259964e-01 -4.78548527e-01 -3.82478982e-01 4.13339108e-01
-5.70393316e-02 4.43883091e-01 1.32687044e+00 -1.21348238e+00
-6.93196774e-01 -8.69562179e-02 1.18465967e-01 3.58807266e-01
-4.82574522e-01 3.63980234e-02 -3.21476758e-01 -1.18407631e+00
2.55650245e-02 -4.83458728e-01 -1.37258291e-01 1.02212310e-01
8.32154155e-02 2.88482219e-01 8.71585190e-01 -1.14863122e+00
1.63495076e+00 -2.11991072e+00 1.69634849e-01 -1.59950882e-01
2.85468370e-01 4.02453899e-01 -2.47354805e-01 1.67240977e-01
-2.82164186e-01 -3.39696854e-02 -2.83896625e-01 -2.53537655e-01
-6.75799429e-01 -2.47269310e-02 -3.67940694e-01 6.32452488e-01
4.36614871e-01 5.14470756e-01 -1.00514925e+00 -3.40267688e-01
4.99329329e-01 1.15147877e+00 -7.18486488e-01 2.16254503e-01
2.11569339e-01 7.01025069e-01 -2.16849610e-01 1.37534276e-01
9.12457108e-01 3.21496353e-02 5.07293828e-02 -7.85216451e-01
-6.34429574e-01 2.14025334e-01 -1.22641051e+00 1.75910163e+00
-4.47361797e-01 8.70150149e-01 1.38736159e-01 -8.22869956e-01
7.32902825e-01 3.16743791e-01 6.36706233e-01 -9.43005502e-01
-6.46019280e-02 8.63485113e-02 -1.80963188e-01 -5.35770953e-01
7.53231287e-01 2.83876993e-02 7.85511792e-01 3.94962467e-02
-2.77311444e-01 2.32002631e-01 4.69213463e-02 -1.27690509e-01
7.92150795e-01 3.91111344e-01 2.94618934e-01 -2.91907609e-01
7.16010571e-01 -4.93660152e-01 7.27000296e-01 4.15029764e-01
-9.07358900e-02 1.14787865e+00 -7.10197315e-02 -5.16665637e-01
-1.45888793e+00 -7.28187025e-01 -2.49943033e-01 1.03677547e+00
3.88727099e-01 -2.72170484e-01 -9.78303313e-01 -1.31546229e-01
-2.96142250e-01 5.28489053e-01 -7.12193608e-01 -9.68220830e-02
-9.75646734e-01 -5.95860183e-01 2.03286022e-01 4.48334515e-01
8.48538399e-01 -9.06301558e-01 -9.97530997e-01 4.31906760e-01
-6.03771389e-01 -1.15111589e+00 -8.41305256e-01 -3.55693072e-01
-1.06309783e+00 -8.53538334e-01 -1.03287196e+00 -5.45755327e-01
6.75403237e-01 7.78912067e-01 9.95063961e-01 1.81724906e-01
-1.49063468e-01 -1.67201869e-02 -4.25431758e-01 3.80697548e-01
-8.56838077e-02 -2.54931152e-01 -2.14042157e-01 4.42307353e-01
-1.97064921e-01 -8.74574125e-01 -1.06049645e+00 3.79866600e-01
-1.10210156e+00 4.20891821e-01 4.06395227e-01 1.03762555e+00
5.48006237e-01 1.07378699e-01 3.62903506e-01 -5.23594499e-01
6.02862597e-01 -2.37513348e-01 -3.32128614e-01 -7.27261417e-03
-3.69880348e-01 1.22284077e-01 7.05996752e-01 -6.48385167e-01
-1.25200856e+00 -3.09962183e-01 3.74662615e-02 -7.03543067e-01
1.86615661e-01 1.34337246e-01 7.58074000e-02 -2.64977872e-01
6.02459192e-01 3.24045688e-01 -1.50532395e-01 -4.82056826e-01
3.33493471e-01 2.73422062e-01 6.02356672e-01 -4.15691078e-01
7.90632129e-01 8.30516040e-01 4.49354909e-02 -9.07968402e-01
-6.96628213e-01 -2.62365639e-01 -5.58506310e-01 -3.29256266e-01
8.82711828e-01 -9.53625679e-01 -2.97840923e-01 3.73223633e-01
-1.13060260e+00 -1.66706100e-01 -3.01261127e-01 3.74440551e-01
-4.26602423e-01 4.47289288e-01 -7.16912806e-01 -6.43594027e-01
-4.61868167e-01 -1.07825506e+00 7.79433131e-01 1.63316011e-01
-8.75505880e-02 -7.31850207e-01 -5.74648790e-02 2.20348269e-01
1.01163638e+00 2.39137247e-01 5.11543751e-01 1.97039753e-01
-7.12484658e-01 1.92148626e-01 -4.46119159e-01 4.37269509e-01
3.20195675e-01 -1.08589813e-01 -9.69582200e-01 -3.63713741e-01
1.06746018e-01 2.67458022e-01 8.94192100e-01 6.72041357e-01
1.17671442e+00 -5.12373924e-01 8.11156780e-02 8.62495244e-01
1.54316723e+00 3.37605514e-02 1.13408089e+00 5.07771909e-01
6.52888179e-01 6.04838789e-01 6.86763704e-01 5.34728765e-01
2.04804659e-01 9.62670386e-01 4.71966594e-01 -2.25092128e-01
-5.37484109e-01 -8.71133506e-02 1.64115146e-01 4.81727093e-01
-8.09241295e-01 1.26929551e-01 -3.89522374e-01 6.06526494e-01
-1.76529706e+00 -1.40915692e+00 -2.55315453e-02 2.50769162e+00
1.20498455e+00 -4.10608575e-02 1.84475128e-02 2.22421661e-01
9.62271810e-01 4.73156631e-01 -3.54030520e-01 -5.83088510e-02
-2.90962577e-01 3.76205415e-01 5.92024624e-01 8.45713675e-01
-1.02149832e+00 6.44106030e-01 5.54242134e+00 1.05885684e+00
-1.42595541e+00 3.45828056e-01 7.14415848e-01 -2.22958416e-01
-2.67381817e-01 -1.42641857e-01 -4.98821586e-01 6.34670913e-01
6.92577302e-01 -6.41724989e-02 6.12809598e-01 2.79854774e-01
7.93793738e-01 4.97747622e-02 -7.05982447e-01 8.20504963e-01
-8.37314501e-02 -1.55490315e+00 4.28960584e-02 -2.83513635e-01
8.09435010e-01 -3.59273255e-01 1.44345164e-01 -2.16327548e-01
-3.83536786e-01 -8.89596164e-01 1.02452791e+00 5.68787515e-01
9.95835900e-01 -7.78382242e-01 5.21487772e-01 -7.35397935e-02
-1.43441725e+00 -6.59396946e-02 -3.10697377e-01 -4.96828258e-02
3.41863751e-01 7.64974833e-01 -2.96535969e-01 6.29900277e-01
8.22916746e-01 7.66752422e-01 -2.43680358e-01 9.80287075e-01
-4.74639684e-02 5.22544265e-01 3.01429480e-02 7.19822288e-01
-9.63840261e-02 -2.64550626e-01 7.47099936e-01 1.34212482e+00
3.80632728e-01 2.80927837e-01 -2.23814905e-01 7.11501062e-01
1.66523963e-01 -9.71372426e-02 -1.95069611e-01 3.54894131e-01
4.57793355e-01 1.04602218e+00 -4.02359188e-01 -2.41224304e-01
-4.11811829e-01 1.09668744e+00 3.31655261e-04 5.21326661e-01
-9.28918421e-01 -1.91018015e-01 8.39509606e-01 7.95315385e-01
4.29518282e-01 -3.15172911e-01 -3.80201399e-01 -9.93152559e-01
1.04492381e-01 -7.85327554e-01 5.62770432e-03 -7.61900902e-01
-1.02982152e+00 6.85926795e-01 -7.43140727e-02 -1.52916574e+00
-3.34006511e-02 1.26035541e-01 -3.91302556e-01 1.04562402e+00
-1.90185726e+00 -1.03837776e+00 -5.09594977e-01 6.06359303e-01
8.50228548e-01 3.46203834e-01 3.41042101e-01 7.26790369e-01
-2.64868945e-01 4.33102578e-01 -1.65541992e-01 -3.55899110e-02
1.00468016e+00 -6.78418159e-01 2.34591737e-01 1.23429883e+00
-2.61749655e-01 7.35234499e-01 1.02411711e+00 -6.89786494e-01
-1.12357187e+00 -1.10179508e+00 6.34215653e-01 3.73447947e-02
3.00411642e-01 1.73021480e-02 -1.27994311e+00 3.22363585e-01
4.94505376e-01 4.22907561e-01 1.12066798e-01 -5.11399806e-01
-4.50477540e-01 -3.50893050e-01 -1.33914292e+00 5.86016715e-01
9.92186487e-01 -4.49342221e-01 -2.93112665e-01 -2.59177893e-01
7.62284756e-01 -4.75982368e-01 -8.44390571e-01 5.46158791e-01
7.27267742e-01 -1.47642684e+00 1.40172434e+00 -2.10021973e-01
9.18629766e-01 -5.43021142e-01 -9.85937193e-02 -1.21925724e+00
-7.79384553e-01 -6.30440474e-01 -2.25804150e-01 1.15268052e+00
-1.81454062e-01 -4.40617144e-01 4.08159822e-01 3.63919675e-01
1.41115058e-02 -6.93370402e-01 -9.18216228e-01 -3.64024401e-01
-3.43671620e-01 -3.80191170e-02 4.33613360e-01 1.24397731e+00
-3.80439579e-01 -1.03028961e-01 -7.64202058e-01 2.54757464e-01
8.76657844e-01 -2.45808065e-01 3.38875949e-01 -7.30296075e-01
-1.29327968e-01 -3.98415834e-01 -2.34890684e-01 -1.02734697e+00
-2.15818614e-01 -1.61665872e-01 -1.46993890e-01 -1.24995542e+00
5.29350638e-02 -2.44767681e-01 -2.83699363e-01 2.95975357e-02
-3.35429221e-01 5.35650551e-01 1.51619852e-01 2.66848236e-01
-1.13108784e-01 4.68272895e-01 1.61737466e+00 1.93615288e-01
-2.97833860e-01 -3.22524726e-01 -5.54046273e-01 5.99370539e-01
7.25565016e-01 -9.05871168e-02 -3.59279931e-01 -6.02528274e-01
3.75326537e-02 2.95171469e-01 4.61305350e-01 -9.61919367e-01
1.08354248e-01 -3.49875182e-01 4.24212605e-01 -3.31214607e-01
3.86049539e-01 -7.78942764e-01 2.81245589e-01 2.43962616e-01
-4.72063780e-01 1.51118875e-01 1.66155189e-01 4.29285586e-01
-3.06441039e-01 -1.22349961e-02 1.12746227e+00 -1.40292365e-02
-6.96629763e-01 1.55508846e-01 -1.23827435e-01 -5.41854128e-02
7.58008957e-01 -5.00354648e-01 -1.21940300e-01 -3.77280474e-01
-3.75678003e-01 -2.89021581e-01 5.11883676e-01 4.12083268e-01
7.90430784e-01 -1.20212901e+00 -9.76997554e-01 2.89592266e-01
-4.94184196e-01 -1.37503937e-01 6.37312710e-01 8.67652178e-01
-4.62630361e-01 5.25153652e-02 -5.59316278e-01 -3.61284047e-01
-1.21085155e+00 4.01493162e-01 3.63817036e-01 -2.71612227e-01
-8.96388471e-01 7.50162303e-01 1.18832134e-01 5.18238366e-01
7.49945343e-02 -2.08355457e-01 -4.35930192e-01 7.25212842e-02
9.71427977e-01 5.18681347e-01 -8.44610110e-02 -6.94145322e-01
-5.60969748e-02 7.63905287e-01 3.42091136e-02 -1.63797751e-01
1.53335452e+00 -5.23217976e-01 -1.56551659e-01 1.68435037e-01
8.22101831e-01 2.07327932e-01 -1.72601342e+00 -2.89260745e-01
-4.53937858e-01 -1.14385962e+00 3.55446309e-01 -3.70342612e-01
-1.39059830e+00 7.24574387e-01 6.32833540e-01 1.30211115e-01
1.49640608e+00 -6.57628655e-01 7.88448453e-01 -4.14820850e-01
3.94547969e-01 -9.21790361e-01 8.56547207e-02 2.84892291e-01
1.13176715e+00 -8.69690716e-01 1.87652469e-01 -6.17303252e-01
-3.82671326e-01 1.16635776e+00 5.03594458e-01 -4.18027490e-01
3.44525725e-01 2.16589913e-01 -3.02270383e-01 3.41144919e-01
-5.87906659e-01 5.07261381e-02 3.81944865e-01 5.02630770e-01
7.26808667e-01 -2.16347963e-01 -5.25195897e-01 2.27974802e-01
4.78059351e-02 2.76275247e-01 5.49838662e-01 6.34682417e-01
-3.91165584e-01 -8.38266969e-01 -5.71918130e-01 1.49492323e-01
-5.92054307e-01 -3.03212225e-01 3.84483188e-01 4.70364392e-01
3.09545249e-01 7.58895874e-01 1.87698722e-01 -1.76645651e-01
2.66645312e-01 -3.86424929e-01 6.75749362e-01 -1.22561149e-01
-6.39973402e-01 1.56810626e-01 -9.73735973e-02 -7.77088761e-01
-8.90648186e-01 -4.05338019e-01 -9.89636421e-01 -3.93999189e-01
-1.93921104e-01 -1.29362822e-01 3.50774139e-01 7.07416594e-01
4.51822042e-01 8.69134009e-01 6.49203181e-01 -1.29602385e+00
-3.76851350e-01 -9.56908822e-01 -3.87391448e-01 6.92941308e-01
6.83315098e-01 -6.33394063e-01 -4.29734588e-01 4.50914860e-01] | [11.049919128417969, -1.8806318044662476] |
cf7003d9-6ff0-44ba-af55-508aeec1c8d9 | teamwork-is-not-always-good-an-empirical | 2305.16559 | null | https://arxiv.org/abs/2305.16559v1 | https://arxiv.org/pdf/2305.16559v1.pdf | Teamwork Is Not Always Good: An Empirical Study of Classifier Drift in Class-incremental Information Extraction | Class-incremental learning (CIL) aims to develop a learning system that can continually learn new classes from a data stream without forgetting previously learned classes. When learning classes incrementally, the classifier must be constantly updated to incorporate new classes, and the drift in decision boundary may lead to severe forgetting. This fundamental challenge, however, has not yet been studied extensively, especially in the setting where no samples from old classes are stored for rehearsal. In this paper, we take a closer look at how the drift in the classifier leads to forgetting, and accordingly, design four simple yet (super-) effective solutions to alleviate the classifier drift: an Individual Classifiers with Frozen Feature Extractor (ICE) framework where we individually train a classifier for each learning session, and its three variants ICE-PL, ICE-O, and ICE-PL&O which further take the logits of previously learned classes from old sessions or a constant logit of an Other class as a constraint to the learning of new classifiers. Extensive experiments and analysis on 6 class-incremental information extraction tasks demonstrate that our solutions, especially ICE-O, consistently show significant improvement over the previous state-of-the-art approaches with up to 44.7% absolute F-score gain, providing a strong baseline and insights for future research on class-incremental learning. | ['Lifu Huang', 'Minqian Liu'] | 2023-05-26 | null | null | null | null | ['class-incremental-learning', 'incremental-learning'] | ['computer-vision', 'methodology'] | [ 4.59419042e-01 -2.40872856e-02 -4.42573667e-01 -6.21186852e-01
-5.18576026e-01 -5.74388385e-01 3.07634443e-01 3.19996327e-01
-4.21044737e-01 1.08134520e+00 -3.84976298e-01 -1.95157304e-01
-1.30896643e-01 -6.44084513e-01 -8.23265195e-01 -8.43537867e-01
-3.09156954e-01 4.30035442e-01 4.84565020e-01 -9.28630028e-03
8.53908658e-02 3.56142640e-01 -2.06984019e+00 5.41490614e-01
8.98274839e-01 1.02791011e+00 -1.56204611e-01 6.66908920e-01
1.96918685e-04 7.84395278e-01 -6.68066800e-01 -2.41181105e-01
2.61126041e-01 -3.01162392e-01 -6.92729652e-01 1.18094206e-01
5.78135788e-01 -3.58438492e-01 -1.11741647e-01 6.79414928e-01
2.67873794e-01 1.77594572e-01 3.90377104e-01 -1.52686548e+00
-4.13730621e-01 9.39112902e-01 -4.78236824e-01 3.67933512e-01
8.13142285e-02 4.27542105e-02 5.37238538e-01 -1.03243434e+00
5.48269987e-01 8.50202918e-01 9.07085657e-01 9.11157370e-01
-1.04219496e+00 -1.07272780e+00 8.58682752e-01 5.18499076e-01
-1.25243998e+00 -5.40504515e-01 7.45709717e-01 -3.59961808e-01
8.31151485e-01 2.31099918e-01 7.76415765e-01 1.04290366e+00
3.12243670e-01 1.22188282e+00 9.46625948e-01 -5.04383743e-01
5.94966650e-01 4.93953317e-01 6.98142469e-01 4.57378805e-01
6.26259327e-01 -8.01742263e-03 -1.01029658e+00 8.07392746e-02
-6.29773065e-02 3.43082279e-01 -1.52103096e-01 -5.24753213e-01
-6.09275281e-01 4.71154779e-01 2.35506535e-01 1.00944042e-01
-1.73930034e-01 -2.03725725e-01 3.70700240e-01 6.96509421e-01
5.79400182e-01 1.90928861e-01 -1.04543471e+00 -2.16283038e-01
-9.41963136e-01 3.69052440e-01 8.02192032e-01 1.16662896e+00
1.02442265e+00 -3.37905735e-02 -1.56670392e-01 6.32336259e-01
-2.96651512e-01 1.71097055e-01 8.06937158e-01 -5.88507473e-01
3.52652669e-01 9.31534588e-01 -1.01452405e-02 -2.77388632e-01
-3.28183383e-01 -7.03610659e-01 -6.46291494e-01 2.57271249e-02
1.56839699e-01 -2.41676763e-01 -1.05903661e+00 1.71338797e+00
4.85329449e-01 5.50994992e-01 -9.10071842e-03 2.85942435e-01
4.15630728e-01 3.10794592e-01 -4.92818989e-02 -7.74995387e-01
8.47951472e-01 -8.58449936e-01 -7.40719676e-01 -5.38585007e-01
3.52947146e-01 -2.44009137e-01 9.93542731e-01 7.94466496e-01
-7.96868622e-01 -7.23776579e-01 -1.38545656e+00 3.97037357e-01
-4.96753454e-01 -1.97883546e-01 7.37767518e-01 6.76135659e-01
-6.52280211e-01 8.41694176e-01 -1.03216434e+00 -1.22268818e-01
6.88137472e-01 4.08188820e-01 -1.73148662e-01 -1.99830607e-01
-1.10295045e+00 5.47571659e-01 5.51509678e-01 1.03822760e-01
-1.02825129e+00 -1.08431792e+00 -6.10358655e-01 -3.67222242e-02
6.96121633e-01 -5.01807332e-01 1.59982491e+00 -9.93258655e-01
-1.37065721e+00 3.86537224e-01 -4.67707247e-01 -5.78683376e-01
6.67414665e-01 -6.34800971e-01 -3.95320594e-01 -3.46177548e-01
-1.01641543e-01 4.68304545e-01 1.23476517e+00 -1.19656634e+00
-1.30602312e+00 -4.61460352e-01 -2.23082110e-01 1.30056471e-01
-8.08453977e-01 -5.15770257e-01 -1.45125911e-01 -5.53880215e-01
1.97118282e-01 -9.22272325e-01 1.21528171e-01 -2.13492364e-01
1.31409973e-01 -4.68209952e-01 1.19044554e+00 -1.86478391e-01
1.82834888e+00 -2.09386086e+00 4.48849201e-02 -2.23085657e-01
1.52037457e-01 3.69890749e-01 8.32537860e-02 2.43243054e-02
-1.63379565e-01 -5.05818911e-02 -4.89338309e-01 -3.88310939e-01
-3.52366149e-01 4.54331905e-01 -4.82956111e-01 3.65430005e-02
3.06878239e-01 6.69803679e-01 -1.18713999e+00 -6.40757754e-02
-1.59856766e-01 2.10015625e-01 -6.46560729e-01 1.03724964e-01
-1.37870014e-01 -1.20682446e-02 3.06552798e-02 8.70695174e-01
7.65105426e-01 3.42351496e-02 1.05025321e-01 2.52881259e-01
2.07391270e-02 7.62806311e-02 -1.36049354e+00 1.44482815e+00
-2.24715725e-01 4.66567636e-01 -3.27674896e-01 -9.28279579e-01
7.80572176e-01 5.30090258e-02 4.70469415e-01 -2.77068973e-01
-2.67779708e-01 2.22161815e-01 -6.12708107e-02 -3.71001244e-01
4.48461384e-01 -1.66856959e-01 -7.54058212e-02 5.07499635e-01
2.96991855e-01 1.22145742e-01 3.13158035e-01 4.98349927e-02
1.31205261e+00 7.03886822e-02 1.92470670e-01 1.06904991e-01
3.68391126e-01 -3.07835340e-01 1.06258345e+00 1.08473098e+00
-5.17569721e-01 3.44186485e-01 1.68103546e-01 -8.18705678e-01
-4.95057791e-01 -1.11503196e+00 -3.07661712e-01 1.29497707e+00
-2.11493447e-01 -4.58782375e-01 -6.52467966e-01 -1.22616482e+00
4.81307089e-01 8.53194118e-01 -6.66238070e-01 -8.81629944e-01
-5.98694324e-01 -1.07263494e+00 4.01565880e-02 6.34666741e-01
5.16451240e-01 -1.00361502e+00 -8.09328258e-01 5.36100388e-01
2.88116112e-02 -7.26370752e-01 -3.04035962e-01 8.92180324e-01
-1.32564330e+00 -1.21576583e+00 -2.37135693e-01 -5.63057065e-01
6.68971956e-01 1.92875683e-01 8.91603351e-01 -1.01828493e-01
-3.76449823e-01 3.81139874e-01 -4.71965581e-01 -9.01416779e-01
-1.55999422e-01 6.38340712e-01 4.96565402e-01 2.26877853e-01
7.64021873e-01 -4.83661860e-01 -4.21761543e-01 1.48116440e-01
-8.88422728e-01 -2.41883397e-01 4.96831000e-01 1.03259635e+00
3.94753218e-01 4.41653818e-01 1.03419733e+00 -1.51173759e+00
4.71603602e-01 -5.64109266e-01 -2.37452745e-01 4.15685922e-01
-1.33124053e+00 2.92959195e-02 5.44812500e-01 -8.27373266e-01
-1.20381200e+00 2.90731102e-01 1.74155489e-01 -2.90586770e-01
5.21773919e-02 3.51706177e-01 -1.42491892e-01 2.10355565e-01
6.97651803e-01 2.98017353e-01 -3.90001267e-01 -4.29302335e-01
2.88612723e-01 9.36309516e-01 6.52470648e-01 -5.48589468e-01
7.66730011e-01 2.46475369e-01 -4.70419556e-01 -6.03580654e-01
-1.28929937e+00 -4.49130028e-01 -1.18778574e+00 -1.77703857e-01
-4.86731678e-02 -9.54257429e-01 -3.13294977e-01 8.98494780e-01
-6.26862228e-01 -4.56727028e-01 -7.67238915e-01 1.09257020e-01
-2.67658234e-01 -2.16628239e-02 -3.56590569e-01 -9.61098850e-01
-3.32352668e-01 -6.68085635e-01 6.36833906e-01 4.76762623e-01
-2.25008368e-01 -6.62201107e-01 -3.17863678e-03 3.78030688e-02
4.18238014e-01 8.81592110e-02 9.17251647e-01 -6.01944387e-01
-4.46268767e-01 -4.22812194e-01 3.60063553e-01 5.35924077e-01
4.30983931e-01 -9.55135971e-02 -1.29661298e+00 -8.40511441e-01
6.82086945e-02 -2.17359155e-01 1.28698313e+00 8.82487372e-03
1.16237330e+00 -3.88733447e-01 -4.88600731e-01 5.46782672e-01
1.11094797e+00 6.19250894e-01 2.28667870e-01 4.53800648e-01
4.60539520e-01 1.82638720e-01 6.89915538e-01 7.89997339e-01
4.52594787e-01 3.08403552e-01 1.58851907e-01 5.11146784e-01
-2.88865209e-01 -3.46764445e-01 5.37379920e-01 9.02620494e-01
2.30311900e-01 4.45733145e-02 -7.70797491e-01 5.94289124e-01
-1.93307960e+00 -7.89700389e-01 3.13762784e-01 2.60273957e+00
1.36678016e+00 6.31119490e-01 -1.16521090e-01 7.65554070e-01
4.88386303e-01 -2.83730686e-01 -1.30263388e+00 -1.47716701e-01
8.94543231e-02 4.99552548e-01 2.32045576e-01 4.53325629e-01
-1.28058040e+00 8.88889313e-01 6.67937565e+00 2.99387306e-01
-1.06464386e+00 1.64216366e-02 4.58425641e-01 -4.68176991e-01
1.75992087e-01 1.55418083e-01 -1.46483779e+00 5.30895829e-01
1.13571703e+00 -3.61678034e-01 4.01409119e-01 1.00092840e+00
-4.35153723e-01 -2.39145100e-01 -1.59050107e+00 8.61008406e-01
2.23101556e-01 -8.95355821e-01 1.30335435e-01 -5.41013360e-01
8.67498577e-01 -2.35046580e-01 1.02946527e-01 9.24796164e-01
2.77580678e-01 -3.01796079e-01 8.29072773e-01 6.64631724e-01
6.64596498e-01 -8.58611286e-01 7.21012414e-01 6.62928641e-01
-1.12376142e+00 -6.60038710e-01 -2.43613377e-01 -2.78949738e-01
-3.67537946e-01 8.01862955e-01 -8.44144106e-01 4.76281255e-01
1.16738784e+00 9.37966645e-01 -1.06623900e+00 1.09093738e+00
-2.65030712e-01 1.01909399e+00 -2.73725301e-01 1.87124595e-01
-3.40645105e-01 4.74341005e-01 3.65538448e-01 1.06039667e+00
1.01246156e-01 2.76001934e-02 1.54741928e-01 1.97694227e-01
-1.20224357e-01 -3.43637615e-01 -3.16563487e-01 3.32658052e-01
7.91680634e-01 9.04404402e-01 -5.78948915e-01 -5.54241598e-01
-1.69367254e-01 1.02644992e+00 5.35552859e-01 3.05168658e-01
-4.60557848e-01 -7.22353697e-01 6.20349348e-01 2.15061039e-01
3.92711431e-01 4.30495292e-02 -2.50784993e-01 -1.39892316e+00
2.74316937e-01 -7.98334181e-01 7.90006220e-01 -2.29538381e-01
-1.38793051e+00 4.22390729e-01 1.39017805e-01 -1.14826715e+00
-3.34324330e-01 -3.39190900e-01 -3.17053944e-01 2.52660990e-01
-1.68662548e+00 -7.18647659e-01 -5.04340291e-01 4.30452049e-01
8.41313362e-01 -2.97828853e-01 6.43308640e-01 3.86285782e-01
-8.38520944e-01 1.09644639e+00 1.95509195e-01 -2.46731132e-01
1.07625008e+00 -1.23290014e+00 1.47635192e-01 7.77440190e-01
6.66477233e-02 5.90791821e-01 5.89341998e-01 -6.71645880e-01
-1.40395224e+00 -1.27765286e+00 9.54918146e-01 -8.03381681e-01
2.79433399e-01 -6.60106659e-01 -1.36492145e+00 9.16074157e-01
-2.23296747e-01 1.77038163e-01 7.03861833e-01 2.66779512e-01
-4.29273129e-01 -8.41240764e-01 -1.06364107e+00 1.97858468e-01
1.19111633e+00 -2.50196666e-01 -6.76276207e-01 4.15400192e-02
7.86352694e-01 -4.53504413e-01 -5.31083226e-01 6.19017005e-01
7.39732921e-01 -9.00711238e-01 6.15986943e-01 -7.60268331e-01
-8.02049637e-02 -2.44382188e-01 1.07138932e-01 -1.35650480e+00
-2.47565687e-01 -7.09819734e-01 -7.32842386e-01 1.38840795e+00
3.89388651e-01 -6.88817859e-01 9.17369008e-01 6.94910228e-01
-1.44074887e-01 -7.60213971e-01 -9.22944963e-01 -9.87759829e-01
1.50787741e-01 -4.93117332e-01 7.24082649e-01 1.09464347e+00
-2.68745609e-02 3.78092498e-01 -2.50230759e-01 -2.87994370e-02
6.14193678e-01 4.60473672e-02 7.11530745e-01 -1.69848001e+00
-1.23914341e-02 -1.98551595e-01 -1.45306662e-01 -8.24141920e-01
2.16948371e-02 -9.15347040e-01 1.07365251e-01 -9.28617477e-01
3.27591419e-01 -6.47415042e-01 -8.44983637e-01 1.01069403e+00
-6.04147375e-01 -2.31090099e-01 8.19000006e-02 2.89855152e-01
-8.08306873e-01 6.22431576e-01 6.76164389e-01 -2.63239026e-01
-7.47635484e-01 4.93369907e-01 -8.86422694e-01 5.02938509e-01
5.03771484e-01 -8.63318324e-01 -6.16422653e-01 -1.36866897e-01
2.02631623e-01 -4.56315666e-01 -1.98149279e-01 -1.25466299e+00
5.55101037e-01 -7.29249865e-02 7.83293366e-01 -6.72706842e-01
-1.11973375e-01 -9.27352726e-01 -1.85569841e-02 6.43595099e-01
-4.41401839e-01 -1.08891234e-01 3.79845202e-01 8.24128926e-01
-1.11069359e-01 -3.48590702e-01 6.78003907e-01 1.65539347e-02
-7.22136080e-01 4.46183681e-01 -3.16705644e-01 2.50399709e-02
1.29554415e+00 -2.45472729e-01 -2.74385154e-01 6.55843467e-02
-9.74585414e-01 3.93338829e-01 8.57049301e-02 8.73650730e-01
6.43189609e-01 -1.32684839e+00 -5.71120501e-01 6.38698995e-01
2.86634296e-01 3.00766498e-01 9.31311548e-02 2.31897056e-01
1.13373816e-01 1.38769567e-01 -7.07266033e-02 -7.22282767e-01
-1.14411414e+00 6.63441777e-01 1.10974222e-01 -2.61821508e-01
-5.09574473e-01 9.90987003e-01 -2.89981544e-01 -4.18234318e-01
7.59217083e-01 -6.10410690e-01 -3.37444954e-02 5.71874917e-01
7.35645354e-01 4.67730969e-01 5.24814427e-01 1.86928526e-01
-4.36140537e-01 2.71429330e-01 -7.69078791e-01 2.87697673e-01
1.51555789e+00 -1.11567490e-01 1.45553932e-01 1.12099802e+00
7.46764719e-01 -4.80030149e-01 -1.73910153e+00 -5.82748353e-01
4.42840874e-01 -3.37054670e-01 -1.08116008e-01 -1.09002912e+00
-8.79134417e-01 7.70880580e-01 9.59276378e-01 -6.08046353e-02
1.38244510e+00 -2.45720938e-01 6.13776982e-01 6.72556818e-01
6.27491117e-01 -1.27796936e+00 2.14741185e-01 6.08826578e-01
6.48947477e-01 -1.23507774e+00 2.04763696e-01 -3.26292217e-02
-3.96549553e-01 1.19774413e+00 1.00612986e+00 1.59565479e-01
1.03829491e+00 3.55408758e-01 -1.65859759e-01 3.86755258e-01
-1.39215815e+00 1.14485413e-01 1.13804020e-01 5.35688460e-01
2.35918820e-01 3.38083308e-04 8.21335465e-02 9.58237886e-01
-1.74943164e-01 4.06587005e-01 4.92889643e-01 1.46685016e+00
-6.03299439e-01 -1.16245377e+00 -9.55558196e-02 6.19460881e-01
2.10190448e-03 1.68165743e-01 -3.84725422e-01 6.56659901e-01
5.28286755e-01 7.42558777e-01 6.79182261e-02 -5.73661208e-01
7.56541133e-01 7.66602159e-01 5.21643281e-01 -8.36538792e-01
-6.68918669e-01 -3.60557854e-01 -4.87033248e-01 -3.03478926e-01
-1.76980451e-01 -1.15451884e+00 -1.09221590e+00 -2.88401209e-02
-4.80602622e-01 7.16188550e-02 4.40510154e-01 6.78575993e-01
3.85672301e-01 6.88659847e-01 8.85588408e-01 -5.21172047e-01
-7.66328335e-01 -1.04942000e+00 -5.01672447e-01 1.75186872e-01
7.42803454e-01 -8.54244530e-01 -6.91808641e-01 2.63956070e-01] | [9.86865520477295, 3.393603563308716] |
d1cabad5-36af-4e94-8751-3d5610f263ed | effective-distant-supervision-for-temporal | 2010.12755 | null | https://arxiv.org/abs/2010.12755v2 | https://arxiv.org/pdf/2010.12755v2.pdf | Effective Distant Supervision for Temporal Relation Extraction | A principal barrier to training temporal relation extraction models in new domains is the lack of varied, high quality examples and the challenge of collecting more. We present a method of automatically collecting distantly-supervised examples of temporal relations. We scrape and automatically label event pairs where the temporal relations are made explicit in text, then mask out those explicit cues, forcing a model trained on this data to learn other signals. We demonstrate that a pre-trained Transformer model is able to transfer from the weakly labeled examples to human-annotated benchmarks in both zero-shot and few-shot settings, and that the masking scheme is important in improving generalization. | ['Greg Durrett', 'Shih-ting Lin', 'Xinyu Zhao'] | 2020-10-24 | null | https://aclanthology.org/2021.adaptnlp-1.20 | https://aclanthology.org/2021.adaptnlp-1.20.pdf | eacl-adaptnlp-2021-4 | ['temporal-relation-extraction'] | ['natural-language-processing'] | [ 2.82614172e-01 3.02453697e-01 -4.96278793e-01 -7.07494497e-01
-9.54566956e-01 -7.47142196e-01 8.91232133e-01 2.44588435e-01
-6.14659905e-01 9.29167211e-01 5.05064309e-01 -9.53614488e-02
-1.56065404e-01 -5.22183776e-01 -5.56093693e-01 -3.35883468e-01
-7.12251186e-01 7.78923035e-01 6.00886583e-01 -3.46243739e-01
-3.33911777e-01 2.93221295e-01 -1.32331181e+00 5.23795009e-01
3.65516901e-01 8.11163306e-01 -2.27700651e-01 5.96879482e-01
2.01861799e-01 1.22224379e+00 -4.33812320e-01 -4.86695051e-01
2.19649464e-01 -5.71095586e-01 -1.14197600e+00 -1.27985030e-01
-6.48609921e-02 -1.93145931e-01 -6.04941487e-01 4.25741524e-01
2.54335493e-01 5.10256767e-01 7.92681575e-01 -1.12785447e+00
-7.87422955e-01 8.87894869e-01 -3.37555259e-01 1.00905669e+00
4.08753604e-01 1.64883047e-01 1.41964746e+00 -7.39697814e-01
1.13976824e+00 1.05214524e+00 7.76552081e-01 5.47193646e-01
-1.53732443e+00 -5.86525381e-01 3.40993762e-01 4.34456587e-01
-1.16214573e+00 -6.32625878e-01 7.60147691e-01 -5.44015169e-01
1.52761555e+00 2.54552886e-02 4.31069672e-01 1.64565706e+00
-2.33691692e-01 5.54482937e-01 8.84047389e-01 -4.26635891e-01
2.18260571e-01 -6.75309226e-02 5.35406709e-01 2.23735437e-01
-2.45384455e-01 5.89619040e-01 -9.71772194e-01 -3.88516933e-02
5.37889898e-01 -1.84113204e-01 -1.74153209e-01 -1.83298737e-01
-1.32109940e+00 4.03899372e-01 3.70884359e-01 5.73685646e-01
-1.63144037e-01 7.19538778e-02 6.34922564e-01 5.83902240e-01
6.98565245e-01 7.09283531e-01 -8.45559239e-01 -3.68266106e-01
-8.36978197e-01 1.15188837e-01 6.48447454e-01 1.04166937e+00
5.58416605e-01 -3.62313092e-01 -3.76188606e-01 7.76804686e-01
-3.90642136e-01 -1.17067687e-01 5.14781177e-01 -8.62912476e-01
5.67291558e-01 3.57245266e-01 1.88444585e-01 -4.74232167e-01
-5.30591130e-01 -1.23354360e-01 -1.93338290e-01 -2.47224420e-01
5.64910531e-01 -1.62019044e-01 -9.99687374e-01 2.04391623e+00
1.22497141e-01 2.12834358e-01 2.33259916e-01 6.81681514e-01
7.40914702e-01 5.10897219e-01 2.81383812e-01 -5.18256426e-01
1.15903509e+00 -9.49655950e-01 -8.48302841e-01 -6.60173714e-01
7.49426484e-01 -3.80841494e-01 1.07306933e+00 7.64291035e-03
-9.42717314e-01 -3.77572298e-01 -9.63303268e-01 -3.12010556e-01
-4.57974732e-01 -1.46491170e-01 8.85076642e-01 -3.02943379e-01
-6.29784346e-01 9.15964425e-01 -1.10159457e+00 -6.44047499e-01
4.87282217e-01 1.36346936e-01 -5.16071916e-01 1.48051381e-01
-1.61539125e+00 1.32611370e+00 5.12990594e-01 -2.25693107e-01
-1.07437074e+00 -7.01448381e-01 -1.01473451e+00 -9.83709935e-03
5.34931123e-01 -1.93052381e-01 1.65556490e+00 -8.06393027e-01
-8.87678981e-01 1.05384505e+00 -3.68969738e-01 -5.55136740e-01
3.34580243e-01 -1.00751542e-01 -7.56576657e-01 1.92134529e-01
3.40862930e-01 5.16582727e-01 5.08615494e-01 -9.72314477e-01
-6.75045133e-01 -1.03018552e-01 -5.28816134e-03 3.30211557e-02
-3.42043936e-01 3.18201423e-01 -2.67860532e-01 -7.71349072e-01
-5.73276281e-02 -7.15705991e-01 -1.35290891e-01 -1.95471972e-01
-1.60708487e-01 -5.84467411e-01 5.81399441e-01 -4.96807039e-01
1.01291203e+00 -2.26224518e+00 -1.45064630e-02 -5.47005832e-02
-2.12385021e-02 -2.51449734e-01 -1.08650222e-01 5.00628948e-01
-3.92824113e-01 -6.65674657e-02 -1.90736026e-01 -2.93039799e-01
-2.03198582e-01 5.72676301e-01 -5.19963145e-01 2.16060340e-01
6.74192309e-01 9.93502378e-01 -1.45976722e+00 -5.09758353e-01
-2.13338792e-01 9.39224809e-02 -2.00200826e-01 2.93714851e-01
-4.09928858e-01 6.22689724e-01 -1.88213438e-01 4.11886334e-01
-5.47448322e-02 -4.21293676e-01 1.80061623e-01 -1.16711557e-01
2.12798208e-01 1.25167525e+00 -8.21863949e-01 1.66890335e+00
-3.49525750e-01 8.65511894e-01 -6.19240582e-01 -8.82117152e-01
6.26627922e-01 7.24887311e-01 5.24585962e-01 -6.29643917e-01
6.12759665e-02 3.02513316e-03 1.93569899e-01 -7.18708158e-01
2.31373325e-01 -5.22703946e-01 -2.06834942e-01 6.65889621e-01
6.41108692e-01 -1.22710161e-01 4.94301915e-01 2.92205751e-01
1.44035017e+00 3.71961534e-01 3.83034110e-01 3.88755426e-02
-2.01532304e-01 2.78607935e-01 8.83641958e-01 4.39929396e-01
-2.98472166e-01 4.85448390e-01 6.44446135e-01 -5.65172553e-01
-1.05765820e+00 -1.01182425e+00 -1.94998965e-01 1.35856247e+00
-1.04988202e-01 -7.66435266e-01 9.76333246e-02 -1.02401876e+00
-1.86480641e-01 8.97407353e-01 -9.13390458e-01 -2.01584309e-01
-6.76538110e-01 -5.28325737e-01 5.97599268e-01 8.72207940e-01
1.02186631e-02 -1.16459608e+00 -4.89488572e-01 3.56979996e-01
-3.95909905e-01 -1.40593493e+00 -3.61111760e-01 9.97099221e-01
-7.79705286e-01 -9.50156331e-01 -1.98788807e-01 -8.47992182e-01
3.94797742e-01 -1.55985445e-01 1.46783996e+00 -3.68948698e-01
-1.60898507e-01 -6.19616956e-02 -4.40073967e-01 -3.63423586e-01
-2.56551296e-01 -2.65784469e-02 -4.35237922e-02 -3.73886704e-01
8.71831834e-01 -1.02297163e+00 -9.65832397e-02 3.04801941e-01
-4.09786254e-01 -1.34082600e-01 2.22781077e-01 1.04096031e+00
3.11578602e-01 6.72295168e-02 6.72351182e-01 -9.52621937e-01
4.36211377e-01 -5.03653049e-01 -3.07356179e-01 1.30609363e-01
-5.19613266e-01 2.00000942e-01 3.25395525e-01 -9.22962844e-01
-1.03261399e+00 2.00288802e-01 2.07389563e-01 -3.64835322e-01
-2.00691223e-01 4.78183359e-01 1.97958991e-01 5.40336311e-01
1.29235411e+00 -2.34691143e-01 -4.17639583e-01 -2.79320061e-01
6.19817793e-01 1.18019491e-01 7.85295486e-01 -8.28967512e-01
7.56096125e-01 4.76665139e-01 -2.68830478e-01 -3.81864905e-01
-1.50920582e+00 -4.56812739e-01 -9.89485025e-01 8.70606154e-02
8.45846534e-01 -9.60631311e-01 -1.17354967e-01 -3.08889989e-02
-1.25295269e+00 -7.94830739e-01 -8.16069245e-01 6.30400538e-01
-6.05507851e-01 -1.36259824e-01 -9.49156523e-01 -6.96938157e-01
1.49682343e-01 -5.23032069e-01 8.65206003e-01 -8.57819095e-02
-9.59237516e-01 -8.79523039e-01 3.46493900e-01 -3.11931595e-02
7.21102580e-04 2.29415864e-01 8.71247947e-01 -1.01579809e+00
-3.79989594e-01 -1.45038784e-01 6.57870397e-02 -2.09759343e-02
2.99994826e-01 -1.69561341e-01 -1.36089683e+00 2.30164438e-01
-7.83660784e-02 -7.95398116e-01 8.35153043e-01 1.22721577e-02
6.86878204e-01 -2.80616581e-01 -7.02928126e-01 2.91089088e-01
8.08739960e-01 2.64313817e-01 1.52425379e-01 2.74844587e-01
3.93227428e-01 8.54475141e-01 6.60156727e-01 1.13121271e-01
3.97660673e-01 6.54448628e-01 -1.57186106e-01 -3.37210223e-02
-3.26703489e-02 -3.55293334e-01 1.82021588e-01 4.98456061e-01
-5.91446832e-03 -1.22593425e-03 -1.06304955e+00 1.20223844e+00
-1.98751271e+00 -1.29001379e+00 1.43376201e-01 1.63472652e+00
1.69168019e+00 8.07728946e-01 1.51718229e-01 2.44935617e-01
5.06922722e-01 1.66737676e-01 -4.28075075e-01 -9.79066342e-02
-2.03498825e-01 4.70639110e-01 3.14366341e-01 5.10550022e-01
-1.28920507e+00 1.17265570e+00 7.58978081e+00 3.25385153e-01
-6.84740365e-01 2.43261397e-01 4.43788499e-01 -4.07521188e-01
-1.26797348e-01 2.40670919e-01 -6.62792444e-01 1.91303819e-01
1.26159465e+00 -3.04409742e-01 4.09929693e-01 6.83812976e-01
6.80295005e-02 9.22945961e-02 -2.05826569e+00 7.23623753e-01
-1.99463084e-01 -1.18259323e+00 -4.88161385e-01 -3.03645492e-01
6.23168647e-01 1.23316228e-01 -2.92982489e-01 5.46072245e-01
8.37287903e-01 -1.12685370e+00 7.52217174e-01 2.13774770e-01
5.56594729e-01 -2.69900680e-01 4.96142507e-01 4.08446014e-01
-1.11223304e+00 -2.73443796e-02 -4.28593904e-02 -4.35423315e-01
4.08922374e-01 6.29783213e-01 -9.87473369e-01 3.43338132e-01
8.12205911e-01 1.16363132e+00 -5.39113343e-01 6.56324625e-01
-6.37313485e-01 8.21730256e-01 -5.73757410e-01 8.85756612e-02
2.71823332e-02 3.56013417e-01 4.70082641e-01 1.30599952e+00
6.67282492e-02 5.14367819e-01 6.36425763e-02 9.24273252e-01
-1.90281421e-02 -3.24495375e-01 -9.31313753e-01 -2.81483650e-01
5.93817413e-01 1.11637461e+00 -6.34286761e-01 -4.04978573e-01
-5.12474298e-01 8.78056169e-01 6.92135513e-01 4.93591219e-01
-7.06566095e-01 -8.76170695e-02 5.22596180e-01 -1.38998311e-03
3.96456122e-01 -2.33522832e-01 -2.94532776e-01 -1.26078689e+00
9.92470980e-02 -6.65560842e-01 9.73960996e-01 -9.49684501e-01
-1.65861130e+00 6.90930009e-01 1.96927845e-01 -1.01057363e+00
-6.04860663e-01 -3.64878923e-01 -6.40397787e-01 8.73718202e-01
-1.22773206e+00 -1.02824473e+00 -1.06933579e-01 4.63601232e-01
5.30317008e-01 4.54672687e-02 8.56425822e-01 3.89280200e-01
-3.16398680e-01 4.04267907e-01 -6.53560758e-01 4.28650320e-01
1.09968734e+00 -1.32961702e+00 4.66516703e-01 9.48616683e-01
5.84214985e-01 7.63260543e-01 9.66566682e-01 -5.20922422e-01
-6.63272917e-01 -9.59399462e-01 1.33470416e+00 -1.02418351e+00
1.15622592e+00 -6.82493329e-01 -1.11034966e+00 1.44291127e+00
1.50947094e-01 3.27241123e-01 7.89814591e-01 7.31145978e-01
-8.22224736e-01 -2.93375105e-02 -8.22451770e-01 6.31537497e-01
1.50565076e+00 -1.07018876e+00 -1.37586522e+00 4.84121740e-01
1.05429769e+00 -4.20865715e-01 -8.55025291e-01 4.97765630e-01
9.72450078e-02 -4.56134379e-01 8.43291283e-01 -1.12139344e+00
5.55251598e-01 -8.30429345e-02 -6.33817492e-03 -1.35896087e+00
-2.85250515e-01 -8.14879358e-01 -3.19764018e-01 1.47068954e+00
9.41891253e-01 -2.44046912e-01 5.54634809e-01 7.73143768e-01
2.48766243e-02 -4.53452170e-01 -8.75464559e-01 -9.61128235e-01
1.44922361e-03 -6.22020721e-01 2.87551761e-01 1.31960642e+00
6.61158919e-01 9.61422801e-01 -3.38311970e-01 1.48719028e-01
1.74104616e-01 -2.49756016e-02 4.06047106e-01 -1.13542807e+00
-6.64578140e-01 -2.48211414e-01 -3.19277465e-01 -8.52096021e-01
3.21647167e-01 -9.38202560e-01 1.72126383e-01 -1.36046422e+00
1.48590356e-01 -3.95418108e-01 -4.76096451e-01 8.87020826e-01
-2.60925055e-01 4.68279459e-02 -3.26910406e-01 3.16787332e-01
-6.82653844e-01 4.14237708e-01 1.05272961e+00 -1.52533084e-01
-2.55886078e-01 -1.28936604e-01 -5.22199631e-01 6.04831934e-01
4.15161461e-01 -7.97167897e-01 -7.32807934e-01 -3.78059953e-01
1.95640028e-01 3.81430835e-02 3.47064108e-01 -6.26331449e-01
2.56514370e-01 -2.74222732e-01 3.05543721e-01 -5.31398177e-01
4.92965281e-01 -8.23129535e-01 -6.18529841e-02 -6.17073327e-02
-1.00506878e+00 -7.88665861e-02 3.32133591e-01 5.69605827e-01
-3.38742852e-01 -8.74772891e-02 6.58418775e-01 -1.92507997e-01
-8.29262078e-01 9.41673890e-02 -1.05458826e-01 6.50793791e-01
9.03601289e-01 2.10666180e-01 -4.02947217e-01 -4.71279442e-01
-1.22405958e+00 3.50369602e-01 2.31290311e-01 4.63861406e-01
1.59967631e-01 -1.42710555e+00 -4.66926306e-01 -1.12468354e-01
4.93540972e-01 6.43734485e-02 -3.01839471e-01 7.71962047e-01
2.25687355e-01 2.48923600e-01 4.96930778e-02 -4.95132953e-01
-8.89048755e-01 9.27380383e-01 3.88792783e-01 -4.54513729e-01
-9.91747022e-01 1.04310322e+00 7.43174478e-02 -2.51230001e-01
4.61954743e-01 -5.79303801e-01 -4.60978560e-02 1.90246284e-01
3.99769008e-01 -1.57231346e-01 5.09123914e-02 -1.73045948e-01
-5.25512934e-01 7.59379789e-02 -1.40880048e-01 -6.55755997e-01
1.71229780e+00 1.10188663e-01 8.63908827e-02 1.05681849e+00
1.03429890e+00 -1.63906842e-01 -1.36317146e+00 -6.77939534e-01
7.38103390e-01 -1.40425548e-01 -3.31653982e-01 -8.49486113e-01
-4.32133138e-01 5.60396433e-01 9.72595215e-02 3.78619313e-01
9.01761174e-01 6.15119815e-01 6.28482103e-01 5.23313403e-01
3.28009665e-01 -1.09821844e+00 3.16778064e-01 7.81348228e-01
6.69622421e-01 -1.28036940e+00 -1.60527632e-01 -4.00621057e-01
-8.68112564e-01 7.39613354e-01 5.78556538e-01 -3.95604633e-02
7.60216475e-01 6.98327184e-01 3.89993265e-02 -4.58243281e-01
-1.33522034e+00 -4.85423446e-01 4.14153010e-01 6.84464693e-01
6.26918793e-01 -2.97277421e-01 -1.87312439e-01 8.55742276e-01
-2.36083269e-01 1.20360911e-01 2.46695280e-01 1.06837475e+00
-7.88971409e-03 -1.05626464e+00 2.06875503e-01 3.35869372e-01
-3.77097636e-01 -2.79944152e-01 -6.08157933e-01 7.78622448e-01
2.32994109e-01 8.53247523e-01 9.94207859e-02 -2.61661947e-01
4.60641026e-01 5.99082112e-01 4.67980534e-01 -9.51373100e-01
-4.57298875e-01 1.02182895e-01 7.65622139e-01 -5.67089617e-01
-6.65654480e-01 -8.55126917e-01 -1.28641403e+00 2.96387792e-01
-9.50649679e-02 3.49684238e-01 -9.46709365e-02 1.18811464e+00
5.01443408e-02 7.33964682e-01 4.00317401e-01 -5.84856510e-01
-3.48854363e-01 -1.13332283e+00 -2.89465129e-01 9.27600682e-01
5.55042922e-01 -9.19995546e-01 -5.47466278e-01 4.57042485e-01] | [9.18089771270752, 9.128061294555664] |
d161c846-cb92-4417-a290-c4ab18b7ab01 | robust-graph-structure-learning-with-the | 2307.02126 | null | https://arxiv.org/abs/2307.02126v1 | https://arxiv.org/pdf/2307.02126v1.pdf | Robust Graph Structure Learning with the Alignment of Features and Adjacency Matrix | To improve the robustness of graph neural networks (GNN), graph structure learning (GSL) has attracted great interest due to the pervasiveness of noise in graph data. Many approaches have been proposed for GSL to jointly learn a clean graph structure and corresponding representations. To extend the previous work, this paper proposes a novel regularized GSL approach, particularly with an alignment of feature information and graph information, which is motivated mainly by our derived lower bound of node-level Rademacher complexity for GNNs. Additionally, our proposed approach incorporates sparse dimensional reduction to leverage low-dimensional node features that are relevant to the graph structure. To evaluate the effectiveness of our approach, we conduct experiments on real-world graphs. The results demonstrate that our proposed GSL method outperforms several competitive baselines, especially in scenarios where the graph structures are heavily affected by noise. Overall, our research highlights the importance of integrating feature and graph information alignment in GSL, as inspired by our derived theoretical result, and showcases the superiority of our approach in handling noisy graph structures through comprehensive experiments on real-world datasets. | ['Ming Li', 'Linsen Wei', 'Shiyu Liu', 'Gang Wen', 'Shaogao Lv'] | 2023-07-05 | null | null | null | null | ['graph-structure-learning'] | ['graphs'] | [ 2.45762527e-01 3.39519948e-01 -6.59661144e-02 -1.10778771e-01
-5.51760674e-01 -4.98625189e-01 5.44299245e-01 3.24061900e-01
-9.69848111e-02 5.01859426e-01 2.03760847e-01 -1.45827264e-01
-5.01224101e-01 -9.76296782e-01 -8.27830851e-01 -7.22249329e-01
-4.83153015e-01 -9.14495364e-02 -3.62022384e-03 -3.24237078e-01
5.11891730e-02 5.29296815e-01 -1.06040144e+00 -3.55702847e-01
9.18172836e-01 6.63944066e-01 2.02803850e-01 4.35074598e-01
-9.20195505e-02 8.84064496e-01 -5.25150836e-01 -5.54515123e-01
4.65659201e-01 -2.87790388e-01 -5.99994838e-01 2.25714430e-01
6.14845634e-01 8.58212113e-02 -8.54128838e-01 1.50091946e+00
4.80891109e-01 2.09802076e-01 4.15530980e-01 -1.33896291e+00
-8.09834957e-01 1.02526295e+00 -6.38863504e-01 2.10740864e-01
1.41179189e-01 5.00585213e-02 1.45694888e+00 -6.15017891e-01
6.91324592e-01 1.38112581e+00 9.58430231e-01 2.22235158e-01
-1.33043480e+00 -5.98958194e-01 8.38194847e-01 1.35210827e-01
-1.41514659e+00 -2.99571246e-01 1.37833929e+00 -9.02502388e-02
7.74378955e-01 5.30197024e-02 4.74564135e-01 1.34476316e+00
4.47129533e-02 7.40751922e-01 7.00898051e-01 -3.35875839e-01
1.99612319e-01 -3.78821820e-01 2.80919433e-01 9.76212502e-01
1.12973046e+00 9.32794809e-02 -2.52004325e-01 -1.07197545e-01
7.74236023e-01 -3.16664800e-02 -2.72729456e-01 -6.75297201e-01
-8.08171153e-01 8.46259534e-01 9.69364047e-01 3.97393107e-01
-3.60087633e-01 3.47181857e-01 4.73243237e-01 3.36417675e-01
4.72474247e-01 3.61117274e-01 -5.12253456e-02 3.25441301e-01
-5.98858416e-01 -4.07140851e-02 8.69310737e-01 1.14008093e+00
6.63211107e-01 3.81225944e-01 -1.00867368e-01 7.59161890e-01
3.34204793e-01 3.52604181e-01 2.10020214e-01 -4.39389676e-01
7.68482447e-01 9.74946439e-01 -6.02796137e-01 -1.77217829e+00
-5.03494084e-01 -9.06390309e-01 -1.60943234e+00 -4.35658783e-01
1.18651293e-01 -6.71048462e-02 -8.80172193e-01 2.07966900e+00
1.31103441e-01 6.48003578e-01 1.20297268e-01 6.03121161e-01
1.21987927e+00 2.34262586e-01 -8.30744728e-02 -3.12471092e-02
7.91339338e-01 -8.24731767e-01 -6.75210118e-01 -3.94702464e-01
7.98193991e-01 -1.50942981e-01 9.40767527e-01 1.46885917e-01
-6.27976954e-01 -3.29677820e-01 -1.05177784e+00 1.26631990e-01
-2.14066416e-01 -5.84188662e-02 8.46191406e-01 7.30759740e-01
-1.29376411e+00 7.72923172e-01 -8.26924920e-01 -5.44260859e-01
5.41152358e-01 2.67022341e-01 -5.97091615e-01 -3.59117538e-01
-1.03276801e+00 3.04531962e-01 6.23346567e-01 4.39491957e-01
-7.06421196e-01 -2.53752172e-01 -1.16576815e+00 3.92339170e-01
8.83008599e-01 -6.45967722e-01 6.00836515e-01 -5.51472425e-01
-1.05521548e+00 3.82920742e-01 1.53778628e-01 -5.84995687e-01
1.29645944e-01 3.15487795e-02 -2.81488717e-01 2.55515516e-01
4.14431058e-02 -9.21425521e-02 7.27743149e-01 -1.35622168e+00
2.26387382e-02 -4.85555530e-01 2.10078627e-01 1.39181972e-01
-6.03512824e-01 -5.16590953e-01 -7.23749936e-01 -9.51826572e-01
4.56910610e-01 -8.48618269e-01 -6.97805583e-01 -4.59623039e-01
-8.06714356e-01 -2.10047930e-01 5.24108887e-01 -3.87516379e-01
1.40827477e+00 -1.97140646e+00 1.94287732e-01 6.72441542e-01
8.61154258e-01 3.32217306e-01 -6.43088937e-01 7.82280743e-01
-1.20207004e-01 3.32897425e-01 -1.81120440e-01 -3.57710361e-01
8.59645605e-02 3.45673442e-01 -8.60718414e-02 5.90502083e-01
2.06565529e-01 1.14694071e+00 -9.77193356e-01 -2.28604287e-01
1.02273775e-02 5.42819560e-01 -4.33019221e-01 1.01001136e-01
9.07986239e-02 9.47656706e-02 -6.56095624e-01 6.80237651e-01
7.87402034e-01 -7.21420646e-01 5.94205141e-01 -2.42789552e-01
6.79140329e-01 6.67683259e-02 -1.42440975e+00 1.51694274e+00
-1.80021331e-01 1.57334536e-01 4.13866550e-01 -1.34632194e+00
1.08960295e+00 -4.07540463e-02 3.35286468e-01 -6.41776145e-01
1.45410046e-01 -1.63409561e-01 -1.50070544e-02 -1.54812172e-01
2.31762916e-01 3.55692089e-01 -1.21727427e-02 3.20017129e-01
2.20196009e-01 6.44806549e-02 4.02821183e-01 8.61740947e-01
1.70585918e+00 -3.02989781e-01 5.21992564e-01 -3.47339571e-01
4.42851782e-01 -6.04728639e-01 5.44602454e-01 1.19177747e+00
-2.68632352e-01 4.05769557e-01 8.12100232e-01 -2.91782409e-01
-5.43585956e-01 -8.23083341e-01 4.76421267e-01 8.17696154e-01
2.51869887e-01 -8.96031976e-01 -7.15639472e-01 -9.82143760e-01
-3.94431241e-02 1.56131819e-01 -6.72368050e-01 -4.38136458e-01
-5.95068455e-01 -8.12679410e-01 4.91071135e-01 4.30237889e-01
4.17550892e-01 -8.05543244e-01 3.50489020e-01 2.53587633e-01
2.30149291e-02 -1.37361038e+00 -3.56079459e-01 1.38500165e-02
-1.04726470e+00 -1.36219227e+00 -2.53437668e-01 -8.95884037e-01
1.01289952e+00 7.31409550e-01 1.28795755e+00 7.28196859e-01
-7.18826428e-02 5.27142644e-01 -4.54570860e-01 3.56349610e-02
-3.06439400e-01 5.58441520e-01 -1.52900051e-02 -8.39307497e-04
3.22650224e-02 -1.05726504e+00 -3.04815173e-01 -4.89125773e-02
-1.05299413e+00 -1.37025371e-01 8.47341061e-01 7.87419856e-01
5.31432152e-01 2.37435043e-01 7.31398821e-01 -1.28480899e+00
1.02499533e+00 -6.50094748e-01 -7.37323403e-01 3.19178671e-01
-8.42423439e-01 3.65006864e-01 9.40728545e-01 -7.68203586e-02
-5.44494331e-01 -6.28705993e-02 5.71225677e-03 -4.21552628e-01
6.90506995e-02 9.58738029e-01 -4.52965856e-01 -4.25136745e-01
5.39177299e-01 1.54560208e-01 2.77975667e-02 -5.28589666e-01
5.21037042e-01 -3.04251239e-02 4.69134748e-01 -5.43302059e-01
1.32304502e+00 3.63715053e-01 4.67251301e-01 -7.53414035e-01
-8.02894235e-01 -4.02914912e-01 -6.07332945e-01 5.80731109e-02
7.00562298e-02 -9.40510869e-01 -5.70396066e-01 3.65690887e-01
-7.93842614e-01 3.98130678e-02 1.01931421e-02 2.93123305e-01
-1.58422038e-01 1.00173116e+00 -6.78556442e-01 -6.75153732e-01
-5.21639824e-01 -1.05011535e+00 8.10358822e-01 1.05177285e-02
3.33122104e-01 -1.32384741e+00 -8.66898894e-03 -6.29444644e-02
4.13231611e-01 5.50654948e-01 9.26181972e-01 -9.36481237e-01
-8.31541121e-01 -2.76434571e-01 -5.68755269e-01 2.70218372e-01
3.64324272e-01 -3.03789258e-01 -6.34349644e-01 -8.38003159e-01
-1.81729689e-01 -1.95697650e-01 1.24704194e+00 3.02672505e-01
1.04044032e+00 -4.80450481e-01 -3.49447399e-01 7.86026001e-01
1.62522900e+00 -4.67223078e-01 2.81037688e-01 8.72932523e-02
1.36251974e+00 2.71744430e-01 2.03488290e-01 3.05839092e-01
4.68126118e-01 2.95573086e-01 8.71109307e-01 -2.20459253e-01
-3.45749438e-01 -5.01908064e-01 2.25372121e-01 1.25237894e+00
3.00529059e-02 -4.96499151e-01 -6.38635874e-01 5.26432633e-01
-1.95879745e+00 -6.81553602e-01 -1.16243750e-01 1.90333569e+00
2.93096185e-01 1.82830155e-01 3.58083658e-03 8.56488347e-02
8.83955240e-01 6.17609799e-01 -5.96644998e-01 4.65263091e-02
-3.95242840e-01 1.99956432e-01 6.41938865e-01 3.86561215e-01
-1.04940343e+00 1.02271426e+00 6.01467752e+00 8.26011479e-01
-6.81889415e-01 -3.48696589e-01 3.76893520e-01 2.03532964e-01
-3.72396171e-01 -9.98866782e-02 -5.45042634e-01 1.66254729e-01
8.15418303e-01 -4.87297952e-01 6.26718462e-01 8.45012248e-01
-5.23073412e-03 3.94524366e-01 -9.88392532e-01 9.67291772e-01
1.81664541e-01 -1.30654061e+00 2.18437672e-01 2.65244901e-01
8.31417739e-01 1.88759595e-01 -3.45955119e-02 4.72234011e-01
7.77995467e-01 -1.12095141e+00 1.84250027e-02 2.51694024e-01
3.59804958e-01 -7.81895995e-01 7.97871649e-01 2.10024551e-01
-1.72191739e+00 -6.08527735e-02 -5.53381920e-01 -5.75096011e-02
-1.90574199e-01 8.00912678e-01 -1.07740486e+00 1.16773176e+00
4.67756569e-01 1.06219423e+00 -1.06378615e+00 9.95124638e-01
-4.73788410e-01 9.13505733e-01 -3.79057586e-01 4.47807312e-02
3.53293449e-01 -4.16824102e-01 6.46215856e-01 1.00810516e+00
1.57107815e-01 -1.99768990e-01 4.24842894e-01 8.86684775e-01
-5.60970187e-01 3.22299421e-01 -1.07633471e+00 -2.88368374e-01
6.54636562e-01 1.42548835e+00 -9.12327647e-01 -8.94278958e-02
-4.32914525e-01 6.05577648e-01 8.30659628e-01 5.08276880e-01
-4.85390574e-01 -4.95657504e-01 4.34511393e-01 -1.40661672e-01
5.15605330e-01 -3.84448647e-01 -5.88202439e-02 -1.23144591e+00
4.03644860e-01 -8.67869556e-01 6.12925053e-01 -1.02564104e-01
-1.72398615e+00 7.97484636e-01 -1.74791962e-01 -8.75711262e-01
-1.09000504e-01 -3.85184526e-01 -6.87270939e-01 4.63701814e-01
-1.58285952e+00 -1.28207862e+00 -4.41453189e-01 6.03894353e-01
3.83470654e-02 -1.44678131e-01 5.84199429e-01 1.96657062e-01
-7.96740532e-01 7.87601352e-01 2.38295585e-01 4.75142688e-01
2.80322045e-01 -1.26146114e+00 1.07875657e+00 1.38956594e+00
4.69615459e-01 7.72070944e-01 4.30608034e-01 -9.58178818e-01
-1.82731354e+00 -1.42910624e+00 3.84216458e-01 -9.55287665e-02
8.38685870e-01 -6.25477195e-01 -1.09105027e+00 7.81104684e-01
-1.70662344e-01 4.74630505e-01 4.17921543e-01 3.51448119e-01
-5.49181700e-01 -1.66449755e-01 -1.03382695e+00 5.91918945e-01
1.65609026e+00 -4.17514652e-01 -3.12292576e-01 4.04102713e-01
8.99095356e-01 -2.57125616e-01 -7.07265675e-01 5.66797197e-01
4.06013727e-02 -7.61213362e-01 9.42288399e-01 -6.26311839e-01
2.03229021e-03 -2.19429284e-01 -1.52920321e-01 -1.52827418e+00
-5.60543239e-01 -7.33674586e-01 -2.63833374e-01 1.23447645e+00
1.31125554e-01 -7.09537208e-01 9.71849024e-01 2.79769719e-01
-1.35292010e-02 -7.74713337e-01 -8.93809676e-01 -1.06138897e+00
-1.93586379e-01 -2.96672732e-01 6.39448583e-01 8.86640251e-01
-4.09433484e-01 4.94654924e-01 -5.28929174e-01 5.61216056e-01
8.44374657e-01 1.46550700e-01 9.77060974e-01 -1.56799233e+00
-2.92465210e-01 -4.07249302e-01 -8.01530361e-01 -8.73185039e-01
5.05786955e-01 -1.43059278e+00 -2.30290234e-01 -1.83921719e+00
1.84678569e-01 -1.37144968e-01 -5.29233515e-01 4.17494625e-01
-4.50467259e-01 1.00610420e-01 3.35980892e-01 -7.54992887e-02
-9.83337104e-01 8.06359351e-01 1.17279303e+00 -1.68242753e-01
-5.64388223e-02 -3.37630021e-03 -1.21316791e+00 5.87581992e-01
5.47148764e-01 -4.07831520e-01 -7.05257773e-01 -3.60077262e-01
3.03871989e-01 -3.34155232e-01 3.30325007e-01 -1.03421140e+00
2.56117582e-01 9.52478722e-02 -7.32856691e-02 -2.92744935e-01
8.29804167e-02 -8.35283041e-01 -5.18242531e-02 3.63791913e-01
-2.06226036e-01 6.77796453e-02 -5.16274758e-02 1.18449724e+00
-1.28069311e-01 3.86469215e-02 5.01702428e-01 -1.08275995e-01
-6.74708307e-01 6.03262901e-01 2.68152118e-01 3.23092341e-01
6.30017042e-01 -1.49376690e-01 -4.69427913e-01 -6.15643620e-01
-6.39352381e-01 3.41638386e-01 3.29451829e-01 3.94658625e-01
7.17091858e-01 -1.36423147e+00 -7.31028676e-01 3.65610898e-01
1.41025946e-01 1.53455585e-01 2.05800787e-01 6.64969563e-01
-1.34458452e-01 8.48927125e-02 1.68219835e-01 -3.70436937e-01
-1.20068336e+00 7.47984171e-01 2.07106229e-02 -7.08602965e-01
-9.48531687e-01 8.81814659e-01 1.56792656e-01 -5.62550724e-01
5.32383144e-01 -4.70527589e-01 -2.07289413e-01 -3.13339323e-01
2.27652788e-01 1.84156314e-01 3.21066558e-01 -6.36740327e-01
-2.96728700e-01 4.19171512e-01 -3.15880567e-01 6.32530808e-01
1.60510457e+00 -1.89923197e-01 -1.49935022e-01 6.38565794e-02
1.15420449e+00 1.02787115e-01 -1.13263822e+00 -7.50367999e-01
3.56696278e-01 -3.77599001e-01 -2.07291357e-02 -3.47874284e-01
-1.51899242e+00 4.72441673e-01 1.37295857e-01 4.71452057e-01
1.12720191e+00 -8.82439595e-03 6.41116917e-01 7.92047203e-01
4.52704966e-01 -7.22701252e-01 1.00571595e-01 6.27857089e-01
7.93761671e-01 -1.27397060e+00 2.08112866e-01 -7.48275459e-01
-2.88114607e-01 8.63629997e-01 5.36104500e-01 -5.15530169e-01
6.37415349e-01 -3.82358022e-02 -3.42948079e-01 -4.65608686e-01
-7.01312780e-01 -3.73411834e-01 3.00706565e-01 8.27968001e-01
6.67886287e-02 -4.41749431e-02 -6.64022341e-02 7.97418296e-01
-1.89945042e-01 -4.17990953e-01 5.86157680e-01 7.70663798e-01
-2.21329898e-01 -1.13285458e+00 8.03123489e-02 5.89792073e-01
-2.84095615e-01 -2.44728088e-01 -7.25956559e-01 9.02728617e-01
-5.68122745e-01 9.65828955e-01 -4.92828578e-01 -5.13052583e-01
3.35799873e-01 -4.54705805e-01 3.38572770e-01 -7.79480278e-01
-3.88545483e-01 -6.41373619e-02 8.10719207e-02 -8.74481082e-01
-3.37530792e-01 -1.97884366e-01 -1.05347455e+00 -2.55942822e-01
-3.77869844e-01 1.71337128e-01 3.80614549e-01 7.51444936e-01
7.65599132e-01 6.00937009e-01 6.79893076e-01 -7.44433641e-01
-6.84781253e-01 -8.73703003e-01 -8.98668289e-01 6.20245159e-01
3.54723066e-01 -6.49861455e-01 -6.61874175e-01 -5.81941903e-01] | [7.077305793762207, 6.205443382263184] |
61b6546a-59cb-4bbb-b18b-4bb9ac96455d | exploiting-low-rank-tensor-train-deep-neural | 2203.06031 | null | https://arxiv.org/abs/2203.06031v1 | https://arxiv.org/pdf/2203.06031v1.pdf | Exploiting Low-Rank Tensor-Train Deep Neural Networks Based on Riemannian Gradient Descent With Illustrations of Speech Processing | This work focuses on designing low complexity hybrid tensor networks by considering trade-offs between the model complexity and practical performance. Firstly, we exploit a low-rank tensor-train deep neural network (TT-DNN) to build an end-to-end deep learning pipeline, namely LR-TT-DNN. Secondly, a hybrid model combining LR-TT-DNN with a convolutional neural network (CNN), which is denoted as CNN+(LR-TT-DNN), is set up to boost the performance. Instead of randomly assigning large TT-ranks for TT-DNN, we leverage Riemannian gradient descent to determine a TT-DNN associated with small TT-ranks. Furthermore, CNN+(LR-TT-DNN) consists of convolutional layers at the bottom for feature extraction and several TT layers at the top to solve regression and classification problems. We separately assess the LR-TT-DNN and CNN+(LR-TT-DNN) models on speech enhancement and spoken command recognition tasks. Our empirical evidence demonstrates that the LR-TT-DNN and CNN+(LR-TT-DNN) models with fewer model parameters can outperform the TT-DNN and CNN+(TT-DNN) counterparts. | ['Javier Tejedor', 'Pin-Yu Chen', 'Chao-Han Huck Yang', 'Jun Qi'] | 2022-03-11 | null | null | null | null | ['tensor-networks', 'spoken-command-recognition'] | ['methodology', 'speech'] | [-3.11710894e-01 2.71285884e-02 2.62346774e-01 -5.27181566e-01
-7.37214625e-01 -3.06398094e-01 3.46619666e-01 -6.90454185e-01
-4.43718135e-01 -1.36025427e-02 4.13768649e-01 -6.78606391e-01
6.76929653e-02 -2.84997404e-01 -5.88609815e-01 -5.86001039e-01
-4.18067515e-01 1.86651275e-01 -1.54800832e-01 -3.09626937e-01
-1.36047555e-02 4.09306973e-01 -1.13518500e+00 3.93740177e-01
7.50464320e-01 1.50952864e+00 1.93639070e-01 9.00295138e-01
9.83989686e-02 1.15396857e+00 -2.29386836e-01 -5.68808138e-01
5.22160530e-01 -1.68427378e-01 -7.08640218e-01 4.21727709e-02
4.75019991e-01 -7.28738010e-01 -8.67904067e-01 7.89403677e-01
5.61527193e-01 3.95326972e-01 3.20998907e-01 -1.02134800e+00
-8.83504748e-01 6.50107026e-01 -3.19532812e-01 1.56014279e-01
-1.66240260e-01 3.67031634e-01 1.43886566e+00 -1.47058773e+00
2.17959702e-01 1.44388139e+00 7.15093672e-01 5.04597068e-01
-1.06072688e+00 -3.21334749e-01 1.77550405e-01 1.70510203e-01
-1.21305203e+00 -3.35845381e-01 7.79353261e-01 -4.31668729e-01
9.25957203e-01 2.03370124e-01 4.56415623e-01 1.18735957e+00
-2.62823738e-02 1.26721585e+00 8.88924658e-01 -1.06976442e-02
2.00977787e-01 -4.52656090e-01 3.23935390e-01 6.67622507e-01
-3.52752984e-01 1.82627857e-01 -4.75463271e-01 1.58063918e-01
7.66279221e-01 -1.13518663e-01 -1.14546590e-01 6.89867064e-02
-1.29661596e+00 7.47883677e-01 7.38070965e-01 3.34326446e-01
-4.38863993e-01 2.47783825e-01 8.06826472e-01 4.57507998e-01
6.19597673e-01 3.94335032e-01 -7.83552527e-01 -2.70092368e-01
-7.85649121e-01 1.65591806e-01 6.77841425e-01 9.07712400e-01
6.44007504e-01 5.59859097e-01 -4.11314845e-01 1.32789838e+00
2.76505768e-01 -1.85851362e-02 4.79204059e-01 -9.10027504e-01
8.11397552e-01 5.88469326e-01 -4.52119589e-01 -5.54913461e-01
-4.08900887e-01 -7.97364652e-01 -1.10147321e+00 4.93225604e-02
1.23546772e-01 -4.73189294e-01 -9.85955894e-01 1.56243324e+00
1.02351159e-01 -5.81935607e-03 2.01418564e-01 1.35418653e+00
8.40758264e-01 5.84020495e-01 -1.34346560e-01 4.86669272e-01
1.31679153e+00 -1.44812930e+00 -5.09476542e-01 -1.85021698e-01
1.17896509e+00 -7.59156227e-01 1.40133607e+00 3.13063532e-01
-9.52805281e-01 -5.74733377e-01 -7.35203981e-01 -5.86761117e-01
-2.06283599e-01 8.26461971e-01 6.36977196e-01 7.98005760e-02
-1.41927850e+00 6.01497829e-01 -7.73804665e-01 -2.63112068e-01
2.24043250e-01 4.44817215e-01 -2.86070645e-01 -1.18967414e-01
-1.07865238e+00 8.18595886e-01 4.32847291e-02 1.05701494e+00
-1.43655074e+00 -6.74243093e-01 -7.72494435e-01 -1.40501752e-01
5.99779598e-02 -5.01541317e-01 1.42040658e+00 -7.86060810e-01
-1.88339424e+00 4.18929368e-01 -2.96826549e-02 -2.32962757e-01
4.12059069e-01 -3.97795141e-01 -2.35649973e-01 -3.26120332e-02
-2.76053786e-01 6.55017018e-01 6.97226822e-01 -9.34949100e-01
-4.36389804e-01 -2.70827472e-01 2.75733620e-01 2.65894830e-01
-4.15976256e-01 2.89541304e-01 -3.61577392e-01 -7.79748380e-01
4.35099781e-01 -1.07478869e+00 -3.79134536e-01 -1.87360942e-01
-9.40820575e-01 -5.88202357e-01 1.00170875e+00 -8.41628432e-01
1.17269254e+00 -2.20547628e+00 6.37544155e-01 -1.36510506e-01
5.79632342e-01 5.33885896e-01 -6.72899127e-01 3.42097849e-01
-2.34309062e-01 -3.72299552e-02 -9.47492123e-02 -9.37213778e-01
1.01034731e-01 5.02938688e-01 -2.37550259e-01 1.98105574e-01
5.39039195e-01 8.26423526e-01 -7.33528018e-01 -4.46744934e-02
9.11020041e-02 4.38712895e-01 -8.31370115e-01 5.92527986e-01
-2.22893327e-01 4.20488715e-01 -4.30577874e-01 7.16119647e-01
7.22312391e-01 2.05582567e-02 5.71181066e-03 -6.60338461e-01
-4.54077095e-01 7.01441884e-01 -6.62403941e-01 1.53294253e+00
-6.00661039e-01 9.18252826e-01 1.97636038e-01 -9.89876509e-01
8.62638950e-01 3.62956196e-01 2.38462254e-01 -4.74909723e-01
1.04393989e-01 4.94016588e-01 3.12193334e-01 -7.66239107e-01
8.00897360e-01 1.81703746e-01 9.00544077e-02 4.00260329e-01
3.36732149e-01 7.87066966e-02 -4.48590741e-02 1.86340347e-01
1.17996275e+00 3.22123915e-02 -7.67261863e-01 -1.62740022e-01
3.75987619e-01 -4.98716801e-01 5.48695505e-01 3.35883617e-01
-2.48055696e-01 7.29260564e-01 7.61399448e-01 -5.25029659e-01
-1.22687149e+00 -8.71349037e-01 -8.14036131e-02 1.54674220e+00
-5.21491170e-01 -4.69965219e-01 -7.44548798e-01 -6.86679006e-01
-1.92096502e-01 4.96383548e-01 -6.62650049e-01 -4.23078723e-02
-6.00050092e-01 -7.05303729e-01 9.07663584e-01 5.49037039e-01
6.17840171e-01 -7.47553170e-01 1.23789184e-01 1.86322674e-01
-3.02745402e-01 -1.35710740e+00 -8.00109386e-01 4.54941094e-01
-1.01759529e+00 -6.21955633e-01 -6.94262803e-01 -7.15020418e-01
6.13896728e-01 1.71562865e-01 7.97060132e-01 7.35418797e-02
2.99563736e-01 5.14305010e-02 -6.28622532e-01 2.05563530e-02
-1.01157568e-01 3.90908957e-01 1.38845831e-01 4.83844817e-01
2.55584627e-01 -4.60545897e-01 -6.76195502e-01 3.31572533e-01
-8.23809147e-01 -2.01735459e-02 7.56918013e-01 9.67259169e-01
2.91474491e-01 -5.32859504e-01 -1.07136056e-01 -3.84935200e-01
6.91789627e-01 -2.14800760e-01 -4.47383523e-01 4.89011928e-02
-1.42144471e-01 1.87981576e-01 5.42264342e-01 -7.23696828e-01
-4.86737013e-01 -2.61070067e-03 -4.15460497e-01 -1.05439067e+00
2.03180939e-01 8.18695426e-01 -3.51067558e-02 -3.18511128e-02
2.60266751e-01 1.98482245e-01 -8.05482119e-02 -9.15707707e-01
4.74095613e-01 7.27516830e-01 4.15355027e-01 -4.92451966e-01
8.39527607e-01 6.04720600e-03 -2.95571715e-01 -9.52930748e-01
-1.32521212e+00 -9.19304639e-02 -8.97336245e-01 -1.66915268e-01
1.19973922e+00 -1.01739597e+00 -7.13861287e-01 8.72530162e-01
-1.37102604e+00 -5.86563408e-01 -4.38529141e-02 9.55259621e-01
-1.80530995e-01 2.30549097e-01 -1.11062098e+00 -7.68163979e-01
-3.85021806e-01 -1.44452858e+00 1.20569849e+00 -1.37472287e-01
2.19584763e-01 -1.07987261e+00 -1.85211092e-01 5.92640519e-01
5.36033750e-01 -1.30387038e-01 6.72371030e-01 -7.55231678e-01
-6.42637968e-01 -8.78940001e-02 -5.82402945e-01 1.14548433e+00
-2.49672085e-01 1.16568804e-01 -1.33467078e+00 -2.84792244e-01
-9.10381675e-02 -5.09220660e-01 8.33166778e-01 3.40528548e-01
1.31880760e+00 -5.55864334e-01 4.46898401e-01 9.55232382e-01
8.32149148e-01 -1.30095690e-01 7.77644157e-01 3.17779422e-01
1.35887659e+00 5.82945824e-01 3.23825121e-01 1.50910929e-01
1.00178814e+00 7.29676127e-01 4.58202153e-01 -4.87248957e-01
-2.03525901e-01 -5.96194044e-02 7.31764436e-01 1.69037843e+00
-2.03265250e-01 -7.93698989e-03 -8.58716071e-01 2.88699806e-01
-1.78606665e+00 -5.09320676e-01 -2.78015614e-01 1.88433886e+00
7.01614916e-01 2.02130396e-02 1.54681280e-01 6.87489882e-02
4.84032512e-01 2.73748338e-01 -4.95586336e-01 -7.39217818e-01
-1.73396707e-01 -7.33954906e-02 1.28398940e-01 4.34500694e-01
-1.09688544e+00 1.22817469e+00 5.62809420e+00 6.11658216e-01
-1.38585711e+00 3.49651687e-02 7.36366153e-01 -7.19903857e-02
-1.73595935e-01 8.57240409e-02 -7.25820780e-01 1.49229601e-01
1.10712290e+00 5.78209400e-01 7.75480032e-01 8.62398863e-01
4.80643570e-01 5.10445595e-01 -1.37773097e+00 8.21804821e-01
-2.27419674e-01 -1.13482606e+00 3.04331183e-01 2.87025243e-01
5.05701482e-01 6.01241469e-01 2.46089637e-01 7.37279296e-01
5.16964674e-01 -9.08462286e-01 8.49488139e-01 4.54509050e-01
7.54882693e-01 -6.82430387e-01 7.81777620e-01 6.13420457e-03
-1.26856947e+00 -1.19500555e-01 -5.29871643e-01 9.19357687e-03
-1.73429325e-01 9.35198128e-01 -6.78593755e-01 6.60675287e-01
8.39637160e-01 8.19868743e-01 -4.52468872e-01 8.09439898e-01
-2.15449855e-01 9.00620043e-01 -1.85326666e-01 1.26090765e-01
8.94355178e-01 -4.91000295e-01 4.69873250e-01 1.44805408e+00
2.36631900e-01 1.41572729e-02 2.33293071e-01 9.41058755e-01
-3.66785169e-01 -2.96298176e-01 -3.47271860e-01 -1.75090715e-01
2.70669281e-01 1.61583924e+00 -2.74544023e-02 -1.85080677e-01
-3.99968177e-01 8.84706914e-01 7.35509872e-01 6.32107377e-01
-6.44292474e-01 -6.27498746e-01 1.21569002e+00 -3.28302830e-01
4.74231303e-01 -7.30024338e-01 -1.88126549e-01 -1.44104838e+00
4.27147746e-02 -9.50294256e-01 -7.94390142e-02 -8.06262732e-01
-1.49131119e+00 1.22209024e+00 -4.68058497e-01 -1.43332791e+00
-9.97964442e-02 -9.70361352e-01 -5.78296125e-01 1.02021277e+00
-1.57896757e+00 -1.32539153e+00 -1.44280512e-02 6.51828766e-01
5.71606517e-01 -8.60374793e-02 5.36667824e-01 5.89535594e-01
-1.26711476e+00 8.21928024e-01 1.21030256e-01 8.27184260e-01
2.71964490e-01 -1.33250237e+00 4.96978223e-01 8.94475460e-01
-2.71743119e-01 8.56220245e-01 1.79201350e-01 -1.74776971e-01
-1.71824229e+00 -1.52451777e+00 1.05865335e+00 -3.25656414e-01
8.92193019e-01 -8.27679157e-01 -9.16089475e-01 6.86417758e-01
1.57866582e-01 3.60204577e-01 4.38115656e-01 3.13663542e-01
-7.75830686e-01 -3.17755312e-01 -6.05589807e-01 8.63774538e-01
9.11219060e-01 -1.24356031e+00 -4.44167227e-01 4.11403239e-01
1.09607649e+00 -5.64272642e-01 -1.01534414e+00 1.51196986e-01
3.85612160e-01 -8.64315033e-01 7.35885978e-01 -6.92342103e-01
6.51896477e-01 -1.85409427e-01 -5.74511349e-01 -1.32975805e+00
-1.45767957e-01 -8.42797875e-01 -1.65294297e-02 1.12516832e+00
7.19993651e-01 -3.86505634e-01 4.75074738e-01 6.86510205e-01
-9.45589304e-01 -9.28488016e-01 -1.17641664e+00 -8.39460492e-01
8.15676451e-02 -8.04089785e-01 3.42193574e-01 9.45075035e-01
-4.14478958e-01 3.92735183e-01 -3.59766304e-01 2.95512408e-01
4.94860888e-01 -2.87706822e-01 6.38211608e-01 -8.79557669e-01
-1.05302982e-01 -3.96477044e-01 -1.62167192e-01 -1.50922287e+00
1.96779430e-01 -1.01194191e+00 1.57613322e-01 -1.34868312e+00
-1.88128352e-01 -4.27185893e-01 -5.11633635e-01 5.97991407e-01
-9.15826559e-02 1.46569937e-01 3.00097048e-01 3.51081252e-01
-5.44402361e-01 9.32756782e-01 1.51368117e+00 -3.31359580e-02
-1.56939894e-01 -1.98911294e-01 -2.96570182e-01 6.41896605e-01
4.01683867e-01 -2.37038463e-01 -5.89101575e-03 -9.58381116e-01
-8.65878761e-02 -5.94983296e-03 4.92291272e-01 -8.43916297e-01
2.71318674e-01 2.79882550e-01 1.31337106e-01 -7.72060812e-01
5.13282359e-01 -5.00602603e-01 -5.95168829e-01 1.56077728e-01
-6.21968150e-01 3.78657520e-01 -1.77184731e-01 4.30757441e-02
-3.58485758e-01 5.97547181e-03 4.86440599e-01 8.97495225e-02
-3.54166716e-01 6.09254539e-01 -6.50778890e-01 -1.78989530e-01
2.82450557e-01 -2.57697050e-02 -3.37325037e-01 -3.15134227e-01
-7.38681376e-01 2.45613992e-01 -2.90165007e-01 5.06226063e-01
9.08156037e-01 -1.40908062e+00 -7.85875678e-01 2.60867596e-01
-7.15727955e-02 2.18872190e-01 1.89990938e-01 1.12712240e+00
-4.27382529e-01 4.37048942e-01 1.43424302e-01 -7.66789317e-01
-8.05680931e-01 1.54845148e-01 5.59116244e-01 -4.20658588e-01
-3.96144062e-01 1.25296068e+00 3.72092962e-01 -1.12312675e+00
6.15899384e-01 -8.99099946e-01 3.72013599e-02 -9.16107371e-02
1.17890224e-01 4.10872579e-01 2.14139864e-01 -5.88701904e-01
-3.62817049e-01 4.20109630e-01 -3.36645812e-01 -1.66020393e-01
1.60356724e+00 -1.47302181e-01 -3.73605341e-01 4.11412865e-01
1.69674730e+00 -5.05478740e-01 -1.46925843e+00 -3.22602719e-01
2.49534041e-01 -3.60580198e-02 6.22880220e-01 -5.46717882e-01
-1.56009972e+00 1.23335469e+00 5.12916982e-01 -1.81645662e-01
1.09302211e+00 -1.24409005e-01 9.54564929e-01 6.56591833e-01
-4.12496291e-02 -9.10142541e-01 5.41297197e-01 1.17725813e+00
1.27617550e+00 -1.02778137e+00 -3.85642588e-01 2.18333919e-02
-9.47284460e-01 1.45723093e+00 6.60613298e-01 -1.42053097e-01
7.30265737e-01 7.24953264e-02 2.80669630e-01 -2.92849481e-01
-1.03805792e+00 -3.31951320e-01 5.71143031e-01 2.73164243e-01
6.98639572e-01 1.14479192e-01 3.44601065e-01 3.68808031e-01
-5.12113452e-01 -9.10358503e-02 4.47410852e-01 5.72543859e-01
1.29439101e-01 -1.02711737e+00 -3.74390334e-02 5.32712936e-01
-2.86900461e-01 -4.51254904e-01 -2.79393286e-01 4.65131819e-01
-6.52971864e-02 1.08973038e+00 1.41566381e-01 -1.29905510e+00
5.01033843e-01 -1.87362626e-01 2.79637016e-02 -5.22818327e-01
-1.25347376e+00 2.95925856e-01 3.40681911e-01 -6.80756450e-01
-1.32621937e-02 -5.54350495e-01 -1.02620840e+00 -4.69481975e-01
-3.94150496e-01 3.71308066e-02 9.43764687e-01 8.48312497e-01
5.09324491e-01 7.29299307e-01 1.01284671e+00 -1.11184537e+00
-8.58206511e-01 -1.27951860e+00 -6.23030782e-01 4.07886207e-02
8.10118973e-01 -3.37586999e-01 -6.19358003e-01 -2.76818424e-01] | [14.244793891906738, 6.225135326385498] |
0d6a9d1c-0b04-427d-a9c5-9f6a017af783 | multivariate-prediction-intervals-for-random | 2205.02260 | null | https://arxiv.org/abs/2205.02260v2 | https://arxiv.org/pdf/2205.02260v2.pdf | Multivariate Prediction Intervals for Random Forests | Accurate uncertainty estimates can significantly improve the performance of iterative design of experiments, as in Sequential and Reinforcement learning. For many such problems in engineering and the physical sciences, the design task depends on multiple correlated model outputs as objectives and/or constraints. To better solve these problems, we propose a recalibrated bootstrap method to generate multivariate prediction intervals for bagged models and show that it is well-calibrated. We apply the recalibrated bootstrap to a simulated sequential learning problem with multiple objectives and show that it leads to a marked decrease in the number of iterations required to find a satisfactory candidate. This indicates that the recalibrated bootstrap could be a valuable tool for practitioners using machine learning to optimize systems with multiple competing targets. | ['Maxwell Hutchinson', 'Brendan Folie'] | 2022-05-04 | null | null | null | null | ['prediction-intervals'] | ['miscellaneous'] | [ 2.66972333e-01 -1.32251695e-01 -3.01420689e-01 -2.97690302e-01
-9.84389842e-01 -5.13842523e-01 4.02296424e-01 7.87692145e-02
-3.64802122e-01 1.55157089e+00 -4.15663421e-01 -6.74513280e-01
-6.98495626e-01 -5.22945940e-01 -9.11298990e-01 -8.29535723e-01
1.98595785e-02 6.46552503e-01 2.56310049e-02 2.61786789e-01
4.80015099e-01 4.73766744e-01 -1.43223512e+00 -9.31200944e-03
7.81782448e-01 9.64499712e-01 2.40440816e-02 6.21668935e-01
4.52019244e-01 5.11814535e-01 -7.00669289e-01 6.50958866e-02
1.68988645e-01 -7.61589587e-01 -4.18149173e-01 -4.77112941e-02
-1.09740578e-01 -5.57277724e-02 4.70950425e-01 8.46781075e-01
5.34885406e-01 3.15503031e-01 1.00862265e+00 -1.17623794e+00
8.85254592e-02 7.91309476e-01 -8.09065402e-01 -1.46901887e-03
6.63792947e-03 2.51054615e-01 7.99545646e-01 -6.29077077e-01
-1.53120890e-01 1.34264028e+00 6.05166793e-01 -3.96909900e-02
-1.78019547e+00 -7.77554512e-01 -6.93728626e-02 -1.31354019e-01
-1.20196915e+00 -3.23341966e-01 5.60855329e-01 -5.40448070e-01
5.44286966e-01 1.70879796e-01 6.25260949e-01 7.46763051e-01
5.27968824e-01 2.55877316e-01 1.42556000e+00 -7.12832570e-01
8.02334309e-01 3.36857766e-01 -9.78337079e-02 4.48647946e-01
7.09308505e-01 8.38495314e-01 -3.44797045e-01 -3.84660810e-01
7.51057267e-01 -2.85206050e-01 6.32259948e-03 -6.22918189e-01
-9.30188179e-01 1.23707747e+00 1.46670491e-01 -8.47557336e-02
-3.67991984e-01 5.06961107e-01 7.00675324e-02 2.50029892e-01
3.70314270e-01 1.20065653e+00 -6.57700598e-01 -6.87450022e-02
-9.46901560e-01 3.80793393e-01 7.84286082e-01 5.79129815e-01
3.29552829e-01 2.62391746e-01 -1.15811050e-01 8.06334913e-01
3.80228549e-01 5.28189123e-01 2.08824888e-01 -1.09221470e+00
2.88992345e-01 1.34167343e-01 9.54473138e-01 -4.45566475e-01
-5.33685923e-01 -6.56731665e-01 -4.00761575e-01 5.23631871e-01
5.41079223e-01 -5.83220899e-01 -9.89051640e-01 1.52061176e+00
4.32373554e-01 2.29401235e-02 -5.41067906e-02 5.52798271e-01
-4.11406904e-02 5.99224091e-01 -1.21975411e-02 -6.75882101e-01
7.79557407e-01 -5.67221344e-01 -5.00530243e-01 -1.88665032e-01
4.65725869e-01 -6.06036127e-01 8.88230622e-01 8.03033888e-01
-1.02924287e+00 -4.20733064e-01 -1.17821729e+00 1.05454087e+00
7.43760392e-02 3.35215330e-02 4.95558351e-01 9.61998403e-01
-3.52166533e-01 9.52624023e-01 -8.42465341e-01 1.95157424e-01
3.42627287e-01 5.34502208e-01 2.10732669e-01 1.63841844e-02
-1.00387681e+00 1.23216939e+00 7.44982004e-01 1.15166046e-01
-9.57462311e-01 -6.41584218e-01 -6.60747707e-01 8.71548653e-02
5.07384777e-01 -4.47836637e-01 1.40613520e+00 -5.38054407e-01
-1.63386905e+00 -3.13529596e-02 1.09218970e-01 -4.61867034e-01
5.20619392e-01 -4.63300273e-02 4.43893708e-02 -3.93048555e-01
-2.07548141e-01 8.83865878e-02 9.46844459e-01 -1.24297023e+00
-6.09275222e-01 -2.19727620e-01 -3.40161651e-01 -8.32399577e-02
2.08272159e-01 -1.29291177e-01 3.63375902e-01 -4.68857020e-01
8.06685016e-02 -1.13515496e+00 -7.34345376e-01 -6.09546006e-01
-1.87171936e-01 8.90721381e-02 3.14117134e-01 -5.54462433e-01
1.17017794e+00 -1.74720967e+00 5.65674193e-02 7.04923034e-01
-2.61771381e-01 -1.73498645e-01 1.08477525e-01 4.36864436e-01
-1.85427740e-01 1.35242537e-01 -2.79049575e-01 2.85036743e-01
-1.02167733e-01 6.97274506e-02 -1.11689553e-01 5.13518393e-01
1.96791530e-01 5.70393562e-01 -8.01798701e-01 -3.06513309e-01
2.74270386e-01 -1.31020963e-01 -4.43892807e-01 3.28040570e-01
-3.17833871e-01 4.77529675e-01 -4.82687354e-01 2.78838426e-01
2.79981673e-01 -3.15956682e-01 2.82217026e-01 8.42372775e-02
9.29407850e-02 -6.68021962e-02 -1.43451691e+00 7.94102550e-01
-6.11853719e-01 2.47460604e-01 -2.11941451e-01 -1.20820701e+00
9.58645403e-01 1.18500255e-01 5.01345873e-01 -3.50571841e-01
2.85412371e-01 1.01105265e-01 1.47080064e-01 -2.04264149e-01
8.50598440e-02 -6.64883614e-01 -1.85489193e-01 4.63409334e-01
-1.55575886e-01 -8.65719795e-01 9.28220004e-02 -4.13561255e-01
8.89187992e-01 2.17027813e-01 6.51610494e-01 -4.33966190e-01
1.28052786e-01 1.73872456e-01 5.46676755e-01 1.09848678e+00
1.95099086e-01 2.72639304e-01 5.84909260e-01 -9.00663882e-02
-1.14647388e+00 -9.67011929e-01 -3.38089257e-01 6.75411820e-01
-2.31516451e-01 2.17394307e-01 -4.45823967e-01 -4.18907702e-01
5.27664959e-01 1.32019305e+00 -4.35946345e-01 -4.08288836e-01
-1.31115705e-01 -1.12074554e+00 1.25248614e-03 5.96083045e-01
-1.22180581e-01 -6.40105605e-01 -6.88564360e-01 5.39289832e-01
2.83139139e-01 -5.17878652e-01 -4.37342823e-02 7.74900019e-01
-9.93912756e-01 -1.10352468e+00 -5.16058683e-01 -2.85350025e-01
5.39495945e-01 -2.62304574e-01 9.88303542e-01 -3.79544288e-01
-7.53040314e-02 1.78476438e-01 -2.75915060e-02 -7.45951653e-01
-7.04863667e-01 -4.14183468e-01 2.83908308e-01 -4.61810976e-01
-7.44628757e-02 -3.58127981e-01 -2.42390946e-01 7.63187826e-01
-4.64363098e-01 -1.05320841e-01 4.41646576e-01 1.37802970e+00
4.61829722e-01 4.72162306e-01 9.92341101e-01 -7.24887431e-01
9.31393206e-01 -3.23453397e-01 -1.61743736e+00 5.21494985e-01
-1.03254890e+00 6.24470055e-01 5.96180677e-01 -7.92078793e-01
-9.21752930e-01 1.82612732e-01 4.71768469e-01 -3.37733179e-01
2.42476434e-01 8.38199258e-01 1.79614484e-01 -2.46384084e-01
8.76684666e-01 -3.28730911e-01 1.36080787e-01 -2.55025893e-01
2.01848730e-01 4.38236952e-01 -9.97158438e-02 -7.73688197e-01
5.76657474e-01 -4.26373214e-01 4.95992601e-01 -5.46748579e-01
-6.39303029e-01 -1.04455207e-03 -1.18401513e-01 -4.53282028e-01
2.79365331e-01 -7.56603181e-01 -8.43208969e-01 4.41549681e-02
-6.13050997e-01 -5.72059631e-01 -3.16954404e-01 9.33932126e-01
-1.03691697e+00 -4.97872569e-02 -1.83679890e-02 -1.22530019e+00
1.72380775e-01 -1.20003247e+00 5.49177706e-01 3.15124214e-01
-4.55186844e-01 -8.60789061e-01 2.43506014e-01 1.67185545e-01
2.26822108e-01 4.67785448e-01 1.13755584e+00 -6.91031277e-01
-2.74572611e-01 -3.23966593e-01 2.91472465e-01 3.83226126e-01
2.25263178e-01 1.72360167e-01 -5.16943574e-01 -5.28833389e-01
9.71053615e-02 -5.98082781e-01 6.03675067e-01 9.10912573e-01
1.08847618e+00 -3.37262183e-01 -2.94335812e-01 -1.88092934e-04
1.19846141e+00 7.56810725e-01 1.60833120e-01 1.73878893e-01
2.52335239e-02 6.41717434e-01 1.08602405e+00 8.30485284e-01
-4.77657020e-01 5.38956642e-01 1.55272633e-01 1.52659833e-01
6.87402308e-01 -2.51520455e-01 1.26298696e-01 9.75846499e-02
3.06949288e-01 -1.95644513e-01 -9.03199911e-01 3.57239932e-01
-1.95687222e+00 -8.15916657e-01 2.33943418e-01 2.67772245e+00
1.06216311e+00 4.13755566e-01 1.54002503e-01 8.03240687e-02
9.09047723e-01 -4.33325112e-01 -9.67885017e-01 -5.58706343e-01
3.26564133e-01 3.88152808e-01 7.72832870e-01 5.05007267e-01
-7.69571066e-01 4.42658246e-01 7.69202995e+00 9.42803442e-01
-6.76907182e-01 -3.59858930e-01 1.00858533e+00 -2.17210665e-01
-6.07880391e-02 1.08918309e-01 -6.94265127e-01 4.15284485e-01
1.40539551e+00 -6.23208880e-01 6.34405851e-01 9.60069418e-01
7.19790041e-01 -6.98442459e-01 -1.25666142e+00 7.05169797e-01
-5.35183430e-01 -1.22374582e+00 -3.10965240e-01 -4.02168520e-02
1.13500583e+00 -5.17010868e-01 7.89625719e-02 3.12955439e-01
9.27976549e-01 -1.38176668e+00 5.63547552e-01 5.68334401e-01
3.79051358e-01 -1.06786680e+00 6.67419970e-01 4.05443728e-01
-5.74827969e-01 -3.69993418e-01 -2.29515225e-01 -5.32408506e-02
1.04103878e-01 8.81446838e-01 -1.20071292e+00 2.26048678e-01
3.14522773e-01 8.26316774e-02 -2.03922659e-01 1.49886549e+00
-3.80426049e-02 9.74598885e-01 -6.67032480e-01 -5.45538664e-01
-2.30177976e-02 -3.66036236e-01 1.96361110e-01 5.59030533e-01
4.45523143e-01 -1.21860325e-01 3.03404331e-02 9.20762062e-01
4.39954251e-01 -1.19654968e-01 -3.41054648e-01 -2.64192432e-01
5.80056369e-01 6.04227185e-01 -7.26306379e-01 -2.51627922e-01
5.39498776e-02 -5.05863838e-02 -1.12373844e-01 4.90756482e-01
-1.08707833e+00 -3.25696111e-01 2.99332500e-01 -2.04907179e-01
4.36038762e-01 -2.78405011e-01 -4.98814970e-01 -5.73882759e-01
-3.51467192e-01 -1.14034438e+00 3.18187773e-01 -7.07176208e-01
-1.16510236e+00 4.02182825e-02 5.20335317e-01 -1.14431965e+00
-7.33990371e-01 -6.91720366e-01 -3.46871167e-01 9.58665609e-01
-8.17246020e-01 -5.35465539e-01 2.31639370e-01 8.86400864e-02
3.71423423e-01 -1.65964454e-01 4.92979258e-01 -5.61482430e-01
-5.88951230e-01 2.75218397e-01 7.86095977e-01 -5.86328983e-01
6.96318030e-01 -1.15571928e+00 -8.33760351e-02 5.49255431e-01
-8.98583606e-02 4.39074337e-01 1.37161887e+00 -7.63300776e-01
-1.14969850e+00 -9.07318294e-01 -6.87945411e-02 -2.75394529e-01
7.42497325e-01 7.57406279e-02 -4.14776146e-01 3.89235258e-01
-1.81674331e-01 -3.82330239e-01 5.31726182e-01 1.43788666e-01
2.80541122e-01 -2.28556454e-01 -1.32458198e+00 4.19625491e-01
1.95089281e-01 -6.84704334e-02 -4.96948749e-01 4.58232880e-01
6.47430122e-01 -3.78893048e-01 -1.08010757e+00 6.23452902e-01
6.63536906e-01 -6.53077185e-01 7.52120674e-01 -9.07308221e-01
2.68389106e-01 -1.85956523e-01 -5.45265749e-02 -1.73726571e+00
-2.70630866e-01 -5.81755221e-01 -3.37865725e-02 9.26299512e-01
8.50539207e-01 -8.30050826e-01 7.34981000e-01 9.52025115e-01
3.57417226e-01 -8.08659673e-01 -8.94903123e-01 -1.08708978e+00
1.78056225e-01 -2.96266615e-01 4.09768134e-01 4.13170725e-01
-9.66388080e-03 2.92588532e-01 -5.34266174e-01 4.23066504e-02
9.26302731e-01 1.34173051e-01 6.90354228e-01 -1.13108182e+00
-7.57344544e-01 -3.09029609e-01 -1.08220819e-02 -5.57989240e-01
-9.37506706e-02 -5.36550395e-02 4.47372586e-01 -9.92651045e-01
-8.22032627e-04 -7.25392520e-01 -4.57686037e-01 2.44535238e-01
-2.82118708e-01 -2.88708210e-01 -1.64112791e-01 -3.32184881e-01
-9.55970064e-02 6.99661791e-01 8.19174707e-01 -4.52223271e-02
-4.84791994e-01 7.47541606e-01 -6.52745962e-01 5.31277418e-01
9.06883955e-01 -8.27595592e-01 -5.43801904e-01 4.52617019e-01
3.17791551e-01 6.28968239e-01 1.75523505e-01 -8.85878384e-01
-9.82592553e-02 -8.62964571e-01 6.74376488e-01 -6.07996643e-01
1.23253942e-01 -9.89976823e-01 6.02098048e-01 7.13667214e-01
-6.46519244e-01 1.12446606e-01 5.79630077e-01 8.36133778e-01
3.67640145e-02 -5.94817400e-01 1.04570997e+00 -1.60660312e-01
-7.11881667e-02 -2.99331576e-01 -5.18282831e-01 3.49455401e-02
1.17846394e+00 1.71563085e-02 7.32112974e-02 -4.71448541e-01
-6.80127501e-01 5.72253883e-01 2.15458512e-01 3.46447825e-02
4.37315851e-01 -1.05987239e+00 -5.16448438e-01 -8.31672102e-02
-2.65025109e-01 -2.31625736e-01 -1.81167692e-01 6.96516991e-01
-2.32411519e-01 4.85276043e-01 -8.40488672e-02 -6.80713475e-01
-1.12299657e+00 5.61139822e-01 4.71605003e-01 -3.59368950e-01
2.16006979e-01 6.25469029e-01 -1.94319189e-01 -1.16969883e-01
1.65233731e-01 -5.02247512e-01 1.08820580e-01 4.86897714e-02
2.83754081e-01 4.86717939e-01 1.11511543e-01 2.75558174e-01
-2.32424244e-01 2.98251748e-01 1.94077417e-01 -4.11331743e-01
1.41977978e+00 1.24470077e-01 1.67621925e-01 7.61506915e-01
6.76814020e-01 -2.65351325e-01 -1.49904740e+00 -3.41488533e-02
2.38693625e-01 -4.97222811e-01 2.81212687e-01 -1.05491412e+00
-5.90400755e-01 3.51562262e-01 5.76706111e-01 8.36808980e-02
9.38550293e-01 -3.43421221e-01 -2.11203575e-01 6.23888493e-01
6.01283729e-01 -1.18475544e+00 1.33040860e-01 1.71619833e-01
1.04885495e+00 -1.07175195e+00 5.75007379e-01 2.06418812e-01
-7.34949052e-01 1.08778548e+00 5.42063773e-01 -5.67674637e-02
7.50494480e-01 3.54453206e-01 -4.25499111e-01 3.50230128e-01
-8.88264239e-01 1.94838986e-01 3.01908791e-01 2.95540422e-01
1.25192598e-01 1.47406876e-01 -4.71100271e-01 5.11310697e-01
-1.32977515e-01 1.19916029e-01 5.79933763e-01 9.87177670e-01
-7.04095006e-01 -1.05855906e+00 -7.63340712e-01 9.28890109e-01
-1.38454856e-02 7.24251866e-02 -4.85903136e-02 7.55144477e-01
-5.05110562e-01 1.12748325e+00 -3.45370203e-01 -3.07037830e-01
3.39425623e-01 1.83991536e-01 6.15057588e-01 -4.82839644e-01
-3.85024875e-01 2.21822947e-01 3.48932743e-01 -3.70906085e-01
-9.97665524e-02 -7.12212920e-01 -8.63543153e-01 -1.08226284e-01
-1.09295201e+00 5.82287431e-01 7.48405874e-01 9.18125451e-01
4.37251218e-02 5.79174638e-01 8.91429186e-01 -8.56913269e-01
-1.33863902e+00 -9.34762061e-01 -7.14020729e-01 -3.03934753e-01
2.55847722e-01 -1.15505207e+00 -5.95251024e-01 -2.85938919e-01] | [5.032345771789551, 2.718290328979492] |
ffbec38f-342b-455f-9c86-5c229fc79360 | evaluation-strategy-of-time-series-anomaly | 2305.09691 | null | https://arxiv.org/abs/2305.09691v1 | https://arxiv.org/pdf/2305.09691v1.pdf | Evaluation Strategy of Time-series Anomaly Detection with Decay Function | Recent algorithms of time-series anomaly detection have been evaluated by applying a Point Adjustment (PA) protocol. However, the PA protocol has a problem of overestimating the performance of the detection algorithms because it only depends on the number of detected abnormal segments and their size. We propose a novel evaluation protocol called the Point-Adjusted protocol with decay function (PAdf) to evaluate the time-series anomaly detection algorithm by reflecting the following ideal requirements: detect anomalies quickly and accurately without false alarms. This paper theoretically and experimentally shows that the PAdf protocol solves the over- and under-estimation problems of existing protocols such as PA and PA\%K. By conducting re-evaluations of SOTA models in benchmark datasets, we show that the PA protocol only focuses on finding many anomalous segments, whereas the score of the PAdf protocol considers not only finding many segments but also detecting anomalies quickly without delay. | ['Kyushik Min', 'Yongwan Gim'] | 2023-05-15 | null | null | null | null | ['time-series-anomaly-detection'] | ['time-series'] | [ 5.50110340e-02 -3.29798609e-01 3.69703770e-01 -1.18954450e-01
-6.05408192e-01 -6.05002284e-01 4.10598278e-01 7.59672284e-01
-3.44357759e-01 2.66484082e-01 -5.23243964e-01 -4.38053966e-01
-4.92854536e-01 -7.72376418e-01 -3.76042247e-01 -5.99039614e-01
-9.25152063e-01 5.14458418e-01 9.70751464e-01 4.67119440e-02
3.70756269e-01 6.63574517e-01 -1.65744126e+00 2.50190329e-02
6.56878173e-01 1.32367039e+00 -9.23983574e-01 1.01207602e+00
-2.49567583e-01 4.67411071e-01 -1.06164765e+00 -9.13474802e-03
4.08576876e-01 -3.82327855e-01 -4.35500503e-01 -1.64507806e-01
5.85410222e-02 -4.01623189e-01 1.04371198e-01 8.33181620e-01
3.86143148e-01 -6.54007569e-02 3.90834719e-01 -1.98695123e+00
3.08030903e-01 4.55828607e-01 -6.99502051e-01 1.00346708e+00
4.76581156e-01 8.36350322e-02 5.97192645e-01 -3.61488372e-01
1.38491526e-01 8.71705711e-01 9.00084794e-01 1.92557395e-01
-1.15649903e+00 -6.15325987e-01 1.72955915e-01 2.01255202e-01
-1.26859641e+00 -1.25593022e-01 5.44089019e-01 -2.50830472e-01
1.11981571e+00 6.08031690e-01 3.68646741e-01 6.89172089e-01
-4.16258425e-02 5.08864105e-01 4.72030640e-01 -2.76526034e-01
3.98382723e-01 -2.60454416e-01 6.33287847e-01 2.63754934e-01
4.43619788e-01 2.39826754e-01 -1.92786917e-01 -1.01006484e+00
3.97430599e-01 1.18153043e-01 -4.89882566e-02 5.37770577e-02
-1.12047768e+00 4.56569344e-01 -3.90433341e-01 3.83235246e-01
-7.75369108e-01 2.41388846e-02 8.21034253e-01 6.76447749e-01
4.95004773e-01 4.71890628e-01 -6.13900244e-01 -5.87203205e-01
-8.69897544e-01 3.57766390e-01 8.83441508e-01 6.86808169e-01
3.73788238e-01 2.68419147e-01 -2.52785355e-01 4.65079755e-01
3.08930613e-02 6.23218060e-01 4.08977330e-01 -7.53136337e-01
2.76015103e-01 9.20815051e-01 2.00275108e-01 -9.98267293e-01
-5.54271638e-01 -2.21056640e-01 -6.17453456e-01 2.29347944e-01
6.39087737e-01 -2.81672746e-01 -5.31786859e-01 1.65128613e+00
3.14907581e-01 6.66190863e-01 -1.41349867e-01 2.92841882e-01
-1.10750221e-01 8.24643612e-01 1.27629369e-01 -7.58040726e-01
8.55442822e-01 -3.41453969e-01 -7.33534992e-01 3.24522495e-01
9.36560154e-01 -6.71857774e-01 8.44898641e-01 6.44572377e-01
-1.10864913e+00 -2.97341466e-01 -8.53479683e-01 9.89266872e-01
-2.68784255e-01 -4.67681974e-01 1.99035391e-01 8.75807166e-01
-9.97501016e-01 6.09998286e-01 -1.01009154e+00 -5.77445328e-01
5.97388335e-02 2.77519763e-01 -1.29922420e-01 4.69324440e-01
-8.39124560e-01 5.48429787e-01 2.23522827e-01 -2.57834256e-01
-6.62438631e-01 -8.21753681e-01 -3.40724170e-01 8.08497220e-02
2.27394640e-01 -1.87723100e-01 1.11685467e+00 -7.96342611e-01
-1.09103763e+00 5.07762134e-01 -7.85643756e-02 -9.98357117e-01
7.77902305e-01 -4.05267090e-01 -1.06617451e+00 2.45938897e-01
-9.01816487e-02 -3.78759772e-01 7.02436209e-01 -8.30584705e-01
-1.09997487e+00 -4.03685033e-01 -2.14158610e-01 -3.46585900e-01
-1.20387480e-01 3.03310782e-01 9.36488435e-02 -4.67650533e-01
1.48960054e-01 -5.56452513e-01 -3.65458608e-01 -1.73082113e-01
-3.37411225e-01 -3.75929415e-01 1.52855647e+00 -3.19786847e-01
1.77199197e+00 -2.52029490e+00 -8.42993855e-01 9.60765243e-01
1.72003120e-01 5.09627342e-01 -7.39938673e-03 5.47303498e-01
-3.30011696e-01 4.52701934e-02 -5.24337709e-01 1.06488457e-02
-2.55491018e-01 4.63543355e-01 -6.86807215e-01 6.36382103e-01
1.16957769e-01 1.02562554e-01 -9.17701364e-01 -1.84249490e-01
2.04088226e-01 -1.50914654e-01 -5.45976400e-01 4.62375969e-01
-3.88343111e-02 1.29564315e-01 -3.86462539e-01 6.58723295e-01
7.15168059e-01 -8.05525109e-02 -3.17297578e-01 2.88298100e-01
-2.61074245e-01 -2.76783761e-02 -1.44795191e+00 6.80472195e-01
1.90400615e-01 3.17261368e-01 -3.08284223e-01 -1.14650607e+00
1.10126948e+00 5.59236646e-01 9.57195044e-01 -7.30984926e-01
-8.06396231e-02 3.57179314e-01 -4.86238413e-02 -6.51624739e-01
1.15483314e-01 4.91583586e-01 3.65458019e-02 8.49412084e-01
-4.27528858e-01 5.84627986e-01 4.56402928e-01 7.12313950e-02
1.90305603e+00 -2.57609725e-01 4.18252110e-01 -1.31888852e-01
7.64244080e-01 5.39301336e-02 5.18312752e-01 1.12574852e+00
-6.16830766e-01 3.48386943e-01 8.89667511e-01 -6.79388523e-01
-8.20620418e-01 -1.53524911e+00 -3.57601605e-02 8.63133013e-01
-4.68625613e-02 -3.05698723e-01 -6.04258418e-01 -7.97127426e-01
5.05253151e-02 8.19707811e-01 -4.64908779e-01 -2.90512949e-01
-7.25700021e-01 -1.05586874e+00 9.23397064e-01 2.80397564e-01
3.13221902e-01 -1.01416576e+00 -1.05207455e+00 5.24339199e-01
-6.39676824e-02 -1.13772976e+00 -1.60889208e-01 1.98713085e-03
-9.38912988e-01 -1.42108369e+00 -1.75647184e-01 -5.37797399e-02
6.96684599e-01 9.33205411e-02 1.12034595e+00 3.22550505e-01
-1.41541481e-01 6.85989320e-01 -6.50742531e-01 -7.71617234e-01
-6.05488896e-01 -4.00054514e-01 1.12601072e-01 2.42122695e-01
6.86776280e-01 -8.77897203e-01 -5.47815859e-01 5.63531935e-01
-9.60669458e-01 -9.99674976e-01 3.07979047e-01 3.02695453e-01
6.35760605e-01 1.33594513e-01 9.62079048e-01 -9.42209542e-01
7.65850723e-01 -7.25836873e-01 -8.06473434e-01 -3.21383588e-02
-1.10369194e+00 -1.31903946e-01 5.55642664e-01 -5.96132159e-01
-4.39143807e-01 -1.21415034e-01 -2.14617386e-01 -4.20313686e-01
-4.41113114e-01 -1.91908218e-02 2.77217686e-01 1.22458436e-01
7.72340298e-01 3.61105084e-01 -1.24630081e-02 -2.40550533e-01
-3.70502263e-01 3.04083377e-01 5.58961749e-01 -4.23412144e-01
8.25208366e-01 5.19234359e-01 1.23552242e-02 -8.88783574e-01
-3.96547288e-01 -9.33219969e-01 -2.71241575e-01 -2.79357493e-01
1.66802973e-01 -3.57068807e-01 -6.18031323e-01 6.51862860e-01
-9.63128984e-01 8.96932483e-02 -5.78158557e-01 2.34693870e-01
-2.73973078e-01 6.59435213e-01 -2.94436246e-01 -1.14568365e+00
-5.47315240e-01 -4.95399296e-01 8.43574166e-01 -1.92025676e-01
-4.03630584e-01 -6.94320023e-01 5.12182415e-01 -6.40612006e-01
6.20161831e-01 8.76941144e-01 8.62377584e-01 -1.46086490e+00
-1.46282732e-01 -7.47623444e-01 -1.32493004e-01 3.00143719e-01
-4.09746058e-02 4.06598121e-01 -8.83982420e-01 -2.60885775e-01
-1.03945192e-02 7.75570452e-01 2.45460808e-01 3.17501277e-01
1.28166413e+00 -3.62087578e-01 -2.27083892e-01 4.55404907e-01
1.31107450e+00 5.60788810e-01 8.45242620e-01 4.96792227e-01
8.12680051e-02 2.39399090e-01 6.69824362e-01 8.04598510e-01
-5.35312900e-03 5.09944499e-01 6.15906179e-01 -1.10318903e-02
5.65763295e-01 2.20049873e-01 6.24813020e-01 3.32481056e-01
-1.57792047e-01 -2.69109756e-01 -1.05211437e+00 6.99283659e-01
-1.92059612e+00 -1.21754098e+00 -6.08629584e-01 2.63481688e+00
3.58658105e-01 5.95131993e-01 7.75606513e-01 1.06878650e+00
6.38417065e-01 -9.64743271e-02 -3.74973029e-01 -8.24436545e-01
2.53398549e-02 2.48242229e-01 2.80842334e-01 -7.38827465e-03
-1.04137468e+00 1.94081113e-01 7.29457951e+00 4.11406755e-01
-9.07361031e-01 4.73968871e-02 2.08126560e-01 1.37723789e-01
3.33418429e-01 -2.51916144e-02 -4.73133773e-01 7.29389966e-01
1.57378554e+00 -5.14962196e-01 -1.17249168e-01 1.07100976e+00
3.67585927e-01 -6.43840358e-02 -9.87080634e-01 7.35130131e-01
-2.92123586e-01 -7.61539042e-01 3.21427621e-02 -1.18260957e-01
3.11010182e-01 3.65242139e-02 -3.62627566e-01 2.88916409e-01
-1.07800581e-01 -4.20070231e-01 3.75367045e-01 4.33312982e-01
2.21745595e-01 -7.77601540e-01 1.08233726e+00 1.81862772e-01
-1.21446228e+00 -2.64153361e-01 5.40773347e-02 -1.04315609e-01
2.26666048e-01 1.08186495e+00 -8.36118042e-01 3.42311740e-01
9.04503703e-01 3.47468406e-01 -5.41319549e-01 1.42613697e+00
2.91793436e-01 1.03161252e+00 -9.92658913e-01 2.07317844e-01
1.79293782e-01 1.16283014e-01 1.09396195e+00 1.19739747e+00
6.17292225e-01 -9.39442813e-02 1.41154855e-01 4.76542294e-01
6.15221500e-01 1.74220890e-01 -4.55367267e-01 1.81942493e-01
6.42262042e-01 8.51330042e-01 -7.19991326e-01 -3.74691278e-01
-2.86867410e-01 4.24823850e-01 -3.12105328e-01 2.60693431e-01
-8.08735311e-01 -5.08744776e-01 5.72328746e-01 5.92184424e-01
2.92368438e-02 -2.20063105e-01 -1.94233105e-01 -6.11821175e-01
1.65728256e-01 -6.64625585e-01 1.04260731e+00 -2.28005096e-01
-1.29670465e+00 6.18749082e-01 1.44884676e-01 -1.67233825e+00
-2.29611740e-01 -3.09166044e-01 -1.18653488e+00 4.46044236e-01
-1.12088478e+00 -5.20031154e-01 -3.05013686e-01 9.37304378e-01
2.10729793e-01 2.54267380e-02 8.86840463e-01 4.51701522e-01
-7.43364036e-01 8.07694316e-01 -1.98926374e-01 -4.29101102e-02
5.45704901e-01 -1.34527123e+00 5.41981280e-01 1.33496487e+00
-2.77435333e-01 1.64025873e-01 1.15823483e+00 -3.12038541e-01
-7.21521318e-01 -1.13250887e+00 6.62415445e-01 -5.47829747e-01
7.44194567e-01 3.03019941e-01 -1.40324986e+00 5.23049355e-01
-3.47099692e-01 1.63633376e-01 5.12289762e-01 7.75288194e-02
-2.96828866e-01 -3.15315574e-01 -1.47786832e+00 2.08270758e-01
7.85881042e-01 -2.45358869e-02 -5.66185892e-01 1.79686144e-01
3.64259690e-01 -1.06141672e-01 -7.16597855e-01 7.34898388e-01
2.50901014e-01 -1.30554175e+00 8.13010097e-01 -3.49543601e-01
-3.89701813e-01 -4.26538289e-01 6.07402474e-02 -8.81784379e-01
1.23582976e-02 -7.69384027e-01 -5.54229379e-01 1.22528315e+00
4.10363883e-01 -1.22014534e+00 3.37659836e-01 1.23197913e-01
1.08399943e-01 -4.92121220e-01 -1.05496883e+00 -1.06499457e+00
-6.06689334e-01 -7.73000836e-01 7.91761756e-01 9.51188505e-01
-1.14984885e-01 -4.10246700e-01 -8.32949504e-02 7.10667133e-01
6.03905380e-01 -6.31937608e-02 7.73114681e-01 -1.52648246e+00
-1.55525014e-01 -4.82090920e-01 -7.82849908e-01 -2.56251097e-01
-5.91133237e-01 -1.72699183e-01 -1.49255097e-01 -6.93679571e-01
-2.80877590e-01 -3.76077563e-01 -7.63049424e-01 4.27231669e-01
-5.94537379e-03 -4.02262993e-02 -4.75691378e-01 3.00896704e-01
-8.17081034e-01 -3.33696790e-02 2.06846327e-01 5.00546634e-01
-4.87068772e-01 5.70589304e-01 7.12925717e-02 7.62688041e-01
9.47853744e-01 -8.84414315e-01 -1.81620613e-01 1.46085039e-01
2.23475650e-01 -2.85139009e-02 3.54004264e-01 -1.41025829e+00
1.60375401e-01 1.08003020e-02 3.86854932e-02 -9.65326786e-01
-3.57308865e-01 -9.23675895e-01 1.20568424e-01 8.00441980e-01
-1.44759476e-01 1.03054535e+00 3.74940485e-01 7.39031196e-01
-3.05085152e-01 1.13890193e-01 6.84708416e-01 2.48451486e-01
-7.45330751e-01 4.38623548e-01 -5.48367620e-01 9.71663371e-02
1.43620956e+00 -2.24130243e-01 -4.01053607e-01 -3.37561190e-01
-6.93309486e-01 4.10966903e-01 2.07336470e-01 2.79986829e-01
5.46988189e-01 -1.07851160e+00 -5.88571608e-01 5.90568185e-01
4.11814362e-01 -2.84591377e-01 1.11903720e-01 1.15275657e+00
-7.67888367e-01 -2.74418723e-02 -2.44359404e-01 -9.49761450e-01
-1.33985901e+00 6.52466595e-01 4.71498370e-01 -5.45894861e-01
-7.87677824e-01 3.00509006e-01 -4.04424638e-01 -4.08195294e-02
4.99244958e-01 -2.94050336e-01 -1.42232612e-01 -2.41007864e-01
9.23020363e-01 8.80969226e-01 1.33227959e-01 -2.26626813e-01
-4.99259323e-01 4.14905518e-01 -4.13335934e-02 2.07379580e-01
1.06665647e+00 -1.05628796e-01 -1.25169709e-01 5.31839252e-01
6.03621781e-01 -8.37639645e-02 -7.16677070e-01 -2.75719285e-01
6.34841323e-01 -2.66006500e-01 -3.05682689e-01 -5.40052295e-01
-7.56516516e-01 5.87205946e-01 1.01075590e+00 1.01967347e+00
1.57599354e+00 -3.69623214e-01 8.91740322e-01 3.18816245e-01
2.92425305e-01 -8.89473915e-01 -1.11873649e-01 4.24492955e-01
5.60284019e-01 -7.70765662e-01 -2.81340033e-01 -1.37254700e-01
-3.80969375e-01 1.02817202e+00 6.55322313e-01 -4.13588285e-01
9.63283956e-01 4.30028379e-01 6.24922924e-02 -1.70398980e-01
-8.94245744e-01 -3.51116732e-02 -8.83168727e-02 4.80186194e-01
-1.73701152e-01 -4.80977707e-02 -5.76320171e-01 5.71297765e-01
1.54553548e-01 -8.63955319e-02 5.56609690e-01 9.07400012e-01
-5.67523122e-01 -8.52104008e-01 -4.57694918e-01 6.00788951e-01
-7.20225036e-01 4.57743913e-01 -2.62460351e-01 8.18588018e-01
-3.04523576e-03 9.52120125e-01 3.90644759e-01 -4.46035177e-01
7.72332788e-01 2.21438915e-01 -2.68826067e-01 -1.02026194e-01
-6.66050732e-01 -1.33342057e-01 4.62328055e-04 -1.08291173e+00
-2.73079723e-01 -5.59470832e-01 -1.23593175e+00 -4.18778479e-01
-1.42765835e-01 5.57920516e-01 4.16184157e-01 1.03352332e+00
4.52127337e-01 5.70940256e-01 8.42768013e-01 -1.62015766e-01
-7.45688856e-01 -8.34502161e-01 -6.16393626e-01 5.62456787e-01
5.94203353e-01 -4.20087010e-01 -9.78489518e-01 -2.80360699e-01] | [7.366203308105469, 2.7445459365844727] |
e2c795ee-10a7-4ba9-97c0-ec662fbee10c | structurallm-structural-pre-training-for-form | 2105.11210 | null | https://arxiv.org/abs/2105.11210v1 | https://arxiv.org/pdf/2105.11210v1.pdf | StructuralLM: Structural Pre-training for Form Understanding | Large pre-trained language models achieve state-of-the-art results when fine-tuned on downstream NLP tasks. However, they almost exclusively focus on text-only representation, while neglecting cell-level layout information that is important for form image understanding. In this paper, we propose a new pre-training approach, StructuralLM, to jointly leverage cell and layout information from scanned documents. Specifically, we pre-train StructuralLM with two new designs to make the most of the interactions of cell and layout information: 1) each cell as a semantic unit; 2) classification of cell positions. The pre-trained StructuralLM achieves new state-of-the-art results in different types of downstream tasks, including form understanding (from 78.95 to 85.14), document visual question answering (from 72.59 to 83.94) and document image classification (from 94.43 to 96.08). | ['Luo Si', 'Fei Huang', 'Songfang Huang', 'Wei Wang', 'Ming Yan', 'Bin Bi', 'Chenliang Li'] | 2021-05-24 | null | https://aclanthology.org/2021.acl-long.493 | https://aclanthology.org/2021.acl-long.493.pdf | acl-2021-5 | ['document-image-classification'] | ['computer-vision'] | [ 2.85912126e-01 -1.40607739e-02 -1.50501072e-01 -3.12778533e-01
-1.00513649e+00 -8.14767897e-01 7.05074549e-01 3.78797561e-01
-2.24311933e-01 4.50326145e-01 2.73787975e-01 -6.18099511e-01
2.58367002e-01 -9.46465671e-01 -1.05173540e+00 -5.83700120e-01
3.20798278e-01 4.87176448e-01 4.49863411e-02 2.46964954e-02
6.83194399e-01 5.61527967e-01 -1.10742199e+00 1.17209136e+00
8.90857935e-01 9.85131860e-01 3.34433585e-01 1.01664793e+00
-6.74242318e-01 8.92323613e-01 -5.75237572e-01 1.33251408e-02
-2.35165894e-01 -3.16426635e-01 -9.57335472e-01 2.14698017e-01
5.50530434e-01 -3.33770841e-01 -2.42873400e-01 7.23551989e-01
3.82047296e-01 -1.84844092e-01 1.01001239e+00 -5.46314716e-01
-1.07460749e+00 3.02863091e-01 -8.47357750e-01 2.87985685e-03
2.71826893e-01 7.90766478e-02 1.02516377e+00 -8.71290326e-01
7.60163248e-01 1.37095344e+00 2.89544374e-01 4.44453597e-01
-1.18185413e+00 -3.80187094e-01 5.59356928e-01 2.62695849e-01
-1.33016682e+00 -3.40867400e-01 4.31764573e-01 -5.09463847e-01
1.38936853e+00 1.09004311e-01 4.64019537e-01 9.59581256e-01
1.69373304e-01 1.25440454e+00 1.18835640e+00 -6.43202424e-01
-9.72768813e-02 4.28782031e-02 1.61634475e-01 8.92178655e-01
1.06793262e-01 -4.94789034e-01 -4.10915792e-01 3.18318069e-01
8.03508043e-01 -5.88747188e-02 -1.43311456e-01 -1.86251700e-02
-9.22906578e-01 6.46899104e-01 4.69923347e-01 4.45843786e-01
3.60631905e-02 3.22817415e-01 2.58806318e-01 2.53110137e-02
4.56790030e-01 5.29170096e-01 -4.73443449e-01 1.42982632e-01
-8.38527620e-01 3.75588238e-02 6.07945144e-01 1.12297177e+00
8.25846314e-01 -2.77964860e-01 -5.66951811e-01 9.28420186e-01
5.78523219e-01 5.78448057e-01 4.08411622e-01 -7.00158715e-01
7.23759830e-01 7.63072371e-01 -2.80327231e-01 -6.03020966e-01
-2.65651107e-01 -6.08153999e-01 -7.35530615e-01 -3.31812531e-01
4.49033946e-01 2.60031223e-01 -1.41199672e+00 1.41626108e+00
-6.25604987e-02 -2.84984738e-01 -2.77464539e-02 6.13572061e-01
8.80452931e-01 1.10779333e+00 8.05566460e-02 1.90182298e-01
1.41618967e+00 -1.54603314e+00 -6.07161283e-01 -4.77800667e-01
1.13481545e+00 -9.38254833e-01 1.40251327e+00 4.76038784e-01
-9.99612629e-01 -6.57013416e-01 -9.35649157e-01 -6.31950498e-01
-7.27339864e-01 4.10628378e-01 3.74673486e-01 2.18938872e-01
-9.90324438e-01 1.68513253e-01 -5.04453123e-01 -3.98487955e-01
5.96345127e-01 2.12954089e-01 -1.31206051e-01 -5.50266981e-01
-6.16098464e-01 5.78493893e-01 9.32486579e-02 -2.05994114e-01
-9.27900732e-01 -7.39005864e-01 -7.77102351e-01 1.77094236e-01
4.00629818e-01 -8.58604491e-01 1.30298483e+00 -6.80827498e-01
-1.33668339e+00 1.17782962e+00 -6.38675511e-01 -4.74018455e-02
2.30896488e-01 -2.74107277e-01 5.05832359e-02 3.28010291e-01
1.05964683e-01 9.71118033e-01 4.56737429e-01 -1.61778450e+00
-4.48687524e-01 -6.33550465e-01 4.55976203e-02 2.59819150e-01
-3.89120787e-01 -2.81385899e-01 -9.84613121e-01 -6.84941888e-01
-3.46197411e-02 -6.98183477e-01 8.52592513e-02 2.52588630e-01
-5.87832093e-01 -2.00162768e-01 9.58655655e-01 -7.44441390e-01
1.15689123e+00 -1.89313197e+00 4.78275977e-02 -7.54931122e-02
7.34225437e-02 4.20047283e-01 -5.31312168e-01 5.90753317e-01
2.45225504e-01 4.45210427e-01 -1.34617284e-01 -6.39886081e-01
-1.25621125e-01 -2.09190510e-02 -3.77688468e-01 -4.23480233e-04
2.16398448e-01 1.35867226e+00 -5.26022017e-01 -6.61672235e-01
2.02989504e-01 3.66105020e-01 -4.08514380e-01 1.25880614e-01
-6.81729257e-01 3.13070834e-01 -6.27799153e-01 9.12224472e-01
6.66019022e-01 -5.87407649e-01 2.09600523e-01 -2.17342898e-01
-1.13064848e-01 2.19221234e-01 -4.61818039e-01 1.95785248e+00
-6.96641862e-01 7.21738994e-01 2.15576172e-01 -8.25672746e-01
7.77221024e-01 -8.22439045e-02 5.96093014e-02 -1.04769182e+00
-8.46018940e-02 -1.17829349e-02 -2.14845315e-02 -4.57347631e-01
2.26947337e-01 6.58716783e-02 -6.57575354e-02 5.57040989e-01
-3.81485634e-02 -3.02699685e-01 2.45212063e-01 2.54125297e-01
9.99181151e-01 5.11596911e-02 1.40491053e-01 -2.94931471e-01
6.80590212e-01 -1.12790897e-01 -2.11949781e-01 6.91442907e-01
2.36099005e-01 8.45161617e-01 6.74584270e-01 -1.98434711e-01
-9.66261327e-01 -9.11759555e-01 1.06164925e-01 1.22956300e+00
1.33978099e-01 -4.98991549e-01 -9.32735682e-01 -6.34624720e-01
9.02100205e-02 6.92964494e-01 -6.53207898e-01 -1.28977224e-01
-6.13192320e-01 -4.54571247e-01 7.22602844e-01 9.03761208e-01
6.35373116e-01 -9.52468455e-01 -6.33179629e-03 -1.15014039e-01
-2.77544037e-02 -1.24171066e+00 -5.59748352e-01 1.30203754e-01
-8.95734251e-01 -7.39356041e-01 -6.83784246e-01 -1.05098665e+00
9.60347354e-01 4.52916652e-01 9.06470954e-01 3.56415570e-01
-2.83709258e-01 1.14375904e-01 -2.40977913e-01 -3.25181663e-01
-1.13876849e-01 4.10170823e-01 -6.94663107e-01 -5.91956042e-02
1.81808144e-01 -1.07495271e-01 -5.12966216e-01 1.35309666e-01
-8.08793426e-01 2.47170150e-01 1.06386352e+00 7.16949522e-01
9.54291344e-01 -3.95213187e-01 2.46502325e-01 -1.04120636e+00
5.68072557e-01 -1.09224662e-01 -1.69697940e-01 6.59500957e-01
-6.00772858e-01 1.88829824e-01 9.31339025e-01 -2.53680974e-01
-1.18895173e+00 -1.77177697e-01 -2.24526018e-01 -8.10912345e-03
-4.72279757e-01 3.60387683e-01 -4.48159486e-01 5.82272448e-02
2.44380936e-01 3.87263715e-01 -3.36807042e-01 -7.14181602e-01
5.42328894e-01 8.31579983e-01 4.71875817e-01 -6.86047792e-01
3.31948489e-01 6.93238616e-01 -8.66273716e-02 -1.06704926e+00
-1.16854644e+00 -5.14410317e-01 -8.05530965e-01 1.90924436e-01
1.04135728e+00 -8.19135129e-01 -5.46943247e-01 7.62877226e-01
