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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
0650e92a-d76e-434a-9c74-43d0f8c03802 | polarnet-an-improved-grid-representation-for | 2003.14032 | null | https://arxiv.org/abs/2003.14032v2 | https://arxiv.org/pdf/2003.14032v2.pdf | PolarNet: An Improved Grid Representation for Online LiDAR Point Clouds Semantic Segmentation | The need for fine-grained perception in autonomous driving systems has resulted in recently increased research on online semantic segmentation of single-scan LiDAR. Despite the emerging datasets and technological advancements, it remains challenging due to three reasons: (1) the need for near-real-time latency with limited hardware; (2) uneven or even long-tailed distribution of LiDAR points across space; and (3) an increasing number of extremely fine-grained semantic classes. In an attempt to jointly tackle all the aforementioned challenges, we propose a new LiDAR-specific, nearest-neighbor-free segmentation algorithm - PolarNet. Instead of using common spherical or bird's-eye-view projection, our polar bird's-eye-view representation balances the points across grid cells in a polar coordinate system, indirectly aligning a segmentation network's attention with the long-tailed distribution of the points along the radial axis. We find that our encoding scheme greatly increases the mIoU in three drastically different segmentation datasets of real urban LiDAR single scans while retaining near real-time throughput. | ['Xiangyu Yue', 'Philip David', 'Zixiang Zhou', 'Zerong Xi', 'Yang Zhang', 'Boqing Gong', 'Hassan Foroosh'] | 2020-03-31 | polarnet-an-improved-grid-representation-for-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Zhang_PolarNet_An_Improved_Grid_Representation_for_Online_LiDAR_Point_Clouds_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Zhang_PolarNet_An_Improved_Grid_Representation_for_Online_LiDAR_Point_Clouds_CVPR_2020_paper.pdf | cvpr-2020-6 | ['robust-3d-semantic-segmentation', 'lidar-semantic-segmentation'] | ['computer-vision', 'computer-vision'] | [ 3.86393249e-01 -1.18510850e-01 -3.85634676e-02 -7.62036383e-01
-6.68798625e-01 -6.40381575e-01 3.87773782e-01 9.85573754e-02
-5.19394875e-01 6.46648169e-01 -3.39722157e-01 -5.54202378e-01
-3.90546471e-01 -1.02728319e+00 -5.93898177e-01 -2.93708146e-01
1.91962063e-01 9.86320198e-01 8.48643661e-01 -2.10065439e-01
5.80419362e-01 1.02686036e+00 -2.24891210e+00 -2.05335215e-01
9.63085651e-01 1.19582462e+00 2.70450145e-01 5.06185770e-01
-4.10221070e-01 4.91489023e-02 -1.35897502e-01 -1.98503688e-01
5.08597672e-01 6.39294535e-02 -5.09479165e-01 7.25106522e-02
8.25261474e-01 2.46236734e-02 -9.90582556e-02 1.10863996e+00
3.62983495e-01 2.01681823e-01 4.89479274e-01 -1.55810618e+00
-2.36585867e-02 -1.13616295e-01 -8.04787993e-01 8.78552198e-02
-8.65915120e-02 1.12064615e-01 9.93452430e-01 -8.25818956e-01
4.82874602e-01 1.09602904e+00 7.71812916e-01 7.40059987e-02
-1.17263448e+00 -6.25593543e-01 2.70551264e-01 2.76586205e-01
-1.79695392e+00 -2.92370200e-01 6.31024003e-01 -4.51150119e-01
7.05451846e-01 3.53637427e-01 6.34039521e-01 1.60525531e-01
2.66655702e-02 2.09217206e-01 1.08446312e+00 1.33945227e-01
3.64983290e-01 1.71133094e-02 5.16659431e-02 4.73421633e-01
4.04723346e-01 2.74039656e-02 -6.41501307e-01 -6.17379732e-02
5.85923553e-01 1.46324642e-03 1.73547551e-01 -9.39078927e-01
-1.09920633e+00 7.93413341e-01 5.00113070e-01 -2.99693286e-01
-9.45755914e-02 1.54543921e-01 2.48851806e-01 -6.41153678e-02
4.64617461e-01 2.40861982e-01 -6.27594829e-01 -2.86968738e-01
-1.21155930e+00 3.99312645e-01 6.61827147e-01 1.25846374e+00
1.33643866e+00 -2.65068710e-01 5.30438244e-01 7.64352798e-01
1.96834743e-01 6.86462045e-01 -1.83963999e-02 -1.12283552e+00
5.01776516e-01 6.07869864e-01 -4.29276489e-02 -9.16575432e-01
-5.53390920e-01 -3.44115406e-01 -6.97891235e-01 4.88686353e-01
5.57000458e-01 2.30769292e-01 -9.21470881e-01 1.45093513e+00
4.84066606e-01 -1.39097825e-01 -1.98840588e-01 8.53294313e-01
3.96555424e-01 3.42242599e-01 -1.33416846e-01 1.66531011e-01
1.50431430e+00 -6.42680168e-01 -6.86797351e-02 -4.68150139e-01
1.78931072e-01 -7.00238705e-01 9.04555142e-01 1.68434247e-01
-8.79601479e-01 -5.28335452e-01 -9.59519982e-01 -4.38356519e-01
-6.53414130e-01 -3.56816798e-01 6.06508315e-01 7.25246310e-01
-1.04386961e+00 4.08724099e-01 -6.76037908e-01 -4.26308036e-01
6.23499453e-01 4.35440689e-01 -1.92197055e-01 -2.25314051e-01
-6.34554267e-01 6.34609103e-01 3.75156850e-02 -1.52811105e-03
-3.83340120e-02 -9.06284869e-01 -8.45989823e-01 1.67102658e-03
6.74909234e-01 -5.79994500e-01 1.03822505e+00 -3.27057093e-01
-1.06068802e+00 8.14113915e-01 -5.48728287e-01 -4.10503149e-01
6.06635392e-01 -2.32828781e-02 -4.43923995e-02 -2.27752272e-02
4.70250607e-01 1.16194105e+00 3.40614796e-01 -1.33492982e+00
-1.03175271e+00 -8.64083946e-01 -1.82122707e-01 4.70378190e-01
2.89090276e-01 -5.62990129e-01 -4.43431675e-01 -5.00834212e-02
8.33420634e-01 -1.10976303e+00 -3.75532001e-01 4.02145654e-01
-2.83042639e-01 -2.42006868e-01 1.23511326e+00 1.40666753e-01
7.06885874e-01 -2.19343019e+00 -3.27319890e-01 3.62345159e-01
2.10438475e-01 1.09351790e-02 2.17389718e-01 2.85560787e-01
1.58788249e-01 9.59955081e-02 -3.59702379e-01 -1.88709974e-01
-5.69857238e-03 7.12369144e-01 -2.77532011e-01 4.85225141e-01
2.24990010e-01 6.03674650e-01 -9.58241463e-01 -6.39381409e-01
4.75791812e-01 3.66829991e-01 -5.26187062e-01 -2.53072172e-01
-1.50853902e-01 3.66136909e-01 -3.66713673e-01 6.41448617e-01
1.19960678e+00 5.13504520e-02 -1.89236119e-01 -2.56554872e-01
-5.45503616e-01 3.04013580e-01 -1.33704102e+00 1.52961659e+00
-2.10396796e-01 7.22899258e-01 2.23792613e-01 -8.78625453e-01
1.00029385e+00 -3.75859290e-01 6.20481014e-01 -9.67687547e-01
-2.26333618e-01 4.48327869e-01 -1.04167096e-01 -1.26562893e-01
1.04047668e+00 -2.08855450e-01 -1.68664843e-01 2.65780240e-01
-3.99434298e-01 -7.21408546e-01 2.08651379e-01 9.95835587e-02
8.00166190e-01 -5.19973785e-02 1.51362449e-01 -4.89132583e-01
2.87602663e-01 3.21646065e-01 5.21185279e-01 6.28050506e-01
-3.16800028e-01 7.63629854e-01 2.53201455e-01 -4.40950662e-01
-8.80955040e-01 -1.22901320e+00 -5.62019467e-01 9.48922813e-01
6.72226787e-01 -2.31074601e-01 -6.08881116e-01 -3.65454704e-01
4.58508700e-01 6.32905900e-01 -2.20809951e-01 2.92087257e-01
-5.85745692e-01 -3.56794298e-01 4.44387108e-01 4.69465971e-01
5.61651766e-01 -6.17266595e-01 -1.12051558e+00 8.62948298e-02
-1.61309496e-01 -1.22946501e+00 -4.17320400e-01 5.47933936e-01
-8.93337905e-01 -1.12529135e+00 -2.38853157e-01 -3.84654254e-01
5.77905416e-01 8.67095232e-01 1.16812193e+00 -3.25645864e-01
-3.24592382e-01 2.09640369e-01 6.98538274e-02 -4.51735884e-01
3.66528928e-01 2.51822531e-01 -1.14773646e-01 -3.19347560e-01
6.07367396e-01 -7.16612637e-01 -6.57330453e-01 6.08644903e-01
-5.29244006e-01 4.89084534e-02 3.29553455e-01 3.09548557e-01
1.07998753e+00 2.72232175e-01 1.77859560e-01 -7.47303545e-01
6.40007704e-02 -4.84676242e-01 -1.10662007e+00 -3.76523435e-01
-7.16636956e-01 8.19707382e-03 2.64407575e-01 1.42217949e-01
-5.78711569e-01 3.15443188e-01 -2.12513909e-01 -3.77705805e-02
-3.67864072e-01 9.31588784e-02 -8.41179937e-02 -3.33221167e-01
3.26179147e-01 1.18961424e-01 -9.10422206e-02 -2.73803234e-01
6.34413958e-01 6.69802606e-01 8.11803877e-01 -5.87578535e-01
7.72951305e-01 9.09571648e-01 4.62953001e-01 -1.05667055e+00
-8.28944206e-01 -7.54354656e-01 -9.82178688e-01 -1.81741074e-01
1.03424883e+00 -8.35882723e-01 -7.95887113e-01 4.16641861e-01
-1.11861932e+00 -1.30360037e-01 -6.20411098e-01 1.82664186e-01
-7.20646858e-01 2.73588985e-01 1.47160888e-01 -8.80223155e-01
9.73175243e-02 -1.42877948e+00 1.43251896e+00 2.59515405e-01
-9.59390923e-02 -4.55996871e-01 -1.76030084e-01 4.79323298e-01
3.01590711e-01 8.78564119e-02 7.82448292e-01 -2.62310356e-01
-1.17767966e+00 4.50804830e-02 -8.88644874e-01 9.31483805e-02
-5.32117076e-02 4.30235043e-02 -9.78431046e-01 1.35508597e-01
-2.61727333e-01 -3.42950225e-02 6.02221966e-01 3.84785950e-01
1.22349560e+00 2.62256652e-01 -4.32862997e-01 8.24387133e-01
1.46872461e+00 1.61650643e-01 5.14785290e-01 2.80015945e-01
7.10385323e-01 7.54697204e-01 9.08888817e-01 2.83617526e-01
1.05398297e+00 7.12812781e-01 8.33575189e-01 -6.80052042e-02
-1.01597413e-01 -2.90096164e-01 -3.90043378e-01 5.10297358e-01
1.47656813e-01 -2.25397825e-01 -1.08946729e+00 7.19147682e-01
-1.72824204e+00 -8.72908294e-01 -4.69515115e-01 2.37296534e+00
3.61108869e-01 2.95842797e-01 -4.10306491e-02 1.87842354e-01
5.78852713e-01 2.42635056e-01 -8.80250394e-01 -5.05480587e-01
7.25373672e-03 2.12436885e-01 1.16055202e+00 5.19399881e-01
-9.20335352e-01 9.07791555e-01 5.88429022e+00 9.11492705e-01
-1.26395965e+00 -2.77393889e-02 3.07242930e-01 -7.51626417e-02
-2.88433909e-01 3.22851613e-02 -1.11022615e+00 5.33210695e-01
7.43790448e-01 -3.68655175e-02 3.30242395e-01 8.42943430e-01
3.26149575e-02 -5.96802294e-01 -7.86679685e-01 1.03725970e+00
-1.18361868e-01 -1.25706983e+00 -2.84178913e-01 5.92458725e-01
4.83558416e-01 7.82446742e-01 -6.42633140e-02 -1.21285759e-01
3.04374695e-01 -8.66941512e-01 1.12098753e+00 2.05956504e-01
1.12548196e+00 -8.20277214e-01 2.69636452e-01 6.33813620e-01
-1.62457609e+00 1.11987084e-01 -5.40778041e-01 -2.09818363e-01
2.42738292e-01 8.07464480e-01 -7.21544385e-01 3.49020571e-01
8.19485009e-01 4.05492127e-01 -3.83928984e-01 9.57499623e-01
1.19676068e-01 1.65467903e-01 -8.47499728e-01 1.19368739e-01
3.98101985e-01 -3.81699443e-01 3.40314388e-01 1.05151820e+00
4.07255888e-01 -4.35862727e-02 1.73867673e-01 6.69911683e-01
1.66526377e-01 -2.82803118e-01 -7.76689231e-01 3.05919051e-01
7.86013246e-01 1.27734613e+00 -1.17125118e+00 -1.92882866e-01
-3.64215612e-01 5.65789223e-01 8.00262243e-02 9.37875137e-02
-6.67262197e-01 -3.76365423e-01 1.07829416e+00 6.00213408e-01
5.31997085e-01 -7.36135721e-01 -9.11484957e-01 -6.58164978e-01
1.17484465e-01 -3.82234126e-01 5.21437004e-02 -8.78031611e-01
-9.88802433e-01 5.77021182e-01 3.11351456e-02 -1.07202780e+00
-1.24235921e-01 -5.01179993e-01 -1.68333828e-01 9.19705391e-01
-2.02363253e+00 -9.47884798e-01 -5.46288610e-01 5.88928759e-01
5.88141322e-01 3.53637218e-01 5.53219914e-01 4.69381481e-01
1.73737891e-02 3.81254964e-02 -1.16926402e-01 -4.25318271e-01
4.90426123e-01 -1.28725159e+00 7.15200245e-01 5.89900911e-01
5.46844304e-03 2.69901663e-01 6.14004910e-01 -5.05436242e-01
-1.08709621e+00 -1.06729424e+00 9.95104730e-01 -4.36994791e-01
5.21176100e-01 -5.08336842e-01 -7.91717768e-01 4.41819072e-01
-1.89350620e-01 1.99030995e-01 3.40439618e-01 -2.05175411e-02
-4.92053002e-01 -4.09474075e-01 -1.29686296e+00 4.53104287e-01
1.28820491e+00 -5.00413239e-01 -2.28625610e-01 1.88783467e-01
4.61004138e-01 -5.66914499e-01 -4.80257303e-01 5.41658223e-01
6.34820521e-01 -1.22025597e+00 1.06741023e+00 -1.75373927e-02
6.75280690e-02 -6.67230725e-01 -4.40835595e-01 -9.30363417e-01
-1.58757702e-01 -2.58259565e-01 5.14587820e-01 7.87458241e-01
1.24643132e-01 -8.83293927e-01 1.10449767e+00 4.67114031e-01
-4.09084916e-01 -8.15950751e-01 -1.43589389e+00 -5.38527191e-01
-1.61484286e-01 -8.06717753e-01 7.19958246e-01 5.08496404e-01
-4.34053063e-01 3.83907497e-01 2.16019839e-01 3.26690108e-01
7.43854225e-01 3.77720654e-01 1.01045609e+00 -1.70628870e+00
4.42784607e-01 -3.88135135e-01 -6.76324427e-01 -1.58360672e+00
-1.73861340e-01 -6.51080847e-01 4.04461086e-01 -1.48424435e+00
-1.27492607e-01 -1.01814246e+00 2.42052853e-01 3.19610000e-01
3.37608278e-01 6.26650512e-01 4.71658744e-02 1.63569734e-01
-6.38525307e-01 3.37694228e-01 1.12545204e+00 2.15886369e-01
1.41452968e-01 4.14625667e-02 -6.81775868e-01 8.96048188e-01
5.93139231e-01 -5.08854866e-01 -6.04303837e-01 -5.46222448e-01
5.11095822e-01 -1.40762523e-01 5.06679654e-01 -1.20078576e+00
6.52617276e-01 -3.74029487e-01 1.52831338e-02 -1.52467287e+00
7.77742028e-01 -1.02547443e+00 2.30517648e-02 1.94062710e-01
3.48096102e-01 1.67752787e-01 2.06065848e-01 5.40846705e-01
-2.09149495e-01 -7.08530797e-03 1.02017081e+00 -2.40048587e-01
-8.14191341e-01 3.72165918e-01 -3.07834208e-01 1.71760261e-01
1.01181245e+00 -9.60957587e-01 -3.35164845e-01 -1.41156837e-01
-3.10366362e-01 4.29357618e-01 8.50331843e-01 2.60507017e-01
2.89803624e-01 -9.64461982e-01 -3.18115890e-01 6.54038191e-01
2.01801792e-01 6.97358370e-01 3.49208504e-01 7.44240999e-01
-6.52708292e-01 6.16664648e-01 -1.44086435e-01 -1.16633356e+00
-1.19211388e+00 9.87846479e-02 1.62933722e-01 9.25618336e-02
-6.21267617e-01 9.43967700e-01 2.71563351e-01 -7.58010864e-01
-9.83617678e-02 -3.23493600e-01 2.70664424e-01 2.64893591e-01
-4.76666689e-02 6.51801646e-01 3.10238481e-01 -9.61542904e-01
-5.84738135e-01 1.06720471e+00 1.72950238e-01 -9.78544950e-02
1.14884901e+00 -4.33374971e-01 -3.70680913e-02 4.51040089e-01
8.60977590e-01 2.16025040e-01 -1.44628763e+00 -5.36744781e-02
-3.84587832e-02 -7.13070393e-01 1.15760185e-01 -4.91374344e-01
-1.00946307e+00 1.20293486e+00 5.27666867e-01 4.88380998e-01
7.43825912e-01 -4.41652490e-03 9.84462440e-01 2.21379668e-01
8.10757518e-01 -1.24068177e+00 -5.37480772e-01 7.67980993e-01
3.33802313e-01 -1.13936663e+00 2.97059506e-01 -8.92000735e-01
-4.79924917e-01 1.04921627e+00 3.95683110e-01 8.91164020e-02
8.13544929e-01 4.58623588e-01 2.15407908e-01 -4.40300047e-01
-3.72263223e-01 -4.60488766e-01 9.69182402e-02 8.74515772e-01
8.18246379e-02 1.11647204e-01 -3.51855718e-02 -1.07414350e-01
-6.74294651e-01 -9.35046077e-02 2.44366631e-01 7.66383767e-01
-8.16144943e-01 -9.76051390e-01 -2.93360382e-01 5.35822272e-01
4.46072146e-02 9.50235203e-02 -9.96967405e-02 7.96811044e-01
5.04547000e-01 8.78662646e-01 7.03018427e-01 -1.18313655e-01
5.70681095e-01 -1.81993619e-01 2.71095723e-01 -6.21650636e-01
1.02458954e-01 -1.38677299e-01 -3.87750864e-02 -9.09086406e-01
-2.26709977e-01 -1.03006160e+00 -1.36847460e+00 -4.24358815e-01
-2.10414246e-01 -8.11743885e-02 1.20774758e+00 7.85819948e-01
7.06867039e-01 2.12094262e-01 4.21572983e-01 -1.14487898e+00
-3.18291843e-01 -4.79697108e-01 -7.15504408e-01 -5.14895841e-02
2.62458593e-01 -7.79302239e-01 -3.15871745e-01 -2.69789308e-01] | [8.038702011108398, -2.742123603820801] |
95931c1f-fb55-48fc-a276-f898d22eba45 | hierarchical-object-representation-for-open | null | null | http://papers.nips.cc/paper/6539-hierarchical-object-representation-for-open-ended-object-category-learning-and-recognition | http://papers.nips.cc/paper/6539-hierarchical-object-representation-for-open-ended-object-category-learning-and-recognition.pdf | Hierarchical Object Representation for Open-Ended Object Category Learning and Recognition | Most robots lack the ability to learn new objects from past experiences. To migrate a robot to a new environment one must often completely re-generate the knowledge- base that it is running with. Since in open-ended domains the set of categories to be learned is not predefined, it is not feasible to assume that one can pre-program all object categories required by robots. Therefore, autonomous robots must have the ability to continuously execute learning and recognition in a concurrent and interleaved fashion. This paper proposes an open-ended 3D object recognition system which concurrently learns both the object categories and the statistical features for encoding objects. In particular, we propose an extension of Latent Dirichlet Allocation to learn structural semantic features (i.e. topics) from low-level feature co-occurrences for each category independently. Moreover, topics in each category are discovered in an unsupervised fashion and are updated incrementally using new object views. The approach contains similarities with the organization of the visual cortex and builds a hierarchy of increasingly sophisticated representations. Results show the fulfilling performance of this approach on different types of objects. Moreover, this system demonstrates the capability of learning from few training examples and competes with state-of-the-art systems. | ['Luís Seabra Lopes', 'Ana Maria Tomé', 'Seyed Hamidreza Kasaei'] | 2016-12-01 | null | null | null | neurips-2016-12 | ['3d-object-recognition'] | ['computer-vision'] | [ 4.23987694e-02 2.68820047e-01 -6.49705529e-02 -7.01398313e-01
4.60187793e-02 -5.48848391e-01 8.87371004e-01 1.32704929e-01
-5.01673758e-01 6.30348086e-01 -2.78930515e-01 2.18710780e-01
-2.63147622e-01 -9.63333964e-01 -6.66598380e-01 -7.11863518e-01
-3.68068576e-01 1.12133813e+00 5.60380042e-01 3.25980522e-02
4.36574161e-01 8.34148228e-01 -2.37885118e+00 3.84490877e-01
4.56651419e-01 7.08948612e-01 9.45584297e-01 3.76712501e-01
-5.68415821e-01 4.70175833e-01 -3.88218313e-01 2.00207829e-01
1.82689786e-01 -1.40952632e-01 -1.02716553e+00 6.25095606e-01
1.24183387e-01 -1.87461630e-01 -1.62491828e-01 8.44595015e-01
-1.48852125e-01 2.17092931e-01 1.08500946e+00 -1.23056853e+00
-5.55703282e-01 2.74876654e-01 -2.86543872e-02 6.16823472e-02
2.70129591e-01 -5.13555110e-02 7.48673379e-01 -9.65216577e-01
8.99064839e-01 1.16310573e+00 3.55454028e-01 6.41839564e-01
-1.21336460e+00 -1.57804936e-01 5.57899773e-01 4.74480927e-01
-1.18960762e+00 -2.22986937e-01 6.81278765e-01 -6.86011910e-01
1.19857132e+00 -3.44486952e-01 7.30795562e-01 7.81618476e-01
7.62015134e-02 6.36616647e-01 9.60079730e-01 -5.63229918e-01
7.00708985e-01 4.80382323e-01 1.80862769e-01 5.95179558e-01
4.19051528e-01 -9.51055884e-02 -6.31595492e-01 1.01015948e-01
7.45313048e-01 1.80032268e-01 2.42556542e-01 -1.22055316e+00
-1.22838545e+00 6.50681317e-01 3.39133918e-01 6.95342302e-01
-5.67554474e-01 7.75794238e-02 1.97780326e-01 3.37717503e-01
6.66157678e-02 3.97907197e-01 -6.31693959e-01 3.99903767e-02
-5.23018301e-01 5.58340028e-02 9.98867214e-01 1.15060151e+00
1.29558051e+00 -1.85368568e-01 4.88171756e-01 7.28606820e-01
4.96864438e-01 2.25420937e-01 8.68648708e-01 -8.85493517e-01
-1.40582725e-01 8.44174027e-01 1.71362441e-02 -8.62203717e-01
-3.88073564e-01 -1.99266240e-01 -3.86621118e-01 4.71655518e-01
1.60754576e-01 4.04910028e-01 -1.16423631e+00 1.62434316e+00
3.30416977e-01 -2.61245877e-01 4.74272430e-01 4.88977969e-01
4.01342303e-01 6.26559019e-01 -2.01871973e-02 -1.00944571e-01
1.17841196e+00 -7.66512275e-01 -3.63876909e-01 -4.43215787e-01
3.38159591e-01 -2.61650413e-01 7.35327780e-01 5.25880635e-01
-6.04627013e-01 -9.02731836e-01 -1.06947207e+00 3.03477142e-02
-7.94014871e-01 -1.16229625e-02 8.16150963e-01 3.87849182e-01
-1.09213841e+00 5.02081096e-01 -9.73822594e-01 -8.25517297e-01
4.17750269e-01 5.44538081e-01 -5.22155762e-01 -1.40711114e-01
-6.28499866e-01 1.01471293e+00 9.28414464e-01 -1.34869695e-01
-1.56192935e+00 -1.09058395e-01 -6.95580661e-01 -2.02646211e-01
2.29538336e-01 -5.90120792e-01 1.28557444e+00 -1.09193373e+00
-1.62286603e+00 9.69022393e-01 -7.08698705e-02 -3.33606035e-01
2.85674501e-02 -1.02136828e-01 4.28845100e-02 3.20525438e-01
2.56788969e-01 1.02506304e+00 9.65731740e-01 -1.71797228e+00
-9.75639105e-01 -5.73684812e-01 1.30187973e-01 4.59350944e-01
-4.29671615e-01 -4.47468191e-01 -2.72396535e-01 -1.51926190e-01
6.77816153e-01 -9.62918699e-01 -2.85306305e-01 4.52829041e-02
2.20905557e-01 -4.53055203e-01 8.02271187e-01 -1.39038503e-01
3.99459779e-01 -2.25400329e+00 5.22995293e-01 1.28654256e-01
9.05712023e-02 -2.46407807e-01 6.00722246e-03 3.73147458e-01
7.27016404e-02 -3.69301766e-01 -2.35009715e-01 -2.06834763e-01
1.65651590e-01 8.07198346e-01 -2.79374987e-01 3.30757201e-01
1.09750636e-01 4.49015707e-01 -9.92133796e-01 -3.66043150e-01
9.32265595e-02 1.02197662e-01 -6.05782926e-01 2.61893094e-01
-4.78164613e-01 2.39826068e-01 -4.73763257e-01 4.12544310e-01
3.38203669e-01 -2.88921803e-01 4.89904374e-01 2.24114835e-01
-3.16299498e-01 2.22424805e-01 -1.26841521e+00 1.89607680e+00
-4.30118382e-01 4.89089787e-01 -2.60149181e-01 -1.50500214e+00
1.33110249e+00 2.63891280e-01 4.69069660e-01 -2.08622307e-01
-1.38631433e-01 2.65721619e-01 -1.67272538e-01 -6.22910917e-01
4.46598530e-01 -3.16062748e-01 -1.79571301e-01 4.30720806e-01
7.14815497e-01 -2.45981008e-01 2.99610585e-01 1.13767141e-03
1.04608750e+00 2.97576278e-01 3.42209548e-01 -2.26774424e-01
2.75583565e-01 3.60312223e-01 3.43334824e-01 6.87801480e-01
1.40879631e-01 2.07993880e-01 2.69447565e-01 -6.04784608e-01
-1.08334100e+00 -1.39824438e+00 -1.62598103e-01 1.25593495e+00
2.17282966e-01 3.25409621e-02 -4.02154028e-01 -7.29697645e-01
2.04372406e-02 7.17722774e-01 -6.75765395e-01 -1.15512803e-01
-1.92296579e-01 -4.40883845e-01 -1.67490821e-02 3.24567944e-01
3.72410983e-01 -1.31919825e+00 -1.29044926e+00 4.32402760e-01
2.23410055e-01 -7.45780230e-01 5.12816429e-01 7.69028664e-01
-1.22820687e+00 -9.65919018e-01 -5.10810137e-01 -1.19874310e+00
1.07160628e+00 4.13183987e-01 8.97628725e-01 -3.64138186e-01
-3.17519128e-01 9.92997348e-01 -6.11795127e-01 -4.33990479e-01
-4.17805582e-01 2.36194268e-01 2.65287936e-01 -7.59617686e-02
5.47996283e-01 -8.58207822e-01 -7.25990385e-02 2.78616309e-01
-1.00430977e+00 -6.74321204e-02 9.58468318e-01 7.97608316e-01
6.43438995e-01 3.59404504e-01 6.74618125e-01 -5.91386437e-01
1.44826069e-01 -7.37497211e-01 -6.38739407e-01 1.73312575e-01
-4.46862549e-01 3.07148844e-01 3.36782694e-01 -6.54477656e-01
-1.25804627e+00 5.28738439e-01 4.29854542e-01 -1.70586362e-01
-8.80065978e-01 5.06445169e-01 -2.48388857e-01 2.29187131e-01
6.55347884e-01 4.04151678e-01 9.58873108e-02 -6.22160673e-01
5.79413354e-01 6.93772793e-01 3.42532814e-01 -5.19490302e-01
5.87268054e-01 5.28007805e-01 -3.35454047e-01 -9.83700514e-01
-5.46407998e-01 -6.45287693e-01 -1.25157309e+00 -3.29801708e-01
7.92422056e-01 -8.38889897e-01 -3.10203791e-01 6.84945166e-01
-1.27084649e+00 -3.19845378e-01 -7.33687222e-01 6.41165376e-01
-9.72268283e-01 2.07808554e-01 -9.65090916e-02 -5.49291134e-01
4.77860242e-01 -8.06538045e-01 8.21545839e-01 1.65359750e-01
-7.93297589e-02 -7.42011249e-01 2.40817696e-01 -6.54190257e-02
1.38968825e-01 -1.81190148e-01 1.00556171e+00 -9.10460413e-01
-7.30720162e-01 -1.12761930e-02 1.09621517e-01 2.90413290e-01
3.63874793e-01 -3.47208768e-01 -9.69056308e-01 -2.08418489e-01
1.68353349e-01 -3.63942832e-01 7.76486278e-01 -2.00567488e-03
7.33417571e-01 -7.81979859e-02 -6.74139619e-01 1.72149613e-02
1.30380893e+00 6.22045040e-01 4.83543396e-01 4.87451196e-01
2.01510757e-01 9.89735484e-01 4.57216412e-01 5.06442249e-01
5.07416129e-01 5.87865710e-01 4.53482062e-01 5.10217965e-01
6.19236864e-02 1.14991432e-02 3.12506646e-01 8.38502228e-01
4.95649548e-03 2.33423039e-01 -1.13389540e+00 8.94256890e-01
-1.72905898e+00 -9.70148563e-01 3.40983093e-01 1.93836772e+00
9.23691630e-01 1.45412445e-01 -1.47586584e-01 6.87191039e-02
6.57605112e-01 -3.96535784e-01 -7.20604539e-01 -2.18663573e-01
1.68349713e-01 -5.63805550e-02 7.38196746e-02 1.92607865e-01
-1.08720326e+00 9.89003301e-01 6.33911753e+00 3.05189490e-01
-7.84357250e-01 5.03510498e-02 9.25133601e-02 3.60513896e-01
-1.45236999e-01 1.88264310e-01 -1.03408945e+00 5.07524833e-02
7.20682502e-01 -2.14762419e-01 3.56756687e-01 1.33284950e+00
-4.54016030e-01 -4.68663692e-01 -1.47650647e+00 7.98439860e-01
4.40026760e-01 -1.04230320e+00 1.09320499e-01 1.97657958e-01
6.37538850e-01 1.35756046e-01 -6.90933764e-02 3.94777328e-01
5.02932370e-01 -6.42485082e-01 9.88154590e-01 9.04635549e-01
4.06659931e-01 -3.07495654e-01 5.93752027e-01 6.19836628e-01
-9.41407144e-01 -3.93871009e-01 -7.57579088e-01 -1.91964015e-01
-1.74536586e-01 4.50159431e-01 -1.36453795e+00 4.17651504e-01
9.11348283e-01 7.32610881e-01 -6.26505971e-01 1.21252620e+00
-2.21227273e-01 1.98605992e-02 -4.77407664e-01 -2.01218918e-01
2.15867743e-01 1.22782439e-01 3.12095016e-01 9.00543988e-01
5.24931550e-01 2.19148487e-01 2.38001585e-01 7.00994730e-01
3.02945197e-01 -1.60359085e-01 -9.67460871e-01 4.90322895e-02
4.33394879e-01 9.00879264e-01 -1.12541628e+00 -5.62646985e-01
-3.56244445e-01 8.01057637e-01 3.73516262e-01 1.42954096e-01
-2.12312952e-01 -3.27791989e-01 4.35601026e-01 -2.07957089e-01
8.31209779e-01 -7.22380996e-01 -3.95134799e-02 -9.64519083e-01
-1.15486555e-01 -5.31425118e-01 2.64593959e-01 -7.20121801e-01
-1.37393689e+00 7.09878862e-01 4.05619919e-01 -1.27124345e+00
-5.60670555e-01 -7.34093606e-01 1.20145626e-01 3.60272646e-01
-1.46696317e+00 -8.83244276e-01 -3.72142613e-01 5.88394642e-01
7.22617984e-01 -5.20648003e-01 1.13632429e+00 -1.18693367e-01
2.45743454e-01 -1.91909477e-01 2.40707964e-01 -2.05828816e-01
3.71035397e-01 -1.10019433e+00 -7.48523623e-02 3.28681558e-01
4.73317325e-01 4.22013968e-01 5.78428805e-01 -6.86648011e-01
-1.29116845e+00 -8.14706087e-01 7.93855250e-01 -3.50771934e-01
5.27353644e-01 -3.90915841e-01 -1.08070016e+00 7.25207388e-01
-9.21411440e-03 -1.94393560e-01 7.13315248e-01 1.87661558e-01
-3.47613543e-01 -2.73720890e-01 -9.52422976e-01 1.13205753e-01
1.04435194e+00 -3.45364809e-01 -1.39400506e+00 1.61300033e-01
4.45819169e-01 1.20540984e-01 -5.82619190e-01 3.27363759e-01
4.09698218e-01 -9.18906689e-01 6.60861731e-01 -5.17130077e-01
-4.67206314e-02 -4.70849752e-01 -4.33681309e-01 -1.21059394e+00
-4.31624621e-01 -7.55281001e-02 9.22241360e-02 1.09479034e+00
2.71294266e-01 -5.76256573e-01 8.81964386e-01 2.82063991e-01
-3.16864133e-01 -1.54510677e-01 -9.49145913e-01 -9.23120975e-01
-1.20990165e-01 -3.66468400e-01 4.54571903e-01 8.37988257e-01
9.17654485e-02 3.14354211e-01 2.73727149e-01 3.23523641e-01
7.22958803e-01 3.94415915e-01 8.61976266e-01 -1.72011268e+00
-1.92405179e-01 -2.58861333e-01 -8.84812891e-01 -1.05390382e+00
1.97726771e-01 -1.05262816e+00 5.43436527e-01 -1.81292391e+00
1.18155167e-01 -8.36343110e-01 -2.96851486e-01 7.96262383e-01
5.13712168e-01 -1.94247171e-01 1.35411605e-01 5.23600698e-01
-8.26789856e-01 7.33722568e-01 7.42738783e-01 -1.21632889e-01
-3.83997679e-01 8.34040716e-03 -2.95490205e-01 9.93868828e-01
8.98127079e-01 -7.23700523e-01 -5.60994208e-01 -4.74975407e-01
2.04251364e-01 -4.22362655e-01 3.52528930e-01 -1.36560512e+00
5.23534536e-01 -1.81747392e-01 4.41127598e-01 -6.37882948e-01
3.87275815e-01 -1.11062551e+00 2.37217829e-01 4.50285792e-01
-1.76379949e-01 -2.35664427e-01 9.74849984e-02 7.78957844e-01
-3.00767332e-01 -7.56845057e-01 5.28751254e-01 -6.49270356e-01
-1.49383521e+00 -1.36199417e-02 -9.78867590e-01 -4.77795094e-01
1.37974620e+00 -6.14076078e-01 8.67039412e-02 1.38025478e-01
-1.14277005e+00 2.82888860e-02 6.55548275e-01 5.60858071e-01
8.15205812e-01 -1.07429934e+00 -1.83049500e-01 4.82388079e-01
4.00183260e-01 1.40304357e-01 1.14727236e-01 1.91262797e-01
-3.39359224e-01 4.05097336e-01 -7.85377145e-01 -8.56523752e-01
-7.09398925e-01 7.73104012e-01 1.96583513e-02 7.93006644e-02
-5.07115364e-01 7.79311240e-01 3.16558242e-01 -6.36736870e-01
4.02749032e-01 -2.16845393e-01 -6.19439006e-01 4.31470156e-01
3.01844746e-01 1.26850545e-01 -7.60159343e-02 -4.68524128e-01
-2.83552349e-01 5.65685451e-01 -8.11274499e-02 -2.19801441e-01
1.77251780e+00 -2.93793172e-01 -4.76881057e-01 1.15349245e+00
8.55690122e-01 -7.03644574e-01 -1.55128026e+00 -3.39066744e-01
5.12537181e-01 -3.42121512e-01 -4.00010228e-01 -5.87160587e-01
-5.57017803e-01 8.74148250e-01 7.63040662e-01 2.52300978e-01
1.01297772e+00 5.62392712e-01 -1.22676760e-01 1.12447464e+00
9.76989388e-01 -1.26532555e+00 6.29717350e-01 7.17332780e-01
8.33643258e-01 -9.13550377e-01 -9.34568327e-03 -1.47922397e-01
-4.25182641e-01 1.35518837e+00 4.45489079e-01 -1.03757076e-01
8.72128069e-01 7.12246224e-02 -1.63220644e-01 -3.22434247e-01
-9.64852870e-01 -3.42136741e-01 5.37554212e-02 1.02008760e+00
-2.03949884e-01 -3.39751095e-02 7.62076899e-02 3.03037226e-01
-4.03320640e-02 -9.31489933e-03 5.33958733e-01 1.29902554e+00
-1.16408837e+00 -1.15278673e+00 -2.92887002e-01 2.88434476e-01
1.62265897e-01 4.25216228e-01 -1.65493041e-01 7.40914762e-01
5.37976980e-01 6.59355462e-01 2.92410493e-01 -1.13709182e-01
2.04226539e-01 3.85926455e-01 6.54813647e-01 -1.24349916e+00
8.82716700e-02 -1.87247485e-01 -1.70580640e-01 -3.38612407e-01
-8.73149633e-01 -1.00200975e+00 -1.49602735e+00 5.10489881e-01
-6.65780678e-02 3.24203432e-01 1.05419672e+00 9.40933406e-01
3.37267905e-01 4.31395888e-01 5.87908745e-01 -1.23139012e+00
-3.36700529e-01 -8.55721414e-01 -6.84547842e-01 1.54298216e-01
8.13712329e-02 -1.25242817e+00 -3.08284163e-01 7.36993968e-01] | [7.5819783210754395, -1.1154332160949707] |
db6c3bbe-4ec1-4a47-b56d-fc0e80f7c9b7 | medml-fusing-medical-knowledge-and-machine | 2207.12283 | null | https://arxiv.org/abs/2207.12283v1 | https://arxiv.org/pdf/2207.12283v1.pdf | MedML: Fusing Medical Knowledge and Machine Learning Models for Early Pediatric COVID-19 Hospitalization and Severity Prediction | The COVID-19 pandemic has caused devastating economic and social disruption, straining the resources of healthcare institutions worldwide. This has led to a nationwide call for models to predict hospitalization and severe illness in patients with COVID-19 to inform distribution of limited healthcare resources. We respond to one of these calls specific to the pediatric population. To address this challenge, we study two prediction tasks for the pediatric population using electronic health records: 1) predicting which children are more likely to be hospitalized, and 2) among hospitalized children, which individuals are more likely to develop severe symptoms. We respond to the national Pediatric COVID-19 data challenge with a novel machine learning model, MedML. MedML extracts the most predictive features based on medical knowledge and propensity scores from over 6 million medical concepts and incorporates the inter-feature relationships between heterogeneous medical features via graph neural networks (GNN). We evaluate MedML across 143,605 patients for the hospitalization prediction task and 11,465 patients for the severity prediction task using data from the National Cohort Collaborative (N3C) dataset. We also report detailed group-level and individual-level feature importance analyses to evaluate the model interpretability. MedML achieves up to a 7% higher AUROC score and up to a 14% higher AUPRC score compared to the best baseline machine learning models and performs well across all nine national geographic regions and over all three-month spans since the start of the pandemic. Our cross-disciplinary research team has developed a method of incorporating clinical domain knowledge as the framework for a new type of machine learning model that is more predictive and explainable than current state-of-the-art data-driven feature selection methods. | ['the N3C consortium', 'Jimeng Sun', 'Adam Cross', 'Sara Warfield', 'Mary Stapel', 'Elise Albers', 'Scott Barrows', 'George Heintz', 'Chaoqi Yang', 'Junyi Gao'] | 2022-07-25 | null | null | null | null | ['severity-prediction'] | ['computer-vision'] | [ 1.57276139e-01 1.93428442e-01 -4.03932005e-01 -4.42255318e-01
-6.95587456e-01 -4.62228477e-01 2.48137377e-02 1.28612673e+00
-3.07649076e-01 6.06508076e-01 9.51027811e-01 -4.01413053e-01
-6.71342790e-01 -8.36174369e-01 -3.86120319e-01 -2.15758905e-01
-5.12541175e-01 1.17207181e+00 -4.77504104e-01 1.33859590e-01
-4.04277891e-01 3.25277179e-01 -9.21134949e-01 4.41908687e-01
1.03202498e+00 4.79435146e-01 -2.86342025e-01 7.03046679e-01
4.24869806e-01 7.63585865e-01 -2.38238990e-01 -1.89378992e-01
3.28666269e-04 -3.95809352e-01 -7.45682001e-01 -4.20270681e-01
-3.01669142e-03 -3.19885820e-01 -7.68233761e-02 2.64487565e-01
6.05357289e-01 -1.80089518e-01 1.11810827e+00 -9.91659343e-01
-4.43786353e-01 7.79714525e-01 -2.15539202e-01 3.75139594e-01
5.05679846e-01 1.21517301e-01 1.04616868e+00 -4.41312730e-01
8.58183563e-01 8.61483097e-01 1.12675929e+00 5.25802374e-01
-1.36836338e+00 -5.01003325e-01 4.37766276e-02 -2.15016052e-01
-1.08050585e+00 2.95475990e-01 -2.53590569e-02 -9.81510460e-01
1.35746992e+00 3.18261117e-01 9.84400094e-01 7.87443399e-01
1.69477820e-01 2.32155532e-01 5.89088798e-01 1.40165523e-01
-2.02741604e-02 -3.88296187e-01 2.71006376e-01 8.11034083e-01
6.53035700e-01 1.73783228e-02 -7.12012053e-02 -1.00569141e+00
2.73080885e-01 7.93602586e-01 -2.36445397e-01 3.14632840e-02
-1.26688588e+00 1.07575774e+00 2.01699302e-01 5.84835559e-02
-7.56381929e-01 -2.95826554e-01 2.88161606e-01 4.96189892e-02
5.68711817e-01 7.88189054e-01 -1.06674981e+00 -5.58639057e-02
-6.75545335e-01 3.27753901e-01 7.93118954e-01 3.54305089e-01
1.26572922e-01 -1.64153904e-01 -2.61969507e-01 8.29853714e-01
2.26657450e-01 7.27772415e-01 1.79786041e-01 -4.68633503e-01
6.01585805e-01 9.30133581e-01 -2.68500984e-01 -8.05862129e-01
-1.31585801e+00 -5.81713378e-01 -9.34147179e-01 -7.11889863e-01
-4.27874140e-02 -7.61026740e-01 -9.04011965e-01 1.88910890e+00
4.95273381e-01 1.97098687e-01 1.05827376e-01 4.78846341e-01
9.73214269e-01 5.06631732e-01 3.66861969e-01 -2.19257593e-01
1.31022215e+00 -4.84179676e-01 -2.06587061e-01 2.30861425e-01
1.18909693e+00 -3.24173629e-01 2.27066636e-01 3.03074151e-01
-9.32301044e-01 2.21773103e-01 -4.82892185e-01 5.18699408e-01
-4.42500353e-01 -4.14908439e-01 6.85563266e-01 4.07778114e-01
-8.34802032e-01 5.28283834e-01 -7.72607923e-01 -7.62046516e-01
4.80166614e-01 4.57899570e-01 -4.82566714e-01 -3.49740475e-01
-1.24679351e+00 9.52019811e-01 4.33888257e-01 -4.67009515e-01
-7.27471232e-01 -1.39290357e+00 -8.60261440e-01 9.50576514e-02
-6.19483255e-02 -1.37745011e+00 6.52146161e-01 -2.15461418e-01
-2.72029102e-01 7.83004284e-01 7.19084889e-02 -3.95379394e-01
1.06252477e-01 -3.97990607e-02 -6.07041180e-01 1.87175542e-01
2.07068622e-01 3.66892338e-01 1.25074670e-01 -6.73470557e-01
-7.91845202e-01 -4.63404357e-01 -5.40875375e-01 2.34472200e-01
-3.25345874e-01 2.42756575e-01 -5.72932400e-02 -8.04530442e-01
-2.23439559e-01 -9.84057903e-01 -5.30476689e-01 -7.35461175e-01
-4.16880399e-01 -2.00140730e-01 2.71374106e-01 -8.32602084e-01
1.24535584e+00 -1.73715961e+00 8.88842046e-02 4.21870917e-01
7.31709242e-01 2.17180386e-01 -1.65450349e-01 7.06127942e-01
-3.38643253e-01 3.12794834e-01 -2.76129574e-01 1.10597275e-02
-5.44930041e-01 -6.62666047e-04 2.26515368e-01 4.19131786e-01
3.42600524e-01 8.98682535e-01 -1.01541364e+00 -2.59388745e-01
1.50249019e-01 6.65862083e-01 -1.21071684e+00 5.77604890e-01
-2.72289496e-02 6.18207753e-01 -4.92481828e-01 7.29962289e-01
2.04339162e-01 -8.47541749e-01 4.00327086e-01 3.27618361e-01
1.63137972e-01 2.14012325e-01 -4.18538988e-01 1.17970192e+00
-1.47242369e-02 -1.49221271e-01 -1.03067137e-01 -7.54565775e-01
4.75652486e-01 5.24840176e-01 1.25238216e+00 2.42882892e-02
6.50196820e-02 -6.29720688e-02 6.03408553e-02 -7.01644421e-01
-2.89700687e-01 -6.56318739e-02 -8.25938955e-02 5.95236838e-01
-2.77243525e-01 -1.78553592e-02 -7.87211359e-02 2.74052918e-01
1.67276609e+00 -5.78306258e-01 5.22589624e-01 -3.78880888e-01
-3.57547365e-02 3.36976290e-01 9.73984778e-01 6.70437455e-01
1.67460740e-01 6.80371583e-01 3.56355488e-01 -9.04292703e-01
-6.21671557e-01 -1.17316604e+00 -3.95763248e-01 8.55402291e-01
-7.63006270e-01 -4.07297194e-01 -4.02296603e-01 -6.72347069e-01
3.88867259e-01 5.88073313e-01 -7.82908022e-01 -1.07522413e-01
-2.73604721e-01 -1.29595935e+00 6.81943357e-01 4.38512772e-01
-3.29916984e-01 -8.02866936e-01 -5.05149126e-01 4.95373785e-01
-2.67172605e-01 -7.91738153e-01 -5.15480638e-01 8.57423246e-02
-7.47033119e-01 -1.55527008e+00 -6.45871103e-01 -6.46918893e-01
6.65087223e-01 -4.97309387e-01 1.23099780e+00 3.56015921e-01
-7.20568955e-01 5.77268839e-01 -3.74645352e-01 -7.23383129e-01
-2.77577579e-01 3.10861915e-01 4.03610975e-01 -4.08820957e-01
5.54278433e-01 -3.50428641e-01 -9.33861732e-01 -2.04507232e-01
-8.08824182e-01 4.28483598e-02 2.35733077e-01 7.52615094e-01
5.25006771e-01 -2.85377562e-01 8.62299979e-01 -1.11830354e+00
4.44291085e-01 -1.33542538e+00 -3.16244513e-01 1.51123270e-01
-1.27230942e+00 -2.27253556e-01 5.32948375e-01 -5.48823178e-02
-4.39455986e-01 -1.53640330e-01 -1.32036239e-01 -4.08795960e-02
-2.54220307e-01 8.64387691e-01 6.05148017e-01 6.33446217e-01
4.32458431e-01 -2.60862112e-01 -1.59501940e-01 -5.19505918e-01
-1.20667569e-01 6.23559356e-01 1.06746107e-01 -3.12031746e-01
5.98180950e-01 1.68642879e-01 2.47038856e-01 -6.42183959e-01
-9.03697729e-01 -6.49998605e-01 -3.39181781e-01 2.10466206e-01
1.46694803e+00 -1.14735699e+00 -8.32563996e-01 2.68111557e-01
-8.54026675e-01 -2.23278105e-01 4.02998403e-02 9.10857737e-01
-1.52534932e-01 -1.22381739e-01 -8.04289520e-01 -3.07442397e-01
-8.56728613e-01 -1.06450737e+00 7.28778541e-01 -1.04829393e-01
-6.99177742e-01 -1.35491657e+00 8.48922491e-01 6.00524306e-01
3.22069675e-01 9.17409480e-01 1.74002481e+00 -1.27949691e+00
-8.47136304e-02 -4.81224544e-02 -2.78975874e-01 -5.55815287e-02
4.21772718e-01 1.14150532e-01 -4.49015886e-01 -5.81027389e-01
-6.00395977e-01 -1.75830260e-01 8.66714060e-01 6.94897473e-01
1.02622914e+00 -4.34651464e-01 -5.66433668e-01 9.45069134e-01
1.42156577e+00 3.89194459e-01 -1.44835487e-01 -1.75862238e-01
9.10666406e-01 7.94298828e-01 1.89980641e-01 6.94529474e-01
1.05806077e+00 3.74594867e-01 2.66422331e-01 -4.10162747e-01
3.55310351e-01 -3.82848620e-01 -3.15367281e-01 8.77724886e-01
-2.69540846e-01 -2.79211432e-01 -1.76954198e+00 8.04187477e-01
-1.72175121e+00 -7.29254544e-01 -1.49570927e-01 2.10214591e+00
7.98600197e-01 -2.37401441e-01 2.54447341e-01 -3.89681816e-01
4.81066287e-01 -1.31796107e-01 -5.29263854e-01 -7.31918991e-01
6.45771995e-02 1.65939048e-01 3.41756135e-01 1.39318347e-01
-9.82307792e-01 4.29295570e-01 7.31457138e+00 1.19400956e-01
-8.39950621e-01 4.10028473e-02 1.00395119e+00 -4.17574972e-01
-3.91702354e-01 -3.81850630e-01 -5.35550952e-01 3.69166225e-01
1.17375290e+00 3.23388763e-02 4.57306564e-01 5.21578968e-01
3.33982319e-01 3.67335856e-01 -1.24956703e+00 6.54332995e-01
6.12862073e-02 -1.58438277e+00 1.50326844e-02 2.66184006e-02
1.20891213e+00 7.55496621e-01 -4.19399068e-02 9.30049866e-02
7.67499268e-01 -1.42285597e+00 -1.24185167e-01 6.52081966e-01
1.25114381e+00 -8.21344733e-01 7.81391799e-01 1.30888864e-01
-9.87199008e-01 -1.87727079e-01 1.35231882e-01 9.09933671e-02
1.62654832e-01 7.39967823e-01 -1.58516359e+00 4.88225013e-01
8.63450646e-01 7.15636730e-01 -1.86630830e-01 8.49145114e-01
2.88659006e-01 1.00624037e+00 -2.49258474e-01 1.39275208e-01
1.25802204e-01 1.11867987e-01 3.27932030e-01 1.39155602e+00
2.37212911e-01 6.48233831e-01 1.17539316e-01 5.05279005e-01
-1.98256269e-01 4.59426761e-01 -8.71860147e-01 -3.94239604e-01
2.85017669e-01 1.02680707e+00 -4.30329353e-01 -4.34208453e-01
-5.90737939e-01 2.91979373e-01 3.46494526e-01 2.88343817e-01
-5.89637041e-01 -1.94496602e-01 8.91591787e-01 2.74931312e-01
6.46501854e-02 3.92648369e-01 -2.86626905e-01 -1.08046710e+00
-8.40895414e-01 -9.40174758e-01 1.07354820e+00 -3.52755010e-01
-1.50909162e+00 4.47787464e-01 -5.76731972e-02 -7.65154421e-01
-6.47156715e-01 -1.55910775e-01 -5.26009619e-01 8.57400715e-01
-1.35541463e+00 -8.84753883e-01 7.05664456e-02 5.60190916e-01
-3.47749814e-02 -1.24514423e-01 1.26759720e+00 2.14326501e-01
-6.81409895e-01 4.78776127e-01 6.95558414e-02 2.85700083e-01
5.56696653e-01 -1.03268313e+00 2.14773104e-01 3.23930532e-01
-4.06134516e-01 6.77310526e-01 2.87140369e-01 -1.06428695e+00
-1.07411253e+00 -1.53979051e+00 1.45571661e+00 -8.30122411e-01
3.92096341e-01 3.00660711e-02 -6.72704518e-01 8.09833825e-01
-2.10665494e-01 -4.42145914e-01 1.38340580e+00 2.95706987e-01
-3.95494103e-01 2.35262454e-01 -1.49841154e+00 2.07852170e-01
1.01891589e+00 -3.21868718e-01 -7.99069524e-01 5.89530945e-01
1.06483984e+00 -5.31763360e-02 -1.52951968e+00 9.51941729e-01
5.11376321e-01 -6.17387950e-01 9.57961202e-01 -1.36110973e+00
6.50285363e-01 1.56762585e-01 1.82462689e-02 -1.49770916e+00
-6.45585954e-01 -3.59469503e-01 1.02880247e-01 5.93350887e-01
1.02595973e+00 -1.02513433e+00 5.57236850e-01 8.77541780e-01
1.34355769e-01 -1.32297683e+00 -5.57621360e-01 -1.94944650e-01
1.78675383e-01 -2.02482536e-01 1.05462980e+00 1.38679874e+00
1.33568227e-01 6.65212274e-02 -1.55877501e-01 4.08139944e-01
4.82083589e-01 2.71793008e-01 2.64449894e-01 -1.49184740e+00
-4.13752645e-01 -3.07984471e-01 -2.60845721e-01 -4.13604155e-02
-3.00947666e-01 -1.15770841e+00 -6.22741878e-01 -2.19993758e+00
8.39747310e-01 -6.40386462e-01 -8.20635855e-01 8.77247095e-01
-3.26166213e-01 -1.40894786e-01 -6.55810982e-02 -5.50899245e-02
-3.71279091e-01 -3.45123932e-02 7.57646382e-01 -8.13201442e-02
-3.68695468e-01 1.04955666e-01 -6.52479947e-01 6.23702943e-01
7.35350966e-01 -6.90971613e-01 -4.05708820e-01 -5.72853446e-01
3.71420711e-01 4.90261167e-01 -2.80852467e-02 -6.78577185e-01
-1.66428506e-01 -5.12063861e-01 5.62527776e-01 -6.66330338e-01
6.31922856e-02 -6.99754477e-01 4.90166932e-01 1.15326262e+00
-3.79992068e-01 6.13424242e-01 7.27836117e-02 3.02700043e-01
1.59463704e-01 4.82418925e-01 4.18973386e-01 1.13879144e-01
1.67528331e-01 8.13653529e-01 -5.55520177e-01 5.65106034e-01
9.86027122e-01 2.52804995e-01 -5.63911915e-01 -2.53341943e-01
-6.67717636e-01 5.21313608e-01 3.98678660e-01 4.60122824e-01
5.49005866e-01 -8.56405258e-01 -1.17618585e+00 3.62422764e-01
3.39985281e-01 -2.12821038e-03 4.49654281e-01 1.05136991e+00