-1.30457711e+00 -6.79624975e-01 2.02291366e-02 -4.96377237e-02
-2.53642946e-01 4.41050828e-01 -1.06786740e+00 -6.14132643e-01
-5.22970140e-01 -9.75130677e-01 1.50208771e+00 2.22038731e-01
-7.46412352e-02 -1.04061830e+00 -2.23800644e-01 8.14795792e-01
1.98167473e-01 6.98406845e-02 1.51698530e+00 -4.01015759e-01
-8.66284907e-01 -8.23826417e-02 -1.00784481e+00 4.49093655e-02
4.65167835e-02 -1.36670485e-01 -1.27781034e+00 -1.44420803e-01
-3.37113440e-01 -4.24884260e-01 1.32094836e+00 4.14142311e-01
1.71714950e+00 -6.41568154e-02 -7.40223825e-01 6.75649822e-01
1.30088007e+00 3.46262842e-01 7.54493296e-01 3.23990077e-01
1.06669962e+00 7.27836132e-01 2.93859243e-01 1.40011296e-01
5.27427673e-01 4.73862737e-01 2.40139052e-01 -3.87449861e-01
-3.61648470e-01 -7.93633163e-01 1.81143239e-01 6.69668496e-01
3.15175414e-01 -8.31357896e-01 -9.39487696e-01 4.83508021e-01
-1.81531513e+00 -3.70107949e-01 -3.46047014e-01 1.68244994e+00
7.16043830e-01 2.25413397e-01 -3.34976643e-01 -1.83779255e-01
5.35765290e-01 3.09019417e-01 -7.13322341e-01 -4.51067984e-01
-2.85715193e-01 9.86005813e-02 2.86794633e-01 5.53531229e-01
-9.52383459e-01 1.42079163e+00 5.96755743e+00 1.16130829e+00
-1.02683902e+00 -5.57255298e-02 1.11661744e+00 -3.98483537e-02
-4.48751986e-01 -6.26647621e-02 -9.93468344e-01 2.47817993e-01
5.80907643e-01 4.22035635e-01 3.00544471e-01 4.82686192e-01
1.45760685e-01 -2.45081723e-01 -1.12402916e+00 1.01131761e+00
3.44910353e-01 -1.50986075e+00 5.43232441e-01 3.01835865e-01
9.18558478e-01 -3.07636410e-01 1.50612488e-01 1.98363140e-01
-5.64083233e-02 -1.58363605e+00 8.02406490e-01 4.88650799e-01
1.04388595e+00 -4.69180733e-01 6.71321809e-01 4.78668362e-01
-1.28879845e+00 -3.36543396e-02 -2.05131426e-01 -4.88820896e-02
4.30243790e-01 6.86704934e-01 -6.05122209e-01 2.74760544e-01
4.09197867e-01 7.61853755e-01 -6.59787118e-01 4.47744936e-01
-4.88513082e-01 5.18376470e-01 -1.77361920e-01 -1.33500725e-01
4.71090674e-01 7.86542073e-02 -1.83150411e-01 1.64061952e+00
2.25222751e-01 4.00614925e-02 1.89395681e-01 1.04389989e+00
-3.71511757e-01 6.66350424e-02 -6.49608850e-01 -4.12888706e-01
3.60426068e-01 1.26049280e+00 -8.36468577e-01 -5.37927508e-01
-3.34148049e-01 1.01895201e+00 5.68607330e-01 4.17574227e-01
-6.00916982e-01 -3.47964585e-01 3.11207235e-01 2.91313708e-01
5.03728449e-01 -3.27894181e-01 -7.15586245e-01 -9.23056722e-01
-5.66666089e-02 -7.30861366e-01 2.93002456e-01 -1.00435209e+00
-1.07244086e+00 3.19453090e-01 -4.16682184e-01 -4.65180904e-01
5.14643192e-02 -1.16696763e+00 -6.36644542e-01 7.24649966e-01
-1.69066811e+00 -1.67845285e+00 -3.94441903e-01 6.84082732e-02
8.65301251e-01 1.55798588e-02 7.25332201e-01 1.48245171e-01
-5.58351696e-01 5.96094191e-01 2.09541604e-01 1.13298781e-01
8.43443632e-01 -1.32100952e+00 3.86560321e-01 4.62313056e-01
2.56311536e-01 6.84069693e-01 6.16516955e-02 -5.58454454e-01
-1.49668527e+00 -1.12161875e+00 9.55786169e-01 -5.87856114e-01
4.00655836e-01 -7.50030041e-01 -8.30974102e-01 4.93886173e-01
2.28881612e-01 -1.94354802e-01 5.94367683e-01 -9.54325404e-03
-5.77744961e-01 -4.15278785e-02 -8.43689859e-01 5.59220910e-01
1.21434522e+00 -9.62668061e-01 -2.17882782e-01 3.08568925e-01
8.35310102e-01 -2.76570886e-01 -6.08893752e-01 2.53561825e-01
5.97097337e-01 -7.18875468e-01 9.56196010e-01 -4.51515526e-01
8.10509026e-01 -2.91406840e-01 -2.02355936e-01 -8.74952853e-01
-3.50223184e-01 -2.84578204e-01 -3.20981860e-01 1.18782532e+00
7.13683188e-01 -2.90341824e-01 8.24464440e-01 1.23299785e-01
-3.74684930e-01 -1.27735949e+00 -2.48370349e-01 -3.95208955e-01
3.53712380e-01 -1.86983734e-01 1.89343676e-01 6.48535669e-01
-2.33007252e-01 8.83629143e-01 4.31499183e-02 -1.74454629e-01
2.09943786e-01 4.37622756e-01 6.59847677e-01 -8.33235860e-01
-3.66717160e-01 -5.88557243e-01 2.95819551e-01 -1.91351378e+00
2.05216989e-01 -9.13316429e-01 -1.52709082e-01 -2.06182933e+00
6.79824829e-01 -1.92052070e-02 -5.68043068e-02 4.95147049e-01
-4.02540639e-02 2.02540502e-01 2.61915863e-01 2.29492351e-01
-8.38809907e-01 3.65871578e-01 1.54963136e+00 -4.82456207e-01
4.97623757e-02 -4.80010539e-01 -1.04333198e+00 6.56341374e-01
7.82126248e-01 -9.81462970e-02 -3.43439907e-01 -1.05011094e+00
5.55861443e-02 -3.26186776e-01 1.91572815e-01 -5.25371909e-01
1.77431792e-01 3.15282196e-02 7.40804613e-01 -9.20657694e-01
3.29081982e-01 -2.15364069e-01 -5.39736331e-01 5.02615869e-01
-5.76988399e-01 -6.44565001e-02 4.15524989e-01 5.32955229e-01
-2.87580013e-01 -3.39155644e-01 5.96356928e-01 -4.21866328e-01
-6.46087408e-01 7.67939836e-02 -4.88045961e-01 4.74144183e-02
7.57551074e-01 -2.42139548e-01 -8.43956172e-01 -4.29755837e-01
-4.03616726e-01 3.46729308e-01 5.28203905e-01 3.56582165e-01
4.81514722e-01 -9.32281792e-01 -4.76000071e-01 1.26936734e-01
1.46622255e-01 3.56056333e-01 3.06297094e-01 5.78029513e-01
-7.21054375e-01 1.08845055e+00 2.64947563e-01 -6.36492670e-01
-1.16191828e+00 5.72698653e-01 2.93696183e-03 -8.40288222e-01
-3.15891355e-01 9.72773135e-01 8.18347454e-01 -4.75717753e-01
1.56616643e-01 -6.51663423e-01 -1.12181015e-01 1.15110509e-01
4.66270089e-01 1.20970979e-01 1.36994883e-01 -2.14877650e-01
-2.93614715e-01 1.20015621e+00 -3.55636448e-01 -1.97551295e-01
1.10677421e+00 -1.77906409e-01 -2.49642417e-01 2.88069248e-01
1.54479659e+00 -2.59019081e-02 -1.17578030e+00 -1.38850771e-02
2.46916637e-02 -1.14209287e-01 -4.58370820e-02 -1.20654690e+00
-7.00216234e-01 1.49821210e+00 2.36494109e-01 -1.77910656e-01
9.15558040e-01 1.43250078e-01 9.58335400e-01 5.03175914e-01
6.45480454e-02 -1.15736711e+00 6.60002232e-01 8.91564369e-01
8.36470068e-01 -1.09565651e+00 -1.52269648e-02 -6.21575594e-01
-4.28984493e-01 1.11918485e+00 9.05897558e-01 1.68421298e-01
2.73974955e-01 4.07366812e-01 2.44510956e-02 -2.56836236e-01
-9.78343248e-01 -2.20583007e-01 4.89610076e-01 4.38678503e-01
8.54576170e-01 -1.36378065e-01 -8.94575119e-02 6.80303514e-01
1.09118357e-01 -2.67819047e-01 5.19736335e-02 9.69999433e-01
-5.78383029e-01 -1.10916483e+00 -1.27649814e-01 5.18865883e-01
-3.08502227e-01 -3.06565642e-01 -1.03124678e+00 6.21800840e-01
7.54092261e-02 7.68290758e-01 2.37343803e-01 -3.10288578e-01
1.34254530e-01 8.56617242e-02 6.90589011e-01 -9.08433735e-01
-2.73308218e-01 4.55478638e-01 1.95094615e-01 -3.95371526e-01
-1.57964732e-02 -2.21727356e-01 -1.57751679e+00 -1.70197979e-01
-3.70449871e-01 -2.15068534e-01 6.17096364e-01 1.15827060e+00
7.10247636e-01 7.73000896e-01 1.20734490e-01 -5.97979426e-01
-3.14584076e-01 -1.07967007e+00 -3.17055553e-01 2.53487110e-01
1.87922254e-01 -1.11911997e-01 -3.76340374e-02 2.80762643e-01] | [11.504968643188477, 2.2591936588287354] |
f5407a76-f7d4-42ce-9fe1-cb05c72fa077 | an-end-to-end-network-for-emotion-cause-pair | 2103.01544 | null | https://arxiv.org/abs/2103.01544v2 | https://arxiv.org/pdf/2103.01544v2.pdf | An End-to-End Network for Emotion-Cause Pair Extraction | The task of Emotion-Cause Pair Extraction (ECPE) aims to extract all potential clause-pairs of emotions and their corresponding causes in a document. Unlike the more well-studied task of Emotion Cause Extraction (ECE), ECPE does not require the emotion clauses to be provided as annotations. Previous works on ECPE have either followed a multi-stage approach where emotion extraction, cause extraction, and pairing are done independently or use complex architectures to resolve its limitations. In this paper, we propose an end-to-end model for the ECPE task. Due to the unavailability of an English language ECPE corpus, we adapt the NTCIR-13 ECE corpus and establish a baseline for the ECPE task on this dataset. On this dataset, the proposed method produces significant performance improvements (~6.5 increase in F1 score) over the multi-stage approach and achieves comparable performance to the state-of-the-art methods. | ['Ashutosh Modi', 'Saim Wani', 'Shreeshail Hingane', 'Aaditya Singh'] | 2021-03-02 | an-end-to-end-network-for-emotion-cause-pair-1 | https://arxiv.org/abs/2103.01544 | https://arxiv.org/pdf/2103.01544.pdf | null | ['emotion-cause-pair-extraction', 'emotion-cause-extraction'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.96847782e-01 2.96035439e-01 6.27293959e-02 -5.70000589e-01
-1.23310077e+00 -7.22731054e-01 6.47678196e-01 1.84464201e-01
-6.30046129e-01 7.78181255e-01 3.61999512e-01 -1.61781475e-01
3.59398663e-01 -2.07967758e-01 -4.54714566e-01 -3.04155260e-01
-7.36038461e-02 2.36621976e-01 7.93230534e-03 -2.54488587e-01
3.46624255e-02 5.44518828e-02 -1.57543850e+00 5.07352173e-01
7.80213177e-01 9.23627794e-01 -1.71497352e-02 7.05474794e-01
-1.74649283e-01 1.05626833e+00 -7.56172240e-01 -6.77978218e-01
-1.57151237e-01 -6.36783004e-01 -9.93775845e-01 -2.51698941e-01
-1.13808475e-01 5.81433177e-02 1.27283365e-01 6.83991313e-01
5.44322252e-01 -5.28979972e-02 6.63803399e-01 -1.42647791e+00
-2.55340815e-01 5.75536609e-01 -6.56672895e-01 -9.95731950e-02
5.22906661e-01 -5.50063968e-01 1.36967587e+00 -1.00952625e+00
7.90891051e-01 1.07340586e+00 4.46947753e-01 7.31775105e-01
-8.76318634e-01 -7.31630445e-01 2.23265827e-01 2.88150042e-01
-1.28094029e+00 -2.59389073e-01 6.89706326e-01 1.73534658e-02
1.55592740e+00 3.19322318e-01 5.23660481e-01 1.05742276e+00
1.51382059e-01 1.14549327e+00 1.31089973e+00 -6.45725310e-01
3.11866343e-01 5.91222309e-02 2.26097003e-01 3.35930854e-01
-2.29169801e-01 -2.85179049e-01 -6.25565171e-01 -1.37982264e-01
1.41813591e-01 -7.62700975e-01 -3.68837088e-01 1.05372310e-01
-1.09223795e+00 7.00306952e-01 1.90710187e-01 3.87967080e-01
-7.17486858e-01 6.46847636e-02 6.15929663e-01 3.28565001e-01
5.71828723e-01 5.89221656e-01 -8.03304553e-01 -5.30821800e-01
-8.22503209e-01 4.96221662e-01 1.13476467e+00 8.67210209e-01
2.46660247e-01 -3.73180151e-01 -1.05293296e-01 8.78796339e-01
7.20541691e-03 1.21397249e-01 2.15729758e-01 -4.70652610e-01
5.09611845e-01 5.65219104e-01 1.94905788e-01 -8.32439363e-01
-4.65403646e-01 -3.36436599e-01 -4.55443710e-01 -2.22285062e-01
2.74276733e-02 -5.80540836e-01 -7.70567119e-01 1.94020760e+00
4.16862637e-01 -2.69162636e-02 4.21822578e-01 9.03491378e-01
9.90526795e-01 9.10564840e-01 3.24475259e-01 -4.29140806e-01
1.46464324e+00 -9.76849377e-01 -1.10785913e+00 -4.41854745e-01
6.93697393e-01 -9.62095737e-01 8.69166613e-01 7.17075646e-01
-9.21754420e-01 3.20968926e-02 -1.05698645e+00 -1.65907919e-01
-3.31460744e-01 3.41024399e-01 7.96555400e-01 2.78089017e-01
-8.75463784e-01 5.27963005e-02 -5.27130246e-01 -2.40009248e-01
-6.94686770e-02 3.97089720e-01 -4.20736343e-01 6.17688037e-02
-1.51867521e+00 9.72891271e-01 4.60870951e-01 1.47432640e-01
-4.29039419e-01 -5.22940040e-01 -9.49060857e-01 -1.01230212e-03
5.50430536e-01 -2.11763620e-01 1.69071758e+00 -1.16343212e+00
-1.29267466e+00 9.41640437e-01 -4.95872408e-01 -2.36879095e-01
3.40103172e-02 -7.04507530e-01 -5.45914888e-01 1.54560441e-02
1.97492149e-02 9.07749474e-01 4.13942039e-01 -1.16879773e+00
-7.68257737e-01 6.91371933e-02 5.14280535e-02 3.08240235e-01
-2.26084650e-01 8.31904590e-01 -7.81632006e-01 -4.99201626e-01
-3.21891785e-01 -1.03250289e+00 -2.09233582e-01 -5.84724128e-01
-3.94279003e-01 -6.30522430e-01 8.28499913e-01 -6.65308595e-01
1.53156888e+00 -2.14531088e+00 1.11198880e-01 -1.51227653e-01
1.26727834e-01 1.24494962e-01 -2.05273807e-01 4.17718887e-01
-4.69838679e-01 2.71727324e-01 -1.10932156e-01 -5.32162189e-01
1.99312314e-01 2.68657178e-01 -2.21298099e-01 3.12941261e-02
7.05841124e-01 7.93027997e-01 -8.88051271e-01 -6.58911407e-01
-1.47135034e-01 4.73918527e-01 -5.30130208e-01 4.00111437e-01
-3.34879786e-01 -3.04997358e-02 -3.56972992e-01 5.11624277e-01
5.14159262e-01 -2.45885905e-02 3.22950274e-01 -1.04863159e-01
-1.83708072e-01 8.11370850e-01 -9.33773279e-01 1.35753632e+00
-6.78903401e-01 7.28279412e-01 1.76528499e-01 -4.54984576e-01
8.69183660e-01 8.35012496e-01 4.14052665e-01 -6.52030766e-01
2.99533337e-01 4.35734987e-01 1.82675377e-01 -3.71566772e-01
7.20191360e-01 -3.11023414e-01 -7.78707623e-01 3.67982656e-01
1.47260770e-01 -1.51030168e-01 3.45612675e-01 1.82076246e-01
1.29753327e+00 1.22198656e-01 6.25631511e-01 -1.26782179e-01
5.06573915e-01 2.09174290e-01 9.61686909e-01 1.13785811e-01
-3.41674268e-01 4.77335870e-01 9.42876995e-01 -1.57110929e-01
-7.14124203e-01 -5.74519098e-01 8.81974623e-02 8.80750835e-01
-1.93931222e-01 -8.44337642e-01 -8.08558166e-01 -8.96031737e-01
-6.04846299e-01 1.08170164e+00 -4.81133610e-01 2.65361130e-01
-6.52568102e-01 -8.06416869e-01 5.78413129e-01 4.54680502e-01
5.50878048e-01 -1.23659909e+00 -7.01377392e-01 4.20398355e-01
-7.62426436e-01 -1.49525666e+00 -1.36025786e-01 4.32646453e-01
-1.49905249e-01 -9.18056071e-01 -3.87344658e-01 -8.26879323e-01
1.91980124e-01 -3.42217892e-01 1.50885832e+00 -2.24065512e-01
-1.12989366e-01 -6.95981756e-02 -6.19332612e-01 -8.93105984e-01
-2.94230938e-01 1.18040785e-01 -2.94199377e-01 -2.52434015e-01
9.22051489e-01 -2.43480682e-01 -3.56472105e-01 -5.95801324e-03
-6.48134470e-01 2.47414991e-01 6.85631394e-01 6.45833790e-01
7.47852087e-01 3.77865881e-02 7.99024403e-01 -8.49717259e-01
7.68894732e-01 -3.51535589e-01 -3.06526423e-01 1.91571921e-01
-5.43244243e-01 -6.50009885e-02 5.37660599e-01 -2.32242897e-01
-1.30582976e+00 2.76893646e-01 -3.71330172e-01 -1.30774483e-01
-3.00365329e-01 7.98798025e-01 -3.35960150e-01 5.63300788e-01
2.81295180e-01 -2.33682826e-01 -5.57762623e-01 -2.81289756e-01
3.96331012e-01 9.42130625e-01 6.77632213e-01 -3.99434894e-01
2.88933188e-01 -3.93339358e-02 -2.01908827e-01 -3.91263455e-01
-1.00665200e+00 -6.04302406e-01 -4.21539068e-01 -2.21039876e-01
1.07452738e+00 -1.17883182e+00 -5.00568867e-01 3.11790258e-01
-1.47646523e+00 -2.12164760e-01 1.92028701e-01 4.40315038e-01
-3.17585349e-01 -8.60256478e-02 -8.03858459e-01 -8.48151088e-01
-4.93217856e-01 -6.96836531e-01 1.23260593e+00 1.45355403e-01
-7.70140529e-01 -6.52963042e-01 7.10045025e-02 7.07205087e-02
6.95601180e-02 5.90517044e-01 9.52364802e-01 -7.98903942e-01
-2.45879032e-02 -1.96714327e-01 -1.77703276e-01 3.76457125e-02
6.64120764e-02 1.76349282e-01 -1.00057232e+00 2.21613735e-01
6.74533173e-02 -7.51168072e-01 5.95493138e-01 -2.43527181e-02
7.30176628e-01 -3.59292775e-01 -1.76437020e-01 1.12763636e-01
1.48559880e+00 3.33684802e-01 6.79101229e-01 3.93086582e-01
4.26286906e-01 8.98081481e-01 1.08942938e+00 4.13657635e-01
7.02593148e-01 6.11197472e-01 3.60046625e-01 -5.12912452e-01
2.14014411e-01 -4.22659107e-02 4.98896390e-01 1.02817988e+00
8.43865648e-02 -5.93004704e-01 -1.00632834e+00 9.15911436e-01
-1.85020602e+00 -6.82947040e-01 -5.62188625e-01 1.53684032e+00
1.15412807e+00 2.99886148e-02 -5.34489788e-02 1.91308737e-01
5.22537708e-01 1.97065249e-01 -8.71747285e-02 -1.03207445e+00
-3.70791070e-02 3.25057983e-01 6.35386854e-02 2.52972096e-01
-1.36760819e+00 1.11544955e+00 5.94417763e+00 7.88041413e-01
-1.21942985e+00 1.72599524e-01 6.05171382e-01 -2.00184226e-01
-1.86380655e-01 -1.11798890e-01 -6.24702871e-01 -5.20678191e-03
1.03405511e+00 -1.16225332e-02 1.14141405e-01 8.42347264e-01
1.71118870e-01 -1.32746965e-01 -1.16569781e+00 8.79353285e-01
3.96302566e-02 -6.16908908e-01 -4.67028141e-01 -1.20898597e-01
7.00659633e-01 4.26725075e-02 -4.52168256e-01 5.67376196e-01
2.00282738e-01 -8.86864185e-01 8.05964589e-01 3.19487974e-02
8.07300031e-01 -1.21006298e+00 1.10576177e+00 1.13643005e-01
-1.19854522e+00 1.69500753e-01 -5.54839429e-03 -3.51187497e-01
4.92121607e-01 6.41882420e-01 -6.87137604e-01 5.74674189e-01
7.43892610e-01 5.01221418e-01 -4.21614796e-01 6.07680440e-01
-8.62255633e-01 9.65681791e-01 -3.24954689e-01 -3.92400622e-01
3.92646015e-01 2.96341985e-01 5.55630445e-01 1.62631309e+00
1.17821559e-01 3.81966293e-01 -9.64460298e-02 7.51408458e-01
-2.40149260e-01 3.67395669e-01 -2.84908950e-01 -2.43477300e-01
3.69600892e-01 1.72779381e+00 -7.01294243e-01 -3.51291746e-01
-4.64827329e-01 1.15189242e+00 5.11268616e-01 8.88302848e-02
-1.00040352e+00 -6.54809773e-01 5.36607027e-01 -4.74489093e-01
6.39102519e-01 5.10503873e-02 -1.35277316e-01 -9.46683526e-01
2.15430617e-01 -1.13789654e+00 2.60536402e-01 -9.00688350e-01
-1.31809914e+00 1.13395250e+00 6.01073867e-03 -9.44376886e-01
-8.67728770e-01 -5.56589365e-01 -7.92087018e-01 6.83216035e-01
-1.54382801e+00 -1.09654117e+00 8.33517220e-03 3.44688416e-01
4.09971744e-01 3.81688565e-01 1.10858476e+00 3.30467314e-01
-5.75812459e-01 4.77186322e-01 -5.15343666e-01 2.30770245e-01
8.85330439e-01 -1.46477306e+00 3.23602915e-01 1.11881888e+00
-3.07759140e-02 4.86354470e-01 8.20459485e-01 -6.66138291e-01
-1.11453187e+00 -1.05233419e+00 1.95367372e+00 -3.50926578e-01
6.92055523e-01 -8.06627929e-01 -6.56965911e-01 4.77966219e-01
7.28799403e-01 -1.61795616e-01 7.02928185e-01 5.35026729e-01
-3.79075974e-01 1.20951176e-01 -8.92285645e-01 6.29214942e-01
7.28315055e-01 -3.81422758e-01 -8.00809443e-01 -8.17412362e-02
7.98592567e-01 -4.46101993e-01 -7.58671880e-01 4.96285886e-01
3.89877468e-01 -8.08713317e-01 3.21745366e-01 -4.11053598e-01
1.01392484e+00 -1.88775167e-01 -8.80244840e-03 -1.28733814e+00
1.59312226e-02 -8.01053107e-01 1.32967994e-01 1.81798744e+00
7.82222033e-01 -1.70686215e-01 2.75973737e-01 9.58447218e-01
-2.03193218e-01 -1.10209870e+00 -8.67026210e-01 -4.14196551e-01
-2.39293035e-02 -6.40240312e-01 4.98800874e-01 9.22546864e-01
2.87405342e-01 8.99394929e-01 -3.10118854e-01 -3.64996307e-02
1.67944968e-01 4.09068018e-01 4.37551081e-01 -8.38896751e-01
-4.17825043e-01 -3.57466727e-01 8.06561187e-02 -5.02807677e-01
4.77251410e-01 -6.35415316e-01 6.40453935e-01 -1.58775795e+00
3.50234240e-01 -1.93308309e-01 -2.65473127e-01 8.76516044e-01
-6.91878080e-01 1.61094725e-01 3.07628095e-01 -1.40344515e-01
-8.24977160e-01 5.58220327e-01 7.92899132e-01 2.76314020e-01
-2.36409903e-01 -4.00907934e-01 -7.55822539e-01 7.44059205e-01
7.47302771e-01 -6.98981762e-01 -2.96203583e-01 -1.43113881e-01
5.05450487e-01 4.92008999e-02 3.53898853e-02 -5.67414522e-01
8.05516262e-03 6.37225583e-02 1.21494763e-01 -6.80430830e-01
1.61082581e-01 -6.48802519e-01 -1.17273100e-01 -6.04860112e-03
-2.27266565e-01 4.15281326e-01 4.51830268e-01 5.63061200e-02
-6.94436729e-01 -2.15523541e-01 2.61192620e-01 1.13924786e-01
-7.75118768e-01 -3.13686371e-01 -8.08792889e-01 1.61115006e-01
8.48687768e-01 3.47678274e-01 -2.65695512e-01 -4.94157046e-01
-2.07856551e-01 2.96513587e-01 1.67092457e-01 5.04127085e-01
5.23264229e-01 -1.16496468e+00 -7.77130961e-01 -4.40558285e-01
1.63818181e-01 -8.64691660e-03 -1.11897536e-01 8.76700044e-01
8.78541823e-03 5.11423945e-01 3.01196184e-02 -7.51709342e-02
-1.66052365e+00 5.34638286e-01 8.18682909e-02 -9.54551876e-01
-2.05050528e-01 9.18523669e-01 1.70776486e-01 -4.18868661e-01
-3.92957078e-03 -2.29975015e-01 -3.95973831e-01 8.51853117e-02
4.17192250e-01 -1.52709246e-01 2.18804345e-01 -6.94960237e-01
-7.02306688e-01 1.58062175e-01 -1.95395529e-01 -4.37311977e-01
1.42960346e+00 -1.21340703e-03 -3.73704940e-01 4.65311080e-01
1.25066793e+00 2.51113474e-01 -7.17879415e-01 1.78318128e-01
3.56143475e-01 2.27838367e-01 1.78396732e-01 -1.21281719e+00
-8.25329840e-01 6.16892934e-01 1.22102641e-01 -1.05985343e-01
1.56648147e+00 1.50244921e-01 9.26762164e-01 1.77753329e-01
2.41535202e-01 -1.24907291e+00 -2.34053046e-01 8.62081110e-01
1.16219246e+00 -1.18742573e+00 -1.72334746e-01 -9.02087212e-01
-8.86070848e-01 7.61389911e-01 7.68171668e-01 6.25374392e-02
3.48492175e-01 7.44551837e-01 3.86897892e-01 -3.61569554e-01
-1.41557705e+00 -4.07810807e-01 5.08430779e-01 1.11454718e-01
9.66773212e-01 5.16965128e-02 -8.49129796e-01 1.19347191e+00
-3.20080549e-01 1.05654687e-01 3.75873268e-01 1.05938184e+00
1.11148208e-01 -1.36951995e+00 -7.04014674e-02 1.66547000e-01
-1.05594623e+00 -3.68057549e-01 -1.16908813e+00 1.13120997e+00
1.79411769e-01 1.33133221e+00 -2.43163556e-02 -4.93109643e-01
3.63307536e-01 2.12771520e-01 2.92268544e-01 -5.16039491e-01
-1.03623569e+00 4.38481748e-01 8.61908436e-01 -7.38082290e-01
-7.51649022e-01 -7.10758150e-01 -1.77033341e+00 1.39430061e-01
-3.63071084e-01 2.38373682e-01 6.70946598e-01 1.11507356e+00
3.72563839e-01 6.94370329e-01 5.63082516e-01 -5.22740126e-01
4.68030758e-02 -9.93886054e-01 -2.10550904e-01 4.78088289e-01
9.66896210e-03 -4.43807632e-01 -1.71930641e-01 -2.70582759e-03] | [12.646896362304688, 6.212346076965332] |
3144e125-573c-44cb-9990-2d658dd9e3ae | gauss-guided-encoder-decoder-architecture-for | 2204.07713 | null | https://arxiv.org/abs/2204.07713v2 | https://arxiv.org/pdf/2204.07713v2.pdf | GAUSS: Guided Encoder-Decoder Architecture for Hyperspectral Unmixing with Spatial Smoothness | In recent hyperspectral unmixing (HU) literature, the application of deep learning (DL) has become more prominent, especially with the autoencoder (AE) architecture. We propose a split architecture and use a pseudo-ground truth for abundances to guide the `unmixing network' (UN) optimization. Preceding the UN, an `approximation network' (AN) is proposed, which will improve the association between the centre pixel and its neighbourhood. Hence, it will accentuate spatial correlation in the abundances as its output is the input to the UN and the reference for the `mixing network' (MN). In the Guided Encoder-Decoder Architecture for Hyperspectral Unmixing with Spatial Smoothness (GAUSS), we proposed using one-hot encoded abundances as the pseudo-ground truth to guide the UN; computed using the k-means algorithm to exclude the use of prior HU methods. Furthermore, we release the single-layer constraint on MN by introducing the UN generated abundances in contrast to the standard AE for HU. Secondly, we experimented with two modifications on the pre-trained network using the GAUSS method. In GAUSS$_\textit{blind}$, we have concatenated the UN and the MN to back-propagate the reconstruction error gradients to the encoder. Then, in the GAUSS$_\textit{prime}$, abundance results of a signal processing (SP) method with reliable abundance results were used as the pseudo-ground truth with the GAUSS architecture. According to quantitative and graphical results for four experimental datasets, the three architectures either transcended or equated the performance of existing HU algorithms from both DL and SP domains. | ['Dulantha Wickramasinghe', 'Neranjan Senarath', 'Lakshitha Ramanayake', 'Dhananjaya Jayasundara', 'Parakrama Ekanayake', 'Vijitha Herath', 'Roshan Godaliyadda', 'Kavinga Weerasooriya', 'Yasiru Ranasinghe'] | 2022-04-16 | null | null | null | null | ['hyperspectral-unmixing'] | ['computer-vision'] | [ 5.85053921e-01 -1.87009364e-01 2.19287798e-01 -1.77339211e-01
-5.31376898e-01 -1.20735668e-01 6.79019511e-01 -1.33954674e-01
-4.61945593e-01 7.94103086e-01 1.61711141e-01 6.45504817e-02
-1.89835653e-01 -9.05729711e-01 -8.61156881e-01 -1.44260442e+00
7.77673796e-02 2.25656405e-01 -3.19961131e-01 -9.19999108e-02
-5.39351068e-02 2.88791299e-01 -1.76960456e+00 3.50344807e-01
9.67283905e-01 1.31008112e+00 4.32255864e-01 5.51261604e-01
-1.65575475e-01 9.82586086e-01 -4.22702581e-01 -2.84911305e-01
6.99883401e-01 -8.96023333e-01 -3.38509798e-01 2.36711040e-01
4.58585918e-01 -3.34360033e-01 -4.18253779e-01 1.54048347e+00
5.54837108e-01 3.15113455e-01 6.84327662e-01 -1.06323314e+00
-8.20665717e-01 8.93724084e-01 -7.02097297e-01 -1.94833696e-01
-2.27405861e-01 3.39026004e-01 9.01104212e-01 -6.95203245e-01
5.53094625e-01 8.20124984e-01 7.62099028e-01 1.90008387e-01
-1.18053687e+00 -6.64495647e-01 -7.62876570e-02 2.43406698e-01
-1.62922418e+00 -4.63608801e-01 8.02540421e-01 -4.89592642e-01
8.04627538e-01 3.20495278e-01 7.83903122e-01 8.19801569e-01
-1.45389259e-01 6.42690897e-01 1.15500283e+00 -7.27766097e-01
3.61651063e-01 5.58117405e-02 -3.82173628e-01 5.82940459e-01
-3.73863168e-02 3.76949817e-01 -4.05674875e-01 1.26913697e-01
7.48284280e-01 -4.55079749e-02 -6.47031784e-01 -1.87551826e-01
-1.23700559e+00 8.66058171e-01 8.49420965e-01 4.63250160e-01
-1.02622485e+00 6.23924062e-02 -5.91969397e-03 2.09418729e-01
5.39668322e-01 5.18828690e-01 -2.37337083e-01 5.14073670e-01
-1.44369960e+00 -3.64385769e-02 4.10709053e-01 6.02399707e-01
1.46464765e+00 3.37269068e-01 -1.28281489e-01 8.67400110e-01
5.40299773e-01 6.17415130e-01 7.48575449e-01 -9.63475406e-01
3.95961910e-01 5.57621956e-01 -1.29170880e-01 -9.29664493e-01
-1.71644077e-01 -8.15433145e-01 -1.33197677e+00 3.46778303e-01
2.74320364e-01 -3.50948095e-01 -1.16212046e+00 1.68156457e+00
8.85284543e-02 5.52397430e-01 3.00732285e-01 1.03198338e+00
5.98321199e-01 1.01914489e+00 -3.85275558e-02 -4.55628455e-01
9.48113084e-01 -9.75638747e-01 -7.61847138e-01 -3.52831006e-01
4.64653432e-01 -7.67591119e-01 6.23912394e-01 4.27650213e-01
-8.13618660e-01 -6.84683800e-01 -1.31857264e+00 6.26890361e-02
-5.38621187e-01 3.77723634e-01 4.51755285e-01 3.11237633e-01
-1.15280330e+00 9.97847140e-01 -5.70745289e-01 -3.61347765e-01
2.59022206e-01 3.08407903e-01 -3.07808012e-01 -3.62862013e-02
-1.34595799e+00 8.50630522e-01 7.74046004e-01 4.52572107e-01
-9.59803581e-01 -6.54710293e-01 -8.30144644e-01 -6.10113665e-02
-4.28273380e-02 -7.89438188e-01 6.76687002e-01 -1.71027851e+00
-1.70004511e+00 7.05376983e-01 8.29295889e-02 -4.44696337e-01
2.23377749e-01 6.68872371e-02 -5.10429084e-01 2.24451855e-01
4.21745703e-02 9.78638053e-01 1.11011088e+00 -1.47036922e+00
-6.67840242e-01 -1.63850158e-01 -2.88350165e-01 3.91171724e-01
-2.51306593e-01 -3.66912484e-01 -1.88746408e-01 -8.38622391e-01
5.30313015e-01 -7.07844555e-01 -3.03646147e-01 -1.26881540e-01
-2.41846368e-01 3.95206600e-01 4.06145841e-01 -9.92154419e-01
1.33073914e+00 -2.41516256e+00 4.24068123e-01 4.76754695e-01
3.04445252e-03 1.95277795e-01 -2.02576637e-01 2.25602150e-01
-5.74389756e-01 -2.53604949e-01 -1.01650476e+00 -3.19026887e-01
4.86225784e-02 3.12479019e-01 -8.19175988e-02 7.47451901e-01
1.46585047e-01 5.75048447e-01 -9.73145425e-01 -1.84055552e-01
3.36059421e-01 5.52066147e-01 -4.91693974e-01 5.04443884e-01
-4.19289082e-01 4.67759639e-01 2.25258425e-01 3.93748909e-01
8.27063203e-01 -6.81802779e-02 2.11691279e-02 -5.55075943e-01
-5.07635236e-01 -1.44515093e-02 -1.42999446e+00 1.90389526e+00
-2.91274101e-01 2.27917060e-01 4.97704178e-01 -9.30263340e-01
9.64024782e-01 3.79482567e-01 5.67317486e-01 -6.13257647e-01
1.81048006e-01 4.62790221e-01 1.06260337e-01 -3.12154084e-01
2.74891049e-01 -4.24520105e-01 6.71145797e-01 3.20107728e-01
3.85397404e-01 -3.20275456e-01 1.54920235e-01 -2.10860342e-01
5.33316076e-01 4.11664993e-01 2.44878843e-01 -2.81592786e-01
7.75442064e-01 -1.93244517e-02 3.29569906e-01 5.74621916e-01
5.83797283e-02 8.65053535e-01 1.06651306e-01 -1.64577827e-01
-1.12099731e+00 -8.08259428e-01 -2.09491044e-01 7.52084434e-01
-2.65729010e-01 -6.61411434e-02 -7.50062883e-01 -2.62992799e-01
-3.78195107e-01 7.10870683e-01 -4.91593927e-01 -3.54837961e-02
-2.09491774e-01 -1.35867321e+00 3.97282660e-01 8.54668990e-02
8.33186686e-01 -1.02238107e+00 -6.99679181e-02 3.33997309e-01
-2.56787658e-01 -6.76744759e-01 -1.30933344e-01 7.99508452e-01
-6.04498804e-01 -7.12217569e-01 -9.23847139e-01 -5.33658981e-01
6.55267417e-01 -1.14442565e-01 5.90951085e-01 -1.70813605e-01
1.24432422e-01 -2.69300997e-01 -3.16007406e-01 -2.23674729e-01
-4.61436689e-01 -1.48315430e-01 4.87568835e-03 5.68387628e-01
4.10602510e-01 -8.54353845e-01 -5.46748996e-01 -1.05036043e-01
-1.41121054e+00 1.05688497e-01 7.53005743e-01 8.01394224e-01
6.51326895e-01 4.08445626e-01 2.18777165e-01 -6.26294732e-01
2.24386379e-01 -6.07978523e-01 -6.70339644e-01 -2.48749498e-02
-6.57143712e-01 2.20155045e-01 6.92875862e-01 -1.62804708e-01
-9.16606963e-01 3.27439696e-01 -5.52772224e-01 -7.30094135e-01
-1.74887285e-01 6.51967406e-01 -4.17773485e-01 -6.14957744e-03
7.85932541e-01 4.01611567e-01 1.46480978e-01 -3.37002039e-01
4.46862996e-01 8.35847974e-01 6.61460817e-01 -3.59752387e-01
8.88792455e-01 4.47187543e-01 -6.18610047e-02 -7.67664194e-01
-7.40207136e-01 -3.92285824e-01 -7.23731041e-01 -3.05343159e-02
1.10877740e+00 -1.15581691e+00 -1.38690487e-01 7.65465438e-01
-9.92726326e-01 -4.61708993e-01 -5.08357346e-01 8.00371289e-01
-4.35127944e-01 5.27939498e-01 -5.26015341e-01 -8.07120681e-01
-3.12133312e-01 -1.19229257e+00 8.19005728e-01 2.33415917e-01
1.15808420e-01 -9.20888543e-01 1.10443585e-01 -8.06144625e-02
4.67702389e-01 2.13434637e-01 8.54557037e-01 -4.08371985e-01
-3.03206235e-01 -3.24142352e-02 -3.44278842e-01 9.28603292e-01
1.31372333e-01 -3.00379377e-02 -1.56506622e+00 -5.51556014e-02
3.92990947e-01 -1.17758643e-02 1.04101431e+00 5.43794036e-01
9.36703622e-01 -4.14223671e-01 1.55917645e-01 1.11319947e+00
1.72779822e+00 2.08610505e-01 1.01298070e+00 4.42772537e-01
8.42427313e-01 6.02360964e-01 2.82812379e-02 4.23808128e-01
-4.65371879e-03 3.78961205e-01 5.80427587e-01 -3.44034076e-01
-1.35244444e-01 -2.10167825e-01 6.28718913e-01 1.05101001e+00
-3.62196565e-01 -1.65937141e-01 -5.30504704e-01 3.84444088e-01
-1.84149718e+00 -1.05837607e+00 -3.18340808e-01 2.36712289e+00
9.45038915e-01 -3.10464919e-01 -3.59192669e-01 2.80348778e-01
6.32224381e-01 4.31558758e-01 -2.16878518e-01 1.64480729e-03
-6.06748819e-01 4.32742238e-01 7.04826891e-01 7.69538403e-01
-1.26746023e+00 7.87505925e-01 5.38998175e+00 7.94887543e-01
-1.36097419e+00 1.19031206e-01 5.39597631e-01 1.05389386e-01
-3.57696593e-01 -1.25028282e-01 -4.28316593e-01 5.08384824e-01
1.02132499e+00 4.51627225e-01 9.63276029e-01 4.60364074e-01
1.91548020e-01 -2.20030487e-01 -7.71950722e-01 1.09264004e+00
1.67024121e-01 -9.99924541e-01 1.30859405e-01 -5.64183667e-02
9.42045867e-01 1.83026239e-01 -2.25981846e-02 1.07268132e-01
3.04739088e-01 -8.41919243e-01 7.70113826e-01 6.76924467e-01
8.25373113e-01 -6.12460017e-01 9.40376163e-01 3.41415614e-01
-1.08232403e+00 -5.95046729e-02 -6.54798508e-01 -1.07847273e-01
5.75086921e-02 9.55746949e-01 -5.28977156e-01 1.01534939e+00
5.52622378e-01 6.93896294e-01 -3.45207959e-01 6.97969079e-01
-4.93282199e-01 4.85400438e-01 -4.48498666e-01 5.19725382e-01
5.83714068e-01 -8.64140332e-01 4.99131560e-01 1.11876714e+00
5.89001477e-01 1.47792362e-02 -9.40559581e-02 9.10580754e-01
5.95014989e-02 2.10172594e-01 -5.01657963e-01 -2.00481117e-01
3.75373056e-03 1.27939844e+00 -5.37677526e-01 -5.34337282e-01
-1.93854004e-01 1.25843835e+00 3.83199863e-02 5.91550291e-01
-6.95139885e-01 -1.85272142e-01 5.65634668e-01 -1.23943826e-02
3.14755082e-01 2.86389198e-02 -3.36196154e-01 -1.09982991e+00
-3.02673161e-01 -1.01268923e+00 3.71251494e-01 -1.06551838e+00
-1.31613648e+00 8.08699369e-01 -1.91864565e-01 -1.20891726e+00
-1.12017885e-01 -8.05164278e-01 -2.06262529e-01 1.37620449e+00
-1.70125926e+00 -1.07327247e+00 -2.98585236e-01 5.56458414e-01
6.04366213e-02 -1.40337899e-01 8.24626982e-01 6.67568564e-01
-6.44823909e-01 7.16233999e-02 2.52585024e-01 -1.71407275e-02
6.94250941e-01 -1.25242925e+00 -2.89428085e-01 1.29503989e+00
1.72235146e-01 4.43594187e-01 6.55781984e-01 -6.32281184e-01
-9.66606736e-01 -1.34453678e+00 7.92076886e-01 4.95949127e-02
7.69143283e-01 1.65786408e-02 -8.94397676e-01 7.18892336e-01
4.95518744e-01 -8.87112394e-02 6.71767712e-01 -5.44103205e-01
-2.96655819e-02 -3.18497986e-01 -1.08331287e+00 4.69719082e-01
7.29700267e-01 -6.05925858e-01 -3.28538537e-01 3.32919598e-01
5.55446684e-01 -2.24098727e-01 -7.58617580e-01 3.96154523e-01
2.25855201e-01 -1.03672647e+00 7.80327737e-01 -1.28377512e-01
7.20083117e-01 -8.37237358e-01 -5.92948675e-01 -1.59716249e+00
-6.67912185e-01 -3.25033277e-01 -8.04424472e-03 1.14552712e+00
5.96170783e-01 -5.40864289e-01 4.37114239e-01 1.20166950e-01
-4.87080455e-01 -3.07898343e-01 -7.00043857e-01 -2.71302700e-01
-1.84581876e-01 -4.06781644e-01 8.07102799e-01 1.14975965e+00
-3.83480906e-01 1.26926214e-01 -5.74415922e-01 6.15368962e-01
5.97096443e-01 -1.71229064e-01 3.44569594e-01 -8.97357166e-01
-6.18788123e-01 -5.11200070e-01 -3.76778468e-02 -1.02630854e+00
1.52240619e-01 -1.13954341e+00 2.88792372e-01 -1.49245560e+00
-1.00264795e-01 -2.08322302e-01 -5.30505717e-01 5.10990143e-01
-2.27946669e-01 5.21942854e-01 1.76700979e-01 3.23419929e-01
2.00714231e-01 6.81022525e-01 1.05281353e+00 -3.92016977e-01
-1.05169214e-01 -2.27607504e-01 -5.69106758e-01 5.26739955e-01
6.89703107e-01 -3.61806184e-01 -1.87312141e-01 -4.65581387e-01
2.08266601e-01 -3.12752783e-01 2.09021524e-01 -1.28027797e+00
2.03092381e-01 -1.25561729e-02 5.12610912e-01 -3.69336694e-01
2.71497697e-01 -1.08341169e+00 6.26868725e-01 2.98374891e-01
4.32473086e-02 -3.02174836e-01 4.62012663e-02 1.06654279e-01
-3.84653926e-01 -5.85731387e-01 1.05995989e+00 -2.60178745e-01
-6.84456944e-01 3.23450923e-01 -1.98723704e-01 -7.37388611e-01
6.76017106e-01 -2.66946733e-01 -5.45223616e-02 -2.17825696e-01
-6.79308653e-01 -4.58953455e-02 4.74935830e-01 -2.55142540e-01
3.29913855e-01 -1.23842347e+00 -7.60941386e-01 5.46516001e-01
-1.13584556e-01 2.41667897e-01 2.86309123e-01 1.13044083e+00
-8.12131941e-01 6.65544793e-02 -1.12740181e-01 -4.81103629e-01
-4.13658738e-01 7.70194948e-01 7.62063742e-01 -1.55727670e-01
-3.12952101e-01 9.05656457e-01 1.69962436e-01 -6.16396725e-01
1.11296058e-01 -2.58820206e-01 -2.08941087e-01 3.38088125e-01
6.33619368e-01 1.87225953e-01 1.82456851e-01 -7.72328436e-01
6.54023548e-04 3.28805983e-01 6.01536512e-01 -3.25855017e-01
1.51687348e+00 -1.71420127e-01 -6.38699293e-01 4.67531413e-01
1.23998713e+00 -9.66253579e-02 -1.33777463e+00 -1.69818267e-01
-2.11520374e-01 -1.59564719e-01 4.48179543e-01 -7.96129763e-01
-1.36613786e+00 8.79154503e-01 8.39159548e-01 1.61274560e-02
1.39366210e+00 -4.77585763e-01 3.07146043e-01 6.10449165e-03
1.71212293e-02 -8.81239593e-01 -5.49398184e-01 4.57369685e-01
8.08574915e-01 -1.08870459e+00 4.49488461e-02 -8.39892179e-02
-5.24756312e-01 1.14205480e+00 4.33914512e-01 5.76185919e-02
6.13554060e-01 1.86363712e-01 4.58179235e-01 -1.73114508e-01
-6.22872747e-02 -4.72921133e-01 2.10005522e-01 4.74071652e-01
4.14913416e-01 -4.47872020e-02 -3.99321541e-02 3.96462798e-01
-1.67268828e-01 -5.91410920e-02 1.99117333e-01 5.18294930e-01
-5.15384138e-01 -9.48480010e-01 -7.41947234e-01 3.02986473e-01
-1.52065754e-01 -5.68371594e-01 -1.65035322e-01 3.23595881e-01
7.69749105e-01 7.11296320e-01 1.03547066e-01 -3.18851531e-01
-2.66348701e-02 4.73420918e-01 1.89691275e-01 -4.86281037e-01
-5.95253050e-01 3.55973929e-01 -2.85294294e-01 -2.22178951e-01
-7.45021760e-01 -4.82991278e-01 -1.01620042e+00 -1.53782055e-01
-3.13662946e-01 2.19252944e-01 6.67128503e-01 9.08867002e-01
-2.96366010e-02 5.32383621e-01 6.37114286e-01 -1.14584076e+00
-3.04766566e-01 -1.34145463e+00 -9.53842163e-01 4.03969258e-01
2.54513830e-01 -3.17455322e-01 -7.02339768e-01 2.43924335e-01] | [10.078027725219727, -2.0335850715637207] |
1b2b0aeb-1729-43b1-8e39-6996f8e94535 | feature-less-stitching-of-cylindrical-tunnel | 1806.10278 | null | http://arxiv.org/abs/1806.10278v1 | http://arxiv.org/pdf/1806.10278v1.pdf | Feature-less Stitching of Cylindrical Tunnel | Traditional image stitching algorithms use transforms such as homography to
combine different views of a scene. They usually work well when the scene is
planar or when the camera is only rotated, keeping its position static. This
severely limits their use in real world scenarios where an unmanned aerial
vehicle (UAV) potentially hovers around and flies in an enclosed area while
rotating to capture a video sequence. We utilize known scene geometry along
with recorded camera trajectory to create cylindrical images captured in a
given environment such as a tunnel where the camera rotates around its center.
The captured images of the inner surface of the given scene are combined to
create a composite panoramic image that is textured onto a 3D geometrical
object in Unity graphical engine to create an immersive environment for end
users. | ['Shaohui Foong', 'Karianto Leman', 'Ramanpreet Singh Pahwa', 'Wei Kiat Leong', 'Minh N. Do'] | 2018-06-27 | null | null | null | null | ['image-stitching'] | ['computer-vision'] | [ 8.21137726e-01 -4.05450165e-01 2.64924675e-01 2.53750116e-01
2.88586766e-01 -1.07921767e+00 4.96222585e-01 -4.38266605e-01
-2.40416273e-01 2.69227445e-01 -2.04346672e-01 -4.21892822e-01
2.54264683e-01 -7.67716229e-01 -5.46386361e-01 -4.74784851e-01
2.48090431e-01 4.33526561e-02 2.04714596e-01 -1.29774764e-01
2.72184819e-01 5.01540363e-01 -1.55866599e+00 -2.80806392e-01
4.07990366e-01 5.08659720e-01 6.27284110e-01 9.92173016e-01
4.98556793e-01 2.33004287e-01 -5.64108968e-01 -3.18908006e-01
8.34227562e-01 -2.63619035e-01 -3.11839819e-01 8.09229136e-01
5.34880400e-01 -6.81130528e-01 -4.96612847e-01 1.19742167e+00
-1.97440237e-01 2.68146005e-02 2.44203061e-01 -1.13902557e+00
2.02594977e-03 -1.96425065e-01 -1.08381736e+00 -2.83973277e-01
9.75506961e-01 -2.14015381e-04 3.02663475e-01 -5.41797996e-01
1.01586354e+00 7.15584695e-01 3.82267326e-01 -1.97611842e-02
-1.13601816e+00 -4.38699663e-01 -3.96910757e-02 -3.10785264e-01
-1.30665743e+00 -2.03607976e-01 1.05498993e+00 -1.74734086e-01
5.39043486e-01 5.09057224e-01 9.82053518e-01 6.32041633e-01
6.89896882e-01 1.30688503e-01 8.49808872e-01 -5.98896682e-01
1.23543758e-02 -3.20991948e-02 -2.66925126e-01 5.37843049e-01
5.67332506e-01 2.66772270e-01 -1.99737966e-01 -1.29230961e-01
1.29888189e+00 4.80696380e-01 -8.38948488e-01 -1.01522481e+00
-1.69206524e+00 3.64595920e-01 1.84111103e-01 1.28824323e-01
-1.57183796e-01 -2.64634073e-01 -9.95408520e-02 3.68470937e-01
3.14931851e-03 4.96436507e-01 -6.04668334e-02 -1.01642564e-01
-8.44563246e-01 2.63361096e-01 7.27412522e-01 1.07835972e+00
7.02067435e-01 1.70451596e-01 9.12731111e-01 3.33615303e-01
8.57827812e-02 6.83080733e-01 1.16855487e-01 -1.08768165e+00
5.17955363e-01 5.81804395e-01 2.59180874e-01 -1.26936877e+00
-4.20865975e-02 -3.36368456e-02 -6.03763819e-01 6.97201729e-01
3.21656525e-01 -3.34388822e-01 -6.17381155e-01 1.04001665e+00
5.68178177e-01 3.95431459e-01 1.16470672e-01 1.09533942e+00
3.49931061e-01 5.78135431e-01 -8.39267015e-01 -1.95809305e-01
1.42374802e+00 -7.03227818e-01 -6.51955128e-01 -4.83856499e-01
3.48578751e-01 -1.14242899e+00 7.26104558e-01 7.33636856e-01
-9.45313156e-01 -4.02501822e-01 -1.48733294e+00 9.77031142e-02
-1.77524418e-01 3.65551487e-02 1.60327777e-01 7.00304866e-01
-7.57256329e-01 2.20604375e-01 -7.11475015e-01 -3.48400295e-01
-3.85120779e-01 2.26249024e-01 -9.04108346e-01 -3.04948002e-01
-6.87075913e-01 7.40086496e-01 1.77549750e-01 -3.59782986e-02
-7.62077987e-01 -3.96172941e-01 -1.00894117e+00 -3.03690165e-01
5.34460843e-01 -9.24676359e-01 1.03074181e+00 -7.54458606e-01
-1.62839484e+00 8.42576206e-01 3.73331569e-02 -5.93127720e-02
5.36331177e-01 -3.53576243e-01 -3.93305153e-01 3.17462236e-01
-1.55755103e-01 6.57433495e-02 1.08679104e+00 -1.57178319e+00
-4.64740455e-01 -6.47671580e-01 3.65232944e-01 8.77741396e-01
3.66663992e-01 -2.25127563e-01 -5.08970916e-01 -4.16905165e-01
6.20910645e-01 -1.09192967e+00 -1.84294298e-01 -1.79789558e-01
-4.17318404e-01 1.05465674e+00 1.95860469e+00 -3.97328436e-01
1.01262295e+00 -2.07357192e+00 3.07088643e-01 5.28329909e-02
1.25544652e-01 2.03139827e-01 2.41729081e-01 8.46590459e-01
-2.56285608e-01 -3.67869735e-01 -2.20619053e-01 -2.01029360e-01
-8.95909548e-01 3.50864157e-02 -2.33548775e-01 7.56333649e-01
-7.52055883e-01 2.99913496e-01 -9.23513889e-01 -1.67241022e-01
7.73971260e-01 4.04245406e-01 -4.00348276e-01 3.52650821e-01
3.01880419e-01 6.56299233e-01 -3.88988167e-01 4.84004498e-01
1.02800381e+00 2.73957551e-01 2.80903935e-01 1.99989509e-02
-4.56982583e-01 -2.30529547e-01 -1.50883043e+00 1.77350390e+00
-4.50183183e-01 9.04659331e-01 3.07482630e-01 -4.94467914e-01
9.05515432e-01 3.73297364e-01 4.78100926e-01 -1.59168467e-01
2.73281038e-01 -1.59477413e-01 -4.46172744e-01 -4.75550383e-01
1.00799930e+00 -5.02310582e-02 1.20298624e-01 5.15058100e-01
-4.42913771e-01 -8.80346835e-01 -8.92272815e-02 2.42141634e-01
9.70704854e-01 1.84105590e-01 5.03763974e-01 3.02610826e-02
2.66795933e-01 1.57569900e-01 -9.02548898e-03 1.89175293e-01
2.46354505e-01 1.05870044e+00 1.74445882e-01 -5.05282223e-01
-1.39550436e+00 -1.07315755e+00 -1.73432082e-02 4.10868563e-02
7.69128144e-01 -4.48211968e-01 -9.27321255e-01 -2.05815881e-01
-4.24599618e-01 5.63619971e-01 -3.63789201e-01 -3.67363617e-02
-4.50842083e-01 -9.94481519e-03 -1.18930288e-01 -1.93896145e-01
9.16168869e-01 -3.41820210e-01 -1.62379360e+00 -1.42098200e-02
-2.65692949e-01 -1.24983346e+00 -5.00852227e-01 -2.80923158e-01
-1.00006926e+00 -1.53865814e+00 -7.15784431e-01 -5.79163671e-01
9.48888659e-01 1.37109160e+00 6.97671771e-01 -3.73713881e-01
-2.10450664e-01 7.20113754e-01 -3.53687495e-01 -1.03836454e-01
-5.08684032e-02 -6.24385953e-01 5.03139980e-02 8.92711207e-02
-2.14822069e-02 -5.35389006e-01 -6.60911918e-01 6.25224173e-01
-1.15129375e+00 6.21551871e-01 2.18946263e-02 6.39347970e-01
1.14321120e-01 5.69982529e-01 -6.87666774e-01 -5.53911805e-01
2.41192639e-01 -1.05875716e-01 -7.95015752e-01 1.53181240e-01
1.78132921e-01 -6.55420721e-01 5.32252610e-01 -4.30511951e-01
-1.09588647e+00 3.22770029e-01 7.89301693e-01 -6.87757075e-01
-1.19671747e-01 3.22201103e-01 -1.85164779e-01 4.05992195e-02
3.71088237e-01 1.66218892e-01 1.25260442e-01 -1.71125889e-01
9.09250528e-02 6.47206485e-01 8.62481594e-01 7.13545904e-02
1.08942831e+00 9.51997221e-01 1.44168306e-02 -1.24231386e+00
-1.29202425e-01 -4.03343350e-01 -1.08555555e+00 -4.95534092e-01
9.23218548e-01 -7.65049219e-01 -6.13553703e-01 5.84235787e-01
-1.14107752e+00 -1.28136113e-01 1.62426278e-01 7.38694489e-01
-5.23283362e-01 5.93898833e-01 -7.90377427e-03 -5.90935409e-01
-8.78803153e-03 -1.21709514e+00 1.26399565e+00 3.65995049e-01
-1.98620394e-01 -9.95500028e-01 3.21693271e-01 4.02364790e-01
-8.94003510e-02 6.23357356e-01 3.47877204e-01 5.52527428e-01
-6.47638559e-01 -4.81232613e-01 2.22269058e-01 -4.15403880e-02
4.74237025e-01 4.81060505e-01 -8.15562308e-01 -3.66016477e-01
4.76415038e-01 2.09585935e-01 1.17829405e-01 5.24066925e-01
5.43876946e-01 -6.49039075e-02 -5.45017481e-01 7.28103280e-01
1.62787271e+00 6.61016703e-01 7.96088755e-01 4.31279004e-01
6.10885680e-01 4.78499115e-01 7.32905209e-01 3.96215618e-01
3.12191565e-02 9.29308355e-01 5.28934836e-01 -2.29836062e-01
2.86189973e-01 -4.35447395e-01 2.85455912e-01 5.47653377e-01
-2.86069185e-01 -3.06168318e-01 -7.77858377e-01 4.77516726e-02
-1.31180942e+00 -8.86049271e-01 -2.79376805e-01 2.79147363e+00
2.53470466e-02 -2.26730376e-01 -4.03489739e-01 3.35903168e-01
7.65283048e-01 2.96166778e-01 -2.24040210e-01 -2.75844127e-01
1.08917236e-01 -2.72230685e-01 7.54221201e-01 6.26250327e-01
-8.70494425e-01 7.75244832e-01 5.87552404e+00 8.97685066e-02
-1.55568790e+00 -4.81424659e-01 -5.14448760e-03 1.05289169e-01
-1.05453394e-01 5.53545296e-01 -9.55255702e-02 2.23907739e-01
3.81567448e-01 -2.57759541e-01 4.86889035e-01 6.46056056e-01
2.45858297e-01 -9.41048741e-01 -7.51437128e-01 1.29539871e+00
1.38667896e-01 -8.41901481e-01 -9.77732018e-02 4.20679837e-01
8.45137775e-01 -2.90761143e-01 3.83138150e-01 -5.74157834e-01
1.20591767e-01 -7.22522914e-01 5.10923624e-01 2.68596560e-01
1.01353264e+00 -4.12332952e-01 2.88968146e-01 5.33071041e-01
-1.22299385e+00 1.93405107e-01 -3.25239062e-01 -4.89726096e-01
6.22043848e-01 1.25813961e-01 -1.14839172e+00 7.68923998e-01
4.03374493e-01 5.07275999e-01 -3.22749525e-01 9.52264011e-01
-2.95288842e-02 -1.28814876e-02 -3.94413292e-01 5.96701562e-01
3.53086323e-01 -1.03612542e+00 1.03841615e+00 3.08486372e-01
7.65022397e-01 4.47300196e-01 -1.62758172e-01 4.10604835e-01
3.74533683e-01 -2.27814347e-01 -1.64850473e+00 2.16955796e-01
6.47385046e-02 1.26630259e+00 -1.08657682e+00 -2.40331337e-01
-5.48258603e-01 1.25981605e+00 -4.15605187e-01 5.93203664e-01
-7.33755052e-01 -4.77959752e-01 4.06078100e-01 2.03589037e-01
1.27411678e-01 -7.44157076e-01 -8.38268846e-02 -1.48792338e+00
-5.35433507e-03 -7.79535055e-01 -6.83512613e-02 -1.41906035e+00
-7.51761021e-03 7.12122083e-01 1.65713832e-01 -1.98071945e+00
-2.19510086e-02 -4.39407974e-01 -7.08148301e-01 5.83478272e-01
-6.61447942e-01 -9.31624770e-01 -6.86293364e-01 5.93130946e-01
7.52306879e-01 -5.91063797e-02 9.04285848e-01 -1.76688924e-01
-3.12038362e-01 -2.58046776e-01 2.17141628e-01 -1.31337479e-01
6.50048196e-01 -8.36710513e-01 2.03540131e-01 1.19349051e+00
2.66107529e-01 8.01669359e-01 9.20048773e-01 -7.66248822e-01
-1.75017154e+00 -4.84687269e-01 2.59411126e-01 -5.58688581e-01
2.34933108e-01 -4.31204200e-01 -4.66754735e-01 7.90753901e-01
5.16367972e-01 -1.04706243e-01 4.50434715e-01 -5.78770757e-01
-1.43474638e-01 8.27734992e-02 -1.07485867e+00 9.71458435e-01
5.92360437e-01 -4.34250444e-01 -5.38179517e-01 -1.04018271e-01
2.93955505e-01 -9.43681657e-01 -4.91048127e-01 1.56334251e-01
9.94916320e-01 -1.28524089e+00 7.51401961e-01 -6.80305660e-02
5.40742159e-01 -7.44666994e-01 -2.56907016e-01 -1.37124813e+00
2.18734503e-01 -1.03241765e+00 4.97332752e-01 4.61547136e-01
-7.46840015e-02 -7.07477450e-01 6.36805832e-01 5.41286469e-01
1.41499177e-01 -2.36953124e-01 -6.17587447e-01 -4.23794627e-01
-7.06584096e-01 -1.68565363e-01 4.50860620e-01 9.73342121e-01
2.92156786e-01 2.05541179e-01 -5.42433321e-01 5.46163440e-01
5.65837264e-01 1.62733421e-01 1.36454332e+00 -9.74937916e-01
6.99994620e-03 1.16738044e-01 -5.49752653e-01 -1.16711390e+00
-4.70823079e-01 -2.86051631e-01 -6.04747951e-01 -1.01283956e+00
1.57478109e-01 -1.14076346e-01 6.99157000e-01 -4.25599217e-01
2.20089823e-01 3.15798610e-01 2.28839785e-01 2.64228344e-01
1.10188857e-01 1.87967584e-01 1.51074862e+00 3.40174109e-01
-4.71772969e-01 2.58652419e-01 -3.50437671e-01 9.46896017e-01
6.31829858e-01 -9.01101604e-02 -7.05220461e-01 -5.27141333e-01
1.16885580e-01 8.46501231e-01 2.24503353e-01 -1.11874366e+00
2.89009929e-01 -1.88499913e-01 3.47243935e-01 -7.90784419e-01
6.76337838e-01 -1.46653974e+00 9.55808043e-01 4.72340733e-01
2.68258601e-01 6.04642332e-01 1.08930156e-01 4.93605018e-01
-3.44832599e-01 -3.80607784e-01 5.58425248e-01 -2.47923508e-01
-2.30953842e-01 1.27274334e-01 -3.16594601e-01 -5.60382962e-01
1.61349773e+00 -1.04822123e+00 -1.99012145e-01 -7.47131050e-01
-2.29878098e-01 -2.63870329e-01 1.45634127e+00 2.48631299e-01
9.59345579e-01 -1.01910388e+00 -2.53400058e-01 8.61241221e-01
-1.26491502e-01 2.06742764e-01 5.48415959e-01 4.88558173e-01
-1.41122341e+00 4.10051018e-01 -3.94199997e-01 -7.66261935e-01
-1.83761716e+00 7.10290015e-01 1.41306356e-01 1.43050075e-01
-9.90784466e-01 1.43873766e-01 7.36133337e-01 -2.38389194e-01
-2.56462336e-01 -4.07631874e-01 -1.02095231e-01 -3.40659529e-01
6.05119944e-01 3.19386750e-01 -8.52041617e-02 -8.80740583e-01
2.46516578e-02 1.02663648e+00 1.34984881e-01 -6.05542839e-01
1.06224585e+00 -3.98725450e-01 2.85203243e-03 2.55963475e-01
1.16401374e+00 6.60899758e-01 -1.31253612e+00 -4.12858799e-02
-6.91323638e-01 -1.18408763e+00 5.85272722e-02 -1.86225668e-01
-9.04473424e-01 7.83553958e-01 2.84745961e-01 3.24267074e-02
1.21239519e+00 -5.30315936e-01 5.20362437e-01 3.27300906e-01
8.17549288e-01 -6.47464991e-01 -5.36091663e-02 3.49382043e-01
6.82672739e-01 -7.04912186e-01 3.78723025e-01 -5.21132290e-01
-9.29781854e-01 1.50528014e+00 2.58331597e-01 -1.91335678e-01
4.92296159e-01 4.66045201e-01 1.85486719e-01 -2.05227286e-01
-4.28471148e-01 2.87056327e-01 4.93446104e-02 7.45712161e-01
1.06504202e-01 1.10574311e-03 2.66837835e-01 -4.29231465e-01
-6.57031119e-01 -3.36246967e-01 1.25707889e+00 1.15163064e+00
-3.24407607e-01 -8.38977695e-01 -8.65327060e-01 -1.43752709e-01
-2.29612350e-01 2.86467940e-01 -3.61641884e-01 9.60918963e-01
-5.82745112e-02 8.33499968e-01 2.42647484e-01 -2.90894657e-01
5.21948874e-01 -3.52868199e-01 6.01704180e-01 -5.11366069e-01
-8.85729864e-02 3.03806692e-01 -1.89708546e-01 -4.84915137e-01
-3.78146827e-01 -7.73161054e-01 -7.79252231e-01 -3.12860847e-01
-2.64513761e-01 -1.50848359e-01 7.05900669e-01 4.73473817e-01
-1.01370670e-01 9.75698903e-02 9.54380691e-01 -1.06919682e+00
2.08288506e-01 -4.67420638e-01 -8.12148273e-01 5.31386547e-02
6.67709053e-01 -7.23181486e-01 -2.75868088e-01 5.34011185e-01] | [9.17900562286377, -2.4804835319519043] |
7d5413fe-531e-44a0-a707-798e661e4eed | text2human-text-driven-controllable-human | 2205.15996 | null | https://arxiv.org/abs/2205.15996v1 | https://arxiv.org/pdf/2205.15996v1.pdf | Text2Human: Text-Driven Controllable Human Image Generation | Generating high-quality and diverse human images is an important yet challenging task in vision and graphics. However, existing generative models often fall short under the high diversity of clothing shapes and textures. Furthermore, the generation process is even desired to be intuitively controllable for layman users. In this work, we present a text-driven controllable framework, Text2Human, for a high-quality and diverse human generation. We synthesize full-body human images starting from a given human pose with two dedicated steps. 1) With some texts describing the shapes of clothes, the given human pose is first translated to a human parsing map. 2) The final human image is then generated by providing the system with more attributes about the textures of clothes. Specifically, to model the diversity of clothing textures, we build a hierarchical texture-aware codebook that stores multi-scale neural representations for each type of texture. The codebook at the coarse level includes the structural representations of textures, while the codebook at the fine level focuses on the details of textures. To make use of the learned hierarchical codebook to synthesize desired images, a diffusion-based transformer sampler with mixture of experts is firstly employed to sample indices from the coarsest level of the codebook, which then is used to predict the indices of the codebook at finer levels. The predicted indices at different levels are translated to human images by the decoder learned accompanied with hierarchical codebooks. The use of mixture-of-experts allows for the generated image conditioned on the fine-grained text input. The prediction for finer level indices refines the quality of clothing textures. Extensive quantitative and qualitative evaluations demonstrate that our proposed framework can generate more diverse and realistic human images compared to state-of-the-art methods. | ['Ziwei Liu', 'Chen Change Loy', 'Wayne Wu', 'Haonan Qiu', 'Shuai Yang', 'Yuming Jiang'] | 2022-05-31 | null | null | null | null | ['human-parsing'] | ['computer-vision'] | [ 4.27253664e-01 1.38309106e-01 1.73844665e-01 -2.37047315e-01
-3.48841369e-01 -3.33988905e-01 4.50820386e-01 -3.61839950e-01
2.28605852e-01 5.56958914e-01 3.00995439e-01 4.58396226e-01
3.01723868e-01 -1.18830752e+00 -9.20201659e-01 -8.31854403e-01
3.90677094e-01 7.76421547e-01 2.23102584e-01 -4.19634074e-01
-2.55634069e-01 3.55744779e-01 -1.81102228e+00 6.08482003e-01
8.10209692e-01 1.09167194e+00 4.89609301e-01 6.94342732e-01
-1.68182477e-02 5.31035423e-01 -3.93577725e-01 -4.82115477e-01
3.16580832e-01 -8.09417665e-01 -2.14123592e-01 8.49026084e-01
3.58577728e-01 -3.69924903e-01 -9.33101028e-02 1.10284448e+00
4.38169599e-01 4.20681294e-03 8.15128922e-01 -9.04870987e-01
-8.71806979e-01 3.77097845e-01 -4.76554573e-01 -5.43753028e-01
3.45175892e-01 3.17868590e-01 7.63854623e-01 -8.27672780e-01
9.04646993e-01 1.45702791e+00 2.86536455e-01 7.46568143e-01
-1.26805151e+00 -5.46834648e-01 1.31500885e-01 -2.59344935e-01
-1.42664325e+00 -3.71398926e-01 9.03450847e-01 -5.49317598e-01
1.72429919e-01 3.08588207e-01 1.08460844e+00 1.34559214e+00
3.16470712e-01 7.22236156e-01 1.29269910e+00 -2.98291743e-01
3.30734909e-01 2.89147705e-01 -4.81689930e-01 9.73936319e-01
4.29682210e-02 1.67563125e-01 -5.13798714e-01 2.79892180e-02
1.35357225e+00 -3.60615104e-02 2.60003973e-02 -5.58796823e-01
-1.25035679e+00 7.70207286e-01 3.13921899e-01 -3.48577239e-02
-7.36996770e-01 -3.36832702e-02 5.67775592e-02 -1.29786164e-01
3.38201135e-01 5.88101149e-02 5.68637140e-02 4.12901729e-01
-1.03973532e+00 5.29705405e-01 7.60107219e-01 1.29525542e+00
6.86176181e-01 2.48428792e-01 -6.00700676e-01 1.05342007e+00
3.51254463e-01 8.95426393e-01 1.47790359e-02 -8.41979623e-01
3.36995125e-01 4.59090799e-01 2.61753440e-01 -1.26863134e+00
8.79610926e-02 -4.23048913e-01 -1.31649029e+00 1.18390001e-01
2.20542446e-01 -4.48848829e-02 -1.28584135e+00 1.58545983e+00
5.93990564e-01 -2.46816754e-01 -2.08286777e-01 1.22682393e+00
6.11091018e-01 9.30585623e-01 1.28699332e-01 -6.53388351e-02
1.48404503e+00 -1.09637499e+00 -6.45392895e-01 -1.44958913e-01
-4.88241106e-01 -8.07488799e-01 1.07511842e+00 2.84867853e-01
-1.21533251e+00 -1.11748433e+00 -9.08339620e-01 -2.86098942e-02
-1.05794175e-02 5.07071018e-01 2.35631064e-01 5.31877041e-01
-9.41092610e-01 2.71527439e-01 -7.16430843e-01 -3.96786392e-01
2.85530120e-01 1.02407902e-01 -1.21634146e-02 -2.33778059e-01
-1.08525980e+00 5.06748855e-01 2.36044437e-01 1.75420463e-01
-1.22687066e+00 -2.17173994e-01 -1.09155059e+00 -1.74001798e-01
1.67286724e-01 -1.20019054e+00 6.95971906e-01 -1.12552392e+00
-1.75576699e+00 7.96977222e-01 -1.33772820e-01 -4.99146916e-02
6.61979377e-01 7.52289295e-02 -1.38325170e-01 3.78805250e-02
2.57678658e-01 1.08919179e+00 1.50039124e+00 -1.62786460e+00
-5.95016062e-01 -1.96911767e-01 -1.20237134e-01 4.61712271e-01
3.03136441e-03 -3.28609794e-01 -8.96504402e-01 -9.55373466e-01
-4.00636941e-02 -9.84970868e-01 -3.49256784e-01 1.62813395e-01
-6.87590778e-01 2.74332523e-01 3.49197537e-01 -7.84511685e-01
9.74633276e-01 -2.08980250e+00 5.88247836e-01 3.18959028e-01
1.12721927e-01 -2.61246532e-01 -6.85601383e-02 2.48149514e-01
3.35173935e-01 -3.39841127e-01 -1.75111726e-01 -5.18176794e-01
1.27264455e-01 2.52206862e-01 -1.66013464e-01 1.52068079e-01
2.82266468e-01 8.06477547e-01 -8.04399550e-01 -6.91467464e-01
5.05001307e-01 6.34142578e-01 -7.77620792e-01 4.77164596e-01
-5.12947738e-01 8.93675327e-01 -6.78406060e-01 6.42698705e-01
5.43698192e-01 -2.67073929e-01 2.29130685e-01 -6.93639755e-01
7.07166940e-02 -3.62517267e-01 -1.29452944e+00 1.61120892e+00
-3.19151521e-01 -1.17398165e-01 1.60441235e-01 -5.51617444e-01
1.08546388e+00 1.38041094e-01 2.41374567e-01 -6.12527311e-01
2.10603207e-01 -7.27760047e-02 -2.77853549e-01 -3.44321400e-01
5.43087721e-01 -2.81517565e-01 -3.03651839e-01 1.15206338e-01
-1.07936122e-01 -4.93794590e-01 2.14886233e-01 -1.32655621e-01
4.30901885e-01 6.16885841e-01 8.84716585e-02 -2.23489374e-01
5.10954022e-01 7.48580247e-02 3.60853165e-01 5.37471294e-01
3.33023876e-01 9.10848916e-01 1.36384070e-01 -6.18376911e-01
-1.62489021e+00 -1.26801324e+00 2.02779278e-01 1.07770848e+00
2.62286216e-01 -1.06699087e-01 -1.18282938e+00 -2.47118503e-01
-9.23232362e-02 5.33275008e-01 -8.31705153e-01 6.73588961e-02
-3.82645041e-01 -5.26078105e-01 1.58672735e-01 3.23985457e-01
7.50611484e-01 -1.33897519e+00 -5.95857739e-01 2.49326289e-01
-3.69289488e-01 -1.13048649e+00 -7.34902620e-01 -2.12224051e-01
-5.31508565e-01 -6.37083888e-01 -1.07202697e+00 -9.64205980e-01
1.16559362e+00 -2.66394585e-01 9.86031771e-01 -2.01242402e-01
-2.92234361e-01 1.10458903e-01 -2.83501685e-01 -5.99411726e-02
-7.15351760e-01 -1.11918874e-01 4.61754942e-04 6.54017746e-01
-2.22949728e-01 -3.49087566e-01 -6.00595057e-01 3.42728466e-01
-9.54778552e-01 8.37645054e-01 7.24554718e-01 9.24525976e-01
1.06373334e+00 3.21323574e-01 4.98977564e-02 -9.11860824e-01
4.81920302e-01 -2.61410058e-01 -4.51145411e-01 2.56146491e-01
6.65281294e-03 2.52879858e-01 8.93334270e-01 -7.88427174e-01
-1.13536596e+00 3.55941892e-01 -1.81540281e-01 -5.53267598e-01
-3.69453222e-01 -6.32758588e-02 -4.68351960e-01 3.57984692e-01
5.48053503e-01 6.16020083e-01 -1.19548962e-01 -2.75266588e-01
4.90144134e-01 4.77916300e-01 4.45550919e-01 -9.89155769e-01
8.50824773e-01 3.40437442e-01 -2.31602341e-01 -8.64307046e-01
-8.14917147e-01 2.68315613e-01 -7.03265965e-01 -4.87634808e-01
1.31636822e+00 -9.39142644e-01 -3.09867471e-01 6.84568346e-01
-9.42042112e-01 -5.61088741e-01 -4.41881508e-01 2.39973376e-03
-8.30566406e-01 -7.63066858e-02 -8.67506742e-01 -7.46674716e-01
-3.18186462e-01 -1.47949064e+00 1.54055262e+00 1.93635225e-01
-2.31058836e-01 -7.52756774e-01 -2.86035150e-01 5.72456598e-01
3.68002117e-01 5.53925455e-01 1.04638147e+00 7.06422403e-02
-7.27567494e-01 5.50728552e-02 4.95857373e-03 3.05102617e-01
2.47032300e-01 -2.24875763e-01 -6.84096992e-01 -2.15581939e-01
-1.77701890e-01 -3.37358057e-01 2.45806292e-01 4.91158307e-01
1.05584049e+00 -3.78802568e-01 -2.23333642e-01 5.51617742e-01
1.34230447e+00 1.30086884e-01 6.28756106e-01 -1.72903817e-02
9.49352801e-01 6.33599818e-01 5.97149253e-01 4.89404649e-01
5.63961923e-01 6.34909928e-01 -3.21802534e-02 -3.70853126e-01
-3.59090984e-01 -7.97324777e-01 2.28163376e-01 9.67517316e-01
-2.06714019e-01 -2.02433228e-01 -3.76865715e-01 4.32808906e-01
-1.53861928e+00 -7.27885067e-01 2.65668452e-01 2.11169982e+00
1.04349577e+00 1.29436910e-01 1.92841455e-01 -1.72104105e-01
9.69618917e-01 9.10422653e-02 -6.07933044e-01 -2.64485240e-01
7.92252179e-03 1.45494029e-01 1.54726937e-01 3.79845440e-01
-8.32643449e-01 1.16284561e+00 5.78051805e+00 9.60407555e-01
-9.62198317e-01 -1.07549451e-01 8.65538836e-01 1.14535064e-01
-2.99839705e-01 -2.46578112e-01 -9.85240459e-01 6.26301348e-01
4.63079989e-01 1.67102188e-01 6.77091241e-01 7.47694492e-01
2.42727578e-01 2.81962063e-02 -1.04017937e+00 9.43337917e-01
1.81387186e-01 -1.13340485e+00 6.76985621e-01 4.99186665e-02
1.08214450e+00 -7.09382236e-01 1.67919219e-01 1.17655359e-01
4.54852372e-01 -9.28255200e-01 1.29213262e+00 9.47910607e-01
1.10031521e+00 -7.58485794e-01 2.99227178e-01 5.58374524e-01
-1.33676028e+00 2.01227039e-01 -5.02251089e-01 2.93141693e-01
3.60997409e-01 5.77925384e-01 -4.07435864e-01 4.49030161e-01
5.78271985e-01 3.92883033e-01 -3.96843731e-01 3.36742282e-01
-1.83750600e-01 2.90135294e-01 -1.78675890e-01 1.60910562e-02
-1.23125706e-02 -4.45884407e-01 2.62295842e-01 1.08902693e+00
4.25085068e-01 2.08346561e-01 6.07851803e-01 1.24751973e+00
1.02356009e-01 1.50180161e-01 -5.94978273e-01 -2.64639612e-02
2.72409290e-01 1.21504843e+00 -7.79950380e-01 -6.77494943e-01
-6.60279319e-02 1.50370097e+00 1.06704451e-01 3.21936578e-01
-9.96647835e-01 9.17107388e-02 2.36697346e-01 4.87402827e-01
4.13176835e-01 -1.74982935e-01 -2.28398442e-01 -1.02533484e+00
-1.13576628e-01 -1.34768021e+00 -4.47171144e-02 -9.84367251e-01
-1.32909822e+00 9.13261592e-01 4.16476466e-03 -1.17264104e+00
-2.42252782e-01 -3.32979023e-01 2.49442011e-02 1.08504450e+00
-7.97117829e-01 -1.65837801e+00 -4.47925478e-01 7.32112229e-01
9.05040026e-01 -1.61759481e-01 8.28192174e-01 1.82976618e-01
-3.26441795e-01 4.08998877e-01 -3.62609059e-01 2.50043720e-01
3.93934101e-01 -9.68482316e-01 4.37535018e-01 6.43730760e-01
-1.12715028e-01 5.84973156e-01 7.67590284e-01 -1.09939563e+00
-1.32204986e+00 -1.34459329e+00 5.78901768e-01 -1.85572624e-01
2.59724945e-01 -6.50997460e-01 -4.30845857e-01 3.62259239e-01
8.43577236e-02 -2.44395733e-01 2.82090038e-01 -4.96305436e-01
3.56580354e-02 -1.78392574e-01 -1.13792050e+00 8.48863959e-01
9.60824907e-01 -3.73389512e-01 -2.68525690e-01 1.75662056e-01
4.88991827e-01 -7.98668683e-01 -1.01935136e+00 3.56854558e-01
7.21603334e-01 -8.99525166e-01 9.50051785e-01 -1.14760406e-01
8.12724173e-01 -4.65065956e-01 -3.04883659e-01 -1.33488965e+00
-6.51872694e-01 -2.96599865e-01 1.14092134e-01 1.11546338e+00
1.94992378e-01 -2.42624786e-02 7.86061466e-01 4.42024440e-01
1.85841471e-01 -7.10295379e-01 -3.06544274e-01 -3.26754957e-01
-1.99875161e-01 -3.96745093e-02 6.42569661e-01 5.46038508e-01
-6.22086048e-01 5.04287302e-01 -9.23576415e-01 2.79135585e-01
1.03785503e+00 3.35142404e-01 8.81514728e-01 -8.03338826e-01
-4.73227382e-01 -1.36690568e-02 -1.25004724e-01 -1.29136586e+00
-2.24197865e-01 -5.93495786e-01 3.86768013e-01 -1.59809744e+00
3.62977743e-01 -5.03199399e-01 3.69364351e-01 1.75992921e-01
-1.48596317e-01 4.79020953e-01 2.78699011e-01 9.76198763e-02
-3.41414422e-01 6.50454283e-01 1.97229445e+00 -2.37168193e-01
-8.96902084e-02 -1.28067415e-02 -4.49326515e-01 6.18479908e-01
4.44747031e-01 -1.64743170e-01 -5.46009004e-01 -3.73460591e-01
-2.12004799e-02 4.26169455e-01 4.79336619e-01 -1.21013713e+00
8.87717083e-02 -3.12206268e-01 9.55669999e-01 -5.56886256e-01
5.16993284e-01 -6.98202968e-01 7.36797929e-01 3.39519083e-01
-3.94181103e-01 -1.22702830e-01 -1.14876233e-01 4.80101138e-01
-3.22385766e-02 2.35218719e-01 1.07746851e+00 -4.50067610e-01
-3.50685954e-01 6.06403649e-01 -3.34749192e-01 -1.02973402e-01
8.42547953e-01 -2.79902071e-01 3.18863541e-01 -4.74576384e-01
-1.12116683e+00 -1.20248906e-01 6.95821941e-01 6.05439007e-01
8.17255378e-01 -1.60304594e+00 -7.60415554e-01 6.13851845e-01
4.82296012e-02 1.79131493e-01 3.82404953e-01 3.03689778e-01
-5.20245135e-01 -2.59310976e-02 -5.49100935e-01 -7.03970790e-01
-9.45411682e-01 6.52227521e-01 2.51636118e-01 -3.02209556e-01
-6.73945665e-01 5.46686411e-01 7.25836217e-01 -4.32919800e-01
-7.80350622e-03 -3.14038843e-01 -7.03005269e-02 -2.09749311e-01
3.38513851e-01 -1.61347520e-02 -4.53911513e-01 -9.90358055e-01
1.78300843e-01 8.45333755e-01 1.82300314e-01 -2.88023740e-01
1.16290545e+00 -1.83870301e-01 -9.36119482e-02 3.99915546e-01
7.08709836e-01 5.26175760e-02 -1.56731558e+00 -3.44976597e-02
-7.02616811e-01 -3.84492338e-01 -3.83244067e-01 -7.25523293e-01
-1.13418400e+00 7.70321965e-01 5.14180481e-01 -1.49579942e-01
1.25682902e+00 -1.15741782e-01 9.24481928e-01 -2.50861704e-01
9.15548265e-01 -9.85775471e-01 1.84720740e-01 2.91252106e-01
1.03842449e+00 -6.65793598e-01 -2.46704668e-01 -6.10139191e-01
-1.12483907e+00 8.30917299e-01 5.35163283e-01 -3.22977811e-01
3.62539411e-01 4.18767303e-01 -3.23342793e-02 -1.91300496e-01
-4.41419363e-01 -1.71960488e-01 5.58632553e-01 7.40310252e-01
6.81359097e-02 2.59386182e-01 3.98383588e-02 6.49235547e-01
-4.56375211e-01 1.05106026e-01 7.94939175e-02 4.04505670e-01
-4.04003620e-01 -1.08366024e+00 -6.41604066e-01 3.15557450e-01
-1.64762780e-01 -6.05443791e-02 -9.13700610e-02 4.22478765e-01
6.98734403e-01 7.04160571e-01 -6.45218045e-02 -4.25135881e-01
4.43458229e-01 -1.88586578e-01 7.14901984e-01 -7.23290622e-01
-3.85117322e-01 4.15072173e-01 1.56518351e-02 -4.56892341e-01
-2.62540936e-01 -5.52695692e-01 -9.10008848e-01 -1.69998780e-01
1.02480799e-01 1.06816333e-04 3.82551223e-01 7.66252041e-01
1.43852562e-01 6.18688941e-01 5.74146092e-01 -1.16103220e+00
-2.68376470e-01 -8.02853286e-01 -7.99515784e-01 7.71466076e-01
-3.55215371e-02 -6.49775326e-01 4.31147143e-02 7.15792477e-01] | [11.935050010681152, -0.7985972166061401] |
023d14eb-0b1f-4e60-8e12-4363dac01b35 | data-efficient-image-recognition-with | 1905.09272 | null | https://arxiv.org/abs/1905.09272v3 | https://arxiv.org/pdf/1905.09272v3.pdf | Data-Efficient Image Recognition with Contrastive Predictive Coding | Human observers can learn to recognize new categories of images from a handful of examples, yet doing so with artificial ones remains an open challenge. We hypothesize that data-efficient recognition is enabled by representations which make the variability in natural signals more predictable. We therefore revisit and improve Contrastive Predictive Coding, an unsupervised objective for learning such representations. This new implementation produces features which support state-of-the-art linear classification accuracy on the ImageNet dataset. When used as input for non-linear classification with deep neural networks, this representation allows us to use 2-5x less labels than classifiers trained directly on image pixels. Finally, this unsupervised representation substantially improves transfer learning to object detection on the PASCAL VOC dataset, surpassing fully supervised pre-trained ImageNet classifiers. | ['Aaron van den Oord', 'Olivier J. Hénaff', 'S. M. Ali Eslami', 'Jeffrey De Fauw', 'Carl Doersch', 'Ali Razavi', 'Aravind Srinivas'] | 2019-05-22 | null | https://proceedings.icml.cc/static/paper_files/icml/2020/3694-Paper.pdf | https://proceedings.icml.cc/static/paper_files/icml/2020/3694-Paper.pdf | icml-2020-1 | ['self-supervised-image-classification'] | ['computer-vision'] | [ 7.91264355e-01 2.37634584e-01 -2.38275409e-01 -8.28006029e-01
-4.52186495e-01 -6.56038284e-01 9.57908332e-01 -7.37951770e-02
-6.20906115e-01 6.67525947e-01 -1.60894826e-01 -3.04965258e-01
3.90678048e-02 -6.35615945e-01 -9.40439224e-01 -6.13310874e-01
-1.62658274e-01 3.28876257e-01 2.35306129e-01 1.00083232e-01
2.15261266e-01 6.39277697e-01 -2.07898760e+00 8.56118977e-01
3.17489892e-01 1.57538188e+00 6.25581890e-02 8.74715090e-01
-8.73887017e-02 1.05878401e+00 -3.21493655e-01 -2.19263062e-02
3.78288686e-01 -2.70415306e-01 -9.11745071e-01 3.60468365e-02
1.15376914e+00 -6.43382594e-02 -1.65137097e-01 8.53077769e-01
6.16793633e-02 1.60621300e-01 1.15623271e+00 -1.05194211e+00
-1.04583883e+00 3.84018123e-01 1.21271990e-01 4.09996808e-01
2.42517352e-01 4.19753730e-01 1.11044753e+00 -8.75250220e-01
3.76376927e-01 1.04352951e+00 7.93163896e-01 6.94789052e-01
-1.66980910e+00 -6.30422652e-01 2.12239977e-02 3.75225991e-01
-1.18675637e+00 -5.13315916e-01 4.34792727e-01 -7.53090978e-01
1.34407222e+00 2.83264458e-01 5.94048917e-01 1.12188470e+00
1.03524514e-01 6.54479384e-01 1.45939934e+00 -5.78877628e-01
2.96366494e-02 3.27472776e-01 2.08073139e-01 6.89332783e-01
1.22673057e-01 7.53291845e-01 -5.09248853e-01 3.72479290e-01
5.93814313e-01 1.36301890e-01 -1.73219919e-01 -3.20108086e-01
-1.21043730e+00 7.84089267e-01 8.73622477e-01 2.94030666e-01
-2.47115344e-01 4.26614165e-01 2.32402012e-01 7.18030572e-01
3.71644586e-01 9.69664454e-01 -8.14767838e-01 -8.70028138e-02
-6.92563534e-01 -2.64143735e-01 6.64503753e-01 8.32916915e-01
1.01136470e+00 3.76047969e-01 5.97156696e-02 7.68365979e-01
9.58354864e-03 5.90923071e-01 7.03095794e-01 -9.32859421e-01
-1.85473114e-01 5.75999081e-01 -4.04238522e-01 -7.72875369e-01
-4.12343144e-01 -6.46696508e-01 -8.84164453e-01 5.08527994e-01
4.28337038e-01 2.82386541e-01 -1.18679762e+00 1.26897144e+00
-4.12370622e-01 5.67663424e-02 1.29385024e-01 4.77099508e-01
7.27428377e-01 7.09953010e-01 2.03883469e-01 2.08216310e-02
1.04406309e+00 -9.63832200e-01 -1.99355602e-01 -4.14391726e-01
4.43749547e-01 -5.05654752e-01 1.03061247e+00 7.50209272e-01
-4.06527251e-01 -1.10880363e+00 -1.11237514e+00 -8.16090312e-03
-7.42225528e-01 5.82671948e-02 9.31927621e-01 5.53481758e-01
-1.01192951e+00 6.99017882e-01 -6.86328530e-01 -3.47608089e-01
9.33240592e-01 5.37122607e-01 -4.88650709e-01 -1.18766107e-01
-6.04218423e-01 1.13155341e+00 7.56599367e-01 -2.06663415e-01
-1.16369379e+00 -7.40112424e-01 -8.64617825e-01 1.12308621e-01
1.67807366e-03 -1.86000600e-01 1.26455522e+00 -1.66521871e+00
-1.58055079e+00 1.26018167e+00 1.70676127e-01 -8.39470565e-01
1.35317534e-01 -1.77021638e-01 -4.28007007e-01 1.65750206e-01
-3.38387012e-01 1.25005603e+00 1.11826324e+00 -1.34279180e+00
-6.49979770e-01 1.42549679e-01 -1.33705258e-01 -2.00709477e-01
-4.70982432e-01 -3.45015943e-01 3.89258564e-01 -4.65475649e-01
-1.23964353e-02 -1.03970921e+00 -2.20266759e-01 3.36615592e-01
-1.93186924e-01 -3.66654932e-01 6.73541427e-01 -3.00716728e-01
3.50222051e-01 -2.17012239e+00 -9.54198614e-02 1.12078123e-01
2.50094056e-01 4.37695712e-01 -4.82510060e-01 -9.50838551e-02
-3.39243740e-01 -1.22922853e-01 -2.28115126e-01 -9.08748806e-02
-1.56768546e-01 4.50253755e-01 -5.94535112e-01 4.19455320e-01
5.90246975e-01 1.07903373e+00 -7.25932419e-01 -8.67991373e-02
4.18854326e-01 1.64495081e-01 -5.17010629e-01 5.15905380e-01
-2.41137221e-01 3.06404054e-01 2.11152598e-01 5.79893351e-01
3.84404063e-01 -3.40716243e-01 -1.08085826e-01 -1.65967643e-01
-5.21801636e-02 2.20032558e-01 -6.31310761e-01 1.42388070e+00
-5.72498202e-01 1.18919849e+00 -7.03373790e-01 -1.50742722e+00
1.12893832e+00 -1.27691086e-02 1.04936995e-01 -6.76583409e-01
7.15140924e-02 1.02803186e-01 3.31626803e-01 -3.49268287e-01
1.32199749e-01 -2.69242942e-01 2.08262846e-01 1.27475828e-01
7.72163808e-01 -3.43264580e-01 -1.74378991e-01 -3.65587831e-01
9.68439221e-01 -6.78311288e-02 4.90820497e-01 -3.72286171e-01
3.17899883e-01 1.78951308e-01 1.26203269e-01 1.03118026e+00
-1.24598831e-01 5.29766977e-01 1.89604089e-01 -1.06228292e+00
-7.94220567e-01 -1.22473645e+00 -4.92884129e-01 1.49558079e+00
-1.78252444e-01 -5.59781305e-02 -1.01005167e-01 -8.42887223e-01
2.18206942e-01 8.06608737e-01 -8.47330987e-01 -3.64431739e-01
-2.05409274e-01 -5.11047006e-01 3.78430277e-01 7.52517700e-01
3.21735233e-01 -1.28949428e+00 -7.02981174e-01 -1.56894256e-03
5.59069216e-01 -1.07253325e+00 2.46757984e-01 9.57472980e-01
-9.24447060e-01 -1.08600032e+00 -3.32759559e-01 -1.00454462e+00
8.43811333e-01 3.58856544e-02 1.30101895e+00 5.47847450e-02
-5.93974054e-01 6.47578061e-01 -2.74372727e-01 -5.29645681e-01
-5.42627335e-01 1.17692463e-01 1.66144714e-01 -1.14222346e-02
5.68531334e-01 -5.66713333e-01 -3.85560006e-01 2.07251534e-01
-7.88214922e-01 5.21608181e-02 9.68301237e-01 1.17945933e+00
6.04789495e-01 -1.96333483e-01 5.13914883e-01 -9.36316073e-01
2.32822135e-01 -2.08174065e-01 -7.03944147e-01 7.10393190e-02
-7.62387216e-01 3.68653208e-01 1.00345230e+00 -8.52860868e-01
-6.90096676e-01 5.94038367e-01 -3.35825607e-02 -3.00954282e-01
-6.28924489e-01 2.94197410e-01 4.25235510e-01 -5.17660379e-01
1.15972829e+00 4.33002412e-01 1.53448746e-01 -2.76296049e-01
7.11616814e-01 5.93896687e-01 7.93077290e-01 -2.52284557e-01
8.99047673e-01 2.11485565e-01 4.80770767e-02 -8.69694114e-01
-1.02944911e+00 -3.48895133e-01 -1.01689804e+00 -5.13731800e-02
8.21860015e-01 -9.15092766e-01 -6.43839240e-01 2.58533299e-01
-9.42357898e-01 -5.87357402e-01 -6.67013705e-01 5.35907328e-01
-6.12500608e-01 -2.36729532e-01 -3.62169981e-01 -5.71691334e-01
6.26391098e-02 -8.54929686e-01 6.99411511e-01 8.26709494e-02
-2.31782094e-01 -9.47372437e-01 6.07010769e-03 3.06723937e-02
6.64296627e-01 9.10218880e-02 7.85143435e-01 -1.02769136e+00
-6.88682914e-01 -3.72624636e-01 -4.40609664e-01 1.09202266e+00
7.85375237e-02 2.86064632e-02 -1.48848093e+00 -2.82227397e-01
-3.37916374e-01 -8.77522171e-01 1.40578163e+00 1.16352938e-01
1.63090229e+00 -4.09580231e-01 -2.57240683e-01 8.23406935e-01
1.34486830e+00 1.58535063e-01 4.62859631e-01 2.80580729e-01
5.50825179e-01 4.72648531e-01 -1.37664462e-02 -2.47489437e-02
4.23035733e-02 4.49216574e-01 4.83853787e-01 -1.19958229e-01
-2.83250988e-01 -9.66243912e-03 2.26184502e-01 7.54497647e-01
-1.62064016e-01 8.33979845e-02 -9.77090597e-01 4.21297878e-01
-1.28604460e+00 -9.03314114e-01 1.13022417e-01 1.99772775e+00
8.87659490e-01 4.35511231e-01 -2.37167016e-01 8.86058658e-02
2.75874048e-01 -6.52178330e-03 -5.69387197e-01 -5.00740767e-01
-7.85783306e-02 6.60978615e-01 6.97910190e-01 2.92645335e-01
-1.42078674e+00 1.00631297e+00 7.53725004e+00 5.05950630e-01
-1.58407640e+00 -7.07552135e-02 8.71917605e-01 1.98503286e-01
1.15138575e-01 -2.10057855e-01 -6.50021255e-01 1.92076564e-01
1.29881942e+00 5.21284826e-02 5.85341156e-01 1.41973829e+00
-6.38401151e-01 1.03373587e-01 -1.77052963e+00 1.17349589e+00
2.89278597e-01 -1.57560098e+00 1.27791747e-01 -8.24999884e-02
9.23624635e-01 3.02138180e-01 4.95311677e-01 7.63806105e-01
3.67889553e-01 -1.71305168e+00 4.81451780e-01 6.06644213e-01
1.07282221e+00 -3.16153139e-01 6.96513593e-01 2.42900461e-01
-7.56416976e-01 -3.65265578e-01 -7.71734834e-01 -4.57038015e-01
-4.68064755e-01 4.00294036e-01 -1.18605220e+00 -2.18881890e-01
6.71070635e-01 1.09521687e+00 -1.11813021e+00 9.93379593e-01
-3.96622092e-01 9.40390944e-01 -2.65132308e-01 -1.36841863e-01
2.20994338e-01 4.23115164e-01 6.69138040e-03 1.45352960e+00
1.14328191e-01 7.67420651e-03 1.72116742e-01 7.83657312e-01
-4.42105502e-01 -2.07470477e-01 -9.45376039e-01 -1.81857213e-01
1.78140804e-01 1.10928237e+00 -4.54305142e-01 -5.62908292e-01
-3.59770954e-01 8.86535823e-01 6.47242308e-01 2.66295642e-01
-3.37355494e-01 -3.01157653e-01 5.11577487e-01 -1.23862423e-01
4.47587520e-01 -2.71247506e-01 -3.99148822e-01 -1.03262317e+00
-3.48031491e-01 -8.86002958e-01 1.24987870e-01 -6.90243900e-01
-1.52412271e+00 8.86701047e-01 -3.88757139e-02 -1.36547363e+00
-5.28092802e-01 -1.45001733e+00 -3.39259326e-01 5.02283335e-01
-1.58680677e+00 -9.70044792e-01 -5.92778921e-01 3.54366302e-01
4.60766166e-01 -6.98215842e-01 1.28658795e+00 -1.60637870e-01
-1.89062506e-02 5.73003054e-01 2.14881659e-01 2.38746300e-01
6.61489248e-01 -1.37891996e+00 2.39777461e-01 6.24685287e-01
8.13948989e-01 2.67657995e-01 4.11823988e-01 1.46639153e-01
-1.21036351e+00 -1.19240367e+00 4.96340215e-01 -7.07646608e-01
5.61969042e-01 -3.37665647e-01 -9.67954397e-01 8.67206812e-01
3.02162349e-01 7.04112649e-01 8.24323416e-01 2.53753901e-01
-9.53453183e-01 -4.21838880e-01 -9.29381192e-01 1.80657282e-01
9.48674858e-01 -7.86958158e-01 -9.99417663e-01 4.63653624e-01
5.71369052e-01 -7.42140189e-02 -6.81278527e-01 4.23724651e-01
6.33448899e-01 -8.49074721e-01 9.94567573e-01 -8.69561911e-01
6.17379189e-01 6.92497299e-04 -3.12335759e-01 -1.56324029e+00
-4.84862208e-01 -1.55991688e-01 3.94041575e-02 6.73957169e-01
5.23564637e-01 -8.12677383e-01 6.47992790e-01 2.62770206e-01
-1.77310407e-01 -5.80399990e-01 -9.61265862e-01 -1.05431032e+00
1.49593979e-01 -5.48347652e-01 1.62211694e-02 1.04894102e+00
-2.72849143e-01 4.04332072e-01 -1.76678404e-01 -8.02272037e-02
5.81669629e-01 5.65767772e-02 6.74972951e-01 -1.51040781e+00
-3.93588245e-01 -6.10313714e-01 -1.14485276e+00 -1.17643332e+00
3.95967662e-01 -1.24226153e+00 2.99413234e-01 -1.05812705e+00
9.37665999e-03 -5.30250371e-01 -5.97255826e-01 8.87514710e-01
2.19138637e-01 7.31678665e-01 2.91351080e-01 2.24805489e-01
-6.61032498e-01 3.99627507e-01 8.70478570e-01 -5.08996189e-01
1.33108124e-01 -7.76091591e-02 -5.43427885e-01 7.88778305e-01
6.73214734e-01 -5.76444328e-01 -1.82310984e-01 -4.20911968e-01
-1.89692691e-01 -4.72761035e-01 5.32475650e-01 -1.59723461e+00
3.97556089e-02 -2.29544818e-01 9.36550498e-01 -2.97491431e-01
3.68448377e-01 -1.03935683e+00 -2.21386686e-01 5.63046157e-01
-1.00180805e+00 -3.68142635e-01 4.84222889e-01 5.46879947e-01
-3.20363671e-01 -2.94005126e-01 1.09592342e+00 -7.28954449e-02
-1.17731035e+00 1.42213359e-01 -5.36149442e-01 -3.81921679e-02
8.89588773e-01 -2.70258516e-01 -5.05476236e-01 -2.67900109e-01
-6.88077211e-01 -2.78617531e-01 1.03611164e-01 4.72216398e-01
7.90498137e-01 -1.32387030e+00 -6.31774902e-01 4.85800058e-01
4.59477305e-01 -2.05851302e-01 -1.89681619e-01 4.08616662e-01
-6.37149394e-01 5.80554247e-01 -8.06413412e-01 -9.61280704e-01
-1.10104752e+00 6.58722103e-01 4.40957963e-01 7.29262307e-02
-3.29770386e-01 1.26546395e+00 2.32026398e-01 -4.59598482e-01
2.21013755e-01 -6.85378730e-01 -2.03185335e-01 -1.08211830e-01
6.17088437e-01 -1.22127652e-01 -1.29840195e-01 -4.61918652e-01
-2.77204186e-01 4.44938570e-01 -1.93232954e-01 3.85267526e-01
1.50309002e+00 3.44649673e-01 4.42019626e-02 7.56676376e-01
1.65642071e+00 -4.45826173e-01 -1.42910492e+00 -3.23275834e-01
2.12211683e-01 -3.29236060e-01 -4.57205288e-02 -1.07439339e+00
-8.95069659e-01 1.11835635e+00 1.10859621e+00 2.06085637e-01
1.01321948e+00 6.37817532e-02 1.85311779e-01 1.24630249e+00
3.24575931e-01 -7.54311085e-01 4.44334775e-01 6.96094692e-01
9.68064308e-01 -1.77741408e+00 1.31257579e-01 -1.60658672e-01
-4.97383446e-01 1.56441629e+00 7.45738566e-01 -3.97297055e-01
7.10098743e-01 5.68946637e-02 8.71449858e-02 -1.08491136e-02
-9.73942339e-01 -3.30867678e-01 8.89394462e-01 9.68789995e-01
4.08561170e-01 2.84415871e-01 4.02858227e-01 2.72311985e-01
-2.86936879e-01 -7.93663785e-02 3.37675422e-01 6.57953382e-01
-6.63846552e-01 -8.07382584e-01 -6.08375333e-02 9.32713211e-01
-8.49673450e-02 -3.34338367e-01 -2.66788661e-01 7.45908499e-01
3.36921215e-01 6.63586318e-01 4.15315181e-01 -6.35836780e-01
1.40700862e-01 2.99627870e-01 5.81907332e-01 -9.43919897e-01
-4.00858611e-01 -6.26317978e-01 -7.87228793e-02 -4.77742642e-01
-5.44513822e-01 -4.02337551e-01 -8.40173662e-01 1.95808381e-01
-8.64324421e-02 -2.56057113e-01 5.97044647e-01 9.60825264e-01
3.40641104e-02 5.39254904e-01 5.60295582e-01 -1.18135500e+00
-6.97840452e-01 -1.06644011e+00 -3.17140788e-01 5.27784407e-01
4.58489925e-01 -7.20974505e-01 -6.11284435e-01 6.26695156e-01] | [9.593189239501953, 2.460481643676758] |
b81e3576-3a7f-49df-bc74-73ed62149bc4 | a-maximum-entropy-approach-to-massive-graph | 1912.09068 | null | https://arxiv.org/abs/1912.09068v1 | https://arxiv.org/pdf/1912.09068v1.pdf | A Maximum Entropy approach to Massive Graph Spectra | Graph spectral techniques for measuring graph similarity, or for learning the cluster number, require kernel smoothing. The choice of kernel function and bandwidth are typically chosen in an ad-hoc manner and heavily affect the resulting output. We prove that kernel smoothing biases the moments of the spectral density. We propose an information theoretically optimal approach to learn a smooth graph spectral density, which fully respects the moment information. Our method's computational cost is linear in the number of edges, and hence can be applied to large networks, with millions of nodes. We apply our method to the problems to graph similarity and cluster number learning, where we outperform comparable iterative spectral approaches on synthetic and real graphs. | ['Michael Osborne', 'Xiaowen Dong', 'Stefan Zohren', 'Stephen Roberts', 'Diego Granziol', 'Robin Ru'] | 2019-12-19 | null | null | null | null | ['graph-similarity'] | ['graphs'] | [ 1.79951444e-01 1.62949115e-01 -3.08918923e-01 -1.59547225e-01
-6.79794848e-01 -7.01111734e-01 4.17987794e-01 6.17850840e-01
-3.15890431e-01 3.51013988e-01 -1.14950530e-01 -6.16920412e-01
-2.41448402e-01 -7.55465686e-01 -5.07278979e-01 -5.72654665e-01
-7.69242227e-01 6.16299987e-01 5.54044545e-01 -5.27990200e-02
5.15234828e-01 4.41880137e-01 -8.99408340e-01 -2.92963833e-01
8.91732216e-01 7.34470785e-01 -1.10767901e-01 1.11693406e+00
5.62019972e-03 4.29341733e-01 -2.16290057e-01 -3.16155225e-01
3.99594635e-01 -5.15133798e-01 -9.07585621e-01 3.12754989e-01
4.99596834e-01 2.13623330e-01 -4.19219047e-01 1.41195238e+00
4.09331203e-01 2.48645246e-01 1.06065607e+00 -1.39963663e+00
-3.86508733e-01 7.66556680e-01 -7.40622640e-01 4.20359254e-01
2.94407636e-01 -2.30070725e-01 1.17072868e+00 -6.85667217e-01
4.07780975e-01 1.21746171e+00 1.27656412e+00 -4.85666804e-02
-1.79737544e+00 -3.95742238e-01 3.88697535e-02 -8.28758404e-02
-1.72929955e+00 -3.82374018e-01 7.57026851e-01 -7.00254738e-01
3.92834395e-01 4.01381813e-02 5.95531583e-01 4.80373353e-01
-7.47714937e-02 1.47443309e-01 9.69112813e-01 -5.62644184e-01
4.26435411e-01 7.67062604e-02 9.73796844e-02 1.02185333e+00
3.25576156e-01 -1.37816533e-01 -1.41638648e-02 -7.70649076e-01
9.31087554e-01 -2.87097782e-01 -1.89079851e-01 -8.23292851e-01
-9.43383873e-01 1.15830576e+00 2.76566595e-01 1.85842767e-01
1.39945205e-02 5.28245270e-01 4.26157236e-01 7.00942457e-01
7.11133122e-01 2.58586287e-01 -2.38868967e-01 1.47622854e-01
-8.76642406e-01 -8.10039788e-02 1.28041446e+00 1.05725229e+00
1.04792285e+00 -6.85405433e-02 4.64383781e-01 7.67817199e-01
8.84796679e-02 4.18179810e-01 3.72011177e-02 -1.00378263e+00
9.75139588e-02 3.86299759e-01 -4.47441824e-02 -1.37835419e+00
-5.84387481e-01 -1.01018451e-01 -1.04656887e+00 -9.09232944e-02
7.82488823e-01 -4.03295308e-01 -6.68523371e-01 1.56156027e+00
3.33285898e-01 3.85098666e-01 -3.27763408e-01 6.04504645e-01
2.93706357e-01 3.54530483e-01 -9.00923982e-02 -2.71824390e-01
9.31379855e-01 -3.98013711e-01 -2.87514120e-01 -1.44949168e-01
8.40467989e-01 -8.64133477e-01 1.05829608e+00 8.14918876e-02
-1.02199483e+00 -3.03381886e-02 -8.59679878e-01 4.85818267e-01
-2.76021898e-01 -1.44748792e-01 9.06685174e-01 1.00040734e+00
-1.39768910e+00 1.05390263e+00 -7.05443203e-01 -5.62938094e-01
2.47214317e-01 5.74314535e-01 -1.95038795e-01 2.29419768e-01
-9.33385611e-01 6.88606143e-01 6.19852245e-01 -2.89964855e-01
-1.49872527e-01 -6.66210651e-01 -8.16322327e-01 1.73386529e-01
4.04943109e-01 -5.23292005e-01 9.97258365e-01 -9.82519448e-01
-1.29494846e+00 7.38382399e-01 2.06415311e-01 -5.51888406e-01
2.93441117e-01 5.71699619e-01 -2.31337875e-01 3.99589419e-01
-8.34782645e-02 1.48522049e-01 1.11135137e+00 -1.05634546e+00
-2.04877242e-01 -1.35221943e-01 -3.09467286e-01 -1.17292494e-01
-3.69447172e-01 -2.56213039e-01 -2.16201752e-01 -5.96251965e-01
1.76368788e-01 -1.08942318e+00 -6.50903046e-01 -1.94274738e-01
-2.95146585e-01 -1.43610656e-01 4.65929270e-01 -3.14645618e-01
1.24019623e+00 -1.95776415e+00 -1.71357140e-01 1.15487182e+00
4.98276025e-01 -2.71693498e-01 -1.21855950e-02 6.31311834e-01
-1.52856961e-01 2.79380888e-01 -3.04270893e-01 1.99248672e-01
4.05136123e-03 -6.73992857e-02 1.13181382e-01 1.04834282e+00
-3.83088030e-02 5.91352761e-01 -1.05748439e+00 -7.33259141e-01
8.71990323e-02 1.23205356e-01 -5.79490602e-01 -2.87831668e-02
1.13684312e-01 -2.44038388e-01 -3.34863573e-01 1.81171685e-01
5.38329482e-01 -9.48299706e-01 5.13229907e-01 -2.13527799e-01
2.76051402e-01 2.29106192e-02 -1.48465073e+00 1.39918303e+00
-3.47979575e-01 6.35229409e-01 3.63049477e-01 -1.25462580e+00
8.84754777e-01 1.38644755e-01 6.57229722e-01 4.41045202e-02
1.45496354e-01 1.19210958e-01 -1.28615182e-02 -4.67444025e-02
1.41395092e-01 -2.60372907e-01 -7.90788606e-02 6.78929329e-01
-1.31431771e-02 -2.10696578e-01 3.23538363e-01 7.07966328e-01
1.39202583e+00 -7.22583890e-01 6.29923642e-01 -8.67159784e-01
2.33843610e-01 -1.12426154e-01 1.11758858e-01 7.67265141e-01
-4.24613338e-03 1.77217498e-01 1.09630561e+00 -4.90457527e-02
-1.36796355e+00 -1.31760824e+00 5.30898497e-02 1.16246676e+00
9.87855867e-02 -5.45540035e-01 -8.96973312e-01 -6.13952100e-01
2.93238163e-01 1.70671731e-01 -6.04745150e-01 -1.80525154e-01
-3.33465248e-01 -5.27930260e-01 4.40897286e-01 2.98628449e-01
-1.97504878e-01 -4.20569479e-01 2.94166237e-01 1.54129043e-01
3.39738071e-01 -1.10309875e+00 -1.17744005e+00 1.71089306e-01
-9.58902895e-01 -1.32579279e+00 -3.72931540e-01 -8.90629649e-01
1.06575191e+00 2.88986892e-01 1.19821870e+00 3.45285416e-01
-3.35785866e-01 7.12952197e-01 -3.70876603e-02 -5.64498361e-03
-6.19253993e-01 3.21538299e-01 8.78350213e-02 -1.52172774e-01
1.01946346e-01 -1.11354327e+00 -3.98257464e-01 1.93807870e-01
-7.59686828e-01 -3.98396820e-01 4.57188666e-01 7.73498774e-01
1.88523233e-01 2.88437337e-01 6.49534106e-01 -1.41388047e+00
1.18851840e+00 -6.41334116e-01 -9.31178987e-01 1.88049406e-01
-1.00949979e+00 3.30293417e-01 7.68423021e-01 -6.19492948e-01
-3.45041394e-01 3.14290375e-01 4.13895786e-01 -3.08518082e-01
1.91629320e-01 4.66602772e-01 3.55107456e-01 -8.60284686e-01
9.37982500e-01 -8.39778930e-02 3.19182247e-01 -1.25934020e-01
7.38158703e-01 2.85019785e-01 6.19689584e-01 -5.69019675e-01
1.00725555e+00 3.04262280e-01 4.09390897e-01 -1.03687775e+00
-5.92785776e-01 -7.80162990e-01 -7.63908446e-01 -1.42764404e-01
2.39304274e-01 -7.11650610e-01 -1.08957183e+00 -1.39490869e-02
-8.76600921e-01 -3.40739936e-01 -6.30684048e-02 3.03135574e-01
-6.99922740e-01 7.06294537e-01 -7.16044962e-01 -8.12972248e-01
-2.50711381e-01 -4.02173758e-01 8.11411738e-01 -7.36825466e-02
-1.52906030e-01 -1.69338000e+00 2.73758680e-01 -3.81875336e-01
2.49857813e-01 2.41912782e-01 9.51467693e-01 -6.85539961e-01
-3.26704949e-01 -2.70507812e-01 -5.24939060e-01 7.65834674e-02
9.49029252e-02 1.50699452e-01 -4.51303571e-01 -5.10137558e-01
-4.92492348e-01 -1.57167450e-01 7.37994790e-01 6.54654264e-01
1.01105368e+00 -5.82219064e-01 -3.76425296e-01 4.84994173e-01
1.35545659e+00 -5.28371990e-01 2.87582934e-01 -2.53739148e-01
8.22978735e-01 3.65360856e-01 1.13337353e-01 5.23769796e-01
1.12267546e-01 3.22580934e-01 -5.64972218e-03 -2.27483600e-01
3.04515272e-01 -1.55749515e-01 1.36981562e-01 1.08087015e+00
4.70793666e-03 1.09311774e-01 -9.99441087e-01 4.35678810e-01
-1.96072161e+00 -1.14520490e+00 -1.41390190e-01 2.29276657e+00
8.86664808e-01 2.34256119e-01 8.30097258e-01 1.47665599e-02
1.20687687e+00 4.85593602e-02 -3.91688257e-01 -3.03752303e-01
1.88738689e-01 1.11229308e-01 1.27065313e+00 8.43472421e-01
-9.74046946e-01 9.04983878e-01 7.86971760e+00 9.85442340e-01
-7.14861691e-01 -2.44803756e-01 4.65609521e-01 2.93715864e-01
-2.09813565e-01 3.34895313e-01 -1.98148713e-01 3.43648285e-01
1.18316066e+00 -6.98134542e-01 8.77967715e-01 9.35584605e-01
1.49881124e-01 -1.64095704e-02 -1.08404374e+00 1.06746256e+00
-2.77477890e-01 -1.38188958e+00 -3.45638067e-01 2.08689094e-01
7.80414522e-01 -7.87040405e-03 -1.49197221e-01 -2.23242685e-01
8.71082783e-01 -1.12375605e+00 -4.36870083e-02 4.02585000e-01
5.77402174e-01 -1.15116119e+00 4.03617978e-01 8.87543038e-02
-1.53316438e+00 2.79538125e-01 -5.83371282e-01 8.31631944e-02
-5.04761077e-02 1.10167348e+00 -1.30663729e+00 8.94428492e-02
2.95345843e-01 5.64964533e-01 -5.88089943e-01 1.08909416e+00
2.40349293e-01 8.12147617e-01 -6.34140015e-01 -2.71692753e-01
7.33393952e-02 -6.03702545e-01 4.46196824e-01 1.45873189e+00
1.38283268e-01 1.22577194e-02 7.40210295e-01 6.61912620e-01
-1.74363732e-01 4.49809819e-01 -9.17525053e-01 -3.46646726e-01
8.89277935e-01 1.38502896e+00 -1.26126552e+00 -2.55960941e-01
-4.00812775e-01 7.21113622e-01 4.78570282e-01 3.70598555e-01
-2.78524429e-01 -7.54297197e-01 4.65668529e-01 4.46783453e-01
4.22109365e-01 -4.74856913e-01 -5.86875565e-02 -7.64017940e-01
-2.18215123e-01 -6.57517374e-01 5.42875051e-01 -1.23616271e-01
-1.96606576e+00 1.85303539e-01 4.50014398e-02 -8.78609300e-01
-4.10480410e-01 -7.02325225e-01 -7.53057539e-01 5.75940549e-01
-7.42515266e-01 -5.30240953e-01 -4.63446788e-02 6.54140294e-01
-3.41079652e-01 7.46343806e-02 6.15965128e-01 1.24757089e-01
-1.52784526e-01 4.41813797e-01 2.39950851e-01 2.43784726e-01
6.45655930e-01 -1.66890192e+00 6.74569786e-01 6.11133814e-01
1.78787097e-01 5.60449541e-01 9.43407059e-01 -5.75497210e-01
-1.36682165e+00 -9.15631533e-01 4.59122539e-01 -1.03496552e-01
1.39851642e+00 -5.22751093e-01 -9.30001140e-01 5.55115223e-01
-7.39364773e-02 3.25058579e-01 8.14373493e-01 5.37978292e-01
-3.73991668e-01 5.12755960e-02 -1.07571959e+00 6.47923589e-01
1.13254142e+00 -7.35289633e-01 8.24469179e-02 6.98452055e-01
5.02696633e-01 3.35597363e-03 -1.35840166e+00 8.80118832e-02
1.83386654e-01 -7.89437413e-01 9.35212195e-01 -6.90422654e-01
-3.79374087e-01 -1.97600096e-01 1.83798358e-01 -1.47863162e+00
-5.69118500e-01 -1.21018469e+00 -5.24321608e-02 8.02872777e-01
4.58146036e-01 -7.94060469e-01 1.01415157e+00 2.31915712e-01
4.13372725e-01 -5.77731550e-01 -6.62623107e-01 -8.99585664e-01
-7.23883286e-02 -1.62791386e-01 2.28870764e-01 1.10957754e+00
5.19657671e-01 7.16825366e-01 -2.90952116e-01 1.71311378e-01
1.00669718e+00 -1.44980503e-02 8.09587717e-01 -1.74907994e+00
-4.42535937e-01 -6.83192074e-01 -7.91189790e-01 -7.49307454e-01
2.64619321e-01 -1.00758815e+00 -1.44765392e-01 -9.71632302e-01
1.83928356e-01 -4.81446564e-01 9.44301933e-02 6.23662323e-02
-1.48828432e-01 4.45120484e-02 -2.13817850e-01 -2.16063727e-02
-7.70766079e-01 1.49205551e-01 8.59310150e-01 1.84092283e-01
-3.30786914e-01 2.12214202e-01 -5.81196666e-01 1.00672925e+00
8.50380838e-01 -4.96437579e-01 -6.19898021e-01 4.64967221e-01
4.46703196e-01 1.54980212e-01 3.11686575e-01 -7.19115913e-01
4.35230285e-01 -2.51681298e-01 2.09074512e-01 -2.37191468e-01
-9.60825756e-02 -6.32751644e-01 3.48242223e-02 3.01069081e-01
-4.05248493e-01 2.96268940e-01 -2.03047663e-01 9.11770940e-01
1.61882803e-01 -2.45578453e-01 1.14215744e+00 -9.22579169e-02
-1.16390117e-01 6.22163355e-01 -3.09387743e-01 5.46745241e-01
8.15272927e-01 -1.88449264e-01 -1.56829923e-01 -7.85026848e-01
-7.62940407e-01 2.01855630e-01 6.83318555e-01 -1.84920043e-01
3.57950211e-01 -1.58814692e+00 -5.86469412e-01 -1.62313178e-01
1.13692125e-02 -4.31700677e-01 -2.60135442e-01 9.15536284e-01
-4.73202854e-01 -1.56385586e-01 2.02024460e-01 -5.86315155e-01
-1.31472874e+00 7.67984688e-01 2.24486455e-01 -1.27515167e-01
-6.28566444e-01 4.30627674e-01 -1.74056023e-01 -3.10660511e-01
1.29630134e-01 -5.34796752e-02 2.61774063e-01 -2.79572345e-02
2.59929240e-01 6.03086472e-01 -2.19890684e-01 -2.58868515e-01
-2.72533387e-01 5.56555867e-01 -3.25812772e-02 -4.60195467e-02
1.15145922e+00 -2.87441164e-01 -2.47122526e-01 5.51586688e-01
1.50803995e+00 7.42596388e-02 -1.01719761e+00 -4.06333476e-01
4.68096882e-01 -4.66155946e-01 -6.61245137e-02 2.44398579e-01
-8.93133640e-01 3.79026264e-01 2.90079653e-01 9.78566945e-01
8.11670840e-01 2.79351532e-01 5.47587574e-01 6.96815491e-01
2.27843791e-01 -1.37989795e+00 9.73098129e-02 3.35028142e-01
2.47979730e-01 -1.13487983e+00 3.85113180e-01 -8.10671508e-01
-2.08187759e-01 1.30530739e+00 8.94671157e-02 -6.23387456e-01
1.14919734e+00 5.25066674e-01 -2.41295263e-01 -3.28118294e-01
-6.56444311e-01 -2.25004151e-01 4.87902433e-01 7.75856018e-01
5.54332018e-01 2.58555263e-01 -2.14339927e-01 -1.25489399e-01
-3.79943669e-01 -5.48407912e-01 5.65600932e-01 5.51286101e-01
-9.71630394e-01 -1.00106013e+00 -2.82458782e-01 9.04985666e-01
-3.28433841e-01 -2.13209540e-01 -7.01435030e-01 7.46269584e-01
-5.75221300e-01 7.50258148e-01 2.01596823e-02 -2.25747064e-01
2.56321803e-02 -4.41111587e-02 7.05283940e-01 -4.99058932e-01
-5.00984937e-02 -3.85166742e-02 2.17910081e-01 -4.44664687e-01
-3.51206869e-01 -5.49149334e-01 -1.32450211e+00 -1.00471783e+00
-4.91857737e-01 4.00426447e-01 4.26522583e-01 7.00700104e-01
1.08379193e-01 -1.14452923e-02 9.33207452e-01 -7.12570667e-01
-6.24341369e-01 -7.54617989e-01 -1.02356982e+00 3.72427553e-01
1.78094789e-01 -4.84724760e-01 -7.88970292e-01 4.02338728e-02] | [6.97224760055542, 5.396354675292969] |
fa417104-dba7-4900-8e65-a05963ec0d69 | navya3dseg-navya-3d-semantic-segmentation | 2302.08292 | null | https://arxiv.org/abs/2302.08292v2 | https://arxiv.org/pdf/2302.08292v2.pdf | Navya3DSeg -- Navya 3D Semantic Segmentation Dataset & split generation for autonomous vehicles | Autonomous driving (AD) perception today relies heavily on deep learning based architectures requiring large scale annotated datasets with their associated costs for curation and annotation. The 3D semantic data are useful for core perception tasks such as obstacle detection and ego-vehicle localization. We propose a new dataset, Navya 3D Segmentation (Navya3DSeg), with a diverse label space corresponding to a large scale production grade operational domain, including rural, urban, industrial sites and universities from 13 countries. It contains 23 labeled sequences and 25 supplementary sequences without labels, designed to explore self-supervised and semi-supervised semantic segmentation benchmarks on point clouds. We also propose a novel method for sequential dataset split generation based on iterative multi-label stratification, and demonstrated to achieve a +1.2% mIoU improvement over the original split proposed by SemanticKITTI dataset. A complete benchmark for semantic segmentation task was performed, with state of the art methods. Finally, we demonstrate an Active Learning (AL) based dataset distillation framework. We introduce a novel heuristic-free sampling method called ego-pose distance based sampling in the context of AL. A detailed presentation on the dataset is available here https://www.youtube.com/watch?v=5m6ALIs-s20. | ['B Ravi Kiran', 'Anh Duong', 'Léo Lemarié', 'Alexandre Almin'] | 2023-02-16 | null | null | null | null | ['semi-supervised-semantic-segmentation'] | ['computer-vision'] | [ 3.27896208e-01 3.57121587e-01 -3.84304464e-01 -7.71997869e-01
-1.23735976e+00 -7.08362699e-01 6.04576707e-01 9.75216776e-02
-5.63660383e-01 5.74187636e-01 -2.62334764e-01 -2.66561776e-01
7.98765279e-04 -7.94199467e-01 -8.76084268e-01 -5.74732482e-01
5.73623627e-02 1.13949966e+00 5.37412643e-01 -2.79646307e-01
3.88218999e-01 5.84991515e-01 -1.85381758e+00 1.02961227e-01
1.09268737e+00 1.22050941e+00 5.65787613e-01 5.02518177e-01
-3.72110963e-01 4.30251896e-01 -3.45190287e-01 -1.92278385e-01
6.87856317e-01 5.25176600e-02 -9.92977083e-01 2.00029761e-01
7.73841262e-01 -1.40036032e-01 -3.46304961e-02 1.09522784e+00
4.48946089e-01 3.59394461e-01 5.61393499e-01 -1.62450719e+00
5.20125143e-02 2.02414960e-01 -2.48999640e-01 -4.36028689e-02
3.46609093e-02 2.82110035e-01 8.62729549e-01 -8.80854070e-01
1.00954509e+00 1.21109450e+00 7.65814483e-01 5.29423296e-01
-1.05978310e+00 -8.46716225e-01 9.81330276e-02 6.58741474e-01
-1.25296581e+00 -4.51607138e-01 8.71422231e-01 -7.26166368e-01
8.29298258e-01 6.02336563e-02 6.42734706e-01 1.31291080e+00
-1.91053793e-01 1.00563121e+00 9.76870596e-01 -7.45005310e-02
6.52016819e-01 -4.83743548e-02 2.42953435e-01 8.51807415e-01
1.26559108e-01 8.87929741e-03 -3.19268942e-01 9.89232138e-02
2.40794197e-01 -9.93606523e-02 3.18267763e-01 -1.12616837e+00
-1.16935527e+00 9.77081478e-01 5.00674725e-01 -4.16951329e-01
-9.39886048e-02 1.04429327e-01 5.76449931e-01 -1.14915995e-02
7.08824933e-01 3.96095425e-01 -7.53061473e-01 -2.17116252e-01
-7.62143493e-01 5.54482222e-01 4.88736302e-01 1.30609679e+00
1.31614220e+00 -6.07275032e-02 1.72808886e-01 8.97576571e-01
4.03835773e-01 6.65263593e-01 7.59635344e-02 -1.57900894e+00
4.78165358e-01 7.96962202e-01 1.81979656e-01 -6.66765630e-01
-5.82331061e-01 -1.74771577e-01 -2.97142416e-01 5.29361844e-01
2.37452760e-01 9.07479320e-03 -1.38057947e+00 1.36214817e+00
5.98132312e-01 3.56151015e-01 -2.09037010e-02 7.99132824e-01
1.15340352e+00 3.15903127e-01 9.42507908e-02 2.68771559e-01
8.90341997e-01 -1.28995359e+00 -3.84739697e-01 -3.77936721e-01
1.02676260e+00 -3.91352475e-01 9.72764790e-01 3.30562115e-01
-6.60271049e-01 -7.71133125e-01 -1.06999624e+00 -3.11334997e-01
-7.92101264e-01 -1.50334716e-01 4.39485997e-01 4.58206594e-01
-1.01132071e+00 4.20651257e-01 -8.91227603e-01 -3.64935786e-01
1.08383596e+00 1.80903926e-01 -3.71938109e-01 -7.90384114e-02
-1.09941959e+00 6.82677925e-01 5.44268310e-01 -9.92427319e-02
-1.41694951e+00 -5.95999897e-01 -1.07725799e+00 -6.25669956e-01
5.72185755e-01 -2.72793502e-01 1.36856771e+00 -6.68155730e-01
-1.37478256e+00 1.37004220e+00 -4.69823256e-02 -7.71768034e-01
7.59250283e-01 -3.55753273e-01 -2.39605963e-01 1.57507733e-01
4.75953400e-01 1.38997364e+00 5.74845910e-01 -1.42844868e+00
-8.87494445e-01 -4.97808695e-01 7.06708431e-02 3.29069525e-01
2.59596288e-01 -4.22373652e-01 -4.62009877e-01 -1.36340156e-01
3.45366091e-01 -1.29844844e+00 -5.35975337e-01 -7.12235197e-02
-3.60196322e-01 -3.87665093e-01 1.24334407e+00 -4.69562143e-01
4.10874933e-01 -2.02084565e+00 4.61762026e-02 -4.61552627e-02
4.39624429e-01 2.50938445e-01 -1.24076106e-01 -1.15216281e-02
1.14521042e-01 -3.97362709e-02 -7.26782441e-01 -6.77750111e-01
2.91499078e-01 2.72698730e-01 -1.56790838e-01 5.30220509e-01
2.15766966e-01 1.05835783e+00 -1.22634125e+00 -5.72495818e-01
4.85433787e-01 8.55893940e-02 -3.62158239e-01 6.98352680e-02
-5.56181133e-01 6.04546368e-01 -2.74315506e-01 8.95860016e-01
8.26930344e-01 1.03521936e-01 -3.59160781e-01 6.18644021e-02
-1.63832292e-01 1.90568179e-01 -9.35376704e-01 2.31823254e+00
-2.90076613e-01 8.96681607e-01 1.05418958e-01 -1.26143420e+00
1.29925537e+00 -1.59098923e-01 7.55384922e-01 -9.80652809e-01
1.45738319e-01 5.10245204e-01 -3.97425622e-01 -5.18057406e-01
7.42441773e-01 4.49351490e-01 -3.34729582e-01 -6.30299374e-02
1.18944176e-01 -6.14336967e-01 3.71011257e-01 3.06020617e-01
1.04372919e+00 6.64956808e-01 -1.93381444e-01 -4.87044573e-01
4.01375771e-01 7.56986082e-01 7.60092914e-01 6.42307222e-01
-9.37369466e-01 5.90043247e-01 3.80650163e-01 -3.85071993e-01
-1.24150848e+00 -1.09810209e+00 -3.47035617e-01 8.88187170e-01
6.20673358e-01 -2.36427546e-01 -8.71719360e-01 -1.20303166e+00
9.16379541e-02 8.71368289e-01 -5.34493506e-01 3.10651641e-02
-6.12508297e-01 -2.91693181e-01 5.22652268e-01 4.26371515e-01
8.18551958e-01 -1.08498919e+00 -6.85547352e-01 -7.75241032e-02
-2.21540496e-01 -1.37865460e+00 2.09151730e-01 3.43171299e-01
-8.26932847e-01 -1.10914743e+00 -4.70385164e-01 -8.79846394e-01
3.42154652e-01 3.05212557e-01 1.20688200e+00 -5.68271697e-01
-2.75821507e-01 3.84004086e-01 -4.51450884e-01 -7.04286695e-01
-3.46598566e-01 4.87683058e-01 -4.62445728e-02 -3.71508211e-01
6.86275303e-01 -1.43756643e-01 -5.80006361e-01 5.03534973e-01
-4.15268004e-01 2.38747403e-01 3.16505402e-01 4.34903532e-01
9.41067040e-01 -3.52154434e-01 5.19476831e-01 -9.25785720e-01
-1.15873413e-02 -6.06709659e-01 -7.74709702e-01 -3.38256687e-01
-5.17799973e-01 -4.05491918e-01 2.01061547e-01 1.46079421e-01
-8.63722384e-01 5.69195867e-01 -5.01900315e-01 -4.65906262e-01
-7.93868124e-01 -2.25984618e-01 -3.49596113e-01 -1.01546422e-01
7.86479592e-01 -7.30964914e-02 1.18219398e-01 -3.80279779e-01
5.81894100e-01 8.80360126e-01 5.20046592e-01 -3.59013319e-01
6.19729996e-01 7.43064761e-01 1.98182426e-02 -7.40744948e-01
-9.89864647e-01 -7.53291965e-01 -1.08279157e+00 -3.89679551e-01
1.17244172e+00 -1.17538476e+00 -3.75213504e-01 5.28289974e-01
-9.78718698e-01 -9.06530678e-01 -4.41644281e-01 3.61603916e-01
-1.10586131e+00 1.72013313e-01 -1.18326403e-01 -4.26841289e-01
-1.51939452e-01 -1.42896664e+00 1.45122111e+00 -4.22269627e-02
-1.98529482e-01 -8.52649808e-01 1.19868577e-01 9.03249621e-01
1.57920703e-01 5.66406131e-01 3.84472638e-01 -8.72116506e-01
-7.61638880e-01 3.33528258e-02 -2.49224454e-01 4.57793087e-01
-1.17628261e-01 -3.19262564e-01 -9.19837952e-01 -1.64610147e-01
-3.57116669e-01 -5.60893178e-01 1.07894325e+00 1.60522252e-01
1.12892151e+00 3.66731077e-01 -4.15234566e-01 7.23971307e-01
1.30530238e+00 3.13144773e-01 4.40876633e-01 6.54323041e-01
1.11183679e+00 8.06667626e-01 1.23585248e+00 1.68130174e-01
4.69215244e-01 6.59049571e-01 8.80759597e-01 -1.96635887e-01
-2.69706160e-01 -1.83751131e-03 4.51109894e-02 5.85200191e-01
2.97638893e-01 1.10728526e-02 -1.52061296e+00 7.85810173e-01
-1.87594855e+00 -5.71765065e-01 -5.12147725e-01 1.97701585e+00
5.60031712e-01 6.05071664e-01 1.79389656e-01 1.45689264e-01
5.52173853e-01 9.67565998e-02 -9.69436228e-01 -3.61353934e-01
-8.28529745e-02 9.05155838e-02 9.89103556e-01 5.91281235e-01
-1.53798056e+00 1.17127371e+00 5.21719265e+00 9.57394183e-01
-8.02013993e-01 3.54733855e-01 5.95209599e-01 1.26306593e-01
-1.44792110e-01 -8.66665840e-02 -1.07114840e+00 4.73310411e-01
9.05679226e-01 2.54030555e-01 -1.30292833e-01 1.24148142e+00