-6.66976154e-01 7.97454298e-01 1.61240175e-01 -3.72873783e-01
-1.31484210e+00 7.20170081e-01 1.07388012e-01 -4.95317549e-01
-5.50101221e-01 8.36142659e-01 1.78531185e-01 -6.18881345e-01
4.85166758e-02 -3.97107184e-01 -2.57872075e-01 1.13883503e-01
5.24548650e-01 5.75003505e-01 -1.50383621e-01 -5.74418426e-01
-7.38976002e-01 5.11717319e-01 -2.79718246e-02 4.24546123e-01
1.91850328e+00 2.93511212e-01 -2.90394008e-01 5.36265485e-02
1.30319798e+00 -3.85684893e-02 -5.66430211e-01 5.61432466e-02
-8.96883085e-02 1.21783309e-01 -3.60675126e-01 -1.11372519e+00
-9.58116114e-01 3.76384228e-01 3.82012695e-01 1.60325035e-01
1.15488756e+00 1.46952823e-01 7.44830310e-01 7.39619434e-02
9.01044086e-02 -6.85707808e-01 -4.42540467e-01 3.91436815e-01
3.96411240e-01 -1.22613311e+00 -1.43763702e-02 -1.18447766e-01
-6.27262414e-01 5.53753912e-01 3.02552044e-01 1.87076405e-01
9.46840942e-01 2.25130692e-01 1.60545900e-01 -5.89455187e-01
-1.25416815e+00 1.06503867e-01 4.67371315e-01 7.16527879e-01
4.63116050e-01 6.97874010e-01 -3.91904324e-01 7.44180501e-01
-4.68174331e-02 -1.01778612e-01 2.40891173e-01 6.10190809e-01
-2.18883321e-01 -1.03313911e+00 -1.46577835e-01 1.34947288e+00
-8.97253215e-01 -5.42479038e-01 -2.73432642e-01 3.43603134e-01
4.30177242e-01 1.08550549e+00 -4.73941863e-02 -4.87538815e-01
1.94273308e-01 1.88694149e-01 -1.07891202e-01 -9.16721880e-01
-9.11731243e-01 -2.85802782e-01 3.24522406e-01 -5.45934856e-01
-2.41905391e-01 -8.37997913e-01 -1.62909126e+00 -2.48568520e-01
1.77290872e-01 1.48889557e-01 4.28139120e-01 7.28711545e-01
7.49464214e-01 5.37916005e-01 5.58946133e-01 -1.57005787e-02
-2.44326979e-01 -5.61366677e-01 -3.90194416e-01 4.29283559e-01
5.03095508e-01 -2.58087546e-01 -2.47151837e-01 -3.98511201e-01] | [7.966822147369385, 6.135936737060547] |
bf624395-2f6d-4431-88ff-cc201b027333 | detecting-anomalous-microflows-in-iot | 2304.04987 | null | https://arxiv.org/abs/2304.04987v1 | https://arxiv.org/pdf/2304.04987v1.pdf | Detecting Anomalous Microflows in IoT Volumetric Attacks via Dynamic Monitoring of MUD Activity | IoT networks are increasingly becoming target of sophisticated new cyber-attacks. Anomaly-based detection methods are promising in finding new attacks, but there are certain practical challenges like false-positive alarms, hard to explain, and difficult to scale cost-effectively. The IETF recent standard called Manufacturer Usage Description (MUD) seems promising to limit the attack surface on IoT devices by formally specifying their intended network behavior. In this paper, we use SDN to enforce and monitor the expected behaviors of each IoT device, and train one-class classifier models to detect volumetric attacks. Our specific contributions are fourfold. (1) We develop a multi-level inferencing model to dynamically detect anomalous patterns in network activity of MUD-compliant traffic flows via SDN telemetry, followed by packet inspection of anomalous flows. This provides enhanced fine-grained visibility into distributed and direct attacks, allowing us to precisely isolate volumetric attacks with microflow (5-tuple) resolution. (2) We collect traffic traces (benign and a variety of volumetric attacks) from network behavior of IoT devices in our lab, generate labeled datasets, and make them available to the public. (3) We prototype a full working system (modules are released as open-source), demonstrates its efficacy in detecting volumetric attacks on several consumer IoT devices with high accuracy while maintaining low false positives, and provides insights into cost and performance of our system. (4) We demonstrate how our models scale in environments with a large number of connected IoTs (with datasets collected from a network of IP cameras in our university campus) by considering various training strategies (per device unit versus per device type), and balancing the accuracy of prediction against the cost of models in terms of size and training time. | ['Vijay Sivaraman', 'Gustavo Batista', 'Theophilus A. Benson', 'Hassan Habibi Gharakheili', 'Ayyoob Hamza'] | 2023-04-11 | null | null | null | null | ['one-class-classifier'] | ['methodology'] | [ 1.46650165e-01 -1.42902091e-01 -4.63351429e-01 -2.08706871e-01
-2.13088766e-01 -1.03556406e+00 2.11353943e-01 7.60658011e-02
1.11893274e-01 4.64107037e-01 -1.55105636e-01 -1.07091057e+00
-2.82907933e-01 -8.57692242e-01 -3.79894465e-01 -8.29805061e-02
-4.70521808e-01 5.08290708e-01 6.47161841e-01 3.03160846e-01
6.68300837e-02 9.37125206e-01 -1.39136922e+00 2.47886091e-01
3.03786725e-01 1.45006955e+00 -8.93297732e-01 7.75838017e-01
7.16688111e-02 7.93842196e-01 -1.03620994e+00 -4.44306374e-01
6.66377068e-01 8.61562416e-02 -7.33322084e-01 2.13955730e-01
2.87536502e-01 -6.29419684e-01 -2.21540302e-01 8.31679404e-01
2.13298112e-01 -5.48322856e-01 9.88819730e-03 -2.07605052e+00
1.14404313e-01 8.72774780e-01 -3.40942472e-01 7.34846830e-01
2.58209109e-01 7.32832611e-01 9.64639843e-01 1.50348902e-01
3.02434593e-01 1.12242723e+00 5.34519851e-01 2.90835023e-01
-1.36509323e+00 -1.26853669e+00 1.33175597e-01 7.51127163e-03
-8.67755413e-01 -4.99069571e-01 5.83325267e-01 -2.90580601e-01
1.02942383e+00 4.81947809e-01 4.40205246e-01 1.76837003e+00
-8.28299448e-02 1.15654290e-01 6.34403527e-01 2.08545968e-01
5.96756637e-01 6.78040758e-02 3.23256820e-01 3.58299822e-01
7.94750690e-01 1.75172240e-01 2.68672258e-02 -8.98601294e-01
5.13779163e-01 3.50604892e-01 6.65187389e-02 -8.41798037e-02
-1.12333095e+00 4.80541706e-01 -1.12329967e-01 1.74676716e-01
-3.96275431e-01 1.88389093e-01 8.31765950e-01 4.85315442e-01
1.29443243e-01 4.56364900e-01 -1.06483829e+00 -6.56675577e-01
-5.03824174e-01 -3.58669341e-01 1.41467500e+00 1.01102829e+00
5.14400959e-01 3.25267971e-01 3.37632298e-01 8.09314251e-02
2.74243623e-01 7.43962705e-01 2.63756420e-02 -1.19237244e+00
3.72880876e-01 6.12803638e-01 -1.29657239e-01 -8.94323409e-01
-4.55802351e-01 -3.75242919e-01 -4.86679643e-01 8.41699541e-02
4.44346160e-01 -4.68037844e-01 -3.19645286e-01 1.56353414e+00
1.69071794e-01 1.01612139e+00 -2.54566878e-01 2.22415090e-01
-5.29604182e-02 2.49306247e-01 3.60117584e-01 -3.55975062e-01
1.26880217e+00 -2.65755087e-01 -3.58788878e-01 2.18095947e-02
7.20831037e-01 -4.75733966e-01 7.93469250e-01 5.15304506e-01
-7.89735973e-01 -2.16115445e-01 -8.06089759e-01 8.95973146e-01
-5.97536087e-01 -6.50122523e-01 6.20783150e-01 1.59238410e+00
-6.20633066e-01 3.01098257e-01 -1.08345520e+00 -5.04713893e-01
7.94760764e-01 6.40584290e-01 1.27059609e-01 2.54655957e-01
-7.75113344e-01 2.74242312e-01 2.55707413e-01 -5.77657878e-01
-9.39612210e-01 -1.11429834e+00 -5.70802808e-01 6.73861131e-02
4.20489550e-01 -4.18610841e-01 1.08725262e+00 -2.93797195e-01
-1.35944164e+00 2.60814786e-01 2.26380005e-01 -5.92561066e-01
1.97365090e-01 2.01476291e-01 -1.33347082e+00 2.13072106e-01
1.15318649e-01 -2.07375102e-02 6.62016571e-01 -9.71390188e-01
-7.36467123e-01 -1.97666392e-01 3.39578182e-01 -9.88098323e-01
-7.02107370e-01 4.03909713e-01 3.82131457e-01 -1.80515066e-01
-2.41714999e-01 -7.42715061e-01 -1.16963752e-01 -1.63870662e-01
-5.43621004e-01 -1.19155701e-02 1.71815407e+00 1.92431305e-02
1.19360960e+00 -2.07954311e+00 -9.59020138e-01 6.85824573e-01
4.10840273e-01 6.36256158e-01 -1.08033754e-01 2.39507526e-01
-3.88960205e-02 7.30423868e-01 2.26444289e-01 -1.03659935e-01
1.17279887e-01 4.54424888e-01 -7.71532774e-01 2.67194301e-01
1.03360727e-01 6.72896206e-01 -9.46835637e-01 -2.20072642e-01
6.61996782e-01 1.52159885e-01 -6.99179351e-01 8.35169181e-02
-3.80639225e-01 6.66855335e-01 -6.82432413e-01 1.08392215e+00
6.44007504e-01 -5.02777517e-01 3.20013314e-01 -3.89439285e-01
-6.62527978e-02 6.03754342e-01 -1.22654343e+00 7.54084289e-01
-5.24489403e-01 2.50258565e-01 6.66249245e-02 -7.35879123e-01
6.41662598e-01 4.01351392e-01 9.18621302e-01 -4.92616028e-01
3.14614713e-01 1.55858204e-01 -4.85808924e-02 -4.59677368e-01
-4.92273629e-01 4.61496025e-01 -2.09791988e-01 1.07811296e+00
-2.50878215e-01 3.37270528e-01 2.78120339e-01 -3.51433791e-02
2.07044411e+00 -7.02991903e-01 2.18747109e-01 2.91229505e-02
4.85847652e-01 -1.36240706e-01 6.25077546e-01 1.00662076e+00
-6.92849338e-01 -3.18536371e-01 8.65182281e-01 -7.77304113e-01
-8.19800735e-01 -1.52065754e+00 -8.14285800e-02 7.77103722e-01
4.24515717e-02 -5.30662715e-01 -3.55010450e-01 -1.15899146e+00
5.81937991e-02 7.92525291e-01 -1.05145112e-01 -4.10572952e-03
-8.97603095e-01 -7.89933741e-01 1.15683377e+00 4.39918488e-01
6.84166193e-01 -8.71567309e-01 -5.98300517e-01 1.92804977e-01
1.74550358e-02 -2.03906608e+00 -1.22007668e-01 3.28976437e-02
-8.62661242e-01 -1.52622867e+00 8.86167526e-01 -9.82063040e-02
3.60691518e-01 6.17273003e-02 1.13211465e+00 1.11701131e-01
-5.61604321e-01 7.56956577e-01 -1.44292504e-01 -3.51221800e-01
-7.15163291e-01 -5.02799898e-02 4.39137697e-01 8.48082602e-02
7.63862550e-01 -1.24654126e+00 -4.02208716e-01 8.27443421e-01
-8.34393024e-01 -1.16216385e+00 3.44697088e-01 -9.26005468e-02
2.11321160e-01 5.24540246e-01 5.01250148e-01 -8.43623519e-01
5.64723909e-01 -8.55365813e-01 -9.26762640e-01 -1.98219150e-01
-8.00110400e-01 -2.69991875e-01 9.48373973e-01 -6.43795252e-01
-4.32681918e-01 -4.42203820e-01 -4.90626507e-02 -4.85982597e-01
-8.30780745e-01 -3.94579381e-01 -3.52110147e-01 -2.53954142e-01
7.12218046e-01 -4.01312828e-01 -2.49675706e-01 -2.39137322e-01
1.71412323e-02 7.27452815e-01 1.44552618e-01 -7.43724763e-01
1.11397266e+00 8.18937778e-01 1.97747067e-01 -8.18090320e-01
-7.16837943e-01 -2.81841367e-01 -2.82413840e-01 -4.35846373e-02
3.72760057e-01 -4.47921097e-01 -1.59975386e+00 9.56632048e-02
-9.20725524e-01 -1.77189291e-01 -6.29830062e-01 3.69010270e-01
-3.22881341e-02 5.15522897e-01 -9.56774116e-01 -6.63578331e-01
-2.33799487e-01 -1.13559842e+00 7.76225269e-01 -9.27335620e-02
-4.73571151e-01 -1.13791525e+00 -1.82483196e-01 3.68863165e-01
6.97619915e-01 3.33952844e-01 9.21384454e-01 -1.30693829e+00
-7.46584117e-01 -1.79247439e-01 -5.44720232e-01 4.03304607e-01
3.49850923e-01 2.63895601e-01 -1.00285220e+00 -2.89971203e-01
1.45440415e-01 4.45205644e-02 -1.28983095e-01 3.79363373e-02
1.63779843e+00 -6.30496919e-01 -3.20160776e-01 6.63944185e-01
1.18700683e+00 3.29207867e-01 5.67345023e-01 2.83299714e-01
6.93238020e-01 2.14397863e-01 2.66937558e-02 7.30690420e-01
-6.57653511e-02 3.65704238e-01 9.28006589e-01 2.94033110e-01
5.83988950e-02 -1.16284646e-01 5.28626204e-01 2.53155500e-01
2.77033150e-01 -4.59225148e-01 -7.21988380e-01 1.60056874e-01
-1.29292858e+00 -1.11960900e+00 -3.43063653e-01 1.98207843e+00
-4.84957658e-02 7.75419235e-01 7.15893626e-01 5.40283918e-01
7.15802550e-01 9.34165530e-03 -7.21818268e-01 -6.52018607e-01
1.72300860e-01 5.81087232e-01 8.95063460e-01 1.53643996e-01
-8.93461585e-01 5.37415326e-01 6.75787401e+00 3.04755330e-01
-1.01152956e+00 1.73401505e-01 3.98372620e-01 4.47785370e-02
1.69523984e-01 6.78657219e-02 -7.07960308e-01 8.47884953e-01
1.78596449e+00 1.73101425e-02 4.70796883e-01 1.02387190e+00
5.86436749e-01 6.79007232e-01 -1.11141527e+00 8.53203773e-01
-4.75343019e-01 -1.16288376e+00 1.78990106e-03 6.41135931e-01
3.50618005e-01 3.51182520e-01 -6.46846928e-03 1.98019639e-01
7.14911878e-01 -6.09228730e-01 1.57273978e-01 -1.50289774e-01
5.14585972e-01 -6.54542267e-01 6.74457967e-01 1.44992312e-02
-1.25004506e+00 -7.90621877e-01 2.47501016e-01 -1.60612956e-01
1.90707088e-01 8.41764629e-01 -9.03744340e-01 1.92150682e-01
7.18296707e-01 7.25618184e-01 -3.06254596e-01 7.47396886e-01
-5.02702892e-02 1.18727374e+00 -6.63999259e-01 1.86803594e-01
-1.11403145e-01 1.50619879e-01 8.66581202e-01 1.04090571e+00
1.93709314e-01 -2.25999191e-01 4.18194741e-01 1.07840383e+00
-1.54661193e-01 -5.42932034e-01 -4.11123812e-01 -8.06725696e-02
9.39901769e-01 1.25474131e+00 -7.28856325e-01 -3.68901081e-02
-4.65922028e-01 5.02982140e-01 -5.32412648e-01 2.72488922e-01
-9.60118234e-01 -2.51957119e-01 1.40733707e+00 5.83108246e-01
2.26774693e-01 -6.60134032e-02 -4.64330643e-01 -1.11443782e+00
-1.02488354e-01 -9.73836005e-01 7.21707821e-01 -3.26616496e-01
-1.67223263e+00 4.56437171e-01 -2.49299482e-02 -1.48733377e+00
-1.80928782e-01 -7.98414588e-01 -8.16783845e-01 9.48405117e-02
-1.12422371e+00 -8.88901889e-01 -3.22318196e-01 9.46343780e-01
1.44253746e-01 -9.07698572e-02 9.59069252e-01 7.54029393e-01
-8.38227212e-01 6.35755956e-01 -6.26356602e-01 3.50889146e-01
2.55355716e-01 -7.68375456e-01 6.63410842e-01 9.87572849e-01
-9.54542682e-03 4.73190278e-01 5.26401818e-01 -6.59155488e-01
-1.20065582e+00 -1.29681408e+00 2.81915635e-01 -8.00148368e-01
1.41304851e+00 -4.48145211e-01 -3.47752392e-01 1.13747454e+00
-3.61908138e-01 5.68041205e-01 7.59346247e-01 -1.22035630e-01
-7.64133155e-01 -4.87788022e-01 -1.70195520e+00 3.76626730e-01
1.29616189e+00 -4.89380926e-01 4.54388484e-02 2.69900262e-01
8.90675247e-01 2.40672335e-01 -1.14772606e+00 5.56965828e-01
3.93010318e-01 -1.28676558e+00 8.86433244e-01 -7.01363385e-01
-7.29365826e-01 -4.46841538e-01 -1.70507029e-01 -6.62586212e-01
4.53475892e-04 -1.08902657e+00 -6.65935099e-01 1.32450008e+00
2.87320375e-01 -1.40663528e+00 8.88655365e-01 4.90130424e-01
1.93376318e-01 -2.72232831e-01 -7.78667867e-01 -1.08100307e+00
-6.03387415e-01 -9.04150426e-01 1.13556659e+00 1.08379781e+00
1.96196549e-02 1.68188050e-01 7.66547844e-02 6.85631156e-01
8.20041180e-01 -1.76420078e-01 7.93224573e-01 -1.73790658e+00
-3.13928783e-01 -4.67850953e-01 -9.61929321e-01 -6.94542766e-01
1.02488138e-01 -3.74928474e-01 -8.97190928e-01 -5.16213596e-01
-2.95772702e-01 -6.48831069e-01 -3.69204909e-01 6.80325687e-01
8.29750478e-01 2.56558806e-01 -2.48829067e-01 2.83215672e-01
-6.89093232e-01 -3.57084244e-01 3.57271403e-01 -8.73942673e-02
-2.30700091e-01 5.34227133e-01 -5.66939235e-01 7.48871982e-01
1.28778422e+00 -6.01515293e-01 -6.32341027e-01 -1.27354696e-01
-2.42990758e-02 -1.72464356e-01 7.52748311e-01 -1.24893725e+00
-5.94357364e-02 -1.45505697e-01 1.49677262e-01 -3.55072707e-01
1.11735277e-01 -1.37506926e+00 1.21338658e-01 6.57832265e-01
5.88452034e-02 3.00882459e-01 1.22530028e-01 6.88109457e-01
3.14740837e-01 3.84716898e-01 5.94138563e-01 1.24629505e-01
-3.09923291e-01 8.04012775e-01 -4.74411130e-01 2.95960695e-01
1.27588701e+00 -3.05927813e-01 -8.50540876e-01 -3.71472031e-01
-6.55739188e-01 -9.92513224e-02 4.41027761e-01 4.06493098e-01
2.22428635e-01 -8.23256493e-01 -8.35470930e-02 6.87271535e-01
1.41556755e-01 -6.30635083e-01 -3.07772756e-01 8.20940018e-01
-3.51238191e-01 2.62376845e-01 -1.72040775e-01 -7.00112879e-01
-1.01003075e+00 7.24562526e-01 3.56690109e-01 -4.41806823e-01
-4.77223188e-01 1.42777011e-01 -5.34852743e-01 -4.29256558e-01
4.39787030e-01 -4.00850832e-01 3.86306673e-01 -4.45733637e-01
7.23490179e-01 9.58818078e-01 2.35252231e-01 -3.06693733e-01
-6.18974984e-01 1.89538538e-01 1.65828064e-01 2.94846743e-01
1.03925788e+00 -1.35545418e-01 -1.90872550e-01 9.79178622e-02
1.25457406e+00 2.97433734e-01 -8.27721059e-01 3.39077041e-02
2.04709679e-01 -4.58282560e-01 -2.33819768e-01 -7.09617138e-01
-1.37724602e+00 5.11247039e-01 6.11838520e-01 1.07933521e+00
1.13374543e+00 -1.77541620e-03 1.38669193e+00 3.15622926e-01
6.22777224e-01 -4.29595321e-01 1.65454835e-01 2.83285141e-01
-5.37388682e-01 -8.69625032e-01 -3.19088727e-01 -8.51313293e-01
2.79231071e-02 1.15871477e+00 6.91483617e-01 -4.04361151e-02
1.18914044e+00 7.99837232e-01 -6.04368635e-02 -3.76132578e-01
-7.57336140e-01 3.54799442e-02 -6.90030396e-01 8.36857855e-01
-2.65620172e-01 2.12760329e-01 4.85841811e-01 2.51862764e-01
-9.41752940e-02 -1.55536413e-01 8.35204840e-01 8.11299741e-01
-2.38148883e-01 -1.46418941e+00 -1.88554242e-01 7.02887714e-01
-7.87569046e-01 1.87377572e-01 -2.46744901e-01 6.82436049e-01
2.32305750e-01 1.73325360e+00 2.46955171e-01 -9.07868862e-01
3.31687301e-01 -8.54386687e-02 -1.00099564e-01 -4.11983222e-01
-6.14233375e-01 -5.42457461e-01 3.30717593e-01 -1.13899708e+00
3.17918137e-02 -5.34148693e-01 -9.49813664e-01 -9.11165059e-01
4.63388162e-03 1.77119318e-02 6.37394786e-01 9.28569317e-01
7.75412202e-01 4.49652493e-01 1.04902780e+00 -2.04625949e-01
-5.49853683e-01 -5.10499716e-01 -4.74136353e-01 5.97196102e-01
3.30779582e-01 -5.53061247e-01 -1.21236324e+00 -2.61791974e-01] | [5.182248592376709, 7.218328952789307] |
4ea75edc-04bc-4fb2-aeb7-e25a9e8c3a45 | congruity-of-genomic-and-epidemiological-data | 2210.01956 | null | https://arxiv.org/abs/2210.01956v2 | https://arxiv.org/pdf/2210.01956v2.pdf | Congruity of genomic and epidemiological data in modeling of local cholera outbreaks | Cholera continues to be a global health threat. Understanding how cholera spreads between locations is fundamental to the rational, evidence-based design of intervention and control efforts. Traditionally, cholera transmission models have utilized cholera case count data. More recently, whole genome sequence data has qualitatively described cholera transmission. Integrating these data streams may provide much more accurate models of cholera spread, however no systematic analyses have been performed so far to compare traditional case-count models to the phylodynamic models from genomic data for cholera transmission. Here, we use high-fidelity case count and whole genome sequencing data from the 1991-1998 cholera epidemic in Argentina to directly compare the epidemiological model parameters estimated from these two data sources. We find that phylodynamic methods applied to cholera genomics data provide comparable estimates that are in line with established methods. Our methodology represents a critical step in building a framework for integrating case-count and genomic data sources for cholera epidemiology and other bacterial pathogens. | ['Andrey Y. Lokhov', 'Daryl Domman', 'Ethan Romero-Severson', 'Josefina Campos', 'Carrie Manore', 'Jeffrey Keithley', 'Lauren Castro', 'Mateusz Wilinski'] | 2022-10-04 | null | null | null | null | ['epidemiology'] | ['medical'] | [ 3.80389720e-01 -5.18408120e-01 -5.22194169e-02 -2.13611946e-01
-7.97095835e-01 -6.42338395e-01 3.93810451e-01 1.02520573e+00
-5.34494638e-01 5.74879169e-01 4.21246201e-01 -9.54907894e-01
-4.43704903e-01 -8.74805391e-01 -5.90638995e-01 -9.57865536e-01
-6.16244674e-01 7.77566910e-01 -1.64022818e-01 -3.60922575e-01
9.77859497e-02 2.48591349e-01 -9.54731762e-01 4.35287766e-02
1.03862202e+00 -2.64082968e-01 6.09784782e-01 1.21664178e+00
-1.30044967e-01 8.11272264e-02 -5.74514627e-01 1.46536678e-01
-1.65915281e-01 -7.02556789e-01 -2.54486769e-01 -6.67727828e-01
-4.59938943e-01 -5.18964648e-01 7.24310754e-03 5.94700694e-01
6.04435265e-01 -6.62467897e-01 8.13206971e-01 -9.40115988e-01
-4.13624257e-01 4.47295964e-01 -4.62398291e-01 3.63035291e-01
5.96729279e-01 3.04796159e-01 3.32339436e-01 -7.83922747e-02
9.91133273e-01 1.34343100e+00 1.18478322e+00 1.34150498e-02
-1.40741324e+00 -4.10473824e-01 -3.72852027e-01 -4.54543903e-02
-1.25828075e+00 -1.04127206e-01 1.06125429e-01 -9.15239751e-01
1.32408404e+00 1.52500153e-01 1.25962007e+00 7.79042900e-01
8.61789823e-01 2.04123527e-01 9.57472265e-01 -1.52968779e-01
6.30976707e-02 -3.10767621e-01 8.45360570e-03 2.63962358e-01
9.72083807e-01 3.77373606e-01 8.36544409e-02 -1.06593037e+00
6.68513715e-01 6.01385415e-01 -2.37641871e-01 -1.05575517e-01
-9.42629933e-01 1.39692473e+00 1.21930316e-01 2.43716747e-01
-6.66894674e-01 2.67547071e-01 5.13297319e-01 4.12476093e-01
5.01906753e-01 1.64476737e-01 -8.57001543e-01 -9.98743922e-02
-3.65164816e-01 3.76353055e-01 8.43931317e-01 5.91613829e-01
5.67886651e-01 -1.86542034e-01 5.49083352e-01 6.51132822e-01
7.74898708e-01 1.21092474e+00 -2.52543002e-01 -4.09174591e-01
-1.64932907e-02 4.45113271e-01 1.52315140e-01 -8.73910010e-01
-7.44299471e-01 2.21762136e-01 -3.68307203e-01 -3.11408371e-01
6.89775050e-01 -3.58943701e-01 -9.61980045e-01 1.58032572e+00
4.61153120e-01 1.49200723e-01 3.92425776e-01 3.67715567e-01
5.46951950e-01 7.84513712e-01 4.60489780e-01 -3.79490346e-01
1.76425993e+00 1.58031672e-01 -4.61932838e-01 2.59616613e-01
1.04717815e+00 -9.11327422e-01 4.75397408e-01 -6.88296929e-02
-3.44102144e-01 2.13295192e-01 -6.69271588e-01 4.20353949e-01
-2.51481026e-01 -8.17600369e-01 4.57658261e-01 9.68051851e-01
-1.07601202e+00 -1.49230942e-01 -1.40643084e+00 -1.19014311e+00
1.61157891e-01 2.64351845e-01 -2.24418759e-01 -5.00719845e-01
-9.53941107e-01 7.25805819e-01 1.58064768e-01 1.16984714e-02
-9.38258350e-01 -7.92757452e-01 -7.85763621e-01 -2.37242997e-01
5.05600497e-02 -7.76302457e-01 9.19877231e-01 -6.29189089e-02
-8.30951154e-01 5.01928389e-01 -3.18157732e-01 -1.65276811e-01
8.20794851e-02 4.78386767e-02 -1.14046946e-01 1.23991095e-01
1.93715289e-01 2.54751235e-01 -2.13407740e-01 -1.26021242e+00
-6.51987076e-01 -7.75863647e-01 -5.33328533e-01 -1.37460679e-01
4.85363841e-01 5.49386442e-01 1.41061500e-01 -4.64555889e-01
-1.71868116e-01 -9.98717427e-01 -7.28621602e-01 -5.07790267e-01
2.49739125e-01 -1.91171959e-01 4.64354843e-01 -9.70931351e-01
7.71524191e-01 -1.84836566e+00 -3.15448791e-01 2.83812266e-02
1.53243646e-01 9.15006399e-02 -1.89945057e-01 1.33334231e+00
5.87249279e-01 3.18565071e-01 -5.90725780e-01 6.47276819e-01
-4.23366249e-01 3.00478041e-01 2.74765920e-02 8.61965060e-01
5.32633245e-01 6.41512632e-01 -1.35694027e+00 -1.16060384e-01
2.12002113e-01 1.02983844e+00 -9.92794573e-01 2.63929635e-01
-5.78190498e-02 6.03578150e-01 -7.14300752e-01 4.59659129e-01
9.04282749e-01 -9.99153480e-02 8.78994167e-01 4.66751426e-01
-3.36641818e-01 2.65001535e-01 -4.19657677e-01 1.28151131e+00
-9.05186832e-02 2.83681840e-01 3.09118778e-01 -6.45246267e-01
7.24626660e-01 5.54384708e-01 6.19939923e-01 -6.79807425e-01
4.07903939e-02 3.51458818e-01 5.40215671e-01 -8.45587671e-01
1.41491801e-01 -3.10499489e-01 -6.60026222e-02 3.72720093e-01
-3.07833940e-01 -2.38385022e-01 1.21865511e-01 -6.88474206e-03
1.03081894e+00 2.22457677e-01 4.91902828e-01 -6.75569773e-01
-1.16872810e-01 1.01771867e+00 6.91750467e-01 8.93700004e-01
-1.26202479e-01 2.71267265e-01 4.58051324e-01 -4.64110583e-01
-1.36081016e+00 -1.06679761e+00 -7.36642301e-01 9.64788616e-01
-1.91807941e-01 -2.10191667e-01 -5.62606454e-01 3.64721030e-01
2.85903662e-02 2.18700886e-01 -5.55092752e-01 1.24185100e-01
-8.12598526e-01 -1.85884500e+00 1.03295279e+00 2.48014852e-02
-1.57591447e-01 -6.97342753e-01 -9.66397703e-01 8.92512977e-01
-1.88340232e-01 -4.19980079e-01 -2.82890387e-02 3.93350512e-01
-8.22375298e-01 -1.77237940e+00 -6.39279366e-01 -8.10364425e-01
3.83564532e-01 2.12587565e-01 9.04457331e-01 5.01819491e-01
-7.74456739e-01 7.32556954e-02 -4.38383728e-01 -7.33208120e-01
-1.03996611e+00 -2.11877882e-01 -6.02001883e-02 -9.46815968e-01
9.67769563e-01 5.88415749e-03 -8.78491819e-01 4.08294022e-01
-9.92242455e-01 -4.94703442e-01 3.61854255e-01 8.00023317e-01
2.90264279e-01 -3.88215005e-01 1.09684551e+00 -6.95090353e-01
5.81145883e-01 -1.32237232e+00 -7.06679046e-01 2.60875344e-01
-3.88922751e-01 -3.33146155e-01 2.23547250e-01 -2.85227865e-01
-1.16026056e+00 -1.83881879e-01 -5.97732961e-01 8.81481469e-01
-2.22935468e-01 8.93007576e-01 5.16329527e-01 6.46637321e-01
5.53649247e-01 -1.91532057e-02 4.15351540e-01 -4.40316409e-01
-4.07395475e-02 9.07950282e-01 1.66474711e-02 -4.07565296e-01
8.57923031e-02 5.34353852e-01 -3.08683097e-01 -1.28996611e+00
2.48576790e-01 -8.98024201e-01 -3.24506581e-01 3.11853468e-01
1.22682095e+00 -1.14941287e+00 -1.19142270e+00 6.75649107e-01
-1.01028776e+00 -7.12718844e-01 4.79879886e-01 9.63637948e-01
-4.02563035e-01 -7.96235800e-02 -1.06756699e+00 -9.17108655e-01
-1.87306516e-02 -1.47379136e+00 1.11876845e+00 -1.26080617e-01
-5.87010495e-02 -1.44754672e+00 1.16492236e+00 -1.73873156e-02
6.12390280e-01 7.06723928e-01 1.03798199e+00 -5.81209958e-01
-3.53326172e-01 2.17443213e-01 -1.20413870e-01 -5.67864120e-01
5.82734227e-01 4.10734743e-01 -6.13442481e-01 -5.40103555e-01
2.11207271e-01 -1.16389528e-01 8.82693052e-01 1.02322459e+00
-3.39137942e-01 -2.97672629e-01 -7.81819165e-01 6.36332095e-01
1.70928085e+00 9.91371393e-01 4.73516375e-01 1.89121529e-01
5.49322009e-01 1.03040695e+00 2.52307534e-01 2.98195064e-01
7.39349723e-01 4.05896246e-01 -1.54769391e-01 -3.46967839e-02
3.40036035e-01 -1.10083818e-01 8.28400105e-02 8.30207586e-01
-1.27793029e-01 -5.02072334e-01 -1.56355798e+00 7.00773180e-01
-1.35624111e+00 -1.26181257e+00 -6.16343737e-01 2.01836157e+00
8.52737188e-01 -5.03966391e-01 2.07112759e-01 -4.43136841e-01
3.68946820e-01 -3.64300311e-01 -3.45758379e-01 -5.72851956e-01
-1.17447607e-01 -1.04603060e-01 6.55421913e-01 6.37929857e-01
-5.54160893e-01 3.62382323e-01 7.90899086e+00 2.75849421e-02
-9.57887053e-01 9.34803858e-02 5.23547828e-01 2.83201188e-01
-5.82530260e-01 6.49999306e-02 -4.46257114e-01 1.12027479e-02
1.46116471e+00 -1.53895959e-01 2.60915458e-01 1.31477147e-01
6.03513300e-01 -2.74191707e-01 -7.19226122e-01 1.64345592e-01
-2.67367989e-01 -1.27460909e+00 -4.13360476e-01 4.26862478e-01
6.99510992e-01 7.24820971e-01 -3.88000578e-01 -2.18449295e-01
1.04976249e+00 -9.72320735e-01 5.09899855e-02 1.46620676e-01
5.96078515e-01 -8.14173281e-01 1.09095669e+00 7.37902746e-02
-1.08255208e+00 2.14473516e-01 -5.36823988e-01 8.30956642e-03
6.59237087e-01 2.64728338e-01 -1.59120977e+00 4.23675627e-01
8.15838814e-01 3.41041237e-01 -8.71010125e-02 1.04650092e+00
3.65269363e-01 9.30788279e-01 -5.04641771e-01 -2.00923625e-02
1.78337589e-01 -2.86043853e-01 2.81079113e-01 1.39129388e+00
4.54293519e-01 1.90724835e-01 8.69613960e-02 4.79585588e-01
5.31012475e-01 1.59299091e-01 -8.52361202e-01 -2.30740666e-01
5.69038749e-01 3.76098126e-01 -9.80387032e-01 -2.66321272e-01
-6.21095657e-01 3.30351233e-01 -2.06011370e-01 5.53522646e-01
-1.28586397e-01 -3.64745021e-01 1.10348129e+00 -9.96654481e-02
2.19781399e-01 -2.42970273e-01 4.47676450e-01 -6.72873974e-01
-7.45933056e-01 -8.83499205e-01 5.80780327e-01 -1.71718895e-01
-1.41876173e+00 3.19568992e-01 5.22964001e-01 -2.23535493e-01
-6.77912056e-01 -4.92930353e-01 -4.74449337e-01 1.07307148e+00
-1.31770658e+00 -7.85741925e-01 9.89216194e-02 5.55623583e-02
3.26269299e-01 6.10834420e-01 8.43629241e-01 1.18261971e-01
-2.43050650e-01 -6.35266164e-03 1.13231540e+00 -2.00034194e-02
2.68129647e-01 -8.44187915e-01 7.50530303e-01 1.89443320e-01
-5.69498479e-01 1.07629323e+00 9.05901670e-01 -1.36459363e+00
-1.70576966e+00 -9.72319007e-01 6.79268122e-01 -5.83489835e-01
6.07430398e-01 -4.31223840e-01 -1.08749378e+00 9.28782463e-01
3.23915362e-01 -8.77151489e-01 1.00590503e+00 -6.23426214e-02
-1.82833910e-01 5.71528494e-01 -1.36026406e+00 2.77367651e-01
6.39986515e-01 -3.12935561e-01 -6.04355872e-01 1.58727303e-01
1.07646608e+00 1.23287857e-01 -1.22336960e+00 2.17946097e-01
9.21218276e-01 -6.67229533e-01 1.09668505e+00 -5.43385923e-01
-1.98028520e-01 -3.53970647e-01 -2.87548691e-01 -1.39273143e+00
-1.53144136e-01 -2.64811367e-01 1.22164142e+00 7.79225767e-01
3.15834552e-01 -1.12417948e+00 2.21277088e-01 -2.57074207e-01
-2.04248041e-01 -2.90843189e-01 -7.63249218e-01 -6.16679311e-01
5.05780876e-01 -1.90606471e-02 1.04689562e+00 1.01075709e+00
-1.72199205e-01 -1.48955062e-01 -1.58411860e-02 4.28564042e-01
7.13871062e-01 -1.95609137e-01 7.04191983e-01 -1.30830371e+00
-9.78303701e-02 -2.38059327e-01 -3.68433058e-01 -2.48004094e-01
-5.80597341e-01 -5.62694192e-01 3.35742414e-01 -1.90809381e+00
7.52959788e-01 -8.73419702e-01 8.33035558e-02 1.35515556e-01
-4.12777185e-01 1.94651455e-01 -3.24358940e-01 1.54893994e-01
3.68066967e-01 8.43948647e-02 7.28130698e-01 1.67121395e-01
-7.30197310e-01 -5.41670084e-01 -4.80716616e-01 4.47491229e-01
1.00984800e+00 -9.58321750e-01 -4.49817091e-01 -4.94264901e-01
4.71743673e-01 2.96870887e-01 1.57809541e-01 -1.19942993e-01
-1.66869968e-01 -7.44064510e-01 8.76262933e-02 -6.88901186e-01
-2.49459341e-01 -5.97836792e-01 9.90813553e-01 1.61335540e+00
1.80572122e-01 5.34197509e-01 4.83286530e-01 6.97121739e-01
1.60708264e-01 2.10261256e-01 2.02325016e-01 -1.11924224e-01
-5.84562682e-03 7.35854506e-02 -1.06499922e+00 1.45315021e-01
7.66599774e-01 -1.06654912e-01 -8.24334264e-01 7.30056018e-02
-2.05474734e-01 3.21037054e-01 8.77910972e-01 2.13613853e-01
3.37589502e-01 -7.70943761e-01 -1.51182902e+00 6.24427907e-02
3.07479858e-01 -2.58475304e-01 4.33725089e-01 9.99396026e-01
-1.39241421e+00 7.79476643e-01 -6.49621785e-02 -1.02837777e+00
-1.03498816e+00 1.03217506e+00 2.27202103e-01 -3.98886167e-02
-5.76025903e-01 2.46319577e-01 5.78848779e-01 -8.84022057e-01
-4.50996965e-01 -2.81186163e-01 -1.04599386e-01 -1.21286418e-02
6.53099716e-01 3.91932815e-01 -5.24893641e-01 -7.82691538e-01
-7.54512668e-01 5.14890432e-01 2.90314138e-01 -4.06326652e-02
1.71593893e+00 -3.01840127e-01 -2.16352403e-01 5.70131838e-01
1.13754845e+00 -1.07756868e-01 -1.01358330e+00 2.89896250e-01
1.39110297e-01 -2.71873742e-01 -6.41196311e-01 -6.22497797e-01
-2.56676376e-01 6.19354844e-01 7.13950217e-01 2.90690958e-01
7.21981585e-01 2.11727843e-01 3.10601562e-01 2.18883783e-01
3.92920583e-01 -2.44357139e-01 -5.59758902e-01 -3.76347452e-03
5.86251378e-01 -1.15977931e+00 -3.08659345e-01 -3.63503665e-01
9.31265429e-02 5.52422643e-01 -1.88618243e-01 8.46605897e-02
5.34859300e-01 5.03595352e-01 4.63606358e-01 -6.28381670e-01
-8.96898150e-01 1.36390239e-01 -4.06343222e-01 1.02087891e+00
7.49666870e-01 5.48306108e-01 -6.64518476e-01 -2.06894666e-01
-3.23205553e-02 4.76781577e-02 6.86886430e-01 1.10578871e+00
-6.35136247e-01 -1.26304138e+00 -9.15343761e-01 7.55741596e-01
-7.49798775e-01 -4.70451355e-01 -1.03051186e-01 9.96945798e-01
-2.30151489e-01 1.42970395e+00 4.97671038e-01 2.33237639e-01
-2.37028468e-02 9.64310169e-02 3.72488618e-01 -3.43613923e-01
-9.27904546e-01 5.44828296e-01 2.22570866e-01 8.48755613e-02
-6.14442527e-01 -9.98421788e-01 -1.54557681e+00 -9.02146220e-01
-3.11829150e-01 3.18939716e-01 9.82402027e-01 5.60266614e-01
4.33083564e-01 3.80590081e-01 3.86617750e-01 -4.71010357e-01
-4.95529883e-02 -9.54912722e-01 -3.77547055e-01 3.21973413e-01
5.23322463e-01 -1.21953152e-01 -2.26039603e-01 6.80667087e-02] | [5.62645149230957, 4.507153511047363] |
41acca22-2a60-4b8a-bb0d-856680f53cb4 | data-adaptive-discriminative-feature | 2211.10061 | null | https://arxiv.org/abs/2211.10061v1 | https://arxiv.org/pdf/2211.10061v1.pdf | Data-Adaptive Discriminative Feature Localization with Statistically Guaranteed Interpretation | In explainable artificial intelligence, discriminative feature localization is critical to reveal a blackbox model's decision-making process from raw data to prediction. In this article, we use two real datasets, the MNIST handwritten digits and MIT-BIH Electrocardiogram (ECG) signals, to motivate key characteristics of discriminative features, namely adaptiveness, predictive importance and effectiveness. Then, we develop a localization framework based on adversarial attacks to effectively localize discriminative features. In contrast to existing heuristic methods, we also provide a statistically guaranteed interpretability of the localized features by measuring a generalized partial $R^2$. We apply the proposed method to the MNIST dataset and the MIT-BIH dataset with a convolutional auto-encoder. In the first, the compact image regions localized by the proposed method are visually appealing. Similarly, in the second, the identified ECG features are biologically plausible and consistent with cardiac electrophysiological principles while locating subtle anomalies in a QRS complex that may not be discernible by the naked eye. Overall, the proposed method compares favorably with state-of-the-art competitors. Accompanying this paper is a Python library dnn-locate (https://dnn-locate.readthedocs.io/en/latest/) that implements the proposed approach. | ['Wei Pan', 'Chunlin Li', 'Lin Yee Chen', 'Xiaotong Shen', 'Ben Dai'] | 2022-11-18 | null | null | null | null | ['visual-localization'] | ['computer-vision'] | [ 3.78814429e-01 3.59372526e-01 1.41786799e-01 -4.05614316e-01
-7.30275273e-01 -7.39241719e-01 2.07751170e-01 8.18173289e-02
-2.68486198e-02 1.01491117e+00 -1.41677976e-01 -2.73918450e-01
-4.45276856e-01 -3.50000858e-01 -7.25130975e-01 -8.40386450e-01
-3.25427055e-01 -1.10918619e-01 -3.72174859e-01 1.75148025e-01
4.41953063e-01 6.13945901e-01 -1.04950166e+00 2.75201052e-01
6.56435549e-01 1.44251251e+00 1.67334285e-02 8.18400145e-01
6.46410227e-01 8.93839121e-01 -6.31170928e-01 -3.26407164e-01
3.73627305e-01 -7.85282493e-01 -7.29879618e-01 -1.93253011e-01
2.82885849e-01 -1.33692175e-01 -6.16820812e-01 1.00520682e+00
7.16340959e-01 -2.06547126e-01 6.53279543e-01 -1.28013921e+00
-9.52319324e-01 5.07796645e-01 -2.64949203e-01 5.97738445e-01
1.57323584e-01 2.54721075e-01 1.10223150e+00 -8.82663131e-01
5.93303740e-01 8.74308884e-01 8.02118659e-01 7.10371733e-01
-1.22790110e+00 -4.94951665e-01 -8.77243727e-02 3.26288283e-01
-1.60907912e+00 -3.98500383e-01 1.15285540e+00 -4.80556756e-01
5.12458265e-01 6.01929009e-01 5.58599055e-01 1.19325757e+00
6.06114089e-01 7.25065768e-01 1.11056757e+00 -2.75151014e-01
3.90293658e-01 1.28206596e-01 -7.04199215e-03 8.64966571e-01
1.76032111e-01 3.94790411e-01 -5.41594863e-01 -2.94047505e-01
9.26462233e-01 2.31323376e-01 -5.29297829e-01 -3.47662657e-01
-1.39550281e+00 6.20690823e-01 3.92705917e-01 8.60039368e-02
-4.86052811e-01 4.65347379e-01 3.62470269e-01 2.64011323e-01
2.08618045e-02 6.61928058e-01 -6.03654802e-01 -7.66494945e-02
-7.78002739e-01 1.73939914e-01 4.60216850e-01 8.51055622e-01
5.43333471e-01 2.91702271e-01 -5.13534173e-02 3.09266537e-01
2.24363923e-01 3.53863239e-01 4.65995193e-01 -9.95951056e-01
1.02657311e-01 4.13640440e-01 -9.55047384e-02 -1.16234791e+00
-5.16549587e-01 -7.59400606e-01 -1.13531613e+00 1.38266861e-01
2.46818975e-01 -2.81602353e-01 -5.77803612e-01 1.62848365e+00
-9.17747896e-03 2.49779731e-01 1.72167301e-01 1.01514316e+00
8.89133513e-01 2.96253234e-01 -1.89077005e-01 -7.53214136e-02
1.18629193e+00 -5.63720703e-01 -5.13631284e-01 -1.68260783e-02
3.52466673e-01 -4.06402498e-01 8.24085236e-01 3.86366695e-01
-9.08092022e-01 -7.61674285e-01 -1.22296894e+00 6.98456541e-02
-1.46588221e-01 3.28408569e-01 7.57667363e-01 4.88293946e-01
-7.28951752e-01 9.64433014e-01 -8.25715780e-01 -1.92185342e-01
7.76802540e-01 2.51060158e-01 -3.63169611e-01 2.55529910e-01
-1.16689467e+00 7.03007996e-01 2.03501731e-01 2.41807401e-01
-1.17702317e+00 -5.79894781e-01 -7.35640764e-01 1.78434141e-02
-1.34104311e-01 -7.80037165e-01 7.88399041e-01 -1.01012623e+00
-1.12731040e+00 7.84739077e-01 -2.88945362e-02 -7.81197727e-01
5.90792477e-01 -1.02986008e-01 -3.32621962e-01 5.52290142e-01
-1.19946234e-01 7.57478714e-01 9.33082342e-01 -1.11071503e+00
-3.65439892e-01 -2.86660403e-01 1.06309094e-01 -1.90133438e-01
3.76965478e-02 -3.14119697e-01 1.18797980e-01 -1.06407034e+00
2.31972501e-01 -7.20198631e-01 -2.91177809e-01 3.54132205e-01
-7.44673669e-01 2.84155011e-01 7.59501517e-01 -6.21122539e-01
1.09658778e+00 -2.31069684e+00 6.87412033e-03 3.27421188e-01
6.16273999e-01 -1.86800197e-01 1.38833687e-01 2.24791333e-01
-3.21873546e-01 1.30021244e-01 -4.15348589e-01 4.12082896e-02
-4.87134270e-02 1.03631131e-01 -3.94751579e-01 9.10222054e-01
4.21910524e-01 1.17322195e+00 -6.99419320e-01 -4.80647475e-01
2.09851563e-01 5.01672447e-01 -3.25608820e-01 -6.54684603e-02
3.18008304e-01 7.97214746e-01 -6.15348399e-01 1.04033840e+00
4.05190140e-01 -3.72611612e-01 1.41588561e-02 -2.09576592e-01
2.81965345e-01 -2.91891396e-03 -9.67283189e-01 1.57098520e+00
1.46069322e-02 8.51545990e-01 -3.96740258e-01 -1.38051867e+00
1.12300050e+00 3.25371712e-01 4.41805184e-01 -4.97379690e-01
2.60546934e-02 1.47486702e-01 2.13779718e-01 -5.00491858e-01
-9.54536721e-02 3.96290980e-03 -1.32467104e-02 2.74081290e-01
4.69957814e-02 9.85740274e-02 -3.16929907e-01 5.16100191e-02
1.35938513e+00 1.10076793e-01 6.87318921e-01 -5.46007633e-01
4.37990397e-01 -2.17957497e-01 8.36449265e-01 9.06045616e-01
-4.99120474e-01 8.74008656e-01 6.97973251e-01 -8.01417112e-01
-1.06030834e+00 -1.12133825e+00 -5.56821764e-01 3.43842596e-01
2.85574317e-01 -2.40679175e-01 -7.48519421e-01 -8.90481830e-01
7.20741451e-02 5.56538284e-01 -9.62144196e-01 -3.68441910e-01
-2.89579391e-01 -4.64442641e-01 1.04148805e+00 8.28486800e-01
3.82175237e-01 -1.23762631e+00 -1.21960711e+00 1.65472835e-01
-5.03647067e-02 -7.91999042e-01 -1.81053653e-01 5.13287306e-01
-9.58578110e-01 -1.08450222e+00 -6.18174314e-01 -6.04886472e-01
7.18019545e-01 -4.06269908e-01 1.07215929e+00 4.23385091e-02
-9.31618989e-01 3.54622006e-01 -1.91400215e-01 -5.30075014e-01
-1.29466224e-02 -3.24048996e-01 1.68194726e-01 7.70767871e-03
1.18480682e-01 -7.37465739e-01 -1.05791795e+00 2.89687693e-01
-6.05907738e-01 8.79125223e-02 7.17537820e-01 1.15187013e+00
9.86533105e-01 -1.39362246e-01 7.74855494e-01 -7.22265840e-01
3.16052109e-01 -3.73825252e-01 -2.58783400e-01 1.28655091e-01
-6.39957190e-01 -6.88513145e-02 8.93655300e-01 -3.62129748e-01
-3.64089906e-01 3.43298107e-01 1.36870816e-01 -7.40477920e-01
-3.97535264e-01 4.08184201e-01 -3.06397825e-01 -9.80904996e-02
9.22027349e-01 4.91784722e-01 -2.05306023e-01 -2.44916841e-01
1.49092674e-01 4.86331910e-01 8.23377430e-01 -4.65838134e-01
6.71545506e-01 5.39799571e-01 1.90873042e-01 -5.76524138e-01
-4.91360098e-01 2.59508714e-02 -6.80566669e-01 -3.23010646e-02
5.27770519e-01 -6.63033366e-01 -7.12194264e-01 9.02674496e-02
-1.07929397e+00 4.27247398e-02 -6.67985380e-01 4.73185390e-01
-9.71506894e-01 3.86531234e-01 -4.69827831e-01 -7.50287950e-01
-5.19424081e-01 -9.90062714e-01 8.84076655e-01 1.82434484e-01
-3.94050717e-01 -1.04728949e+00 -1.86911121e-01 6.02465728e-03
1.89215660e-01 7.65073478e-01 9.12489176e-01 -9.36529994e-01
-3.97748113e-01 -3.52544785e-01 -8.08834136e-02 5.44118345e-01
1.68942958e-01 -1.37818009e-02 -1.39771509e+00 -1.01275988e-01
8.15267265e-02 -2.00892165e-01 7.61217475e-01 7.26830602e-01
1.62109864e+00 -4.12232399e-01 -3.65484744e-01 9.62184310e-01
1.19187284e+00 3.91445726e-01 7.52598047e-01 1.97249323e-01
4.43628848e-01 2.48502195e-01 5.14995158e-01 7.96119153e-01
1.50635205e-02 2.68084347e-01 5.59998751e-01 -3.95693868e-01
1.16313092e-01 -1.31629586e-01 1.42921314e-01 3.50973576e-01
-2.04591364e-01 -3.76540907e-02 -8.40041041e-01 4.01130199e-01
-1.87667263e+00 -9.51583505e-01 1.30261078e-01 1.82852721e+00
5.33056855e-01 1.30370826e-01 -2.23089874e-01 3.96478981e-01
6.88062012e-01 -7.69046769e-02 -8.80404830e-01 -3.48004222e-01
-4.44645584e-01 3.11272681e-01 3.55849326e-01 1.61131710e-01
-1.32422543e+00 4.61207688e-01 6.27238750e+00 6.15230203e-01
-1.31126308e+00 -1.44112572e-01 1.09118330e+00 9.94444266e-02
-2.13472962e-01 -3.65717411e-02 -2.63946533e-01 4.17915523e-01
7.93620467e-01 -3.24299484e-01 2.93755919e-01 1.00889564e+00
2.76540875e-01 3.67305607e-01 -1.55102313e+00 1.28468311e+00
5.64484745e-02 -1.60962486e+00 -9.75582898e-02 -9.43900719e-02
3.00325572e-01 -3.28137159e-01 3.90514344e-01 -1.00327909e-01
-3.58070970e-01 -1.51982176e+00 8.83017480e-01 8.33024442e-01
1.02954936e+00 -7.19151497e-01 7.37429380e-01 5.48830256e-02
-9.92392838e-01 -1.84014618e-01 -5.35047054e-01 -6.63177110e-03
-3.24764162e-01 6.13045633e-01 -7.03436792e-01 5.28040111e-01
7.29712129e-01 9.12765145e-01 -6.05922997e-01 1.00094175e+00
-2.75863469e-01 7.63435900e-01 -2.77750138e-02 1.83577448e-01
1.27163893e-02 1.60503566e-01 6.25952125e-01 1.26823366e+00
2.93460935e-01 1.63574249e-01 -1.73011959e-01 1.32453048e+00
-5.93122207e-02 -1.24182589e-01 -7.86304414e-01 -4.81532440e-02
3.79409701e-01 1.32694483e+00 -6.56262100e-01 -1.95208758e-01
-2.76242137e-01 1.15165460e+00 -7.65539855e-02 3.62415165e-01
-1.03611135e+00 -7.59592533e-01 4.58038360e-01 -6.09953096e-03
2.96764612e-01 2.08379343e-01 -6.67749047e-01 -1.04030085e+00
2.02857882e-01 -9.13350999e-01 3.14685971e-01 -7.79146731e-01
-1.31679523e+00 7.65597343e-01 -3.41211081e-01 -1.54030883e+00
-2.83760071e-01 -6.67642891e-01 -5.91316104e-01 8.56684148e-01
-1.23223317e+00 -1.03927863e+00 -2.73027837e-01 6.80368423e-01
3.77310276e-01 -2.49255374e-01 1.14010465e+00 9.31302682e-02
-4.09715444e-01 8.38742793e-01 1.92835212e-01 4.10769224e-01
4.03346568e-01 -1.30623269e+00 4.71674204e-01 7.57370174e-01
3.58294219e-01 7.37579882e-01 6.17406785e-01 -3.02913040e-01
-1.33411622e+00 -1.12407875e+00 4.09285158e-01 -4.86521751e-01
3.97204310e-01 -2.41114974e-01 -6.14066899e-01 6.70719445e-01
1.57220997e-02 6.82904899e-01 7.73144424e-01 -3.27916533e-01
-1.96044698e-01 -1.56741694e-01 -1.45984185e+00 3.28175485e-01
8.96916151e-01 -4.57506746e-01 -6.92493558e-01 1.61027968e-01
1.93021655e-01 -3.00366104e-01 -9.81625855e-01 4.69677091e-01
7.70785630e-01 -9.12664890e-01 9.00860965e-01 -7.07501113e-01
5.94192028e-01 -3.50798130e-01 -4.30516422e-01 -1.07184958e+00
-4.33172435e-01 -8.46122205e-01 -3.73302042e-01 8.97475719e-01
5.27299821e-01 -6.82563782e-01 6.49194360e-01 3.60855788e-01
-1.41240358e-01 -1.16049731e+00 -1.11808038e+00 -6.31088972e-01
-2.64417827e-02 -3.93386334e-01 3.40843558e-01 9.74071920e-01
2.13885084e-01 1.48625737e-02 -3.85741532e-01 3.77312750e-01
7.13318229e-01 2.45860904e-01 4.20288354e-01 -1.19412673e+00
-4.81126040e-01 -1.70774534e-01 -1.08403993e+00 -8.27024519e-01
-5.90628348e-02 -8.20859671e-01 -1.98516741e-01 -1.01806331e+00
3.00430000e-01 -2.93512523e-01 -9.01632786e-01 6.07230723e-01
-9.10909399e-02 3.32727909e-01 8.07112083e-02 2.26648390e-01
-4.64154959e-01 2.30179444e-01 1.10821676e+00 -2.17440918e-01
1.07156664e-01 1.41556573e-03 -9.96602178e-01 9.35543478e-01
8.76101732e-01 -5.56796491e-01 -2.71348208e-01 5.47291487e-02
-1.05786197e-01 1.68724209e-01 8.82497847e-01 -1.01490772e+00
-3.94878983e-02 6.31487072e-02 1.03161335e+00 -2.08282262e-01
2.42727995e-01 -7.29598999e-01 8.52960795e-02 6.61862731e-01
-6.06633604e-01 1.86137006e-01 1.48885489e-01 5.27122915e-01
-1.75029173e-01 -8.36710259e-02 9.17852640e-01 -1.03734143e-01
-5.48988640e-01 2.71089226e-01 -2.21738711e-01 1.33155018e-01
1.15681243e+00 -4.15976554e-01 -2.60747701e-01 -4.15992022e-01
-8.51664782e-01 -2.20461890e-01 2.30780393e-01 5.95165603e-02
1.16578615e+00 -1.13096070e+00 -7.25956142e-01 3.11890244e-01
1.57907501e-01 -2.65695304e-01 2.65658975e-01 1.01642406e+00
-7.09891975e-01 3.42893243e-01 -3.71642888e-01 -7.14339197e-01
-1.00326419e+00 6.55006826e-01 6.10382795e-01 1.67559505e-01
-9.71128643e-01 8.30394268e-01 4.30046648e-01 4.20176703e-03
2.46374577e-01 -5.18735230e-01 2.87460424e-02 -3.94657224e-01
4.50905442e-01 2.33502164e-01 -1.97930336e-01 -3.59918088e-01
-6.80829883e-01 4.32433397e-01 1.49301335e-01 1.78776547e-01
1.36309540e+00 -1.28650218e-01 2.42728993e-01 3.80319208e-01
1.03163290e+00 -7.46173263e-02 -1.31935668e+00 1.76554352e-01
-1.34845763e-01 -3.49758714e-01 -1.47177488e-01 -9.74585772e-01
-1.29328144e+00 1.06300128e+00 1.00534940e+00 9.88320038e-02
1.28383505e+00 -3.67347635e-02 3.45433414e-01 4.52328175e-01
1.56167850e-01 -7.50095606e-01 -1.14706546e-01 -7.50520080e-02
1.01040030e+00 -1.06306624e+00 -1.55616207e-02 -1.59622833e-01
-1.01930952e+00 1.32905841e+00 3.57010841e-01 -3.25216830e-01
6.84914887e-01 3.42372745e-01 2.11221755e-01 -2.29077175e-01
-7.09201813e-01 2.25507632e-01 3.93401146e-01 7.24097848e-01
3.32166523e-01 7.49156773e-02 -3.86489034e-02 1.12245977e+00
-2.26439327e-01 -9.16291401e-02 3.77718925e-01 8.54825079e-01
-1.11283027e-01 -4.71217483e-01 -2.48561651e-01 4.60825026e-01
-8.05827856e-01 -3.14510465e-01 -4.25759286e-01 7.87741721e-01
1.97296351e-01 6.52136803e-01 -1.08985260e-01 -5.14809906e-01
1.43341988e-01 1.14043571e-01 3.73962373e-01 -2.01395869e-01
-3.55440617e-01 1.96806081e-02 -2.48414040e-01 -6.11899495e-01
-9.69306156e-02 -6.08352661e-01 -1.37391555e+00 -4.12229784e-02
-1.11970395e-01 7.53933564e-02 4.66327041e-01 7.22609282e-01
5.26049495e-01 7.13456392e-01 6.70816004e-01 -6.37873650e-01
-5.60048640e-01 -8.35333109e-01 -8.65818322e-01 4.00186270e-01
5.08617163e-01 -5.27624130e-01 -4.28403467e-01 4.06329215e-01] | [14.276988983154297, 3.2860891819000244] |
adb0d989-4960-4ac6-8ae3-c704e7c32e15 | unconstrained-iris-segmentation-using | 1812.08245 | null | http://arxiv.org/abs/1812.08245v1 | http://arxiv.org/pdf/1812.08245v1.pdf | Unconstrained Iris Segmentation using Convolutional Neural Networks | The extraction of consistent and identifiable features from an image of the
human iris is known as iris recognition. Identifying which pixels belong to the
iris, known as segmentation, is the first stage of iris recognition. Errors in
segmentation propagate to later stages. Current segmentation approaches are
tuned to specific environments. We propose using a convolution neural network
for iris segmentation. Our algorithm is accurate when trained in a single
environment and tested in multiple environments. Our network builds on the Mask
R-CNN framework (He et al., ICCV 2017). Our approach segments faster than
previous approaches including the Mask R-CNN network. Our network is accurate
when trained on a single environment and tested with a different sensors
(either visible light or near-infrared). Its accuracy degrades when trained
with a visible light sensor and tested with a near-infrared sensor (and vice
versa). A small amount of retraining of the visible light model (using a few
samples from a near-infrared dataset) yields a tuned network accurate in both
settings. For training and testing, this work uses the Casia v4 Interval, Notre
Dame 0405, Ubiris v2, and IITD datasets. | ['Sohaib Ahmad', 'Benjamin Fuller'] | 2018-12-19 | null | null | null | null | ['iris-segmentation'] | ['medical'] | [ 4.61566150e-01 -2.47060969e-01 -2.47837707e-01 -5.14279366e-01
-3.58186036e-01 -7.07280815e-01 2.60817975e-01 -4.26804185e-01
-4.76495206e-01 1.60227597e-01 -9.27808881e-02 -4.88232553e-01
-1.56154425e-03 -4.75121319e-01 -5.51450491e-01 -6.49721324e-01
2.33641997e-01 2.90225297e-01 -8.68878961e-02 1.14918239e-01
1.71295583e-01 6.80449665e-01 -1.79568040e+00 2.44285583e-01
7.03441679e-01 8.12449753e-01 -5.10975361e-01 1.24466443e+00
3.11230600e-01 5.15602887e-01 -5.68884134e-01 -1.43646792e-01
9.91448343e-01 -4.60594028e-01 -9.46401536e-01 3.74742538e-01
1.24495542e+00 -5.03818810e-01 9.63486955e-02 1.07659042e+00
5.91332138e-01 -8.81964862e-02 3.71591859e-02 -6.60933912e-01
-7.27688730e-01 2.63399631e-01 -7.79261470e-01 1.60166666e-01
2.15883687e-01 7.58534849e-01 5.25649965e-01 -4.10573661e-01
3.02163363e-01 8.60095918e-01 8.63867760e-01 7.45195568e-01
-1.28481793e+00 -5.99136591e-01 -2.76176572e-01 -3.96472543e-01
-1.32657623e+00 -5.42445838e-01 1.84082478e-01 -4.54485685e-01
1.07820010e+00 4.88684833e-01 7.24106312e-01 7.67408073e-01
-1.16149217e-01 5.65837920e-01 1.65058613e+00 -5.36627412e-01
-2.98666824e-02 8.64406824e-02 9.92103294e-02 5.03480971e-01
2.05870360e-01 8.80917072e-01 -2.68006027e-01 8.39248598e-02
9.04608011e-01 3.58669013e-02 -2.13429376e-01 3.19323599e-01
-9.95852411e-01 3.13486636e-01 5.68775237e-01 1.48622975e-01
-4.51952852e-02 -1.46987587e-01 -1.07175179e-01 5.56841969e-01
6.17326677e-01 8.80709529e-01 -4.66953069e-01 -1.14373498e-01
-1.30550289e+00 -1.84375465e-01 5.09260237e-01 5.55372000e-01
8.50451410e-01 -3.16331953e-01 -3.15547794e-01 6.89665079e-01
3.20388794e-01 3.27364743e-01 1.82220787e-01 -7.68697679e-01
-4.18201387e-02 7.36479342e-01 2.07924500e-01 -3.90246451e-01
-7.41565526e-01 -7.36598313e-01 -4.84640867e-01 9.00206208e-01
8.57889235e-01 -5.13072729e-01 -1.82874048e+00 1.09240210e+00
4.69105959e-01 7.12197661e-01 -6.29766425e-03 1.16885197e+00
1.08099091e+00 -3.62229235e-02 -2.17926294e-01 2.28315845e-01
1.17476845e+00 -1.03734064e+00 -2.03526646e-01 -3.71659428e-01
4.24356908e-01 -1.14137256e+00 7.28852630e-01 6.61115766e-01
-9.39022124e-01 -7.23712087e-01 -9.50439036e-01 -3.29970479e-01
-4.27037537e-01 4.61901844e-01 5.00115871e-01 1.09706783e+00
-1.48335087e+00 6.57906473e-01 -8.46432030e-01 -5.08919716e-01
4.26794291e-01 8.94729435e-01 -2.22323060e-01 -4.02376242e-02
-6.47168279e-01 6.42485380e-01 1.02207102e-01 3.28088105e-01
-1.71602175e-01 -5.71819484e-01 -7.86531031e-01 -4.29971576e-01
2.74265576e-02 -4.29602891e-01 1.10459113e+00 -1.26756680e+00
-1.83338571e+00 1.54387927e+00 -1.38451010e-01 -4.46321189e-01
3.52016658e-01 1.63211524e-02 -5.10021210e-01 5.35506569e-02
-3.88748437e-01 6.16885662e-01 9.77189541e-01 -9.35087919e-01
-7.48250306e-01 -5.38663268e-01 2.12166086e-01 1.40974363e-02
1.89619914e-01 5.37570953e-01 -6.69837356e-01 -1.87577605e-01
1.40763313e-01 -1.18283355e+00 -1.97534040e-01 -5.33238836e-02
-6.86727762e-01 4.14561927e-02 3.79763633e-01 -7.25659490e-01
7.82275975e-01 -2.20793700e+00 -4.65229750e-01 5.13625801e-01
3.35241467e-01 7.61864603e-01 -2.36959204e-01 -2.14859620e-01
-4.40709651e-01 6.95584342e-02 1.95206925e-02 -4.95355308e-01
-4.21275795e-01 1.23005852e-01 1.25850141e-01 8.96610916e-01
1.12395532e-01 7.60923862e-01 -6.67604089e-01 -3.05184841e-01
4.61597919e-01 4.85078514e-01 -3.88197094e-01 -7.68180238e-03
5.84920757e-02 8.03202629e-01 1.16897210e-01 1.04835773e+00
8.99222016e-01 -4.26773459e-01 -4.76051092e-01 -2.77222469e-02
-4.00535464e-01 -6.21567406e-02 -1.16667306e+00 1.66202903e+00
-1.85684651e-01 6.95635796e-01 1.08956434e-01 -6.57238245e-01
9.42791462e-01 3.71389151e-01 2.55606771e-01 -6.54797971e-01
2.00268239e-01 2.43727982e-01 2.14165613e-01 -6.27478361e-01
2.00595304e-01 4.00932692e-02 6.69266284e-01 5.55977941e-01
-2.11680010e-01 1.07477419e-02 8.51167962e-02 -4.84018415e-01
9.19443965e-01 2.08949849e-01 5.78447953e-02 1.44033924e-01
1.79752246e-01 1.28288403e-01 5.08663416e-01 9.56113279e-01
-5.93496263e-01 9.36687350e-01 7.89940506e-02 -9.06940758e-01
-8.54307830e-01 -7.00809181e-01 -7.71779478e-01 7.97275066e-01
1.93131298e-01 3.09892558e-02 -6.82588935e-01 -4.93904710e-01
2.21644953e-01 1.12376206e-01 -8.95692110e-01 2.54328638e-01
-4.08128321e-01 -8.85587037e-01 4.64327782e-01 3.03335428e-01
6.23736560e-01 -9.73697484e-01 -6.72012985e-01 -1.60441801e-01
5.40938318e-01 -9.13999379e-01 -2.66080230e-01 -9.68868211e-02
-5.94939768e-01 -1.38880432e+00 -6.09261215e-01 -6.45445108e-01
7.49900579e-01 -2.60846806e-03 1.02388704e+00 4.04625267e-01
-8.45718145e-01 2.66417056e-01 5.05223218e-03 -6.64628744e-01
-2.09263563e-01 -1.48775354e-01 -4.42636311e-02 4.32724983e-01
1.10687411e+00 -5.78499250e-02 -8.51977408e-01 2.41599053e-01
-4.62725878e-01 5.89177720e-02 4.29945081e-01 7.87647069e-01
6.04174972e-01 -9.09743980e-02 -3.70878577e-01 -9.87931252e-01
1.37496501e-01 -1.14783756e-01 -9.72939312e-01 1.74681097e-01
-7.74716258e-01 -3.02340418e-01 3.07962835e-01 -4.08071429e-01
-7.34396100e-01 2.63538897e-01 3.21340933e-02 -5.36855161e-01
-6.58360481e-01 1.02126367e-01 5.76264203e-01 -6.30127072e-01
1.21236360e+00 -1.61752596e-01 2.08701000e-01 -6.35504007e-01
2.74578147e-02 1.10596371e+00 7.60937214e-01 -1.59235939e-01
7.91761100e-01 7.23533034e-01 1.14249796e-01 -5.68584502e-01
-9.58927453e-01 -8.08013499e-01 -6.94315493e-01 -1.27853841e-01
8.47858727e-01 -8.54375362e-01 -8.82162035e-01 8.91256928e-01
-5.52957296e-01 -5.50706983e-01 -3.64297062e-01 7.42356002e-01
-1.93856314e-01 9.59099364e-03 -5.77175021e-01 -5.46038568e-01
-2.77114540e-01 -1.45140791e+00 1.27789903e+00 9.45982873e-01
-8.85365829e-02 -8.49613190e-01 2.85260081e-01 7.11669028e-01
4.16555315e-01 3.76642644e-01 4.32988554e-02 -5.48507750e-01
-5.11797130e-01 -3.34624916e-01 -5.05047619e-01 3.56324136e-01
3.94614041e-01 6.20811522e-01 -1.56517649e+00 -5.12185812e-01
-2.08903715e-01 -2.94421375e-01 1.00966084e+00 8.10960710e-01
1.33900058e+00 -3.58264148e-02 -3.17532569e-01 1.45635319e+00
1.46030736e+00 1.76169127e-01 6.89182997e-01 4.80090052e-01
4.56838578e-01 4.91936564e-01 3.77965599e-01 -1.30563930e-01
-1.37098372e-01 4.95205939e-01 3.57392609e-01 -1.06828594e+00
-4.05282438e-01 2.75812000e-01 -2.75789887e-01 -2.24997655e-01
-4.18735832e-01 1.14844434e-01 -1.23840702e+00 6.55705392e-01
-1.44994497e+00 -9.19933379e-01 -1.63159221e-01 2.44163942e+00
1.26471615e+00 -2.31335402e-01 1.89169139e-01 -1.69670418e-01
6.44523919e-01 -3.05857748e-01 -8.16986203e-01 -5.79277158e-01
-3.54344100e-02 7.43634999e-01 8.20470273e-01 5.65362096e-01
-1.33844149e+00 8.62143397e-01 7.31828070e+00 1.59494922e-01
-1.71343994e+00 -2.48595566e-01 9.09745991e-01 -3.91643792e-01
3.06008548e-01 -5.87429702e-02 -7.66158581e-01 3.74944091e-01
9.78301942e-01 5.86874306e-01 8.21112931e-01 4.54852611e-01
3.77993286e-02 -3.90392810e-01 -1.05830777e+00 1.27858567e+00
8.94616023e-02 -1.18300200e+00 -8.54196429e-01 2.06713423e-01
9.30619955e-01 5.72848678e-01 5.59580564e-01 -2.50406265e-01
4.16614175e-01 -1.66796255e+00 -1.53092951e-01 6.27988160e-01
1.24792814e+00 -5.93588114e-01 6.91992819e-01 3.18808258e-02
-6.19382977e-01 -1.68704558e-02 -1.33584663e-01 2.34725643e-02
-6.15538657e-01 4.31158513e-01 -1.07047868e+00 1.26586378e-01
1.02776420e+00 9.26334560e-01 -8.65974844e-01 1.35969245e+00
-1.11915134e-01 8.32461059e-01 -4.53828394e-01 4.32269007e-01
3.21765766e-02 -3.91003877e-01 3.91825497e-01 9.35750544e-01
-7.79248253e-02 -7.25360140e-02 1.65356413e-01 1.03426445e+00
4.32256237e-03 -1.51428357e-01 -3.43001604e-01 5.74443862e-02
-2.52105538e-02 1.34419096e+00 -5.35989702e-01 -2.46049747e-01
-6.88830853e-01 7.49167621e-01 -2.65508872e-02 7.02112079e-01
-2.49853730e-01 -3.26316506e-01 9.02512372e-01 -5.46350814e-02
1.21946655e-01 2.23766714e-01 -7.10641742e-01 -8.51627588e-01
-1.29166260e-01 -1.05062437e+00 1.55794680e-01 -7.39473224e-01
-1.16929066e+00 4.39786315e-01 -4.22706395e-01 -1.29414499e+00
1.25005860e-02 -9.23890591e-01 -6.89058125e-01 1.60414851e+00
-1.82188725e+00 -1.28426611e+00 -6.30272031e-01 6.30678356e-01
-2.42479611e-02 -3.41278553e-01 7.96677113e-01 1.23602077e-01
-9.01191711e-01 8.17453027e-01 6.29745349e-02 6.31240964e-01
1.14927423e+00 -1.47769058e+00 6.13858104e-01 1.04291618e+00
7.26459622e-02 7.66632199e-01 4.80160177e-01 -4.90488708e-01
-1.06950855e+00 -1.12197077e+00 6.65358484e-01 -5.30055463e-01
1.53234094e-01 1.81788489e-01 -6.77607000e-01 9.35572863e-01
2.50010520e-01 5.04764855e-01 1.01554930e+00 4.08390492e-01
-2.51866490e-01 -1.34632930e-01 -1.47603369e+00 4.64422107e-01
7.38496363e-01 -7.30364025e-01 -1.30165890e-01 6.70785964e-01
2.47026965e-01 -1.27097344e+00 -9.45318043e-01 3.81171972e-01
6.24645412e-01 -1.26200891e+00 8.63968611e-01 -8.00959587e-01
1.99657932e-01 -4.95428830e-01 5.06424546e-01 -1.09763992e+00
-4.38848436e-02 -9.31709528e-01 2.86047041e-01 5.58568537e-01
4.59100276e-01 -9.69105184e-01 8.77734601e-01 9.38738167e-01
1.43854469e-01 -6.38614953e-01 -6.83773637e-01 -6.33018494e-01
-1.02757230e-01 -2.05168650e-01 8.74936521e-01 9.70043898e-01
-3.82394940e-01 -3.25117737e-01 -1.43228590e-01 5.84929109e-01
7.54957855e-01 4.14361358e-01 9.99608755e-01 -1.34484303e+00
-3.94940287e-01 -3.20210993e-01 -5.90069890e-01 -7.85151064e-01
-3.00535649e-01 -6.17250562e-01 -1.23567313e-01 -1.13424945e+00
-1.02658756e-01 -6.00322843e-01 -1.44799948e-01 9.78997707e-01
-2.04052925e-01 8.56020212e-01 -2.35634506e-01 2.67003179e-01
9.97454375e-02 -6.45370424e-01 1.35456002e+00 -2.43654907e-01
-5.76654673e-01 5.47547340e-01 -5.60398161e-01 5.01893938e-01
9.07667637e-01 -8.67528766e-02 -8.27516243e-02 -3.56810063e-01
2.10139871e-01 -1.64215252e-01 4.11523283e-01 -1.00693333e+00
4.92254496e-01 -1.20759495e-01 6.80027962e-01 -3.08962733e-01
2.11148098e-01 -7.63211489e-01 3.73513065e-02 3.49361628e-01
-3.25696021e-01 -4.86894369e-01 4.71885681e-01 3.29703023e-03
-7.29543120e-02 -7.58947507e-02 1.18247485e+00 -1.09820381e-01
-4.89346176e-01 4.56672549e-01 3.32343340e-01 -7.79209435e-02
8.85446012e-01 -7.48115778e-01 -6.14419758e-01 3.76693569e-02
-9.12426710e-01 2.65383959e-01 9.94466484e-01 4.07425582e-01
3.35873693e-01 -6.82559907e-01 -8.03323030e-01 1.07785809e+00
2.47800782e-01 1.45175099e-01 -4.87929061e-02 1.08572030e+00
-8.61144245e-01 2.21534416e-01 -8.61491859e-02 -1.12366974e+00
-1.78255928e+00 3.71029735e-01 1.14275014e+00 2.11249754e-01
-5.76657951e-01 1.22143281e+00 -2.59099066e-01 -3.96446019e-01
2.84351230e-01 -5.54639995e-01 -2.98157901e-01 -4.57737297e-01
1.01835239e+00 1.23735510e-01 3.05631787e-01 -4.53692287e-01
-9.20654014e-02 9.37539458e-01 -1.19442992e-01 1.98841840e-01
1.06526637e+00 9.66124311e-02 -4.61014122e-01 6.69974014e-02
9.17237520e-01 4.80364729e-03 -1.06330287e+00 -5.35231292e-01
-4.84930634e-01 -8.18422079e-01 4.86991197e-01 -1.15647650e+00
-1.24636197e+00 6.49177790e-01 1.15403092e+00 1.16957696e-02
1.43947184e+00 -3.29000950e-01 4.41706836e-01 -6.91548660e-02
-9.67973769e-02 -1.13520682e+00 -5.54062009e-01 1.23081744e-01
3.81713480e-01 -1.78154635e+00 5.38253263e-02 -8.72921869e-02
-3.22297156e-01 1.26859164e+00 7.33352423e-01 1.54131040e-01
7.35891163e-01 2.63588786e-01 7.77406573e-01 -3.79404724e-01
-1.32057875e-01 -5.59907258e-01 8.21886122e-01 6.16978943e-01
5.09931028e-01 1.37106478e-01 3.59009564e-01 -5.04594505e-01
-1.92773521e-01 2.77334064e-01 4.57338095e-01 7.65733898e-01
-9.37076285e-02 -1.00781572e+00 -6.76162720e-01 7.67058790e-01
-5.49788475e-01 -1.85198277e-01 -5.51237822e-01 5.05407631e-01
5.75809598e-01 1.26544058e+00 4.40243930e-01 -3.71101856e-01
-6.89119697e-02 -5.86059280e-02 4.00497854e-01 -9.80729043e-01
-1.05409539e+00 1.80224657e-01 -2.32851520e-01 -9.00604367e-01
-8.95467222e-01 -5.77600777e-01 -8.32288027e-01 -3.24004531e-01
-3.01818252e-01 -4.84547108e-01 7.06242442e-01 9.49540794e-01
3.72364372e-01 1.86546773e-01 7.32228041e-01 -4.71988559e-01
-3.49323273e-01 -9.37491596e-01 -9.05783772e-01 3.25443625e-01
1.15615451e+00 -1.27609029e-01 -5.07804930e-01 3.75671089e-02] | [3.74202036857605, -3.6325972080230713] |
41c7628d-82c0-40b8-9343-54ee67fcbe6e | a-mixed-hierarchical-attention-based-encoder | 1804.07790 | null | http://arxiv.org/abs/1804.07790v1 | http://arxiv.org/pdf/1804.07790v1.pdf | A Mixed Hierarchical Attention based Encoder-Decoder Approach for Standard Table Summarization | Structured data summarization involves generation of natural language
summaries from structured input data. In this work, we consider summarizing
structured data occurring in the form of tables as they are prevalent across a
wide variety of domains. We formulate the standard table summarization problem,
which deals with tables conforming to a single predefined schema. To this end,
we propose a mixed hierarchical attention based encoder-decoder model which is
able to leverage the structure in addition to the content of the tables. Our
experiments on the publicly available WEATHERGOV dataset show around 18 BLEU (~
30%) improvement over the current state-of-the-art. | ['Mitesh M. Khapra', 'Preksha Nema', 'Shreyas Shetty', 'Parag Jain', 'Karthik Sankaranarayanan', 'Anirban Laha'] | 2018-04-20 | a-mixed-hierarchical-attention-based-encoder-1 | https://aclanthology.org/N18-2098 | https://aclanthology.org/N18-2098.pdf | naacl-2018-6 | ['data-summarization'] | ['miscellaneous'] | [ 4.34185416e-01 6.12913013e-01 -2.72442222e-01 -4.17382747e-01
-1.31306016e+00 -6.66877627e-01 6.65211499e-01 1.04127431e+00
-1.44255564e-01 1.07495022e+00 1.41698897e+00 -4.94001992e-02
2.89435685e-01 -7.75701225e-01 -1.09574354e+00 -2.08648667e-01
3.19877788e-02 3.76911700e-01 6.63440488e-03 -4.43130940e-01
2.91567981e-01 -1.68128267e-01 -1.41644561e+00 8.89690578e-01
1.18544114e+00 8.27076852e-01 -4.58640531e-02 6.84256852e-01
-3.54699761e-01 1.23229432e+00 -9.17677522e-01 -8.61748457e-01
-1.26741752e-01 -5.69969296e-01 -8.25629950e-01 9.13031474e-02
7.77304709e-01 -3.73500645e-01 -4.09795195e-01 9.56820667e-01
4.44303840e-01 7.92947933e-02 5.56104243e-01 -7.21352518e-01
-6.16660058e-01 1.32155430e+00 -4.27744836e-01 1.97552398e-01
6.51477873e-01 -1.48516133e-01 1.52615595e+00 -3.98744076e-01
9.17475820e-01 1.14989746e+00 3.64764124e-01 3.39172810e-01
-1.18834603e+00 -5.66575862e-02 3.99762869e-01 -4.18431275e-02
-8.36263359e-01 -7.40795195e-01 6.37066841e-01 -1.50648072e-01
1.39459205e+00 3.22466969e-01 3.85533839e-01 9.80689764e-01
5.69418013e-01 9.74066854e-01 2.84236431e-01 -1.27126560e-01
3.80818635e-01 -1.26346320e-01 4.62771475e-01 3.61824274e-01
9.71750796e-01 -7.27632940e-01 -6.92251563e-01 -4.58964258e-02
-2.27259681e-01 -3.65344524e-01 -2.82712430e-01 -2.70826370e-01
-1.35694122e+00 7.24103451e-01 5.05310535e-01 -9.56463888e-02
-7.98107445e-01 1.64314374e-01 1.12741363e+00 1.87876582e-01
5.96619725e-01 7.22786546e-01 -2.40479514e-01 -6.26654327e-02
-9.12746727e-01 6.41205132e-01 9.86622691e-01 1.28134644e+00
2.88730383e-01 9.20258909e-02 -6.31472945e-01 3.95032793e-01
-9.39416587e-02 3.63105983e-01 3.26096714e-01 -5.92105567e-01
1.27553582e+00 8.77428174e-01 2.95762569e-01 -7.79952765e-01
-3.03772569e-01 -4.15353239e-01 -1.16098368e+00 -5.01871288e-01
-1.57753587e-01 -2.73252755e-01 -7.72081614e-01 1.64578402e+00
2.80978292e-01 -5.76074660e-01 6.53026521e-01 3.57970834e-01
1.41550684e+00 1.14686346e+00 -1.90912157e-01 -4.98322219e-01
1.24485517e+00 -8.07960391e-01 -1.13535619e+00 -1.78200498e-01
5.22731602e-01 -1.66680902e-01 1.07125938e+00 3.38457972e-01
-1.29188454e+00 -3.91106427e-01 -1.42966318e+00 -6.49307847e-01
-3.47176015e-01 -7.10491613e-02 2.93833971e-01 2.91975051e-01
-9.90414977e-01 3.32608789e-01 -7.28711486e-01 -4.51384008e-01
3.40064615e-01 1.56785864e-02 -3.81802082e-01 2.46987149e-01
-1.04760134e+00 5.59811294e-01 8.25442374e-01 -3.14283103e-01
-5.29782355e-01 -7.32085586e-01 -1.11534739e+00 4.99614596e-01
7.87333429e-01 -1.05689228e+00 1.48117483e+00 -1.74426094e-01
-1.13011599e+00 6.31684303e-01 -4.86868024e-01 -1.06525385e+00
3.20735395e-01 -5.40541172e-01 -1.61112949e-01 -1.30020054e-02
2.20606819e-01 4.34562683e-01 3.48782986e-02 -1.14855838e+00
-6.22223973e-01 -2.31769741e-01 6.40098751e-02 2.40097106e-01
-3.71142402e-02 -2.07369089e-01 -4.14982140e-01 -6.75726950e-01
-4.39788640e-01 -4.18697566e-01 -2.43456751e-01 -8.11568737e-01
-1.18960226e+00 -2.18111426e-01 3.02823633e-01 -1.01294994e+00
1.98410761e+00 -1.70257258e+00 3.96800429e-01 -3.73091936e-01
2.75500894e-01 9.24704820e-02 8.72088298e-02 1.04403365e+00
2.28620246e-01 2.97853500e-01 -6.63555205e-01 -3.91658336e-01
2.24804908e-01 -1.76715091e-01 -7.99424946e-01 -5.37256198e-03
3.06452066e-01 9.81409669e-01 -7.89129853e-01 -3.57518524e-01
-2.83938706e-01 -4.17876765e-02 -6.69249833e-01 3.56923550e-01
-7.34755516e-01 1.88346803e-01 -3.36097449e-01 4.60241050e-01
3.77224773e-01 -1.63656145e-01 1.41959697e-01 -2.07297921e-01
-1.41220182e-01 9.91614580e-01 -9.23745811e-01 1.84260702e+00
-1.88849896e-01 5.17668128e-01 -2.28436902e-01 -6.75490618e-01
7.83765435e-01 1.91311866e-01 2.72666425e-01 -5.93989909e-01
3.20049375e-02 -4.81342152e-03 -2.31952995e-01 -3.21906239e-01
1.20587301e+00 3.43681991e-01 -7.73278713e-01 3.21754366e-01
7.21997768e-02 -1.84102908e-01 8.50767612e-01 6.93184614e-01
1.24713218e+00 -3.39586467e-01 1.12230873e+00 -2.91086018e-01
4.38872606e-01 2.57027179e-01 4.48487014e-01 8.62381041e-01
3.18782479e-01 5.53643346e-01 1.06202030e+00 -3.91925454e-01
-1.31222868e+00 -8.52802336e-01 1.57833070e-01 7.31656849e-01
-1.33440956e-01 -1.00158143e+00 -8.50662231e-01 -6.84381545e-01
5.53677045e-02 1.17341924e+00 -7.61425674e-01 -1.27817497e-01
-6.67402804e-01 -5.46332538e-01 3.50913465e-01 5.16557872e-01
5.28159380e-01 -1.07162642e+00 -9.16632891e-01 4.54722941e-01
-3.51689935e-01 -1.16018343e+00 -5.91473639e-01 2.18042344e-01
-7.60549247e-01 -8.30402315e-01 -3.54426473e-01 -4.34890121e-01
1.64056361e-01 -2.75461413e-02 1.56001091e+00 -4.95417774e-01
2.55387783e-01 1.15127200e-02 -4.69403058e-01 -8.86633277e-01
-5.84619284e-01 6.68648005e-01 -3.10908288e-01 -2.35655233e-02
2.31149839e-03 -2.48889148e-01 -1.18137635e-01 -6.45013213e-01
-1.19332445e+00 3.80096793e-01 5.22910118e-01 7.30280697e-01
7.90389895e-01 -1.61525831e-01 8.73737991e-01 -1.37261093e+00
9.72619593e-01 -7.34301150e-01 -3.84103000e-01 6.92645371e-01
-4.07468438e-01 5.87118268e-01 1.00159025e+00 2.05524772e-01
-1.03859091e+00 -1.16701713e-02 -9.74087641e-02 2.30248019e-01
-7.01164156e-02 1.03316069e+00 -6.05597913e-01 1.06217647e+00
5.13240397e-01 3.85243118e-01 -4.85122174e-01 -4.73833233e-01
6.15809023e-01 7.49760091e-01 9.16943669e-01 -2.21148148e-01
4.47244734e-01 2.80349016e-01 -2.77144253e-01 -6.81596100e-01
-1.03924537e+00 -1.89527750e-01 -4.72271740e-01 2.95491442e-02
8.85701716e-01 -1.02087891e+00 -4.03719693e-01 3.00520416e-02
-1.29230142e+00 -7.00828359e-02 -5.31092286e-01 5.01950271e-02
-5.70030332e-01 3.35266024e-01 -3.54025632e-01 -5.91063499e-01
-1.16180611e+00 -9.07500863e-01 1.24368334e+00 1.80553675e-01
-4.90444660e-01 -6.73465788e-01 2.14872003e-01 5.58584221e-02
2.52635658e-01 6.96431220e-01 1.12518215e+00 -9.89392221e-01
-5.11055470e-01 -2.34651402e-01 -4.53280024e-02 -1.24632791e-01
1.71684787e-01 -4.89136279e-02 -5.55459082e-01 -1.76464140e-01
-1.79685190e-01 -3.57305735e-01 1.34192169e+00 2.91877985e-01
1.11354208e+00 -8.85854304e-01 -4.08686735e-02 3.40242624e-01
1.46528792e+00 2.60944575e-01 5.37305057e-01 1.90773278e-01
8.22452247e-01 6.80735409e-01 4.14457828e-01 6.95636451e-01
8.18374634e-01 4.97362405e-01 3.11224431e-01 2.43140385e-01
-4.08913381e-02 -7.20728338e-01 3.54154199e-01 8.58145475e-01
5.76181531e-01 -9.46443081e-01 -8.16072404e-01 7.46744096e-01
-1.89426148e+00 -1.08661449e+00 -1.78830236e-01 2.18685770e+00
1.04690814e+00 3.79302680e-01 8.13184753e-02 4.25985940e-02
5.14346838e-01 7.12276518e-01 -7.61101007e-01 -7.95795381e-01
-2.28613913e-01 -1.53263092e-01 5.89422703e-01 3.48509401e-01
-1.16307020e+00 8.37285578e-01 6.12273884e+00 3.73569191e-01
-8.48562181e-01 -3.60737771e-01 6.40403271e-01 -2.45577842e-01
-7.20459461e-01 -3.05462152e-01 -8.91344607e-01 3.87471348e-01
1.33461797e+00 -9.47293103e-01 5.90408109e-02 5.74374914e-01
8.87720287e-02 -2.19838738e-01 -1.33396888e+00 6.77961290e-01
2.89231628e-01 -1.65690589e+00 6.85607195e-01 -1.85423836e-01
8.06870401e-01 1.85794923e-02 -2.41731852e-01 3.08360875e-01
4.52113062e-01 -7.47721374e-01 9.19933796e-01 5.78796506e-01
9.09035921e-01 -8.54784310e-01 8.73509586e-01 2.19793037e-01
-1.11860526e+00 1.94326162e-01 -2.61223078e-01 1.82177261e-01
3.09418023e-01 6.38595402e-01 -7.84949660e-01 1.03158402e+00
5.77187181e-01 9.45048392e-01 -6.87799335e-01 9.07227695e-01
-1.56552941e-01 6.67450547e-01 -1.13318548e-01 -1.02518260e-01
2.01060578e-01 8.48250687e-02 6.73134327e-01 1.42777824e+00
2.56331414e-01 2.52536815e-02 1.08517371e-01 5.65993190e-01
-6.54036105e-01 2.57198781e-01 -8.55012536e-01 -1.62348419e-01
3.61587584e-01 6.56510472e-01 -4.20293331e-01 -8.77402663e-01
-3.47333968e-01 7.63665199e-01 3.80835712e-01 7.70710111e-02
-6.14326596e-01 -6.16654873e-01 4.63659346e-01 -1.42899573e-01
4.17679548e-01 1.21828608e-01 -7.82188833e-01 -1.51243150e+00
4.56116438e-01 -1.00799417e+00 6.74571097e-01 -6.27577603e-01
-9.39306498e-01 8.89268577e-01 2.71840960e-01 -1.15140390e+00
-4.53399360e-01 1.88758135e-01 -7.20534325e-01 4.47875261e-01
-1.23448229e+00 -8.33524942e-01 -2.61752903e-01 -1.82880778e-02
9.05773640e-01 -5.37955351e-02 6.81757689e-01 -1.91228867e-01
-8.26380789e-01 5.52919924e-01 4.16670263e-01 4.37046029e-02
6.26331091e-01 -1.64159691e+00 1.26052296e+00 1.25198925e+00
1.83638856e-01 5.36716640e-01 1.02402043e+00 -9.29912031e-01
-1.61049402e+00 -1.34269452e+00 1.45722139e+00 -2.08316281e-01
3.81842166e-01 -5.66280901e-01 -1.10378695e+00 7.12195814e-01
9.10000622e-01 -6.64073288e-01 6.11111641e-01 -8.17405879e-02
-3.73463988e-01 -3.23348373e-01 -9.36933815e-01 5.83060741e-01
7.84793317e-01 -2.09832594e-01 -8.79430175e-01 1.54963925e-01
1.21160865e+00 -6.68938220e-01 -7.36137748e-01 4.07631665e-01
2.29831517e-01 -6.94202960e-01 5.15902281e-01 -7.83640683e-01
1.00141668e+00 -1.73707664e-01 -2.69284219e-01 -1.57965767e+00
-1.27677783e-01 -7.00109363e-01 -6.68818355e-01 1.47278619e+00
6.23939216e-01 -1.12841509e-01 5.74711740e-01 3.38997245e-01
-4.59994286e-01 -7.04948306e-01 -7.62202322e-01 -3.99399042e-01
-1.33595821e-02 -3.04972101e-02 8.62585664e-01 4.18499112e-01
3.97500306e-01 8.28583837e-01 -5.02539575e-01 4.23166081e-02
5.75766325e-01 3.00749749e-01 9.57288027e-01 -1.11687768e+00
9.29694325e-02 -3.94110858e-01 -4.90506319e-03 -1.01194537e+00
-8.35241936e-03 -8.78993094e-01 8.64734054e-02 -2.45040441e+00
3.18064928e-01 5.56983769e-01 -6.78234994e-02 2.59610087e-01
-4.58008081e-01 -4.47828203e-01 2.52733320e-01 -7.12324604e-02
-1.19492173e+00 8.85936797e-01 7.67323673e-01 -4.26940084e-01
-3.38724971e-01 -1.94096521e-01 -1.31906772e+00 1.16870008e-01
8.12517524e-01 -4.60520178e-01 -4.53096956e-01 -6.18355751e-01
3.68057162e-01 4.82203960e-01 -3.19219202e-01 -1.13635111e+00
3.13545227e-01 -4.39056680e-02 6.68756887e-02 -1.14670420e+00
-1.71150595e-01 -3.46085042e-01 -5.26723340e-02 3.30273271e-01
-9.93064523e-01 4.87563729e-01 3.68859679e-01 7.67189741e-01
-5.14566600e-01 1.96643755e-01 4.07883227e-01 -8.40643644e-02
-3.81539673e-01 8.24449509e-02 -4.77188379e-01 5.18019378e-01
8.13496768e-01 2.49845684e-01 -6.17941976e-01 -6.36411786e-01
-9.29773003e-02 5.98180234e-01 3.87173355e-01 4.07459080e-01
4.26893890e-01 -1.09000790e+00 -1.20888698e+00 -1.09534688e-01
5.29876709e-01 2.84712523e-01 2.08556041e-01 3.83097768e-01
-5.22654235e-01 9.47397411e-01 -2.47179061e-01 -6.63007423e-02
-9.46526825e-01 7.17859685e-01 -8.07637423e-02 -7.40769386e-01
-8.63322139e-01 3.52870613e-01 5.80312423e-02 -1.20187283e-01
4.94399428e-01 -1.03096616e+00 -4.22099024e-01 3.94735992e-01
8.36885393e-01 3.00356686e-01 4.72904265e-01 -3.27867270e-01
-2.48864934e-01 -7.52429888e-02 -4.39553350e-01 -4.09303717e-02
1.39572334e+00 -2.30997533e-01 5.27831689e-02 6.81251049e-01
1.11953795e+00 3.04565340e-01 -1.00402021e+00 -3.93923074e-01
3.90083134e-01 8.97775218e-02 -2.14040339e-01 -8.81103635e-01
-5.08940101e-01 5.39051354e-01 -2.59020120e-01 4.74886954e-01
1.19896901e+00 1.05251253e-01 1.03205657e+00 8.08399141e-01
1.18593439e-01 -8.38959634e-01 -4.45966482e-01 5.68314910e-01
1.18542564e+00 -1.18419242e+00 2.32791066e-01 -1.03716314e-01
-1.11934519e+00 8.71839345e-01 3.30937117e-01 -1.74500421e-01
5.27666882e-03 2.29833424e-01 -3.27935785e-01 -2.59475172e-01
-1.44440734e+00 -2.78299302e-01 4.61282402e-01 1.13126919e-01
6.18593514e-01 1.02395803e-01 -5.18081009e-01 9.08692241e-01
-5.15187562e-01 -2.12922320e-01 1.00815225e+00 8.86219442e-01
-6.29849136e-01 -7.46289194e-01 -9.04758126e-02 8.45966339e-01
-6.77951157e-01 -9.83631462e-02 -8.17554891e-01 3.90554279e-01
-4.79229033e-01 1.14035249e+00 7.95499235e-02 -3.68495524e-01
7.77567804e-01 1.12504914e-01 -2.76729539e-02 -8.15378726e-01
-9.08477426e-01 -1.72325045e-01 4.28460062e-01 -4.67081040e-01
-1.85799167e-01 -7.87545025e-01 -1.14461589e+00 -3.60809922e-01
3.89584363e-01 3.73137385e-01 3.35555464e-01 4.82523173e-01
6.05283558e-01 8.76037598e-01 3.62014323e-01 -4.30447668e-01
-6.01050019e-01 -1.05920959e+00 -1.70838162e-01 2.60244787e-01
7.22907603e-01 -1.63953658e-02 1.43905550e-01 2.11437181e-01] | [12.535480499267578, 9.408080101013184] |
23648d7a-8738-4f1c-b20d-9abeb0dbe93f | local-region-perception-and-relationship | 2303.08545 | null | https://arxiv.org/abs/2303.08545v2 | https://arxiv.org/pdf/2303.08545v2.pdf | Local Region Perception and Relationship Learning Combined with Feature Fusion for Facial Action Unit Detection | Human affective behavior analysis plays a vital role in human-computer interaction (HCI) systems. In this paper, we introduce our submission to the CVPR 2023 Competition on Affective Behavior Analysis in-the-wild (ABAW). We propose a single-stage trained AU detection framework. Specifically, in order to effectively extract facial local region features related to AU detection, we use a local region perception module to effectively extract features of different AUs. Meanwhile, we use a graph neural network-based relational learning module to capture the relationship between AUs. In addition, considering the role of the overall feature of the target face on AU detection, we also use the feature fusion module to fuse the feature information extracted by the backbone network and the AU feature information extracted by the relationship learning module. We also adopted some sampling methods, data augmentation techniques and post-processing strategies to further improve the performance of the model. | ['Wangyuan Zhu', 'Jichao Zhu', 'Guochen Xie', 'Gongpeng Zhao', 'Zhongpeng Cai', 'Renda Li', 'Jun Yu'] | 2023-03-15 | null | null | null | null | ['action-unit-detection', 'facial-action-unit-detection', 'relational-reasoning'] | ['computer-vision', 'computer-vision', 'natural-language-processing'] | [ 4.58540112e-01 2.82959014e-01 7.42986500e-02 -4.14449841e-01
-4.27919418e-01 -8.56189802e-02 2.41853699e-01 1.50511160e-01
-2.26747110e-01 1.04848608e-01 2.36208588e-01 5.18872440e-01
2.39034086e-01 -6.44668221e-01 -1.29187495e-01 -7.34698951e-01
-3.16353738e-02 -1.01362430e-01 1.00915013e-02 -4.72698122e-01
-2.89396912e-01 6.25086188e-01 -1.50548780e+00 5.04916251e-01
3.73244911e-01 1.56204045e+00 -4.70823675e-01 4.76370007e-01
1.79424033e-01 1.14754760e+00 -3.04917455e-01 -3.87501389e-01
8.14342499e-02 -5.25577903e-01 -5.59656441e-01 2.75120676e-01
1.74542770e-01 -3.04931998e-01 -3.79593194e-01 8.65758598e-01
6.16707265e-01 4.56394732e-01 7.23382652e-01 -1.41267216e+00
-1.36905059e-01 2.68968612e-01 -9.67584491e-01 -5.01211919e-02
5.82749009e-01 -4.05200906e-02 1.07578385e+00 -9.62189496e-01
5.72758198e-01 1.10480595e+00 5.24510384e-01 6.75231695e-01
-7.38459349e-01 -6.16620481e-01 2.64699399e-01 5.43512344e-01
-1.40315127e+00 -5.55686176e-01 1.24422908e+00 -1.93167180e-01
7.12339461e-01 2.96286225e-01 8.48580420e-01 1.15884769e+00
-2.50666142e-02 1.02235961e+00 7.10612357e-01 -3.97245973e-01
-4.03727815e-02 1.01653174e-01 2.32967421e-01 1.12390983e+00
-5.78869641e-01 -2.99093515e-01 -6.40746593e-01 -2.81061471e-01
3.97420913e-01 1.99036598e-02 5.19958027e-02 -3.47640395e-01
-3.40695173e-01 6.54783607e-01 6.64035022e-01 2.86984861e-01
-6.36388779e-01 5.21007888e-02 5.46146631e-01 4.15317863e-02
5.37758827e-01 5.68046570e-02 -6.78746998e-02 7.17657954e-02
-3.95194322e-01 -1.11339860e-01 5.59649467e-01 5.72521567e-01
8.31216395e-01 -2.52544165e-01 -6.01759791e-01 1.00459421e+00
4.15269703e-01 8.18760544e-02 1.23653382e-01 -8.81762087e-01
1.64705575e-01 1.12658417e+00 -2.59543568e-01 -1.33080077e+00
-5.37659228e-01 4.95168827e-02 -9.04377520e-01 -2.51845866e-01
-8.06672424e-02 -2.50654608e-01 -5.95654130e-01 1.76315641e+00
5.07337272e-01 2.15267196e-01 -6.52302429e-02 9.52912867e-01
1.02864361e+00 4.60588336e-01 2.14741886e-01 -5.48094332e-01
1.43229961e+00 -1.03397512e+00 -8.51958394e-01 -9.75970775e-02
8.23850513e-01 -4.41568404e-01 6.56787515e-01 2.35573217e-01
-9.59133208e-01 -5.07847190e-01 -1.08074427e+00 1.05878301e-02
-4.54199582e-01 5.07460296e-01 5.74882150e-01 2.92004406e-01
-7.77662933e-01 -8.94793775e-03 -4.34528559e-01 -4.75703984e-01
5.64766526e-01 5.78952491e-01 -4.96779799e-01 -4.98197377e-02
-1.27590191e+00 7.28542209e-01 1.57546923e-02 5.68552911e-01
-6.62480712e-01 -2.06785649e-01 -9.96436417e-01 -5.55551127e-02
5.71757793e-01 -3.16018045e-01 9.24276471e-01 -1.38423479e+00
-1.44882214e+00 8.45156252e-01 -3.04345965e-01 -9.22126547e-02
4.56690677e-02 -2.46082395e-01 -4.72292423e-01 4.50345963e-01
-3.48915726e-01 5.96643984e-01 1.00774693e+00 -1.18825054e+00
-3.98458958e-01 -6.49643302e-01 3.11633982e-02 4.69114274e-01
-7.02700794e-01 2.78136909e-01 -7.74654090e-01 -1.59691900e-01
-1.63874164e-01 -8.84929776e-01 -1.71503752e-01 -8.45181495e-02
-1.87894210e-01 -7.13937521e-01 8.84380817e-01 -6.48788393e-01
1.34815407e+00 -2.49363637e+00 2.31593624e-01 7.42993355e-01
7.08043039e-01 2.91245103e-01 -3.23027670e-01 2.55655229e-01
4.91253585e-02 -4.54683661e-01 1.00927636e-01 -6.13989711e-01
-2.84003288e-01 -5.32506332e-02 1.71585426e-01 3.84450883e-01
3.71228278e-01 8.94411027e-01 -6.78811133e-01 -8.06665421e-01
1.58098176e-01 5.00236452e-01 -1.01393983e-01 5.43663263e-01
1.68971628e-01 1.00856274e-01 -5.18734872e-01 7.81293929e-01
4.20385331e-01 1.27395168e-01 2.84491926e-01 -6.21283710e-01
4.37050879e-01 -2.94858992e-01 -7.65507102e-01 1.46833730e+00
-4.29969966e-01 2.83917159e-01 3.85972172e-01 -8.56615603e-01
1.18440259e+00 4.45429951e-01 1.04158378e+00 -7.61071146e-01
6.84255004e-01 -3.86988044e-01 1.66091684e-03 -5.36419690e-01
1.93058282e-01 2.35796064e-01 2.36824583e-02 2.34592229e-01
7.30604529e-02 3.68229032e-01 -8.11417475e-02 3.39418560e-01
1.18750250e+00 1.09262824e-01 5.33955216e-01 2.04457834e-01
8.42014551e-01 -3.01859498e-01 6.72701001e-01 2.17979595e-01
-7.76697695e-01 4.12413388e-01 5.70230007e-01 -4.75684226e-01
-3.58792722e-01 -7.21871853e-01 3.08993012e-01 1.41414094e+00
3.02607626e-01 -7.71734893e-01 -9.21489179e-01 -1.03813100e+00
-3.33749443e-01 1.85805768e-01 -1.09224892e+00 -8.14118624e-01
-3.43276590e-01 -6.79590762e-01 4.41063434e-01 6.18145287e-01
6.08510852e-01 -1.38511527e+00 -5.11377573e-01 -1.88678615e-02
-2.39122227e-01 -1.27618170e+00 -4.81219232e-01 3.01986169e-02
-1.43846348e-01 -1.21928561e+00 -2.07991064e-01 -7.29784369e-01
6.95573926e-01 2.58920729e-01 6.85888290e-01 6.61329404e-02
-6.30357623e-01 6.47752643e-01 -5.56912005e-01 -4.64050204e-01
-8.80535915e-02 -9.28423926e-02 1.30290827e-02 6.94009364e-01
7.80307710e-01 -1.99758008e-01 -6.21760845e-01 1.86399102e-01
-4.82235402e-01 -1.97618291e-01 5.23489833e-01 4.80276346e-01
4.17734802e-01 -4.01260518e-02 4.63209480e-01 -7.18161762e-01
8.23999345e-01 -3.33974391e-01 -1.49927260e-02 4.80275542e-01
-2.03720823e-01 -5.72886646e-01 3.52456063e-01 -3.58434707e-01
-1.18894947e+00 5.66154838e-01 -6.79259375e-02 -5.84568501e-01
-7.42610171e-02 3.29918176e-01 -4.36127812e-01 -1.75756291e-01
5.42982817e-01 -9.51527953e-02 2.65822202e-01 1.27841026e-01
2.81170130e-01 8.58485818e-01 2.44605511e-01 -2.45467767e-01
2.99819112e-01 3.70002538e-01 1.22689679e-01 -8.61967981e-01
-1.00506341e+00 -7.17290580e-01 -6.27181828e-01 -8.32335055e-01
1.22352135e+00 -9.85004783e-01 -1.11703336e+00 3.54668409e-01
-9.79556739e-01 -6.03966750e-02 -2.05829702e-02 1.43330485e-01
-3.18316996e-01 2.53260344e-01 -7.84240901e-01 -1.17767859e+00
-7.79545188e-01 -7.77532220e-01 1.16659331e+00 3.70740563e-01
-5.01759052e-01 -6.70542538e-01 5.77917658e-02 5.36409318e-01
-4.39451262e-02 2.28311434e-01 5.71617961e-01 -5.59840262e-01
2.31759474e-01 -4.70205188e-01 -4.62771565e-01 3.72310668e-01
2.84133673e-01 1.29632324e-01 -1.17412198e+00 -8.49794317e-03
-1.02777682e-01 -6.93270206e-01 7.64663339e-01 1.50419742e-01
1.13697147e+00 -3.01104579e-02 -3.18219215e-01 2.70567328e-01
9.16683316e-01 2.10801765e-01 5.82006633e-01 -2.97568470e-01
1.16123188e+00 1.06945479e+00 8.82154047e-01 5.96527994e-01
4.19383973e-01 6.24239564e-01 4.34318364e-01 -3.26074958e-01
2.01260652e-02 2.53020637e-02 5.08874178e-01 6.10621929e-01
-2.77472824e-01 -4.57311459e-02 -5.70446908e-01 2.45033562e-01
-2.25697756e+00 -6.98588431e-01 6.05686307e-02 1.65316641e+00
3.53477478e-01 -1.79366350e-01 3.89210492e-01 -1.82979807e-01
5.13206422e-01 1.09797992e-01 -3.37751508e-01 -6.75918281e-01
6.62523508e-02 -2.55849827e-02 -1.39008984e-01 3.01339179e-01
-1.13605058e+00 1.24158871e+00 5.74365425e+00 7.59254277e-01
-9.03651476e-01 1.77338403e-02 7.21056342e-01 -2.41684634e-02
1.10879391e-01 -5.12077093e-01 -5.32160461e-01 -6.74174726e-02
6.45436227e-01 3.61964226e-01 3.30659479e-01 8.52092206e-01
2.04759911e-01 -1.79587856e-01 -9.11572814e-01 1.20634079e+00
4.83925253e-01 -5.89193046e-01 -1.62101075e-01 5.23345545e-02
2.72875279e-01 -3.62416446e-01 -1.31549478e-01 3.56666803e-01
1.07842339e-02 -9.86444473e-01 2.91207600e-02 7.25997925e-01
4.27028328e-01 -1.05554724e+00 9.62166429e-01 6.87363595e-02
-1.47474992e+00 -1.52485803e-01 -3.11717272e-01 -3.64050493e-02
-2.47927025e-01 3.59480470e-01 -8.22937191e-01 5.47037721e-01
8.27925980e-01 8.73624682e-01 -7.12912917e-01 3.06420058e-01
-1.28968611e-01 2.93028265e-01 -1.93511412e-01 -1.69327602e-01
4.79581580e-02 -3.16375822e-01 1.08997598e-01 1.01577377e+00
-1.66069850e-01 4.06376332e-01 2.58666307e-01 5.01700997e-01
-1.40879750e-01 4.22786921e-01 -8.51562500e-01 -1.56586021e-02
1.13447167e-01 1.99286592e+00 -4.06500399e-01 -1.95424974e-01
-5.85833490e-01 1.15416515e+00 7.48476803e-01 2.96227515e-01
-9.12973821e-01 -2.72369921e-01 6.05861127e-01 -3.11387271e-01
-1.38630308e-02 5.15491031e-02 6.33834451e-02 -9.06012833e-01
-1.71082318e-01 -7.65100062e-01 4.95591253e-01 -7.56351113e-01
-1.27183318e+00 7.59896398e-01 -2.67010480e-01 -9.85035300e-01
-2.08054811e-01 -5.06451309e-01 -7.37764478e-01 5.79937041e-01
-8.92729878e-01 -1.54088914e+00 -4.66022789e-01 9.18372214e-01
2.66116768e-01 -6.82326630e-02 8.88222933e-01 1.32495657e-01
-9.60942447e-01 6.70098841e-01 -7.59187996e-01 3.58828545e-01
7.43126571e-01 -8.13436627e-01 -3.79416734e-01 6.02836788e-01
1.12330660e-01 4.14117962e-01 2.00965822e-01 -5.00808716e-01
-1.29550719e+00 -1.09205842e+00 5.67357421e-01 -2.47565806e-01
4.65515465e-01 -5.96266806e-01 -7.11829245e-01 4.93796974e-01
3.09707493e-01 1.41691014e-01 8.56557131e-01 3.59702319e-01
-2.45401233e-01 -4.60082591e-01 -1.01074541e+00 7.63480246e-01
1.01518452e+00 -7.12015748e-01 -1.90680683e-01 -9.82020944e-02
3.60830128e-01 3.11697386e-02 -9.18168128e-01 7.90349126e-01
7.03381956e-01 -8.73326302e-01 7.24985957e-01 -4.59313184e-01
3.24383527e-01 3.05757392e-02 -1.41628027e-01 -9.52436149e-01
-4.19216394e-01 -3.95899594e-01 -3.97664905e-01 1.37981558e+00
1.74385414e-01 -1.48200944e-01 9.21405911e-01 6.78007662e-01
1.68991983e-01 -1.11085987e+00 -5.44023395e-01 1.41995713e-01
-7.17633784e-01 -3.64849925e-01 1.56267583e-01 6.68143451e-01
4.64059860e-01 8.06813419e-01 -5.66388428e-01 2.76156869e-02
4.07662362e-01 -5.41454256e-02 6.48986638e-01 -1.10826242e+00
-3.34491059e-02 -1.81897536e-01 -4.49594349e-01 -4.30209577e-01
4.01704758e-01 -4.41001266e-01 2.94883043e-01 -1.32376826e+00
3.51072550e-01 1.75370142e-01 -6.79646254e-01 6.92013860e-01