2.88259685e-01 -1.08729981e-01 -8.22407126e-01 1.10375905e+00
-1.68759413e-02 -1.43892860e+00 -1.58735469e-01 2.99259424e-02
9.87072587e-01 7.63084590e-01 -1.98582575e-01 4.40327793e-01
4.84221220e-01 -7.55504131e-01 1.05814242e+00 3.18302393e-01
8.11403275e-01 -8.66716206e-01 6.47328734e-01 3.44077796e-01
-1.13426793e+00 -1.64800808e-01 -2.59095997e-01 1.18472777e-01
2.81714827e-01 5.15924275e-01 -9.83302176e-01 5.40455222e-01
9.36452389e-01 1.09100640e+00 -6.51928246e-01 1.07893050e+00
-1.22309357e-01 5.27547121e-01 -2.02216685e-01 1.79330662e-01
5.94577014e-01 -3.84097070e-01 8.72188389e-01 1.20573354e+00
1.94660023e-01 -3.17235857e-01 4.45786268e-01 7.16053426e-01
8.52222147e-05 -1.27675205e-01 -7.60575831e-01 1.79372340e-01
6.25457942e-01 1.30452418e+00 -1.05284131e+00 -5.08646905e-01
-1.27213269e-01 8.84279847e-01 9.69453305e-02 1.04753003e-01
-9.49559391e-01 -5.31721175e-01 8.90978158e-01 1.63706973e-01
2.87471235e-01 -6.51125014e-01 -4.93668795e-01 -6.45640254e-01
-3.41264814e-01 -5.28546333e-01 5.52215539e-02 -6.12920463e-01
-7.66089737e-01 3.70836288e-01 1.53563514e-01 -1.35634637e+00
-1.12468883e-01 -5.71989119e-01 3.96856293e-02 4.74440068e-01
-1.50583816e+00 -9.83437240e-01 -8.25947642e-01 3.17464411e-01
1.10102439e+00 -3.14342231e-01 6.02024972e-01 4.64573234e-01
-5.61755538e-01 2.14950204e-01 2.20358506e-01 1.00221343e-01
5.25332630e-01 -1.38483858e+00 9.14878130e-01 5.75679481e-01
-8.01480114e-02 -1.59060687e-01 4.84035462e-01 -6.88184083e-01
-1.08424854e+00 -1.65290833e+00 5.52056789e-01 -7.65795887e-01
4.49809968e-01 -8.59562576e-01 -5.50545454e-01 6.77732348e-01
-1.59854248e-01 -6.28637820e-02 2.90010303e-01 -3.78373027e-01
-1.42992020e-01 -9.36961547e-02 -1.41376877e+00 3.49595845e-01
1.67438853e+00 -2.48998657e-01 -3.25939983e-01 5.39065659e-01
1.19043422e+00 -6.88070536e-01 -4.91016567e-01 8.29911411e-01
1.12437807e-01 -9.76644874e-01 1.07508588e+00 -2.38541886e-01
3.47797871e-01 -3.56727690e-01 -3.35402638e-01 -9.89430547e-01
-3.21821600e-01 -3.03337693e-01 -5.07774502e-02 9.76447940e-01
3.88746858e-01 -5.74974835e-01 1.33638680e+00 5.93014359e-02
-1.01543212e+00 -7.70531416e-01 -9.99764681e-01 -8.82541835e-01
-1.65684819e-02 -7.74721622e-01 4.85784411e-01 9.33085918e-01
-7.42141426e-01 2.67782480e-01 2.63613462e-01 -6.67975619e-02
1.08410490e+00 -8.81819874e-02 9.83446360e-01 -1.47039139e+00
4.01455730e-01 -3.66409451e-01 -8.04662585e-01 -1.15237737e+00
3.40384036e-01 -1.08980858e+00 3.64884824e-01 -1.92402387e+00
-2.87505627e-01 -8.09872448e-01 -2.77941935e-02 4.95361328e-01
3.02145541e-01 6.01615548e-01 3.33682681e-03 2.43852094e-01
-1.03462696e+00 5.60297012e-01 1.13990366e+00 -3.87515426e-01
-1.52914792e-01 -7.85906762e-02 -1.94209993e-01 8.29987586e-01
1.09802079e+00 -3.86954337e-01 -6.54706776e-01 -3.10697168e-01
1.88516900e-02 -5.13586700e-01 3.85474563e-01 -1.21862292e+00
-6.89698607e-02 -1.36512235e-01 2.57974491e-02 -1.25647056e+00
5.45180380e-01 -6.95135891e-01 -5.30730374e-02 4.50218886e-01
-2.66858757e-01 -2.06645012e-01 2.22923353e-01 5.53907692e-01
-2.36008346e-01 -1.78374738e-01 8.81583154e-01 -3.08954358e-01
-1.73533416e+00 5.06263673e-01 -2.11898088e-01 3.44306797e-01
1.31671131e+00 -5.74104548e-01 -1.50052279e-01 1.09422103e-01
-7.40770996e-01 5.39306998e-01 5.68541050e-01 7.41219878e-01
5.41629732e-01 -1.25545108e+00 -7.98255026e-01 1.07029021e-01
5.44892371e-01 7.38841414e-01 2.88238436e-01 6.83217764e-01
-9.64415789e-01 4.56005186e-01 -3.10791403e-01 -1.19914436e+00
-1.10072148e+00 2.23435074e-01 2.41900980e-01 2.45925978e-01
-7.03929365e-01 1.03034234e+00 4.27797548e-02 -1.05435979e+00
3.34749043e-01 -1.47071138e-01 -2.52251297e-01 3.40704590e-01
-2.16243323e-02 7.85589695e-01 4.10408705e-01 -8.79420698e-01
-4.78190124e-01 4.83111829e-01 1.19874038e-01 7.30319545e-02
1.16831875e+00 -3.65810603e-01 1.74075246e-01 6.39354825e-01
1.50036526e+00 -4.85171705e-01 -1.64606953e+00 -1.63262144e-01
2.29683921e-01 -3.84810209e-01 -1.49669275e-02 -5.77579200e-01
-8.56815279e-01 8.04854333e-01 1.08872175e+00 7.35877007e-02
6.74343407e-01 1.22815162e-01 9.81088877e-01 5.66964626e-01
6.18328452e-01 -1.43393636e+00 3.12405545e-02 7.70353258e-01
6.33075714e-01 -1.66996026e+00 -1.64718628e-01 -4.36881870e-01
-6.56895518e-01 6.52222991e-01 6.13193750e-01 -1.53193191e-01
5.50418437e-01 4.09103744e-03 3.09921503e-01 -2.65656918e-01
-3.53108466e-01 -5.35381317e-01 2.65689082e-02 8.23346436e-01
-1.05223656e-01 2.86070287e-01 -6.65049478e-02 7.11817667e-02
-4.66192067e-01 -2.77976662e-01 3.16900462e-01 9.75348651e-01
-7.42898703e-01 -9.21810210e-01 -1.19341336e-01 3.38163167e-01
2.48565614e-01 3.40025276e-01 -4.55110490e-01 6.78600669e-01
7.24562407e-01 8.59364390e-01 5.00633597e-01 -3.93717110e-01
2.88077116e-01 3.22442353e-02 2.65746087e-01 -6.66170299e-01
-2.54910048e-02 -3.76072645e-01 3.94725949e-01 -7.73412406e-01
-5.06761789e-01 -8.76387656e-01 -1.54850149e+00 -2.60547027e-02
5.49319312e-02 -1.09152086e-01 1.15111279e+00 6.62442923e-01
5.88692546e-01 3.89524281e-01 6.79696083e-01 -1.24644303e+00
-5.41562475e-02 -8.49590778e-01 -3.57394516e-01 6.06539369e-01
1.40462160e-01 -1.07655430e+00 -4.30621445e-01 2.57178582e-03] | [8.146706581115723, -2.5464861392974854] |
855b43ab-b630-4603-9710-7eeec2be305f | zero-shot-cross-lingual-transfer-language | 2301.13720 | null | https://arxiv.org/abs/2301.13720v1 | https://arxiv.org/pdf/2301.13720v1.pdf | Zero-shot cross-lingual transfer language selection using linguistic similarity | We study the selection of transfer languages for different Natural Language Processing tasks, specifically sentiment analysis, named entity recognition and dependency parsing. In order to select an optimal transfer language, we propose to utilize different linguistic similarity metrics to measure the distance between languages and make the choice of transfer language based on this information instead of relying on intuition. We demonstrate that linguistic similarity correlates with cross-lingual transfer performance for all of the proposed tasks. We also show that there is a statistically significant difference in choosing the optimal language as the transfer source instead of English. This allows us to select a more suitable transfer language which can be used to better leverage knowledge from high-resource languages in order to improve the performance of language applications lacking data. For the study, we used datasets from eight different languages from three language families. | ['Fumito Masui', 'Michal Ptaszynski', 'Juuso Eronen'] | 2023-01-31 | null | null | null | null | ['zero-shot-cross-lingual-transfer', 'dependency-parsing', 'cross-lingual-transfer'] | ['natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [-2.28800118e-01 -2.63719976e-01 -4.50251520e-01 -6.38878107e-01
-6.97671115e-01 -8.77458870e-01 3.81634563e-01 5.18931210e-01
-7.69438386e-01 6.32769644e-01 2.09094644e-01 -4.44324195e-01
5.92382140e-02 -7.08671629e-01 -5.18487573e-01 -3.74832571e-01
2.46310055e-01 3.33735496e-01 1.96245328e-01 -3.34824175e-01
4.53552067e-01 4.83323485e-01 -1.06080937e+00 4.18722034e-01
1.22008562e+00 4.42433268e-01 3.74603540e-01 7.74116116e-03
-5.14837205e-01 3.89034778e-01 -2.85088390e-01 -6.79980278e-01
2.66290307e-01 -5.02072632e-01 -9.51368988e-01 -2.30272084e-01
1.14247695e-01 2.85528541e-01 6.70693457e-01 1.06455553e+00
4.20291871e-01 5.20156249e-02 8.05497646e-01 -8.76077771e-01
-6.03191137e-01 9.82901454e-01 -3.26974660e-01 1.90573514e-01
4.02488530e-01 -1.39437437e-01 1.19004560e+00 -8.77700448e-01
8.26735198e-01 1.28898454e+00 5.37688196e-01 4.43341941e-01
-1.07347190e+00 -7.30650365e-01 2.21708074e-01 -5.24870194e-02
-1.51497698e+00 -4.81053710e-01 8.84494662e-01 -4.29687977e-01
7.66725361e-01 -2.83938110e-01 1.37337610e-01 8.24138939e-01
9.34053510e-02 4.80648965e-01 1.40793204e+00 -9.93109286e-01
1.85829684e-01 6.46399021e-01 2.01649845e-01 8.98260474e-01
4.36124474e-01 -3.09855431e-01 -6.89929545e-01 -5.27177528e-02
2.38967746e-01 -4.46617365e-01 -1.88495174e-01 -1.84401378e-01
-1.23301804e+00 1.11524236e+00 2.57809937e-01 8.60779464e-01
-3.13839167e-01 -3.33149821e-01 6.09936297e-01 5.39504707e-01
5.56359768e-01 6.49129570e-01 -7.49160945e-01 7.10676685e-02
-5.20556808e-01 -1.73858404e-01 1.02480268e+00 8.03037465e-01
1.04157591e+00 -2.13702068e-01 2.90567335e-02 1.06825399e+00
2.56036967e-01 5.61760843e-01 7.52687693e-01 -5.85572541e-01
7.34199464e-01 7.14879870e-01 -1.39799938e-01 -6.76504970e-01
-3.00261885e-01 -6.02601729e-02 -3.50125968e-01 -1.40900612e-01
5.68736315e-01 -2.98527271e-01 -5.62734306e-01 1.89840734e+00
1.75183803e-01 -4.29160446e-01 3.82057577e-01 6.70765817e-01
5.21295965e-01 4.32679117e-01 2.86401689e-01 -6.02287278e-02
1.69075155e+00 -5.62999249e-01 -2.91518420e-01 -2.58168906e-01
1.23416531e+00 -8.54457140e-01 1.50104451e+00 1.73483193e-01
-7.19608963e-01 -4.76768553e-01 -9.47376311e-01 2.17407122e-02
-6.79402709e-01 3.32884967e-01 6.67641878e-01 6.19387865e-01
-9.84731555e-01 5.42150438e-01 -7.77342498e-01 -8.63416970e-01
-8.68804380e-02 3.19412857e-01 -6.55177295e-01 -1.01382501e-01
-1.30922377e+00 1.06865191e+00 6.32855773e-01 -3.98052156e-01
-1.82995439e-01 -5.03267884e-01 -8.10531020e-01 -1.50631264e-01
1.68950409e-02 -4.83834773e-01 7.97399461e-01 -1.43288887e+00
-1.49970973e+00 1.32598138e+00 -2.15637043e-01 -2.63495177e-01
2.30983287e-01 -9.08429325e-02 -2.85555542e-01 -1.83229223e-01
4.30215061e-01 5.10085702e-01 3.96729410e-01 -8.20624948e-01
-7.55605936e-01 -3.79069865e-01 -6.63546249e-02 2.25152507e-01
-7.85083413e-01 5.34391820e-01 -4.01605844e-01 -6.04397118e-01
-1.38570756e-01 -1.24754024e+00 -2.48300415e-02 -6.38846993e-01
-1.22368239e-01 -3.96778524e-01 1.28058136e-01 -7.02463686e-01
9.18816209e-01 -2.06616545e+00 1.28627300e-01 2.13141963e-01
-3.20167959e-01 4.20216881e-02 -2.08486408e-01 6.53183818e-01
9.37260240e-02 4.49973732e-01 -1.62336618e-01 -2.17944250e-01
-1.83055773e-01 1.18422985e-01 -9.23411474e-02 2.89031029e-01
3.60991567e-01 4.58390266e-01 -7.15204477e-01 -5.50089121e-01
-1.70590729e-01 3.52917820e-01 -5.65033734e-01 2.31422976e-01
7.45487362e-02 7.33424485e-01 -7.70604014e-01 3.54694933e-01
4.12397444e-01 -1.34794964e-02 2.65239596e-01 -1.63410559e-01
-2.70861655e-01 7.53336787e-01 -9.28609133e-01 1.66634619e+00
-1.03827322e+00 3.15572619e-01 -2.10568547e-01 -9.13157582e-01
1.12460911e+00 2.31555164e-01 1.21485643e-01 -5.83362758e-01
1.92461789e-01 5.08349895e-01 3.71792912e-01 -2.08358452e-01
3.46106410e-01 -3.51696312e-01 -3.00465733e-01 5.57731032e-01
1.97913393e-01 2.99159642e-02 3.71920317e-01 -2.11754963e-01
7.97614038e-01 -4.74654436e-02 5.67841589e-01 -8.81069422e-01
8.46705377e-01 1.12309642e-01 6.45451725e-01 4.99666870e-01
-1.23028107e-01 1.37751088e-01 2.70323843e-01 -7.30519146e-02
-6.53342187e-01 -8.54517519e-01 -2.02238798e-01 1.59457231e+00
-4.33379449e-02 -1.86104938e-01 -7.20444381e-01 -8.08460295e-01
-1.52451649e-01 9.55577791e-01 -3.08809400e-01 -5.91130480e-02
-7.72751510e-01 -5.24427295e-01 6.31137431e-01 3.61464649e-01
3.67104650e-01 -1.17225039e+00 -2.58196920e-01 -3.07273213e-02
-1.83425128e-01 -1.27361321e+00 -3.76645863e-01 5.16766012e-01
-8.74166369e-01 -6.10848308e-01 -6.52048230e-01 -1.07612920e+00
6.72165573e-01 -1.16325207e-02 1.27015018e+00 -4.06752005e-02
4.27565068e-01 3.04834604e-01 -6.89704001e-01 -4.67631191e-01
-8.69967639e-01 8.05674314e-01 2.51578659e-01 -6.00345209e-02
6.69469535e-01 -3.24147820e-01 -2.59955168e-01 1.10604376e-01
-7.60792494e-01 -2.21593961e-01 5.74211717e-01 4.53549922e-01
2.79098004e-01 -1.96537167e-01 5.38702548e-01 -1.18413866e+00
7.71142364e-01 -5.59828639e-01 -6.34096682e-01 6.02880836e-01
-5.09233296e-01 6.04878545e-01 9.88565564e-01 -4.18429255e-01
-1.12820399e+00 -1.27334744e-01 -1.17484946e-02 7.16800019e-02
-2.43391618e-01 6.64004982e-01 -1.58793569e-01 -4.01013419e-02
4.14184630e-01 -2.04382569e-01 -4.31507856e-01 -6.39074624e-01
3.52315813e-01 7.41821706e-01 -6.13607392e-02 -9.58465576e-01
4.87794399e-01 1.57011211e-01 -3.49638373e-01 -8.41242075e-01
-6.72216356e-01 -3.54254246e-01 -1.03878951e+00 2.72854567e-01
1.07723689e+00 -9.06038284e-01 -4.11423296e-01 9.68789533e-02
-1.40449572e+00 -1.14531733e-01 2.76282281e-01 7.86584318e-01
-2.78544109e-02 1.47255301e-01 -7.08424747e-01 -4.09995705e-01
-4.46868002e-01 -1.20578575e+00 8.94033968e-01 1.01270832e-01
-3.47502202e-01 -1.37498748e+00 3.47753018e-01 1.88587770e-01
3.84593338e-01 -2.16029629e-01 1.21742272e+00 -1.17986953e+00
-1.14961840e-01 8.58284757e-02 3.30149457e-02 4.99633312e-01
5.01059771e-01 1.22965217e-01 -7.28932083e-01 -4.11727160e-01
-4.93921861e-02 -3.29952836e-01 6.90606356e-01 6.82359412e-02
4.95010853e-01 6.81305677e-02 -2.67551839e-01 4.66803312e-01
1.44844103e+00 2.06315503e-01 1.82196140e-01 3.92635167e-01
7.50754476e-01 1.10940349e+00 6.25178158e-01 2.15997640e-02
5.61151803e-01 5.94589233e-01 -4.97630328e-01 -2.11815536e-02
1.56692475e-01 -1.96986824e-01 9.19526458e-01 1.47067750e+00
4.08268347e-02 -2.46300906e-01 -1.24747503e+00 5.94704986e-01
-1.51002157e+00 -3.14615130e-01 3.81910563e-01 2.41881418e+00
9.92503285e-01 5.43662943e-02 9.69517604e-02 -4.37981129e-01
7.47055650e-01 -1.28559425e-01 -6.57450780e-02 -7.72202849e-01
-1.46403596e-01 4.07788098e-01 6.93919420e-01 4.60905761e-01
-9.06442583e-01 1.47872567e+00 5.64222383e+00 7.15056300e-01
-1.58731842e+00 9.18303281e-02 5.03474891e-01 6.28018975e-01
-2.51880348e-01 2.11565852e-01 -1.09963500e+00 3.47260624e-01
1.10310292e+00 -3.07304323e-01 1.77629516e-01 6.03280604e-01
1.18439600e-01 1.59225941e-01 -1.29133856e+00 5.36903560e-01
-1.79941242e-03 -7.73384511e-01 1.70020208e-01 -2.69790947e-01
4.73228931e-01 3.49332839e-01 -3.04846168e-01 1.96823731e-01
4.56804276e-01 -8.11100185e-01 4.68788654e-01 1.04020566e-01
4.21355158e-01 -7.31775045e-01 6.51835024e-01 3.51586610e-01
-1.10256505e+00 2.77102560e-01 -4.07995582e-01 6.93048090e-02
-3.21822688e-02 4.28848207e-01 -1.07677555e+00 5.43594539e-01
6.37817860e-01 3.96750540e-01 -7.53722966e-01 4.33465034e-01
-2.95910269e-01 8.40169132e-01 -2.86803991e-01 -2.19074816e-01
2.59455629e-02 -4.39378738e-01 3.49458337e-01 1.57649255e+00
5.55806875e-01 -3.29141796e-01 5.26052356e-01 6.51154697e-01
-1.87370390e-01 1.12736690e+00 -1.01790690e+00 -2.76414663e-01
5.16314507e-01 1.11765313e+00 -1.07412267e+00 -1.53705508e-01
-7.24492788e-01 1.07384026e+00 6.56186402e-01 4.70173582e-02
-5.43452740e-01 -4.46707577e-01 5.21310329e-01 6.02259254e-03
3.77550900e-01 -4.27655697e-01 -1.59808040e-01 -1.41840267e+00
-3.75306569e-02 -8.79178643e-01 5.00496864e-01 -1.93711519e-01
-1.64088035e+00 8.76305699e-01 1.47247970e-01 -1.08246779e+00
-3.27278525e-01 -8.54043305e-01 -6.01891398e-01 1.00481462e+00
-1.56641865e+00 -1.14503920e+00 1.55504644e-01 5.82377672e-01
2.70088911e-01 -3.73842508e-01 8.42980146e-01 2.68839538e-01
-5.11473477e-01 6.59624875e-01 -1.51164159e-01 4.24069941e-01
1.12990022e+00 -1.03028500e+00 2.33950362e-01 7.93162644e-01
3.70236784e-01 8.67358923e-01 4.87327993e-01 -5.04939079e-01
-1.20185542e+00 -9.03708816e-01 1.17339742e+00 -2.89201945e-01
8.04554939e-01 -5.12140214e-01 -1.06488681e+00 6.09961987e-01
5.12974918e-01 -2.53363699e-01 1.08579135e+00 4.42011625e-01
-4.46274430e-01 -2.69114196e-01 -1.05918717e+00 5.64476967e-01
8.40942204e-01 -7.63496637e-01 -6.68640316e-01 4.94003706e-02
8.68075609e-01 2.46514171e-01 -1.01376128e+00 3.33817422e-01
2.89480567e-01 -6.96377099e-01 4.98014569e-01 -5.99582911e-01
1.85922682e-01 -2.08083346e-01 -3.09142649e-01 -1.58084321e+00
-2.19000533e-01 -2.86368102e-01 9.52628553e-01 1.90094662e+00
8.24799895e-01 -9.84071553e-01 3.67693305e-01 4.86745983e-01
2.36986995e-01 -3.48120928e-01 -6.62750065e-01 -7.17368424e-01
5.88939786e-01 -3.35557520e-01 4.19826627e-01 1.22735202e+00
-1.06862344e-01 7.98828483e-01 5.54365665e-02 1.81332722e-01
1.60410255e-01 4.32276845e-01 6.99650764e-01 -1.18502355e+00
-1.96855485e-01 -3.17952365e-01 -2.94300914e-01 -6.47139668e-01
8.17058802e-01 -1.31306911e+00 7.71398842e-02 -9.98189628e-01
-3.92994434e-02 -7.97860444e-01 -6.55751765e-01 5.27553022e-01
-1.04815736e-01 1.13517039e-01 2.37142771e-01 1.29766375e-01
-3.27573538e-01 3.77488315e-01 8.15956533e-01 2.91881382e-01
-5.64418197e-01 -1.67112108e-02 -9.73594427e-01 9.38436389e-01
8.83156240e-01 -7.33883142e-01 -2.13004619e-01 -6.86753571e-01
3.42590004e-01 -1.74983665e-01 -2.13629544e-01 -7.07339346e-01
-1.25797063e-01 -4.68610883e-01 1.41757559e-02 5.91952167e-02
-2.36512527e-01 -8.14807892e-01 -4.35559064e-01 4.06322032e-01
-4.58406478e-01 4.07869458e-01 2.34146595e-01 -6.67757988e-02
-3.50818515e-01 -5.04639268e-01 8.08963656e-01 -1.10687420e-01
-7.21093059e-01 -4.24390174e-02 -1.81564957e-01 2.89522856e-01
7.58492589e-01 1.38834089e-01 3.00558433e-02 -1.82848871e-01
-2.27461174e-01 -1.37969793e-03 7.10198045e-01 7.25748599e-01
4.64506336e-02 -1.15403628e+00 -9.56514359e-01 1.17425956e-01
3.98439109e-01 -6.30423903e-01 -4.39406693e-01 6.36645913e-01
-5.55994332e-01 4.88525093e-01 -3.24131995e-01 -4.37233478e-01
-1.29132950e+00 3.40562522e-01 3.41393724e-02 -3.95065367e-01
-7.02043623e-02 4.91189837e-01 2.39153922e-01 -7.37675428e-01
-1.42749339e-01 -5.36209762e-01 -4.31206554e-01 1.38938025e-01
1.06385276e-01 5.73047847e-02 3.01417619e-01 -9.24719930e-01
-7.55720615e-01 6.95048690e-01 -7.53259063e-02 -2.63500750e-01
1.23634040e+00 -1.59231961e-01 -3.13692510e-01 6.78712547e-01
1.34781146e+00 7.44890571e-01 -4.09344792e-01 -3.09308589e-01
7.20580757e-01 -1.85693562e-01 -2.16812804e-01 -4.60456043e-01
-1.02523732e+00 8.93559635e-01 4.63180691e-01 4.58235899e-03
1.06117523e+00 8.73678476e-02 5.67821324e-01 7.07808673e-01
6.75223589e-01 -1.09891033e+00 -3.64739031e-01 8.58178735e-01
5.69984853e-01 -1.40661740e+00 -1.75306290e-01 -4.45306242e-01
-9.40587103e-01 1.03902459e+00 4.66508567e-01 -1.94282040e-01
8.67059827e-01 1.45406842e-01 3.99210781e-01 -8.56884103e-03
-5.52466333e-01 -4.15975451e-01 4.97433782e-01 3.88767451e-01
1.19325721e+00 1.57742992e-01 -7.69730151e-01 6.13171518e-01
-4.13749367e-01 -1.26650691e-01 1.50539309e-01 8.05127621e-01
-1.68239787e-01 -1.78486586e+00 -9.47788879e-02 8.16411525e-02
-8.10671985e-01 -3.31903219e-01 -4.38150495e-01 7.47929275e-01
2.89615482e-01 8.32351804e-01 -2.28208192e-02 -1.32141232e-01
2.84944654e-01 3.68796915e-01 6.18565977e-01 -8.68154764e-01
-8.27315509e-01 -2.29987428e-01 2.13638783e-01 -1.67789325e-01
-7.12875724e-01 -4.38096851e-01 -1.30180109e+00 -1.31486639e-01
-4.05013055e-01 3.93517941e-01 7.14121342e-01 1.04136300e+00
3.82472068e-01 -6.49780110e-02 6.47479892e-01 -4.91579890e-01
-3.40857357e-01 -8.72274935e-01 -3.24776322e-01 6.90311611e-01
-2.16376707e-01 -6.44323528e-01 -2.38196209e-01 2.27770340e-02] | [10.819241523742676, 9.9459867477417] |
33ee490d-b13e-4d97-b233-c2dd27b42dd5 | a-simple-and-effective-method-of-cross | 2304.01352 | null | https://arxiv.org/abs/2304.01352v2 | https://arxiv.org/pdf/2304.01352v2.pdf | A Simple and Effective Method of Cross-Lingual Plagiarism Detection | We present a simple cross-lingual plagiarism detection method applicable to a large number of languages. The presented approach leverages open multilingual thesauri for candidate retrieval task and pre-trained multilingual BERT-based language models for detailed analysis. The method does not rely on machine translation and word sense disambiguation when in use, and therefore is suitable for a large number of languages, including under-resourced languages. The effectiveness of the proposed approach is demonstrated for several existing and new benchmarks, achieving state-of-the-art results for French, Russian, and Armenian languages. | ['Arutyun Avetisyan', 'Tsolak Ghukasyan', 'Arthur Malajyan', 'Karen Avetisyan'] | 2023-04-03 | null | null | null | null | ['word-sense-disambiguation'] | ['natural-language-processing'] | [-3.19593251e-01 -3.60044152e-01 -4.53583926e-01 3.40363652e-01
-1.59044814e+00 -1.01533234e+00 1.09340870e+00 7.41441548e-01
-8.07613134e-01 7.27221251e-01 2.22579002e-01 -6.41901553e-01
3.43013257e-02 -6.86262906e-01 -4.73920017e-01 -8.57871547e-02
1.19365461e-01 6.55090988e-01 2.75089055e-01 -7.50375092e-01
5.48363447e-01 4.78936911e-01 -1.18071413e+00 3.77727956e-01
1.06837237e+00 2.78138727e-01 -7.58211240e-02 2.77189463e-01
-4.35040712e-01 3.77835631e-01 -5.72816253e-01 -7.30326235e-01
1.96082026e-01 7.38257915e-02 -1.08288598e+00 -4.32541728e-01
6.12713873e-01 4.03818488e-01 -1.36623889e-01 1.14968264e+00
5.44276834e-01 -3.62272784e-02 5.23366392e-01 -3.23154837e-01
-7.34822989e-01 5.21903455e-01 -3.33237052e-01 2.38328591e-01
9.09646094e-01 -9.29035321e-02 1.21582472e+00 -1.13053739e+00
1.00792694e+00 1.41218817e+00 3.76452148e-01 -9.52253398e-03
-8.75017464e-01 -4.68684286e-01 -2.91450441e-01 4.00810391e-02
-1.63758337e+00 -3.33457589e-01 4.95625913e-01 -3.61173630e-01
1.18892372e+00 1.42983839e-01 1.18574806e-01 9.20841277e-01
3.47649276e-01 5.24848223e-01 1.37178731e+00 -1.08480740e+00
-2.31511667e-02 5.67140996e-01 2.89368302e-01 6.03156447e-01
5.27070642e-01 -2.38250792e-01 -5.24851143e-01 -6.34070754e-01
6.67553544e-02 -3.90550315e-01 -2.40517959e-01 -1.01290964e-01
-1.23704815e+00 9.25071120e-01 -1.12654142e-01 8.49735618e-01
-2.99306631e-01 -4.89262313e-01 8.77240300e-01 5.91186404e-01
7.61102974e-01 8.53927016e-01 -3.68202597e-01 2.79766135e-02
-1.31344748e+00 2.29061440e-01 1.02117217e+00 8.82541478e-01
7.73968875e-01 -3.72323096e-01 -9.04192030e-02 8.61282110e-01
2.52122581e-01 7.97244191e-01 8.25459063e-01 -4.36808199e-01
8.21827531e-01 8.25251520e-01 2.27271646e-01 -9.59039986e-01
1.66244749e-02 -3.47753584e-01 -1.89574182e-01 -3.70604545e-01
3.37376028e-01 3.94010574e-01 -5.23002863e-01 1.30448937e+00
1.32704452e-01 -3.67666185e-01 6.86569691e-01 6.10893726e-01
8.13234091e-01 7.87441015e-01 -8.62950534e-02 -2.30680391e-01
1.72429001e+00 -8.43760908e-01 -6.73876882e-01 -4.11695778e-01
8.09546888e-01 -1.59411871e+00 1.18917251e+00 2.62350827e-01
-1.11272037e+00 -2.40821347e-01 -9.69222665e-01 -2.69603163e-01
-8.08085740e-01 2.60827333e-01 2.39492625e-01 8.40416431e-01
-9.36589897e-01 2.75413871e-01 -4.43800122e-01 -8.55946720e-01
-1.08224049e-01 5.44649595e-03 -5.21656036e-01 -3.29067647e-01
-1.64683068e+00 1.17419684e+00 5.46965420e-01 -2.25862637e-01
-6.23322964e-01 -2.27454379e-01 -8.87201011e-01 -9.31706578e-02
2.72884727e-01 -2.45898396e-01 9.47586298e-01 -6.32059216e-01
-1.20729208e+00 1.36923850e+00 -8.99147466e-02 -3.31496298e-01
3.65178436e-01 -5.76933324e-01 -7.77211249e-01 2.42058292e-01
3.09190065e-01 -3.69497552e-03 5.38598895e-01 -7.30139613e-01
-1.88994303e-01 -1.66878894e-01 -4.51970659e-02 2.00331509e-01
-7.07449019e-01 8.96458089e-01 -5.92284560e-01 -6.93843901e-01
-1.83379147e-02 -9.43510294e-01 -3.74586508e-02 -6.11109555e-01
-1.27925694e-01 -3.93518448e-01 5.61980128e-01 -1.10608768e+00
1.47275221e+00 -1.83945930e+00 -2.00370446e-01 2.39220947e-01
-3.15283984e-01 8.23758423e-01 -2.98841417e-01 1.17864740e+00
2.84680516e-01 1.83904797e-01 -1.27688631e-01 -7.01567158e-02
-4.38754223e-02 -2.58304417e-01 -5.04495084e-01 6.08483076e-01
9.90610048e-02 7.55381942e-01 -1.08082533e+00 -5.90515018e-01
1.26311392e-01 3.82122308e-01 -1.18088074e-01 5.37809804e-02
-1.49264187e-01 1.46400914e-01 -5.43600202e-01 7.39829540e-01
5.47202051e-01 2.38154326e-02 2.85699785e-01 2.17587009e-01
-4.12556916e-01 1.04623246e+00 -1.07340109e+00 2.00813317e+00
-7.34615088e-01 4.64125693e-01 -1.93802282e-01 -4.99318480e-01
1.24200833e+00 4.72027540e-01 3.31921550e-03 -6.50333583e-01
4.99410406e-02 1.04770339e+00 -3.35358590e-01 -4.68643218e-01
8.74603212e-01 1.61113963e-01 -5.30905902e-01 6.66208804e-01
2.01266468e-01 7.23827034e-02 3.95591795e-01 4.06318575e-01
9.42436457e-01 1.30432144e-01 8.16074789e-01 -7.10627496e-01
1.35247922e+00 2.52710432e-01 2.15532333e-01 5.60689867e-01
-6.81628436e-02 8.27451050e-02 7.22892657e-02 -3.54260802e-01
-9.35069323e-01 -8.16241741e-01 -6.68997038e-03 1.06661320e+00
1.29940644e-01 -6.26395583e-01 -5.58628500e-01 -6.87219918e-01
-2.53968656e-01 8.20522010e-01 4.46689576e-02 -8.49294513e-02
-9.14918005e-01 -3.67693365e-01 1.16969180e+00 -4.29292589e-01
2.28121772e-01 -1.20918965e+00 -2.12909237e-01 1.41680226e-01
-4.84081417e-01 -1.33327055e+00 -4.25269872e-01 -5.01776040e-01
-6.41330957e-01 -1.06987894e+00 -8.11388195e-01 -1.09509051e+00
2.06597671e-01 2.94287056e-01 1.35349286e+00 -2.92479843e-01
-1.95158318e-01 4.66539264e-01 -2.67709136e-01 5.48047461e-02
-1.00634849e+00 4.62883532e-01 9.36894864e-02 -1.94313005e-01
7.67843425e-01 -5.96318245e-02 -1.41329080e-01 6.38795346e-02
-9.10661459e-01 -7.27766752e-01 5.43891311e-01 5.73759854e-01
3.40739101e-01 -5.00863373e-01 6.23578250e-01 -9.02763128e-01
9.50589538e-01 -4.57490087e-01 -7.36993849e-01 8.56761456e-01
-6.73417389e-01 2.89138496e-01 6.82234287e-01 -3.44398677e-01
-1.06306970e+00 -3.47261786e-01 -9.69962180e-02 -1.07040063e-01
-8.88480470e-02 6.85326993e-01 -4.61644977e-02 -3.38457763e-01
7.27165759e-01 2.30611369e-01 -3.05375546e-01 -8.78617465e-01
6.22641087e-01 1.01197362e+00 5.40220499e-01 -6.04154110e-01
7.46082246e-01 2.05940232e-01 -3.27178091e-01 -7.88733602e-01
-4.32289928e-01 -9.07505155e-01 -6.01077855e-01 1.22242436e-01
3.33062023e-01 -1.29150689e+00 -1.30573228e-01 3.22982371e-01
-1.52443135e+00 5.24247587e-01 5.27892634e-02 6.23690426e-01
-4.49783690e-02 8.72661710e-01 -7.71013439e-01 -5.85202515e-01
-9.50563669e-01 -1.03002679e+00 1.02140403e+00 -1.55466557e-01
-2.71025926e-01 -1.28001797e+00 6.87855244e-01 3.57577980e-01
2.36529961e-01 -2.09731102e-01 9.08608854e-01 -1.22078788e+00
-1.93430796e-01 -5.73949277e-01 -1.01798907e-01 1.09803863e-01
-3.63424048e-02 -2.99145252e-01 -7.22528040e-01 -6.49485290e-01
-1.33851945e-01 -6.56781852e-01 6.76694214e-01 -3.08362544e-01
5.53536043e-03 -3.82821411e-01 -1.97451457e-01 -1.28996328e-01
1.56283712e+00 -4.90808666e-01 5.48044384e-01 7.99385905e-01
4.57735300e-01 7.89064705e-01 8.72893155e-01 1.92640610e-02
3.43818367e-01 7.15810359e-01 -7.06020668e-02 2.15684250e-01
-8.79118890e-02 -4.20350373e-01 6.81597531e-01 1.23832619e+00
4.78239357e-01 -2.80953288e-01 -1.21055770e+00 9.64291275e-01
-1.66833758e+00 -6.44690335e-01 -1.42298818e-01 2.55573392e+00
9.49420810e-01 -1.85622741e-02 -1.53486133e-01 -3.15535754e-01
8.84541392e-01 1.94913924e-01 2.12050140e-01 -7.92146027e-01
-4.82563645e-01 5.20892084e-01 5.64260066e-01 8.01557958e-01
-9.35168386e-01 1.56997252e+00 6.29155064e+00 1.22814989e+00
-1.12793422e+00 3.92684877e-01 -1.21866748e-01 2.56961942e-01
-3.81527573e-01 4.36951667e-01 -9.21394229e-01 3.86609659e-02
1.15892768e+00 -3.92090917e-01 2.61461616e-01 5.71509004e-01
5.70519790e-02 9.96812508e-02 -5.41782081e-01 7.75613725e-01
4.95569974e-01 -1.20853209e+00 1.81898758e-01 -2.63867527e-01
6.14246488e-01 5.00203609e-01 -2.22280934e-01 3.77505630e-01
1.68863937e-01 -6.25891984e-01 5.80979764e-01 7.40794688e-02
7.49869108e-01 -7.01940596e-01 7.57597208e-01 4.49570000e-01
-9.29181516e-01 4.30170476e-01 -4.23157245e-01 1.14017457e-01
1.28380358e-02 4.93699729e-01 -7.24494576e-01 8.70886385e-01
2.75169253e-01 5.68478882e-01 -8.22561085e-01 8.21830273e-01
-3.85798365e-01 6.02120578e-01 -3.69387329e-01 -1.27968758e-01
4.44265932e-01 -1.05245426e-01 8.89239967e-01 1.81289423e+00
5.32347023e-01 -5.50292313e-01 4.66422826e-01 3.97001475e-01
-1.91682845e-01 1.03621435e+00 -7.62181818e-01 -2.91339636e-01
4.72033948e-01 1.25096750e+00 -4.09272522e-01 -4.58727241e-01
-6.01774335e-01 7.96065092e-01 4.94415522e-01 6.11654148e-02
-2.82219559e-01 -7.22323477e-01 2.43106708e-01 2.17232611e-02
8.81469101e-02 -2.60884553e-01 2.40333304e-01 -1.57057023e+00
1.91252008e-01 -1.24528527e+00 9.50138390e-01 -2.59782255e-01
-1.23481011e+00 9.13827240e-01 -7.39754885e-02 -1.44239140e+00
-5.40499568e-01 -4.71990287e-01 -2.32317924e-01 1.10663998e+00
-1.92207885e+00 -1.23930287e+00 4.28799450e-01 6.15624785e-01
3.99492890e-01 -5.18781066e-01 1.12246048e+00 7.13095248e-01
-2.16119215e-01 6.11835122e-01 3.70959669e-01 7.87974969e-02
1.32557881e+00 -6.30453527e-01 6.22337997e-01 1.16189897e+00
4.68126297e-01 1.03624904e+00 6.64248407e-01 -7.36001909e-01
-1.43784797e+00 -9.97639000e-01 1.99264634e+00 -2.03615159e-01
1.17554641e+00 -3.69277954e-01 -1.02663207e+00 2.66206443e-01
6.57683909e-01 -7.81413987e-02 6.15247905e-01 1.22753762e-01
-9.90398288e-01 1.31124005e-01 -1.07915294e+00 3.88974696e-01
4.16330695e-01 -9.29237187e-01 -9.84692037e-01 6.80754781e-01
5.69933236e-01 -2.76591182e-01 -8.59200239e-01 1.17960326e-01
3.21075201e-01 -4.16210055e-01 8.68693590e-01 -7.05805719e-01
1.29082590e-01 -2.30497271e-01 -1.69667304e-01 -9.48191345e-01
5.89846773e-03 -1.00215030e+00 4.87694219e-02 1.26540577e+00
5.13420105e-01 -7.55886018e-01 1.55900776e-01 6.24282807e-02
3.38407829e-02 -6.93920329e-02 -1.20253360e+00 -9.89238679e-01
4.75074351e-01 -2.85985500e-01 2.66628474e-01 1.07966650e+00
2.99884617e-01 5.65385878e-01 -2.42950439e-01 1.55258149e-01
4.44816828e-01 3.68787080e-01 5.36844671e-01 -1.04073048e+00
-1.19414866e-01 -3.09491605e-01 -3.46708179e-01 -6.49151027e-01
6.23917103e-01 -1.22768199e+00 -2.98361629e-01 -1.08365870e+00
1.19569622e-01 -2.27731794e-01 -3.52182090e-01 2.62265593e-01
-2.40938395e-01 4.45521951e-01 5.22866733e-02 6.68227315e-01
-8.88984561e-01 4.19046074e-01 5.35207212e-01 -1.45382285e-01
-2.09165946e-01 -1.39240161e-01 -5.66543102e-01 5.93857706e-01
5.54514468e-01 -7.70182371e-01 1.98998868e-01 -4.15022522e-01
4.34085578e-01 -2.34108806e-01 1.71339035e-01 -7.96390057e-01
1.77617520e-01 -8.02037586e-03 -4.51755643e-01 -3.45896572e-01
-4.56036665e-02 -4.38366324e-01 -2.17877761e-01 5.88103592e-01
-4.13613290e-01 4.22513902e-01 3.30005884e-01 2.38522813e-01
-4.86186296e-01 -4.68840897e-01 6.00309610e-01 -2.85603076e-01
-6.33173168e-01 -2.15953186e-01 -4.94966328e-01 4.67382401e-01
6.84205711e-01 3.26682478e-01 -5.87532759e-01 -2.06202850e-01
-5.53067364e-02 -4.75508198e-02 7.00459898e-01 8.32984626e-01
2.57302850e-01 -1.20451224e+00 -1.09093308e+00 5.40030375e-02
6.73032105e-01 -9.16997075e-01 -2.85445839e-01 5.66533923e-01
-6.65011704e-01 9.36519384e-01 -3.50410957e-03 -3.89807820e-01
-1.22193241e+00 6.00909173e-01 -7.31426328e-02 -9.42883372e-01
-3.68349731e-01 1.41857520e-01 -4.29676205e-01 -9.38485563e-01
-1.08427830e-01 2.37886786e-01 -3.39084864e-01 -2.02240481e-04
5.14810741e-01 3.50372016e-01 5.47060192e-01 -1.21653342e+00
-5.28360307e-01 5.25884807e-01 -2.72423178e-01 -3.83237839e-01
8.60174239e-01 -3.51675004e-01 -6.48092628e-01 5.07411778e-01
1.32555568e+00 5.95990539e-01 1.64910376e-01 -7.17892945e-01
4.42712665e-01 -4.38116997e-01 2.51365211e-02 -5.82352340e-01
-3.86917472e-01 8.49764168e-01 3.80117595e-01 -2.58156568e-01
8.23114216e-01 -9.22988951e-02 9.63617742e-01 7.97070265e-01
7.49746025e-01 -1.01510561e+00 -2.68311620e-01 8.65890741e-01
6.08674586e-01 -1.13592386e+00 2.28046879e-01 -3.66733521e-01
-3.91227901e-01 1.20637834e+00 5.43041453e-02 -1.73934475e-01
4.31712985e-01 -2.91672796e-01 3.41054022e-01 -1.27333358e-01
-5.61391771e-01 -6.73169717e-02 6.80586874e-01 8.38523954e-02
7.90688932e-01 -8.25132653e-02 -1.13734806e+00 2.08111122e-01
3.92269008e-02 -2.45970577e-01 4.93523061e-01 9.32326138e-01
-2.71006405e-01 -1.52320409e+00 -6.53447807e-01 -1.09767064e-01
-1.01299167e+00 -5.42023957e-01 -5.23823380e-01 7.45160162e-01
-2.38889262e-01 9.83315349e-01 -3.26115280e-01 4.29567285e-02
2.23234579e-01 2.76775420e-01 2.92039961e-01 -7.00292230e-01
-1.23305631e+00 2.90868104e-01 2.75678992e-01 -4.35323209e-01
-5.00966907e-01 -8.39581072e-01 -6.79656923e-01 -1.75778076e-01
-5.53239226e-01 5.22469282e-01 7.12884188e-01 9.48460042e-01
5.91051400e-01 -3.56594145e-01 4.87051696e-01 -2.63492405e-01
-6.31493211e-01 -9.78599668e-01 -2.05970347e-01 3.84782344e-01
8.49370030e-04 -1.93800136e-01 -2.12707669e-01 -3.17486852e-01] | [11.15715503692627, 10.014444351196289] |
109196b1-51bd-46c3-a557-e7b40ca08fc1 | a-dataset-for-audio-video-based-vehicle-speed | 2212.01651 | null | https://arxiv.org/abs/2212.01651v1 | https://arxiv.org/pdf/2212.01651v1.pdf | A dataset for audio-video based vehicle speed estimation | Accurate speed estimation of road vehicles is important for several reasons. One is speed limit enforcement, which represents a crucial tool in decreasing traffic accidents and fatalities. Compared with other research areas and domains, the number of available datasets for vehicle speed estimation is still very limited. We present a dataset of on-road audio-video recordings of single vehicles passing by a camera at known speeds, maintained stable by the on-board cruise control. The dataset contains thirteen vehicles, selected to be as diverse as possible in terms of manufacturer, production year, engine type, power and transmission, resulting in a total of $ 400 $ annotated audio-video recordings. The dataset is fully available and intended as a public benchmark to facilitate research in audio-video vehicle speed estimation. In addition to the dataset, we propose a cross-validation strategy which can be used in a machine learning model for vehicle speed estimation. Two approaches to training-validation split of the dataset are proposed. | ['Ivana Čavor', 'Nikola Bulatović', 'Slobodan Djukanović'] | 2022-12-03 | null | null | null | null | ['vehicle-speed-estimation'] | ['computer-vision'] | [-8.03706869e-02 -3.77875537e-01 -6.67504370e-01 -4.62830603e-01
-7.73122311e-01 -4.89647508e-01 4.83119071e-01 1.66948348e-01
-5.61614692e-01 5.90587974e-01 -2.78512716e-01 -6.00563347e-01
-1.98020250e-01 -8.08757961e-01 -7.98296809e-01 -5.00893891e-01
-1.75598204e-01 2.09333450e-01 6.26035333e-01 -1.53608933e-01
2.28263557e-01 7.95806408e-01 -2.06390834e+00 8.37432817e-02
4.08937037e-01 1.11530066e+00 1.40079737e-01 8.59562218e-01
2.40222991e-01 8.13600063e-01 -4.38450187e-01 -6.56644881e-01
2.04486251e-01 -7.93232322e-02 -6.39090478e-01 -1.82965726e-01
3.33220810e-01 -4.92061526e-01 -6.30921900e-01 6.86912715e-01
1.72171116e-01 1.80896327e-01 6.67034268e-01 -1.88125563e+00
3.87865841e-01 4.01630551e-01 -1.45725086e-01 5.23814440e-01
2.33171687e-01 7.18759513e-03 7.72498667e-01 -4.29700851e-01
4.44515198e-01 9.06171858e-01 6.26058936e-01 4.32163149e-01
-9.51397061e-01 -1.16532981e+00 -2.46578455e-01 9.40280735e-01
-1.69097042e+00 -8.58840466e-01 1.00148308e+00 -6.03902459e-01
7.87935376e-01 2.40904257e-01 7.53021896e-01 9.24220264e-01
4.62339669e-02 5.39104462e-01 5.35430729e-01 -9.85792354e-02
2.10711420e-01 4.43890780e-01 9.40437540e-02 2.03021944e-01
2.00539783e-01 1.90594569e-01 -3.26732546e-01 -4.92209643e-02
3.24186087e-01 -3.79499137e-01 -2.55143978e-02 -2.52118081e-01
-9.94995654e-01 8.02966416e-01 -1.37071803e-01 -1.27480598e-02
-2.36280233e-01 2.26814613e-01 8.83225203e-01 7.51616299e-01
2.80138791e-01 -1.25994265e-01 -6.60349011e-01 -7.33494759e-01
-8.99816692e-01 5.22574604e-01 5.99999726e-01 1.09179282e+00
8.19387853e-01 1.47734404e-01 4.10833806e-01 8.03220928e-01
8.72531384e-02 7.47760057e-01 -6.11310489e-02 -1.05242574e+00
8.11066806e-01 1.63082466e-01 2.70087476e-04 -1.05033755e+00
-3.73180687e-01 1.66767314e-01 -4.78818893e-01 1.28605366e-01
4.67753708e-01 -5.64655550e-02 -1.07034288e-01 1.30868721e+00
2.19239563e-01 4.88660663e-01 -1.93370387e-01 7.05006659e-01
7.28967547e-01 8.54395747e-01 1.23180158e-01 -4.87116277e-01
1.19465518e+00 -5.70057631e-01 -6.51150227e-01 8.77055302e-02
6.71315312e-01 -6.90708637e-01 5.20041347e-01 7.40354240e-01
-9.69981611e-01 -7.42343426e-01 -9.01314139e-01 2.27712244e-01
-5.31201005e-01 9.75995511e-02 3.69425684e-01 1.02885520e+00
-6.20628476e-01 4.20248240e-01 -6.01428747e-01 -9.11427140e-02
4.24201727e-01 3.03057343e-01 -4.38266337e-01 -2.68596467e-02
-1.25898981e+00 9.53993082e-01 2.52790868e-01 -1.33213267e-01
-1.01091278e+00 -9.30048466e-01 -9.58287239e-01 -4.35469776e-01
4.48944092e-01 -1.43654227e-01 1.37892652e+00 -7.09819436e-01
-1.47782338e+00 5.95128953e-01 -6.61963895e-02 -6.51342571e-01
7.89402306e-01 -1.16539657e-01 -9.37828243e-01 2.82506198e-01
-8.42271522e-02 5.14364839e-01 9.51191247e-01 -9.50019479e-01
-1.18719816e+00 -1.86731257e-02 -2.98295040e-02 -2.86322862e-01
-1.32227167e-01 2.71013379e-01 -6.18026078e-01 -3.23126376e-01
-6.22166455e-01 -9.42733049e-01 1.17258243e-01 -3.06395918e-01
-2.14699194e-01 -3.77928883e-01 1.25729191e+00 -7.21338451e-01
1.49479079e+00 -2.10530162e+00 -2.93200672e-01 3.07093590e-01
-1.02364711e-01 3.76903385e-01 -3.08351070e-02 3.48439008e-01
-8.13807249e-02 1.19218953e-01 3.01091224e-01 -2.37850323e-01
-1.80386186e-01 7.50195682e-02 -5.11019006e-02 7.94516385e-01
2.09358379e-01 4.62420911e-01 -8.47172856e-01 -6.49664044e-01
7.68225491e-01 3.93046409e-01 -3.81186694e-01 6.17140457e-02
2.41759300e-01 2.06349313e-01 -4.57542568e-01 5.36114335e-01
7.86570489e-01 8.16346824e-01 -3.65691125e-01 -3.02922100e-01
-4.45955575e-01 1.87437490e-01 -1.29374194e+00 1.12704921e+00
-6.54164314e-01 1.29010785e+00 -1.01470500e-02 -1.14854538e+00
7.56059587e-01 3.96541744e-01 6.85776174e-01 -7.59024501e-01
3.55521530e-01 2.15757340e-01 -2.06956580e-01 -7.45598257e-01
6.16409004e-01 3.11832011e-01 -2.37415254e-01 -7.34259039e-02
-1.66161910e-01 -2.63745278e-01 7.60126948e-01 1.36925448e-02
9.00467455e-01 -3.58753234e-01 -5.20838797e-02 -7.55281970e-02
8.04317951e-01 9.61497426e-02 3.17065507e-01 1.66548491e-01
-3.73109967e-01 1.81912690e-01 6.81881905e-01 -3.29871505e-01
-1.37978530e+00 -9.05726850e-01 -3.74163151e-01 8.89753759e-01
1.96190059e-01 -3.16170037e-01 -8.04370642e-01 -2.52940476e-01
6.42217845e-02 6.43728912e-01 -3.35883707e-01 -1.61121234e-01
-6.24492168e-01 9.13798511e-02 7.32350707e-01 5.46965182e-01
2.66846955e-01 -6.12843156e-01 -6.02017283e-01 7.90339112e-02
-1.42928213e-01 -1.47474802e+00 -3.06183755e-01 -1.39101103e-01
-5.80404222e-01 -1.38827372e+00 -3.84693354e-01 -5.99259555e-01
1.73408255e-01 5.21069705e-01 1.01071060e+00 -2.95676091e-05
-1.13404714e-01 2.02630311e-01 -4.37057823e-01 -7.80082405e-01
-6.94896221e-01 2.18832359e-01 3.00004542e-01 1.06089719e-01
4.99779820e-01 -8.91881138e-02 -4.75590020e-01 8.88343513e-01
-4.70330626e-01 -2.44490847e-01 2.86392480e-01 2.40768433e-01
4.51525778e-01 5.66373289e-01 7.36071348e-01 -4.80738580e-01
3.72595280e-01 -7.70714939e-01 -9.73549426e-01 -3.58751774e-01
-2.90937841e-01 -4.69347388e-01 7.14008629e-01 -4.14684862e-01
-6.41859472e-01 7.68574774e-02 -4.23347682e-01 -4.93249834e-01
-4.74629521e-01 9.96203572e-02 -3.38290900e-01 -1.05425008e-01
2.06216887e-01 -4.33611088e-02 2.05254793e-01 -3.64733845e-01
2.83518992e-02 8.59059215e-01 6.59034431e-01 -3.50821693e-03
8.59466076e-01 1.74866989e-01 3.65281999e-02 -1.44198596e+00
-6.76515028e-02 -8.50252926e-01 -4.89320576e-01 -7.30632424e-01
6.80138648e-01 -9.85817611e-01 -1.06503570e+00 5.77827811e-01
-9.77458417e-01 -2.80582637e-01 -1.05980933e-02 8.09482098e-01
-7.55258501e-01 2.14290053e-01 -1.86152562e-01 -9.25942957e-01
1.52652755e-01 -1.12708557e+00 8.97926033e-01 -3.23972851e-03
-1.69909358e-01 -8.20211768e-01 -1.61834151e-01 4.97665793e-01
3.61972183e-01 1.39616475e-01 5.59703648e-01 -3.58789563e-01
-3.62997830e-01 -6.28519058e-01 -1.05061799e-01 6.28384590e-01
-1.50623113e-01 4.84706342e-01 -9.48005319e-01 -1.75300092e-01
-3.73383582e-01 -1.92779198e-01 6.10089719e-01 6.53496802e-01
1.16924071e+00 -3.88312899e-02 -4.53787684e-01 1.46429762e-01
1.21034586e+00 2.36241654e-01 8.67632926e-01 4.00184155e-01
4.23065782e-01 7.55164266e-01 9.99709547e-01 4.86416519e-01
3.54819447e-01 9.52594161e-01 6.52944326e-01 1.79744005e-01
3.42188440e-02 -1.67899355e-01 3.38763386e-01 7.06293046e-01
-3.01604271e-01 -1.15241237e-01 -8.40448022e-01 6.67972088e-01
-1.29104006e+00 -1.38626719e+00 -5.98368764e-01 2.48829389e+00
2.79651433e-01 4.54210520e-01 8.45596075e-01 9.87037122e-01
6.03142679e-01 -8.20029527e-02 -1.31985500e-01 -5.80825627e-01
2.93594807e-01 -4.70995694e-01 1.03749800e+00 3.28343481e-01
-1.16340673e+00 3.73077065e-01 6.84765005e+00 1.14162302e+00
-1.05163491e+00 7.67779648e-02 4.67428923e-01 -7.05794320e-02
1.05069987e-01 -4.66202289e-01 -9.79571581e-01 8.04259062e-01
1.75426078e+00 -1.75771788e-01 3.03887874e-01 8.95986319e-01
8.11374187e-01 -2.48382688e-01 -1.21543002e+00 1.19638932e+00
-1.23401210e-01 -1.26787519e+00 -2.29259431e-01 2.33863108e-02
2.75845975e-01 9.71401781e-02 -1.61808088e-01 3.77062380e-01
-5.17828763e-01 -7.99964964e-01 1.12271369e+00 4.28603649e-01
9.29993212e-01 -1.48350906e+00 1.01482558e+00 5.34035623e-01
-1.53110647e+00 -2.85252869e-01 -1.55310497e-01 -7.17984214e-02
4.79276121e-01 3.02320421e-01 -4.74114656e-01 5.27651846e-01
7.98508584e-01 6.27641439e-01 -4.35829997e-01 1.23885632e+00
4.00483310e-01 8.61541510e-01 -2.95612544e-01 -1.88672826e-01
9.89435706e-03 -1.19028829e-01 3.58167589e-01 1.33315659e+00
2.59894490e-01 -2.88492084e-01 -7.25773945e-02 3.05527329e-01
5.88543043e-02 1.68413758e-01 -9.02771294e-01 1.48735270e-01
6.71097755e-01 1.02542508e+00 -4.93499160e-01 -2.62263101e-02
-7.43055880e-01 2.58200794e-01 -3.56018305e-01 1.92480683e-01
-1.43143034e+00 -8.28118384e-01 9.19858456e-01 6.47091508e-01
1.65385172e-01 -3.59439701e-01 8.42824280e-02 -3.02934319e-01
-6.62395507e-02 -4.86086756e-01 -6.75257370e-02 -4.67263103e-01
-7.45797694e-01 3.66772592e-01 6.41087294e-01 -1.68883896e+00
-2.36898616e-01 -7.31289804e-01 -3.94635260e-01 5.12365103e-01
-1.77094233e+00 -8.02974403e-01 -3.18425149e-01 5.28563559e-01
9.69228327e-01 -4.32889551e-01 2.25328729e-01 1.13550639e+00
-6.74731970e-01 7.29395390e-01 4.45783138e-02 1.30943760e-01
3.90340865e-01 -5.62208056e-01 3.42130899e-01 4.34864581e-01
-2.89664835e-01 4.64807525e-02 9.01363730e-01 -2.36109957e-01
-1.75035787e+00 -1.36443853e+00 8.37970614e-01 -5.99800766e-01
9.13121521e-01 -3.35793287e-01 -7.08806813e-01 2.65620679e-01
-4.41202186e-02 5.61757162e-02 5.18018126e-01 -3.95608008e-01
1.00409977e-01 -5.61067164e-01 -9.61305618e-01 1.61600664e-01
6.15286052e-01 -5.34028590e-01 -1.24836966e-01 1.31910250e-01
3.26335996e-01 -2.57607877e-01 -8.35420728e-01 5.68673126e-02
7.79917955e-01 -8.02458763e-01 9.43054855e-01 -3.28806221e-01
1.68355376e-01 -2.19014779e-01 -4.25546244e-02 -1.06159806e+00
-5.93189672e-02 -5.98126411e-01 -1.50872350e-01 1.46125150e+00
4.20832634e-01 -4.44320202e-01 8.76253903e-01 3.53165299e-01
-2.45723039e-01 -5.37814736e-01 -1.28790486e+00 -1.13812554e+00
-7.70635530e-02 -1.44559014e+00 5.92148721e-01 3.00057530e-01
-2.48005226e-01 2.39298880e-01 -5.20830095e-01 1.13311820e-01
5.37684619e-01 -5.78604460e-01 1.12944591e+00 -1.38366318e+00
3.80442202e-01 -6.16354525e-01 -9.51117396e-01 -9.45344687e-01
3.92958969e-01 -4.08138067e-01 -3.76329087e-02 -1.12702584e+00
1.58115625e-02 -4.58343834e-01 1.18866943e-01 -9.25077945e-02
5.80586910e-01 4.45608586e-01 -1.08894616e-01 -1.13030393e-02
-4.47156996e-01 1.35068700e-01 7.35126734e-01 -3.36177289e-01
-4.27686423e-02 5.21664739e-01 5.85054010e-02 7.09592104e-01
9.00689960e-01 -4.44037706e-01 -5.83219051e-01 -1.06631465e-01
1.42961591e-01 8.30997303e-02 4.00980383e-01 -1.12019157e+00
1.30213410e-01 -3.83243293e-01 -2.47504205e-01 -9.57147062e-01
3.12728673e-01 -1.31409264e+00 4.59163219e-01 4.68273342e-01
-1.22616209e-01 1.22712053e-01 3.35017949e-01 7.15120912e-01
-4.25612301e-01 -2.49569282e-01 8.32712293e-01 5.08783340e-01
-1.04740119e+00 3.50963175e-01 -8.77305031e-01 -1.03461944e-01
1.49893200e+00 -6.12226963e-01 -1.20019585e-01 -3.89381856e-01
-1.02783531e-01 2.29708061e-01 3.35528910e-01 7.63153136e-01
4.19912070e-01 -1.46209073e+00 -8.32681894e-01 2.14933008e-01
3.73064935e-01 -4.19738621e-01 4.18078005e-01 8.61385226e-01
-8.13701093e-01 7.05197513e-01 -3.05002540e-01 -6.29087985e-01
-1.40952063e+00 7.95832992e-01 6.74491152e-02 2.88991690e-01
-3.44862819e-01 2.98236340e-01 -5.48403561e-01 1.36901021e-01
2.80718863e-01 -4.71783608e-01 -4.51798201e-01 2.20419377e-01
8.37085247e-01 1.18086088e+00 4.72300410e-01 -1.30082095e+00
-4.24465925e-01 6.62106514e-01 3.34661782e-01 3.02282900e-01
9.33714151e-01 -4.39185739e-01 4.52703208e-01 6.55289829e-01
1.61029434e+00 -6.76481873e-02 -1.13557172e+00 2.11051539e-01
-1.59355953e-01 -5.25064111e-01 2.70481974e-01 2.73973551e-02
-1.25807452e+00 7.96473861e-01 7.19841897e-01 5.30262768e-01
9.88536775e-01 -7.73103489e-03 8.78596961e-01 3.53589535e-01
5.57273328e-01 -1.47934186e+00 -3.17159265e-01 3.15460861e-01
6.44952118e-01 -1.38354671e+00 -1.07533567e-01 -5.69266677e-01
-4.78878468e-01 1.03533840e+00 2.67696559e-01 1.62396252e-01
7.75709212e-01 3.50439131e-01 -1.22795522e-01 3.09275836e-01
-6.61970198e-01 4.39711660e-02 1.79129928e-01 9.29424942e-01
2.78487831e-01 1.19602941e-02 -1.66887984e-01 4.42991227e-01
-2.34607011e-01 1.26849860e-01 5.16677558e-01 6.58167303e-01
-4.42720950e-01 -6.95598722e-01 -3.31562519e-01 5.80211341e-01
-3.80761802e-01 4.68667775e-01 1.77942589e-01 9.95722950e-01
1.96681678e-01 1.40118003e+00 2.45418057e-01 -5.19273877e-01
8.58074903e-01 -3.39145333e-01 1.09770738e-01 8.93168151e-03
-8.58487040e-02 -5.55915296e-01 5.94417751e-01 -5.89277983e-01
-3.78133178e-01 -8.37588072e-01 -1.00588882e+00 -9.36446846e-01
-3.90914083e-01 2.74320900e-01 9.42912221e-01 7.56787658e-01
-8.96133333e-02 5.13853133e-01 1.05822265e+00 -1.22432947e+00
-2.08733633e-01 -7.13995039e-01 -7.73726225e-01 5.27613997e-01
5.85257947e-01 -9.95794594e-01 -3.42403859e-01 3.12423736e-01] | [7.853906631469727, -0.9286969304084778] |
d1b3eafd-adc7-4eeb-b056-7d0cb21123ec | video-matting-via-sparse-and-low-rank | null | null | http://openaccess.thecvf.com/content_iccv_2015/html/Zou_Video_Matting_via_ICCV_2015_paper.html | http://openaccess.thecvf.com/content_iccv_2015/papers/Zou_Video_Matting_via_ICCV_2015_paper.pdf | Video Matting via Sparse and Low-Rank Representation | We introduce a novel method of video matting via sparse and low-rank representation. Previous matting methods [10, 9] introduced a nonlocal prior to estimate the alpha matte and have achieved impressive results on some data. However, on one hand, searching inadequate or excessive samples may miss good samples or introduce noise; on the other hand, it is difficult to construct consistent nonlocal structures for pixels with similar features, yielding spatially and temporally inconsistent video mattes. In this paper, we proposed a novel video matting method to achieve spatially and temporally consistent matting result. Toward this end, a sparse and low-rank representation model is introduced to pursue consistent nonlocal structures for pixels with similar features. The sparse representation is used to adaptively select best samples and accurately construct the nonlocal structures for all pixels, while the low-rank representation is used to globally ensure consistent nonlocal structures for pixels with similar features. The two representations are combined to generate consistent video mattes. Experimental results show that our method has achieved high quality results in a variety of challenging examples featuring illumination changes, feature ambiguity, topology changes, transparency variation, dis-occlusion, fast motion and motion blur. | ['Xiaogang Wang', 'Xiaowu Chen', 'Guangying Cao', 'Dongqing Zou'] | 2015-12-01 | null | null | null | iccv-2015-12 | ['video-matting'] | ['computer-vision'] | [ 1.08415216e-01 -6.68460727e-01 -7.66180605e-02 -6.92071468e-02
-6.29146039e-01 -2.90502489e-01 2.41659507e-01 -3.22350800e-01
7.50704482e-02 8.79753053e-01 4.10500318e-01 2.98539490e-01
-8.39697719e-02 -4.27347332e-01 -7.44094610e-01 -8.47698092e-01
8.83971974e-02 1.13336250e-01 2.71658003e-01 9.78380889e-02
2.36080825e-01 2.79860944e-01 -1.46810222e+00 5.67112744e-01
1.08031440e+00 8.27843308e-01 4.03743416e-01 3.36825937e-01
-2.53573060e-01 7.36134529e-01 -3.07402223e-01 1.29058268e-02
5.17665446e-01 -5.96428752e-01 -4.28700775e-01 6.52395308e-01
1.00614607e+00 -3.67176324e-01 -5.55882514e-01 1.34835577e+00
1.06167391e-01 1.06992364e-01 3.80652130e-01 -1.04455268e+00
-5.87469757e-01 3.64377469e-01 -1.23292482e+00 1.19970158e-01
5.01246870e-01 2.26579949e-01 6.55357718e-01 -1.21520102e+00
5.78468740e-01 1.47551751e+00 5.14517367e-01 9.56687778e-02
-1.22019660e+00 -5.73221803e-01 4.54423845e-01 2.72005528e-01
-1.56956398e+00 -5.33223927e-01 9.12577868e-01 -3.68827105e-01
2.30223492e-01 5.57279110e-01 8.46397460e-01 6.13740325e-01
3.36319149e-01 7.32705772e-01 1.24831057e+00 -3.16950649e-01
1.24985492e-02 -2.42517889e-01 -1.03393041e-01 9.05846357e-01
4.23157573e-01 -1.22612044e-01 -8.05303752e-01 -3.19579154e-01
1.21823037e+00 3.91326934e-01 -6.06853664e-01 -4.89713967e-01
-1.65666342e+00 4.55574393e-01 3.68760496e-01 4.09829378e-01
-6.10762596e-01 1.83026120e-01 4.27160296e-04 2.84706086e-01
3.48369956e-01 1.70144856e-01 9.86168254e-03 -1.32298678e-01
-1.23263919e+00 4.09678109e-02 4.29455280e-01 8.84088516e-01
1.11870801e+00 4.09880579e-01 -2.48203456e-01 1.04340184e+00
3.65787208e-01 6.62864864e-01 4.33606148e-01 -1.18266463e+00
7.04920292e-01 4.58450884e-01 3.25803757e-01 -1.69401789e+00
1.18824452e-01 -2.68251926e-01 -1.19807470e+00 2.02034965e-01
2.56793857e-01 2.51146317e-01 -1.22276425e+00 1.46593511e+00
2.58785069e-01 6.06952488e-01 -1.76658571e-01 1.18833649e+00
7.31087387e-01 1.01615632e+00 -4.26752955e-01 -6.44829392e-01
1.05681527e+00 -1.12016702e+00 -1.16490722e+00 -1.94008544e-01
-1.16429225e-01 -1.23346412e+00 8.38642776e-01 3.69117081e-01
-1.57563186e+00 -5.55400968e-01 -1.00601363e+00 1.67415142e-01
3.11772972e-01 8.97701830e-02 3.35359097e-01 1.53818026e-01
-1.05958164e+00 3.79779726e-01 -8.11490059e-01 -2.78811119e-02
2.80760884e-01 2.09572256e-01 -2.85063088e-01 -6.63287640e-01
-8.01899135e-01 4.74074304e-01 3.84895839e-02 3.87982696e-01
-9.26960051e-01 -3.10261399e-01 -8.01901519e-01 -1.60272613e-01
3.37355167e-01 -7.55109370e-01 5.63005209e-01 -1.24786806e+00
-1.18769205e+00 4.06044394e-01 -6.17256820e-01 -1.21174417e-01
4.93242770e-01 -3.68804485e-01 -3.49335164e-01 2.47303829e-01
2.19571277e-01 4.92983967e-01 1.34041965e+00 -1.61867011e+00
-4.11510110e-01 -7.54240975e-02 -3.57332468e-01 4.80310082e-01
-4.09332842e-01 -2.52696007e-01 -6.61144316e-01 -1.15885103e+00
9.14864242e-01 -7.26698339e-01 -3.52433026e-01 2.51713216e-01
-3.12534213e-01 4.80673522e-01 1.13431156e+00 -8.08463395e-01
1.29935980e+00 -2.28125501e+00 4.81824726e-01 4.30923790e-01
3.67328078e-01 8.86973441e-02 -2.23095968e-01 2.84732699e-01
2.37842146e-02 -2.57440925e-01 -4.31965888e-01 -1.58632293e-01
-6.56445503e-01 2.92535424e-01 -2.83846945e-01 6.34944737e-01
-5.26772179e-02 5.92058778e-01 -1.05809295e+00 -5.94330072e-01
3.46612483e-01 4.46435660e-01 -4.17135596e-01 1.82636991e-01
4.47399728e-02 5.25747895e-01 -4.67192531e-01 8.54074001e-01
9.60449159e-01 -2.04288691e-01 -8.84341002e-02 -7.44905710e-01
-1.09514207e-01 -1.32376477e-01 -1.61082292e+00 1.82086158e+00
-1.69227511e-01 5.19731998e-01 5.59229195e-01 -7.26950288e-01
8.28970373e-01 2.33608171e-01 8.33675981e-01 -3.76670480e-01
-1.12333149e-01 3.65073293e-01 -2.40414575e-01 -4.56416160e-01
6.65087879e-01 6.50750771e-02 4.97279823e-01 2.21603021e-01
-5.16362190e-01 -3.41693275e-02 1.00969926e-01 3.38176161e-01
9.33988869e-01 1.59952454e-02 -2.92441435e-03 -2.22256243e-01
6.42556310e-01 -2.12607197e-02 9.25461352e-01 4.95999038e-01
-7.25530386e-02 1.05142355e+00 -1.88168958e-01 -5.00162184e-01
-8.36851776e-01 -1.23605132e+00 1.88996166e-01 4.60932761e-01
8.10583770e-01 -5.24717748e-01 -3.49902749e-01 -2.40240559e-01
-2.11417168e-01 7.82414824e-02 -2.51676917e-01 -3.03920172e-02
-9.17516708e-01 -4.76076126e-01 -2.41497196e-02 2.14721128e-01
9.18771029e-01 -6.41943872e-01 -1.35037243e-01 3.32763553e-01
-5.29104888e-01 -9.18101013e-01 -9.51680720e-01 -3.25992465e-01
-1.12564230e+00 -9.33566153e-01 -1.01252830e+00 -1.04891109e+00
1.25064588e+00 1.06444573e+00 8.98298442e-01 3.34683031e-01
-1.97887778e-01 3.10739994e-01 -2.82076061e-01 3.94815654e-01
-1.33434191e-01 -6.05206311e-01 7.24881291e-02 4.37317550e-01
-3.25140715e-01 -6.48661613e-01 -7.35105634e-01 6.23058319e-01
-1.10552514e+00 4.35367167e-01 6.30786598e-01 1.05818832e+00
8.46669197e-01 1.89717308e-01 2.99423099e-01 -5.40002823e-01
2.82956570e-01 -3.22019309e-01 -4.28093970e-01 2.23836437e-01
-2.78138220e-01 -8.50823969e-02 4.82370853e-01 -5.91401935e-01
-1.08472657e+00 2.09132969e-01 3.32261771e-01 -1.00023365e+00
2.78383046e-01 6.09151244e-01 -2.04020455e-01 -1.91244170e-01
4.07079905e-01 5.84974468e-01 2.12096408e-01 -5.89922488e-01
3.13312858e-01 1.94383577e-01 6.42539918e-01 -6.73466742e-01
1.22112036e+00 7.33827710e-01 9.26114768e-02 -7.31705844e-01
-5.67829192e-01 -4.67541546e-01 -4.84541148e-01 -2.34726310e-01
5.01967371e-01 -1.23669922e+00 5.74568547e-02 6.19105875e-01
-9.33451951e-01 -2.41767932e-02 -1.55135974e-01 5.68349361e-01
-3.43287319e-01 9.02573228e-01 -7.94418395e-01 -5.94425738e-01
-2.61720508e-01 -1.25761795e+00 7.93379486e-01 1.70059115e-01
-7.78933614e-02 -7.35442877e-01 -1.85324758e-01 3.40682864e-01
5.11378467e-01 2.66534805e-01 5.11324525e-01 5.09769380e-01
-1.10495698e+00 3.93398814e-02 -2.74720013e-01 1.28770903e-01
7.18187571e-01 1.92635790e-01 -5.22337794e-01 -5.65727472e-01
1.07437707e-01 -3.69286686e-02 1.03582716e+00 6.44849062e-01
1.07682383e+00 -5.76321006e-01 -3.12378258e-01 7.71526515e-01
1.21848285e+00 6.69770241e-02 7.55341351e-01 2.49621823e-01
1.14945221e+00 3.72669607e-01 9.91270423e-01 4.62824047e-01
1.26725182e-01 6.79401755e-01 3.33953232e-01 -2.20260903e-01
-2.75974751e-01 -1.27719522e-01 5.67182422e-01 1.03803933e+00
-5.94292171e-02 -1.77248009e-02 -3.55054975e-01 6.30755424e-01
-2.23130989e+00 -1.09514165e+00 -3.57827336e-01 2.22153926e+00
9.04586375e-01 -1.59681201e-01 5.11924513e-02 2.74102911e-02
1.01431692e+00 4.48014200e-01 -3.17735672e-01 1.68736637e-01
-4.57589984e-01 -2.30581626e-01 2.74066508e-01 5.64735949e-01
-7.92515159e-01 8.24201643e-01 6.12641764e+00 8.59400272e-01
-1.01276314e+00 -7.83639103e-02 6.28220916e-01 -1.83039457e-01
-7.06684828e-01 2.45021150e-01 -4.38398182e-01 6.75769866e-01
6.71919212e-02 -1.89492889e-02 5.17044306e-01 4.50500607e-01
4.59140629e-01 -2.32209399e-01 -6.34744406e-01 1.44855261e+00
2.48363569e-01 -1.42013943e+00 3.00055087e-01 -1.78659081e-01
1.21916723e+00 -4.66159850e-01 1.38736963e-01 -3.55183244e-01
1.25398800e-01 -8.67811620e-01 7.63201475e-01 6.60524249e-01
7.21537173e-01 -5.94915569e-01 3.05947542e-01 5.97228184e-02
-1.41546071e+00 1.14766173e-01 -4.77042615e-01 1.06139086e-01
3.65150928e-01 1.04320920e+00 -3.47441465e-01 4.54653829e-01
5.50162613e-01 1.17378199e+00 -3.83553416e-01 1.32676923e+00
3.99715602e-02 4.81669098e-01 -3.77476841e-01 4.68384832e-01
2.12008342e-01 -6.08297050e-01 8.27064753e-01 8.13894331e-01
5.43901742e-01 1.69754833e-01 5.40155411e-01 7.01446772e-01
5.29410318e-02 7.95933828e-02 -5.43252230e-01 1.53117701e-01
4.77507710e-01 1.23400497e+00 -7.04276443e-01 -3.87073785e-01
-3.35425287e-01 1.19447887e+00 1.64259207e-02 6.44872069e-01
-7.04088509e-01 1.61121756e-01 6.34983182e-01 2.86903888e-01
1.64553210e-01 -6.00503445e-01 -3.14439505e-01 -1.53736675e+00
2.59872288e-01 -1.08103609e+00 4.14433628e-01 -7.31446445e-01
-1.31788492e+00 3.05993885e-01 -1.51076823e-01 -1.74060619e+00
2.75206077e-03 -3.86782759e-03 -7.99106359e-01 7.32003570e-01
-1.35505724e+00 -1.09385240e+00 -6.61485851e-01 8.28142762e-01
7.75586128e-01 -1.74084708e-01 2.35483781e-01 4.38526332e-01
-4.57237393e-01 2.31411561e-01 3.03879678e-01 -2.47222140e-01
9.36884999e-01 -8.59030664e-01 -1.30178839e-01 1.13907683e+00
1.94827989e-01 7.70656645e-01 5.95075786e-01 -1.03686869e+00
-1.74965847e+00 -1.02505469e+00 3.59067470e-01 1.38284326e-01
2.76841849e-01 9.55611691e-02 -1.09115458e+00 4.28656995e-01
2.72428453e-01 1.84449196e-01 2.99015999e-01 -3.07746261e-01
-2.01731950e-01 -3.45296055e-01 -1.04445422e+00 7.16734707e-01
1.03255439e+00 -1.00191534e-01 -2.35395029e-01 3.07537824e-01
3.49738926e-01 -5.35733223e-01 -6.00503683e-01 3.39593977e-01
5.79153836e-01 -9.71013069e-01 1.05298781e+00 3.14056836e-02
3.67898464e-01 -1.03243434e+00 -2.05164477e-01 -1.27810717e+00
-6.69944227e-01 -8.93983364e-01 -3.49491239e-01 1.25489974e+00
1.91329550e-02 -3.44985813e-01 7.43151546e-01 3.55088800e-01
-2.53941149e-01 -6.22821748e-01 -8.85045767e-01 -7.05004334e-01
-6.09372437e-01 1.10736638e-01 2.77903676e-01 1.05641544e+00
-2.81885684e-01 2.60948855e-02 -1.03705168e+00 2.08607152e-01
7.74878979e-01 5.33763170e-01 8.25660229e-01 -6.69979274e-01
-2.14437589e-01 -1.29165888e-01 -4.07516301e-01 -1.48257661e+00
-8.26670304e-02 -4.59644288e-01 3.30168754e-01 -1.64506185e+00
4.29345816e-01 -6.49107456e-01 -1.75136074e-01 2.37907499e-01
-4.58910048e-01 3.77368957e-01 1.29643261e-01 6.73401654e-01
-5.85603297e-01 7.17226982e-01 1.63394833e+00 -3.95818681e-01
-2.52399296e-01 -2.66210645e-01 -4.21728104e-01 6.71401799e-01
4.76928443e-01 -1.87117383e-01 -4.86860275e-01 -6.69086456e-01
3.05460840e-02 1.33849159e-01 2.06209451e-01 -1.07506096e+00
2.04586074e-01 -6.11980736e-01 7.46862769e-01 -8.09189916e-01
6.45654857e-01 -9.75936592e-01 5.62138915e-01 4.07545596e-01
2.90299896e-02 1.67911634e-01 3.75046842e-02 8.63709509e-01
-5.11475801e-01 -5.45936227e-02 8.83312166e-01 -2.31879786e-01
-6.84186101e-01 5.14873803e-01 -2.90301770e-01 -1.61733031e-01
9.81934786e-01 -4.89339650e-01 -4.51385900e-02 -7.55658388e-01
-6.15426302e-01 2.91170299e-01 8.91315818e-01 3.10365111e-01
1.23780394e+00 -1.64106297e+00 -8.31369460e-01 4.72510099e-01
-1.77898705e-01 1.52908310e-01 3.27815443e-01 8.52220476e-01
-7.00006604e-01 -2.46763706e-01 -3.19274068e-01 -8.39485824e-01
-1.55683935e+00 3.56648743e-01 3.10667008e-02 -1.54908514e-02
-6.31318390e-01 8.03318799e-01 3.76273245e-01 3.96918893e-01
7.89387599e-02 -2.11661801e-01 1.50986969e-01 -2.06525847e-01
6.26334429e-01 3.29737335e-01 -2.99171925e-01 -7.90012658e-01
-1.06123969e-01 8.74468863e-01 -2.14526221e-01 -1.43286526e-01
1.24346721e+00 -5.39381862e-01 -3.91363502e-01 3.92313391e-01
9.51307237e-01 2.31295213e-01 -1.44480503e+00 -3.62093508e-01
-4.60973829e-01 -1.34731472e+00 3.72092053e-02 -1.89460114e-01
-1.44663274e+00 5.10279000e-01 4.77363199e-01 -1.49051994e-01
1.25701427e+00 -4.29545790e-01 1.16939867e+00 4.21451181e-02
4.56892371e-01 -7.45317519e-01 5.71284831e-01 3.37053806e-01
1.01371086e+00 -1.06661952e+00 4.76988465e-01 -7.92590141e-01
-4.74147022e-01 1.10984695e+00 7.89154947e-01 -1.51235342e-01
3.84683967e-01 1.76206231e-01 -6.41924189e-03 -2.05904424e-01
-5.39642334e-01 1.19393848e-01 4.15502220e-01 3.62970263e-01
2.83189386e-01 -1.99984848e-01 -3.85584354e-01 -2.14244261e-01
3.36443782e-01 -2.80478448e-01 5.54103494e-01 1.07859063e+00
-6.36375487e-01 -1.10690904e+00 -9.23728645e-01 5.22579670e-01
-2.75226653e-01 -4.14804779e-02 -1.90198123e-01 2.60599554e-01
-6.08186573e-02 9.19407547e-01 -8.50216523e-02 -3.41521859e-01
-1.44527927e-01 -3.70544016e-01 5.95164299e-01 -5.06816745e-01
-2.20259041e-01 7.10417449e-01 3.45987105e-03 -5.29771268e-01
-6.15358233e-01 -6.06309235e-01 -1.05966711e+00 -2.49182329e-01
-3.33923548e-01 7.30252191e-02 1.71947539e-01 6.22865438e-01
2.69782603e-01 1.66371882e-01 8.08977962e-01 -9.38921690e-01
-1.55397013e-01 -6.89743757e-01 -6.78044736e-01 7.17206120e-01
3.77772063e-01 -7.23415971e-01 -3.38407457e-01 4.11404669e-01] | [10.856633186340332, -1.717824935913086] |
0b1deb24-bf95-45e8-8db2-25d466e9841c | hybrid-transducer-and-attention-based-encoder | 2305.03101 | null | https://arxiv.org/abs/2305.03101v1 | https://arxiv.org/pdf/2305.03101v1.pdf | Hybrid Transducer and Attention based Encoder-Decoder Modeling for Speech-to-Text Tasks | Transducer and Attention based Encoder-Decoder (AED) are two widely used frameworks for speech-to-text tasks. They are designed for different purposes and each has its own benefits and drawbacks for speech-to-text tasks. In order to leverage strengths of both modeling methods, we propose a solution by combining Transducer and Attention based Encoder-Decoder (TAED) for speech-to-text tasks. The new method leverages AED's strength in non-monotonic sequence to sequence learning while retaining Transducer's streaming property. In the proposed framework, Transducer and AED share the same speech encoder. The predictor in Transducer is replaced by the decoder in the AED model, and the outputs of the decoder are conditioned on the speech inputs instead of outputs from an unconditioned language model. The proposed solution ensures that the model is optimized by covering all possible read/write scenarios and creates a matched environment for streaming applications. We evaluate the proposed approach on the \textsc{MuST-C} dataset and the findings demonstrate that TAED performs significantly better than Transducer for offline automatic speech recognition (ASR) and speech-to-text translation (ST) tasks. In the streaming case, TAED outperforms Transducer in the ASR task and one ST direction while comparable results are achieved in another translation direction. | ['Juan Pino', 'Paden D. Tomasello', 'Xutai Ma', 'Ning Dong', 'Xinyue Chen', 'Hirofumi Inaguma', 'Anna Y. Sun', 'Yun Tang'] | 2023-05-04 | null | null | null | null | ['speech-to-text-translation'] | ['natural-language-processing'] | [ 7.49038339e-01 1.15601383e-01 5.79101071e-02 -2.86537081e-01
-1.10384762e+00 -2.95219302e-01 6.29774392e-01 -2.11073846e-01
-3.02879751e-01 2.99058437e-01 5.41174591e-01 -6.99925184e-01
2.38176063e-01 -2.12363571e-01 -7.24898577e-01 -6.39265120e-01
2.31447101e-01 3.96545291e-01 2.22537532e-01 -3.85298580e-01
4.11439948e-02 -4.13454659e-02 -1.36528802e+00 5.23941755e-01
7.54195571e-01 9.68598545e-01 8.55519176e-01 1.03773355e+00
-2.40591154e-01 1.07065392e+00 -4.25083458e-01 -4.52076524e-01
1.45202398e-01 -5.85494995e-01 -5.46199441e-01 -2.22947463e-01
5.19752577e-02 -2.42605180e-01 -6.82164431e-01 7.50073731e-01
9.98445451e-01 5.49223050e-02 6.19485736e-01 -8.46948087e-01
-7.23939478e-01 7.59286106e-01 -3.99771601e-01 3.89162868e-01
4.31948990e-01 2.34148148e-02 1.26463044e+00 -1.01758969e+00
1.78404152e-01 1.28822517e+00 2.34713376e-01 6.16397142e-01
-9.72114801e-01 -4.00764525e-01 1.41205788e-01 1.18988656e-01
-1.24410713e+00 -1.28578687e+00 5.19219041e-01 -2.99344033e-01
1.72199059e+00 4.14078921e-01 1.62485778e-01 1.23132515e+00
3.67252290e-01 1.05370235e+00 7.82766521e-01 -6.45013452e-01
1.48639798e-01 1.58882856e-01 -1.74438536e-01 2.45888799e-01
-4.80684102e-01 3.16197962e-01 -1.07208943e+00 2.13852286e-01
4.00996685e-01 -3.18321168e-01 -3.28915089e-01 2.47224405e-01
-1.27731764e+00 5.41322112e-01 -2.77866006e-01 3.19155306e-01
-5.04626274e-01 5.81814311e-02 7.18426108e-01 6.64688110e-01
3.85564178e-01 -4.39262576e-03 -5.36709309e-01 -6.75993145e-01
-9.91269529e-01 -8.47994611e-02 8.00884008e-01 1.26869571e+00
2.02350616e-01 5.75285673e-01 -5.88639557e-01 9.67124760e-01
4.00611371e-01 9.93505538e-01 8.95819604e-01 -3.14748883e-01
1.16909218e+00 -4.91461046e-02 -2.14906171e-01 -5.52004278e-01
1.28849834e-01 -6.06200933e-01 -8.16267610e-01 -4.94949281e-01
-2.01261491e-01 -8.86626393e-02 -9.45740581e-01 1.65235853e+00
2.25161016e-02 1.71816453e-01 3.02658707e-01 5.60619354e-01
7.03857780e-01 1.20748758e+00 -6.28937334e-02 -4.61333066e-01
1.12803185e+00 -1.06760168e+00 -1.00606263e+00 -4.83008534e-01
8.91756535e-01 -8.09394062e-01 1.09038031e+00 8.15880150e-02
-1.40841699e+00 -5.67647338e-01 -8.22631836e-01 -6.49314895e-02
1.37376502e-01 2.32124045e-01 -3.04535389e-01 5.34434140e-01
-1.31703651e+00 1.54122129e-01 -8.41152668e-01 -4.32752043e-01
-1.01007111e-01 2.95833796e-01 3.22925448e-02 2.29146376e-01
-1.33402574e+00 1.02248120e+00 2.07079634e-01 6.52813315e-02
-1.03991604e+00 -2.05184355e-01 -9.09986913e-01 4.23054218e-01
2.79208928e-01 -4.76119637e-01 1.64586473e+00 -8.58195305e-01
-2.02059555e+00 4.87000316e-01 -7.13633239e-01 -7.53151715e-01
2.87017196e-01 -2.25340202e-01 -5.33955932e-01 -3.86031456e-02
-1.53413713e-01 2.38621265e-01 1.03436923e+00 -7.59081602e-01
-6.76254153e-01 -1.95997074e-01 -5.10318875e-01 5.37241757e-01
-4.70516920e-01 3.91712904e-01 -4.17741984e-01 -7.15135753e-01
3.61762829e-02 -7.45947480e-01 2.68057585e-01 -6.85202062e-01
-4.09045130e-01 -1.33882523e-01 8.80708218e-01 -1.04064405e+00
1.74139071e+00 -2.26818585e+00 3.16193044e-01 -1.24752291e-01
-2.46537760e-01 5.32477498e-01 -3.41580391e-01 1.00009239e+00
-4.32015061e-02 -5.32443896e-02 -2.20546737e-01 -7.95298576e-01
5.91857322e-02 3.85263890e-01 -5.43285072e-01 2.25168154e-01
2.53826380e-01 1.14878595e+00 -5.08046806e-01 -5.75561762e-01
2.83823192e-01 2.91040987e-01 -2.89858371e-01 6.10822022e-01
-1.15307897e-01 4.27391618e-01 -4.52916056e-01 3.92824411e-01
2.35582992e-01 7.61541724e-02 2.53205001e-02 3.53865474e-01
-1.61066204e-01 1.00391960e+00 -8.12770963e-01 1.51411247e+00
-8.04299891e-01 6.61733270e-01 1.72212839e-01 -1.16790855e+00
1.06278896e+00 1.04597545e+00 1.21747769e-01 -1.21703470e+00
1.88342497e-01 3.80061805e-01 2.99723983e-01 -8.50410342e-01
5.44142663e-01 -2.59973705e-01 1.60317719e-01 4.88318354e-01
1.19401582e-01 -3.51562351e-02 -2.22786874e-01 2.11827219e-01
1.07641470e+00 -9.72650293e-03 2.79642969e-01 5.78457005e-02
6.23061180e-01 -5.40930986e-01 3.39381009e-01 8.01637232e-01
-6.46890029e-02 5.44873118e-01 2.42934778e-01 3.11269462e-01
-1.19249916e+00 -7.45384455e-01 2.99181402e-01 1.46499765e+00
-2.07177505e-01 -2.60006875e-01 -6.15267336e-01 -4.11231458e-01
-2.74499714e-01 1.07410860e+00 -2.59331256e-01 -1.50805756e-01
-6.72245264e-01 -9.81329679e-02 7.35244870e-01 5.05772650e-01
1.68013141e-01 -1.09227896e+00 -3.63819808e-01 4.39772576e-01
-4.71606463e-01 -1.24262381e+00 -9.99914706e-01 3.95539016e-01
-7.11224556e-01 -2.39769235e-01 -8.31720829e-01 -9.80056286e-01
2.44482160e-01 3.24234068e-01 8.07247043e-01 -3.39223385e-01
3.62060249e-01 2.21269369e-01 -7.31711507e-01 -3.86370629e-01
-1.01856041e+00 2.34655172e-01 4.61707488e-02 3.74013096e-01
1.29591063e-01 -5.54023683e-01 -2.18644172e-01 2.31153578e-01
-8.15929353e-01 2.17883170e-01 7.68908441e-01 8.71852517e-01
2.45413840e-01 -3.07166994e-01 8.06126297e-01 -5.12296975e-01
7.44578242e-01 -5.19232094e-01 -2.77698964e-01 2.79852509e-01
-6.20335102e-01 2.03496039e-01 8.67701471e-01 -5.23187220e-01
-1.14973962e+00 -5.47581762e-02 -5.82598865e-01 -4.88623649e-01
7.59073421e-02 6.54642761e-01 -2.80736625e-01 4.15575743e-01
2.89160550e-01 1.04848742e+00 -1.42277461e-02 -5.51717162e-01
1.96295798e-01 1.28684342e+00 4.07654047e-01 -3.33621204e-01
6.60101891e-01 -1.82660773e-01 -5.11718333e-01 -8.24084997e-01
-3.18626016e-01 -5.48851609e-01 -3.59521091e-01 -1.02859110e-01
7.26350009e-01 -8.89412284e-01 -4.95407283e-01 5.33723772e-01
-1.38750052e+00 -2.62725890e-01 -2.08146080e-01 6.26797616e-01
-6.77061617e-01 3.76293808e-01 -5.73840916e-01 -1.41769254e+00
-7.03883648e-01 -1.32351649e+00 1.38595629e+00 -3.14648509e-01
-1.60948634e-01 -9.14373815e-01 -9.73570049e-02 2.04198301e-01
7.56345868e-01 -5.73583543e-01 9.05889034e-01 -1.19736171e+00
-2.70492047e-01 -1.49139881e-01 4.66603935e-02 6.41981423e-01
-1.60348527e-02 -3.37320775e-01 -1.15323830e+00 -4.03490007e-01
3.49493891e-01 -2.14523837e-01 6.59099698e-01 3.58520210e-01
7.24879622e-01 -6.80880904e-01 -5.92123382e-02 2.81504631e-01
1.15187311e+00 6.49253488e-01 6.90729439e-01 -3.54744382e-02
5.30526519e-01 5.25266111e-01 3.50012571e-01 3.82307619e-01
3.83391649e-01 9.04794693e-01 1.59644112e-01 1.26105309e-01
-1.91330850e-01 -4.82157320e-01 1.12258208e+00 1.75808096e+00
4.55621243e-01 -8.23890686e-01 -7.48132765e-01 7.18086123e-01
-1.81073093e+00 -8.59170377e-01 -1.27295569e-01 2.36093593e+00
8.27672124e-01 1.97677493e-01 -6.16084673e-02 2.22901568e-01
5.66636503e-01 2.59066254e-01 -3.24689299e-01 -8.72953176e-01
-1.20776549e-01 4.08819318e-01 2.90626287e-01 6.16386712e-01
-4.90207434e-01 9.37185347e-01 5.83650064e+00 1.10248458e+00
-1.23638010e+00 4.37785327e-01 3.83742303e-01 -2.87156645e-02
-3.17634016e-01 2.41310941e-03 -7.21198738e-01 5.58699608e-01
1.57767105e+00 -2.78119475e-01 4.60233063e-01 4.39846605e-01
6.96687341e-01 1.22546457e-01 -1.40755057e+00 9.03303206e-01
1.08737588e-01 -8.71514916e-01 7.10484385e-02 -9.76745039e-02
2.81367630e-01 2.26232067e-01 1.40579805e-01 4.00638312e-01
-3.26113170e-03 -1.08890796e+00 1.02819753e+00 2.11513504e-01
1.10917461e+00 -2.69994289e-01 7.27166116e-01 7.35202849e-01
-1.25628996e+00 -2.09269021e-02 3.26453261e-02 -9.24196318e-02
4.57709998e-01 1.88707665e-01 -1.10502601e+00 6.49793088e-01
1.98138058e-01 6.04115605e-01 4.40683290e-02 5.03762901e-01
-3.81947192e-03 1.26865113e+00 -2.07015783e-01 -2.19945312e-01
2.89171338e-01 1.21573061e-02 9.07811344e-01 1.53839111e+00
6.43607736e-01 -1.61899775e-01 -1.55143857e-01 5.46048582e-01
2.73303483e-02 3.82459700e-01 -6.94235146e-01 -2.46956974e-01
8.35393727e-01 6.47081912e-01 -5.96849844e-02 -3.69294494e-01
-5.11708558e-01 1.08943677e+00 1.12201691e-01 4.96800154e-01
-8.69337916e-01 -3.07859421e-01 3.13323468e-01 3.56052294e-02
6.03150368e-01 -3.23147684e-01 -2.50605971e-01 -9.49902356e-01
3.17751795e-01 -1.19884789e+00 1.00662664e-01 -7.89102614e-01
-8.70510995e-01 7.89499462e-01 -2.75898486e-01 -1.19520020e+00
-4.84243393e-01 -2.22057462e-01 -6.59400702e-01 1.21300197e+00
-1.64166260e+00 -1.04484630e+00 3.07895601e-01 5.22334218e-01
1.07704139e+00 -3.19981039e-01 8.02639306e-01 4.50795263e-01
-6.53269768e-01 8.43815327e-01 4.35969174e-01 -1.43679343e-02
4.51946735e-01 -9.49041665e-01 8.28509212e-01 1.08740377e+00
2.01054752e-01 2.90826052e-01 6.57071888e-01 -5.32205164e-01
-1.81929314e+00 -1.13617480e+00 1.39634705e+00 -2.09013656e-01
6.12173736e-01 -6.56131327e-01 -9.32208300e-01 7.96424031e-01
4.76223558e-01 -5.96644282e-01 4.11644220e-01 -3.27391386e-01
-1.04619116e-01 -1.11371703e-01 -6.59140050e-01 5.22021651e-01
8.09763253e-01 -8.97202492e-01 -5.60810089e-01 1.70994371e-01
1.17149472e+00 -4.58870858e-01 -7.15689182e-01 -2.70845536e-02
3.61921549e-01 -6.05473518e-01 4.78399515e-01 -4.91262197e-01
4.50454146e-01 -7.24410862e-02 -5.42754114e-01 -1.24949789e+00
-6.00087717e-02 -1.23271871e+00 -3.39871317e-01 1.41531217e+00
6.79145753e-01 -8.74190807e-01 1.45600706e-01 2.66190618e-01
-6.44655287e-01 -7.40892529e-01 -1.39785123e+00 -1.03272963e+00
-3.10124774e-02 -6.88796461e-01 5.17668962e-01 3.62088948e-01
5.67722097e-02 6.62132263e-01 -7.24762797e-01 1.70906842e-01
1.32595465e-01 -3.88797849e-01 6.06444657e-01 -4.25269395e-01
-6.16886854e-01 -3.09670806e-01 -3.45616788e-02 -1.69173539e+00
3.88265848e-02 -1.07053006e+00 4.12409276e-01 -1.41955578e+00
6.00868501e-02 -2.55410910e-01 -1.51060253e-01 3.01497489e-01
-4.00407128e-02 -4.07598168e-01 3.32096308e-01 2.77974606e-01
-3.07981610e-01 9.17688549e-01 1.07187200e+00 -1.30448136e-02
-1.58833131e-01 1.99730128e-01 -4.07220334e-01 -2.95295250e-02
5.94063759e-01 -5.23223281e-01 -4.68920171e-01 -8.28741252e-01
-8.65108743e-02 6.27887964e-01 -1.28733292e-01 -7.24264622e-01
5.16416132e-01 5.90304546e-02 -4.36620981e-01 -5.83218455e-01
4.47273821e-01 -6.74585998e-01 -6.38393611e-02 2.76587397e-01
-7.41406500e-01 2.55125403e-01 5.30238487e-02 5.77037454e-01
-3.62050325e-01 -1.69124454e-01 5.75097740e-01 2.31071621e-01
-1.60747603e-01 1.62468076e-01 -7.68964767e-01 1.34866163e-01
6.02318943e-01 -2.62670189e-01 -5.80542311e-02 -9.62711811e-01
-4.37893778e-01 2.45070487e-01 -1.18837893e-01 5.54598212e-01
8.37791026e-01 -1.00178683e+00 -1.19183946e+00 4.17774916e-01
7.63762817e-02 -1.41309485e-01 4.06013392e-02 1.18190908e+00
1.02159359e-01 8.12702775e-01 4.39808398e-01 -6.31594658e-01
-1.36621606e+00 4.18156415e-01 3.35566670e-01 -3.84969890e-01
-4.66772825e-01 8.02066565e-01 3.37410718e-01 -2.72096813e-01
5.18414080e-01 -2.82929391e-01 9.91756245e-02 -3.16331714e-01
2.44614050e-01 1.97580978e-01 4.07299906e-01 -8.67306471e-01
-2.40524590e-01 1.15400314e-01 -2.46034682e-01 -4.86385852e-01
1.24986470e+00 -5.63587487e-01 1.63000450e-01 6.91809475e-01
1.28223395e+00 -1.78345665e-02 -7.77598798e-01 -6.49028301e-01
3.91555130e-02 -1.46267951e-01 1.09264925e-01 -7.30080664e-01
-8.56188953e-01 1.28070760e+00 4.26981509e-01 3.08991909e-01
1.22705424e+00 -8.94864947e-02 1.29266226e+00 5.01623005e-02
3.11858878e-02 -1.14572918e+00 -2.42646504e-02 9.29547369e-01
8.96633148e-01 -9.18320417e-01 -6.50573909e-01 -1.29463181e-01
-9.93490636e-01 9.01664734e-01 2.84433991e-01 4.92044091e-01
3.32527846e-01 6.50817335e-01 -2.53473390e-02 1.27932161e-01
-1.35888720e+00 -1.58227637e-01 1.87315702e-01 6.00797236e-01
6.99940443e-01 1.10656023e-02 -9.86291468e-02 4.78155375e-01
-2.96904951e-01 -1.78515926e-01 3.05476636e-01 9.05733049e-01
-4.62553501e-01 -1.02433670e+00 -2.67424524e-01 4.24058437e-01
-4.46063370e-01 -6.36403024e-01 -2.13660985e-01 1.79446355e-01
-5.10613143e-01 1.23851228e+00 -3.59170139e-02 -5.96562147e-01
2.83967912e-01 4.70269203e-01 6.34943098e-02 -6.50341809e-01
-7.59784698e-01 4.20011073e-01 3.57219815e-01 -2.79105455e-01
-3.89753245e-02 -6.23208582e-01 -1.07991743e+00 -5.48605658e-02
-6.97307944e-01 1.57299012e-01 7.54554987e-01 1.03648567e+00
6.54398680e-01 7.15102017e-01 1.00164890e+00 -4.44099724e-01
-1.06237125e+00 -1.28248417e+00 -4.87817287e-01 1.46372736e-01
5.53442240e-01 -1.20805055e-01 -2.56371230e-01 2.22354561e-01] | [14.47297477722168, 6.9319281578063965] |
34580d53-2e51-4fbc-9a79-59ec0ab6b3de | dual-domain-image-synthesis-using | 2204.09015 | null | https://arxiv.org/abs/2204.09015v1 | https://arxiv.org/pdf/2204.09015v1.pdf | Dual-Domain Image Synthesis using Segmentation-Guided GAN | We introduce a segmentation-guided approach to synthesise images that integrate features from two distinct domains. Images synthesised by our dual-domain model belong to one domain within the semantic mask, and to another in the rest of the image - smoothly integrated. We build on the successes of few-shot StyleGAN and single-shot semantic segmentation to minimise the amount of training required in utilising two domains. The method combines a few-shot cross-domain StyleGAN with a latent optimiser to achieve images containing features of two distinct domains. We use a segmentation-guided perceptual loss, which compares both pixel-level and activations between domain-specific and dual-domain synthetic images. Results demonstrate qualitatively and quantitatively that our model is capable of synthesising dual-domain images on a variety of objects (faces, horses, cats, cars), domains (natural, caricature, sketches) and part-based masks (eyes, nose, mouth, hair, car bonnet). The code is publicly available at: https://github.com/denabazazian/Dual-Domain-Synthesis. | ['Dima Damen', 'Andrew Calway', 'Dena Bazazian'] | 2022-04-19 | null | null | null | null | ['caricature'] | ['computer-vision'] | [ 6.84453428e-01 5.25883675e-01 7.77963921e-02 -3.77830803e-01
-8.02837789e-01 -7.85640955e-01 1.03814638e+00 -6.67712510e-01
7.10758194e-02 6.96523845e-01 1.26515225e-01 1.85929477e-01
2.16061428e-01 -8.70045602e-01 -7.63976097e-01 -7.48008430e-01
4.36258405e-01 3.59219283e-01 2.35277370e-01 -4.43633378e-01
1.93476509e-02 4.65267211e-01 -1.51273143e+00 5.74180841e-01
3.95867586e-01 9.64775324e-01 1.46467015e-01 7.99317896e-01
-2.90945508e-02 6.90747142e-01 -3.72836173e-01 -5.37696183e-01
5.89643240e-01 -1.02881575e+00 -7.88085818e-01 5.66609204e-01
4.24239337e-01 -2.73660094e-01 -1.57577708e-01 1.01015759e+00
3.36905658e-01 8.49485248e-02 8.33414912e-01 -1.35806119e+00
-7.87399650e-01 1.92423478e-01 -6.12293899e-01 -2.64730781e-01
3.77132505e-01 7.64318883e-01 9.33022201e-01 -8.35676849e-01
1.06383777e+00 1.53309214e+00 5.27488649e-01 9.84596550e-01
-1.67447817e+00 -6.30634010e-01 -4.89615798e-02 -4.23019946e-01
-1.13473606e+00 -7.99989283e-01 7.22005010e-01 -4.65221286e-01
6.19828939e-01 -6.63782295e-04 5.92410207e-01 1.46527421e+00
-3.41078527e-02 7.40303218e-01 1.35577583e+00 -3.98999512e-01
2.57651269e-01 4.24790025e-01 -6.24990821e-01 6.08576775e-01
-1.03346594e-01 5.07073700e-01 -2.83810407e-01 1.32395834e-01
1.04815876e+00 -6.65098280e-02 1.20567441e-01 -5.95204711e-01
-9.86568928e-01 8.26871634e-01 2.39281416e-01 2.44889110e-01
-3.73400569e-01 3.58718187e-01 6.82927817e-02 3.61912489e-01
5.47819495e-01 4.36456650e-01 -3.14224243e-01 1.79832488e-01
-1.25818801e+00 3.81816804e-01 6.54261351e-01 1.13182771e+00
9.26626563e-01 2.45085239e-01 -1.77435756e-01 9.39352393e-01
1.63652197e-01 4.57133651e-01 2.84154534e-01 -1.44656599e+00
-8.37670714e-02 3.44914466e-01 4.74442728e-02 -6.19932055e-01
1.86753064e-01 1.95995837e-01 -5.97026169e-01 7.13598609e-01
3.92680794e-01 -3.57273012e-01 -1.49028718e+00 1.79093742e+00
3.45547318e-01 1.42101839e-01 1.17988683e-01 8.74598324e-01
9.18505073e-01 5.85930765e-01 1.91370204e-01 4.06733274e-01
1.55022156e+00 -1.00925124e+00 -4.05595481e-01 -4.89533007e-01
1.02818005e-01 -7.54773557e-01 1.02469981e+00 1.88133374e-01
-1.56127238e+00 -6.78885579e-01 -1.00992823e+00 -1.25445977e-01
-5.53277850e-01 -2.22461104e-01 2.24924356e-01 6.89891577e-01
-1.24600577e+00 5.43984652e-01 -4.14172381e-01 -4.29765940e-01
8.25325370e-01 1.28540233e-01 -5.49115181e-01 -2.20322892e-01
-1.13485968e+00 9.38776016e-01 4.29649830e-01 -6.44255936e-01
-1.48902190e+00 -7.20589697e-01 -1.16985762e+00 -5.70981503e-02
2.20595151e-01 -8.52549911e-01 1.33576488e+00 -1.68638182e+00
-1.64843285e+00 1.42742443e+00 1.03187464e-01 -4.22943294e-01
6.11813724e-01 2.65229613e-01 -2.19621420e-01 4.70074028e-01
2.68471301e-01 1.49349427e+00 1.35952997e+00 -1.56574738e+00
-5.90009928e-01 -3.76321934e-02 1.64965644e-01 2.23346755e-01
3.86010259e-01 1.53290898e-01 -3.14365357e-01 -7.44605660e-01
-4.02692109e-01 -7.88687468e-01 -2.37195745e-01 3.76511157e-01
-3.87503803e-01 2.30251536e-01 9.31866527e-01 -6.98585987e-01
3.69532496e-01 -2.38528919e+00 2.03127429e-01 -4.69619110e-02
1.35677680e-01 1.91338733e-01 -3.72023404e-01 3.89779627e-01
-2.35222533e-01 2.01180100e-01 -8.78291368e-01 -5.43731987e-01
-3.49603477e-03 3.63485813e-01 -1.43674329e-01 4.17953193e-01
7.74878502e-01 1.15152168e+00 -9.14807558e-01 -4.95121211e-01
3.87114882e-01 5.19169211e-01 -4.33958828e-01 2.76961654e-01
-5.97202301e-01 6.96193099e-01 -1.11102648e-01 7.52450943e-01
7.76796043e-01 -9.50124040e-02 2.31588185e-02 2.18114078e-01
7.54898265e-02 -7.52643794e-02 -1.02851284e+00 1.83928037e+00
-4.15133685e-01 6.66937292e-01 5.57763457e-01 -8.54601860e-01
7.93734550e-01 3.91672671e-01 2.01605916e-01 -8.10063481e-01
2.03151181e-01 2.70453036e-01 -2.37041086e-01 -2.69252092e-01
1.49518669e-01 -7.16959536e-01 -2.81017125e-01 4.95343715e-01
5.95026433e-01 -8.23227108e-01 1.29005760e-01 9.20280665e-02
8.12252998e-01 4.55163956e-01 2.63318628e-01 -2.11073592e-01
3.77651960e-01 6.62576184e-02 1.88976169e-01 5.09408593e-01
-2.84573674e-01 1.12861216e+00 7.36573040e-01 -1.12230986e-01
-1.56498373e+00 -1.26569104e+00 9.41522419e-02 1.08749115e+00
1.15336865e-01 2.47525617e-01 -1.05786431e+00 -5.09858966e-01
5.85136190e-02 8.79305124e-01 -7.86570966e-01 -2.44445130e-01
-3.48692656e-01 3.11745908e-02 7.42907584e-01 2.39267200e-01
4.75528657e-01 -1.42014194e+00 -7.74501204e-01 -5.29008992e-02
1.70787752e-01 -1.07421124e+00 -4.17676032e-01 1.09599084e-01
-6.15438282e-01 -8.36755157e-01 -1.30143368e+00 -1.03922462e+00
5.65137029e-01 -1.80284996e-02 1.27898347e+00 -2.36152917e-01
-4.40241992e-01 5.94086230e-01 -2.21158966e-01 -2.77460605e-01
-5.18125296e-01 -4.07161206e-01 -4.01415229e-01 2.11856395e-01
9.96253192e-02 -5.66622019e-01 -7.05725610e-01 3.64819437e-01
-1.09400344e+00 1.03091180e-01 5.62122881e-01 9.15510297e-01
3.95716846e-01 -3.30972403e-01 4.77079958e-01 -1.07380962e+00
5.00829756e-01 -5.79484403e-01 -4.33463275e-01 -5.21893986e-02
-1.13847673e-01 -1.91698700e-01 3.82359326e-01 -3.80364925e-01
-1.28903103e+00 3.25537682e-01 1.92647465e-02 -6.19098604e-01
-6.05339110e-01 -2.39314899e-01 -2.64758348e-01 -1.02668144e-01
7.54420400e-01 2.86105752e-01 4.77153718e-01 -2.01011986e-01
1.05868208e+00 4.17642891e-01 5.99848986e-01 -6.02181554e-01
8.39284837e-01 8.53901565e-01 -2.97517449e-01 -8.39038610e-01
-4.62357104e-01 -2.14994699e-01 -6.15105510e-01 -7.77870715e-02
1.18913245e+00 -1.08094752e+00 3.42228040e-02 4.90704775e-01
-1.16779602e+00 -6.77976251e-01 -9.73176360e-01 -1.73321486e-01
-1.08870971e+00 2.15298925e-02 -3.76820207e-01 -6.00497842e-01
-1.34458989e-01 -1.09745193e+00 1.37201917e+00 2.63109505e-01
-4.10447270e-01 -1.01212358e+00 -1.04439363e-01 1.44098774e-01
2.67079949e-01 6.55708432e-01 5.80639541e-01 -4.75949496e-01
-6.14016175e-01 8.84127691e-02 -4.58474159e-01 6.94243073e-01
5.12936339e-02 -3.66133898e-02 -1.19230926e+00 -1.55378357e-01
1.29966112e-02 -6.17605388e-01 8.80011141e-01 5.50356627e-01
6.62489176e-01 -2.51208365e-01 -1.66591659e-01 6.64278924e-01
1.47769499e+00 3.23832124e-01 9.14257467e-01 -1.51486799e-01
4.03566390e-01 1.03630042e+00 5.45868337e-01 1.91995546e-01
-2.84010619e-02 3.93075258e-01 3.40167254e-01 -5.53986430e-01
-6.33511007e-01 -3.30952615e-01 3.55518699e-01 -1.82137728e-01
3.07372123e-01 -2.18336940e-01 -7.04492450e-01 1.15706801e+00
-1.48370838e+00 -9.44352865e-01 2.70463049e-01 1.63723767e+00
8.86277020e-01 4.80195358e-02 4.47165370e-01 -2.26055041e-01
8.65484893e-01 2.54194587e-01 -5.69441736e-01 -7.89245844e-01
-2.43848726e-01 5.99332094e-01 4.17834520e-01 5.15513003e-01
-9.63729799e-01 1.28618157e+00 6.51310062e+00 8.45729232e-01
-9.60795283e-01 3.03109974e-01 9.12888288e-01 -1.21364430e-01
-4.52140868e-01 1.13163747e-01 -3.63024801e-01 5.44767261e-01
8.50908697e-01 -4.65421379e-02 4.98444766e-01 7.92028725e-01
-8.14169943e-02 -1.09067494e-02 -1.09895277e+00 7.45064259e-01
7.94046745e-02 -1.51902461e+00 -1.33633316e-01 -2.33908836e-02
1.17060280e+00 -2.18780085e-01 4.80623662e-01 1.10451832e-01
7.60344267e-01 -1.26287985e+00 1.04554605e+00 4.22285229e-01
1.52918243e+00 -4.76730406e-01 1.52447656e-01 1.07862391e-01
-8.23187709e-01 6.92380965e-02 -9.98513475e-02 4.10457514e-02
1.56135395e-01 7.95601904e-02 -7.21588194e-01 1.90744027e-01
5.76564312e-01 6.93236887e-01 -1.53845489e-01 4.18664068e-01
-3.72070312e-01 1.68830216e-01 -1.60737962e-01 4.05635357e-01
5.46437442e-01 -2.50840366e-01 5.03767550e-01 1.22730780e+00
1.53131470e-01 1.58977851e-01 -1.25588015e-01 1.53844464e+00
-7.16247261e-02 -4.72659320e-01 -1.06553328e+00 -2.38048583e-01
2.26856649e-01 1.13802195e+00 -8.04340482e-01 -6.02679849e-01
-3.69356632e-01 1.30115414e+00 -2.55336553e-01 4.47059751e-01
-7.61638582e-01 -5.22635102e-01 9.67133105e-01 2.31852040e-01
6.57189488e-01 8.63212943e-02 -5.84813654e-01 -7.68985450e-01
-4.40105051e-01 -9.92398202e-01 9.42193568e-02 -9.55481589e-01
-1.29789925e+00 4.16299939e-01 1.34411752e-01 -1.14694369e+00
-3.94140452e-01 -5.84649980e-01 -7.12089181e-01 1.14609885e+00
-1.34030783e+00 -1.46269095e+00 -1.57337561e-01 6.66096628e-01
7.59903967e-01 -2.11630747e-01 7.70606399e-01 1.94220215e-01
-6.90044761e-02 4.03318614e-01 -8.27333778e-02 1.63509205e-01
6.64338708e-01 -1.06922090e+00 8.92464161e-01 6.85859978e-01
-1.93439364e-01 3.80030364e-01 6.47533834e-01 -5.70170879e-01
-8.34933996e-01 -1.06665456e+00 4.99934822e-01 -4.73255873e-01
3.61009270e-01 -5.51140428e-01 -4.51983601e-01 7.69695938e-01
6.42445385e-01 4.76810075e-02 4.07516778e-01 -5.73863626e-01
-3.08822900e-01 2.18873382e-01 -1.71198082e+00 6.61168218e-01
9.62289214e-01 -6.55745566e-01 -6.47474289e-01 1.21241458e-01
7.40861177e-01 -2.05631375e-01 -6.84086144e-01 4.20019999e-02
6.21684909e-01 -1.14104402e+00 1.11801684e+00 -5.32933354e-01
9.20649529e-01 -1.06983252e-01 -1.21425077e-01 -1.34478438e+00
-1.90457910e-01 -6.52610838e-01 2.59846270e-01 1.21153688e+00
3.70917380e-01 -5.64209342e-01 7.82455504e-01 2.57340401e-01
-2.65972875e-02 -3.80549043e-01 -7.72014380e-01 -7.15790451e-01
2.48439521e-01 -1.21606037e-01 6.53542459e-01 7.78785288e-01
-4.55377311e-01 2.59599954e-01 -3.60491931e-01 -3.51054460e-01
5.99792123e-01 3.90210785e-02 7.98081815e-01 -8.42810571e-01
-2.65707403e-01 -5.83236277e-01 -3.45528752e-01 -6.85579121e-01
-1.21929608e-02 -8.10639083e-01 1.26408577e-01 -1.48315883e+00
-8.46646205e-02 -2.48544946e-01 5.87137090e-03 5.64440310e-01
3.41681927e-01 6.42607152e-01 3.47569615e-01 -3.41518223e-02
-3.08795720e-01 2.80472577e-01 1.62235975e+00 -1.54034987e-01
6.19472265e-02 -3.51752102e-01 -9.70582902e-01 6.76944375e-01
7.32475162e-01 -5.42204082e-01 -4.37040985e-01 -2.14484364e-01
-2.90437669e-01 1.35956600e-01 6.46559536e-01 -8.78138006e-01
-1.56055823e-01 -2.23235518e-01 5.49096763e-01 -1.28388964e-02
7.73284495e-01 -6.94903970e-01 3.98643583e-01 5.17053187e-01
-3.98876101e-01 -6.32111907e-01 2.31834218e-01 5.75713694e-01
-2.31102332e-01 -8.73890370e-02 1.52784061e+00 -8.07413936e-01
-1.00022471e+00 1.77153319e-01 -2.35378578e-01 1.85446680e-01
1.38878369e+00 -7.74178028e-01 -2.86659636e-02 -5.08841276e-01
-9.15416539e-01 3.81764509e-02 9.86071110e-01 3.76029849e-01
7.90558100e-01 -1.28442967e+00 -8.27231050e-01 6.29329503e-01
9.10062641e-02 3.35099809e-02 2.41104424e-01 5.82684755e-01
-6.67865455e-01 1.82710320e-01 -7.91312993e-01 -6.74653947e-01
-1.01161516e+00 4.78064448e-01 3.97886187e-01 5.65062426e-02
-4.29677874e-01 1.32975268e+00 7.75757611e-01 -5.87202907e-01
-1.59883320e-01 -9.21100155e-02 3.34336430e-01 1.38628870e-01
2.50544399e-01 8.68362039e-02 -4.64948893e-01 -7.71608174e-01
-2.13578805e-01 6.21929169e-01 8.00496042e-02 -5.11105537e-01
1.23387384e+00 -9.87048745e-02 1.39863966e-02 3.14707041e-01
1.24984646e+00 -3.53832066e-01 -1.78082645e+00 -8.76783133e-02
-2.54922569e-01 -5.37568271e-01 -2.48670429e-01 -9.98968422e-01
-9.28415775e-01 9.94350910e-01 5.28263688e-01 8.83560181e-02
1.24604952e+00 1.95471346e-01 8.64269555e-01 -5.38346231e-01
1.75557405e-01 -1.03534234e+00 3.49492699e-01 3.32855254e-01
8.80272746e-01 -1.17548144e+00 -2.42668748e-01 -2.08252534e-01
-1.17983556e+00 7.05526888e-01 3.67834955e-01 -6.58613980e-01
4.23318595e-01 7.50696659e-02 -5.23609146e-02 -2.92429537e-01
-6.81201756e-01 -4.67923820e-01 2.33350456e-01 1.04184711e+00
1.38826251e-01 8.26303214e-02 8.39633420e-02 2.51015335e-01
6.35161698e-02 -8.24628919e-02 3.68193746e-01 9.26107705e-01
-3.46157044e-01 -1.05382192e+00 -3.32321733e-01 2.52490610e-01
-3.88225108e-01 -1.28727496e-01 -8.05340886e-01 9.02902663e-01
5.15578628e-01 8.56057763e-01 2.20179066e-01 -1.30652770e-01
1.19108975e-01 2.81245530e-01 6.44372106e-01 -9.17250931e-01
-4.94812101e-01 2.06154048e-01 1.36766180e-01 -5.55814505e-01
-3.78386497e-01 -5.80807745e-01 -9.95544553e-01 -2.36750081e-01
9.54333097e-02 -4.56563324e-01 7.17230260e-01 6.19247496e-01
4.20390427e-01 4.73480970e-01 4.57882971e-01 -1.14175212e+00
-1.50823206e-01 -6.72102988e-01 -7.23656952e-01 6.08923018e-01
4.73145336e-01 -6.85647130e-01 -2.77311891e-01 4.71432626e-01] | [11.654657363891602, -0.43046432733535767] |
d92c46b3-b198-47ce-a6fc-06dd3568a007 | a-temporal-learning-approach-to-inpainting | 2203.17013 | null | https://arxiv.org/abs/2203.17013v1 | https://arxiv.org/pdf/2203.17013v1.pdf | A Temporal Learning Approach to Inpainting Endoscopic Specularities and Its effect on Image Correspondence | Video streams are utilised to guide minimally-invasive surgery and diagnostic procedures in a wide range of procedures, and many computer assisted techniques have been developed to automatically analyse them. These approaches can provide additional information to the surgeon such as lesion detection, instrument navigation, or anatomy 3D shape modeling. However, the necessary image features to recognise these patterns are not always reliably detected due to the presence of irregular light patterns such as specular highlight reflections. In this paper, we aim at removing specular highlights from endoscopic videos using machine learning. We propose using a temporal generative adversarial network (GAN) to inpaint the hidden anatomy under specularities, inferring its appearance spatially and from neighbouring frames where they are not present in the same location. This is achieved using in-vivo data of gastric endoscopy (Hyper-Kvasir) in a fully unsupervised manner that relies on automatic detection of specular highlights. System evaluations show significant improvements to traditional methods through direct comparison as well as other machine learning techniques through an ablation study that depicts the importance of the network's temporal and transfer learning components. The generalizability of our system to different surgical setups and procedures was also evaluated qualitatively on in-vivo data of gastric endoscopy and ex-vivo porcine data (SERV-CT, SCARED). We also assess the effect of our method in computer vision tasks that underpin 3D reconstruction and camera motion estimation, namely stereo disparity, optical flow, and sparse point feature matching. These are evaluated quantitatively and qualitatively and results show a positive effect of specular highlight inpainting on these tasks in a novel comprehensive analysis. | ['Danail Stoyanov', 'Francisco Vasconcelos', 'Rema Daher'] | 2022-03-31 | null | null | null | null | ['3d-shape-modeling'] | ['computer-vision'] | [ 3.72499466e-01 1.93044424e-01 1.97836161e-01 2.13830441e-01
-6.76825881e-01 -7.96634674e-01 4.08864230e-01 -1.55379340e-01
-3.73488188e-01 4.41011369e-01 2.26870507e-01 -2.05526814e-01
-3.14783156e-02 -3.66094232e-01 -7.51266003e-01 -9.60991204e-01
-4.04389948e-01 5.92545234e-02 1.55882537e-01 -1.36771485e-01
3.61342286e-03 6.29586101e-01 -1.25173247e+00 4.82651234e-01
5.14818907e-01 7.51028955e-01 1.70056507e-01 9.79102671e-01
2.49374956e-01 7.62387097e-01 -4.62386280e-01 -1.47105619e-01
7.02728570e-01 -5.76015115e-01 -4.00458217e-01 2.09498987e-01
4.17336524e-01 -4.57961738e-01 -2.90631443e-01 8.94380450e-01
5.64663768e-01 9.11717936e-02 6.03852153e-01 -6.19399667e-01
5.82591631e-02 1.45536344e-02 -4.22529787e-01 3.12774807e-01
6.03795707e-01 3.22289169e-01 2.08206668e-01 -5.76591849e-01
8.68956208e-01 7.22880244e-01 9.93750453e-01 3.82244557e-01
-1.19990325e+00 -3.47647876e-01 -3.99668813e-01 -1.07475184e-01
-6.63511574e-01 -4.81059551e-02 8.58866394e-01 -5.75689316e-01
6.02236271e-01 3.25938374e-01 9.05593991e-01 9.04175758e-01
6.30362749e-01 5.56612551e-01 1.19286025e+00 -7.21458673e-01
-1.25953071e-02 3.69800895e-01 -6.90334380e-01 1.00317967e+00
4.21017446e-02 8.77249002e-01 -3.06667566e-01 -1.94058016e-01
1.20387721e+00 2.99956679e-01 -8.80929172e-01 -7.34371364e-01
-1.26551032e+00 8.48666668e-01 4.87981021e-01 3.19867432e-01
-5.98122299e-01 -9.61427689e-02 3.92158926e-01 2.86498487e-01
1.35319963e-01 5.23319364e-01 -1.00969821e-01 7.59523213e-02
-6.64629698e-01 -3.66829216e-01 1.13328648e+00 7.12039053e-01
5.10042191e-01 5.25856391e-02 1.53041556e-01 5.87127566e-01
1.78632885e-01 1.49747521e-01 8.87562096e-01 -1.02429640e+00
-1.32709965e-01 4.30075496e-01 8.87891874e-02 -1.00302875e+00
-4.13026035e-01 -3.50500792e-01 -8.81033063e-01 7.20014513e-01
2.90460140e-01 -5.06073013e-02 -1.06618941e+00 1.15538538e+00
4.64192808e-01 5.43011129e-01 1.46101922e-01 1.09919846e+00
9.42679822e-01 3.26257437e-01 -2.36407623e-01 -1.99711174e-01
1.22399271e+00 -9.31635976e-01 -5.66671312e-01 -7.93535635e-02
7.33255744e-01 -1.11507297e+00 6.23608291e-01 6.40972495e-01
-1.23630464e+00 -3.62400442e-01 -8.92647505e-01 1.40050858e-01
-9.59184486e-03 7.20786080e-02 6.17158771e-01 7.34417319e-01
-1.00042236e+00 6.17819309e-01 -1.14991379e+00 -1.46847978e-01
1.62415430e-01 5.37476063e-01 -5.80615282e-01 -1.52934566e-01
-7.91784346e-01 9.52500582e-01 1.32647052e-01 2.26588979e-01
-6.36694431e-01 -8.83060455e-01 -1.15926349e+00 -2.68046767e-01
1.07146293e-01 -1.03644955e+00 9.56098437e-01 -1.33756113e+00
-1.72936463e+00 1.21295321e+00 9.90132242e-02 -6.11056209e-01
7.01260090e-01 -2.29084641e-02 -2.05237374e-01 6.66648865e-01
-3.03165495e-01 5.20535529e-01 9.56967950e-01 -1.27859867e+00
-4.88135308e-01 -1.13572426e-01 1.44119695e-01 3.20577949e-01
1.94793284e-01 -1.92403510e-01 -3.86439919e-01 -8.24167311e-01
8.28176588e-02 -1.37788951e+00 -4.03654337e-01 4.89455402e-01
-1.28283322e-01 8.52672756e-01 6.53711736e-01 -8.67777050e-01
7.44181514e-01 -2.10508275e+00 5.49459718e-02 2.23433018e-01
-3.36414529e-03 4.61219847e-01 3.81965935e-02 1.24499358e-01
-1.90751612e-01 -3.92186582e-01 -3.50707740e-01 -6.10658042e-02
-5.59672832e-01 2.80405253e-01 -1.83926877e-02 9.26337004e-01
-9.00466368e-02 8.50696385e-01 -1.00888228e+00 -4.43192154e-01
9.60071981e-01 7.22006679e-01 -7.05917239e-01 2.72557616e-01
1.01040684e-01 9.87475038e-01 -9.55731422e-02 5.90527475e-01
5.19857645e-01 1.00547470e-01 1.83611915e-01 -4.93384331e-01
-1.82915896e-01 -1.01355352e-01 -9.44720805e-01 2.03055286e+00
-8.61295402e-01 4.83096451e-01 4.29045409e-01 -8.50860715e-01
4.46645111e-01 5.96446395e-01 7.24628270e-01 -8.24318647e-01
2.39502072e-01 4.80403066e-01 1.45721212e-01 -6.88926935e-01
-5.51652163e-02 -3.65718901e-01 3.49170327e-01 8.56480524e-02
-1.70330238e-02 -3.37416440e-01 -1.44699439e-01 -1.86968923e-01
1.03374004e+00 2.81840235e-01 3.17008495e-01 -2.59223878e-01
6.15896702e-01 2.30097994e-01 6.45535290e-02 4.96443242e-01
-6.93396032e-02 9.93771017e-01 2.28433728e-01 -5.48396528e-01
-8.40906143e-01 -8.53599072e-01 -6.33810312e-02 1.50712609e-01
3.11057597e-01 1.55285135e-01 -4.90599155e-01 -6.99223101e-01
-1.95618883e-01 3.73819321e-01 -8.22153449e-01 -2.88131416e-01
-6.93149507e-01 -6.65192485e-01 1.13253668e-02 1.00983165e-01
-1.35692686e-01 -1.21632433e+00 -1.24233186e+00 4.78538781e-01
-1.09746821e-01 -1.25362802e+00 -2.75807589e-01 3.08441699e-01
-1.13263106e+00 -1.40958321e+00 -9.69411492e-01 -1.00568283e+00
8.11131477e-01 3.52755398e-01 1.04555917e+00 3.50377746e-02
-1.04636431e+00 7.62700856e-01 -2.75574356e-01 -4.41146910e-01
-8.81815493e-01 -4.42393363e-01 -4.28562284e-01 -1.16685860e-01
-2.18538344e-01 -5.62081039e-01 -1.17160666e+00 4.14539844e-01
-1.13708448e+00 4.98930439e-02 8.36466253e-01 1.14672709e+00
5.82905650e-01 -1.43841356e-01 -2.00870886e-01 -1.15020680e+00
1.91162527e-01 -2.97288060e-01 -5.62738538e-01 -1.72875419e-01
-1.66090086e-01 -8.82987529e-02 5.97710133e-01 -4.30865705e-01
-1.04875970e+00 2.06160173e-01 2.62198877e-02 -8.19848180e-01
-2.09040016e-01 3.94532084e-01 7.44742334e-01 -5.36615670e-01
7.71600246e-01 1.52130097e-01 6.30904436e-01 2.94356216e-02
4.01972001e-03 5.51832188e-03 5.72886527e-01 -6.31524548e-02
7.14569032e-01 1.12548292e+00 3.05050850e-01 -9.28663373e-01
-6.07695699e-01 -9.01374400e-01 -5.89911282e-01 -1.63892224e-01
6.31900489e-01 -8.73634219e-01 -4.91884708e-01 3.45270455e-01
-9.73777294e-01 -2.65183121e-01 -4.29388791e-01 9.15407717e-01
-7.48229623e-01 5.10355473e-01 -8.21347356e-01 -4.24260139e-01
-3.70569348e-01 -1.32017028e+00 9.82500970e-01 6.34361580e-02
-1.46539450e-01 -1.65088785e+00 2.93007791e-01 1.29081115e-01
3.95258397e-01 8.60911310e-01 7.86460161e-01 -2.31902316e-01
-7.54598856e-01 -3.11401427e-01 2.17277795e-01 4.78276283e-01
2.90824324e-01 -1.94898590e-01 -9.07547712e-01 -4.63500410e-01
3.36071879e-01 1.24800377e-01 6.37237549e-01 9.22607601e-01
8.12890172e-01 -2.08060384e-01 -2.25726992e-01 1.12298834e+00
1.58304143e+00 2.49311715e-01 7.64614522e-01 5.10252535e-01
5.15045345e-01 8.31259727e-01 6.50188982e-01 1.06916413e-01
-5.00707030e-01 6.84945822e-01 7.66893923e-01 -6.83946013e-01
-4.48644608e-01 7.03234971e-02 2.43213266e-01 5.17927766e-01
-1.97430670e-01 1.53526157e-01 -4.33152825e-01 4.56627011e-01
-1.36434352e+00 -7.83307731e-01 -4.01722547e-03 2.41512513e+00
5.63260257e-01 -1.38163373e-01 -1.99573502e-01 -1.21455953e-01
3.75207424e-01 -2.01866075e-01 -2.22550526e-01 -3.73842388e-01
7.61691928e-02 4.68704373e-01 6.37276351e-01 4.98355865e-01
-1.03144407e+00 2.62698740e-01 5.69753647e+00 4.07270014e-01
-1.53468895e+00 4.31160163e-03 4.42038089e-01 -1.19131878e-01
-2.30307087e-01 -2.47521520e-01 -1.09366514e-01 4.47497129e-01
6.01813078e-01 2.88758427e-01 2.46511489e-01 4.12599653e-01
2.90163368e-01 -4.29068029e-01 -1.00595725e+00 1.05910945e+00
3.77953678e-01 -1.37756252e+00 -1.50205508e-01 1.48557737e-01
8.62282991e-01 -6.63083345e-02 1.27646670e-01 -2.65639648e-02
-2.47618780e-01 -9.81834531e-01 1.92860544e-01 5.41688561e-01
8.19343150e-01 -5.38832724e-01 8.32966566e-01 4.87494320e-02
-6.07128024e-01 7.82571211e-02 7.68068060e-02 5.19463539e-01
1.64324477e-01 4.41082060e-01 -1.25529790e+00 5.18084466e-01
5.15031576e-01 6.67996883e-01 -1.09566689e-01 1.32723999e+00
-1.15712769e-01 2.21373916e-01 -3.50882590e-01 3.77067298e-01
3.29243720e-01 -3.76069099e-01 8.79229784e-01 1.25288570e+00
4.33667004e-01 7.70588070e-02 -2.58324087e-01 6.87662184e-01
3.39130640e-01 5.10962047e-02 -6.22685969e-01 5.24823725e-01
-2.97555655e-01 1.40945876e+00 -8.06541204e-01 7.94525817e-02
-5.81912458e-01 1.16976655e+00 -4.09581184e-01 2.67962813e-01
-6.54211223e-01 9.54273567e-02 2.12222800e-01 4.94708866e-01
2.52229482e-01 1.76781803e-01 1.29205674e-01 -9.99630928e-01
-2.80413218e-03 -7.55902469e-01 4.14045870e-01 -8.41470301e-01
-8.43467653e-01 4.78000045e-01 -2.02406347e-01 -1.88554561e+00
-6.16074324e-01 -8.79258096e-01 -7.90696859e-01 7.58568764e-01
-1.97340393e+00 -1.03507257e+00 -5.15625954e-01 6.58892989e-01
3.90417188e-01 1.53930530e-01 1.04043508e+00 1.52094081e-01
1.96115181e-01 1.71389267e-01 2.17878535e-01 -9.82380211e-02
8.53797138e-01 -1.30747592e+00 -2.76895791e-01 6.68663800e-01
4.33463082e-02 4.41355199e-01 7.09708452e-01 -2.74323493e-01
-1.36236215e+00 -8.36454988e-01 3.72008123e-02 -2.69899040e-01
3.57449263e-01 1.96156725e-01 -5.71534753e-01 6.35841072e-01
2.17212141e-01 3.26448768e-01 6.37168050e-01 -6.08493149e-01
1.68319181e-01 1.68449014e-01 -1.43312526e+00 4.36730117e-01
4.67377365e-01 -1.13817900e-01 -4.79300618e-01 3.35823655e-01
2.22221762e-01 -9.76279974e-01 -8.53763759e-01 5.55565834e-01
7.07582474e-01 -1.49056089e+00 1.22193730e+00 -1.82333551e-02
5.83590031e-01 -1.12730950e-01 4.29499269e-01 -1.39976847e+00
1.74065083e-01 -9.66225147e-01 -5.00859246e-02 2.18378559e-01
1.04869500e-01 -7.07364559e-01 9.68966544e-01 2.27431059e-01
-5.04598737e-01 -6.09473348e-01 -8.13371420e-01 -2.71821886e-01
-2.27464572e-01 4.38815393e-02 -4.47526604e-01 9.34399009e-01
-2.59281784e-01 -3.36746693e-01 -4.86104712e-02 3.41780156e-01
6.68876529e-01 1.90001830e-01 6.89810753e-01 -8.83288622e-01
-5.50588191e-01 -2.91394681e-01 -5.86653173e-01 -6.79415822e-01
-1.79446444e-01 -7.38988936e-01 -1.00891091e-01 -1.31586456e+00
-8.76542777e-02 -3.37631255e-01 -1.47289932e-01 -2.68257281e-04
1.35319084e-01 4.68835652e-01 -1.20609745e-01 4.06224012e-01
4.92490493e-02 -1.21368133e-02 1.63700771e+00 1.60086110e-01
-3.31493497e-01 3.90076816e-01 -1.75037995e-01 9.04718041e-01
4.67865288e-01 -3.26277882e-01 -2.41233781e-01 -2.08185595e-02
-1.17936775e-01 4.32857662e-01 8.85724127e-01 -1.07100070e+00
2.35730603e-01 6.25281513e-01 4.35524434e-01 -3.08306932e-01
4.28827584e-01 -1.26837444e+00 3.64523679e-01 9.11750078e-01
2.47147065e-02 -1.52507633e-01 5.00721037e-01 6.05487943e-01
-5.45183659e-01 -3.16466630e-01 9.74916339e-01 -6.59149766e-01
-6.58905625e-01 7.79587701e-02 -2.51261830e-01 -1.30067483e-01
1.11456430e+00 -7.21924663e-01 2.55644649e-01 -4.89811599e-01
-9.61027741e-01 -3.56458277e-01 5.61161697e-01 7.34696090e-02
7.73137033e-01 -6.76742136e-01 -6.27230942e-01 7.01380789e-01
-4.66093868e-02 3.16682547e-01 6.03361249e-01 1.56120455e+00
-1.33575952e+00 3.15588593e-01 -1.80562362e-01 -9.08174574e-01
-1.27710438e+00 5.42618036e-01 9.07639146e-01 -3.71864498e-01
-8.91720712e-01 5.76702833e-01 5.89011788e-01 -2.92381734e-01
1.39884979e-01 -3.89310569e-01 -2.51597222e-02 -3.52651238e-01
2.14930862e-01 -5.46673909e-02 3.68302166e-01 -3.75635117e-01
1.54567540e-01 8.06562185e-01 -8.22879151e-02 1.79255351e-01
1.11702216e+00 -1.96683213e-01 1.10448830e-01 1.11935996e-01
1.25396240e+00 1.76459223e-01 -1.36202466e+00 -9.43992566e-03
-4.44794416e-01 -4.51478988e-01 3.11808616e-01 -9.31300640e-01
-1.18722796e+00 7.88914979e-01 8.38250160e-01 7.04020411e-02
1.26909435e+00 -3.29014659e-01 7.77563393e-01 -3.43856186e-01
3.32241058e-01 -4.54629779e-01 6.33087307e-02 -9.54059511e-02
7.43689895e-01 -1.32253921e+00 3.67836037e-04 -7.81307161e-01
-4.61296976e-01 1.33750021e+00 8.60120133e-02 -3.48311990e-01
6.26223922e-01 5.30704081e-01 4.67620671e-01 -1.67844415e-01
-8.62130821e-02 1.27018884e-01 3.73411626e-01 6.08760178e-01
3.65733266e-01 -2.69478768e-01 2.24867817e-02 -2.81868428e-01
1.15874551e-01 -4.58204076e-02 6.68621421e-01 1.09580696e+00
1.00607887e-01 -8.97029519e-01 -3.28190893e-01 7.68048093e-02
-9.40672219e-01 -2.45717600e-01 2.63973296e-01 1.18409073e+00
-2.08548337e-01 4.39716667e-01 -2.19938233e-01 3.33547473e-01
4.08456236e-01 -4.42385852e-01 8.19421113e-01 -6.49560034e-01
-9.55329478e-01 4.51715261e-01 -2.24073172e-01 -6.24776006e-01
-9.22606289e-01 -6.47066653e-01 -9.82557654e-01 3.15746963e-01
-2.67252177e-01 -3.42236534e-02 8.62174451e-01 4.36327487e-01
3.46737392e-02 6.70486808e-01 6.34244144e-01 -1.18109155e+00
-2.37845883e-01 -5.31430900e-01 -5.35766721e-01 8.66729259e-01
6.58902466e-01 -3.77152771e-01 -9.79808867e-01 4.08949137e-01] | [13.871456146240234, -3.110417127609253] |
2d69106e-f73b-4428-833c-bebaf04e6edb | feedback-gradient-descent-efficient-and | 2205.08385 | null | https://arxiv.org/abs/2205.08385v1 | https://arxiv.org/pdf/2205.08385v1.pdf | Feedback Gradient Descent: Efficient and Stable Optimization with Orthogonality for DNNs | The optimization with orthogonality has been shown useful in training deep neural networks (DNNs). To impose orthogonality on DNNs, both computational efficiency and stability are important. However, existing methods utilizing Riemannian optimization or hard constraints can only ensure stability while those using soft constraints can only improve efficiency. In this paper, we propose a novel method, named Feedback Gradient Descent (FGD), to our knowledge, the first work showing high efficiency and stability simultaneously. FGD induces orthogonality based on the simple yet indispensable Euler discretization of a continuous-time dynamical system on the tangent bundle of the Stiefel manifold. In particular, inspired by a numerical integration method on manifolds called Feedback Integrators, we propose to instantiate it on the tangent bundle of the Stiefel manifold for the first time. In the extensive image classification experiments, FGD comprehensively outperforms the existing state-of-the-art methods in terms of accuracy, efficiency, and stability. | ['Dong Eui Chang', 'Fanchen Bu'] | 2022-05-12 | null | null | null | null | ['numerical-integration'] | ['miscellaneous'] | [-4.39035624e-01 -7.10916072e-02 1.48222610e-01 -1.92460790e-01
1.81126699e-01 -3.80107105e-01 5.22441447e-01 -2.55481064e-01
-6.73054576e-01 7.60746598e-01 -2.24675417e-01 -4.46699589e-01
-2.19407976e-01 -4.48049068e-01 -5.99648535e-01 -8.33509922e-01
-1.49951931e-02 -1.08728848e-01 1.24419607e-01 -3.83592308e-01
2.51594216e-01 6.71096146e-01 -1.20966804e+00 -6.49270773e-01
1.14256978e+00 9.08333778e-01 6.79204762e-02 2.82312095e-01
8.56691599e-02 4.20585960e-01 3.90887298e-02 -3.24245572e-01
4.05248731e-01 -3.58487010e-01 -6.35862470e-01 -4.65085395e-02
4.78217453e-01 -4.55087990e-01 -5.48941553e-01 1.22952294e+00
4.61268157e-01 2.80469000e-01 7.49008179e-01 -1.20331120e+00
-7.11372375e-01 2.30876490e-01 -1.89307347e-01 2.18552798e-01
-2.34848693e-01 -1.43627271e-01 9.81073916e-01 -1.18211019e+00
8.27015817e-01 7.16122806e-01 7.08255708e-01 7.73430884e-01
-1.33875394e+00 -5.00878692e-01 2.35018134e-01 2.59537518e-01
-1.27674711e+00 -2.97223419e-01 8.82055223e-01 -5.31593084e-01
6.06098354e-01 2.15856060e-01 7.85683393e-01 7.15615630e-01
8.69170129e-02 6.80482388e-01 8.99426162e-01 -1.80497125e-01
2.22660631e-01 2.13945016e-01 4.43195015e-01 1.11336732e+00
3.06195438e-01 -3.03143635e-02 -1.31205410e-01 1.23707011e-01
8.51247251e-01 -5.18471934e-02 -5.07919967e-01 -6.24484897e-01
-1.04390824e+00 9.77837682e-01 5.50516129e-01 3.62406075e-01
-1.85173705e-01 -9.90934391e-03 2.39360854e-01 1.89805999e-01
4.61310804e-01 3.52176011e-01 -1.07075959e-01 -9.62710306e-02
-7.23656237e-01 6.90771863e-02 8.13463032e-01 5.81106484e-01
7.03328729e-01 3.07984352e-01 5.85366376e-02 5.67651868e-01
4.48331863e-01 1.48829564e-01 5.04695058e-01 -8.06645572e-01
3.02651137e-01 6.55244231e-01 1.60818417e-02 -1.16491151e+00
-6.57020628e-01 -6.74201548e-01 -1.10514641e+00 3.38672310e-01
3.20958734e-01 -3.79499614e-01 -3.39483112e-01 1.97041380e+00
5.39942741e-01 9.87655297e-02 2.00564042e-01 1.15074503e+00
5.94611645e-01 3.70683402e-01 -4.73677278e-01 -1.91380218e-01
7.99887955e-01 -8.19948256e-01 -7.79129267e-01 5.08285701e-01
9.13464248e-01 -5.58410347e-01 8.91568363e-01 2.93256968e-01
-8.46638858e-01 -2.55470067e-01 -1.42582905e+00 4.23626509e-03
-2.25068435e-01 6.93672657e-01 6.33414090e-01 6.72887266e-01
-1.09065771e+00 1.24637938e+00 -9.08192813e-01 -2.11691558e-01
2.03108341e-01 3.21083039e-01 -5.35179555e-01 4.58226383e-01
-1.17473817e+00 1.01163638e+00 5.63172325e-02 3.08596343e-01
-5.83567619e-01 -6.41628921e-01 -6.45862520e-01 -2.41132751e-01
1.23722158e-01 -5.90401649e-01 8.05499077e-01 -5.90099752e-01
-2.02322817e+00 5.02081335e-01 2.22518489e-01 -5.26795387e-01
9.31031048e-01 -4.23164994e-01 -1.91752939e-03 1.87203184e-01
-2.53480434e-01 5.29722929e-01 8.06400597e-01 -8.76503408e-01
-1.44111320e-01 -3.10876250e-01 2.28258789e-01 2.38936990e-01
-1.08346653e+00 -4.41187054e-01 -1.62976626e-02 -5.44313729e-01
4.46820945e-01 -9.50781465e-01 -1.65947676e-01 2.56907165e-01
-3.94730717e-01 -4.41505671e-01 8.63475919e-01 -7.13393509e-01
1.26694643e+00 -2.05159998e+00 4.85287726e-01 -8.84215832e-02
1.95202962e-01 4.18338984e-01 3.08605641e-01 1.37567207e-01
-3.63754258e-02 -1.86505485e-02 -3.07892114e-01 -5.62175333e-01
1.38854995e-01 6.33193851e-02 -2.61829883e-01 9.10373032e-01
3.28718156e-01 5.50198972e-01 -7.96910226e-01 -2.89379507e-01
3.69284824e-02 6.61753833e-01 -7.86305845e-01 -1.18081495e-02
1.54211655e-01 5.67688525e-01 -2.66611516e-01 -7.76300281e-02
6.79293931e-01 1.33416234e-02 -2.10594505e-01 -3.84658486e-01
-5.64935744e-01 2.87889004e-01 -1.30271995e+00 1.50826180e+00
-2.94642866e-01 6.37286782e-01 1.92520887e-01 -1.28878391e+00
8.96844387e-01 2.47350335e-01 6.03336871e-01 -1.94238007e-01
2.61358410e-01 4.04207140e-01 8.90953392e-02 -3.89794409e-01
3.43009442e-01 4.35626134e-02 5.64848840e-01 3.53539586e-01
1.28066912e-01 4.31549624e-02 2.88465828e-01 2.21698135e-01
6.33303940e-01 1.87067762e-01 -3.29194032e-02 -6.92874968e-01
8.02169144e-01 -3.40235054e-01 5.21886051e-01 2.08401516e-01
-1.76335737e-01 5.01890898e-01 5.43619275e-01 -2.78614581e-01
-9.80360210e-01 -7.67426193e-01 -7.11622238e-01 4.90016073e-01
2.37392094e-02 -1.60837039e-01 -1.10685062e+00 -5.84157944e-01
-1.47622332e-01 3.74800861e-01 -4.64685887e-01 -2.82518238e-01
-4.05513644e-01 -6.58770084e-01 4.04494047e-01 2.17754528e-01
9.35791790e-01 -5.29102743e-01 -5.16225755e-01 7.45067224e-02
1.73786953e-01 -1.24697793e+00 -6.48066580e-01 -1.15564831e-01
-1.14791024e+00 -8.58939707e-01 -9.39702749e-01 -6.23456597e-01
8.87849212e-01 1.25151098e-01 3.86629075e-01 -1.61746785e-01
-1.33161992e-01 2.89452046e-01 -5.20601943e-02 9.82027352e-02
-2.13738084e-02 1.68637797e-01 5.18600523e-01 4.27855104e-01
-2.50436723e-01 -9.03602183e-01 -5.61258197e-01 3.99433255e-01
-7.82833278e-01 2.74535984e-01 2.59099424e-01 7.82255530e-01
3.01850736e-01 -1.79679200e-01 4.63064134e-01 -4.13114607e-01
4.66190368e-01 -2.62232542e-01 -8.38572145e-01 -5.74899279e-02
-6.75369322e-01 5.04956722e-01 8.56973171e-01 -4.39485282e-01
-7.96707332e-01 3.75454463e-02 -2.11482957e-01 -6.27884984e-01
3.65426540e-01 4.04574960e-01 4.47103903e-02 -6.25466406e-01
6.08618796e-01 3.08004409e-01 1.02537371e-01 -6.28009260e-01
3.95993352e-01 3.46663922e-01 3.19634944e-01 -3.36622059e-01
6.47547781e-01 8.09781432e-01 4.39603209e-01 -1.01568711e+00
-7.18073905e-01 -2.46668741e-01 -8.14897060e-01 -3.22928488e-01
8.24363291e-01 -4.34487104e-01 -7.98158288e-01 6.46316469e-01
-1.15730429e+00 -1.74303621e-01 -1.07579893e-02 9.01296139e-01
-3.93863022e-01 6.36110663e-01 -6.84346795e-01 -8.75572920e-01
-4.55830753e-01 -1.08922422e+00 4.02662963e-01 1.76474959e-01
1.01160325e-01 -1.06476927e+00 8.86152089e-02 -1.23934880e-01
4.28478211e-01 2.20232010e-01 4.64633822e-01 -2.80267328e-01
-3.89390618e-01 -1.29107699e-01 -3.07608366e-01 6.39762223e-01
7.55364075e-02 2.54102796e-01 -7.04171717e-01 -3.54182094e-01
3.67036641e-01 -2.41498612e-02 1.00170863e+00 1.59986347e-01
8.25003028e-01 -3.58802438e-01 -1.21605709e-01 8.47520411e-01
1.33429134e+00 -1.36030808e-01 2.96753705e-01 2.38595396e-01
9.16904330e-01 5.60176849e-01 3.18986237e-01 3.95355463e-01
4.33626473e-01 5.47047198e-01 5.96120119e-01 3.33860740e-02
8.93417820e-02 6.95658401e-02 6.56063437e-01 1.34024394e+00
-3.18367332e-01 5.36074996e-01 -5.41482568e-01 4.19530809e-01
-1.85567069e+00 -5.44921696e-01 -2.91255653e-01 2.38359213e+00
7.87541628e-01 1.50108591e-01 1.09462775e-01 2.27189466e-01
6.54896498e-01 -2.78824735e-02 -6.63219988e-01 -3.56668293e-01
-1.10311784e-01 -2.21354768e-01 5.36020339e-01 6.82051361e-01
-1.19777918e+00 6.67476535e-01 5.22331142e+00 6.95246696e-01
-1.49147928e+00 2.00794294e-01 2.61300474e-01 1.00611605e-01
-1.24652311e-01 1.07930275e-02 -1.15419984e+00 3.15804571e-01
4.66846585e-01 -1.20384030e-01 4.78022397e-01 8.60927165e-01
2.32557774e-01 2.37671271e-01 -9.62363303e-01 9.03991699e-01
1.83979589e-02 -1.32945502e+00 -1.72583476e-01 1.91913828e-01
9.37480390e-01 6.66443929e-02 7.88822398e-02 3.64518315e-02
-1.44150749e-01 -7.02912092e-01 6.98234618e-01 6.26025856e-01
3.18017751e-01 -6.20156288e-01 6.22472167e-01 3.66470963e-01
-1.14314270e+00 1.95012122e-01 -4.07694012e-01 -1.75978586e-01
1.05085529e-01 7.62441874e-01 -3.23107183e-01 6.49986207e-01
3.55584770e-01 1.10066593e+00 -3.05183142e-01 9.71258879e-01
-2.32238725e-01 3.48171115e-01 -5.84099174e-01 -4.11300540e-01
4.98966813e-01 -1.07288194e+00 8.01913381e-01 9.77200449e-01
3.91965091e-01 4.14600261e-02 -2.78684646e-01 1.02316427e+00
-2.72297487e-03 4.67738003e-01 -7.76731670e-01 -4.30398546e-02
1.52391437e-02 1.54950583e+00 -6.82256877e-01 -2.61802763e-01
-2.79163629e-01 8.06174219e-01 6.20992422e-01 3.17292988e-01
-8.27156961e-01 -4.30217385e-01 7.64510334e-01 -2.24574283e-01
4.81012404e-01 -6.74377501e-01 -3.70620936e-01 -1.44869292e+00
4.38119322e-01 -3.18036705e-01 9.21627060e-02 -2.96893120e-01
-9.81745899e-01 6.74932599e-01 -2.74674863e-01 -1.30788803e+00
1.26475409e-01 -1.00583112e+00 -6.03895187e-01 6.78355753e-01
-1.29428375e+00 -5.49259305e-01 -1.45504594e-01 6.38283491e-01
-4.03653905e-02 -9.42245349e-02 6.34806693e-01 5.07927656e-01
-9.37431395e-01 5.21368146e-01 4.15809870e-01 1.32025421e-01
1.55156001e-01 -1.26743567e+00 7.17486739e-02 8.13459396e-01
3.41919884e-02 7.97165692e-01 6.11219645e-01 -1.70064941e-01
-1.40219331e+00 -8.38125229e-01 5.32049179e-01 1.22536689e-01
8.84716570e-01 -3.61350864e-01 -9.94602799e-01 3.26155782e-01
6.93143010e-02 -2.79347692e-02 2.73755968e-01 -2.47481704e-01
-9.65887830e-02 -3.20810169e-01 -8.69716823e-01 1.04163837e+00
1.00596905e+00 -3.87798637e-01 -2.40483001e-01 3.61807197e-01
5.65017402e-01 -3.73383075e-01 -9.56337035e-01 4.85893041e-01
4.88076717e-01 -9.57313240e-01 6.91044748e-01 -5.99897504e-01
2.03860417e-01 -4.30628747e-01 -6.09824769e-02 -1.27779496e+00
5.09531908e-02 -9.52118158e-01 -2.04126745e-01 1.20583272e+00
2.20422551e-01 -1.14328039e+00 5.72998226e-01 4.20371264e-01
-3.44230443e-01 -1.29383397e+00 -1.14186502e+00 -9.52321291e-01
2.85952419e-01 -3.41396481e-01 1.05890945e-01 9.60819542e-01
1.54299960e-01 3.31282318e-01 -2.88126111e-01 1.52969360e-01
6.54873013e-01 -1.00022815e-01 4.61462736e-01 -1.12620926e+00
-6.06699921e-02 -7.89908350e-01 -5.40485084e-01 -1.15484166e+00
1.54235110e-01 -8.90889287e-01 -2.56115347e-01 -1.24647629e+00
-1.58848599e-01 -4.38383430e-01 -1.37402430e-01 1.29895911e-01
9.90047082e-02 3.20591964e-02 2.13504851e-01 2.14190885e-01
-2.15603366e-01 1.07360661e+00 1.43651152e+00 2.38456249e-01
-2.73532987e-01 3.95148210e-02 -2.50448316e-01 8.21234882e-01
1.00054085e+00 -1.27316102e-01 -3.16037595e-01 -3.38550538e-01
2.14551792e-01 -3.22505295e-01 2.84003556e-01 -1.14790905e+00
3.20066780e-01 2.14094341e-01 -1.98942363e-01 -4.07894999e-01
4.53011066e-01 -5.10176063e-01 -1.85466394e-01 7.01114535e-01
-6.27720356e-02 2.08355375e-02 -2.05240166e-03 3.47321093e-01
-1.14201240e-01 -5.91867983e-01 8.89015436e-01 3.18988770e-01
-3.12734663e-01 6.07610226e-01 -2.98748881e-01 -8.77950899e-03
7.68542051e-01 1.10286213e-01 -4.34521660e-02 -2.64354676e-01
-6.20693088e-01 1.25628570e-02 2.80810356e-01 8.73260126e-02
5.67353427e-01 -1.59192991e+00 -4.11666632e-01 2.67592013e-01
-4.89064783e-01 -7.07954392e-02 1.45359139e-04 1.44038212e+00
-6.40254319e-01 4.41257447e-01 -2.06963629e-01 -6.71269536e-01
-1.14159691e+00 2.65715361e-01 6.32322788e-01 -2.41818666e-01
-5.09654820e-01 7.70422637e-01 2.72026993e-02 -5.86883783e-01
2.78849810e-01 -4.14631784e-01 -4.28433299e-01 1.25935137e-01
4.67963368e-02 5.91494560e-01 1.37932912e-01 -6.08821094e-01
-2.75566906e-01 8.56723189e-01 1.49779528e-01 -3.66508216e-01
1.29823112e+00 7.54896086e-03 -3.24018449e-01 4.58671778e-01
1.54042661e+00 -2.15239108e-01 -1.58450031e+00 -1.41304702e-01
-1.06759571e-01 -8.55523720e-02 2.39592761e-01 1.74703732e-01
-1.20024657e+00 1.21596849e+00 5.77959418e-01 4.86136019e-01
7.38594294e-01 -5.41845441e-01 7.21884191e-01 6.42188489e-01
1.76745087e-01 -1.07148254e+00 1.76752284e-01 7.46555686e-01
1.07526517e+00 -1.03609180e+00 -5.41283116e-02 -3.25861394e-01
-3.97387117e-01 1.44168472e+00 5.98147452e-01 -5.48269510e-01
1.03398764e+00 -1.84316441e-01 -1.51616395e-01 1.58868879e-01
-3.66965801e-01 -4.53641340e-02 4.02807653e-01 1.21775389e-01
4.79293376e-01 -1.14093892e-01 -7.23938704e-01 3.73209059e-01
-4.03677195e-01 -1.23785667e-01 2.77184218e-01 5.87196171e-01
-2.72400111e-01 -9.46493566e-01 -1.18019871e-01 9.03022513e-02
-3.72432351e-01 2.63994746e-03 -1.89744264e-01 7.78571844e-01
-8.98854584e-02 5.66157997e-01 -9.75952521e-02 -2.66965985e-01
2.04328358e-01 -1.38036117e-01 5.80557883e-01 -1.39668122e-01
-1.96532175e-01 -2.80217171e-01 -1.70781776e-01 -4.25859660e-01
-3.71802509e-01 -7.47505248e-01 -1.51617980e+00 -1.53053299e-01
-5.35135150e-01 1.99283123e-01 9.56174135e-01 1.12383366e+00
3.26483518e-01 3.23209375e-01 9.60898817e-01 -9.61944342e-01
-9.85901892e-01 -1.06105030e+00 -4.99584317e-01 3.44042301e-01
3.68377060e-01 -9.91716683e-01 -6.62589133e-01 -2.42842719e-01] | [7.528221607208252, 3.9472169876098633] |
aa2351be-3fad-4e79-98b3-aee9b10c098c | propandl-a-modular-architecture-for | 2304.08645 | null | https://arxiv.org/abs/2304.08645v1 | https://arxiv.org/pdf/2304.08645v1.pdf | ProPanDL: A Modular Architecture for Uncertainty-Aware Panoptic Segmentation | We introduce ProPanDL, a family of networks capable of uncertainty-aware panoptic segmentation. Unlike existing segmentation methods, ProPanDL is capable of estimating full probability distributions for both the semantic and spatial aspects of panoptic segmentation. We implement and evaluate ProPanDL variants capable of estimating both parametric (Variance Network) and parameter-free (SampleNet) distributions quantifying pixel-wise spatial uncertainty. We couple these approaches with two methods (Temperature Scaling and Evidential Deep Learning) for semantic uncertainty estimation. To evaluate the uncertainty-aware panoptic segmentation task, we address limitations with existing approaches by proposing new metrics that enable separate evaluation of spatial and semantic uncertainty. We additionally propose the use of the energy score, a proper scoring rule, for more robust evaluation of spatial output distributions. Using these metrics, we conduct an extensive evaluation of ProPanDL variants. Our results demonstrate that ProPanDL is capable of estimating well-calibrated and meaningful output distributions while still retaining strong performance on the base panoptic segmentation task. | ['Steven Waslander', 'Chang Won Lee', 'Jacob Deery'] | 2023-04-17 | null | null | null | null | ['panoptic-segmentation'] | ['computer-vision'] | [-2.49093905e-01 -8.52243602e-02 -2.11722583e-01 -6.59587085e-01
-1.08220553e+00 -1.04767716e+00 9.11088943e-01 2.78285772e-01
-2.97712415e-01 9.37879920e-01 2.38380283e-01 -5.37401855e-01
-4.95251298e-01 -9.47073698e-01 -5.21746576e-01 -5.92667878e-01
-5.06155729e-01 6.78567648e-01 3.38674486e-01 3.01485419e-01
8.74575004e-02 6.46071553e-01 -1.21962905e+00 -2.07430154e-01
1.21032405e+00 1.42623520e+00 -2.06500798e-01 6.42104864e-01
-1.10599436e-01 5.55416286e-01 -6.48242295e-01 4.41707782e-02
3.69772822e-01 -8.07982981e-02 -6.29231274e-01 -4.88192707e-01
6.91320837e-01 -4.62699264e-01 -1.45557430e-02 1.06320810e+00
3.09017658e-01 4.46216941e-01 1.04310811e+00 -1.31770539e+00
-2.26837665e-01 1.17187274e+00 -4.57076758e-01 4.81524825e-01
-3.56842786e-01 4.04053152e-01 1.37437356e+00 -5.29883087e-01
1.44141719e-01 1.48490191e+00 8.94439518e-01 -2.11199731e-01
-1.47687018e+00 -7.62773514e-01 1.45205021e-01 -3.99981052e-01
-1.38457739e+00 -2.54205257e-01 3.72564316e-01 -4.69282627e-01
1.00404131e+00 2.45108813e-01 4.32794303e-01 6.00897193e-01
2.71584392e-01 6.50210977e-01 1.31962895e+00 1.75336316e-01
7.40549922e-01 -1.33908838e-01 3.29423994e-02 1.28849462e-01
4.07244146e-01 5.76925993e-01 -1.32043347e-01 -3.27825457e-01
8.45153272e-01 -4.76249874e-01 -8.87176618e-02 -1.62349567e-01
-9.94153082e-01 7.74771094e-01 8.16037953e-01 8.50013196e-02
-2.77922899e-01 8.64522099e-01 1.39495611e-01 -1.26002967e-01
1.07292914e+00 9.08473432e-01 -6.40838027e-01 -8.04770514e-02
-1.75095820e+00 4.61593091e-01 8.37342560e-01 6.17052436e-01
9.50367272e-01 2.39829794e-01 -3.79931003e-01 6.03561640e-01
5.54194212e-01 9.58284020e-01 -2.76407421e-01 -1.37819552e+00
3.46805960e-01 2.35990331e-01 4.00748879e-01 -7.34724045e-01
-8.50207686e-01 -5.22598803e-01 -6.89453602e-01 3.32280606e-01
4.59427595e-01 -6.04822218e-01 -1.38174605e+00 1.76886284e+00
1.66362062e-01 2.81861544e-01 -1.25874907e-01 6.93211019e-01
4.62986737e-01 1.01654041e+00 5.75938165e-01 2.83182651e-01
1.13410461e+00 -6.17901385e-01 -3.48130554e-01 -2.55125314e-01
-2.59653851e-02 -3.14721137e-01 8.11134756e-01 -1.82766486e-02
-7.99827695e-01 -2.74483055e-01 -1.02724254e+00 2.90887058e-01
-6.50952995e-01 -1.63937539e-01 6.75430059e-01 6.34971142e-01
-1.28262138e+00 9.89967704e-01 -1.03854489e+00 -6.31998703e-02
3.86564761e-01 1.12866677e-01 2.58548647e-01 5.19734502e-01
-1.46121573e+00 8.89232159e-01 7.02343106e-01 6.89160079e-02
-1.14856923e+00 -1.16520417e+00 -1.10222459e+00 4.15039241e-01
1.87555701e-01 -4.88377601e-01 1.30675423e+00 -4.25087839e-01
-1.27565372e+00 2.50329524e-01 5.41590512e-01 -7.82354355e-01
4.98153299e-01 -2.40691036e-01 -3.87073696e-01 4.09221113e-01
2.54021108e-01 1.24134243e+00 8.15967858e-01 -1.31065035e+00
-6.04431689e-01 -3.13200653e-02 -3.63189057e-02 1.16472423e-01
2.46083409e-01 -4.00219768e-01 -1.67449370e-01 -6.54837668e-01
1.55751720e-01 -5.96298873e-01 -4.61004615e-01 1.67400748e-01
-6.35520756e-01 -7.08441809e-02 4.49378103e-01 -5.88936329e-01
1.09134829e+00 -1.92713082e+00 -2.70288885e-01 6.79724753e-01
2.49703273e-01 1.16341390e-01 -1.31425321e-01 1.33330747e-01
3.05673242e-01 6.74122393e-01 -1.11791432e+00 -2.79940665e-01
3.89966726e-01 2.55377322e-01 -4.61369395e-01 3.77483964e-01
5.82802474e-01 9.21315610e-01 -9.07977343e-01 -4.48517978e-01
6.61037147e-01 3.83398324e-01 -3.32473665e-01 1.70914620e-01
-9.38473701e-01 3.35970610e-01 -4.89275694e-01 6.73552930e-01
1.19000244e+00 -1.19559176e-01 -1.14159808e-01 -8.71858820e-02
-2.34101623e-01 3.61250997e-01 -1.17574918e+00 1.46170151e+00
-6.76452816e-01 6.24126196e-01 2.45506644e-01 -2.85782337e-01
8.14930856e-01 -1.87848241e-03 2.59723216e-01 -3.82040381e-01
7.77583569e-02 3.31343025e-01 -2.55793631e-01 2.93431021e-02
8.32596719e-01 -1.63770348e-01 -3.60692054e-01 6.31943703e-01
1.89550877e-01 -8.89127910e-01 -5.15690707e-02 1.71235591e-01
6.09666169e-01 3.86043012e-01 1.47822499e-01 -9.15437162e-01
-7.45053636e-04 4.87199239e-02 2.21984446e-01 1.09132934e+00
-4.76568818e-01 7.21249282e-01 6.67792559e-01 -1.71435162e-01
-9.29390728e-01 -1.71543300e+00 -7.22999752e-01 9.32677090e-01
1.77428842e-01 -2.34139301e-02 -5.59885204e-01 -5.29772639e-01
5.49356699e-01 1.22873533e+00 -6.95714056e-01 2.59824216e-01
-8.40270519e-02 -1.27489841e+00 9.14968193e-01 6.82115197e-01
8.05630863e-01 -7.69602835e-01 -6.36110783e-01 2.20909700e-01
5.90166226e-02 -8.88664186e-01 -1.39350697e-01 4.82936651e-01
-8.16007614e-01 -7.86022305e-01 -5.82154393e-01 1.01154678e-01
2.89269667e-02 -3.72074932e-01 1.38847220e+00 -5.73009372e-01
-4.62179594e-02 3.79303813e-01 1.55410782e-01 -1.98593482e-01
-2.89040387e-01 2.77734101e-01 -1.85691386e-01 -5.69972217e-01
1.80850342e-01 -8.55933130e-01 -7.97607899e-01 1.53590277e-01
-1.04357088e+00 -3.59553605e-01 3.55862170e-01 2.25817680e-01
5.40679276e-01 1.58387586e-01 4.63234574e-01 -5.06541014e-01
5.39752066e-01 -9.73738134e-01 -1.05926073e+00 1.87947780e-01
-8.44571292e-01 3.93251866e-01 1.77152693e-01 1.12032846e-01
-1.27380490e+00 -3.08606416e-01 -3.49043518e-01 -1.48406222e-01
-1.54390529e-01 5.99626422e-01 2.37977460e-01 2.21054867e-01
8.53250146e-01 -2.38499835e-01 -3.35248441e-01 -3.48106146e-01
9.26502049e-01 3.92215669e-01 7.47420728e-01 -8.86468887e-01
6.75797641e-01 4.99810368e-01 -2.40965635e-02 -5.63018084e-01
-9.60394621e-01 -4.22257066e-01 -4.61658031e-01 -9.78627130e-02
7.82817841e-01 -1.13122427e+00 -1.74004614e-01 4.40738589e-01
-8.56067240e-01 -6.36027098e-01 -6.36070013e-01 4.17222708e-01
-3.89078826e-01 1.76610216e-01 -5.26776373e-01 -1.01425779e+00
-4.61890996e-01 -9.93274271e-01 1.28693736e+00 2.33502284e-01
-1.24750026e-01 -1.44912827e+00 5.03890693e-01 -3.26569736e-01
8.19233119e-01 5.36559403e-01 8.16142619e-01 -6.03920579e-01
-6.76108897e-01 1.84718922e-01 -8.00635278e-01 4.36624199e-01
-8.79711583e-02 3.32500368e-01 -1.40729547e+00 2.18754917e-01
-2.77026236e-01 -2.48122752e-01 1.59293187e+00 1.02971268e+00
1.07151365e+00 -2.23816738e-01 -2.19019994e-01 8.29948843e-01
1.77387762e+00 -1.75163910e-01 5.84476411e-01 1.83283269e-01
4.30350482e-01 5.84387124e-01 1.94217414e-01 4.45408553e-01
3.39458972e-01 1.42330658e-02 7.82640815e-01 6.59997342e-03
-1.27284490e-02 -3.38607013e-01 5.66720739e-02 2.65993532e-02
2.67894804e-01 -6.64011598e-01 -1.21231496e+00 7.72960067e-01
-1.80459249e+00 -6.02995813e-01 -6.67541614e-03 1.88727319e+00
8.91651750e-01 4.90074269e-02 -7.47129619e-02 -4.92721856e-01
5.81583738e-01 8.11275363e-01 -8.46406758e-01 -4.01278496e-01
-2.13487923e-01 3.65680188e-01 1.02674246e+00 8.98643732e-01
-1.50787294e+00 1.00287819e+00 7.84165144e+00 1.03482938e+00
-8.42660069e-01 -2.97186263e-02 1.09059918e+00 -1.47621741e-03
-9.27597582e-01 1.22894190e-01 -7.15771258e-01 3.45755935e-01
1.21979988e+00 9.43545997e-02 2.45298684e-01 7.78142571e-01
9.67610255e-02 -5.63316703e-01 -8.63605082e-01 4.53332394e-01
-6.55239463e-01 -1.54211640e+00 4.97515835e-02 -2.35379517e-01
1.06625473e+00 7.02471793e-01 1.54211774e-01 2.12241635e-01
1.09959722e+00 -1.27403820e+00 6.57740235e-01 4.89534348e-01
9.09925103e-01 -6.63586974e-01 6.97916090e-01 -1.44765541e-01
-1.16186345e+00 1.36103243e-01 -3.47183973e-01 1.29906416e-01
2.66590059e-01 1.22569931e+00 -4.90610898e-01 3.64113569e-01
8.22249413e-01 7.53697038e-01 -3.77141654e-01 1.08626533e+00
-5.06209373e-01 8.46962512e-01 -1.10435522e+00 2.63777703e-01
8.36029649e-01 -1.58538595e-01 6.72313571e-01 1.43935978e+00
4.11159247e-01 -2.84868836e-01 5.20889880e-03 1.68427420e+00
-1.95408821e-01 -2.57380903e-01 -4.28714812e-01 -6.16866052e-02
7.57606685e-01 1.24860466e+00 -8.36907625e-01 -2.92069465e-01
3.42901975e-01 3.57167423e-01 -1.03194294e-02 4.95164841e-01
-8.64512622e-01 -2.27559596e-01 8.53785634e-01 -2.63734341e-01
2.44783863e-01 -2.03326508e-01 -6.97387636e-01 -8.66880655e-01
-4.49628770e-01 -1.06459491e-01 4.49224889e-01 -8.47466826e-01
-1.47978699e+00 4.03895855e-01 7.75497496e-01 -6.95774376e-01
-2.97117054e-01 -6.21840656e-01 -9.79041815e-01 9.71947908e-01
-1.86289847e+00 -1.05820012e+00 -1.18729331e-01 -5.69360442e-02
1.17737345e-01 2.42088884e-01 4.33531076e-01 -2.16664210e-01
-3.39036673e-01 -2.40948908e-02 5.09692430e-01 -3.42278123e-01
4.85755503e-01 -1.80782151e+00 6.94577575e-01 8.28956127e-01
-1.86159939e-01 2.18714684e-01 8.12247455e-01 -6.83297873e-01
-4.18726832e-01 -1.47216618e+00 3.96579474e-01 -5.36074996e-01
1.05833697e+00 -1.79073915e-01 -7.81974554e-01 5.62528610e-01
1.62490994e-01 -1.78184822e-01 2.78354049e-01 1.11925378e-01
-4.22211528e-01 -1.10587865e-01 -1.43189979e+00 3.51272553e-01
5.25717914e-01 -5.38237154e-01 -5.22242308e-01 1.88115761e-01
1.02348137e+00 -6.83796182e-02 -1.01712644e+00 7.02216387e-01
4.92308378e-01 -1.09033132e+00 8.87058854e-01 -5.96662937e-03
4.75633979e-01 -4.31234419e-01 -3.57969999e-01 -1.62676549e+00
-1.42282858e-01 -4.55302805e-01 3.47573869e-02 1.23597479e+00
6.89593375e-01 -7.57683814e-01 4.53553230e-01 5.55712402e-01
2.44830083e-02 -3.36156547e-01 -1.09760344e+00 -7.86587477e-01
6.52054489e-01 -8.92632902e-01 9.94976044e-01 6.08994842e-01
-3.48508626e-01 -1.39574885e-01 -8.97276625e-02 4.79958028e-01
7.70823359e-01 1.88973397e-01 7.91722015e-02 -1.27585721e+00
-2.51692403e-02 -8.07233632e-01 2.08532829e-02 -8.65947723e-01
6.17097951e-02 -5.84447801e-01 6.02486372e-01 -1.70032358e+00
-1.91731200e-01 -8.29666376e-01 -5.60284257e-01 3.17414969e-01
5.93464496e-03 2.23024175e-01 -1.98096126e-01 1.42469212e-01
-3.52702409e-01 6.53715968e-01 9.33789551e-01 -2.66063213e-01
-2.75920689e-01 -9.84145477e-02 -4.01902795e-01 6.42537057e-01
1.10374403e+00 -4.65824008e-01 -6.05296552e-01 -4.62783098e-01
4.29330468e-01 -1.89792693e-01 6.36253238e-01 -1.25158381e+00
-1.28362194e-01 -3.65846783e-01 3.26139033e-01 -9.87327218e-01
1.15438581e-01 -6.10906899e-01 5.64814322e-02 1.32143036e-01
-3.38582009e-01 -2.67148256e-01 8.33937645e-01 5.40542543e-01
-8.63358080e-02 -1.12440243e-01 9.53802645e-01 -2.37024039e-01
-9.90105391e-01 4.25767809e-01 -4.33141589e-01 3.83059412e-01
6.04794741e-01 2.98341900e-01 -6.93445086e-01 -4.04797405e-01
-6.11527264e-01 7.55441010e-01 3.31769168e-01 1.02723921e-02
2.13389188e-01 -8.74258101e-01 -6.94895625e-01 -1.09733291e-01
1.47436815e-03 3.18873376e-01 1.84010580e-01 4.46553171e-01
-1.03250194e+00 4.77798849e-01 -4.08367626e-02 -8.52702618e-01
-3.13891564e-04 5.63178807e-02 8.45845044e-01 -3.27583462e-01
-3.09313595e-01 9.80404437e-01 3.91170770e-01 -8.52048516e-01
1.27105296e-01 -1.00621223e+00 6.41500950e-02 1.91433370e-01
2.31496856e-01 4.29908991e-01 -3.42970461e-01 -1.38712630e-01
-3.82384002e-01 2.17277527e-01 5.29602885e-01 -5.98109603e-01
1.28539979e+00 -3.34460944e-01 -2.42999807e-01 5.48065960e-01
8.76727641e-01 -4.04801518e-01 -1.82216716e+00 -1.89027339e-01
1.50769828e-02 1.04393372e-02 5.83007991e-01 -1.39775991e+00
-1.00105953e+00 1.15816736e+00 6.63643956e-01 1.91159666e-01
8.35942924e-01 7.58011192e-02 5.42165995e-01 2.02619433e-01
2.62494870e-02 -1.30195570e+00 -5.07515073e-01 4.61095273e-01
7.51782954e-01 -1.26914370e+00 2.88544148e-01 -8.65842700e-02
-4.22287136e-01 8.95286083e-01 4.23786849e-01 -1.71335638e-01
1.24918520e+00 4.47340310e-01 1.42262697e-01 -3.74601275e-01
-3.72435480e-01 -3.88947666e-01 6.19433880e-01 4.19522077e-01
-1.28848208e-02 4.41283643e-01 3.83432984e-01 1.80270806e-01
-2.13733986e-01 -3.48131895e-01 1.19415537e-01 4.85544652e-01
-7.90459991e-01 -4.17641789e-01 -3.45469475e-01 5.26335537e-01
-1.88534290e-01 -4.75919008e-01 -4.04356495e-02 8.58377039e-01
-2.55707745e-03 8.38686764e-01 6.50054574e-01 -4.51739505e-02
-2.32964069e-01 1.84435826e-02 6.58082142e-02 -4.18187082e-01
-4.61592525e-01 7.04282336e-03 2.55012363e-01 -5.07539093e-01
-4.62247878e-01 -3.60662818e-01 -9.84886408e-01 -2.93043166e-01
-9.65713337e-02 2.52942264e-01 8.46809447e-01 8.08161795e-01
1.15725003e-01 3.92883897e-01 4.33098078e-01 -1.03506982e+00
-7.58783937e-01 -1.15827382e+00 -9.90021646e-01 -2.05918223e-01
5.06461143e-01 -5.93212783e-01 -7.60378540e-01 -5.92479110e-01] | [7.272225856781006, 3.4526915550231934] |
b0167dbf-02bf-4dc0-a3e0-3a4386d0727d | spin-road-mapper-extracting-roads-from-aerial | 2109.07701 | null | https://arxiv.org/abs/2109.07701v1 | https://arxiv.org/pdf/2109.07701v1.pdf | SPIN Road Mapper: Extracting Roads from Aerial Images via Spatial and Interaction Space Graph Reasoning for Autonomous Driving | Road extraction is an essential step in building autonomous navigation systems. Detecting road segments is challenging as they are of varying widths, bifurcated throughout the image, and are often occluded by terrain, cloud, or other weather conditions. Using just convolution neural networks (ConvNets) for this problem is not effective as it is inefficient at capturing distant dependencies between road segments in the image which is essential to extract road connectivity. To this end, we propose a Spatial and Interaction Space Graph Reasoning (SPIN) module which when plugged into a ConvNet performs reasoning over graphs constructed on spatial and interaction spaces projected from the feature maps. Reasoning over spatial space extracts dependencies between different spatial regions and other contextual information. Reasoning over a projected interaction space helps in appropriate delineation of roads from other topographies present in the image. Thus, SPIN extracts long-range dependencies between road segments and effectively delineates roads from other semantics. We also introduce a SPIN pyramid which performs SPIN graph reasoning across multiple scales to extract multi-scale features. We propose a network based on stacked hourglass modules and SPIN pyramid for road segmentation which achieves better performance compared to existing methods. Moreover, our method is computationally efficient and significantly boosts the convergence speed during training, making it feasible for applying on large-scale high-resolution aerial images. Code available at: https://github.com/wgcban/SPIN_RoadMapper.git. | ['Vishal M. Patel', 'Jeya Maria Jose Valanarasu', 'Wele Gedara Chaminda Bandara'] | 2021-09-16 | null | null | null | null | ['road-segementation'] | ['computer-vision'] | [ 6.23653680e-02 -4.39346172e-02 7.87710994e-02 -3.15673053e-01
-1.26559749e-01 -7.45739996e-01 4.47112501e-01 -9.07181799e-02
-3.16718161e-01 5.29615343e-01 -5.30610196e-02 -6.32325768e-01
-3.38712752e-01 -1.62302172e+00 -7.60369360e-01 -3.31873596e-01
-1.35784462e-01 1.92708910e-01 8.63779843e-01 -5.02349079e-01
8.87099728e-02 9.85694170e-01 -1.65036452e+00 -7.43666962e-02
1.14484322e+00 8.15265894e-01 5.00593364e-01 6.94116652e-01
-3.56238186e-01 5.88532984e-01 -1.11406758e-01 -1.96613863e-01
2.92651623e-01 -6.95617199e-02 -9.02171910e-01 3.20535041e-02
6.47702992e-01 -3.55734527e-01 -5.21116078e-01 1.19014347e+00
-9.80440825e-02 9.12350118e-02 5.12093902e-01 -1.18912733e+00
-4.72667187e-01 2.50404656e-01 -6.32233322e-01 8.70240256e-02
-1.02733627e-01 -3.15994471e-02 1.08340943e+00 -8.28962982e-01
4.13531482e-01 9.98789728e-01 6.96431577e-01 -3.67843583e-02
-9.36114073e-01 -5.31431079e-01 4.80802625e-01 1.47229478e-01
-1.48010135e+00 5.89239597e-02 7.47735500e-01 -3.93640429e-01
7.55675435e-01 4.08100098e-01 7.70830035e-01 3.10704827e-01
1.14332847e-01 7.47489452e-01 7.87609756e-01 1.52923800e-02
-1.92611832e-02 -3.99930567e-01 2.56061524e-01 9.81015503e-01
1.52538285e-01 -2.48389214e-01 -1.94332615e-01 2.34955475e-01
1.24043846e+00 4.03662831e-01 -3.69191796e-01 -3.93665016e-01
-1.09561503e+00 7.61201322e-01 1.20808029e+00 8.50927830e-03
-4.90404189e-01 1.50614262e-01 -7.79756671e-03 -6.99672401e-02
1.97060034e-01 1.51310802e-01 -3.45819056e-01 4.07671779e-01
-8.76235187e-01 1.64772585e-01 5.52007794e-01 9.09573555e-01
1.25129700e+00 -9.32372361e-02 3.22566539e-01 9.13007021e-01
1.22616202e-01 5.91768682e-01 -9.62726176e-02 -1.07491374e+00
5.38875997e-01 1.23230743e+00 -1.41778573e-01 -1.26431668e+00
-7.83342481e-01 -1.66865915e-01 -9.69397902e-01 5.12286544e-01
3.89592230e-01 -1.66572422e-01 -1.37476122e+00 1.26793432e+00
3.75301570e-01 5.32903820e-02 6.88222796e-02 9.29636121e-01
1.02632928e+00 6.78099871e-01 -1.34168088e-01 7.12990224e-01
1.49303031e+00 -9.56846416e-01 -3.74826849e-01 -5.49918354e-01
5.53911626e-01 -5.06058097e-01 1.12894320e+00 -1.63210914e-01
-5.92955351e-01 -3.71656835e-01 -1.11502028e+00 -2.49013543e-01
-9.41242695e-01 3.43709677e-01 7.51952648e-01 6.37505725e-02
-1.20140672e+00 5.49128175e-01 -7.13149071e-01 -3.81609172e-01
5.35846412e-01 3.26015979e-01 -4.75941211e-01 -8.56917799e-02
-1.33129716e+00 7.21901417e-01 3.69980723e-01 5.79719365e-01
-2.84803271e-01 -3.20346415e-01 -1.21124029e+00 5.74396253e-02
5.32424152e-01 -4.96906072e-01 6.94912374e-01 -6.47290170e-01
-1.09710550e+00 5.81239939e-01 -1.10549830e-01 -3.88626516e-01
4.06618088e-01 -2.64967769e-01 -4.09225821e-01 2.45414704e-01
2.55403280e-01 8.53503048e-01 6.00693882e-01 -9.92681503e-01
-9.98065710e-01 -3.59471977e-01 5.85497260e-01 4.82484609e-01
4.67383899e-02 -4.36821759e-01 -7.17184365e-01 -3.20142955e-01
6.23936653e-01 -1.00426674e+00 -3.64763647e-01 8.40174314e-03
-7.43698299e-01 4.86068986e-02 1.18238842e+00 -7.64134467e-01
1.18058741e+00 -1.96776319e+00 -1.54039532e-01 5.08017242e-01
3.70967865e-01 2.50462800e-01 -9.50700641e-02 3.73864740e-01
1.10112697e-01 2.82626361e-01 -7.43919432e-01 4.74844903e-01
-2.31153533e-01 3.72213423e-01 -2.52346277e-01 2.09539205e-01
3.16638649e-01 9.49288309e-01 -7.30309486e-01 -3.50270987e-01
5.49333394e-01 6.19725108e-01 -2.63369352e-01 -1.64686024e-01
5.48707396e-02 3.02553862e-01 -5.51311314e-01 6.24348283e-01
1.01276553e+00 -1.87047213e-01 -1.45166755e-01 -1.62168846e-01
-5.50065696e-01 5.86281084e-02 -1.18194139e+00 1.22695363e+00
-3.86074156e-01 8.37062061e-01 1.01925740e-02 -8.32623780e-01
1.11264896e+00 -1.84860632e-01 1.81261390e-01 -8.14358830e-01
-1.34122595e-01 6.95195943e-02 -1.75894156e-01 -3.42871249e-01
6.02012515e-01 4.67974514e-01 -5.20199910e-03 -7.26572946e-02
-3.61217439e-01 -4.09129381e-01 3.19480538e-01 2.11686090e-01
1.05807865e+00 1.13461778e-01 1.91282835e-02 -2.59229481e-01
5.32883286e-01 3.61214995e-01 5.50720692e-01 5.09357810e-01
7.15870708e-02 6.26451552e-01 5.37252903e-01 -9.17527139e-01
-8.86643469e-01 -1.10941708e+00 -2.42746860e-01 7.81370580e-01
6.78717196e-01 -2.74609864e-01 -7.19626129e-01 -7.27115571e-01
-6.07352646e-04 2.12781578e-01 -6.00731552e-01 3.19734454e-01
-6.91568136e-01 -6.57446027e-01 3.71834934e-01 7.80092776e-01
1.12606204e+00 -9.27915752e-01 -7.63389766e-01 3.32647264e-02
-3.03000271e-01 -1.32219446e+00 -1.63852468e-01 2.05575943e-01
-7.40429759e-01 -1.33089185e+00 -4.89396989e-01 -7.73162961e-01
9.33537602e-01 6.49336219e-01 8.81668031e-01 2.20084801e-01
-2.92660803e-01 -1.15198344e-01 -1.60519660e-01 -1.42549202e-01
3.23922575e-01 2.83723533e-01 -5.82009852e-01 -1.44810036e-01
1.08181767e-01 -5.52487373e-01 -7.52032757e-01 6.71226144e-01
-8.04735899e-01 5.24989367e-01 5.87654650e-01 6.77106738e-01
7.19236910e-01 4.64875132e-01 -1.36695942e-02 -9.15613174e-01
2.44875520e-01 -4.60240394e-01 -1.03004038e+00 1.47582203e-01
-1.68423235e-01 -4.50952761e-02 5.90104461e-01 2.15589017e-01
-8.86529028e-01 2.98399091e-01 -2.17441127e-01 5.67872450e-02
-3.02877933e-01 5.92601895e-01 -1.14064753e-01 -1.67753607e-01
4.60879385e-01 9.55681801e-02 -1.54475540e-01 -6.11935928e-02
4.26688343e-01 4.91141081e-01 5.11749625e-01 -4.92531657e-02
9.27275777e-01 8.13533187e-01 1.47924691e-01 -1.16193628e+00
-8.13160181e-01 -4.46138024e-01 -1.07736516e+00 -2.54669935e-01
9.38769162e-01 -9.27983165e-01 -3.98305118e-01 5.04390717e-01
-8.70724022e-01 -7.12251782e-01 7.62438327e-02 3.16170096e-01
-2.54969329e-01 5.31071313e-02 -4.35339749e-01 -4.48900223e-01
-2.18608126e-01 -1.13335335e+00 1.01978958e+00 7.32315958e-01
8.77926126e-02 -1.06517208e+00 -2.69147873e-01 1.93498388e-01
3.91223341e-01 4.45400447e-01 7.30136693e-01 -4.45497222e-02
-1.06063664e+00 -2.32089162e-01 -8.43026817e-01 9.85694304e-02
1.35201424e-01 4.83647138e-01 -7.88450003e-01 2.68232554e-01
-7.03538537e-01 1.09420255e-01 1.13466978e+00 6.11514747e-01
1.11292732e+00 -9.24214944e-02 -3.31370533e-01 8.19207072e-01
1.44495010e+00 8.34212899e-02 7.87845373e-01 5.87283492e-01
1.20920169e+00 9.22421098e-01 5.97024322e-01 1.93821937e-02
6.47511780e-01 3.71427506e-01 7.80225933e-01 -7.06188679e-01
-1.55881837e-01 -1.33608907e-01 -4.38360572e-02 2.97784358e-01
-1.69986233e-01 3.79355694e-03 -1.29433072e+00 7.65121818e-01
-2.05388165e+00 -8.22963476e-01 -7.28894830e-01 2.06108570e+00
2.00936317e-01 2.04190120e-01 -1.31174549e-01 -9.38225761e-02
8.06311011e-01 2.08810747e-01 -7.80543149e-01 -2.63723880e-01
-3.07502806e-01 1.24267429e-01 1.16201425e+00 7.65891254e-01
-1.37077928e+00 1.35965145e+00 5.35379982e+00 5.58467746e-01
-1.21304691e+00 -3.35913718e-01 3.66214007e-01 3.96128803e-01
-2.37816229e-01 1.74683839e-01 -8.20044100e-01 6.89917654e-02
4.78461921e-01 3.15279424e-01 4.03971702e-01 8.75026584e-01
1.51362613e-01 -4.33538347e-01 -2.93774247e-01 6.26636624e-01
-3.57279301e-01 -1.37139571e+00 -3.05364523e-02 8.45583230e-02
6.70391679e-01 4.93140131e-01 -1.79090872e-01 7.84027576e-02
7.71571279e-01 -1.02770889e+00 4.67847168e-01 3.19175154e-01
6.36109829e-01 -8.22484970e-01 7.03389406e-01 2.49876469e-01
-1.87717533e+00 -1.43032283e-01 -3.88044119e-01 -5.36196046e-02
1.00838825e-01 5.96067011e-01 -7.64893293e-01 6.06763303e-01
9.08682287e-01 8.07989478e-01 -6.68321192e-01 9.78689492e-01
-6.80781245e-01 2.88030356e-01 -4.37165856e-01 3.04724514e-01
5.95561624e-01 -6.24346435e-01 2.52911836e-01 1.05419588e+00
2.31241494e-01 -6.68924674e-02 1.66121289e-01 7.34343767e-01
8.04715604e-02 2.33461577e-02 -1.00695181e+00 2.31182858e-01
4.46773350e-01 1.57564485e+00 -1.34607232e+00 -2.57452190e-01
-5.10984778e-01 8.04379642e-01 4.42328393e-01 6.97264194e-01
-9.08170164e-01 -9.66150165e-01 8.06174755e-01 2.12922603e-01
4.06172842e-01 -4.71134633e-01 -3.52364957e-01 -9.10759568e-01
1.08912461e-01 -1.97967023e-01 2.56143719e-01 -8.46266806e-01
-7.79311895e-01 7.75094271e-01 -1.10306889e-01 -1.18320954e+00
3.14182043e-01 -6.76405907e-01 -7.07565069e-01 1.04467261e+00
-1.88486469e+00 -1.32378387e+00 -9.26726103e-01 6.26598299e-01
5.78452706e-01 3.32974106e-01 6.16065919e-01 5.08835949e-02
-6.90794408e-01 -2.23534405e-01 -9.38508958e-02 4.94514287e-01
1.73168361e-01 -1.36340654e+00 8.46326947e-01 1.13619351e+00
4.43577133e-02 4.96148467e-01 1.40012011e-01 -7.21405864e-01
-9.28503275e-01 -1.45664096e+00 5.78401625e-01 -1.26459494e-01
6.92595780e-01 -2.75164843e-01 -1.09932852e+00 5.35496175e-01
-9.29088425e-03 1.56496450e-01 2.19557121e-01 1.12905152e-01
-2.59496272e-01 -2.29654640e-01 -7.93631494e-01 9.37530816e-01
1.20472324e+00 -6.04628563e-01 -1.66857541e-01 1.53509453e-01
4.91145968e-01 -5.06282926e-01 -6.58984005e-01 6.74111724e-01
6.16320729e-01 -1.10036230e+00 1.03489935e+00 3.65291499e-02
4.23774719e-01 -7.41366267e-01 -3.43734100e-02 -1.36034238e+00
-3.42607260e-01 -6.97814524e-02 3.57073307e-01 8.42289329e-01
6.30151093e-01 -1.04051328e+00 7.69943595e-01 5.27672708e-01
-2.64952719e-01 -7.38389671e-01 -4.32862312e-01 -5.92049539e-01
-1.54066697e-01 -3.04075807e-01 8.41162622e-01 8.87412906e-01
-3.91009092e-01 3.64931434e-01 6.52722716e-02 7.02252805e-01
5.57056487e-01 4.40878212e-01 8.50403488e-01 -1.39668679e+00
5.00915945e-01 -6.57241464e-01 -5.37686646e-01 -9.43749189e-01
2.45544184e-02 -7.67473042e-01 -2.02412717e-02 -2.18986177e+00
-3.18273425e-01 -6.41733706e-01 -1.97463986e-02 7.91503489e-01
-6.96551129e-02 4.23600078e-01 -7.18994215e-02 3.20561320e-01
-1.72632858e-01 4.44816560e-01 1.29959631e+00 -1.42233044e-01
-3.49071741e-01 -2.16571055e-02 -4.69020665e-01 9.65258121e-01
1.16935289e+00 -1.13077730e-01 -6.04974389e-01 -5.20515621e-01
3.82342994e-01 -1.08462520e-01 5.64824820e-01 -1.18412781e+00
3.48733932e-01 -1.49127811e-01 3.69609088e-01 -1.10384023e+00
1.27714172e-01 -8.31570625e-01 2.39125878e-01 2.25669757e-01
8.29381198e-02 1.27491057e-01 3.55204523e-01 3.61297607e-01
-3.75545204e-01 7.43956864e-02 6.30743265e-01 -3.16880643e-01
-1.14626920e+00 4.75353658e-01 -3.06056172e-01 -1.39182702e-01
9.72005367e-01 -5.29358447e-01 -5.14199615e-01 -1.37543336e-01
-6.04468703e-01 6.90210223e-01 6.58578575e-01 3.87322545e-01
8.53623390e-01 -1.02847445e+00 -4.07710075e-01 2.99052984e-01
2.47011304e-01 5.94623625e-01 4.32954788e-01 9.16570246e-01
-1.22028482e+00 3.44304532e-01 -4.28350568e-01 -6.96424961e-01
-1.34005523e+00 1.67428888e-02 3.68164629e-01 -2.91083790e-02
-1.01662421e+00 7.02857375e-01 6.16840541e-01 -7.54292011e-01
-1.70587376e-01 -6.34156644e-01 -5.76816559e-01 4.64053936e-02
3.90820056e-01 2.39567131e-01 -2.59074513e-02 -8.74846101e-01
-2.95316428e-01 1.00966954e+00 9.91327614e-02 2.95738466e-02
1.27820313e+00 -2.55629390e-01 -2.64921457e-01 2.47721076e-01
8.55140626e-01 -1.18733183e-01 -1.55041790e+00 -1.89258695e-01
-2.36401632e-02 -4.07576680e-01 3.09414268e-01 -4.28819120e-01
-1.28839743e+00 9.14786398e-01 4.02524590e-01 4.88697559e-01
1.10170603e+00 -1.30570516e-01 8.59080911e-01 4.66152191e-01
2.05076531e-01 -1.05895412e+00 -2.74700165e-01 8.57117116e-01
8.12402368e-01 -1.32093012e+00 -4.10302822e-03 -7.50131667e-01
-7.03447938e-01 1.30237854e+00 7.44886339e-01 -1.78896755e-01
7.85169601e-01 1.72119111e-01 6.88530952e-02 -6.37054324e-01
1.78983305e-02 -9.49387372e-01 3.23927462e-01 4.63995963e-01
1.02580478e-02 3.89770806e-01 1.89157709e-01 -1.52981564e-01
-2.72150755e-01 -3.78524154e-01 4.82666016e-01 8.73034239e-01
-7.20570326e-01 -7.91810572e-01 -3.15724909e-01 5.89471340e-01
2.16783956e-04 -2.12798476e-01 -4.29559588e-01 8.24708521e-01
1.49022833e-01 7.28111506e-01 2.60997325e-01 -4.23332453e-01
4.87152100e-01 -3.04540008e-01 -6.36452064e-02 -4.67648923e-01
-5.14835082e-02 -2.05661595e-01 1.11935303e-01 -6.57939553e-01
-2.90923625e-01 -3.55495214e-01 -1.68425727e+00 -2.58842349e-01
-2.02922791e-01 -1.00715145e-01 8.16063702e-01 7.64906824e-01
4.84545618e-01 6.17946386e-01 5.21682143e-01 -8.91502619e-01
5.93547821e-01 -5.26684165e-01 -5.24094760e-01 -1.37455040e-03
3.89359266e-01 -8.49364519e-01 -6.66382164e-02 -4.98967879e-02] | [8.917546272277832, -1.5981851816177368] |
3e1a9443-4a13-4bd6-ab4f-fa79b249f946 | through-the-looking-glass-neural-3d | 2004.10904 | null | https://arxiv.org/abs/2004.10904v2 | https://arxiv.org/pdf/2004.10904v2.pdf | Through the Looking Glass: Neural 3D Reconstruction of Transparent Shapes | Recovering the 3D shape of transparent objects using a small number of unconstrained natural images is an ill-posed problem. Complex light paths induced by refraction and reflection have prevented both traditional and deep multiview stereo from solving this challenge. We propose a physically-based network to recover 3D shape of transparent objects using a few images acquired with a mobile phone camera, under a known but arbitrary environment map. Our novel contributions include a normal representation that enables the network to model complex light transport through local computation, a rendering layer that models refractions and reflections, a cost volume specifically designed for normal refinement of transparent shapes and a feature mapping based on predicted normals for 3D point cloud reconstruction. We render a synthetic dataset to encourage the model to learn refractive light transport across different views. Our experiments show successful recovery of high-quality 3D geometry for complex transparent shapes using as few as 5-12 natural images. Code and data are publicly released. | ['Yu-Ying Yeh', 'Zhengqin Li', 'Manmohan Chandraker'] | 2020-04-22 | through-the-looking-glass-neural-3d-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Li_Through_the_Looking_Glass_Neural_3D_Reconstruction_of_Transparent_Shapes_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Li_Through_the_Looking_Glass_Neural_3D_Reconstruction_of_Transparent_Shapes_CVPR_2020_paper.pdf | cvpr-2020-6 | ['3d-point-cloud-reconstruction', 'transparent-objects', 'point-cloud-reconstruction'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 3.97052974e-01 -2.69653834e-02 7.51402259e-01 -5.52089095e-01
-3.47937077e-01 -4.95978504e-01 5.35561681e-01 -5.39984584e-01
-4.18865122e-03 3.21089864e-01 -2.37756688e-02 -2.71231439e-02
1.22276269e-01 -9.08915877e-01 -8.97380829e-01 -4.09616053e-01
-1.23318121e-01 1.06758583e+00 3.47227961e-01 -1.30554497e-01
2.38945708e-01 9.38257754e-01 -1.79187596e+00 4.35941398e-01
7.76179373e-01 9.91927922e-01 4.33302462e-01 8.68253291e-01
-2.57192552e-01 4.70151424e-01 1.73477098e-01 -1.02347955e-01
9.15921330e-01 3.47484857e-01 -5.25455475e-01 3.21135014e-01
1.05705464e+00 -9.45789456e-01 -8.85217264e-02 7.48277962e-01
3.61986905e-01 8.76738727e-02 6.86330616e-01 -7.59243429e-01
-5.08975923e-01 -4.48500752e-01 -4.70022529e-01 -4.42119300e-01
6.35876238e-01 3.51261735e-01 7.94890165e-01 -1.28985143e+00
9.08385098e-01 1.21013081e+00 8.45979214e-01 6.25885904e-01
-1.37062967e+00 -4.25060540e-01 -3.67219858e-02 -3.36129904e-01
-1.16414380e+00 -5.34857392e-01 9.48772669e-01 -6.06920421e-01
9.63201940e-01 4.52287376e-01 1.16601694e+00 7.17737198e-01
1.76878460e-03 5.20591289e-02 1.16337729e+00 -2.42604375e-01
7.54261091e-02 2.28612408e-01 -1.85383365e-01 8.21624815e-01
1.07621606e-02 2.23319367e-01 -7.11656749e-01 -4.14609045e-01
1.26380455e+00 1.77162170e-01 -5.44786513e-01 -8.24906826e-01
-1.08154380e+00 3.37612361e-01 4.96448666e-01 -6.38911843e-01
-3.53723198e-01 -4.60069440e-02 -3.74158859e-01 1.30368799e-01
9.31193709e-01 1.54132113e-01 -6.24966741e-01 2.53421515e-01
-5.17927706e-01 2.86247939e-01 8.83216977e-01 9.62712586e-01
1.23481345e+00 9.46548730e-02 5.03065705e-01 7.49421060e-01
7.44196475e-01 1.08231080e+00 -4.91436511e-01 -1.64449632e+00
4.75686908e-01 6.49816215e-01 4.54017997e-01 -9.35385406e-01
-3.48630160e-01 -1.17994279e-01 -7.28266537e-01 8.75112712e-01
2.69867420e-01 1.86508521e-01 -8.42688501e-01 1.14837813e+00
8.92530501e-01 2.22876325e-01 -1.67313978e-01 1.08302951e+00
9.22844291e-01 5.21979511e-01 -9.85867321e-01 -1.35221053e-03
8.92154038e-01 -6.43322885e-01 -6.02851659e-02 -1.67948216e-01
-6.64400309e-02 -8.61970425e-01 1.03451800e+00 5.77693284e-01
-1.59576190e+00 -9.04067084e-02 -6.64119720e-01 -3.61053377e-01
1.67645931e-01 -4.11253273e-01 6.49577677e-01 2.14492574e-01
-1.35417259e+00 5.84671319e-01 -7.58132041e-01 -2.94874869e-02
5.11799872e-01 4.41325992e-01 -3.83069515e-01 -5.08342505e-01
-3.28740418e-01 4.28821057e-01 -4.45228904e-01 2.60763943e-01
-8.78651679e-01 -1.24001217e+00 -7.41694868e-01 -3.77644658e-01
9.12223682e-02 -1.28709197e+00 8.23024511e-01 -7.09460735e-01
-1.54885757e+00 1.06449902e+00 -3.21527600e-01 9.17088166e-02
6.21801496e-01 -1.43027782e-01 1.33540884e-01 5.87179601e-01
-1.80354550e-01 4.33488429e-01 8.41338038e-01 -2.03576446e+00
-6.38062209e-02 -6.20428264e-01 6.45098388e-02 6.29325271e-01
1.49851337e-01 -2.51330674e-01 -4.25682753e-01 -5.94415106e-02
6.40100479e-01 -8.07393730e-01 -2.70992249e-01 8.37859154e-01
-4.98933166e-01 5.84281921e-01 7.15101600e-01 -6.96031690e-01
-1.25185996e-01 -1.51424932e+00 5.69779649e-02 3.79847765e-01
4.41427648e-01 -1.50650024e-01 -2.35469699e-01 1.86974466e-01
1.23952277e-01 -1.77294612e-01 -3.12104642e-01 -8.68501127e-01
-2.77265370e-01 1.01970643e-01 -2.58421183e-01 5.80992162e-01
-2.33058944e-01 6.38969719e-01 -5.42711854e-01 -2.89891455e-02
4.78466541e-01 1.13968372e+00 -9.00905311e-01 4.88770753e-01
-5.21194339e-01 8.34181666e-01 -2.82001555e-01 7.07053721e-01
1.25570786e+00 -4.00837421e-01 -2.10117564e-01 -3.22023153e-01
-3.58523428e-01 3.43351901e-01 -1.27792954e+00 1.61707997e+00
-7.22337365e-01 3.93407524e-01 7.51255929e-01 -6.04361035e-02
7.06146061e-01 2.98897445e-01 5.50798118e-01 -6.93401754e-01
-1.44087091e-01 1.63273901e-01 -4.87011790e-01 -3.24500769e-01
3.78341198e-01 -2.20285565e-01 8.55145752e-01 7.32111156e-01
-5.37715375e-01 -9.86265481e-01 -6.64154053e-01 1.89326569e-01
8.55622649e-01 3.88646364e-01 -4.63056445e-01 -9.91726294e-02
2.31605366e-01 -9.56974849e-02 5.22826314e-01 4.83666390e-01
6.08759701e-01 1.18710721e+00 -3.85806859e-01 -9.59382057e-01
-1.33802855e+00 -1.38388526e+00 -1.77134484e-01 3.99274886e-01
3.64834219e-01 -3.44919274e-03 -2.98950464e-01 -4.12263647e-02
7.59128900e-03 3.05494756e-01 -2.05094427e-01 3.81978363e-01
-7.05331624e-01 -5.06542981e-01 -4.16854709e-01 -7.92230666e-02
5.22221029e-01 -9.97650802e-01 -6.46759033e-01 -3.07681441e-01
-2.55526543e-01 -1.32718742e+00 -2.45631546e-01 -1.29212022e-01
-1.16122472e+00 -1.19478011e+00 -5.85390985e-01 -5.84880352e-01
1.03600609e+00 8.45353544e-01 1.44874513e+00 3.51198882e-01
-3.85806501e-01 8.79732490e-01 2.04392433e-01 -4.02447402e-01
-4.26598698e-01 -7.25793898e-01 5.83184324e-02 5.52746095e-02
-3.81921917e-01 -1.18097782e+00 -8.29227269e-01 3.54607284e-01
-6.22823536e-01 6.27146363e-01 -1.16313964e-01 2.04346880e-01
6.99394941e-01 -6.08351469e-01 -5.19793153e-01 -7.88051367e-01
-1.06709898e-01 -1.37182355e-01 -1.11943209e+00 1.68535747e-02
-3.30692470e-01 -2.65560538e-01 4.22126681e-01 -1.65859908e-01
-1.38234782e+00 2.48736009e-01 -2.42450256e-02 -5.58595181e-01
-1.51472628e-01 -3.34171027e-01 2.00302303e-02 -5.82545459e-01
5.63118398e-01 1.96211785e-01 7.22924620e-02 -6.51692629e-01
6.22019805e-02 1.06043369e-01 2.19758540e-01 -5.44419229e-01
1.33745205e+00 1.36628926e+00 4.16041821e-01 -1.25029063e+00
-7.59708047e-01 -4.59594190e-01 -8.26274157e-01 -5.48630595e-01
6.52986348e-01 -1.14776587e+00 -1.10877681e+00 6.12520099e-01
-1.45982301e+00 -7.54726827e-01 -3.17054033e-01 3.80958587e-01
-6.26615107e-01 3.63706082e-01 -6.76958084e-01 -6.89963222e-01
-5.52130163e-01 -9.65603769e-01 1.44408178e+00 -1.30615965e-01
1.82580292e-01 -9.99780536e-01 2.23469198e-01 6.85004413e-01
3.97807002e-01 1.03599034e-01 8.71521175e-01 7.75002480e-01
-1.61795640e+00 4.14369017e-01 -3.20708752e-01 2.26287454e-01
1.15260191e-01 1.82454869e-01 -1.37642109e+00 -1.21898308e-01
3.29232186e-01 -3.66360545e-01 5.77615619e-01 5.48089981e-01
9.89521325e-01 -2.50856847e-01 4.69392817e-03 1.54102015e+00
1.69215727e+00 -3.10829431e-01 5.41371524e-01 9.23127159e-02
1.34313178e+00 8.91043782e-01 1.41638249e-01 5.96136212e-01
7.68233180e-01 7.48255074e-01 1.17212307e+00 -1.82432950e-01
-2.42019087e-01 1.21073104e-01 -1.08610308e-02 9.72412407e-01
-6.68602109e-01 6.02875352e-02 -9.29999709e-01 5.46134412e-02
-1.09748411e+00 -7.18552828e-01 -7.10706770e-01 2.54277492e+00
7.05830753e-01 -2.74810702e-01 -2.87583560e-01 -2.95947671e-01
1.16962329e-01 1.40895543e-03 -7.24934101e-01 -1.06646843e-01
-1.77752301e-01 1.04627302e-02 4.27993864e-01 1.17379606e+00
-3.58204424e-01 8.17199707e-01 6.14089298e+00 -4.83966991e-02
-9.65282619e-01 -3.61875370e-02 2.48855814e-01 -1.66985705e-01
-1.16525543e+00 2.28854199e-03 -5.93605638e-01 -1.07129991e-01
3.77093047e-01 4.95140404e-01 9.08808768e-01 1.94908202e-01
4.62955326e-01 -1.88803613e-01 -9.01337981e-01 1.16716611e+00
1.30400077e-01 -1.59391761e+00 4.29933012e-01 3.93315524e-01
9.55667675e-01 7.93771565e-01 7.02834800e-02 -7.98317015e-01
4.99155194e-01 -8.03469062e-01 7.63017058e-01 1.04519010e+00
8.27809155e-01 -3.02470684e-01 -1.51473850e-01 5.69427073e-01
-9.98721302e-01 3.42173100e-01 -4.22371566e-01 7.91581720e-03
3.47040415e-01 9.80646014e-01 -8.39470744e-01 3.97257417e-01
8.81767571e-01 8.38532329e-01 -2.70892262e-01 1.07549047e+00
6.75424337e-02 1.43821105e-01 -5.58908045e-01 5.89449227e-01
-2.88437665e-01 -8.31958532e-01 8.80813837e-01 4.20002043e-01
3.64415973e-01 1.65374711e-01 1.79395825e-02 1.30760849e+00
5.15716560e-02 -2.43175372e-01 -8.00713837e-01 7.61770904e-01
1.16795927e-01 1.28034258e+00 -5.91128707e-01 7.73976222e-02
-3.36347878e-01 1.06421268e+00 3.15105319e-01 6.85007811e-01
-2.33126417e-01 4.31690425e-01 8.56411397e-01 5.61887085e-01
-1.27287935e-02 -5.31696796e-01 -4.92478997e-01 -1.40400946e+00
5.79652250e-01 -4.30577189e-01 -5.01354396e-01 -1.44317031e+00
-1.23768723e+00 3.40477467e-01 -2.19908655e-01 -1.39862800e+00
1.22856282e-01 -6.90978944e-01 -4.93602276e-01 1.23241913e+00
-1.97288680e+00 -1.15352976e+00 -6.81217492e-01 8.51912260e-01
2.73042709e-01 2.37445876e-01 8.18243265e-01 2.84985993e-02
1.15773879e-01 -4.73305106e-01 7.64518455e-02 -3.69573236e-01
4.24437314e-01 -1.00745130e+00 6.23897016e-01 4.09633994e-01
-3.51228565e-02 6.08651757e-01 5.25974870e-01 -6.66544259e-01
-1.84905088e+00 -9.09104764e-01 4.53905702e-01 -7.20257938e-01
-2.37363726e-02 -6.88315511e-01 -9.13950920e-01 6.90831602e-01
-1.81760877e-01 5.47385275e-01 2.48701140e-01 -3.67487013e-01
-5.44150352e-01 -1.64712563e-01 -1.33757353e+00 6.04231119e-01
1.48829710e+00 -6.81174636e-01 6.85795546e-02 5.69354892e-01
7.14274108e-01 -6.95713818e-01 -6.27164066e-01 2.30053216e-01
7.59867370e-01 -1.52884519e+00 1.56335962e+00 5.58923744e-02
3.87555569e-01 -2.12728411e-01 -2.48874798e-01 -1.29290032e+00
2.26266131e-01 -8.59183967e-01 2.71983426e-02 7.36338973e-01
1.82128504e-01 -7.42996037e-01 1.00462806e+00 8.18008602e-01
-3.08612913e-01 -4.35565531e-01 -6.54248118e-01 -3.48763287e-01
-2.58631587e-01 -7.16343224e-01 5.27772963e-01 8.81144345e-01
-9.31586862e-01 4.44248728e-02 -2.91994452e-01 6.46975040e-01
1.34866428e+00 4.77484882e-01 1.02253580e+00 -1.79146576e+00
-3.38615686e-01 1.75790250e-01 2.80719437e-02 -1.38119423e+00
4.20278050e-02 -8.73028874e-01 6.97827339e-02 -1.61130810e+00
2.79885143e-01 -8.71262491e-01 8.29316974e-01 -4.11590151e-02
4.96971101e-01 4.95689362e-01 -1.50602870e-02 3.69463235e-01
-1.16455905e-01 7.02697456e-01 1.71314025e+00 1.26858562e-01
-3.11735034e-01 1.54905617e-01 -1.83388785e-01 1.23426306e+00
4.88011360e-01 -3.07087064e-01 -4.58765656e-01 -1.18881786e+00
9.10279512e-01 1.03115991e-01 8.49862635e-01 -7.00346529e-01
7.50280544e-03 -2.72131264e-01 4.30434704e-01 -8.91345501e-01
1.18253553e+00 -1.33047771e+00 2.96450466e-01 -3.20643559e-02
-3.13112475e-02 -6.47375658e-02 -6.82683438e-02 5.30333102e-01
4.29600954e-01 1.56377599e-01 6.71644688e-01 -6.33602738e-01
-1.74914271e-01 7.91164517e-01 9.51182246e-02 1.05327427e-01
4.67285067e-01 -5.76420009e-01 -2.34515458e-01 -4.41205114e-01
-5.24085581e-01 6.12383224e-02 1.18846548e+00 -4.98908423e-02
1.28265059e+00 -1.12139702e+00 -5.57214797e-01 6.51620567e-01
9.02635511e-03 8.10293674e-01 3.37780088e-01 4.29911733e-01
-1.21529973e+00 -1.30047530e-01 4.91581559e-02 -9.37626779e-01
-1.53447545e+00 -2.76231289e-01 7.35731006e-01 3.40491235e-01
-1.22926140e+00 9.06693399e-01 5.43644905e-01 -1.29312873e+00
5.14078178e-02 -4.49597806e-01 2.85341173e-01 -6.37180507e-01
5.35075545e-01 5.14624536e-01 1.97790563e-01 -8.30449641e-01
-6.11330383e-02 1.20282137e+00 4.24523443e-01 -2.73072422e-01
1.82877898e+00 -4.73581612e-01 -4.95409697e-01 6.14912510e-01
1.21021032e+00 2.56390721e-01 -1.78115511e+00 -3.11609328e-01
-8.52595985e-01 -8.01396906e-01 1.23486392e-01 -5.62105715e-01
-1.07272506e+00 1.04286802e+00 1.89889297e-01 -1.91534743e-01
7.44644701e-01 -8.66099373e-02 8.67661238e-01 5.78199863e-01
7.99309611e-01 -6.71932757e-01 -2.57411283e-02 8.01225960e-01
9.92212176e-01 -1.31704724e+00 1.62145406e-01 -9.13100302e-01
-7.01436028e-03 1.19900048e+00 4.32989359e-01 -7.26443529e-02
9.51402068e-01 3.88266295e-01 1.97728530e-01 -5.83473504e-01
-6.17034376e-01 3.50584656e-01 4.26063180e-01 8.91918957e-01
-1.59113705e-02 -1.77705288e-01 9.08472121e-01 -5.49848199e-01
-8.57269242e-02 -1.78991064e-01 7.04898357e-01 6.38226330e-01
-2.67670274e-01 -7.89777815e-01 -5.09424090e-01 3.84865075e-01
7.49314651e-02 -2.51083523e-01 -2.80308992e-01 2.86611050e-01
-5.59528582e-02 6.33221388e-01 3.98477256e-01 3.34198251e-02
2.20325932e-01 -4.65391546e-01 7.63383329e-01 -1.02874875e+00
-3.63493174e-01 1.83694661e-02 -3.23658586e-02 -9.00118172e-01
-7.45376587e-01 -7.06427693e-01 -1.28526247e+00 -1.77645251e-01
-4.79270564e-03 -3.89510453e-01 8.86402249e-01 6.51528656e-01
3.65166277e-01 -1.00502588e-01 8.59487712e-01 -1.72377717e+00
-1.41511902e-01 -3.48895669e-01 -7.44638324e-01 2.60515511e-01
7.73386180e-01 -3.38529825e-01 -7.67522931e-01 7.36485943e-02] | [9.564826965332031, -3.079742431640625] |
fd1b6e9e-ab85-49e3-89b6-06761bf4668e | the-devil-is-in-the-detail-simple-tricks | 2108.12284 | null | https://arxiv.org/abs/2108.12284v4 | https://arxiv.org/pdf/2108.12284v4.pdf | The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers | Recently, many datasets have been proposed to test the systematic generalization ability of neural networks. The companion baseline Transformers, typically trained with default hyper-parameters from standard tasks, are shown to fail dramatically. Here we demonstrate that by revisiting model configurations as basic as scaling of embeddings, early stopping, relative positional embedding, and Universal Transformer variants, we can drastically improve the performance of Transformers on systematic generalization. We report improvements on five popular datasets: SCAN, CFQ, PCFG, COGS, and Mathematics dataset. Our models improve accuracy from 50% to 85% on the PCFG productivity split, and from 35% to 81% on COGS. On SCAN, relative positional embedding largely mitigates the EOS decision problem (Newman et al., 2020), yielding 100% accuracy on the length split with a cutoff at 26. Importantly, performance differences between these models are typically invisible on the IID data split. This calls for proper generalization validation sets for developing neural networks that generalize systematically. We publicly release the code to reproduce our results. | ['Jürgen Schmidhuber', 'Kazuki Irie', 'Róbert Csordás'] | 2021-08-26 | null | https://aclanthology.org/2021.emnlp-main.49 | https://aclanthology.org/2021.emnlp-main.49.pdf | emnlp-2021-11 | ['systematic-generalization'] | ['reasoning'] | [-4.17299904e-02 7.85463974e-02 -2.08973184e-01 -4.28615421e-01
-4.68179017e-01 -9.10630584e-01 7.12381184e-01 -3.10262050e-02
-6.62895620e-01 5.78795493e-01 1.28986955e-01 -7.87494183e-01
-4.55102861e-01 -7.79296577e-01 -6.77658021e-01 -4.44645852e-01
-1.81440935e-01 3.60627085e-01 3.92244488e-01 -3.40088814e-01
1.46194428e-01 5.02348363e-01 -1.36627221e+00 5.58678284e-02
9.82116103e-01 9.58014727e-01 4.99123000e-02 6.23396456e-01
3.19581956e-01 2.09249780e-01 -3.04978579e-01 -7.04869986e-01
3.25019836e-01 1.93505958e-01 -9.08722162e-01 -6.02573693e-01
9.73256946e-01 -3.92467707e-01 -4.98141885e-01 8.09481919e-01
6.81981623e-01 1.26496524e-01 8.34792852e-01 -1.36122227e+00
-1.25691986e+00 1.10305607e+00 -1.30176514e-01 4.48459476e-01
-3.81288305e-02 1.98556632e-01 1.30191100e+00 -1.01743925e+00
5.10551095e-01 1.11463487e+00 1.32055402e+00 6.33605897e-01
-1.36358380e+00 -8.92487407e-01 1.80861354e-01 1.27663299e-01
-1.39733875e+00 -3.90067041e-01 3.59881490e-01 -2.84772813e-01
1.31057823e+00 1.39841095e-01 6.07826710e-01 1.55948043e+00
2.03814376e-02 6.34301662e-01 9.89937127e-01 -2.74574161e-01
7.46571720e-02 3.55931409e-02 6.09578788e-01 6.00164771e-01
5.36675930e-01 3.21156353e-01 -5.27396917e-01 1.85370594e-02
6.61604464e-01 -3.07298809e-01 -4.79520679e-01 -2.84524322e-01
-1.03863096e+00 7.69143403e-01 5.22461355e-01 2.34648794e-01
4.93161008e-02 6.29503354e-02 4.58692580e-01 5.04336536e-01
3.85821939e-01 1.00681579e+00 -8.77531528e-01 -3.16962332e-01
-6.96210027e-01 5.86007357e-01 6.93386436e-01 1.20894408e+00
4.67548072e-01 -3.10548618e-02 -4.06118780e-02 1.00455081e+00
-6.66314065e-02 2.62842298e-01 7.30426133e-01 -1.12885618e+00
7.80886114e-01 5.76444924e-01 -2.10444465e-01 -6.75102890e-01
-7.15329707e-01 -1.03204489e+00 -6.64481342e-01 -1.35870993e-01
4.51015830e-01 -1.51701242e-01 -8.32353532e-01 2.05140662e+00
-1.69257611e-01 -1.98828816e-01 3.21017541e-02 5.02391696e-01
7.50497162e-01 3.96734744e-01 1.40238360e-01 4.13946509e-01
9.60354865e-01 -8.56115818e-01 -2.99458295e-01 -1.69482872e-01
1.15647030e+00 -3.16675812e-01 1.64768553e+00 4.82857972e-01
-1.19913709e+00 -5.50042212e-01 -1.39148104e+00 -2.85145909e-01
-6.96833551e-01 2.27549747e-02 7.51265347e-01 8.56730163e-01
-1.55217254e+00 1.09383440e+00 -8.30725610e-01 -5.35398662e-01
5.54401219e-01 5.06197810e-01 -4.89446163e-01 -1.70963019e-01
-1.42844141e+00 1.04703808e+00 5.58057308e-01 -3.04330498e-01
-6.81717992e-01 -1.23574197e+00 -4.97181326e-01 2.83378303e-01
-1.69315428e-01 -6.42948389e-01 1.21684349e+00 -3.30280155e-01
-1.18656087e+00 7.66884565e-01 3.52722019e-01 -7.31246948e-01
4.39311773e-01 -4.36161906e-01 -3.79740417e-01 -1.97922334e-01
-7.68864229e-02 1.11249423e+00 2.25022435e-01 -7.26751804e-01
-4.54623520e-01 -4.16322052e-01 7.93587491e-02 1.32109433e-01
-1.04580820e+00 -2.28708088e-01 -1.33689642e-01 -5.85570037e-01
5.92749417e-02 -9.94918227e-01 5.94759025e-02 -2.75324285e-01
-2.55205363e-01 -6.12024426e-01 5.52626252e-01 -4.58361566e-01
1.21561527e+00 -2.07522964e+00 4.28920910e-02 -1.71352893e-01
1.63273320e-01 1.25522837e-01 -5.85873723e-01 3.90725732e-01
-3.61479193e-01 5.59705019e-01 -1.83603223e-02 -3.02498698e-01
3.71619761e-01 8.04928988e-02 -4.90849197e-01 1.98922604e-01
3.99637759e-01 9.82729256e-01 -6.50056601e-01 9.27051976e-02
-2.28015110e-01 1.54012427e-01 -1.04020333e+00 -1.21000551e-01
8.25730190e-02 -1.98047042e-01 8.36980566e-02 5.63064694e-01
6.37180984e-01 -2.86710799e-01 -1.00117721e-01 -6.57505915e-02
-8.40523839e-02 6.41922474e-01 -6.57804251e-01 1.72570968e+00
-3.19536924e-01 7.57286370e-01 -4.82046157e-01 -8.95642281e-01
7.66629338e-01 3.70857865e-02 -4.12567146e-02 -7.58029580e-01
-1.98266014e-01 2.83645689e-01 3.51130158e-01 -9.75438133e-02
7.38360286e-01 1.50623530e-01 5.53289242e-02 2.88479388e-01
4.36062425e-01 -9.21216831e-02 2.79773414e-01 1.89471453e-01
1.31224751e+00 -3.20819430e-02 -3.30552310e-01 -6.70212209e-01
7.58920982e-03 -4.32407558e-02 3.88043016e-01 8.24623704e-01
-1.46461487e-01 7.06861258e-01 7.19837844e-01 -4.57597584e-01
-1.22198367e+00 -1.48613048e+00 -5.58817744e-01 1.51497579e+00
-3.55205029e-01 -6.84116840e-01 -7.60895371e-01 -7.09195554e-01
2.20957920e-01 1.00673449e+00 -9.21726584e-01 -5.92245162e-01
-5.00691175e-01 -9.25517738e-01 1.06061900e+00 9.89362776e-01
5.68765163e-01 -6.20780349e-01 -2.40012780e-01 -8.31401423e-02
2.83067226e-01 -1.10600019e+00 -1.71312585e-01 4.30470377e-01
-1.17990649e+00 -8.09508562e-01 -6.78811133e-01 -7.23320603e-01
2.82040030e-01 -1.10990509e-01 1.10451567e+00 -9.14796442e-02
1.66964233e-02 1.38208330e-01 -2.08022997e-01 -2.58615077e-01
-4.63141873e-02 9.06377554e-01 2.97696412e-01 -8.73794556e-01
1.59349695e-01 -8.20171118e-01 -4.17082518e-01 4.46442336e-01
-6.25954986e-01 2.18228493e-02 5.33077955e-01 8.05332541e-01
2.84784045e-02 -1.91007540e-01 9.24526691e-01 -7.13486016e-01
9.17849243e-01 -4.67769414e-01 -3.72566581e-01 1.32925957e-01
-1.12960601e+00 2.16015831e-01 7.73925841e-01 -6.14813924e-01
-5.63245058e-01 -5.97248971e-01 -2.46135890e-01 -4.08308089e-01
4.90206853e-02 4.08503830e-01 8.84458423e-02 -5.84779941e-02
1.11039615e+00 1.88379839e-01 -2.60221273e-01 -7.07827806e-01
3.67577940e-01 4.77701545e-01 5.42041183e-01 -8.91634822e-01
6.84235454e-01 -2.69763559e-01 -2.17375055e-01 -5.05240917e-01
-7.46181369e-01 6.59245476e-02 -5.73654592e-01 4.12860155e-01
4.20550376e-01 -9.02918994e-01 -3.29519570e-01 3.37960213e-01
-8.83554220e-01 -7.93116033e-01 -2.85766393e-01 6.48064673e-01
-3.54587406e-01 -1.14432976e-01 -7.62984157e-01 -2.22208381e-01
-3.12244177e-01 -1.04858649e+00 8.97081494e-01 8.05667490e-02
-3.72633010e-01 -1.15850317e+00 4.21543717e-02 2.85755415e-02
8.26140344e-01 -8.85409042e-02 1.36225402e+00 -9.49142218e-01
-2.16398954e-01 9.78197902e-02 -4.28415120e-01 4.84227717e-01
-4.50968929e-02 1.06529661e-01 -1.15552855e+00 -4.54994500e-01
-6.06591761e-01 -6.24853015e-01 1.07433510e+00 1.90403491e-01
1.41840076e+00 -1.36770353e-01 -3.61429870e-01 1.00200391e+00
1.00433815e+00 8.52491055e-03 5.71765304e-01 5.50232768e-01
4.12754208e-01 2.42834240e-01 7.42448941e-02 -3.12941265e-03
3.91038477e-01 3.12463969e-01 2.79550046e-01 3.58303547e-01
-2.43230671e-01 -3.66324872e-01 4.79937226e-01 7.25542605e-01
-7.56122395e-02 -3.25891823e-01 -1.12282240e+00 6.71217084e-01
-1.30042517e+00 -5.95215976e-01 5.62975183e-02 2.18420959e+00
9.51765835e-01 4.67397839e-01 6.91137239e-02 3.29173058e-01
3.18437248e-01 -7.93595463e-02 -5.13256013e-01 -5.91743350e-01
-1.30629003e-01 4.40417230e-01 5.90237439e-01 3.49286914e-01
-9.22288716e-01 1.03048635e+00 7.67662477e+00 7.27566242e-01
-9.97063100e-01 7.79159665e-02 6.05172932e-01 -2.23920554e-01
-4.22351360e-01 -3.20190132e-01 -1.23625159e+00 2.29118168e-01
1.47844279e+00 -1.89928085e-01 5.02059281e-01 7.31205225e-01
-3.51535738e-01 4.39676791e-01 -1.45822144e+00 7.39265680e-01
-1.45283654e-01 -1.27239215e+00 1.88464001e-01 2.94788601e-03
8.76485407e-01 4.40816909e-01 6.34729505e-01 8.51964831e-01
3.43535990e-01 -1.38192463e+00 6.63595080e-01 1.30252123e-01
9.84134436e-01 -7.55057573e-01 4.83921111e-01 1.06574975e-01
-7.62202561e-01 -2.81383216e-01 -6.86526597e-01 -2.07651570e-01
-4.84482288e-01 3.97886395e-01 -1.01922774e+00 3.64120543e-01
8.72795582e-01 8.25467288e-01 -1.19804370e+00 8.46794963e-01
-2.27873385e-01 9.15540278e-01 -4.54384923e-01 -6.90202489e-02
2.67288476e-01 1.27280697e-01 1.62775680e-01 1.26434767e+00
5.24399102e-01 -1.54063836e-01 -3.32483917e-01 9.15085971e-01
-4.19275433e-01 -7.82821104e-02 -6.94631040e-01 -3.07862550e-01
7.00731814e-01 1.01247072e+00 -3.72213006e-01 -1.60933688e-01
-1.44326016e-01 5.27705550e-01 7.96889424e-01 5.44526100e-01
-9.16250706e-01 -6.75999939e-01 6.27656698e-01 1.80414319e-01
2.82758862e-01 -2.29716793e-01 -5.16348183e-01 -8.82564306e-01
-1.21330984e-01 -7.53444195e-01 2.67327011e-01 -7.68123925e-01
-1.20858359e+00 5.65477848e-01 2.30808005e-01 -7.88772881e-01
-1.45598203e-01 -1.18087649e+00 -5.51607192e-01 8.46769512e-01
-1.16314089e+00 -9.00110722e-01 -1.52402192e-01 3.41697335e-01
1.85434923e-01 -3.56364340e-01 8.80264580e-01 4.07460272e-01
-7.61564791e-01 1.27273583e+00 2.26301476e-01 1.50987685e-01
6.91826344e-01 -1.34916735e+00 8.39318693e-01 5.60308754e-01
-8.07071328e-02 1.14967811e+00 4.79884624e-01 -1.89777195e-01
-1.15176249e+00 -1.18128920e+00 7.71663547e-01 -5.84577680e-01
8.26918244e-01 -5.49809992e-01 -9.71903682e-01 1.14328253e+00
3.13979387e-01 -8.91255438e-02 4.64918345e-01 8.17657471e-01
-7.69235730e-01 -2.00555384e-01 -9.57039714e-01 7.71125674e-01
1.71964312e+00 -3.99070203e-01 -7.65364170e-01 3.42251033e-01
1.08904028e+00 -3.75381678e-01 -1.25283873e+00 5.79996407e-01
6.79377854e-01 -1.00043309e+00 9.77673054e-01 -9.80030358e-01
6.42174840e-01 4.44655001e-01 -2.58523107e-01 -1.79961336e+00
-6.41426027e-01 -3.04932654e-01 2.75211204e-02 1.16535676e+00
7.28394151e-01 -1.07078302e+00 6.86872423e-01 4.42051321e-01
-5.48632979e-01 -1.01040232e+00 -8.52934539e-01 -1.16386926e+00
8.03413689e-01 -5.60066402e-01 7.37284541e-01 1.06634033e+00
2.82454286e-02 3.40361834e-01 3.79319102e-01 -1.11201955e-02
2.88381964e-01 -2.54542470e-01 5.38223505e-01 -1.34155965e+00
-5.48825078e-02 -8.06626201e-01 -3.47771227e-01 -9.66294408e-01
1.25018448e-01 -1.25236857e+00 -5.10920525e-01 -1.33820617e+00
1.43116415e-01 -7.22940743e-01 -4.93126720e-01 6.26759648e-01
7.27237985e-02 1.51636615e-01 -1.13962712e-02 -1.50684699e-01
-2.75853217e-01 4.76118773e-01 1.18307507e+00 -1.91335808e-02
2.60594189e-02 -4.01575416e-01 -1.06925905e+00 5.66515863e-01
1.08751869e+00 -2.06407472e-01 -7.38760650e-01 -8.44843864e-01
3.34146202e-01 -6.47963345e-01 3.94461870e-01 -1.34725320e+00
5.66285253e-02 1.18943378e-01 6.89465344e-01 -3.71205956e-01
3.08864236e-01 -3.21600378e-01 -1.09367669e-01 5.09886682e-01
-7.21399009e-01 6.94253325e-01 6.88632905e-01 8.55555236e-02
2.26517618e-01 -1.00088015e-01 6.37115479e-01 2.95587212e-01
-4.40775394e-01 3.99438381e-01 3.82965542e-02 6.13315344e-01
5.81802130e-01 -1.69442266e-01 -8.09309304e-01 4.02630493e-03
-5.63182771e-01 2.50137091e-01 4.78862107e-01 6.23357177e-01
4.28779900e-01 -1.50295758e+00 -6.73902571e-01 3.68626565e-01
7.27725327e-02 -1.54303804e-01 4.51078117e-02 8.87333751e-01
-4.07793790e-01 6.77645266e-01 -3.82954150e-01 -4.70311612e-01
-8.21565866e-01 3.64872009e-01 4.89167184e-01 -1.80206910e-01
-4.04164553e-01 1.21080732e+00 1.72379568e-01 -8.22562337e-01
3.05006921e-01 -1.00458479e+00 5.00171445e-02 -1.02558616e-03
3.55643809e-01 6.35115147e-01 3.43903214e-01 1.00686975e-01
-3.79538268e-01 4.36649173e-01 -3.37134123e-01 -1.12276979e-01
1.51004040e+00 2.70509064e-01 2.88545460e-01 3.42393667e-01
1.61214423e+00 -1.06559440e-01 -1.05394471e+00 -4.56993990e-02
-1.15677819e-01 -5.60366958e-02 9.30663049e-02 -1.04761076e+00
-9.95125532e-01 1.05373037e+00 6.59132600e-01 3.01789522e-01
9.88902688e-01 -2.55060494e-01 6.50879920e-01 7.00773060e-01
1.53268844e-01 -9.65852559e-01 -1.91170856e-01 8.82579863e-01
1.03098977e+00 -8.36493969e-01 -1.59166440e-01 1.86428756e-01
-2.67493904e-01 1.10040188e+00 8.75211239e-01 -2.37491712e-01
3.82300258e-01 2.24636719e-01 -4.86911058e-01 1.43055590e-02
-1.16124022e+00 2.85414070e-01 2.05533028e-01 6.66623116e-01
4.58703309e-01 1.33553632e-02 -7.21221045e-02 8.29035759e-01
-8.96399438e-01 -3.09201349e-02 3.88998866e-01 4.76889998e-01
-4.74693626e-01 -8.90052795e-01 -3.40569317e-02 6.25059247e-01
-3.35826010e-01 -4.90300983e-01 -3.97839606e-01 1.23986685e+00
3.40769766e-03 5.12224078e-01 1.85736328e-01 -7.49664009e-01
4.74022478e-01 4.83370245e-01 6.71823025e-01 -4.30093914e-01
-5.01514316e-01 -7.41430879e-01 2.31945053e-01 -4.42103595e-01
2.90096998e-01 -7.20507860e-01 -1.08641350e+00 -6.26123726e-01
-1.33347526e-01 1.13817630e-02 2.71919876e-01 6.62440777e-01
4.76870924e-01 5.16800404e-01 2.40023628e-01 -6.80412292e-01
-1.11254549e+00 -1.29150510e+00 -4.70555782e-01 2.51004934e-01
1.78706795e-01 -8.47607017e-01 -6.12516820e-01 -4.01569426e-01] | [9.169183731079102, 3.36297607421875] |
4d8a6ade-7d8d-4e92-ada5-57e3fc0a221d | learning-structure-guided-diffusion-model-for | 2306.17074 | null | https://arxiv.org/abs/2306.17074v1 | https://arxiv.org/pdf/2306.17074v1.pdf | Learning Structure-Guided Diffusion Model for 2D Human Pose Estimation | One of the mainstream schemes for 2D human pose estimation (HPE) is learning keypoints heatmaps by a neural network. Existing methods typically improve the quality of heatmaps by customized architectures, such as high-resolution representation and vision Transformers. In this paper, we propose \textbf{DiffusionPose}, a new scheme that formulates 2D HPE as a keypoints heatmaps generation problem from noised heatmaps. During training, the keypoints are diffused to random distribution by adding noises and the diffusion model learns to recover ground-truth heatmaps from noised heatmaps with respect to conditions constructed by image feature. During inference, the diffusion model generates heatmaps from initialized heatmaps in a progressive denoising way. Moreover, we further explore improving the performance of DiffusionPose with conditions from human structural information. Extensive experiments show the prowess of our DiffusionPose, with improvements of 1.6, 1.2, and 1.2 mAP on widely-used COCO, CrowdPose, and AI Challenge datasets, respectively. | ['Jingdong Wang', 'Errui Ding', 'Junyu Han', 'Kun Yao', 'Dongmei Fu', 'Chang Xu', 'Xiyu Wang', 'Jian Wang', 'Qiansheng Yang', 'Zhongwei Qiu'] | 2023-06-29 | null | null | null | null | ['pose-estimation', '2d-human-pose-estimation'] | ['computer-vision', 'computer-vision'] | [-1.24410085e-01 1.92618221e-01 3.26311350e-01 -3.89166802e-01
-8.28073800e-01 -3.94818753e-01 4.98122483e-01 -2.74489552e-01
-3.89228851e-01 5.51062346e-01 4.27201003e-01 5.24859130e-01
1.78620383e-01 -8.08428168e-01 -1.13919806e+00 -6.66526794e-01
2.61103243e-01 8.83850396e-01 1.13968499e-01 -4.35202152e-01
1.09960556e-01 2.21910194e-01 -1.45301127e+00 6.84637949e-02
7.23422825e-01 1.04645097e+00 1.73967153e-01 7.42089212e-01
-4.99554724e-02 6.64664805e-01 -7.10877001e-01 -6.43416107e-01
5.53571284e-01 -1.22507647e-01 -4.86934602e-01 -1.47137672e-01
9.11478519e-01 -5.87426126e-01 -6.16060317e-01 1.12266552e+00
8.20969880e-01 4.97249961e-02 6.37061656e-01 -1.30182278e+00
-7.43659675e-01 3.04750830e-01 -7.80798852e-01 -2.94172108e-01
5.14153540e-01 3.59431922e-01 5.09575367e-01 -1.03836012e+00
9.74893749e-01 1.59920311e+00 1.02801180e+00 6.66885972e-01
-1.22116995e+00 -5.54801583e-01 4.46534418e-02 2.35513777e-01
-1.46580982e+00 -2.10253283e-01 9.38755929e-01 -4.58486468e-01
3.48627031e-01 2.38457084e-01 7.74794400e-01 1.41039085e+00
1.70857590e-02 1.05603623e+00 1.20799112e+00 -1.11998804e-01
2.57864177e-01 -1.72027156e-01 -3.21725875e-01 8.52486253e-01
1.85966104e-01 -8.10378343e-02 -7.49208093e-01 -1.30549669e-01
1.11422467e+00 -7.65744655e-04 -3.75576705e-01 -6.85785055e-01
-1.38365710e+00 5.35080254e-01 9.00831521e-01 -3.05734873e-01
-7.73072541e-01 3.35641861e-01 1.43296629e-01 -9.75717306e-02
6.25757277e-01 4.33541417e-01 -1.24820381e-01 -1.13860451e-01
-8.80632222e-01 7.54637837e-01 4.23808783e-01 1.03416455e+00
9.70411897e-01 -2.55121171e-01 -3.69771302e-01 6.45208597e-01
2.72024184e-01 9.54584301e-01 3.26832294e-01 -1.13156331e+00
6.80799723e-01 5.95652103e-01 4.45715934e-01 -1.62624872e+00
-4.50380951e-01 -2.27915779e-01 -1.19633043e+00 9.04322043e-02
5.28565407e-01 -2.40057200e-01 -1.28592110e+00 1.72303295e+00
5.83696961e-01 2.46004432e-01 -2.52180248e-01 1.24564207e+00
8.40014994e-01 7.11569369e-01 -2.83922464e-01 2.47730210e-01
1.06574655e+00 -1.09875774e+00 -7.47137368e-01 -2.85308003e-01
1.19428694e-01 -4.69983667e-01 9.70253408e-01 5.60175300e-01
-1.21054780e+00 -9.47493255e-01 -8.69352221e-01 -3.12295467e-01
-2.03372017e-01 1.35401472e-01 3.84384722e-01 1.79850012e-01
-9.92058098e-01 5.17041624e-01 -8.37511241e-01 -1.08810619e-01
6.03255808e-01 -1.59353256e-01 -5.55784523e-01 -2.84753561e-01
-1.10257018e+00 6.66090786e-01 2.88926512e-01 4.52318847e-01
-1.09495068e+00 -8.90547931e-01 -8.39114368e-01 -4.25378561e-01
4.25063491e-01 -1.04491484e+00 8.84249091e-01 -3.58957738e-01
-1.34459972e+00 8.02695274e-01 7.29545057e-02 -4.42872763e-01
1.21477818e+00 -8.29020500e-01 1.29715741e-01 3.41396987e-01
2.99461067e-01 9.81723607e-01 1.18170428e+00 -1.53002369e+00
-4.71902609e-01 -5.56761146e-01 -2.45893821e-01 4.14498985e-01
1.40618414e-01 -6.17761910e-01 -8.71191561e-01 -7.75015652e-01
2.10266322e-01 -1.05736482e+00 -4.14866358e-01 3.16238612e-01
-8.95979404e-01 1.15998229e-02 3.97465438e-01 -9.59370553e-01
9.20016468e-01 -2.00077271e+00 7.09680915e-01 4.30268764e-01
6.19507015e-01 -2.71214303e-02 -7.49728903e-02 1.84522718e-02
3.03207815e-01 -1.48005575e-01 -1.63722977e-01 -6.44606829e-01
2.03112006e-01 1.25830904e-01 -1.82697937e-01 4.71577555e-01
8.82173106e-02 1.24811077e+00 -8.60212445e-01 -3.76553804e-01
2.48464301e-01 7.23636687e-01 -4.69841152e-01 5.61471403e-01
-1.98017120e-01 4.72529441e-01 -2.95853317e-01 4.68720883e-01
9.88812506e-01 -4.46507111e-02 -2.91685581e-01 -6.05798185e-01
1.71635404e-01 -3.76545846e-01 -1.42358339e+00 2.41393328e+00
-3.72563414e-02 1.98961437e-01 7.50906672e-03 -5.84174573e-01
1.12793720e+00 -9.87642780e-02 4.04611409e-01 -5.29875875e-01
3.45754949e-03 -1.50289506e-01 -6.75925672e-01 -5.52248597e-01
6.97210550e-01 1.13579385e-01 -1.53181642e-01 2.73739338e-01
5.13563352e-03 -3.31572443e-01 -1.89254314e-01 2.36779436e-01
9.20233011e-01 4.37541038e-01 -2.51026809e-01 1.26570687e-01
5.38068637e-02 1.34710178e-01 4.84144181e-01 8.06395888e-01
-2.01307058e-01 1.31957674e+00 3.43676090e-01 -8.45207751e-01
-1.37502789e+00 -1.26708806e+00 1.62027180e-01 7.34591722e-01
3.84660721e-01 -4.61470872e-01 -1.06632698e+00 -8.28720510e-01
3.83522138e-02 2.79288173e-01 -1.04672718e+00 -1.41255125e-01
-7.81025767e-01 -7.74239421e-01 6.00504041e-01 6.41036034e-01
1.07952631e+00 -7.51808465e-01 -5.97692728e-01 1.24364272e-02
-7.01217115e-01 -8.26119781e-01 -5.38304925e-01 -2.77066618e-01
-4.20652032e-01 -1.05338478e+00 -1.39888585e+00 -6.42170906e-01
9.37522769e-01 8.18309262e-02 1.15560305e+00 -1.83167547e-01
-3.96064281e-01 2.13747516e-01 -2.72532940e-01 -1.79231152e-01
-4.24838588e-02 -9.61419847e-03 2.24608541e-01 1.76164761e-01
1.23623274e-01 -4.07193214e-01 -9.47066903e-01 4.16228205e-01
-5.73020279e-01 2.83487231e-01 7.94236064e-01 6.92962945e-01
9.00371194e-01 -5.81509322e-02 1.50380641e-01 -7.31217265e-01
6.08016849e-01 -3.67956191e-01 -2.83450305e-01 2.10537940e-01
-2.36039549e-01 2.92573720e-01 2.51124263e-01 -3.59727561e-01
-1.19073522e+00 2.70514607e-01 -1.37671560e-01 -8.86684597e-01
-3.36423904e-01 1.02107532e-01 -1.04251795e-01 3.83498780e-02
1.01479685e+00 2.45234936e-01 1.75506771e-01 -6.05678022e-01
9.04509485e-01 2.34878331e-01 1.03069711e+00 -6.87377632e-01
1.17046702e+00 6.25113606e-01 -4.13325056e-02 -3.23154896e-01
-7.07211137e-01 -2.68418074e-01 -6.84360445e-01 -2.91724652e-01
1.11161125e+00 -1.09416687e+00 -6.23352170e-01 7.86180258e-01
-1.00572681e+00 -2.87051231e-01 -2.36392647e-01 1.53953731e-01
-6.24252379e-01 3.09604496e-01 -6.09793365e-01 -4.96435016e-01
-4.43343937e-01 -8.23198557e-01 1.46784914e+00 2.57837862e-01
-1.32874727e-01 -6.08146131e-01 3.62036973e-01 4.04136389e-01
3.53065640e-01 7.80240417e-01 4.44108307e-01 -9.68295112e-02
-5.32226086e-01 -2.60333955e-01 -1.69828415e-01 1.71503469e-01
-1.66236833e-01 -3.98594677e-01 -9.28804219e-01 -2.39487872e-01
-1.74080133e-01 -5.88257790e-01 8.22630584e-01 4.43354011e-01
1.12554014e+00 -4.56905514e-01 -3.90999526e-01 8.75164807e-01
1.14513457e+00 -3.22651953e-01 7.61477232e-01 3.52432936e-01
1.08039355e+00 6.62646890e-01 5.81926882e-01 4.34135824e-01
6.69104815e-01 6.54443085e-01 5.64166784e-01 -8.51027220e-02
-2.07849205e-01 -8.90005350e-01 3.68754491e-02 6.21601701e-01
-3.84953529e-01 2.08469257e-01 -8.18973541e-01 3.57885361e-01
-2.02533269e+00 -9.10513163e-01 7.00500086e-02 1.96685290e+00
8.38506877e-01 2.26784244e-01 2.83257782e-01 -1.03969105e-01
8.30571532e-01 3.00668746e-01 -6.34347260e-01 4.59726632e-01
-1.77705273e-01 -1.02903694e-01 2.00858340e-01 4.10992712e-01
-1.13526726e+00 8.62740695e-01 5.52184629e+00 6.47767484e-01
-7.91624725e-01 -1.37705788e-01 6.63118958e-01 -1.42773181e-01
-2.03156158e-01 -4.46231782e-01 -6.66529477e-01 4.57500726e-01
1.34153068e-01 2.02084959e-01 3.84636492e-01 9.84993279e-01
-1.10554554e-01 7.43001103e-02 -8.24131370e-01 1.49677157e+00
2.76900649e-01 -1.21152186e+00 3.52860875e-02 -2.06610009e-01
8.97805274e-01 -2.27532998e-01 2.42144451e-01 2.06468165e-01
4.45205778e-01 -9.24316108e-01 8.79087985e-01 9.43242967e-01
7.54404247e-01 -7.09127545e-01 7.54457355e-01 2.89363474e-01
-1.05218339e+00 2.38910377e-01 -4.90320772e-01 2.43725657e-01
3.75490636e-01 6.68024719e-01 -7.83296764e-01 6.88735485e-01
1.08011138e+00 7.24848092e-01 -9.04349566e-01 1.05081189e+00
-4.70941752e-01 1.85263723e-01 -4.45500672e-01 2.75321454e-02
-1.02942018e-02 3.01691592e-02 4.69642878e-01 9.22953367e-01
1.91032976e-01 -1.15644872e-01 2.33102247e-01 1.03932393e+00
-1.34203807e-01 -2.88396746e-01 -4.44058955e-01 4.43521351e-01
5.49361169e-01 1.34443188e+00 -4.04141307e-01 -3.84047478e-01
2.83245504e-01 1.28619826e+00 4.12415385e-01 4.79911387e-01
-8.48080695e-01 -5.90012014e-01 4.84288305e-01 2.15043396e-01
9.54682231e-02 -2.07284093e-01 -1.78573817e-01 -1.19461966e+00
2.21958578e-01 -6.64053202e-01 9.80338082e-02 -1.09399521e+00
-1.42730355e+00 7.43665099e-01 2.25631058e-01 -1.04809725e+00
-2.40038767e-01 -3.88272166e-01 -4.98458654e-01 7.57308006e-01
-1.05891347e+00 -1.22981679e+00 -9.67150509e-01 6.43043935e-01
4.19863611e-01 1.31597906e-01 4.08337921e-01 1.71836559e-02
-3.33149999e-01 4.91236150e-01 -1.74492493e-01 2.86229998e-01
8.12568128e-01 -1.60365999e+00 1.04849398e+00 7.22102523e-01
1.59139745e-02 4.98018533e-01 7.56976128e-01 -9.18623090e-01
-1.46189213e+00 -1.19247687e+00 3.16744417e-01 -7.59363770e-01
1.02388486e-01 -6.58171475e-01 -8.04236531e-01 5.20341873e-01
2.53600832e-02 9.58127305e-02 -5.13091646e-02 -3.47498222e-03
-3.16317201e-01 -6.83607683e-02 -1.23654008e+00 6.61913931e-01
1.35504568e+00 -3.11515123e-01 -6.59883559e-01 2.66409963e-01
7.58196950e-01 -8.30151021e-01 -8.68430138e-01 2.93160766e-01
5.40727019e-01 -8.16597402e-01 1.29265797e+00 -4.55207199e-01
3.78373206e-01 -5.57786644e-01 -1.11805275e-01 -1.78253317e+00
-4.39016581e-01 -7.51245618e-01 -2.58479267e-01 9.49503005e-01
7.22948089e-03 -1.27624229e-01 1.08636010e+00 5.95347464e-01
2.29867157e-02 -6.21712208e-01 -7.97574878e-01 -3.83535206e-01
-1.44213840e-01 -2.41739199e-01 7.58286238e-01 7.63795316e-01
-4.90155548e-01 2.71570534e-01 -6.79170668e-01 8.06580111e-02
1.20503461e+00 -2.80351341e-01 1.33927262e+00 -9.96620297e-01
-1.34326234e-01 -4.40735184e-02 -4.81591463e-01 -1.34989023e+00
-2.12956205e-01 -3.79013717e-01 5.10380328e-01 -1.60099959e+00
4.95136268e-02 -4.09637660e-01 5.84877282e-02 3.88683677e-01
-5.24644911e-01 4.46025550e-01 3.35712850e-01 2.60517120e-01
-6.57677412e-01 6.90069556e-01 1.45537293e+00 -2.09978208e-01
-1.45645171e-01 -1.80918813e-01 -4.58776563e-01 7.91360319e-01
6.47413790e-01 -2.96859264e-01 -4.13725048e-01 -5.84064841e-01
5.45638025e-01 -2.47281477e-01 6.65345192e-01 -1.21155965e+00
3.92142147e-01 5.72679937e-02 8.71557057e-01 -7.91566312e-01
5.47339201e-01 -5.90228260e-01 3.84339839e-01 3.30266148e-01
-3.42448175e-01 3.60628814e-01 -6.77566603e-02 8.65249097e-01
3.95789631e-02 2.27940127e-01 4.78721470e-01 -4.76229161e-01
-7.03337371e-01 4.19392407e-01 2.35586345e-01 3.74987781e-01
8.50870728e-01 -1.29528418e-01 -3.99577409e-01 -4.58991438e-01
-8.31585050e-01 2.07357228e-01 4.90336865e-01 5.56758106e-01
9.43957090e-01 -1.80707800e+00 -8.10354531e-01 3.75380069e-01
-3.88114084e-03 9.34970021e-01 4.91835773e-01 6.91271544e-01
-7.32143879e-01 -5.58434799e-02 -3.63375604e-01 -7.54207551e-01
-8.45453441e-01 6.11596406e-01 2.46310622e-01 4.81532104e-02
-8.78393710e-01 1.10893917e+00 3.03059310e-01 -5.34128666e-01
4.66595113e-01 -3.06374013e-01 1.02884911e-01 -5.81972823e-02
7.06908882e-01 5.70573032e-01 -2.05955997e-01 -6.47855520e-01
-1.96519330e-01 8.81104827e-01 -1.72687992e-01 -1.93684265e-01
1.29817390e+00 -1.47502661e-01 1.57852784e-01 1.77549779e-01
1.14078844e+00 -1.43334985e-01 -1.89770687e+00 -2.36700073e-01
-4.88552868e-01 -6.58737063e-01 -3.01941514e-01 -7.78144836e-01
-1.12964511e+00 7.37475753e-01 7.44854271e-01 -1.23673193e-01
9.47616041e-01 -4.04218845e-02 8.83003235e-01 3.13540936e-01
4.19501781e-01 -1.19821775e+00 5.78146517e-01 3.45065556e-02
1.17274594e+00 -1.09008908e+00 -8.97451863e-02 -4.26907800e-02
-8.15466583e-01 8.03572655e-01 8.25931728e-01 -3.79152536e-01
3.31860065e-01 1.23077743e-01 2.48796642e-01 -3.17952871e-01
-3.39438528e-01 1.99116371e-03 4.16667879e-01 9.48165894e-01
-4.59774494e-01 -3.56703848e-02 3.59682649e-01 5.23080587e-01
-4.51723039e-01 -3.86743583e-02 1.00110590e-01 7.93223321e-01
-5.47804356e-01 -4.38618332e-01 -8.19188833e-01 1.66014135e-01
-1.75465774e-02 1.67623654e-01 -4.95821863e-01 4.89250183e-01
2.68452346e-01 4.41089422e-01 -2.20750533e-02 -7.86659420e-01
6.09292328e-01 -1.18423462e-01 4.32191789e-01 -9.73213986e-02
-2.56040961e-01 -1.24631651e-01 -8.98345411e-02 -8.85954916e-01
-2.83948600e-01 -3.72447968e-01 -9.95032728e-01 -4.52071249e-01
-6.26043379e-02 1.04473354e-02 6.50689006e-01 6.27081454e-01
6.04731500e-01 3.40745568e-01 3.58477443e-01 -1.30784345e+00
-4.17285919e-01 -1.06696701e+00 -2.07490072e-01 9.51440871e-01
2.89373666e-01 -8.00975084e-01 -1.94606587e-01 1.60820365e-01] | [7.046763896942139, -0.9290841817855835] |
ab443413-ac1c-4e0a-9ea6-6253ab53c846 | beyond-the-meta-leveraging-game-design | 2305.18477 | null | https://arxiv.org/abs/2305.18477v2 | https://arxiv.org/pdf/2305.18477v2.pdf | Beyond the Meta: Leveraging Game Design Parameters for Patch-Agnostic Esport Analytics | Esport games comprise a sizeable fraction of the global games market, and is the fastest growing segment in games. This has given rise to the domain of esports analytics, which uses telemetry data from games to inform players, coaches, broadcasters and other stakeholders. Compared to traditional sports, esport titles change rapidly, in terms of mechanics as well as rules. Due to these frequent changes to the parameters of the game, esport analytics models can have a short life-spam, a problem which is largely ignored within the literature. This paper extracts information from game design (i.e. patch notes) and utilises clustering techniques to propose a new form of character representation. As a case study, a neural network model is trained to predict the number of kills in a Dota 2 match utilising this novel character representation technique. The performance of this model is then evaluated against two distinct baselines, including conventional techniques. Not only did the model significantly outperform the baselines in terms of accuracy (85% AUC), but the model also maintains the accuracy in two newer iterations of the game that introduced one new character and a brand new character type. These changes introduced to the design of the game would typically break conventional techniques that are commonly used within the literature. Therefore, the proposed methodology for representing characters can increase the life-spam of machine learning models as well as contribute to a higher performance when compared to traditional techniques typically employed within the literature. | ['Anders Drachen', 'James Walker', 'Florian Block', 'Alan Pedrassoli Chitayat'] | 2023-05-29 | null | null | null | null | ['dota-2'] | ['playing-games'] | [ 5.50580285e-02 7.58691356e-02 -1.02483273e-01 5.74600883e-02
-4.74567890e-01 -7.16667771e-01 5.27297318e-01 4.42639977e-01
-7.74252772e-01 3.78957927e-01 1.34969711e-01 -2.87399024e-01
-4.10808384e-01 -1.11713982e+00 -5.70452392e-01 -2.93573886e-01
5.60070612e-02 6.56093895e-01 7.24337935e-01 -6.06796205e-01
5.94743848e-01 3.08029473e-01 -1.92350769e+00 3.85886014e-01
4.08186913e-01 8.45542967e-01 -1.02275111e-01 6.81273699e-01
-1.24211200e-01 1.09448469e+00 -9.58057880e-01 -8.71565819e-01
4.98413920e-01 -2.34478340e-01 -6.84551060e-01 -2.08400816e-01
3.81020397e-01 -1.62637960e-02 -5.02792537e-01 7.32087314e-01
4.31532264e-01 3.34191829e-01 4.10150111e-01 -1.07346225e+00
-5.38682714e-02 8.60443711e-01 -5.15470564e-01 2.47754991e-01
4.00839090e-01 1.00664519e-01 1.06860495e+00 -8.26838464e-02
5.72180986e-01 8.54154646e-01 1.23283100e+00 2.32960746e-01
-1.12307787e+00 -7.01357484e-01 -1.12153523e-01 3.83376956e-01
-1.36802626e+00 7.89768919e-02 5.93300521e-01 -4.39581305e-01
9.01756763e-01 5.89944124e-01 1.13229585e+00 9.10848141e-01
4.10466269e-02 5.72480679e-01 1.21091890e+00 -3.46216351e-01
2.36100435e-01 1.52457967e-01 2.10620791e-01 1.72024637e-01
2.63402343e-01 2.00296685e-01 -6.31033063e-01 -1.62283063e-01
3.75776917e-01 -5.60324974e-02 3.80339116e-01 -1.64276421e-01
-5.74133635e-01 1.11854267e+00 8.48808959e-02 1.76541895e-01
-3.71424943e-01 2.52847821e-01 7.95456827e-01 1.80428773e-01
4.84905183e-01 8.14174652e-01 -8.27898607e-02 -1.13800895e+00
-1.25224853e+00 7.64339626e-01 8.83027792e-01 4.79138702e-01
5.56419790e-01 1.20480582e-01 4.75830212e-02 1.07164657e+00
-9.50248688e-02 -5.30598946e-02 5.22710979e-01 -7.51099467e-01
3.44019532e-01 9.70034838e-01 -3.34900200e-01 -1.42405951e+00
-5.13366520e-01 -6.71603620e-01 -1.82217494e-01 8.90961364e-02
3.95135611e-01 1.25169665e-01 -8.12475145e-01 1.33588588e+00
1.84436738e-01 8.70782360e-02 -4.83415395e-01 7.29784727e-01
7.10821569e-01 3.51035833e-01 8.83620828e-02 3.38822603e-01
1.24566758e+00 -5.48035383e-01 -3.11526090e-01 -3.76043379e-01
5.72552681e-01 -7.68271148e-01 8.33465874e-01 8.05981576e-01
-1.10117412e+00 -4.04321760e-01 -1.03002703e+00 3.65746915e-01
-5.73145449e-01 -3.99259716e-01 8.92162979e-01 1.22752178e+00
-6.76211596e-01 8.05969656e-01 -5.39197147e-01 -4.44653928e-01
3.48729968e-01 5.97238719e-01 -2.32125834e-01 2.65700340e-01
-1.24818230e+00 9.36202943e-01 6.61080062e-01 -3.87034744e-01
-3.43542397e-01 -8.78384054e-01 -5.15404046e-01 -7.35285953e-02
6.72690034e-01 -2.26682112e-01 1.26863182e+00 -8.37083638e-01
-1.37951970e+00 8.80180240e-01 6.16219223e-01 -7.02041864e-01
6.71679378e-01 -6.17201328e-02 -3.61109734e-01 -9.62163210e-02
1.97068185e-01 2.48719588e-01 4.91125554e-01 -9.09144104e-01
-1.18849957e+00 -1.12015575e-01 2.73098141e-01 3.01574059e-02
-2.55474955e-01 1.52166665e-01 -6.56119108e-01 -7.90459096e-01
-4.15806398e-02 -1.13953662e+00 -2.03327134e-01 -7.71340668e-01
-1.04335368e-01 -3.13267149e-02 3.88126612e-01 -6.36420429e-01
1.78342533e+00 -2.06182432e+00 -1.56912178e-01 5.76781511e-01
3.04219931e-01 4.20923501e-01 3.75899076e-01 8.89819562e-01
-1.88017607e-01 2.03252703e-01 7.85496309e-02 -1.26893204e-02
2.41399370e-02 2.37789035e-01 4.35889997e-02 3.96029830e-01
-2.65962690e-01 5.33239007e-01 -6.85119808e-01 -1.36741340e-01
1.68505296e-01 -9.48030278e-02 -5.88109970e-01 -1.92976639e-01
8.00515637e-02 -3.75132829e-01 2.15796288e-04 4.49015379e-01
4.30488467e-01 2.93996215e-01 1.38746994e-02 1.94236830e-01
-2.88728178e-01 3.24515522e-01 -1.33494079e+00 1.35659230e+00
-1.13763414e-01 4.81053561e-01 -4.38362241e-01 -1.11567497e+00
1.04210615e+00 -7.69321918e-02 6.59111142e-01 -7.12883770e-01
2.39211082e-01 1.73361316e-01 3.45122933e-01 -3.43001395e-01
1.10095620e+00 -2.46296301e-01 -5.02301753e-01 4.77286428e-01
-1.06521428e-01 1.90635361e-02 4.80316311e-01 3.06145042e-01
1.44934356e+00 5.01096295e-03 2.17499226e-01 4.49725017e-02
1.17030390e-01 5.10840714e-01 5.59315324e-01 1.03214240e+00
2.18589907e-03 6.22608304e-01 5.79890609e-01 -5.06913543e-01
-1.14581335e+00 -8.60907197e-01 1.40869349e-01 1.24958253e+00
-9.81339216e-02 -9.10837233e-01 -7.71364272e-01 -4.50391978e-01
1.88337445e-01 4.86744404e-01 -6.93619967e-01 -5.86001337e-01
-4.54100490e-01 -7.20116377e-01 1.06522155e+00 3.18082482e-01
4.66756225e-01 -9.20850456e-01 -8.57405543e-01 4.59131777e-01
-1.83452100e-01 -8.73421788e-01 -8.99054483e-02 2.25453064e-01
-7.23092318e-01 -1.07009089e+00 -3.32085937e-01 -2.92615414e-01
-1.52763605e-01 1.19018257e-01 1.18919122e+00 2.83317447e-01
-5.06936669e-01 3.28003645e-01 -8.44401717e-01 -7.78717637e-01
-5.75929105e-01 6.09445870e-01 -1.27181737e-02 -2.56544556e-02
7.77734399e-01 -4.69752580e-01 -4.02764946e-01 2.63763785e-01
-9.66075838e-01 -9.71130803e-02 6.51878595e-01 4.86279875e-01
1.21719152e-01 2.38079786e-01 5.89897335e-01 -1.24067438e+00
1.01788926e+00 -6.98481143e-01 -2.28021607e-01 -2.60643750e-01
-7.04421163e-01 -3.72841686e-01 4.06949580e-01 -6.12713337e-01
-5.57345271e-01 -1.83285370e-01 -1.14849724e-01 -1.82121053e-01
-1.64676294e-01 8.28128636e-01 2.00710490e-01 -6.72789589e-02
9.47866440e-01 1.16649203e-01 2.28876665e-01 -5.49520791e-01
4.68365615e-03 1.01742435e+00 5.97717285e-01 -5.16955912e-01
8.53963077e-01 2.46144563e-01 -1.50136352e-01 -7.86551356e-01
-3.98867607e-01 -9.07487690e-01 -3.99546385e-01 -6.36773765e-01
3.67780328e-01 -6.44776106e-01 -8.43508422e-01 6.28191233e-01
-5.67907155e-01 6.70321658e-02 -5.89390576e-01 4.26799238e-01
-5.33015132e-01 3.50590616e-01 -6.01042449e-01 -9.39153850e-01
-8.06645006e-02 -7.12077796e-01 4.34807688e-01 2.70911604e-01
-7.72969604e-01 -6.55667245e-01 2.44259804e-01 6.79301083e-01
3.23580980e-01 3.04215938e-01 7.31238842e-01 -1.01283884e+00
-2.35426016e-02 -8.10675383e-01 1.30078048e-01 3.89081270e-01
-1.11088611e-01 -7.91073292e-02 -7.25736439e-01 1.50404721e-02
-1.53314635e-01 -1.67493392e-02 6.40159428e-01 1.31834313e-01
7.46213019e-01 4.21478646e-03 -1.33180887e-01 3.28299582e-01
1.07741094e+00 4.04399872e-01 8.71851623e-01 1.06842268e+00
4.28255200e-01 9.01953340e-01 7.94049561e-01 5.98675728e-01
2.45691076e-01 8.60046446e-01 3.84053111e-01 1.08677372e-01
4.17021662e-02 -4.48814154e-01 2.70350426e-01 8.23521078e-01
-4.32620466e-01 9.24375579e-02 -1.02946973e+00 5.90569019e-01
-1.94388163e+00 -1.24939215e+00 -3.60790193e-01 2.23802209e+00
5.33361912e-01 5.53094983e-01 9.15711284e-01 7.03725338e-01
5.70791364e-01 2.96065435e-02 -2.24407241e-01 -9.89909530e-01
-3.61327007e-02 5.14309406e-01 9.72277224e-01 -1.42700091e-01
-1.02585697e+00 1.03232396e+00 6.67598295e+00 1.36043572e+00
-9.28376615e-01 -7.78526068e-02 2.32465371e-01 -4.66397673e-01
1.29320517e-01 6.67956322e-02 -5.13710260e-01 7.48203337e-01
1.19539034e+00 -2.29500517e-01 4.17357147e-01 8.54438126e-01
2.82880515e-01 -4.03165817e-01 -5.43711543e-01 1.06914973e+00
3.13322484e-01 -1.55797780e+00 -9.54519585e-02 5.30342638e-01
2.67000139e-01 -1.15583800e-01 -5.40198684e-02 6.13486886e-01
4.85083431e-01 -1.04524386e+00 1.06483066e+00 5.45404196e-01
4.98374462e-01 -1.06488073e+00 1.00216269e+00 3.93735945e-01
-9.31940198e-01 -3.08753699e-01 -2.39724934e-01 -6.85090125e-01
1.16235659e-01 4.94613051e-02 -7.47110724e-01 5.68705201e-01
1.08664548e+00 4.92167801e-01 -8.81158113e-01 1.49210036e+00
2.47389719e-01 1.09631383e+00 -4.92213160e-01 -3.17979515e-01
4.43569183e-01 -3.34449470e-01 5.37934184e-01 1.10314322e+00
7.31766596e-02 -1.36818469e-01 -2.89339423e-02 4.73972857e-01
1.68356717e-01 3.14575583e-01 -3.87690127e-01 -1.91264182e-01
4.67338890e-01 1.06736791e+00 -9.28601921e-01 -1.58586606e-01
-3.49271089e-01 6.02672696e-01 1.08590074e-01 -1.79203868e-01
-9.11644697e-01 -3.90500486e-01 6.72890484e-01 8.87769878e-01
2.12149978e-01 1.08190171e-01 -3.71998250e-01 -4.14848417e-01
-1.09162025e-01 -1.38423693e+00 5.67503035e-01 -2.97247887e-01
-1.33620775e+00 2.91495323e-01 1.78846225e-01 -1.43602788e+00
-2.42379785e-01 -4.25742030e-01 -5.04291415e-01 5.76533616e-01
-5.35953820e-01 -1.06802821e+00 -9.13403034e-02 2.56820410e-01
3.35804254e-01 -5.45991302e-01 4.99792933e-01 4.95139420e-01
-3.62985164e-01 7.75199354e-01 1.51169270e-01 2.16650724e-01
5.63515902e-01 -1.11508083e+00 4.75897729e-01 6.65444732e-01
2.92746425e-01 4.68490869e-01 8.52127194e-01 -8.65699887e-01
-1.00255942e+00 -5.86557031e-01 3.42022389e-01 -4.04292315e-01
9.59816754e-01 -4.64953810e-01 -6.19683266e-01 3.63707364e-01
-3.20346028e-01 -7.73119926e-01 1.12728131e+00 3.84577781e-01
-1.26166448e-01 3.98425478e-03 -9.36948717e-01 6.07166708e-01
8.43719840e-01 -4.22116995e-01 -6.59409761e-01 -1.49622530e-01
8.97696018e-02 -3.85266840e-01 -8.14768016e-01 -3.87060852e-03
7.99000382e-01 -1.11286223e+00 8.46309245e-01 -8.04978907e-01
4.79682118e-01 -5.07486127e-02 -3.14550027e-02 -1.16460681e+00
-3.73068690e-01 -6.35328114e-01 1.25324875e-01 1.25595856e+00
2.92548031e-01 -4.50896651e-01 1.44446898e+00 6.33092046e-01
-1.39251068e-01 -7.27790058e-01 -1.08210123e+00 -9.37660336e-01
2.58763582e-01 -9.87443626e-01 6.02447510e-01 6.66143060e-01
1.59726366e-01 2.03321323e-01 -5.45016587e-01 -4.54937756e-01
4.13419783e-01 -3.58842582e-01 1.20027423e+00 -1.62346983e+00
-5.84585249e-01 -8.10882509e-01 -1.28026915e+00 -6.80798650e-01
-3.52859408e-01 -8.78091693e-01 -2.44434759e-01 -1.29725647e+00
2.31573567e-01 -5.61090589e-01 1.52323559e-01 2.62058049e-01
2.66349260e-02 6.30861998e-01 4.81778353e-01 1.90636411e-01