-3.29299957e-01 5.08179069e-01 1.91723689e-01 -2.53764112e-02
-8.87954533e-01 8.15382063e-01 1.18864834e+00 -8.66468474e-02
-3.68991762e-01 8.16797186e-03 -7.08704770e-01 7.25429833e-01
8.96366656e-01 -3.13179232e-02 -5.79549074e-01 2.13454440e-01
1.95751548e-01 -3.04287598e-02 2.69522727e-01 -8.23517501e-01
1.67915642e-01 -6.38250485e-02 5.78146815e-01 -3.27243537e-01
8.47761989e-01 -9.92543161e-01 -5.78526676e-01 1.70213968e-01
-3.12913299e-01 -1.63752571e-01 7.11574256e-02 3.96002501e-01
-2.55338073e-01 1.39903501e-01 6.95694804e-01 6.18482288e-03
-8.42996180e-01 5.44068456e-01 -6.39536083e-01 -4.59139884e-01
1.41974688e+00 1.02803344e-03 -2.83102710e-02 -5.84593177e-01
-7.77428329e-01 4.17000830e-01 -5.82795870e-03 4.84947443e-01
9.43638384e-01 -1.25123203e+00 -3.23776335e-01 2.44464263e-01
5.83210766e-01 -3.70286793e-01 4.39769715e-01 1.19646466e+00
-2.51171663e-02 -2.06561282e-01 -1.94366038e-01 -2.72707105e-01
-1.79631460e+00 5.58483481e-01 3.03528428e-01 -9.72738788e-02
-1.38359472e-01 1.02129602e+00 2.45540783e-01 -2.61019707e-01
3.34712684e-01 3.39397043e-01 -8.89851987e-01 4.42530632e-01
5.03403425e-01 2.87349164e-01 -1.40954077e-01 -9.06639695e-01
-5.91811538e-01 4.96588767e-01 -1.74957544e-01 1.45196440e-02
9.99517918e-01 -2.18414679e-01 -5.02482474e-01 4.65435237e-01
1.34658635e+00 -2.95682643e-02 -9.43718851e-01 -3.03757608e-01
-1.58979684e-01 1.70360386e-01 3.18815671e-02 -5.17573655e-01
-1.06548655e+00 8.49266827e-01 7.45319486e-01 5.90368621e-02
1.51586998e+00 1.67428136e-01 6.59619510e-01 3.08625072e-01
-6.92215189e-02 -1.47029734e+00 3.67129475e-01 1.63766742e-01
7.24567890e-01 -1.30310786e+00 -8.56421888e-02 -8.19133043e-01
-1.00356114e+00 1.02246988e+00 1.25644553e+00 -1.11400224e-01
7.95131803e-01 1.16799615e-01 3.24982643e-01 -3.86236012e-01
-8.11755359e-01 -5.24606228e-01 6.70699298e-01 5.09775043e-01
4.02076304e-01 -7.88889006e-02 -7.18935803e-02 6.94106996e-01
3.30360800e-01 -1.06606826e-01 -4.17216234e-02 6.48582935e-01
-3.81887764e-01 -8.97619367e-01 4.88757193e-02 4.00490195e-01
-1.72164097e-01 7.69845024e-02 -1.21109080e+00 4.88521874e-01
2.45028868e-01 1.17133033e+00 7.08858366e-04 -1.02651060e+00
2.33560547e-01 1.39394403e-01 5.37370682e-01 -4.15732205e-01
-8.45060408e-01 2.90424079e-01 2.54364789e-01 -1.05420101e+00
-6.49615169e-01 -5.84399045e-01 -1.40245807e+00 9.09142792e-02
-7.40826651e-02 -4.72184084e-02 3.17835301e-01 1.09928620e+00
4.24974769e-01 6.20657861e-01 7.50957966e-01 -5.73216856e-01
7.90770724e-02 -1.19196463e+00 -6.56402588e-01 4.08298284e-01
2.62593497e-02 -4.44042087e-01 -1.09332427e-01 -2.20986336e-01] | [13.63210678100586, 1.8648643493652344] |
665c5b4d-e868-4a89-9de4-1949d3fa2f7b | gnn3dmot-graph-neural-network-for-3d-multi-1 | 2006.07327 | null | https://arxiv.org/abs/2006.07327v1 | https://arxiv.org/pdf/2006.07327v1.pdf | GNN3DMOT: Graph Neural Network for 3D Multi-Object Tracking with Multi-Feature Learning | 3D Multi-object tracking (MOT) is crucial to autonomous systems. Recent work uses a standard tracking-by-detection pipeline, where feature extraction is first performed independently for each object in order to compute an affinity matrix. Then the affinity matrix is passed to the Hungarian algorithm for data association. A key process of this standard pipeline is to learn discriminative features for different objects in order to reduce confusion during data association. In this work, we propose two techniques to improve the discriminative feature learning for MOT: (1) instead of obtaining features for each object independently, we propose a novel feature interaction mechanism by introducing the Graph Neural Network. As a result, the feature of one object is informed of the features of other objects so that the object feature can lean towards the object with similar feature (i.e., object probably with a same ID) and deviate from objects with dissimilar features (i.e., object probably with different IDs), leading to a more discriminative feature for each object; (2) instead of obtaining the feature from either 2D or 3D space in prior work, we propose a novel joint feature extractor to learn appearance and motion features from 2D and 3D space simultaneously. As features from different modalities often have complementary information, the joint feature can be more discriminate than feature from each individual modality. To ensure that the joint feature extractor does not heavily rely on one modality, we also propose an ensemble training paradigm. Through extensive evaluation, our proposed method achieves state-of-the-art performance on KITTI and nuScenes 3D MOT benchmarks. Our code will be made available at https://github.com/xinshuoweng/GNN3DMOT | ['Xinshuo Weng', 'Kris Kitani', 'Yunze Man', 'Yongxin Wang'] | 2020-06-12 | null | null | null | null | ['3d-multi-object-tracking'] | ['computer-vision'] | [ 3.83705385e-02 -4.43338424e-01 -5.60053475e-02 -2.35904768e-01
-4.94209796e-01 -5.29311299e-01 5.31911314e-01 -6.00099415e-02
-4.04720098e-01 4.30908978e-01 -1.96901575e-01 5.46445660e-02
-7.65400305e-02 -7.26410747e-01 -7.11769879e-01 -9.70710397e-01
1.09140880e-01 6.73344493e-01 7.94969440e-01 1.18341066e-01
1.75369196e-02 8.45296800e-01 -1.74308705e+00 -7.74225146e-02
3.10196131e-01 1.07913685e+00 3.21470827e-01 6.26327813e-01
6.25715926e-02 3.98162395e-01 -4.70683724e-01 1.41442373e-01
5.20412505e-01 -5.33671379e-01 -4.95136589e-01 2.12181509e-01
6.75627649e-01 -3.43512714e-01 -3.89845133e-01 1.17756164e+00
3.97908479e-01 2.00267151e-01 7.94228077e-01 -1.33926272e+00
-2.16300771e-01 1.52729675e-01 -6.71502292e-01 2.00489640e-01
1.30364209e-01 3.00863236e-01 1.06277335e+00 -8.42072606e-01
6.66651487e-01 1.14801109e+00 3.00988019e-01 5.76731741e-01
-1.16372550e+00 -7.09938347e-01 3.01636428e-01 2.60208160e-01
-1.41964662e+00 -2.96024919e-01 9.30860817e-01 -4.06513959e-01
5.55866480e-01 2.65700370e-01 1.00358880e+00 8.28428090e-01
1.19293638e-01 9.02951419e-01 8.70836318e-01 -1.29737243e-01
-9.58778244e-03 7.36840069e-02 9.80879366e-02 8.42207849e-01
4.95283604e-01 3.15193683e-01 -4.79304969e-01 -5.39209694e-02
4.82889265e-01 3.04941118e-01 2.35594083e-02 -8.35367560e-01
-1.23097181e+00 6.05625272e-01 7.16321170e-01 2.94446170e-01
-3.00051332e-01 2.12835252e-01 2.48848628e-02 2.45861545e-01
9.59586911e-03 -4.71883863e-02 -3.36453646e-01 9.02080834e-02
-6.14095807e-01 4.01683241e-01 5.25394380e-01 1.01928663e+00
1.24867308e+00 -3.11051816e-01 -3.21838498e-01 4.17484671e-01
7.10073888e-01 7.49446988e-01 2.48055860e-01 -7.54762292e-01
2.04044849e-01 9.16702569e-01 -6.27420028e-04 -8.59004617e-01
-4.89112318e-01 -2.70615458e-01 -5.89373469e-01 6.29134834e-01
5.92349529e-01 5.34512587e-02 -9.47972417e-01 1.57961261e+00
8.66446972e-01 1.21438496e-01 -7.37662092e-02 1.18500400e+00
9.06836569e-01 2.86747456e-01 -2.59583741e-01 1.17077649e-01
1.49809480e+00 -9.47974086e-01 -2.44518951e-01 -2.93640226e-01
6.33881211e-01 -6.89843893e-01 4.97980326e-01 1.43245071e-01
-7.22085714e-01 -7.61070907e-01 -1.01353264e+00 2.06394047e-02
-3.92510355e-01 1.10497652e-03 4.98504549e-01 3.49951506e-01
-6.99837267e-01 4.22108203e-01 -9.61437941e-01 -5.40555418e-01
4.23900127e-01 5.31607985e-01 -4.34560597e-01 -6.08934015e-02
-7.05244005e-01 8.13249707e-01 5.80262244e-01 -5.84269762e-02
-8.99944246e-01 -2.55635828e-01 -8.29427719e-01 -2.49709904e-01
5.35889745e-01 -8.59341443e-01 9.75467026e-01 -7.53031373e-01
-1.36031485e+00 7.01028049e-01 -2.08078340e-01 -9.67667773e-02
3.86629552e-01 -3.29180300e-01 -2.13018119e-01 -3.12833115e-02
7.28424937e-02 6.57112062e-01 1.04118276e+00 -1.29656136e+00
-1.07776785e+00 -5.24479389e-01 -1.21621579e-01 3.22634578e-01
-1.11818932e-01 -8.52466002e-03 -7.87159801e-01 -1.96955815e-01
2.95288891e-01 -1.25270736e+00 4.49602585e-03 3.28708082e-01
-3.74696165e-01 -4.24004138e-01 1.26078415e+00 -1.14497907e-01
7.61072695e-01 -2.18353295e+00 3.85399222e-01 3.18426847e-01
4.95505005e-01 2.25687280e-01 -8.47441703e-02 -2.82889828e-02
3.77139986e-01 -3.72722059e-01 2.48708203e-02 -3.35353434e-01
-7.24353045e-02 3.29249322e-01 2.58587211e-01 7.53342688e-01
5.10955751e-01 8.87173116e-01 -9.84722137e-01 -6.69181466e-01
3.83053154e-01 3.55341166e-01 -4.30737644e-01 1.57807708e-01
-1.51287526e-01 6.51673257e-01 -8.28033507e-01 9.01352167e-01
6.37755752e-01 -3.44717503e-01 -7.19197169e-02 -4.38007355e-01
-2.35125571e-01 1.42533481e-01 -1.39138818e+00 1.50448787e+00
5.97542897e-03 3.91039819e-01 -5.87679893e-02 -7.70458519e-01
1.03513086e+00 -6.42652288e-02 7.01280653e-01 -4.84430611e-01
2.73980796e-01 1.40153483e-01 2.97202379e-01 -3.23319763e-01
4.19061005e-01 1.70389339e-01 -5.60809448e-02 3.59236926e-01
2.05895722e-01 -1.00333067e-02 7.91909024e-02 5.09287864e-02
1.12782705e+00 4.05753583e-01 2.17258528e-01 7.74350539e-02
5.16467214e-01 -5.85702509e-02 7.78968513e-01 8.15658808e-01
-4.38220888e-01 4.27185178e-01 4.90671843e-02 -2.96838343e-01
-8.63641679e-01 -1.06872308e+00 -8.71849358e-02 9.54573214e-01
6.77813768e-01 -2.43115306e-01 3.28710265e-02 -1.16246045e+00
4.11990881e-01 1.58515990e-01 -5.84599733e-01 -2.67488897e-01
-5.68767130e-01 -2.64618069e-01 2.29996800e-01 5.32881141e-01
3.81358236e-01 -8.70422721e-01 -8.58058810e-01 9.07042399e-02
1.85291797e-01 -9.37567830e-01 -4.91430730e-01 4.91453946e-01
-7.57917404e-01 -1.04683781e+00 -3.87331545e-01 -5.62826693e-01
7.05810845e-01 6.59923494e-01 7.63691962e-01 3.70040536e-01
-1.57880381e-01 3.64300609e-01 -5.42791486e-01 -3.16897631e-01
-1.34329632e-01 -1.32640330e-02 3.07359248e-01 2.24724367e-01
6.05686665e-01 -2.68406451e-01 -5.87789178e-01 5.35977066e-01
-6.21572554e-01 -1.63545348e-02 8.39044511e-01 8.04372728e-01
8.34112406e-01 -1.46620333e-01 -7.66213462e-02 -3.42344910e-01
-1.62501693e-01 -4.41439867e-01 -7.62776732e-01 6.80404380e-02
-2.16485247e-01 1.76999018e-01 3.93750221e-01 -6.77626193e-01
-6.00787044e-01 6.25975192e-01 1.62303910e-01 -8.32170308e-01
-4.40519571e-01 1.31293207e-01 -3.35583836e-01 -2.88674861e-01
3.85537058e-01 2.42185622e-01 7.06113949e-02 -4.98147875e-01
2.13406235e-01 4.90516543e-01 2.74660438e-01 -4.06535119e-01
1.22932136e+00 6.53603077e-01 3.03218514e-01 -7.09282279e-01
-7.44434953e-01 -7.23482907e-01 -9.64570045e-01 -4.92607057e-01
9.72411036e-01 -8.35399985e-01 -6.38740182e-01 3.97031754e-01
-8.42811227e-01 -1.16873011e-01 -2.00032145e-01 8.93170536e-01
-2.49660760e-01 3.23599398e-01 -1.65403694e-01 -7.40768909e-01
-9.60543156e-02 -1.18843508e+00 1.23419964e+00 5.81784904e-01
-4.14997973e-02 -7.71814406e-01 1.60314322e-01 7.07264468e-02
1.41072810e-01 1.00010693e-01 2.92500645e-01 -9.25250828e-01
-8.05435121e-01 -2.90308982e-01 -3.06786567e-01 4.29797843e-02
4.22925442e-01 1.60615593e-01 -8.10311854e-01 -5.33507109e-01
-1.22245342e-01 -2.25598156e-01 8.59897792e-01 2.11628214e-01
6.95178986e-01 1.29452601e-01 -7.42315590e-01 6.48550808e-01
1.19571757e+00 7.47953281e-02 2.22914249e-01 4.36434060e-01
9.72513437e-01 3.26806724e-01 8.66193652e-01 2.80688733e-01
4.05258209e-01 9.41614807e-01 6.03234947e-01 1.07991770e-01
-2.93118030e-01 -2.61399373e-02 5.74659884e-01 4.46868360e-01
-1.25949681e-01 -5.45697212e-02 -7.13683724e-01 4.36576128e-01
-2.04312348e+00 -8.47749650e-01 -1.78330541e-01 2.17747593e+00
3.29984248e-01 2.26603553e-01 4.60859239e-01 -2.56856948e-01
6.66981399e-01 -5.46196885e-02 -7.63916016e-01 2.76359022e-01
-1.00699626e-01 -1.97669089e-01 5.59616268e-01 9.86579880e-02
-1.13426793e+00 8.31998348e-01 4.48892784e+00 5.35450041e-01
-1.04876304e+00 -1.07981600e-01 -1.32486790e-01 -1.84024930e-01
8.24245140e-02 2.78494924e-01 -1.32430851e+00 4.65050936e-01
3.54880333e-01 8.16731676e-02 1.07524499e-01 7.60138929e-01
-2.76789457e-01 -2.00442344e-01 -1.17055154e+00 9.17375624e-01
2.18402129e-02 -8.54044318e-01 4.37397556e-03 3.04202199e-01
3.35696667e-01 2.79528767e-01 -2.00588107e-01 2.39703536e-01
2.75510371e-01 -4.88418281e-01 7.83278346e-01 5.73410630e-01
2.93016762e-01 -4.43268657e-01 6.42314613e-01 4.15192604e-01
-1.61303532e+00 6.38515800e-02 -4.64265317e-01 8.42644796e-02
1.32515179e-02 4.56418306e-01 -6.52126670e-01 9.25959826e-01
8.67360473e-01 9.92058814e-01 -6.77633405e-01 1.25711799e+00
-1.73259467e-01 2.67657667e-01 -6.69761181e-01 -1.19831078e-01
1.38951227e-01 -8.56182873e-02 9.43442822e-01 9.71117079e-01
3.77894521e-01 -1.33425295e-01 5.57104528e-01 7.85975039e-01
8.12856033e-02 -1.22502774e-01 -7.00175226e-01 2.14898318e-01
4.42935795e-01 1.53372478e+00 -7.64295757e-01 -3.43608350e-01
-7.46154368e-01 9.61170435e-01 4.12812501e-01 7.13093728e-02
-8.69913459e-01 -1.66570559e-01 7.87874758e-01 -1.49642110e-01
9.08001661e-01 -3.41870546e-01 1.37605920e-01 -1.06079352e+00
3.69979292e-02 -4.10485119e-01 5.95966101e-01 -3.93581450e-01
-1.34081769e+00 4.47179317e-01 -1.14545956e-01 -1.67799687e+00
7.48326723e-03 -6.52422607e-01 -5.20481169e-01 8.41534436e-01
-1.46663749e+00 -1.34534407e+00 -5.34162700e-01 7.75062263e-01
1.78709760e-01 -6.13384843e-02 3.69989246e-01 2.45664075e-01
-4.49308902e-01 4.54342544e-01 -2.72619314e-02 1.84461847e-01
8.15913975e-01 -1.14131904e+00 1.39160469e-01 7.89437890e-01
2.85224885e-01 2.71119446e-01 4.80526537e-01 -8.79770160e-01
-1.75163114e+00 -1.18265760e+00 5.28274298e-01 -5.42786539e-01
6.16626143e-01 -3.07370752e-01 -9.14739668e-01 5.68403602e-01
-2.03550503e-01 3.95292014e-01 4.51019287e-01 -1.21417539e-02
-1.50230646e-01 -1.24079838e-01 -9.33816016e-01 3.93045425e-01
1.25369143e+00 -2.57437199e-01 -5.32229602e-01 1.28250495e-01
4.11517650e-01 -4.70238715e-01 -8.36478949e-01 5.05743802e-01
7.28258431e-01 -8.38267863e-01 8.87680769e-01 -4.59001780e-01
-1.90533355e-01 -1.06071401e+00 -2.37738848e-01 -1.01983500e+00
-6.37142062e-01 -2.12693259e-01 -5.89204550e-01 1.08746064e+00
1.75910115e-01 -5.29260159e-01 6.96838975e-01 2.38486782e-01
-1.97754934e-01 -5.07446706e-01 -1.11356223e+00 -1.07423615e+00
-3.27271849e-01 -2.47288883e-01 5.49025953e-01 6.59250438e-01
-5.03304660e-01 4.25732106e-01 -4.22390729e-01 4.84500825e-01
8.20994914e-01 5.69069326e-01 1.25555301e+00 -1.44196534e+00
-3.41242254e-01 -4.09313202e-01 -7.90215552e-01 -1.39907598e+00
-3.34370211e-02 -1.06979573e+00 2.71915436e-01 -1.31051815e+00
4.17959690e-01 -5.86043835e-01 -4.52435851e-01 6.63333654e-01
-2.86276281e-01 4.14969444e-01 3.88428956e-01 3.51192474e-01
-9.08442080e-01 6.47931099e-01 1.36671245e+00 -2.36302763e-01
-3.42414916e-01 8.83667171e-02 -3.37410569e-01 5.17237604e-01
7.92128325e-01 -7.51764953e-01 1.53141603e-01 -3.55335802e-01
-1.63204744e-01 -2.87336677e-01 8.38553965e-01 -1.25458205e+00
4.33355629e-01 -1.39811978e-01 6.98990762e-01 -9.05378103e-01
4.70658362e-01 -1.21414661e+00 2.80691296e-01 6.61447942e-01
2.78892130e-01 -1.69359833e-01 1.34048760e-01 5.57592809e-01
3.90390605e-02 -1.69149891e-01 7.27846742e-01 1.14332967e-01
-9.47674870e-01 5.13236046e-01 -2.35787421e-01 -3.25006306e-01
1.20779920e+00 -4.87060457e-01 -2.16825068e-01 2.12622751e-02
-5.37965357e-01 5.00087857e-01 8.06210756e-01 5.82813323e-01
6.06449425e-01 -1.54864323e+00 -7.41069436e-01 3.57695669e-01
3.78025740e-01 1.39413625e-01 3.32933404e-02 1.10196388e+00
1.29497141e-01 9.79811847e-02 -2.70253569e-01 -1.16239166e+00
-1.65653944e+00 5.64374864e-01 3.36238772e-01 -6.26872927e-02
-8.08639050e-01 6.81601584e-01 2.24068284e-01 -3.48123223e-01
3.23661678e-02 -2.17451379e-01 -1.62819326e-01 6.16607629e-02
3.97935063e-01 1.81180894e-01 -7.25914612e-02 -9.13435996e-01
-7.33694077e-01 9.12999094e-01 -3.48759651e-01 1.66229397e-01
1.20227456e+00 -9.55619141e-02 1.32230282e-01 5.03727973e-01
1.04751563e+00 2.15744339e-02 -1.47515535e+00 -4.98500794e-01
-1.17600031e-01 -7.01666057e-01 -1.36813603e-03 -3.88013572e-01
-1.10262656e+00 5.06983042e-01 9.20403004e-01 1.67836770e-01
9.26642895e-01 5.43744981e-01 5.03742099e-01 3.95543575e-01
3.71602178e-01 -8.77437651e-01 1.93073869e-01 6.92768276e-01
4.57139164e-01 -1.44386446e+00 6.93065003e-02 -3.90826948e-02
-5.50564647e-01 1.02657497e+00 8.00233006e-01 -3.03930696e-02
4.98357832e-01 1.36141643e-01 5.60869463e-02 -4.99583155e-01
-4.39233214e-01 -7.55752146e-01 6.59190774e-01 4.86943811e-01
5.48583977e-02 -1.16000220e-01 1.06180489e-01 2.06564546e-01
2.32893810e-01 -2.58046091e-01 -2.36568972e-02 1.20737398e+00
-6.26708269e-01 -1.22740626e+00 -5.16934156e-01 5.74170291e-01
-2.93231346e-02 3.77528489e-01 -6.09218061e-01 7.60414600e-01
2.69411683e-01 7.33692586e-01 7.69155920e-02 -6.87268913e-01
3.57866794e-01 -3.47925313e-02 6.78406537e-01 -5.27392209e-01
-4.49623346e-01 2.13533446e-01 -1.34592175e-01 -6.35209918e-01
-7.36175954e-01 -9.56506312e-01 -1.41545701e+00 -4.26790342e-02
-6.06467843e-01 -7.61819184e-02 5.24101555e-01 9.15726423e-01
4.06620413e-01 5.23899913e-01 6.25306666e-01 -1.17616987e+00
-2.09619612e-01 -7.58266389e-01 -3.82056177e-01 4.95565385e-01
4.52547908e-01 -1.20381558e+00 -2.87737966e-01 -2.23356098e-01] | [6.443276882171631, -2.213313102722168] |
0126b7db-aa2f-4a25-8509-924e3669f8e6 | geometry-of-deep-learning-for-magnetic | 1809.01749 | null | http://arxiv.org/abs/1809.01749v2 | http://arxiv.org/pdf/1809.01749v2.pdf | Geometry of Deep Learning for Magnetic Resonance Fingerprinting | Current popular methods for Magnetic Resonance Fingerprint (MRF) recovery are
bottlenecked by the heavy storage and computation requirements of a
dictionary-matching (DM) step due to the growing size and complexity of the
fingerprint dictionaries in multi-parametric quantitative MRI applications. In
this paper we study a deep learning approach to address these shortcomings.
Coupled with a dimensionality reduction first layer, the proposed MRF-Net is
able to reconstruct quantitative maps by saving more than 60 times in memory
and computations required for a DM baseline. Fine-grid manifold enumeration
i.e. the MRF dictionary is only used for training the network and not during
image reconstruction. We show that the MRF-Net provides a piece-wise affine
approximation to the Bloch response manifold projection and that rather than
memorizing the dictionary, the network efficiently clusters this manifold and
learns a set of hierarchical matched-filters for affine regression of the NMR
characteristics in each segment. | ['Dong-Dong Chen', 'Mike E. Davies', 'Marion I. Menzel', 'Pedro A. Gómez', 'Mohammad Golbabaee'] | 2018-09-05 | null | null | null | null | ['magnetic-resonance-fingerprinting'] | ['medical'] | [ 4.11693126e-01 1.62285924e-01 -5.60322776e-03 -3.29389244e-01
-9.75437284e-01 -3.10781896e-01 3.84811968e-01 2.91669339e-01
-7.60904312e-01 4.83881742e-01 2.56883800e-01 -1.40677169e-01
-3.58843595e-01 -6.50783241e-01 -9.08885241e-01 -1.01496780e+00
-3.16224694e-01 9.56196308e-01 1.33869573e-02 6.47631139e-02
3.19418758e-01 7.47733176e-01 -1.36771142e+00 2.80480921e-01
4.52070326e-01 9.81563687e-01 6.14589334e-01 5.10365725e-01
1.88367903e-01 6.87377095e-01 -1.27644360e-01 1.28706634e-01
2.98552603e-01 -3.02474022e-01 -8.06980908e-01 -1.86511517e-01
7.87325978e-01 -8.67432117e-01 -8.01765025e-01 1.00849271e+00
9.32195842e-01 3.13264370e-01 7.72222459e-01 -5.47946334e-01
-1.96740985e-01 5.95782459e-01 -3.96881074e-01 3.98780167e-01
-1.25318408e-01 -3.75440419e-01 6.43162012e-01 -1.17232239e+00
9.64604020e-01 6.25960171e-01 8.04962814e-01 4.99611795e-01
-1.50990307e+00 -3.90313148e-01 -4.84073609e-01 2.29426727e-01
-1.37580740e+00 -7.28678644e-01 6.94849312e-01 -5.16833544e-01
9.00284290e-01 2.91630656e-01 4.54982072e-01 7.71149158e-01
4.05127257e-01 2.06738353e-01 1.30417871e+00 -3.55301082e-01
2.67613947e-01 -6.34454414e-02 9.80236288e-03 8.89612496e-01
-5.71703874e-02 1.96383655e-01 -5.22587836e-01 -3.70472491e-01
1.30586147e+00 -3.90207358e-02 -1.95702359e-01 -7.55984008e-01
-1.43463266e+00 9.00423706e-01 2.47928396e-01 4.36870128e-01
-6.40835404e-01 2.26744026e-01 4.27985162e-01 3.10224712e-01
2.98900418e-02 5.28671145e-01 -2.91317534e-02 2.23964050e-01
-1.37930310e+00 2.25256547e-01 5.45073211e-01 5.93307197e-01
7.16906965e-01 -5.38146719e-02 3.91329639e-02 7.65450776e-01
7.09185824e-02 5.83417416e-01 5.39598823e-01 -1.12694967e+00
3.54587808e-02 -5.56418598e-02 -7.75732324e-02 -1.30389392e+00
-9.07128274e-01 -5.19238472e-01 -1.21774280e+00 -7.52149895e-02
4.72864300e-01 3.67838830e-01 -7.92428970e-01 1.60181355e+00
4.27764982e-01 1.18482418e-01 -5.64408362e-01 1.11522770e+00
6.20100915e-01 3.08857411e-01 -7.53936172e-02 -3.72438341e-01
1.20416832e+00 -4.73372996e-01 -5.17148972e-01 1.87055796e-01
6.80654824e-01 -6.27655745e-01 6.73083365e-01 3.92687649e-01
-1.27450967e+00 -4.77481186e-01 -1.11398494e+00 -3.14348131e-01
-4.52911891e-02 3.49773094e-02 4.83482152e-01 2.67437011e-01
-1.29294991e+00 1.06898355e+00 -1.01166797e+00 1.42447010e-01
3.35214972e-01 7.74612427e-01 -5.78951836e-01 -2.68257380e-01
-9.82913256e-01 1.10375059e+00 1.88595146e-01 -5.36280349e-02
-1.19628763e+00 -9.60731268e-01 -5.84826171e-01 -1.31564751e-01
-2.34020486e-01 -3.84793997e-01 6.78636611e-01 -3.98094177e-01
-1.23979986e+00 1.04170215e+00 -2.13978421e-02 -5.71903706e-01
3.97987783e-01 3.11177313e-01 -2.25887507e-01 6.80935383e-01
-1.15863709e-02 7.44059503e-01 1.05964172e+00 -1.00958765e+00
-1.90767780e-01 -6.63790882e-01 -3.13180208e-01 2.68735960e-02
-2.40886316e-01 -3.18381220e-01 -8.23153406e-02 -6.96090102e-01
5.78448832e-01 -9.64713156e-01 -4.71816510e-01 -8.59665275e-02
-1.00905970e-01 2.87820339e-01 4.53680217e-01 -1.14047110e+00
1.07672954e+00 -1.96191740e+00 3.66048455e-01 5.55136621e-01
6.93462431e-01 -1.04241751e-01 -2.18782485e-01 1.65945306e-01
-2.95217812e-01 -5.67224920e-01 -2.20427528e-01 -3.16243052e-01
-2.28886321e-01 1.56120703e-01 -2.53672004e-01 1.13174415e+00
-2.37465188e-01 7.79369712e-01 -7.42100775e-01 -3.93523425e-01
2.11782828e-01 6.39693379e-01 -6.85833812e-01 4.33737040e-02
1.76215813e-01 8.88337553e-01 -1.21001415e-01 2.52248049e-01
9.04852927e-01 -3.32113177e-01 5.21301806e-01 -7.33495176e-01
-1.36794895e-01 3.05492520e-01 -1.12846208e+00 2.35792804e+00
-3.95727575e-01 3.37345541e-01 3.16594064e-01 -1.53639531e+00
7.51962960e-01 3.96741152e-01 1.07762623e+00 -1.06694925e+00
1.81908265e-01 6.06523275e-01 -1.53613701e-01 -1.20919518e-01
2.69189537e-01 -4.74071711e-01 1.80094138e-01 7.46738374e-01
4.37351108e-01 3.33187193e-01 -2.22564861e-01 2.51574945e-02
1.11079788e+00 -3.89572233e-02 -1.24700733e-01 -8.45364869e-01
6.02582514e-01 -1.07708059e-01 2.58455336e-01 7.86416471e-01
3.85678327e-03 5.08544505e-01 7.08183423e-02 -8.67958665e-01
-1.51845884e+00 -1.00074244e+00 -5.91856122e-01 1.02013409e+00
-8.39785486e-02 -1.26497641e-01 -7.84397900e-01 -2.05549151e-01
-3.01879714e-03 2.55589455e-01 -6.35467708e-01 -2.17978284e-01
-1.08814633e+00 -8.46052170e-01 4.49994415e-01 2.44460776e-01
2.76442859e-02 -7.92908847e-01 -5.67497194e-01 6.39412761e-01
-3.31169069e-01 -9.20044124e-01 -5.13356507e-01 4.76861000e-01
-1.33666515e+00 -8.32757413e-01 -7.91927874e-01 -8.68229866e-01
8.32582057e-01 1.37916386e-01 1.04865825e+00 -1.50858779e-02
-4.31199521e-01 1.25761524e-01 1.91493571e-01 1.45440221e-01
-4.96613622e-01 1.32263273e-01 2.11952224e-01 -5.54720573e-02
2.18835935e-01 -1.16708839e+00 -8.07829320e-01 1.36144191e-01
-8.77535284e-01 4.46020365e-02 6.78646564e-01 1.01822996e+00
1.15158510e+00 -1.66871324e-01 4.48591501e-01 -7.20183909e-01
3.20736557e-01 -3.63250762e-01 -6.60883188e-01 -4.47570067e-03
-6.69034362e-01 3.28143716e-01 5.61217189e-01 -4.76570010e-01
-4.12523896e-01 3.33406538e-01 -1.26803532e-01 -4.98008937e-01
1.66420579e-01 3.65614533e-01 2.86413878e-01 -7.01992154e-01
7.90269256e-01 5.42595804e-01 3.45126450e-01 -5.87410808e-01
3.17963511e-01 3.64969641e-01 8.55935395e-01 -6.84330523e-01
5.62775314e-01 7.02306271e-01 4.83323872e-01 -8.12988281e-01
-5.20756781e-01 -4.46528703e-01 -1.00021064e+00 -2.58360833e-01
9.13345814e-01 -8.15256894e-01 -8.20532382e-01 1.95718378e-01
-9.94590402e-01 -1.31989047e-01 -3.34251970e-01 6.67534351e-01
-9.44794178e-01 4.71669406e-01 -8.52510333e-01 -1.92141235e-01
-5.24940252e-01 -1.33101511e+00 1.15738392e+00 -4.81337845e-01
-1.43520087e-02 -8.04574430e-01 2.41771981e-01 2.70618200e-01
6.34816825e-01 3.33889812e-01 1.13213003e+00 -2.10089609e-01
-6.01597548e-01 -1.50578499e-01 -1.06770642e-01 8.52822736e-02
-3.47359538e-01 -1.07123601e+00 -9.55899656e-01 -6.16158605e-01
5.27804196e-01 -2.79258430e-01 7.09257364e-01 6.79225087e-01
1.23943055e+00 -2.65791833e-01 -1.06737591e-01 9.66417849e-01
1.52503133e+00 -1.46564664e-02 5.98871827e-01 2.65582293e-01
7.89208710e-01 5.69808960e-01 8.97670761e-02 4.28952366e-01
3.17538857e-01 7.78747737e-01 1.32028833e-01 -1.52246058e-01
-3.25529128e-01 -1.26067474e-01 3.43399569e-02 1.38306403e+00
-2.98826154e-02 6.73658907e-01 -9.16632056e-01 4.81345505e-01
-1.49455118e+00 -9.10736322e-01 -5.64906225e-02 2.38242960e+00
9.82901692e-01 -3.64384919e-01 1.21436357e-01 2.13376492e-01
4.71083730e-01 -2.75687985e-02 -6.14334106e-01 3.00824232e-02
-6.22493364e-02 6.26770854e-01 7.79786527e-01 7.13051856e-01
-1.00967360e+00 3.75899255e-01 6.62877607e+00 8.16411674e-01
-1.32309830e+00 5.59310794e-01 4.62509066e-01 -5.11513837e-03
-1.95758387e-01 -3.11625540e-01 -3.72315884e-01 1.49437204e-01
1.24790931e+00 4.31350946e-01 9.80960488e-01 4.57788199e-01
-8.88951309e-03 1.26558110e-01 -1.13003922e+00 1.32801855e+00
-1.73396081e-01 -1.83747041e+00 1.85940377e-02 3.86289567e-01
4.38384682e-01 4.26682740e-01 7.84072652e-02 -1.59193814e-01
-2.82792807e-01 -1.15539467e+00 6.68744683e-01 6.29389465e-01
1.15895510e+00 -7.48388708e-01 4.11635727e-01 3.28494996e-01
-9.02262509e-01 1.54403821e-02 -7.84048438e-01 4.50029522e-01
2.13791803e-03 6.54626608e-01 -7.13571787e-01 2.91610301e-01
5.53386867e-01 2.73085654e-01 -1.23863980e-01 6.62884712e-01
5.02257407e-01 2.32979894e-01 -2.90483445e-01 6.26193523e-01
7.44287968e-02 -2.92034507e-01 3.89031112e-01 1.10138798e+00
2.81003922e-01 8.70688632e-02 1.29825681e-01 8.87101710e-01
-4.14948678e-03 1.91776529e-01 -2.70591110e-01 -1.36727002e-02
3.15784276e-01 1.36801159e+00 -7.43279099e-01 -2.80420333e-01
-1.97359234e-01 9.17678118e-01 5.03805041e-01 5.23214713e-02
-4.97760475e-01 -6.61557093e-02 2.25137293e-01 5.94849050e-01
3.65225762e-01 -5.11798084e-01 -2.28516564e-01 -1.04563725e+00
-1.34902671e-01 -9.80751097e-01 8.48662332e-02 -4.26491857e-01
-1.11383510e+00 5.60269773e-01 -1.67878300e-01 -8.48146081e-01
-8.90676528e-02 -5.27092457e-01 1.22208461e-01 1.09473312e+00
-1.42393196e+00 -9.65348959e-01 -6.73308596e-02 8.68276119e-01
-2.59642929e-01 -6.16431385e-02 1.24005353e+00 8.95857930e-01
4.68658283e-02 4.20432299e-01 3.84973347e-01 8.67161006e-02
5.40740371e-01 -1.12614930e+00 1.61033347e-01 2.96099275e-01
-4.35431711e-02 8.25673759e-01 5.64399540e-01 -4.28209931e-01
-1.87995398e+00 -7.25836635e-01 7.93687940e-01 -1.99172810e-01
6.10017002e-01 -5.56561530e-01 -9.27591980e-01 3.93920034e-01
-2.79736578e-01 3.63287926e-01 7.88445771e-01 -3.13389122e-01
-3.10391754e-01 -1.90469146e-01 -1.32785273e+00 8.75174105e-02
9.07125831e-01 -1.00555706e+00 -2.25719869e-01 6.15891159e-01
2.89427191e-01 -6.56198561e-01 -1.52158344e+00 7.16455430e-02
7.70158112e-01 -7.43605912e-01 1.20647025e+00 -4.25040543e-01
1.71882033e-01 -2.89455861e-01 -3.90333742e-01 -8.70049238e-01
-6.20363891e-01 -4.67565596e-01 -3.27549666e-01 4.39435720e-01
2.30242927e-02 -3.48305076e-01 8.28625441e-01 3.25545341e-01
-2.30378687e-01 -7.45760620e-01 -1.32097328e+00 -3.97715777e-01
2.20421746e-01 -1.01425231e-01 3.97087932e-01 1.13176060e+00
-3.16906601e-01 1.36717021e-01 -5.28150976e-01 1.81450158e-01
1.22094238e+00 2.03170568e-01 2.32309684e-01 -1.17650032e+00
-5.94623923e-01 -1.30639046e-01 -3.29071015e-01 -1.06820655e+00
-1.11715831e-01 -1.37432659e+00 -1.52659407e-02 -1.06847215e+00
3.03307146e-01 -7.42929697e-01 -5.34632266e-01 -3.37374769e-02
5.39012313e-01 3.99318278e-01 4.42796946e-02 6.60490155e-01
-2.46564478e-01 1.39323086e-01 1.48621166e+00 -7.18582720e-02
9.39933732e-02 -4.75383550e-01 -3.84818077e-01 2.75903732e-01
3.69344890e-01 -8.59459937e-01 -2.30056882e-01 -4.89170194e-01
2.77066410e-01 6.00463390e-01 4.75452185e-01 -1.06025970e+00
3.32440555e-01 3.44352901e-01 6.59223318e-01 -6.13852441e-01
2.49441579e-01 -7.99405158e-01 4.34489518e-01 5.68952799e-01
-4.41226214e-01 8.40137079e-02 -4.99121919e-02 4.30628508e-01
-4.30829637e-02 -1.73679024e-01 1.08827436e+00 -2.50347137e-01
-1.00449502e-01 5.82520604e-01 -2.26425260e-01 -1.75870463e-01
3.77333075e-01 -7.71755949e-02 1.23864233e-01 -3.49062122e-02
-1.14039576e+00 -4.23012942e-01 3.41120601e-01 -3.66191417e-02
6.02322340e-01 -1.57720876e+00 -5.27986765e-01 3.23556274e-01
-3.53961259e-01 -1.09455191e-01 5.67476809e-01 1.42455971e+00
-6.71403527e-01 4.86241072e-01 -5.94399333e-01 -6.54648304e-01
-7.45495677e-01 6.54580832e-01 7.29976952e-01 -5.92706501e-01
-1.04695570e+00 6.61740482e-01 1.52272731e-01 -6.54925227e-01
1.46009624e-01 -1.48211598e-01 -1.03175333e-02 5.69000542e-02
7.30816185e-01 4.34525222e-01 5.25103390e-01 -8.87862027e-01
-3.19907725e-01 6.42604470e-01 -8.96826200e-03 -1.79159060e-01
1.71159220e+00 -1.93386264e-02 -5.08082330e-01 1.84878856e-01
1.53302908e+00 -1.38988405e-01 -9.96540666e-01 -5.26633263e-01
1.11705065e-01 -7.14408755e-02 5.96272290e-01 -5.61758578e-01
-1.11094916e+00 9.47489738e-01 1.05905461e+00 -2.51958549e-01
1.00371897e+00 -1.68356925e-01 1.06625283e+00 4.48710948e-01
6.72485948e-01 -1.10691226e+00 -2.40657687e-01 2.25826800e-01
9.23426151e-01 -8.18836927e-01 1.36543319e-01 5.41482233e-02
-5.79848662e-02 1.36144960e+00 -1.88768491e-01 -4.54027414e-01
9.37242627e-01 3.40762883e-01 -2.11993009e-01 -5.81387520e-01
-1.11079879e-01 4.97426182e-01 4.59731430e-01 5.94934464e-01
3.49991679e-01 1.60375908e-01 -3.22340846e-01 3.04555207e-01
-1.76999167e-01 -1.23250775e-01 1.66253105e-01 7.72487521e-01
-5.43911397e-01 -1.07268584e+00 -2.90546507e-01 6.27223909e-01
-4.23736751e-01 -1.87755704e-01 1.80266038e-01 4.27328259e-01
1.37003914e-01 2.91525185e-01 1.06095783e-01 -3.47791523e-01
1.29820123e-01 -1.45532027e-01 1.00677538e+00 -3.01944852e-01
-5.19534707e-01 3.40973258e-01 -2.48823419e-01 -9.84129071e-01
-4.76226479e-01 -8.06147397e-01 -1.41037810e+00 -3.30442011e-01
-1.01782225e-01 -5.28958961e-02 9.17563975e-01 8.16131413e-01
3.10824156e-01 2.47556627e-01 5.96934140e-01 -1.11977613e+00
-8.59053254e-01 -7.71547556e-01 -8.19228649e-01 3.83278102e-01
4.26061332e-01 -6.93906903e-01 1.92337390e-02 -1.81150422e-01] | [13.473820686340332, -2.400557518005371] |
39b3783e-6e0c-4107-9e72-3e7546ace371 | exploring-word-alignment-towards-an-efficient | null | null | https://aclanthology.org/2022.loresmt-1.13 | https://aclanthology.org/2022.loresmt-1.13.pdf | Exploring Word Alignment towards an Efficient Sentence Aligner for Filipino and Cebuano Languages | Building a robust machine translation (MT) system requires a large amount of parallel corpus which is an expensive resource for low-resourced languages. The two major languages being spoken in the Philippines which are Filipino and Cebuano have an abundance in monolingual data that this study took advantage of attempting to find the best way to automatically generate parallel corpus out from monolingual corpora through the use of bitext alignment. Byte-pair encoding was applied in an attempt to optimize the alignment of the source and target texts. Results have shown that alignment was best achieved without segmenting the tokens. Itermax alignment score is best for short-length sentences and match or argmax alignment score are best for long-length sentences. | ['Kristine Mae M. Adlaon', 'Jenn Leana Fernandez'] | null | null | null | null | loresmt-coling-2022-10 | ['word-alignment'] | ['natural-language-processing'] | [ 2.52380162e-01 -1.24929167e-01 -2.22682923e-01 -4.16446537e-01
-1.09192157e+00 -7.73966074e-01 7.81938136e-01 3.95884104e-02
-7.00519204e-01 1.09690332e+00 4.63324249e-01 -7.06280470e-01
3.06004137e-01 -5.40903866e-01 -4.03770030e-01 -4.77976590e-01
1.83871850e-01 8.49400520e-01 -1.71102360e-01 -6.56330526e-01
5.98263741e-01 3.79052937e-01 -7.56656170e-01 2.13564828e-01
8.77778769e-01 -1.94366321e-01 7.05896080e-01 7.60910451e-01
-1.71266973e-01 5.49166240e-02 -5.85668921e-01 -6.38158798e-01
7.32468724e-01 -7.99125969e-01 -1.13970900e+00 -2.35516191e-01
2.41942555e-01 2.09920295e-02 7.82568678e-02 9.18269932e-01
6.83419704e-01 -2.31098652e-01 5.42923629e-01 -7.21709907e-01
-4.28463489e-01 8.55760336e-01 -4.22118485e-01 6.39392018e-01
4.03803527e-01 1.44791931e-01 9.49983180e-01 -1.02191353e+00
8.09054494e-01 1.24146020e+00 3.38584065e-01 1.98172837e-01
-1.07841015e+00 -5.04612565e-01 -6.58756912e-01 1.57488901e-02
-1.35511529e+00 -5.75695634e-01 4.53682303e-01 -2.83589542e-01
1.56516469e+00 2.27568850e-01 4.64984745e-01 7.60161817e-01
6.75742805e-01 1.68385878e-01 1.43704391e+00 -1.09634757e+00
-4.06273931e-01 7.78322369e-02 -1.87011823e-01 3.54153812e-01
1.42078340e-01 9.08367783e-02 -3.74918759e-01 -9.52573270e-02
5.42906582e-01 -7.52483487e-01 3.51247221e-01 4.88037229e-01
-1.68937993e+00 1.01523638e+00 -2.32741937e-01 8.92272234e-01
-3.35314751e-01 -4.47511345e-01 5.52454412e-01 8.61580789e-01
3.95982414e-01 4.70328331e-01 -4.11304414e-01 -5.72239280e-01
-9.12513614e-01 1.32187158e-01 6.99448168e-01 9.92862046e-01
5.65602660e-01 1.68946516e-02 2.32773140e-01 1.14229226e+00
2.07157835e-01 8.95513356e-01 7.62737334e-01 -4.67014581e-01
1.16252851e+00 2.93330729e-01 -1.71636030e-01 -5.77834427e-01
1.60673000e-02 1.20920025e-01 -3.39570910e-01 -8.79148841e-02
5.17049074e-01 -4.00638223e-01 -6.95911705e-01 1.36440408e+00
2.88023233e-01 -7.68433690e-01 5.21361113e-01 5.43859184e-01
1.13323405e-01 8.96556258e-01 -7.51747340e-02 -5.13391912e-01
1.22579634e+00 -9.18315113e-01 -5.71770728e-01 -2.90889621e-01
8.16896498e-01 -1.93743312e+00 1.00214267e+00 -1.11326151e-01
-1.19489920e+00 -4.20311928e-01 -1.16939008e+00 -1.07473396e-01
-1.02919914e-01 -6.23172671e-02 2.89294004e-01 8.25353861e-01
-9.79231536e-01 4.98896778e-01 -7.42880344e-01 -7.17730999e-01
-1.82038531e-01 5.68138957e-01 -7.78569758e-01 6.59607118e-03
-1.04021049e+00 1.57214022e+00 5.51546395e-01 -5.16852625e-02
-2.51647413e-01 1.36987433e-01 -5.53001702e-01 -4.86336589e-01
-3.04128259e-01 -2.87811339e-01 1.12403762e+00 -1.17742693e+00
-1.69244194e+00 1.04599369e+00 -3.85090977e-01 -1.62087381e-01
4.65948135e-01 5.09579666e-02 -4.27675217e-01 -1.49427012e-01
3.46324593e-01 5.16770005e-01 4.74713951e-01 -5.74660838e-01
-8.54459882e-01 -2.58733630e-01 -6.26225412e-01 5.89303136e-01
9.29708108e-02 8.38242531e-01 5.20802774e-02 -6.68091357e-01
1.67833686e-01 -1.21325195e+00 -1.68497786e-01 -1.12189245e+00
-4.04480100e-02 -1.44996062e-01 5.17974675e-01 -1.34489250e+00
1.18439746e+00 -1.73767257e+00 1.65532500e-01 2.34213248e-01
-6.48097396e-01 1.88340828e-01 -3.18255872e-01 1.17270041e+00
-1.48305267e-01 1.34788781e-01 -9.78060141e-02 1.77332327e-01
-1.79350063e-01 5.85168362e-01 -1.35401353e-01 4.17104751e-01
9.76322889e-02 7.32794523e-01 -7.13778675e-01 -9.22801495e-01
4.76893932e-02 7.01175928e-02 -6.41070008e-02 2.66244382e-01
1.98325008e-01 4.57757920e-01 -1.81626156e-01 5.67587733e-01
3.15782219e-01 5.31441391e-01 2.86446273e-01 4.40188736e-01
-5.69740951e-01 7.83165932e-01 -6.42454147e-01 1.61814749e+00
-3.92113984e-01 7.44718552e-01 -3.41137677e-01 -5.40692270e-01
1.19860625e+00 6.75692081e-01 2.71043450e-01 -6.38162673e-01
2.12436140e-01 7.56482422e-01 6.82550490e-01 -6.20270610e-01
8.04019332e-01 -4.14997280e-01 -2.13067487e-01 7.53135920e-01
-1.18251354e-01 -3.23000431e-01 6.64754748e-01 -1.07373014e-01
6.36641562e-01 1.83665335e-01 6.38848484e-01 -5.50327659e-01
5.83385706e-01 4.27334905e-01 5.50670683e-01 1.47075549e-01
-3.62271853e-02 5.83980739e-01 1.18774936e-01 -4.86119896e-01
-1.82768023e+00 -7.96380818e-01 -1.82208031e-01 9.65452135e-01
-5.94614625e-01 -1.26232743e-01 -7.54382491e-01 -5.03372908e-01
-6.51995540e-01 7.23756671e-01 4.21394594e-02 5.24116218e-01
-1.33262253e+00 -9.10978973e-01 7.80941367e-01 -5.65761663e-02
1.07305110e-01 -1.17989385e+00 -1.17643669e-01 4.86799330e-01
-6.55362129e-01 -8.36049736e-01 -8.05221319e-01 2.67068129e-02
-1.12724602e+00 -7.33763397e-01 -7.69849360e-01 -1.29805636e+00
8.06649327e-01 -6.08676486e-02 9.12603378e-01 -2.35026240e-01
4.38634790e-02 -4.37096685e-01 -2.92434126e-01 -5.41520238e-01
-1.05583382e+00 5.20383120e-01 -5.54892607e-03 -6.40664816e-01
8.28538060e-01 -5.58822632e-01 -2.63807308e-02 2.70633668e-01
-6.40688777e-01 1.72193244e-01 8.64318967e-01 8.38600814e-01
3.33435625e-01 -5.32739460e-01 4.30708289e-01 -6.39330268e-01
8.57797384e-01 -3.80420893e-01 -4.33768034e-01 3.26621830e-01
-5.70413351e-01 3.16137344e-01 5.12352645e-01 -2.97920942e-01
-7.59654880e-01 2.94473991e-02 -4.31191534e-01 5.24455369e-01
-9.72784534e-02 4.63208854e-01 -1.30371964e-02 1.14852957e-01
7.08281338e-01 2.97804683e-01 7.00518042e-02 -3.82694364e-01
-2.48745698e-02 1.28596413e+00 4.64764833e-01 -7.58074582e-01
7.46749878e-01 -1.69602484e-01 -1.10450327e-01 -1.06974292e+00
5.17933406e-02 -4.38459933e-01 -1.06425095e+00 -9.47653204e-02
5.84043443e-01 -6.38517797e-01 4.65604924e-02 3.14987153e-01
-1.22915995e+00 -1.75762326e-01 9.09720138e-02 8.28765750e-01
-5.22912860e-01 3.92611742e-01 -5.91211081e-01 -4.33378756e-01
-7.20429182e-01 -1.27601492e+00 7.07047403e-01 -5.96350320e-02
-7.34394729e-01 -9.24136043e-01 7.06544459e-01 4.41612452e-01
3.52280408e-01 -6.70918031e-04 1.15565491e+00 -8.95006180e-01
-2.28054807e-01 -1.95313454e-01 7.41369799e-02 3.19877297e-01
4.37645763e-01 1.59379423e-01 -2.66932487e-01 -3.67450923e-01
-5.27173951e-02 -6.52230456e-02 -1.88425571e-01 -1.60115045e-02
-2.78166294e-01 -5.50265789e-01 3.62724327e-02 9.09956694e-02
1.42012072e+00 4.79077488e-01 5.93497694e-01 6.42561913e-01
4.95425493e-01 7.56699324e-01 6.06931806e-01 -7.35562593e-02
3.96450788e-01 7.54048169e-01 -3.93125057e-01 1.49983630e-01
-1.00141332e-01 -3.01919997e-01 6.85337305e-01 1.72156858e+00
-1.63670599e-01 1.33212581e-02 -1.15349352e+00 8.36022496e-01
-1.54425681e+00 -8.87395978e-01 -3.64152670e-01 2.33995962e+00
1.27158785e+00 -7.93187767e-02 1.95477113e-01 8.01331922e-02
7.28331506e-01 -8.22890773e-02 2.85148293e-01 -1.10404730e+00
-1.56262904e-01 4.95425910e-01 7.89413571e-01 1.02998602e+00
-5.37357807e-01 1.15219164e+00 7.12501764e+00 7.16515660e-01
-1.31462657e+00 1.82924941e-01 2.59597868e-01 -1.27304673e-01
-2.39637196e-01 4.55556899e-01 -8.99942160e-01 5.37440777e-01
1.49837995e+00 -4.41424489e-01 5.84163845e-01 3.08508337e-01
4.54305887e-01 -9.33958590e-02 -1.06896496e+00 6.73834264e-01
1.39698358e-02 -1.02064645e+00 8.57879058e-04 1.97223216e-01
9.47877645e-01 5.98066568e-01 -3.33115608e-01 -2.74077673e-02
3.92237961e-01 -9.52111006e-01 6.74704671e-01 -7.28643313e-02
6.76371455e-01 -8.66796076e-01 7.90391505e-01 5.22972643e-01
-7.66109407e-01 4.62791115e-01 -5.52067876e-01 -3.04645538e-01
3.56272906e-01 1.25616267e-01 -1.30666673e+00 3.85207921e-01
9.48696807e-02 1.29286602e-01 -3.34840655e-01 7.51568437e-01
-3.19185495e-01 7.27075577e-01 -5.27150512e-01 -1.83037817e-01
5.65240979e-01 -6.39093876e-01 8.19631994e-01 1.57656276e+00
6.54502630e-01 -1.75276697e-01 1.95821729e-02 1.92162879e-02
1.85458809e-01 9.57375944e-01 -8.53700995e-01 -3.17839593e-01
3.70789558e-01 8.46038580e-01 -7.15377152e-01 -1.68082699e-01
-3.85190398e-01 9.76165891e-01 2.56310910e-01 5.59441671e-02
-2.46735305e-01 -4.33980435e-01 4.11311328e-01 -5.28039075e-02
-8.18324164e-02 -6.62132382e-01 -5.33887446e-01 -9.80668783e-01
7.62655139e-02 -1.47599566e+00 2.62564927e-01 -2.91653395e-01
-9.87328827e-01 9.61776197e-01 1.03841774e-01 -9.45793033e-01
-9.36933756e-01 -4.23803985e-01 -5.51964760e-01 1.64272928e+00
-9.65469360e-01 -1.41001678e+00 5.92955887e-01 3.07628006e-01
8.71941805e-01 -5.74149668e-01 9.60396349e-01 3.09329867e-01
-3.36806685e-01 5.52016318e-01 3.26544195e-01 2.17942178e-01
8.73410344e-01 -1.05383766e+00 8.08449209e-01 1.03734541e+00
3.81865501e-01 7.86508858e-01 8.57810259e-01 -7.94547498e-01
-1.10712564e+00 -4.97705340e-01 2.13529253e+00 -4.32843983e-01
6.54042423e-01 -2.23139927e-01 -3.81318837e-01 7.59518027e-01
8.78588438e-01 -7.77397692e-01 8.34422529e-01 -3.10516711e-02
-8.79195109e-02 1.12082273e-01 -9.62204397e-01 7.86220372e-01
4.18901980e-01 -5.24573624e-01 -8.91287208e-01 5.32817900e-01
2.11495161e-01 -3.28253329e-01 -9.96848345e-01 -1.23290876e-02
6.10715091e-01 -3.32398921e-01 4.29832965e-01 -4.69232678e-01
6.58914328e-01 -2.42768660e-01 -2.02315807e-01 -1.36592579e+00
-6.75221831e-02 -1.04984641e+00 9.23036873e-01 1.25991488e+00
9.54106987e-01 -6.92941725e-01 5.49812078e-01 4.31980103e-01
-1.21273652e-01 -2.53523618e-01 -1.02154064e+00 -7.83945441e-01
4.94814634e-01 -1.27060683e-02 5.13008118e-01 8.70185077e-01
3.54702473e-01 8.14913392e-01 -4.31624591e-01 -1.38360292e-01
4.37756538e-01 -2.93470733e-02 8.19545567e-01 -5.35772443e-01
-2.84014881e-01 -4.07281905e-01 -1.65514141e-01 -5.55285394e-01
1.01247160e-02 -1.06751680e+00 2.51591038e-02 -1.34448850e+00
1.85653493e-01 -6.15059733e-01 2.79610455e-01 3.95419806e-01
-1.98315531e-01 3.15155149e-01 2.10797444e-01 5.57539105e-01
4.49884504e-01 5.41897938e-02 9.90382969e-01 1.22651622e-01
-2.46931106e-01 -1.05532240e-02 -2.64725268e-01 2.53589094e-01
1.08322251e+00 -9.14158046e-01 -2.37615973e-01 -6.70335174e-01
2.91586012e-01 2.18059540e-01 -7.11680710e-01 -4.84437972e-01
-4.24605906e-02 -5.87394059e-01 9.34922099e-02 -5.16137064e-01
-9.97514352e-02 -6.68705225e-01 5.21620035e-01 5.04829168e-01
-2.85874665e-01 1.04647028e+00 1.56166032e-01 -2.85284162e-01
-3.76472145e-01 -7.18805015e-01 7.82614410e-01 -3.03449124e-01
-3.35877985e-01 -1.89357266e-01 -6.19957447e-01 -1.22338004e-01
9.99538600e-01 -4.02023315e-01 -3.77855860e-02 -7.19705224e-02
-9.84712094e-02 -1.96092948e-01 6.23835087e-01 6.24331355e-01
2.45035775e-02 -1.28087783e+00 -1.34745884e+00 2.51423270e-01
-2.24073883e-02 -5.68832517e-01 -4.11969543e-01 6.65748656e-01
-1.28069687e+00 7.64948666e-01 -8.79932582e-01 -3.26862842e-01
-1.73314977e+00 1.23155408e-01 -8.72305855e-02 -4.59486604e-01
-3.90025765e-01 5.40371120e-01 -5.33687651e-01 -5.72068751e-01
-3.72641206e-01 2.64536887e-01 -1.63966976e-02 -2.67142743e-01
2.46265516e-01 3.94159406e-01 1.61025062e-01 -1.31229365e+00
-8.09650421e-02 3.57991457e-01 -2.96924323e-01 -8.00104618e-01
1.16780126e+00 -3.09829772e-01 -7.00386703e-01 3.24697822e-01
1.26441121e+00 3.51474196e-01 -4.84228581e-01 -1.00648990e-02
2.74967581e-01 -6.49014533e-01 -3.06230098e-01 -4.82841134e-01
-3.59987229e-01 7.21136987e-01 3.43665659e-01 -3.52207750e-01
7.16235220e-01 -5.06188750e-01 9.01864767e-01 3.67012113e-01
4.99390483e-01 -1.28922403e+00 -5.98151386e-01 8.98258150e-01
7.57253408e-01 -1.05472922e+00 -2.46538948e-02 -2.04455052e-02
-5.17364740e-01 1.19961452e+00 1.52056515e-01 8.10750276e-02
1.16570733e-01 2.71987438e-01 5.14353633e-01 2.53520221e-01
-6.36056900e-01 2.29916930e-01 1.30254135e-01 3.73171031e-01
1.02093995e+00 2.41618484e-01 -1.51534390e+00 -3.06836367e-01
-8.90125632e-01 -3.64954770e-01 5.11655450e-01 1.10942495e+00
-3.11725676e-01 -2.12797570e+00 -6.06017947e-01 2.77685970e-01
-1.05786800e+00 -5.16172767e-01 -5.01756132e-01 7.50555098e-01
-4.24244702e-02 9.24898982e-01 5.10937832e-02 -1.63160846e-01
-4.40368131e-02 3.18040937e-01 6.71105146e-01 -5.99516749e-01
-1.11473548e+00 2.86198586e-01 4.68736678e-01 6.41395077e-02
-1.64616942e-01 -8.81706476e-01 -9.68916476e-01 -6.84867442e-01
-3.15442175e-01 5.64263284e-01 1.21767223e+00 1.19164944e+00
-2.02900037e-01 -3.22694242e-01 6.65444672e-01 -5.10696709e-01
-6.35319471e-01 -1.41517365e+00 -1.64780766e-01 1.26369506e-01
-1.28394708e-01 2.72703588e-01 6.30417168e-02 3.71350706e-01] | [11.446974754333496, 10.394775390625] |
a4127199-8c64-4093-9fd3-abb758e3fd2b | unsupervised-visual-program-induction-with | null | null | https://openreview.net/forum?id=t14vYukzfvF | https://openreview.net/pdf?id=t14vYukzfvF | Unsupervised Visual Program Induction with Function Modularization | Program induction serves as one way to analog the ability of human thinking. However, existing methods could only tackle the task under simple scenarios (Fig~\ref{fig:task_examples}(a),(b)). When it comes to complex scenes, e.g., the visual scenes, current program induction methods fail due to the huge program action space. In this paper, to the best of our knowledge, we are the first to tackle this problem. We propose a novel task named {\it unsupervised visual program induction} in complex visual scenes that require complex primitive functions. Solving this task faces two challenges: i) modeling complex primitive functions for complex visual scenes is very difficult, and ii) employing complex functions in the unsupervised program induction suffers from a huge and heterogeneous program action space. To tackle these challenges, we propose the Self-Exploratory-Modularized-Function (SEMF) model, which can jointly model individual function selection and its parameters through a unified modular block. Moreover, a Monto-Carlo-Tree-Search (MCTS) based Self-Exploratory algorithm is proposed to explore program space with modularized function as prior. The exploratory results, in turn, guide the training of these modularized functions. Our extensive experiments demonstrate that the proposed SEFM model outperforms all the existing baselines in model performance, training efficiency, and model generalization ability. | ['Wenwu Zhu', 'Ziwei Zhang', 'Xin Wang', 'Xuguang Duan'] | 2021-09-29 | null | null | null | null | ['program-induction'] | ['computer-code'] | [ 2.77383149e-01 -3.46847445e-01 -4.72981870e-01 -4.01602864e-01
-2.59950668e-01 -6.38507009e-01 4.71547008e-01 4.81035523e-02
-2.52159148e-01 3.57559711e-01 -2.65671045e-01 -6.91364944e-01
1.33759771e-02 -6.77877009e-01 -7.78905749e-01 -4.45280284e-01
-1.06791630e-01 2.60215670e-01 2.99620211e-01 1.02506183e-01
5.78869045e-01 1.78917646e-01 -1.83433020e+00 5.54377198e-01
1.58526444e+00 6.10125899e-01 7.79967964e-01 6.22788429e-01
-3.13608110e-01 1.13847327e+00 -4.37417328e-01 -1.23360008e-01
-2.41737947e-01 -2.65399396e-01 -9.89184976e-01 9.25842375e-02
1.13290861e-01 -3.17038685e-01 2.50036288e-02 1.13783622e+00
3.85194942e-02 3.55613887e-01 5.67216933e-01 -1.37254596e+00
-4.46589589e-01 6.68067396e-01 -6.20088279e-01 9.60137248e-02
4.27667707e-01 4.92877275e-01 1.02876937e+00 -9.67075348e-01
6.26398802e-01 1.18572164e+00 3.72846991e-01 4.18241411e-01
-1.59145689e+00 -5.58797479e-01 5.24538100e-01 3.15608859e-01
-1.23906875e+00 -5.87812774e-02 9.95815933e-01 -8.12473178e-01
1.05906355e+00 2.42308587e-01 7.56503105e-01 7.92393386e-01
-1.75456002e-01 1.41111708e+00 1.25893104e+00 -3.54616970e-01
2.32710093e-01 1.95158347e-01 4.41276252e-01 1.08401203e+00
-1.00459576e-01 -7.19847456e-02 -5.66686273e-01 -1.17874011e-01
6.11898243e-01 1.63276672e-01 -3.62586468e-01 -5.55598557e-01
-1.18269372e+00 8.70165706e-01 5.70478141e-01 3.13992232e-01
-6.19243346e-02 2.46808171e-01 3.87189031e-01 4.53580290e-01
9.55232158e-02 5.83456159e-01 -4.71489042e-01 -2.43480831e-01
-9.34669495e-01 2.62436301e-01 6.80703342e-01 1.20157743e+00
9.79702175e-01 1.90214887e-01 -1.44083083e-01 7.09367692e-01
3.53201360e-01 9.78288651e-02 3.06277990e-01 -6.37198269e-01
5.45382619e-01 1.04366469e+00 -2.47839108e-01 -8.60657990e-01
-2.87375003e-01 -3.76173764e-01 -6.90224648e-01 3.01331669e-01
2.61720359e-01 1.65456101e-01 -8.71245384e-01 1.64724565e+00
3.19244832e-01 -1.30511403e-01 -4.62071896e-01 3.91570836e-01
1.02987289e+00 7.76586592e-01 2.61501223e-01 -2.73930401e-01
1.20762932e+00 -1.35759819e+00 -3.96515757e-01 -3.35245997e-01
7.84504235e-01 -5.28099418e-01 1.71033311e+00 5.17080545e-01
-9.64479268e-01 -9.06242311e-01 -8.76632690e-01 1.49779588e-01
-2.31818065e-01 3.17115903e-01 1.27587938e+00 6.96985245e-01
-1.05527091e+00 3.25497597e-01 -8.54153514e-01 -1.80853277e-01
6.40590429e-01 3.83281916e-01 -6.86013997e-02 -2.74066836e-01
-5.63334167e-01 4.45726335e-01 6.24382675e-01 -2.69323081e-01
-1.16324079e+00 -6.09963059e-01 -9.79465306e-01 3.05567116e-01
5.64079165e-01 -7.10061491e-01 1.22384548e+00 -1.11100173e+00
-1.31876349e+00 7.19553113e-01 -5.12894809e-01 -5.26171178e-02
1.02737397e-01 -1.02489322e-01 1.97941139e-01 4.39954661e-02
9.64228436e-02 6.35159612e-01 8.15836608e-01 -1.54677618e+00
-3.88921738e-01 -2.29516521e-01 3.44620287e-01 9.03509557e-02
-4.02896851e-01 3.15642774e-01 -5.63569725e-01 -7.33737111e-01
2.50441744e-03 -6.97390676e-01 -2.51142859e-01 -1.44308984e-01
-4.62478757e-01 -5.04354835e-01 6.07519507e-01 -5.05587220e-01
1.83099282e+00 -2.18747878e+00 2.84538686e-01 1.04686409e-01
3.77983958e-01 9.06075314e-02 4.07458991e-02 1.81370348e-01
-7.01776966e-02 3.23518753e-01 -3.46843392e-01 -2.58396059e-01
1.25352204e-01 1.32933781e-01 -1.72653779e-01 3.57458666e-02
6.16109446e-02 1.03521335e+00 -9.35747802e-01 -8.47340822e-01
4.57746647e-02 -7.43462741e-02 -1.07641530e+00 4.18644071e-01
-6.49555683e-01 6.02245271e-01 -5.37279546e-01 1.00927997e+00
5.70757747e-01 -4.87365216e-01 1.67521805e-01 7.89775513e-03
-2.80893892e-01 1.43685833e-01 -1.06241775e+00 1.77056003e+00
-4.72891301e-01 4.93675590e-01 -5.54470606e-02 -1.24611235e+00
8.47368062e-01 -7.05330521e-02 1.96565390e-01 -5.43741643e-01
-9.29536521e-02 -5.70593514e-02 2.43969932e-01 -7.80956984e-01
2.78864920e-01 1.54048935e-01 -9.73626599e-02 3.78885061e-01
1.96595654e-01 -2.20738813e-01 4.80812371e-01 1.49167120e-01
1.18357384e+00 6.20328963e-01 4.00954038e-01 -2.37826005e-01
5.54347336e-01 2.62272179e-01 5.41367173e-01 9.84644294e-01
-3.98235396e-02 1.28538355e-01 7.72284746e-01 -5.02805412e-01
-7.12431312e-01 -8.56907547e-01 9.93494987e-02 1.52481806e+00
-9.58680082e-03 -7.50461161e-01 -6.33489907e-01 -7.37263799e-01
-2.25548431e-01 8.63422692e-01 -4.03258324e-01 8.21755901e-02
-6.97190940e-01 -6.90747142e-01 2.86290616e-01 5.66363692e-01
6.95633113e-01 -1.20836568e+00 -7.67337501e-01 6.97348639e-02
-1.65269047e-01 -7.84395516e-01 -3.33394200e-01 3.42058301e-01
-1.04312813e+00 -1.08307350e+00 -2.65171230e-01 -1.15707588e+00
1.01001048e+00 3.06692988e-01 1.45901895e+00 4.33047503e-01
-2.63585806e-01 3.26709390e-01 -3.76574337e-01 -2.55573392e-01
-1.88396573e-01 -1.20654225e-01 -5.24897218e-01 -2.40846857e-01
1.94029525e-01 -7.18290508e-01 -5.22771239e-01 2.71503806e-01
-8.33448946e-01 5.10312319e-01 7.15827942e-01 1.04020905e+00
5.78416705e-01 5.08639812e-01 2.21371412e-01 -1.12715983e+00
5.45216560e-01 -5.31743944e-01 -6.53220356e-01 3.86029333e-01
-4.76463616e-01 1.26862064e-01 9.72319543e-01 -6.59400105e-01
-1.27287591e+00 3.13907564e-01 1.52847782e-01 -2.95016170e-01
-3.64204884e-01 8.09823632e-01 -3.79680634e-01 -2.48377889e-01
7.45293200e-01 7.57987618e-01 -4.33367521e-01 -5.15222251e-01
2.02629477e-01 2.78460920e-01 3.05152118e-01 -1.17409635e+00
9.17986929e-01 1.94548711e-01 -1.04522280e-01 -6.69595003e-01
-4.27439779e-01 -3.95779043e-01 -6.35468066e-01 -1.79158837e-01
7.80246854e-01 -7.25342691e-01 -9.01146233e-01 4.87992167e-01
-1.24193072e+00 -7.87897468e-01 2.25140199e-01 1.24567382e-01
-6.51472747e-01 6.10050380e-01 -3.95992041e-01 -9.00325894e-01
-1.44120529e-01 -1.58887732e+00 8.01488101e-01 2.22395897e-01
-2.32827485e-01 -9.23162222e-01 -2.42656097e-01 4.48417008e-01
2.61614114e-01 2.37315387e-01 1.46982372e+00 -4.19487566e-01
-9.59795356e-01 1.34973288e-01 -1.48261309e-01 1.50100544e-01
-2.56825775e-01 1.64961815e-01 -7.05674887e-01 -1.34916380e-01
1.05144255e-01 -6.55103981e-01 8.01432490e-01 2.06364155e-01
1.76221168e+00 -2.69608527e-01 -4.09422547e-01 9.11758244e-01
1.29658830e+00 4.62937653e-01 5.17903864e-01 4.61712152e-01
9.03284311e-01 4.45762128e-01 7.62401521e-01 5.78931808e-01
5.42504668e-01 4.78180647e-01 2.44640991e-01 2.01357454e-02
1.27233908e-01 -3.60805124e-01 3.97601783e-01 7.88704157e-01
9.91271902e-03 -6.64499477e-02 -1.15174913e+00 5.00031710e-01
-1.78030729e+00 -9.19966280e-01 -1.62086308e-01 2.03682804e+00
9.90115583e-01 2.12115988e-01 4.35539037e-02 9.20205470e-03
4.93146122e-01 1.05613887e-01 -5.25953948e-01 -3.12230408e-01
2.22067222e-01 4.34362203e-01 -1.75372407e-01 2.13771150e-01
-1.18884170e+00 1.09599626e+00 5.64674377e+00 9.20702219e-01
-9.57549691e-01 -8.48424714e-03 5.60871601e-01 2.12745801e-01
-4.25195724e-01 4.47801501e-01 -7.42991328e-01 5.93196809e-01
3.44110399e-01 -5.00024594e-02 7.06493199e-01 1.30413973e+00
-1.08540699e-01 -3.26578945e-01 -1.27670825e+00 1.05580497e+00
5.78019060e-02 -1.01978517e+00 1.02072671e-01 -2.96565890e-01
6.85193360e-01 -3.78830016e-01 6.00652061e-02 9.36805189e-01
3.65739226e-01 -1.10976934e+00 5.87603509e-01 3.09473038e-01
6.22660339e-01 -7.50072539e-01 1.30814388e-01 6.89588547e-01
-1.52413857e+00 -3.74473274e-01 -6.01890460e-02 -1.01041801e-01
-3.88267338e-01 3.06996047e-01 -6.42229259e-01 2.44612306e-01
7.97749400e-01 4.47423518e-01 -8.90500903e-01 1.24579763e+00
-5.71631074e-01 7.63500690e-01 -5.32337055e-02 -3.56496215e-01
1.06806427e-01 1.28449779e-02 4.22894508e-01 1.22059131e+00
1.23650663e-01 -1.78369880e-01 6.11455083e-01 1.42269623e+00
1.71137214e-01 1.23735599e-01 -6.47492349e-01 -1.35396793e-01
3.73616785e-01 1.13777411e+00 -9.27396953e-01 -3.92002344e-01
-6.39637530e-01 7.25107789e-01 6.62285566e-01 5.82227588e-01
-8.18602204e-01 -4.56067175e-01 2.07928210e-01 -7.44272321e-02
3.32776845e-01 -3.88309211e-01 -6.18654191e-01 -1.38606393e+00
1.40047088e-01 -1.25286472e+00 4.85679209e-01 -7.12447941e-01
-7.54904032e-01 3.68280083e-01 2.90889025e-01 -1.11948049e+00
-1.15505040e-01 -5.47771811e-01 -8.91936183e-01 6.58309460e-01
-1.31203258e+00 -1.20317495e+00 -4.32739377e-01 7.97007263e-01
8.08343828e-01 -1.85706660e-01 7.11848736e-01 1.08653471e-01
-7.16675222e-01 7.47553766e-01 -1.36735454e-01 -2.15534009e-02
3.23790699e-01 -1.44842350e+00 3.07835430e-01 9.57339287e-01
-1.36572093e-01 1.10249555e+00 5.33665001e-01 -6.90037131e-01
-1.81565058e+00 -8.99612784e-01 5.27212799e-01 -2.65461206e-01
6.38944209e-01 -5.35003662e-01 -9.14100111e-01 5.97970366e-01
2.39474401e-02 -1.44873917e-01 8.14627826e-01 2.59539515e-01
-4.07497853e-01 9.65783745e-02 -8.20646822e-01 7.07946062e-01
1.06691778e+00 -4.32077616e-01 -5.31267285e-01 3.19208175e-01
7.26449370e-01 -2.03399211e-01 -7.07894683e-01 5.21120787e-01
2.05460817e-01 -9.28408384e-01 9.98655677e-01 -5.17661154e-01
6.26024008e-01 -5.07519126e-01 2.05656216e-02 -1.10423565e+00
-3.91685367e-01 -6.57315731e-01 -3.83096486e-01 1.22748506e+00
1.58813789e-01 -3.86879057e-01 6.01392090e-01 3.03194642e-01
-4.26850408e-01 -9.50246513e-01 -5.02056956e-01 -6.64888561e-01
-1.26415566e-01 -5.25276065e-01 4.22693729e-01 8.85231674e-01
1.80284977e-01 4.51582462e-01 -1.35955989e-01 5.10536926e-03
7.10884213e-01 7.28921115e-01 1.09903800e+00 -1.21412361e+00
-9.25869823e-01 -6.98836565e-01 -2.59286650e-02 -1.45689082e+00
3.12280208e-01 -9.77077782e-01 3.44412662e-02 -1.44818020e+00
6.63020432e-01 -5.47275126e-01 -9.05222911e-03 5.70240676e-01
-6.63177371e-01 -5.15650392e-01 2.40956247e-01 3.17211241e-01
-9.53005672e-01 4.51026082e-01 1.29881525e+00 -2.97115356e-01
-3.86217296e-01 9.80150774e-02 -7.00717807e-01 8.53254855e-01
6.54262543e-01 -3.87733579e-01 -7.48064399e-01 -4.09156829e-01
3.09829503e-01 2.68137664e-01 4.01464522e-01 -1.04095280e+00
3.76357496e-01 -5.05303621e-01 3.01591426e-01 -7.60075092e-01
-5.73358499e-02 -5.64163148e-01 -1.04591586e-01 6.09995425e-01
-1.43936425e-01 1.58805847e-01 2.44274512e-01 4.48570490e-01
-3.38420123e-01 -6.24971807e-01 6.00924492e-01 -5.40439725e-01
-1.12232280e+00 2.27244943e-01 -5.62420666e-01 3.09251808e-03
9.15123999e-01 -2.91370511e-01 -3.53281140e-01 -9.99169052e-02
-5.01890004e-01 4.58642960e-01 5.33984005e-01 1.33883640e-01
6.12473667e-01 -1.20984352e+00 -2.58662254e-01 2.22666889e-01
3.30409527e-01 5.17094016e-01 3.82815689e-01 9.63169098e-01
-5.40270269e-01 1.90120608e-01 -1.55032843e-01 -8.17284942e-01
-1.47638023e+00 9.73843157e-01 -1.56475142e-01 -6.72083914e-01
-3.29122663e-01 9.83195186e-01 7.75559425e-01 -6.58985674e-01
3.75914931e-01 -3.62636238e-01 -3.03889662e-01 -1.12792343e-01
4.25457150e-01 1.17298715e-01 -4.29657608e-01 -1.73775539e-01
-1.30894870e-01 3.81248057e-01 -1.84068352e-01 2.91374058e-01
1.45606709e+00 2.90462822e-01 -4.65852529e-01 3.79508197e-01
9.52372432e-01 -2.61546642e-01 -1.15393400e+00 -3.18903059e-01
1.01732276e-01 -5.68174899e-01 -1.48610219e-01 -6.84374690e-01
-7.38690972e-01 1.10431123e+00 2.40401462e-01 3.84811871e-02
1.39515817e+00 -2.04650432e-01 2.46056393e-01 6.98078990e-01
4.33422625e-01 -8.93792510e-01 5.68831265e-01 7.42594182e-01
6.43756509e-01 -1.13769984e+00 6.53236285e-02 -6.10275209e-01
-5.55781901e-01 1.17545652e+00 1.19686544e+00 1.97250277e-01
2.89947450e-01 1.86140686e-01 -6.43789172e-01 -2.98241705e-01
-8.57814252e-01 -1.95472136e-01 2.61934459e-01 6.56308472e-01
5.78044176e-01 -9.91755500e-02 -1.60214335e-01 6.37423337e-01
-7.76407644e-02 9.31496546e-02 1.27561241e-02 1.38314819e+00
-5.31663835e-01 -9.99011636e-01 -4.23740447e-01 5.21232367e-01
-2.26815585e-02 -3.20511013e-01 -1.21803969e-01 8.01693916e-01
2.21338391e-01 6.92891181e-01 -2.86780030e-01 -4.99123096e-01
2.25111857e-01 2.34409839e-01 5.63888371e-01 -7.89209485e-01
-6.95707917e-01 8.40831175e-02 -1.59894526e-01 -4.92522538e-01
-2.04525739e-01 -5.21857500e-01 -1.19889474e+00 -1.45921379e-01
-2.48206139e-01 -1.40131608e-01 4.04919714e-01 7.32669353e-01
1.28375679e-01 7.25971937e-01 6.36940181e-01 -8.63368213e-01
-5.55979848e-01 -7.90059984e-01 -1.67543918e-01 5.04185498e-01
6.86505511e-02 -7.32325554e-01 -1.62330732e-01 2.30867773e-01] | [7.768241882324219, 7.779454708099365] |
4d86a62f-60fc-4ac0-aa45-51f604933bcc | probabilistic-word-association-for-dialogue | null | null | https://uwe-repository.worktribe.com/output/1434940/probabilistic-word-association-for-dialogue-act-classification-with-recurrent-neural-networks | https://uwe-repository.worktribe.com/OutputFile/1434957 | Probabilistic Word Association for Dialogue Act Classification with Recurrent Neural Networks | The identification of Dialogue Act’s (DA) is an important aspect in determining the meaning of an utterance for many applications that require natural language understanding, and recent work using recurrent neural networks (RNN) has shown promising results when applied to the DA classification problem. This work presents a novel probabilistic method of utterance representation and describes a RNN sentence model for out-of-context DA Classification. The utterance representations are generated from keywords selected for their frequency association with certain DA’s. The proposed probabilistic representations are applied to the Switchboard DA corpus and performance is compared with pre-trained word embeddings using the same baseline RNN model. The results indicate that the probabilistic method achieves 75.48% overall accuracy and an improvement over the word embedding representations of 1.8%. This demonstrates the potential utility of using statistical utterance representations, that are able to capture word-DA relationships, for the purpose of DA classification. | ['Steve Battle', 'Nathan Duran'] | 2018-04-20 | null | null | null | engineering-applications-of-neural-networks | ['dialog-act-classification', 'dialogue-act-classification'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.43951946e-01 1.58128157e-01 -4.24960330e-02 -5.40433764e-01
-4.82709557e-01 -2.62605935e-01 1.00392461e+00 4.64706331e-01
-6.43799663e-01 5.51799715e-01 1.02701867e+00 -1.07303970e-01
-8.08612108e-02 -6.63205147e-01 2.32131302e-01 -7.21373379e-01
-6.21253662e-02 5.19686043e-01 -1.25138447e-01 -5.62301815e-01
4.33831155e-01 4.73959714e-01 -1.65085495e+00 3.20060641e-01
1.53246775e-01 8.57607543e-01 1.86684132e-01 8.68258119e-01
-7.04442382e-01 9.98093486e-01 -1.13704419e+00 1.35667771e-01
-3.94646555e-01 -1.25672504e-01 -9.92256701e-01 -1.28928542e-01
-4.14284021e-02 -2.94623464e-01 -3.56206030e-01 5.47594130e-01
4.92448539e-01 6.43673718e-01 1.08613944e+00 -6.87073290e-01
-6.32817626e-01 8.19672883e-01 3.23932059e-02 4.56899047e-01
7.66602695e-01 -3.96479398e-01 1.52215195e+00 -9.59763646e-01
3.97146493e-01 1.38439441e+00 3.04871440e-01 6.73956215e-01
-1.32468081e+00 -8.01056400e-02 -1.34346128e-01 1.73214003e-01
-1.18238544e+00 -2.49984682e-01 1.04466736e+00 -4.16611165e-01
1.75049615e+00 1.01257734e-01 1.88659042e-01 1.47071993e+00
2.81997144e-01 9.37113345e-01 7.67246902e-01 -9.24614847e-01
1.16474636e-01 2.62635440e-01 9.72871542e-01 2.41926536e-01
-3.72505307e-01 -2.14386806e-01 -7.15805948e-01 -5.27166843e-01
3.00321788e-01 -1.38081731e-02 -2.26246286e-02 2.40158420e-02
-9.40259635e-01 1.46441126e+00 -7.72721916e-02 9.51873839e-01
-7.19972789e-01 9.07696113e-02 9.15305793e-01 1.14082545e-01
6.60329401e-01 5.91283619e-01 -4.54392642e-01 -7.37206638e-01
-4.57504570e-01 5.30951977e-01 7.28687525e-01 4.67556179e-01
2.10996583e-01 3.98662865e-01 -5.16645551e-01 1.40741730e+00
3.70387465e-01 2.02943042e-01 1.13963223e+00 -4.61597055e-01
6.33552000e-02 7.35776842e-01 -9.08118561e-02 -1.00512302e+00
-5.70930958e-01 6.24459721e-02 -3.60185355e-01 -2.33470395e-01
1.35626301e-01 -2.30797708e-01 -4.78347450e-01 1.57094264e+00
6.37347549e-02 -3.65053624e-01 5.68582833e-01 4.32366461e-01
1.06709921e+00 1.33329320e+00 3.37319404e-01 -1.72468483e-01
1.56198967e+00 -5.96491277e-01 -1.21212161e+00 1.36218011e-01
7.49295294e-01 -7.09870696e-01 1.08848822e+00 3.02493483e-01
-6.71868086e-01 -5.83485425e-01 -1.09723938e+00 -1.77871451e-01
-6.86867237e-01 2.40698874e-01 4.05321509e-01 8.75890136e-01
-8.70266378e-01 2.19603211e-01 -4.06952351e-01 -4.74354237e-01
-2.05428913e-01 2.85173297e-01 -3.82792018e-02 2.54784077e-01
-1.50230014e+00 1.34309506e+00 5.64438045e-01 8.05656016e-02
-6.10740125e-01 -2.41125360e-01 -1.26869690e+00 2.44028211e-01
-6.31481037e-02 -3.95384766e-02 1.33589637e+00 -4.58066344e-01
-2.01515460e+00 4.96145427e-01 -4.39613700e-01 -7.50473082e-01
-5.53160965e-01 -4.12597835e-01 -6.19045198e-01 3.78793143e-02
-3.03360522e-01 4.62401092e-01 6.95846140e-01 -6.09706104e-01
-4.79217440e-01 -1.93974063e-01 1.01625472e-01 1.60231590e-01
-5.96247315e-01 2.95691162e-01 4.56540167e-01 -6.90095007e-01
-2.56387413e-01 -6.98217273e-01 1.26293510e-01 -9.37586844e-01
-6.05211928e-02 -1.17213118e+00 9.20071065e-01 -6.90956056e-01
1.46531355e+00 -1.93909025e+00 2.61541635e-01 9.47275311e-02
-6.45527840e-02 4.20580924e-01 -4.34411392e-02 9.48041081e-01
6.45430293e-03 -2.75576532e-01 -1.26834393e-01 -3.82831126e-01
1.98591262e-01 5.06019294e-01 -5.74235439e-01 1.39848351e-01
3.85494620e-01 5.03598332e-01 -7.39909410e-01 8.84774476e-02
5.28640926e-01 6.88149631e-01 -1.72781512e-01 4.78974700e-01
-2.18516171e-01 -1.25501618e-01 -3.34697634e-01 1.76966995e-01
-3.93521637e-02 4.70833868e-01 6.45555556e-01 -1.79295883e-01
-7.34045506e-02 8.69008958e-01 -5.73879838e-01 1.48269606e+00
-7.54541278e-01 1.05740750e+00 -5.58083892e-01 -8.86678338e-01
1.38186729e+00 7.74493814e-01 2.07651019e-01 -3.12284678e-01
4.47048753e-01 -1.41676031e-02 1.59450799e-01 -3.93026739e-01
1.11676490e+00 -2.21424580e-01 -4.02630121e-01 6.30639493e-01
4.93114412e-01 -1.54644728e-01 -6.65040268e-03 8.55594873e-02
1.03326404e+00 -3.24311815e-02 7.25134015e-01 -2.44322270e-01
9.51160014e-01 -1.84980571e-01 6.55847937e-02 3.17402184e-01
-1.84995830e-01 2.24678159e-01 6.69725299e-01 -5.69423974e-01
-7.76181519e-01 -7.90653467e-01 -2.35484004e-01 1.39320266e+00
-5.04772902e-01 -5.87797701e-01 -5.11636734e-01 -6.03428423e-01
-1.34086162e-01 1.48402369e+00 -6.49121404e-01 -8.42301026e-02
-5.36588788e-01 -5.21068931e-01 7.59398401e-01 5.53916097e-01
-8.35488662e-02 -1.39701426e+00 -4.46152598e-01 5.77014089e-01
3.19819199e-03 -9.41408873e-01 -1.72697380e-01 1.95249945e-01
-5.18755376e-01 -6.59398198e-01 -6.05212510e-01 -7.49047041e-01
2.73985207e-01 9.84542966e-02 9.17977273e-01 -2.96407819e-01
-5.87579794e-02 7.32609570e-01 -7.02971697e-01 -3.78677666e-01
-8.59164655e-01 8.38008523e-02 4.27091032e-01 -1.07836850e-01
1.13094306e+00 -4.99300987e-01 1.12615474e-01 -8.24651495e-02
-8.85524273e-01 -6.73871040e-01 2.01538369e-01 1.16581607e+00
7.95969553e-03 -2.07856029e-01 9.32460964e-01 -9.13431704e-01
1.44130468e+00 -5.04837990e-01 -1.00573249e-01 9.30217430e-02
-5.08064091e-01 2.96152353e-01 3.71576011e-01 -4.10087615e-01
-1.20984626e+00 -1.00759961e-01 -4.60564107e-01 -2.70885956e-02
-4.43423063e-01 5.73008299e-01 1.71128407e-01 5.38296580e-01
6.51452780e-01 2.95914441e-01 3.46837848e-01 -4.12849784e-01
5.31857133e-01 1.06399930e+00 -8.59498084e-02 -5.07022023e-01
-1.31472899e-03 -3.76836509e-01 -4.54498529e-01 -1.24118114e+00
-5.24200320e-01 -8.05830657e-01 -5.66447496e-01 -1.42471492e-01
1.11432445e+00 -4.96234864e-01 -5.57813466e-01 1.27359286e-01
-1.41080821e+00 1.53849646e-01 -1.78940855e-02 6.43124700e-01
-5.75726330e-01 4.02851492e-01 -7.59432971e-01 -1.16821146e+00
-6.15659893e-01 -1.03019357e+00 7.04890966e-01 7.95441419e-02
-9.65632439e-01 -1.14006186e+00 4.80539531e-01 3.26292783e-01
5.54627776e-01 -1.24822296e-01 1.25646234e+00 -1.32187629e+00
4.75621909e-01 -3.29743296e-01 2.18852088e-01 7.00259149e-01
4.49473500e-01 -6.56620041e-02 -1.34368968e+00 1.83332860e-01
1.59387305e-01 -3.40762764e-01 5.42104125e-01 3.98853540e-01
6.77999377e-01 -2.18744054e-01 9.20267552e-02 -6.95479035e-01
9.89826679e-01 5.98321497e-01 4.24197316e-01 -7.47363120e-02
3.23841542e-01 7.51776278e-01 5.69345951e-01 4.50237870e-01
2.58049905e-01 7.05937505e-01 8.34932253e-02 4.29670751e-01
1.14235170e-01 4.93251868e-02 8.04703414e-01 1.16798115e+00
2.70562887e-01 -2.90825635e-01 -8.73998523e-01 7.57606030e-01
-1.68585467e+00 -7.82775819e-01 1.23186141e-01 1.73753905e+00
7.87878931e-01 2.48926982e-01 3.34473580e-01 2.40247548e-01
5.09567201e-01 6.92964375e-01 1.30437493e-01 -1.54104328e+00
9.20987278e-02 5.21036386e-01 1.82928666e-02 5.35934806e-01
-1.03092074e+00 9.40604627e-01 6.23893595e+00 7.88372278e-01
-7.62516499e-01 5.50311022e-02 3.88437092e-01 2.77441561e-01
-6.56749755e-02 -6.46208286e-01 -8.88082981e-01 1.40796483e-01
1.63536263e+00 -1.77894846e-01 6.47732019e-02 9.68813837e-01
3.94024909e-01 -1.38761297e-01 -1.15024137e+00 7.47596025e-01
4.79952395e-01 -1.33079350e+00 2.46799111e-01 -1.78476915e-01
2.65173763e-01 -2.78135359e-01 -8.89559910e-02 7.22024679e-01
2.05572307e-01 -1.06193340e+00 3.73423025e-02 4.29983139e-01
1.39294863e-01 -1.18079960e+00 1.32958102e+00 2.64037967e-01
-8.33519697e-01 9.25646797e-02 -6.22435689e-01 -3.01936716e-01
1.80252865e-01 2.23170102e-01 -1.70987511e+00 2.89218873e-01
1.30922019e-01 6.53704762e-01 -1.46134287e-01 8.43253955e-02
-1.82557702e-01 8.69015098e-01 -8.33868086e-02 -8.79386306e-01
5.91204941e-01 -2.97706246e-01 6.70848966e-01 1.52693665e+00
6.53293282e-02 4.47860509e-02 -1.78826302e-01 6.59689605e-01
5.44268154e-02 5.70268929e-01 -9.33571875e-01 -4.60942298e-01
4.55271095e-01 9.52260256e-01 -2.18333170e-01 -2.97776788e-01
-3.04318696e-01 7.13082552e-01 2.56735444e-01 -1.47837102e-01
-4.39238220e-01 -4.30653930e-01 1.00837338e+00 -5.91785550e-01
3.93986046e-01 -4.60381746e-01 1.05637997e-01 -4.75837171e-01
-1.78820178e-01 -7.02214837e-01 2.82792628e-01 -5.55888891e-01
-1.46861064e+00 9.47118163e-01 1.39651507e-01 -9.49845195e-01
-9.49892938e-01 -8.38801205e-01 -6.69616938e-01 9.76759732e-01
-9.58927155e-01 -1.02930343e+00 4.99803662e-01 -2.79653654e-03
1.14872682e+00 -6.59276545e-01 1.78776097e+00 -3.76703404e-02
-3.95994157e-01 4.04863149e-01 1.52846277e-01 1.94504425e-01
3.71621042e-01 -1.37706792e+00 8.07398483e-02 4.20072615e-01
4.61768001e-01 7.71889687e-01 8.37539494e-01 -3.29275459e-01
-1.03919005e+00 -6.96841955e-01 1.28926373e+00 -5.57516158e-01
7.44003236e-01 -3.67593706e-01 -9.13631022e-01 3.65603238e-01
5.87619603e-01 -7.23854184e-01 1.20481694e+00 4.76991534e-01
-2.29287699e-01 1.56658933e-01 -1.17531908e+00 4.72688943e-01
1.77781284e-01 -8.93632054e-01 -1.42198086e+00 1.47671938e-01
8.55625451e-01 3.91820148e-02 -9.41110492e-01 -6.23274073e-02
6.27715409e-01 -5.36619425e-01 9.63416696e-01 -8.18932295e-01
3.14753443e-01 5.26055247e-02 -4.58308458e-01 -1.48854518e+00
-3.57763395e-02 -2.85143673e-01 -2.14785248e-01 1.40245378e+00
5.26192486e-01 -4.74011809e-01 3.65025878e-01 5.44653177e-01
-8.79738480e-02 -6.09408915e-01 -1.17139733e+00 -3.92918140e-01
-2.73179356e-03 -6.27329886e-01 3.00248235e-01 8.69407177e-01
4.04763371e-01 8.84835541e-01 -5.61617613e-01 -3.91570963e-02
-2.40408078e-01 -4.00721401e-01 5.05235255e-01 -1.03168511e+00
-7.76187517e-03 -5.34317970e-01 -9.16113138e-01 -1.14855623e+00
5.87301016e-01 -6.11721218e-01 1.34122625e-01 -1.22013748e+00
-4.04032528e-01 7.01196417e-02 -3.28931332e-01 1.44094393e-01
-4.49383520e-02 -2.62004554e-01 4.09744941e-02 -1.92220286e-01
5.74412309e-02 9.53719199e-01 3.02618027e-01 -2.46451214e-01
-3.36095870e-01 8.88862610e-02 -6.89080954e-02 5.72207808e-01
9.84159172e-01 -3.66412282e-01 -4.91482794e-01 9.87198800e-02
-2.94322252e-01 -7.64277279e-02 -3.06113124e-01 -6.32930279e-01
-1.59265161e-01 1.31000996e-01 -3.01832948e-02 -9.35945392e-01
8.29401433e-01 -7.37388670e-01 -5.24158776e-01 2.80531853e-01
-8.81368876e-01 2.54554063e-01 2.19169021e-01 5.90475619e-01
-4.32860762e-01 -8.35746884e-01 2.59662539e-01 2.35426068e-01
-8.82970691e-01 -5.01712799e-01 -1.33641934e+00 -4.32832718e-01
7.90619314e-01 1.12877600e-02 6.70367405e-02 -5.89497387e-01
-7.75256455e-01 -1.79362535e-01 -4.51714337e-01 7.26517379e-01
8.32641542e-01 -1.24615002e+00 -5.52512348e-01 9.16570053e-03
1.56814247e-01 -4.76037443e-01 6.04470074e-02 3.25938523e-01
-3.28567624e-01 7.17899859e-01 -1.58228472e-01 -3.77859831e-01
-1.52077830e+00 1.57432377e-01 1.05775595e-01 -3.92112017e-01
-5.81212819e-01 8.08364153e-01 -2.03836799e-01 -5.86798191e-01
3.02361786e-01 -6.08427286e-01 -1.03178394e+00 2.83388734e-01
7.87048519e-01 2.98848063e-01 -4.38405834e-02 -9.33208048e-01
-2.69651383e-01 -2.11881269e-02 -3.36619526e-01 -2.47157410e-01
1.32766080e+00 -2.12272070e-02 -1.55941501e-01 1.18253720e+00
1.28827477e+00 1.30837947e-01 -2.81810284e-01 -5.75555980e-01
5.18120944e-01 -8.92240256e-02 1.88937917e-01 -5.17077565e-01
-2.72211015e-01 9.56967652e-01 7.81631708e-01 5.45374513e-01
5.21227181e-01 -1.03366643e-01 7.92917430e-01 8.11444640e-01
4.75875922e-02 -1.26152885e+00 -2.67597809e-02 1.07376182e+00
1.06657422e+00 -1.00669718e+00 -1.00225285e-01 -2.68242545e-02
-9.14895952e-01 1.79132974e+00 2.05568701e-01 -3.59112889e-01
6.40043378e-01 -2.88763583e-01 2.22849891e-01 -4.92286325e-01
-7.56110251e-01 5.56606334e-03 2.76893824e-01 4.33047771e-01
9.27903473e-01 3.01816136e-01 -7.22967207e-01 5.27464330e-01
-1.78646147e-01 -6.53844893e-01 7.40762413e-01 9.01002645e-01
-4.22155499e-01 -1.42225039e+00 -1.24763362e-01 5.07793427e-01
-3.41702074e-01 -9.87858102e-02 -5.53168952e-01 7.00248361e-01
-4.28656459e-01 1.07008207e+00 3.20627034e-01 -5.05942047e-01
2.00963482e-01 9.35687780e-01 5.06458133e-02 -1.05081260e+00
-8.19498241e-01 -1.54165074e-01 7.18202889e-01 -1.05287120e-01
-7.92595923e-01 -6.43428266e-01 -1.27203536e+00 -5.10145025e-03
-3.72461438e-01 3.83640915e-01 8.27469528e-01 1.16012216e+00
1.56643301e-01 6.37652576e-01 8.21014762e-01 -6.89590096e-01
-6.77678645e-01 -1.66492951e+00 -4.12623227e-01 1.20965414e-01
1.75582126e-01 -8.95550013e-01 -2.74805576e-01 -2.48966217e-01] | [12.791044235229492, 7.563652515411377] |
b8a60a71-4016-403c-be6a-e368a9ca4f6d | fpga-based-binocular-image-feature-extraction | 1905.04890 | null | https://arxiv.org/abs/1905.04890v2 | https://arxiv.org/pdf/1905.04890v2.pdf | FPGA-based Binocular Image Feature Extraction and Matching System | Image feature extraction and matching is a fundamental but computation intensive task in machine vision. This paper proposes a novel FPGA-based embedded system to accelerate feature extraction and matching. It implements SURF feature point detection and BRIEF feature descriptor construction and matching. For binocular stereo vision, feature matching includes both tracking matching and stereo matching, which simultaneously provide feature point correspondences and parallax information. Our system is evaluated on a ZYNQ XC7Z045 FPGA. The result demonstrates that it can process binocular video data at a high frame rate (640$\times$480 @ 162fps). Moreover, an extensive test proves our system has robustness for image compression, blurring and illumination. | ['Ziwei Zhao', 'Peng Gao', 'Fei Wang', 'Qi Ni'] | 2019-05-13 | null | null | null | null | ['stereo-matching'] | ['computer-vision'] | [ 3.21535617e-02 -8.53468597e-01 -6.89072385e-02 -1.82017267e-01
-1.37837492e-02 -3.11421573e-01 3.66356403e-01 8.94834194e-03
-6.54630840e-01 1.19699299e-01 -2.06210658e-01 -4.28491145e-01
1.14885978e-01 -6.91976190e-01 -4.56961930e-01 -2.50569493e-01
1.31240889e-01 -3.76201808e-01 6.26500785e-01 -7.48593500e-03
9.24797237e-01 9.02458966e-01 -1.84547353e+00 -8.87857005e-02
3.25726688e-01 1.31057930e+00 5.20628691e-01 8.91062260e-01
4.25866991e-01 4.60875601e-01 -5.23362219e-01 -1.44875571e-01
7.53667593e-01 2.52666444e-01 -3.40160221e-01 4.00418788e-01
8.49913359e-01 -7.94497848e-01 -9.07303214e-01 1.23101807e+00
4.81490314e-01 -1.59369260e-01 2.48074651e-01 -1.21761155e+00
-2.42403805e-01 -4.16662037e-01 -9.58381832e-01 4.08161968e-01
7.33655870e-01 2.29382649e-01 2.85011292e-01 -9.51013267e-01
4.10651237e-01 9.35073853e-01 7.55248189e-01 -8.39602277e-02
-4.99737263e-01 -8.15446794e-01 -8.15987349e-01 5.16295373e-01
-1.64270794e+00 -3.18674415e-01 3.45411092e-01 -2.99578995e-01
1.22878325e+00 2.83080101e-01 8.94800842e-01 -2.35859361e-02
1.01854014e+00 1.09939359e-01 8.39187264e-01 -5.03549039e-01
-2.16557235e-01 -1.91689804e-01 3.63940448e-01 8.62372100e-01
4.22530681e-01 5.14521182e-01 -6.98278308e-01 -8.86352733e-02
1.34162235e+00 6.60119355e-01 -5.81400156e-01 1.58211544e-01
-1.41547000e+00 5.89832485e-01 4.07396674e-01 -1.76983595e-01
-2.82389820e-01 2.23854110e-01 3.88672858e-01 6.33384049e-01
-4.69567716e-01 -1.72817305e-01 -1.48494288e-01 -1.55121312e-01
-6.79294527e-01 -4.57364693e-03 4.65936512e-01 1.73327243e+00
7.65823007e-01 -7.26494119e-02 1.55314162e-01 3.80658656e-01
4.75529909e-01 1.09531319e+00 7.39631116e-01 -9.05273616e-01
3.40454966e-01 4.59794343e-01 1.85020879e-01 -1.32233381e+00
-2.67976522e-01 2.48215273e-01 -6.89303100e-01 5.10219574e-01
-3.02949362e-02 9.74670798e-02 -5.65554261e-01 5.02361119e-01
1.77616313e-01 1.67846397e-01 4.12235521e-02 1.09407473e+00
1.16957557e+00 5.77348590e-01 -5.38184762e-01 3.15678343e-02
1.75554585e+00 -8.72750640e-01 -6.78181410e-01 -2.01481842e-02
-2.31705303e-03 -1.64898896e+00 7.09385097e-01 2.82077163e-01
-9.74266410e-01 -1.22465789e+00 -1.45082855e+00 -4.45304334e-01
1.57623157e-01 3.67527246e-01 7.83098459e-01 6.33426368e-01
-8.81747127e-01 3.39539498e-01 -7.73758769e-01 -2.39496678e-01
-9.30127874e-02 8.56973112e-01 -4.94688183e-01 -4.65922244e-02
-7.85178304e-01 6.68301582e-01 2.48055816e-01 5.34015372e-02
3.07091186e-03 -3.74876887e-01 -9.88632083e-01 1.27571914e-02
-1.93264619e-01 -7.47100055e-01 1.07562780e+00 -5.54592609e-01
-1.41081750e+00 8.10050368e-01 -4.11982089e-01 -4.19828504e-01
1.59082696e-01 -6.83824569e-02 -5.91200709e-01 4.45791602e-01
-9.09634158e-02 6.02158368e-01 1.04787123e+00 -1.27932042e-01
-1.07990897e+00 -3.57024044e-01 -3.59044731e-01 2.12776899e-01
-8.99616722e-03 3.43381017e-01 -6.33361459e-01 -2.41423428e-01
3.86530101e-01 -5.33692539e-01 4.96210456e-02 3.44945341e-01
5.84983975e-02 -7.94987828e-02 1.30017722e+00 -2.83926040e-01
1.01124561e+00 -2.21702695e+00 -9.25314248e-01 3.48206669e-01
-5.71557991e-02 4.60227489e-01 3.79090309e-01 4.19994205e-01
2.33169094e-01 -6.91235244e-01 7.20599294e-01 3.06409895e-01
-3.48493278e-01 -3.00889641e-01 -3.34955841e-01 9.57890153e-01
-1.04407094e-01 7.87368596e-01 -3.54215235e-01 -5.96322119e-01
8.31159055e-01 6.54524863e-01 -4.23185855e-01 -1.11753553e-01
8.43383491e-01 -1.09414317e-01 -3.59347850e-01 1.02733302e+00
1.10550427e+00 -1.22319674e-02 -3.02254170e-01 -7.09910750e-01
-6.54533267e-01 -2.68109679e-01 -1.45500278e+00 1.38764429e+00
-2.50586927e-01 1.28034914e+00 -1.39935911e-01 -2.62510300e-01
1.31165993e+00 4.25662436e-02 1.87123179e-01 -7.11379588e-01
7.00001538e-01 3.97713006e-01 -2.47017264e-01 -4.74059880e-01
9.79901731e-01 2.86210030e-01 2.29446799e-01 1.05551258e-01
-9.41799730e-02 -3.32666039e-01 -8.24831426e-02 -1.76972300e-01
1.07125223e+00 -1.22962043e-01 7.56722510e-01 -4.88091171e-01
8.51965785e-01 1.42723113e-01 3.92952949e-01 5.22428453e-01
-3.55166078e-01 4.98384118e-01 -4.37777400e-01 -7.89994597e-01
-1.15373409e+00 -1.11113429e+00 -5.32230377e-01 1.43458098e-01
1.05680263e+00 -5.60349882e-01 -3.21521223e-01 2.12203693e-02
1.93261757e-01 -4.89660263e-01 1.68420762e-01 -4.31512184e-02
-5.71888685e-01 1.27748951e-01 1.33420333e-01 4.98846591e-01
9.35088038e-01 -8.55210125e-01 -1.53724051e+00 1.07149437e-01
4.85790551e-01 -1.09182405e+00 -7.66836286e-01 -4.05395001e-01
-9.16377962e-01 -1.41813982e+00 -5.34980655e-01 -1.37005270e+00
8.34251761e-01 1.12599862e+00 7.29016066e-01 3.66297275e-01
-1.00862920e+00 3.74977618e-01 -1.17882706e-01 -3.73148531e-01
3.10580194e-01 -6.93196476e-01 7.89790004e-02 -4.22026604e-01
1.03692102e+00 -1.64419100e-01 -1.16469240e+00 7.23900616e-01
-5.33867061e-01 -6.89436542e-03 4.65411723e-01 8.40695500e-01
5.84221840e-01 -4.62269671e-02 -1.67461097e-01 -7.12917149e-02
3.54452312e-01 3.79382163e-01 -1.38032103e+00 -2.45565735e-02
-3.93212467e-01 -4.80248094e-01 4.37132746e-01 -1.32093430e-01
-5.31241775e-01 4.52023029e-01 2.46153641e-02 -3.92074764e-01
1.09107457e-01 -1.79473532e-03 1.87261179e-01 -8.94899309e-01
5.49229562e-01 3.21529716e-01 6.66981220e-01 -5.49225286e-02
-1.95753470e-01 1.31057394e+00 1.14076853e+00 1.61849156e-01
8.78797770e-01 7.41422892e-01 2.91744471e-01 -1.25360322e+00
2.36532122e-01 -9.70220327e-01 -4.04085100e-01 -2.78658658e-01
4.65844244e-01 -1.45985115e+00 -1.65941083e+00 7.27866709e-01
-1.09741020e+00 4.54909116e-01 1.30775616e-01 9.59298134e-01
-6.01102114e-01 6.61600173e-01 -6.91490650e-01 -9.09511149e-02
-8.04911137e-01 -1.33105934e+00 1.08177805e+00 8.35919619e-01
5.55680096e-02 -6.58159673e-01 -3.63935500e-01 -7.06776306e-02
3.78809571e-01 1.06426194e-01 -4.30607535e-02 2.35926270e-01
-9.70888674e-01 -6.91640258e-01 -7.02209949e-01 -4.46801297e-02
4.02433634e-01 3.05185318e-01 -6.11460328e-01 -4.73097116e-01
4.71306384e-01 1.12924673e-01 6.40302598e-01 3.74131083e-01
7.67665565e-01 1.39116302e-01 -4.11972940e-01 1.01392961e+00
1.89781356e+00 5.67071140e-01 9.01743591e-01 5.92786849e-01
3.71616006e-01 1.46730337e-04 1.15948117e+00 5.15698731e-01
2.82548100e-01 6.84638262e-01 1.27925217e-01 -2.35808447e-01
-4.47791480e-02 2.31850058e-01 2.17796013e-01 5.48265219e-01
7.35581815e-02 3.25664282e-01 -6.94132924e-01 -1.09354667e-01
-1.42226887e+00 -8.09600949e-01 -6.25000477e-01 2.18954396e+00
3.44662338e-01 -2.70981610e-01 -3.36391717e-01 4.07654017e-01
1.00619090e+00 -3.61598372e-01 -2.50917822e-01 -4.39249456e-01
1.13857528e-02 4.50674593e-01 8.98978770e-01 5.51797569e-01
-1.17539442e+00 6.96286380e-01 5.95549059e+00 5.23178756e-01
-1.55587554e+00 -4.60006326e-01 -4.05892096e-02 2.25180894e-01
3.50273699e-01 -7.79994950e-02 -1.05561161e+00 4.54417139e-01
3.49620581e-01 -4.08557326e-01 -1.84914414e-02 9.19500113e-01
-8.65477696e-02 -4.61133897e-01 -8.49637032e-01 1.78806031e+00
2.05892548e-01 -1.48489618e+00 -3.42004076e-02 9.67003554e-02
3.33850801e-01 -4.84559173e-03 1.31222205e-02 -3.38543981e-01
-5.81296086e-01 -6.10211253e-01 5.32235444e-01 2.67949551e-01
9.30605233e-01 -9.15874779e-01 9.40515757e-01 1.03737526e-01
-1.49616182e+00 1.87774077e-02 -9.40540254e-01 -2.52810359e-01
2.17699539e-02 1.71216235e-01 -6.87810063e-01 3.23928028e-01
7.18034148e-01 8.36909413e-01 -5.89971066e-01 1.49426425e+00
2.91336358e-01 -4.16871220e-01 -2.85516262e-01 -1.18872017e-01
5.00965538e-03 -2.71021068e-01 2.49794990e-01 1.09155011e+00
8.00593317e-01 4.01671201e-01 6.37800172e-02 1.26053020e-01
3.68536174e-01 1.14729710e-01 -7.09180534e-01 6.06076360e-01
5.46303093e-01 1.16580200e+00 -8.13850999e-01 -2.84332991e-01
-6.56820834e-01 9.51131046e-01 -4.60877448e-01 -2.45245576e-01
-8.16840231e-01 -1.28142500e+00 6.97414815e-01 7.78090535e-03
3.51511180e-01 -5.62461197e-01 -9.27596837e-02 -1.07888746e+00
4.15039696e-02 -6.68360949e-01 7.42700547e-02 -8.39937091e-01
-5.89194953e-01 4.71141249e-01 -4.31733251e-01 -1.92347443e+00
-9.00949687e-02 -8.77265275e-01 -4.95412827e-01 9.04693127e-01
-1.59846091e+00 -7.76337326e-01 -9.11816776e-01 1.13391900e+00
5.48359990e-01 -5.55228472e-01 5.53613663e-01 3.48908335e-01
-1.12611510e-01 5.87080836e-01 1.22648112e-01 1.53079420e-01
6.08172417e-01 -6.44505262e-01 5.93747377e-01 9.76843059e-01
-1.04585961e-01 8.79143417e-01 5.03451169e-01 -5.13015330e-01
-2.25637674e+00 -7.10223377e-01 8.63275886e-01 5.11560217e-02
2.48735830e-01 6.04177117e-02 -4.25610751e-01 3.06816459e-01
2.69578010e-01 5.02404690e-01 3.23602796e-01 -8.63139987e-01
-1.00587636e-01 -4.01980251e-01 -1.36771703e+00 4.58265632e-01
6.97713912e-01 -5.79928160e-01 -4.88030344e-01 4.85241227e-02
3.63890737e-01 -9.50545490e-01 -7.99530566e-01 4.23754752e-01
8.65992904e-01 -1.38497138e+00 1.06214952e+00 3.74487817e-01
-8.53371322e-02 -7.95375645e-01 -3.27483535e-01 -2.93037504e-01
-2.39756688e-01 -8.32661688e-01 3.26539755e-01 8.74412417e-01
-3.62178952e-01 -8.49208355e-01 7.53378987e-01 1.53551027e-01
-2.55264360e-02 -4.51791465e-01 -8.66253436e-01 -8.31219256e-01
-1.03549027e+00 -9.27510411e-02 5.57290912e-01 3.36945504e-01
1.68371186e-01 2.89478172e-02 -1.17550336e-01 3.68720829e-01