-3.62164736e-01 4.68305536e-02 4.51552421e-01 -2.41204351e-02
-1.56577215e-01 3.37962210e-01 -9.34484243e-01 7.10121989e-01
8.30912828e-01 -6.99477613e-01 -1.60462603e-01 9.91435423e-02
7.74378777e-01 -5.24184823e-01 1.86628416e-01 -1.28777564e+00
3.35304499e-01 3.07863206e-02 -1.37953972e-02 -3.74354064e-01
3.26319873e-01 -7.75291741e-01 5.30088663e-01 3.96191567e-01
2.38217693e-03 9.99829248e-02 3.41670007e-01 5.18174708e-01
-2.36645356e-01 -5.92980623e-01 4.42195266e-01 -1.93270132e-01
-7.34192610e-01 -1.04088247e-01 -9.92768347e-01 1.40002489e-01
1.28068328e+00 -1.15365684e+00 -1.80666298e-01 -4.93508041e-01
-6.35958374e-01 -2.08166063e-01 6.05157673e-01 6.69583261e-01
1.80820510e-01 -9.18236911e-01 -5.23414612e-01 -8.35349485e-02
1.64008051e-01 -3.83052051e-01 4.51909155e-01 7.32526958e-01
-1.12985587e+00 5.69488667e-02 -4.53562289e-01 -3.39817107e-01
-1.43273854e+00 2.29853943e-01 2.02430874e-01 -2.77348161e-01
-7.06331313e-01 5.53288519e-01 -5.50774872e-01 -4.82362747e-01
1.72609642e-01 1.68798402e-01 -3.83763522e-01 4.51313227e-01
3.27256680e-01 6.78973615e-01 3.10526192e-01 -8.95829022e-01
-1.57486945e-01 2.68014759e-01 -3.29841286e-01 -1.69596136e-01
1.33412671e+00 2.33883202e-01 1.90942302e-01 7.88061619e-01
6.60022974e-01 1.95273817e-01 -7.18105137e-01 1.77459717e-02
4.16657209e-01 -6.97754502e-01 -6.27974197e-02 -7.36672819e-01
-1.01212287e+00 5.45612633e-01 6.02434695e-01 4.79981154e-01
6.79985523e-01 -7.60136470e-02 7.72621155e-01 -2.64214557e-02
4.93248582e-01 -1.64322507e+00 -1.28604937e-02 5.32816827e-01
2.69873261e-01 -7.91779995e-01 1.66214660e-01 -3.16438228e-01
-7.88607299e-01 8.16693604e-01 4.19883609e-01 -2.38502756e-01
4.07168269e-01 2.95147330e-01 9.51410234e-02 -4.02850568e-01
-4.36046958e-01 -2.19109386e-01 -6.12846948e-02 7.59676039e-01
2.87545532e-01 9.80545878e-02 -6.52165949e-01 1.17937577e+00
-9.31924403e-01 -9.14717615e-02 8.55809331e-01 1.05283904e+00
-4.00233805e-01 -1.23170722e+00 -5.41276395e-01 1.02615166e+00
-7.08714962e-01 3.09283775e-03 -8.29249382e-01 1.10234582e+00
4.49518740e-01 9.99946415e-01 1.77465931e-01 -9.87411857e-01
7.38269925e-01 -4.94846404e-02 7.44405612e-02 -7.57854700e-01
-1.53958011e+00 -2.19906881e-01 2.95368195e-01 -3.67472500e-01
-1.60761580e-01 -7.90326536e-01 -9.68651116e-01 -1.02577364e+00
-4.62682128e-01 4.23507392e-01 6.50020063e-01 7.74457812e-01
1.60641044e-01 6.00116789e-01 1.49689257e-01 -5.61816990e-01
-4.12756473e-01 -1.07668233e+00 -9.58927691e-01 6.74558342e-01
-7.51232743e-01 -9.85368133e-01 -2.47334495e-01 -2.83084661e-01] | [6.550948143005371, 0.37324920296669006] |
3ee1965a-8060-4b2b-aa71-0c9343452de7 | scalable-educational-question-generation-with | 2305.07871 | null | https://arxiv.org/abs/2305.07871v1 | https://arxiv.org/pdf/2305.07871v1.pdf | Scalable Educational Question Generation with Pre-trained Language Models | The automatic generation of educational questions will play a key role in scaling online education, enabling self-assessment at scale when a global population is manoeuvring their personalised learning journeys. We develop \textit{EduQG}, a novel educational question generation model built by adapting a large language model. Our extensive experiments demonstrate that \textit{EduQG} can produce superior educational questions by further pre-training and fine-tuning a pre-trained language model on the scientific text and science question data. | ['Emine Yilmaz', 'Hamze Muse', 'Sahan Bulathwela'] | 2023-05-13 | null | null | null | null | ['question-generation'] | ['natural-language-processing'] | [ 1.02472469e-01 3.27236593e-01 1.50951326e-01 -3.07941645e-01
-1.06295609e+00 -8.22331071e-01 9.44266498e-01 6.27828240e-01
-3.80042315e-01 7.34139502e-01 6.52402282e-01 -1.15804195e+00
-4.56629664e-01 -1.37735081e+00 -6.77930236e-01 1.49637923e-01
1.68873936e-01 4.69101489e-01 3.46533418e-01 -5.78982770e-01
3.84373248e-01 7.10080862e-02 -1.60010421e+00 7.62623325e-02
1.55218482e+00 4.72967833e-01 1.88438028e-01 1.27682066e+00
-6.36277378e-01 1.08724034e+00 -8.26522827e-01 -7.64699340e-01
-1.55213892e-01 -8.06199193e-01 -1.21389937e+00 -2.96883881e-01
9.24512327e-01 -1.78799376e-01 -3.79372209e-01 9.09718156e-01
4.20643181e-01 6.04637623e-01 5.35964549e-01 -7.89269865e-01
-9.94985580e-01 7.12021708e-01 2.57281333e-01 3.84979993e-01
9.24323201e-01 1.02741331e-01 1.05042934e+00 -4.09550041e-01
4.72927153e-01 1.08620071e+00 4.51867014e-01 6.57842875e-01
-1.02097237e+00 -6.87497497e-01 2.63334922e-02 2.24729300e-01
-9.95754898e-01 -2.05047116e-01 4.25655186e-01 -5.36441863e-01
4.98806238e-01 5.45085847e-01 9.23115015e-01 9.10513163e-01
4.50423539e-01 6.15316987e-01 1.30074584e+00 -2.89561898e-01
2.43755922e-01 9.08819884e-02 1.44933071e-02 1.01468325e+00
1.09105468e-01 -3.25584888e-01 -6.16131425e-01 -1.95463538e-01
7.43097663e-01 -2.64840454e-01 -6.88872188e-02 3.96249205e-01
-9.66912866e-01 9.75463986e-01 1.20965913e-01 1.93923697e-01
8.71657804e-02 1.56180054e-01 -1.05436631e-01 8.53852749e-01
2.72815794e-01 1.16990960e+00 -3.71413469e-01 -5.72904229e-01
-1.10775983e+00 6.26448572e-01 1.03388202e+00 1.14921808e+00
4.78362322e-01 -2.97680467e-01 -5.19034624e-01 7.10879862e-01
2.07000315e-01 3.04990828e-01 7.85856307e-01 -1.24633217e+00
2.26730376e-01 9.82653618e-01 -1.43867120e-01 -3.88981193e-01
-8.74058753e-02 -5.54792881e-01 -3.66154283e-01 -1.41805619e-01
5.53601086e-01 -4.81278896e-01 -8.43243182e-01 1.44134533e+00
1.89466774e-01 2.34556213e-01 6.99858144e-02 1.91738456e-01
1.53051162e+00 7.00730324e-01 4.62936282e-01 3.19563776e-01
1.55020595e+00 -9.11038637e-01 -5.09811997e-01 -1.31014645e-01
5.86872220e-01 -9.46502090e-01 1.35441744e+00 4.85911906e-01
-1.29334152e+00 -6.39095962e-01 -6.97922170e-01 -4.99568731e-01
-5.44306517e-01 -2.56484479e-01 6.33570969e-01 9.94234443e-01
-1.37223291e+00 4.34958696e-01 -3.39999855e-01 -2.23432124e-01
3.63969296e-01 1.60961330e-01 -7.55373016e-02 -1.61819309e-01
-1.18422031e+00 3.99700344e-01 3.11505795e-02 -5.08580446e-01
-7.96591520e-01 -1.14220929e+00 -8.77255321e-01 7.91076049e-02
3.64950061e-01 -9.75724459e-01 1.42204130e+00 -8.44937861e-02
-1.87340212e+00 8.98201406e-01 -5.98810241e-02 -5.80077395e-02
4.44638461e-01 -1.72166467e-01 -4.13373441e-01 -3.97352055e-02
8.06518793e-02 6.11455739e-01 3.97218555e-01 -5.17818272e-01
-7.03814209e-01 -1.02339372e-01 3.65328074e-01 1.49787560e-01
-6.00235283e-01 -6.54102564e-02 -2.08712533e-01 -6.23029768e-01
-6.39880151e-02 -5.72559476e-01 -4.97964919e-01 -2.71833628e-01
-2.12642662e-02 -8.74373078e-01 4.17662621e-01 -7.30887294e-01
1.27344120e+00 -1.25721014e+00 -3.07177544e-01 3.22861262e-02
4.16857570e-01 2.25501865e-01 -3.46082300e-01 6.12229526e-01
4.17645335e-01 3.96349698e-01 2.75664806e-01 -3.09717469e-02
2.71890551e-01 -3.97939421e-02 -1.72217056e-01 -3.48626673e-01
-1.58825666e-01 1.06999123e+00 -1.45537364e+00 -6.50305033e-01
-4.60420325e-02 8.10644403e-02 -9.96382356e-01 7.44186521e-01
-6.46353304e-01 3.45949322e-01 -7.86469519e-01 2.70585418e-01
1.32343709e-01 -5.06195009e-01 -2.82401651e-01 7.68888593e-01
-1.48784056e-01 9.35698628e-01 -9.63328540e-01 1.93266308e+00
-7.52620816e-01 6.83887005e-01 -2.50897229e-01 -3.17992866e-01
1.02112150e+00 2.16780290e-01 3.93174201e-01 -9.24541593e-01
1.75457746e-02 3.39513980e-02 -1.51893124e-01 -8.42517316e-01
9.49796021e-01 -7.35971928e-02 -2.25449070e-01 6.65752649e-01
4.34453815e-01 -7.58945405e-01 5.62432766e-01 7.44043231e-01
1.30863142e+00 3.07456017e-01 -1.37780309e-01 -4.86503005e-01
7.44169414e-01 -1.52219325e-01 -1.30985722e-01 1.14555931e+00
-1.09904371e-01 3.98205489e-01 2.83698738e-01 -1.39304131e-01
-6.61680460e-01 -1.09524441e+00 -1.23958535e-01 1.62250400e+00
-4.04935241e-01 -8.07029247e-01 -7.17680335e-01 -7.95243740e-01
-6.57135919e-02 9.12610650e-01 -3.28611642e-01 -2.94842254e-02
-2.51631826e-01 -3.19935560e-01 4.59943831e-01 1.62269071e-01
3.75690252e-01 -9.53312695e-01 -1.92016229e-01 4.27261651e-01
-1.39550537e-01 -9.72840190e-01 -5.67140579e-01 -2.43877783e-01
-5.22625983e-01 -6.32073641e-01 -7.25983739e-01 -9.33421195e-01
5.85440934e-01 6.09066570e-03 1.56816661e+00 2.87963748e-01
-1.54588267e-01 7.74186194e-01 -2.99250484e-01 -5.72361767e-01
-6.97637796e-01 5.52605450e-01 -3.44365269e-01 -7.40625560e-01
4.58592206e-01 -7.16979563e-01 -8.38507116e-01 -2.32821986e-01
-1.14665759e+00 2.03644589e-01 3.09669137e-01 3.93841952e-01
1.23935364e-01 -1.90748334e-01 9.30527568e-01 -1.24048889e+00
1.31457543e+00 -6.20295942e-01 -6.01307094e-01 2.60811716e-01
-7.63485372e-01 2.84853131e-01 6.80971384e-01 -3.75243753e-01
-1.06528401e+00 -7.16014683e-01 -9.90011454e-01 3.52134049e-01
-2.24880800e-01 6.32813215e-01 -6.19200058e-04 -3.19179595e-01
5.02852976e-01 2.88853586e-01 -4.31871921e-01 -6.39609158e-01
5.77777088e-01 5.82691491e-01 5.69875896e-01 -9.93037522e-01
9.26988542e-01 -2.35997319e-01 -3.11451149e-03 -9.12987113e-01
-9.27706897e-01 -4.46196437e-01 -2.51906991e-01 -3.41204077e-01
6.18260562e-01 -8.19165409e-01 -1.07323170e+00 1.45254284e-01
-6.16815031e-01 -5.20432651e-01 -4.86106038e-01 2.76754677e-01
-1.16129600e-01 2.60389805e-01 -6.13415539e-01 -4.42950100e-01
-2.39477456e-01 -7.93828905e-01 8.77594829e-01 9.72787738e-01
-1.87660426e-01 -1.47067404e+00 4.17553484e-01 1.05376756e+00
5.33368587e-01 -9.29513052e-02 1.00477099e+00 -9.55705583e-01
-5.56521833e-01 -4.07413952e-02 5.93773685e-02 1.55477375e-01
1.41768903e-02 -2.65689522e-01 -6.03719294e-01 -1.93623647e-01
-2.63174862e-01 -6.20558560e-01 6.04412258e-01 -1.76888183e-01
1.29482698e+00 -7.40036666e-01 2.18643293e-01 4.93269533e-01
1.25274396e+00 -3.16348195e-01 2.99756348e-01 3.21501046e-01
5.60350001e-01 5.01637816e-01 -4.09471467e-02 9.90439653e-02
1.28558135e+00 -7.47359321e-02 2.37772316e-02 2.74549603e-01
-4.44258928e-01 -7.25398421e-01 3.10822189e-01 1.31242383e+00
1.57552138e-01 -2.51768589e-01 -1.01752496e+00 8.12620938e-01
-1.11872828e+00 -7.85227418e-01 -4.19373304e-01 2.14757562e+00
1.36100924e+00 6.72793686e-02 -1.33026540e-01 -4.67664063e-01
-7.03897998e-02 9.72098038e-02 -2.91387916e-01 -5.70235491e-01
2.44128287e-01 1.12577283e+00 2.46692181e-01 8.63692522e-01
-4.58969772e-01 9.22477067e-01 6.74726295e+00 8.13979268e-01
-8.19129586e-01 1.15241325e-02 5.61650932e-01 1.00837037e-01
-1.07486129e+00 7.86557719e-02 -1.09448838e+00 4.84202236e-01
1.41621459e+00 -7.88240612e-01 1.35315314e-01 3.66836488e-01
9.63154212e-02 -1.01187110e-01 -8.63415897e-01 3.42110604e-01
1.51497141e-01 -1.37971008e+00 3.38558733e-01 6.34870008e-02
1.17446005e+00 -2.33309627e-01 -6.04561456e-02 7.35336602e-01
1.19998538e+00 -1.19650495e+00 1.23900957e-01 6.95443988e-01
5.56079209e-01 -4.87986833e-01 2.66930103e-01 5.28914988e-01
-8.67230296e-01 4.30063382e-02 -2.01237336e-01 -2.22767264e-01
-9.28407758e-02 4.30403620e-01 -1.17931414e+00 5.19473255e-01
4.09710884e-01 -3.46070938e-02 -1.15390348e+00 1.13646972e+00
-4.97996122e-01 1.17712486e+00 -1.29911065e-01 -5.63680232e-01
2.33905941e-01 -3.77081215e-01 5.09136856e-01 1.15965629e+00
5.80357969e-01 4.82152253e-01 1.79981723e-01 8.47727180e-01
-4.96903896e-01 3.78617555e-01 -1.65824041e-01 -3.37121278e-01
3.70043010e-01 1.36339164e+00 -2.98339814e-01 -3.63514483e-01
-4.65880364e-01 8.06787431e-01 1.31212056e-01 2.71502584e-01
-2.41289750e-01 -7.89964676e-01 5.33698022e-01 3.91417116e-01
-2.09545225e-01 -5.26337862e-01 3.61903049e-02 -1.26808548e+00
-3.73037934e-01 -1.17826390e+00 5.24626851e-01 -7.64975667e-01
-1.09587872e+00 3.44387025e-01 -3.54297817e-01 -5.45211315e-01
-6.11802161e-01 -5.24284124e-01 -7.83258021e-01 1.11718774e+00
-1.59349644e+00 -1.15727103e+00 -5.36472857e-01 4.35666829e-01
4.94405985e-01 -7.33060464e-02 7.88519323e-01 1.46348342e-01
-2.49013379e-01 7.75750577e-01 6.00232407e-02 6.66991621e-02
9.03617620e-01 -1.74941921e+00 5.27755678e-01 7.22433090e-01
1.66979898e-02 7.60054410e-01 7.41925299e-01 -7.18828499e-01
-1.75230217e+00 -1.22178483e+00 1.34888566e+00 -7.75366247e-01
9.74483192e-01 -3.70308310e-01 -9.34899449e-01 6.34765267e-01
7.05245614e-01 -5.80783248e-01 1.23358798e+00 2.41877049e-01
-7.54689202e-02 1.31561160e-01 -9.15612638e-01 6.89886570e-01
1.01422489e+00 -6.01745188e-01 -9.20961738e-01 4.24704373e-01
9.91235912e-01 -4.76747423e-01 -1.56913531e+00 6.76249415e-02
3.48389655e-01 -6.05255544e-01 9.45409060e-01 -9.53859746e-01
8.82815063e-01 2.50467211e-01 4.30869758e-01 -1.26193941e+00
-2.82753944e-01 -1.15505314e+00 -2.16277167e-01 1.49051344e+00
4.03062671e-01 -4.62353468e-01 1.03745520e+00 7.93905079e-01
-5.75098582e-02 -4.68494534e-01 -5.87090135e-01 -4.53810513e-01
7.66542733e-01 -1.27127618e-01 8.48658144e-01 1.04725623e+00
9.31017660e-03 5.06625593e-01 2.33511269e-01 -1.75561488e-01
5.53029418e-01 2.61758156e-02 8.48723650e-01 -1.37301922e+00
-1.97505593e-01 -6.89125836e-01 -1.95333108e-01 -1.33266258e+00
-1.02014691e-01 -1.19067621e+00 -1.98883340e-01 -1.95781898e+00
4.27008141e-03 -4.07981306e-01 -1.33264646e-01 1.78140581e-01
-8.13640654e-01 1.36070192e-01 -1.60509810e-01 -4.89380777e-01
-9.35626328e-01 5.14355183e-01 1.93294179e+00 1.32734224e-01
-5.18867262e-02 1.01446435e-01 -1.37339497e+00 3.97893667e-01
5.23896039e-01 -1.69562712e-01 -5.27930796e-01 -3.73288721e-01
8.83438110e-01 2.46065155e-01 -3.73051665e-03 -1.06706405e+00
7.08945751e-01 -5.51357031e-01 3.53133261e-01 -1.79570943e-01
-1.83262154e-01 -4.16014679e-02 -5.30503452e-01 2.49127612e-01
-8.71764839e-01 1.19315051e-01 2.16509417e-01 7.25363046e-02
4.62580696e-02 -5.60733199e-01 3.75509024e-01 -3.79070818e-01
-4.18206394e-01 3.32278281e-01 -7.04115629e-01 7.22740471e-01
4.05643791e-01 1.14907682e-01 -4.89078581e-01 -8.46912980e-01
-4.65121388e-01 7.84705341e-01 9.37823653e-02 6.68482244e-01
4.26544428e-01 -1.13758481e+00 -1.12363088e+00 9.94015560e-02
-5.09871989e-02 1.22122236e-01 1.86835706e-01 -1.20573333e-02
-5.76033354e-01 6.37433946e-01 3.84339206e-02 -2.26967633e-02
-1.05156624e+00 -1.06835030e-01 1.27451718e-01 -6.70320332e-01
-2.44670153e-01 1.28642690e+00 -1.64683938e-01 -1.01546395e+00
5.05401231e-02 -5.34602523e-01 -5.27236581e-01 -1.83677003e-01
8.03432047e-01 5.49506485e-01 5.47267534e-02 7.24146068e-02
5.55800259e-01 8.53649899e-02 -4.10115905e-02 -3.20083916e-01
1.32538903e+00 -1.26753479e-01 -3.35590765e-02 8.03895220e-02
1.00656426e+00 5.97788453e-01 -7.82533586e-01 -3.14423025e-01
5.69812059e-02 -1.81454316e-01 2.28632361e-01 -1.22055280e+00
-4.70218271e-01 7.59797871e-01 2.79702157e-01 4.47374225e-01
8.94895971e-01 -8.17957669e-02 1.06729496e+00 6.77991807e-01
1.77178040e-01 -8.06516469e-01 3.71594876e-01 7.56922483e-01
5.36114037e-01 -1.04790652e+00 2.52548866e-02 -8.52767900e-02
5.15636541e-02 9.70237672e-01 8.98968518e-01 1.75293967e-01
4.73412991e-01 -4.53198748e-03 3.19738798e-02 -2.28679135e-01
-8.54153454e-01 -2.70885438e-01 7.60596573e-01 2.11906224e-01
9.76279557e-01 1.33917019e-01 -5.05689800e-01 7.66596496e-01
-1.20184553e+00 2.60413855e-01 8.65631521e-01 9.68907595e-01
-7.72991002e-01 -1.62003219e+00 -2.29581594e-01 7.32422471e-01
-7.12065637e-01 -3.63907933e-01 -5.50438106e-01 3.14879388e-01
4.47033718e-02 1.09639525e+00 -1.21418737e-01 -8.10675919e-02
2.91479141e-01 3.10920477e-01 7.47848272e-01 -8.84618044e-01
-1.15231109e+00 -5.10120928e-01 1.06256865e-01 -1.09734185e-01
2.40166709e-01 -5.11122108e-01 -1.08002400e+00 -3.98223102e-01
-3.13346758e-02 8.37774694e-01 6.23307586e-01 9.37796772e-01
4.84509200e-01 8.77954125e-01 3.35574895e-01 1.43571556e-01
-5.96439302e-01 -9.75365281e-01 -9.27930027e-02 2.23204955e-01
3.48950386e-01 1.84349447e-01 -1.35365948e-01 1.89411372e-01] | [10.906654357910156, 7.786113262176514] |
9ea0757f-4c80-44a1-8f62-68c18e6bcc89 | language-independent-sentence-level | null | null | https://aclanthology.org/Y12-1018 | https://aclanthology.org/Y12-1018.pdf | Language Independent Sentence-Level Subjectivity Analysis with Feature Selection | null | ['Vasudeva Varma', 'Aditya Mogadala'] | 2012-11-01 | language-independent-sentence-level-1 | https://aclanthology.org/Y12-1018 | https://aclanthology.org/Y12-1018.pdf | paclic-2012-11 | ['subjectivity-analysis'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.322773456573486, 3.6696999073028564] |
f41b6c13-6f30-4207-8766-d38570855322 | training-end-to-end-dialogue-systems-with-the | null | null | https://core.ac.uk/download/pdf/230775856.pdf | https://core.ac.uk/download/pdf/230775856.pdf | Training End-to-End Dialogue Systems with the Ubuntu Dialogue Corpus | In this paper, we analyze neural network-based dialogue systems trained in an end-to-end manner using an updated version of the recent Ubuntu Dialogue Corpus, a dataset containing almost 1 million multi-turn dialogues, with a total of over 7 million utterances and 100 million words. This dataset is interesting because of its size, long context lengths, and technical nature; thus, it can be used to train large models directly from data with minimal feature engineering. We provide baselines in two different environments: one where models are trained to select the correct next response from a list of candidate responses, and one where models are trained to maximize the loglikelihood of a generated utterance conditioned on the context of the conversation. These are both evaluated on a recall task that we call next utterance classification (NUC), and using vector-based metrics that capture the topicality of the responses. We observe that current end-to-end models are unable to completely solve these tasks; thus, we provide a qualitative error analysis to determine the primary causes of error for end-to-end models evaluated on NUC, and examine sample utterances from the generative models. As a result of this analysis, we suggest some promising directions for future research on the Ubuntu Dialogue Corpus, which can also be applied to end-to-end dialogue systems in general. | ['Joelle Pineau', 'Chia-Wei Liu', 'Laurent Charlin', 'Iulian Vlad Serban', 'Nissan Pow', 'Ryan Lowe'] | 2017-01-01 | null | null | null | null | ['conversation-disentanglement'] | ['natural-language-processing'] | [-4.70208675e-02 4.41474557e-01 -1.14513887e-02 -7.94664800e-01
-1.09502828e+00 -7.07909822e-01 7.99557507e-01 1.03337012e-01
-5.59667587e-01 8.38194788e-01 7.81564593e-01 -3.69721591e-01
3.28754038e-01 -5.23990333e-01 -1.78955734e-01 -2.06064790e-01
5.99192008e-02 8.57189834e-01 -5.48661985e-02 -6.98562920e-01
2.58423239e-01 -7.07039684e-02 -1.09769118e+00 5.91336727e-01
6.54882312e-01 9.05104041e-01 1.04967907e-01 1.13837636e+00
-3.18497755e-02 1.09336555e+00 -9.13563430e-01 -5.17739832e-01
-1.91315651e-01 -8.36172879e-01 -1.43624687e+00 3.61702628e-02
3.34100932e-01 -7.91278481e-01 -2.68109024e-01 5.28316021e-01
6.44388497e-01 5.37098646e-01 6.38379872e-01 -7.79424250e-01
-4.58107650e-01 7.22764909e-01 2.02313066e-01 9.41877887e-02
7.36679375e-01 3.19069326e-01 1.32230008e+00 -8.24315727e-01
6.34929419e-01 1.52180910e+00 3.66605371e-01 9.74291921e-01
-1.23712325e+00 -8.46060440e-02 -2.69293729e-02 -1.31470859e-01
-5.74776351e-01 -9.86312568e-01 5.34747899e-01 -3.31301242e-01
1.33604372e+00 3.37562799e-01 2.82829583e-01 1.34889340e+00
-4.61917324e-03 1.08582282e+00 9.16070938e-01 -5.26135683e-01
2.80891240e-01 2.28246182e-01 4.85911310e-01 5.70333242e-01
-7.81579256e-01 -1.41189113e-01 -5.53370655e-01 -5.07155061e-01
2.84377724e-01 -4.61575955e-01 -3.29381168e-01 3.92568484e-02
-1.01334846e+00 1.30898356e+00 2.23781377e-01 1.86895534e-01
-3.34267527e-01 -2.46261597e-01 7.23140776e-01 5.15285015e-01
7.39573658e-01 7.22704589e-01 -6.28484964e-01 -8.09318721e-01
-5.85793376e-01 5.48241735e-01 1.44813156e+00 7.18423188e-01
6.20347977e-01 -5.59964627e-02 -4.36805815e-01 1.41745853e+00
1.99896142e-01 1.06013022e-01 5.68766594e-01 -1.16246128e+00
6.07554376e-01 3.59302163e-01 1.06289923e-01 -4.47895646e-01
-5.47420204e-01 4.41416129e-02 -3.60081136e-01 -2.77282834e-01
6.52718604e-01 -7.24599779e-01 -4.07575995e-01 1.92317498e+00
2.10136279e-01 -4.61915582e-01 3.69192481e-01 8.27919006e-01
1.06979632e+00 8.11376989e-01 -3.12309191e-02 -2.22918406e-01
1.20401645e+00 -1.22927511e+00 -7.58428633e-01 -3.07490855e-01
1.10297549e+00 -7.47448504e-01 1.26848423e+00 2.76833773e-01
-1.18529165e+00 -4.12604570e-01 -7.72711515e-01 -2.26840928e-01
-1.10204577e-01 2.32408009e-02 5.28633237e-01 4.85132188e-01
-1.14561212e+00 4.25731212e-01 -5.40854692e-01 -4.32834655e-01
-3.21273357e-01 8.37626830e-02 6.03506006e-02 2.07814634e-01
-1.48761547e+00 1.27078819e+00 2.82374561e-01 -5.35261482e-02
-7.47686982e-01 -9.82554555e-02 -8.61575186e-01 1.01353511e-01
2.95191467e-01 -3.04667920e-01 2.04736924e+00 -7.92903841e-01
-1.95896637e+00 5.61086118e-01 -2.55130380e-01 -5.84347248e-01
2.22597331e-01 -1.38793573e-01 -2.10191403e-02 -3.77634540e-02
-1.37247369e-01 6.44839346e-01 1.96181089e-01 -9.46255386e-01
-7.13092566e-01 -1.86910734e-01 3.51936698e-01 4.39029574e-01
-3.38909864e-01 9.87100452e-02 -1.64265469e-01 -1.74203113e-01
-2.58901387e-01 -1.07615447e+00 -4.93665375e-02 -7.15916395e-01
-4.04417306e-01 -6.39545560e-01 4.66228724e-01 -8.16912055e-01
1.26338339e+00 -1.85786903e+00 2.67566532e-01 -2.54921257e-01
3.19762193e-02 1.94555789e-01 -3.17139417e-01 7.88562417e-01
4.11584765e-01 7.72682428e-02 -1.71466470e-01 -6.29499614e-01
8.23922530e-02 -4.12760526e-02 -2.86643147e-01 -5.89746162e-02
1.18869066e-01 7.74664581e-01 -9.87872005e-01 -1.46205634e-01
1.29936010e-01 1.10150270e-01 -5.50234556e-01 7.15954185e-01
-5.72757244e-01 3.26394796e-01 -2.07108185e-01 1.66293383e-01
9.59861726e-02 2.12286394e-02 3.22967291e-01 4.63946499e-02
1.60616681e-01 1.01389277e+00 -4.28198487e-01 1.56564248e+00
-8.07663023e-01 8.09003592e-01 2.38928348e-01 -5.68510056e-01
8.91232729e-01 5.90346515e-01 1.31146237e-01 -5.80797791e-01
2.17267320e-01 1.18881859e-01 2.66262382e-01 -5.65637946e-01
1.01588273e+00 -2.48009134e-02 -4.09384340e-01 9.55740511e-01
4.18541580e-01 -1.59778044e-01 4.13545340e-01 4.69435751e-01
1.03135514e+00 -1.70793429e-01 3.02808374e-01 3.04419044e-02
2.44603455e-01 -9.50089470e-02 1.22501291e-01 8.63411963e-01
-2.29202643e-01 4.52020556e-01 6.82908654e-01 -3.76890153e-01
-9.06389475e-01 -8.03931773e-01 7.32230768e-03 1.54374397e+00
-4.68901247e-01 -4.33939755e-01 -9.34814334e-01 -9.59520400e-01
-3.53734404e-01 1.10970271e+00 -4.05930966e-01 -7.78598292e-03
-5.13827741e-01 -5.27138412e-01 6.88151121e-01 3.27199548e-01
3.16074640e-01 -1.25678253e+00 -3.57748032e-01 3.32673311e-01
-7.25420833e-01 -9.78554726e-01 -6.16229773e-01 1.71656653e-01
-6.90795898e-01 -9.23681378e-01 -5.63941419e-01 -6.76699460e-01
8.15767497e-02 4.70108688e-02 1.39485502e+00 3.26463431e-02
4.41226423e-01 3.94559801e-01 -6.07790947e-01 -2.81094253e-01
-1.07029784e+00 3.67150724e-01 -1.14286747e-02 -1.91357270e-01
6.35271907e-01 -1.87086612e-01 -2.43884072e-01 2.74391443e-01
-5.27679801e-01 1.19952112e-01 2.38719448e-01 1.19396353e+00
-2.06836879e-01 -6.15120947e-01 9.35036540e-01 -9.36702073e-01
1.47047925e+00 -5.62897801e-01 -1.49262488e-01 3.49094301e-01
-3.08463782e-01 -1.30502938e-03 6.71388626e-01 -3.40874940e-01
-1.18169177e+00 -2.71130711e-01 -5.57710290e-01 2.23895490e-01
-2.10299537e-01 6.19461298e-01 1.68368354e-01 4.98454928e-01
8.37625861e-01 2.53652871e-01 2.72187054e-01 -3.62455696e-01
4.43234831e-01 1.41131234e+00 2.13541746e-01 -7.20489860e-01
-1.88161001e-01 -3.23813260e-01 -8.53746235e-01 -1.06426156e+00
-8.49781334e-01 -4.53424364e-01 -4.19354081e-01 -3.07871014e-01
6.72378778e-01 -7.84812033e-01 -7.48870790e-01 3.99100035e-01
-1.47669649e+00 -8.37283313e-01 1.02559865e-01 4.15711462e-01
-7.20726967e-01 3.28089714e-01 -1.17202663e+00 -1.05169094e+00
-5.94695091e-01 -1.25714302e+00 8.23543549e-01 1.89344838e-01
-9.35263991e-01 -1.22102404e+00 3.15686911e-01 5.72133660e-01
4.78925407e-01 -3.01716536e-01 1.08051574e+00 -1.25118816e+00
4.94399257e-02 -9.32043418e-02 7.23776370e-02 5.39380491e-01
1.61266223e-01 -1.18605243e-02 -1.13899624e+00 -4.53299761e-01
3.28106195e-01 -1.15089643e+00 5.94254434e-01 1.63473859e-01
5.76657653e-01 -6.07751966e-01 8.37211013e-02 -2.18612224e-01
7.95804441e-01 3.05303156e-01 3.55073482e-01 4.48011374e-03
3.54237229e-01 1.03139257e+00 6.85051143e-01 4.49701339e-01
6.58980548e-01 8.50341439e-01 1.50611162e-01 2.52119809e-01
1.41530171e-01 -2.30211347e-01 3.35869044e-01 1.01564229e+00
3.77675265e-01 -6.33335233e-01 -8.61606300e-01 6.56845689e-01
-1.89454186e+00 -7.56918013e-01 2.02139914e-01 2.14295101e+00
1.17669201e+00 1.41357392e-01 4.89613205e-01 -3.86675388e-01
5.53022563e-01 4.05015230e-01 -3.07283968e-01 -9.17762220e-01
2.51243919e-01 -4.20068353e-02 -1.67837873e-01 9.21685696e-01
-8.56822789e-01 9.39620733e-01 6.82766438e+00 5.85431397e-01
-1.05735600e+00 5.50809614e-02 9.02723730e-01 -1.90095171e-01
-1.21836364e-01 6.05006628e-02 -7.60519326e-01 4.64508295e-01
1.48397589e+00 2.99427547e-02 5.69239855e-01 8.72281313e-01
2.72923887e-01 -2.12998599e-01 -1.49491942e+00 5.18558264e-01
8.93359035e-02 -1.06512511e+00 -5.88034019e-02 -7.72806853e-02
3.81335825e-01 7.63605386e-02 -2.03652322e-01 8.29415023e-01
5.68952680e-01 -9.06807780e-01 3.56356472e-01 1.45592868e-01
4.48952377e-01 -6.80068254e-01 7.98935473e-01 7.59413064e-01
-4.47137475e-01 5.03855087e-02 -3.33723903e-01 -3.66172433e-01
3.21403414e-01 1.18096665e-01 -1.53316641e+00 1.33272633e-01
1.47254124e-01 2.30598390e-01 -2.90108532e-01 4.26270515e-01
-1.24889631e-02 7.97316492e-01 -2.14540035e-01 -6.91082180e-01
5.03294349e-01 6.34879470e-02 4.00533170e-01 1.41560328e+00
-7.15192109e-02 9.65788811e-02 3.02273154e-01 6.66107059e-01
-3.24692667e-01 2.23554015e-01 -6.21033847e-01 -5.11214398e-02
7.86916673e-01 1.17072678e+00 -2.53322184e-01 -3.04529548e-01
-3.72964442e-01 7.96012878e-01 8.02322328e-01 3.53302717e-01
-4.05415237e-01 -3.68712008e-01 5.03168523e-01 -3.99140626e-01
-1.90284222e-01 -1.33439705e-01 -1.53593317e-01 -1.03159308e+00
-9.07972306e-02 -1.29073250e+00 1.24810629e-01 -4.92397338e-01
-1.28704298e+00 8.31188262e-01 -5.64333238e-02 -7.04025030e-01
-1.15543294e+00 -4.23500657e-01 -7.10571766e-01 8.93308282e-01
-8.81122828e-01 -6.78558588e-01 2.20117047e-01 1.17027350e-01
1.14905751e+00 -2.56121129e-01 1.29765761e+00 -3.34867910e-02
-5.43114245e-01 6.79344177e-01 1.87106550e-01 5.07212102e-01
8.63089800e-01 -1.26701105e+00 5.67654729e-01 3.79241049e-01
5.46943620e-02 7.03837276e-01 7.97100902e-01 -4.41824704e-01
-1.00483286e+00 -6.19598866e-01 1.17385542e+00 -6.74984992e-01
3.13808113e-01 -4.58017081e-01 -9.11047995e-01 6.58708394e-01
5.64600885e-01 -7.09622979e-01 8.12117219e-01 7.28045106e-01
4.85161729e-02 3.53882343e-01 -9.78355050e-01 7.41955996e-01
4.82823223e-01 -6.68432295e-01 -6.81360781e-01 3.83300722e-01
7.91091740e-01 -6.47539020e-01 -9.82446432e-01 1.32701546e-01
6.07502103e-01 -1.17816913e+00 4.37513798e-01 -7.84897268e-01
7.70862281e-01 6.98271573e-01 -2.29376137e-01 -1.78962934e+00
-3.93804489e-03 -7.55609393e-01 -1.56143889e-01 1.31329501e+00
7.94661701e-01 -3.25174123e-01 5.86410284e-01 1.11635268e+00
-1.21802554e-01 -1.12414205e+00 -9.06103790e-01 -3.55200410e-01
3.30877900e-01 -2.38556698e-01 4.68189806e-01 7.13937700e-01
4.77247387e-01 1.18626785e+00 -6.31459832e-01 -5.93040466e-01
-5.61375581e-02 -1.08542934e-01 1.03163838e+00 -8.11894476e-01
-4.06226844e-01 -4.41127449e-01 2.39846423e-01 -1.65455544e+00
3.29038233e-01 -4.00905818e-01 4.68658268e-01 -1.35107803e+00
-3.54397781e-02 -3.02804857e-01 2.99182922e-01 2.96735615e-01
-3.43593866e-01 -2.76175857e-01 3.04870874e-01 1.77757889e-01
-6.45332932e-01 8.82237613e-01 9.10785317e-01 -6.51093796e-02
-4.05876577e-01 2.64445782e-01 -5.70204139e-01 5.93164206e-01
7.14214385e-01 -2.35314548e-01 -5.16993403e-01 -4.43770140e-01
-2.02507272e-01 7.56524861e-01 -2.37237960e-01 -5.16176105e-01
5.48302643e-02 -1.45325348e-01 -1.14406988e-01 -4.29695308e-01
7.64344931e-01 -1.84974894e-01 -7.05671906e-01 7.55580664e-02
-9.66834247e-01 -3.23906243e-02 -3.95285860e-02 2.69723862e-01
-3.39075059e-01 -5.68159282e-01 6.30797744e-01 -3.52307886e-01
-4.28163975e-01 -2.15560213e-01 -8.33702385e-01 3.39410424e-01
5.35994112e-01 1.26860812e-01 -3.95005912e-01 -1.17043638e+00
-5.84363878e-01 4.41937268e-01 2.84200877e-01 5.31095862e-01
5.27219892e-01 -1.09209418e+00 -9.47647154e-01 -1.51051596e-01
1.08378537e-01 -2.18724191e-01 2.02663749e-01 4.19101387e-01
-2.56885022e-01 7.06682622e-01 9.61147919e-02 -5.43931782e-01
-1.32022941e+00 4.10040319e-02 3.67776155e-01 -5.41048169e-01
-1.86241776e-01 8.09419870e-01 3.03440653e-02 -9.77478802e-01
4.56133634e-01 -5.44417500e-02 -3.77597362e-01 7.35248849e-02
5.89711666e-01 2.72410095e-01 1.31618371e-02 -6.26442909e-01
1.20214768e-01 -3.07050347e-01 -3.96849602e-01 -6.26789570e-01
1.01398158e+00 -3.37878048e-01 -3.53909656e-02 6.95612371e-01
1.30954504e+00 -2.01235145e-01 -1.18325615e+00 -4.53446627e-01
-8.62730369e-02 -2.93408900e-01 -3.03542823e-01 -1.01155007e+00
-4.80398029e-01 8.86587083e-01 1.69109955e-01 6.06758237e-01
6.16043329e-01 -1.81655243e-01 9.53678370e-01 8.52357388e-01
2.67536670e-01 -1.22111690e+00 3.23274404e-01 1.24074972e+00
9.80169296e-01 -1.41057622e+00 -3.75082105e-01 8.53542611e-02
-1.02694237e+00 1.18401384e+00 8.83883893e-01 2.06761867e-01
1.74885944e-01 -3.74385007e-02 3.92204702e-01 -1.17491171e-01
-1.30942369e+00 9.01949331e-02 -1.95697267e-02 3.97767901e-01
1.08914125e+00 2.80922316e-02 -3.41006339e-01 4.32115883e-01
-5.10367215e-01 -3.44669223e-01 7.42443264e-01 8.07600260e-01
-5.50542295e-01 -1.17476869e+00 -4.46296521e-02 6.32950306e-01
-3.52960378e-01 -1.16751723e-01 -9.04089332e-01 4.22983021e-01
-7.53062487e-01 1.59534025e+00 -6.77841157e-02 -6.63392961e-01
3.78576010e-01 5.55401504e-01 2.40570888e-01 -9.09381270e-01
-8.87141466e-01 -1.10835060e-01 9.10222352e-01 -2.05875948e-01
-1.12587005e-01 -4.78222728e-01 -9.91525590e-01 -4.01763111e-01
-5.81584156e-01 5.24562836e-01 6.20697021e-01 1.10218906e+00
1.77094057e-01 2.99239218e-01 8.38427305e-01 -9.14182007e-01
-1.14253223e+00 -1.75956452e+00 -1.53204769e-01 3.98672968e-01
3.19254309e-01 -2.57055283e-01 -1.80262402e-01 -3.59525830e-01] | [12.778046607971191, 8.02440357208252] |
664fef10-e365-4428-9d89-043ef999a12a | dimba-discretely-masked-black-box-attack-in | 2207.08044 | null | https://arxiv.org/abs/2207.08044v1 | https://arxiv.org/pdf/2207.08044v1.pdf | DIMBA: Discretely Masked Black-Box Attack in Single Object Tracking | The adversarial attack can force a CNN-based model to produce an incorrect output by craftily manipulating human-imperceptible input. Exploring such perturbations can help us gain a deeper understanding of the vulnerability of neural networks, and provide robustness to deep learning against miscellaneous adversaries. Despite extensive studies focusing on the robustness of image, audio, and NLP, works on adversarial examples of visual object tracking -- especially in a black-box manner -- are quite lacking. In this paper, we propose a novel adversarial attack method to generate noises for single object tracking under black-box settings, where perturbations are merely added on initial frames of tracking sequences, which is difficult to be noticed from the perspective of a whole video clip. Specifically, we divide our algorithm into three components and exploit reinforcement learning for localizing important frame patches precisely while reducing unnecessary computational queries overhead. Compared to existing techniques, our method requires fewer queries on initialized frames of a video to manipulate competitive or even better attack performance. We test our algorithm in both long-term and short-term datasets, including OTB100, VOT2018, UAV123, and LaSOT. Extensive experiments demonstrate the effectiveness of our method on three mainstream types of trackers: discrimination, Siamese-based, and reinforcement learning-based trackers. | ['Jonathan Fieldsend', 'Wenjie Ruan', 'Xiangyu Yin'] | 2022-07-17 | null | null | null | null | ['visual-object-tracking', 'miscellaneous'] | ['computer-vision', 'miscellaneous'] | [ 1.08704753e-01 -3.15652430e-01 4.35855687e-02 1.45361900e-01
-6.72610819e-01 -1.13557398e+00 4.02149916e-01 -5.87910354e-01
-3.82754862e-01 5.70935786e-01 -2.08575591e-01 -4.47929621e-01
2.96438903e-01 -4.82541889e-01 -1.20720744e+00 -8.68537784e-01
-3.12401563e-01 -2.15431288e-01 4.50289786e-01 -1.68741763e-01
-1.57523572e-01 7.67999530e-01 -1.08014071e+00 -3.67696658e-02
4.46967542e-01 1.08936119e+00 -2.08440885e-01 9.21002507e-01
5.63677311e-01 9.74600196e-01 -1.20206451e+00 -6.87394679e-01
8.33417356e-01 -1.17027201e-01 -2.37666965e-01 -3.56926583e-02
8.32573533e-01 -6.04411721e-01 -9.29492474e-01 1.60043681e+00
7.83882320e-01 -8.13068449e-02 2.50662029e-01 -1.62352419e+00
-7.55346298e-01 5.33354998e-01 -5.38982511e-01 3.59575480e-01
1.94880545e-01 1.04198074e+00 3.28130811e-01 -2.87528217e-01
2.84747422e-01 1.51654840e+00 8.43359470e-01 9.45097089e-01
-9.81783211e-01 -1.43876445e+00 2.26434097e-01 1.24111548e-01
-1.09921312e+00 -3.82564813e-01 6.72597051e-01 -3.94660741e-01
3.69213462e-01 1.75301120e-01 4.15864170e-01 1.79616237e+00
2.31549799e-01 6.80410981e-01 7.69162714e-01 1.21902563e-01
-6.47256449e-02 -2.14182556e-01 -4.44037825e-01 5.39493084e-01
4.61291015e-01 1.05220020e+00 -3.68738204e-01 -3.39667201e-01
7.19351530e-01 -8.61972421e-02 -5.53336978e-01 -1.29497215e-01
-1.42662203e+00 7.34669864e-01 5.75342357e-01 -2.34702080e-01
-1.32793754e-01 7.02349842e-01 6.04080081e-01 5.05997002e-01
-1.84106324e-02 5.15890956e-01 -3.93863469e-01 -5.18294089e-02
-5.90425313e-01 3.85353476e-01 5.81056654e-01 1.05025423e+00
3.10154706e-01 7.81546295e-01 -5.05756378e-01 4.25860174e-02
1.85496539e-01 1.22145510e+00 2.24934801e-01 -1.02232122e+00
4.28350925e-01 -1.85259134e-01 3.51607949e-01 -1.34987187e+00
-1.11383922e-01 -4.44729239e-01 -8.22185040e-01 5.34228563e-01
6.08555019e-01 -7.62663007e-01 -9.01093304e-01 1.91313672e+00
4.64415997e-01 6.77628934e-01 1.18131652e-01 1.05364680e+00
7.27976084e-01 4.99176711e-01 6.29584212e-03 -1.79660879e-02
1.21011853e+00 -9.79279459e-01 -6.70748591e-01 -1.55871212e-01
1.86262608e-01 -8.30447435e-01 7.12317109e-01 3.27615291e-01
-7.94529676e-01 -8.17371249e-01 -1.07734251e+00 5.44240177e-01
-3.44945341e-01 -8.75311419e-02 2.46962473e-01 9.44222569e-01
-6.59417331e-01 4.84867245e-01 -7.93624878e-01 8.94182250e-02
7.16367900e-01 3.41079324e-01 -2.59255141e-01 3.06206822e-01
-1.51705456e+00 7.22437263e-01 2.98551679e-01 1.76880419e-01
-1.74993289e+00 -9.21583295e-01 -7.38714814e-01 -9.76008251e-02
7.41794527e-01 -5.08783281e-01 1.32103992e+00 -9.64648724e-01
-1.41257560e+00 3.74689132e-01 3.98663819e-01 -1.03439975e+00
8.24660242e-01 -5.12857795e-01 -6.01306319e-01 2.29468614e-01
-2.73951352e-01 7.90451705e-01 1.62908638e+00 -1.16135192e+00
-4.84160334e-01 -4.62757982e-02 4.24600810e-01 -2.69377768e-01
-3.95051599e-01 3.71545672e-01 -2.63957262e-01 -1.35158527e+00
-6.55170500e-01 -1.13274288e+00 -2.57565260e-01 5.33091545e-01
-5.09738863e-01 1.63191065e-01 1.32829177e+00 -4.20961976e-01
8.54590178e-01 -2.49849010e+00 -2.73835033e-01 -6.90902099e-02
2.47465327e-01 8.34579289e-01 -3.92606735e-01 4.15121987e-02
-2.00221241e-01 1.78840995e-01 1.22063369e-01 2.14042008e-01
7.53518566e-02 9.04209260e-03 -1.01129603e+00 7.83702314e-01
3.33401501e-01 1.07795870e+00 -9.63077426e-01 -3.42885286e-01
8.63823146e-02 3.71602327e-01 -5.16172111e-01 3.63752902e-01
-2.71992207e-01 6.38840199e-01 -3.72122616e-01 7.05449641e-01
7.55320072e-01 2.56767333e-01 -4.29626226e-01 -3.03260356e-01
2.58260667e-01 -2.72186786e-01 -1.06487119e+00 1.29339457e+00
3.45998965e-02 9.13370848e-01 3.46892737e-02 -7.24072397e-01
6.89653814e-01 3.78470004e-01 2.30123207e-01 -4.48889136e-01
3.42559218e-01 -2.41142616e-01 2.30143845e-01 -4.03185606e-01
1.67390808e-01 1.77997679e-01 -1.59841180e-01 1.86800361e-01
-2.37794034e-03 2.17371479e-01 -5.27121685e-02 1.83965862e-01
1.33497131e+00 -2.32119374e-02 -1.15410693e-01 2.44204924e-01
4.02217805e-01 -2.38904819e-01 7.19374299e-01 1.30987966e+00
-7.02922583e-01 3.46310794e-01 3.83109450e-01 -4.88724887e-01
-7.80157685e-01 -1.11809361e+00 2.63337970e-01 8.22284579e-01
3.59583348e-01 -2.40727425e-01 -8.06954265e-01 -1.04338288e+00
2.18811855e-01 2.53875285e-01 -6.17738307e-01 -6.16749167e-01
-6.76764011e-01 -3.45692635e-01 1.50128829e+00 5.46131790e-01
7.24301517e-01 -1.23068035e+00 -7.85461783e-01 -4.48213443e-02
1.17034607e-01 -1.54969895e+00 -7.34436512e-01 -1.37446240e-01
-5.36504507e-01 -1.19799578e+00 -6.49396777e-01 -4.67591137e-01
3.88608485e-01 2.63123602e-01 7.98513055e-01 8.05918649e-02
-3.22017401e-01 3.18994284e-01 -2.66320407e-01 -6.77709937e-01
-5.03120840e-01 -3.47628564e-01 3.75745714e-01 -1.42155634e-02
2.64051575e-02 -3.14068317e-01 -4.79839385e-01 6.95630729e-01
-9.95337903e-01 -6.74178123e-01 4.37526315e-01 9.07907248e-01
1.17870919e-01 2.35078350e-01 3.24443698e-01 -4.64296550e-01
3.16337883e-01 -3.24532866e-01 -1.06770265e+00 1.22264378e-01
5.12260338e-03 -1.76330552e-01 8.54192376e-01 -1.23791432e+00
-6.31671667e-01 5.64828999e-02 -8.57580174e-03 -1.22622406e+00
-2.15506941e-01 -8.68428871e-02 -2.51909047e-01 -7.64682889e-01
8.97594929e-01 2.03878731e-01 -6.04770184e-02 -1.22177757e-01
4.22194600e-01 2.66260237e-01 1.15291917e+00 -5.57767510e-01
1.91191900e+00 5.59733927e-01 -8.95986557e-02 -5.15715778e-01
-6.72568440e-01 9.10095051e-02 -8.43178332e-02 -4.87486988e-01
7.30893910e-01 -1.04501140e+00 -1.24384856e+00 7.72587359e-01
-1.27308726e+00 -2.55798787e-01 -1.24016613e-01 3.63015473e-01
-3.19015801e-01 5.49675524e-01 -4.07760590e-01 -5.03573120e-01
-1.83843091e-01 -1.24528611e+00 9.12916601e-01 3.42431039e-01
2.24924043e-01 -6.67679429e-01 -1.34644182e-02 1.11453243e-01
4.72401351e-01 6.45679533e-01 2.04281136e-01 -6.42443717e-01
-8.70856762e-01 -3.46656740e-01 -7.67383724e-02 4.28306550e-01
-1.23228274e-01 2.12397769e-01 -9.51452613e-01 -7.53779590e-01
-2.78450642e-02 -5.62703192e-01 6.50159121e-01 1.04032300e-01
1.36741900e+00 -6.36293113e-01 -2.47701228e-01 1.03549623e+00
1.03710306e+00 3.46676737e-01 6.56330884e-01 4.51526403e-01
7.83315599e-01 -3.05477120e-02 7.69787073e-01 4.00520712e-01
-3.05889577e-01 6.86430633e-01 1.08784032e+00 -1.17982477e-01
-7.04229251e-02 -2.11645633e-01 9.90292788e-01 -3.22061889e-02
2.42543668e-01 -4.30927843e-01 -4.42271680e-01 4.15390402e-01
-1.57891488e+00 -1.20522439e+00 2.46183127e-01 1.98656726e+00
7.21205413e-01 4.24096256e-01 2.60945380e-01 6.31863624e-03
8.99057806e-01 4.68688011e-01 -7.81346500e-01 8.32997337e-02
-9.73543748e-02 4.93111722e-02 9.23004508e-01 7.38565698e-02
-1.51607585e+00 1.15253294e+00 5.96259117e+00 1.01359892e+00
-1.38023388e+00 4.54776660e-02 1.47099188e-02 -3.42901260e-01
3.21819693e-01 -1.74101084e-01 -8.25449765e-01 7.95166910e-01
7.77938485e-01 -7.75254220e-02 4.90425617e-01 9.52383518e-01
2.41259881e-03 4.91066277e-01 -9.39081013e-01 9.94756699e-01
-3.27661037e-02 -1.30482268e+00 -4.41151857e-02 -1.01410083e-01
6.07098758e-01 4.38793562e-02 5.78834832e-01 4.18433726e-01
6.62158489e-01 -9.79192734e-01 9.91821051e-01 2.07596853e-01
7.92924106e-01 -7.42143929e-01 5.85956752e-01 2.65464216e-01
-1.12846327e+00 -1.60688445e-01 -4.38772112e-01 2.28820771e-01
-2.00263225e-02 3.81206931e-03 -5.39126396e-01 4.08125639e-01
1.08587182e+00 5.60387611e-01 -7.95195937e-01 1.11828458e+00
-4.45550203e-01 8.44268262e-01 -2.15603933e-01 1.71486542e-01
4.16098952e-01 6.19306505e-01 1.04125786e+00 1.04250622e+00
1.00008421e-01 -8.06797072e-02 2.71507144e-01 7.43492186e-01
-2.28991300e-01 -6.60117269e-01 -9.39749777e-01 -1.16034470e-01
5.81097603e-01 1.18495607e+00 -3.46964151e-01 -1.29858047e-01
-2.09307671e-01 7.56720424e-01 -2.95060817e-02 5.24305761e-01
-1.79572296e+00 -5.57081282e-01 1.01157010e+00 -3.51077825e-01
6.96241677e-01 -2.05430284e-01 2.30653375e-01 -1.10654616e+00
1.86251048e-02 -1.59666979e+00 2.43597448e-01 -6.50697172e-01
-1.21549451e+00 6.35844886e-01 -2.03500658e-01 -1.55529249e+00
-7.33007491e-02 -5.85361242e-01 -7.32274771e-01 4.81984526e-01
-1.27175295e+00 -9.36535716e-01 -3.46995503e-01 9.84611571e-01
3.39568704e-01 -5.80787182e-01 3.96710455e-01 3.81025672e-01
-6.47564352e-01 1.19650209e+00 -1.01302713e-01 7.28546441e-01
9.51138973e-01 -7.79010832e-01 7.78270841e-01 1.38217783e+00
2.43371949e-01 4.30473804e-01 9.11315143e-01 -6.48189545e-01
-1.43001330e+00 -1.52101398e+00 -5.85818551e-02 -5.29688299e-01
9.19722497e-01 -2.27558106e-01 -8.11232269e-01 6.84887588e-01
3.13850790e-01 5.11777163e-01 2.79552460e-01 -5.71800053e-01
-6.72634065e-01 -2.07242548e-01 -9.76819396e-01 8.06485295e-01
1.00655997e+00 -5.36193371e-01 -5.35237551e-01 3.51062506e-01
1.13861418e+00 -7.88912654e-01 -7.09712684e-01 5.58460414e-01
5.71005821e-01 -7.07889080e-01 1.37328732e+00 -1.01839530e+00
9.53122601e-02 -7.31049955e-01 4.21866355e-03 -1.19078839e+00
-1.99895844e-01 -1.35035646e+00 -4.19402033e-01 1.07178295e+00
6.54213279e-02 -5.72662234e-01 6.16059959e-01 4.59317453e-02
3.02513745e-02 -3.00536364e-01 -8.33971977e-01 -1.36077809e+00
-9.55969095e-02 -5.18663347e-01 7.39279687e-01 8.16748798e-01
-7.43003190e-01 -3.57946843e-01 -9.73918259e-01 9.61961746e-01
7.98723340e-01 -2.87936270e-01 1.32555676e+00 -7.13921428e-01
-3.48415703e-01 -2.37225309e-01 -7.38394558e-01 -9.61278498e-01
3.08845371e-01 -3.38071316e-01 1.53540984e-01 -3.33023906e-01
-2.71070480e-01 -9.02081504e-02 -3.38824153e-01 5.58189988e-01
-3.20896298e-01 5.09223580e-01 6.12306058e-01 8.56821463e-02
-6.87109053e-01 5.06452262e-01 1.10789204e+00 -4.90260988e-01
3.99949253e-01 1.94677740e-01 -4.89873677e-01 8.49001527e-01
5.89488983e-01 -9.95041132e-01 -1.89074904e-01 -5.73935986e-01
-3.98283787e-02 5.61808981e-03 9.87080276e-01 -1.35432780e+00
3.92289340e-01 -1.67700201e-01 3.55973423e-01 -5.05166292e-01
8.98783654e-02 -1.12443900e+00 8.66290554e-02 8.27660441e-01
-2.72195965e-01 1.63060233e-01 5.87919235e-01 8.78392994e-01
-8.73757377e-02 7.13950349e-03 9.88880217e-01 1.34094775e-01
-6.17693245e-01 6.52645230e-01 -2.09377721e-01 2.70702690e-01
1.25436997e+00 -4.88548279e-02 -7.58882165e-01 -4.25583303e-01
-4.57999885e-01 2.78722018e-01 4.02373284e-01 7.36765921e-01
5.52002549e-01 -1.44625545e+00 -5.07641971e-01 1.76176772e-01
-1.57896444e-01 -2.36691445e-01 1.57776877e-01 5.60170412e-01
-5.54218054e-01 2.31309980e-01 -4.02875900e-01 -6.85980678e-01
-1.36106551e+00 1.18193734e+00 5.31036973e-01 4.34213094e-02
-5.69891334e-01 9.14380431e-01 4.48437124e-01 -8.84552747e-02
8.55105698e-01 -2.22310185e-01 1.50280491e-01 -3.39545131e-01
5.99914670e-01 1.91795364e-01 -3.20583194e-01 -5.10173082e-01
-3.93726915e-01 4.55948174e-01 -4.07821313e-02 1.89059153e-01
7.44777918e-01 3.13644022e-01 3.66718620e-01 -1.83614805e-01
8.81278813e-01 4.02096994e-02 -1.81004989e+00 -2.29509190e-01
-4.02069211e-01 -7.79973388e-01 -1.69431433e-01 -6.32668853e-01
-1.53445530e+00 8.60731781e-01 7.70886064e-01 2.58758575e-01
1.14979231e+00 -3.28411013e-01 1.07471991e+00 7.34386444e-01
2.30005056e-01 -5.36818385e-01 4.42363203e-01 3.61987829e-01
8.45910311e-01 -1.23888564e+00 -1.60564899e-01 -1.00450687e-01
-5.44206262e-01 8.60774279e-01 8.89855385e-01 -3.01555574e-01
4.37271118e-01 5.28774500e-01 2.34289989e-01 1.19336762e-01
-7.20072448e-01 -4.85603921e-02 1.71547532e-01 9.34147537e-01
-4.49184448e-01 -3.83064002e-01 4.88434583e-01 3.50925624e-01
-1.64035276e-01 -2.52720833e-01 4.29377615e-01 8.94475996e-01
-2.37094983e-01 -6.63638175e-01 -8.79717171e-01 -1.33160070e-01
-8.86148870e-01 -4.13494045e-03 -3.99954081e-01 9.59077775e-01
2.04058081e-01 9.30642009e-01 -2.58803397e-01 -6.01036668e-01
2.20728591e-01 -4.54764694e-01 3.20464522e-01 -1.06642827e-01
-8.07467818e-01 -3.04464009e-02 -2.47395217e-01 -8.56983483e-01
-4.19341981e-01 -4.99669462e-01 -6.44543648e-01 -4.31058675e-01
-4.95374054e-01 -9.07071158e-02 3.34616393e-01 8.33556235e-01
2.27097422e-01 7.42559671e-01 8.67205977e-01 -9.38175976e-01
-1.05600393e+00 -6.88042521e-01 -1.20248310e-01 3.14302862e-01
7.72453308e-01 -8.69132578e-01 -5.72993577e-01 2.18086615e-01] | [5.408635139465332, 7.956803798675537] |
d04c8453-4c50-4e9d-81f7-7de843c220bb | uboco-unsupervised-boundary-contrastive-1 | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Kang_UBoCo_Unsupervised_Boundary_Contrastive_Learning_for_Generic_Event_Boundary_Detection_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Kang_UBoCo_Unsupervised_Boundary_Contrastive_Learning_for_Generic_Event_Boundary_Detection_CVPR_2022_paper.pdf | UBoCo: Unsupervised Boundary Contrastive Learning for Generic Event Boundary Detection | Generic Event Boundary Detection (GEBD) is a newly suggested video understanding task that aims to find one level deeper semantic boundaries of events. Bridging the gap between natural human perception and video understanding, it has various potential applications, including interpretable and semantically valid video parsing. Still at an early development stage, existing GEBD solvers are simple extensions of relevant video understanding tasks, disregarding GEBD's distinctive characteristics. In this paper, we propose a novel framework for unsupervised/supervised GEBD, by using the Temporal Self-similarity Matrix (TSM) as the video representation. The new Recursive TSM Parsing (RTP) algorithm exploits local diagonal patterns in TSM to detect boundaries, and it is combined with the Boundary Contrastive (BoCo) loss to train our encoder to generate more informative TSMs. Our framework can be applied to both unsupervised and supervised settings, with both achieving state-of-the-art performance by a huge margin in GEBD benchmark. Especially, our unsupervised method outperforms previous state-of-the-art "supervised" model, implying its exceptional efficacy. | ['Seon Joo Kim', 'Taehyun Kim', 'Jinwoo Kim', 'Hyolim Kang'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['boundary-detection'] | ['computer-vision'] | [ 5.48423886e-01 2.14984313e-01 -4.35906559e-01 -4.27519649e-01
-5.80299258e-01 -4.50367033e-01 5.21571040e-01 -3.52640972e-02
-9.22860652e-02 3.01305801e-01 5.85752904e-01 -1.81777835e-01
-1.81224570e-02 -4.88490015e-01 -1.04289234e+00 -5.72981060e-01
-4.58546132e-02 2.66944766e-01 5.19572616e-01 -5.51532730e-02
2.19279453e-01 -2.80833244e-02 -1.60560489e+00 6.00696743e-01
7.64966547e-01 1.07591844e+00 2.92920887e-01 4.35730726e-01
-1.22273855e-01 1.13636005e+00 -2.66195446e-01 -4.76305187e-01
9.97223482e-02 -6.52463019e-01 -1.21942687e+00 3.65383714e-01
3.85422438e-01 -2.70562321e-01 -4.60375994e-01 1.00278938e+00
1.03800207e-01 2.14017019e-01 3.47322077e-01 -1.28623044e+00
-4.97359157e-01 6.54024839e-01 -5.97075999e-01 5.11868596e-01
6.47745907e-01 -8.40288475e-02 1.26304793e+00 -7.99575627e-01
8.67966056e-01 1.39605629e+00 4.94022876e-01 6.27389967e-01
-9.07160819e-01 -4.07159775e-01 6.15178287e-01 7.16112912e-01
-1.08621275e+00 -2.19528273e-01 8.77612829e-01 -5.03674150e-01
1.03191388e+00 1.80936411e-01 5.81792176e-01 1.41239893e+00
-2.63833143e-02 1.26218963e+00 7.22755909e-01 -3.75170678e-01
2.28383526e-01 -2.05180719e-01 3.42484355e-01 7.71313965e-01
-1.23179823e-01 -4.08000574e-02 -8.54737341e-01 2.35082746e-01
5.85688531e-01 -2.12105975e-01 -4.27231431e-01 -4.60242301e-01
-1.39030099e+00 7.85292625e-01 1.46638066e-01 9.27123278e-02
-6.49117455e-02 1.26249120e-01 8.30411613e-01 1.39337197e-01
5.26428819e-01 6.89357147e-03 -3.10804754e-01 -3.83710504e-01
-7.50146389e-01 6.13423390e-03 6.24435425e-01 1.02093911e+00
5.70051551e-01 -2.65102416e-01 -2.28478968e-01 5.78002334e-01
2.69022703e-01 -5.56707755e-02 3.86186570e-01 -1.10476458e+00
5.50176024e-01 6.38335049e-01 -6.22149259e-02 -1.03367734e+00
-3.95636916e-01 3.14637125e-02 -7.08307743e-01 -3.33223701e-01
3.66793841e-01 1.71368290e-02 -8.36872637e-01 1.86281204e+00
4.59429651e-01 8.49042773e-01 1.33643851e-01 1.04479980e+00
9.74488914e-01 9.42794085e-01 2.55788177e-01 -3.35924834e-01
1.42863524e+00 -1.13114476e+00 -7.27941155e-01 -4.70160782e-01
5.82988739e-01 -4.17255640e-01 1.02305448e+00 4.83798891e-01
-9.83492494e-01 -6.16151035e-01 -1.02995408e+00 -2.86297143e-01
-6.82635233e-02 9.98333991e-02 8.74535620e-01 4.30806428e-01
-9.96011913e-01 5.10433197e-01 -8.74870121e-01 -5.22093952e-01
4.30665970e-01 1.25431910e-01 -4.27721679e-01 -1.50280967e-01
-1.27621210e+00 4.32947129e-01 8.18485975e-01 3.13344419e-01
-1.04259837e+00 -6.21036768e-01 -1.20244038e+00 1.82626955e-02
9.84478414e-01 -8.69552553e-01 1.11686623e+00 -1.21168590e+00
-1.30156827e+00 1.02590787e+00 -4.25590932e-01 -7.01690733e-01
4.95736331e-01 -4.84924942e-01 -3.12459469e-01 5.13402522e-01
2.80137032e-01 9.22829390e-01 9.54734683e-01 -1.07265460e+00
-7.16906607e-01 -2.11161926e-01 3.04145459e-02 3.28230977e-01
-4.55400981e-02 3.03102523e-01 -7.71958351e-01 -7.50563145e-01
2.89302975e-01 -6.55343711e-01 -8.95327255e-02 -3.11682165e-01
-4.19589579e-01 -3.20293158e-01 1.00561476e+00 -8.69666874e-01
1.40019619e+00 -2.16985273e+00 5.69270849e-01 -1.93796024e-01
1.01544999e-01 1.22268334e-01 -3.01019028e-02 3.39345008e-01
-3.31103206e-01 1.15823597e-01 -2.81469703e-01 -3.92323285e-01
-7.90368319e-02 2.42054671e-01 -3.43090355e-01 2.39313364e-01
5.42622685e-01 8.84830534e-01 -1.09903002e+00 -5.34603536e-01
2.03528672e-01 1.48338497e-01 -5.84946930e-01 1.49860740e-01
-4.45404321e-01 6.99597239e-01 -3.94165784e-01 5.15419304e-01
5.53604305e-01 -3.41710567e-01 4.50399578e-01 -2.08078638e-01
3.44978236e-02 2.82498866e-01 -1.18467665e+00 1.86206245e+00
-1.54489931e-02 7.34982848e-01 -2.53387213e-01 -1.54087901e+00
7.29949951e-01 2.75664270e-01 4.52985406e-01 -5.55218220e-01
-2.80832890e-02 -1.22273751e-01 -2.52399951e-01 -8.37497354e-01
2.91329563e-01 1.45853329e-02 -1.76203668e-01 8.55588466e-02
1.07698068e-01 5.31627715e-01 3.75121444e-01 2.78531283e-01
1.03654325e+00 4.85833108e-01 4.06177908e-01 -2.55500972e-01
6.83781385e-01 -8.35781544e-02 9.33614492e-01 6.73943043e-01
-2.41134614e-01 6.43712938e-01 9.14298773e-01 -5.81946850e-01
-8.02475452e-01 -1.03699839e+00 2.21395805e-01 1.04163349e+00
6.90410793e-01 -6.16733193e-01 -1.04952395e+00 -8.99102509e-01
-4.13135588e-01 4.70080167e-01 -5.24852693e-01 -1.74328074e-01
-7.84174621e-01 -7.05865622e-01 4.52935278e-01 8.58633339e-01
8.37418079e-01 -1.16533697e+00 -4.38191712e-01 3.24594796e-01
-7.01219380e-01 -1.68232274e+00 -4.23935086e-01 -7.95172900e-02
-9.05893028e-01 -1.11035168e+00 -3.56679857e-01 -1.01703966e+00
4.19258505e-01 5.56834161e-01 1.28172207e+00 -2.59729736e-02
-1.42519772e-01 4.81697977e-01 -7.99187839e-01 -6.10061064e-02
-4.12176281e-01 -2.34523684e-01 -2.12676018e-01 2.19873771e-01
5.76523602e-01 -3.77232879e-01 -6.68494582e-01 5.43677092e-01
-9.34560716e-01 5.45507729e-01 1.78679377e-01 8.35635483e-01
6.59882545e-01 1.39628008e-01 6.88837409e-01 -8.97104383e-01
2.14569166e-01 -5.71909189e-01 -3.96877855e-01 3.76658648e-01
-2.08389208e-01 1.51368100e-02 6.02286100e-01 -2.53985167e-01
-1.48747087e+00 -7.55552053e-02 -1.23148464e-01 -4.41071659e-01
-3.11592132e-01 4.07086879e-01 -5.58372617e-01 3.35716724e-01
8.92906711e-02 2.67771304e-01 -3.85769665e-01 -5.82173705e-01
3.29112619e-01 3.16495240e-01 6.80524170e-01 -6.50645673e-01
3.25148791e-01 6.47167087e-01 -1.25883386e-01 -7.95558989e-01
-1.13976872e+00 -7.29856610e-01 -6.77946508e-01 -3.06134701e-01
1.30406392e+00 -1.05149662e+00 -6.28067374e-01 5.91574609e-01
-1.25105560e+00 -2.31346473e-01 -1.96568901e-03 3.91124457e-01
-7.06538260e-01 8.69482160e-01 -7.00751901e-01 -5.77640772e-01
-1.54831663e-01 -1.17275929e+00 1.31448925e+00 2.09867746e-01
-1.31898686e-01 -9.44907129e-01 -1.87282935e-01 8.76812577e-01
-4.60696816e-01 2.49073803e-01 9.63600099e-01 -6.77935004e-01
-7.79327214e-01 3.08054835e-01 -3.78351539e-01 3.79186511e-01
-1.74666837e-01 -6.92282915e-02 -8.47585261e-01 -4.21957858e-02
2.08863214e-01 -1.56836882e-01 1.16649592e+00 5.63251495e-01
1.32674742e+00 -2.49483377e-01 -3.29564571e-01 6.99425578e-01
1.22959876e+00 4.65282053e-01 8.52204561e-01 5.03036976e-01
8.13218892e-01 7.01821566e-01 9.07991707e-01 5.73187768e-01
4.83381957e-01 7.10465610e-01 5.51867068e-01 1.07112467e-01
-8.70765895e-02 -4.22810197e-01 7.59073138e-01 7.53353059e-01
-1.39615238e-01 -4.85268086e-01 -8.12582552e-01 5.24391532e-01
-2.25899529e+00 -1.12696552e+00 -1.50689960e-01 1.97082436e+00
4.80595112e-01 3.88727069e-01 6.32528514e-02 2.52342876e-02
9.78700936e-01 4.91966486e-01 -5.11533856e-01 -3.67814690e-01
-1.62202775e-01 4.68291305e-02 8.65005180e-02 1.48194417e-01
-1.57623541e+00 1.18537343e+00 5.58187056e+00 8.74199510e-01
-6.72476292e-01 8.13868567e-02 7.68692791e-01 3.41008514e-01
-1.17117710e-01 2.74869144e-01 -8.57976675e-01 5.06810546e-01
8.09187353e-01 2.05909759e-01 2.62096912e-01 7.10807562e-01
5.41593969e-01 -2.51255870e-01 -1.29202247e+00 1.08981371e+00
2.67895162e-01 -1.46827304e+00 1.68595776e-01 -4.42151576e-01
5.55248737e-01 -2.23758921e-01 -2.98660964e-01 2.06056386e-01
-2.70168662e-01 -7.77951479e-01 8.42486620e-01 1.57022431e-01
4.55569386e-01 -5.76055825e-01 6.22806907e-01 3.91359657e-01
-1.56295455e+00 -1.26355007e-01 -3.21601570e-01 -5.02886586e-02
3.96445930e-01 3.37594002e-01 -5.04772246e-01 7.74569571e-01
9.17785943e-01 1.22029793e+00 -4.45299387e-01 7.88835466e-01
-3.06861490e-01 7.47266114e-01 -7.04251081e-02 3.23944360e-01
4.86755013e-01 -4.00522530e-01 7.25829780e-01 1.32510924e+00
1.73771992e-01 1.12771861e-01 2.57096618e-01 6.92680895e-01
-8.47017244e-02 -2.11384501e-02 -4.39804435e-01 5.17010987e-02
2.63200432e-01 8.62357318e-01 -1.07049620e+00 -4.83738124e-01
-6.07510328e-01 1.27367139e+00 2.10685849e-01 2.81421244e-01
-1.15374172e+00 5.82501665e-02 5.80082536e-01 -6.94623813e-02
6.11645103e-01 -8.25355127e-02 -7.77984411e-02 -1.50576437e+00
1.67324573e-01 -9.89145160e-01 7.99162686e-01 -7.74484217e-01
-1.10315120e+00 3.63209099e-01 2.67225385e-01 -1.28320348e+00
-1.97482809e-01 -6.05292022e-01 -6.82349145e-01 4.36789310e-03
-1.40528667e+00 -1.09647155e+00 -4.88326460e-01 4.81324613e-01
1.24765468e+00 1.99529633e-01 3.40021402e-01 3.75908047e-01
-1.00136364e+00 3.22331667e-01 -3.37148197e-02 2.48450398e-01
4.76032108e-01 -1.15014660e+00 5.95176041e-01 1.27578139e+00
2.16152817e-01 2.48796806e-01 6.48931742e-01 -7.04566181e-01
-1.24307013e+00 -1.10691106e+00 6.58986509e-01 -2.63731718e-01
6.17461562e-01 -4.62399215e-01 -9.55476761e-01 1.04818487e+00
1.41175017e-01 -1.55800238e-01 5.01124799e-01 -1.16840027e-01
-4.41090643e-01 1.64905429e-01 -6.84987187e-01 7.05237567e-01
1.66554618e+00 -3.15790713e-01 -7.76234150e-01 2.94283569e-01
1.04054737e+00 -4.11995947e-01 -7.70483077e-01 5.18036485e-01
2.50232041e-01 -1.24313068e+00 1.07670176e+00 -7.37495899e-01
6.95867419e-01 -2.46762365e-01 3.87803428e-02 -7.85882711e-01
-1.47761509e-01 -9.97488797e-01 -1.73666626e-01 1.52768755e+00
1.71832368e-02 -2.60771900e-01 7.80892193e-01 3.47189963e-01
-4.20595437e-01 -5.89110136e-01 -9.95820999e-01 -8.17255497e-01
-3.82368594e-01 -8.03603530e-01 2.57989019e-01 7.29636788e-01
2.15451956e-01 3.93785357e-01 -6.25682890e-01 2.41396964e-01
5.58479488e-01 1.96576461e-01 5.66682816e-01 -9.61678088e-01
-2.94256777e-01 -3.18993419e-01 -5.85332811e-01 -1.69180739e+00
3.74146760e-01 -8.04950774e-01 7.84209892e-02 -1.40614891e+00
5.02556980e-01 1.52006168e-02 -9.83764827e-02 2.57151634e-01
-2.19938859e-01 -1.47538101e-02 2.22707301e-01 1.86318215e-02
-1.13735080e+00 5.46826720e-01 1.05094945e+00 -1.03101142e-01
-1.44496128e-01 -2.38642871e-01 -6.23156667e-01 1.15221477e+00
5.37928641e-01 -2.38325238e-01 -6.14315629e-01 -4.05008376e-01
1.73045650e-01 1.48515955e-01 5.33664703e-01 -8.83728981e-01
3.51765044e-02 -1.53219044e-01 1.49350688e-02 -5.88718057e-01
2.13797972e-01 -5.35061121e-01 1.74491227e-01 1.46399155e-01
-2.45094433e-01 -1.14000663e-01 1.10181972e-01 8.80273938e-01
-6.10055268e-01 -4.20573115e-01 4.59236532e-01 -1.32866204e-01
-1.64650428e+00 2.25401998e-01 -3.55959922e-01 4.09821689e-01
1.15107751e+00 -5.90731263e-01 -4.00880694e-01 -3.00934464e-01
-7.31277943e-01 2.66750246e-01 2.72245556e-01 5.45923471e-01
8.13052058e-01 -1.07444429e+00 -4.68874842e-01 4.50262502e-02
1.87419236e-01 1.97406918e-01 5.75814784e-01 9.47142780e-01
-5.32841265e-01 2.99673706e-01 -4.46800748e-03 -7.90081441e-01
-1.21567094e+00 7.73277164e-01 8.00504908e-02 -2.80462116e-01
-1.12271774e+00 8.07053387e-01 8.73075724e-01 1.74426749e-01
4.03793335e-01 -4.26222086e-01 -3.10117632e-01 6.92887306e-02
5.99488258e-01 4.06941146e-01 -1.39670819e-01 -6.61361337e-01
-3.98636281e-01 6.84642851e-01 -6.41565174e-02 3.32094461e-01
1.22275233e+00 -3.88241172e-01 9.11928341e-02 2.55566716e-01
1.16803479e+00 -4.32570785e-01 -1.54813218e+00 -1.65742010e-01
4.33262199e-01 -2.97517598e-01 -2.43598506e-01 -2.82726139e-01
-1.02365255e+00 8.99248898e-01 1.79912373e-01 1.27761006e-01
1.49224246e+00 3.77077967e-01 1.12421715e+00 1.25862062e-01
3.08584183e-01 -1.03162777e+00 3.70645374e-01 4.23475772e-01
5.57893693e-01 -1.52639806e+00 -1.86213180e-01 -1.00148499e+00
-9.72613573e-01 1.13968730e+00 5.90489089e-01 -9.43463203e-03
3.45052034e-01 -5.65185025e-02 -4.10818189e-01 -2.71939695e-01
-7.19960749e-01 -2.48832077e-01 3.64914685e-01 4.82862622e-01
2.00862810e-01 -2.22952634e-01 -4.40255880e-01 8.68023515e-01
3.42143744e-01 1.92274414e-02 4.20152038e-01 7.30011940e-01
-3.71092737e-01 -9.58681822e-01 -9.56247523e-02 1.54590175e-01
-5.49249351e-01 -1.24603637e-01 -3.33306253e-01 1.01710367e+00
1.50193915e-01 9.23715591e-01 1.17539741e-01 -3.52011412e-01
1.89994201e-01 5.79111651e-02 3.75108749e-01 -4.69536901e-01
-3.19439352e-01 2.21399188e-01 2.75596201e-01 -8.83162916e-01
-8.98960590e-01 -7.96339989e-01 -1.44417977e+00 7.70500675e-03
-2.31388554e-01 -1.46551087e-01 3.61621380e-02 1.18457055e+00
4.01733488e-01 3.72177035e-01 3.82506460e-01 -6.68929517e-01
-1.01627961e-01 -5.44304609e-01 -3.50122452e-01 8.41566801e-01
-6.09039553e-02 -8.66476357e-01 -2.67118156e-01 5.17536640e-01] | [9.414580345153809, 0.6215787529945374] |
5d89812b-c276-488c-989f-0007a5c9d509 | digital-image-forensics-using-deep-learning | 2210.09052 | null | https://arxiv.org/abs/2210.09052v1 | https://arxiv.org/pdf/2210.09052v1.pdf | Digital Image Forensics using Deep Learning | During the investigation of criminal activity when evidence is available, the issue at hand is determining the credibility of the video and ascertaining that the video is real. Today, one way to authenticate the footage is to identify the camera that was used to capture the image or video in question. While a very common way to do this is by using image meta-data, this data can easily be falsified by changing the video content or even splicing together content from two different cameras. Given the multitude of solutions proposed to this problem, it is yet to be sufficiently solved. The aim of our project is to build an algorithm that identifies which camera was used to capture an image using traces of information left intrinsically in the image, using filters, followed by a deep neural network on these filters. Solving this problem would have a big impact on the verification of evidence used in criminal and civil trials and even news reporting. | ['Bishesh Sinha', 'Yash Mathur', 'Vivek Kapoor', 'Mukund Sood', 'Akash Nagaraj'] | 2022-10-14 | null | null | null | null | ['image-forensics'] | ['computer-vision'] | [ 4.90310282e-01 -7.19576031e-02 1.16778493e-01 -2.44902790e-01
-6.20877922e-01 -8.73401225e-01 7.12753654e-01 2.46844321e-01
-4.80273008e-01 6.88109517e-01 7.79265761e-02 -5.69823146e-01
1.06497519e-01 -7.12472737e-01 -6.61717653e-01 -5.34188688e-01
3.85454267e-01 2.19735757e-01 3.61095577e-01 1.41915813e-01
8.20717573e-01 6.49192095e-01 -1.14949858e+00 6.93590403e-01
1.04517996e-01 8.53015184e-01 -3.93373752e-03 9.01134789e-01
-8.51893499e-02 1.02688897e+00 -7.93612480e-01 -9.30871189e-01
3.65132153e-01 -4.38957304e-01 -9.99155104e-01 1.69958979e-01
4.06582594e-01 -7.09129512e-01 -2.98353344e-01 1.28532255e+00
2.13301316e-01 -4.21424240e-01 4.08236861e-01 -1.03632545e+00
-3.15769017e-01 5.12949049e-01 -4.80024338e-01 7.40175366e-01
7.59098351e-01 1.47731259e-01 3.92784655e-01 -4.61141706e-01
8.60094488e-01 8.71340871e-01 6.51293337e-01 2.50383466e-01
-8.19910288e-01 -6.51391685e-01 -5.42883635e-01 4.51398164e-01
-1.23386025e+00 -6.13298357e-01 4.92047340e-01 -7.82483935e-01
4.15158302e-01 1.52922457e-03 6.61966383e-01 1.29078150e+00
2.81852335e-01 4.70622092e-01 1.17259872e+00 -6.66283429e-01
-2.69481428e-02 2.54010320e-01 1.28408998e-01 4.13978457e-01
5.36432743e-01 -1.32134100e-02 -3.91180247e-01 -2.55035102e-01
5.30928075e-01 -7.59997964e-02 -3.20863456e-01 2.53156245e-01
-7.65446007e-01 7.60887325e-01 -1.63729593e-01 7.53635943e-01
-4.31852579e-01 6.92658201e-02 5.82094908e-01 4.59972680e-01
1.81646913e-01 5.31290591e-01 9.06950086e-02 -4.05394167e-01
-1.56697428e+00 2.87449658e-01 6.37507558e-01 2.05446064e-01
5.51512480e-01 -3.68560046e-01 1.56780526e-01 7.35251233e-03
3.30227613e-01 1.33629888e-01 3.19560617e-01 -8.05731356e-01
7.00368166e-01 6.55047536e-01 1.38320565e-01 -1.47813213e+00
-1.49490476e-01 1.37060925e-01 -3.02456349e-01 5.32844663e-01
9.99451876e-01 -1.81457058e-01 -6.29722357e-01 1.10234964e+00
2.25916401e-01 3.78067762e-01 -2.40055516e-01 1.07760322e+00
4.46493387e-01 5.64269006e-01 -2.59465665e-01 2.38609426e-02
1.51932645e+00 -5.85188977e-02 -7.70896733e-01 -2.79037394e-02
2.29357779e-01 -8.95841002e-01 1.93575576e-01 7.73309231e-01
-8.60065818e-01 -4.00523454e-01 -1.34004009e+00 3.51560931e-03
-3.11901897e-01 -1.94340006e-01 1.26582786e-01 1.03992879e+00
-6.05507970e-01 6.05404437e-01 -4.71406311e-01 -2.66344130e-01
3.90468240e-01 2.65644908e-01 -7.52232015e-01 -3.17917079e-01
-1.33999598e+00 1.07266521e+00 6.31880462e-01 3.75923485e-01
-7.12107480e-01 -2.54307866e-01 -3.74050617e-01 5.80781475e-02
4.13300008e-01 -1.65908590e-01 8.90818596e-01 -1.35713565e+00
-8.49636734e-01 1.20794106e+00 1.35637835e-01 -7.04644442e-01
9.79949772e-01 1.49821453e-02 -6.75562918e-01 5.61821342e-01
2.61679471e-01 6.80835843e-02 1.36438501e+00 -9.74856913e-01
-8.86518478e-01 -4.84854937e-01 2.04923704e-01 -4.28746462e-01
-1.97275385e-01 4.93995041e-01 -3.39510053e-01 -3.31749290e-01
-6.90748394e-02 -9.13458586e-01 3.28442484e-01 -2.91325688e-01
-2.91892529e-01 2.39090234e-01 1.18613768e+00 -1.29233134e+00
1.28516209e+00 -2.11357474e+00 -3.83766800e-01 5.42725623e-01
1.79170534e-01 6.61067665e-01 2.20254779e-01 5.10283530e-01
-4.49488163e-02 2.33637407e-01 -7.93060064e-02 1.57189190e-01
-4.06567186e-01 -2.30268940e-01 -5.63561201e-01 9.13927317e-01
1.69773936e-01 5.58728755e-01 -8.87432635e-01 -5.89193940e-01
2.30354980e-01 4.70284522e-01 -1.10996410e-01 -2.03538835e-02
1.62387595e-01 4.51109290e-01 -4.91629481e-01 3.37895364e-01
6.69583976e-01 7.91807473e-02 3.33959162e-01 -8.27462226e-02
-2.99166590e-01 2.74319500e-01 -1.57549012e+00 1.23631048e+00
8.21216032e-02 1.18655384e+00 1.85479447e-01 -1.10684919e+00
6.78434253e-01 7.58149624e-01 4.06595618e-01 -5.47226131e-01
3.93431187e-01 3.17450255e-01 -2.53649712e-01 -9.07930493e-01
7.33293533e-01 -2.69995511e-01 -5.93497939e-02 7.69582808e-01
-2.37307474e-01 2.71801561e-01 2.02801943e-01 3.47172879e-02
1.34716654e+00 1.97297726e-02 1.94082588e-01 4.91838723e-01
7.35550165e-01 1.24713108e-01 2.42616728e-01 6.54203236e-01
-2.45336756e-01 7.32616901e-01 6.19938850e-01 -7.36968160e-01
-1.22071815e+00 -4.51567799e-01 1.79341864e-02 2.19317719e-01
-8.10554922e-02 -7.43765160e-02 -9.43332195e-01 -6.64601386e-01
-2.24432185e-01 4.00346935e-01 -5.27876973e-01 -2.19782982e-02
-5.66737831e-01 -2.00828463e-01 9.26133275e-01 -1.11924611e-01
6.18064106e-01 -9.48361576e-01 -1.07468021e+00 4.42059219e-01
-3.72574061e-01 -1.20812893e+00 -1.37744293e-01 -1.98203459e-01
-4.61655527e-01 -1.49023557e+00 -2.98044682e-01 3.08484334e-04
4.58296657e-01 4.01966602e-01 4.88483369e-01 4.54146087e-01
-3.63419622e-01 1.24860577e-01 -5.92357635e-01 -2.94661462e-01
-7.39539325e-01 -1.71056047e-01 -3.09187561e-01 4.33078289e-01
4.86994833e-01 -1.47252589e-01 -2.40391806e-01 -9.78412107e-02
-1.47239506e+00 -4.53462124e-01 2.49633640e-01 2.89031565e-01
-1.28442630e-01 5.50756872e-01 7.99292251e-02 -9.20881927e-01
7.20753908e-01 -5.41369081e-01 -6.69562101e-01 2.36116484e-01
-2.06788778e-01 -1.79780096e-01 3.74947309e-01 -1.36068389e-01
-8.46115530e-01 6.66206926e-02 -1.38230950e-01 -4.12326097e-01
-3.50421131e-01 3.88658285e-01 -2.60789115e-02 8.16823170e-02
3.39863896e-01 -5.83936721e-02 -5.65221719e-02 -1.70068786e-01
-1.82243437e-02 9.99348283e-01 7.13363171e-01 -1.37083560e-01
7.20803201e-01 6.40583038e-01 -1.33506373e-01 -8.55341077e-01
-5.88509262e-01 -6.15284860e-01 -7.20933616e-01 -6.01396680e-01
1.13856936e+00 -5.22697926e-01 -7.25227237e-01 4.24255043e-01
-1.60853314e+00 5.05339980e-01 2.89731234e-01 3.49032819e-01
-1.38415188e-01 7.76490271e-01 -2.35800907e-01 -1.01832092e+00
7.26674944e-02 -1.13737464e+00 6.68405235e-01 -3.97570133e-02
-4.98190284e-01 -6.56647384e-01 1.53891355e-01 9.88204718e-01
6.92522675e-02 3.83068174e-01 4.44583505e-01 -4.76581544e-01
-5.86119592e-01 -8.80772412e-01 -2.60665298e-01 3.40407073e-01
-1.89898908e-02 3.64115655e-01 -1.03633142e+00 -1.56916410e-01
4.26718920e-01 -6.14991151e-02 5.44919848e-01 2.77954519e-01
6.85720205e-01 -3.29693645e-01 -1.73045158e-01 2.44119033e-01
1.48687875e+00 3.97093922e-01 1.33248878e+00 7.66167462e-01
2.98130244e-01 7.35158801e-01 2.73983181e-01 3.97932798e-01
-5.94203621e-02 7.59558856e-01 5.05741239e-01 2.93777704e-01
8.76141116e-02 -1.97394371e-01 5.50672472e-01 1.12046957e-01
-1.87113136e-01 -4.42303419e-01 -8.22770417e-01 4.42450553e-01
-1.72308612e+00 -1.94548297e+00 -5.17962515e-01 2.13098097e+00
1.11485779e-01 3.75536531e-01 1.88644767e-01 6.60714507e-01
1.11356044e+00 -1.10678509e-01 2.10006386e-02 -5.95884919e-01
-6.19552992e-02 -1.16697498e-01 4.88345712e-01 3.21961939e-01
-1.16017115e+00 4.55678344e-01 6.34329891e+00 5.25980651e-01
-1.38514328e+00 2.47703433e-01 3.82278681e-01 -9.20387451e-03
1.25538349e-01 2.42452189e-01 -6.26246452e-01 1.03542328e+00
9.91772950e-01 2.47534290e-01 2.12225154e-01 3.91160756e-01
5.87228656e-01 -5.99680066e-01 -8.15642655e-01 8.99821103e-01
4.04799193e-01 -1.61898422e+00 -2.07713246e-01 3.13924372e-01
3.50462288e-01 -4.51298535e-01 -1.29898652e-01 -3.90165061e-01
-3.59481946e-02 -1.05348277e+00 1.16982400e+00 5.99355757e-01
6.62484169e-01 -5.75282991e-01 8.82110119e-01 3.98406386e-01
-6.12146795e-01 -1.47818640e-01 -1.74635127e-01 -1.67671010e-01
2.68691778e-01 5.66109478e-01 -1.00082862e+00 3.47616434e-01
5.02784371e-01 2.69401968e-01 -5.19116282e-01 1.18588960e+00
-4.39718515e-01 7.26103544e-01 -1.46027267e-01 2.43812934e-01
3.73921454e-01 3.62429477e-04 6.19113445e-01 1.21840298e+00
4.62564975e-01 -1.76026672e-02 -3.12209874e-01 8.02851617e-01
8.13336447e-02 -2.85100609e-01 -8.30045938e-01 -1.86566636e-01
2.87309319e-01 1.13890588e+00 -9.63919222e-01 -1.52572170e-01
-4.09928858e-01 1.02981317e+00 -1.94842760e-02 -1.24862283e-01
-7.50759125e-01 -1.61112204e-01 1.45889476e-01 7.21380830e-01
4.73577738e-01 -2.15231493e-01 -8.75241123e-03 -1.01603651e+00
2.22747713e-01 -1.03003299e+00 4.18502182e-01 -1.00417137e+00
-8.21650624e-01 4.63173151e-01 -1.32714450e-01 -1.25896013e+00
-1.64092466e-01 -6.34014010e-01 -5.73914230e-01 7.29530215e-01
-1.14731431e+00 -9.79545534e-01 1.14564680e-01 4.11589503e-01
1.40963316e-01 -2.49062315e-01 2.56471723e-01 5.98770916e-01
-2.43682444e-01 -5.43085067e-03 -1.52601972e-01 5.30240238e-01
5.73544264e-01 -6.30507052e-01 9.93169695e-02 1.37341464e+00
4.66068774e-01 4.23104793e-01 8.49760473e-01 -8.73917043e-01
-1.38905990e+00 -4.35679168e-01 1.10355711e+00 -6.48339510e-01
9.24057841e-01 8.20667222e-02 -9.00068104e-01 3.49512279e-01
1.48752958e-01 -4.44309920e-01 4.78355825e-01 -3.92141879e-01
-3.46717566e-01 1.53083280e-01 -1.18621039e+00 4.69306000e-02
2.90690571e-01 -8.17720830e-01 -9.45948720e-01 3.13609913e-02
-1.23439290e-01 -1.89718634e-01 -3.37030202e-01 -3.10685843e-01
7.41217971e-01 -1.19786751e+00 5.88173509e-01 -4.89601433e-01
6.07320487e-01 -5.39630592e-01 3.72844823e-02 -7.70842671e-01
3.61993350e-02 -4.94496733e-01 4.31927174e-01 1.39620304e+00
2.90242285e-01 -3.30371261e-01 9.11656559e-01 7.44728386e-01
2.05346242e-01 2.26281211e-01 -1.38272834e+00 -4.18342471e-01
-4.00149852e-01 -8.99427414e-01 5.52921295e-01 1.05644739e+00
6.72514737e-02 1.05918460e-02 -6.74279451e-01 2.15267703e-01
3.64215046e-01 -1.11121424e-01 7.50923693e-01 -1.15212619e+00
-1.15217365e-01 -2.59450436e-01 -8.16547155e-01 -2.63704598e-01
-4.10838462e-02 -7.43054688e-01 -3.21659803e-01 -1.33814800e+00
1.50846049e-01 -5.46840914e-02 1.34890661e-01 1.72758400e-01
8.26428756e-02 4.35214430e-01 4.79465246e-01 3.96495521e-01
-2.66336560e-01 -5.06764054e-01 7.48159111e-01 -1.75413266e-01
3.30632746e-01 1.75911225e-02 -5.23501575e-01 7.60467052e-01
6.06374443e-01 -8.38269770e-01 8.26268196e-02 -4.01384532e-01
7.00320363e-01 3.06235462e-01 7.21054494e-01 -1.21440530e+00
4.97340679e-01 -1.18466662e-02 5.52983463e-01 -6.05439126e-01
8.58420581e-02 -1.26192701e+00 6.94969773e-01 4.65810478e-01
-1.92314312e-01 2.66871423e-01 -1.10511996e-01 6.76157355e-01
-3.40162843e-01 -1.04273057e+00 6.71965122e-01 -3.84881020e-01
-7.78384447e-01 -2.09341660e-01 -8.37423444e-01 -1.95436552e-01
1.19238937e+00 -7.70084500e-01 -2.90775299e-01 -3.67658108e-01
-5.40840030e-01 -2.56093711e-01 5.04204690e-01 1.98923126e-01
6.69993103e-01 -9.69535887e-01 -6.83918715e-01 1.87209882e-02
-1.53119743e-01 -6.02315784e-01 1.00758322e-01 5.90597451e-01
-9.07433152e-01 2.59334713e-01 -4.24492002e-01 -2.70478278e-01
-1.60345030e+00 6.12992167e-01 1.88412845e-01 -1.87615991e-01
-5.25754929e-01 3.37464362e-01 -7.63999581e-01 3.04012328e-01
-1.86229706e-01 7.23094866e-02 -3.38032484e-01 5.07137537e-01
1.04390728e+00 4.74923193e-01 2.65909135e-01 -1.02144790e+00
-4.00313258e-01 3.61631453e-01 6.41517108e-03 -3.32463592e-01
1.43657184e+00 -2.37417042e-01 -2.67682135e-01 1.89042896e-01
1.19013333e+00 2.20802739e-01 -9.72368598e-01 2.99918413e-01
8.17673281e-02 -7.94996262e-01 -1.56267844e-02 -4.94594842e-01
-1.04693735e+00 9.30083215e-01 4.60564941e-01 7.90034235e-01
7.24951565e-01 -3.00597191e-01 6.94850206e-01 1.83574766e-01
4.51620132e-01 -1.44730234e+00 -9.65268910e-02 2.41465181e-01
6.29178882e-01 -1.23941863e+00 4.49118853e-01 1.01054996e-01
-4.92885411e-01 1.65281415e+00 -6.84886202e-02 -2.70534009e-02
4.52452570e-01 3.89977872e-01 -4.51738648e-02 -4.34550047e-01
-1.26900911e-01 1.01562269e-01 5.15318401e-02 4.23284143e-01
2.74333805e-01 -1.67737409e-01 -5.01744449e-01 6.23953678e-02
-4.47466373e-02 4.88458365e-01 1.05705225e+00 9.22717631e-01
-5.77673316e-01 -1.15644395e+00 -1.10319269e+00 3.80999207e-01
-1.15216184e+00 2.56275773e-01 -7.15380669e-01 4.61960584e-01
5.61902344e-01 1.30900180e+00 -1.78089708e-01 -2.32187137e-01
3.08036357e-02 4.62105013e-02 3.92206073e-01 -3.25739205e-01
-8.81323576e-01 -2.18954206e-01 2.87593633e-01 -2.93186367e-01
-6.99576795e-01 -9.13360834e-01 -9.06243801e-01 -5.37756860e-01
-3.83732945e-01 1.38452008e-01 9.54685926e-01 1.21478903e+00
-1.11806460e-01 5.56929857e-02 4.28409815e-01 -6.56317294e-01
-7.22121969e-02 -4.76954430e-01 -4.59411949e-01 6.37195885e-01
4.16250795e-01 -5.02254963e-01 -3.37450922e-01 5.21202981e-01] | [12.487969398498535, 1.084869623184204] |
a3d70467-c161-4369-8487-86012a1abc12 | detecting-curve-text-in-the-wild-new-dataset | 1712.02170 | null | http://arxiv.org/abs/1712.02170v1 | http://arxiv.org/pdf/1712.02170v1.pdf | Detecting Curve Text in the Wild: New Dataset and New Solution | Scene text detection has been made great progress in recent years. The
detection manners are evolving from axis-aligned rectangle to rotated rectangle
and further to quadrangle. However, current datasets contain very little curve
text, which can be widely observed in scene images such as signboard, product
name and so on. To raise the concerns of reading curve text in the wild, in
this paper, we construct a curve text dataset named CTW1500, which includes
over 10k text annotations in 1,500 images (1000 for training and 500 for
testing). Based on this dataset, we pioneering propose a polygon based curve
text detector (CTD) which can directly detect curve text without empirical
combination. Moreover, by seamlessly integrating the recurrent transverse and
longitudinal offset connection (TLOC), the proposed method can be end-to-end
trainable to learn the inherent connection among the position offsets. This
allows the CTD to explore context information instead of predicting points
independently, resulting in more smooth and accurate detection. We also propose
two simple but effective post-processing methods named non-polygon suppress
(NPS) and polygonal non-maximum suppression (PNMS) to further improve the
detection accuracy. Furthermore, the proposed approach in this paper is
designed in an universal manner, which can also be trained with rectangular or
quadrilateral bounding boxes without extra efforts. Experimental results on
CTW-1500 demonstrate our method with only a light backbone can outperform
state-of-the-art methods with a large margin. By evaluating only in the curve
or non-curve subset, the CTD + TLOC can still achieve the best results. Code is
available at https://github.com/Yuliang-Liu/Curve-Text-Detector. | ['Zhang Shuaitao', 'Jin Lianwen', 'Zhang Sheng', 'Liu Yuliang'] | 2017-12-06 | null | null | null | null | ['curved-text-detection'] | ['computer-vision'] | [ 2.73786604e-01 -2.88336188e-01 -7.23024160e-02 -2.04116609e-02
-7.69187212e-01 -6.38041496e-01 4.76023823e-01 2.29917299e-02
-2.11436689e-01 -9.22428966e-02 -1.87202886e-01 -3.36109430e-01
1.79965928e-01 -5.76613665e-01 -6.37096882e-01 -6.96295202e-01
4.80681002e-01 3.80372524e-01 8.42370152e-01 -3.00987929e-01
5.44939697e-01 3.52154911e-01 -1.31346714e+00 3.54682744e-01
1.11995232e+00 7.74215102e-01 4.73716229e-01 5.40053189e-01
-2.20808476e-01 3.04442763e-01 -4.53426570e-01 -5.18103838e-01
2.63640791e-01 1.80995688e-02 -1.29038945e-01 2.27736220e-01
4.70062464e-01 -5.49834311e-01 -4.34857637e-01 9.71984029e-01
6.30438924e-01 -2.23163813e-01 6.46434307e-01 -1.01341546e+00
-4.59962130e-01 4.17719364e-01 -1.36088502e+00 -1.15051173e-01
2.49978840e-01 1.84289366e-01 8.09005201e-01 -1.37853336e+00
4.15494919e-01 1.05972922e+00 8.02213669e-01 2.38662511e-01
-7.82279313e-01 -8.97420406e-01 3.16949606e-01 -8.24856088e-02
-1.55799472e+00 1.09630004e-01 9.74004388e-01 -3.61383379e-01
6.66274965e-01 2.27417558e-01 4.86893773e-01 8.52304399e-01
9.17760096e-03 1.33833838e+00 7.12834358e-01 -4.01544690e-01
-3.02670181e-01 2.56066732e-02 1.47359625e-01 7.35532582e-01
2.24977702e-01 -2.00693160e-01 -3.35167795e-01 1.36254383e-02
8.87926579e-01 1.25693172e-01 -2.60098159e-01 -5.28364718e-01
-1.30887306e+00 5.85754156e-01 2.92450756e-01 7.47381821e-02
2.28868902e-01 -6.77896142e-02 4.79455113e-01 -8.84986594e-02
2.81988502e-01 1.30176293e-02 -3.54378343e-01 -1.35312811e-01
-9.02883589e-01 3.51257116e-01 4.86804545e-01 1.38088131e+00
3.75186384e-01 -1.08842969e-01 -2.00488895e-01 1.13402116e+00
1.27923980e-01 9.31341469e-01 4.31563139e-01 -2.66269714e-01
1.15413475e+00 8.58406782e-01 5.39185889e-02 -1.05350411e+00
-5.82491279e-01 -4.17327970e-01 -9.19139385e-01 8.98064971e-02
4.62112039e-01 -6.39588609e-02 -1.00860953e+00 8.50650609e-01
4.78111088e-01 -5.37206046e-02 -3.83434147e-01 8.87539387e-01
6.96357608e-01 7.75830030e-01 -3.72987568e-01 2.01039001e-01
1.46765828e+00 -1.13219500e+00 -6.48773372e-01 -2.66412616e-01
9.80623066e-01 -1.24929976e+00 1.42576492e+00 8.16583514e-01
-7.35545695e-01 -5.28050423e-01 -1.36538994e+00 -3.08681905e-01
-3.39390337e-01 8.07809412e-01 3.23541254e-01 5.42317808e-01
-5.26631355e-01 3.37527990e-01 -8.34646106e-01 -2.00741678e-01
5.35395980e-01 3.23646516e-01 8.99165273e-02 -5.59834316e-02
-6.58568323e-01 4.60901558e-01 3.77106845e-01 2.10067257e-01
-1.67904004e-01 -4.43623334e-01 -5.99638700e-01 -1.45164311e-01
8.14879835e-01 -1.69147924e-01 1.22509694e+00 -7.30540156e-01
-1.49354303e+00 6.43917084e-01 -9.42969695e-02 6.63349405e-02
1.03843892e+00 -4.46959853e-01 -4.16150361e-01 7.51481056e-02
-2.04871353e-02 5.11018634e-01 1.00870061e+00 -1.16209006e+00
-8.75327587e-01 -3.66967887e-01 -4.39713597e-01 4.46864039e-01
-4.48139727e-01 1.23609036e-01 -1.13401282e+00 -9.82221723e-01
4.17255372e-01 -9.70206916e-01 1.33183599e-01 4.03875619e-01
-5.94928741e-01 -4.26079094e-01 1.41874826e+00 -7.04279900e-01
1.44781256e+00 -2.29819131e+00 -2.82665461e-01 2.93263376e-01
1.12086020e-01 4.48995113e-01 -8.28213058e-03 5.21833539e-01
1.56483456e-01 7.40811080e-02 -3.13010395e-01 -4.18407112e-01
-8.30552280e-02 -1.42182931e-01 -4.42079753e-01 5.29108703e-01
3.20256054e-01 8.56125414e-01 -5.68198681e-01 -6.91943467e-01
4.55067009e-01 3.20274621e-01 -3.55957031e-01 -2.66697675e-01
-2.18422472e-01 1.36248142e-01 -7.13460386e-01 8.69639575e-01
1.14386141e+00 -3.15614343e-01 -5.59964702e-02 -1.68640822e-01
-3.40339631e-01 -2.41198525e-01 -1.28201616e+00 1.46762013e+00
-5.62167950e-02 8.06365550e-01 -8.97287950e-02 -6.16784871e-01
1.16298890e+00 -1.67990416e-01 3.13994139e-01 -7.39557981e-01
1.74188346e-01 2.55417079e-01 -1.22584499e-01 -4.44323272e-01
7.66043007e-01 5.75253963e-01 -3.14759873e-02 5.42087071e-02
-6.30943060e-01 -7.85531327e-02 2.66809791e-01 1.26401961e-01
6.70295775e-01 3.55526090e-01 8.10773745e-02 -1.46009132e-01
6.56922221e-01 1.65440634e-01 4.52515721e-01 6.31172717e-01
-8.53032917e-02 1.05131328e+00 5.74645221e-01 -1.81609586e-01
-1.22764456e+00 -8.56957972e-01 -4.09968317e-01 9.47894692e-01
6.33611917e-01 -4.74693388e-01 -7.00157225e-01 -4.99542236e-01
4.58741598e-02 4.67976063e-01 -3.54602784e-01 1.65689826e-01
-1.10537815e+00 -6.89638138e-01 6.68014586e-01 8.34756136e-01
8.38850200e-01 -7.41139889e-01 -3.07204843e-01 5.37849106e-02
-3.39854099e-02 -1.20871425e+00 -1.00764906e+00 -1.09226398e-01
-9.36942339e-01 -9.13186193e-01 -1.04324627e+00 -1.05748022e+00
6.29788280e-01 7.02133417e-01 5.37878931e-01 1.60106465e-01
-4.11528945e-01 1.00357197e-02 -4.50032294e-01 -4.94288027e-01
-1.62741467e-01 9.41096991e-02 -4.20621276e-01 2.30568387e-02
3.21950197e-01 -4.51022647e-02 -7.52495885e-01 8.45679164e-01
-8.05623889e-01 4.06767428e-01 6.69653177e-01 6.66922867e-01
6.19857073e-01 4.33271229e-02 1.00749776e-01 -8.24073136e-01
3.35290670e-01 4.82321382e-02 -7.59458184e-01 1.23731740e-01
-6.16061687e-01 -3.33248943e-01 6.53100491e-01 -6.97649479e-01
-9.90288973e-01 2.96656132e-01 -5.17512597e-02 -4.68764216e-01
7.68544748e-02 1.87321037e-01 -7.11381659e-02 -6.76466227e-02
4.67192322e-01 4.23096210e-01 -1.83052436e-01 -5.83018064e-01
2.81977683e-01 9.36887741e-01 5.24080634e-01 -4.28878605e-01
1.11049795e+00 6.55088842e-01 -2.19432771e-01 -1.10229445e+00
-4.68591303e-01 -8.73519063e-01 -7.56700099e-01 -2.18125582e-01
6.88206434e-01 -9.40851748e-01 -5.77880144e-01 1.01458347e+00
-1.14636469e+00 -2.80989319e-01 2.38153234e-01 1.92490920e-01
-2.06594437e-01 8.81204367e-01 -6.20000184e-01 -6.87670588e-01
-3.24557066e-01 -1.05940402e+00 1.55764413e+00 2.66995460e-01
2.47491390e-01 -5.36998391e-01 -3.70275736e-01 2.84137040e-01
2.96555948e-03 1.32612541e-01 7.57551968e-01 -4.39538866e-01
-7.75253236e-01 -4.00125206e-01 -6.14536524e-01 1.79760665e-01
-1.01167381e-01 2.52960712e-01 -7.66696751e-01 -3.20368707e-01
-3.47773403e-01 7.86646633e-05 8.64678681e-01 1.15335323e-01
1.37319243e+00 4.75690365e-02 -6.05763257e-01 6.31674528e-01
1.29959750e+00 1.62967667e-01 7.66845167e-01 3.37533295e-01
1.14986634e+00 4.64230895e-01 9.36606944e-01 4.65178162e-01
3.48491400e-01 9.30409849e-01 2.30283812e-01 -3.67252856e-01
-1.44879207e-01 -3.81885588e-01 3.56176645e-01 7.56439030e-01
-2.29912456e-02 -2.88423240e-01 -1.00396526e+00 2.21365228e-01
-1.84927797e+00 -6.27332687e-01 -8.72036099e-01 2.21591067e+00
5.98719895e-01 3.95744026e-01 1.68768242e-01 2.05772102e-01
9.31050777e-01 3.42348800e-03 -6.93713546e-01 -8.29093009e-02
-2.88840830e-01 -1.90496758e-01 8.31974030e-01 7.36876652e-02
-1.29880309e+00 1.00397587e+00 4.90238905e+00 1.48025215e+00
-1.15851223e+00 -1.80219546e-01 3.80440444e-01 1.75388992e-01
6.18779510e-02 -1.08566582e-01 -1.14311230e+00 5.43125927e-01
4.30478230e-02 9.02921483e-02 1.99636042e-01 9.64519024e-01
2.95632303e-01 -4.61243168e-02 -7.48581946e-01 1.34069681e+00
1.42542630e-01 -9.87485111e-01 7.34767169e-02 -5.07800877e-02
7.03551531e-01 5.91285825e-02 1.61411732e-01 4.05279040e-01
-2.20915109e-01 -7.33139813e-01 9.46606994e-01 1.43152341e-01
9.21667576e-01 -5.62139034e-01 4.29801047e-01 5.19719720e-01
-1.60113442e+00 -9.72587243e-02 -4.81096894e-01 4.16971594e-01
-5.27041703e-02 5.20558536e-01 -7.92030454e-01 6.26088142e-01
5.99726558e-01 7.98910797e-01 -8.27088952e-01 1.15235257e+00
-1.95149913e-01 6.68217182e-01 -6.53684914e-01 -3.42549086e-01
2.39117593e-01 -3.40484113e-01 5.87680221e-01 1.37523472e+00
3.59609693e-01 -3.83297615e-02 4.01612490e-01 7.30083466e-01
9.66741294e-02 5.85602760e-01 -3.78712207e-01 2.62199938e-01
3.63289773e-01 1.28447413e+00 -1.17465556e+00 -3.03609639e-01
-6.43579781e-01 1.06279874e+00 1.74956787e-02 3.33958387e-01
-1.05406570e+00 -6.31644368e-01 -1.47853836e-01 3.24384987e-01
6.07833207e-01 -2.90099978e-01 -4.57484275e-01 -1.11876750e+00
5.40351510e-01 -8.13306391e-01 2.49245852e-01 -8.94397616e-01
-1.02240777e+00 2.52664089e-01 -1.62446961e-01 -1.73277831e+00
3.99026483e-01 -9.42950726e-01 -6.95014894e-01 6.62538648e-01
-1.43446338e+00 -1.55091596e+00 -5.43597400e-01 4.50242281e-01
9.93475616e-01 1.39685050e-01 1.10676132e-01 3.74947339e-01
-6.98427856e-01 9.30536330e-01 5.54881632e-01 4.58442509e-01
9.39908147e-01 -1.10743582e+00 5.73130608e-01 6.78697407e-01
-2.54001245e-02 3.72363836e-01 4.34874296e-01 -8.25556695e-01
-1.38992763e+00 -1.15662134e+00 3.15904617e-01 -3.88531357e-01
4.32404697e-01 -8.08269203e-01 -1.16341507e+00 4.10388917e-01
-1.70646712e-01 -1.76322892e-01 -3.10242195e-02 -1.18243009e-01
-4.33450997e-01 -1.47603810e-01 -7.45396435e-01 8.16731453e-01
9.86194849e-01 -1.50974959e-01 -2.99462616e-01 4.51248437e-01
6.34065628e-01 -8.34946275e-01 -4.74378079e-01 4.51337725e-01
6.04132712e-01 -9.09817457e-01 7.85539329e-01 1.80824995e-01
4.41500723e-01 -6.31985247e-01 2.58829594e-01 -5.75450301e-01
-8.84621292e-02 -6.67862177e-01 6.12848476e-02 1.24638844e+00
3.55483592e-01 -6.88119113e-01 1.03825307e+00 1.18083514e-01
-4.13073778e-01 -9.41040576e-01 -7.52566576e-01 -8.21325600e-01
2.29496822e-01 -4.36439395e-01 4.96616900e-01 7.62940824e-01
-1.30453661e-01 1.80582941e-01 -2.94795334e-01 2.70580769e-01
4.40980464e-01 2.41900951e-01 1.02244592e+00 -1.15030622e+00
-1.93747878e-01 -6.57282233e-01 -3.09335858e-01 -1.78724384e+00
-4.66333330e-01 -8.16567719e-01 2.39057059e-04 -1.19924462e+00
2.79715180e-01 -3.75137240e-01 1.74312353e-01 3.30903828e-01
-3.18948984e-01 8.89807865e-02 2.11841986e-01 4.41228151e-01
-5.09502470e-01 5.34761608e-01 1.53620315e+00 -2.01395884e-01
-3.22702616e-01 5.93281649e-02 -2.98288494e-01 9.52996731e-01
7.69144356e-01 -2.78817594e-01 -2.20889077e-01 -3.58710051e-01
1.46376058e-01 -1.33364439e-01 3.45612854e-01 -8.50666821e-01
4.08794612e-01 2.37189844e-01 5.26856720e-01 -1.47197795e+00
4.48306873e-02 -6.78996086e-01 -3.62801850e-01 3.31954837e-01
7.05321506e-02 -5.25028892e-02 4.21189129e-01 7.41497517e-01
1.80179067e-02 -2.93413222e-01 7.13869274e-01 3.72787118e-01
-7.06391096e-01 2.40858093e-01 -7.98507333e-02 -7.79807419e-02
1.07659292e+00 -6.45672381e-01 -4.25584793e-01 -9.16081890e-02
-1.82601020e-01 6.25976145e-01 5.88869274e-01 5.55471897e-01
6.71578288e-01 -1.10869920e+00 -7.46865571e-01 2.20271096e-01
4.21090692e-01 3.68266165e-01 3.97071838e-01 9.42832410e-01
-8.27673674e-01 4.98185664e-01 3.62222672e-01 -1.04991794e+00
-1.35426128e+00 5.45529425e-01 1.55336216e-01 -3.36729407e-01
-1.10813534e+00 4.69012022e-01 4.35273767e-01 -4.48125899e-01
3.65186274e-01 -6.25937223e-01 -1.27933949e-01 -1.54227838e-01
3.92033041e-01 3.59528154e-01 1.51109681e-01 -3.78992021e-01
-9.34138969e-02 1.26702356e+00 -4.87177998e-01 2.02710196e-01
7.64301300e-01 -1.17754759e-02 3.25430781e-01 1.91932380e-01
1.10428464e+00 2.71889716e-01 -1.41465592e+00 -1.84763715e-01
-4.30129701e-03 -4.87004071e-01 -8.02618638e-02 -6.71090305e-01
-9.25189674e-01 9.03109908e-01 6.98752820e-01 1.86682045e-02
1.00033498e+00 -2.67957777e-01 8.95058215e-01 3.92186821e-01
1.92238882e-01 -1.35748029e+00 2.63121665e-01 3.61524999e-01
9.16343749e-01 -1.18377733e+00 1.59240201e-01 -9.88712609e-01
-7.20494330e-01 1.47415113e+00 6.61281407e-01 -2.07360461e-01
4.20493126e-01 4.22319740e-01 5.18549839e-03 -1.35195494e-01
-3.50319803e-01 -8.83890912e-02 3.58381450e-01 3.52409244e-01
3.61268848e-01 -1.04844257e-01 -3.32766980e-01 4.64177489e-01
2.89512891e-03 -2.99585044e-01 6.04613006e-01 9.25426424e-01
-4.88829970e-01 -1.05589604e+00 -5.92254043e-01 6.84833944e-01
-2.15161473e-01 -1.30895823e-01 -3.75342339e-01 1.14266539e+00
-1.34149969e-01 6.52037621e-01 -6.45440817e-02 -2.84483969e-01
6.62642419e-01 -2.70621091e-01 1.27644688e-01 -3.01409870e-01
-2.59697735e-01 6.01229072e-01 -1.84483200e-01 -4.17436153e-01
1.28647178e-01 -8.73800695e-01 -1.58087814e+00 -1.38142496e-01
-8.96816611e-01 -4.44572240e-01 6.07226014e-01 4.90217090e-01
1.99383259e-01 3.91540617e-01 6.13387525e-01 -7.80720174e-01
-7.50748456e-01 -9.79824424e-01 -5.25695264e-01 2.70878762e-01
7.65585378e-02 -5.67111671e-01 -3.05406660e-01 1.40322521e-01] | [12.099713325500488, 2.245655059814453] |
d026bd49-fe16-4b8b-a965-e3dd4c5d053a | corporate-culture-and-organizational | 2301.08907 | null | https://arxiv.org/abs/2301.08907v1 | https://arxiv.org/pdf/2301.08907v1.pdf | Corporate Culture and Organizational Fragility | Complex organizations accomplish tasks through many steps of collaboration among workers. Corporate culture supports collaborations by establishing norms and reducing misunderstandings. Because a strong corporate culture relies on costly, voluntary investments by many workers, we model it as an organizational public good, subject to standard free-riding problems, which become severe in large organizations. Our main finding is that voluntary contributions to culture can nevertheless be sustained, because an organization's equilibrium productivity is endogenously highly sensitive to individual contributions. However, the completion of complex tasks is then necessarily fragile to small shocks that damage the organization's culture. | ['Matthieu V. Leduc', 'Benjamin Golub', 'Matthew Elliott'] | 2023-01-21 | null | null | null | null | ['culture'] | ['speech'] | [-5.52598417e-01 6.00574195e-01 -2.25576043e-01 2.86588848e-01
1.23547986e-01 -6.04765594e-01 4.15577620e-01 1.45244915e-02
-6.20603263e-01 1.14937091e+00 5.48804820e-01 -9.11376029e-02
-1.92806304e-01 -5.03901005e-01 -2.29666248e-01 -5.43329418e-01
4.69248801e-01 1.29423589e-01 -3.05024713e-01 -2.71339983e-01
6.40680969e-01 4.06682268e-02 -8.98433805e-01 -2.49593124e-01
4.94442403e-01 3.45825136e-01 2.86351979e-01 4.78146076e-01
6.86197132e-02 1.62825787e+00 -8.92831683e-01 -4.00638759e-01
2.24923730e-01 -1.36545137e-01 -8.99053156e-01 6.90724671e-01
-4.36803907e-01 6.22687936e-02 3.05169970e-01 8.58710349e-01
1.26371175e-01 1.70922354e-02 6.02554798e-01 -1.34395730e+00
-1.46835923e+00 7.57224798e-01 -3.48922670e-01 8.16873312e-02
1.77183345e-01 5.28603554e-01 9.78686154e-01 -7.24407256e-01
9.67352509e-01 1.24204206e+00 7.48313606e-01 1.63724989e-01
-1.77406347e+00 -4.91182417e-01 1.32283330e-01 -1.05260265e+00
-1.05954218e+00 -5.79269886e-01 5.58033168e-01 -1.17225921e+00
9.56859469e-01 -9.80073139e-02 7.10553229e-01 9.01412010e-01
9.52656329e-01 -4.91936296e-01 1.33963215e+00 -2.98626363e-01
-2.66507678e-02 3.85725111e-01 2.74174392e-01 2.00203061e-01
1.30065346e+00 2.42670309e-02 -4.49355543e-01 -4.56980675e-01
1.12432611e+00 3.65097642e-01 -4.56380248e-01 6.45618141e-02
-1.54592299e+00 4.83260602e-01 1.32264197e-01 6.64474010e-01
-9.17199612e-01 7.58705318e-01 4.01380360e-02 1.10699844e+00
-1.90924816e-02 1.13872337e+00 -1.32016718e-01 -2.38746613e-01
5.78748435e-02 1.51343316e-01 8.53224218e-01 7.48037815e-01
1.03413975e+00 2.07824241e-02 -1.11622080e-01 3.11976612e-01
2.11845279e-01 7.74115920e-01 -1.85109749e-01 -1.94021273e+00
2.86840588e-01 4.84098643e-01 1.04399610e+00 -1.46164787e+00
-2.97580808e-01 -9.16473508e-01 -5.86300313e-01 1.33911252e-01
2.90414542e-01 -7.61512876e-01 1.00267470e-01 1.66008639e+00
-4.17608202e-01 -8.01170647e-01 9.58703756e-02 6.40181720e-01
-2.45012492e-01 2.49639228e-01 5.31498156e-02 -9.32741284e-01
5.98261535e-01 -8.87114763e-01 -1.40114927e+00 -4.42813426e-01
6.35088444e-01 -8.79342556e-01 8.86031449e-01 8.26784596e-02
-1.35061848e+00 -4.14853126e-01 -2.85346031e-01 2.79923201e-01
2.24548846e-01 -7.24732339e-01 7.75709331e-01 5.51210999e-01
-1.28399062e+00 3.97537410e-01 -4.38240618e-01 -5.88940233e-02
4.09240901e-01 2.26848513e-01 -3.65707338e-01 2.80819535e-01
-6.23389482e-01 9.06297565e-01 -2.92236924e-01 6.42472357e-02
-7.01606154e-01 -4.00147438e-01 -1.38863340e-01 2.53132969e-01
9.61542308e-01 -1.26171446e+00 1.21893775e+00 -1.42042685e+00
-1.02518404e+00 5.10815918e-01 2.81284422e-01 2.21153602e-01
1.68313533e-01 -1.30763762e-02 -1.58535615e-02 -1.47344396e-01
6.42829120e-01 -1.84251722e-02 5.59973538e-01 -1.79412997e+00
-2.98004270e-01 -4.59037036e-01 -1.63589008e-02 -5.45181632e-02
-4.47871536e-01 3.05410773e-01 5.18026829e-01 -3.54701340e-01
-1.98181972e-01 -1.09901214e+00 -5.52423000e-01 -7.92729974e-01
-3.09217036e-01 3.65713686e-02 -1.09820321e-01 -2.21926913e-01
1.30258536e+00 -2.07593369e+00 -5.38455956e-02 2.37329781e-01
7.77719855e-01 -6.57480359e-01 2.29366243e-01 8.29667389e-01
2.62957513e-01 8.06442142e-01 2.37178579e-01 -3.28226149e-01
1.28397718e-01 2.96797842e-01 2.32607946e-01 3.16850960e-01
2.02019110e-01 8.56057703e-01 -8.45078528e-01 -6.71495590e-03
-7.42840469e-01 -3.26709956e-01 -6.14141881e-01 1.43280044e-01
3.71534109e-01 3.78833860e-01 -3.60384464e-01 5.25332451e-01
3.05802912e-01 -7.21033871e-01 9.15097415e-01 1.12601399e+00
-9.25235868e-01 9.27155018e-02 -6.52305245e-01 7.89051652e-01
-2.44380802e-01 4.61068213e-01 5.97973406e-01 -5.98024607e-01
9.76360798e-01 5.44288337e-01 2.95560569e-01 -3.83385777e-01
2.34703124e-01 3.70843768e-01 4.48023647e-01 -3.68131280e-01
7.66880751e-01 -5.17897964e-01 -1.77392662e-01 1.05856526e+00
-1.44968212e-01 -1.38855308e-01 -7.75752142e-02 4.76630270e-01
1.23741686e+00 -6.58340096e-01 3.93124104e-01 -8.08683753e-01
-7.83492997e-02 3.80240560e-01 1.23596263e+00 9.81746197e-01
-5.29504776e-01 -1.50640592e-01 1.08035088e+00 -1.55203670e-01
-1.13461125e+00 -5.02251029e-01 4.42065150e-01 9.39583838e-01
3.60122994e-02 -1.33623064e-01 -5.01463890e-01 -3.16497594e-01
4.42666113e-01 3.86605382e-01 -5.73491335e-01 7.62189031e-02
-2.87959695e-01 -2.42456689e-01 4.97234613e-02 4.33075100e-01
3.68201435e-01 -1.05747378e+00 -6.55448496e-01 3.78259897e-01
-1.67449892e-01 -6.54732466e-01 -8.51562262e-01 9.62322354e-02
-6.71569228e-01 -1.10037649e+00 -6.84714496e-01 -3.74288619e-01
9.31113243e-01 9.75126803e-01 1.33133423e+00 7.51253068e-01
2.25395858e-01 8.59908044e-01 1.87170357e-01 -9.96362627e-01
-5.33102632e-01 -1.38452321e-01 5.01741946e-01 -1.75212435e-02
1.77524194e-01 -5.73524654e-01 -2.92518198e-01 2.49044299e-01
-1.42403975e-01 -2.25010589e-01 2.58151978e-01 8.32416117e-01
7.48825520e-02 8.04338157e-02 1.28660095e+00 -9.40471411e-01
1.12340724e+00 -5.61991215e-01 -4.44440246e-01 7.48627260e-02
-1.03819239e+00 -5.66255569e-01 2.26887897e-01 -1.88638195e-01
-1.42572200e+00 -6.45924509e-01 1.04159725e+00 1.57734454e-01
2.87724018e-01 6.54661059e-01 5.13751984e-01 9.63092893e-02
8.08380246e-01 -3.85039330e-01 6.16705835e-01 -1.01489246e-01
-2.88832307e-01 4.16969121e-01 4.45323363e-02 -7.31764734e-01
9.74833548e-01 3.49074483e-01 -1.07179701e-01 -7.33141720e-01
-7.49651670e-01 -1.57019570e-01 -3.61890823e-01 -5.74076653e-01
9.07681167e-01 -1.31554902e+00 -1.16981423e+00 1.24594569e-01
-1.18841100e+00 -6.42079890e-01 -3.61322820e-01 7.75298059e-01
-5.05487919e-01 -7.58321136e-02 -3.83265346e-01 -1.29953432e+00
3.87509555e-01 -7.77354777e-01 1.94775611e-01 2.78775692e-01
-3.60254884e-01 -1.06746852e+00 2.24290550e-01 6.26437962e-01
8.23468387e-01 3.66486311e-01 5.19920647e-01 -6.91846386e-02
-1.12690878e+00 2.28199258e-01 6.99416250e-02 4.24809486e-01
4.49133754e-01 2.98170924e-01 -4.03236151e-01 -6.57248423e-02
5.42240560e-01 -3.78292233e-01 5.17004788e-01 3.89756441e-01
2.92601675e-01 -6.58567011e-01 -7.18889153e-03 -2.35475823e-02
1.15265369e+00 2.89016396e-01 2.30011255e-01 4.83244687e-01
3.09605837e-01 1.20114267e+00 6.67163551e-01 7.39177585e-01
4.38569218e-01 1.47437796e-01 4.89875115e-02 3.52794230e-02
5.05581796e-01 1.21622056e-01 2.41788819e-01 9.72640395e-01
-1.06498599e+00 2.37115815e-01 -1.31301272e+00 5.65520465e-01
-1.93919241e+00 -1.51478863e+00 -5.69276035e-01 1.97982538e+00
7.56798565e-01 5.01112521e-01 1.96363717e-01 -1.64285198e-01
6.54908955e-01 -9.32164714e-02 -2.44305376e-02 -5.54437697e-01
-1.19011968e-01 -6.07452512e-01 5.46898723e-01 6.36907756e-01
-7.27970451e-02 6.08583391e-01 7.68353605e+00 -5.81165135e-01
-4.81344193e-01 4.29196149e-01 7.18900919e-01 -2.74878204e-01
-7.24134445e-01 2.36931965e-01 -4.72822428e-01 1.05432548e-01
7.87754238e-01 -1.00162911e+00 5.72998106e-01 7.36870468e-01
6.23759866e-01 -3.47073883e-01 -6.79062665e-01 5.36321580e-01
-2.88867950e-01 -1.72292387e+00 -9.03184474e-01 5.89966476e-01
1.52928782e+00 -5.16372025e-01 2.86182612e-01 1.31951347e-01
8.66623044e-01 -8.60698938e-01 8.61885488e-01 7.65440583e-01
3.79149199e-01 -1.04407072e+00 7.84018576e-01 7.21420169e-01
-3.81696492e-01 -7.69007921e-01 -5.74111581e-01 -1.41278565e+00
1.03975847e-01 7.22180068e-01 -5.45288324e-01 -6.20847270e-02
3.46531481e-01 1.55547902e-01 -2.08019376e-01 -7.56354537e-03
-6.34286702e-02 2.64999032e-01 7.66328275e-01 4.45940077e-01
1.66529834e-01 -4.64225709e-01 1.27759352e-01 7.13659465e-01
-4.45796996e-02 2.46643096e-01 2.55666196e-01 1.25269914e+00
-3.44105363e-01 -1.55887157e-01 -1.29740095e+00 -8.06060135e-01
7.37853348e-01 9.33351755e-01 -5.30280769e-01 -2.22996101e-01
-3.63161504e-01 2.85744280e-01 3.00734341e-01 6.23537421e-01
-1.52362049e-01 -2.86410987e-01 1.09756887e+00 2.16082290e-01
-4.40292120e-01 -7.83558249e-01 -7.43050218e-01 -9.13630486e-01
-2.78586805e-01 -7.72032678e-01 -3.34837347e-01 -4.38322008e-01
-1.07768846e+00 -1.47705361e-01 -7.69213378e-01 -2.85580039e-01
1.90260008e-01 -4.69949283e-02 -4.26067352e-01 9.58755255e-01
-1.27061725e+00 -5.98434687e-01 -5.05867563e-02 3.02916169e-01
5.14327660e-02 -1.69944063e-01 4.63185459e-01 -3.49453598e-01
-8.26664388e-01 2.35103909e-02 9.23186094e-02 -1.21217631e-01
9.94567394e-01 -1.02119720e+00 -1.56961650e-01 6.61085665e-01
-4.67161506e-01 1.41410053e+00 2.59562522e-01 -1.12312174e+00
-1.50689113e+00 -9.65538144e-01 1.53742719e+00 -8.25135589e-01
9.65593338e-01 -2.76304603e-01 -6.76411390e-01 1.19540477e+00
8.51009250e-01 -4.58673954e-01 7.36539602e-01 6.80151522e-01
-1.34199202e-01 -8.18698853e-02 -6.19095266e-01 5.74060857e-01
1.40381765e+00 -6.86140060e-01 -5.56062818e-01 4.66315120e-01
1.25591052e+00 5.52732706e-01 -1.28893089e+00 -5.96933722e-01
1.40773458e-03 -1.13933313e+00 5.67093551e-01 -5.72869301e-01
6.82021439e-01 1.37630522e-01 5.36298454e-02 -1.26331961e+00
-1.34772611e+00 -1.29933345e+00 6.96772397e-01 1.22358131e+00
2.42853746e-01 -1.12090576e+00 9.33643132e-02 7.87827611e-01
-2.89057374e-01 5.85323311e-02 -1.94037080e-01 -1.08455074e+00
6.63584545e-02 4.63572741e-01 5.62334180e-01 1.90632451e+00
3.90642583e-01 5.33784270e-01 -2.89762318e-01 -1.03619331e-02
8.79343152e-01 -1.43525794e-01 1.21239758e+00 -1.72689772e+00
-1.46725446e-01 -3.99536759e-01 3.60842496e-01 1.12711221e-01
1.69634536e-01 -5.58126807e-01 -1.27756029e-01 -1.38792193e+00
4.67151403e-01 -7.24673212e-01 -1.63522556e-01 2.25484893e-01
4.18466367e-02 -1.24131024e-01 6.46819174e-01 6.72425389e-01
-6.49060667e-01 9.71764252e-02 1.60029995e+00 3.56130034e-01
-5.27205050e-01 -3.94402534e-01 -1.67841256e+00 8.37263644e-01
8.69746685e-01 -7.17608213e-01 -1.41374752e-01 -3.62240553e-01
8.34352851e-01 7.67199934e-01 1.72158957e-01 -4.30176735e-01
8.15274194e-02 -1.42552471e+00 -1.83070093e-01 2.70855099e-01
-1.44837826e-01 -7.61340618e-01 9.89084184e-01 8.93955410e-01
-3.06621850e-01 4.17442709e-01 -6.13406897e-01 4.44766343e-01
-1.69333443e-01 -7.26589784e-02 7.18160689e-01 -5.28545499e-01
3.54248017e-01 -3.68825972e-01 -9.02326345e-01 1.78632274e-01
1.10676181e+00 -3.14174443e-01 -8.58207643e-01 -1.98109299e-01
-6.35184944e-01 1.07149832e-01 1.13035369e+00 -1.40770480e-01
-1.14270769e-01 -1.34193408e+00 -5.89492142e-01 -4.16864187e-01
-1.62140019e-02 -3.26683760e-01 1.86423659e-02 1.27958763e+00
-3.64840239e-01 7.84075975e-01 -6.66740119e-01 2.28514299e-02
-7.40791619e-01 6.25823736e-01 1.41763330e-01 -1.05279535e-01
-1.54296875e-01 8.06198061e-01 5.48214674e-01 2.99683720e-01
-2.18757540e-01 -1.48625895e-02 -2.35696230e-02 3.52292061e-01
5.52172959e-01 6.04324937e-01 -6.95233285e-01 -3.83158356e-01
-2.68799335e-01 1.18951008e-01 2.52481878e-01 -2.25198567e-01
1.25415170e+00 -7.07225680e-01 -7.32675374e-01 9.21654522e-01
2.78423786e-01 3.55824947e-01 -1.13736153e+00 -8.95219296e-02
2.95778006e-01 -1.16566682e+00 -1.60599887e-01 -5.17110586e-01
-7.58283734e-01 3.78740370e-01 -5.36500454e-01 6.66663945e-01
4.60876286e-01 -2.65700608e-01 -1.64458141e-01 6.72715485e-01
7.32755423e-01 -1.82941186e+00 1.01211774e+00 6.07293606e-01
1.16148531e+00 -9.64527369e-01 -3.55952680e-01 -5.20746291e-01
-1.11457682e+00 5.00298142e-01 6.80373073e-01 -4.28903587e-02
5.72639883e-01 2.35929251e-01 4.83678654e-02 -2.91619927e-01
-1.47545969e+00 1.42822549e-01 -7.05897570e-01 8.84839654e-01
6.94308996e-01 3.99220675e-01 -8.50086451e-01 6.37215495e-01
1.51854143e-01 2.37701777e-02 1.64489996e+00 8.52139950e-01
-9.85319614e-01 -1.03965724e+00 -1.04710531e+00 3.78578275e-01
-8.76036346e-01 1.29775807e-01 -7.66610622e-01 7.49271214e-01
4.12006080e-01 1.03051305e+00 3.29045057e-01 2.27673575e-01
2.73534685e-01 1.41340211e-01 -2.06582732e-02 -1.05779171e+00
-1.14867127e+00 3.09466630e-01 4.88115370e-01 -3.29665244e-01
-6.13446295e-01 -8.91208172e-01 -1.21949792e+00 -8.91611755e-01
-1.59195483e-01 7.76370838e-02 1.91425130e-01 4.06828761e-01
4.30008829e-01 6.97987437e-01 9.84004200e-01 -2.71590561e-01
-8.26585352e-01 -6.08476996e-01 -1.27715945e+00 4.61379677e-01
2.01742664e-01 -7.92336106e-01 -3.17605466e-01 3.03599119e-01] | [8.997416496276855, 6.15219259262085] |
51a2aab2-bdb1-4d91-80ec-8b47b63d0b2f | target-speaker-voice-activity-detection-with-1 | 2208.13085 | null | https://arxiv.org/abs/2208.13085v3 | https://arxiv.org/pdf/2208.13085v3.pdf | Target Speaker Voice Activity Detection with Transformers and Its Integration with End-to-End Neural Diarization | This paper describes a speaker diarization model based on target speaker voice activity detection (TS-VAD) using transformers. To overcome the original TS-VAD model's drawback of being unable to handle an arbitrary number of speakers, we investigate model architectures that use input tensors with variable-length time and speaker dimensions. Transformer layers are applied to the speaker axis to make the model output insensitive to the order of the speaker profiles provided to the TS-VAD model. Time-wise sequential layers are interspersed between these speaker-wise transformer layers to allow the temporal and cross-speaker correlations of the input speech signal to be captured. We also extend a diarization model based on end-to-end neural diarization with encoder-decoder based attractors (EEND-EDA) by replacing its dot-product-based speaker detection layer with the transformer-based TS-VAD. Experimental results on VoxConverse show that using the transformers for the cross-speaker modeling reduces the diarization error rate (DER) of TS-VAD by 11.3%, achieving a new state-of-the-art (SOTA) DER of 4.57%. Also, our extended EEND-EDA reduces DER by 6.9% on the CALLHOME dataset relative to the original EEND-EDA with a similar model size, achieving a new SOTA DER of 11.18% under a widely used training data setting. | ['Jian Wu', 'Takuya Yoshioka', 'Naoyuki Kanda', 'Xiong Xiao', 'Dongmei Wang'] | 2022-08-27 | null | null | null | null | ['activity-detection'] | ['computer-vision'] | [-2.28236988e-01 1.70811370e-01 4.05181795e-01 -4.52686638e-01
-9.42838252e-01 -5.33719182e-01 4.42913055e-01 -4.70831931e-01
-3.46108042e-02 -1.79033011e-01 3.87861311e-01 -4.28366989e-01
1.26541898e-01 -1.82409748e-01 -4.42525506e-01 -6.37581885e-01
-2.67689824e-01 3.83737892e-01 -2.41794083e-02 -1.65499493e-01
-4.74604666e-01 3.92915547e-01 -1.32983041e+00 4.02761251e-01
5.50526798e-01 1.13007021e+00 -4.59396653e-02 7.46575892e-01
1.88356582e-02 4.05624926e-01 -8.88514280e-01 -2.65203387e-01
2.27643296e-01 -6.23924911e-01 -4.76799548e-01 7.35303834e-02
5.91141701e-01 -2.48670265e-01 -4.60494429e-01 6.36089087e-01
7.97239065e-01 2.07347825e-01 5.17506897e-01 -1.16045654e+00
-5.45455337e-01 1.01844645e+00 -3.35738927e-01 4.59461033e-01
2.07246333e-01 2.15233043e-02 1.03953767e+00 -1.11769402e+00
1.72772974e-01 1.67518222e+00 6.71842158e-01 8.32316220e-01
-1.52672589e+00 -9.26825047e-01 3.42003763e-01 -5.36610596e-02
-1.40468526e+00 -7.87388802e-01 1.02163053e+00 -5.18817008e-01
1.42881060e+00 6.07667327e-01 5.98145783e-01 1.32509208e+00
-2.68773679e-02 8.27269614e-01 7.76001096e-01 -1.69084221e-01
2.60284245e-01 2.40068644e-01 2.31534421e-01 4.92629349e-01
-4.66777623e-01 1.46202937e-01 -8.55986655e-01 6.19583391e-02
5.54881752e-01 -3.62444311e-01 -7.33760893e-02 7.70657696e-03
-1.23014581e+00 7.19903469e-01 3.33064288e-01 5.34632146e-01
-3.78322124e-01 -1.78715348e-01 6.08628392e-01 5.36843300e-01
5.85727155e-01 2.74373651e-01 -3.55721533e-01 -1.04275793e-01
-1.30312777e+00 3.61760557e-02 6.98435426e-01 7.42307246e-01
2.86865067e-02 9.82263088e-01 -3.33918124e-01 1.05256343e+00
5.49164593e-01 7.33512819e-01 1.00463593e+00 -5.91387689e-01
4.89096254e-01 3.26080024e-01 -4.53141958e-01 -5.65137088e-01
-3.95798385e-01 -9.84135628e-01 -8.30672264e-01 3.26908678e-02
2.29949698e-01 -7.57044181e-02 -9.94980633e-01 2.02239442e+00
2.56306797e-01 2.03595906e-01 2.36238614e-01 8.67380619e-01
8.42270732e-01 8.69257092e-01 -3.18624318e-01 -2.23423049e-01
1.35763872e+00 -8.58340979e-01 -9.11042809e-01 -2.32878312e-01
5.25454760e-01 -6.76900625e-01 1.30485451e+00 4.06950831e-01
-1.03304553e+00 -5.28102398e-01 -1.00032127e+00 6.56547844e-02
-1.50979638e-01 1.66977897e-01 4.92854081e-02 9.20478940e-01
-1.26132584e+00 2.37741351e-01 -1.00869930e+00 -2.71936804e-01
5.03475703e-02 4.67836320e-01 -6.46604970e-02 6.90502167e-01
-1.32014918e+00 7.14957833e-01 -1.04530267e-01 2.01461032e-01
-1.33128238e+00 -1.03151095e+00 -6.52692974e-01 2.28906766e-01
-2.86488622e-01 -5.01077712e-01 1.46256196e+00 -7.47863650e-01
-2.21700478e+00 5.97698033e-01 -4.61507529e-01 -7.34150350e-01
3.55711401e-01 -1.88153088e-01 -8.49835157e-01 -1.56193882e-01
-1.80575222e-01 4.27869827e-01 9.20770764e-01 -8.72390509e-01
-1.84302375e-01 -3.66000772e-01 -4.41968739e-01 1.86016023e-01
-6.34626508e-01 1.14179924e-01 -1.47561699e-01 -7.86491752e-01
2.68377960e-01 -1.06157720e+00 3.88364673e-01 -4.99007404e-01
-6.95704937e-01 -2.74180442e-01 9.90515649e-01 -9.61362064e-01
1.50287056e+00 -2.40586495e+00 3.76485348e-01 2.42888406e-02
2.36847997e-01 3.27994436e-01 -2.85314709e-01 3.47864658e-01
-4.04228538e-01 -1.00692376e-01 -1.76089302e-01 -1.00645876e+00
1.78833693e-01 -1.06799304e-01 -5.58965504e-01 3.77718508e-01
1.65204421e-01 3.74548048e-01 -3.93626243e-01 2.61287857e-02
2.68306106e-01 8.76686931e-01 -7.59773254e-01 3.25652868e-01
1.38865009e-01 3.18861455e-01 9.78583917e-02 4.06825572e-01
5.06296754e-01 3.42153788e-01 -2.98812501e-02 2.38917284e-02
-2.59150475e-01 1.04256153e+00 -1.07026350e+00 1.47624826e+00
-7.22482920e-01 8.98144662e-01 2.11083308e-01 -5.14718890e-01
8.89631987e-01 8.38713706e-01 3.75570238e-01 -5.42385161e-01
2.61853516e-01 3.33190620e-01 4.30833280e-01 -3.33146118e-02
5.75885400e-02 -3.68534148e-01 -2.25381646e-02 2.62623191e-01
5.00200868e-01 7.84671679e-02 -2.79907674e-01 1.27216235e-01
7.66550124e-01 -4.01578963e-01 -4.13539708e-01 -4.40673530e-01
4.76847559e-01 -6.30854845e-01 5.34634113e-01 1.99756861e-01
-2.01778471e-01 4.93498266e-01 3.34930092e-01 5.35729565e-02
-8.91391337e-01 -1.28174222e+00 -3.85354042e-01 1.04433477e+00
-6.32269561e-01 -4.57705587e-01 -1.14436483e+00 -3.04459810e-01
-1.43233791e-01 1.19789195e+00 -5.25358856e-01 -2.54617780e-01
-6.17054105e-01 -5.99387765e-01 1.02210808e+00 4.38180059e-01
4.64606553e-01 -5.75917125e-01 -1.41608417e-01 4.14056987e-01
-3.63322645e-01 -1.04698277e+00 -1.00116229e+00 3.95261258e-01
-9.24110651e-01 -2.56452113e-01 -7.58022070e-01 -6.80370748e-01
1.02497280e-01 2.77765803e-02 6.59215689e-01 -8.65321755e-01
2.69474775e-01 1.92331374e-01 7.96200708e-02 -3.44549328e-01
-7.07016706e-01 2.25198805e-01 6.66166008e-01 4.94295776e-01
4.46573526e-01 -7.83946455e-01 -2.20747858e-01 4.14301187e-01
-5.93742847e-01 -2.17285007e-01 2.10102275e-01 9.08550143e-01
2.33280495e-01 -1.89939260e-01 7.03825891e-01 -2.85767645e-01
5.55419445e-01 -2.61325479e-01 -5.97697794e-01 -9.71368104e-02
-6.44409180e-01 3.19863588e-01 4.91473913e-01 -8.56494308e-01
-1.04466987e+00 -1.55071974e-01 -3.75304967e-01 -1.05101442e+00
1.71806470e-01 1.98672295e-01 -3.99622828e-01 4.90003407e-01
6.05499268e-01 3.77964109e-01 4.28252406e-02 -7.73643851e-01
2.41155773e-01 8.94875407e-01 3.32589537e-01 1.05643645e-01
7.10781157e-01 1.63659416e-02 -5.07719398e-01 -1.15570879e+00
-5.35062015e-01 -4.02968287e-01 -4.87717330e-01 -7.61799812e-02
8.11774492e-01 -1.16881514e+00 -8.38139892e-01 7.22136497e-01
-1.12376046e+00 -3.58029276e-01 -3.02201897e-01 8.61974478e-01
-2.55688936e-01 -6.67327344e-02 -7.93613791e-01 -1.01066232e+00
-6.11920536e-01 -1.16728902e+00 9.57052767e-01 -3.20411921e-01
-3.89505237e-01 -9.82580900e-01 2.15493187e-01 3.43868673e-01
7.48246253e-01 -1.94149077e-01 7.19560504e-01 -1.02027285e+00
-7.44300857e-02 2.52230056e-02 5.66442192e-01 7.31155872e-01
1.51348084e-01 -1.55246451e-01 -1.68982077e+00 -3.10071170e-01
5.28914213e-01 3.53179932e-01 6.84141517e-01 5.29196024e-01
6.58799827e-01 -4.87075865e-01 -9.48373452e-02 6.25432372e-01
6.76238120e-01 2.91806579e-01 3.03792298e-01 -4.58855145e-02
6.72500074e-01 4.09954876e-01 -9.93239805e-02 1.57936528e-01
4.17712271e-01 9.31333601e-01 2.39865914e-01 -1.18124433e-01
-6.58142388e-01 -3.47699583e-01 9.80922341e-01 1.68036211e+00
2.61368215e-01 -3.27745944e-01 -8.20977449e-01 7.11282909e-01
-1.18254256e+00 -1.09947956e+00 -1.41950309e-01 2.41925454e+00
7.87918627e-01 2.16129318e-01 5.43107986e-01 4.54565555e-01
6.27087414e-01 1.31400719e-01 -6.03853047e-01 -7.57338047e-01
-9.30937678e-02 -8.52334350e-02 1.25549406e-01 7.83448875e-01
-8.88882339e-01 8.59180152e-01 6.18947172e+00 3.94324094e-01
-1.65238309e+00 3.77671719e-01 1.73923805e-01 -5.37450254e-01
-1.87008888e-01 -4.55997169e-01 -1.04964662e+00 3.90433878e-01
1.81156504e+00 -1.75025240e-01 6.92671299e-01 7.77897239e-01
3.88545990e-01 6.02189243e-01 -1.39072967e+00 1.19590509e+00
2.96049863e-01 -6.47323966e-01 -4.17775400e-02 2.47637331e-01
2.73024499e-01 1.90812096e-01 5.14652014e-01 6.03426337e-01
9.21194777e-02 -7.75881290e-01 1.09077215e+00 1.15107097e-01
8.57198954e-01 -5.40707588e-01 4.31970894e-01 1.62916347e-01
-1.24622679e+00 -1.11639157e-01 3.99756208e-02 1.48773417e-01
7.19732940e-02 5.57778478e-01 -1.53525162e+00 1.00693800e-01
7.14853346e-01 4.73072588e-01 -3.22419405e-01 5.83706021e-01
1.81108221e-01 1.21758342e+00 -5.64798832e-01 1.39064834e-01
1.79938599e-01 1.03626758e-01 1.12804556e+00 1.27067769e+00
4.48192060e-01 -3.91476184e-01 -4.93300408e-01 8.67228448e-01
-2.50033677e-01 -1.61871061e-01 -3.60967010e-01 -5.23629412e-02
6.52570546e-01 8.75627816e-01 -5.11534847e-02 -2.34927222e-01
-9.50914528e-03 1.15358996e+00 6.24306165e-02 4.35553610e-01
-9.40356016e-01 -4.56372917e-01 1.12060726e+00 2.99783975e-01
5.13079703e-01 -1.58101112e-01 -1.08462662e-01 -1.14570189e+00
3.74854095e-02 -1.04700541e+00 1.23891048e-01 -4.51763898e-01
-1.22223783e+00 1.30928087e+00 -1.22006878e-01 -1.27738583e+00
-6.16984010e-01 -4.44997311e-01 -6.13196611e-01 1.18210590e+00
-1.13981497e+00 -1.11553061e+00 2.76449949e-01 7.50611365e-01
8.35540056e-01 -3.63859981e-01 1.10107112e+00 4.92529213e-01
-8.99012148e-01 1.15345991e+00 1.93330497e-01 1.78019240e-01
6.23016894e-01 -1.30394983e+00 6.95504069e-01 9.07518446e-01
1.71350747e-01 7.59861708e-01 7.99152911e-01 -1.34558618e-01
-1.39766216e+00 -1.08356345e+00 1.19857895e+00 -6.03143513e-01
5.72569311e-01 -1.09486210e+00 -9.84520972e-01 1.02043784e+00
3.18409175e-01 -1.63457707e-01 8.01801324e-01 4.05792385e-01
-6.54525101e-01 -6.19546652e-01 -9.39601302e-01 4.89449888e-01
8.36655021e-01 -9.65022266e-01 -8.29510868e-01 2.84785591e-02
9.03495252e-01 -3.12938839e-01 -7.67100394e-01 -1.16558395e-01
4.59073693e-01 -7.64763892e-01 8.34898472e-01 -3.05291325e-01
-3.10567617e-01 -3.01459432e-01 -3.51281106e-01 -1.66278255e+00
-4.09330010e-01 -8.13827157e-01 -1.64189428e-01 1.59135258e+00
6.03050530e-01 -7.46443808e-01 1.32817313e-01 4.01483536e-01
-5.15048265e-01 -3.58151734e-01 -1.45630801e+00 -1.18807745e+00
2.05295190e-01 -6.72957122e-01 7.27541924e-01 9.33561921e-01
1.42555028e-01 8.71311367e-01 -3.02500665e-01 4.96305227e-01
4.30142462e-01 -2.62250513e-01 3.23098987e-01 -1.30164456e+00
-3.31325710e-01 -5.64467490e-01 -3.15342128e-01 -1.10816252e+00
-3.50578092e-02 -1.16648388e+00 -1.28915031e-02 -1.17829132e+00
-3.71692866e-01 -3.69591936e-02 -4.28784519e-01 3.82827252e-01
2.61482090e-01 -8.06689709e-02 1.94904312e-01 9.58539173e-02
-2.42984295e-02 8.42595756e-01 6.93464935e-01 -1.44209534e-01
-7.05727875e-01 6.15448020e-02 -2.42912471e-01 5.45474648e-01
6.21929407e-01 -5.75878143e-01 -4.37045336e-01 -4.96615112e-01
-4.57885444e-01 1.67121381e-01 1.29543737e-01 -1.21389592e+00
1.04656123e-01 5.62331200e-01 1.52876347e-01 -5.67055166e-01
8.64719570e-01 -5.82519352e-01 2.41220891e-01 5.53992927e-01
-5.34144759e-01 -7.87201920e-04 4.99917746e-01 2.57142484e-01
-3.15595239e-01 4.60142344e-01 7.00615108e-01 4.09688920e-01
1.09423071e-01 -2.85681412e-02 -8.14818919e-01 -7.68054575e-02
6.24719918e-01 -6.58547739e-03 4.87319045e-02 -3.68275136e-01
-7.64806867e-01 -1.33932024e-01 -2.21855372e-01 6.16114914e-01
4.81794059e-01 -1.37522984e+00 -8.32480311e-01 6.39722764e-01
4.16906029e-02 -4.91248101e-01 3.89795721e-01 9.01880682e-01
8.98453519e-02 8.26829612e-01 1.37611568e-01 -7.53315806e-01
-1.44223440e+00 4.88742292e-01 8.54806066e-01 -1.14940695e-01
-5.27896285e-01 1.18565083e+00 5.34632564e-01 -6.26606643e-01
4.93410230e-01 -7.23623812e-01 -3.18686292e-02 1.49203330e-01
4.49850023e-01 3.37485433e-01 4.75719631e-01 -9.50571954e-01
-7.53320396e-01 1.93200007e-01 -1.86072346e-02 -5.86065292e-01
1.31693292e+00 -2.27628827e-01 3.44295472e-01 1.17741668e+00
1.29621768e+00 2.36575052e-01 -1.14914823e+00 -3.60437900e-01
-3.25794399e-01 5.46958707e-02 6.27059877e-01 -9.84513819e-01
-1.12114024e+00 1.42603290e+00 1.11310053e+00 3.00552160e-01
1.05665088e+00 -5.34555539e-02 8.14317763e-01 4.00184616e-02
-2.09448859e-01 -7.69841969e-01 4.87508103e-02 5.95227182e-01
1.27936959e+00 -7.81844854e-01 -7.37905025e-01 -2.88438406e-02
-8.86752248e-01 9.86379981e-01 3.31673443e-01 2.23823160e-01
8.21858585e-01 2.35389605e-01 2.94784814e-01 -7.11963549e-02
-9.07900810e-01 1.96234152e-01 4.75726783e-01 3.34038883e-01
6.16312802e-01 2.40427539e-01 3.75981569e-01 8.74489665e-01
-6.80548310e-01 -6.12365901e-01 2.58372873e-01 2.41347417e-01
-3.51965576e-02 -8.43461633e-01 -5.32382071e-01 -2.63286587e-02
-4.21497673e-01 -2.80520171e-01 -3.20751786e-01 3.63886595e-01
-2.97522247e-01 1.17147005e+00 3.30863148e-01 -8.19839001e-01
6.00649655e-01 7.01973796e-01 -1.97239948e-04 -4.22058195e-01
-1.05738163e+00 5.96358001e-01 1.06317259e-01 -2.95640022e-01
1.37057146e-02 -8.25881064e-01 -1.11236835e+00 -3.38969260e-01
-4.35142964e-01 1.09699801e-01 8.97501647e-01 6.64236605e-01
6.92785323e-01 1.08979332e+00 9.33047712e-01 -5.88731050e-01
-7.21377730e-01 -1.48945868e+00 -7.62058973e-01 -1.29647274e-03
8.77758503e-01 -4.75859791e-01 -7.38662779e-01 3.42516303e-02] | [14.47159194946289, 6.153413772583008] |
a12f9aa7-cc53-4c1b-929d-33adce26b125 | biomedgpt-a-unified-and-generalist-biomedical | 2305.17100 | null | https://arxiv.org/abs/2305.17100v1 | https://arxiv.org/pdf/2305.17100v1.pdf | BiomedGPT: A Unified and Generalist Biomedical Generative Pre-trained Transformer for Vision, Language, and Multimodal Tasks | In this paper, we introduce a unified and generalist Biomedical Generative Pre-trained Transformer (BiomedGPT) model, which leverages self-supervision on large and diverse datasets to accept multi-modal inputs and perform a range of downstream tasks. Our experiments demonstrate that BiomedGPT delivers expansive and inclusive representations of biomedical data, outperforming the majority of preceding state-of-the-art models across five distinct tasks with 20 public datasets spanning over 15 unique biomedical modalities. Through the ablation study, we also showcase the efficacy of our multi-modal and multi-task pretraining approach in transferring knowledge to previously unseen data. Overall, our work presents a significant step forward in developing unified and generalist models for biomedicine, with far-reaching implications for improving healthcare outcomes. | ['Lichao Sun', 'Hongfang Liu', 'Yong Chen', 'Quanzheng Li', 'Brian D. Davison', 'Lifang He', 'Xiang Li', 'Yuyin Zhou', 'Chen Chen', 'Xun Chen', 'Sunyang Fu', 'Eashan Adhikarla', 'Yixin Liu', 'Zhiling Yan', 'Jun Yu', 'Kai Zhang'] | 2023-05-26 | null | null | null | null | ['image-captioning', 'text-summarization'] | ['computer-vision', 'natural-language-processing'] | [ 6.49213791e-01 4.55019236e-01 -8.69021788e-02 -6.03749156e-01
-1.45588040e+00 -4.08135504e-01 6.22109592e-01 6.65364116e-02
-2.87062019e-01 1.21428406e+00 4.41336572e-01 -5.18628180e-01
4.34267595e-02 -4.82902378e-01 -8.80495191e-01 -7.25848496e-01
1.96600333e-01 7.96103418e-01 -4.40653831e-01 -1.14984430e-01
-3.81767422e-01 2.22135335e-01 -6.74227655e-01 8.26772988e-01
7.31476545e-01 6.86771989e-01 9.79064107e-02 5.56637824e-01
3.08999211e-01 3.53995472e-01 -5.05308867e-01 -6.90293431e-01
-2.80190527e-01 -7.00396061e-01 -1.00899041e+00 -2.87577093e-01
2.36751989e-01 -1.45795614e-01 -1.62566960e-01 5.91616094e-01
1.09247446e+00 -2.31435984e-01 9.57269847e-01 -7.54180968e-01
-1.35504329e+00 7.28894353e-01 -2.74358004e-01 4.03066874e-01
3.12631041e-01 3.87340337e-01 8.26976180e-01 -8.55340898e-01
8.99588704e-01 1.15345240e+00 1.01150453e+00 1.08654630e+00
-1.38470483e+00 -8.34396660e-01 8.37259553e-03 -2.00763419e-01
-1.02181935e+00 -4.68852311e-01 1.72942817e-01 -4.80049938e-01
1.17013896e+00 9.97440964e-02 3.79418969e-01 1.91905284e+00
6.01499498e-01 8.53638828e-01 1.16182971e+00 -1.65158212e-01
-1.14059955e-01 -9.46397632e-02 -3.64994779e-02 6.56722784e-01
2.18391135e-01 -1.69819146e-02 -5.44379830e-01 -3.76982838e-01
6.50145471e-01 1.76325887e-01 -4.12047952e-01 3.25241327e-01
-1.64541459e+00 6.86106265e-01 3.56170684e-01 5.10326982e-01
-5.18124759e-01 1.22621335e-01 4.03679579e-01 1.19626589e-01
7.06130743e-01 5.09247184e-01 -7.69427121e-01 1.30924374e-01
-1.01110470e+00 -1.04327001e-01 3.83439004e-01 8.21193278e-01
2.80866861e-01 4.40743193e-02 -4.33248073e-01 6.79207623e-01
1.36226326e-01 5.02437294e-01 6.36065304e-01 -4.67361927e-01
2.61039436e-01 3.81000161e-01 -5.63373148e-01 -2.90836036e-01
-6.04554355e-01 -8.88019860e-01 -1.13011408e+00 -5.03545105e-01
8.43343958e-02 -4.58482504e-01 -1.53361022e+00 2.06122065e+00
1.13590650e-01 2.32334912e-01 4.73169774e-01 4.81548488e-01
1.38692796e+00 3.71745318e-01 8.61542404e-01 2.80156374e-01
1.55698323e+00 -6.19502723e-01 -6.75926566e-01 -3.47881347e-01
6.06526971e-01 -5.20391405e-01 6.47295117e-01 2.62921840e-01
-1.12641895e+00 -4.82588917e-01 -6.24771595e-01 -3.20017010e-01
-5.08336186e-01 2.17779815e-01 7.99795687e-01 2.66999692e-01
-1.25072384e+00 4.38015550e-01 -1.10020220e+00 -5.67247868e-01
8.96335065e-01 1.85320169e-01 -6.09363794e-01 -3.66636902e-01
-1.45630157e+00 1.09703660e+00 4.48410213e-01 1.44954741e-01
-1.57170677e+00 -1.36837053e+00 -9.48678076e-01 -1.20706379e-01
-3.07209998e-01 -1.65119863e+00 9.72094536e-01 -3.93921226e-01
-1.02177823e+00 1.28450906e+00 -6.04067817e-02 -4.02815849e-01
3.91384393e-01 -2.63434052e-01 -4.76769120e-01 1.77201331e-01
2.73804456e-01 8.78752530e-01 3.48843068e-01 -9.68693018e-01
-2.04937801e-01 -4.71051306e-01 -5.25483072e-01 7.40244240e-02
-2.13835329e-01 -1.61541820e-01 -2.45229572e-01 -8.48477542e-01
-3.37199658e-01 -7.39264965e-01 -3.79930019e-01 -3.15024227e-01
-8.33437383e-01 -1.80617362e-01 2.90169567e-01 -7.62385368e-01
7.25303352e-01 -1.90683961e+00 4.38383967e-01 -2.24929184e-01
3.52163792e-01 6.55282661e-03 -1.90764591e-01 4.61997837e-01
-3.67319852e-01 1.24649674e-01 -2.38203198e-01 -5.66485882e-01
-1.89160064e-01 1.64207190e-01 -2.91953325e-01 3.18055898e-01
5.58572948e-01 1.61568105e+00 -1.05814373e+00 -3.92884374e-01
5.08656725e-03 9.19024467e-01 -4.40136760e-01 2.27103606e-01
-2.53016874e-03 1.00123549e+00 -6.78233445e-01 9.56363678e-01
3.01215321e-01 -8.70409310e-01 1.19802773e-01 -3.60567868e-01
5.85160971e-01 2.38858923e-01 1.73863936e-02 2.28525782e+00
-3.05685103e-01 1.71394926e-02 -3.02582756e-02 -1.16761422e+00
5.36951244e-01 7.49190152e-01 4.58945751e-01 -2.35456124e-01
2.00125381e-01 2.41114154e-01 -1.80173188e-01 -5.03412604e-01
-2.01219648e-01 -7.89477408e-01 -2.58755773e-01 2.87476271e-01
8.82348239e-01 5.66856414e-02 -2.78610110e-01 2.83017576e-01
1.27760148e+00 4.10710782e-01 2.14439973e-01 -4.08102661e-01
2.89130628e-01 -1.48946330e-01 4.56654668e-01 7.41569161e-01
-1.52783766e-01 7.47629762e-01 1.23609640e-01 -1.70489803e-01
-5.54419935e-01 -1.30376256e+00 -4.31734681e-01 1.04528093e+00
-5.72028100e-01 -8.38986486e-02 -5.38759410e-01 -8.29371631e-01
2.69036084e-01 6.22270525e-01 -1.17400157e+00 -5.19380033e-01
-3.16668689e-01 -1.45574236e+00 1.00336576e+00 8.20254564e-01
1.50810286e-01 -9.02677357e-01 -2.00534120e-01 2.13276848e-01
-2.87986904e-01 -1.23186374e+00 -2.73597777e-01 3.85396063e-01
-1.06823409e+00 -9.06841755e-01 -1.19528902e+00 -8.77628565e-01
6.79181635e-01 -3.15926641e-01 1.36288512e+00 -3.47023398e-01
-6.02002978e-01 4.32342827e-01 -2.73848176e-02 -7.20182002e-01
-6.52035475e-01 4.48455423e-01 -3.07759196e-01 -4.02063906e-01
4.33036655e-01 -4.86321419e-01 -8.80962372e-01 -2.81778842e-01
-7.54261494e-01 2.54526287e-01 9.59713161e-01 1.22121251e+00
8.15721452e-01 -7.18174994e-01 1.32384288e+00 -1.44330585e+00
6.62644029e-01 -9.96854365e-01 3.37932110e-01 4.17438805e-01
-6.15174949e-01 -2.28522681e-02 5.94259024e-01 -2.34241098e-01
-1.22025967e+00 -2.35160187e-01 -4.83244210e-01 -3.23284686e-01
-3.86908710e-01 7.30189264e-01 -7.42331073e-02 2.77133256e-01
7.43955672e-01 4.10807729e-01 -3.07660792e-02 -5.18334329e-01
5.08240521e-01 3.97066176e-01 8.70412171e-01 -6.35211289e-01
3.65504324e-01 4.82938141e-01 1.09590232e-01 -1.11680515e-01
-1.09667087e+00 -1.78469226e-01 -4.36111361e-01 2.55818546e-01
1.23573852e+00 -1.24735403e+00 -2.28817061e-01 4.90458369e-01
-8.28044951e-01 -4.18536246e-01 -1.53649181e-01 3.35086763e-01
-4.19429660e-01 6.54959977e-02 -1.14678502e+00 -2.54671782e-01
-8.44206989e-01 -1.10372424e+00 1.63556969e+00 -1.13194743e-02
-4.65621024e-01 -1.52168763e+00 2.83254474e-01 6.63063109e-01
4.04361904e-01 5.72398543e-01 1.11575878e+00 -9.94223714e-01
-1.67544961e-01 -1.36412017e-03 -4.91106473e-02 9.37637687e-02
5.36638260e-01 -5.06733656e-01 -1.31974506e+00 -4.84177351e-01
-3.81392121e-01 -8.59819889e-01 1.35037196e+00 5.19366860e-01
1.21065664e+00 1.16049305e-01 -9.24878955e-01 1.01958609e+00
1.13322401e+00 5.85107878e-02 5.47915161e-01 -1.94493271e-02
5.94907165e-01 1.88259557e-01 3.10253710e-01 -2.08890671e-03
4.84710276e-01 -9.96406004e-02 1.27236083e-01 -8.34728897e-01
-3.81897211e-01 -1.35563806e-01 -2.05051973e-02 5.14314294e-01
-1.09659903e-01 -3.67595464e-01 -9.62356687e-01 8.83830905e-01
-1.40532100e+00 -8.09886634e-01 4.64049488e-01 1.65063226e+00
1.46850467e+00 -1.22348428e-01 -1.77610219e-01 -4.71103698e-01
4.30551201e-01 -3.12546760e-01 -8.30173433e-01 -1.15902260e-01
-2.14762509e-01 8.05289805e-01 1.05400145e-01 1.17454328e-01
-1.04538405e+00 8.16521406e-01 7.92128229e+00 5.58664083e-01
-1.06008971e+00 4.11008626e-01 1.17592728e+00 -1.66625734e-02
-5.17070413e-01 -5.81153929e-01 -7.58706033e-01 2.82151669e-01
1.27314210e+00 -5.32039143e-02 -1.56811148e-01 4.80349422e-01
-2.29643449e-01 4.87457186e-01 -1.49607289e+00 5.82488179e-01
2.11048678e-01 -1.32613063e+00 3.51525962e-01 -7.14681223e-02
7.79772758e-01 4.66901660e-01 4.33757722e-01 4.09472525e-01
6.96650922e-01 -1.45798469e+00 6.13535866e-02 6.19886398e-01
1.35083675e+00 -2.88225919e-01 8.13197076e-01 -1.20424658e-01
-6.08747423e-01 3.70749325e-01 2.43689902e-02 5.02009213e-01
3.49917442e-01 7.45039940e-01 -1.13398945e+00 1.00318420e+00
6.11028254e-01 9.44679320e-01 -3.35061073e-01 6.66250587e-01
-9.87907499e-02 6.82508886e-01 -1.65156554e-02 5.10437250e-01
1.32486939e-01 3.93497497e-01 1.75057665e-01 1.74979651e+00
4.76417452e-01 1.27491266e-01 -4.77020442e-02 1.01538467e+00
-4.70949471e-01 -1.70415044e-01 -5.30323088e-01 -2.98154712e-01
2.81628609e-01 1.28115201e+00 -3.80020499e-01 -5.93659759e-01
-2.72490293e-01 9.85200882e-01 3.18945915e-01 6.39160514e-01
-8.11259747e-01 -1.18171982e-01 5.46141982e-01 -3.16508591e-01
1.62032500e-01 4.82659310e-01 -3.74912888e-01 -1.18818784e+00
-5.57623506e-01 -9.31905687e-01 9.82100487e-01 -7.83733368e-01
-1.85138094e+00 7.46269405e-01 -5.82615584e-02 -7.17358530e-01
-3.37103873e-01 -6.56453669e-01 -2.60319322e-01 1.14543533e+00
-1.78095078e+00 -1.84360588e+00 -2.57188063e-02 6.74373269e-01
1.62966385e-01 -2.19812810e-01 1.45922112e+00 3.69044423e-01
-6.15906358e-01 8.19835901e-01 6.84135854e-02 1.03781380e-01
1.12039149e+00 -1.26060653e+00 5.40579915e-01 4.18583184e-01
-2.25917548e-01 1.18640769e+00 5.19261301e-01 -6.92387044e-01
-1.17119849e+00 -1.47619140e+00 7.33256817e-01 -9.12800312e-01
4.10253853e-01 -2.99859822e-01 -8.43135059e-01 1.33592451e+00
4.15297031e-01 1.51933003e-02 1.52024055e+00 4.10592705e-01
-3.06710780e-01 1.25891864e-01 -1.21154809e+00 1.73105031e-01
1.14498842e+00 -5.82874417e-01 -8.65185797e-01 4.36253488e-01
5.27690768e-01 -5.25139093e-01 -1.46810102e+00 7.68110096e-01
4.40275848e-01 -2.70341128e-01 1.29222906e+00 -1.32144582e+00
7.14576483e-01 3.35829347e-01 1.59018219e-01 -1.56198776e+00
-7.29935288e-01 -4.78505552e-01 -9.43581462e-02 9.69840884e-01
8.26559007e-01 -7.65118778e-01 5.03962994e-01 2.81421661e-01
-6.13640845e-01 -1.12317824e+00 -9.35637951e-01 -2.44686902e-01
7.00111389e-01 -9.03795585e-02 4.40168500e-01 1.20201576e+00
1.64998874e-01 6.72020853e-01 -3.67180407e-01 -1.79866269e-01
6.00782573e-01 1.17239349e-01 3.37182492e-01 -1.11151934e+00
-5.91472626e-01 -1.39552474e-01 -2.73262441e-01 -6.17479801e-01
1.66382939e-01 -1.39115238e+00 -1.50649965e-01 -1.90670025e+00
8.06779206e-01 -2.18128324e-01 -9.63706136e-01 9.62699831e-01
-6.76013708e-01 6.39457166e-01 -3.91563058e-01 -5.55525981e-02
-4.72722322e-01 3.36983502e-01 1.46403730e+00 -3.35343182e-01
3.27199697e-01 -3.06180716e-01 -1.50896335e+00 2.59698600e-01
5.45544386e-01 -3.45840007e-01 -5.01310349e-01 -7.30576932e-01
-1.97685212e-02 -2.14101613e-01 3.69410843e-01 -6.66425467e-01
-1.73134312e-01 4.78920974e-02 9.87166584e-01 -3.65170538e-01
3.81376922e-01 -1.62511095e-01 2.86551505e-01 5.12449622e-01
-4.51743066e-01 -8.81837681e-02 6.36981189e-01 6.35344326e-01
1.09666258e-01 5.60325742e-01 6.59954131e-01 -3.79241675e-01
-2.63466924e-01 2.98962623e-01 -2.49545902e-01 3.08782846e-01
8.65445137e-01 2.30548471e-01 -7.67094433e-01 -1.61061957e-01
-1.25608933e+00 2.94680238e-01 7.88544565e-02 3.09923410e-01
6.52034044e-01 -1.16399121e+00 -1.19162464e+00 8.34672004e-02
2.79572994e-01 3.01908609e-02 5.93999207e-01 9.76122797e-01
6.16647489e-02 7.42472827e-01 -1.20796345e-01 -7.61375546e-01
-9.10289168e-01 5.21165073e-01 4.28313106e-01 -4.96691495e-01
-6.41288042e-01 1.12054026e+00 6.61327600e-01 -4.86230999e-01
-2.06705034e-01 -4.94045377e-01 9.69277173e-02 -3.21500808e-01
4.12531346e-01 -1.67501897e-01 2.90258825e-01 -3.53036970e-01
-5.69409370e-01 2.06552804e-01 -4.17853117e-01 1.49333596e-01
1.57059443e+00 1.73319265e-01 1.83876436e-02 3.66827339e-01
1.10548961e+00 -5.09511054e-01 -7.69720495e-01 -3.30433287e-02
-4.01934445e-01 3.22511494e-01 -2.82962006e-02 -1.54980803e+00
-1.04436052e+00 8.52505088e-01 3.81425887e-01 -5.95655143e-01
1.21771777e+00 5.67031860e-01 9.45660651e-01 2.59036720e-01
3.09279561e-01 -5.95039129e-01 -8.11061412e-02 1.69420972e-01
7.16078639e-01 -1.13643181e+00 -1.25380665e-01 -2.97225028e-01
-6.72162712e-01 7.11290836e-01 5.53943038e-01 3.26694310e-01
5.58713734e-01 4.45887357e-01 2.73921788e-01 -5.16389132e-01
-9.68170464e-01 9.37129110e-02 4.73474413e-01 7.72716045e-01
1.01831770e+00 -8.70577525e-03 -1.76370338e-01 9.99780118e-01
-5.23679368e-02 4.34126019e-01 -1.80457324e-01 8.96493673e-01
1.29026055e-01 -1.16932237e+00 1.10458180e-01 5.49650252e-01
-9.92925286e-01 -5.22761047e-01 -3.00482959e-01 6.22335672e-01
1.07654549e-01 5.53309679e-01 -2.70015031e-01 -2.13527218e-01
1.53372765e-01 4.36753184e-01 7.19550967e-01 -8.73528659e-01
-6.45353138e-01 1.26708299e-01 8.57651904e-02 -4.27993000e-01
-4.24382329e-01 -6.40447199e-01 -1.24696457e+00 2.97934473e-01
5.25608752e-03 -6.94284588e-02 1.86521098e-01 1.01113129e+00
9.83422935e-01 1.13258410e+00 -1.59501329e-01 -4.48461562e-01
-4.76687074e-01 -1.26624990e+00 -5.05742840e-02 5.67633092e-01
3.85281742e-01 -4.90526408e-01 -6.49861544e-02 2.92523354e-01] | [8.563766479492188, 8.709541320800781] |
8ad30aff-5faf-47b0-a998-217bc507bab9 | progressive-multi-stage-interactive-training | 2112.04223 | null | https://arxiv.org/abs/2112.04223v1 | https://arxiv.org/pdf/2112.04223v1.pdf | Progressive Multi-stage Interactive Training in Mobile Network for Fine-grained Recognition | Fine-grained Visual Classification (FGVC) aims to identify objects from subcategories. It is a very challenging task because of the subtle inter-class differences. Existing research applies large-scale convolutional neural networks or visual transformers as the feature extractor, which is extremely computationally expensive. In fact, real-world scenarios of fine-grained recognition often require a more lightweight mobile network that can be utilized offline. However, the fundamental mobile network feature extraction capability is weaker than large-scale models. In this paper, based on the lightweight MobilenetV2, we propose a Progressive Multi-Stage Interactive training method with a Recursive Mosaic Generator (RMG-PMSI). First, we propose a Recursive Mosaic Generator (RMG) that generates images with different granularities in different phases. Then, the features of different stages pass through a Multi-Stage Interaction (MSI) module, which strengthens and complements the corresponding features of different stages. Finally, using the progressive training (P), the features extracted by the model in different stages can be fully utilized and fused with each other. Experiments on three prestigious fine-grained benchmarks show that RMG-PMSI can significantly improve the performance with good robustness and transferability. | ['Yang Yu', 'Chengkai Zhu', 'Yinqi Zhang', 'Yifeng Liu', 'Qingliang Chen', 'Zhenxin Wu'] | 2021-12-08 | null | null | null | null | ['fine-grained-image-classification'] | ['computer-vision'] | [ 9.14748609e-02 -5.44487119e-01 3.04597765e-02 -2.58091152e-01
-3.01702946e-01 -4.97298390e-01 6.00309014e-01 -3.27836186e-01
-3.12950253e-01 5.36496580e-01 -6.13771789e-02 -1.81477979e-01
-1.28783494e-01 -1.12240398e+00 -6.84289694e-01 -7.80424237e-01
2.35740930e-01 6.89355433e-02 5.94852984e-01 -1.53951317e-01
2.01208174e-01 5.56760490e-01 -1.84080172e+00 6.61603451e-01
9.69222903e-01 1.46274364e+00 4.28609043e-01 5.52016973e-01
-4.00524437e-01 8.34434271e-01 -5.52116871e-01 -1.20954633e-01
2.61478543e-01 -2.08330646e-01 -6.34562910e-01 1.88604429e-01
3.18080932e-01 -3.90707344e-01 -2.04006821e-01 1.05457306e+00
3.97699505e-01 -1.20967664e-02 4.93940234e-01 -1.40327966e+00
-4.46656138e-01 6.06281698e-01 -5.96278608e-01 9.78889540e-02
-1.47096485e-01 3.49705160e-01 6.24481261e-01 -1.10524726e+00
3.93053442e-01 1.21662951e+00 6.56877458e-01 3.31104517e-01
-9.29701686e-01 -8.48453820e-01 3.51257175e-01 3.81374955e-01
-1.65881491e+00 -2.77206391e-01 7.95078039e-01 -3.75029624e-01
6.66307211e-01 4.06661242e-01 8.93067241e-01 8.06279063e-01
-1.00525066e-01 5.58002651e-01 1.18540812e+00 -1.48562208e-01
9.42397956e-03 3.16063873e-02 1.56410530e-01 8.22989821e-01
2.47075945e-01 2.36873087e-02 -4.23354685e-01 3.26101512e-01
9.10557508e-01 3.31861407e-01 -2.75880247e-01 -2.74482965e-01
-1.23401511e+00 6.28320813e-01 7.87953496e-01 3.90911818e-01
-5.45638204e-01 1.11488983e-01 1.48391068e-01 2.37470314e-01
3.10261756e-01 6.65285736e-02 -2.92161793e-01 1.19813293e-01
-1.10247827e+00 -1.95880011e-02 5.74206829e-01 8.12094629e-01
1.15639603e+00 5.64904921e-02 -5.32088697e-01 8.26477706e-01
3.13185602e-01 2.76892453e-01 6.66439116e-01 -6.44397318e-01
5.28771222e-01 1.06942141e+00 -2.11042643e-01 -1.14060283e+00
-2.33724013e-01 -7.34756291e-01 -1.23393786e+00 3.68328393e-01
3.21233213e-01 9.86467209e-03 -1.01978970e+00 1.36051548e+00
4.99110639e-01 3.31662118e-01 -1.67155221e-01 9.86835659e-01
1.20123899e+00 7.15420544e-01 1.25271574e-01 9.55507606e-02
1.38153565e+00 -1.27431595e+00 -1.21914059e-01 -1.51250929e-01
2.99820989e-01 -5.93461752e-01 1.08996367e+00 2.33773917e-01
-7.02748060e-01 -9.44091022e-01 -1.05122697e+00 2.18818374e-02
-5.24867594e-01 4.00719106e-01 6.93909943e-01 4.25672680e-01
-9.96335208e-01 4.65974003e-01 -4.10679758e-01 -4.58200388e-02
7.08946288e-01 3.37450653e-01 -4.00709599e-01 -1.79660171e-02
-1.00694072e+00 4.10036534e-01 6.30827844e-01 4.47725564e-01
-1.07654202e+00 -6.49269640e-01 -7.38443911e-01 2.73711741e-01
3.51583004e-01 -6.50773764e-01 9.09556329e-01 -1.11267316e+00
-1.40484583e+00 5.81539810e-01 4.64720950e-02 -1.53920859e-01
6.94155514e-01 8.01065639e-02 -2.40859717e-01 1.12986445e-01
3.60066034e-02 7.48717904e-01 1.02196348e+00 -1.14918840e+00
-9.74671602e-01 -2.02521995e-01 3.92268360e-01 2.36488938e-01
-5.01306057e-01 -8.85901898e-02 -7.55539656e-01 -6.78526878e-01
-4.32825387e-02 -7.49058843e-01 -1.81056440e-01 -1.11836612e-01
-5.32507658e-01 -2.84677684e-01 9.55051422e-01 -4.46754992e-01
1.28878999e+00 -2.26313281e+00 7.07056969e-02 4.51634437e-01
5.01615286e-01 4.68681812e-01 -2.99024254e-01 4.59174402e-02
-2.07069311e-02 8.87053460e-02 5.57960346e-02 1.05933454e-02
-5.27938753e-02 -1.17479853e-01 -4.89016585e-02 7.65493140e-02
2.71719456e-01 1.26996422e+00 -6.18510187e-01 -6.24185383e-01
4.94917423e-01 2.94234812e-01 -3.54787141e-01 1.03892811e-01
-3.90074439e-02 1.75968617e-01 -4.30526227e-01 8.34405363e-01
8.06389332e-01 -6.03468180e-01 7.63922781e-02 -6.94279373e-01
-2.54308015e-01 -2.65463084e-01 -1.39524615e+00 1.39551318e+00
-6.27455711e-01 5.02226591e-01 -3.73665802e-02 -9.25018191e-01
8.60920370e-01 -1.45909116e-01 9.38680097e-02 -6.97910130e-01
2.89767414e-01 1.87696517e-01 1.12799474e-03 -3.09988081e-01
4.90763158e-01 2.14650497e-01 -1.21527083e-01 2.85965264e-01
3.57777532e-03 1.83829367e-01 2.21520826e-01 1.11421660e-01
9.35920954e-01 1.23979151e-01 2.18695134e-01 -7.30728731e-02
6.81663573e-01 -6.52353317e-02 6.68600798e-01 7.34219849e-01
-6.14737868e-02 6.24169409e-01 1.75819188e-01 -5.58159351e-01
-8.51708353e-01 -8.80868971e-01 1.66077331e-01 1.06278312e+00
5.23159325e-01 -3.74270111e-01 -7.40761757e-01 -9.13624763e-01
3.57036479e-02 -8.25802535e-02 -7.58009493e-01 -1.27851248e-01
-4.06081825e-01 -7.50748575e-01 4.36322242e-01 6.66059077e-01
1.06735909e+00 -1.36281431e+00 -6.95342124e-01 1.60707742e-01
-2.07976177e-01 -8.95455897e-01 -3.01145226e-01 7.14961067e-02
-6.40585065e-01 -9.88201618e-01 -8.18439186e-01 -8.72238278e-01
6.68383956e-01 7.12608218e-01 9.07765090e-01 4.77393985e-01
-1.99685141e-01 -3.01894486e-01 -4.38165516e-01 -1.46638408e-01
-1.41211227e-02 2.22703636e-01 -2.94572979e-01 3.75249684e-01
2.40845487e-01 -5.05413890e-01 -6.90102518e-01 4.00602728e-01
-9.79381740e-01 5.50125957e-01 9.35512662e-01 9.84378040e-01
7.29986787e-01 3.12069774e-01 4.13210362e-01 -8.86931837e-01
4.51959580e-01 -2.96296060e-01 -5.86568952e-01 4.20376986e-01
-4.31278557e-01 -6.15887716e-02 8.47390294e-01 -6.06101930e-01
-1.16639471e+00 3.24325487e-02 -1.21470116e-01 -6.41106009e-01
-3.47829133e-01 4.43950772e-01 -4.86049294e-01 -3.70540738e-01
5.32819688e-01 3.86928797e-01 -3.62231672e-01 -3.92790765e-01
3.62578571e-01 8.65256429e-01 4.61758494e-01 -2.16871470e-01
1.14863706e+00 3.47138584e-01 -2.86907434e-01 -5.68540990e-01
-7.14903414e-01 -1.77508399e-01 -5.38485229e-01 -3.77884895e-01
7.64775038e-01 -1.02360272e+00 -7.58943319e-01 8.76633525e-01
-8.99519324e-01 -5.38447917e-01 -3.30101103e-01 2.45801225e-01
-1.20347783e-01 1.91117018e-01 -5.65670192e-01 -3.20867866e-01
-5.76179981e-01 -1.18070650e+00 1.09745574e+00 7.16487586e-01
1.85437545e-01 -5.85206628e-01 -4.47242975e-01 2.69345045e-01
6.16025567e-01 1.49600431e-01 6.59095287e-01 -1.37621894e-01
-8.76133025e-01 4.12815847e-02 -6.72014654e-01 3.11872452e-01
1.79001674e-01 1.36533126e-01 -9.68810022e-01 -1.61208287e-01
-4.80151772e-01 -2.50606596e-01 1.04028010e+00 1.90204114e-01
1.61099136e+00 -2.86888152e-01 -4.18459684e-01 1.03896511e+00
1.54907763e+00 1.69603780e-01 6.47296607e-01 5.56012571e-01
1.18257618e+00 3.39697808e-01 7.16310203e-01 3.56151223e-01
4.55352247e-01 5.03906012e-01 4.93901193e-01 -3.27738196e-01
-4.05472040e-01 -1.01579666e-01 1.57947406e-01 7.25839496e-01
-2.90097475e-01 -3.74732800e-02 -7.46335566e-01 5.04159749e-01
-1.82215738e+00 -8.96550000e-01 -5.85566387e-02 1.80002248e+00
5.70637405e-01 1.09183192e-01 -5.84746748e-02 2.45061234e-01
8.73910487e-01 2.00900450e-01 -5.67041576e-01 5.35711413e-03
-1.88859940e-01 3.19372475e-01 3.50254178e-01 -4.33996283e-02
-1.11899757e+00 1.11136651e+00 4.92614508e+00 1.40367782e+00
-1.39770269e+00 9.30842012e-02 6.64902329e-01 1.06241502e-01
-2.02913374e-01 -1.12777278e-01 -7.11346567e-01 6.74369216e-01
3.92915994e-01 9.19200256e-02 5.62899172e-01 9.32203770e-01
-1.10794842e-01 -1.68700330e-02 -6.79130077e-01 1.14387727e+00
-1.73660636e-01 -1.58438718e+00 2.16015711e-01 -2.83279810e-02
7.25624084e-01 7.73567241e-03 -4.03848998e-02 3.45103621e-01
1.60765335e-01 -9.09797192e-01 9.98915076e-01 5.82951903e-01
1.17656720e+00 -8.78839374e-01 7.60741949e-01 4.47154820e-01
-1.77792382e+00 -4.14727718e-01 -5.25059164e-01 1.08505040e-01
-4.90472466e-02 7.69240618e-01 -3.76653045e-01 9.18383062e-01
9.98861670e-01 6.51437461e-01 -7.91344285e-01 1.04892957e+00
-2.57724822e-01 4.33278024e-01 -1.70516759e-01 -1.39688611e-01
1.82242125e-01 -8.17888677e-02 1.77462891e-01 1.19636476e+00
3.73006433e-01 -1.04436345e-01 1.41430601e-01 6.24806404e-01
-2.47272372e-01 6.63555264e-02 -3.89170080e-01 3.60338166e-02
4.35340881e-01 1.91510129e+00 -9.00427878e-01 -6.77666426e-01
-4.32094306e-01 9.65472639e-01 4.90576565e-01 2.43862599e-01
-8.57106388e-01 -6.82118833e-01 4.89453375e-01 -5.78935258e-02
6.57946408e-01 1.54358428e-03 -2.26755187e-01 -1.42540920e+00
1.10343978e-01 -1.10100210e+00 3.02186877e-01 -7.52296448e-01
-1.22477746e+00 9.14391458e-01 -4.45216119e-01 -1.48923945e+00
1.28535996e-03 -4.55255300e-01 -6.81211948e-01 9.04998660e-01
-1.63497710e+00 -1.52848351e+00 -1.12560213e+00 9.42788601e-01
5.71782291e-01 -8.21304768e-02 4.99292314e-01 4.70290244e-01
-6.72573745e-01 6.82784677e-01 -2.66811848e-01 3.31117809e-01
2.59906054e-01 -9.69781101e-01 2.97787368e-01 9.39786732e-01
8.92705768e-02 4.12926942e-01 -6.80470392e-02 -5.26679814e-01
-1.14055049e+00 -1.69781005e+00 5.75611591e-01 3.28194499e-02
3.25905621e-01 -4.15490419e-01 -8.21958005e-01 3.10253143e-01
3.17368917e-02 2.40872145e-01 2.93395638e-01 -2.43995219e-01
-3.75466973e-01 -5.09944856e-01 -9.52889323e-01 5.32161593e-01
1.22368026e+00 -5.31768680e-01 -3.45391661e-01 6.33571893e-02
6.26049280e-01 -2.66292036e-01 -6.44051433e-01 5.05672812e-01
7.80386806e-01 -1.08735657e+00 9.91315663e-01 -3.22144449e-01
4.21112299e-01 -6.51169419e-01 -1.24948382e-01 -1.26578438e+00
-5.89831293e-01 -1.61493853e-01 -2.46585384e-02 1.35699487e+00
8.89576748e-02 -6.70010746e-01 7.20169902e-01 9.03777629e-02
-2.63123289e-02 -7.74905682e-01 -5.39461076e-01 -6.88609183e-01
-3.44767004e-01 -4.47837055e-01 1.07891035e+00 8.37330639e-01
-5.18804491e-01 2.36022592e-01 -3.07486027e-01 1.15058675e-01
5.15805483e-01 7.42790759e-01 9.35131073e-01 -1.35303187e+00
-3.41025889e-01 -4.14310455e-01 -4.15473610e-01 -9.67092216e-01
-6.89795837e-02 -8.22238147e-01 1.42037258e-01 -1.58658564e+00
4.48452502e-01 -6.44321501e-01 -3.64563316e-01 5.87542415e-01
-2.75896013e-01 8.37209404e-01 5.33968508e-01 2.69338906e-01
-7.40443051e-01 5.12758732e-01 1.36847770e+00 -3.90319794e-01
-1.71918780e-01 -1.42690539e-01 -8.31701398e-01 8.59153986e-01
8.01548302e-01 -2.45707795e-01 -3.35652441e-01 -3.08481693e-01
-1.90560110e-02 -2.88369507e-01 5.53366482e-01 -1.40541267e+00
2.76762933e-01 -2.00443774e-01 8.00438881e-01 -7.48401940e-01
1.63292080e-01 -7.09973693e-01 4.81337607e-01 4.33979750e-01
1.02125719e-01 -1.08002119e-01 1.02832310e-01 3.07087272e-01
-4.14845973e-01 6.30332576e-03 5.87530732e-01 -2.11545095e-01
-1.03467262e+00 6.50437832e-01 -2.49216706e-01 -2.43858591e-01
1.16904724e+00 -3.61095458e-01 -4.30574059e-01 7.91912246e-03
-5.43898463e-01 8.74937475e-02 4.66386169e-01 3.98327172e-01
6.86865866e-01 -1.63884485e+00 -4.78015304e-01 3.29860568e-01
1.48141682e-01 1.94018498e-01 6.54143512e-01 7.47241080e-01
-4.81225580e-01 1.82807297e-01 -4.45590258e-01 -6.91121697e-01
-1.30487645e+00 5.08704960e-01 2.76677012e-01 -4.18238759e-01
-4.32186514e-01 7.95325100e-01 5.98361194e-01 -3.45567852e-01
-4.52421941e-02 -2.88430512e-01 -5.22731721e-01 2.98910975e-01
5.42839348e-01 3.19819510e-01 1.27081513e-01 -6.79265440e-01
-3.39652866e-01 7.46489167e-01 -9.57286060e-02 1.69959351e-01
1.37110436e+00 -1.07045181e-01 -1.35752574e-01 5.20185679e-02
1.08375025e+00 -9.95705351e-02 -1.43552899e+00 -3.41186583e-01
-3.60810339e-01 -5.62686980e-01 -8.46043304e-02 -5.77435315e-01
-1.46000040e+00 9.41858590e-01 7.04973996e-01 2.98108608e-01
1.58595622e+00 -1.36947364e-01 7.31964231e-01 1.96201622e-01
4.52285767e-01 -1.01483929e+00 6.13784343e-02 4.95269656e-01
7.19188392e-01 -9.65718448e-01 -3.49240035e-01 -5.30719936e-01
-4.33637500e-01 1.06679022e+00 9.31904674e-01 -8.92974883e-02
6.56004310e-01 1.71783507e-01 1.11734755e-01 -9.82288718e-02
-5.18034756e-01 -4.02745754e-01 3.99639487e-01 5.34605861e-01
-5.58503121e-02 9.89135429e-02 -1.96456537e-01 7.91629970e-01
-9.48708877e-02 2.81958252e-01 1.90384582e-01 7.84382403e-01
-4.23742533e-01 -9.46036100e-01 -2.45632336e-01 7.01845407e-01
-2.20647201e-01 -2.98329920e-01 -1.29872173e-01 5.92839777e-01
6.93537891e-01 9.09455836e-01 9.30352882e-02 -9.21533883e-01
2.19776973e-01 -2.82629818e-01 1.93661615e-01 -5.25189698e-01
-7.78062224e-01 -2.76288688e-02 -1.14178129e-01 -9.03124988e-01
-4.76087481e-01 -2.68673122e-01 -1.07379544e+00 -3.78675789e-01
-3.81164789e-01 -1.71088148e-02 4.93438333e-01 9.73910868e-01
5.27676344e-01 8.64280641e-01 7.62026489e-01 -1.12849593e+00
-1.53178811e-01 -9.69585359e-01 -5.22792995e-01 3.75673503e-01
1.97184786e-01 -7.74893463e-01 -1.09414749e-01 -4.57790233e-02] | [9.644554138183594, 1.9609624147415161] |
076c1ce9-6369-4232-ab67-270087904985 | a-multi-stage-deep-architecture-for-summary | 2205.00694 | null | https://arxiv.org/abs/2205.00694v1 | https://arxiv.org/pdf/2205.00694v1.pdf | A Multi-stage deep architecture for summary generation of soccer videos | Video content is present in an ever-increasing number of fields, both scientific and commercial. Sports, particularly soccer, is one of the industries that has invested the most in the field of video analytics, due to the massive popularity of the game and the emergence of new markets. Previous state-of-the-art methods on soccer matches video summarization rely on handcrafted heuristics to generate summaries which are poorly generalizable, but these works have yet proven that multiple modalities help detect the best actions of the game. On the other hand, machine learning models with higher generalization potential have entered the field of summarization of general-purpose videos, offering several deep learning approaches. However, most of them exploit content specificities that are not appropriate for sport whole-match videos. Although video content has been for many years the main source for automatizing knowledge extraction in soccer, the data that records all the events happening on the field has become lately very important in sports analytics, since this event data provides richer context information and requires less processing. We propose a method to generate the summary of a soccer match exploiting both the audio and the event metadata. The results show that our method can detect the actions of the match, identify which of these actions should belong to the summary and then propose multiple candidate summaries which are similar enough but with relevant variability to provide different options to the final editor. Furthermore, we show the generalization capability of our work since it can transfer knowledge between datasets from different broadcasting companies, different competitions, acquired in different conditions, and corresponding to summaries of different lengths | ['Thomas Menguy', 'Pierre-Alexandre Mattei', 'Frédéric Precioso', 'Melissa Sanabria'] | 2022-05-02 | null | null | null | null | ['sports-analytics'] | ['computer-vision'] | [ 5.46186976e-02 -1.56630114e-01 -4.30229723e-01 -6.65312856e-02
-7.90044487e-01 -4.32958812e-01 4.94774312e-01 5.38714588e-01
-4.52488929e-01 7.13683784e-01 5.75388372e-01 3.64569813e-01
-4.24337059e-01 -7.78356194e-01 -7.71017730e-01 -5.39768457e-01
-5.92632480e-02 4.39041227e-01 7.20017552e-01 -5.09756446e-01
4.08887714e-01 3.69400114e-01 -2.24308705e+00 7.67602921e-01
5.42454243e-01 1.13015985e+00 3.30763370e-01 6.53673649e-01
-5.71041517e-02 8.66010725e-01 -8.22175026e-01 -5.38226306e-01
1.79194644e-01 -5.73472381e-01 -5.15573800e-01 -1.08304610e-02
3.70197594e-01 -3.46171595e-02 -5.47701061e-01 8.48152399e-01
5.74258566e-01 2.37056434e-01 2.39961296e-01 -1.21088302e+00
2.76544780e-01 1.01943219e+00 -1.87230766e-01 3.85473520e-01
7.86833286e-01 -2.08043009e-02 1.03288627e+00 -3.29104096e-01
9.45731640e-01 7.94150770e-01 5.65353751e-01 2.04619065e-01
-7.24477470e-01 -6.04253769e-01 5.82501292e-02 9.52872932e-01
-1.03679192e+00 -3.02793771e-01 8.27811420e-01 -3.74585807e-01
5.86611927e-01 4.69507873e-01 1.18598735e+00 1.25069952e+00
1.57501757e-01 9.86998379e-01 8.40003431e-01 -2.76518762e-01
2.79271305e-01 8.33063852e-03 -3.96459475e-02 1.50525182e-01
1.26120299e-01 -3.20976526e-01 -1.07753146e+00 1.20703742e-01
2.88433433e-01 -6.98185945e-03 -3.31402689e-01 -1.68879405e-01
-1.19211483e+00 7.39975214e-01 -1.70639828e-01 5.08560777e-01
-6.02128029e-01 -1.00500725e-01 9.40182269e-01 1.99972421e-01
2.22224578e-01 5.21302581e-01 -9.05985311e-02 -6.49402380e-01
-1.40232134e+00 8.63370478e-01 9.00304317e-01 6.88115299e-01
5.50654233e-01 -3.20449062e-02 -3.37624431e-01 6.72608554e-01
-4.11452055e-01 1.38248175e-01 6.35189235e-01 -7.91581631e-01
5.48304200e-01 6.76399112e-01 -1.34710908e-01 -1.33984947e+00
-4.52297956e-01 -6.46062493e-01 -5.91542959e-01 -1.68080673e-01
3.55410069e-01 3.86002325e-02 -3.30142885e-01 1.41487956e+00
8.94409344e-02 3.07202548e-01 -1.08540416e-01 9.24801230e-01
1.05518305e+00 7.32338905e-01 -1.87816054e-01 -2.75923789e-01
1.46496367e+00 -5.48067868e-01 -5.89893043e-01 -2.31338143e-01
2.71047562e-01 -7.03420341e-01 7.00233698e-01 8.63286614e-01
-1.05914724e+00 -6.06415629e-01 -1.14577937e+00 2.02129543e-01
-2.28308693e-01 7.82090724e-02 5.63375473e-01 3.58917922e-01
-6.50732040e-01 7.52346516e-01 -4.55847919e-01 -5.18605888e-01
2.40745157e-01 1.01779111e-01 -4.89421368e-01 -1.31895483e-01
-1.26009095e+00 7.48387158e-01 8.90092134e-01 -9.69182625e-02
-7.77301788e-01 -6.15078270e-01 -6.03686512e-01 2.68736839e-01
9.47301805e-01 -3.80206496e-01 9.72996652e-01 -1.21030307e+00
-1.36165154e+00 6.55388236e-01 1.85971975e-01 -8.67434323e-01
6.17276490e-01 -2.72116244e-01 -4.93468285e-01 3.70835781e-01
8.36919546e-02 3.40471685e-01 6.56906009e-01 -7.69305766e-01
-1.13990045e+00 -1.72482893e-01 2.44768560e-01 2.43238300e-01
-4.97905970e-01 1.49515137e-01 -6.57268763e-01 -7.35983372e-01
-9.40886736e-02 -7.88594186e-01 -7.11813867e-02 -6.74881697e-01
-3.93511742e-01 -1.33039057e-01 5.71294487e-01 -8.38037252e-01
1.64919019e+00 -2.13502812e+00 5.29485583e-01 4.60506864e-02
-1.33092143e-02 3.26321483e-01 1.90619364e-01 8.33060026e-01
1.95495114e-01 -1.66629881e-01 1.64169818e-01 -1.01278853e-02
-2.56905090e-02 1.28975242e-01 -4.28389311e-01 1.90271214e-01
-2.13878423e-01 4.07277256e-01 -9.78674054e-01 -6.68464780e-01
8.46509784e-02 1.25065371e-01 -5.98494887e-01 8.66251364e-02
-3.94284487e-01 4.51697856e-01 -3.98919076e-01 3.20539594e-01
2.26593405e-01 2.89930224e-01 2.00508356e-01 -4.31005508e-01
-2.96899050e-01 1.99165002e-01 -1.44597399e+00 1.87429702e+00
-1.24417059e-01 7.95038283e-01 -1.89292505e-01 -1.41886854e+00
8.20998251e-01 3.82457703e-01 8.70927393e-01 -6.48901582e-01
2.24404246e-01 2.06741378e-01 -1.72026470e-01 -8.83312285e-01
1.06696987e+00 -8.43579136e-03 -3.39201689e-01 1.33949876e-01
3.04908305e-01 1.45413808e-03 1.08036292e+00 1.43963620e-01
1.07082820e+00 3.79641354e-01 3.90958935e-01 2.69034594e-01
5.22918522e-01 1.82070136e-01 8.00180733e-01 8.50924611e-01
1.52853891e-01 8.16670001e-01 6.95077956e-01 -4.17536229e-01
-9.80335951e-01 -6.72593176e-01 3.02840799e-01 1.00344491e+00
1.88115761e-01 -7.48300314e-01 -7.26571679e-01 -2.97537863e-01
-3.39290589e-01 4.43309665e-01 -2.65040368e-01 -2.14957073e-01
-6.71838164e-01 -4.07480210e-01 5.09721398e-01 2.46036977e-01
5.22084653e-01 -1.21078515e+00 -8.86946142e-01 4.71273959e-01
-7.11807966e-01 -1.29445994e+00 -5.97776882e-02 -1.35898620e-01
-7.55333006e-01 -1.12750363e+00 -5.23447514e-01 -3.97799730e-01
-2.29817569e-01 3.61530706e-02 1.17565036e+00 -1.27264127e-01
-2.94702023e-01 4.26647514e-01 -7.66212761e-01 -5.35393655e-01
-6.15006089e-01 4.30019855e-01 -3.46632227e-02 3.50549221e-01
3.50626647e-01 -5.82028091e-01 -2.80785263e-01 1.96050286e-01
-1.11291146e+00 1.82563722e-01 5.90824485e-01 4.22413886e-01
5.78670561e-01 2.94230551e-01 6.62353396e-01 -6.93123460e-01
4.74646986e-01 -5.64911008e-01 -2.80762315e-01 1.08654082e-01
7.19341785e-02 -9.00424868e-02 6.53848708e-01 -4.18010086e-01
-7.92105079e-01 -2.22443342e-02 -2.71515399e-01 -3.08414102e-01
-3.25134724e-01 7.31829524e-01 -2.25808844e-01 5.60167074e-01
7.28863895e-01 4.13934022e-01 -1.59666643e-01 -4.65210527e-01
8.84839147e-02 4.78485405e-01 6.82389677e-01 -5.09121478e-01
5.06881595e-01 4.35161382e-01 4.81141023e-02 -1.07021630e+00
-7.23591864e-01 -6.67843401e-01 -3.87502313e-01 -7.88743854e-01
8.44962656e-01 -7.98667252e-01 -6.81777477e-01 3.84422183e-01
-1.06152499e+00 3.63621771e-01 -4.97408777e-01 8.19211364e-01
-7.40291417e-01 5.25322318e-01 -3.96951944e-01 -5.40634453e-01
-1.00828297e-01 -1.17069018e+00 7.47270465e-01 4.12246317e-01
-3.30463260e-01 -3.95503402e-01 2.29197070e-02 4.71763730e-01
7.68222660e-02 4.72349346e-01 6.52774036e-01 -9.31707203e-01
-6.54056072e-01 -4.92102832e-01 3.00301939e-01 2.43520036e-01
-9.55666006e-02 6.93668351e-02 -7.44674146e-01 3.87662575e-02
-1.49693117e-01 -1.15043186e-02 8.79537821e-01 4.80187386e-01
1.07002342e+00 -1.59120828e-01 -1.66278034e-01 3.04563433e-01
1.19320798e+00 1.24599608e-02 8.92818630e-01 5.82480788e-01
4.08710122e-01 7.14623511e-01 9.21826601e-01 6.85838223e-01
2.39252552e-01 9.71688390e-01 6.30190670e-01 3.15681756e-01
-7.81818405e-02 -2.87633806e-01 6.55625224e-01 8.26186717e-01
-6.26416385e-01 -1.84557810e-01 -5.43268740e-01 6.64291799e-01
-2.14085293e+00 -1.62243807e+00 -8.52976143e-02 2.38621974e+00
5.49282372e-01 4.10466701e-01 6.53085470e-01 5.83389044e-01
7.28394866e-01 2.49692425e-01 -1.15126789e-01 -3.77331614e-01
-3.14304441e-01 2.73447365e-01 4.00470734e-01 -2.24043056e-01
-1.11313665e+00 7.02039778e-01 5.15674496e+00 1.30651987e+00
-1.14417744e+00 -1.25870585e-01 1.43080488e-01 -4.44169104e-01
-7.67429173e-02 -8.43464285e-02 -8.33109617e-01 5.31484306e-01
8.38511229e-01 -2.62542099e-01 1.45781055e-01 8.82375658e-01
4.32387292e-01 -3.42079401e-01 -1.02762961e+00 9.96988714e-01
4.28869992e-01 -1.68093801e+00 1.27419263e-01 -8.34721401e-02
5.66686153e-01 -2.62242287e-01 -2.62054175e-01 3.70725691e-01
-2.60994613e-01 -6.43940151e-01 1.16728377e+00 7.47613072e-01
1.81736022e-01 -8.50612164e-01 9.55251276e-01 4.01430845e-01
-1.11338246e+00 -2.99783915e-01 -2.57951736e-01 -1.22015998e-01
3.70167583e-01 6.97075725e-01 -5.94238639e-01 9.83946621e-01
9.25202489e-01 8.28558564e-01 -5.99381685e-01 1.45888722e+00
-6.82427734e-02 5.95386207e-01 -2.23554030e-01 -1.29017800e-01
-1.19069666e-02 -8.02240297e-02 9.53283250e-01 1.18241405e+00
7.04221845e-01 -1.42655045e-01 3.45419943e-01 3.20074826e-01
1.89845577e-01 4.89801526e-01 -5.81093192e-01 -1.92735001e-01
1.76630139e-01 1.02609229e+00 -8.72175634e-01 -2.61059761e-01
-3.47549498e-01 5.38972080e-01 -1.45489007e-01 -1.13075329e-02
-9.92474318e-01 -2.99950629e-01 5.09296298e-01 5.62737584e-01
3.84392649e-01 -9.24933553e-02 1.22066878e-01 -1.13436723e+00
7.79602230e-02 -1.17917824e+00 6.01274252e-01 -5.76349795e-01
-9.14953947e-01 5.68943799e-01 3.18349808e-01 -1.68935633e+00
-5.35136163e-01 -3.31907660e-01 -3.90976548e-01 1.57784343e-01
-1.06373358e+00 -9.62767243e-01 -3.12037826e-01 3.81403178e-01
8.62622857e-01 -3.63672942e-01 4.14069206e-01 5.28519511e-01
-3.40210378e-01 1.88595593e-01 -7.73061216e-02 1.01803215e-02
7.59681165e-01 -7.05947042e-01 -2.28214040e-01 9.15385306e-01
4.14486200e-01 7.02183768e-02 1.04067254e+00 -6.14947498e-01
-1.28257918e+00 -6.29322290e-01 8.19617689e-01 -4.10209149e-02
6.45605624e-01 -1.69838071e-01 -7.17648327e-01 3.80632281e-01
1.59513935e-01 -5.31441212e-01 5.55181384e-01 5.33013744e-03
-8.81650252e-04 -5.36993921e-01 -5.67275167e-01 5.71712315e-01
8.97691488e-01 -2.29377478e-01 -7.10326195e-01 1.42496333e-01
3.32389474e-01 -5.04024565e-01 -6.66469693e-01 3.27106208e-01
5.94923675e-01 -1.40362287e+00 8.52200210e-01 -4.85025525e-01
7.33291268e-01 -2.52591759e-01 -8.08657333e-02 -1.28453636e+00
8.28079879e-02 -3.90577018e-01 7.13924915e-02 1.41429043e+00
1.25378132e-01 -3.02267224e-02 9.03442740e-01 7.40737021e-02
-4.02437419e-01 -4.99127895e-01 -9.47430730e-01 -5.98876953e-01
-5.16144753e-01 -7.01344550e-01 5.70164144e-01 5.37685215e-01
1.47229448e-01 1.22821271e-01 -6.75411105e-01 -1.47760138e-01
3.85509104e-01 2.71608323e-01 9.82372880e-01 -1.52482331e+00
-5.30863047e-01 -6.76782608e-01 -9.79024470e-01 -7.73411572e-01
4.20449786e-02 -7.53849626e-01 -5.33295572e-02 -1.56872809e+00
1.81718543e-01 6.86630281e-03 -1.27668768e-01 2.08959073e-01
1.19645312e-01 2.41814539e-01 4.44349557e-01 4.23181802e-02
-7.55529046e-01 1.16234280e-01 8.57494771e-01 -2.31443644e-01
-3.13134789e-01 3.38822424e-01 -3.73218775e-01 7.28182018e-01
6.19525373e-01 -4.42020416e-01 -2.26904094e-01 -3.94166857e-02
8.26661885e-01 1.50635958e-01 2.73668349e-01 -1.53614235e+00
3.09807718e-01 -1.71452269e-01 -7.78474510e-02 -6.98150218e-01
4.06454951e-01 -7.11178839e-01 5.40806949e-01 3.07478786e-01
-2.88008481e-01 -2.03478247e-01 1.49062291e-01 5.42513549e-01
-6.84231341e-01 -3.98302287e-01 3.97883445e-01 -3.00511569e-01
-1.11863828e+00 2.07254007e-01 -6.28249109e-01 5.12848720e-02
9.48332429e-01 -4.41249162e-01 -8.74356460e-03 -6.53801382e-01
-7.90488183e-01 -1.17313996e-01 2.03621432e-01 5.64849675e-01
3.84557813e-01 -1.00701439e+00 -9.29059088e-01 -2.43811339e-01
2.17653230e-01 -3.50650430e-01 6.75650477e-01 8.44080746e-01
-6.70004427e-01 2.35992193e-01 -5.43187916e-01 -6.93566263e-01
-1.38492882e+00 5.21358073e-01 -1.10521778e-01 -4.59987491e-01
-7.74036467e-01 3.75772208e-01 -2.55007744e-01 3.52483660e-01
2.66316175e-01 -2.67016709e-01 -7.96611786e-01 7.30160713e-01
6.53025925e-01 6.33716643e-01 3.13348621e-01 -7.55901635e-01
-1.27532750e-01 3.91303569e-01 1.24457642e-01 -3.20729092e-02
1.52612484e+00 1.26787022e-01 1.17728926e-01 5.51391602e-01
6.54060304e-01 2.18143910e-01 -9.71405864e-01 -1.18175283e-01
4.98300456e-02 -3.71505439e-01 -1.49118185e-01 -3.92383635e-01
-1.21692348e+00 7.62873948e-01 2.58543968e-01 4.92352992e-01
1.16202676e+00 -3.48459929e-02 8.72307181e-01 1.74345851e-01
5.98204434e-01 -1.41142178e+00 6.96068779e-02 3.50325286e-01
7.55714655e-01 -8.26328576e-01 2.65249819e-01 -3.68545473e-01
-7.77019441e-01 1.19980216e+00 1.87728956e-01 -7.25440532e-02
2.90167481e-01 1.54771343e-01 -2.71811306e-01 -2.87815705e-02
-5.09289086e-01 -3.56738716e-01 2.12462023e-01 5.25288463e-01
3.94922644e-02 -9.06097144e-02 -8.20605814e-01 9.71022248e-01
-4.25685346e-01 2.02160776e-01 8.54610145e-01 7.12656736e-01
-5.66965640e-01 -1.18894935e+00 -4.42449689e-01 4.75030214e-01
-6.68519557e-01 2.62462974e-01 -2.08199129e-01 8.10951173e-01
3.70110303e-01 8.46292019e-01 -1.35048434e-01 -4.77243513e-01
6.04990065e-01 1.04171462e-01 5.12243748e-01 -5.32861590e-01
-9.33535397e-01 -1.27324485e-04 2.38141671e-01 -5.92791498e-01
-6.26756251e-01 -8.78600776e-01 -9.88849878e-01 -3.54438663e-01
1.99088734e-02 4.03479367e-01 7.20348597e-01 1.00320673e+00
2.62120754e-01 5.92805922e-01 3.33194226e-01 -1.08870769e+00
-1.51359633e-01 -8.84780765e-01 -6.36761487e-01 6.76537633e-01
-1.73616871e-01 -6.75878167e-01 -1.04281388e-01 1.89435035e-01] | [8.14765739440918, 0.10782019048929214] |
6e84057b-9c21-402c-9d86-725d514ab43b | learning-language-specific-layers-for | 2305.02665 | null | https://arxiv.org/abs/2305.02665v1 | https://arxiv.org/pdf/2305.02665v1.pdf | Learning Language-Specific Layers for Multilingual Machine Translation | Multilingual Machine Translation promises to improve translation quality between non-English languages. This is advantageous for several reasons, namely lower latency (no need to translate twice), and reduced error cascades (e.g., avoiding losing gender and formality information when translating through English). On the downside, adding more languages reduces model capacity per language, which is usually countered by increasing the overall model size, making training harder and inference slower. In this work, we introduce Language-Specific Transformer Layers (LSLs), which allow us to increase model capacity, while keeping the amount of computation and the number of parameters used in the forward pass constant. The key idea is to have some layers of the encoder be source or target language-specific, while keeping the remaining layers shared. We study the best way to place these layers using a neural architecture search inspired approach, and achieve an improvement of 1.3 chrF (1.5 spBLEU) points over not using LSLs on a separate decoder architecture, and 1.9 chrF (2.2 spBLEU) on a shared decoder one. | ['Stephan Peitz', 'Yi-Hsiu Liao', 'Robin M. Schmidt', 'Telmo Pessoa Pires'] | 2023-05-04 | null | null | null | null | ['architecture-search'] | ['methodology'] | [-6.65254099e-03 2.71564186e-01 -2.12289557e-01 -2.45274305e-01
-7.84901798e-01 -6.00347340e-01 6.23281240e-01 2.20217794e-01
-7.02730179e-01 8.62178504e-01 8.49918574e-02 -5.95695972e-01
3.23793501e-01 -8.73277485e-01 -9.65855718e-01 -5.23241103e-01
3.99162203e-01 5.74583232e-01 1.23576978e-02 -3.22593540e-01
5.07246796e-03 3.43827099e-01 -1.16436422e+00 2.14775294e-01
1.06033766e+00 7.30851293e-01 4.25466359e-01 3.32125664e-01
-2.27898434e-01 5.71217656e-01 -3.64532411e-01 -9.58845854e-01
9.91006419e-02 -5.92058003e-01 -8.10154319e-01 -4.69292521e-01
2.98456430e-01 -2.27166668e-01 -8.49450231e-02 9.22367513e-01
4.94027674e-01 -1.49223357e-01 3.49065930e-01 -8.23433936e-01
-5.85042477e-01 1.06079435e+00 -4.47755873e-01 2.97447592e-02
7.80326575e-02 -9.17827487e-02 1.01202106e+00 -7.76298583e-01
5.69067061e-01 1.28656995e+00 5.19785404e-01 6.15052640e-01
-1.27563918e+00 -7.77724028e-01 1.41239092e-01 8.91491771e-02
-1.37212420e+00 -7.47820735e-01 4.24011707e-01 -1.80071786e-01
1.19459033e+00 2.87184238e-01 6.54016495e-01 1.09046972e+00
3.39059919e-01 7.29320347e-01 1.16800606e+00 -6.42628014e-01
-5.21897525e-02 4.70095873e-01 -2.87497252e-01 7.09998727e-01
3.68545115e-01 -1.33209914e-01 -4.50174868e-01 1.51050031e-01
7.38272071e-01 -3.44212025e-01 -1.20512724e-01 -6.81152344e-02
-1.17831802e+00 8.39873075e-01 4.83045667e-01 5.79020560e-01
-2.61796415e-01 6.45588487e-02 3.91770422e-01 5.79192758e-01
4.58341897e-01 6.81087911e-01 -6.64782345e-01 -2.71063000e-01
-1.17127573e+00 7.35111460e-02 7.32674539e-01 7.89003074e-01
8.86902273e-01 -1.19769517e-02 1.15131037e-02 1.05863023e+00
1.07578225e-01 5.72981179e-01 5.45656264e-01 -5.97562253e-01
7.36045480e-01 6.28081799e-01 -2.67918408e-01 -6.63302779e-01
-4.75060821e-01 -7.37862229e-01 -8.97824049e-01 -3.77706856e-01
5.20149529e-01 -1.26341760e-01 -6.44007027e-01 1.98217332e+00
-1.75369158e-01 -4.44306642e-01 -7.31151402e-02 6.50010526e-01
2.60578305e-01 7.82412052e-01 2.03308463e-02 -8.75936002e-02
1.44261479e+00 -1.06434286e+00 -4.03498411e-01 -6.12095773e-01
1.17497134e+00 -9.99497652e-01 1.27299130e+00 1.87193811e-01
-1.23625588e+00 -3.91277343e-01 -9.50066209e-01 -3.72247607e-01
-3.83452296e-01 5.17162979e-01 5.09937227e-01 5.87852180e-01
-1.21073496e+00 5.49888551e-01 -8.65203142e-01 -3.50697100e-01
1.21856734e-01 6.60733521e-01 -3.96691829e-01 -7.85435364e-02
-1.32739484e+00 1.29955328e+00 5.27542472e-01 -2.61726081e-02
-3.88490498e-01 -5.41383564e-01 -8.78227890e-01 3.57991576e-01
1.89581245e-01 -7.22149730e-01 1.02268505e+00 -1.11654985e+00
-1.70585191e+00 8.82172346e-01 -3.56372893e-01 -6.19365454e-01
5.77484012e-01 -1.80094570e-01 -2.21480519e-01 -3.30557227e-01
-1.12334099e-02 9.60433841e-01 3.76903385e-01 -8.05322170e-01
-5.52635014e-01 -2.49019861e-01 1.67741314e-01 1.68994263e-01
-6.49497390e-01 1.77502722e-01 -5.59416413e-01 -5.85617721e-01
-7.64709711e-02 -1.24466515e+00 -9.38313901e-02 -3.80907029e-01
-2.32744128e-01 -7.75549114e-02 1.87445030e-01 -1.11717248e+00
1.33016503e+00 -1.97768080e+00 4.10065114e-01 5.64337336e-02
-1.03488542e-01 2.92570561e-01 -2.46777698e-01 3.90939146e-01
1.83538586e-01 2.18927979e-01 -2.29741305e-01 -3.74726862e-01
-2.50610113e-01 1.20918408e-01 -2.24045496e-02 2.54933029e-01
3.65033209e-01 1.09134662e+00 -5.60447693e-01 -3.66786778e-01
-1.00567907e-01 5.13231099e-01 -7.80186176e-01 -2.07723647e-01
-1.42332062e-01 4.15972143e-01 -4.04905565e-02 2.84289539e-01
6.13825262e-01 8.00954029e-02 4.38576669e-01 7.12168887e-02
-2.84551710e-01 1.05493224e+00 -1.00027370e+00 1.72407186e+00
-1.20976436e+00 7.01398015e-01 -5.96187590e-03 -7.53511548e-01
1.12458241e+00 2.09399536e-01 2.50084884e-02 -1.11882162e+00
2.18863428e-01 7.03143775e-01 3.39430153e-01 9.41777527e-02
3.93685192e-01 -7.49863982e-02 -2.01887354e-01 5.88449776e-01
1.94181621e-01 1.87309846e-01 2.35101879e-01 -1.91070318e-01
6.34446383e-01 9.05706212e-02 1.33772939e-01 -6.19329035e-01
6.99093044e-01 -2.28537962e-01 5.52470565e-01 3.12687069e-01
3.33483130e-01 2.42341205e-01 6.40765309e-01 -3.20968837e-01
-1.29484844e+00 -6.70267045e-01 6.35387823e-02 1.09506536e+00
-1.73461616e-01 -5.57299435e-01 -8.71246338e-01 -3.69812846e-01
-3.37849975e-01 8.98281574e-01 -2.60495305e-01 -4.44478393e-01
-1.13140607e+00 -7.48626530e-01 8.14900637e-01 4.50093895e-01
5.37021220e-01 -6.80199742e-01 -3.51983815e-01 3.28515828e-01
-4.87941384e-01 -1.06316853e+00 -3.14737856e-01 2.39358351e-01
-1.06670141e+00 -3.76770824e-01 -9.19394791e-01 -7.34658003e-01
5.63348293e-01 -1.34820312e-01 1.18407631e+00 2.46298127e-02
2.41287902e-01 -6.21765852e-01 -9.30477232e-02 -3.08016390e-01
-5.21186531e-01 9.01455343e-01 4.87414487e-02 -1.27547011e-01
2.76400328e-01 -4.35835838e-01 -2.18767926e-01 1.15551017e-01
-5.66421926e-01 4.87533569e-01 7.77303278e-01 9.30274308e-01
2.33977288e-01 -2.84423500e-01 3.04272413e-01 -7.48309374e-01
4.44101989e-01 -2.16413483e-01 -6.42641783e-01 2.39479229e-01
-7.68340111e-01 3.56386840e-01 9.87183809e-01 -3.59472811e-01
-8.32695127e-01 -1.48412108e-01 -2.68715948e-01 -3.06598097e-02
3.06042910e-01 3.52850378e-01 -1.96678087e-01 1.07335672e-01
4.45865124e-01 4.03161675e-01 -6.66283146e-02 -6.98074043e-01
2.08420306e-01 7.01974869e-01 4.22397479e-02 -4.70744133e-01
4.72041309e-01 2.97604408e-03 -3.25993896e-01 -4.35620457e-01
-5.02256453e-01 -1.17522320e-02 -7.04717159e-01 2.21036285e-01
5.17904699e-01 -1.09102190e+00 -5.36420524e-01 3.07945162e-01
-1.17810440e+00 -3.64830345e-01 1.14193469e-01 5.97711384e-01
-3.40318859e-01 -8.57151672e-02 -8.87023985e-01 -4.61533666e-01
-4.41684633e-01 -1.31606162e+00 9.87197518e-01 3.48572358e-02
-4.04554129e-01 -8.74249756e-01 -2.22763106e-01 3.65464926e-01
6.30686462e-01 -3.25801283e-01 1.19471598e+00 -5.04596710e-01
-5.00630021e-01 4.67269914e-03 -2.92688608e-01 3.59301597e-01
-9.13093686e-02 -3.05853933e-01 -7.24185646e-01 -5.48118174e-01
-2.42983531e-02 -8.67094696e-02 7.78925240e-01 6.13472089e-02
7.21728504e-01 -5.41948199e-01 -2.87054628e-01 6.01660371e-01
1.37181401e+00 5.13031194e-03 5.38826048e-01 5.60503781e-01
7.14897752e-01 5.79079330e-01 2.64626533e-01 3.27844843e-02
6.34782732e-01 9.59705293e-01 3.73051986e-02 -3.38103712e-01
-3.04338932e-01 -4.25801814e-01 7.21362770e-01 1.19567704e+00
-1.88608855e-01 -1.84619516e-01 -1.11987138e+00 6.09370470e-01
-1.53334284e+00 -4.76414263e-01 1.16861099e-02 2.29772997e+00
1.07524920e+00 2.25903317e-01 8.33378732e-02 -1.16843600e-02
5.48798740e-01 -2.31882960e-01 -6.99944422e-02 -9.18519557e-01
-1.75779328e-01 3.14971268e-01 7.46584237e-01 7.60413945e-01
-6.29594803e-01 1.30698204e+00 5.40553665e+00 8.47753763e-01
-1.61090815e+00 1.55882418e-01 8.03526640e-01 -2.70403594e-01
-3.25073659e-01 1.65152520e-01 -1.03083944e+00 5.60257614e-01
1.35250521e+00 -1.21938586e-01 8.47509980e-01 5.21004915e-01
4.89478000e-03 -6.04206026e-02 -1.29516017e+00 8.00692141e-01
-4.68018614e-02 -1.16310310e+00 2.41474926e-01 2.78151304e-01
5.70873737e-01 2.55094737e-01 -2.24267781e-01 4.28169161e-01
-2.23500393e-02 -9.44629788e-01 9.30407524e-01 2.65472949e-01
8.16232741e-01 -1.03462768e+00 8.52155447e-01 5.86705625e-01
-8.10984790e-01 7.08089024e-03 -3.84530872e-01 -2.90124923e-01
-1.60193034e-02 5.42290449e-01 -8.31354201e-01 5.38588762e-01
6.79542482e-01 3.64081919e-01 -5.79049170e-01 5.67631602e-01
-2.51856506e-01 5.54525793e-01 -4.82427955e-01 -2.85332184e-02
3.59469652e-01 -2.21157506e-01 2.98001856e-01 1.34568417e+00
4.78861898e-01 -5.78961551e-01 -8.31158385e-02 7.33639061e-01
-4.08212036e-01 4.86522347e-01 -4.30720448e-01 -1.82723366e-02
4.85893786e-01 1.02692747e+00 -5.62867939e-01 -2.89550841e-01
-3.76960546e-01 1.19036984e+00 5.05654037e-01 -7.29496107e-02
-7.88908005e-01 -3.67939174e-01 4.38351274e-01 3.69804323e-01
2.08357617e-01 -3.43562126e-01 -4.52537686e-01 -1.23665869e+00
2.10874990e-01 -1.05133951e+00 -4.12408300e-02 -2.43959919e-01
-6.70718670e-01 7.09585667e-01 -2.96318829e-01 -1.00062442e+00
-5.11702538e-01 -5.11425197e-01 -1.56013668e-01 1.22747290e+00
-1.57169068e+00 -1.32111537e+00 3.82972121e-01 2.79956222e-01
4.95952547e-01 1.06932940e-02 1.00366378e+00 8.07364702e-01
-7.08954871e-01 1.19167423e+00 1.02500834e-01 2.20792949e-01
7.58714378e-01 -8.44524205e-01 6.34731591e-01 8.59361887e-01
1.58592135e-01 9.76427794e-01 3.47930938e-01 -2.75274247e-01
-1.38894844e+00 -8.78685653e-01 1.92152810e+00 -2.97519326e-01
5.82657754e-01 -8.41444194e-01 -7.32599497e-01 6.76530302e-01
2.55099893e-01 -5.26267469e-01 5.19118965e-01 4.93249595e-01
-3.88212770e-01 -2.61311799e-01 -8.38622630e-01 9.32870150e-01
8.72730494e-01 -6.13732815e-01 -2.24940136e-01 3.17926109e-02
5.73411167e-01 -3.35115582e-01 -8.82629275e-01 2.96376646e-01
6.11317158e-01 -6.65597558e-01 7.31176317e-01 -2.62647808e-01
6.22756839e-01 -9.97095704e-02 -4.76175621e-02 -1.49294150e+00
-3.99145544e-01 -4.64977026e-01 -3.72145474e-02 1.27784872e+00
9.48916435e-01 -8.46962512e-01 4.33512539e-01 3.11293542e-01
2.26748101e-02 -1.00721860e+00 -1.02326989e+00 -8.97090852e-01
4.26429808e-01 -4.26574759e-02 7.27281868e-01 9.60113227e-01
-1.00383587e-01 8.14829290e-01 -4.86085057e-01 -2.70450532e-01
-6.01823926e-02 2.50076819e-02 5.61362803e-01 -1.07364213e+00
-4.70002651e-01 -7.10089087e-01 -6.67156503e-02 -1.03441525e+00
2.89307963e-02 -1.19967377e+00 -2.19434977e-01 -1.39078891e+00
1.42309487e-01 -7.04082847e-01 -2.67787427e-01 8.12755466e-01
-1.47004098e-01 3.52218658e-01 5.40261745e-01 2.87594199e-01
-9.35099274e-02 3.78602415e-01 1.08722091e+00 8.10615811e-03
-1.79917365e-01 -1.00892060e-01 -8.57595265e-01 5.55766106e-01
1.06012154e+00 -5.32648504e-01 -2.35532716e-01 -1.05869722e+00
4.50080723e-01 -1.48732841e-01 -1.74747854e-02 -8.71167064e-01
2.90614143e-02 2.14163855e-01 3.79189610e-01 -9.98048112e-02
3.00840735e-01 -6.43158376e-01 7.52515271e-02 7.07084656e-01
-2.05503166e-01 2.32468158e-01 4.45011497e-01 -1.69831544e-01
-2.64522463e-01 -2.18539953e-01 7.52071679e-01 -1.88938886e-01
-1.80980340e-01 -1.03236578e-01 -3.47707897e-01 -2.58370817e-01
6.24294460e-01 -2.75942590e-02 3.82039472e-02 -1.30549401e-01
-3.42233628e-01 8.07620734e-02 6.53128684e-01 7.10288942e-01
-1.08548932e-01 -1.19847596e+00 -8.36966038e-01 3.92095327e-01
-4.83889952e-02 -2.53513098e-01 -4.86895926e-02 1.07262540e+00
-6.11968040e-01 8.58157396e-01 -3.60312074e-01 -3.99136156e-01
-1.04544735e+00 2.39552349e-01 2.90103674e-01 -7.70457983e-01
-4.49585974e-01 9.40666795e-01 -1.60386898e-02 -7.80692339e-01
5.21428771e-02 -2.12281123e-01 7.49989077e-02 2.19842762e-01
3.00076634e-01 3.35769087e-01 5.18593907e-01 -5.31152725e-01
-4.52380359e-01 4.78211194e-01 -4.18937117e-01 -1.71247140e-01
1.19842672e+00 -5.35023436e-02 -4.04359430e-01 3.19092005e-01
1.18027985e+00 2.77082533e-01 -8.41997445e-01 -2.09876016e-01
1.42090484e-01 -9.63784978e-02 1.26609966e-01 -1.02235115e+00
-9.43069756e-01 1.03783000e+00 3.80071104e-01 -1.87158227e-01
1.15132749e+00 -8.69913846e-02 1.11868191e+00 3.55403572e-01
5.67048788e-01 -1.12776148e+00 -6.18656933e-01 8.08021069e-01
6.08174562e-01 -9.31209087e-01 -2.03774169e-01 -3.42946291e-01
-4.11812186e-01 9.81449366e-01 5.26357889e-01 7.34107429e-03
1.55869564e-02 3.10893327e-01 -1.77688170e-02 3.27292651e-01
-7.64707267e-01 -4.50505922e-03 6.38978034e-02 -2.54983995e-02
8.83882523e-01 3.34296763e-01 -7.02857375e-01 6.04997098e-01
-5.90967178e-01 -1.59640089e-01 1.96814239e-01 5.99330425e-01
-1.51391432e-01 -1.76138783e+00 -1.57639489e-01 3.38015169e-01
-6.45587027e-01 -5.07838428e-01 -4.19771373e-01 7.50202000e-01
3.62251282e-01 7.31946945e-01 2.90457666e-01 -3.33854347e-01
2.07897067e-01 2.13897467e-01 6.21119976e-01 -4.68097180e-01
-8.18893135e-01 1.19220950e-01 1.90645203e-01 -3.63654405e-01
7.48160630e-02 -5.55862546e-01 -9.65056241e-01 -5.97509205e-01
-3.86419058e-01 8.14970210e-02 8.61116648e-01 8.11359882e-01
5.17805874e-01 4.42976952e-01 5.34722865e-01 -4.47020024e-01
-4.18699741e-01 -1.01173460e+00 -4.41487581e-02 -1.08381920e-01
3.83198336e-02 -2.83135265e-01 6.49634078e-02 -8.52257609e-02] | [11.552321434020996, 10.279289245605469] |
e0d02b38-147a-4efd-b4dc-71ed8f8032af | illumination-aware-multi-task-gans-for | null | null | https://ieeexplore.ieee.org/document/8606933 | https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8606933 | Illumination-Aware Multi-Task GANs for Foreground Segmentation | Foreground-background segmentation has been an active research area over the years. However, conventional models fail to produce accurate results when challenged with the videos of challenging illumination conditions. In this paper, we present a robust model that allows accurately extracting the foreground even in exceptionally dark or bright scenes and in continuously varying illumination in a video sequence. This is accomplished by a triple multi-task generative adversarial network (TMT-GAN) that effectively models the semantic relationship between the dark and bright images and performs binary segmentation end-to-end. Our contribution is twofold: first, we show that by jointly optimizing the GAN loss and the segmentation loss, our network simultaneously learns both tasks that mutually benefit each other. Second, fusing features of images with varying illumination into the segmentation branch vastly improve the performance of the network. Comparative evaluations on highly challenging real and synthetic benchmark datasets (ESI and SABS) demonstrate the robustness of TMT-GAN and its superiority over state-of-the-art approaches. | ['Hubert P. H.', 'Edmond S. L. Ho', 'Dimitrios Sakkos'] | 2019-01-10 | null | null | null | ieee-access-2019-1 | ['video-background-subtraction', 'foreground-segmentation'] | ['computer-vision', 'computer-vision'] | [ 6.79330587e-01 -2.29253292e-01 2.67479390e-01 -2.72314638e-01
-8.47926974e-01 -6.41933680e-01 4.47598755e-01 -8.32394302e-01
-2.89444566e-01 8.43454957e-01 -3.98800790e-01 -1.99928150e-01
3.97624254e-01 -5.43898940e-01 -1.05599117e+00 -1.01034153e+00
2.68355370e-01 2.65400797e-01 4.12308306e-01 1.05278850e-01
-6.99681491e-02 2.44308844e-01 -1.30256152e+00 2.21175492e-01
1.08583701e+00 1.06387460e+00 7.53359199e-02 9.19579685e-01
7.99984634e-02 9.29946721e-01 -8.15437913e-01 -6.05481088e-01
6.51387155e-01 -5.90985775e-01 -6.64581358e-01 4.11247104e-01
7.25673556e-01 -5.63495278e-01 -3.99089485e-01 1.03727174e+00
4.76031929e-01 -2.91382298e-02 4.37136620e-01 -1.49045193e+00
-4.72941488e-01 2.20525026e-01 -8.98956120e-01 1.69246614e-01
6.10850677e-02 4.33693379e-01 5.70202529e-01 -5.42020917e-01
5.22808969e-01 1.13547111e+00 6.29653037e-01 5.76040804e-01
-1.22956717e+00 -6.67421401e-01 3.56272787e-01 7.32121915e-02
-1.22766805e+00 -1.19927131e-01 9.42356050e-01 -4.14140075e-01
3.85512769e-01 2.29015961e-01 5.59966147e-01 1.37673533e+00
2.17671826e-01 1.14145446e+00 1.38939035e+00 -1.46663234e-01
-4.36911359e-03 -2.20122427e-01 -2.89501011e-01 6.13730907e-01
9.99310240e-02 5.29468656e-02 -3.45893860e-01 2.92408109e-01
7.64561713e-01 5.90513721e-02 -3.45713675e-01 -3.29937309e-01
-1.05876267e+00 5.16696811e-01 3.44663739e-01 4.20857295e-02
-3.81396145e-01 3.30427825e-01 1.99818417e-01 3.29764769e-03
5.72397411e-01 9.76029783e-02 -2.82234430e-01 6.83248490e-02
-1.27888477e+00 1.87495202e-01 5.16717494e-01 9.02516067e-01
6.14732146e-01 4.38343734e-01 -4.02509630e-01 6.47468328e-01
2.43399888e-01 6.71988904e-01 1.01959288e-01 -1.08831835e+00
3.10198069e-01 3.46332014e-01 1.24773331e-01 -8.16949248e-01
4.50814515e-02 -5.89820504e-01 -9.27741170e-01 3.42903018e-01
5.08831382e-01 -4.68290120e-01 -1.27061784e+00 1.84405220e+00
5.41858196e-01 5.38462996e-01 1.89631164e-01 9.23607945e-01
7.19307780e-01 7.23615050e-01 1.31125078e-01 -8.65648091e-02
1.03195643e+00 -1.37149549e+00 -8.50898504e-01 -7.44033039e-01
-1.49582893e-01 -8.05212617e-01 8.63342047e-01 4.19408590e-01
-1.30712843e+00 -6.23883367e-01 -1.09959555e+00 4.87473309e-02
-1.91943482e-01 4.60710339e-02 6.36914849e-01 6.57335341e-01
-9.77025092e-01 3.51414472e-01 -9.67627883e-01 4.62257043e-02
8.48948419e-01 2.73876727e-01 -8.17835913e-04 -3.13323855e-01
-9.78071332e-01 5.47992468e-01 3.27521533e-01 3.21749389e-01
-1.35731888e+00 -6.83472514e-01 -7.27851212e-01 -7.32157901e-02
6.37278914e-01 -8.57053399e-01 1.15401614e+00 -1.60470808e+00
-1.54689145e+00 1.00536132e+00 -6.62236959e-02 -5.36902905e-01
1.05812645e+00 -4.75708455e-01 -1.87275723e-01 2.93961257e-01
1.41945213e-01 7.84516573e-01 1.15141749e+00 -1.65524971e+00
-6.42785251e-01 -2.92859674e-01 -4.51517776e-02 1.33247674e-01
7.67234117e-02 6.17293045e-02 -8.76309037e-01 -8.06178033e-01
-2.99041063e-01 -9.30561483e-01 -1.53251335e-01 -5.99916950e-02
-5.93679667e-01 2.62638509e-01 1.22819352e+00 -8.77164960e-01
7.78303683e-01 -2.11173940e+00 2.43536219e-01 -2.78262287e-01
6.84312060e-02 5.83604634e-01 9.01883096e-03 -2.25921571e-02
2.01495558e-01 2.57983916e-02 -4.72987026e-01 -4.55782592e-01
-6.16642125e-02 4.44786221e-01 -2.82424837e-01 4.83818918e-01
4.16693658e-01 1.20055890e+00 -7.52613723e-01 -6.38795316e-01
3.57297242e-01 6.28274322e-01 -2.52803266e-01 6.21173263e-01
-2.97764719e-01 7.20180035e-01 -2.84739763e-01 8.03123832e-01
8.86724770e-01 -3.54837716e-01 8.24160129e-03 -1.33693457e-01
2.24756300e-01 -4.67960685e-01 -9.11847651e-01 1.57013488e+00
-2.01051012e-01 6.81047857e-01 4.21493977e-01 -9.01384234e-01
4.95794833e-01 1.57885060e-01 3.96976441e-01 -7.57092237e-01
2.80957997e-01 1.20590232e-01 -2.00483337e-01 -4.00972813e-01
2.01635718e-01 -1.58290207e-01 3.63620035e-02 1.02161625e-02
1.57447413e-01 -2.75463343e-01 2.54302889e-01 5.33747636e-02
7.16626465e-01 5.10213554e-01 -4.93693128e-02 1.33932188e-01
4.84740824e-01 -3.14353555e-01 9.07004714e-01 7.02620089e-01
-3.85926276e-01 8.51316333e-01 4.45879638e-01 -2.33797640e-01
-9.73842204e-01 -1.21854866e+00 1.81170508e-01 9.53448176e-01
4.81328636e-01 2.31990397e-01 -1.10335135e+00 -8.55263352e-01
-2.07021430e-01 6.90830708e-01 -7.33520806e-01 -6.73910528e-02
-6.98300183e-01 -9.08492982e-01 6.36050463e-01 6.81127965e-01
1.00097477e+00 -9.23446000e-01 -5.85886896e-01 2.61514843e-03
-3.57879639e-01 -1.59672415e+00 -4.81420219e-01 1.58984989e-01
-5.07740200e-01 -1.01526690e+00 -9.47838664e-01 -6.79237723e-01
5.99301457e-01 2.46653646e-01 1.25961077e+00 8.78511113e-04
-4.11686212e-01 2.94810772e-01 -2.05980361e-01 -4.00908828e-01
-4.60961431e-01 -1.35919407e-01 -6.46165311e-01 4.41358387e-01
-1.16696581e-01 -4.38703924e-01 -7.96187460e-01 5.20717740e-01
-1.31280637e+00 3.19751561e-01 6.21586800e-01 8.65207076e-01
7.34736383e-01 1.05091766e-01 4.46873307e-01 -1.05465817e+00
-3.08618043e-02 -3.80289167e-01 -6.78564847e-01 3.46902221e-01
-2.26205498e-01 -3.22306365e-01 6.46660328e-01 -3.24458033e-01
-1.28348911e+00 2.48472944e-01 -2.38786042e-01 -6.23538375e-01
-2.89728284e-01 -2.84552455e-01 -5.32870173e-01 -2.62634516e-01
1.45464346e-01 4.57004577e-01 -1.88986942e-01 -1.30480140e-01
3.59053820e-01 3.29947889e-01 9.19653535e-01 -5.27590990e-01
9.12279546e-01 7.75343955e-01 6.36644941e-03 -5.85809350e-01
-1.21660697e+00 -2.84415305e-01 -5.85622907e-01 -4.39524204e-01
1.29647470e+00 -1.03638983e+00 -4.64701205e-01 1.04901171e+00
-9.56332922e-01 -6.85060084e-01 -1.00143582e-01 -1.23546338e-02
-6.85064793e-01 4.31483239e-01 -5.20313084e-01 -7.70604193e-01
-4.66649503e-01 -1.38444412e+00 1.25596690e+00 5.47776997e-01
3.70488077e-01 -1.10134697e+00 -3.43647391e-01 8.01854253e-01
4.02779937e-01 9.29063857e-01 5.04742742e-01 -3.96958053e-01
-9.50536370e-01 -4.02065404e-02 -3.42303693e-01 7.91064858e-01
7.10422844e-02 2.04741463e-01 -1.20017087e+00 -2.82989174e-01
-5.43621485e-04 -3.31193715e-01 1.04740047e+00 5.13657272e-01
1.32894969e+00 -1.92079827e-01 -2.05272332e-01 1.14200139e+00
1.59925652e+00 3.65537137e-01 7.90256083e-01 3.07192057e-01
9.83824551e-01 1.79785475e-01 6.37269437e-01 5.64698689e-02
1.49604425e-01 4.88767743e-01 7.45565295e-01 -6.81769788e-01
-4.14530337e-01 -8.39450955e-02 2.91683108e-01 2.40431383e-01
7.93132484e-02 -6.87077761e-01 -7.43697882e-01 5.28734446e-01
-1.78101599e+00 -9.02857542e-01 -3.08089498e-02 1.83281493e+00
8.29567730e-01 2.07952708e-01 7.41932392e-02 -1.43923815e-02
7.70820618e-01 3.08274746e-01 -7.82208085e-01 -8.97333473e-02
-5.16972184e-01 9.27716047e-02 6.57871485e-01 1.76241085e-01
-1.40583181e+00 1.00429666e+00 6.58157921e+00 8.36089313e-01
-1.15337753e+00 2.53446668e-01 1.05643249e+00 -3.27430815e-02
5.53187579e-02 -2.38810137e-01 -6.13637388e-01 7.76285470e-01
6.41761422e-01 2.22289816e-01 5.40370464e-01 7.07929730e-01
-8.44430253e-02 -6.47967979e-02 -8.51797283e-01 7.22808361e-01
2.64598668e-01 -1.06359673e+00 -1.59044005e-02 -2.29123794e-02
1.26319885e+00 9.71933419e-04 3.62956673e-01 1.91107094e-01
4.49955702e-01 -1.06564403e+00 9.09664333e-01 4.53205764e-01
6.67202473e-01 -7.08870471e-01 7.89525330e-01 2.33357951e-01
-8.93909097e-01 2.35155784e-02 7.45147094e-02 3.44006419e-01
4.05756593e-01 5.41136026e-01 -5.49794257e-01 7.43833780e-01
6.41063273e-01 6.56202257e-01 -6.26422048e-01 9.35889542e-01
-4.62245226e-01 8.05719554e-01 -1.17212653e-01 5.39815843e-01
4.17543322e-01 -3.05097908e-01 4.55503106e-01 1.34406829e+00
3.49613167e-02 -1.10511556e-01 3.37327719e-01 1.08867896e+00
-3.01675975e-01 -3.54791284e-01 -4.54662293e-01 -2.54458208e-02
5.42641710e-03 1.26064527e+00 -9.80835438e-01 -3.69829237e-01
-4.57527697e-01 1.35871720e+00 8.84746984e-02 6.64056003e-01
-1.37476420e+00 -6.97753802e-02 6.36683822e-01 -1.46144673e-01
6.89063370e-01 -7.55118504e-02 -3.66475582e-01 -1.10337389e+00
8.81153792e-02 -9.83673334e-01 2.17418686e-01 -8.11551392e-01
-1.22747052e+00 5.75060308e-01 -2.11893216e-01 -9.61551070e-01
3.71280313e-02 -5.80527365e-01 -6.09041095e-01 6.66569233e-01
-1.85339355e+00 -1.53274357e+00 -5.34427822e-01 5.41247129e-01
7.84012794e-01 7.31169954e-02 3.02436978e-01 4.73734379e-01
-8.88877928e-01 5.54986179e-01 2.23423854e-01 3.36016923e-01
6.56028450e-01 -1.50761032e+00 4.62756217e-01 1.38145399e+00
-8.28005150e-02 1.40724212e-01 6.46493435e-01 -5.93770087e-01
-1.18692482e+00 -1.46985102e+00 2.27477737e-02 -2.72568285e-01
3.65034640e-01 -4.24146831e-01 -8.54194999e-01 7.32184231e-01
6.22538447e-01 3.20734382e-01 4.42988068e-01 -6.06257379e-01
-2.23042816e-01 -1.63212582e-01 -1.14012504e+00 5.23228884e-01
8.63127828e-01 -1.35511935e-01 -2.65132636e-01 3.24611336e-01
8.04373205e-01 -6.82835817e-01 -6.74772978e-01 5.61547637e-01
3.05412799e-01 -1.12207019e+00 1.23404288e+00 -3.82170111e-01
6.67538285e-01 -4.14315313e-01 -1.65443644e-01 -1.22200644e+00
1.69174105e-01 -8.34594488e-01 -1.63266912e-01 1.31163156e+00
1.61314383e-02 -5.25156438e-01 6.00287318e-01 3.72040123e-01
-9.68169346e-02 -8.20063353e-01 -7.46981800e-01 -7.55763352e-01
1.00858271e-01 -1.07485093e-01 4.47867453e-01 6.84064806e-01
-1.08256042e+00 2.41965070e-01 -6.84796929e-01 1.66057572e-01
9.42801595e-01 2.56375402e-01 9.10919368e-01 -8.86977911e-01
-3.55193108e-01 -3.81038755e-01 -3.06790203e-01 -9.84397888e-01
2.64624655e-01 -4.94740695e-01 2.69139022e-01 -1.53688693e+00
2.23663479e-01 -1.22058094e-01 -2.08285943e-01 3.64825070e-01
-6.60836101e-01 5.43813884e-01 3.12676132e-01 -1.79326728e-01
-8.75453472e-01 4.88457888e-01 1.59440351e+00 -2.40926012e-01
2.46564507e-01 5.01440763e-02 -5.84063113e-01 8.16103637e-01
5.98222136e-01 -3.88898551e-01 -3.53546709e-01 -6.11913264e-01
-1.42267391e-01 -9.96540189e-02 7.20273435e-01 -9.12531018e-01
-6.23845942e-02 -1.78855941e-01 7.07730711e-01 -5.49607098e-01
3.70575100e-01 -8.54408622e-01 1.75504342e-01 2.82971084e-01
-3.98739241e-02 -1.80636421e-01 3.29262376e-01 7.68499970e-01
-2.94615686e-01 8.50361586e-02 1.12663269e+00 -1.55353412e-01
-7.53345609e-01 3.33005607e-01 -1.18687712e-02 5.10441005e-01
1.40521932e+00 -3.28949094e-01 -2.55839735e-01 -2.02602029e-01
-4.17037278e-01 2.97101945e-01 6.59075201e-01 4.00348336e-01
4.06295180e-01 -1.01240647e+00 -7.06061006e-01 2.85046518e-01
-3.18823218e-01 3.56765985e-01 4.18027490e-01 8.75089228e-01
-7.55640209e-01 -5.59028797e-03 -2.67869949e-01 -7.83741236e-01
-1.35643685e+00 6.21028543e-01 5.68446577e-01 -3.40459883e-01
-4.32957172e-01 9.39793229e-01 5.64826071e-01 -1.17676839e-01
3.01464349e-01 -1.21711053e-01 2.58556545e-01 -3.00478637e-01
3.38684738e-01 2.59359866e-01 -1.14784978e-01 -6.35711670e-01
-1.95526779e-01 4.70378697e-01 2.55691148e-02 1.22827172e-01
1.34968972e+00 -1.33433715e-01 1.20402172e-01 2.29821965e-01
1.17603421e+00 -2.47924089e-01 -1.96137822e+00 -7.08616003e-02
-4.98351157e-01 -7.98380256e-01 3.78622636e-02 -9.94845569e-01
-1.66420698e+00 9.50534463e-01 5.46419859e-01 1.53730527e-01
1.55913925e+00 -3.45860988e-01 1.13037753e+00 -1.96538299e-01
6.33567497e-02 -9.81707275e-01 2.37704664e-01 2.21439406e-01
4.85820442e-01 -1.37731969e+00 -2.14108407e-01 -5.32240510e-01
-9.11107183e-01 7.55964875e-01 7.45162427e-01 -9.36613306e-02
1.22695155e-01 4.84579176e-01 2.81145573e-01 7.77952299e-02
-4.94147569e-01 -2.70770133e-01 3.56338739e-01 6.57600939e-01
1.78415090e-01 -1.75519854e-01 7.54823610e-02 3.51416022e-01
2.46890545e-01 -9.88076627e-02 3.21401626e-01 9.90303934e-01
-1.18447065e-01 -8.26467097e-01 -4.29456741e-01 1.19027093e-01
-9.44029093e-01 1.22782454e-01 -5.40262938e-01 7.59766936e-01
2.96094447e-01 9.31938946e-01 -4.74124961e-02 -1.52526617e-01
1.30208135e-01 1.54726226e-02 4.64156657e-01 -1.66569471e-01
-7.01046765e-01 3.15487802e-01 -2.94027895e-01 -7.02622652e-01
-6.57176733e-01 -6.98146224e-01 -1.09367549e+00 -3.41621526e-02
-2.28397235e-01 -2.21818209e-01 6.24941409e-01 9.58129287e-01
1.73050538e-01 1.07020271e+00 6.01726234e-01 -9.63660002e-01
-4.33940321e-01 -6.42359495e-01 -4.26889181e-01 6.50140882e-01
5.21421254e-01 -6.23062074e-01 -4.35801059e-01 4.70323533e-01] | [10.50694751739502, -0.8590356111526489] |
34c5ea8a-ba35-49e2-ad01-0ef0ce4bad18 | sound-source-detection-localization-and | 1908.00766 | null | https://arxiv.org/abs/1908.00766v2 | https://arxiv.org/pdf/1908.00766v2.pdf | Sound source detection, localization and classification using consecutive ensemble of CRNN models | In this paper, we describe our method for DCASE2019 task3: Sound Event Localization and Detection (SELD). We use four CRNN SELDnet-like single output models which run in a consecutive manner to recover all possible information of occurring events. We decompose the SELD task into estimating number of active sources, estimating direction of arrival of a single source, estimating direction of arrival of the second source where the direction of the first one is known and a multi-label classification task. We use custom consecutive ensemble to predict events' onset, offset, direction of arrival and class. The proposed approach is evaluated on the TAU Spatial Sound Events 2019 - Ambisonic and it is compared with other participants' submissions. | ['Sławomir Kapka', 'Mateusz Lewandowski'] | 2019-08-02 | null | null | null | null | ['sound-event-localization-and-detection'] | ['audio'] | [ 3.63599285e-02 -4.36753780e-01 4.14050430e-01 -2.65145004e-01
-1.16004705e+00 -6.33970320e-01 5.41641653e-01 4.78707463e-01
-4.99544263e-01 6.10720217e-01 4.19979960e-01 -2.76494175e-01
-1.31112067e-02 -4.91673797e-01 -5.82738161e-01 -5.36670685e-01
-1.99764669e-01 2.07158491e-01 4.94986385e-01 1.73403680e-01
1.86796665e-01 5.24268746e-01 -1.63995981e+00 7.03760862e-01
4.27466668e-02 1.23807776e+00 8.12284499e-02 1.33049321e+00
8.02099183e-02 1.13279164e+00 -1.14110327e+00 2.58426636e-01
-3.26715678e-01 -5.11225998e-01 -6.12506688e-01 -9.16068017e-01
1.88069344e-01 -3.30246128e-02 -2.35103257e-02 5.41097105e-01
1.12635565e+00 3.25015455e-01 8.50522339e-01 -1.38910735e+00
2.22168583e-02 7.07594514e-01 -3.09964448e-01 1.10903013e+00
3.41642499e-01 -3.13893676e-01 7.36096621e-01 -9.85209048e-01
3.76329035e-01 8.86408508e-01 8.76057863e-01 2.81889975e-01
-6.49901867e-01 -1.09943831e+00 -1.79512814e-01 5.90628684e-01
-1.36360323e+00 -6.03993416e-01 8.03712368e-01 -5.13004243e-01
1.21915126e+00 1.61103606e-01 1.83160946e-01 1.56730783e+00
5.85801974e-02 2.68013626e-01 9.72161114e-01 -4.18798953e-01
4.47189689e-01 -7.79604539e-02 4.28364575e-01 -5.56066749e-04
-6.17956042e-01 3.26020509e-01 -1.07344687e+00 -2.17701375e-01
-5.41897342e-02 -2.85672337e-01 1.02052771e-01 8.76995683e-01
-1.18767166e+00 3.91109884e-01 3.16196352e-01 4.60771352e-01
-6.80149853e-01 4.22206879e-01 4.71986383e-01 1.21019788e-01
6.99291945e-01 -5.42914197e-02 -4.47689086e-01 -4.13122118e-01
-1.21870148e+00 3.51888180e-01 6.54071808e-01 7.91122437e-01
4.10007119e-01 1.21116109e-01 -4.96670872e-01 8.50052476e-01
3.15996855e-01 6.01844132e-01 3.29095483e-01 -4.79021460e-01
3.88962954e-01 -1.77619571e-03 5.00062183e-02 -9.46663797e-01
-7.56167769e-01 -5.86541891e-01 -5.57015121e-01 -4.87753600e-02
6.73899204e-02 -5.27010024e-01 -7.13849902e-01 1.74044549e+00
2.81608075e-01 1.21181631e+00 -7.59497061e-02 7.10354686e-01
1.02663171e+00 1.13482332e+00 5.40757537e-01 -2.41405174e-01
1.45709670e+00 -7.17342734e-01 -9.28492129e-01 1.12842433e-01
-2.02123802e-02 -9.52251077e-01 1.67643726e-01 4.97699648e-01
-7.12595701e-01 -7.17828155e-01 -7.15696216e-01 2.81522363e-01
-5.38640857e-01 5.64954400e-01 3.26734096e-01 4.77108508e-01
-6.95197701e-01 4.32539135e-01 -6.00462258e-01 5.07845320e-02
9.33937505e-02 4.13984098e-02 4.70196456e-02 6.25105023e-01
-1.40830350e+00 6.56299472e-01 2.00144812e-01 3.74774963e-01
-1.39672565e+00 -8.08141708e-01 -4.23317373e-01 -1.31491255e-02
-1.53764084e-01 -2.16466650e-01 1.38592005e+00 -3.62840503e-01
-1.16648912e+00 5.53354204e-01 -3.87188852e-01 -6.20379865e-01
2.12434486e-01 -1.08588398e-01 -1.21425855e+00 7.20146969e-02
3.18690598e-01 2.61742711e-01 6.99371338e-01 -8.12878907e-01
-1.04822147e+00 9.50149540e-03 -5.97737730e-01 4.90955040e-02
2.16086552e-01 6.66093349e-01 2.06858441e-01 -6.20984137e-01
-8.41851458e-02 -5.53638041e-01 1.83149114e-01 -5.53477883e-01
-6.48764133e-01 -4.90306407e-01 5.44675946e-01 -7.08453953e-01
1.45878243e+00 -2.26067448e+00 -3.34341139e-01 6.57616034e-02
-1.09801749e-02 -1.13523044e-01 8.73675719e-02 6.53571904e-01
-4.10821527e-01 -2.97261588e-02 3.89841348e-02 -4.78868753e-01
-1.30269662e-01 -3.57755750e-01 -9.71704066e-01 3.51984769e-01
7.57584646e-02 1.50993183e-01 -8.15775692e-01 -3.30008507e-01
2.18904346e-01 6.50823891e-01 -2.40547538e-01 4.86620784e-01
7.83384070e-02 5.76670170e-01 -2.09169999e-01 5.40814519e-01
7.55917609e-01 3.55558008e-01 -4.71240193e-01 -3.72551590e-01
-5.60670316e-01 6.88199520e-01 -1.60782361e+00 1.31119883e+00
-7.44823337e-01 7.99527347e-01 -1.65127158e-01 -5.39558232e-01
8.18331182e-01 8.01858485e-01 3.89729977e-01 -4.28368032e-01
1.92363948e-01 2.62195885e-01 -2.37905860e-01 -6.60384238e-01
3.19492012e-01 -2.62082636e-01 -1.67337060e-01 6.94032431e-01
3.40455562e-01 3.35982084e-01 -2.31745392e-02 1.34401128e-01
1.39652717e+00 -1.34894013e-01 5.50953597e-02 2.26284042e-02
4.06520605e-01 -4.17296171e-01 4.32620049e-01 9.72182393e-01
-2.89322257e-01 7.95788229e-01 3.23415160e-01 -4.22697932e-01
-3.89470994e-01 -1.34674525e+00 -1.61303848e-01 1.30784523e+00
-4.97643277e-02 -3.12566757e-01 -4.85222846e-01 -5.78471899e-01
-3.41970503e-01 1.17240047e+00 -5.00778913e-01 -2.90966360e-04
-6.20831549e-01 -7.92654753e-01 1.11646914e+00 4.97295827e-01
2.14643463e-01 -1.77600288e+00 -7.31852949e-01 4.40130532e-01
-5.55289745e-01 -9.88471806e-01 -4.17739391e-01 7.06196070e-01
-2.89622527e-02 -9.14544880e-01 -3.41840982e-01 -6.53533757e-01
-1.06128268e-01 -2.21167222e-01 1.04785419e+00 -3.84945899e-01
-3.11816424e-01 1.95821568e-01 -3.96317989e-01 -9.05405700e-01
-2.38288000e-01 -1.50651559e-01 8.26909095e-02 3.81766647e-01
2.92689830e-01 -9.68654394e-01 -5.60733557e-01 4.03832570e-02
-4.60727721e-01 -3.96994740e-01 1.43661916e-01 1.68797597e-01
5.65951884e-01 9.98707861e-03 1.12142181e+00 -5.34511030e-01
7.75812864e-01 -1.05554688e+00 -6.25886098e-02 5.63492216e-02
3.68549190e-02 -4.76022840e-01 4.61364269e-01 -5.63063800e-01
-1.04849982e+00 2.47887224e-02 -5.75851560e-01 -2.23910332e-01
-7.69249499e-01 4.52396311e-02 1.71111777e-01 5.05639613e-01
6.94701552e-01 2.72686660e-01 -1.07516205e+00 -5.92184663e-01
1.86688238e-04 1.11199880e+00 6.90967083e-01 -4.63025510e-01
2.92699456e-01 1.60867214e-01 -2.48867214e-01 -7.00997591e-01
-7.90437222e-01 -7.50359356e-01 -4.97135252e-01 -6.82476342e-01
1.06681955e+00 -1.11015844e+00 -8.73946548e-01 8.14644039e-01
-1.66509306e+00 4.77683172e-02 -9.24895778e-02 6.61329031e-01
-7.22521096e-02 -4.41263407e-01 -3.78720134e-01 -1.13877630e+00
-4.63200688e-01 -7.54334867e-01 1.12695134e+00 1.93111330e-01
-1.44592151e-01 -5.85061252e-01 7.39502072e-01 -7.55878165e-02
4.79629040e-01 2.51052707e-01 5.34900069e-01 -1.17897558e+00
-1.22305527e-01 -1.74239486e-01 1.03557892e-01 2.35766843e-01
-2.37525344e-01 -1.17036626e-01 -1.45828927e+00 4.21068400e-01
-6.57906104e-03 1.56995784e-02 7.86992073e-01 5.72186053e-01
1.08684289e+00 -2.98579801e-02 -3.86924386e-01 3.93135667e-01
1.15648472e+00 5.65895259e-01 3.35419059e-01 -4.68792543e-02
3.86157542e-01 2.12706566e-01 2.68728912e-01 4.73144740e-01
4.46859300e-01 6.44103825e-01 3.38289469e-01 1.68686330e-01
-4.54869181e-01 -2.44567201e-01 3.35346788e-01 9.03255045e-01
1.10942245e-01 -8.23440433e-01 -1.01135933e+00 8.22999358e-01
-1.31435454e+00 -1.27062094e+00 -5.69805026e-01 1.82321870e+00
5.23013353e-01 -7.95463920e-02 1.61322564e-01 4.07789677e-01
1.04892099e+00 1.19047850e-01 -1.62157893e-01 -5.54917634e-01
-7.05018863e-02 4.76170391e-01 9.45943892e-02 4.63707179e-01
-1.10657501e+00 5.71705759e-01 6.52775383e+00 8.76775742e-01
-1.13727307e+00 5.93402684e-01 4.14298594e-01 -4.16052252e-01
8.60071629e-02 -2.81521499e-01 -9.16259766e-01 8.04758370e-01
1.60620558e+00 4.16005135e-01 3.35686624e-01 3.63250464e-01
2.58964598e-01 -2.13465497e-01 -1.02035415e+00 8.72986436e-01
1.41300157e-01 -1.06853855e+00 -2.39989355e-01 -4.34433937e-01
2.15388089e-01 2.80013740e-01 -3.43730748e-01 3.90081763e-01
1.12551991e-02 -7.06710756e-01 1.21435595e+00 8.43449235e-01
6.01843059e-01 -7.57991016e-01 4.34846520e-01 3.97332400e-01
-1.30025959e+00 -7.97071233e-02 7.72673488e-02 -1.57752439e-01
4.74934906e-01 7.84809232e-01 -9.64550078e-01 4.60190743e-01
1.01630330e+00 4.18873340e-01 -4.42405760e-01 1.14443421e+00
-2.91404635e-01 1.38530576e+00 -5.28472424e-01 -5.73408566e-02
-9.65350643e-02 3.40442955e-01 9.97636139e-01 1.67076719e+00
6.10802054e-01 1.74987242e-01 -2.27104262e-01 7.93891430e-01
-1.06583163e-03 -2.05925237e-02 -3.81334692e-01 5.35296857e-01
9.51723456e-01 1.26885617e+00 -8.17812145e-01 -1.98574707e-01
4.34113666e-02 7.05772579e-01 1.21705398e-01 3.32541883e-01
-1.14128244e+00 -6.44566655e-01 1.65036246e-01 -2.94664264e-01
2.07316428e-01 3.34321171e-01 -8.75147060e-02 -5.02453566e-01
-1.82206810e-01 -3.45560879e-01 4.89079684e-01 -1.14762330e+00
-1.22274220e+00 1.09558904e+00 -4.44743484e-02 -1.14815140e+00
-1.48680195e-01 -2.01855719e-01 -1.24787378e+00 9.40523863e-01
-1.29259109e+00 -1.08001661e+00 -4.72705625e-02 5.14953017e-01
5.05711854e-01 -2.24070072e-01 8.54176819e-01 8.81588280e-01
-6.99676812e-01 3.18862706e-01 -2.71441966e-01 2.50306148e-02
8.72483432e-01 -1.26064980e+00 1.31164283e-01 8.00037503e-01
4.31248426e-01 8.98394585e-02 6.40974939e-01 -6.04763448e-01
-3.78901720e-01 -1.20821857e+00 1.52944696e+00 -5.06007314e-01
5.77556074e-01 -5.96113682e-01 -4.22856390e-01 5.15959680e-01
2.50661612e-01 1.04731284e-01 5.34852445e-01 5.13722189e-02
-2.18500957e-01 -3.03310364e-01 -9.56462622e-01 6.83023185e-02
7.21047044e-01 -6.76086128e-01 -5.09355307e-01 2.92907864e-01
6.02234304e-01 -2.99832284e-01 -5.41214645e-01 1.93363085e-01
3.87506247e-01 -8.87585282e-01 1.08801639e+00 -2.34411076e-01
3.72166902e-01 -4.40900892e-01 -2.41046935e-01 -1.21815777e+00
-1.75117627e-02 -4.60873961e-01 -3.65986452e-02 1.73672068e+00
4.45513725e-01 -2.94566661e-01 -2.01013818e-01 -2.43808776e-01
-4.52168792e-01 -3.46539348e-01 -1.39580691e+00 -3.34023446e-01
-3.72299910e-01 -1.15263462e+00 5.93398571e-01 6.59187973e-01
-4.00560915e-01 4.93323296e-01 -5.97963929e-01 6.35512829e-01
4.43064898e-01 -2.07190681e-02 5.07598855e-02 -1.15832269e+00
-1.86937392e-01 -1.15310565e-01 -6.22919165e-02 -3.51178378e-01
6.95902556e-02 -1.02718115e+00 4.75834846e-01 -1.03843808e+00
-1.33330390e-01 -5.18625259e-01 -9.61896420e-01 4.52946872e-01
-7.38392817e-03 3.88361692e-01 -1.28189221e-01 -1.14801368e-02
-7.06790626e-01 1.88011318e-01 3.12059253e-01 1.37639806e-01
-2.88904369e-01 6.29274666e-01 -2.83696115e-01 7.14886367e-01
6.12074792e-01 -1.24774814e+00 1.49909789e-02 -2.84085214e-01
4.07011896e-01 4.88023937e-01 8.04726362e-01 -1.45894265e+00
5.48670769e-01 2.14195177e-01 3.81929964e-01 -1.15798581e+00
2.90671676e-01 -6.30837858e-01 3.02135378e-01 -4.77497466e-02
-6.61643028e-01 6.47802055e-02 3.95206153e-01 6.22517765e-01
-2.58972764e-01 -3.72570306e-01 3.82071257e-01 2.47063875e-01
-4.15343136e-01 -6.23846911e-02 -9.60193872e-01 6.36448711e-02
9.29432750e-01 3.93965721e-01 -2.61103541e-01 -2.08224371e-01
-8.53027046e-01 -2.76696593e-01 -8.93996060e-01 6.69191957e-01
4.90622699e-01 -1.33411705e+00 -8.87262762e-01 8.60709175e-02
-1.64367318e-01 -5.24923742e-01 6.28025711e-01 7.83126235e-01
-2.22951561e-01 1.37405038e-01 6.56044576e-03 -3.99359554e-01
-1.06484330e+00 1.58274442e-01 6.62649214e-01 -6.81857243e-02
-4.12867010e-01 1.07588542e+00 -1.62493169e-01 -4.12381530e-01
4.47659403e-01 -2.88360298e-01 -6.98154569e-01 3.73721987e-01
7.50540853e-01 8.51431787e-01 3.35047483e-01 -6.88498259e-01
-8.32862377e-01 3.84093314e-01 2.62327403e-01 -5.01923263e-01
1.35857582e+00 -9.67245921e-02 -1.25844508e-01 9.27157164e-01
1.03899193e+00 3.48239601e-01 -7.90929139e-01 1.39551774e-01
-1.02841817e-01 3.40092242e-01 3.94293219e-01 -1.39902270e+00
-8.95165980e-01 1.21744525e+00 1.06746709e+00 1.79605588e-01
1.24733460e+00 1.21453647e-02 7.69369483e-01 -1.49177864e-01
8.25342664e-04 -9.65082586e-01 -2.72538334e-01 5.03918827e-01
8.74484837e-01 -6.97335124e-01 -5.59895396e-01 1.27270415e-01
-5.19145489e-01 1.04298306e+00 3.27672780e-01 -1.28290415e-01
1.21693122e+00 5.63996494e-01 5.83423115e-02 -3.75074118e-01
-9.18137312e-01 -1.04977548e-01 3.14821035e-01 5.28548181e-01
2.62074828e-01 9.39497352e-02 3.97636630e-02 1.11858356e+00
-2.18515202e-01 -8.67597461e-02 3.40247482e-01 5.81330717e-01
-3.13919663e-01 -5.13914227e-01 -4.74594414e-01 2.28469431e-01
-8.34718347e-01 -3.64638239e-01 -2.64481008e-01 1.31424680e-01
8.31813633e-01 1.40665996e+00 3.35313678e-01 -4.80120391e-01
5.73485196e-01 3.90704304e-01 -1.20672159e-01 -4.30680215e-01
-1.08973014e+00 1.64667100e-01 2.45976344e-01 -2.64390796e-01
-2.67460257e-01 -8.47728670e-01 -1.30753934e+00 2.71248847e-01
-2.89760768e-01 2.52386659e-01 9.36388910e-01 9.93816137e-01
4.35707241e-01 1.13026762e+00 7.00377166e-01 -7.25915492e-01
-1.09741628e-01 -1.31085193e+00 -4.34309393e-01 1.37284234e-01
6.99319363e-01 -4.78736401e-01 -6.09167218e-01 2.05867469e-01] | [15.144868850708008, 5.206925868988037] |
29575f57-52c4-477c-9e5f-819bebde8950 | collaborative-deep-reinforcement-learning-for-1 | 1702.05573 | null | http://arxiv.org/abs/1702.05573v1 | http://arxiv.org/pdf/1702.05573v1.pdf | Collaborative Deep Reinforcement Learning for Joint Object Search | We examine the problem of joint top-down active search of multiple objects
under interaction, e.g., person riding a bicycle, cups held by the table, etc..