7.71591246e-01 4.92390692e-01 9.93726015e-01 -9.31761622e-01
3.29915658e-02 -1.88294604e-01 -1.39630949e+00 -1.30227888e+00
-2.80451536e-01 -3.19660395e-01 -4.09320444e-01 -1.00574219e+00
4.65742499e-02 -1.27699181e-01 1.49266809e-01 -4.42762934e-02
2.00906828e-01 7.85046995e-01 2.31689706e-01 2.74747968e-01
-2.59172559e-01 9.18574333e-02 1.08757079e+00 2.81764399e-02
-2.61662658e-02 4.23582084e-02 -1.04183294e-01 7.31204331e-01
7.45057583e-01 3.88017409e-02 -2.59024769e-01 -3.49427044e-01
-1.21900007e-01 2.56298482e-01 5.42789519e-01 -1.36159050e+00
7.10804701e-01 3.48577276e-02 7.58378863e-01 -9.85831916e-01
4.89292055e-01 -1.17837584e+00 2.98169404e-01 1.21686709e+00
5.06619573e-01 7.76416957e-01 2.09288463e-01 1.10950328e-01
-6.51082337e-01 -1.94926828e-01 6.61238432e-01 1.91309720e-01
-1.21264493e+00 3.77936095e-01 -2.59979039e-01 -5.91735840e-01
1.33864498e+00 -9.84424531e-01 -4.83097225e-01 -7.90969133e-02
7.86065608e-02 -1.52177466e-02 8.09413850e-01 3.23254317e-01
1.41639233e+00 -1.51208079e+00 -4.15604204e-01 1.02456129e+00
6.57222122e-02 -2.37815827e-01 1.69304997e-01 6.16792440e-01
-1.52423060e+00 7.01503277e-01 -6.87466085e-01 -1.01485813e+00
-2.09511089e+00 3.66595328e-01 7.83469900e-02 7.36784160e-01
-8.50118995e-01 6.59071803e-01 -2.97105581e-01 5.22397220e-01
1.90827802e-01 -3.27179015e-01 -6.66081980e-02 -2.72158831e-01
9.23764229e-01 6.30516410e-01 6.43278360e-02 -6.69282019e-01
-4.57545370e-01 1.30944204e+00 -5.11650890e-02 8.63430575e-02
6.51071250e-01 -3.33571762e-01 -1.28820926e-01 -3.64763916e-01
1.48902965e+00 1.31821856e-01 -1.13347554e+00 -5.72610423e-02
-2.92857856e-01 -1.31404126e+00 1.63031612e-02 1.11152694e-01
-8.87135923e-01 7.94394493e-01 9.46116924e-01 -1.04702123e-01
1.26718211e+00 -5.21632195e-01 8.96948040e-01 5.37537217e-01
8.49661291e-01 -9.39105690e-01 -1.84870899e-01 6.22592092e-01
6.23945117e-01 -1.30634463e+00 5.19000709e-01 -5.97729623e-01
-1.23980828e-01 1.72316360e+00 5.57302237e-01 -6.08834326e-01
8.55669439e-01 4.98337001e-01 3.18003297e-01 -8.76396820e-02
-4.17751282e-01 -1.10928826e-01 1.66547567e-01 6.38296247e-01
2.55024135e-01 -2.17530251e-01 -5.03334641e-01 -1.62607267e-01
-5.45704782e-01 5.87782152e-02 7.44226992e-01 1.25791872e+00
-8.14006567e-01 -9.23167050e-01 -6.67839289e-01 2.20215395e-01
-5.79923987e-01 -1.50692537e-01 2.94335455e-01 9.73640382e-01
-1.12872072e-01 9.25453663e-01 5.49764156e-01 -5.80262601e-01
3.79803777e-01 -8.62515211e-01 7.74775982e-01 -1.13463521e-01
-8.02435875e-01 2.14292020e-01 -4.77891237e-01 -7.97513664e-01
-3.11466187e-01 -4.82502997e-01 -1.34189916e+00 -6.10384524e-01
-3.59050453e-01 -3.14323232e-02 9.53068137e-01 2.41749719e-01
3.68638128e-01 7.69320801e-02 9.69843566e-01 -7.91846335e-01
-4.44163799e-01 -4.19531941e-01 -6.21174991e-01 8.41586962e-02
5.72201788e-01 -4.49371666e-01 -8.15782920e-02 2.14427575e-01] | [8.9647855758667, -2.143216371536255] |
e742e48d-72f3-4f1f-ba5e-c2f206a0bc6a | learning-embedding-adaptation-for-few-shot | 1812.03664 | null | https://arxiv.org/abs/1812.03664v6 | https://arxiv.org/pdf/1812.03664v6.pdf | Few-Shot Learning via Embedding Adaptation with Set-to-Set Functions | Learning with limited data is a key challenge for visual recognition. Many few-shot learning methods address this challenge by learning an instance embedding function from seen classes and apply the function to instances from unseen classes with limited labels. This style of transfer learning is task-agnostic: the embedding function is not learned optimally discriminative with respect to the unseen classes, where discerning among them leads to the target task. In this paper, we propose a novel approach to adapt the instance embeddings to the target classification task with a set-to-set function, yielding embeddings that are task-specific and are discriminative. We empirically investigated various instantiations of such set-to-set functions and observed the Transformer is most effective -- as it naturally satisfies key properties of our desired model. We denote this model as FEAT (few-shot embedding adaptation w/ Transformer) and validate it on both the standard few-shot classification benchmark and four extended few-shot learning settings with essential use cases, i.e., cross-domain, transductive, generalized few-shot learning, and low-shot learning. It archived consistent improvements over baseline models as well as previous methods and established the new state-of-the-art results on two benchmarks. | ['De-Chuan Zhan', 'Han-Jia Ye', 'Hexiang Hu', 'Fei Sha'] | 2018-12-10 | few-shot-learning-via-embedding-adaptation | http://openaccess.thecvf.com/content_CVPR_2020/html/Ye_Few-Shot_Learning_via_Embedding_Adaptation_With_Set-to-Set_Functions_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Ye_Few-Shot_Learning_via_Embedding_Adaptation_With_Set-to-Set_Functions_CVPR_2020_paper.pdf | cvpr-2020-6 | ['generalized-few-shot-learning'] | ['methodology'] | [ 3.31656694e-01 -8.08255151e-02 -4.73661810e-01 -5.45766771e-01
-9.96963978e-01 -4.05534774e-01 9.37627256e-01 -1.60141677e-01
-3.47455472e-01 6.16857827e-01 2.87292749e-01 9.57304761e-02
-2.83536404e-01 -7.32468188e-01 -7.29738474e-01 -8.88513982e-01
5.29892109e-02 6.42151117e-01 5.91036201e-01 -2.00843170e-01
1.04076073e-01 2.42384896e-01 -1.90542424e+00 3.58981103e-01
4.08643991e-01 1.05559194e+00 -4.20710519e-02 7.04663634e-01
-1.02510974e-01 8.58736694e-01 -4.72881675e-01 -1.82484552e-01
3.44199985e-01 -5.33180475e-01 -6.49306297e-01 2.39744738e-01
7.39157557e-01 -3.39152038e-01 -4.40679371e-01 7.61648118e-01
7.00419843e-01 5.16612470e-01 1.24426293e+00 -1.54678047e+00
-1.24213934e+00 1.46586914e-02 -3.30990881e-01 5.57922363e-01
5.19986488e-02 3.54791015e-01 1.12959063e+00 -1.37333262e+00
8.53987813e-01 1.09185231e+00 5.39253354e-01 1.03228509e+00
-1.46254122e+00 -4.22162235e-01 -1.65499337e-02 6.57313764e-01
-1.17929769e+00 -6.25126898e-01 6.30221188e-01 -7.02039957e-01
1.18807912e+00 6.96597248e-02 3.18324208e-01 1.58331966e+00
-1.36223659e-01 8.32010627e-01 1.12146854e+00 -6.43408000e-01
6.76726282e-01 3.59418124e-01 6.27857566e-01 4.97004896e-01
-3.82539758e-04 3.21302295e-01 -5.66888630e-01 -1.98574364e-01
2.32153162e-01 4.92182344e-01 -3.36223900e-01 -9.02007341e-01
-9.07730103e-01 1.07657504e+00 3.82211328e-01 4.20728713e-01
-2.18432187e-03 5.46209030e-02 5.06024957e-01 5.97018540e-01
7.03430414e-01 3.81512702e-01 -4.16883051e-01 -1.08542040e-01
-4.42750067e-01 -5.86835593e-02 8.20775568e-01 1.05515194e+00
1.18797398e+00 2.36984342e-01 -7.40937352e-01 1.18664002e+00
-1.73064083e-01 1.63695633e-01 8.49045157e-01 -5.39987028e-01
1.67350098e-01 3.11931789e-01 8.20711721e-03 -3.57758224e-01
7.99065363e-03 -1.58327311e-01 -4.67205763e-01 3.06121081e-01
1.95792705e-01 3.61156464e-02 -1.44434583e+00 1.65616238e+00
1.79419890e-01 8.44049394e-01 1.82540253e-01 8.42401683e-01
1.12142062e+00 7.96151280e-01 2.25806057e-01 -1.41894624e-01
1.18690419e+00 -1.21373522e+00 -4.97014463e-01 -5.49387515e-01
7.13689029e-01 -2.28507638e-01 1.38124275e+00 -1.32560089e-01
-4.53697741e-01 -6.40690565e-01 -1.24558401e+00 6.49860501e-02
-8.84899676e-01 -3.12559396e-01 3.22297126e-01 4.91162688e-01
-8.74016881e-01 6.46338463e-01 -4.94732499e-01 -8.00329030e-01
5.61640620e-01 1.87221710e-02 -4.97841895e-01 -5.58404803e-01
-1.12417603e+00 1.04882491e+00 4.70687389e-01 -6.19975984e-01
-1.29782879e+00 -1.03030252e+00 -1.13395452e+00 2.16454640e-01
3.94770414e-01 -5.12944877e-01 1.18106079e+00 -6.58625841e-01
-1.21039414e+00 1.19619322e+00 -1.98051129e-02 -3.58379811e-01
1.43934608e-01 -8.71858001e-02 -4.97257352e-01 4.39738967e-02
1.77084729e-01 5.96802592e-01 1.25891137e+00 -1.20659029e+00
-4.62099344e-01 -2.89806724e-01 -5.56264073e-02 8.57797042e-02
-9.86890256e-01 -8.64035711e-02 -1.49293482e-01 -4.06619668e-01
-5.91930270e-01 -5.82535505e-01 2.61211097e-01 3.52261245e-01
5.32926396e-02 -5.28061569e-01 1.21413350e+00 -6.89664930e-02
9.27998006e-01 -2.44136739e+00 1.13461599e-01 -3.30229372e-01
2.64082789e-01 5.92588365e-01 -5.12515604e-01 5.59706807e-01
-1.65674806e-01 -1.50795832e-01 -3.01614434e-01 -3.26264650e-01
1.99303582e-01 6.58625066e-01 -4.16399688e-01 5.40599227e-01
4.02550370e-01 1.12600076e+00 -1.18834662e+00 -4.68452632e-01
4.87770766e-01 3.26096475e-01 -8.08057189e-02 7.20537126e-01
4.72692819e-03 -2.61070728e-01 -1.12660907e-01 8.52310896e-01
2.27888286e-01 -3.94475400e-01 -2.72449553e-01 -3.25187109e-02
2.90705889e-01 -4.07811821e-01 -8.43653679e-01 1.56595707e+00
-3.77226442e-01 7.11773336e-01 -5.79752028e-01 -1.38881803e+00
1.08597553e+00 3.01598489e-01 2.05592364e-01 -5.76932132e-01
1.98374167e-02 9.18516219e-02 -1.48264065e-01 -7.11198092e-01
-8.27944651e-02 -5.30894458e-01 -3.35654095e-02 3.29957426e-01
9.57393050e-01 1.21968083e-01 9.69088748e-02 9.69834700e-02
1.32778680e+00 -2.52102520e-02 8.71528864e-01 -1.19869843e-01
-8.71015936e-02 -1.28109321e-01 5.38898468e-01 9.64867771e-01
-6.55277312e-01 9.14006889e-01 2.81304538e-01 -5.32260597e-01
-1.09810543e+00 -1.27440786e+00 -1.87321469e-01 1.72691000e+00
1.21490344e-01 -2.84127176e-01 -2.62609839e-01 -1.01012564e+00
1.67515740e-01 9.65123951e-01 -1.30583310e+00 -7.14087963e-01
-4.07861546e-02 -5.40651381e-01 1.39302835e-01 8.66138279e-01
-2.42642425e-02 -1.13329911e+00 -6.00643933e-01 5.40739149e-02
4.68920469e-01 -1.12206435e+00 -5.04889786e-01 5.76473594e-01
-4.76747423e-01 -1.26839972e+00 -9.35155213e-01 -9.65339303e-01
4.76339519e-01 4.50882494e-01 1.11246264e+00 -2.99887210e-01
-5.86771905e-01 7.07022667e-01 -6.78301811e-01 -4.17386800e-01
1.88354740e-03 -3.06303144e-01 3.70324291e-02 3.83075744e-01
8.49537909e-01 -4.32812542e-01 -3.96108627e-01 2.97569603e-01
-9.06216979e-01 -5.25604725e-01 4.77230728e-01 1.44388127e+00
4.02560085e-01 -5.89848101e-01 6.58892035e-01 -1.04341280e+00
5.52751720e-01 -7.87296355e-01 3.61160636e-02 5.24340630e-01
-5.67409098e-01 5.84789254e-02 7.39868641e-01 -8.52088809e-01
-8.64549339e-01 -3.96136492e-02 2.74069130e-01 -1.23624098e+00
-4.00684148e-01 2.00447999e-02 -1.57639280e-01 -1.32504895e-01
1.11446726e+00 3.24077308e-01 -1.10458069e-01 -4.70775902e-01
5.72712541e-01 7.02919245e-01 2.99499005e-01 -4.37901288e-01
8.65805030e-01 4.46339905e-01 -2.51726806e-01 -1.06900167e+00
-1.14354503e+00 -9.65979576e-01 -7.00834870e-01 -7.09720924e-02
8.03694546e-01 -6.69665158e-01 8.14932510e-02 1.88914701e-01
-1.02557373e+00 -3.04677486e-01 -9.09104288e-01 3.64292651e-01
-7.97459781e-01 1.45758484e-02 -3.62795383e-01 -5.47164321e-01
-2.10080400e-01 -8.87921393e-01 1.12494040e+00 1.32505938e-01
-1.36106730e-01 -1.24836028e+00 4.92759109e-01 -3.49086523e-01
3.83748144e-01 2.45721236e-01 1.03161335e+00 -1.24161804e+00
2.64595803e-02 -2.60563374e-01 -2.74942935e-01 4.23820972e-01
1.76845849e-01 -2.13604182e-01 -1.41993678e+00 -4.38059717e-01
-1.89876348e-01 -1.07226729e+00 1.20542896e+00 2.31055528e-01
9.90736842e-01 -1.01824559e-01 -3.78119558e-01 7.46228456e-01
1.65985942e+00 4.23110649e-02 5.88606238e-01 1.85118690e-01
6.07049644e-01 3.63274097e-01 5.96254230e-01 4.05611485e-01
-9.13894996e-02 7.86102235e-01 3.88651401e-01 2.09846318e-01
-3.63660067e-01 -3.55533928e-01 3.23143780e-01 4.71984386e-01
1.36476099e-01 -3.05831823e-02 -8.04035068e-01 8.82408023e-01
-2.07795501e+00 -1.11893523e+00 5.07620275e-01 2.02032995e+00
6.80984497e-01 6.12444356e-02 1.48073077e-01 2.13952623e-02
7.24700451e-01 5.44607460e-01 -7.04848528e-01 -4.69571918e-01
9.49097574e-02 4.47037637e-01 1.23208724e-01 1.59638360e-01
-1.23597813e+00 9.42305803e-01 6.11916924e+00 9.98180091e-01
-8.46600771e-01 5.42114556e-01 3.46598387e-01 -1.30040556e-01
-5.52686341e-02 -2.58826073e-02 -1.01637375e+00 3.33528847e-01
9.07544613e-01 -3.82259071e-01 1.63297117e-01 1.23971248e+00
-5.34650147e-01 5.23519933e-01 -1.66080022e+00 1.08016121e+00
6.44364476e-01 -1.43687654e+00 8.19609761e-02 -1.28133953e-01
7.82463670e-01 2.36961525e-02 6.97375368e-03 1.05715370e+00
2.06753731e-01 -9.20085549e-01 2.85420924e-01 4.52242523e-01
1.11447775e+00 -3.91300738e-01 5.34275353e-01 3.36268961e-01
-1.10172677e+00 -4.93560642e-01 -7.51749337e-01 -4.05481942e-02
-9.08916146e-02 9.71578807e-02 -8.29941511e-01 1.44151658e-01
6.05162680e-01 1.11922038e+00 -6.56967640e-01 1.14454484e+00
-1.38803512e-01 6.51093423e-01 1.96615219e-01 -1.02175593e-01
4.61075693e-01 1.57267660e-01 5.50630152e-01 1.19232988e+00
1.84884503e-01 6.88082054e-02 3.35285842e-01 7.42287755e-01
-1.32849172e-01 -8.72003883e-02 -1.13181496e+00 -5.86236082e-02
5.17846644e-01 1.24032605e+00 -3.93762171e-01 -6.73681319e-01
-6.68481588e-01 1.00419676e+00 8.28723609e-01 5.24473906e-01
-5.92907906e-01 -5.98651648e-01 7.16347337e-01 3.67298126e-02
6.83420002e-01 2.63162345e-01 2.28248507e-01 -1.27972972e+00
-1.85492203e-01 -6.24941528e-01 8.03055763e-01 -6.24335706e-01
-1.96447146e+00 4.48662132e-01 2.42123067e-01 -1.50674629e+00
-4.05395657e-01 -9.34208095e-01 -9.71539080e-01 6.40817702e-01
-1.56549680e+00 -1.22932553e+00 -3.77715290e-01 7.96668410e-01
9.59304571e-01 -5.74326038e-01 1.22372198e+00 9.93275568e-02
-3.54194611e-01 6.29478574e-01 1.93657041e-01 1.63763762e-01
1.01632059e+00 -1.40779841e+00 2.02657819e-01 5.90272188e-01
5.02493262e-01 1.47496402e-01 6.16856873e-01 -7.54231811e-02
-1.37975705e+00 -1.25419188e+00 5.83502889e-01 -6.01930141e-01
9.72230315e-01 -5.50956726e-01 -1.11281896e+00 8.31635177e-01
2.58000612e-01 8.85222793e-01 9.43924785e-01 1.32374257e-01
-8.45024824e-01 -2.48069495e-01 -1.02523601e+00 1.98698431e-01
1.19134235e+00 -7.14754760e-01 -1.12404668e+00 3.49932164e-01
7.78642416e-01 1.04850605e-01 -6.32382452e-01 2.80734718e-01
3.58471900e-01 -8.25913727e-01 1.01100278e+00 -1.36570811e+00
4.45772171e-01 1.17905900e-01 -4.98117417e-01 -1.83922803e+00
-6.41301572e-01 -1.83748528e-01 -8.18326414e-01 1.08142948e+00
1.55883491e-01 -5.90338588e-01 5.84461212e-01 4.10639793e-01
-2.66429543e-01 -7.70005703e-01 -9.13635850e-01 -1.33932257e+00
-2.17233375e-02 -3.44673507e-02 1.99550450e-01 1.23995245e+00
2.19238866e-02 6.86984062e-01 -6.57126904e-01 -1.71644509e-01
9.25781727e-01 2.45978102e-01 7.24949002e-01 -1.23096097e+00
-5.29480159e-01 -1.91179395e-01 -9.21364427e-01 -5.93149602e-01
4.19996917e-01 -9.93040562e-01 5.20392284e-02 -1.57937026e+00
5.13688207e-01 1.32662937e-01 -6.57387972e-01 8.25757444e-01
-2.62403697e-01 3.21663231e-01 1.98737919e-01 1.86859250e-01
-7.79194057e-01 8.01636159e-01 9.14523423e-01 -5.58783889e-01
-7.75202038e-03 -1.72688544e-01 -4.65626419e-01 4.18598413e-01
4.59868908e-01 -4.95800018e-01 -6.52616024e-01 -1.06422439e-01
-5.07862985e-01 -2.84506679e-01 4.87697661e-01 -9.68797147e-01
1.74894467e-01 -2.83621907e-01 2.99561381e-01 -8.37229043e-02
7.39366710e-01 -7.33209252e-01 -5.28176427e-01 3.33300382e-01
-4.68409181e-01 -6.75590634e-01 -1.01728365e-01 8.20061743e-01
-1.94678664e-01 -4.60116178e-01 1.12087667e+00 -1.98212154e-02
-1.72183466e+00 6.10365152e-01 2.26395711e-01 6.27116978e-01
1.57603419e+00 -3.97969782e-01 -6.31103456e-01 -2.36893758e-01
-9.52377260e-01 3.04291900e-02 2.45260403e-01 7.17322111e-01
9.53373015e-01 -1.67447364e+00 -7.37474144e-01 1.75610974e-01
1.24008763e+00 -6.92525744e-01 9.44114998e-02 5.44391334e-01
2.82512188e-01 1.50071278e-01 -4.94800478e-01 -7.68768907e-01
-1.08476317e+00 1.07006812e+00 3.14414263e-01 1.33596256e-01
-8.50274205e-01 8.27819109e-01 3.42520058e-01 -3.95254582e-01
2.73623884e-01 9.25205871e-02 -2.24277169e-01 3.07379097e-01
8.77881587e-01 4.17541444e-01 -6.21180683e-02 -4.13648933e-01
-2.95281649e-01 5.05946338e-01 -1.68623462e-01 2.85872310e-01
1.49836338e+00 2.50176936e-01 6.25642180e-01 1.06873417e+00
1.73907602e+00 -8.78142953e-01 -1.55088079e+00 -6.61053121e-01
2.09420267e-02 -8.17605555e-01 -1.34163454e-01 -5.44350386e-01
-7.08857596e-01 1.20735824e+00 6.72313988e-01 2.62362182e-01
6.25337780e-01 3.53447378e-01 4.24826294e-01 4.98721957e-01
2.48026595e-01 -1.18554986e+00 5.32044649e-01 4.20747042e-01
7.18088329e-01 -1.64217257e+00 -3.23326826e-01 4.11776751e-02
-8.15839469e-01 1.02452278e+00 8.92550051e-01 -3.72519732e-01
8.45402002e-01 4.13218886e-02 -2.58953303e-01 -2.86290735e-01
-1.15307677e+00 -6.32454753e-01 5.83800316e-01 1.08361506e+00
1.22250512e-01 -8.23969096e-02 1.03467651e-01 3.31249893e-01
6.19515538e-01 1.35313004e-01 2.06247911e-01 1.08476126e+00
-7.72487938e-01 -6.65826440e-01 1.18118925e-02 7.31483698e-01
3.37460697e-01 -1.33209433e-02 -4.75456655e-01 8.96057308e-01
-3.29827145e-02 7.40515709e-01 7.87573159e-02 -3.91862899e-01
5.06230593e-01 4.91771102e-01 5.26002407e-01 -1.24898231e+00
-8.11108425e-02 -3.32132608e-01 1.05113655e-01 -4.45193648e-01
-2.95820385e-01 -2.67143250e-01 -7.02070832e-01 1.51796788e-01
-1.16748750e-01 -7.81352147e-02 6.71674609e-02 1.07961512e+00
3.16297650e-01 3.75850260e-01 8.15185905e-01 -1.00712001e+00
-1.16261566e+00 -1.09232497e+00 -8.40163767e-01 8.86554778e-01
3.95638853e-01 -1.10562551e+00 -6.69196665e-01 7.25053549e-02] | [9.96606731414795, 2.775993824005127] |
31670e70-1677-4b18-85ce-48a6053f78d2 | demelange-deconvolution-et-debruitage | 2307.01761 | null | https://arxiv.org/abs/2307.01761v1 | https://arxiv.org/pdf/2307.01761v1.pdf | Démélange, déconvolution et débruitage conjoints d'un modèle convolutif parcimonieux avec dérive instrumentale, par pénalisation de rapports de normes ou quasi-normes lissées (PENDANTSS) | Denoising, detrending, deconvolution: usual restoration tasks, traditionally decoupled. Coupled formulations entail complex ill-posed inverse problems. We propose PENDANTSS for joint trend removal and blind deconvolution of sparse peak-like signals. It blends a parsimonious prior with the hypothesis that smooth trend and noise can somewhat be separated by low-pass filtering. We combine the generalized pseudo-norm ratio SOOT/SPOQ sparse penalties $\ell_p/\ell_q$ with the BEADS ternary assisted source separation algorithm. This results in a both convergent and efficient tool, with a novel Trust-Region block alternating variable metric forward-backward approach. It outperforms comparable methods, when applied to typically peaked analytical chemistry signals. Reproducible code is provided: https://github.com/paulzhengfr/PENDANTSS. | ['Laurent Duval', 'Emilie Chouzenoux', 'Paul Zheng'] | 2023-07-04 | null | null | null | null | ['denoising'] | ['computer-vision'] | [ 3.16710263e-01 -4.43466872e-01 5.35884500e-01 -1.04158558e-01
-1.01846254e+00 -5.51601887e-01 5.23439288e-01 -1.50154546e-01
-1.58295393e-01 1.05091786e+00 3.33727032e-01 -1.23883039e-01
-2.60100961e-01 -8.45387131e-02 -6.93706095e-01 -1.11963296e+00
2.32233247e-03 3.11780989e-01 -1.51916653e-01 -1.28051624e-01
3.30374599e-01 4.47335839e-01 -1.27747869e+00 1.54307872e-01
9.63105440e-01 7.76985228e-01 2.63879895e-01 7.05131412e-01
7.43589103e-02 9.10302877e-01 7.16002285e-02 -3.18139464e-01
2.80433327e-01 -5.64755976e-01 -2.55058050e-01 -1.64156899e-01
7.71088079e-02 -5.28576076e-02 -2.16710180e-01 1.23301196e+00
6.72607720e-01 3.96841526e-01 7.35296190e-01 -8.26308906e-01
-6.82461143e-01 3.00289661e-01 -8.48227978e-01 3.97217155e-01
4.85285819e-01 2.03762710e-01 6.69929564e-01 -1.55082977e+00
5.48875749e-01 1.04394591e+00 1.19153845e+00 -1.07818528e-03
-1.78377318e+00 -5.48553288e-01 -1.45691305e-01 -1.35980137e-02
-1.09386349e+00 -1.10051680e+00 7.81065643e-01 -8.00391197e-01
7.21280694e-01 5.27480543e-01 3.58924508e-01 1.04592240e+00
1.64409384e-01 5.43717146e-01 1.35678899e+00 -3.51879746e-02
2.29106426e-01 -2.65259117e-01 2.60230333e-01 2.93970555e-01
3.28160882e-01 1.75229698e-01 -8.48295510e-01 -6.14352286e-01
6.02857113e-01 1.58973217e-01 -6.17535353e-01 -3.32409181e-02
-1.29096234e+00 7.18135774e-01 -1.55611292e-01 1.25271931e-01
-6.99046671e-01 1.88942537e-01 1.50649399e-01 2.29854882e-01
9.21593308e-01 5.12257278e-01 -3.38819474e-01 -1.06153712e-01
-1.59012544e+00 5.06055236e-01 8.17082882e-01 8.72137368e-01
6.03609800e-01 3.97063047e-01 -1.85977116e-01 8.02594244e-01
6.53061032e-01 7.06904829e-01 2.19545700e-02 -1.61435628e+00
-1.86487913e-01 -2.58455724e-01 5.46873987e-01 -7.62087762e-01
-4.28262144e-01 -6.27696753e-01 -9.54338312e-01 1.42657712e-01
5.84118009e-01 -3.66539776e-01 -7.74031937e-01 1.24368393e+00
2.49349654e-01 5.59272051e-01 -5.02415597e-02 1.05872047e+00
8.15258741e-01 6.64097786e-01 -3.38833332e-01 -7.90944815e-01
1.33891904e+00 -7.17965961e-01 -1.18575478e+00 -2.98772931e-01
-1.92552730e-01 -1.33707058e+00 4.30308968e-01 7.51827419e-01
-1.60330796e+00 -2.53395349e-01 -9.42548692e-01 -3.95507812e-01
-8.89875516e-02 8.32805857e-02 4.23965394e-01 3.71234328e-01
-9.30121899e-01 1.12327945e+00 -9.70142603e-01 2.72346675e-01
3.23460311e-01 1.75746679e-01 -1.23121835e-01 -2.09372208e-01
-7.82932281e-01 6.35084927e-01 -3.82913172e-01 6.50791943e-01
-8.73862207e-01 -1.13781834e+00 -6.46292686e-01 -2.46119037e-01
1.57110184e-01 -7.98933923e-01 1.14691329e+00 -6.71460211e-01
-1.63243198e+00 6.18879437e-01 -7.86348760e-01 -3.04907054e-01
7.04438627e-01 -4.24421996e-01 -4.04431403e-01 3.01947892e-01
2.51130372e-01 -1.36251107e-01 1.37288547e+00 -1.27164817e+00
7.96461254e-02 -3.40073645e-01 -1.07732165e+00 -9.04264376e-02
3.91742855e-01 1.65392593e-01 -8.44055712e-02 -9.15717840e-01
6.67424738e-01 -6.27475739e-01 -5.61161697e-01 -1.07877769e-01
-3.11228722e-01 5.33780456e-01 3.16215098e-01 -1.26610351e+00
8.42503548e-01 -2.15893197e+00 2.42809474e-01 2.54628867e-01
4.18962926e-01 -8.69540349e-02 -4.43352796e-02 6.90531492e-01
-4.70941484e-01 -2.81039536e-01 -8.63541007e-01 -5.51674724e-01
-7.05804452e-02 -2.24191427e-01 -4.36229467e-01 9.89123166e-01
9.35943350e-02 5.44227481e-01 -1.06443763e+00 2.54843503e-01
8.59799385e-02 7.48086035e-01 -3.67498368e-01 1.65818721e-01
1.35604337e-01 8.25341642e-01 1.75384268e-01 7.21706688e-01
1.15975428e+00 -1.93969235e-01 1.58800883e-03 -4.26970363e-01
-7.43555725e-01 7.99891949e-02 -1.46343374e+00 1.75484133e+00
2.63931364e-01 6.03898823e-01 1.01068687e+00 -9.58644867e-01
9.46999967e-01 4.62502837e-01 7.31390119e-01 -3.16084921e-01
-9.02947690e-03 6.94545150e-01 -3.53924692e-01 -5.68171918e-01
5.42758048e-01 -6.31647468e-01 5.73814452e-01 -1.45650059e-01
1.28655985e-01 -3.59040231e-01 4.28039908e-01 1.21519148e-01
1.02529216e+00 6.26812279e-01 5.78812808e-02 -6.42549813e-01
4.50521082e-01 2.13373691e-01 5.36579907e-01 6.48349881e-01
-1.52387917e-01 1.05422211e+00 4.46845025e-01 4.81840447e-02
-9.81668890e-01 -9.41917121e-01 -3.74650985e-01 8.10464323e-01
-1.69881463e-01 -1.65940568e-01 -3.82169873e-01 4.45117742e-01
2.25363418e-01 6.82117343e-01 -3.49034250e-01 1.93464160e-01
-3.42809647e-01 -1.22355068e+00 3.76065105e-01 1.19076386e-01
-1.35950908e-01 -4.48195726e-01 -1.79013163e-01 6.72935545e-01
-2.11867586e-01 -8.70323062e-01 -5.01803041e-01 5.86045504e-01
-9.81781125e-01 -8.64604354e-01 -1.27846301e+00 -3.03548247e-01
3.57432485e-01 4.24278677e-01 8.99674177e-01 -4.48975414e-01
-3.80689949e-01 2.43234694e-01 -6.06707111e-02 -2.38947541e-01
-1.85281202e-01 -9.67349708e-01 1.50607631e-01 1.62669063e-01
2.73681670e-01 -8.25896680e-01 -8.65804076e-01 4.80996184e-02
-4.41716403e-01 -2.34130859e-01 3.04024875e-01 9.87486959e-01
6.74264312e-01 -4.42852527e-01 1.97183400e-01 -5.01316905e-01
8.44109714e-01 -6.47454798e-01 -5.86486995e-01 -3.36672485e-01
-5.59424520e-01 -4.26761657e-02 3.45057756e-01 -1.88981473e-01
-1.08806884e+00 3.17473337e-02 -2.52410442e-01 -7.67884851e-01
5.43361306e-02 7.38183081e-01 3.56649667e-01 -2.43762378e-02
7.82000959e-01 3.61039609e-01 4.53173161e-01 -8.56126785e-01
4.56579298e-01 3.39610606e-01 8.68549228e-01 -4.92505014e-01
5.69248438e-01 7.78722107e-01 4.74904478e-02 -1.29620504e+00
-2.01514885e-01 -8.35294485e-01 -3.36260438e-01 -2.32092187e-01
6.29594088e-01 -1.21083367e+00 -8.06447864e-01 5.81073821e-01
-1.14787173e+00 -1.87585548e-01 -4.08557922e-01 6.99268460e-01
-4.63462830e-01 7.82072961e-01 -7.00838268e-01 -1.06980669e+00
-3.90471280e-01 -1.04693007e+00 8.63497555e-01 8.07213970e-03
-3.20816368e-01 -7.50892818e-01 2.43360847e-01 2.87227511e-01
4.89063054e-01 4.53041762e-01 -1.82938855e-02 -4.40931052e-01
-1.88633785e-01 3.16040292e-02 -2.24357337e-01 3.94591153e-01
-1.41458539e-02 2.19040528e-01 -1.07775450e+00 -3.13616514e-01
6.19302213e-01 2.21925527e-01 1.20663214e+00 1.11482024e+00
4.10936177e-01 -3.08533132e-01 -6.49667978e-02 9.11302269e-01
1.36447644e+00 9.99584645e-02 6.25075638e-01 1.16006143e-01
5.67852974e-01 5.44772148e-01 1.62637591e-01 8.27755332e-01
-1.47103027e-01 1.72946230e-01 1.98754221e-01 -1.19113445e-01
-1.51958540e-01 3.68802935e-01 5.54786265e-01 8.47780764e-01
-5.52089155e-01 1.13204017e-01 -9.10297453e-01 5.87951362e-01
-1.86716688e+00 -1.19478214e+00 -1.30908096e+00 2.04357123e+00
7.65732288e-01 -2.74066061e-01 -5.87803870e-02 1.71141967e-01
6.59200072e-01 1.70067891e-01 -5.41829109e-01 -1.65647417e-02
-2.75162965e-01 2.47212291e-01 8.21054161e-01 8.90300870e-01
-8.94143164e-01 2.31140718e-01 6.15672731e+00 5.62622130e-01
-8.12129438e-01 4.62664187e-01 1.21314317e-01 -2.90491700e-01
-4.01012570e-01 -1.63707789e-02 -3.89029413e-01 6.45207763e-01
1.04384804e+00 -1.63590014e-01 6.79423809e-01 2.72745073e-01
7.95475543e-01 -1.27966225e-01 -7.47662902e-01 1.02247214e+00
-1.74405441e-01 -1.25078118e+00 -7.05025494e-01 -1.74023315e-01
5.39057791e-01 2.48668790e-01 1.93099864e-02 -4.03780788e-01
2.59393811e-01 -7.76189685e-01 1.06475544e+00 1.01907742e+00
5.85316598e-01 -2.56506920e-01 3.16128939e-01 7.72739649e-02
-1.03333783e+00 7.47520253e-02 -1.56122848e-01 -2.41240755e-01
4.48795468e-01 1.35891736e+00 -2.58364648e-01 8.30984414e-01
6.98944569e-01 1.16395378e+00 2.46793211e-01 1.12209439e+00
-9.22733545e-02 5.62295675e-01 -3.68206769e-01 5.46420336e-01
3.37631702e-02 -1.14288473e+00 1.11214399e+00 1.48214221e+00
7.22137570e-01 4.65186745e-01 -2.40464702e-01 1.14229202e+00
3.91030341e-01 -1.39564037e-01 -2.29477838e-01 -1.14154793e-01
1.81128770e-01 1.08822608e+00 -7.61918485e-01 -3.80636722e-01
-4.35618937e-01 9.37881172e-01 -4.47447300e-01 8.74144435e-01
-4.45855439e-01 -2.77744889e-01 8.62573385e-01 1.77220777e-01
5.10780811e-01 -3.89893025e-01 -4.33370560e-01 -1.26240087e+00
-1.98765635e-01 -9.69953179e-01 2.26996973e-01 -1.04995096e+00
-1.53863239e+00 1.77040786e-01 -2.60193020e-01 -1.18530023e+00
2.30810136e-01 -6.50268674e-01 -4.97230768e-01 1.29784679e+00
-1.82562292e+00 -6.19074643e-01 -7.88727403e-02 6.34399593e-01
3.47144753e-01 2.25840524e-01 5.48217297e-01 4.38131660e-01
-7.39772618e-01 -2.17022866e-01 1.04838324e+00 -3.96756142e-01
8.49362373e-01 -1.23402810e+00 1.77980185e-01 1.17727804e+00
-5.90164185e-01 1.07925522e+00 1.55005240e+00 -7.98891664e-01
-1.70553732e+00 -7.89572537e-01 5.93042850e-01 -3.69734280e-02
9.64032710e-01 -2.21486185e-02 -1.31623936e+00 5.48114359e-01
4.80851382e-01 7.39310235e-02 7.66576290e-01 -1.28272384e-01
-1.52574122e-01 1.67171583e-02 -1.13213408e+00 2.58599907e-01
6.90082431e-01 -4.01086122e-01 -6.34763360e-01 5.51924944e-01
1.53547168e-01 -4.83715177e-01 -1.12708533e+00 1.58441469e-01
5.13672531e-01 -1.09470761e+00 1.31980431e+00 -1.63404807e-01
5.03489196e-01 -5.67647159e-01 -1.34563386e-01 -1.08998275e+00
-6.37934089e-01 -1.35486650e+00 -3.31452549e-01 9.88739192e-01
1.97977737e-01 -8.38956594e-01 3.48892838e-01 6.03664219e-01
-6.93101883e-01 -1.80169240e-01 -1.05254769e+00 -8.20373356e-01
-1.77733064e-01 -3.54977578e-01 4.55863513e-02 1.13420761e+00
8.78931731e-02 7.31700584e-02 -5.65494537e-01 4.00796890e-01
1.19427073e+00 -3.20623219e-02 2.37814143e-01 -1.47689390e+00
-2.44465217e-01 -5.35497427e-01 2.76245445e-01 -8.88516605e-01
-3.60793114e-01 -7.18648493e-01 2.85638630e-01 -1.50825536e+00
-1.66161239e-01 1.70656219e-01 -2.43781000e-01 3.12809855e-01
-1.69944257e-01 2.97609538e-01 -3.98840494e-02 5.42709053e-01
1.54743092e-02 5.18505454e-01 1.05566430e+00 -8.93540084e-02
-3.03111345e-01 -1.51506394e-01 -6.28036737e-01 7.09863305e-01
5.99455893e-01 -7.07300067e-01 2.37340666e-02 -6.88445419e-02
3.67906868e-01 4.38754290e-01 4.99674231e-01 -7.52083480e-01
3.32037449e-01 -5.86006567e-02 2.74161816e-01 -5.07142186e-01
6.37241364e-01 -5.53295553e-01 8.85694861e-01 5.35254240e-01
6.65737092e-02 -3.47613722e-01 2.47387230e-01 6.04192317e-01
-1.67094514e-01 -3.62944901e-01 7.79765129e-01 -3.97442967e-01
-2.46196225e-01 -4.87870649e-02 -5.65103352e-01 -2.32704952e-01
3.84690046e-01 -3.56864303e-01 -2.34785020e-01 -1.93927795e-01
-1.06382871e+00 2.10161656e-02 3.78172785e-01 -3.12083125e-01
5.48673093e-01 -9.87935185e-01 -1.18349719e+00 1.74840689e-01
-5.08412540e-01 -2.03738645e-01 6.39678299e-01 1.81299984e+00
-5.22861481e-01 6.11417927e-02 6.25413433e-02 -4.80960250e-01
-1.01127994e+00 5.31946659e-01 4.36706632e-01 1.52037323e-01
-5.93893588e-01 1.07064390e+00 -2.67917216e-01 1.07664168e-01
-2.29378417e-01 -1.96782038e-01 9.62098241e-02 5.84095716e-01
7.25695610e-01 1.19560003e+00 1.40104651e-01 -6.26186907e-01
-4.17664886e-01 4.96852547e-01 3.36888194e-01 -3.41099113e-01
1.80225241e+00 -2.81860799e-01 -6.92536056e-01 3.56780231e-01
1.06770003e+00 3.16883832e-01 -1.59563613e+00 4.11442593e-02
-3.14130858e-02 -2.82789558e-01 3.88556987e-01 -6.23937011e-01
-7.25619912e-01 6.31045103e-01 5.15265286e-01 4.69078004e-01
9.93810892e-01 -5.34288287e-01 4.79332596e-01 -6.93892092e-02
-4.05212581e-01 -8.72485459e-01 -3.21613252e-01 5.53574264e-01
1.25022352e+00 -9.92536128e-01 5.61499953e-01 -5.18303931e-01
-3.65427375e-01 1.06476152e+00 -3.43923718e-01 -3.29077244e-01
5.85903347e-01 3.50496411e-01 2.64428724e-02 -5.24374962e-01
-3.44513297e-01 -2.56272316e-01 2.27193028e-01 4.91924405e-01
6.91064239e-01 -2.50249177e-01 -6.01340175e-01 7.24354684e-01
1.75096855e-01 2.31042847e-01 6.67910397e-01 7.83782780e-01
-3.09659868e-01 -8.74650121e-01 -1.05810249e+00 4.64512676e-01
-5.49228013e-01 -4.97579306e-01 9.35854465e-02 1.68887213e-01
-1.73025340e-01 1.20045030e+00 -3.53807926e-01 2.43173346e-01
3.87839466e-01 3.95449668e-01 2.30443403e-01 -1.76237032e-01
-5.50363004e-01 1.10108924e+00 1.68514684e-01 -5.01689315e-01
-7.95451701e-01 -1.34244490e+00 -1.04510748e+00 -2.69331455e-01
-1.59624428e-01 3.90010178e-01 6.44103348e-01 6.14877462e-01
4.44107860e-01 6.38013601e-01 3.45459968e-01 -1.45537126e+00
-3.24213207e-01 -9.35734510e-01 -8.86034250e-01 9.33772400e-02
8.61211479e-01 -5.15535057e-01 -8.54331493e-01 4.49519783e-01] | [11.63428783416748, -2.6514251232147217] |
7ec42276-b1c9-4aff-a5aa-a4f4de133e08 | distributed-consensus-algorithm-for-decision | 2306.05998 | null | https://arxiv.org/abs/2306.05998v1 | https://arxiv.org/pdf/2306.05998v1.pdf | Distributed Consensus Algorithm for Decision-Making in Multi-agent Multi-armed Bandit | We study a structured multi-agent multi-armed bandit (MAMAB) problem in a dynamic environment. A graph reflects the information-sharing structure among agents, and the arms' reward distributions are piecewise-stationary with several unknown change points. The agents face the identical piecewise-stationary MAB problem. The goal is to develop a decision-making policy for the agents that minimizes the regret, which is the expected total loss of not playing the optimal arm at each time step. Our proposed solution, Restarted Bayesian Online Change Point Detection in Cooperative Upper Confidence Bound Algorithm (RBO-Coop-UCB), involves an efficient multi-agent UCB algorithm as its core enhanced with a Bayesian change point detector. We also develop a simple restart decision cooperation that improves decision-making. Theoretically, we establish that the expected group regret of RBO-Coop-UCB is upper bounded by $\mathcal{O}(KNM\log T + K\sqrt{MT\log T})$, where K is the number of agents, M is the number of arms, and T is the number of time steps. Numerical experiments on synthetic and real-world datasets demonstrate that our proposed method outperforms the state-of-the-art algorithms. | ['Setareh Maghsudi', 'Xiaotong Cheng'] | 2023-06-09 | null | null | null | null | ['change-point-detection'] | ['time-series'] | [-3.00444156e-01 4.23293002e-02 -6.99556649e-01 -1.58187106e-01
-9.96223569e-01 -8.49984169e-01 1.41128972e-01 3.68381649e-01
-4.75648552e-01 1.04410887e+00 -2.18281433e-01 -6.11381710e-01
-7.89919019e-01 -6.98412120e-01 -8.71991158e-01 -9.96348619e-01
-5.29902279e-01 1.06109679e+00 1.55328259e-01 5.78056127e-02
8.63251016e-02 2.25617111e-01 -3.74465078e-01 -1.66735381e-01
8.50420058e-01 1.39826024e+00 -3.11209634e-02 7.08538651e-01
1.98618263e-01 1.03349888e+00 -4.91602540e-01 -7.74822414e-01
7.67053425e-01 -4.76723820e-01 -6.35656536e-01 2.72655934e-01
-5.24461269e-01 -5.79777300e-01 -2.54895806e-01 1.27347326e+00
3.32865715e-01 3.49428535e-01 5.98069727e-01 -1.41012108e+00
-3.25518548e-01 1.10740662e+00 -1.46470618e+00 4.96541321e-01
-1.27133265e-01 -7.32329264e-02 9.72648323e-01 2.33978912e-01
2.90545285e-01 1.51669478e+00 3.22836608e-01 3.71001244e-01
-1.15301669e+00 -5.88658750e-01 7.69885480e-01 1.84371322e-01
-9.18683589e-01 1.33165624e-02 6.10472500e-01 -1.98552400e-01
3.74847591e-01 4.83511329e-01 7.62686372e-01 3.55640084e-01
1.22664630e-01 1.17654634e+00 1.17641723e+00 -5.23231328e-01
6.32515728e-01 -1.49185151e-01 3.28844637e-01 6.76011741e-01
7.56851196e-01 1.96312338e-01 -1.48557991e-01 -4.88494456e-01
6.73642993e-01 2.99859196e-01 1.03301108e-02 -3.73046875e-01
-8.30547333e-01 9.74945247e-01 2.09122866e-01 -2.07691804e-01
-7.72455513e-01 4.60076720e-01 -7.89559931e-02 7.16361165e-01
6.09618068e-01 -1.76484197e-01 -4.40001965e-01 7.14609772e-02
-2.11664349e-01 2.86453366e-01 7.14561462e-01 1.14156997e+00
5.26192129e-01 -2.58965999e-01 -4.53784376e-01 6.53180540e-01
4.29844141e-01 7.38195479e-01 -5.38356508e-05 -1.24806237e+00
9.81807530e-01 2.62086838e-01 1.05674076e+00 -7.58176923e-01
-4.29142356e-01 -6.89232647e-01 -7.29702055e-01 2.52841860e-01
5.61160445e-01 -8.24618042e-01 -2.73600042e-01 1.70154488e+00
6.30037129e-01 -2.89543152e-01 -4.82047955e-03 8.11018646e-01
-2.13156447e-01 6.93608344e-01 -4.62472081e-01 -1.02360976e+00
1.00036311e+00 -1.24310839e+00 -7.38949180e-01 -3.03170234e-01
4.44844633e-01 -3.16179603e-01 6.37699291e-02 4.17673230e-01
-1.36703277e+00 4.50185061e-01 -9.44864810e-01 9.11140263e-01
1.73162982e-01 -5.24708211e-01 7.84183562e-01 9.25983429e-01
-4.90495384e-01 7.22211599e-02 -8.29729438e-01 3.14417541e-01
4.60734725e-01 4.17483568e-01 3.61872792e-01 -9.43035111e-02
-4.34672982e-01 5.15534401e-01 3.13022077e-01 2.19969556e-01
-1.19020092e+00 -4.29613560e-01 -2.39670500e-01 -1.43268883e-01
1.30273998e+00 -8.12865496e-01 1.65313983e+00 -1.00161123e+00
-1.72022200e+00 1.81769386e-01 5.69380969e-02 -4.93658483e-01
9.86079633e-01 3.06383997e-01 -9.59080644e-03 -1.21105269e-01
2.86983013e-01 -1.37944326e-01 6.88767552e-01 -1.27825618e+00
-1.33068681e+00 -7.34760404e-01 3.17361891e-01 4.03217554e-01
1.26851976e-01 -2.99027096e-02 -1.36782959e-01 -5.48299909e-01
1.34112369e-02 -1.09793150e+00 -7.40147173e-01 -6.31743729e-01
-3.41423690e-01 -2.77650118e-01 1.66029185e-01 -2.87775695e-01
9.67811108e-01 -1.79781365e+00 -1.43722147e-02 4.95196193e-01
-8.71795192e-02 -2.59186506e-01 -1.92687631e-01 3.96296024e-01
5.02651036e-01 3.65922041e-02 2.76706487e-01 -3.54478866e-01
3.13145846e-01 -2.15330962e-02 -1.72726959e-01 7.31832981e-01
-8.89837801e-01 6.56083822e-01 -7.76328027e-01 -5.90899959e-02
-1.53632477e-01 -5.97109318e-01 -3.21444362e-01 -4.51742262e-02
-4.66670245e-01 5.96490651e-02 -1.08451247e+00 6.52587295e-01
8.58195066e-01 -3.29987824e-01 3.75290573e-01 5.44446468e-01
-1.20933942e-01 -1.87993675e-01 -1.56434214e+00 1.09502161e+00
-1.64126202e-01 2.05360409e-02 6.10190809e-01 -1.17205620e+00
2.51113385e-01 7.99413994e-02 4.72160429e-01 -4.67867881e-01
4.40388262e-01 1.02454297e-01 -6.73478423e-03 -4.68412414e-02
1.09357342e-01 1.87466778e-02 -7.07515478e-02 9.41733181e-01
-4.67330337e-01 2.61345357e-01 3.58599752e-01 2.82200605e-01
1.11591876e+00 -5.67789376e-01 4.53841269e-01 -2.69480199e-02
1.37921423e-01 1.16561450e-01 7.97009289e-01 1.55705631e+00
-4.44067985e-01 -3.73941749e-01 6.78045154e-01 -3.85927767e-01
-5.88825405e-01 -8.31015587e-01 4.61207926e-01 1.44209123e+00
6.58681750e-01 4.46761727e-01 -5.18205166e-01 -6.34121180e-01
4.57301348e-01 7.39353240e-01 -8.29389930e-01 3.39480728e-01
-7.89998546e-02 -1.25657332e+00 -3.59517217e-01 4.80575860e-02
7.28232265e-01 -5.43737948e-01 -4.67100739e-01 6.84045911e-01
-1.13118127e-01 -7.04822838e-01 -8.52415860e-01 -1.34575531e-01
-6.71686947e-01 -1.00206769e+00 -6.55552506e-01 -4.05606240e-01
8.40842843e-01 7.39043474e-01 4.51835603e-01 -4.98257548e-01
1.24676749e-01 7.08881021e-01 -3.89216244e-01 -1.00357080e+00
-1.81596756e-01 -2.33233944e-01 -3.61113325e-02 4.03293103e-01
-1.01481035e-01 -1.65812582e-01 -1.06397557e+00 5.50195634e-01
-5.99561155e-01 6.98969662e-02 6.48529291e-01 5.91461182e-01
5.90134680e-01 2.57577658e-01 7.97941148e-01 -5.21436930e-01
9.35323954e-01 -4.27673876e-01 -1.39364123e+00 7.15886474e-01
-4.27594483e-01 -1.77676797e-01 2.65271008e-01 -6.14882410e-01
-1.11731231e+00 -2.32901007e-01 8.75712752e-01 6.69674501e-02
4.64725912e-01 5.21545112e-01 1.85578510e-01 -6.58058077e-02
8.93702209e-02 1.14161074e-01 1.56040043e-01 -2.05956802e-01
3.42988580e-01 7.84807622e-01 5.50025441e-02 -6.75936878e-01
4.04462606e-01 6.86475217e-01 2.15926915e-01 -1.40575066e-01
-9.52857733e-01 -2.49417216e-01 3.03382665e-01 -3.42739493e-01
4.94045578e-02 -7.99888074e-01 -1.73583531e+00 4.59102094e-01
-9.44960475e-01 -4.96027589e-01 -1.25682980e-01 6.80479765e-01
-8.96673858e-01 2.36035779e-01 -4.37577248e-01 -1.66066241e+00
-2.84538954e-01 -7.33573079e-01 3.81566696e-02 3.77281785e-01
6.20468736e-01 -7.76079476e-01 1.14743918e-01 8.16129148e-01
7.69612864e-02 1.50694534e-01 7.27527738e-01 -5.86591423e-01
-8.90902042e-01 -1.10251546e-01 5.97996898e-02 -1.71690390e-01
2.05391139e-01 -5.41145980e-01 3.05276904e-02 -8.53896797e-01
4.28750403e-02 -1.64494529e-01 4.30055290e-01 1.01186144e+00
8.36116254e-01 -1.12302303e+00 -6.47135079e-01 -2.31177025e-02
1.17844391e+00 9.96946692e-01 -2.12096959e-01 5.45331657e-01
-1.66985318e-01 1.82421803e-01 9.57999468e-01 1.10621977e+00
4.82218057e-01 5.30946732e-01 9.09499943e-01 3.77213806e-01
7.24094868e-01 1.23016402e-01 4.53398883e-01 2.06713572e-01
-1.41437635e-01 -5.95239222e-01 -4.71014678e-01 6.48874164e-01
-2.64517546e+00 -1.01708937e+00 7.01198727e-02 2.40122485e+00
6.82016015e-01 1.28161579e-01 6.75719738e-01 -2.14971557e-01
1.09407091e+00 -2.19689980e-01 -1.07906830e+00 -3.81094456e-01
-4.69023399e-02 -7.03147888e-01 1.05999887e+00 6.46947682e-01
-7.96764433e-01 4.49531287e-01 5.41996479e+00 1.04436827e+00
-3.21930349e-01 4.98118281e-01 1.04020023e+00 -7.36999333e-01
1.89592138e-01 -2.92800665e-01 -5.74516833e-01 6.69636130e-01
5.16907334e-01 -4.90872473e-01 1.15603209e+00 8.37824166e-01
5.26042283e-01 -5.00128984e-01 -9.32429910e-01 8.63798440e-01
-3.43806118e-01 -1.40761507e+00 -3.66609186e-01 2.17825115e-01
1.25923824e+00 4.93775941e-02 1.40589327e-01 1.00106612e-01
1.46316242e+00 -3.13981801e-01 9.57271874e-01 2.82502234e-01
1.31453797e-01 -1.26704931e+00 7.13123441e-01 7.53631711e-01
-9.02526796e-01 -8.36072385e-01 -2.62945652e-01 -8.98475498e-02
1.39536887e-01 3.41466904e-01 -6.44579113e-01 7.89520144e-01
5.35654724e-01 2.77413446e-02 2.79386908e-01 1.25546229e+00
1.72614679e-02 2.91424572e-01 -5.35009503e-01 -6.78851783e-01
6.48405075e-01 -7.12193191e-01 7.48966694e-01 5.54385543e-01
1.75918251e-01 7.64817178e-01 5.50703347e-01 2.26011425e-01
-1.17995158e-01 3.91579121e-02 6.47438839e-02 6.35030940e-02
7.03372061e-01 9.30481434e-01 -7.66014218e-01 -3.04562986e-01
-2.07160011e-01 5.23875654e-01 4.16947126e-01 4.28889692e-01
-8.91905665e-01 -1.02290429e-01 4.54526514e-01 -4.75622475e-01
5.62400460e-01 4.88184541e-02 -9.93763581e-02 -6.26007557e-01
3.43585648e-02 -4.24545854e-01 9.52919722e-01 -4.75717187e-01
-1.55759060e+00 -1.10182360e-01 -1.22952692e-01 -9.28704679e-01
-2.22631678e-01 -5.77283204e-02 -4.03966814e-01 4.07305628e-01
-1.25582516e+00 -8.23147297e-01 3.09666812e-01 7.13072956e-01
3.93376589e-01 -3.04087818e-01 2.59034723e-01 -1.80004880e-01
-8.60338926e-01 5.29876828e-01 1.12556684e+00 -1.68926250e-02
-2.48076338e-02 -9.47915494e-01 -5.68472564e-01 6.67820334e-01
-4.19888824e-01 7.08232224e-02 7.54174173e-01 -4.81977165e-01
-1.66615582e+00 -1.03714919e+00 -3.27760607e-01 2.31004119e-01
1.02387869e+00 9.42354798e-02 1.04299299e-01 1.01760769e+00
1.39306694e-01 -2.68232282e-02 6.40611947e-01 6.72854707e-02
-5.15842922e-02 -6.87706172e-01 -1.46166551e+00 6.78342402e-01
8.96258235e-01 3.60059351e-01 -3.30910474e-01 5.46199501e-01
6.09674335e-01 -3.40700239e-01 -5.49534440e-01 6.57235980e-02
5.19762993e-01 -6.23113811e-01 4.43513900e-01 -7.89523065e-01
-4.02840286e-01 -9.53740627e-02 -6.07663170e-02 -1.64247131e+00
-5.50557554e-01 -1.38666105e+00 -2.19942078e-01 8.57619762e-01