Such objects under interaction often can provide contextual cues to each other
to facilitate more efficient search. By treating each detector as an agent, we
present the first collaborative multi-agent deep reinforcement learning
algorithm to learn the optimal policy for joint active object localization,
which effectively exploits such beneficial contextual information. We learn
inter-agent communication through cross connections with gates between the
Q-networks, which is facilitated by a novel multi-agent deep Q-learning
algorithm with joint exploitation sampling. We verify our proposed method on
multiple object detection benchmarks. Not only does our model help to improve
the performance of state-of-the-art active localization models, it also reveals
interesting co-detection patterns that are intuitively interpretable. | ['Bo Xin', 'Xiangyu Kong', 'Yizhou Wang', 'Gang Hua'] | 2017-02-18 | collaborative-deep-reinforcement-learning-for-2 | http://openaccess.thecvf.com/content_cvpr_2017/html/Kong_Collaborative_Deep_Reinforcement_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Kong_Collaborative_Deep_Reinforcement_CVPR_2017_paper.pdf | cvpr-2017-7 | ['active-object-localization'] | ['computer-vision'] | [-2.81833783e-02 2.54780412e-01 -5.37068665e-01 -1.50912991e-02
-1.07904959e+00 -4.90977049e-01 5.85562944e-01 3.36644828e-01
-6.60079062e-01 6.87329829e-01 5.88706881e-02 1.50189772e-01
-4.99542743e-01 -5.92610419e-01 -9.23946917e-01 -8.18904817e-01
-5.08714557e-01 8.65009785e-01 4.50229704e-01 -8.77272040e-02
2.55914032e-01 3.95088911e-01 -1.17525458e+00 1.43607751e-01
6.97025299e-01 9.78971124e-01 5.51784933e-01 7.30242670e-01
2.43753299e-01 1.14913201e+00 -8.62119079e-01 -1.50123909e-01
2.70978570e-01 -2.33429581e-01 -7.94407964e-01 2.69435853e-01
3.37793157e-02 -5.95295072e-01 -1.35209352e-01 9.38205302e-01
5.14545381e-01 4.73902792e-01 5.16445935e-01 -1.40563548e+00
-6.73344791e-01 7.97539473e-01 -9.43652570e-01 4.17934656e-01
3.82156610e-01 5.53238213e-01 1.47108245e+00 -6.36644423e-01
6.73081100e-01 1.51534414e+00 5.14041521e-02 3.59590054e-01
-1.21342862e+00 -6.34750307e-01 5.99906564e-01 3.87745440e-01
-1.19502652e+00 -2.05609068e-01 9.16797698e-01 -2.92916875e-02
9.79658246e-01 -2.69214250e-02 8.28040719e-01 1.33963943e+00
1.31098226e-01 1.58511138e+00 7.30381072e-01 -4.15607601e-01
3.96295547e-01 -1.46837771e-01 -9.85474437e-02 1.05724156e+00
2.07851574e-01 2.21897215e-01 -9.18730855e-01 -4.75483716e-01
7.20447361e-01 2.97534972e-01 3.01889688e-01 -9.06606495e-01
-1.09888947e+00 1.00045967e+00 7.25659788e-01 2.97568198e-02
-5.31764090e-01 6.48615539e-01 1.38353836e-03 6.61822036e-02
3.40020806e-01 6.52746975e-01 -3.42177808e-01 2.15439368e-02
-2.91306823e-01 3.74866724e-01 6.26669645e-01 8.86547387e-01
7.54487455e-01 -2.66878366e-01 -2.48220280e-01 6.53975785e-01
7.35736132e-01 4.27968323e-01 -1.11631244e-01 -1.21432018e+00
6.31793797e-01 8.48123789e-01 6.33327484e-01 -6.48516297e-01
-4.88397241e-01 -5.88883102e-01 -3.38034838e-01 4.25695837e-01
4.59949046e-01 -3.73097450e-01 -4.97883171e-01 1.73264813e+00
4.55656707e-01 1.02862619e-01 -3.08395643e-02 9.66312110e-01
3.63867193e-01 3.67847085e-01 3.05212885e-01 -1.92827642e-01
1.46121991e+00 -1.50549245e+00 -5.74445605e-01 -5.06453276e-01
4.83826578e-01 -2.43650153e-01 6.00538552e-01 2.08726332e-01
-1.15233183e+00 -4.91317242e-01 -8.92024755e-01 2.55836755e-01
-3.74467999e-01 -3.64945340e-03 8.33958268e-01 1.84723109e-01
-8.15891683e-01 1.86029166e-01 -1.05752492e+00 -1.55042589e-01
9.61817980e-01 6.72961593e-01 -1.33612156e-01 1.36419356e-01
-9.22216117e-01 7.79134750e-01 1.40496373e-01 -1.46327972e-01
-1.53662145e+00 -3.90390217e-01 -6.77402198e-01 8.43433589e-02
1.15555549e+00 -6.83308303e-01 1.53364635e+00 -1.13530171e+00
-1.54711545e+00 6.28586769e-01 -2.33509205e-02 -7.16213584e-01
5.84645450e-01 -5.82976937e-01 -1.79726467e-01 3.38288069e-01
4.95645046e-01 9.00634289e-01 8.87710750e-01 -1.34624815e+00
-1.03977501e+00 -5.53393185e-01 5.39830685e-01 6.01858258e-01
-4.33866903e-02 1.64805293e-01 -5.33876896e-01 -3.90483171e-01
-3.66553873e-01 -8.81992936e-01 -4.83500987e-01 2.92375594e-01
-4.64993298e-01 -7.93225348e-01 8.27505231e-01 1.12558544e-01
8.33319902e-01 -1.96916616e+00 2.43153930e-01 1.59831852e-01
3.45126629e-01 -1.82582974e-01 -2.93081880e-01 4.69837517e-01
3.82397234e-01 -1.86384216e-01 1.79577127e-01 -6.94455743e-01
2.49179900e-01 2.27191567e-01 -1.76733494e-01 6.60171986e-01
3.32296371e-01 1.34022069e+00 -1.37222898e+00 -4.56667095e-01
-4.42986004e-02 3.44647318e-02 -3.37471664e-01 3.04711640e-01
-4.68389452e-01 5.11639595e-01 -1.01747799e+00 9.04816747e-01
3.59130472e-01 -7.65756667e-01 2.27947235e-01 3.06877762e-01
3.41172554e-02 4.23521511e-02 -1.30395830e+00 1.88033271e+00
-1.92355335e-01 3.34894150e-01 4.04948205e-01 -1.03198743e+00
4.17567939e-01 8.19491968e-03 6.44924581e-01 -1.20721424e+00
-1.84432358e-01 -7.44886026e-02 2.21249044e-01 -3.14822793e-01
5.65326035e-01 6.10935390e-01 -1.07293881e-01 8.40993226e-01
6.48771152e-02 2.95015156e-01 1.31879255e-01 3.70385736e-01
1.33930290e+00 5.95504791e-02 3.57574999e-01 -1.86792448e-01
7.99685866e-02 -1.57469958e-01 4.09139931e-01 1.61115479e+00
-4.60136950e-01 2.69065518e-02 3.98240179e-01 -2.04253092e-01
-5.96405745e-01 -1.10723495e+00 4.02799577e-01 1.67597246e+00
7.87425578e-01 -1.98801160e-01 -3.24610054e-01 -1.08320343e+00
2.01501921e-01 2.36243188e-01 -6.82025313e-01 -2.40121037e-01
-6.37292743e-01 -6.99954450e-01 3.02453578e-01 6.27724767e-01
5.47858119e-01 -1.44302821e+00 -1.13135993e+00 4.51569855e-01
1.19182505e-01 -7.45815217e-01 -3.99217725e-01 6.26773417e-01
-6.38552308e-01 -1.36062908e+00 -6.62456453e-01 -5.82221270e-01
6.52463138e-01 3.17228734e-01 1.33886039e+00 2.19562039e-01
-2.68069744e-01 8.21023762e-01 -2.09709570e-01 -4.91449416e-01
-9.84170586e-02 1.85432136e-01 9.83773321e-02 4.89660688e-02
1.61408707e-01 -1.65901080e-01 -1.06594920e+00 5.46634436e-01
-5.02641857e-01 -3.37712318e-01 7.00271964e-01 9.34793055e-01
4.63650525e-01 -2.61355583e-02 4.79493290e-01 -6.27058148e-01
7.71442533e-01 -6.26882732e-01 -9.84389186e-01 3.65506083e-01
-3.95388186e-01 1.23487473e-01 8.67648274e-02 -5.75903714e-01
-1.08818471e+00 1.74830005e-01 3.22902501e-01 -1.82322800e-01
-9.87202749e-02 1.26053803e-02 8.90175924e-02 -2.05512881e-01
7.00010538e-01 -8.12748149e-02 -2.93505311e-01 -2.32668757e-01
3.89465451e-01 1.44585729e-01 1.95530862e-01 -6.70328557e-01
4.83269960e-01 7.35390484e-01 -9.55967680e-02 -3.68403047e-01
-7.99763203e-01 -6.08703971e-01 -3.17537814e-01 -9.82326195e-02
7.16695905e-01 -9.74225283e-01 -1.47592294e+00 3.98540527e-01
-1.21507311e+00 -7.38597572e-01 -3.06128323e-01 2.97591448e-01
-7.61035383e-01 2.49415319e-02 -4.53407824e-01 -1.23788786e+00
6.48742914e-02 -1.28111064e+00 1.36192942e+00 5.05280316e-01
-4.61815931e-02 -9.07647729e-01 1.72267288e-01 5.08253753e-01
1.38110727e-01 -2.53806531e-01 4.83110934e-01 -7.22805619e-01
-1.36505687e+00 1.25342578e-01 4.93138880e-02 -3.90232235e-01
2.33046357e-02 -2.25271195e-01 -7.72745967e-01 -4.20674831e-01
-4.45413500e-01 -6.14040613e-01 9.82983470e-01 5.95528364e-01
1.17525339e+00 -3.10753912e-01 -9.36176181e-01 1.33507833e-01
1.07968700e+00 4.64434236e-01 1.93950757e-01 4.50968444e-01
2.42559776e-01 4.74121690e-01 9.18735206e-01 6.09799683e-01
4.29175735e-01 9.75955009e-01 9.87779796e-01 -2.38908947e-01
2.26907134e-01 -1.09707229e-01 2.30432987e-01 -3.06993306e-01
-3.08296718e-02 -6.22682095e-01 -8.21416199e-01 5.80372930e-01
-2.47265649e+00 -9.38285351e-01 4.60615814e-01 1.76094198e+00
5.35966039e-01 3.48619908e-01 3.23606551e-01 -5.52475333e-01
4.82857674e-01 3.52766454e-01 -1.07388532e+00 3.07942688e-01
-7.46035203e-02 -4.34086248e-02 4.68406796e-01 6.19260848e-01
-1.38753498e+00 9.29612994e-01 6.56841993e+00 9.11229789e-01
-2.84484386e-01 4.72820550e-01 5.12856603e-01 -2.75445193e-01
9.97288749e-02 -1.16679020e-01 -9.60171580e-01 2.34152898e-01
3.22798729e-01 4.31740850e-01 4.28691626e-01 9.02361453e-01
3.43953259e-02 -6.85470521e-01 -1.27029121e+00 8.45353127e-01
-1.95002139e-01 -1.49027622e+00 -3.79958063e-01 1.72904700e-01
8.40404689e-01 2.43082926e-01 2.86898166e-01 3.51921201e-01
1.04359400e+00 -7.40421295e-01 7.00365961e-01 3.91062617e-01
-6.07469454e-02 -6.60961509e-01 2.64077097e-01 5.38941503e-01
-1.08946931e+00 -6.31252289e-01 -9.26007256e-02 -5.67744225e-02
6.82429075e-02 -7.51982704e-02 -6.58578217e-01 2.09689796e-01
6.34861469e-01 5.11028171e-01 -5.26595235e-01 1.15668333e+00
-3.49030226e-01 2.20518842e-01 -3.41568828e-01 -4.70340192e-01
4.73298877e-01 -3.05444710e-02 7.65262604e-01 8.57456923e-01
-2.80666232e-01 -8.45071077e-02 4.84295011e-01 1.18682122e+00
-3.18461031e-01 -3.64439011e-01 -5.99693239e-01 1.03991680e-01
4.36228722e-01 1.29054165e+00 -9.74101186e-01 -3.98916811e-01
-4.16003734e-01 1.05676830e+00 7.64561117e-01 6.40079677e-01
-8.73051405e-01 -4.73383963e-02 7.47591496e-01 -1.35139138e-01
3.53823721e-01 -1.47208467e-01 3.10230583e-01 -8.50536466e-01
-5.59129827e-02 -7.29545951e-01 5.77868700e-01 -6.52435124e-01
-1.28807855e+00 -2.34171096e-02 -1.93939973e-02 -1.00198436e+00
-4.13491786e-01 -4.49446321e-01 -5.70675611e-01 4.22004163e-01
-1.29630077e+00 -1.07339430e+00 4.99249734e-02 6.59986556e-01
7.67442048e-01 -4.22320634e-01 6.80973351e-01 1.77262835e-02
-4.34817821e-01 4.30975258e-01 5.83190173e-02 2.38229319e-01
4.05436486e-01 -1.42246830e+00 3.16366464e-01 6.27052128e-01
6.81263685e-01 5.79618275e-01 4.00522798e-01 -5.90542793e-01
-1.56383777e+00 -7.11909771e-01 -3.04048862e-02 -7.85550773e-01
8.09657037e-01 -6.37194753e-01 -5.04620910e-01 6.67396724e-01
4.69831973e-01 3.07019383e-01 4.32647228e-01 3.98757696e-01
-1.47226751e-01 -1.51142180e-01 -8.40892911e-01 6.86074018e-01
1.22334647e+00 -5.00719130e-01 -5.05500495e-01 6.63485646e-01
7.39345253e-01 -5.52221358e-01 -2.66767979e-01 -9.28815976e-02
4.30478930e-01 -8.21030259e-01 1.26125002e+00 -8.59700382e-01
2.12782323e-01 -1.40997022e-01 1.95859566e-01 -1.34497368e+00
-3.30023974e-01 -8.24138999e-01 -8.05214047e-01 5.98581910e-01
5.38125336e-01 -5.04980683e-01 1.02292204e+00 4.41201299e-01
6.77643418e-02 -8.08107555e-01 -1.00550914e+00 -5.44734240e-01
-5.76929688e-01 -1.99428961e-01 3.67906779e-01 4.35880840e-01
1.68914557e-03 3.56823504e-01 -3.97430658e-01 3.29644293e-01
9.07304943e-01 1.83920935e-01 7.03398883e-01 -9.80535030e-01
-8.02468121e-01 -4.62760717e-01 1.15056904e-02 -1.61046886e+00
2.63868481e-01 -4.74220157e-01 1.13817945e-01 -1.43011522e+00
5.54446995e-01 -5.14115393e-01 -5.23414493e-01 5.60759485e-01
-1.18007638e-01 1.15137458e-01 1.54496193e-01 1.64121032e-01
-1.74772286e+00 6.39513552e-01 1.37409651e+00 -5.70932388e-01
-3.82915169e-01 3.87819618e-01 -7.38737822e-01 6.79237604e-01
5.05322933e-01 -7.30801821e-01 -2.90079534e-01 -4.92005318e-01
5.22350073e-01 2.84689665e-01 6.74483657e-01 -6.34964347e-01
7.20085382e-01 -2.22046003e-01 4.91806865e-01 -5.38130522e-01
6.74848378e-01 -9.83812690e-01 -4.80121970e-01 6.73602879e-01
-5.50688803e-01 -3.77187245e-02 2.73830798e-02 1.13902724e+00
8.78326315e-03 -2.63924077e-02 5.42472363e-01 -3.29494596e-01
-9.19355929e-01 3.33868116e-01 -5.37876010e-01 -3.91690526e-03
1.13099873e+00 -2.27882434e-02 -6.54235065e-01 -4.11876023e-01
-8.02228391e-01 8.00180316e-01 -3.32866535e-02 5.97987473e-01
4.44207489e-01 -1.19793272e+00 -3.50428998e-01 5.07689752e-02
2.86740839e-01 1.98331088e-01 5.61195798e-02 7.40417957e-01
8.77671838e-02 2.60798633e-01 -2.03438684e-01 -6.83511615e-01
-1.09834683e+00 4.83366281e-01 5.82352459e-01 -5.99532425e-01
-3.80198359e-01 8.96834552e-01 4.09066558e-01 -2.16489002e-01
5.57197511e-01 -1.13267526e-01 -1.74845070e-01 9.96829271e-02
3.73882651e-01 4.60728973e-01 -3.15788358e-01 -1.09699100e-01
-4.64121729e-01 3.12597603e-01 -3.92544150e-01 -1.78333908e-01
1.26784742e+00 -1.86651826e-01 2.00829625e-01 2.21018836e-01
7.49913514e-01 -1.38682649e-01 -1.79191482e+00 -3.83782148e-01
1.14077561e-01 -2.82448649e-01 -1.46009505e-01 -9.92598414e-01
-9.76263702e-01 5.26553333e-01 7.70885050e-01 4.28060025e-01
6.05301857e-01 5.51557124e-01 8.65041018e-02 9.71773624e-01
6.13212943e-01 -1.41779983e+00 1.02349973e+00 2.41224885e-01
7.65286088e-01 -1.76663768e+00 6.06844872e-02 9.10560340e-02
-6.28384054e-01 7.08284318e-01 8.81812274e-01 -2.49501139e-01
4.53933209e-01 2.08783299e-01 -3.15980643e-01 -5.69364846e-01
-9.38749552e-01 -4.04430509e-01 1.31142154e-01 4.97632056e-01
-1.92158729e-01 8.92585665e-02 2.69579053e-01 3.80024612e-01
7.41905689e-01 -3.31804663e-01 -7.45743737e-02 9.88593698e-01
-5.81159592e-01 -9.13373768e-01 -2.46987671e-01 3.26758623e-01
-4.77129191e-01 2.33078003e-01 -5.09003818e-01 9.53448534e-01
2.87012048e-02 9.75610971e-01 2.79020011e-01 1.94075480e-01
1.42760471e-01 -3.32057565e-01 5.76689303e-01 -6.83310568e-01
-6.12701476e-01 2.02460468e-01 -1.66797042e-01 -8.67400944e-01
-7.25189269e-01 -5.79820395e-01 -1.22743011e+00 2.20563307e-01
-5.13654768e-01 7.90757164e-02 1.85504049e-01 9.47482765e-01
5.60224593e-01 6.59471869e-01 4.90724504e-01 -8.20907652e-01
-6.42902255e-01 -8.89175475e-01 -6.40781224e-01 2.59786457e-01
6.86945021e-01 -1.06976688e+00 -1.04522854e-01 -5.24567604e-01] | [3.819063425064087, 1.9176979064941406] |
ddadd12c-5d0c-4d91-9aca-351d102095be | decoupled-gradient-harmonized-detector-for | 2004.04455 | null | https://arxiv.org/abs/2004.04455v1 | https://arxiv.org/pdf/2004.04455v1.pdf | Decoupled Gradient Harmonized Detector for Partial Annotation: Application to Signet Ring Cell Detection | Early diagnosis of signet ring cell carcinoma dramatically improves the survival rate of patients. Due to lack of public dataset and expert-level annotations, automatic detection on signet ring cell (SRC) has not been thoroughly investigated. In MICCAI DigestPath2019 challenge, apart from foreground (SRC region)-background (normal tissue area) class imbalance, SRCs are partially annotated due to costly medical image annotation, which introduces extra label noise. To address the issues simultaneously, we propose Decoupled Gradient Harmonizing Mechanism (DGHM) and embed it into classification loss, denoted as DGHM-C loss. Specifically, besides positive (SRCs) and negative (normal tissues) examples, we further decouple noisy examples from clean examples and harmonize the corresponding gradient distributions in classification respectively. Without whistles and bells, we achieved the 2nd place in the challenge. Ablation studies and controlled label missing rate experiments demonstrate that DGHM-C loss can bring substantial improvement in partially annotated object detection. | ['Yuanfan Guo', 'Jiancheng Yang', 'Yi Xu', 'Tiancheng Lin', 'Canqian Yang'] | 2020-04-09 | null | null | null | null | ['cell-detection'] | ['computer-vision'] | [ 3.52234751e-01 3.83346558e-01 -5.36766350e-01 -3.35752279e-01
-1.30916893e+00 -4.44911569e-01 3.27064514e-01 1.12864032e-01
-4.62465078e-01 9.26489830e-01 1.38991296e-01 -2.58061618e-01
2.21091807e-01 -1.85290188e-01 -4.19290751e-01 -1.23537767e+00
1.18058175e-01 6.44629002e-02 4.36598480e-01 1.96064606e-01
-1.26411155e-01 3.31705391e-01 -9.60369706e-01 3.91471565e-01
8.48307788e-01 8.16895187e-01 4.68233004e-02 4.64880526e-01
-1.19947813e-01 1.08337188e+00 -5.89266002e-01 -2.82218754e-01
2.82084495e-01 -5.88676214e-01 -5.75847268e-01 -7.94061273e-02
7.23510265e-01 -1.00436911e-01 -2.17380330e-01 1.32157803e+00
5.79320967e-01 -4.39978123e-01 6.77811742e-01 -1.22204077e+00
-1.05410554e-01 5.20854175e-01 -1.25422800e+00 1.77827820e-01
-4.27185595e-01 3.16378534e-01 9.50407743e-01 -7.61181951e-01
8.45282018e-01 8.24006915e-01 8.98396015e-01 7.20980883e-01
-1.12303615e+00 -8.66633892e-01 2.65219659e-01 3.88786830e-02
-1.43186092e+00 -1.90366909e-01 5.77029943e-01 -2.61251271e-01
9.24262553e-02 6.73100889e-01 5.82215250e-01 8.55863690e-01
-1.85536984e-02 1.21485031e+00 1.16030467e+00 -1.72450379e-01
3.48372646e-02 4.65952873e-01 3.41835409e-01 7.00908720e-01
6.33214891e-01 -1.82297081e-01 -3.88305634e-02 3.25378142e-02
5.54860115e-01 1.46190688e-01 -5.05616844e-01 -3.34557712e-01
-1.09571242e+00 4.42775398e-01 6.31577432e-01 2.74198115e-01
1.84573784e-01 1.41100124e-01 5.56686282e-01 2.11112457e-03
1.85843989e-01 -4.16718945e-02 -1.30416289e-01 4.70983118e-01
-9.27004516e-01 1.07313909e-01 4.04331118e-01 8.07457209e-01
3.11538011e-01 -2.12217376e-01 -5.98039508e-01 7.51361132e-01
3.01823884e-01 2.50007153e-01 6.85506105e-01 -3.82652283e-01
1.88302189e-01 9.22161162e-01 -5.17435633e-02 -5.49523771e-01
-5.25021851e-01 -1.11103344e+00 -1.03980696e+00 2.86568105e-01
8.43300223e-01 -1.05512053e-01 -1.26208222e+00 1.68019783e+00
4.85722691e-01 1.42313018e-01 -4.75085415e-02 1.08167291e+00
8.65039468e-01 5.46150580e-02 3.15289676e-01 -2.07328245e-01
1.68594849e+00 -1.18622708e+00 -6.77486598e-01 -1.32431716e-01
1.13788068e+00 -4.84301805e-01 1.06222224e+00 9.09997001e-02
-7.57069409e-01 5.15664257e-02 -9.73726690e-01 -1.03027724e-01
-1.40624344e-01 6.16840601e-01 7.93377340e-01 6.71592057e-01
-8.34707737e-01 1.59828030e-02 -8.52820218e-01 -3.21961969e-01
9.77058887e-01 2.63291866e-01 -3.71651500e-01 -1.18220642e-01
-5.40101171e-01 7.38248646e-01 5.71242087e-02 1.28573433e-01
-9.95531142e-01 -1.06615865e+00 -5.30663431e-01 -2.90365994e-01
4.88114536e-01 -4.48401988e-01 1.22512376e+00 -9.86607254e-01
-1.06990695e+00 1.16301906e+00 -1.14412814e-01 -5.37229538e-01
9.30262923e-01 3.36485118e-01 -2.64802635e-01 6.21580742e-02
-4.48282249e-02 6.64801598e-01 7.84832776e-01 -1.31404531e+00
-9.56228137e-01 -5.33007383e-01 -4.09150004e-01 9.70276669e-02
-1.41259149e-01 -1.29189014e-01 -4.68036801e-01 -7.96313465e-01
2.45055735e-01 -7.67488539e-01 -4.32955533e-01 4.91072178e-01
-4.83553380e-01 -8.45339373e-02 9.66685534e-01 -6.96706593e-01
1.10540617e+00 -2.50323057e+00 -3.81372511e-01 1.81512058e-01
5.04320741e-01 2.43170217e-01 6.93496466e-02 -5.59543133e-01
9.91398916e-02 2.53422499e-01 -2.14108527e-01 -3.23496222e-01
-1.38059095e-01 1.46066397e-01 5.53003475e-02 7.69192100e-01
2.40669698e-01 9.62084651e-01 -9.61722791e-01 -8.21018457e-01
-1.50269911e-01 4.41810638e-01 -5.42110324e-01 -2.64946669e-01
-1.54865971e-02 5.50679147e-01 -3.86059463e-01 1.08908844e+00
8.19518745e-01 -5.62748134e-01 1.38776213e-01 -4.92920518e-01
2.25819111e-01 -1.62737012e-01 -9.91833806e-01 1.24161053e+00
-2.24816916e-03 5.44808626e-01 4.78982896e-01 -7.29519129e-01
6.05408490e-01 1.29634231e-01 3.42942834e-01 -6.36864364e-01
7.52643943e-02 4.17798251e-01 1.47819921e-01 -4.04021621e-01
1.50153562e-01 -2.91959107e-01 1.71873808e-01 -1.32680386e-01
-2.74748027e-01 2.41232365e-01 4.33560945e-02 2.07301125e-01
1.17966270e+00 -1.90534532e-01 3.48918796e-01 -2.98310548e-01
6.04709446e-01 1.49510324e-01 9.38444674e-01 7.19132364e-01
-6.51749015e-01 6.93085194e-01 7.38797784e-01 -8.30234662e-02
-6.83698535e-01 -8.90436709e-01 -4.60621774e-01 8.64366949e-01
3.14496785e-01 7.14745326e-03 -5.10082722e-01 -1.19821858e+00
9.96604040e-02 1.69369921e-01 -9.10825849e-01 8.71160924e-02
-6.64008141e-01 -1.27558959e+00 8.26374173e-01 5.76020241e-01
4.20946717e-01 -5.17468750e-01 -4.14461017e-01 -8.29871893e-02
9.71847028e-02 -9.21053290e-01 -6.18087232e-01 2.81464726e-01
-8.88127685e-01 -1.35217524e+00 -1.17177653e+00 -8.34301472e-01
1.17700160e+00 1.15839124e-01 9.77344930e-01 1.31064713e-01
-9.72432196e-01 -1.76038533e-01 -3.02147359e-01 -4.56866115e-01
-2.50377655e-01 -1.82773426e-01 -5.21237731e-01 2.29041442e-01
2.56157726e-01 -1.76352441e-01 -8.89738381e-01 4.79313165e-01
-5.61243057e-01 2.49520084e-03 9.98688102e-01 9.84231651e-01
7.32957423e-01 -2.41380818e-02 5.21233439e-01 -1.07902718e+00
-5.37651591e-02 -2.62450069e-01 -7.00762630e-01 1.60626814e-01
-4.09281433e-01 -1.62478551e-01 4.43612151e-02 -3.89266074e-01
-1.01946473e+00 3.56869072e-01 1.08847968e-01 -1.18999280e-01
-8.94334093e-02 8.02339986e-02 -7.36011099e-03 -2.50647604e-01
5.71361065e-01 1.17348850e-01 -1.69949587e-02 -4.18332189e-01
5.28760515e-02 4.90627259e-01 7.44239330e-01 -1.32549420e-01
6.33097887e-01 9.26949501e-01 7.13668466e-02 -6.70471370e-01
-1.06998599e+00 -6.61641240e-01 -5.55615760e-02 -2.05504596e-02
7.22906947e-01 -1.03844440e+00 -5.65400302e-01 4.70784545e-01
-6.66898012e-01 -2.27094844e-01 -5.32050312e-01 5.04457057e-01
-8.01050290e-03 7.67506510e-02 -5.31462431e-01 -8.33818257e-01
-4.14153516e-01 -1.02413344e+00 1.28162348e+00 6.20621622e-01
1.05221674e-03 -6.37446463e-01 -9.61688608e-02 3.88086408e-01
5.16064644e-01 5.76453686e-01 8.30218792e-01 -7.38938808e-01
-6.71286345e-01 -6.60166800e-01 -6.72577858e-01 2.61275381e-01
-1.83222927e-02 -1.44407183e-01 -1.05750263e+00 -4.77982461e-01
-2.98127383e-01 -1.74212798e-01 1.25717974e+00 5.28409839e-01
1.20510089e+00 -1.82258964e-01 -7.76318312e-01 7.68125772e-01
1.43392158e+00 -6.00111559e-02 5.71845412e-01 7.86937848e-02
7.33924031e-01 4.50282484e-01 5.34099102e-01 2.12047130e-01
1.70701459e-01 4.81490940e-01 4.45410103e-01 -6.54644191e-01
-6.64633512e-01 -1.38953552e-01 2.35083893e-01 9.03142765e-02
7.53442422e-02 -1.15479991e-01 -7.83520103e-01 6.83565736e-01
-1.51584029e+00 -5.59037030e-01 -3.73471618e-01 1.99763048e+00
9.45768893e-01 1.15639053e-01 6.51132539e-02 1.35102212e-01
8.26026559e-01 -9.11293253e-02 -6.74534440e-01 5.43424070e-01
-2.87711561e-01 -1.78573892e-01 8.34196031e-01 3.76958042e-01
-1.11515677e+00 6.12837672e-01 5.65068531e+00 1.21455801e+00
-1.18166602e+00 4.05135870e-01 9.62589502e-01 -1.93093702e-01
2.41597239e-02 -5.42175137e-02 -1.36801016e+00 3.84570956e-01
2.11197808e-01 8.54330733e-02 -2.43178949e-01 8.48347783e-01
7.62360469e-02 -1.52317435e-01 -1.08388674e+00 9.65559185e-01
1.78130660e-02 -1.26354671e+00 -1.14438668e-01 2.69388884e-01
7.79632628e-01 -6.85183033e-02 1.96751561e-02 5.08801341e-01
2.18449041e-01 -1.07529950e+00 4.57951903e-01 4.16260064e-01
9.35465276e-01 -4.81281549e-01 1.21357107e+00 1.05839297e-01
-1.13606000e+00 -1.14879854e-01 -1.23454407e-01 6.43993795e-01
-1.06718063e-01 6.78554296e-01 -1.01775825e+00 3.02070707e-01
3.95281434e-01 5.91698289e-01 -7.88337350e-01 1.40152156e+00
-8.76351520e-02 6.89047337e-01 -4.59521860e-01 2.69723926e-02
1.02112852e-01 3.16027313e-01 5.82629502e-01 1.25485456e+00
1.93470493e-02 7.59669766e-02 1.03246823e-01 7.81071126e-01
-2.69522607e-01 7.12532029e-02 1.57905191e-01 1.29422516e-01
2.55540103e-01 1.53691781e+00 -9.79847133e-01 -2.95030922e-01
-3.13438088e-01 8.14710259e-01 -1.84384331e-01 3.22073609e-01
-8.40878010e-01 -2.38628045e-01 1.82938874e-01 4.52591985e-01
1.16628058e-01 4.95428771e-01 -4.91893053e-01 -9.69853520e-01
2.60108691e-02 -7.23282099e-01 6.79327428e-01 -1.96801215e-01
-1.28648257e+00 2.21573025e-01 -3.59169245e-01 -1.40836298e+00
4.54148561e-01 -5.14036894e-01 -7.95092583e-01 5.25468647e-01
-1.85505033e+00 -1.25944161e+00 -7.89504111e-01 4.02577162e-01
3.94408196e-01 8.89255777e-02 2.54290998e-01 6.13025904e-01
-8.22774827e-01 1.28007019e+00 -4.59959283e-02 3.23937863e-01
9.33490872e-01 -1.50618088e+00 -3.76620442e-01 5.38196862e-01
-5.10253072e-01 1.92995146e-01 4.35406685e-01 -5.24199009e-01
-1.07172537e+00 -1.35473788e+00 5.62332094e-01 -4.52592552e-01
5.50293565e-01 -1.97397098e-01 -7.29638755e-01 4.08276737e-01
-3.46755892e-01 6.46815062e-01 5.84268987e-01 -4.08118516e-01
-1.39096484e-01 -2.12556139e-01 -1.37967324e+00 5.43582261e-01
6.49965942e-01 -6.92941770e-02 4.61088540e-03 4.85679626e-01
5.01417458e-01 -5.58413982e-01 -6.80213451e-01 8.45297635e-01
4.48276222e-01 -6.42824590e-01 8.23789239e-01 -3.86932611e-01
-7.39556504e-03 -5.83483279e-01 -6.25908002e-02 -5.51320136e-01
9.58652869e-02 -3.97877008e-01 -3.26012610e-03 1.15574205e+00
6.02226675e-01 -4.43028092e-01 1.28345871e+00 3.22451144e-01
-3.14962626e-01 -8.79228115e-01 -9.70594049e-01 -6.71966732e-01
4.70753983e-02 1.03938363e-01 2.22509399e-01 8.78656507e-01
-2.03235403e-01 6.31964905e-03 -1.65410593e-01 2.73791254e-01
8.64818752e-01 -1.41805917e-01 7.58255601e-01 -1.05530143e+00
-3.73813391e-01 -7.79705822e-01 -5.31515658e-01 -7.34312177e-01
-4.53181982e-01 -1.06944764e+00 -2.56896429e-02 -1.23223829e+00
6.21936738e-01 -5.87768793e-01 -5.46889901e-01 7.03847587e-01
-4.32001650e-01 7.36260772e-01 -2.81280354e-02 1.39127284e-01
-7.02439308e-01 3.33231568e-01 1.37747765e+00 -3.16897780e-01
-7.02926591e-02 -7.01724589e-02 -7.12929726e-01 9.18792069e-01
4.94618714e-01 -6.34411633e-01 -7.92405978e-02 2.39327952e-01
-1.08644672e-01 -6.24957494e-02 6.74727201e-01 -9.18030918e-01
2.39563033e-01 -2.05423478e-02 5.46256065e-01 -7.87434697e-01
6.26067743e-02 -6.60609186e-01 -1.22850247e-01 6.90419614e-01
-3.51286888e-01 -6.69387519e-01 2.19354153e-01 8.29293966e-01
-1.53952047e-01 -1.25658497e-01 1.31363463e+00 -5.64526618e-02
-2.77463913e-01 1.80682570e-01 -1.05402350e-01 1.80583507e-01
1.28846240e+00 -4.54522610e-01 -5.75793266e-01 1.78611144e-01
-8.39608908e-01 5.78977406e-01 2.78981507e-01 -4.35401089e-02
2.16192037e-01 -1.07515180e+00 -8.80308747e-01 1.20666355e-01
3.00371468e-01 3.50301683e-01 3.98939610e-01 1.55336583e+00
-3.64379168e-01 2.66940981e-01 3.48652542e-01 -7.57269382e-01
-1.48774266e+00 1.32596806e-01 6.96650565e-01 -6.84846163e-01
-7.84388602e-01 1.30255544e+00 5.34695089e-01 -2.16229543e-01
7.93255329e-01 -4.85648662e-01 -7.62388557e-02 4.34652977e-02
6.17976189e-01 2.43651867e-01 1.32501617e-01 -2.38013878e-01
-4.62600261e-01 2.56218940e-01 -6.11186683e-01 4.10659790e-01
9.39679265e-01 1.84004560e-01 5.33552170e-02 5.16481698e-03
9.81723428e-01 1.06671676e-01 -1.29467046e+00 -1.29775703e-01
2.11109117e-01 -3.55467439e-01 2.67043471e-01 -1.10662115e+00
-1.14285266e+00 5.62579930e-01 1.09517241e+00 -4.13100541e-01
1.01769245e+00 1.51057750e-01 7.76297212e-01 7.98638761e-02
-2.63723675e-02 -8.25250506e-01 -1.78060867e-02 -2.70818193e-02
4.43979859e-01 -1.50112903e+00 5.85537329e-02 -4.65642959e-01
-5.91685593e-01 6.78206205e-01 5.27178168e-01 -2.66843494e-02
3.56680602e-01 5.51532209e-01 2.97315300e-01 -1.71925366e-01
-6.45753741e-01 -2.77941644e-01 2.10570857e-01 3.52206349e-01
3.00987303e-01 2.44590893e-01 -4.49484348e-01 9.68889773e-01
9.53634977e-02 -7.66627043e-02 3.83504361e-01 7.38753200e-01
-5.33052742e-01 -8.13291967e-01 -3.46481144e-01 6.53129935e-01
-7.07299590e-01 -1.03301942e-01 -4.03198570e-01 1.02840841e+00
1.50146678e-01 2.82775283e-01 -1.25480041e-01 1.01389736e-01
2.26096407e-01 -2.03394070e-01 1.80446327e-01 -4.15767431e-01
-5.66152871e-01 4.36193436e-01 9.65112168e-03 -2.70630449e-01
-1.38138950e-01 -5.14187932e-01 -1.45523286e+00 2.83650041e-01
-6.20293975e-01 1.89152092e-01 5.80571413e-01 7.38538980e-01
4.15027402e-02 6.17479503e-01 3.44238013e-01 -2.96114415e-01
-7.80950963e-01 -8.58378291e-01 -7.33970582e-01 2.83412546e-01
6.66017711e-01 -4.74813849e-01 -6.52164102e-01 1.09243579e-01] | [15.004589080810547, -2.9182851314544678] |
c5132add-6e8f-4367-93b2-6bfd2b057805 | permuted-adain-enhancing-the-representation | 2010.05785 | null | https://arxiv.org/abs/2010.05785v3 | https://arxiv.org/pdf/2010.05785v3.pdf | Permuted AdaIN: Reducing the Bias Towards Global Statistics in Image Classification | Recent work has shown that convolutional neural network classifiers overly rely on texture at the expense of shape cues. We make a similar but different distinction between shape and local image cues, on the one hand, and global image statistics, on the other. Our method, called Permuted Adaptive Instance Normalization (pAdaIN), reduces the representation of global statistics in the hidden layers of image classifiers. pAdaIN samples a random permutation $\pi$ that rearranges the samples in a given batch. Adaptive Instance Normalization (AdaIN) is then applied between the activations of each (non-permuted) sample $i$ and the corresponding activations of the sample $\pi(i)$, thus swapping statistics between the samples of the batch. Since the global image statistics are distorted, this swapping procedure causes the network to rely on cues, such as shape or texture. By choosing the random permutation with probability $p$ and the identity permutation otherwise, one can control the effect's strength. With the correct choice of $p$, fixed apriori for all experiments and selected without considering test data, our method consistently outperforms baselines in multiple settings. In image classification, our method improves on both CIFAR100 and ImageNet using multiple architectures. In the setting of robustness, our method improves on both ImageNet-C and Cifar-100-C for multiple architectures. In the setting of domain adaptation and domain generalization, our method achieves state of the art results on the transfer learning task from GTAV to Cityscapes and on the PACS benchmark. | ['Lior Wolf', 'Sagie Benaim', 'Oren Nuriel'] | 2020-10-09 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Nuriel_Permuted_AdaIN_Reducing_the_Bias_Towards_Global_Statistics_in_Image_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Nuriel_Permuted_AdaIN_Reducing_the_Bias_Towards_Global_Statistics_in_Image_CVPR_2021_paper.pdf | cvpr-2021-1 | ['synthetic-to-real-translation'] | ['computer-vision'] | [ 2.60150909e-01 -1.90941706e-01 -2.08714217e-01 -5.34813166e-01
-5.62669098e-01 -5.93067646e-01 6.34885848e-01 2.94734109e-02
-9.12806332e-01 5.84878862e-01 -1.30604699e-01 -2.83659786e-01
-1.11096509e-01 -8.74517918e-01 -1.01583850e+00 -1.13575625e+00
-1.03954978e-01 3.79402816e-01 2.56578624e-01 -6.01558201e-02
1.99673951e-01 5.80195189e-01 -1.35050237e+00 6.16028309e-01
4.36121464e-01 1.44818723e+00 3.31561789e-02 3.60991865e-01
-2.13562651e-03 5.70494592e-01 -5.05903006e-01 -2.98903078e-01
5.17453909e-01 -2.54145026e-01 -8.00084412e-01 -1.14774264e-01
1.00499582e+00 -1.06346667e-01 -8.09379369e-02 1.12165320e+00
4.09711570e-01 3.05220544e-01 7.97838271e-01 -1.11518323e+00
-7.66095817e-01 6.37615502e-01 -8.22236836e-01 3.55055600e-01
-4.04913098e-01 3.07000250e-01 1.00996840e+00 -7.56030083e-01
5.82180440e-01 1.21826971e+00 6.43691778e-01 2.74818838e-01
-1.52739727e+00 -8.89811218e-01 5.66509724e-01 1.37513563e-01
-1.32370675e+00 -3.29640508e-01 5.13160229e-01 -5.38113713e-01
9.57917511e-01 6.13005385e-02 2.94558465e-01 8.99213016e-01
1.52927384e-01 5.69743216e-01 1.11849153e+00 -2.39666149e-01
3.07650089e-01 5.73687181e-02 1.49204388e-01 4.80211377e-01
2.64474332e-01 6.01918399e-02 -3.35577488e-01 9.15276781e-02
6.96783423e-01 -1.84928849e-01 -2.18969539e-01 -3.81046206e-01
-1.12308252e+00 7.00234354e-01 8.77665222e-01 2.17984691e-01
-3.85970294e-01 1.74688801e-01 4.19306189e-01 4.09780800e-01
2.89196581e-01 4.12715167e-01 -7.92095363e-01 3.51484835e-01
-6.86877668e-01 2.53116816e-01 3.63390028e-01 5.77887952e-01
8.45849395e-01 3.49335908e-03 -4.01106775e-01 9.89482939e-01
-1.33573696e-01 6.03961885e-01 6.01743698e-01 -6.80887818e-01
7.08613098e-01 4.35638726e-01 -6.21154346e-02 -1.02883434e+00
-3.63335431e-01 -5.81629455e-01 -8.62641275e-01 5.09612381e-01
8.64699900e-01 -7.54869506e-02 -1.36381900e+00 2.16322780e+00
-1.16160549e-01 2.78708525e-02 -7.31602013e-02 7.24729121e-01
7.59895504e-01 3.23518783e-01 3.93005699e-01 3.28305960e-01
1.35096776e+00 -7.13074148e-01 -5.39522432e-03 -6.03854537e-01
6.00817561e-01 -4.94262010e-01 1.30663919e+00 3.93627852e-01
-7.77701795e-01 -6.38446867e-01 -1.07635915e+00 1.23859711e-01
-5.95271111e-01 2.91772276e-01 2.42172122e-01 4.34414178e-01
-1.09052396e+00 7.10703015e-01 -7.22475767e-01 -1.07214816e-01
7.74946690e-01 5.47589481e-01 -5.21151662e-01 -1.90626413e-01
-9.02382195e-01 7.51854479e-01 3.16108197e-01 -1.87968567e-01
-5.65130651e-01 -8.20215225e-01 -7.46253490e-01 2.47210547e-01
8.14909339e-02 -4.16604072e-01 9.45243716e-01 -1.48910475e+00
-1.08783507e+00 1.00039995e+00 2.56941505e-02 -7.32520163e-01
3.86234879e-01 1.89290807e-01 -6.25124350e-02 5.25412597e-02
9.81384218e-02 1.07306623e+00 9.87177551e-01 -1.03934193e+00
-6.62332952e-01 -5.11685133e-01 -1.88743979e-01 1.75253108e-01
-1.14479937e-01 -1.99689165e-01 -4.76352453e-01 -6.23568356e-01
1.14748925e-01 -1.03727579e+00 -2.96546184e-02 -9.28067639e-02
-1.86176017e-01 -1.34397447e-01 6.13618612e-01 -4.32203442e-01
6.82269454e-01 -2.54916716e+00 -4.81306538e-02 4.74741727e-01
-1.71972904e-02 1.81837201e-01 -5.00509322e-01 -3.64540815e-01
-6.44995213e-01 1.96116403e-01 -3.42244267e-01 -2.33775094e-01
-1.31597832e-01 3.21094804e-02 -2.96561480e-01 4.48774219e-01
2.89516836e-01 8.09790492e-01 -5.90466499e-01 -1.33173466e-01
8.63419175e-02 3.94725889e-01 -7.19449520e-01 -3.81149083e-01
-2.01469839e-01 3.52001458e-01 -1.63779870e-01 7.90439025e-02
8.95276487e-01 -3.95154983e-01 9.76471603e-02 -2.35606790e-01
9.40987393e-02 3.06598663e-01 -1.13109052e+00 1.28214073e+00
-3.84455740e-01 6.03321016e-01 3.97850685e-02 -1.20930600e+00
7.97434390e-01 1.91018581e-01 2.98536271e-01 -9.93247986e-01
1.50808826e-01 1.20000347e-01 5.22117794e-01 -1.28818735e-01
8.50561820e-03 -1.19126454e-01 1.17351655e-02 3.42932105e-01
1.63237110e-01 1.45560592e-01 1.59361854e-01 5.43408357e-02
9.88694727e-01 -2.89812908e-02 1.14594720e-01 -4.29748058e-01
4.14328426e-01 -2.12207183e-01 4.51525390e-01 1.08416247e+00
-2.29297116e-01 7.53758967e-01 7.09474683e-01 -4.74918187e-01
-9.40759063e-01 -1.11000586e+00 -3.32321793e-01 1.39095402e+00
-1.92435205e-01 1.53516278e-01 -8.00228953e-01 -8.58169198e-01
1.73521876e-01 5.92557788e-01 -9.97973621e-01 -2.37609833e-01
-7.43169129e-01 -9.74199772e-01 4.08691406e-01 6.87663019e-01
7.68477142e-01 -1.44449842e+00 -3.14405859e-01 8.37542340e-02
-7.25253522e-02 -1.12591636e+00 -5.29082477e-01 5.78948796e-01
-7.93547928e-01 -9.02470767e-01 -4.74111110e-01 -8.02439868e-01
8.64816129e-01 1.64635599e-01 1.12616992e+00 -6.82411045e-02
-1.10578276e-01 5.42031303e-02 -1.85579300e-01 -3.16883236e-01
9.49448273e-02 3.00412327e-01 -2.80389786e-01 3.22398484e-01
3.03531557e-01 -4.68547165e-01 -7.53307700e-01 3.82198960e-01
-9.93419945e-01 -2.95645177e-01 6.38098896e-01 7.77596295e-01
6.09298527e-01 1.15531728e-01 4.43536460e-01 -1.07517803e+00
2.52781391e-01 -4.30080920e-01 -6.18377030e-01 3.78878713e-02
-2.68033445e-01 2.93291450e-01 8.30101848e-01 -6.43679976e-01
-8.29515159e-01 -4.64380346e-02 -2.44030952e-02 -4.13774371e-01
-3.88919950e-01 4.65797752e-01 -3.20624143e-01 -4.97915559e-02
9.66263473e-01 2.14904379e-02 -4.28622887e-02 -4.29223448e-01
2.56088763e-01 6.60053715e-02 6.12953424e-01 -5.97421110e-01
5.16837537e-01 6.11058772e-01 -1.23731755e-01 -7.24013448e-01
-5.39984405e-01 -2.27505155e-02 -6.29202425e-01 2.91006804e-01
9.59807277e-01 -7.73096859e-01 -4.95706528e-01 7.40816057e-01
-7.97366381e-01 -7.34817147e-01 -2.58177668e-01 3.35137695e-01
-1.55837670e-01 -1.29929990e-01 -3.50834876e-01 -2.80841321e-01
6.08896650e-02 -1.42336786e+00 7.51233876e-01 4.90865521e-02
-1.72929704e-01 -7.06779182e-01 -4.13470745e-01 4.58462052e-02
4.55941141e-01 9.85594988e-02 1.32388639e+00 -9.39841032e-01
-5.99774480e-01 -2.93293893e-01 -5.14181852e-01 6.05774999e-01
9.77719501e-02 -1.84105039e-01 -1.10845196e+00 -3.16717744e-01
-2.51166850e-01 -1.60979167e-01 1.33168960e+00 7.98878312e-01
1.56878507e+00 -2.90138215e-01 -2.35693946e-01 7.98261285e-01
1.25668979e+00 3.78768712e-01 7.78296947e-01 5.41467726e-01
5.48851788e-01 5.19808650e-01 1.71330217e-02 1.02942303e-01
5.91223761e-02 5.92879534e-01 4.16003019e-01 -2.78811693e-01
-1.79483473e-01 1.86620563e-01 1.34442285e-01 -1.72408819e-01
-2.56406572e-02 -1.50672868e-01 -8.16375732e-01 5.79793751e-01
-1.50334883e+00 -8.52019370e-01 2.60049641e-01 2.42936397e+00
7.62320876e-01 3.99482936e-01 -5.56899905e-02 -1.11566382e-02
6.08074486e-01 2.60406882e-01 -6.06462359e-01 -3.53484482e-01
-3.15821707e-01 7.02908337e-01 8.62974107e-01 4.99718994e-01
-1.30981970e+00 9.60383892e-01 5.49393702e+00 8.63334537e-01
-1.58168125e+00 -5.62555343e-02 1.13053679e+00 -2.99096763e-01
3.30072567e-02 -3.90725881e-01 -7.45379448e-01 4.70021665e-01
6.58008814e-01 6.07474387e-01 5.58695495e-01 7.57422924e-01
-9.65306386e-02 -1.67410836e-01 -1.27941215e+00 7.83654869e-01
-8.06897730e-02 -1.21073484e+00 8.45199451e-02 1.27255097e-01
6.78656876e-01 4.66990590e-01 6.36409283e-01 3.46090019e-01
4.89818990e-01 -1.24594724e+00 8.24676991e-01 1.64733693e-01
6.56590223e-01 -6.21358156e-01 6.52647793e-01 9.50443596e-02
-8.68567109e-01 -2.37086385e-01 -1.87002838e-01 1.43832430e-01
-4.84947950e-01 5.04364312e-01 -6.36679590e-01 -4.03236374e-02
1.03133368e+00 4.80420440e-01 -6.68507993e-01 7.88632512e-01
-2.06214204e-01 6.29322052e-01 -4.58648443e-01 2.58467138e-01
3.45699579e-01 -8.97454992e-02 2.98915267e-01 1.20772290e+00
7.53322020e-02 -6.68692291e-02 4.46800201e-04 7.99351990e-01
-3.68580908e-01 6.25690296e-02 -4.49503690e-01 3.49847257e-01
3.93765241e-01 1.06985092e+00 -8.90453398e-01 -4.29263562e-01
-3.93892109e-01 6.53524041e-01 4.33863938e-01 6.58035159e-01
-7.36336529e-01 -2.66102910e-01 8.25014174e-01 2.47823834e-01
8.16659331e-01 -1.08543359e-01 -6.81336284e-01 -8.69190812e-01
9.76504609e-02 -9.39985394e-01 5.52849948e-01 -6.17764771e-01
-1.22481370e+00 7.23280728e-01 1.10954478e-01 -8.76391232e-01
4.44400497e-02 -1.03953707e+00 -6.35055482e-01 1.07972109e+00
-1.55033827e+00 -9.01984990e-01 -9.19203684e-02 6.00147188e-01
1.51185095e-01 -1.47924364e-01 6.19300067e-01 1.86367378e-01
-7.01722860e-01 8.22662830e-01 5.69015853e-02 6.05217993e-01
9.57940400e-01 -1.09194744e+00 4.60498333e-01 7.63405144e-01
6.68838844e-02 5.63357532e-01 4.15542454e-01 -4.13514733e-01
-6.18613422e-01 -1.17065990e+00 6.19968951e-01 -2.63744980e-01
3.97351861e-01 -4.42864418e-01 -1.07127786e+00 8.29283476e-01
8.23762044e-02 3.47664982e-01 3.53086740e-01 1.09041847e-01
-8.69699657e-01 -4.32952553e-01 -1.20639002e+00 6.15929902e-01
7.76777983e-01 -3.96699697e-01 -1.88269138e-01 1.32323608e-01
3.56083810e-01 -2.08281204e-01 -6.29196942e-01 4.70246732e-01
5.27767599e-01 -9.00608003e-01 1.18906248e+00 -6.57404900e-01
3.82946342e-01 -2.33717546e-01 -3.15363258e-01 -1.38903451e+00
-6.28934741e-01 -9.86370957e-04 7.54448116e-01 1.01298213e+00
8.00101101e-01 -8.82483304e-01 7.55660295e-01 5.77459931e-01
-3.75144668e-02 -6.72502041e-01 -9.01012003e-01 -6.01225078e-01
5.15305221e-01 -4.10978824e-01 7.02755868e-01 8.42318892e-01
-6.09899879e-01 1.42004535e-01 1.45458922e-01 1.18276589e-01
3.29459250e-01 -1.50446832e-01 4.78123844e-01 -1.03891933e+00
-4.34037805e-01 -8.93760264e-01 -2.02902779e-01 -7.85156131e-01
1.81881011e-01 -9.86582041e-01 1.19181171e-01 -9.56166029e-01
2.04305183e-02 -6.60261095e-01 -6.65361941e-01 9.23605502e-01
-1.35351583e-01 5.84583998e-01 2.26991445e-01 1.08760044e-01
-2.71890223e-01 6.88507259e-02 1.17315316e+00 -3.09564054e-01
-2.18685523e-01 -1.46827340e-01 -8.66292775e-01 7.73774266e-01
8.54694664e-01 -4.19892520e-01 -3.44724804e-01 -6.55980229e-01
1.13533966e-01 -4.08615738e-01 5.87302327e-01 -9.22162533e-01
-6.40537292e-02 1.23281432e-02 8.07565629e-01 -2.87747324e-01
1.67789251e-01 -7.56243229e-01 -3.13516229e-01 2.10974440e-01
-5.63329279e-01 9.22381729e-02 4.66099948e-01 2.85782844e-01
-2.61414610e-03 -5.81958331e-02 1.35836804e+00 -2.55346149e-01
-5.44079661e-01 4.89225447e-01 -1.93949580e-01 2.29760617e-01
5.32801867e-01 -1.95919290e-01 -4.55560684e-01 -1.28433585e-01
-7.18153298e-01 -7.94541314e-02 2.80000150e-01 3.26324403e-01
2.27320671e-01 -1.15594041e+00 -6.33212209e-01 6.52360618e-01
1.12336971e-01 1.28967702e-01 2.08110169e-01 9.17040408e-01
-2.52095312e-01 2.35527039e-01 -4.17198837e-01 -6.94486856e-01
-9.62954104e-01 3.23861301e-01 6.30257785e-01 -2.09150627e-01
-3.35581660e-01 1.11503863e+00 7.93293715e-01 -3.01546872e-01
2.48020008e-01 -6.98605895e-01 -2.19176158e-01 8.82831365e-02
4.75725144e-01 -1.25447586e-01 5.54446757e-01 -6.24002993e-01
-3.96153778e-01 5.03988385e-01 -3.90818268e-01 -2.80817777e-01
1.35805738e+00 2.13749811e-01 -1.17911272e-01 1.35204539e-01
1.47342277e+00 -2.48494908e-01 -1.46691072e+00 -3.17974836e-01
-1.02344550e-01 -3.50721896e-01 6.07889332e-02 -1.08261120e+00
-1.53742278e+00 8.26745212e-01 7.36202538e-01 -4.73760925e-02
1.14318299e+00 7.19226897e-02 3.70730102e-01 4.21014249e-01
-1.86087620e-02 -8.73444378e-01 2.71777138e-02 7.85931706e-01
8.76501977e-01 -1.19790852e+00 -3.15582037e-01 -9.61729139e-03
-7.49626696e-01 7.12943912e-01 6.99624479e-01 -5.54276168e-01
7.86591351e-01 2.45442688e-01 1.56717807e-01 -3.17518823e-02
-5.34047127e-01 7.07092807e-02 4.45673436e-01 5.75137258e-01
5.06796479e-01 9.13321525e-02 1.73841923e-01 5.76061249e-01
-2.91290939e-01 -3.54136705e-01 -4.69533689e-02 6.05163515e-01
-3.24518740e-01 -1.02170455e+00 -4.73289669e-01 5.94744861e-01
-4.83756661e-01 -3.95438731e-01 -2.78321087e-01 1.03249276e+00
4.88981456e-01 5.04456580e-01 5.62902629e-01 -1.42177492e-01
2.42831662e-01 2.85561442e-01 4.84067351e-01 -4.89734083e-01
-6.96147561e-01 1.02385603e-01 -1.81889147e-01 -3.83613557e-01
-1.54612914e-01 -8.45875263e-01 -1.29322612e+00 -2.11750403e-01
1.03720665e-01 -1.38806745e-01 6.01427615e-01 1.04722214e+00
3.22971672e-01 4.33685601e-01 2.96830446e-01 -8.95040333e-01
-4.23471987e-01 -8.75789881e-01 -5.57219326e-01 7.21278608e-01
3.79831553e-01 -7.20789373e-01 -2.51486003e-01 1.11379966e-01] | [9.297712326049805, 2.2386579513549805] |
b9240d3e-776f-4967-ae1a-0c2ffddf5024 | a-neuromorphic-architecture-for-reinforcement | 2307.02947 | null | https://arxiv.org/abs/2307.02947v1 | https://arxiv.org/pdf/2307.02947v1.pdf | A Neuromorphic Architecture for Reinforcement Learning from Real-Valued Observations | Reinforcement Learning (RL) provides a powerful framework for decision-making in complex environments. However, implementing RL in hardware-efficient and bio-inspired ways remains a challenge. This paper presents a novel Spiking Neural Network (SNN) architecture for solving RL problems with real-valued observations. The proposed model incorporates multi-layered event-based clustering, with the addition of Temporal Difference (TD)-error modulation and eligibility traces, building upon prior work. An ablation study confirms the significant impact of these components on the proposed model's performance. A tabular actor-critic algorithm with eligibility traces and a state-of-the-art Proximal Policy Optimization (PPO) algorithm are used as benchmarks. Our network consistently outperforms the tabular approach and successfully discovers stable control policies on classic RL environments: mountain car, cart-pole, and acrobot. The proposed model offers an appealing trade-off in terms of computational and hardware implementation requirements. The model does not require an external memory buffer nor a global error gradient computation, and synaptic updates occur online, driven by local learning rules and a broadcasted TD-error signal. Thus, this work contributes to the development of more hardware-efficient RL solutions. | ['Saeed Afshar', 'Teresa B. Ludermir', 'Yeshwanth Bethi', 'Sergio F. Chevtchenko'] | 2023-07-06 | null | null | null | null | ['reinforcement-learning-1', 'acrobot', 'decision-making'] | ['methodology', 'playing-games', 'reasoning'] | [ 6.99657276e-02 -3.51887256e-01 -9.50271171e-03 -3.07915919e-02
-5.48764527e-01 -2.24210188e-01 5.44659793e-01 1.80979267e-01
-1.10079670e+00 1.14071405e+00 -5.00615060e-01 2.58956198e-02
-2.12432668e-01 -4.73868638e-01 -9.19557214e-01 -1.22068179e+00
-3.36080730e-01 2.60126919e-01 5.90119779e-01 -3.85061532e-01
4.97204274e-01 5.62241375e-01 -1.74387968e+00 -6.49612322e-02
6.67231441e-01 1.15167177e+00 6.69744015e-01 4.92644399e-01
2.37996861e-01 9.50246871e-01 -6.14260912e-01 2.45618790e-01
2.18451932e-01 -6.05987787e-01 -3.21794264e-02 -5.33699989e-01
-2.92081863e-01 5.41093685e-02 -6.63501024e-02 7.08054423e-01
1.06960893e+00 3.26299697e-01 3.10498625e-01 -1.25706899e+00
-1.81093022e-01 6.02684259e-01 -3.12122762e-01 3.66626978e-01
-1.42206475e-01 5.41124940e-01 5.95910609e-01 -5.48795760e-01
6.23544157e-01 9.40411568e-01 7.04855382e-01 6.08444154e-01
-1.42948985e+00 -7.46167064e-01 3.36656123e-01 2.88112581e-01
-1.26275158e+00 -4.83338892e-01 7.19370723e-01 1.82104886e-01
1.44867766e+00 -1.89016685e-01 1.12688303e+00 1.30233765e+00
7.11605132e-01 8.22887123e-01 1.33493948e+00 -3.38276803e-01
1.27149117e+00 -1.47608861e-01 -2.88942873e-01 5.36730170e-01
2.97334164e-01 3.27040166e-01 -8.79133701e-01 -1.37452513e-01
8.86681736e-01 -2.82520771e-01 -5.76912090e-02 -3.78382444e-01
-8.98456275e-01 5.04456758e-01 3.43812406e-01 2.25173950e-01
-7.69721687e-01 9.34297681e-01 3.55124712e-01 2.00707942e-01
-1.16458297e-01 6.15281641e-01 -2.60984361e-01 -3.45555156e-01
-8.50400388e-01 3.75267625e-01 6.08247876e-01 5.08574069e-01
3.96549225e-01 7.35090077e-01 4.36601900e-02 6.40767097e-01
4.09687936e-01 5.20988941e-01 8.15934896e-01 -1.31159496e+00
-1.32512137e-01 2.86038369e-01 3.28735977e-01 -6.82874858e-01
-7.59757698e-01 -4.41194892e-01 -3.90873969e-01 8.96649718e-01
2.94623464e-01 -1.78824976e-01 -5.68303466e-01 1.86901653e+00
2.06743643e-01 1.97948530e-01 1.60645559e-01 1.01933503e+00
1.41901121e-01 5.36186814e-01 8.41780901e-02 -5.18314302e-01
9.77319956e-01 -6.37180269e-01 -7.97836661e-01 -3.22855294e-01
2.76873261e-02 -1.15960762e-01 9.22603071e-01 5.82822859e-01
-1.35882843e+00 -2.13816583e-01 -1.32784808e+00 3.18267763e-01
-2.14660272e-01 1.05954908e-01 6.73179328e-01 4.41555202e-01
-1.25839031e+00 8.30127001e-01 -1.22720432e+00 -4.23080325e-01
3.18943799e-01 7.56668448e-01 3.01449060e-01 5.51475048e-01
-1.03731596e+00 9.75146174e-01 2.95298755e-01 1.18195012e-01
-1.09756112e+00 -2.99420506e-01 -2.42234170e-01 -2.93062255e-02
2.97736198e-01 -5.36923587e-01 1.19536674e+00 -1.03837109e+00
-2.35209489e+00 4.53478634e-01 -2.55324971e-02 -1.04128945e+00
2.93145239e-01 1.69891745e-01 -1.43016458e-01 1.78506881e-01
-4.09297496e-01 7.63411701e-01 9.22970355e-01 -1.03187525e+00
-4.86741185e-01 -2.97348082e-01 -2.72413135e-01 3.07412833e-01
-5.83941974e-02 -1.29280806e-01 -1.53830439e-01 -6.76472962e-01
-1.15861736e-01 -8.75032604e-01 -4.00017262e-01 1.30464330e-01
4.15718019e-01 -1.72501788e-01 3.45183402e-01 1.34444116e-02
1.03892958e+00 -2.04993343e+00 6.56082258e-02 1.75481871e-01
-1.57773659e-01 2.10124686e-01 -1.02575444e-01 5.08915424e-01
4.90103543e-01 -4.17708158e-01 -2.73256153e-01 -1.83932722e-01
8.72418210e-02 2.91757435e-01 -3.44471335e-01 5.96826732e-01
1.98653251e-01 8.61323595e-01 -9.48999047e-01 -2.91994959e-01
6.87552914e-02 4.02686447e-01 -4.03707981e-01 -7.34253526e-02
-5.40375352e-01 4.64751393e-01 -3.85938108e-01 5.27380049e-01
7.75824264e-02 -7.63106719e-02 4.40594971e-01 2.02402666e-01
-6.93307996e-01 2.32363001e-01 -1.38282025e+00 1.83169484e+00
-3.09723914e-01 5.41434109e-01 1.70685470e-01 -1.01574349e+00
1.23300183e+00 1.98582381e-01 5.79200625e-01 -1.40116370e+00
5.77660561e-01 5.71257651e-01 1.91361070e-01 1.01661170e-02
2.21919417e-01 3.46539259e-01 1.77764222e-01 5.81412971e-01
2.24949628e-01 8.76714066e-02 2.57589072e-01 -1.77066952e-01
1.36982429e+00 5.05257607e-01 2.54570454e-01 -5.20503521e-01
2.94320852e-01 -5.28498366e-02 8.55532646e-01 9.58873332e-01
-4.54205483e-01 -6.24681311e-03 4.03739631e-01 -4.28036302e-01
-7.98277438e-01 -1.01731610e+00 2.67682355e-02 1.12853825e+00
4.86962706e-01 -1.19548097e-01 -6.96724534e-01 1.88702688e-01
-1.11196220e-01 7.01358438e-01 -4.59943712e-01 -3.16584975e-01
-7.80319631e-01 -8.65509748e-01 7.38706470e-01 3.46490771e-01
5.45180142e-01 -1.61624467e+00 -1.67664945e+00 7.42100477e-01
3.34824741e-01 -8.28570366e-01 5.03470451e-02 8.50521028e-01
-1.13026488e+00 -7.48382747e-01 -3.81815612e-01 -7.78924286e-01
4.48037922e-01 -1.48301691e-01 7.96748638e-01 -2.95872867e-01
-4.32875812e-01 3.49706382e-01 -5.83751984e-02 -5.04528940e-01
7.48457313e-02 -3.34796667e-01 3.45115781e-01 -1.05170282e-02
1.37648761e-01 -9.05669153e-01 -7.65214741e-01 2.23824248e-01
-5.51770985e-01 -1.28971204e-01 6.79162085e-01 1.10086811e+00
1.16039443e+00 -2.69337118e-01 9.77313161e-01 -3.08792681e-01
8.85103703e-01 -2.34046564e-01 -1.17417455e+00 1.92068666e-01
-9.18246031e-01 5.90371311e-01 9.37467515e-01 -7.02175021e-01
-9.61339831e-01 2.65084088e-01 4.87073623e-02 -3.78190547e-01
1.65213332e-01 -1.10918768e-02 3.92279446e-01 -4.74187881e-01
8.67473722e-01 6.03765547e-01 -1.06934682e-02 -1.17850117e-01
3.89343426e-02 5.82813956e-02 4.28998560e-01 -7.33639598e-01
-7.51421973e-02 5.56507587e-01 4.72269170e-02 -6.86578870e-01
1.76409081e-01 -3.01606301e-02 -3.85720283e-02 -5.84063470e-01
4.55422878e-01 -7.54618764e-01 -1.42650998e+00 6.36963010e-01
-9.61441576e-01 -9.09221172e-01 -4.46657896e-01 5.81345499e-01
-1.11944973e+00 -2.63257593e-01 -6.38954878e-01 -1.17678261e+00
-5.90576470e-01 -1.05366087e+00 6.34926379e-01 5.01635134e-01
-4.83525954e-02 -5.31084001e-01 4.36334491e-01 -3.47785890e-01
6.60865188e-01 2.26598859e-01 4.62727964e-01 -2.58006662e-01
-5.15518963e-01 3.63872379e-01 2.76830137e-01 -3.32802922e-01
-5.65264702e-01 -1.35492235e-01 -9.66323733e-01 -4.24659073e-01
9.50257704e-02 -6.87526166e-01 7.77359784e-01 6.75064623e-01
9.07947719e-01 -1.05431959e-01 -2.92137414e-01 4.41042662e-01
1.61277092e+00 6.31970763e-01 4.57905114e-01 5.62366009e-01
-7.33216703e-02 1.66358039e-01 4.25484270e-01 9.26022053e-01
3.46311867e-01 6.10069513e-01 6.48039758e-01 3.23561251e-01
8.90192762e-02 -9.30960104e-02 7.54277468e-01 8.02274764e-01
-1.19700603e-01 -8.82641301e-02 -7.00238943e-01 3.97475988e-01
-2.20969296e+00 -1.04283726e+00 2.44731814e-01 2.33505297e+00
8.99575293e-01 2.82846183e-01 1.00297004e-01 8.26441795e-02
4.87462312e-01 -2.11027488e-01 -1.38338077e+00 -6.65054202e-01
-3.20533514e-01 4.88004416e-01 6.98251843e-01 1.36365354e-01
-5.44455707e-01 9.66120064e-01 6.18496084e+00 8.70910585e-01
-1.48838556e+00 9.01975557e-02 1.63125485e-01 -5.51475227e-01
2.15985090e-01 -2.82679141e-01 -8.94136131e-01 6.08691692e-01
1.34173524e+00 -2.19082404e-02 1.14181244e+00 6.50378704e-01
4.96135235e-01 -5.85955381e-01 -7.86299825e-01 1.24086773e+00
-1.16185911e-01 -1.47895730e+00 -5.70742846e-01 -2.08113596e-01
4.88489360e-01 4.18039680e-01 1.61604628e-01 4.08083379e-01
5.94040275e-01 -6.97911978e-01 1.11476350e+00 5.51982343e-01
3.09237689e-01 -8.21610510e-01 3.63828838e-01 3.47833604e-01
-1.15170324e+00 -5.43664813e-01 -4.22598094e-01 -2.00576067e-01
-6.17023855e-02 1.69211447e-01 -4.58258986e-01 -4.47643623e-02
8.98630619e-01 4.72277313e-01 -2.81750649e-01 1.06914437e+00
-1.91027120e-01 5.55224836e-01 -6.29561782e-01 -8.36115241e-01
3.51506025e-01 -1.59873590e-01 3.34229201e-01 1.00289941e+00
1.69864684e-01 9.18068588e-02 -1.23469569e-01 9.89416420e-01
-7.25377724e-02 -9.54896733e-02 -3.71376693e-01 1.58577532e-01
7.47813284e-01 1.04563558e+00 -1.10399485e+00 7.04970509e-02
1.02550350e-02 7.23852336e-01 5.65036952e-01 4.19272363e-01
-9.80006278e-01 -4.43938404e-01 4.57423896e-01 -3.04110855e-01
4.23307687e-01 -3.04445714e-01 -3.86637628e-01 -6.84183061e-01
-1.32075757e-01 -7.59895682e-01 -4.95135449e-02 -6.94314778e-01
-7.37558901e-01 5.21341205e-01 -4.85182345e-01 -9.99303997e-01
-3.14812452e-01 -3.86278570e-01 -4.60592449e-01 3.33960056e-01
-1.58723044e+00 -3.72380078e-01 -6.66039856e-03 8.81799996e-01
4.38211888e-01 -2.40897432e-01 8.15775454e-01 1.12103917e-01
-7.29563117e-01 4.97728556e-01 5.38482428e-01 -3.79420221e-01
4.81765002e-01 -1.00687730e+00 -3.17055136e-02 6.18569434e-01
3.68341431e-02 3.62614363e-01 7.45353997e-01 -3.51614058e-01
-1.87696373e+00 -8.37006986e-01 2.70601481e-01 1.71906486e-01
5.42388976e-01 -4.46584761e-01 -6.93597674e-01 1.05024584e-01
2.70880014e-01 1.15488857e-01 3.04143190e-01 -3.72794807e-01
5.94708063e-02 -5.27106345e-01 -1.16502953e+00 9.67023611e-01
9.08832312e-01 -1.95855290e-01 -3.73005360e-01 -1.57909587e-01
1.75212398e-01 -1.94223598e-01 -5.02682149e-01 5.79052093e-03
6.76553369e-01 -8.81485403e-01 6.75801158e-01 -1.46700412e-01
-2.79139102e-01 -5.01129985e-01 6.27330616e-02 -1.28321314e+00
-1.77780673e-01 -9.14122105e-01 -4.50329959e-01 6.48629248e-01
2.97513932e-01 -7.04378605e-01 5.84373832e-01 3.23558599e-01
-1.75900131e-01 -1.09731758e+00 -1.23528397e+00 -9.46980238e-01
-2.32738033e-01 -3.25192779e-01 2.45009977e-02 3.45680267e-01
1.82190329e-01 5.03529608e-02 -1.45363271e-01 6.09433278e-03
7.27182567e-01 -5.92183955e-02 2.73738116e-01 -7.99540699e-01
-4.00033653e-01 -6.81070149e-01 -2.75713354e-01 -7.23120272e-01
8.26145411e-02 -7.23966002e-01 5.24422586e-01 -1.30627358e+00
-3.44985455e-01 -5.59156239e-01 -7.26765752e-01 5.35435319e-01
3.35699767e-01 4.78635076e-03 1.78308859e-01 2.56296664e-01
-9.77331817e-01 9.53354955e-01 7.32499897e-01 1.08540379e-01
-4.84442741e-01 -3.59669089e-01 -6.96481317e-02 6.49710834e-01
1.27196193e+00 -8.39314222e-01 -4.61716920e-01 -3.89075547e-01
4.62341130e-01 1.32662252e-01 2.72169501e-01 -1.43317711e+00
9.37098444e-01 -1.01455376e-01 4.33875591e-01 -2.26858884e-01
6.27047181e-01 -6.93681121e-01 1.28611876e-02 1.02763605e+00
-4.83125597e-01 3.67031068e-01 4.62715834e-01 8.56985927e-01
1.31761134e-01 -3.46188322e-02 1.12638283e+00 -2.55468991e-02
-8.77747893e-01 -7.65943676e-02 -9.30721462e-01 1.70122683e-01
1.28175449e+00 -3.74038577e-01 -3.70055646e-01 7.15897009e-02
-3.88805330e-01 3.28653425e-01 2.78651327e-01 4.14005667e-02
7.78746784e-01 -9.37604666e-01 -3.91518235e-01 2.36129194e-01
-2.92449385e-01 -5.61839998e-01 -6.07220195e-02 9.80979562e-01
-4.66590255e-01 3.86159748e-01 -9.39483404e-01 -4.78623092e-01
-6.25190973e-01 3.01999837e-01 5.01593232e-01 -2.69155383e-01
-4.58399862e-01 7.33303010e-01 -4.27497208e-01 -2.22875789e-01
6.00908995e-01 -2.30105296e-01 -1.27940223e-01 -6.46321252e-02
4.20269877e-01 5.02613068e-01 9.24775302e-02 1.04141079e-01
-4.32028532e-01 3.71952266e-01 2.86231488e-01 -5.39399683e-01
1.49570858e+00 7.24428594e-02 1.10583983e-01 8.92002642e-01
3.64618927e-01 -5.90175688e-01 -1.75931478e+00 6.07105158e-02
1.79440007e-01 2.18239591e-01 2.40103930e-01 -9.03274059e-01
-8.75852108e-01 7.28784025e-01 9.03351068e-01 -3.89582753e-01
1.16951859e+00 -4.40518320e-01 5.45976460e-01 8.75944436e-01
9.54999030e-01 -1.68753994e+00 2.97842532e-01 6.05758905e-01
7.81117916e-01 -9.40837443e-01 -1.67193413e-01 6.27792716e-01
-6.38079286e-01 1.09651458e+00 6.96110249e-01 -6.99511647e-01
2.78762192e-01 5.08490980e-01 -2.25856438e-01 1.13726050e-01
-1.36265361e+00 -2.10935086e-01 -6.01303041e-01 5.68495750e-01
1.77857820e-02 -1.50035888e-01 -6.33939803e-01 5.11626005e-01
3.74711335e-01 2.58425325e-01 3.27740401e-01 1.22248602e+00
-7.24482596e-01 -8.11125696e-01 -2.57988960e-01 2.69919902e-01
-3.56164396e-01 -2.77017001e-02 -8.51652771e-02 6.99314713e-01
-2.70661451e-02 7.39594281e-01 5.80871999e-02 -7.85043612e-02
1.05903186e-01 -9.63050723e-02 6.67585254e-01 -2.67745033e-02
-1.03179491e+00 2.07866132e-01 -3.04747194e-01 -9.60170627e-01
-3.60744536e-01 -6.07206523e-01 -1.90763283e+00 -1.84362814e-01
-5.77784851e-02 -2.81883739e-02 1.09786665e+00 6.63542986e-01
7.64658213e-01 7.81006277e-01 4.30121481e-01 -1.01788855e+00
-6.26265109e-01 -4.37248141e-01 -6.64831817e-01 -1.78863496e-01
7.71640688e-02 -8.60581279e-01 -2.11088642e-01 -1.78268805e-01] | [8.104851722717285, 2.462430000305176] |
95f61d8a-67d2-4027-82ed-68f55ca0346b | time-series-clustering-via-community | 1508.04757 | null | http://arxiv.org/abs/1508.04757v1 | http://arxiv.org/pdf/1508.04757v1.pdf | Time Series Clustering via Community Detection in Networks | In this paper, we propose a technique for time series clustering using
community detection in complex networks. Firstly, we present a method to
transform a set of time series into a network using different distance
functions, where each time series is represented by a vertex and the most
similar ones are connected. Then, we apply community detection algorithms to
identify groups of strongly connected vertices (called a community) and,
consequently, identify time series clusters. Still in this paper, we make a
comprehensive analysis on the influence of various combinations of time series
distance functions, network generation methods and community detection
techniques on clustering results. Experimental study shows that the proposed
network-based approach achieves better results than various classic or
up-to-date clustering techniques under consideration. Statistical tests confirm
that the proposed method outperforms some classic clustering algorithms, such
as $k$-medoids, diana, median-linkage and centroid-linkage in various data
sets. Interestingly, the proposed method can effectively detect shape patterns
presented in time series due to the topological structure of the underlying
network constructed in the clustering process. At the same time, other
techniques fail to identify such patterns. Moreover, the proposed method is
robust enough to group time series presenting similar pattern but with time
shifts and/or amplitude variations. In summary, the main point of the proposed
method is the transformation of time series from time-space domain to
topological domain. Therefore, we hope that our approach contributes not only
for time series clustering, but also for general time series analysis tasks. | ['Liang Zhao', 'Leonardo N. Ferreira'] | 2015-08-19 | null | null | null | null | ['time-series-clustering'] | ['time-series'] | [-8.78183693e-02 -4.14528221e-01 2.10485488e-01 1.07618891e-01
-7.22009502e-03 -1.00979471e+00 6.47511959e-01 7.07259297e-01
-9.44537669e-02 3.54629278e-01 -1.30921885e-01 -2.54703254e-01
-8.27555716e-01 -1.01808679e+00 -1.04679391e-01 -8.27901185e-01
-7.36465991e-01 4.10355628e-01 3.59951079e-01 -1.66500002e-01
3.98544371e-01 7.59024858e-01 -1.40076005e+00 -1.70937374e-01
8.69785786e-01 7.36327410e-01 -1.01754472e-01 3.00787032e-01
-9.88305584e-02 3.83114636e-01 -7.46524990e-01 1.28005877e-01
1.34429574e-01 -5.06884396e-01 -4.32371110e-01 1.65308997e-01
-5.20323694e-01 5.18624604e-01 -1.41051725e-01 8.83560479e-01
2.42614314e-01 2.93966532e-01 8.18741798e-01 -1.66140854e+00
-2.21290946e-01 8.60399783e-01 -9.11277950e-01 2.24352360e-01
3.20380896e-01 -1.61639094e-01 6.73709333e-01 -4.11776692e-01
8.59027684e-01 1.10825658e+00 7.33784795e-01 -2.33067885e-01
-1.48112130e+00 -6.40808105e-01 1.03230692e-01 2.73939729e-01
-1.70372987e+00 5.73299173e-03 1.30897141e+00 -5.95470250e-01
5.57517886e-01 4.57052678e-01 8.31589699e-01 4.68604892e-01
-2.94396374e-02 2.16384172e-01 1.19613159e+00 -2.95382142e-01
3.17172825e-01 -2.17982709e-01 2.98479378e-01 2.07256824e-01
2.91322976e-01 -4.04720157e-01 1.30821437e-01 -4.85555112e-01
4.54069614e-01 3.23072523e-01 -2.91106254e-01 -4.60056722e-01
-1.52264369e+00 6.65540755e-01 3.47724706e-01 1.11357784e+00
-4.41564322e-01 -1.32419482e-01 5.38473845e-01 4.76483613e-01
4.54507560e-01 2.28522643e-01 -1.17968954e-01 1.64303929e-01
-9.83082294e-01 7.13574067e-02 7.56443262e-01 7.32478440e-01
5.60350776e-01 -9.61789191e-02 1.05617218e-01 3.78804475e-01
1.99991614e-01 2.75727332e-01 4.49041992e-01 -6.47063076e-01
2.45532691e-01 8.70187700e-01 -1.91577837e-01 -1.91661251e+00
-6.70432329e-01 -3.66253316e-01 -1.13926899e+00 -2.14754581e-01
4.97808784e-01 3.58163454e-02 -3.69135886e-01 1.55099165e+00
4.51997072e-01 4.19840068e-01 -3.16763320e-03 4.62836623e-01
4.84856367e-01 7.04025269e-01 -1.47895083e-01 -9.43130374e-01
1.13492453e+00 -2.63919175e-01 -8.95929039e-01 8.03342402e-01
3.16852510e-01 -8.82881463e-01 2.93121040e-01 3.36942703e-01
-7.63845026e-01 -5.07802010e-01 -6.93277657e-01 8.97455335e-01
-5.11663079e-01 -1.14759572e-01 3.66220206e-01 3.98975134e-01
-1.13126945e+00 8.00751448e-01 -8.45492542e-01 -8.19312871e-01
-2.51312912e-01 2.77919114e-01 -3.15521181e-01 2.23004505e-01
-9.89685655e-01 3.10015529e-01 5.99692047e-01 1.35204166e-01
-2.80756146e-01 -2.54113019e-01 -4.39766526e-01 -8.65542097e-04
2.38203689e-01 -2.34028518e-01 4.28694516e-01 -7.67752945e-01
-8.66884947e-01 4.83761996e-01 -6.67199045e-02 -4.02906507e-01
4.58090305e-01 7.47472048e-01 -9.32000220e-01 4.32567745e-01
1.03586435e-01 -3.19833076e-03 7.35545516e-01 -1.22431827e+00
-3.61512005e-01 -3.12946230e-01 -4.06419486e-01 -2.76810467e-01
-4.65421438e-01 1.85123786e-01 -2.23234206e-01 -8.47179592e-01
6.52742743e-01 -9.07449365e-01 -2.33715028e-01 -3.86752576e-01
-4.84695733e-01 -5.50116897e-01 1.09689856e+00 -3.74980271e-01
1.68982685e+00 -2.26298714e+00 2.74839520e-01 1.11548746e+00
5.14385343e-01 -1.11275345e-01 1.88812733e-01 1.18351018e+00
-4.22593951e-01 3.19949389e-01 -3.54309499e-01 1.06392652e-01
-1.75049797e-01 -1.61104277e-02 -1.02157257e-01 7.86274493e-01
-1.36967197e-01 3.25507700e-01 -1.02360845e+00 -6.06010973e-01
2.87103027e-01 4.58440900e-01 -1.04415655e-01 -2.84481347e-01
2.20392689e-01 5.66819072e-01 -3.44483256e-01 4.33643192e-01
6.00758970e-01 -2.02051058e-01 6.04853094e-01 -3.48624676e-01
-5.29387474e-01 -4.11334723e-01 -1.46467626e+00 1.07765651e+00
2.47373506e-01 6.01876140e-01 -9.40687060e-02 -1.40608597e+00
1.39614904e+00 5.83729446e-01 1.07476354e+00 -2.38687798e-01
1.95662841e-01 2.75788784e-01 3.63994330e-01 -3.98527920e-01
1.79815590e-01 1.78036839e-01 1.04419947e-01 5.55685520e-01
-3.01807851e-01 9.30310786e-02 6.43289685e-01 1.85421422e-01
1.09435558e+00 -5.04598677e-01 2.31439516e-01 -3.75722557e-01
7.58689761e-01 6.51473477e-02 4.62969810e-01 1.99101672e-01
-2.33081847e-01 2.27200121e-01 7.46836960e-01 -1.65886760e-01
-1.05154657e+00 -9.86188173e-01 -1.13495409e-01 5.11252940e-01
-2.08390206e-02 -5.18031895e-01 -6.10517383e-01 -2.56113589e-01
5.90591319e-02 1.88372880e-01 -5.90473890e-01 1.55243888e-01
-5.68848491e-01 -7.91154563e-01 3.98197144e-01 5.23424149e-03
2.82877952e-01 -1.04254925e+00 -2.26392716e-01 3.80226374e-01
-3.00370187e-01 -6.98384285e-01 -4.16614503e-01 -6.76193684e-02
-1.30960691e+00 -1.34538114e+00 -5.88471770e-01 -9.17956531e-01
7.73901105e-01 5.28001487e-01 7.31723130e-01 1.34867445e-01
-1.48172557e-01 4.19880182e-01 -7.86428988e-01 -4.21876870e-02
-3.51590425e-01 -1.92076460e-01 9.28266272e-02 5.02995610e-01
2.55450189e-01 -1.07189369e+00 -6.29181802e-01 5.05030155e-01
-9.05851364e-01 -6.17816746e-01 1.20516635e-01 2.40037233e-01
4.70050871e-01 1.06012404e+00 7.24379897e-01 -4.26967919e-01
9.90468264e-01 -8.29790473e-01 -5.15476048e-01 1.89470425e-01
-6.68886960e-01 -3.30483258e-01 8.79132390e-01 -5.91645896e-01
-4.21386302e-01 -5.84085584e-02 5.68866014e-01 -5.18450618e-01
-1.86893299e-01 8.94845188e-01 1.56705052e-01 -1.94706097e-02
5.99931657e-01 4.01367009e-01 1.95806429e-01 -4.68140513e-01
1.77631155e-01 5.92009485e-01 2.90829867e-01 -3.26984048e-01
9.74957049e-01 8.55125248e-01 1.72381625e-01 -9.01283443e-01
2.83072203e-01 -8.25972319e-01 -8.82480681e-01 -6.29737914e-01
6.94326878e-01 -5.10315776e-01 -7.95407891e-01 4.07698005e-01
-9.37381923e-01 2.95875639e-01 1.85444560e-02 5.00972629e-01
-2.07631066e-01 7.05714107e-01 -5.24803221e-01 -1.05978334e+00
-1.54469535e-01 -5.38544238e-01 4.82230842e-01 -5.60920350e-02
-3.24144453e-01 -1.34486270e+00 1.85632661e-01 -1.21208139e-01
1.59474820e-01 1.08581436e+00 9.61525500e-01 -8.13161612e-01
-2.35474110e-01 -3.46257716e-01 6.01345301e-02 -2.53330261e-01
4.30197656e-01 6.64632082e-01 -4.64184165e-01 -5.30542195e-01
7.02438205e-02 6.62260532e-01 3.62613946e-01 5.47382474e-01
9.01923597e-01 -1.88415036e-01 -6.61566377e-01 2.18019962e-01
1.40819585e+00 7.68221915e-01 4.81321931e-01 3.45609218e-01
5.86213946e-01 9.52491462e-01 4.99263614e-01 6.06061578e-01
3.12497646e-01 4.60001528e-01 1.64285406e-01 -2.04926789e-01
3.28481734e-01 9.76751819e-02 8.54408890e-02 1.39648342e+00
-4.80024934e-01 -1.59811109e-01 -1.14277887e+00 7.02425659e-01
-2.04130030e+00 -1.43368876e+00 -8.63056421e-01 2.19534349e+00
4.16383207e-01 4.17879783e-02 6.66237831e-01 9.29434836e-01
1.32453811e+00 -3.90485413e-02 -1.37851134e-01 -1.02848724e-01
-2.48106539e-01 4.59376955e-03 2.58824259e-01 5.51689006e-02
-1.03974485e+00 3.18700045e-01 5.95849848e+00 6.27367735e-01
-1.09796667e+00 -1.23525418e-01 2.64239222e-01 3.23987275e-01
-2.85933465e-01 -1.27327414e-02 8.87424573e-02 4.98745114e-01
8.50860834e-01 -8.05041790e-01 3.49892616e-01 4.67648417e-01
7.43708193e-01 1.53207853e-01 -7.72417605e-01 9.04066801e-01
-2.25249738e-01 -9.16867852e-01 -2.68775940e-01 1.93390608e-01
6.69011712e-01 -2.88907886e-01 -2.49464676e-01 -2.03144431e-01
1.15158651e-02 -7.52012193e-01 4.30600494e-01 5.24780989e-01
2.82382309e-01 -9.52979743e-01 5.32614827e-01 2.74228960e-01
-1.95921874e+00 -1.92838386e-01 3.50451022e-02 -1.23987302e-01
1.45590842e-01 8.56005430e-01 -7.06597090e-01 9.88846004e-01
6.08697355e-01 9.73913789e-01 -5.28979480e-01 1.43665373e+00
2.84046233e-01 5.89114666e-01 -3.69385153e-01 -7.75647834e-02
1.74607322e-01 -6.41524971e-01 6.24992311e-01 9.92221773e-01
6.35236025e-01 2.13583022e-01 2.57119119e-01 7.66426504e-01
5.12652278e-01 3.54253650e-01 -8.62488508e-01 -2.44532198e-01
9.02060270e-01 1.31787145e+00 -1.58209777e+00 -3.16152215e-01
-2.74893075e-01 4.30370897e-01 -2.70821571e-01 4.75132048e-01
-6.91783249e-01 -6.47703230e-01 2.39343598e-01 3.94921035e-01
2.08863318e-01 -5.46761394e-01 -4.35264967e-02 -8.55579436e-01
3.71623226e-02 -7.14954317e-01 6.64390743e-01 -5.87381780e-01
-1.37514031e+00 7.12244332e-01 2.82460898e-01 -1.79311645e+00
-1.95356756e-01 -6.24063239e-02 -8.43888640e-01 5.46934903e-01
-8.59951377e-01 -7.61066675e-01 -2.47584641e-01 1.05414212e+00
1.23061053e-01 1.65390391e-02 5.31859040e-01 4.48226839e-01
-4.26789135e-01 3.02483849e-02 5.21000147e-01 3.21967214e-01
4.39705670e-01 -1.11174810e+00 1.72383741e-01 9.17559385e-01
-9.77478363e-03 9.12018955e-01 7.46397376e-01 -8.59798491e-01
-1.11678207e+00 -9.93329644e-01 9.55238461e-01 -8.20805505e-02
1.06446207e+00 -2.70734578e-01 -9.06977057e-01 2.52550215e-01
2.76185751e-01 -2.82599717e-01 6.48299038e-01 3.40471417e-03
-1.82752386e-01 -2.16533944e-01 -1.20605958e+00 4.83326256e-01
8.01566184e-01 -4.25630659e-01 -5.55769384e-01 3.91223550e-01
4.74928379e-01 4.11266774e-01 -1.37085438e+00 2.82903314e-01
3.59172374e-01 -1.04097295e+00 9.24442053e-01 -1.05798610e-01
-3.85892503e-02 -7.44696259e-01 1.74905300e-01 -1.26075637e+00
-6.60886824e-01 -7.93094635e-01 1.17272109e-01 1.47287393e+00
1.25520587e-01 -7.92088509e-01 4.16650504e-01 -6.66719079e-02
2.03590363e-01 -1.59150615e-01 -7.83484459e-01 -1.02201509e+00
-1.72712713e-01 -2.32919797e-01 6.09817445e-01 1.71699870e+00
4.26979780e-01 5.72282150e-02 6.54424503e-02 1.08782984e-01
1.00030804e+00 3.55125159e-01 4.11687076e-01 -1.76032436e+00
1.08564019e-01 -7.15774000e-01 -6.53489709e-01 4.95881699e-02
-7.84674957e-02 -7.84885764e-01 -3.52487743e-01 -1.37197554e+00
-1.66500106e-01 -6.71325684e-01 -2.90635794e-01 3.06400675e-02
1.62621543e-01 1.07203811e-01 2.21588343e-01 6.82388484e-01
-3.11827391e-01 2.65563056e-02 1.05438221e+00 -3.43200378e-02
-5.45132935e-01 1.19700164e-01 -2.07411379e-01 4.89355177e-01
8.63679171e-01 -3.95002246e-01 -5.83904803e-01 2.56329715e-01
5.00111990e-02 3.25949043e-01 2.23237813e-01 -1.06331670e+00
4.55512553e-01 -4.16763611e-02 6.24211058e-02 -8.08011770e-01
-2.39490375e-01 -1.21820354e+00 9.32532370e-01 6.67863071e-01
9.82721448e-02 6.19396627e-01 3.37388739e-02 7.57433832e-01
-4.24661875e-01 1.06275275e-01 4.57049668e-01 -8.50566011e-03
-5.06784916e-01 2.24384442e-02 -7.16470838e-01 -2.62446225e-01
1.30676842e+00 -4.30235952e-01 -2.35205457e-01 -3.83403063e-01
-1.07304549e+00 2.76050001e-01 4.27389324e-01 3.24991405e-01
3.75129610e-01 -1.54433060e+00 -5.98743618e-01 -5.91297224e-02
1.12201817e-01 -3.43040198e-01 1.40597478e-01 1.27601159e+00
-5.26096702e-01 2.59610683e-01 -3.29673499e-01 -9.19841707e-01
-1.45756352e+00 9.65764582e-01 -2.87309103e-02 -8.44113082e-02
-5.69613814e-01 4.13728170e-02 -2.49476954e-01 -2.80473500e-01
2.10140005e-01 -4.91419971e-01 -6.39343500e-01 6.29565299e-01
1.50447845e-01 6.42272413e-01 -9.47923437e-02 -8.43568027e-01
-5.14809847e-01 1.00516510e+00 5.45958877e-01 -3.82595249e-02
1.39487553e+00 -3.22609156e-01 -6.54474437e-01 7.62400627e-01
1.15075159e+00 -2.08003353e-02 -4.68779236e-01 6.27995655e-02
3.30718786e-01 -2.21548542e-01 -4.75647777e-01 -2.42791981e-01
-1.06949306e+00 3.80781651e-01 5.02351344e-01 1.29427063e+00
1.44303286e+00 -9.40021798e-02 2.89638609e-01 1.52428150e-01
3.55336338e-01 -9.43238497e-01 -1.20915860e-01 1.01812035e-01
6.55013025e-01 -6.73600674e-01 -1.47525147e-01 -6.29472435e-01
-2.18568414e-01 1.27387893e+00 1.06429547e-01 -2.48724923e-01
9.27962720e-01 2.04062834e-02 -1.57482326e-01 -4.49845433e-01
-5.02112448e-01 -3.53372693e-01 1.90509483e-01 7.28107512e-01
5.16553998e-01 2.20698953e-01 -8.28664422e-01 7.43382424e-02
-2.13653564e-01 -2.36740693e-01 5.28299809e-01 7.40574837e-01
-4.00283933e-01 -1.00966108e+00 -9.08704042e-01 2.15869680e-01
-9.77476612e-02 3.01239967e-01 -6.40450656e-01 1.12808836e+00
9.89675373e-02 1.42369998e+00 2.90683448e-01 -5.94962895e-01
4.74994570e-01 4.41157483e-02 1.02637112e-01 -9.77614671e-02
-8.21734488e-01 6.17148221e-01 -8.66745263e-02 -2.53692478e-01
-8.90653670e-01 -9.24671412e-01 -1.21338880e+00 -6.72670603e-01
-3.87556702e-01 6.29625261e-01 4.18468773e-01 5.14467120e-01
2.30292872e-01 5.30844510e-01 1.10285985e+00 -5.40844202e-01
1.95913345e-01 -9.00289893e-01 -9.04791474e-01 6.18104935e-01
8.56171921e-02 -5.54394841e-01 -6.61212683e-01 1.74120486e-01] | [7.253254413604736, 3.4229860305786133] |
977a2de3-38a9-4529-9a01-5e16101e32b9 | referee-reference-free-sentence-summarization | 2210.13800 | null | https://arxiv.org/abs/2210.13800v1 | https://arxiv.org/pdf/2210.13800v1.pdf | Referee: Reference-Free Sentence Summarization with Sharper Controllability through Symbolic Knowledge Distillation | We present Referee, a novel framework for sentence summarization that can be trained reference-free (i.e., requiring no gold summaries for supervision), while allowing direct control for compression ratio. Our work is the first to demonstrate that reference-free, controlled sentence summarization is feasible via the conceptual framework of Symbolic Knowledge Distillation (West et al., 2022), where latent knowledge in pre-trained language models is distilled via explicit examples sampled from the teacher models, further purified with three types of filters: length, fidelity, and Information Bottleneck. Moreover, we uniquely propose iterative distillation of knowledge, where student models from the previous iteration of distillation serve as teacher models in the next iteration. Starting off from a relatively modest set of GPT3-generated summaries, we demonstrate how iterative knowledge distillation can lead to considerably smaller, but better summarizers with sharper controllability. A useful by-product of this iterative distillation process is a high-quality dataset of sentence-summary pairs with varying degrees of compression ratios. Empirical results demonstrate that the final student models vastly outperform the much larger GPT3-Instruct model in terms of the controllability of compression ratios, without compromising the quality of resulting summarization. | ['Yejin Choi', 'Yulia Tsvetkov', 'Sachin Kumar', 'Peter West', 'Melanie Sclar'] | 2022-10-25 | null | null | null | null | ['abstractive-sentence-summarization'] | ['natural-language-processing'] | [ 5.57764709e-01 7.58573771e-01 -3.81803066e-01 -1.48082748e-01
-1.24550474e+00 -7.46420443e-01 5.90236247e-01 4.82153177e-01
-1.58668667e-01 1.12038171e+00 8.39949369e-01 -3.77571523e-01
-2.53933609e-01 -7.81876326e-01 -8.56818557e-01 -2.76406050e-01
2.11326238e-02 5.34912288e-01 7.65826553e-02 -2.61420369e-01
5.80480456e-01 1.32472068e-01 -1.30595160e+00 2.38522694e-01
1.64271319e+00 3.15967470e-01 2.95218647e-01 1.20894957e+00
2.76208650e-02 1.02576220e+00 -9.03939605e-01 -6.10645711e-01
-1.41806230e-01 -7.72331893e-01 -1.10236692e+00 -1.33804465e-02
7.00432122e-01 -5.27886927e-01 -5.93901515e-01 7.75493443e-01
5.14752984e-01 2.84409165e-01 7.46329784e-01 -5.52133918e-01
-8.34492624e-01 1.21704352e+00 -1.54671133e-01 1.76633716e-01
5.34340441e-01 3.44804078e-01 1.19228017e+00 -5.02258897e-01
4.73886430e-01 1.26826036e+00 5.67259908e-01 5.12323916e-01
-1.46726131e+00 -3.85354698e-01 6.95820525e-02 -5.64412922e-02
-8.64553511e-01 -7.26901889e-01 2.66283184e-01 -1.40139818e-01
1.22405517e+00 5.94512165e-01 7.62783051e-01 8.00494313e-01
-4.16431911e-02 1.03711438e+00 4.59672570e-01 -6.06720984e-01
1.75180659e-01 1.77694663e-01 3.99872899e-01 8.08558047e-01
5.74137866e-01 -2.74857014e-01 -8.15970719e-01 -1.83546588e-01
4.17870462e-01 -3.40526313e-01 -6.03040993e-01 -1.23363659e-01
-1.11807299e+00 7.56472409e-01 1.83532298e-01 3.98823805e-02
-2.73537397e-01 1.98292807e-01 5.11342466e-01 4.45733130e-01
5.91636598e-01 1.05776548e+00 -4.48497206e-01 -6.39680386e-01
-1.58778644e+00 4.43458289e-01 1.25038624e+00 1.29187608e+00
5.83420157e-01 1.77459225e-01 -7.38965511e-01 5.73101997e-01
-2.07196698e-01 5.90105176e-01 4.99729961e-01 -1.42318869e+00
7.63767004e-01 3.85616392e-01 7.61353597e-02 -7.45403290e-01
2.00108454e-01 -5.41205049e-01 -6.78188443e-01 -5.32583773e-01
5.11487722e-02 -3.45908999e-01 -7.14473069e-01 1.66916192e+00
-1.47319645e-01 9.46501195e-02 4.87832248e-01 1.67102545e-01
8.82123590e-01 9.57113743e-01 -2.36756548e-01 -5.73371708e-01
9.78107870e-01 -1.25328493e+00 -5.79555869e-01 -2.52723277e-01
9.09549713e-01 -4.86377060e-01 1.06309116e+00 4.47554111e-01
-1.87834799e+00 -3.26820403e-01 -1.22024369e+00 -3.28283966e-01
2.09504098e-01 1.02323115e-01 6.03938043e-01 5.75662315e-01
-1.41426730e+00 1.12163281e+00 -9.17539358e-01 -3.59995812e-01
4.58988518e-01 2.04876751e-01 -1.59042925e-02 2.50251442e-02
-9.70576286e-01 1.02870083e+00 5.31602740e-01 -3.38213831e-01
-8.09367716e-01 -1.08164465e+00 -8.13241005e-01 5.45145392e-01
5.01585424e-01 -1.20029461e+00 1.73009193e+00 -6.09785438e-01
-1.70278573e+00 3.21135998e-01 -2.44758010e-01 -7.52272010e-01
4.55487937e-01 -5.62877834e-01 3.10643524e-01 5.53094447e-01
1.04306519e-01 6.00863159e-01 6.56879187e-01 -1.12048888e+00
-4.27870572e-01 1.65267974e-01 -1.27913533e-02 4.56677794e-01
-6.46891415e-01 -3.34078848e-01 -3.87875885e-01 -4.85155791e-01
-2.70718962e-01 -5.34964740e-01 -2.82514483e-01 -5.65580487e-01
-7.69131064e-01 -1.57225981e-01 3.76498371e-01 -8.85109246e-01
1.63365483e+00 -1.67604434e+00 2.93192893e-01 -1.20406091e-01
3.34228158e-01 5.57959974e-01 -3.32075953e-01 7.09161282e-01
2.83394903e-01 4.01027203e-01 -4.53832150e-01 -4.57329571e-01
1.97376087e-02 1.11598156e-01 -6.21508539e-01 -9.73212570e-02
3.66047651e-01 1.13028359e+00 -1.29959035e+00 -6.24761462e-01
-1.09555855e-01 6.00235648e-02 -9.47313666e-01 2.50049502e-01
-4.42202985e-01 9.41586867e-02 -4.63369638e-01 6.68387562e-02
1.47997662e-01 -2.46981680e-01 9.65266749e-02 1.46786841e-02
-3.96049879e-02 8.09886754e-01 -7.50878215e-01 1.94679952e+00
-4.75790679e-01 7.76073813e-01 -1.28534213e-01 -9.58914697e-01
7.88881421e-01 2.92080551e-01 2.53655645e-03 -1.73661783e-01
-3.89035381e-02 1.49743602e-01 -2.30114400e-01 -5.24301887e-01
1.27143967e+00 1.02617964e-01 -1.06911317e-01 6.79622948e-01
3.10999185e-01 -9.29900169e-01 6.94418013e-01 9.19427693e-01
1.28781581e+00 -1.62528157e-01 2.28150651e-01 -2.89560139e-01
1.98885366e-01 1.32315844e-01 3.43407243e-01 1.18873835e+00
1.88947901e-01 7.20929205e-01 6.91492975e-01 2.61711836e-01
-1.36280119e+00 -1.04864717e+00 7.99709037e-02 8.60913515e-01
-3.55303913e-01 -7.77677834e-01 -1.06062889e+00 -4.57734823e-01
7.72079313e-03 1.32361412e+00 -2.75172919e-01 -4.66659486e-01
-6.13781691e-01 -1.92599580e-01 7.50924230e-01 5.93311667e-01
4.80515778e-01 -8.63127589e-01 -5.31960070e-01 1.65500224e-01
-1.86109349e-01 -7.67119348e-01 -7.54418850e-01 5.30919209e-02
-1.29892886e+00 -7.22004294e-01 -6.20564699e-01 -5.89378834e-01
5.47739148e-01 2.27643862e-01 1.40903103e+00 6.75378516e-02
1.54860243e-01 6.06647015e-01 -1.49274349e-01 -2.72882223e-01
-9.49239731e-01 5.73454618e-01 -1.62732601e-01 -7.14068115e-01
-8.71462151e-02 -8.34956527e-01 -5.20438850e-01 -4.92417157e-01
-8.95699084e-01 4.36509192e-01 6.66661322e-01 8.71978998e-01
2.20056310e-01 -2.08106041e-01 8.34102690e-01 -9.65042412e-01
1.16790557e+00 -2.61037409e-01 -1.92373678e-01 5.23758471e-01
-6.48119450e-01 4.73718047e-01 7.90726781e-01 -2.67828315e-01
-1.09642756e+00 -5.65237999e-01 9.73869115e-02 -1.65532723e-01
1.77311972e-01 6.65166736e-01 7.35308230e-02 4.26701576e-01
9.29110885e-01 6.29080415e-01 1.08591139e-01 -2.37181231e-01
9.00564969e-01 6.08995438e-01 9.48018432e-01 -7.32604861e-01
6.35143995e-01 -1.20221086e-01 -3.30098540e-01 -8.25178087e-01
-1.00699770e+00 -9.64653119e-02 -5.88909268e-01 2.45366260e-01
5.92230618e-01 -9.70378220e-01 -3.82951915e-01 9.34598148e-02
-1.15526903e+00 -5.49257517e-01 -8.68361294e-01 2.57323772e-01
-8.42978179e-01 5.60955107e-01 -8.64073455e-01 -5.56140661e-01
-9.74047780e-01 -7.43663251e-01 8.03006351e-01 3.95118803e-01
-5.30311167e-01 -1.11647940e+00 4.22754064e-02 4.08578753e-01
5.09516776e-01 -7.33212456e-02 1.11093628e+00 -8.38731408e-01
-6.45622134e-01 -1.73062205e-01 -3.67795154e-02 7.77548969e-01
4.54417393e-02 1.41625315e-01 -7.26352096e-01 -3.83470237e-01
7.38424296e-03 -5.64653039e-01 1.19912052e+00 3.57102513e-01
9.70157444e-01 -9.36103940e-01 -1.34297624e-01 4.22200590e-01
1.10508835e+00 -1.49564222e-01 4.43349928e-01 3.00960485e-02
6.71493888e-01 3.24096322e-01 2.99877405e-01 3.44596028e-01
3.81402016e-01 3.41370776e-02 -2.50983298e-01 3.32743317e-01
-2.92478830e-01 -7.00988412e-01 4.89139944e-01 1.59936070e+00
1.14161067e-01 -3.06475610e-01 -6.62644684e-01 7.84624338e-01
-1.82383752e+00 -1.10832822e+00 2.24350303e-01 2.23717904e+00
1.52270293e+00 2.90305048e-01 -4.36812043e-02 3.36368419e-02
5.27713418e-01 1.68284148e-01 -5.71152687e-01 -7.54149914e-01
1.11429460e-01 4.11748707e-01 3.97863626e-01 8.12424660e-01
-5.38201511e-01 9.15501714e-01 6.93117237e+00 1.01050580e+00
-6.27129078e-01 -2.33467981e-01 5.32802284e-01 -4.98503745e-01
-7.84454584e-01 1.22138180e-01 -7.13264406e-01 3.37232143e-01
1.45150495e+00 -9.41332102e-01 3.59855831e-01 5.40881217e-01
3.05834681e-01 -1.36742219e-01 -1.39037585e+00 4.14156944e-01
1.49916962e-01 -1.68248761e+00 3.93338263e-01 -2.93957204e-01
1.11446369e+00 -2.59000540e-01 -1.12867959e-01 5.74996293e-01
6.50677562e-01 -9.39912081e-01 5.93745530e-01 6.66686296e-01
8.02106857e-01 -9.58224058e-01 3.09193224e-01 6.67861998e-01
-6.27766252e-01 1.23531232e-02 -4.31717098e-01 -1.17237784e-01
1.90348506e-01 6.33442283e-01 -1.04058707e+00 7.38721609e-01
-4.60240841e-02 8.17074060e-01 -5.09012401e-01 8.65796208e-01
-4.36550170e-01 1.09860289e+00 -1.51053339e-01 -2.18940094e-01
1.78597569e-02 -7.53198266e-02 7.83389449e-01 1.55109823e+00
3.44327688e-01 3.21010560e-01 3.03842481e-02 9.07117069e-01
-3.05394500e-01 -1.23021163e-01 -5.66364408e-01 -1.80372760e-01
8.41117203e-01 8.52012753e-01 -1.77980602e-01 -9.55934167e-01
6.35825321e-02 8.56316984e-01 4.53394562e-01 3.74828011e-01
-3.67762506e-01 -6.69700921e-01 1.95600554e-01 -3.84520590e-02
4.66361821e-01 -1.81253165e-01 -5.54551899e-01 -1.26934719e+00
-1.78303182e-01 -9.30669606e-01 4.45210375e-03 -7.29584575e-01
-9.66816962e-01 1.88856915e-01 3.01828593e-01 -7.26209998e-01
-5.86652696e-01 1.35248005e-01 -9.30029929e-01 9.86267090e-01
-1.37260854e+00 -6.48997247e-01 4.73016948e-02 6.46559969e-02
8.00620973e-01 4.22415026e-02 8.56896222e-01 -2.84150809e-01
-5.17516553e-01 8.73349726e-01 5.09163916e-01 -2.07666650e-01
4.93117392e-01 -1.53641963e+00 6.24120653e-01 9.29493785e-01
-5.22006489e-02 8.58916819e-01 1.04664099e+00 -6.78971410e-01
-1.46634853e+00 -1.12435985e+00 1.05472648e+00 -6.19183183e-01
6.01595283e-01 -9.04365405e-02 -9.80004013e-01 6.77309871e-01
5.55179358e-01 -8.82712424e-01 4.90844876e-01 1.36803493e-01
-5.32827638e-02 -2.02652961e-02 -9.20057535e-01 6.42672896e-01
9.27596271e-01 -4.74465489e-01 -1.07811010e+00 4.76184100e-01
1.32354641e+00 -4.94034886e-01 -9.71679449e-01 1.17165640e-01
1.95446491e-01 -7.57690191e-01 8.35417569e-01 -5.98418713e-01
1.06377506e+00 2.48160839e-01 2.16654763e-01 -1.60123420e+00
-2.08004147e-01 -9.79611874e-01 -6.46821856e-01 1.58046591e+00
5.34611762e-01 -3.34817261e-01 8.40613782e-01 7.37754643e-01
-6.08336449e-01 -9.00938511e-01 -4.75208908e-01 -9.31214929e-01
5.20085752e-01 -1.01886289e-02 5.20381033e-01 5.33624411e-01
5.11474967e-01 6.86747372e-01 -1.38673201e-01 -3.26161087e-01
4.94797409e-01 6.12915978e-02 9.39295232e-01 -8.89221311e-01
-5.63242972e-01 -5.53809822e-01 1.13316931e-01 -1.69695544e+00
-4.14457917e-02 -9.95358467e-01 1.86982885e-01 -1.99905849e+00
4.96361613e-01 -1.35210752e-01 1.93772614e-01 3.57076287e-01
-5.79283893e-01 -4.08636034e-01 3.64806533e-01 1.76736161e-01
-6.87568665e-01 8.56741488e-01 1.53556597e+00 -1.29195422e-01
-5.30848145e-01 -3.09826788e-02 -1.23241150e+00 3.37675929e-01
7.16417313e-01 -3.13964307e-01 -8.48370910e-01 -5.29882133e-01
-1.14674075e-02 4.97006565e-01 -3.87101285e-02 -9.70340312e-01
5.27659833e-01 3.45019475e-02 -1.51540101e-01 -5.97230315e-01
6.84542656e-02 8.39151815e-02 -3.47204357e-01 4.14659470e-01
-9.08300877e-01 9.96084139e-03 2.65279353e-01 5.11009932e-01
-1.34985134e-01 -6.00256979e-01 4.44523126e-01 -2.21247151e-01
1.50732184e-02 -4.47391607e-02 -2.61719018e-01 5.98035872e-01
4.75248277e-01 -2.73948550e-01 -5.86747646e-01 -6.27512515e-01
-4.18965101e-01 4.33260947e-01 3.95171940e-01 -1.17645316e-01
4.95671064e-01 -9.01159823e-01 -1.10976315e+00 -9.34078097e-02
-4.35198456e-01 4.74738926e-01 9.92557183e-02 5.83358884e-01
-4.13062334e-01 6.13101065e-01 2.86812633e-01 -5.05670369e-01
-1.05248833e+00 3.23466688e-01 -6.31810650e-02 -5.96038043e-01
-7.10613668e-01 8.82717490e-01 -1.26031935e-01 -3.39778900e-01
2.38504648e-01 -6.08560383e-01 1.51995763e-01 -2.44265795e-02
5.15044212e-01 5.69331527e-01 5.95888831e-02 1.58332288e-01
1.94358408e-01 9.16597322e-02 -4.38521326e-01 -1.95558831e-01
1.28715837e+00 -6.16859831e-02 -3.91492918e-02 2.90852934e-01
1.04528761e+00 8.77637491e-02 -1.27478397e+00 -1.83982223e-01
-1.44460276e-01 -2.32378066e-01 -2.75384765e-02 -8.07918668e-01
-6.26124084e-01 7.85367668e-01 -3.19316655e-01 3.13677788e-01
1.10389352e+00 -2.11267799e-01 9.67376709e-01 7.63642669e-01
1.77830569e-02 -8.91774714e-01 2.94306308e-01 6.78562820e-01
7.66617239e-01 -6.71470702e-01 2.81845212e-01 -1.87278599e-01
-6.41283870e-01 1.03282082e+00 3.48361522e-01 -1.28867611e-01
5.29108718e-02 1.42388448e-01 -4.62916732e-01 -2.08925009e-02
-1.30001116e+00 2.36411363e-01 1.23543069e-01 5.09690523e-01
4.29979026e-01 -3.09231579e-02 -3.58839452e-01 5.49527228e-01
-9.17973459e-01 3.67492773e-02 9.11541879e-01 9.03049648e-01
-9.81197834e-01 -8.53655338e-01 -7.54435435e-02 9.34272110e-01
-2.01857626e-01 -5.44328034e-01 -1.90484375e-01 4.85579520e-01
-6.15650356e-01 1.10078645e+00 -4.21254449e-02 -2.30697662e-01
4.33582395e-01 1.28101334e-01 5.95238984e-01 -9.46592033e-01
-7.93523312e-01 -4.04123873e-01 4.43839937e-01 -1.36203915e-01
1.52710332e-02 -6.73182309e-01 -1.10967278e+00 -6.73876643e-01
-5.42054534e-01 7.17527151e-01 2.56038308e-01 8.76178622e-01
6.29158676e-01 6.38800740e-01 5.42009830e-01 -8.19390595e-01
-1.15263712e+00 -1.18853557e+00 -2.25829974e-01 9.08496752e-02
5.07326841e-01 4.73401472e-02 -4.47131425e-01 2.51711935e-01] | [12.367201805114746, 9.379453659057617] |
6c7d6069-8cff-4614-ab7d-e0393d9f8b30 | sample-factory-egocentric-3d-control-from | 2006.11751 | null | https://arxiv.org/abs/2006.11751v2 | https://arxiv.org/pdf/2006.11751v2.pdf | Sample Factory: Egocentric 3D Control from Pixels at 100000 FPS with Asynchronous Reinforcement Learning | Increasing the scale of reinforcement learning experiments has allowed researchers to achieve unprecedented results in both training sophisticated agents for video games, and in sim-to-real transfer for robotics. Typically such experiments rely on large distributed systems and require expensive hardware setups, limiting wider access to this exciting area of research. In this work we aim to solve this problem by optimizing the efficiency and resource utilization of reinforcement learning algorithms instead of relying on distributed computation. We present the "Sample Factory", a high-throughput training system optimized for a single-machine setting. Our architecture combines a highly efficient, asynchronous, GPU-based sampler with off-policy correction techniques, allowing us to achieve throughput higher than $10^5$ environment frames/second on non-trivial control problems in 3D without sacrificing sample efficiency. We extend Sample Factory to support self-play and population-based training and apply these techniques to train highly capable agents for a multiplayer first-person shooter game. The source code is available at https://github.com/alex-petrenko/sample-factory | ['Gaurav Sukhatme', 'Vladlen Koltun', 'Tushar Kumar', 'Zhehui Huang', 'Aleksei Petrenko'] | 2020-06-21 | null | https://proceedings.icml.cc/static/paper_files/icml/2020/3314-Paper.pdf | https://proceedings.icml.cc/static/paper_files/icml/2020/3314-Paper.pdf | icml-2020-1 | ['fps-games'] | ['playing-games'] | [-1.48530960e-01 -3.06883454e-01 8.86133537e-02 -3.25214081e-02
-8.22431087e-01 -5.45309067e-01 4.81077433e-01 1.65885404e-01
-1.08641231e+00 1.05014408e+00 -5.43666363e-01 -4.28571314e-01
1.74119294e-01 -7.92520642e-01 -8.66854846e-01 -7.13077009e-01
-5.24800241e-01 9.54315126e-01 4.29633826e-01 -2.86760062e-01
1.89585790e-01 3.58688623e-01 -1.92692423e+00 -2.09498703e-02
5.55827618e-01 5.13123393e-01 4.23208922e-01 1.34150445e+00
5.11692464e-01 7.22332835e-01 -7.04636753e-01 1.74443424e-01
5.92399240e-01 -6.28987253e-01 -4.50286180e-01 -2.29421109e-01
3.74201387e-01 -5.68047881e-01 -1.64972484e-01 7.86360562e-01
1.03771222e+00 4.30756867e-01 1.60025477e-01 -1.33961153e+00
4.13007110e-01 3.06836784e-01 -4.21913296e-01 3.68690014e-01
2.47839093e-01 6.75149322e-01 6.14134908e-01 -8.96548256e-02
5.45196414e-01 1.08640134e+00 6.06706679e-01 7.88657427e-01
-1.15064299e+00 -7.94077039e-01 -1.15852296e-01 7.41721466e-02
-1.11603606e+00 -3.54891330e-01 -5.13038933e-02 -2.28184134e-01
1.35640752e+00 6.65340526e-03 1.29944015e+00 9.40192461e-01
2.58013338e-01 7.91927397e-01 1.16797900e+00 -2.93008775e-01
7.21103847e-01 -3.59324425e-01 -6.02385521e-01 1.05929363e+00
7.03846142e-02 6.03519499e-01 -6.44150972e-01 -2.21942589e-01
1.39237249e+00 -1.57626569e-01 8.60132575e-02 -3.16193461e-01
-1.15892136e+00 1.04504907e+00 3.00898045e-01 -2.42327288e-01
-4.36195999e-01 8.25083792e-01 5.25674105e-01 5.47159374e-01
2.87279218e-01 3.74263912e-01 -6.24953091e-01 -9.54668522e-01
-6.70638740e-01 9.78805542e-01 1.06024897e+00 7.87058175e-01
6.09090745e-01 2.07073495e-01 2.82539099e-01 4.57683772e-01
1.98286138e-02 5.47030628e-01 4.49599713e-01 -1.45324349e+00
2.12052375e-01 1.85547918e-01 2.11112797e-01 -2.60506541e-01
-6.40599251e-01 -2.19943464e-01 -5.24480581e-01 9.86296296e-01
6.51850462e-01 -7.40494132e-01 -5.65536797e-01 1.65918732e+00
9.70386147e-01 3.50848675e-01 -2.24046670e-02 1.03667963e+00
-2.75390055e-02 6.24887526e-01 -1.17080212e-01 -1.14605483e-02
1.29440427e+00 -1.03824520e+00 1.45115405e-01 -2.27008894e-01
7.61481464e-01 -6.04208350e-01 1.12781966e+00 5.44274330e-01
-1.16142714e+00 -1.37016013e-01 -8.83023262e-01 3.14203024e-01
5.91818541e-02 -1.77445322e-01 1.19352782e+00 7.36470282e-01
-1.12571442e+00 1.07963073e+00 -1.58643579e+00 -4.47520554e-01
5.51417053e-01 7.77786374e-01 -1.45175204e-01 1.38048947e-01
-4.98538703e-01 7.42027760e-01 2.70178527e-01 -4.67909366e-01
-1.26466513e+00 -7.49863923e-01 -4.48665977e-01 -1.80414557e-01
4.77533251e-01 -1.05981398e+00 1.77933347e+00 -8.52595329e-01
-2.29850173e+00 4.88004386e-01 3.41764450e-01 -6.62491739e-01
6.74644589e-01 -2.26059303e-01 4.98872697e-01 -2.88980883e-02
1.66169368e-02 7.28434503e-01 7.22315133e-01 -6.37156665e-01
-9.53192055e-01 -3.98528010e-01 1.84507236e-01 6.76286042e-01
1.08772524e-01 6.49175569e-02 -4.23611654e-03 -1.69893757e-01
-4.67570335e-01 -1.25034821e+00 -6.17991090e-01 -1.80897210e-02
4.51987743e-01 -7.74110630e-02 5.20216584e-01 -1.99831948e-01
3.20589483e-01 -1.67946947e+00 2.25428611e-01 -8.73284936e-02
1.41665488e-01 2.02318475e-01 -4.57617790e-02 5.70875883e-01
5.36514997e-01 -5.76072395e-01 -1.58290684e-01 -2.19059736e-01
1.14928178e-01 4.74653959e-01 1.53921902e-01 6.67267680e-01
-1.61170408e-01 5.56056380e-01 -1.21784663e+00 -4.93021786e-01
2.87912577e-01 2.74390489e-01 -1.35541296e+00 3.10924739e-01
-4.48382199e-01 8.44188750e-01 -4.34193432e-01 3.47189754e-01
1.78109840e-01 -1.42541863e-02 3.04406941e-01 7.24403441e-01
-3.39790255e-01 2.12120160e-01 -1.22786093e+00 1.96147954e+00
-6.42801464e-01 4.82755005e-01 5.86780369e-01 -9.26391065e-01
5.63739002e-01 1.08552895e-01 6.70969307e-01 -6.22539699e-01
4.37946588e-01 2.91895300e-01 9.44504961e-02 -2.39140213e-01
5.11194468e-01 -2.21319750e-01 -2.88615581e-02 5.90221703e-01
1.97433978e-01 -6.65912271e-01 5.91177285e-01 -1.87026896e-02
1.45281279e+00 5.34849405e-01 7.44941011e-02 -3.46900523e-01
-1.55496553e-01 3.12441081e-01 3.88898373e-01 9.64698255e-01
-2.37896353e-01 2.45386343e-02 3.60516638e-01 -5.32038689e-01
-1.33977020e+00 -9.83237863e-01 2.16114298e-01 1.59289467e+00
-2.06371229e-02 -5.14399350e-01 -8.28334451e-01 -1.91555619e-01
1.59341469e-01 3.65669042e-01 -1.71717837e-01 5.80536835e-02
-7.01999307e-01 -7.20418334e-01 5.90658724e-01 2.19920889e-01
4.14297134e-01 -1.13235283e+00 -1.44727361e+00 4.05473202e-01
2.75712907e-01 -6.56447053e-01 -1.82523131e-01 3.50776672e-01
-1.01123488e+00 -9.55922723e-01 -6.50004983e-01 -6.40750825e-01
4.01116282e-01 6.46149665e-02 1.17153323e+00 2.58482456e-01
-7.02541530e-01 4.37160432e-01 -1.85291424e-01 -4.70268339e-01
-3.02843720e-01 2.19928980e-01 3.87319267e-01 -7.78482556e-01
-1.63314223e-01 -8.99061739e-01 -6.80959880e-01 1.59899414e-01
-5.75990081e-01 3.29878658e-01 4.25550371e-01 1.00513744e+00
3.59882712e-01 -2.13849932e-01 2.56477088e-01 -5.10091484e-01
6.11966074e-01 -2.67915308e-01 -1.33819675e+00 -4.05315757e-01
-2.53467411e-01 2.25768581e-01 7.66734779e-01 -6.53346181e-01
-6.03163898e-01 3.09616655e-01 -3.96299362e-01 -2.90450990e-01
-3.97348180e-02 -8.02210160e-03 6.05810046e-01 -3.53522271e-01
8.88224185e-01 2.64426202e-01 3.87878895e-01 -2.75521334e-02
2.05802634e-01 3.26529264e-01 2.03488559e-01 -1.20708287e+00
3.64939570e-01 2.57142782e-01 1.11692637e-01 -9.14511263e-01
-2.34204695e-01 -1.44089118e-01 -1.57953687e-02 -3.97270441e-01
5.81747830e-01 -9.97049510e-01 -1.65618801e+00 4.14409727e-01
-7.83279538e-01 -1.38434446e+00 -5.62639356e-01 7.16199338e-01
-1.22137630e+00 -1.21779963e-01 -8.66855443e-01 -8.61104071e-01
-3.23032320e-01 -1.17915428e+00 1.08237731e+00 6.49171472e-01
1.37150139e-01 -6.91298723e-01 8.08279276e-01 1.51420206e-01
6.09761059e-01 3.80984321e-02 1.42261118e-01 -7.78371990e-02
-6.46122336e-01 2.00993381e-02 2.49469057e-01 -1.96596563e-01
-3.94525111e-01 -2.23928139e-01 -5.74915648e-01 -7.90848136e-01
-4.13050085e-01 -8.53441834e-01 5.24414480e-01 3.14778268e-01
9.16063368e-01 -1.72037303e-01 -4.94302213e-02 6.84438884e-01
1.29522479e+00 -1.20129131e-01 2.93071955e-01 4.47188705e-01
2.94414669e-01 1.29849166e-02 5.60884714e-01 1.09121883e+00
4.18796688e-01 7.04225838e-01 4.96555954e-01 1.96901947e-01
2.66657829e-01 -2.64035732e-01 5.67158401e-01 6.87508762e-01
-6.06167316e-01 -3.22822519e-02 -8.25920582e-01 2.89695829e-01
-2.06261063e+00 -1.12875247e+00 2.47725353e-01 2.45864320e+00
9.66185749e-01 -8.14127922e-02 7.33383119e-01 -3.77407223e-01
4.80935454e-01 -3.45935643e-01 -6.63263381e-01 -4.38061953e-01
4.61834788e-01 6.27310574e-01 7.72808373e-01 6.54169679e-01
-9.14004385e-01 1.25113475e+00 5.59630442e+00 9.38197076e-01
-1.06068194e+00 2.10457027e-01 3.60712498e-01 -7.59408236e-01
3.66120636e-01 -1.79968774e-02 -7.73562491e-01 3.31249297e-01
1.16860390e+00 -2.74891615e-01 1.18323755e+00 1.10513699e+00
1.49475545e-01 -6.54677808e-01 -8.34544003e-01 1.08242071e+00
-4.38144714e-01 -1.33281136e+00 -5.91941655e-01 3.72128367e-01
6.84217274e-01 6.74418092e-01 -1.90341711e-01 6.85623169e-01
1.06369543e+00 -7.76834667e-01 6.05835557e-01 -5.80264861e-03
4.90586877e-01 -9.76911187e-01 3.77676100e-01 8.05997789e-01
-1.03202212e+00 6.75016120e-02 -5.92147231e-01 -7.50687182e-01
9.58347991e-02 5.11991493e-02 -8.66795182e-01 6.11714981e-02
7.06980050e-01 1.82435811e-01 -6.12482578e-02 1.06481385e+00
2.98889168e-02 4.53753114e-01 -1.08080053e+00 -6.57105744e-01
3.52373123e-01 -3.76364142e-01 4.00405884e-01 8.92653525e-01
4.13163632e-01 2.89237142e-01 3.96046042e-01 3.89004529e-01
9.35205966e-02 -4.00908440e-02 -4.86683697e-01 -6.20738640e-02
2.86758304e-01 1.37508905e+00 -9.95871961e-01 -3.14890206e-01
4.53627743e-02 1.08261418e+00 6.49879932e-01 -2.18970969e-01
-9.84315574e-01 -3.13488781e-01 1.04099345e+00 -7.47030377e-02
3.55634660e-01 -8.78478706e-01 2.03431934e-01 -1.29750383e+00
-3.25986356e-01 -1.09131503e+00 1.94209188e-01 -2.17141896e-01
-8.37033451e-01 2.70528495e-01 -1.57219246e-01 -8.84738982e-01
-5.34061491e-01 -5.44948041e-01 -4.77237433e-01 3.62702668e-01
-1.12834728e+00 -6.18632674e-01 -1.92043483e-01 8.22605610e-01
5.39858103e-01 -3.31765711e-01 9.34800088e-01 3.29412043e-01
-4.70280260e-01 4.44275260e-01 3.41778904e-01 -2.93594718e-01
4.45063561e-01 -1.30973279e+00 5.28094888e-01 5.11595726e-01
1.06920458e-01 1.49917290e-01 9.14293587e-01 -4.72199112e-01
-1.98810375e+00 -6.52373791e-01 2.19738595e-02 -2.14175582e-01
7.34840393e-01 -4.75740612e-01 -1.73650697e-01 5.16926169e-01
1.62004158e-01 8.61280411e-02 4.44610983e-01 2.63563573e-01
1.27895385e-01 -1.41535196e-02 -1.11947322e+00 7.77594745e-01
1.12917137e+00 7.12509616e-04 6.53944239e-02 6.30672634e-01
2.35048994e-01 -9.51515317e-01 -7.52380788e-01 -3.36242825e-01
5.35970151e-01 -1.03868139e+00 8.50938261e-01 -5.68698823e-01
3.33803982e-01 -2.03323618e-01 8.27264041e-02 -1.46671212e+00
-2.53880713e-02 -9.62151349e-01 5.07148840e-02 4.92343187e-01
3.40368338e-02 -8.13517332e-01 1.28835452e+00 2.00641736e-01
-1.99330539e-01 -5.77017844e-01 -1.21782029e+00 -8.25276434e-01
2.23105922e-01 -3.39091808e-01 2.94324219e-01 4.50894415e-01
3.21247458e-01 3.08223099e-01 -4.06853974e-01 1.06720150e-01
8.41590881e-01 1.05961889e-03 1.23951781e+00 -6.68739855e-01
-1.14073408e+00 -4.62392658e-01 -4.09374863e-01 -1.17022204e+00
-5.71968891e-02 -6.45611823e-01 3.27053696e-01 -1.03758991e+00
-6.28047511e-02 -7.61839688e-01 4.09233093e-01 4.87302303e-01
1.25628904e-01 3.53663117e-01 4.39018279e-01 1.70708392e-02
-9.32757854e-01 7.69194305e-01 1.18477786e+00 2.66664326e-01
-2.27465004e-01 1.83942035e-01 5.44719361e-02 6.59883797e-01
1.16929126e+00 -6.02085710e-01 -2.55409181e-01 -6.21706486e-01
2.73570627e-01 4.05463666e-01 3.46800029e-01 -1.60486424e+00
3.14975023e-01 -4.98611838e-01 2.33540669e-01 1.59713432e-01
5.73454678e-01 -4.69544679e-01 -1.26466930e-01 9.89739597e-01
-9.77822766e-02 3.53830546e-01 3.82286042e-01 3.75501543e-01
3.35906088e-01 -1.15736797e-01 9.28850651e-01 -5.05544722e-01
-4.43251997e-01 4.33993816e-01 -8.34245145e-01 3.17852855e-01
1.19735110e+00 1.65966809e-01 -2.90181994e-01 -3.33747178e-01
-4.24192280e-01 3.31906706e-01 6.42607093e-01 -1.51243523e-01
2.47771904e-01 -9.44667935e-01 -7.52740145e-01 2.85750985e-01
-3.61606032e-01 7.24226087e-02 2.37109810e-01 6.25186861e-01
-1.34000397e+00 -4.84948270e-02 -6.25260532e-01 -6.18921638e-01
-1.32943034e+00 2.06675127e-01 4.12923723e-01 -3.34725469e-01
-8.57498229e-01 1.06212890e+00 -3.20315540e-01 -6.74081326e-01
5.44361025e-02 -1.19741246e-01 5.41896462e-01 -5.94159782e-01
5.32127559e-01 4.79690909e-01 -1.96857840e-01 9.12658796e-02
-2.60343611e-01 2.34329738e-02 1.10359333e-01 -6.06263578e-01
1.51970613e+00 3.91638994e-01 2.02424198e-01 1.54049173e-01
6.38400853e-01 -3.84320438e-01 -1.87073839e+00 2.18673676e-01
-4.77030635e-01 -5.53597748e-01 1.14393249e-01 -3.63746434e-01
-8.46979856e-01 6.69860840e-01 7.57361233e-01 -3.23681906e-02
7.64480054e-01 -2.38742813e-01 7.76567280e-01 8.08153927e-01
1.10436916e+00 -1.15341413e+00 8.32947418e-02 8.22879493e-01
3.73373747e-01 -1.14695811e+00 1.46768078e-01 2.60267943e-01
-4.77812618e-01 1.04010630e+00 6.76044166e-01 -8.46994042e-01
7.13229850e-02 8.22806537e-01 -1.54630587e-01 3.62787060e-02
-1.10783887e+00 -2.68600732e-01 -7.61244059e-01 6.54181480e-01
1.64441437e-01 1.73098966e-01 -7.80194402e-02 -4.63624522e-02
-5.36667705e-01 2.12340057e-01 6.18811011e-01 1.40389931e+00
-7.63417661e-01 -1.28372467e+00 -2.18885109e-01 3.31091940e-01
-2.50710249e-01 7.03964010e-02 2.59217203e-01 6.75606132e-01
-1.57111123e-01 5.30223250e-01 2.74809331e-01 -1.68594882e-01
2.91619301e-02 -3.07320207e-01 1.28991199e+00 -5.61696887e-01
-9.41611886e-01 -4.77995425e-02 1.04430512e-01 -8.65681589e-01
-1.09342679e-01 -9.03180897e-01 -1.24060416e+00 -1.06191862e+00
-2.87197679e-02 2.31644437e-01 1.04887962e+00 7.58524537e-01
4.60521996e-01 4.12524700e-01 3.04037988e-01 -1.69949841e+00
-7.30898380e-01 -6.45266056e-01 -4.95034993e-01 -8.28937888e-02
7.40137026e-02 -6.51531816e-01 1.22099873e-02 -2.25794360e-01] | [3.9664306640625, 1.4691931009292603] |
ff79313b-49eb-4280-bda5-2a0c509f9d7f | edge-based-tensor-prediction-via-graph-neural | 2201.05770 | null | https://arxiv.org/abs/2201.05770v1 | https://arxiv.org/pdf/2201.05770v1.pdf | Edge-based Tensor prediction via graph neural networks | Message-passing neural networks (MPNN) have shown extremely high efficiency and accuracy in predicting the physical properties of molecules and crystals, and are expected to become the next-generation material simulation tool after the density functional theory (DFT). However, there is currently a lack of a general MPNN framework for directly predicting the tensor properties of the crystals. In this work, a general framework for the prediction of tensor properties was proposed: the tensor property of a crystal can be decomposed into the average of the tensor contributions of all the atoms in the crystal, and the tensor contribution of each atom can be expanded as the sum of the tensor projections in the directions of the edges connecting the atoms. On this basis, the edge-based expansions of force vectors, Born effective charges (BECs), dielectric (DL) and piezoelectric (PZ) tensors were proposed. These expansions are rotationally equivariant, while the coefficients in these tensor expansions are rotationally invariant scalars which are similar to physical quantities such as formation energy and band gap. The advantage of this tensor prediction framework is that it does not require the network itself to be equivariant. Therefore, in this work, we directly designed the edge-based tensor prediction graph neural network (ETGNN) model on the basis of the invariant graph neural network to predict tensors. The validity and high precision of this tensor prediction framework were shown by the tests of ETGNN on the extended systems, random perturbed structures and JARVIS-DFT datasets. This tensor prediction framework is general for nearly all the GNNs and can achieve higher accuracy with more advanced GNNs in the future. | ['Hongjun Xiang', 'Xingao Gong', 'Hongyu Yu', 'Yang Zhong'] | 2022-01-15 | null | null | null | null | ['formation-energy'] | ['miscellaneous'] | [-1.44946352e-02 -4.74953577e-02 -1.04496121e-01 -1.17864385e-01
3.11191082e-01 7.18222633e-02 4.60445464e-01 8.33668858e-02
9.17738944e-04 8.10364306e-01 -5.14861867e-02 -3.82109582e-01
-4.35097814e-01 -1.20813382e+00 -7.98083842e-01 -1.22108269e+00
-5.63308716e-01 3.11974466e-01 1.28644958e-01 -5.16327381e-01
1.83209494e-01 8.17389369e-01 -1.63593578e+00 9.22085494e-02
8.67370963e-01 1.19528389e+00 -1.43812463e-01 3.89479011e-01
7.42556080e-02 7.08473086e-01 5.23307137e-02 -2.36037090e-01
2.78724805e-02 -2.09416956e-01 -8.36849868e-01 -3.88970524e-01
2.53029227e-01 -1.70186818e-01 -7.06912160e-01 9.55155253e-01
3.83993089e-01 2.65659064e-01 8.20925355e-01 -1.17986929e+00
-8.89707804e-01 8.05069804e-01 -5.54705076e-02 -3.14219713e-01
2.66582012e-01 1.49094211e-02 1.33819187e+00 -8.25607419e-01
8.05925071e-01 8.92211556e-01 5.76219320e-01 2.74358183e-01
-1.02054358e+00 -4.73984331e-01 -2.50777602e-01 7.19081402e-01
-1.20576787e+00 1.50883377e-01 1.17447162e+00 -4.13565606e-01
1.38279414e+00 3.31320196e-01 8.95626128e-01 7.94512033e-01
7.61914134e-01 2.07751974e-01 6.26187921e-01 -5.78221023e-01
2.98491776e-01 -2.65706956e-01 5.49611866e-01 7.75473356e-01
6.06495380e-01 6.48100823e-02 -6.31736457e-01 -2.43146494e-01
4.45771396e-01 -2.03913208e-02 -2.02871099e-01 -5.50151050e-01
-1.23041010e+00 6.93961859e-01 6.73606753e-01 5.58186710e-01
-5.89927018e-01 2.70785153e-01 5.39142251e-01 -1.17057497e-02
4.34945524e-01 4.89286840e-01 -2.45130807e-01 4.20405045e-02
-4.40352917e-01 4.55702513e-01 8.53084922e-01 2.95986861e-01
1.02112198e+00 3.96066010e-01 1.66601874e-02 4.43380058e-01
3.22038710e-01 4.81007874e-01 2.81153947e-01 -7.35652685e-01
1.95277870e-01 8.73958647e-01 -2.04274788e-01 -1.41027200e+00
-6.38264477e-01 -4.16222692e-01 -1.34703541e+00 -3.27316411e-02
1.98993236e-01 -2.71872338e-02 -5.09857237e-01 1.52083230e+00
4.11662281e-01 -4.07752730e-02 -4.45218422e-02 7.08768010e-01
9.70111787e-01 8.32391739e-01 -3.22396368e-01 -2.16601133e-01
9.23931956e-01 -6.22383714e-01 -7.45905519e-01 6.14263892e-01
9.95041192e-01 -4.01833236e-01 5.38327336e-01 2.07129404e-01
-8.61536860e-01 -3.57818216e-01 -1.20383930e+00 3.42476070e-02
-4.40377831e-01 4.04455140e-02 1.19487870e+00 3.54161590e-01
-1.05751848e+00 1.19285166e+00 -9.28480446e-01 2.08254308e-02
-1.98023602e-01 7.12713122e-01 -3.80147249e-01 1.29355207e-01
-1.37087190e+00 6.72859490e-01 5.85231304e-01 5.00323772e-01
-3.93317282e-01 -7.13122129e-01 -5.37756801e-01 1.10837668e-01
-1.08832285e-01 -6.23602152e-01 6.66008949e-01 -5.03162026e-01
-1.61314273e+00 1.70447558e-01 -4.83129919e-02 -1.67725235e-01
-1.43641800e-01 4.41123813e-01 -4.62671578e-01 7.75318444e-02
-2.62592256e-01 2.51178414e-01 2.60712564e-01 -8.44517648e-01
2.73800492e-01 -2.19593376e-01 4.78891470e-02 -1.17951870e-01
-4.53065842e-01 -3.13103527e-01 2.69340605e-01 -2.36893758e-01
4.59889561e-01 -1.03679872e+00 -2.31698915e-01 -2.04017058e-01
-6.92890644e-01 -4.98087913e-01 8.41591656e-01 -6.09181345e-01
1.02697039e+00 -1.97852612e+00 3.67588639e-01 6.27036631e-01
5.34615099e-01 1.68860823e-01 -1.35408211e-02 9.39981937e-01
-5.39450824e-01 -1.39616430e-01 -6.66190963e-03 3.10229897e-01
-2.72504073e-02 1.91308901e-01 -1.62511729e-02 4.68311369e-01
-2.46503633e-02 8.52151394e-01 -5.94157755e-01 -8.19565132e-02
1.54096127e-01 7.53283560e-01 -7.82220542e-01 -1.50691822e-01
-3.80042732e-01 2.51401126e-01 -5.48701167e-01 4.11580831e-01
8.62347722e-01 -4.57115412e-01 2.44082287e-01 -8.49027038e-01
-2.27885455e-01 4.23065096e-01 -1.16406131e+00 1.03912866e+00
4.39415239e-02 4.25611943e-01 -9.42467526e-02 -1.37030792e+00
9.67602134e-01 3.16950321e-01 9.74368393e-01 -7.01544523e-01
-7.14879110e-03 3.64808500e-01 6.77331686e-01 -4.91507828e-01
5.06588101e-01 2.69590095e-02 2.75319397e-01 5.57154477e-01
1.77868947e-01 1.01478331e-01 4.12861496e-01 2.33532980e-01
8.38725805e-01 -6.84707686e-02 -1.00335948e-01 -2.98652917e-01
8.43464375e-01 -1.57088950e-01 1.28426209e-01 3.19486737e-01
2.89496481e-01 2.66929846e-02 7.30686486e-01 -8.40910733e-01
-1.35096073e+00 -6.17183924e-01 -3.01539421e-01 6.09065950e-01
7.40618706e-02 -7.77892590e-01 -8.04955602e-01 -4.42725904e-02
-8.88091624e-02 5.03978848e-01 -2.96771824e-01 -3.52402329e-01
-5.67424953e-01 -1.09591711e+00 3.53915244e-01 1.66500717e-01
5.44956148e-01 -9.82103944e-01 2.01346457e-01 1.79978207e-01
2.90479865e-02 -1.05660939e+00 1.07243031e-01 3.26251090e-02
-8.88631225e-01 -1.00703979e+00 -1.66775763e-01 -4.40007120e-01
6.77965999e-01 -3.08624506e-02 4.78113174e-01 2.68654495e-01
-1.37594536e-01 1.92886174e-01 -1.06301479e-01 5.06091379e-02
-5.46771348e-01 1.74078345e-02 5.36791682e-01 2.66526967e-01
4.24641222e-02 -8.49788249e-01 -5.07485092e-01 1.45594358e-01
-7.89140642e-01 2.38102779e-01 3.48624945e-01 6.62942410e-01
5.61817527e-01 3.27864736e-01 4.30090316e-02 -3.62571865e-01
6.11734331e-01 -2.77403206e-01 -4.94567961e-01 1.91069782e-01
-7.20680654e-01 4.45817351e-01 1.00267410e+00 -2.74605930e-01
-8.16387415e-01 -2.82285154e-01 -3.41932088e-01 1.37318596e-01
2.32302591e-01 8.46657395e-01 9.77377221e-03 -7.55446136e-01
3.40067357e-01 2.88973838e-01 -6.75004423e-02 -3.60832214e-01
-7.76018528e-03 4.89699721e-01 -2.77315336e-03 -9.13289666e-01
7.86561489e-01 2.93682486e-01 9.29206252e-01 -1.16192138e+00
-2.47829035e-01 2.02198885e-02 -5.04594922e-01 -5.72310090e-01
7.48820126e-01 -4.06767577e-01 -1.45958686e+00 7.55093157e-01
-1.39364731e+00 1.69118568e-01 -8.45177174e-02 9.00381088e-01
-7.37048686e-02 7.85134137e-01 -8.81933689e-01 -6.24642670e-01
-5.59416294e-01 -1.44959283e+00 5.28717339e-01 4.23346944e-02
8.81363824e-02 -1.13179600e+00 8.17226321e-02 2.80999333e-01
6.58751070e-01 2.67589360e-01 1.49967456e+00 -4.50101614e-01
-9.64924812e-01 -3.07419330e-01 -2.21886739e-01 5.18234134e-01
-2.39340797e-01 5.12946784e-01 -5.17930925e-01 -1.69255197e-01
-1.32695973e-01 9.41669941e-02 6.29447162e-01 4.59465206e-01
1.16820621e+00 -3.66415650e-01 -2.53428221e-01 5.94254851e-01
1.50130951e+00 1.60490498e-01 6.39839828e-01 5.96309043e-02
1.30399895e+00 2.75911868e-01 -2.49615327e-01 3.07250559e-01
1.86119929e-01 7.62229204e-01 5.92306137e-01 1.77671403e-01
9.40971076e-02 -8.06191415e-02 6.37822270e-01 1.58316565e+00
-1.00253987e+00 -9.05904621e-02 -9.02392268e-01 -1.54862195e-01
-1.59613538e+00 -1.01922214e+00 -9.21373725e-01 2.08804965e+00
3.36380988e-01 1.43585028e-02 -2.04758674e-01 1.70059785e-01
5.29401183e-01 1.77551240e-01 -5.83441079e-01 -7.32864499e-01
-1.89165086e-01 3.43649358e-01 6.36980593e-01 4.27812546e-01
-5.34317970e-01 4.00063962e-01 6.50294971e+00 7.84185231e-01
-1.43244255e+00 -1.11356795e-01 2.09198534e-01 4.01726276e-01
-5.75344503e-01 2.33610332e-01 -7.68557310e-01 3.86058241e-01
1.06048250e+00 1.00023150e-02 5.97392797e-01 7.05958724e-01
1.57773137e-01 1.23675093e-01 -9.41415966e-01 6.82341516e-01
-2.29864269e-01 -1.65083265e+00 3.79850119e-01 3.55285376e-01
6.18803799e-01 1.42659515e-01 -1.52222872e-01 5.88173270e-02
-2.56990492e-01 -8.77794147e-01 2.93176800e-01 7.06547558e-01
5.07585943e-01 -5.35783827e-01 7.69967377e-01 1.60999253e-01
-1.02630544e+00 2.19117925e-01 -4.08660501e-01 -4.54503715e-01
6.07645661e-02 1.08730495e+00 -7.06734717e-01 8.89740765e-01
4.06009614e-01 8.34190607e-01 -2.81926900e-01 6.95699096e-01
8.85068998e-02 5.58495522e-01 -3.57861161e-01 -5.51851034e-01
2.57290781e-01 -8.58075857e-01 5.75018406e-01 5.92977464e-01
4.93086785e-01 -2.29969814e-01 -2.66214937e-01 9.53190744e-01
-1.28505126e-01 2.43002251e-01 -6.28151000e-01 -3.59583765e-01
-6.00548722e-02 1.31602871e+00 -5.03625214e-01 -1.08405434e-01
-3.17840844e-01 3.43095988e-01 1.76329017e-01 5.08125484e-01
-7.31121778e-01 -4.04275060e-01 6.87731385e-01 1.99492186e-01
2.44413674e-01 -5.62121749e-01 -6.35369271e-02 -1.10804105e+00
2.90186107e-01 -4.92489129e-01 -3.23558867e-01 -7.92327523e-01
-1.17341375e+00 3.03484291e-01 -3.18308063e-02 -9.73522365e-01
-5.26815541e-02 -1.36579263e+00 -6.05991185e-01 6.92850649e-01
-1.10542333e+00 -1.10121536e+00 -1.52973561e-02 3.39801401e-01
-5.91390252e-01 -1.63218826e-01 9.39273953e-01 4.61872578e-01
-8.34503710e-01 1.89548701e-01 5.89106143e-01 1.01476282e-01
4.57806401e-02 -9.78686571e-01 -3.30921775e-03 5.23355067e-01
-2.56818235e-01 7.24887609e-01 7.79486895e-01 -5.93172133e-01
-1.95649993e+00 -7.50286758e-01 9.26379442e-01 2.44994417e-01
9.46564555e-01 -4.03898180e-01 -9.03212309e-01 3.99994075e-01
1.37002282e-02 8.94487798e-02 5.06069839e-01 1.05397396e-01
-1.40055507e-01 -4.17734623e-01 -6.47091568e-01 5.30282795e-01
9.53837931e-01 -5.12698114e-01 1.66163504e-01 8.42900932e-01
7.45720685e-01 -3.05153400e-01 -1.29292178e+00 5.45008659e-01
6.12378120e-01 -1.17924130e+00 1.10564542e+00 -5.01763344e-01
4.74718064e-01 -2.62921184e-01 -1.90040946e-01 -1.06276011e+00
-4.27630574e-01 -3.83312285e-01 -2.79552758e-01 7.18009353e-01
4.36980158e-01 -9.95441079e-01 6.97532654e-01 6.57192886e-01
-3.67479235e-01 -9.67670441e-01 -1.11730063e+00 -6.91187739e-01
-4.17875759e-02 -5.70238471e-01 6.92066014e-01 9.02689338e-01
2.66432613e-01 3.92092049e-01 -3.87550384e-01 1.10530511e-01
4.54750568e-01 -1.35868579e-01 2.49835759e-01 -1.40657973e+00
-3.42383355e-01 -3.38648856e-01 -6.94379926e-01 -6.24548912e-01
3.43487829e-01 -1.41004527e+00 -6.53053880e-01 -1.42951858e+00
1.39368609e-01 -3.51678699e-01 -1.71303749e-01 3.60865235e-01
5.96695900e-01 -2.17427909e-01 -1.49444386e-01 2.65468985e-01
-3.00132602e-01 7.63119400e-01 1.67435241e+00 -1.87598974e-01
-1.36304513e-01 -1.90043375e-01 -8.14248547e-02 4.95628119e-01
6.79488361e-01 -3.22363436e-01 1.94813367e-02 5.25868014e-02
9.55474317e-01 1.33411288e-01 4.43883032e-01 -1.16874456e+00
2.99374074e-01 3.26371975e-02 1.35071203e-01 -6.99815691e-01
3.88849229e-01 -7.62926936e-01 6.53357029e-01 7.07781196e-01
-3.02063152e-02 2.93998457e-02 -1.38375401e-01 2.03585401e-01
-3.02143633e-01 -2.31856599e-01 2.95505047e-01 7.50506446e-02
-2.69721270e-01 5.43891013e-01 -3.71648520e-01 -6.17733240e-01
8.41848552e-01 -2.47792602e-01 -4.28128093e-01 -1.60285115e-01
-3.49301070e-01 -3.87474954e-01 2.96027541e-01 -1.63227126e-01
6.23782456e-01 -1.51141322e+00 -3.31343174e-01 3.93543452e-01
-2.00433537e-01 -4.95136715e-02 5.81385136e-01 1.03796196e+00
-9.31129158e-01 7.46183574e-01 -3.63840669e-01 -5.61163008e-01
-9.92488086e-01 4.17591810e-01 5.93221545e-01 -3.83490264e-01
-3.69104326e-01 3.08097929e-01 -2.39334390e-01 -7.11892784e-01
-1.93655461e-01 -5.23342133e-01 -2.34028488e-01 -2.76247054e-01
8.92187804e-02 6.11280203e-01 3.68270874e-01 -9.79530036e-01
-3.21621805e-01 7.66170859e-01 1.62693132e-02 3.36199939e-01
1.60821736e+00 4.74544823e-01 -1.02420676e+00 1.38303533e-01
1.46436119e+00 -7.59305879e-02 -3.85039061e-01 -5.73552074e-03
-4.52569515e-01 2.63613045e-01 2.23958731e-01 -2.38822564e-01
-1.11539197e+00 6.82837188e-01 4.07154620e-01 4.97921050e-01
7.02365994e-01 -1.05345346e-01 9.20711219e-01 7.49224901e-01
2.53638744e-01 -9.33841527e-01 -1.15879998e-01 7.16069162e-01
6.64681017e-01 -4.82696056e-01 7.97611922e-02 -5.91416121e-01
6.37896582e-02 1.61027849e+00 4.37918156e-01 -5.37499338e-02
8.58352542e-01 -9.93493199e-02 -7.04889774e-01 -4.68779296e-01
-5.09013712e-01 3.38701397e-01 5.36577761e-01 3.37450892e-01
7.70577967e-01 3.68641585e-01 -5.26297331e-01 1.12613559e-01
-3.67502838e-01 -1.08184122e-01 6.67676330e-01 4.40643698e-01
-2.83729374e-01 -1.40831685e+00 -2.96645373e-01 5.15450299e-01
-7.56101534e-02 -6.60052150e-02 -1.43929556e-01 6.24101162e-01
3.20639998e-01 5.22861242e-01 -1.26192063e-01 -7.13165641e-01
1.56204745e-01 7.52841607e-02 4.96488273e-01 -1.02313131e-01
-1.92885831e-01 -6.45470917e-01 1.75467178e-01 -4.70084757e-01
-6.65474236e-01 -4.24570382e-01 -1.40562856e+00 -7.53076017e-01
-4.43550855e-01 3.39906752e-01 9.93801355e-01 1.09906590e+00
4.53094661e-01 5.40396750e-01 5.36049128e-01 -1.01389670e+00
-3.21339488e-01 -8.17810595e-01 -7.96689272e-01 3.29410076e-01
5.62022962e-02 -6.76173270e-01 -4.36725497e-01 -5.46802759e-01] | [5.543237686157227, 5.36561918258667] |
7f914d35-5864-4712-be53-c78d04b11b55 | 190409380 | 1904.09380 | null | http://arxiv.org/abs/1904.09380v1 | http://arxiv.org/pdf/1904.09380v1.pdf | Repurposing Entailment for Multi-Hop Question Answering Tasks | Question Answering (QA) naturally reduces to an entailment problem, namely,
verifying whether some text entails the answer to a question. However, for
multi-hop QA tasks, which require reasoning with multiple sentences, it remains
unclear how best to utilize entailment models pre-trained on large scale
datasets such as SNLI, which are based on sentence pairs. We introduce Multee,
a general architecture that can effectively use entailment models for multi-hop
QA tasks. Multee uses (i) a local module that helps locate important sentences,
thereby avoiding distracting information, and (ii) a global module that
aggregates information by effectively incorporating importance weights.
Importantly, we show that both modules can use entailment functions pre-trained
on a large scale NLI datasets. We evaluate performance on MultiRC and
OpenBookQA, two multihop QA datasets. When using an entailment function
pre-trained on NLI datasets, Multee outperforms QA models trained only on the
target QA datasets and the OpenAI transformer models. The code is available at
https://github.com/StonyBrookNLP/multee. | ['Ashish Sabharwal', 'Tushar Khot', 'Niranjan Balasubramanian', 'Heeyoung Kwon', 'Harsh Trivedi'] | 2019-04-20 | repurposing-entailment-for-multi-hop-question | https://aclanthology.org/N19-1302 | https://aclanthology.org/N19-1302.pdf | naacl-2019-6 | ['multi-hop-question-answering'] | ['knowledge-base'] | [ 1.78660303e-01 3.24840158e-01 3.55233282e-01 -6.27632916e-01
-1.81853652e+00 -7.58880615e-01 4.95239168e-01 2.54295826e-01
-4.55575377e-01 7.63153195e-01 6.32902741e-01 -7.69469500e-01
-1.51369274e-01 -8.66022646e-01 -7.80273557e-01 -4.35960256e-02
2.35843569e-01 8.64475310e-01 3.46627504e-01 -5.92038989e-01
2.09253266e-01 -6.26473203e-02 -1.22137105e+00 8.57389450e-01
1.09811413e+00 8.82260263e-01 5.84999844e-02 1.01609993e+00
-2.84492761e-01 1.56763792e+00 -6.63280189e-01 -8.61503959e-01
2.53762826e-02 -3.68612170e-01 -1.74583459e+00 -4.23491240e-01
7.51377642e-01 -4.49710429e-01 -2.64335304e-01 6.66810930e-01
3.29672694e-01 6.30080476e-02 3.48773599e-01 -1.23466849e+00
-8.28143477e-01 8.68690968e-01 -6.82703704e-02 4.59758073e-01
8.93998861e-01 3.33557397e-01 1.94441020e+00 -8.41196597e-01
5.23389220e-01 1.32351696e+00 6.36338651e-01 2.23428488e-01
-9.39352989e-01 -1.44072592e-01 -1.56208187e-01 8.55074823e-01
-1.00157547e+00 -6.06154501e-01 4.71241117e-01 1.49759084e-01
1.44106388e+00 4.47370648e-01 2.17248239e-02 6.99794710e-01
1.97220504e-01 1.11359775e+00 7.68356800e-01 -4.06345874e-01
-2.81542186e-02 -3.79278779e-01 6.72967017e-01 8.14725935e-01
-2.99493372e-01 -5.71813583e-01 -4.23424214e-01 -3.69989634e-01
-3.30253035e-01 -3.42758656e-01 -3.02985251e-01 9.85683724e-02
-1.27392709e+00 1.04165113e+00 4.10390496e-01 3.15740913e-01
-2.65379220e-01 1.95629969e-01 4.08374041e-01 9.65939045e-01
2.04161584e-01 8.90803158e-01 -7.25251198e-01 -2.44158790e-01
-7.46210754e-01 2.79584736e-01 1.11970270e+00 8.92725706e-01
8.91729891e-01 -7.47384012e-01 -4.51711982e-01 6.80685401e-01
2.33794019e-01 4.88508224e-01 1.70863390e-01 -1.42094350e+00
9.98784900e-01 7.41910994e-01 5.09138256e-02 -5.93344748e-01
-3.73185724e-01 -2.05191240e-01 -2.81723291e-01 -3.48259598e-01
6.77529037e-01 -2.41748661e-01 -5.34486234e-01 1.70894849e+00
2.42435038e-01 -2.31904075e-01 4.66010749e-01 7.84292340e-01
1.07721555e+00 8.16121042e-01 -1.10363439e-01 1.93451628e-01
1.66911066e+00 -1.26893532e+00 -5.83352566e-01 -5.69574535e-01
1.18527448e+00 -9.52784121e-01 1.47007120e+00 2.76292861e-01
-1.45863748e+00 -1.85307533e-01 -9.41489279e-01 -9.42602277e-01
-2.82286227e-01 -1.82118252e-01 3.77930731e-01 1.20093286e-01
-1.32978177e+00 2.98792988e-01 -3.28565419e-01 -2.89755553e-01
2.10589051e-01 -1.74383968e-02 -4.53646392e-01 -7.92720675e-01
-1.79806197e+00 1.34644353e+00 -1.31758958e-01 9.66798812e-02
-7.58378863e-01 -6.88902080e-01 -1.29991865e+00 2.34786585e-01
4.81985182e-01 -9.00918007e-01 1.79189599e+00 -5.11966348e-01
-1.19114602e+00 6.35560572e-01 -5.72879255e-01 -5.15863836e-01
1.34142667e-01 -2.71461606e-01 -2.12544635e-01 5.55153906e-01
5.56652248e-01 4.86727715e-01 4.87646818e-01 -7.51954317e-01
-5.19893408e-01 -4.49067324e-01 8.72624934e-01 3.86415571e-01
-8.71943980e-02 1.76833346e-01 -2.59814650e-01 9.85433832e-02
-5.51497266e-02 -5.50373375e-01 -3.30458470e-02 -5.53255528e-02
-3.95640969e-01 -6.04286253e-01 6.37449861e-01 -9.85646605e-01
1.09376967e+00 -1.80347049e+00 2.70725936e-02 6.85361102e-02
2.03062728e-01 6.62421510e-02 -6.30834639e-01 7.49463141e-01
1.77254915e-01 -9.52254310e-02 -4.95330513e-01 -2.65757859e-01
1.92608893e-01 4.91029143e-01 -2.86343843e-01 1.31683901e-01
6.49731755e-01 1.20061314e+00 -1.14131081e+00 -7.23448634e-01
-1.87098891e-01 -5.03459498e-02 -6.39122963e-01 1.63230926e-01
-5.80308557e-01 1.22367032e-02 -3.98825020e-01 5.30234516e-01
5.57507157e-01 -6.74695432e-01 8.96054581e-02 -2.54393190e-01
2.97968954e-01 1.09652734e+00 -7.16366827e-01 1.96565306e+00
-8.83390665e-01 6.66444719e-01 2.54006505e-01 -1.08122289e+00
4.54113156e-01 4.37903702e-01 2.24260166e-02 -9.66264725e-01
-1.10624388e-01 2.53785849e-01 1.60660744e-02 -7.13284969e-01
6.54713452e-01 2.28434503e-02 -3.31059486e-01 9.39670205e-01
2.62529939e-01 -3.21496844e-01 7.15392530e-01 9.13938522e-01
1.83116698e+00 -3.64214897e-01 2.31863298e-02 -1.38083175e-01
8.32867622e-01 4.57014591e-01 1.26218811e-01 7.73116112e-01
-2.78566778e-01 4.89421874e-01 6.69227839e-01 -1.72701284e-01
-7.15005159e-01 -1.13651216e+00 -1.44252911e-01 1.17910278e+00
1.49776444e-01 -6.12106919e-01 -4.43964630e-01 -9.06488657e-01
-1.40588492e-01 9.84749138e-01 -3.57249022e-01 -7.36739263e-02
-7.63285935e-01 -3.38948339e-01 8.65047634e-01 4.45442557e-01
4.55948144e-01 -9.43786025e-01 -2.59884536e-01 1.16475351e-01
-1.14389586e+00 -1.13354087e+00 -4.75235462e-01 1.43694282e-01
-5.03825605e-01 -1.19144917e+00 -2.15532154e-01 -6.20271504e-01
4.25862432e-01 3.49886537e-01 1.85141635e+00 5.09075761e-01
-1.11230239e-01 6.61192477e-01 -5.79098344e-01 -5.14445007e-02
-5.48664749e-01 3.07114124e-01 -4.94303554e-01 -3.77297878e-01
5.72286665e-01 -3.43032181e-01 -4.63210464e-01 1.94766372e-01
-8.78183007e-01 -1.21776946e-01 5.96792161e-01 8.40068281e-01
1.44389570e-01 -3.88951391e-01 7.52121985e-01 -7.30266333e-01
8.58619392e-01 -7.19035029e-01 -1.47077709e-01 6.40649498e-01
-2.82153130e-01 4.24506485e-01 6.24386430e-01 2.45259687e-01
-1.24567926e+00 -5.20003498e-01 -8.36127043e-01 2.95877993e-01
4.39254642e-02 7.36393750e-01 -1.96689323e-01 3.34663212e-01
7.06771672e-01 6.46072552e-02 1.14697985e-01 -1.57494321e-01
8.46923828e-01 8.29291582e-01 5.56745648e-01 -7.94168413e-01
6.97592258e-01 2.88110971e-01 -2.03680664e-01 -4.76505071e-01
-1.44634819e+00 -6.50963783e-01 -4.34926689e-01 1.27352670e-01
7.99911320e-01 -8.75622332e-01 -7.52855241e-01 -1.06353097e-01
-1.39481497e+00 -4.53815013e-01 -3.24902505e-01 1.97568074e-01
-4.89978909e-01 6.73000276e-01 -1.06412303e+00 -4.17991132e-01
-8.05817962e-01 -9.64038312e-01 1.07115519e+00 -2.20577829e-02
-4.93588984e-01 -1.02040172e+00 4.73123938e-02 1.17958224e+00
4.70754385e-01 -4.71833795e-01 1.16371429e+00 -8.11035514e-01
-5.84948957e-01 -1.63077265e-01 -2.79845804e-01 4.52025056e-01
-1.17284037e-01 -1.53059840e-01 -1.02300620e+00 -3.69665995e-02
1.24721033e-02 -1.14503467e+00 8.16018343e-01 -9.98888388e-02
8.05056095e-01 -4.31494892e-01 1.63184106e-01 -3.33100557e-02
9.93219078e-01 -3.73038471e-01 7.48582602e-01 2.84521431e-01
2.63128251e-01 6.23871922e-01 7.66852915e-01 9.87260863e-02
1.10596395e+00 2.80903012e-01 5.32455087e-01 1.32026732e-01
-8.26290697e-02 -1.35997608e-01 4.19373065e-01 9.32067275e-01
5.42333186e-01 -2.49794975e-01 -9.76109743e-01 8.03173006e-01
-1.88382959e+00 -1.16637945e+00 -3.75834346e-01 1.62120140e+00
1.12209380e+00 4.22478057e-02 -2.86000073e-01 8.35541934e-02
1.79859444e-01 2.00014561e-01 -3.71772289e-01 -6.16175115e-01
-2.32723981e-01 4.24338847e-01 -1.43265799e-01 1.00065637e+00
-8.82042885e-01 7.23129451e-01 5.72674179e+00 7.51031280e-01
-3.80943418e-01 3.37583721e-01 4.50832576e-01 -7.07149282e-02
-8.22629988e-01 1.66279927e-01 -5.42324960e-01 2.17189938e-01
1.13585448e+00 -2.47614637e-01 3.47350627e-01 5.18307924e-01
-1.76397592e-01 -2.24020198e-01 -1.27057314e+00 2.87012011e-01
2.35397100e-01 -1.39120972e+00 5.94723411e-02 -5.59121013e-01
2.68547952e-01 3.61753374e-01 -2.75803387e-01 7.35010624e-01
6.68719172e-01 -9.79764402e-01 4.23656762e-01 2.99669147e-01
4.17715937e-01 -5.01103222e-01 1.06262088e+00 4.18262392e-01
-1.07746911e+00 -7.59599358e-02 -3.90066147e-01 -2.85750747e-01
4.74840730e-01 5.80360770e-01 -7.88941622e-01 8.78360271e-01
6.75548732e-01 5.39588094e-01 -7.54596233e-01 6.51513875e-01
-7.08003819e-01 6.18118227e-01 -4.16473061e-01 -9.82258916e-02
5.19475877e-01 5.20455875e-02 3.62171412e-01 1.04053187e+00
1.28735498e-01 3.53315310e-03 -1.91337958e-01 7.46663511e-01
-2.67785668e-01 -1.50460884e-01 -4.54245031e-01 1.47877038e-01
5.41414261e-01 1.43317282e+00 1.66281089e-01 -4.52333540e-01
-8.44944239e-01 1.10153377e+00 7.14279950e-01 1.87432855e-01
-7.13592291e-01 -5.89583457e-01 6.58361197e-01 -2.42085025e-01
1.31284833e-01 -4.00858335e-02 -8.72964785e-02 -1.31360435e+00
3.62445474e-01 -1.22395051e+00 1.03495491e+00 -1.27746487e+00
-1.46875501e+00 6.45565331e-01 -2.62194991e-01 -8.22234869e-01
-5.76507032e-01 -6.66607738e-01 -8.31860602e-01 8.84541869e-01
-1.81359482e+00 -1.19643164e+00 -1.63218409e-01 8.99946332e-01
5.60723066e-01 3.61107320e-01 8.29364598e-01 3.14793617e-01
-2.65148044e-01 5.59917152e-01 -3.92698109e-01 2.19642207e-01
8.90679598e-01 -1.41242778e+00 5.01044571e-01 9.43899632e-01
4.20462906e-01 7.69373357e-01 6.62605524e-01 -1.67141125e-01
-1.66235614e+00 -8.70198131e-01 1.59421587e+00 -1.07910776e+00
9.91841793e-01 -1.47576645e-01 -1.09177387e+00 1.02744353e+00
6.70556366e-01 -1.81766897e-01 8.64984632e-01 3.87199193e-01
-5.46022713e-01 -2.57184654e-02 -1.09066784e+00 5.01487851e-01
9.32908773e-01 -8.21512043e-01 -1.27859318e+00 6.03522539e-01
1.14142740e+00 -3.72954786e-01 -1.08168793e+00 2.50965536e-01
1.89574406e-01 -7.29684174e-01 1.00602531e+00 -7.90531993e-01
8.23934853e-01 -3.40833694e-01 -3.25184315e-01 -1.32380581e+00
-1.49509773e-01 -3.28328699e-01 -1.80184558e-01 1.05459273e+00
1.10634220e+00 -6.15725696e-01 3.81539077e-01 8.45774114e-01
-3.20811152e-01 -7.60182619e-01 -1.09477115e+00 -5.44380665e-01
3.04277092e-01 -5.53402781e-01 6.36760771e-01 7.05445290e-01
4.97698754e-01 1.03459609e+00 2.25958955e-02 2.15849683e-01
3.57791871e-01 4.15119290e-01 5.94165027e-01 -8.93690705e-01
-2.86894828e-01 -2.44601164e-02 1.36052296e-01 -1.27597654e+00
3.64569187e-01 -1.07568336e+00 2.37418473e-01 -2.08256435e+00
2.13364407e-01 -2.92408198e-01 1.00129813e-01 7.05943167e-01
-4.92649287e-01 5.12324572e-02 4.58534770e-02 5.84914698e-04
-1.13664651e+00 6.26927793e-01 1.21879268e+00 -3.71722639e-01
4.66069728e-01 -2.00322181e-01 -1.00774217e+00 5.19913852e-01
9.42003191e-01 -2.53638446e-01 -6.37251079e-01 -1.13898313e+00
6.35861993e-01 1.49476901e-01 3.61987352e-01 -7.78444707e-01
4.87453997e-01 1.37770608e-01 -7.78748468e-02 -5.44757962e-01
4.76725310e-01 -5.49743116e-01 -6.45952046e-01 2.69029945e-01
-5.92024982e-01 3.85030419e-01 1.31799608e-01 2.38052055e-01
-4.81366515e-01 -5.39659262e-01 2.76002288e-01 -4.00093555e-01
-6.55013442e-01 7.55323917e-02 -4.52949077e-01 7.62332022e-01
5.57343423e-01 5.07013798e-01 -1.04453504e+00 -7.87013948e-01
-3.76084715e-01 9.27374065e-01 1.17056414e-01 1.61723837e-01
7.66838133e-01 -1.07396722e+00 -9.61139262e-01 -1.96720675e-01
3.88969451e-01 5.59485033e-02 3.35256994e-01 9.93828475e-01
-4.42903072e-01 5.61264753e-01 2.02266783e-01 -3.65421802e-01
-1.18363059e+00 3.87287527e-01 3.12109977e-01 -8.02084923e-01
-2.79650569e-01 9.98118818e-01 -1.57758638e-01 -1.07667136e+00
-1.61028683e-01 -4.32470769e-01 -1.57763183e-01 -7.95797482e-02
7.13740706e-01 1.73726439e-01 3.17820191e-01 -4.10514832e-01
-5.47203362e-01 8.76915529e-02 -1.29271418e-01 -2.84105867e-01
1.00619686e+00 -4.47544426e-01 -5.77482462e-01 1.30431831e-01
1.29258204e+00 -1.55657962e-01 -6.02932751e-01 -6.09032393e-01
2.04269573e-01 -1.23469360e-01 -1.55967742e-01 -9.25487161e-01
-5.40327609e-01 1.06870961e+00 -2.81664342e-01 2.47815788e-01
1.01957500e+00 3.19130242e-01 1.21090150e+00 9.73287165e-01
4.18349028e-01 -8.54887724e-01 3.32435817e-01 9.61166739e-01
1.22330093e+00 -1.38623726e+00 -3.86112034e-01 -1.27725020e-01
-7.67534256e-01 9.66600239e-01 6.10177934e-01 3.31771642e-01
3.11211467e-01 2.16891855e-01 2.61736035e-01 -4.54881400e-01
-1.26324821e+00 -4.18278843e-01 1.37241513e-01 3.05843771e-01
4.60583895e-01 -1.48736045e-01 -1.64487079e-01 6.15628958e-01
-4.80334759e-01 -1.54858053e-01 5.47098577e-01 1.09979486e+00
-4.83811349e-01 -1.12151325e+00 -2.57048845e-01 6.55802310e-01
-4.72329617e-01 -6.39035881e-01 -3.79744858e-01 5.04460275e-01
-3.54397357e-01 1.67396891e+00 7.72539303e-02 -2.28705674e-01
3.43122929e-01 3.43702316e-01 3.78532708e-01 -5.63041508e-01
-6.16246045e-01 -1.01920986e+00 8.59052360e-01 -7.54603207e-01
-1.94933116e-01 -6.05357587e-01 -1.27589679e+00 -6.07352436e-01
-9.89612266e-02 4.92588252e-01 1.15104415e-01 1.29408109e+00
5.09077132e-01 3.12145591e-01 3.46995115e-01 5.78664988e-02
-8.58902037e-01 -1.09423208e+00 -9.39446464e-02 5.67294180e-01
4.03202444e-01 5.83731756e-02 -4.11361188e-01 -2.14662015e-01] | [11.136187553405762, 7.943655967712402] |
c909bec6-ec1d-4ad9-b45c-fdb9ff3786d0 | classification-based-financial-markets | 1603.08604 | null | http://arxiv.org/abs/1603.08604v2 | http://arxiv.org/pdf/1603.08604v2.pdf | Classification-based Financial Markets Prediction using Deep Neural Networks | Deep neural networks (DNNs) are powerful types of artificial neural networks
(ANNs) that use several hidden layers. They have recently gained considerable
attention in the speech transcription and image recognition community
(Krizhevsky et al., 2012) for their superior predictive properties including
robustness to overfitting. However their application to algorithmic trading has
not been previously researched, partly because of their computational
complexity. This paper describes the application of DNNs to predicting
financial market movement directions. In particular we describe the
configuration and training approach and then demonstrate their application to
backtesting a simple trading strategy over 43 different Commodity and FX future
mid-prices at 5-minute intervals. All results in this paper are generated using
a C++ implementation on the Intel Xeon Phi co-processor which is 11.4x faster
than the serial version and a Python strategy backtesting environment both of
which are available as open source code written by the authors. | ['Matthew Dixon', 'Diego Klabjan', 'Jin Hoon Bang'] | 2016-03-29 | null | null | null | null | ['algorithmic-trading'] | ['time-series'] | [-2.12515414e-01 -3.92312199e-01 -1.55149335e-02 -4.71415609e-01
-7.94838667e-02 -8.43859971e-01 6.36715770e-01 -3.15656155e-01
-5.83412588e-01 6.05606616e-01 -1.97639048e-01 -1.06561100e+00
-7.81899393e-02 -8.50439012e-01 -5.76644182e-01 -4.44010079e-01
-3.66087973e-01 6.86577260e-01 8.74553770e-02 -2.77687162e-01
1.76526427e-01 7.20672309e-01 -1.46022975e+00 2.06856743e-01
3.31995457e-01 1.45961761e+00 -1.76660925e-01 8.98774385e-01
8.46874788e-02 1.04979205e+00 -7.21736550e-01 -4.20097083e-01
7.88017571e-01 -3.04773957e-01 -3.40223014e-01 -2.79714972e-01
-2.33173929e-02 -5.56638420e-01 -2.72234768e-01 1.13214016e+00
6.73967242e-01 3.61627415e-02 1.77125275e-01 -1.03750885e+00
-5.10066092e-01 8.26808572e-01 -2.02086106e-01 7.16628850e-01
-3.44418406e-01 1.76511616e-01 8.87184739e-01 -7.59090185e-01
1.59539342e-01 1.06498456e+00 7.28741825e-01 1.83470190e-01
-1.03138185e+00 -1.00013447e+00 -4.15074438e-01 2.18698457e-01
-8.70014489e-01 -2.49009907e-01 6.54259682e-01 -3.83012801e-01
1.45021892e+00 2.78785497e-01 8.24951470e-01 1.05520666e+00
2.59507328e-01 8.20191264e-01 1.12405229e+00 -4.48843390e-01
3.82052213e-01 1.66542351e-01 1.05925992e-01 3.81541431e-01
1.18398502e-01 7.67472982e-01 -2.00675786e-01 -3.15382570e-01
1.05832100e+00 -2.88121253e-02 2.71912903e-01 -8.16677362e-02
-1.03959620e+00 1.32791638e+00 3.79706174e-01 3.90527070e-01
-6.97592258e-01 2.27931395e-01 4.80492234e-01 8.49108398e-01
3.95028055e-01 4.27081317e-01 -9.42877889e-01 -6.01426065e-01
-8.96800697e-01 5.64351082e-01 1.12495029e+00 4.91099447e-01
1.74112707e-01 8.57453763e-01 3.71452570e-01 8.19911659e-01
3.62001173e-02 3.01895678e-01 1.11479878e+00 -9.38047707e-01
3.04578602e-01 2.55356491e-01 -8.04103538e-02 -9.49825227e-01
-4.89584208e-01 -7.17248917e-01 -8.41904581e-01 5.74132860e-01
5.17540693e-01 -7.27653086e-01 -8.04239750e-01 1.26493311e+00
-7.56163150e-02 2.68491864e-01 2.16429353e-01 5.97640038e-01
4.69142377e-01 7.91889787e-01 -2.82553047e-01 9.76707116e-02
1.26567614e+00 -9.16301727e-01 -2.65870839e-01 -1.57046363e-01
2.80708730e-01 -7.87759602e-01 7.24076748e-01 6.63066268e-01
-1.13371336e+00 -5.93578696e-01 -1.12785101e+00 4.07373488e-01
-6.43428683e-01 -2.03997083e-02 9.25539374e-01 9.89399076e-01
-1.04616630e+00 8.54716778e-01 -9.47179377e-01 1.28168032e-01
2.91515708e-01 6.21032238e-01 3.44994605e-01 8.01494300e-01
-1.48768318e+00 8.25304210e-01 6.37401402e-01 1.91916585e-01
-5.94771802e-01 -4.74887341e-01 -4.78692025e-01 1.82834238e-01
2.74905503e-01 -3.17199051e-01 1.88734627e+00 -1.34936059e+00
-2.02280426e+00 4.23832923e-01 6.39857888e-01 -1.32259893e+00
6.92430794e-01 -8.18364397e-02 -6.62941813e-01 -2.04205588e-01
-3.93865913e-01 5.88022113e-01 7.33575404e-01 -1.46776468e-01
-6.15264714e-01 -2.00415760e-01 -3.86687852e-02 -1.18856601e-01
-2.21741214e-01 3.50862652e-01 1.08764000e-01 -1.43405044e+00
-2.41948321e-01 -9.88279402e-01 -1.95337191e-01 -4.50538665e-01
-2.45563850e-01 -6.35584295e-02 7.05795288e-01 -7.54719913e-01
9.27914560e-01 -1.95962119e+00 -4.15815413e-01 3.65394324e-01
-3.41371417e-01 4.53744352e-01 2.05164224e-01 2.65152276e-01
-5.27413726e-01 -3.30453031e-02 -1.17005818e-01 2.04909429e-01
3.13242257e-01 2.45842665e-01 -5.09158611e-01 3.19208741e-01
-9.59329978e-02 1.01214695e+00 -3.14995348e-01 1.76700890e-01
2.61216044e-01 2.62190104e-01 -2.93950081e-01 -5.49183711e-02
-3.76294345e-01 7.49914441e-03 -3.09226543e-01 6.14866138e-01
4.63118464e-01 -4.50216122e-02 7.34198317e-02 2.92576581e-01
-3.08082581e-01 3.11682761e-01 -1.10021293e+00 8.96744013e-01
-8.13729092e-02 8.56988966e-01 -1.28931656e-01 -1.28066540e+00
9.63825822e-01 4.28459853e-01 2.13880122e-01 -9.40647483e-01
3.82136673e-01 4.96551037e-01 3.92072529e-01 -5.12579978e-02
3.44366729e-01 -3.04677278e-01 1.07303359e-01 9.02688801e-01
-1.47558853e-01 1.60444781e-01 2.49801934e-01 -4.42063510e-01
7.68025219e-01 -1.40088707e-01 1.83495298e-01 -1.21681377e-01
2.04666466e-01 8.21928084e-02 5.13569653e-01 8.76192927e-01
-3.83124091e-02 2.22062454e-01 5.40813208e-01 -9.74425137e-01
-1.26538277e+00 -8.19561839e-01 -2.63973445e-01 1.00351238e+00
-6.94731593e-01 1.76016346e-01 -7.04336762e-01 -1.73178222e-02
7.71346912e-02 8.85702133e-01 -4.20988470e-01 1.61309958e-01
-6.46095455e-01 -8.69797468e-01 7.85190701e-01 7.70633042e-01
9.30027306e-01 -1.66253197e+00 -1.15382087e+00 5.22742271e-01
6.12777114e-01 -7.57096410e-01 5.60552999e-02 4.41592991e-01
-1.08226538e+00 -9.28188562e-01 -9.25027192e-01 -6.53146207e-01
-8.14821124e-02 -3.46681058e-01 1.07244432e+00 -1.15643695e-01
-1.64071366e-01 -1.35647357e-01 -1.83616765e-02 -8.18504870e-01
-4.98902589e-01 1.25598580e-01 1.59852251e-01 -1.83568239e-01
5.42804301e-01 -4.89332974e-01 -3.17953408e-01 1.06701039e-01
-8.66866529e-01 1.00143284e-01 6.65112972e-01 9.61007118e-01
2.56093591e-01 3.87102157e-01 4.01916236e-01 -8.18950534e-01
1.00138927e+00 -4.21723545e-01 -1.50119829e+00 -1.45112826e-02
-8.83751631e-01 -6.50707856e-02 5.48406959e-01 -4.08817202e-01
-8.64582539e-01 -8.32259655e-02 -3.60607177e-01 -4.07591432e-01
-5.87608106e-02 7.35766768e-01 4.13350135e-01 3.02203298e-02
1.83130398e-01 1.75513759e-01 1.96743339e-01 -6.84698701e-01
-7.33880252e-02 4.66427714e-01 5.71420193e-01 -2.29946271e-01
6.35253191e-01 -5.93920015e-02 -1.65263593e-01 -6.11724079e-01
-2.29379579e-01 1.78088188e-01 -2.14977369e-01 2.01830780e-03
7.15263426e-01 -6.99338377e-01 -9.55368996e-01 9.05287325e-01
-7.48890698e-01 -4.43518251e-01 -7.37206712e-02 6.37585104e-01
-4.82056707e-01 -3.39958705e-02 -1.13007450e+00 -7.30051279e-01
-5.71532786e-01 -1.18122208e+00 2.03729525e-01 3.77251476e-01
-2.36573920e-01 -1.14326489e+00 1.40978590e-01 -5.88559397e-02
7.89739847e-01 2.78263390e-01 8.93884838e-01 -1.15693331e+00
-2.94370890e-01 -3.34964424e-01 -4.29964587e-02 6.69728994e-01
-1.41055375e-01 9.60483029e-03 -9.96046424e-01 -2.78553903e-01
2.93910623e-01 -8.23190659e-02 7.36223876e-01 7.11056232e-01
9.55089092e-01 -5.56849539e-01 2.69527823e-01 5.71231425e-01
1.24087334e+00 8.92509401e-01 3.99472743e-01 1.01481616e+00
1.27698362e-01 3.26995194e-01 1.36352628e-01 5.14145374e-01
-1.82102337e-01 5.19487143e-01 3.47192585e-01 -1.98211536e-01
6.53889358e-01 2.07265466e-01 4.67783362e-01 4.30149376e-01
-4.67843749e-02 -7.81800896e-02 -1.11526525e+00 1.09108031e-01
-1.68807065e+00 -1.16723824e+00 1.66300423e-02 1.93275690e+00
7.53864527e-01 5.67221940e-01 5.71607709e-01 2.88179904e-01
5.94643652e-01 -3.68871428e-02 -7.71324337e-01 -8.17233264e-01
-3.35908651e-01 6.21588767e-01 6.83443248e-01 4.28261906e-01
-1.20143390e+00 7.00234950e-01 6.46823311e+00 6.32914603e-01
-1.19573975e+00 -2.01837569e-02 1.13267553e+00 -2.44504541e-01
6.75754249e-02 -2.56322920e-01 -5.77758372e-01 5.36999524e-01
1.55377865e+00 -2.69025534e-01 5.71564078e-01 1.06182730e+00
2.16812789e-02 6.97783083e-02 -7.27114856e-01 8.89864981e-01
-4.38688397e-01 -1.65365088e+00 -2.66618907e-01 2.92404234e-01
4.88773257e-01 5.35408199e-01 4.96697724e-01 4.70399976e-01
6.04495704e-01 -1.04824877e+00 9.83173668e-01 1.78192973e-01
1.33014187e-01 -1.30323756e+00 1.04541922e+00 2.57370770e-01
-6.29354298e-01 -3.88982058e-01 -3.57255369e-01 -3.28083903e-01
-1.35973878e-02 4.19242084e-01 -8.44015598e-01 1.18209273e-01
1.02847195e+00 3.55967253e-01 -5.60803235e-01 7.84676373e-01
5.05079292e-02 8.76773238e-01 -6.10526323e-01 -3.42648774e-01
7.17940569e-01 -2.37394348e-01 2.70525843e-01 1.07346356e+00
4.03326541e-01 -2.25616284e-02 -2.68101960e-01 8.59406650e-01
5.31089827e-02 -2.89009452e-01 -3.33911777e-01 -4.28555399e-01
2.66503841e-01 9.48630989e-01 -9.15724576e-01 -4.12536770e-01
-5.22854924e-01 6.13913655e-01 -1.62091613e-01 3.70274782e-01
-8.32617521e-01 -7.04034388e-01 8.07379007e-01 -2.46275753e-01
6.24408364e-01 -1.29312679e-01 -4.89674240e-01 -8.85920584e-01
-1.30795121e-01 -1.04254138e+00 5.32852888e-01 -7.62777328e-01
-1.01680398e+00 8.31874371e-01 1.45304119e-02 -1.05319428e+00
-9.98153269e-01 -1.05375063e+00 -4.28401768e-01 9.15120125e-01
-1.21070170e+00 -3.26619893e-01 2.97383308e-01 4.04225916e-01
4.45448548e-01 -9.03885782e-01 7.79598355e-01 1.99879110e-01
-6.15050077e-01 3.62187773e-01 5.24906695e-01 4.50849771e-01
-4.01667543e-02 -1.29723692e+00 1.05242050e+00 8.28630567e-01
3.84611994e-01 5.50611079e-01 8.33297491e-01 -4.44700658e-01
-1.12006247e+00 -7.81447172e-01 8.13434899e-01 7.31096938e-02
9.49772477e-01 -2.41782904e-01 -7.93738782e-01 9.98805463e-01
6.47070348e-01 -2.62554377e-01 6.30921245e-01 -3.20974588e-01
-6.51874915e-02 -1.56135902e-01 -1.05034983e+00 6.69293046e-01
2.32766718e-01 -2.82356262e-01 -4.40644532e-01 1.60132527e-01
2.93542176e-01 -5.19584715e-01 -7.08111882e-01 1.74882188e-01
5.19081175e-01 -1.50900912e+00 9.46221113e-01 -4.59747553e-01
3.95428091e-02 9.75711569e-02 -7.24190939e-03 -1.11847818e+00
-1.53448567e-01 -8.49080563e-01 -2.27140397e-01 8.55251431e-01
6.07857049e-01 -1.12296665e+00 1.03042293e+00 5.31236351e-01
1.60265714e-01 -7.48470902e-01 -1.08765602e+00 -9.62543964e-01
1.04601495e-01 -7.33378828e-01 9.75416303e-01 8.46153140e-01
-1.68978870e-01 1.35383144e-01 -3.57710779e-01 -1.87735304e-01
2.53682792e-01 1.70835063e-01 4.23613042e-01 -1.29598713e+00
-7.85313666e-01 -8.91796291e-01 -3.04908127e-01 -7.31754541e-01
5.19338176e-02 -5.38271964e-01 -4.03995782e-01 -6.63779557e-01
-3.71282667e-01 -1.67962715e-01 -4.03499216e-01 6.85666978e-01
5.44342577e-01 2.65307546e-01 1.54615402e-01 1.71889469e-01
1.50075331e-01 1.92637160e-01 3.49084884e-01 -1.08129852e-01
-8.38705897e-02 4.63735670e-01 -4.46680069e-01 7.52302885e-01
1.29686284e+00 -3.27263445e-01 -2.92780966e-01 -2.82131463e-01
1.38855979e-01 1.16424806e-01 3.93004894e-01 -1.03541279e+00
1.44561544e-01 7.13381395e-02 7.60374665e-01 -6.96578920e-01
2.90496171e-01 -7.57133543e-01 4.97785509e-01 1.04697859e+00
-2.64541715e-01 1.11392117e+00 5.58893979e-01 1.89914018e-01
-3.69130850e-01 -2.99842149e-01 7.83608913e-01 -3.55509281e-01
-7.95417726e-01 2.45588794e-01 -7.21307993e-01 -2.64732152e-01
1.12962365e+00 -2.95683593e-01 -6.97206706e-03 -6.81224287e-01
-7.08209813e-01 -3.09361108e-02 3.50817367e-02 4.93415236e-01
4.58546728e-01 -1.29523110e+00 -6.17058575e-01 5.37218630e-01
-5.47190309e-01 -6.69605136e-01 -2.44206324e-01 5.66607416e-01
-1.10002148e+00 9.18092906e-01 -3.61818105e-01 -1.97249353e-01
-1.08409834e+00 4.59296972e-01 6.86883807e-01 -1.93804339e-01
-6.72475994e-01 9.08574045e-01 -2.49595001e-01 -2.65896112e-01
4.77533162e-01 -5.55673122e-01 -7.28579983e-02 6.27498850e-02
7.21084595e-01 4.91829485e-01 2.97331303e-01 -3.03685397e-01
-3.29893231e-01 1.35938540e-01 -1.62483007e-01 -5.29805958e-01
1.68893206e+00 4.72625613e-01 -3.02049033e-02 4.54277277e-01
1.01266921e+00 -6.16031766e-01 -1.17697656e+00 -1.05349220e-01
5.67706048e-01 -8.01905692e-02 2.92329729e-01 -8.55020821e-01
-1.55997694e+00 8.36283863e-01 8.03692758e-01 6.01502419e-01
1.10499060e+00 -5.06296337e-01 9.36305940e-01 5.25677621e-01
5.05705141e-02 -1.25973427e+00 -3.70289326e-01 6.53538704e-01
7.20402777e-01 -7.88521767e-01 -2.33182043e-01 4.18498456e-01
-6.02748990e-01 1.55876911e+00 7.43804574e-02 -2.97595143e-01
8.74731779e-01 5.80472767e-01 3.45418364e-01 3.65222059e-02
-8.97215664e-01 2.04333305e-01 -2.21834093e-01 2.74224430e-01
3.52789223e-01 1.02444358e-01 -3.91924083e-02 4.75183725e-01
-6.00782037e-01 1.37082845e-01 3.30587476e-01 1.03533328e+00
-2.16777503e-01 -1.01674068e+00 -2.51300216e-01 4.29521352e-01
-9.90519702e-01 -2.64596343e-01 -3.80745769e-01 1.03565240e+00
-2.66101480e-01 6.97077990e-01 5.24803221e-01 -4.82950270e-01
2.86155999e-01 5.85663356e-02 8.14101025e-02 -2.78379470e-01
-1.23628712e+00 1.76524445e-01 1.65069103e-01 -3.46297741e-01
-4.29661572e-02 -9.32347119e-01 -1.07164609e+00 -8.10654521e-01
3.35777104e-02 2.64930576e-01 7.15734601e-01 5.09115696e-01
2.96815246e-01 3.49891812e-01 5.45414865e-01 -8.84579301e-01
-1.08519936e+00 -1.08700931e+00 -8.55658174e-01 -1.27490357e-01
3.47039044e-01 -3.49838734e-01 -3.18481892e-01 -2.03881279e-01] | [4.451401710510254, 4.189846038818359] |
354e8f07-1114-46e6-9c1f-f42ec701a2e1 | the-mucow-word-sense-disambiguation-test | null | null | https://aclanthology.org/2020.wmt-1.40 | https://aclanthology.org/2020.wmt-1.40.pdf | The MUCOW word sense disambiguation test suite at WMT 2020 | This paper reports on our participation with the MUCOW test suite at the WMT 2020 news translation task. We introduced MUCOW at WMT 2019 to measure the ability of MT systems to perform word sense disambiguation (WSD), i.e., to translate an ambiguous word with its correct sense. MUCOW is created automatically using existing resources, and the evaluation process is also entirely automated. We evaluate all participating systems of the language pairs English -> Czech, English -> German, and English -> Russian and compare the results with those obtained at WMT 2019. While current NMT systems are fairly good at handling ambiguous source words, we could not identify any substantial progress - at least to the extent that it is measurable by the MUCOW method - in that area over the last year. | ['Jörg Tiedemann', 'Alessandro Raganato', 'Yves Scherrer'] | null | null | null | null | wmt-emnlp-2020-11 | ['word-sense-disambiguation'] | ['natural-language-processing'] | [ 2.33156934e-01 2.48477757e-01 -2.65192270e-01 -1.28242880e-01
-1.07315445e+00 -9.79151964e-01 1.13001263e+00 3.66415977e-01
-8.16175640e-01 1.32342875e+00 5.69531739e-01 -7.16508687e-01
-1.57209337e-02 -4.44860399e-01 -3.83706391e-01 4.17184122e-02
3.39491457e-01 1.04064202e+00 2.34783754e-01 -7.73661852e-01
3.56184870e-01 6.52947277e-02 -8.85497332e-01 3.51423502e-01
1.19749594e+00 1.92467809e-01 2.64923453e-01 5.30225277e-01
-4.44325000e-01 2.16208771e-01 -7.51104057e-01 -6.74490988e-01
2.64201969e-01 -6.26353979e-01 -1.38339615e+00 -4.25351471e-01
3.03582340e-01 4.95751709e-01 1.72805250e-01 1.15988362e+00
5.09047568e-01 -4.27382812e-02 5.77619314e-01 -8.78548384e-01
-7.09731042e-01 9.64611948e-01 -4.09091003e-02 6.02446020e-01
8.67127776e-01 -8.89411345e-02 1.13625264e+00 -1.19415474e+00
1.24510610e+00 1.12311006e+00 4.34868604e-01 5.00261962e-01
-1.01083720e+00 -4.15203780e-01 1.76011547e-02 1.42559215e-01
-1.41241419e+00 -4.98004168e-01 -8.74288231e-02 -1.92314610e-01
1.63247907e+00 5.00844121e-01 4.83529240e-01 9.74750221e-01
4.33751523e-01 5.60952961e-01 1.40930378e+00 -8.47335994e-01
-4.70560789e-02 1.70488313e-01 -4.36306745e-02 2.66646296e-01
5.15600979e-01 -7.05155209e-02 -7.62908638e-01 2.02670563e-02
2.48023644e-01 -1.08432901e+00 -1.87877730e-01 2.89590120e-01
-2.04180861e+00 2.57177204e-01 -2.33503848e-01 9.21931744e-01
-3.31299573e-01 -4.72109199e-01 3.49999517e-01 9.17084277e-01
4.85994905e-01 1.14201438e+00 -8.32834065e-01 -4.92612183e-01
-1.01584816e+00 3.31846118e-01 9.21613753e-01 1.24815106e+00
2.99482197e-01 -1.16809271e-01 -1.48556262e-01 9.13576901e-01
-2.92474106e-02 7.38695383e-01 9.27387357e-01 -7.10015476e-01
7.98521399e-01 3.72054130e-01 5.39973915e-01 -6.57875121e-01
-1.40740454e-01 -3.14452142e-01 -2.34083205e-01 -1.44053057e-01
3.03711683e-01 -3.75017047e-01 -8.83241773e-01 1.77896857e+00
-1.77192897e-01 -3.37530047e-01 8.23784530e-01 6.66394413e-01
6.07848585e-01 7.40615189e-01 1.25878543e-01 -5.23703456e-01
1.26871991e+00 -7.55266845e-01 -7.67767727e-01 -6.03709877e-01
7.55278230e-01 -1.55404055e+00 1.08356690e+00 4.81839702e-02
-1.33212900e+00 -4.35484737e-01 -1.18223071e+00 2.20191538e-01
-6.46967053e-01 -1.62732810e-01 2.00475514e-01 4.89160001e-01
-1.17704475e+00 3.08601916e-01 -4.54914927e-01 -8.80920768e-01
-6.69385493e-01 5.37320152e-02 -4.89841044e-01 -1.17200643e-01
-1.82867253e+00 1.56978703e+00 5.99947989e-01 -3.20987761e-01
-1.09897844e-01 -3.90186995e-01 -7.25269020e-01 -4.03536886e-01
1.63707495e-01 -6.82676554e-01 1.53376436e+00 -8.12989771e-01
-1.05313730e+00 9.85503495e-01 -4.18207407e-01 -5.39493442e-01
6.86694026e-01 -3.06502134e-02 -1.19926786e+00 -3.11144948e-01
7.54899383e-01 7.36354411e-01 1.38247713e-01 -1.00472641e+00
-1.07159424e+00 -1.45881534e-01 -6.14371300e-02 6.89294457e-01
5.37729487e-02 4.57419306e-01 -1.31900266e-01 -9.19511020e-01
2.22374991e-01 -1.24736035e+00 -1.26623418e-02 -9.13996935e-01
-3.36982787e-01 -2.22274274e-01 2.95014173e-01 -8.40599775e-01
1.11732078e+00 -1.74432671e+00 -1.41323043e-03 8.28893557e-02
-3.62343282e-01 3.70659918e-01 -3.26594740e-01 7.90972173e-01
-1.79125458e-01 5.39759278e-01 -2.03537330e-01 1.18819915e-01
7.89686367e-02 3.02895159e-01 -3.08495015e-01 -2.64420033e-01
2.52118945e-01 9.43244278e-01 -1.19697201e+00 -4.59189624e-01
-6.76940903e-02 -1.13775268e-01 1.40929654e-01 -3.44500333e-01
-3.14113200e-01 1.15058988e-01 -1.90859303e-01 5.08829713e-01
2.62730896e-01 3.01406801e-01 5.07397771e-01 1.92776784e-01
-5.29299736e-01 9.24308896e-01 -1.00853109e+00 1.53047252e+00
-5.72728336e-01 7.25661457e-01 -3.48595083e-01 -4.32435930e-01
7.94543326e-01 7.89889574e-01 1.53656289e-01 -9.42309022e-01
-3.54777873e-01 9.41510141e-01 2.25706577e-01 -2.77711391e-01
1.20393276e+00 -1.93893045e-01 -3.43778551e-01 1.99467734e-01
1.13909826e-01 -2.86790103e-01 7.09275961e-01 1.01978378e-02
9.57225084e-01 -3.07187811e-02 4.01685596e-01 -8.00821781e-01
3.27876210e-01 7.26174533e-01 4.51507658e-01 5.79733312e-01
1.02270357e-01 5.20938396e-01 -2.61050323e-03 -5.34951948e-02
-1.07876348e+00 -1.14013970e+00 -1.39825940e-01 9.15073633e-01
2.68411487e-01 -5.57659924e-01 -7.03559160e-01 -5.46577573e-01
-3.48262995e-01 1.27512193e+00 -2.15010762e-01 1.04190111e-01
-6.11413419e-01 -7.53246009e-01 8.75292838e-01 2.51052171e-01
4.55516398e-01 -1.06549346e+00 -2.56410569e-01 5.17224431e-01
-7.76811540e-01 -1.37723482e+00 -6.79260731e-01 -1.76004954e-02
-4.53539193e-01 -6.73516810e-01 -8.11443090e-01 -1.06315720e+00
8.06482881e-02 1.88120469e-01 1.31356025e+00 -3.04543555e-01
1.35267124e-01 1.07718445e-01 -5.91780484e-01 -5.14713824e-01
-8.07726741e-01 4.33130205e-01 3.93323511e-01 -6.26827300e-01
6.37608171e-01 -1.68882713e-01 -8.48942176e-02 5.57463944e-01
-8.25449586e-01 -5.33142164e-02 5.45113623e-01 5.23791373e-01
4.45696443e-01 -2.38164321e-01 8.64008546e-01 -7.03204989e-01
1.00418139e+00 -3.02638084e-01 -2.44075134e-01 5.70047319e-01
-1.05313683e+00 1.38143882e-01 2.02422410e-01 -1.64125904e-01
-9.51073408e-01 -3.02076101e-01 -2.77912021e-01 5.55288672e-01
-9.52785164e-02 8.54480803e-01 -1.40694246e-01 1.18635543e-01
6.55752242e-01 1.80542037e-01 -5.54867506e-01 -3.38742644e-01
3.41515452e-01 8.10805976e-01 7.81161785e-01 -6.50291741e-01
8.13696623e-01 -3.76965314e-01 -5.30269384e-01 -4.95468885e-01
-4.96990353e-01 -4.49903458e-01 -5.91085315e-01 2.02474240e-02
8.28544915e-01 -8.98207128e-01 2.96772599e-01 2.08025828e-01
-1.31077075e+00 -1.76463097e-01 -1.70773283e-01 7.13137090e-01
-3.42024833e-01 2.54929602e-01 -4.50074255e-01 -2.79797435e-01
-4.48563904e-01 -1.09246540e+00 7.54295588e-01 -5.82540818e-02
-1.04065979e+00 -1.02544010e+00 1.95657104e-01 3.43793631e-01
3.66203129e-01 -7.60570467e-02 1.19331276e+00 -9.60563183e-01
5.89272846e-03 -1.81239247e-01 -3.86877842e-02 5.71240932e-02
2.27599263e-01 -1.24902062e-01 -3.34454268e-01 -1.67420581e-01
-4.06749487e-01 -1.42624718e-03 4.11269397e-01 -2.07664996e-01
-1.92237779e-01 -3.31757426e-01 -2.57968187e-01 -3.07146579e-01
1.22863543e+00 3.71169955e-01 5.34007490e-01 8.76385629e-01
1.68148965e-01 4.41701859e-01 1.07374763e+00 -2.79435664e-01
5.35557270e-01 8.45260620e-01 -3.43208730e-01 1.94442376e-01
-4.09052879e-01 -3.09319347e-01 4.50982243e-01 1.29640782e+00
-3.76382656e-03 -5.68460286e-01 -1.38891864e+00 8.58473063e-01
-1.64192820e+00 -7.43933797e-01 -1.30214736e-01 2.10903621e+00
1.15365636e+00 5.63771009e-01 -1.26286268e-01 -5.80735095e-02
7.79917717e-01 -1.98705401e-02 2.01510742e-01 -7.86540568e-01
-5.78700423e-01 3.90627533e-01 5.48481524e-01 1.08753085e+00
-5.57979822e-01 1.47140563e+00 7.04254389e+00 7.19605386e-01
-9.54352140e-01 1.16968706e-01 1.43006265e-01 1.99004918e-01
-5.51121473e-01 1.42792672e-01 -7.14101195e-01 4.01535898e-01
1.03812063e+00 -8.32358837e-01 3.11073303e-01 3.01690102e-01
3.80507223e-02 -4.10732301e-03 -7.59938896e-01 4.53346819e-01
-3.85811552e-02 -1.21388543e+00 3.59855413e-01 -7.47514516e-02
9.09447730e-01 3.37400198e-01 -2.59475768e-01 3.96549374e-01
6.49754405e-01 -8.53254557e-01 9.95983422e-01 3.24130893e-01
8.70530188e-01 -6.55044436e-01 1.04636884e+00 2.81203210e-01
-8.84618104e-01 5.58923721e-01 -3.64667922e-01 8.12090412e-02
2.50700593e-01 5.06238163e-01 -1.18781531e+00 9.58525896e-01
2.96926945e-01 3.08625311e-01 -6.48618102e-01 8.43687236e-01
-2.92666316e-01 6.60481393e-01 -1.36608183e-01 -2.55531013e-01
4.72434849e-01 -2.14521028e-02 1.14446461e+00 1.69900036e+00
6.69470549e-01 -7.68237039e-02 1.95485294e-01 1.33485332e-01
-9.73674804e-02 4.88888025e-01 -5.66401303e-01 -1.71797276e-01
7.12319076e-01 7.48660207e-01 -6.01389050e-01 -5.32940030e-01
-1.69811904e-01 1.21224236e+00 -1.25978053e-01 2.06961617e-01
-3.56610179e-01 -4.97782290e-01 7.52966940e-01 -6.65647909e-02
-7.39752576e-02 -4.88480985e-01 -4.34706181e-01 -1.05531347e+00
3.04721475e-01 -1.11489451e+00 2.61536509e-01 -1.03450632e+00
-1.07008016e+00 1.27081442e+00 4.05700430e-02 -1.23626888e+00
-7.11991489e-01 -4.11165088e-01 -1.76656902e-01 1.12241137e+00
-1.23150873e+00 -7.42798388e-01 4.83551204e-01 2.62324929e-01
6.73095882e-01 -3.58891696e-01 1.09346545e+00 1.88205644e-01
-2.56155021e-02 6.54846847e-01 1.71188310e-01 1.30535930e-01
1.12947941e+00 -1.45066226e+00 1.06561780e+00 1.16160917e+00
4.80195105e-01 7.83600688e-01 1.18097782e+00 -1.04335046e+00
-7.80924797e-01 -8.95249546e-01 2.01075983e+00 -7.78233707e-01
9.49909627e-01 -3.63840396e-03 -6.07374191e-01 4.29863483e-01
7.90597320e-01 -6.41274393e-01 5.06297886e-01 1.77785948e-01
-2.97702730e-01 3.15346360e-01 -9.87174809e-01 7.81942785e-01
1.12980437e+00 -4.93080974e-01 -1.32922959e+00 3.98698151e-01
9.43514049e-01 -5.03685653e-01 -1.02031994e+00 5.33140540e-01
3.63361865e-01 -1.97987959e-01 6.57992721e-01 -8.45373511e-01
2.73846626e-01 -4.88574117e-01 -5.38999021e-01 -1.95561922e+00
-2.27897987e-01 -6.48863435e-01 5.99013209e-01 1.23526633e+00
1.28953731e+00 -6.93073153e-01 1.68826252e-01 2.23517492e-01
-3.81229103e-01 -1.91403806e-01 -1.27607870e+00 -1.22327542e+00
3.65759879e-01 -4.58511174e-01 8.57141614e-01 1.43954003e+00
4.42099422e-01 7.03744709e-01 3.49210352e-02 9.61035267e-02
1.33764654e-01 -1.69341207e-01 2.04641312e-01 -1.01910937e+00
-4.90530767e-02 -5.56904316e-01 -4.76325721e-01 -5.18028915e-01
1.06795691e-01 -1.03183603e+00 1.41836926e-01 -1.90352404e+00
-1.35215551e-01 -1.16484813e-01 -3.24366093e-01 6.19108260e-01
-4.06559348e-01 3.80762249e-01 1.58622921e-01 3.49894136e-01
-3.85575086e-01 -2.49442644e-02 1.02319682e+00 -2.38023937e-01
4.11925884e-03 -1.53320238e-01 -8.64894629e-01 3.01385492e-01
9.23775554e-01 -5.07753372e-01 -2.48359352e-01 -7.88871765e-01
2.34728187e-01 1.90062243e-02 -8.30834955e-02 -1.03866613e+00
9.21096429e-02 -4.04249340e-01 1.85982883e-01 -1.13144875e-01
3.93641591e-02 -4.41269010e-01 1.91586196e-01 3.10330242e-01
-4.92860854e-01 9.28305686e-01 3.92500788e-01 -1.65475048e-02
-3.78441304e-01 -2.86208689e-01 4.45727378e-01 -2.33130395e-01
-9.16615009e-01 -1.72650516e-01 -8.87901425e-01 4.65965271e-01
5.40103853e-01 -6.95756152e-02 -4.11987543e-01 -3.08526307e-01
-6.87097013e-01 1.65200040e-01 6.54404998e-01 9.66880083e-01
1.53053671e-01 -1.37828934e+00 -1.11403608e+00 -2.13421941e-01
6.25674188e-01 -7.62515068e-01 -5.88642061e-01 5.68397284e-01
-3.31659585e-01 7.36954391e-01 -2.73516744e-01 8.47273879e-03
-1.22239554e+00 1.57319397e-01 3.03138047e-01 -3.89471859e-01
-2.33163118e-01 4.09477085e-01 -6.67065561e-01 -5.81269860e-01
-1.80196732e-01 -1.12794347e-01 -6.37896731e-02 -6.41471371e-02
4.53618020e-01 6.60227016e-02 6.49214029e-01 -8.52606595e-01
-4.29134011e-01 2.64963806e-01 -1.76891685e-01 -9.49618161e-01
7.49602377e-01 -3.12413424e-01 -1.62769288e-01 4.62314457e-01
7.02259183e-01 4.24593687e-01 1.18334463e-03 -2.65807629e-01
7.01057315e-01 -2.56740123e-01 -2.44609669e-01 -1.58842003e+00
-1.98429570e-01 3.32025230e-01 4.24369633e-01 2.63930261e-01
5.89355767e-01 -1.46617919e-01 9.67007160e-01 6.41485214e-01
6.49087906e-01 -1.49500597e+00 -6.69592679e-01 1.10605800e+00
7.58201241e-01 -1.00353861e+00 -5.09147525e-01 -3.25157106e-01
-7.28720427e-01 8.28903735e-01 3.80324811e-01 4.14421082e-01
1.29414201e-01 6.45303801e-02 5.30780733e-01 5.88816665e-02
-7.19007909e-01 -3.15691948e-01 4.37021106e-01 5.86433768e-01
8.15940201e-01 3.50004047e-01 -1.02879190e+00 2.47924730e-01
-7.52839804e-01 -1.82061076e-01 5.90513229e-01 9.59134758e-01
-6.81000173e-01 -1.50162530e+00 -1.56370282e-01 3.05670738e-01
-4.95738477e-01 -6.21520281e-01 -1.04678488e+00 1.01653016e+00
-2.99341530e-02 1.30303037e+00 -3.24950010e-01 -6.59658790e-01
5.55735230e-01 4.62278515e-01 4.33926076e-01 -7.65120745e-01
-4.72096413e-01 -9.79612544e-02 8.93837988e-01 -9.61312875e-02
-3.70775819e-01 -9.42983210e-01 -1.10144114e+00 -2.86608785e-01
-3.18024978e-02 8.93493176e-01 5.95531940e-01 1.33788514e+00
2.80158725e-02 1.45408839e-01 3.91762376e-01 -1.01178676e-01
-2.05627993e-01 -1.17550993e+00 -1.24815024e-01 3.98766100e-02
-1.11197829e-01 -2.72541791e-01 5.65927848e-03 -4.66103144e-02] | [11.44973087310791, 10.256850242614746] |
3213f0ab-4ad6-4bdc-956a-56cdf3aa283c | mobilevos-real-time-video-object-segmentation | 2303.07815 | null | https://arxiv.org/abs/2303.07815v1 | https://arxiv.org/pdf/2303.07815v1.pdf | MobileVOS: Real-Time Video Object Segmentation Contrastive Learning meets Knowledge Distillation | This paper tackles the problem of semi-supervised video object segmentation on resource-constrained devices, such as mobile phones. We formulate this problem as a distillation task, whereby we demonstrate that small space-time-memory networks with finite memory can achieve competitive results with state of the art, but at a fraction of the computational cost (32 milliseconds per frame on a Samsung Galaxy S22). Specifically, we provide a theoretically grounded framework that unifies knowledge distillation with supervised contrastive representation learning. These models are able to jointly benefit from both pixel-wise contrastive learning and distillation from a pre-trained teacher. We validate this loss by achieving competitive J&F to state of the art on both the standard DAVIS and YouTube benchmarks, despite running up to 5x faster, and with 32x fewer parameters. | ['Albert Saa-Garriga', 'Bruno Manganelli', 'Mehmet Kerim Yucel', 'Roy Miles'] | 2023-03-14 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Miles_MobileVOS_Real-Time_Video_Object_Segmentation_Contrastive_Learning_Meets_Knowledge_Distillation_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Miles_MobileVOS_Real-Time_Video_Object_Segmentation_Contrastive_Learning_Meets_Knowledge_Distillation_CVPR_2023_paper.pdf | cvpr-2023-1 | ['semi-supervised-video-object-segmentation', 'video-object-segmentation', 'video-semantic-segmentation'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 5.36654830e-01 4.00881190e-03 -6.80949092e-01 -1.36045188e-01
-9.58427429e-01 -6.46020830e-01 4.97984499e-01 -2.86477238e-01
-9.71891165e-01 6.41920924e-01 -2.64136016e-01 -7.50392497e-01
2.55219340e-01 -4.10201043e-01 -1.18095064e+00 -5.12422502e-01
-3.09376828e-02 4.17534262e-01 3.97764295e-01 4.25003499e-01
-2.29721610e-02 3.89017284e-01 -1.48449481e+00 4.17499274e-01
6.75815046e-01 1.25912070e+00 3.60419989e-01 1.29632902e+00
5.85411154e-02 1.14250398e+00 -3.38184237e-01 -2.81425446e-01
4.52064633e-01 -1.26163974e-01 -1.31819582e+00 2.32160211e-01
9.21880662e-01 -6.48201227e-01 -6.95830226e-01 8.54686141e-01
3.01911771e-01 1.63594455e-01 3.57608348e-01 -1.05561566e+00
-3.80129427e-01 5.98293483e-01 -6.45968497e-01 7.39030778e-01
-2.55036950e-01 3.76916051e-01 8.76947641e-01 -7.51631260e-01
4.96229053e-01 1.09535730e+00 6.41603768e-01 5.63138723e-01
-1.09862125e+00 -5.14349520e-01 4.24687088e-01 1.63144603e-01
-1.37482524e+00 -6.28375530e-01 2.19863445e-01 -2.37828895e-01
1.44873095e+00 1.36067301e-01 4.63213980e-01 7.41929531e-01
-3.74588192e-01 1.28570211e+00 9.47184026e-01 -2.31835693e-01
8.37359503e-02 2.59491382e-03 1.22397378e-01 9.57573473e-01
2.50014812e-01 -2.19152406e-01 -4.25047219e-01 2.54116088e-01
9.90130246e-01 -1.22451186e-02 -2.03664109e-01 -1.07590057e-01
-9.57565129e-01 5.51209092e-01 4.15078908e-01 7.10031390e-02
-1.04678035e-01 7.57124364e-01 3.96976352e-01 1.01646677e-01
5.46626270e-01 2.37441882e-01 -8.57796609e-01 -5.46397567e-01
-1.46634638e+00 -2.46236995e-01 7.74440587e-01 1.21512949e+00
7.06616938e-01 3.27059627e-01 9.88493189e-02 4.16713983e-01
2.70495504e-01 6.36732578e-01 6.44244909e-01 -1.43123448e+00
6.18636906e-01 3.63287814e-02 4.36785817e-02 -4.23971146e-01
-1.95644289e-01 -3.07381958e-01 -6.13450885e-01 9.46332365e-02
5.88694334e-01 -3.92877400e-01 -1.28119719e+00 1.46613503e+00
1.47345483e-01 9.33556497e-01 -6.22458570e-02 7.61192858e-01
7.17458665e-01 8.79229188e-01 -4.33801562e-02 -2.38047928e-01
1.26124799e+00 -1.71200383e+00 -2.06025705e-01 -5.15239656e-01
8.09688151e-01 -2.97241509e-01 1.13516045e+00 6.14762127e-01
-1.65566969e+00 -8.04104805e-01 -1.29034078e+00 -4.21387255e-01
-2.74993092e-01 -6.71866089e-02 8.49225700e-01 9.92622733e-01
-1.33630693e+00 9.33601201e-01 -1.33308220e+00 -2.86737792e-02
1.07659853e+00 5.99228084e-01 1.15471091e-02 -7.35643357e-02
-5.32406747e-01 3.69363576e-01 4.91155744e-01 -1.99393332e-01
-1.05650842e+00 -9.50716615e-01 -6.60639584e-01 1.33789912e-01
7.52841890e-01 -6.17984235e-01 1.50168169e+00 -1.26565814e+00
-1.56882071e+00 1.01559436e+00 -2.95713786e-02 -1.04960167e+00
5.43024540e-01 -6.99021101e-01 -1.10834964e-01 6.06098354e-01
-2.64234811e-01 1.11182880e+00 1.00154793e+00 -1.00315046e+00
-1.00857747e+00 -1.45860508e-01 3.17466170e-01 2.11250484e-01
-3.98083448e-01 -1.81751847e-01 -1.17442238e+00 -4.98940080e-01
-3.51065069e-01 -1.09259427e+00 -1.96297884e-01 1.51047125e-01
-2.42002085e-01 1.04748636e-01 1.00365674e+00 -6.99190855e-01
9.33067679e-01 -2.15166092e+00 -6.38627559e-02 -1.07188970e-01
3.26523453e-01 7.57729650e-01 -3.14900517e-01 -6.52388215e-01
1.21114865e-01 3.11140716e-01 -7.48383701e-02 -6.90265417e-01
-1.57809168e-01 3.46386969e-01 -3.21286768e-01 4.46518958e-01
1.64823636e-01 1.24269116e+00 -8.74084115e-01 -5.23064435e-01
1.72091261e-01 5.01604080e-01 -7.08008409e-01 -6.97912946e-02
-5.01247346e-01 1.35370433e-01 5.81934303e-03 6.11656606e-01
6.71588659e-01 -7.46533573e-01 4.35802460e-01 -4.36516777e-02
1.27940133e-01 3.63852531e-01 -1.01114368e+00 1.99685061e+00
-5.99838376e-01 1.02891159e+00 2.76218534e-01 -1.06196690e+00
1.66514888e-01 6.66855574e-02 3.48953903e-01 -7.13942349e-01
2.37581924e-01 2.92673320e-01 -3.71804655e-01 -1.51187748e-01
6.66117132e-01 3.60261053e-01 3.84285331e-01 5.31077206e-01
2.29806319e-01 -1.08627595e-01 2.17550948e-01 2.31720030e-01
1.09638762e+00 3.83827031e-01 -2.06304900e-02 -1.23358451e-01
-4.95651588e-02 -2.03342259e-01 3.51294219e-01 1.00643063e+00
-1.97438821e-01 5.28280854e-01 4.52591389e-01 -4.04485583e-01
-1.07513869e+00 -1.13192296e+00 1.75530165e-01 1.44674671e+00
2.41496101e-01 -2.04870224e-01 -9.42151010e-01 -1.00594616e+00
-2.47934297e-01 2.66713172e-01 -3.55016679e-01 2.12260574e-01
-7.54191399e-01 -6.57129645e-01 7.63046443e-01 1.08948088e+00
9.21716094e-01 -7.59143591e-01 -9.00551021e-01 1.49624385e-02
2.71943271e-01 -1.55903482e+00 -4.50693160e-01 3.77220333e-01
-1.25341070e+00 -9.33167636e-01 -8.59536111e-01 -9.07938063e-01
4.73704368e-01 5.78455567e-01 1.55416727e+00 3.38880569e-01
-3.92389655e-01 5.24266720e-01 -3.10077779e-02 -2.76204981e-02
6.58380836e-02 4.50156987e-01 -2.12367430e-01 -4.69628096e-01
1.94435850e-01 -5.08955240e-01 -8.52462590e-01 1.62276730e-01
-1.04572999e+00 1.99248418e-01 5.89321613e-01 7.23495185e-01
8.79212320e-01 1.58593446e-01 2.09661841e-01 -1.14077175e+00
-5.41600473e-02 -3.74887854e-01 -8.07864249e-01 1.26210172e-02
-7.14258552e-01 6.72815815e-02 4.82579172e-01 -5.50893486e-01
-9.34472024e-01 2.45394930e-01 -7.36684799e-02 -5.34576237e-01
-1.56484142e-01 -4.62483522e-03 -7.36424550e-02 -2.53966451e-01
3.60140413e-01 9.17364806e-02 -1.73563153e-01 -2.87712723e-01
7.76989043e-01 4.77891445e-01 8.15112650e-01 -5.92461944e-01
4.45724398e-01 7.20193207e-01 -1.82591721e-01 -7.70587862e-01
-8.39168012e-01 -4.15809661e-01 -5.55432439e-01 2.76527613e-01
8.68990719e-01 -1.30818582e+00 -7.07247198e-01 5.55989265e-01
-9.56712782e-01 -1.05623996e+00 -5.50676048e-01 2.89344221e-01
-7.54818022e-01 4.11487669e-01 -9.63096857e-01 -4.55986142e-01
-3.68756592e-01 -1.05596173e+00 8.69373083e-01 5.40631890e-01
2.23727912e-01 -9.58980560e-01 -4.04029757e-01 6.32158697e-01
4.12727803e-01 -1.07033506e-01 6.43324792e-01 -5.32759845e-01
-1.07548857e+00 1.83733165e-01 -8.07986975e-01 5.52590966e-01
-3.23255897e-01 -1.56820297e-01 -1.16762578e+00 -3.37876916e-01
-6.85646236e-02 -5.47076166e-01 1.46839130e+00 5.64594626e-01
1.58176541e+00 -4.19279665e-01 -3.49708736e-01 1.01332629e+00
1.48320317e+00 9.35378820e-02 7.30122745e-01 2.28853807e-01
1.07669115e+00 -3.54399569e-02 3.60949039e-01 2.86312670e-01
2.64545918e-01 4.36181098e-01 4.53362584e-01 -1.94315895e-01
-5.50626159e-01 3.28255049e-03 1.02807380e-01 4.98857498e-01
-1.91820964e-01 -4.08730537e-01 -7.44403780e-01 7.35073864e-01
-1.88494027e+00 -6.27685726e-01 1.42565355e-01 2.18024659e+00
9.32302237e-01 5.07376134e-01 3.69797349e-01 1.28302544e-01
4.49230194e-01 4.78301346e-01 -8.14394772e-01 -2.93285012e-01
-1.03122644e-01 5.83154738e-01 9.59938407e-01 5.75767338e-01
-1.32713282e+00 1.15402508e+00 6.83561611e+00 1.06136823e+00
-1.09464073e+00 4.07242477e-01 1.21297956e+00 -4.90687162e-01
-7.39857205e-04 -3.36708367e-01 -6.98015392e-01 4.71532553e-01
1.31237900e+00 3.67066741e-01 8.16136479e-01 1.07277095e+00
-3.15385938e-01 -2.15531245e-01 -1.22122657e+00 1.30020571e+00
6.84096888e-02 -1.62890589e+00 -1.27268687e-01 1.75195813e-01
1.19138443e+00 5.22768617e-01 4.49276775e-01 3.99751276e-01
2.19086632e-01 -1.24168670e+00 4.10762966e-01 1.87921882e-01
1.04017675e+00 -7.08700001e-01 5.32540679e-01 3.47849950e-02
-1.15565205e+00 -3.66111659e-02 -2.08849400e-01 -2.31370345e-01
1.10919088e-01 4.01097327e-01 -7.67040014e-01 7.93070197e-02
7.57198751e-01 6.14950418e-01 -4.49479222e-01 9.85190809e-01
-5.86637445e-02 7.65986800e-01 -4.16349947e-01 2.24027276e-01
3.99029136e-01 3.33052069e-01 4.55342792e-02 1.51511312e+00
1.67343467e-01 -2.33104452e-02 1.96111128e-01 4.83791351e-01
-8.40312243e-01 -3.10350537e-01 -1.49067998e-01 -3.08493108e-01
5.03468871e-01 1.02582502e+00 -1.19035339e+00 -8.37381244e-01
-5.30710697e-01 1.42802143e+00 2.31357530e-01 4.19928581e-01
-1.10689211e+00 -2.30422169e-01 5.32366812e-01 1.55097485e-01
1.01991427e+00 -3.92454982e-01 -4.94508743e-01 -9.64447439e-01
-1.80873409e-01 -6.94779813e-01 1.29005671e-01 -6.44121945e-01
-7.54982531e-01 3.87702942e-01 -1.98462203e-01 -7.48979986e-01
-2.20108390e-01 -1.04869235e+00 -2.94126779e-01 3.69750381e-01
-1.91399372e+00 -9.57623839e-01 -2.66837329e-01 5.10579824e-01
7.90365338e-01 -1.10446839e-02 6.03324950e-01 5.30321896e-01
-5.53359985e-01 6.94724143e-01 2.35401213e-01 1.80737078e-01
1.50119558e-01 -1.47282600e+00 8.01398158e-01 8.68594706e-01
5.39804697e-01 1.50823310e-01 2.84261674e-01 -3.20161074e-01
-1.49999535e+00 -1.19268751e+00 4.22516584e-01 -4.20063287e-01
6.49690509e-01 -1.12017795e-01 -6.75733864e-01 8.22558045e-01
2.38787889e-01 7.41644323e-01 5.46648264e-01 1.09663479e-01
-5.26092350e-01 7.74586946e-02 -1.04724121e+00 4.97441679e-01
1.26423860e+00 -7.10563958e-01 -2.07083851e-01 4.49754626e-01
9.51628864e-01 -6.74429357e-01 -6.06326461e-01 1.24729544e-01
4.81451124e-01 -7.33129084e-01 1.13730776e+00 -6.78274155e-01
3.72081637e-01 -1.17931589e-01 -1.74686328e-01 -6.74016953e-01
-5.10144681e-02 -1.01396179e+00 -8.00877392e-01 8.85840237e-01
3.81912261e-01 -1.66783482e-01 1.36085761e+00 4.57925409e-01
-2.92486306e-02 -8.40094090e-01 -1.01679671e+00 -1.01769638e+00
1.35898784e-01 -7.03386247e-01 2.03595445e-01 5.21106660e-01
-3.65899295e-01 4.10839081e-01 -5.04268885e-01 8.51910859e-02
3.28259945e-01 -1.01086952e-01 5.76246500e-01 -9.39082026e-01
-7.23693192e-01 -5.69681406e-01 -3.57214898e-01 -1.99738753e+00
1.88238814e-01 -5.46085954e-01 9.82329920e-02 -1.28578997e+00
3.61889541e-01 -3.71199101e-01 -3.42910558e-01 4.38917369e-01
-1.34886041e-01 7.96853065e-01 3.67947966e-01 3.18530649e-02
-1.41079593e+00 -2.54875813e-02 1.00664675e+00 -3.44132900e-01
-3.10858279e-01 -1.44939020e-01 -7.08317518e-01 9.89569545e-01
5.72997093e-01 -3.60069573e-01 -5.60343266e-01 -9.16425645e-01
1.72648385e-01 -1.61169514e-01 3.26517940e-01 -1.12436640e+00
1.77875221e-01 5.65255471e-02 2.98948258e-01 -3.84720206e-01
5.48572719e-01 -6.94524884e-01 -3.78288001e-01 4.53392118e-01
-2.24583954e-01 -1.91741988e-01 5.34525871e-01 6.30921543e-01
7.34154135e-02 -2.10022107e-01 7.77160406e-01 -1.26461402e-01
-1.08548439e+00 3.13447744e-01 -7.88512230e-02 4.34286118e-01
8.79046977e-01 -3.00762206e-01 -6.91710711e-01 -1.75825149e-01
-6.36679113e-01 2.66347881e-02 3.89951497e-01 1.40873389e-02
4.88452613e-01 -1.04289734e+00 -3.08317572e-01 8.16101283e-02
-5.36361337e-01 3.20019782e-01 4.12393868e-01 8.09588850e-01
-7.36063421e-01 6.01504445e-01 -8.51462111e-02 -6.12973392e-01
-1.26336706e+00 6.11916721e-01 1.86863437e-01 -3.47798407e-01
-4.82517809e-01 1.36087036e+00 4.21881787e-02 6.83122352e-02
5.63596606e-01 -7.53631592e-01 2.97052681e-01 -2.09399596e-01
7.78732300e-01 7.18975842e-01 -5.01413345e-02 -2.37256810e-01
-3.32511187e-01 3.50645065e-01 -2.46118724e-01 -9.82907340e-02
1.01822519e+00 -8.57666135e-02 4.71773714e-01 1.46044225e-01
1.37576532e+00 -1.08017333e-01 -2.05077124e+00 -3.52538705e-01
-1.74023002e-01 -3.92633051e-01 4.42574739e-01 -7.13747859e-01
-1.42566907e+00 8.68224859e-01 8.27304423e-01 1.36993915e-01
1.16782188e+00 2.55770069e-02 1.26215136e+00 6.90690160e-01
1.53841093e-01 -1.37411654e+00 3.02349299e-01 5.57370543e-01
-1.55979590e-02 -1.46415305e+00 9.59175676e-02 -4.11722481e-01
-4.96958882e-01 8.10058594e-01 4.79696572e-01 -3.16443920e-01
6.56929553e-01 5.88381231e-01 -2.62888253e-01 2.44390577e-01
-6.48833096e-01 -3.06853443e-01 2.16648325e-01 6.69321477e-01
2.45330185e-01 -1.42596308e-02 2.39656851e-01 4.95302022e-01
1.80979043e-01 2.32124746e-01 2.66669184e-01 1.01890755e+00
-5.93995810e-01 -6.49038076e-01 1.12284370e-01 4.56169754e-01
-7.78715074e-01 -4.33924347e-01 1.53463170e-01 6.97318792e-01
2.26598606e-01 8.52636397e-01 5.23039997e-01 -2.89380819e-01
-2.02648073e-01 -2.26842612e-01 6.69211209e-01 -5.88713706e-01
-3.54931295e-01 1.10301025e-01 1.40967324e-01 -7.57375598e-01
-6.86430752e-01 -2.95604676e-01 -1.43125260e+00 -5.04916906e-01
-1.92863092e-01 -2.77426362e-01 6.57351315e-01 1.17700648e+00
5.47923625e-01 7.23757923e-01 2.94281751e-01 -1.15726459e+00
-2.29480118e-01 -5.91667116e-01 -4.30736989e-01 -3.96182686e-02
3.33918452e-01 -2.03903511e-01 -1.89806178e-01 3.34702909e-01] | [9.210428237915039, 0.023442238569259644] |
3cd88e20-33b3-449d-84da-57f0c0775b3f | a-two-sub-problem-decomposition-for-the | 2101.01022 | null | https://arxiv.org/abs/2101.01022v1 | https://arxiv.org/pdf/2101.01022v1.pdf | A Two Sub-problem Decomposition for the Optimal Design of Filterless Optical Networks | Filterless optical transport networks relies on passive optical interconnections between nodes, i.e., on splitters/couplers and amplifiers. While different studies have investigated their design, none of them offer a solution for an optimal design. We propose a one step solution scheme, which combines network provisioning, i.e., routing and wavelength assignment within a single mathematical model. Decomposition into two different types sub-problems is then used in order to conceive an exact solution scheme. The first type of subproblem relies on the generation of filterless subnetworks while the second one takes care of their wavelength assignment. Numerical experiments demonstrate the improved performance of the proposed optimization model and algorithm over the state of the art, with the improvement of the solution for several open source data sets. | ['Yan Wang', 'Brigitte Jaumard'] | 2021-01-04 | null | null | null | null | ['problem-decomposition'] | ['miscellaneous'] | [ 2.42121041e-01 1.95005342e-01 3.43415700e-02 -1.26478493e-01
2.51988828e-01 -5.77216685e-01 4.97833975e-02 -2.46876761e-01
-3.44005138e-01 1.18345904e+00 -3.44345152e-01 -3.52189749e-01
-6.50970519e-01 -1.05670440e+00 -2.00381190e-01 -8.96625221e-01
2.09679961e-01 3.45202893e-01 3.01587492e-01 -1.11986853e-01
3.80013496e-01 7.60381699e-01 -1.35287666e+00 -4.43882436e-01
9.28765535e-01 1.18895304e+00 1.39800340e-01 3.25177431e-01
-4.46313173e-01 2.64456630e-01 -5.92019141e-01 -3.79118651e-01
7.60303557e-01 -5.36164284e-01 -7.20110953e-01 6.38008773e-01
6.32796958e-02 -1.03296317e-01 -1.73892736e-01 9.84958887e-01
1.87570512e-01 1.15838408e-01 3.65839034e-01 -1.57735372e+00
-5.53174973e-01 4.08867359e-01 -4.90968734e-01 -1.58388931e-02
-7.88373426e-02 3.32356364e-01 1.14717579e+00 -2.46757939e-01
7.20933199e-01 8.77952993e-01 2.26826131e-01 3.30425620e-01
-1.50487792e+00 -5.35146058e-01 -1.21402279e-01 2.10640833e-01
-1.28358293e+00 -6.22244596e-01 7.07146883e-01 -2.35581174e-01
7.27146864e-01 2.41820008e-01 9.93272543e-01 7.40668029e-02
2.74479073e-02 3.39775458e-02 1.18450010e+00 -4.11003560e-01
1.68829784e-01 4.45723623e-01 1.94076583e-01 7.62557328e-01
7.09845424e-01 -8.29334781e-02 -1.03078514e-01 1.32759854e-01
8.50841284e-01 -2.16983378e-01 -3.47759545e-01 -2.36304894e-01
-7.66021609e-01 5.27069569e-01 4.91483271e-01 4.29073751e-01
-4.33098376e-01 -1.03047326e-01 -1.79042593e-01 5.59366584e-01
4.25107151e-01 8.13457608e-01 -1.24508575e-01 4.39849645e-01
-1.01902997e+00 -1.00742720e-01 8.79389465e-01 1.17300546e+00
1.21094179e+00 -6.23745173e-02 -3.19151916e-02 7.44375169e-01
5.60033739e-01 2.76185066e-01 -2.25428447e-01 -8.15260470e-01
2.44046137e-01 8.07465792e-01 2.42700562e-01 -9.10511196e-01
-5.47153533e-01 -9.30272877e-01 -7.52670586e-01 3.73711556e-01
5.23423672e-01 -5.94816387e-01 -6.83475196e-01 1.50323606e+00
4.16283965e-01 2.92674750e-01 6.08121417e-02 9.10576880e-01
3.76773775e-01 8.67635667e-01 -2.80951202e-01 -7.44095981e-01
1.10295343e+00 -1.21775019e+00 -8.12557578e-01 1.49781749e-01
2.59280026e-01 -1.23761797e+00 1.45470053e-01 3.66775841e-01
-1.26267564e+00 -1.39089689e-01 -9.95982766e-01 1.55456468e-01
-2.05870822e-01 1.33365661e-01 5.70316315e-01 7.58869827e-01
-1.12034452e+00 5.76493800e-01 -3.04824114e-01 -4.86001760e-01
1.65819332e-01 6.05470657e-01 -2.53622085e-01 2.53669977e-01
-5.52296519e-01 5.41587293e-01 2.31511861e-01 3.02319616e-01
-2.23729149e-01 -8.33497465e-01 -8.43047649e-02 2.11972564e-01
6.42161191e-01 -1.07922709e+00 7.26575911e-01 -8.42330813e-01
-1.83138883e+00 6.14720821e-01 -1.47675797e-01 -4.45882469e-01
5.21899164e-01 3.30074042e-01 -3.17228675e-01 2.93116271e-01
-2.91470826e-01 3.06324989e-01 7.10994542e-01 -1.27952039e+00
-8.70025456e-01 9.04467888e-03 3.07564914e-01 -2.42059350e-01
-4.48258370e-01 9.75305587e-02 -3.03207606e-01 -1.08135298e-01
2.94425756e-01 -8.04392219e-01 -4.35020596e-01 2.48919159e-01
-9.53166902e-01 -1.94108948e-01 6.30717993e-01 -1.48686007e-01
1.24400747e+00 -1.75485349e+00 3.46704811e-01 4.51759100e-01
4.53172326e-01 3.28397512e-01 -2.17855856e-01 8.94883156e-01
-2.55920850e-02 1.00742184e-01 -2.12001558e-02 -6.06872141e-01
-4.58931364e-02 7.53877014e-02 -9.37449262e-02 5.32109380e-01
2.18069673e-01 2.42880762e-01 -5.38036764e-01 -3.82126510e-01
3.09659183e-01 2.88060904e-01 -7.43032634e-01 2.91728795e-01
-1.83241591e-01 4.73937511e-01 -4.99751389e-01 4.91153538e-01
1.02322066e+00 -1.86558619e-01 2.94550717e-01 -4.90544319e-01
-9.08433259e-01 7.85259008e-02 -1.57303846e+00 1.20623732e+00
-3.03198874e-01 3.07626814e-01 4.31485891e-01 -1.09408438e+00
1.04218471e+00 4.14773017e-01 9.03733552e-01 -4.13695574e-01
2.54908144e-01 4.96367335e-01 1.96867898e-01 -4.42313701e-01
3.47171217e-01 -4.33162510e-01 5.20505309e-01 3.88150454e-01
2.94261217e-01 1.01988882e-01 8.45729768e-01 -1.12208813e-01
1.12686658e+00 -1.90722570e-01 -6.68845698e-02 -3.11138362e-01
8.00568402e-01 -9.88261476e-02 7.97848880e-01 5.68732321e-01
6.66386485e-02 1.93646878e-01 6.31870389e-01 -1.08430304e-01
-9.99985695e-01 -9.62806046e-01 -5.93236163e-02 1.22525888e-02
4.54625040e-01 -1.06894121e-01 -6.55217886e-01 -1.12943955e-01
1.08979549e-02 2.58108050e-01 1.82120934e-01 2.66965508e-01
-3.39937985e-01 -8.16191196e-01 2.93016527e-02 -3.93259972e-01
5.91957927e-01 -6.22317493e-01 -6.47843480e-01 5.48015594e-01
1.79489627e-01 -1.35983241e+00 6.17396012e-02 -1.39155954e-01
-9.58805621e-01 -1.23217642e+00 -4.69265312e-01 -5.32621861e-01
8.25523317e-01 4.29744065e-01 8.71854186e-01 3.79718840e-01
-4.20224816e-01 2.58769244e-01 -2.87379891e-01 -3.49881724e-02
-1.97878584e-01 3.00853968e-01 3.56288776e-02 5.91097534e-01
-8.68230388e-02 -9.82154608e-01 -7.51991689e-01 3.96291941e-01
-7.08180189e-01 2.06955642e-01 9.37900424e-01 3.18986982e-01
4.40969825e-01 2.66169369e-01 5.56686878e-01 -8.84902060e-01
5.00573516e-01 -6.75764084e-01 -1.25322247e+00 2.66747862e-01
-8.54090393e-01 -1.91474818e-02 9.05292392e-01 2.24026605e-01
-8.57238650e-01 -5.77171817e-02 -7.36450031e-02 -4.64157984e-02
-1.56553715e-01 5.98099753e-02 -1.45399377e-01 -6.09196484e-01
-1.14324167e-01 9.52421203e-02 1.09427735e-01 -8.30765843e-01
1.82390481e-01 7.37539291e-01 1.82294324e-02 -4.45141405e-01
1.24957514e+00 5.45712590e-01 5.88889241e-01 -1.10759449e+00
-4.83415633e-01 -3.41145486e-01 -4.21526194e-01 -4.19159651e-01
6.58542693e-01 -3.50922167e-01 -1.09731102e+00 4.92877841e-01
-1.35015535e+00 1.23391196e-01 -2.61463046e-01 5.09565115e-01
-5.48035145e-01 2.76779622e-01 -4.59926546e-01 -1.02068067e+00
-5.64949624e-02 -1.01122367e+00 3.19022119e-01 6.34435713e-01
3.18565667e-01 -7.24717379e-01 -5.50658964e-02 4.09928292e-01
6.14437342e-01 2.27236092e-01 1.11306608e+00 8.44114739e-03
-1.62431002e+00 2.13351056e-01 -5.25436461e-01 3.99076551e-01
3.68780047e-01 3.58231515e-01 -5.54430187e-01 -3.64452153e-01
-1.85940757e-01 1.90401986e-01 5.23746431e-01 2.44937703e-01
6.58194661e-01 -2.68247873e-01 -3.47545683e-01 7.28364110e-01
2.21017575e+00 2.92286724e-01 7.31783509e-01 -1.23530030e-02
1.36230454e-01 6.64245307e-01 2.70377040e-01 4.51971591e-01
3.05674616e-02 6.87434137e-01 7.52832651e-01 -3.20536941e-01
-2.79133558e-01 2.48106256e-01 -2.97721177e-01 8.29576015e-01
-3.99847567e-01 -7.93666780e-01 -3.94892931e-01 4.42669272e-01
-1.78860688e+00 -8.88851106e-01 -4.30555552e-01 2.03543949e+00
3.34346652e-01 7.67581835e-02 2.60359883e-01 2.12923571e-01
7.50550508e-01 1.28507977e-02 -4.12646741e-01 -2.61760145e-01
-3.99362966e-02 4.07241374e-01 6.72859609e-01 6.73065662e-01
-5.39177299e-01 7.29203999e-01 6.02236223e+00 4.02447373e-01
-1.07968855e+00 -3.10890712e-02 1.10367686e-01 9.24370810e-03
-3.19706559e-01 3.92155409e-01 -7.83846498e-01 5.64115167e-01
5.51045716e-01 -2.22230852e-01 7.37393796e-01 -2.35931147e-02
6.01868689e-01 -2.35479191e-01 -6.45895481e-01 8.33474040e-01
-4.40414131e-01 -1.46393895e+00 -2.28510797e-02 3.27254772e-01
6.18066192e-01 -3.39960665e-01 -2.30774805e-01 -4.64997500e-01
-1.58444345e-01 -4.75897342e-01 7.11719751e-01 9.86887574e-01
2.85215288e-01 -5.33323228e-01 4.72851604e-01 5.36130294e-02
-9.88497794e-01 -2.21731976e-01 -4.04058516e-01 -3.31283987e-01
5.24878561e-01 1.01897979e+00 -2.16668397e-01 9.42275405e-01
1.24232858e-01 5.98611951e-01 -7.10035637e-02 1.94744158e+00
-2.25394517e-01 5.25130928e-01 -3.48921269e-01 -1.54924393e-01
1.63462773e-01 -1.12817001e+00 9.06292379e-01 7.01883137e-01
3.69417638e-01 3.05656577e-03 5.14229685e-02 1.41203415e+00
-1.14226885e-01 3.13342243e-01 -1.44210458e-01 -1.03869520e-01
4.51538295e-01 1.57166493e+00 -8.38132441e-01 1.60425827e-01
-6.71509683e-01 5.82887888e-01 2.47612879e-01 6.42016113e-01
-5.46557486e-01 -8.26938212e-01 1.27010131e+00 2.56221443e-01
1.76582232e-01 -4.75607425e-01 -2.22160220e-01 -9.54268992e-01
9.22798142e-02 -3.63782674e-01 4.34630141e-02 -4.82389599e-01
-1.32617927e+00 6.76380515e-01 -1.72145382e-01 -1.17411196e+00
3.08010817e-01 -7.25816071e-01 -5.49895763e-01 1.14027333e+00
-2.07213187e+00 -6.99523509e-01 -2.30704531e-01 3.98096085e-01
-4.62451950e-02 -1.27030268e-01 5.59164047e-01 7.23079145e-01
-1.05460179e+00 1.51873976e-01 1.57407597e-01 -2.89680809e-01
4.34340298e-01 -8.69893432e-01 -5.24639845e-01 1.13184941e+00
-2.72924393e-01 5.27233899e-01 8.01520944e-01 -3.11887175e-01
-1.51118100e+00 -6.11779988e-01 9.25256133e-01 5.34066081e-01
6.16674006e-01 -1.79363027e-01 -2.22680286e-01 3.65668416e-01
6.43733501e-01 -1.72151700e-01 7.01856434e-01 -1.97497189e-01
2.32464716e-01 -6.65081561e-01 -1.33521152e+00 3.40848684e-01
1.11011541e+00 -6.06521545e-03 -1.69999280e-03 3.94063830e-01
4.44229811e-01 -7.60642886e-02 -8.23113143e-01 1.30613253e-01
4.02807206e-01 -1.23124719e+00 7.98523784e-01 -6.35711253e-01
3.12774360e-01 -5.15313327e-01 1.53050363e-01 -1.15596664e+00
-3.82516086e-01 -1.13069272e+00 3.42707396e-01 1.45609117e+00
4.73590106e-01 -1.15091336e+00 7.36204088e-01 3.58885467e-01
-1.84394613e-01 -6.95972800e-01 -9.19237971e-01 -9.01009738e-01
-3.26583028e-01 -8.65295250e-03 7.47043371e-01 6.31531298e-01
-4.56163108e-01 3.98916632e-01 -3.29532653e-01 4.92164552e-01
8.90602648e-01 3.93394619e-01 6.61929309e-01 -1.50271690e+00
-1.70956269e-01 -6.10256433e-01 -5.75247407e-01 -1.19693446e+00
7.51010254e-02 -8.16389143e-01 -2.93239236e-01 -1.89044642e+00
-3.38000596e-01 -1.02180612e+00 -2.49235526e-01 1.58447996e-01
3.73322934e-01 4.25260901e-01 4.53320265e-01 1.64782390e-01
-2.09653094e-01 2.77394623e-01 1.53997469e+00 5.81281371e-02
-3.04626167e-01 3.07739705e-01 -6.74614310e-01 2.63537109e-01
9.85542655e-01 -4.31869298e-01 -5.19938946e-01 -4.86114502e-01
2.57203788e-01 1.28603846e-01 3.37380171e-01 -1.26855481e+00
2.45183617e-01 -3.28623772e-01 -3.57039094e-01 -1.27565905e-01
3.85383368e-01 -1.19934893e+00 3.67504835e-01 4.65812206e-01
1.73652798e-01 -2.63089657e-01 -1.54875815e-01 4.84967083e-01
-7.58251995e-02 -5.01682043e-01 1.12271881e+00 -5.56343235e-02
-4.05994415e-01 3.26586932e-01 -4.42750782e-01 -2.67450452e-01
1.16724563e+00 -2.99154371e-01 -5.25034308e-01 -8.34605470e-02
-9.02231574e-01 2.77569801e-01 3.12245935e-01 1.75837502e-02
2.00761572e-01 -1.03789258e+00 -4.93179321e-01 5.98235764e-02
-3.89783204e-01 -4.92103040e-01 6.99278861e-02 9.97814953e-01
-6.90426350e-01 5.76142907e-01 -6.39214218e-01 -2.62232572e-01
-1.15236557e+00 2.39582598e-01 6.03402078e-01 -2.16881692e-01
-2.16224790e-01 5.93177736e-01 -4.02205050e-01 5.68921864e-03
7.53439905e-04 -1.77988648e-01 -9.70570520e-02 3.88181955e-02
1.76367331e-02 7.07355559e-01 -1.54200941e-01 -5.63329637e-01
-1.78952381e-01 9.81351316e-01 4.00733262e-01 9.43086967e-02
1.60823488e+00 -5.21751046e-01 -8.82542133e-01 -8.37272312e-03
9.66484129e-01 6.11284040e-02 -7.81679988e-01 -2.17494115e-01
-2.33439773e-01 -7.77108848e-01 1.81012645e-01 -5.89344978e-01
-1.49824882e+00 6.12078130e-01 4.45075929e-01 8.47520769e-01
1.40977919e+00 -4.40669149e-01 9.10962045e-01 3.26417387e-01
5.12698233e-01 -1.09557235e+00 -4.24742401e-01 1.23232007e-01
3.66021335e-01 -7.26127565e-01 -1.87021628e-01 -1.29447222e+00
4.88315552e-01 1.46092582e+00 6.39877141e-01 -3.73626314e-02
7.74612844e-01 6.42027557e-02 -1.02866992e-01 -1.31735072e-01
-5.35726368e-01 -6.82800949e-01 4.92056347e-02 1.57151163e-01
2.93786228e-01 -4.72964533e-02 -9.84875441e-01 5.59180193e-02
9.10363495e-02 1.60654560e-01 6.60303295e-01 7.06131399e-01
-6.22169673e-01 -1.68856943e+00 -1.06080338e-01 4.73434001e-01
-7.11473599e-02 1.30935237e-01 -1.77884609e-01 5.16134560e-01
5.66470683e-01 1.47070527e+00 2.31628612e-01 -1.23001426e-01
5.42264581e-01 -8.15079957e-02 5.98194003e-01 -7.79113472e-01
-4.15702760e-01 -6.07266277e-02 1.46958813e-01 -3.63373131e-01
-8.05091977e-01 -1.66444346e-01 -8.82303119e-01 -5.26717961e-01
-5.85640669e-01 4.71535891e-01 6.19267583e-01 9.83878434e-01
4.38446820e-01 5.33722043e-01 1.09516227e+00 -2.13297576e-01
-3.24141502e-01 -4.43745375e-01 -1.04768348e+00 3.70815024e-02
4.73351836e-01 -5.92948854e-01 -3.25427115e-01 -4.05071855e-01] | [5.961945056915283, 1.6837393045425415] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.