5.27508795e-01 -1.11686814e+00 8.64498973e-01 6.66597843e-01
2.75130033e-01 -4.35560793e-01 -1.65881681e+00 -1.05493784e+00
-3.42741609e-02 -3.13826501e-02 6.23023868e-01 6.95337176e-01
3.33261818e-01 5.18542044e-02 -5.52358031e-01 4.61659849e-01
1.11827064e+00 5.50624311e-01 7.26540565e-01 -8.00851583e-01
-7.41088688e-01 -5.61646283e-01 3.07119995e-01 -1.11689997e+00
-1.30765751e-01 -3.23552847e-01 -2.37910911e-01 -1.47715235e+00
8.27418685e-01 -6.75921977e-01 -5.44426382e-01 4.01914150e-01
1.09125964e-01 -4.53646183e-01 4.96545047e-01 1.28236398e-01
-1.41592276e+00 4.37400013e-01 1.28019035e+00 -6.53066754e-01
-5.08591115e-01 7.20395386e-01 -7.02179074e-01 6.59418762e-01
7.11847365e-01 -7.76355743e-01 -2.71898657e-01 -4.76976931e-01
2.96677113e-01 9.57843721e-01 -1.89638697e-02 -2.24400148e-01
4.05495912e-01 -1.01577711e+00 -1.22382134e-01 -1.02505112e+00
2.60982513e-01 -1.05368400e+00 3.23292673e-01 1.00257170e+00
-3.92313421e-01 -1.76527664e-01 -1.95656970e-01 1.22692943e+00
4.90400404e-01 -5.40069699e-01 5.77792048e-01 -2.56949782e-01
1.49451375e-01 4.72392201e-01 -5.81772268e-01 -2.05128208e-01
1.57236814e+00 1.48737460e-01 -7.02316582e-01 -8.33704054e-01
-6.23803139e-01 9.44695294e-01 -2.19055712e-01 -9.38319862e-02
1.22768611e-01 -1.09908247e+00 -7.73497760e-01 -4.82008010e-01
-3.95783126e-01 -6.26681447e-02 3.84595126e-01 8.26727450e-01
-9.40921996e-03 4.02108520e-01 2.12128744e-01 -2.85559952e-01
-1.22001851e+00 7.47667313e-01 5.43050408e-01 -6.14082575e-01
1.91120267e-01 5.62119067e-01 -1.43802956e-01 1.37999892e-01
4.03416872e-01 -1.24592870e-01 4.04859275e-01 1.91771522e-01
5.70744336e-01 1.03132963e+00 -3.10585797e-01 1.08931474e-01
-4.25371341e-02 8.73311237e-02 -5.25411069e-01 -6.08615518e-01
1.54409003e+00 -6.67553663e-01 -2.20284820e-01 2.09379196e-01
4.22040790e-01 -1.02415502e-01 -1.44545627e+00 -5.54897845e-01
-1.31524399e-01 -6.53218806e-01 1.07229270e-01 -1.12233591e+00
-1.07245457e+00 -1.41488716e-01 4.28968400e-01 8.94092858e-01
8.05071533e-01 -1.32765127e-02 1.73296094e-01 6.67546093e-01
1.12870741e+00 -1.25140190e+00 2.91146953e-02 3.98147404e-01
8.19079340e-01 -1.09142590e+00 1.72345757e-01 -1.90718457e-01
-4.97413903e-01 5.35640121e-01 2.59567857e-01 9.79238078e-02
5.33882082e-01 -1.31897002e-01 -4.02639329e-01 2.11762264e-01
-9.29301620e-01 -1.66612595e-01 -4.13935363e-01 3.22845548e-01
-7.17153192e-01 6.68637455e-01 -5.51205218e-01 9.09921646e-01
1.03142478e-01 -8.10379162e-02 7.22291410e-01 1.10581589e+00
-6.59544230e-01 -7.32645333e-01 -8.15356255e-01 6.40687764e-01
-8.13970864e-01 5.53709865e-01 -1.72112137e-01 5.33368230e-01
-4.75391179e-01 1.55367601e+00 1.44599214e-01 3.51007879e-01
1.19532183e-01 -5.88659346e-01 6.69745803e-01 9.56433117e-02
-2.53768861e-01 5.26692867e-01 -1.95559729e-02 -1.87430367e-01
-5.37426114e-01 -6.07125521e-01 -8.08670938e-01 -6.70015693e-01
-5.38209200e-01 5.41112006e-01 6.00566983e-01 8.63173068e-01
2.70847559e-01 2.48190388e-01 1.43042743e+00 -4.78534728e-01
-1.05387235e+00 -1.01568961e+00 -8.97951901e-01 -1.14997640e-01
5.18608630e-01 -6.66057110e-01 -3.17160547e-01 -5.86987376e-01] | [4.488866806030273, 3.242136240005493] |
33bc2a8a-67f3-4e20-ba77-a2cd208ccbc7 | time-warped-trials | 2302.00583 | null | https://arxiv.org/abs/2302.00583v1 | https://arxiv.org/pdf/2302.00583v1.pdf | Time-warped Trials | I outline a signal resampling strategy for aligning event times between time series trials in contexts where significant event times like onsets and offsets vary between trials. These variations prevent direct comparisons of trials in practical contexts as comparisons require equal-length time series (Salari et al., 2019). Algorithms like dynamic time warping help us quantify these variations locally but do not apply well to continuous transformations of time series signals without interpolating or downsampling to add or remove samples (Jamid, 2004; Eckner, 2014). I show that with consideration for padding and sampling frequency that sinc interpolation is sufficient to resample parts of trial intervals to produce equal-length time-locked trials that correlate to and strongly approximate their unwarped counterparts with minimal interpolation effects. Specifically I show that interpolation effects can be minimized by oversampling, selectively interpolating mis-aligned parts of trials with respect to mean mis-aligned event lengths, and interpolating mis-aligned events with sufficient zero-padding. Interpolated signals then have a bandlimit well below the Nyquist frequency and satisfy the Nyquist-Shannon sampling theorem ensuring perfect reconstructions, and I find that I can track and potentially counteract resampling effects on signal energy quantities. | ['Eric Easthope'] | 2023-02-01 | null | null | null | null | ['dynamic-time-warping'] | ['time-series'] | [ 6.12096846e-01 -3.67021561e-01 -3.12326014e-01 -7.60777295e-02
-8.96388471e-01 -9.45985496e-01 5.29810846e-01 2.95067519e-01
-8.57982516e-01 1.01597857e+00 3.11128914e-01 -2.65710950e-01
-3.43095154e-01 -3.12077999e-01 -5.75072646e-01 -4.03064013e-01
-6.30953610e-01 -4.02343929e-01 2.64743179e-01 2.43483856e-01
4.44790691e-01 4.38925505e-01 -1.50016618e+00 6.46741912e-02
7.17234552e-01 5.37625253e-01 -1.28445160e-02 9.46865976e-01
6.76326931e-01 3.22164744e-01 -6.65150285e-01 2.55306631e-01
3.08643728e-01 -9.51278508e-01 -2.98290431e-01 -5.28181612e-01
-5.41688933e-04 -3.26011211e-01 -3.64747077e-01 9.71928477e-01
6.22662842e-01 3.32904905e-01 4.81790602e-01 -1.17391157e+00
-2.91652113e-01 7.03724802e-01 -6.35645866e-01 5.51250696e-01
5.38375437e-01 -1.55213345e-02 5.06332159e-01 -5.03120720e-01
7.54779696e-01 6.86759412e-01 1.25304568e+00 9.64585990e-02
-1.85836887e+00 -7.70090878e-01 -2.90991873e-01 2.44611576e-01
-1.45363069e+00 -6.49489045e-01 6.76756144e-01 -1.30692050e-01
1.01555943e+00 6.96184993e-01 9.14416075e-01 1.18300283e+00
5.99625051e-01 1.62031911e-02 1.17666161e+00 -6.01295114e-01
3.47366005e-01 -5.30506372e-01 1.96906686e-01 -5.07344365e-01
3.01095843e-01 3.34537655e-01 -5.70193946e-01 -3.03246707e-01
1.06514359e+00 -2.25105464e-01 -4.86230016e-01 2.27918193e-01
-1.80614376e+00 3.79937381e-01 1.34883597e-01 5.77194929e-01
-8.12325418e-01 2.68039048e-01 6.31274343e-01 5.02713561e-01
1.86790586e-01 5.26647270e-01 -3.37135524e-01 -3.90755206e-01
-1.37678230e+00 5.77812850e-01 3.66904289e-01 6.65136874e-01
2.57742703e-01 2.95713931e-01 -2.15921462e-01 4.29560274e-01
-2.96600461e-01 5.44048309e-01 1.00427043e+00 -1.11050320e+00
5.89041002e-02 -3.06012064e-01 3.60217959e-01 -8.22576523e-01
-6.94597721e-01 -3.16733986e-01 -6.67562544e-01 6.29800558e-02
9.17701483e-01 -2.60435849e-01 -4.79158998e-01 2.04535413e+00
5.14903292e-02 5.36665022e-01 -3.17600280e-01 7.23303318e-01
-2.13276595e-01 6.47406578e-01 3.72857809e-01 -1.19320786e+00
1.36542213e+00 2.02177376e-01 -1.18805003e+00 -3.80653203e-01
2.54301280e-01 -9.78012621e-01 9.79832888e-01 3.28602761e-01
-1.42541146e+00 -4.41951245e-01 -1.34531999e+00 4.92178462e-02
1.83144733e-01 -4.52193499e-01 5.63567877e-01 5.22836030e-01
-7.62950420e-01 1.00028658e+00 -1.10471952e+00 6.90591102e-03
4.65390719e-02 1.26454115e-01 -1.96069598e-01 6.76188111e-01
-1.29119992e+00 1.03468478e+00 1.61044359e-01 -2.66500600e-02
-4.32288051e-01 -8.99873793e-01 -5.37575185e-01 -2.04742000e-01
-2.77250737e-01 -1.70657888e-01 1.29559958e+00 -1.02802026e+00
-1.01181424e+00 5.20101190e-01 -3.72807443e-01 -9.73879635e-01
1.95739657e-01 4.17124964e-02 -9.23105597e-01 1.14293106e-01
1.21732883e-01 2.80118972e-01 9.69716251e-01 -6.00897968e-01
-6.95882365e-02 -2.53052741e-01 -5.59422731e-01 4.48100865e-02
9.93917324e-03 2.99731549e-02 2.94918478e-01 -1.02334917e+00
4.49202448e-01 -6.94219828e-01 -1.12085775e-01 -1.34220079e-01
-6.26535565e-02 6.17731452e-01 3.67901862e-01 -7.71628320e-01
1.61019742e+00 -2.23337507e+00 -3.68464261e-01 2.30409577e-01
2.61165537e-02 -1.37164935e-01 -2.40817025e-01 4.93012160e-01
-6.16937280e-01 -1.20389827e-01 -3.41550946e-01 4.19337004e-01
-1.84082374e-01 -1.89980522e-01 -6.30665183e-01 1.08539712e+00
-1.23314887e-01 6.63972557e-01 -9.41591382e-01 -1.03663124e-01
3.60653669e-01 7.61929810e-01 -4.86085773e-01 -3.23666424e-01
3.25741798e-01 5.46332181e-01 1.99897483e-01 2.47601986e-01
6.25790119e-01 2.09423512e-01 2.96276420e-01 -4.34172928e-01
-6.39084876e-01 7.63347030e-01 -1.14978623e+00 1.48373652e+00
-3.06778878e-01 1.22021317e+00 -1.37189060e-01 -8.86215866e-01
6.71155334e-01 5.58312774e-01 5.40980935e-01 -1.03684795e+00
1.08575352e-01 4.13438946e-01 4.16023552e-01 -4.49422151e-02
5.40903568e-01 -4.58742887e-01 2.71855090e-02 4.82762694e-01
-2.56802619e-01 -2.87791312e-01 8.80725458e-02 -2.59281576e-01
8.61582577e-01 1.34201318e-01 8.58330250e-01 -4.25554276e-01
-1.48424789e-01 -1.59175858e-01 3.99528027e-01 6.27973378e-01
-4.20844942e-01 5.01562595e-01 2.81154931e-01 -1.53501928e-01
-1.35659361e+00 -1.17855585e+00 -9.00724649e-01 7.02886105e-01
-6.70792460e-02 -2.70681024e-01 -9.38899815e-01 3.99286985e-01
-2.56076753e-01 1.07727194e+00 -5.64422369e-01 -2.17196494e-01
-7.74033248e-01 -8.08259726e-01 7.53361344e-01 4.06331271e-01
7.03965202e-02 -1.05281103e+00 -1.21377134e+00 7.48165607e-01
-3.83000195e-01 -7.74487615e-01 -7.40335882e-01 5.78912139e-01
-1.20086884e+00 -8.51618648e-01 -6.00504220e-01 -4.57039505e-01
3.32715511e-01 2.13902086e-01 6.78539395e-01 -3.38263959e-01
-3.10776502e-01 5.11431061e-02 -8.69173184e-02 -3.48165095e-01
-1.28602386e-01 -5.70490658e-01 2.56464064e-01 -4.26490873e-01
2.16781884e-01 -1.13510609e+00 -7.66655445e-01 2.96094030e-01
-8.13548684e-01 -1.48422122e-01 3.67267872e-03 7.35425234e-01
7.57674873e-01 4.32886444e-02 9.41598713e-01 -2.27799594e-01
7.28736043e-01 -3.26999217e-01 -5.95107496e-01 -2.47782081e-01
-2.71559447e-01 -1.78205356e-01 6.67441130e-01 -1.05264997e+00
-3.39426935e-01 -4.35575008e-01 2.31594056e-01 -9.88460556e-02
1.25083596e-01 3.08169782e-01 3.65600854e-01 -2.27689240e-02
1.24112773e+00 1.83842838e-01 -1.04483016e-01 1.06507912e-01
2.90464908e-01 4.25125629e-01 9.08258617e-01 -3.66245598e-01
3.81138206e-01 5.70165396e-01 -9.82875451e-02 -8.16422045e-01
-3.40369269e-02 1.56092331e-01 -4.66754854e-01 -2.29295939e-01
7.15399683e-01 -7.71905959e-01 -7.04044998e-01 2.02179819e-01
-9.81793880e-01 -3.66040528e-01 -6.65047705e-01 1.08782661e+00
-9.53905523e-01 1.27318248e-01 -7.06292748e-01 -9.98917162e-01
4.80958202e-04 -6.17568016e-01 5.05604625e-01 2.44779393e-01
-1.23484313e+00 -8.02820206e-01 2.01676145e-01 -5.91671288e-01
5.10492980e-01 6.88725471e-01 5.08928418e-01 -4.97074455e-01
-4.84093092e-02 -3.35598588e-01 3.91227901e-01 -1.52415782e-01
2.54233927e-01 4.49164659e-02 -9.69283462e-01 -2.45267913e-01
5.54481387e-01 2.85298645e-01 1.83242649e-01 1.19704235e+00
8.94334018e-01 -4.27026749e-01 -4.13408846e-01 5.48022985e-01
1.15487099e+00 6.23168886e-01 9.04786229e-01 3.79161507e-01
8.28782767e-02 5.36588788e-01 3.98510635e-01 5.61405003e-01
-1.45284474e-01 4.86063600e-01 -1.51908010e-01 2.43725479e-01
7.19189420e-02 -2.95478791e-01 2.45956913e-01 8.75456572e-01
-9.29435343e-02 5.32556102e-02 -6.13873124e-01 8.27410817e-01
-1.30597484e+00 -1.54901075e+00 -5.17162800e-01 3.02901244e+00
1.19525301e+00 3.67036015e-01 1.05475649e-01 7.36577630e-01
9.53545690e-01 2.80635566e-01 -6.81200624e-01 -4.76082325e-01
-2.47629717e-01 5.94230115e-01 9.25655603e-01 6.71100199e-01
-7.19226837e-01 2.36386433e-01 7.60574389e+00 6.29907608e-01
-1.00276709e+00 -7.17152357e-02 2.05192015e-01 -5.27628064e-01
-5.66774905e-01 2.77586907e-01 -2.06159741e-01 7.30858684e-01
1.82578552e+00 -9.22381520e-01 8.76767814e-01 1.74799368e-01
7.55536079e-01 -3.29820931e-01 -1.10939395e+00 8.76598060e-01
-5.07025242e-01 -9.41987991e-01 -5.83333790e-01 -2.14495808e-01
4.77468133e-01 -2.45003998e-01 -8.92994925e-02 -4.34000641e-02
-2.07403928e-01 -1.07029939e+00 9.95323956e-01 4.98763204e-01
9.40338850e-01 -6.33912086e-01 -3.53666618e-02 2.57347137e-01
-1.33353221e+00 2.49850094e-01 -2.63307542e-01 -4.99997288e-01
4.83221889e-01 9.01164651e-01 -4.39929724e-01 2.25796521e-01
2.20260262e-01 5.09097517e-01 4.83923852e-02 1.00781715e+00
1.40325412e-01 7.55563855e-01 -6.65765285e-01 1.73973173e-01
-4.53355700e-01 -2.20395327e-01 6.07920766e-01 9.65277135e-01
6.70004666e-01 2.15500727e-01 -6.25384688e-01 7.24652052e-01
5.36575258e-01 -3.20508510e-01 -4.27028060e-01 -1.64691031e-01
1.46814573e+00 6.32923126e-01 -7.96922028e-01 -2.57125199e-01
-3.88965428e-01 7.48421967e-01 -4.59220469e-01 6.57492697e-01
-1.00213134e+00 -8.76057029e-01 5.34178615e-01 2.52929509e-01
-2.08917722e-01 -4.18290317e-01 -7.29943156e-01 -7.29317665e-01
1.88775897e-01 -8.99642408e-01 1.60966754e-01 -8.78628850e-01
-9.90082383e-01 3.73630822e-01 2.85203785e-01 -1.76988435e+00
-6.72398329e-01 9.72576514e-02 -5.36745131e-01 1.21863568e+00
-8.41999054e-01 -2.74630785e-01 1.91859663e-01 5.83879709e-01
2.13868052e-01 7.64944732e-01 8.45993400e-01 2.80033201e-01
9.34481546e-02 4.52499598e-01 -4.72215675e-02 -3.57108176e-01
9.37160134e-01 -8.58323216e-01 6.09502554e-01 8.68208349e-01
1.42909791e-02 1.08494055e+00 1.24303508e+00 -7.20753610e-01
-1.23489547e+00 -6.33639574e-01 1.06766355e+00 -2.04677597e-01
9.38481569e-01 -2.20985502e-01 -1.23586464e+00 9.34160352e-01
1.05818406e-01 9.42813903e-02 4.92338121e-01 -2.66004264e-01
-2.21528351e-01 1.01823285e-01 -1.08374667e+00 9.87876236e-01
8.30941617e-01 -9.19599295e-01 -8.94412637e-01 8.15464929e-02
4.89099354e-01 -5.89816034e-01 -1.11344981e+00 3.30964953e-01
1.00649643e+00 -9.01062131e-01 1.16143060e+00 -1.27936646e-01
-9.74127054e-02 -4.68492150e-01 5.94432233e-03 -1.11215627e+00
-2.25257844e-01 -1.06915092e+00 3.09532791e-01 9.50173080e-01
1.00662917e-01 -8.44494283e-01 9.78026986e-02 5.66738307e-01
-6.72251508e-02 8.42445493e-02 -1.21057034e+00 -9.12828028e-01
6.67920783e-02 -7.09726036e-01 2.29900450e-01 1.11034763e+00
6.77831173e-01 -2.96545446e-01 -1.47321597e-01 4.74940939e-03
6.97695434e-01 -2.44098857e-01 1.82393476e-01 -8.77915561e-01
-1.93131223e-01 -4.71703440e-01 -2.60914415e-01 -7.36143053e-01
-4.09315079e-01 -5.56626141e-01 7.47008761e-03 -8.24032724e-01
-2.79525965e-01 -9.88724157e-02 -3.72740597e-01 9.02988538e-02
-2.55926669e-01 3.47705871e-01 -1.19889408e-01 3.63722116e-01
3.65559548e-01 6.97570667e-02 9.71327543e-01 4.78998661e-01
-6.13080740e-01 -2.16233462e-01 -3.83652270e-01 5.65501094e-01
5.49275041e-01 -5.16062081e-01 -6.44018710e-01 1.48819089e-01
2.01185375e-01 7.13360131e-01 7.35729396e-01 -1.14226842e+00
8.26625973e-02 -1.24524213e-01 6.84543014e-01 -6.07222319e-01
8.83108601e-02 -7.20712245e-01 1.04743457e+00 4.29908186e-01
-6.74379647e-01 3.17113072e-01 8.61970127e-01 3.25504035e-01
1.07923197e-02 -2.59328783e-01 7.89788127e-01 3.51293862e-01
-1.54615641e-01 -3.48939896e-01 -8.08611393e-01 6.89498037e-02
1.00510633e+00 -5.41345000e-01 -3.16756934e-01 -2.33528793e-01
-6.88882947e-01 -2.67308652e-01 5.59730589e-01 2.88667046e-02
1.28812641e-01 -1.58918285e+00 -4.31126386e-01 3.72277439e-01
-4.09153968e-01 -6.98546886e-01 6.91396296e-01 1.18864894e+00
-2.81309634e-01 2.47804686e-01 -6.71153069e-01 -4.59595352e-01
-9.02413011e-01 5.43384790e-01 2.16156363e-01 1.78255573e-01
-6.85955286e-01 5.70087790e-01 -2.43459582e-01 5.16222239e-01
-1.25281215e-02 -7.25458503e-01 2.47724608e-01 4.69085068e-01
9.84535098e-01 3.32733274e-01 6.97133914e-02 -2.41723463e-01
-5.78542948e-01 6.79545224e-01 3.06605190e-01 -7.69489884e-01
9.19844806e-01 -2.35773891e-01 9.21939909e-02 9.82979119e-01
1.08709061e+00 2.80386806e-01 -1.41524756e+00 1.90160766e-01
-9.63794068e-02 -3.59111398e-01 -5.68041876e-02 -4.14638370e-01
-2.12285116e-01 5.20636618e-01 4.78183329e-01 5.48437655e-01
1.55480659e+00 -6.32848740e-01 6.09115660e-01 -3.60367179e-01
4.31182146e-01 -9.57434654e-01 -5.04036009e-01 9.31036100e-02
7.85731614e-01 -2.70387858e-01 3.48867267e-01 4.15574126e-02
-2.58684695e-01 1.03661036e+00 -2.36764058e-01 -6.76997721e-01
4.19652522e-01 5.13722956e-01 -3.11156183e-01 4.09155369e-01
-6.27808034e-01 7.95601755e-02 2.76039820e-02 8.15375090e-01
8.13070118e-01 1.70659497e-01 -1.05263269e+00 5.13177752e-01
-5.85365355e-01 2.00544119e-01 6.02522969e-01 9.18863237e-01
-4.55365986e-01 -8.11517239e-01 -9.45372999e-01 3.95599246e-01
-3.33122641e-01 -3.73472124e-01 2.43679121e-01 1.17962861e+00
-4.44985598e-01 8.50464642e-01 6.11390471e-01 -2.97667056e-01
2.75931835e-01 3.87907892e-01 6.98128164e-01 -3.08831837e-02
-6.23977602e-01 6.67581499e-01 1.20396435e-01 -6.16881907e-01
-4.33783501e-01 -9.14512217e-01 -1.44560337e+00 -4.22721475e-01
-3.93466592e-01 -8.84578601e-02 5.89878023e-01 7.71715939e-01
1.86583981e-01 8.27332139e-01 4.16393757e-01 -9.69040394e-01
-4.46046084e-01 -1.03324175e+00 -6.45166755e-01 1.42056137e-01
6.65536225e-01 -2.32755631e-01 -7.97763884e-01 4.41635162e-01] | [7.323195934295654, 3.267007827758789] |
eecf3a25-a90f-41b0-b227-c24ee6321f5e | cross-genre-argument-mining-can-language | 2306.04314 | null | https://arxiv.org/abs/2306.04314v1 | https://arxiv.org/pdf/2306.04314v1.pdf | Cross-Genre Argument Mining: Can Language Models Automatically Fill in Missing Discourse Markers? | Available corpora for Argument Mining differ along several axes, and one of the key differences is the presence (or absence) of discourse markers to signal argumentative content. Exploring effective ways to use discourse markers has received wide attention in various discourse parsing tasks, from which it is well-known that discourse markers are strong indicators of discourse relations. To improve the robustness of Argument Mining systems across different genres, we propose to automatically augment a given text with discourse markers such that all relations are explicitly signaled. Our analysis unveils that popular language models taken out-of-the-box fail on this task; however, when fine-tuned on a new heterogeneous dataset that we construct (including synthetic and real examples), they perform considerably better. We demonstrate the impact of our approach on an Argument Mining downstream task, evaluated on different corpora, showing that language models can be trained to automatically fill in discourse markers across different corpora, improving the performance of a downstream model in some, but not all, cases. Our proposed approach can further be employed as an assistive tool for better discourse understanding. | ['Steffen Eger', 'Jonas Belouadi', 'Henrique Lopes Cardoso', 'Gil Rocha'] | 2023-06-07 | null | null | null | null | ['discourse-parsing', 'argument-mining'] | ['natural-language-processing', 'natural-language-processing'] | [ 3.94797742e-01 7.16930091e-01 -4.05247331e-01 -2.73810148e-01
-8.01658809e-01 -8.51432920e-01 1.13732779e+00 8.00134361e-01
-5.24213433e-01 9.12949562e-01 6.52259052e-01 -6.18282080e-01
-2.28264660e-01 -9.63316500e-01 -5.41995347e-01 -2.85279632e-01
-5.15563786e-02 6.09419107e-01 6.63439929e-01 -6.33725643e-01
4.22484457e-01 7.34250471e-02 -1.50331783e+00 7.82400608e-01
9.79416549e-01 4.27792460e-01 4.45791408e-02 5.13305306e-01
-4.09611404e-01 1.17540991e+00 -9.95396852e-01 -5.60953975e-01
-2.10240990e-01 -3.89657438e-01 -1.27842820e+00 -6.70546964e-02
1.50036151e-02 1.97130024e-01 2.58178562e-01 7.70332515e-01
1.90721706e-01 -2.01051235e-01 6.32882595e-01 -7.28490472e-01
-2.57080317e-01 1.29792821e+00 -2.34678492e-01 3.32431734e-01
5.30280352e-01 -1.20646663e-01 1.29439592e+00 -4.59276021e-01
1.05526805e+00 1.37397254e+00 3.60423356e-01 2.81073868e-01
-1.09317064e+00 -1.91321149e-01 2.20079258e-01 1.20307148e-01
-4.78219718e-01 -5.33668756e-01 8.54589164e-01 -5.54719329e-01
9.19853985e-01 4.37727153e-01 4.66156781e-01 1.16760349e+00
-4.98129815e-01 8.29707026e-01 1.18878198e+00 -9.52073038e-01
7.84038827e-02 1.43699378e-01 4.36281145e-01 2.92967021e-01
1.73318684e-01 -3.34165543e-01 -3.36867243e-01 -3.46063137e-01
1.95398718e-01 -8.21788490e-01 -2.20966011e-01 -8.28676671e-02
-1.22586775e+00 1.10923588e+00 -3.18034063e-03 9.16586399e-01
-1.18656933e-01 -3.11108381e-01 9.04020250e-01 3.29675257e-01
6.78761363e-01 8.16178560e-01 -3.09555531e-01 -5.32186627e-01
-7.11319327e-01 5.17720640e-01 1.12615430e+00 3.54025126e-01
4.73751187e-01 -4.38913286e-01 -3.29603612e-01 1.08692920e+00
1.22350127e-01 7.09000370e-03 5.42918086e-01 -8.67084801e-01
8.60705793e-01 1.05897224e+00 1.48240075e-01 -9.37323570e-01
-4.45807815e-01 -2.21782178e-01 -2.03990236e-01 2.45982632e-02
9.73510385e-01 -2.60236651e-01 -1.77408651e-01 1.77780819e+00
4.81103897e-01 -2.35165328e-01 3.12096179e-01 6.65450692e-01
7.38956928e-01 4.34393138e-01 1.65648967e-01 -2.65098184e-01
1.55096757e+00 -7.21363306e-01 -7.69332349e-01 -2.76744157e-01
1.21479273e+00 -9.54767346e-01 1.03740096e+00 2.34815106e-01
-1.08631098e+00 -3.11558962e-01 -1.10162592e+00 -7.81592429e-02
-1.91108495e-01 1.73687205e-01 5.18960595e-01 6.61239147e-01
-4.70184177e-01 5.67778349e-01 -5.97257733e-01 -2.45359689e-01
3.51827890e-01 4.79346663e-02 -2.18571112e-01 3.10943335e-01
-1.30619371e+00 1.14216399e+00 6.29001021e-01 -1.85076624e-01
-6.28667921e-02 -3.10462683e-01 -8.18564773e-01 -1.85298786e-01
7.50510275e-01 -4.22146618e-01 1.26746976e+00 -9.60517526e-01
-1.38730693e+00 1.14229012e+00 -3.36925201e-02 -8.00429106e-01
8.44597995e-01 -3.97716790e-01 -2.83987731e-01 1.04713745e-01
2.47626245e-01 1.40420005e-01 4.62295562e-01 -1.06794322e+00
-6.15033090e-01 -2.45380387e-01 5.86944938e-01 3.29904258e-03
-4.27501500e-01 4.49848831e-01 6.03985451e-02 -5.90531707e-01
-2.07155481e-01 -7.83538043e-01 3.40992101e-02 -3.96307468e-01
-6.05131149e-01 -6.60675466e-01 8.25119257e-01 -5.78225613e-01
1.55200267e+00 -1.82529235e+00 1.62225783e-01 5.05288504e-02
1.20617524e-02 5.70600629e-01 1.76608920e-01 6.79326713e-01
-2.01870333e-02 3.15492600e-01 -3.73665094e-01 -4.10099104e-02
-7.49951974e-02 3.53601038e-01 -3.08346480e-01 2.35761732e-01
4.80451047e-01 8.08312774e-01 -8.52036297e-01 -5.90687037e-01
1.25329107e-01 1.95392504e-01 -3.86921227e-01 4.65887152e-02
-6.75202072e-01 4.67937559e-01 -6.07525051e-01 1.60768375e-01
-2.76093080e-04 -3.43878835e-01 5.88817835e-01 2.29877174e-01
-2.37571687e-01 1.12391257e+00 -1.22635376e+00 1.41630876e+00
-4.01960135e-01 7.44666159e-01 -3.10481079e-02 -1.35827065e+00
9.08924401e-01 3.89028609e-01 1.53768703e-01 -6.42908514e-01
2.25121349e-01 4.33250785e-01 7.13298142e-01 -7.46986926e-01
5.23772538e-01 1.06882200e-01 -1.91299543e-01 7.25407839e-01
-3.64089906e-01 2.97129512e-01 9.09530580e-01 1.23257875e-01
1.06702626e+00 9.85441506e-02 4.83025610e-01 -2.29451388e-01
9.41546977e-01 3.77570778e-01 2.58923382e-01 4.41064626e-01
2.12368220e-01 4.71638680e-01 1.03002381e+00 -2.77601540e-01
-1.05082583e+00 -2.92167693e-01 -2.42691487e-01 1.11233878e+00
-2.15020284e-01 -6.30132675e-01 -8.55564892e-01 -8.34425747e-01
-2.58148491e-01 7.66733408e-01 -5.90360582e-01 4.87084240e-01
-1.32840931e+00 -9.31543171e-01 8.22387457e-01 2.49754429e-01
2.66913205e-01 -1.37505579e+00 -1.07155824e+00 5.39520919e-01
-3.98737788e-01 -1.21516025e+00 4.53562915e-01 1.07439108e-01
-5.70841432e-01 -1.61885762e+00 -3.86691242e-01 -5.27818501e-01
2.48614892e-01 -1.24524094e-01 1.15535665e+00 6.12512827e-01
2.25461096e-01 -6.31578267e-02 -6.49800897e-01 -4.17852610e-01
-1.23829210e+00 4.34600085e-01 -5.64561486e-01 -2.97101617e-01
1.86896905e-01 -3.56256068e-01 -1.42693713e-01 1.04535542e-01
-8.37423801e-01 9.23543870e-02 7.05235302e-02 8.67915928e-01
5.86110391e-02 -4.63853031e-01 8.26949000e-01 -1.72486413e+00
1.11380565e+00 -3.89576674e-01 -4.05385852e-01 1.55767173e-01
-2.80102372e-01 3.46040070e-01 5.72946250e-01 -4.37358528e-01
-1.01313019e+00 -6.05521083e-01 -5.45060575e-01 3.98824275e-01
-1.26286671e-01 6.32957518e-01 -1.90762803e-01 4.85846102e-01
9.22628403e-01 -3.98537964e-01 1.99675318e-02 -5.20430326e-01
4.99246001e-01 5.02579093e-01 1.87865183e-01 -9.70313668e-01
5.69247186e-01 2.69014806e-01 -3.22483256e-02 -8.18970978e-01
-9.12785828e-01 -2.49137267e-01 -6.42231286e-01 7.56778792e-02
7.21644759e-01 -5.69091082e-01 -5.68232179e-01 -1.24036195e-02
-1.33086383e+00 -3.37004930e-01 -2.86190987e-01 1.98921651e-01
-3.89467984e-01 4.84580547e-01 -6.21928453e-01 -8.45256984e-01
-1.73661679e-01 -1.14087594e+00 8.20397496e-01 1.69417448e-02
-1.00217450e+00 -1.24066901e+00 2.45404318e-01 4.51488644e-01
1.41844243e-01 5.70524931e-01 1.47660625e+00 -1.27861011e+00
-1.16123043e-01 3.76663022e-02 -5.03191054e-02 1.36381596e-01
1.02117196e-01 1.36825010e-01 -1.05947292e+00 1.35818005e-01
-8.52701962e-02 -3.80667984e-01 7.12891340e-01 3.36072315e-03
6.43989801e-01 -4.69937474e-01 -3.23543519e-01 -2.30596006e-01
9.61770177e-01 2.86901742e-02 5.60128987e-01 1.02862632e+00
3.06121051e-01 1.12067914e+00 8.06183219e-01 9.45880488e-02
2.92060822e-01 7.68620670e-01 3.02742124e-01 -6.33002445e-02
-2.25763574e-01 -6.82837330e-04 2.71861047e-01 6.78430021e-01
-1.09781310e-01 -3.70520949e-01 -9.30729270e-01 7.21741974e-01
-2.01799512e+00 -1.07734549e+00 -6.11853957e-01 1.84497070e+00
1.04366565e+00 6.54613256e-01 4.96918947e-01 7.52161026e-01
6.28970444e-01 3.99246633e-01 5.05520888e-02 -5.68197548e-01
-4.10016656e-01 3.63449305e-01 -1.65997997e-01 6.85453653e-01
-1.16137147e+00 7.08499670e-01 5.37697363e+00 6.28481090e-01
-9.68194485e-01 1.10319227e-01 5.94468236e-01 3.10235113e-01
-4.65506434e-01 2.21485928e-01 -5.90424120e-01 4.56916183e-01
8.34191144e-01 -4.68672849e-02 -1.46261990e-01 7.13393271e-01
1.32948816e-01 -2.29553759e-01 -1.27048695e+00 2.78058678e-01
-1.55111998e-01 -1.73084843e+00 -2.34309047e-01 5.24096787e-02
3.68253469e-01 -2.84775466e-01 -3.73409331e-01 2.91749239e-01
2.39833936e-01 -9.27507579e-01 7.41672575e-01 -2.37750113e-01
2.40169629e-01 -4.70787197e-01 6.96275115e-01 7.06967294e-01
-7.22091138e-01 -3.34275374e-03 8.18376988e-02 -5.05743027e-01
2.70814240e-01 4.71370190e-01 -1.26336992e+00 6.62684023e-01
1.54368922e-01 4.04315203e-01 -7.18873441e-01 6.37366951e-01
-6.23753309e-01 8.22160184e-01 -3.51281375e-01 -4.25418168e-01
1.94151163e-01 -1.28002331e-01 7.55285442e-01 1.45453048e+00
-2.62093660e-03 -1.51272081e-02 2.07489982e-01 6.12154007e-01
5.57370558e-02 4.41538870e-01 -5.52145660e-01 -5.19761257e-02
5.38681448e-01 1.17853451e+00 -8.81495595e-01 -3.74799281e-01
-3.10187757e-01 2.71338582e-01 5.34215510e-01 -9.37800631e-02
-6.57139242e-01 -2.04878934e-02 3.54070723e-01 4.76567507e-01
1.81262940e-01 -3.87153239e-03 -3.19052815e-01 -7.76725173e-01
2.72680074e-01 -1.13636756e+00 5.93515694e-01 -3.30934852e-01
-1.04995382e+00 6.83938563e-01 2.20771343e-01 -9.75684047e-01
-8.56149673e-01 -6.02262199e-01 -7.01293826e-01 6.72913969e-01
-1.49465561e+00 -1.13109481e+00 1.29320681e-01 8.12083185e-02
6.26603663e-01 -1.13124736e-02 9.01839852e-01 2.63679892e-01
-2.76206970e-01 2.53585100e-01 -4.08926636e-01 4.76716816e-01
6.09481096e-01 -1.14400983e+00 4.03247684e-01 8.80199432e-01
5.84172547e-01 5.17814696e-01 1.02662408e+00 -4.88394588e-01
-9.01825786e-01 -7.33602405e-01 1.09849012e+00 -5.64218581e-01
9.22204912e-01 -2.80725986e-01 -1.39856708e+00 4.66116071e-01
5.29755533e-01 -4.32964593e-01 5.61692417e-01 5.94666243e-01
-2.51526505e-01 5.87483883e-01 -8.22572052e-01 5.56223810e-01
9.18584824e-01 -3.19468260e-01 -1.21187031e+00 3.16514790e-01
6.25557899e-01 -6.28547966e-01 -7.16280997e-01 4.37946439e-01
2.98739493e-01 -1.17166233e+00 6.91864014e-01 -6.08955562e-01
8.13569307e-01 -2.38028333e-01 8.82869288e-02 -1.04828274e+00
3.75893176e-01 -6.22193933e-01 -7.94983357e-02 1.70012569e+00
7.27972388e-01 -5.41325331e-01 6.13975346e-01 3.20075452e-01
-7.95641020e-02 -8.38777304e-01 -9.01252091e-01 -4.07809108e-01
2.62790859e-01 -5.50081491e-01 3.97493154e-01 8.66633177e-01
5.18961251e-01 9.34215367e-01 1.85762867e-01 -3.49563658e-01
-1.58593971e-02 2.99235314e-01 9.54304338e-01 -1.54654062e+00
-3.92702073e-01 -5.55442214e-01 5.83402961e-02 -8.49210143e-01
3.18866462e-01 -9.11448300e-01 -3.40759397e-01 -1.47163105e+00
-9.47524235e-02 -8.37733090e-01 3.88718069e-01 2.91360885e-01
-2.65194893e-01 -8.09962600e-02 3.51207763e-01 1.66743070e-01
-3.38144004e-01 1.91292822e-01 1.10329628e+00 -7.22901151e-02
-3.10769171e-01 4.30372283e-02 -6.94128513e-01 1.08701277e+00
6.42911851e-01 -5.63360691e-01 -3.05971891e-01 -3.07778776e-01
4.46648926e-01 7.71352947e-02 1.25526205e-01 -5.81958652e-01
-3.32213975e-02 4.40510735e-02 -1.06914327e-01 -3.82479221e-01
3.14084679e-01 -3.67103726e-01 -1.21483564e-01 3.62284660e-01
-5.91847479e-01 2.43459165e-01 1.62956059e-01 1.69976965e-01
-3.31492603e-01 -7.56140888e-01 3.80534261e-01 -1.87917024e-01
-4.17576522e-01 -6.24402046e-01 -2.92129785e-01 6.46871507e-01
7.90122926e-01 -2.00233266e-01 -7.66274095e-01 -1.14130534e-01
-4.12704945e-01 -6.05457462e-02 4.47301447e-01 4.18139249e-01
5.34692518e-02 -7.74467707e-01 -9.17946160e-01 -1.09635204e-01
-3.08895973e-03 1.50460258e-01 -4.42081273e-01 7.82990754e-01
-4.82285202e-01 3.62371504e-01 1.10229343e-01 -7.28238761e-01
-1.44901860e+00 2.43152440e-01 -8.32522810e-02 -9.18350875e-01
-7.89920449e-01 5.14608622e-01 -1.22861572e-01 -1.89161912e-01
2.05465164e-02 -4.46135849e-01 -7.33840704e-01 4.19545919e-01
4.71377939e-01 2.06831321e-01 2.60144651e-01 -6.23571813e-01
-2.04631597e-01 2.39938885e-01 -1.71034764e-02 -3.39769781e-01
1.53257537e+00 3.65468599e-02 -1.47400811e-01 5.99191904e-01
6.42312229e-01 6.90753400e-01 -7.89897144e-01 -8.01616684e-02
7.55926609e-01 -1.90645173e-01 -5.53466320e-01 -7.15134978e-01
-3.10294062e-01 7.32140124e-01 -4.63202931e-02 9.62338686e-01
6.40116572e-01 1.34760305e-01 5.18778086e-01 2.63440579e-01
1.61796555e-01 -9.89418209e-01 -1.75369736e-02 6.81394279e-01
7.98235476e-01 -1.34222305e+00 8.66748542e-02 -7.51456082e-01
-5.78239560e-01 1.28733194e+00 3.35527599e-01 -2.19603419e-01
2.05243573e-01 3.99495453e-01 8.41431841e-02 -3.74500811e-01
-7.74511516e-01 -2.02155441e-01 -4.71400004e-03 4.21660990e-01
9.90765870e-01 -1.32853001e-01 -9.01542127e-01 3.62834781e-01
-3.81696761e-01 -3.70027304e-01 5.37436306e-01 8.18766415e-01
-3.30122858e-01 -1.80372632e+00 -4.97020841e-01 1.61681429e-01
-7.57520556e-01 6.06082231e-02 -6.70068026e-01 1.41607773e+00
1.05242290e-01 1.12735760e+00 -7.33504770e-03 1.90041497e-01
3.21222126e-01 2.82833666e-01 5.30585051e-01 -8.25913668e-01
-1.04645705e+00 4.39574979e-02 1.12986577e+00 4.41540107e-02
-1.02521670e+00 -7.65380561e-01 -1.35721958e+00 -4.49341275e-02
-4.86704916e-01 4.35181767e-01 3.69668543e-01 1.45620906e+00
2.72687674e-02 6.73201621e-01 1.03064813e-01 -4.27630663e-01
-4.54518229e-01 -1.21273112e+00 1.07939631e-01 7.39445210e-01
1.86651707e-01 -8.23978662e-01 -9.93151441e-02 7.26178288e-02] | [10.209208488464355, 9.445548057556152] |
dc77bfc4-f427-47b4-b5d8-842ea81bcb8e | domain-invariant-feature-alignment-using | 2212.01590 | null | https://arxiv.org/abs/2212.01590v1 | https://arxiv.org/pdf/2212.01590v1.pdf | Domain-Invariant Feature Alignment Using Variational Inference For Partial Domain Adaptation | The standard closed-set domain adaptation approaches seek to mitigate distribution discrepancies between two domains under the constraint of both sharing identical label sets. However, in realistic scenarios, finding an optimal source domain with identical label space is a challenging task. Partial domain adaptation alleviates this problem of procuring a labeled dataset with identical label space assumptions and addresses a more practical scenario where the source label set subsumes the target label set. This, however, presents a few additional obstacles during adaptation. Samples with categories private to the source domain thwart relevant knowledge transfer and degrade model performance. In this work, we try to address these issues by coupling variational information and adversarial learning with a pseudo-labeling technique to enforce class distribution alignment and minimize the transfer of superfluous information from the source samples. The experimental findings in numerous cross-domain classification tasks demonstrate that the proposed technique delivers superior and comparable accuracy to existing methods. | ['Hemanth Venkateswara', 'Arunabha Sen', 'Suli Adeniye', 'Sandipan Choudhuri'] | 2022-12-03 | null | null | null | null | ['partial-domain-adaptation'] | ['methodology'] | [ 5.16657352e-01 1.31675228e-01 -3.46351057e-01 -4.63945240e-01
-9.87301350e-01 -9.83509719e-01 4.51071978e-01 -1.31080046e-01
-3.34328145e-01 1.30270302e+00 -2.01578975e-01 3.75722833e-02
-1.04609497e-01 -5.99311590e-01 -6.47151470e-01 -1.00791717e+00
4.95597661e-01 7.08180845e-01 1.09823942e-01 8.46105367e-02
1.45201683e-01 2.36807182e-01 -1.25608337e+00 9.04459283e-02
1.17379618e+00 8.90277326e-01 1.04863398e-01 -1.05812587e-01
-4.71406877e-01 4.35090989e-01 -9.33636367e-01 -6.29059076e-01
6.51747465e-01 -4.45916235e-01 -6.56332493e-01 1.39135420e-01
5.88415504e-01 -1.08540982e-01 8.82112607e-02 1.46818674e+00
5.09004712e-01 2.66609788e-01 1.14391720e+00 -1.61349201e+00
-9.59870219e-01 4.07130659e-01 -7.79670358e-01 1.23313526e-02
-2.80102901e-02 -1.12274460e-01 6.31150246e-01 -5.07438362e-01
6.62650049e-01 1.13866425e+00 5.73664725e-01 9.88693058e-01
-1.50204086e+00 -1.06256676e+00 3.08816940e-01 -1.04911990e-01
-1.68224192e+00 -2.94873744e-01 1.13910198e+00 -5.53045094e-01
2.36186609e-01 9.25354660e-03 -9.97746736e-02 1.51345694e+00
-8.43408257e-02 4.40299064e-01 1.32911861e+00 -3.32709253e-01
5.59359968e-01 9.40934896e-01 -3.00886598e-03 1.85382888e-02
2.84597546e-01 6.29981235e-02 -2.62432128e-01 -4.46933031e-01
5.32262921e-01 -1.40316740e-01 -2.69583911e-01 -1.01637948e+00
-8.27302277e-01 9.00618434e-01 1.53612956e-01 1.20167263e-01
-2.50052243e-01 -3.95393133e-01 4.44345742e-01 3.95176977e-01
7.04306364e-01 2.27024913e-01 -6.23235703e-01 4.01590526e-01
-8.56475830e-01 2.76151597e-01 6.85442865e-01 1.30545938e+00
7.11186945e-01 8.42257887e-02 -1.56238228e-01 1.14495409e+00
3.32206190e-01 6.95245743e-01 3.41460854e-01 -1.04255128e+00
6.41777515e-01 2.77767986e-01 3.37385714e-01 -6.40987873e-01
6.34374544e-02 -4.94915098e-01 -8.09156477e-01 3.64739150e-01
9.17084157e-01 -2.91270226e-01 -8.29626799e-01 2.23914504e+00
7.37682819e-01 1.76727921e-01 2.52036303e-01 8.37859511e-01
3.76781434e-01 4.16263819e-01 4.55768377e-01 -4.08420473e-01
1.07817769e+00 -5.62156737e-01 -9.23854947e-01 -2.19391853e-01
2.94047564e-01 -7.09483564e-01 1.04718113e+00 1.64606392e-01
-7.46472716e-01 -5.27893126e-01 -1.03537512e+00 1.77919686e-01
-4.19894487e-01 -2.52420157e-01 2.41017714e-01 9.39582467e-01
-5.49325705e-01 3.63879621e-01 -2.74953723e-01 -3.30576450e-01
6.86270773e-01 4.11325693e-01 -3.17437947e-01 -1.56717464e-01
-1.38109601e+00 7.46134222e-01 4.68751162e-01 -5.48104167e-01
-6.85295045e-01 -1.05676591e+00 -7.97773004e-01 -1.56823739e-01
3.95322859e-01 -5.00038147e-01 1.08596385e+00 -1.30416703e+00
-1.47705853e+00 1.07403016e+00 -6.19326271e-02 -2.93106705e-01
6.90350771e-01 5.82492203e-02 -4.59564209e-01 -2.58191615e-01
3.32064956e-01 6.78016663e-01 1.20392847e+00 -1.59295726e+00
-5.68476498e-01 -4.30799752e-01 -2.57028073e-01 3.53628039e-01
-4.57486361e-01 -1.70635685e-01 -5.13531342e-02 -8.01320195e-01
-1.48823917e-01 -9.05245543e-01 1.09074533e-01 1.49008662e-01
-2.19671026e-01 -2.35491723e-01 1.04067230e+00 -4.10216838e-01
8.12633574e-01 -2.34160399e+00 2.21976444e-01 1.03879772e-01
-1.52877644e-01 3.03239405e-01 -3.45311724e-02 2.91679278e-02
-2.16110989e-01 8.12049285e-02 -6.60939813e-01 -3.86831731e-01
1.05796203e-01 4.18847650e-01 -5.91116488e-01 5.70765913e-01
1.72967836e-01 3.68836582e-01 -7.62256145e-01 -7.21909583e-01
1.34511650e-01 3.15921694e-01 -5.51057637e-01 2.86200523e-01
-2.52083212e-01 9.38360512e-01 -4.39839363e-01 5.23666322e-01
1.21372294e+00 -3.00266314e-02 2.99517334e-01 6.50111288e-02
5.54879427e-01 -1.56685740e-01 -1.45105195e+00 1.67429090e+00
-1.56245977e-01 2.06715763e-01 2.60893106e-01 -1.25415730e+00
1.08875346e+00 2.94891894e-01 6.93482816e-01 -5.86982906e-01
1.15817897e-01 2.12901130e-01 -1.83702156e-01 -3.02111924e-01
2.48589590e-01 -7.45060027e-01 -2.98863351e-01 4.04575229e-01
2.19771832e-01 -2.23876029e-01 -3.40023190e-01 -9.64492932e-02
4.22545016e-01 3.82089645e-01 4.06801671e-01 -4.28311348e-01
5.65044165e-01 -6.77698897e-03 1.02997804e+00 5.94617546e-01
-7.32483089e-01 6.22761369e-01 3.17604512e-01 1.46125434e-02
-1.09446347e+00 -1.40122890e+00 -4.36559528e-01 9.51173067e-01
3.56745988e-01 4.91818905e-01 -9.28916156e-01 -1.17820776e+00
1.62437528e-01 9.44993556e-01 -6.19053185e-01 -3.58553201e-01
-3.14282209e-01 -6.87809706e-01 5.88186383e-01 2.79918134e-01
5.32411516e-01 -7.49148130e-01 -2.77410238e-03 -9.00366809e-03
-3.64608943e-01 -1.01819539e+00 -6.14471197e-01 1.47756800e-01
-7.82092392e-01 -9.34684098e-01 -9.63526487e-01 -9.65249896e-01
7.89986730e-01 7.83165079e-03 1.11294067e+00 -6.39503539e-01
9.63551849e-02 1.46291897e-01 -1.72893316e-01 -5.28661609e-01
-6.45307004e-01 8.10496584e-02 2.43958592e-01 1.85158610e-01
8.03242624e-01 -4.44478869e-01 -2.82786965e-01 4.55105454e-01
-1.04022491e+00 -4.43590313e-01 2.33517922e-02 8.75835359e-01
6.98223591e-01 3.40086699e-01 1.07907236e+00 -1.17846859e+00
5.83450675e-01 -9.29866731e-01 -5.80585659e-01 2.51158834e-01
-6.17261529e-01 1.93442963e-02 8.68874788e-01 -8.33858788e-01
-1.62113678e+00 8.38318840e-02 2.46792719e-01 -6.09225452e-01
-5.84957480e-01 -1.98858276e-01 -7.51250982e-01 1.01224080e-01
8.76946270e-01 1.28579646e-01 2.15077907e-01 -4.90620047e-01
3.05211097e-01 9.33313489e-01 4.32602763e-01 -8.47098887e-01
9.26639915e-01 6.70656621e-01 -1.90587893e-01 -4.76256818e-01
-8.81673217e-01 -3.51360977e-01 -8.72266173e-01 1.52402371e-01
6.79844677e-01 -9.45414722e-01 -6.20915145e-02 4.47921246e-01
-9.59677219e-01 -1.83211714e-01 -6.63286746e-01 2.69062489e-01
-6.21662617e-01 4.96264964e-01 -3.54511291e-02 -6.47192419e-01
6.70238137e-02 -1.03570545e+00 7.42271185e-01 2.44381815e-01
-2.39873573e-01 -1.30698037e+00 8.37920904e-02 4.27078336e-01
2.99472421e-01 2.81738281e-01 9.26897645e-01 -1.09684885e+00
-2.32849851e-01 1.30623505e-01 -4.79491316e-02 6.69273555e-01
4.99759883e-01 -5.14670134e-01 -1.10335445e+00 -4.38244760e-01
2.41702259e-01 -3.76184911e-01 4.07066762e-01 3.68269026e-01
9.30163383e-01 -6.50432706e-02 -3.21922421e-01 5.42724788e-01
1.43673801e+00 3.46447468e-01 3.87604922e-01 1.22650400e-01
5.41277885e-01 9.80191827e-01 8.27587605e-01 4.27971840e-01
1.12658739e-01 6.81844771e-01 1.53423190e-01 8.58322233e-02
-2.87792414e-01 -2.99224257e-01 1.61617935e-01 2.40980342e-01
6.19505465e-01 -4.36516106e-01 -7.03147233e-01 7.30648160e-01
-1.58397985e+00 -9.30004060e-01 1.70810014e-01 2.40903759e+00
1.06845903e+00 -9.50164199e-02 3.53437923e-02 -1.09024182e-01
1.11276627e+00 -6.54524788e-02 -1.00208724e+00 -1.50312990e-01
-1.59524202e-01 -6.60578683e-02 5.24575472e-01 5.34875751e-01
-1.20619023e+00 8.31303537e-01 6.47044802e+00 1.03516233e+00
-8.48655641e-01 4.18171853e-01 4.64158684e-01 -1.12900741e-01
-4.18672323e-01 -2.41100863e-01 -8.07963014e-01 7.78686523e-01
6.76930666e-01 -8.08335319e-02 3.17073226e-01 8.33877206e-01
-2.20372081e-01 1.31212488e-01 -1.19433665e+00 8.05216551e-01
8.64280313e-02 -6.74439669e-01 -7.69038275e-02 1.90087706e-01
9.44687247e-01 -2.99207747e-01 4.05037522e-01 4.19603616e-01
4.14719850e-01 -7.03172743e-01 5.37873983e-01 1.98757634e-01
1.09918833e+00 -8.74874055e-01 5.95374286e-01 5.22092640e-01
-7.57353723e-01 -4.33554314e-02 -5.28373778e-01 1.62533641e-01
6.33434504e-02 4.43975568e-01 -7.78247714e-01 3.89472008e-01
5.82692802e-01 4.80918378e-01 -1.82578176e-01 6.04346812e-01
4.71739359e-02 3.81648123e-01 -2.20133305e-01 4.04830635e-01
-1.48654670e-01 -3.24936002e-01 7.24520028e-01 8.79266322e-01
1.95671633e-01 -1.09066837e-01 1.17259927e-01 1.16799903e+00
-1.38696924e-01 -2.76100561e-02 -1.02534389e+00 3.25319886e-01
7.89522946e-01 7.31658041e-01 -4.66485292e-01 -2.78525382e-01
-4.05641228e-01 1.09982681e+00 4.04533714e-01 5.91865480e-01
-1.08424926e+00 -2.65726000e-01 1.00174105e+00 6.65637851e-02
9.99475643e-02 1.18863933e-01 -6.01989388e-01 -1.14776635e+00
-7.00052967e-03 -8.22409689e-01 6.24379694e-01 -3.59194607e-01
-1.79896545e+00 2.17697620e-01 3.18416297e-01 -1.47338569e+00
-1.62565038e-01 -4.37473089e-01 -1.68210417e-01 1.12660730e+00
-1.54663122e+00 -1.05391264e+00 -2.12448463e-02 1.01707375e+00
4.82560724e-01 -4.43284005e-01 8.38422477e-01 5.16982138e-01
-3.67693394e-01 1.06508589e+00 6.04818165e-01 -8.48415792e-02
1.20359898e+00 -1.26546490e+00 -1.34248689e-01 6.11072898e-01
-2.57153451e-01 5.50632954e-01 7.06757426e-01 -6.33976340e-01
-6.02005422e-01 -1.26657844e+00 5.21503568e-01 -6.38225973e-01
2.93583602e-01 -3.86128366e-01 -1.32736945e+00 6.93349063e-01
1.68462023e-01 -3.37956510e-02 1.12580049e+00 -1.37883186e-01
-5.48995137e-01 -1.27974272e-01 -1.85480905e+00 4.03443009e-01
9.12923932e-01 -4.78483528e-01 -7.63219476e-01 3.70350599e-01
6.23848557e-01 -2.53371090e-01 -9.89598691e-01 2.76425958e-01
1.79552019e-01 -6.82651997e-01 1.10676622e+00 -5.41213512e-01
1.15943469e-01 -3.63679916e-01 -3.49537104e-01 -1.57351446e+00
-1.94739461e-01 -3.75450760e-01 7.63524324e-02 1.74579465e+00
2.15707615e-01 -7.92997777e-01 8.67846370e-01 9.21643019e-01
2.78396308e-01 4.41525541e-02 -1.11630225e+00 -9.80990589e-01
7.94390082e-01 9.33681708e-03 7.34259546e-01 1.32766736e+00
-2.55424321e-01 2.14254647e-01 -4.28888559e-01 3.22749227e-01
1.19948137e+00 9.80932917e-03 5.45662999e-01 -1.32013083e+00
-7.61288330e-02 -6.06632605e-02 7.73035958e-02 -7.82036185e-01
7.59112239e-01 -8.77701283e-01 1.18671447e-01 -9.88551617e-01
2.38817692e-01 -9.53900158e-01 -4.94374037e-01 2.68772244e-01
-1.87369645e-01 1.43700987e-01 1.03836752e-01 3.53096545e-01
-2.92034864e-01 5.65504551e-01 1.27841234e+00 -2.45231792e-01
-8.97634923e-02 1.85292959e-02 -9.55464184e-01 5.64670503e-01
8.61801326e-01 -8.77769053e-01 -9.25007761e-01 -3.28388005e-01
-3.88665080e-01 -1.43380642e-01 1.47502422e-01 -7.98941135e-01
4.77157831e-02 -4.93765354e-01 3.11910391e-01 -2.04620019e-01
2.40060285e-01 -1.36297929e+00 2.50999093e-01 1.47008449e-01
-3.65685016e-01 -7.18894541e-01 2.45181158e-01 8.88458848e-01
-1.66041017e-01 -4.29793149e-01 1.39935660e+00 6.67367456e-03
-6.74694180e-01 1.90218791e-01 -9.27842185e-02 4.45744783e-01
1.53002119e+00 -3.39756638e-01 -1.18139908e-01 -1.69379607e-01
-6.40666068e-01 1.95260525e-01 6.88211024e-01 5.70620120e-01
2.00672150e-01 -1.48629320e+00 -7.04783022e-01 4.22110349e-01
1.40195847e-01 1.76879987e-01 4.36299086e-01 1.80336922e-01
1.30994678e-01 2.33788580e-01 -3.75017703e-01 -6.02794349e-01
-1.10309708e+00 8.31221879e-01 4.53640580e-01 -1.99666679e-01
-2.52387971e-01 9.13638830e-01 6.29623830e-01 -9.45423841e-01
1.72994047e-01 2.81016886e-01 -1.80182993e-01 1.97228655e-01
3.02984953e-01 2.80237377e-01 -2.08395839e-01 -7.64015973e-01
-2.96407372e-01 6.38234973e-01 -2.14386702e-01 3.32471309e-03
7.96100378e-01 -4.84889239e-01 2.30568692e-01 3.21969599e-01
1.30528450e+00 -1.73610270e-01 -1.67735040e+00 -6.73510194e-01
-1.87178642e-01 -5.88657498e-01 -3.62652957e-01 -9.28247988e-01
-8.73170435e-01 8.16339374e-01 8.92212093e-01 3.71159092e-02
1.13521373e+00 -1.65989585e-02 6.06607676e-01 -1.16088137e-01
4.61183816e-01 -1.35572517e+00 2.10560840e-02 2.68590301e-01
6.89031422e-01 -1.58028233e+00 -1.91302568e-01 -5.86728334e-01
-9.78008628e-01 4.57020760e-01 9.01748896e-01 -2.89402530e-02
6.65666223e-01 1.32190883e-01 2.55564690e-01 2.43600860e-01
-2.44881928e-01 2.28103116e-01 2.06192270e-01 1.23291874e+00
9.83328149e-02 2.31985405e-01 -1.44247249e-01 6.48256421e-01
1.80216983e-01 -1.00889124e-01 2.74839014e-01 7.87364841e-01
-2.13502571e-01 -1.40558779e+00 -8.41034114e-01 1.08270474e-01
-7.04369009e-01 1.73148602e-01 -3.12364966e-01 8.20964754e-01
3.97122145e-01 8.96997631e-01 1.90982103e-01 7.59628862e-02
2.53311038e-01 5.07303834e-01 3.87898624e-01 -6.67880714e-01
-2.41931289e-01 -8.65456823e-04 -3.34905297e-01 -5.25925979e-02
-4.38523233e-01 -7.35036135e-01 -1.04779041e+00 -2.09417462e-01
-2.67564029e-01 1.63421765e-01 3.79268765e-01 7.73679137e-01
4.27022040e-01 4.28868502e-01 5.14477670e-01 -4.48774695e-01
-1.17079771e+00 -9.33212101e-01 -9.87120688e-01 9.22770917e-01
4.05906767e-01 -9.76677179e-01 -4.30377752e-01 3.27267885e-01] | [10.358121871948242, 3.1683826446533203] |
63f22dfd-781e-47fa-a326-4fbb99bbc9ea | a-cohesive-distillation-architecture-for | 2301.08130 | null | https://arxiv.org/abs/2301.08130v2 | https://arxiv.org/pdf/2301.08130v2.pdf | A Cohesive Distillation Architecture for Neural Language Models | A recent trend in Natural Language Processing is the exponential growth in Language Model (LM) size, which prevents research groups without a necessary hardware infrastructure from participating in the development process. This study investigates methods for Knowledge Distillation (KD) to provide efficient alternatives to large-scale models. In this context, KD means extracting information about language encoded in a Neural Network and Lexical Knowledge Databases. We developed two methods to test our hypothesis that efficient architectures can gain knowledge from LMs and extract valuable information from lexical sources. First, we present a technique to learn confident probability distribution for Masked Language Modeling by prediction weighting of multiple teacher networks. Second, we propose a method for Word Sense Disambiguation (WSD) and lexical KD that is general enough to be adapted to many LMs. Our results show that KD with multiple teachers leads to improved training convergence. When using our lexical pre-training method, LM characteristics are not lost, leading to increased performance in Natural Language Understanding (NLU) tasks over the state-of-the-art while adding no parameters. Moreover, the improved semantic understanding of our model increased the task performance beyond WSD and NLU in a real-problem scenario (Plagiarism Detection). This study suggests that sophisticated training methods and network architectures can be superior over scaling trainable parameters. On this basis, we suggest the research area should encourage the development and use of efficient models and rate impacts resulting from growing LM size equally against task performance. | ['Jan Philip Wahle'] | 2023-01-12 | null | null | null | null | ['word-sense-disambiguation'] | ['natural-language-processing'] | [ 1.51958779e-01 4.98005450e-01 -1.45611435e-01 -1.72866866e-01
-6.69483960e-01 -5.78654826e-01 4.20478791e-01 3.51558179e-01
-9.79014277e-01 7.63261020e-01 1.05249025e-01 -5.81955373e-01
-1.55393407e-01 -8.62989485e-01 -6.95018232e-01 -7.34670535e-02
3.64628881e-01 4.56182718e-01 4.48168695e-01 -1.41749248e-01
3.35934550e-01 5.26381493e-01 -1.60731864e+00 4.41772968e-01
9.66668606e-01 6.45877421e-01 8.00686479e-01 4.88411009e-01
-7.05535710e-01 9.13417280e-01 -1.02197373e+00 -3.57338011e-01
-2.41686143e-02 -1.61044344e-01 -1.09589636e+00 -3.86734217e-01
5.02171814e-01 -1.37817532e-01 1.66713670e-01 8.55591238e-01
7.20080376e-01 2.33592436e-01 4.49429721e-01 -8.90502632e-01
-7.41832316e-01 9.61290836e-01 -9.27045420e-02 2.61644959e-01
2.42786482e-01 -1.08584948e-01 6.82440817e-01 -7.82848835e-01
5.76373279e-01 1.39959908e+00 7.70121276e-01 6.83362246e-01
-1.16685307e+00 -7.31136501e-01 6.73547806e-03 2.96206713e-01
-1.46519458e+00 -4.21821505e-01 4.02157187e-01 -3.69315743e-01
1.71670175e+00 -4.93489206e-02 2.50597209e-01 9.80444252e-01
-1.34316504e-01 7.70387590e-01 1.16570365e+00 -1.17687953e+00
1.75326943e-01 8.79163682e-01 4.02147472e-01 6.63254738e-01
5.93322992e-01 -1.24090016e-01 -8.36975694e-01 2.70603993e-03
4.83009547e-01 -6.65723324e-01 -4.78749834e-02 -2.19510458e-02
-7.72492290e-01 7.90611744e-01 -2.13901728e-01 6.73703015e-01
-2.64435410e-01 -1.89615235e-01 3.78588527e-01 5.20685554e-01
5.72350383e-01 1.07530916e+00 -8.50589752e-01 -1.78209811e-01
-1.14258575e+00 -7.14694485e-02 1.18001723e+00 7.72312284e-01
8.58193934e-01 1.75610915e-01 -1.24210268e-02 9.92915690e-01
1.38089582e-01 4.99376476e-01 9.75173533e-01 -8.80022764e-01
3.44632894e-01 7.49180853e-01 -1.23578973e-01 -1.01272726e+00
-5.49458802e-01 -5.12413204e-01 -2.80060530e-01 1.68148410e-02
4.88400161e-01 -2.91434288e-01 -5.85071921e-01 1.80377877e+00
7.53557310e-02 1.99693799e-01 3.86795104e-01 2.78912395e-01
7.85022438e-01 6.27987385e-01 3.16078901e-01 -2.25354180e-01
1.52566075e+00 -6.78184688e-01 -7.19340622e-01 -5.20039439e-01
1.18494403e+00 -9.45475161e-01 1.38106537e+00 6.62062049e-01
-1.04198682e+00 -8.10282052e-01 -1.03974712e+00 -8.65898281e-02
-7.85461307e-01 1.98747590e-01 4.28939879e-01 9.30937886e-01
-1.19021714e+00 4.97069955e-01 -5.56560755e-01 -5.40745199e-01
1.50739923e-01 3.26678902e-01 -6.97805285e-02 5.46453297e-02
-1.56675291e+00 1.49761546e+00 9.23517525e-01 -3.77152890e-01
-5.00302970e-01 -9.00635183e-01 -7.56159127e-01 3.50640208e-01
5.32409310e-01 -5.89055598e-01 1.27988267e+00 -9.99509037e-01
-1.49010444e+00 7.78572559e-01 -1.55309349e-01 -6.86272740e-01
8.28949064e-02 -4.81975466e-01 -3.53292853e-01 1.25245184e-01
8.74018967e-02 8.12248170e-01 5.57173967e-01 -1.16560996e+00
-6.55951679e-01 1.14613980e-01 -1.17824763e-01 6.63257167e-02
-1.01558042e+00 1.46181658e-01 6.31963089e-02 -5.76911330e-01
-3.25334758e-01 -5.30452430e-01 2.67339535e-02 -5.35216033e-01
-5.10940440e-02 -5.53015649e-01 5.26240885e-01 -7.97488987e-01
1.61770582e+00 -1.94375885e+00 -3.45108986e-01 1.61556393e-01
7.03709722e-02 9.32541847e-01 -3.79086137e-01 4.33579355e-01
2.15357039e-02 4.46157813e-01 2.54963219e-01 -2.98552155e-01
7.91615695e-02 3.14670235e-01 -3.91120195e-01 -3.26133847e-01
3.54563385e-01 7.92146742e-01 -8.08658302e-01 -5.22623301e-01
2.60520242e-02 3.65617156e-01 -4.18162704e-01 1.90695524e-01
-3.55086923e-01 -9.25872847e-02 -2.12347254e-01 9.31821316e-02
3.90833855e-01 -2.89415419e-01 1.81292444e-01 -4.23183888e-02
-7.01796031e-03 7.00941682e-01 -1.50651205e+00 1.56195712e+00
-9.31084752e-01 7.78718472e-01 -8.64930972e-02 -9.62166607e-01
1.08044159e+00 4.92135912e-01 -1.47350142e-02 -7.12314963e-01
1.17334403e-01 2.28590056e-01 -4.07118648e-02 -6.45421147e-01
7.42749095e-01 -1.15385406e-01 3.04536343e-01 6.99676394e-01
3.95574450e-01 -1.25236809e-01 2.49801680e-01 2.29813844e-01
9.49283838e-01 -7.20247701e-02 3.73458982e-01 -4.72763270e-01
5.22894144e-01 1.87508479e-01 1.64662033e-01 1.02907157e+00
1.06813602e-01 -3.35744381e-01 1.39640391e-01 -1.59924328e-01
-7.99858749e-01 -8.02855611e-01 1.88921615e-02 1.49712372e+00
-3.38174611e-01 -5.74451268e-01 -9.85426068e-01 -4.39889818e-01
-1.97294801e-01 1.28023529e+00 5.22091612e-02 -3.63696188e-01
-5.40634811e-01 -4.80312675e-01 1.02992117e+00 5.19266546e-01
4.74433959e-01 -1.32205307e+00 -6.90915048e-01 3.77098352e-01
-2.11719260e-01 -1.27932632e+00 8.15813467e-02 3.07206780e-01
-9.61330473e-01 -8.29746783e-01 -3.80164176e-01 -8.44928563e-01
4.19254363e-01 2.20344797e-01 1.23949063e+00 -1.19503886e-02
-2.83647418e-01 6.57752395e-01 -2.97199279e-01 -8.65337431e-01
-9.35902059e-01 3.92600536e-01 3.41055006e-01 -6.21904075e-01
1.12987924e+00 -4.41492617e-01 2.87905298e-02 7.15194643e-02
-1.08488905e+00 -9.13571790e-02 7.85977721e-01 6.05884135e-01
2.26313993e-01 4.09827679e-01 8.85845780e-01 -8.50540876e-01
1.19930148e+00 -3.41695368e-01 -4.81781572e-01 4.99416530e-01
-1.05410159e+00 4.58114922e-01 5.01334131e-01 -8.34525585e-01
-1.32891738e+00 -1.41112685e-01 3.57238902e-03 -1.12964191e-01
-5.35946608e-01 7.75094211e-01 1.35599613e-01 -1.08631356e-02
1.18404913e+00 1.83202133e-01 3.40602100e-02 -6.00103796e-01
4.49367493e-01 8.11885715e-01 1.42746821e-01 -9.42630291e-01
5.68446338e-01 -7.64137805e-02 -3.55772436e-01 -1.22222757e+00
-1.01790535e+00 -5.25673747e-01 -7.54238427e-01 1.75672725e-01
6.60475135e-01 -9.95661736e-01 -4.62255865e-01 2.16970533e-01
-1.35409236e+00 -3.82454932e-01 -3.83784562e-01 6.74047172e-01
-3.24979216e-01 4.54622477e-01 -4.88932908e-01 -9.60718691e-01
-3.80223155e-01 -7.39222527e-01 5.22417247e-01 3.24697524e-01
-7.13867068e-01 -1.18467057e+00 -3.27101089e-02 4.81566250e-01
6.63668394e-01 -5.65073550e-01 1.18415534e+00 -1.29768574e+00
-3.40605170e-01 2.71212496e-03 -1.44538254e-01 7.86009192e-01
-9.41345170e-02 -3.15974534e-01 -1.26438618e+00 -8.16070437e-02
7.87422433e-02 -6.40719235e-01 6.56077921e-01 2.13811710e-01
8.47197652e-01 -3.92894089e-01 -2.34890714e-01 4.28180769e-03
1.28246486e+00 1.06704593e-01 2.32132301e-01 5.78375459e-01
4.55942154e-01 9.58265424e-01 3.92273307e-01 1.84295513e-02
2.71671146e-01 3.30164611e-01 -2.45769948e-01 3.53974663e-02
-3.07085276e-01 -4.21037048e-01 5.37456632e-01 1.02758551e+00
2.92530566e-01 -1.59588516e-01 -1.24510157e+00 6.74408853e-01
-1.46151137e+00 -6.83910549e-01 1.60027876e-01 2.15002465e+00
1.23583269e+00 3.32646847e-01 -1.64521918e-01 -5.15542440e-02
4.92544979e-01 -2.70705462e-01 -2.01732472e-01 -6.90038681e-01
-3.14732753e-02 5.86938620e-01 3.89190406e-01 6.91030622e-01
-5.63488781e-01 1.33099675e+00 6.08887291e+00 1.33016467e+00
-1.01504886e+00 2.93679416e-01 2.99380749e-01 1.26052946e-01
-1.48207590e-01 -1.54389277e-01 -1.44636464e+00 -1.12200566e-02
1.57355177e+00 -3.23685408e-01 2.12954640e-01 1.01084054e+00
1.14155799e-01 -2.89817244e-01 -9.11784887e-01 5.97422540e-01
2.44223014e-01 -1.23601139e+00 2.55355597e-01 -2.58273363e-01
4.76779193e-01 -8.49206373e-02 -2.80003190e-01 6.29754841e-01
4.73139435e-01 -1.02370286e+00 4.04339552e-01 3.51691842e-01
4.69735026e-01 -6.59064353e-01 7.59736717e-01 9.10531163e-01
-8.64695191e-01 -5.08523546e-03 -6.18034542e-01 -2.40777850e-01
-5.61218373e-02 4.45190161e-01 -1.55156934e+00 3.52124453e-01
4.97768641e-01 1.58509180e-01 -7.69977570e-01 7.28988230e-01
-3.40303153e-01 9.35050488e-01 -4.84661758e-01 -4.12484050e-01
8.74402672e-02 3.72858375e-01 3.41758430e-01 1.61623394e+00
1.80092469e-01 -1.26519009e-01 1.76426142e-01 8.89434814e-01
1.53944165e-01 4.12977308e-01 -6.84670925e-01 -2.25935176e-01
8.77095044e-01 9.56208706e-01 -6.66160464e-01 -5.36679029e-01
-5.10728836e-01 6.11127555e-01 4.83143240e-01 2.37072513e-01
-2.47798219e-01 -3.32461774e-01 3.61818105e-01 8.47819820e-02
9.18013975e-02 -2.85402417e-01 -3.65002841e-01 -9.27553117e-01
-1.04066901e-01 -9.91474569e-01 4.22133446e-01 -6.91460490e-01
-1.40736961e+00 6.01250052e-01 3.06225598e-01 -6.50131226e-01
-4.34052557e-01 -9.27276611e-01 -3.29241008e-01 1.04377997e+00
-1.65158737e+00 -9.15902972e-01 1.56062990e-01 4.44288194e-01
6.68839097e-01 -4.14372891e-01 1.13783503e+00 1.38768852e-01
-2.43090674e-01 7.05825031e-01 -1.25780374e-01 1.23435572e-01
9.40705299e-01 -1.05071688e+00 2.11849399e-02 7.38109946e-01
3.46993059e-01 9.85416532e-01 5.00754237e-01 -7.59550869e-01
-7.84715295e-01 -8.25704634e-01 1.40300834e+00 -6.26517415e-01
8.69683623e-01 -2.30866104e-01 -1.13133705e+00 4.17772740e-01
2.84656614e-01 -7.84798443e-01 1.03532159e+00 3.28894973e-01
-3.13637137e-01 6.68715909e-02 -9.14841831e-01 5.00705302e-01
5.85390627e-01 -6.84379399e-01 -1.12836301e+00 2.03670070e-01
9.59503293e-01 -2.37287264e-02 -7.74016738e-01 9.16838199e-02
2.01611921e-01 -5.01717210e-01 1.03593349e+00 -6.99961245e-01
2.54534692e-01 8.19642991e-02 1.57505348e-02 -1.26461351e+00
-1.76441262e-03 -4.41540986e-01 -1.49774864e-01 1.43306208e+00
7.41125762e-01 -6.27811015e-01 6.71139240e-01 7.99061656e-01
4.79369909e-02 -3.48244965e-01 -7.54997969e-01 -1.00896168e+00
4.90206450e-01 -7.59464324e-01 3.48989427e-01 1.22546911e+00
-2.46665347e-03 6.24540389e-01 -8.82001873e-03 1.79231524e-01
2.64485717e-01 -4.01263356e-01 5.49727440e-01 -1.38773811e+00
-2.91678190e-01 -4.87223715e-01 -4.72623706e-02 -9.90531266e-01
5.41447520e-01 -9.56611693e-01 -1.00117467e-01 -1.45169008e+00
-1.38340607e-01 -2.78815508e-01 -2.65060335e-01 7.89978325e-01
-2.55425960e-01 -2.64186352e-01 2.73842722e-01 -6.89696595e-02
-3.62139046e-01 1.16466455e-01 9.28442478e-01 1.27543271e-01
-4.47155803e-01 -1.78740680e-01 -7.57331312e-01 8.99794102e-01
9.37061131e-01 -7.61200428e-01 -7.45271206e-01 -4.40072507e-01
4.69622254e-01 -2.78972119e-01 1.70001537e-01 -1.10312891e+00
5.80954731e-01 4.93639968e-02 2.22750574e-01 -2.72656798e-01
1.21537179e-01 -8.63182127e-01 -3.35389286e-01 3.95965874e-01
-4.63649005e-01 -5.36867827e-02 8.19115341e-01 1.70759469e-01
-1.41952306e-01 -8.87337923e-01 5.22568762e-01 -3.82619828e-01
-9.76698279e-01 -4.88704443e-01 -7.50386894e-01 1.40025571e-01
6.81390762e-01 -2.96796381e-01 -4.35466588e-01 -2.38825694e-01
-6.68532729e-01 9.45545658e-02 -3.92741902e-04 5.05728781e-01
4.17566210e-01 -8.27211320e-01 -5.41972458e-01 2.21147269e-01
-1.56637952e-01 1.00597106e-02 -4.37689722e-02 5.60211539e-01
-3.76041710e-01 8.01517487e-01 -7.41165951e-02 -2.37216905e-01
-1.38173401e+00 5.43089867e-01 1.18206128e-01 -5.11257231e-01
-2.49346420e-01 1.00437486e+00 -1.10899568e-01 -7.56333828e-01
4.89712298e-01 -3.15026224e-01 -4.56656933e-01 2.99156398e-01
7.84866989e-01 4.55107719e-01 1.52556673e-01 -7.21971840e-02
2.55380739e-02 2.42839113e-01 -1.98728904e-01 -1.62740096e-01
1.07837141e+00 -2.23083392e-01 -1.13118447e-01 5.31910479e-01
6.82540774e-01 1.47926539e-01 -5.66475451e-01 -5.60147464e-01
5.28221250e-01 -3.92953642e-02 1.65955544e-01 -1.16353452e+00
-4.32524025e-01 1.00631881e+00 5.58994114e-01 3.07470679e-01
1.00070047e+00 -1.36371121e-01 5.54200709e-01 1.08122969e+00
3.04955721e-01 -1.48982370e+00 1.73025832e-01 7.73547351e-01
3.68421525e-01 -1.06292725e+00 -1.29019320e-01 -3.08233261e-01
-5.44854164e-01 1.27734888e+00 8.16839039e-01 2.58357882e-01
6.51718140e-01 4.07012641e-01 3.68349671e-01 -1.31257936e-01
-7.09136546e-01 -2.79475719e-01 3.21492434e-01 6.83353961e-01
6.19917274e-01 -2.15522364e-01 -3.48864943e-01 9.22294974e-01
-4.14816320e-01 2.30221733e-01 5.42293966e-01 8.91561806e-01
-9.30064082e-01 -1.35911953e+00 -4.85850126e-01 3.90165061e-01
-4.60807890e-01 -5.93534589e-01 -4.64011580e-01 9.59449589e-01
2.34628737e-01 1.23479891e+00 -2.15797648e-01 -2.02889726e-01
2.51785547e-01 8.12554419e-01 1.93468630e-01 -1.06144595e+00
-7.07142651e-01 -2.12383747e-01 3.99650931e-01 -3.18375945e-01
-4.94666070e-01 -1.26313239e-01 -1.14788163e+00 -2.46381804e-01
-7.28769362e-01 4.21378195e-01 7.18937039e-01 1.16303968e+00
4.83838648e-01 5.28200209e-01 -1.60280079e-01 -2.64290571e-01
-7.52510667e-01 -1.19285905e+00 -3.28788102e-01 2.94580460e-02
-2.27209747e-01 -5.63979924e-01 -2.66803831e-01 2.50469059e-01] | [10.618781089782715, 8.770041465759277] |
e2c268a6-7761-4036-bf5a-457a0fc5acbe | domain-generalization-using-causal-matching-1 | 2006.07500 | null | https://arxiv.org/abs/2006.07500v3 | https://arxiv.org/pdf/2006.07500v3.pdf | Domain Generalization using Causal Matching | In the domain generalization literature, a common objective is to learn representations independent of the domain after conditioning on the class label. We show that this objective is not sufficient: there exist counter-examples where a model fails to generalize to unseen domains even after satisfying class-conditional domain invariance. We formalize this observation through a structural causal model and show the importance of modeling within-class variations for generalization. Specifically, classes contain objects that characterize specific causal features, and domains can be interpreted as interventions on these objects that change non-causal features. We highlight an alternative condition: inputs across domains should have the same representation if they are derived from the same object. Based on this objective, we propose matching-based algorithms when base objects are observed (e.g., through data augmentation) and approximate the objective when objects are not observed (MatchDG). Our simple matching-based algorithms are competitive to prior work on out-of-domain accuracy for rotated MNIST, Fashion-MNIST, PACS, and Chest-Xray datasets. Our method MatchDG also recovers ground-truth object matches: on MNIST and Fashion-MNIST, top-10 matches from MatchDG have over 50% overlap with ground-truth matches. | ['Shruti Tople', 'Divyat Mahajan', 'Amit Sharma'] | 2020-06-12 | domain-generalization-using-causal-matching | null | null | arxiv-2020-6 | ['rotated-mnist'] | ['computer-vision'] | [ 5.41882217e-01 3.92170131e-01 -6.99397266e-01 -8.31146777e-01
-6.43758535e-01 -7.06730485e-01 8.64230454e-01 2.05019131e-01
-1.92166179e-01 8.94783020e-01 4.36612815e-01 7.01134726e-02
-5.75978518e-01 -7.88445652e-01 -1.26192081e+00 -6.74246788e-01
1.06346041e-01 6.98508799e-01 2.56070763e-01 6.82081506e-02
5.95345609e-02 7.73255885e-01 -1.45156276e+00 7.66162813e-01
4.76045370e-01 6.96055710e-01 -4.70163487e-02 2.66309619e-01
3.11429888e-01 7.53132105e-01 -4.06126499e-01 -2.31751591e-01
3.06636363e-01 -5.85044324e-01 -1.06556475e+00 4.02245671e-01
7.77882576e-01 -3.32385808e-01 -7.42995679e-01 9.10702288e-01
1.44912735e-01 2.92333931e-01 1.30301118e+00 -1.44456053e+00
-1.22817969e+00 5.54600179e-01 -3.53851408e-01 2.92593062e-01
3.77030849e-01 -8.38032812e-02 1.24726057e+00 -5.77275097e-01
8.05577397e-01 1.43952096e+00 4.33800548e-01 6.93050981e-01
-1.74878550e+00 -8.36474419e-01 4.67798144e-01 3.13689500e-01
-1.10736763e+00 -2.60913938e-01 5.72369158e-01 -5.91289580e-01
7.44638860e-01 1.19470544e-01 2.45571047e-01 1.58774769e+00
1.74636960e-01 6.17276013e-01 1.12173462e+00 -4.50947642e-01
4.63503003e-01 1.76338926e-01 3.66394907e-01 2.04139829e-01
6.01241350e-01 5.40972888e-01 -8.50489736e-01 -3.57607186e-01
7.07306027e-01 1.00116469e-01 -1.93773374e-01 -6.83564007e-01
-1.12057161e+00 8.66948009e-01 7.80705690e-01 1.29325375e-01
-3.86122853e-01 1.21589415e-01 1.92036435e-01 3.48004073e-01
2.13883176e-01 6.52496338e-01 -7.25702345e-01 6.16414428e-01
-5.63615859e-01 6.81142151e-01 3.94218981e-01 1.28813446e+00
6.67299449e-01 -3.43359321e-01 1.27167162e-02 5.47608733e-01
-5.12658469e-02 3.84471774e-01 5.92604816e-01 -7.35941410e-01
1.71240553e-01 5.71137488e-01 4.89139333e-02 -7.45841563e-01
-6.00336790e-01 -5.99818885e-01 -5.99440753e-01 -2.25099213e-02
4.68402356e-01 3.99258196e-01 -1.23225760e+00 2.29130697e+00
1.49707153e-01 1.94241479e-01 1.81111231e-01 8.12165856e-01
8.20432186e-01 9.00367126e-02 4.39276189e-01 -9.18338746e-02
1.24737406e+00 -2.70564377e-01 -4.39491034e-01 -3.21831673e-01
7.20782459e-01 -2.97962010e-01 1.14296114e+00 3.59226525e-01
-6.41318738e-01 -5.28454065e-01 -1.13094532e+00 3.47227678e-02
-3.30924004e-01 -4.72011268e-02 5.99915504e-01 4.76579726e-01
-4.58796591e-01 7.53913820e-01 -7.12547362e-01 -5.81240296e-01
5.77686787e-01 4.04185265e-01 -6.84545279e-01 -2.49761477e-01
-1.06307399e+00 9.83345270e-01 6.41442358e-01 -6.46275282e-01
-1.41012418e+00 -1.03898907e+00 -6.21421814e-01 -9.41517055e-02
2.83316046e-01 -8.40163350e-01 1.32754660e+00 -1.02424073e+00
-5.87929964e-01 1.18771231e+00 -1.78828493e-01 -5.39086163e-01
2.17428252e-01 -2.52963722e-01 -5.93335688e-01 2.11799160e-01
3.92363816e-01 6.95501804e-01 9.15475428e-01 -1.47732365e+00
-5.72358370e-01 -6.53254092e-01 5.78798838e-02 9.55605283e-02
-2.72107363e-01 -3.07722628e-01 9.99658778e-02 -7.48646498e-01
5.76539040e-01 -1.06175721e+00 -7.28637278e-02 2.05567237e-02
-5.54577529e-01 -2.83162951e-01 7.55627811e-01 -2.49752924e-01
6.93390846e-01 -2.30592227e+00 1.28012270e-01 1.36830047e-01
2.76826918e-01 -4.74124879e-01 -5.46131330e-03 7.52773434e-02
-8.31543863e-01 8.97372141e-02 -3.55906665e-01 -7.00060697e-03
-4.60463576e-02 6.93236887e-01 -8.62647414e-01 8.03081036e-01
4.58299160e-01 7.87657082e-01 -8.66328716e-01 -2.30467513e-01
1.45745846e-02 -1.21963732e-01 -7.67822027e-01 -1.57117471e-02
-3.74495178e-01 5.55711925e-01 -4.07292485e-01 4.89808381e-01
6.29437149e-01 -4.94213879e-01 2.22709328e-01 -2.86604613e-01
5.42677164e-01 4.30045992e-01 -1.29032469e+00 1.53530884e+00
-1.40571967e-01 3.15205514e-01 -4.89623249e-01 -1.35192811e+00
7.48703718e-01 1.83762029e-01 3.74923229e-01 -6.19310915e-01
-1.42204776e-01 2.23493114e-01 1.20919488e-01 -3.53506029e-01
3.56985241e-01 -7.29192734e-01 -2.27721125e-01 4.27314252e-01
4.64925915e-01 -1.53292418e-01 -2.96708047e-01 4.07491148e-01
1.17019928e+00 -2.57961825e-03 5.34320652e-01 -3.75638783e-01
5.35955802e-02 3.38338792e-01 6.67974651e-01 1.13424277e+00
2.74314076e-01 7.30007589e-01 4.93620008e-01 -1.92811415e-01
-1.06087649e+00 -1.64062893e+00 -7.11086094e-01 1.12597096e+00
2.83854216e-01 6.24578968e-02 -1.95215553e-01 -9.84692633e-01
4.52052146e-01 1.11503398e+00 -1.17081439e+00 -6.43438578e-01
-3.77485961e-01 -8.32683861e-01 4.55694765e-01 9.62716162e-01
1.21605374e-01 -8.14516187e-01 -3.24860156e-01 2.89188810e-02
1.78668246e-01 -1.05859447e+00 3.91901582e-02 5.79674661e-01
-9.96036768e-01 -1.29388595e+00 -4.22112018e-01 -3.59067321e-01
8.07561874e-01 5.73160201e-02 1.17015183e+00 -3.22677016e-01
-2.86046296e-01 7.00442493e-01 -3.49986911e-01 -4.73705947e-01
-4.16588068e-01 -6.06343672e-02 3.56297016e-01 -8.63377228e-02
4.72510278e-01 -6.57989740e-01 -3.36219609e-01 5.87499619e-01
-1.06182468e+00 -3.11661273e-01 5.58034956e-01 1.03103769e+00
8.15352559e-01 -2.85567865e-02 8.71320248e-01 -1.14549220e+00
2.32618615e-01 -8.68248165e-01 -3.63814443e-01 2.00976834e-01
-5.20026803e-01 1.63548246e-01 4.49456185e-01 -8.79400849e-01
-1.11017144e+00 2.64156252e-01 4.19179440e-01 -5.05880952e-01
-4.70814139e-01 3.40533882e-01 -3.23912382e-01 3.67590576e-01
1.41486990e+00 1.81024913e-02 -2.56449819e-01 -5.90071023e-01
5.02098918e-01 3.16403151e-01 8.97685647e-01 -9.73237276e-01
7.22052872e-01 1.00950110e+00 4.15317416e-01 -5.86867929e-01
-1.20520842e+00 -3.94062012e-01 -6.99571908e-01 2.61845201e-01
8.33755016e-01 -9.53881860e-01 -2.86598802e-01 -3.00341342e-02
-8.60382318e-01 -1.00936159e-01 -7.45218158e-01 7.69772470e-01
-7.59903371e-01 7.63287097e-02 -9.42741409e-02 -4.16350096e-01
4.73286867e-01 -8.31921995e-01 1.07182968e+00 -3.97661962e-02
-4.36439335e-01 -7.81371832e-01 -2.59348929e-01 -1.45575628e-01
-1.64759323e-01 2.86700130e-01 1.33930814e+00 -1.19142461e+00
-5.13741851e-01 -2.45482147e-01 -2.56595522e-01 3.18326980e-01
3.89376491e-01 -5.35515308e-01 -1.04044354e+00 -1.72871947e-01
5.64553961e-03 -3.99322271e-01 1.00659513e+00 5.96868038e-01
1.40240240e+00 -1.74470142e-01 -7.32172072e-01 4.14683968e-01
1.16094363e+00 8.88400152e-02 2.90071160e-01 2.00750947e-01
4.05978084e-01 6.64325655e-01 5.48581660e-01 3.17084730e-01
-4.57290467e-03 8.27331126e-01 4.64178711e-01 -8.99220109e-02
-2.39563152e-01 -4.14945215e-01 -4.19135503e-02 -3.18742126e-01
1.90853268e-01 -1.60981119e-01 -7.85407424e-01 8.48286569e-01
-1.70758331e+00 -8.95617485e-01 -3.50886852e-01 2.16923499e+00
8.63351345e-01 2.01983124e-01 1.66702136e-01 -9.70777199e-02
5.67519069e-01 -3.48445565e-01 -9.02810752e-01 9.60942432e-02
-5.18399298e-01 4.36711133e-01 6.07972026e-01 3.01204294e-01
-1.10605788e+00 5.42962074e-01 6.44875336e+00 4.44208443e-01
-7.47769296e-01 2.79561460e-01 3.85177523e-01 -1.03194475e-01
-4.51464206e-01 1.30766749e-01 -8.75400484e-01 5.74224442e-02
6.67980492e-01 -4.06089008e-01 9.05546248e-02 1.12117326e+00
-2.22996920e-01 2.90227495e-02 -2.01368880e+00 6.21679842e-01
-3.22548039e-02 -1.28367901e+00 1.88759282e-01 2.16183960e-01
8.79786372e-01 -1.28806055e-01 2.76972592e-01 2.80439913e-01
6.68019056e-01 -1.09142220e+00 6.91540658e-01 4.54463810e-01
7.25548327e-01 -4.73651737e-01 2.71503419e-01 1.98644504e-01
-6.31530881e-01 -1.96048647e-01 -5.19259274e-01 5.78207709e-02
-9.37666744e-02 4.59857255e-01 -9.79001164e-01 7.24162042e-01
7.27581441e-01 8.41538131e-01 -4.87055898e-01 6.51835382e-01
-4.99776721e-01 7.04270124e-01 -3.47263068e-01 4.41986203e-01
-2.79168710e-02 5.00844657e-01 5.68864346e-01 1.12030160e+00
-4.14098985e-02 3.82617682e-01 5.28585203e-02 1.10147536e+00
-1.00015007e-01 -3.68981153e-01 -7.58811772e-01 1.86211735e-01
3.46312881e-01 5.70681632e-01 -5.32221556e-01 -5.43903291e-01
-3.75872225e-01 7.46895730e-01 8.75443518e-02 4.10170525e-01
-8.00669551e-01 9.88470614e-02 8.75529170e-01 2.45816469e-01
3.05957407e-01 6.22936413e-02 -6.07058108e-01 -1.04771984e+00
6.65763859e-04 -7.86799312e-01 8.84986341e-01 -8.54265511e-01
-1.83281815e+00 1.76195830e-01 6.89227402e-01 -1.19364643e+00
-3.55224162e-01 -8.88618171e-01 -1.20496601e-02 8.40013206e-01
-1.10076308e+00 -1.12466979e+00 -2.78958499e-01 8.74123156e-01
2.09740147e-01 3.97578329e-02 9.66632068e-01 1.11280963e-01
-8.49686339e-02 6.34234130e-01 -1.38021618e-01 -6.70399591e-02
8.64049077e-01 -1.22461843e+00 8.91513228e-02 7.22541988e-01
2.82145709e-01 8.67096484e-01 8.79700541e-01 -9.17004168e-01
-1.01120567e+00 -1.31852615e+00 5.43581426e-01 -8.82969022e-01
5.44816613e-01 -3.45064819e-01 -1.16126609e+00 1.24243379e+00
-3.67906064e-01 2.48385891e-01 5.39657116e-01 5.17873764e-01
-8.35621715e-01 8.22368339e-02 -1.32709324e+00 3.90791386e-01
1.54406834e+00 -6.34885967e-01 -1.27503574e+00 7.02028990e-01
8.16517472e-01 -4.08589274e-01 -7.28101611e-01 6.12095833e-01
3.95963550e-01 -5.54037333e-01 1.11181760e+00 -1.54319346e+00
5.38353264e-01 -3.58426750e-01 -7.66694427e-01 -1.27870631e+00
-4.98026431e-01 1.04458489e-01 2.28815116e-02 8.91707599e-01
4.08941597e-01 -6.36457503e-01 6.57946885e-01 6.95549071e-01
-2.24894688e-01 -2.56218821e-01 -1.05873287e+00 -1.35115778e+00
3.19456965e-01 -6.45362556e-01 7.09459662e-01 1.33227062e+00
-2.72419184e-01 3.06327343e-01 -1.60229534e-01 6.88543081e-01
7.00571895e-01 2.40488321e-01 6.47898793e-01 -1.44658983e+00
-5.54626107e-01 -7.24263713e-02 -6.87108338e-01 -9.78109002e-01
4.32268977e-01 -1.17739427e+00 -3.13425571e-01 -1.29175925e+00
4.64421868e-01 -4.48981494e-01 -3.40196669e-01 7.91905046e-01
8.76644626e-02 -1.42786652e-02 -7.90554583e-02 3.29454213e-01
-5.36467433e-02 3.54441524e-01 1.05666530e+00 -9.44571048e-02
-1.20122120e-01 -4.02363651e-02 -9.63353515e-01 7.03467309e-01
4.83296037e-01 -9.40542817e-01 -5.21949947e-01 -1.52557597e-01
9.31248292e-02 -1.90931335e-01 9.94999051e-01 -6.87713921e-01
3.63918804e-02 -3.43294591e-01 8.29242945e-01 -5.03936112e-01
4.28926617e-01 -9.27151144e-01 2.36707792e-01 4.47751224e-01
-7.03271270e-01 -2.46711776e-01 2.85534114e-01 9.04363275e-01
6.89068809e-02 -2.30037287e-01 9.14590895e-01 -8.06828961e-02
-7.92943120e-01 9.90828127e-02 -7.55761564e-02 2.30928123e-01
9.78126645e-01 -1.18941255e-01 -4.52034742e-01 -2.11152554e-01
-1.20614552e+00 -1.45467252e-01 2.32649565e-01 4.80070531e-01
4.94251132e-01 -1.33782339e+00 -8.01284015e-01 1.85260892e-01
5.42736888e-01 5.03792539e-02 1.03236981e-01 6.37084723e-01
4.36120182e-01 4.32291538e-01 -1.23514436e-01 -1.00675929e+00
-9.68641102e-01 8.32672775e-01 3.99989933e-01 1.41021132e-01
-6.96946740e-01 9.79216814e-01 1.12990808e+00 -4.77149934e-01
4.75454740e-02 -5.23075819e-01 3.56954515e-01 -2.00407561e-02
2.67511100e-01 6.25074729e-02 3.17693621e-01 -3.43348205e-01
-5.26085079e-01 2.12198913e-01 -4.01137114e-01 -1.53729647e-01
1.41112745e+00 2.69027174e-01 2.78394133e-01 4.38271731e-01
9.62272942e-01 -3.58225048e-01 -1.05594337e+00 -6.04042113e-01
1.20901195e-02 -6.80580199e-01 -1.98721424e-01 -1.01345789e+00
-6.66615784e-01 5.23291886e-01 7.23066151e-01 -7.22645745e-02
1.24137259e+00 8.60289216e-01 -2.11846501e-01 4.02911216e-01
3.12417805e-01 -7.34745324e-01 2.65963256e-01 1.01665549e-01
1.33304644e+00 -9.87448275e-01 3.17135632e-01 -4.35884953e-01
-5.81151307e-01 8.34103882e-01 6.68107867e-01 -4.42341566e-01
5.97819924e-01 -1.39603972e-01 -5.26772857e-01 -4.17314023e-01
-7.66431272e-01 -4.86601368e-02 4.67625797e-01 7.74437964e-01
8.86988491e-02 3.48537922e-01 -1.12809949e-01 9.30081904e-01
-2.17171103e-01 -1.27452999e-01 4.21685427e-01 7.89193749e-01
-1.59783691e-01 -8.82608056e-01 -5.05209446e-01 7.66631246e-01
-2.63865769e-01 8.60476270e-02 -5.44615090e-01 1.36720037e+00
1.25124201e-01 6.32055163e-01 1.69111326e-01 -3.21589500e-01
7.32505918e-01 4.89929169e-02 8.66607547e-01 -9.61786509e-01
-1.48977250e-01 -2.70475090e-01 8.57971236e-02 -4.85325813e-01
-4.51058805e-01 -1.04782271e+00 -1.33101976e+00 1.71367042e-02
-3.32364053e-01 -2.85127699e-01 3.39653105e-01 1.09085357e+00
1.98151454e-01 6.26317859e-01 3.68014276e-01 -3.57461363e-01
-8.60556006e-01 -8.89820218e-01 -6.73932493e-01 9.01393056e-01
4.51866090e-01 -1.31437254e+00 -6.11855745e-01 3.13550562e-01] | [10.253175735473633, 2.918940544128418] |
51487985-7790-41b3-94be-d78af1cd18ab | sst-single-stream-temporal-action-proposals | null | null | http://openaccess.thecvf.com/content_cvpr_2017/html/Buch_SST_Single-Stream_Temporal_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Buch_SST_Single-Stream_Temporal_CVPR_2017_paper.pdf | SST: Single-Stream Temporal Action Proposals | Our paper presents a new approach for temporal detection of human actions in long, untrimmed video sequences. We introduce Single-Stream Temporal Action Proposals (SST), a new effective and efficient deep architecture for the generation of temporal action proposals. Our network can run continuously in a single stream over very long input video sequences, without the need to divide input into short overlapping clips or temporal windows for batch processing. We demonstrate empirically that our model outperforms the state-of-the-art on the task of temporal action proposal generation, while achieving some of the fastest processing speeds in the literature. Finally, we demonstrate that using SST proposals in conjunction with existing action classifiers results in improved state-of-the-art temporal action detection performance.
| ['Juan Carlos Niebles', 'Victor Escorcia', 'Chuanqi Shen', 'Shyamal Buch', 'Bernard Ghanem'] | 2017-07-01 | null | null | null | cvpr-2017-7 | ['temporal-action-proposal-generation'] | ['computer-vision'] | [ 4.03206378e-01 -2.98988819e-01 -3.40665877e-01 -1.35211870e-01
-6.21491730e-01 -3.54298413e-01 8.12067389e-01 -3.82993639e-01
-6.33466363e-01 4.49450821e-01 5.08721948e-01 -5.17678121e-03
1.68545470e-01 -2.50905693e-01 -4.95347559e-01 -5.03662646e-01
-6.50721967e-01 6.29831403e-02 1.04685414e+00 -9.91622172e-03
2.18643159e-01 5.39006352e-01 -1.61002004e+00 8.88452590e-01
1.01722851e-02 1.10172069e+00 -1.48084953e-01 1.17205393e+00
2.11556539e-01 1.31044042e+00 -6.49213135e-01 -1.84588969e-01
6.25417709e-01 -6.93538487e-01 -8.49146962e-01 1.40916944e-01
7.88670003e-01 -9.25785065e-01 -8.80755663e-01 5.98594129e-01
3.16452742e-01 5.25825381e-01 2.81667769e-01 -1.36157703e+00
-1.23960651e-01 6.46092474e-01 -3.67746264e-01 8.32515538e-01
6.65735066e-01 6.29432082e-01 8.12318623e-01 -8.86625528e-01
9.15973067e-01 1.25951600e+00 7.37599790e-01 6.18110240e-01
-9.06700313e-01 -5.73389530e-01 4.12660390e-01 5.64895451e-01
-1.07956100e+00 -6.93989277e-01 3.10841382e-01 -3.62257153e-01
1.61147022e+00 -5.75719029e-03 9.69368398e-01 1.28777063e+00
4.50030804e-01 1.43904901e+00 3.69330585e-01 -1.44086704e-01
1.96401343e-01 -8.73111904e-01 -1.53107271e-01 5.15181899e-01
-3.96546453e-01 3.81766558e-01 -8.86665642e-01 -1.02843512e-02
1.09852934e+00 -6.36024177e-02 7.15968832e-02 -2.63780449e-02
-1.54996860e+00 4.67392594e-01 5.70452735e-02 4.48006332e-01
-7.60519624e-01 6.11072600e-01 1.00405800e+00 3.89114529e-01
3.45595896e-01 3.12892318e-01 -4.51712221e-01 -8.65796864e-01
-1.30591333e+00 4.85618681e-01 2.54223526e-01 9.03535783e-01
-1.79107800e-01 4.07436043e-01 -6.31170869e-01 2.66690522e-01
-1.99199080e-01 1.17424875e-01 5.44491231e-01 -1.64133990e+00
3.83453965e-01 1.33055523e-01 4.21505958e-01 -6.37175083e-01
-3.95424813e-01 2.00389251e-01 -4.25689429e-01 1.76852033e-01
5.26366651e-01 -1.83600277e-01 -9.68356490e-01 1.55484855e+00
2.22614706e-01 5.33382595e-01 -4.76796962e-02 9.48586524e-01
4.78926122e-01 7.40707994e-01 3.38044196e-01 -4.80303943e-01
1.02651978e+00 -1.38322818e+00 -6.21659756e-01 -1.08789943e-01
5.15645146e-01 -5.99545777e-01 5.81332922e-01 7.14662731e-01
-1.39604974e+00 -9.05788720e-01 -8.77692163e-01 2.65840460e-02
7.98782632e-02 3.25700283e-01 8.07486892e-01 1.55293584e-01
-1.21441388e+00 9.71932888e-01 -1.32325399e+00 -5.46923161e-01
5.83055317e-01 2.64030725e-01 -4.49093282e-01 1.70681238e-01
-9.52118039e-01 6.44322336e-01 4.62834835e-01 -9.56718549e-02
-1.20007610e+00 -4.17163432e-01 -7.25179195e-01 -7.55490214e-02
6.25537753e-01 -4.71180797e-01 2.01741862e+00 -1.46949196e+00
-1.58205497e+00 4.83980119e-01 -2.72925407e-01 -1.03369784e+00
9.23049927e-01 -6.24609172e-01 -4.47115660e-01 8.66096497e-01
1.02510542e-01 1.15338588e+00 1.02783537e+00 -1.86991721e-01
-1.00992930e+00 1.42539978e-01 7.86249712e-02 6.49132952e-02
-2.24486478e-02 7.47243822e-01 -5.35088181e-01 -9.82210755e-01
-2.07593039e-01 -8.59292805e-01 -4.96437281e-01 2.73587048e-01
7.60086477e-02 -6.00043595e-01 8.01943600e-01 -5.27262211e-01
1.28378332e+00 -2.20670128e+00 1.79074947e-02 -4.24691588e-01
-6.79883584e-02 4.51752961e-01 -4.31891680e-01 3.90541375e-01
-1.58161342e-01 -1.43572628e-01 1.82931334e-01 -2.92512029e-01
-1.10281207e-01 -2.33243499e-02 -3.56771976e-01 4.02053952e-01
3.26601863e-01 9.23268616e-01 -1.12668633e+00 -6.37756169e-01
4.55018938e-01 2.35919073e-01 -3.73235613e-01 3.00236851e-01
-2.93801099e-01 2.08377302e-01 -3.30683172e-01 5.13458431e-01
1.13157041e-01 -9.72319487e-03 2.13043779e-01 9.65176672e-02
-1.50389791e-01 5.17142117e-01 -1.07489789e+00 1.75753498e+00
2.18071759e-01 9.71765757e-01 -2.42246762e-01 -7.88639009e-01
4.40744877e-01 8.10687304e-01 1.07327950e+00 -5.87960362e-01
2.11273134e-02 -1.07587546e-01 5.85844852e-02 -5.90079963e-01
7.11860955e-01 -4.79138680e-02 1.22821033e-01 8.89757514e-01
1.98837698e-01 3.96576881e-01 7.30664551e-01 2.13012457e-01
1.67450643e+00 6.12505198e-01 4.83798563e-01 2.72430778e-01
3.28384310e-01 3.09058391e-02 5.97708404e-01 8.99191558e-01
-7.94036269e-01 6.13415658e-01 4.67293441e-01 -1.05249703e+00
-1.19034970e+00 -1.02076638e+00 5.46808183e-01 1.58377242e+00
-2.82959849e-01 -5.07178009e-01 -5.46674490e-01 -9.04058695e-01
-2.69801706e-01 2.44204193e-01 -5.93274713e-01 1.49645284e-01
-1.17813253e+00 -1.29245654e-01 6.81348205e-01 1.25740981e+00
5.26145339e-01 -1.74644470e+00 -1.30557156e+00 5.90386212e-01
-1.44715086e-01 -1.35957050e+00 -6.80752754e-01 -1.30283073e-01
-1.10517776e+00 -1.24375045e+00 -7.62478828e-01 -4.88712728e-01
2.71596700e-01 4.60148722e-01 1.05911648e+00 -6.94153830e-02
-4.19397950e-01 3.87887836e-01 -6.56936049e-01 -6.29183650e-02
-3.76142979e-01 -2.92557329e-01 2.22034529e-01 -1.40568584e-01
5.79745829e-01 -3.15588355e-01 -7.56685793e-01 4.88508224e-01
-7.98421264e-01 1.30544767e-01 4.29530740e-01 4.66697276e-01
3.27857554e-01 -7.64422268e-02 2.98704773e-01 -7.64267370e-02
3.14350277e-01 -1.39530078e-01 -4.26838726e-01 1.99243822e-03
1.09521650e-01 -8.25978890e-02 5.70157111e-01 -8.65119219e-01
-1.13897359e+00 4.69876200e-01 -2.13479903e-02 -8.02484691e-01
-3.91398519e-01 4.44794297e-02 6.43468559e-01 1.62866309e-01
7.04889894e-01 3.70526463e-01 -1.17459841e-01 -3.50413233e-01
3.94005954e-01 1.32818639e-01 7.75391459e-01 -2.39967659e-01
2.08993644e-01 7.16307163e-01 -1.48663625e-01 -5.83937585e-01
-5.66664398e-01 -7.12304115e-01 -1.10662425e+00 -6.09097481e-01
1.00265384e+00 -8.15617442e-01 -5.49719214e-01 8.19633603e-01
-1.22420955e+00 -7.16089070e-01 -3.74497116e-01 5.66261530e-01
-9.98020470e-01 4.97477770e-01 -1.17614913e+00 -8.77358019e-01
-2.84818232e-01 -8.24183941e-01 1.10751247e+00 -3.24751735e-02
-6.66616678e-01 -4.59535956e-01 4.83855233e-02 2.09088461e-03
2.11207852e-01 1.09103516e-01 7.64552969e-03 -5.38203776e-01
-7.27372646e-01 -3.49389613e-01 7.53772408e-02 2.51433372e-01
-5.99026680e-02 4.49477434e-01 -5.23229241e-01 -2.39374399e-01
-2.77749866e-01 -4.97503698e-01 1.09350157e+00 6.52119815e-01
1.02171266e+00 -2.22124875e-01 -2.36296982e-01 2.38088399e-01
8.01887214e-01 5.25353312e-01 7.37627864e-01 2.29542553e-01
2.52815008e-01 3.46516430e-01 1.10742319e+00 8.34431529e-01
-1.99099839e-01 7.82301605e-01 1.34944662e-01 1.36836633e-01
-9.31111947e-02 -1.24740586e-01 8.47678006e-01 2.39486024e-01
-5.47246337e-01 -2.45256349e-01 -6.40874267e-01 7.51478553e-01
-2.17142463e+00 -1.72727203e+00 1.22130059e-01 1.92599714e+00
4.50588435e-01 4.52997863e-01 7.96053529e-01 2.11760283e-01
5.23223460e-01 5.14597416e-01 -4.29474682e-01 -3.95201355e-01
1.87862605e-01 3.17102492e-01 3.30171496e-01 1.51861265e-01
-1.81509936e+00 1.19542336e+00 8.11265850e+00 5.63634753e-01
-1.00040627e+00 8.11247155e-02 3.01876724e-01 -6.60481691e-01
6.38338029e-01 -3.34694803e-01 -5.86782634e-01 1.97372049e-01
1.11204934e+00 -1.83132082e-01 1.10061184e-01 8.80219102e-01
7.14542985e-01 -3.20985138e-01 -1.35481191e+00 9.18755829e-01
1.26202136e-01 -1.37281609e+00 -6.16631582e-02 -1.68854296e-01
6.90140069e-01 1.74703807e-01 -3.01691145e-01 2.51867443e-01
4.42191243e-01 -6.89481974e-01 1.01891625e+00 4.76817042e-01
4.91501987e-01 -4.57475573e-01 5.70772111e-01 9.63340551e-02
-1.58129740e+00 -4.96511459e-01 -1.83693394e-01 -3.93309295e-01
6.73364460e-01 -2.36934721e-02 -5.51447034e-01 5.48104718e-02
7.70030737e-01 1.21464121e+00 -4.18998092e-01 1.29951084e+00
-2.03969598e-01 5.86997688e-01 -2.20434606e-01 4.45354059e-02
6.69548452e-01 3.54741752e-01 5.49551964e-01 1.46210814e+00
3.35727215e-01 6.54526532e-01 5.08053839e-01 2.32872158e-01
2.10489124e-01 -1.85788542e-01 -6.49044275e-01 -3.76127958e-01
1.37225851e-01 7.79269814e-01 -9.30181563e-01 -1.00156391e+00
-4.53259140e-01 1.09137058e+00 -6.24527484e-02 2.73787111e-01
-1.01383042e+00 -1.02086082e-01 7.91585267e-01 3.72094065e-02
7.63138235e-01 -5.28438747e-01 2.06020281e-01 -9.65189219e-01
1.72069862e-01 -9.68218327e-01 8.17102909e-01 -8.34315002e-01
-9.79360104e-01 3.81507039e-01 1.77521080e-01 -1.75265002e+00
-6.97039664e-01 -5.80134094e-01 -6.68908894e-01 1.40523300e-01
-6.59562171e-01 -7.93299556e-01 -7.05205426e-02 6.18921757e-01
1.28362501e+00 -1.42213047e-01 4.95588809e-01 1.86506182e-01
-3.14331919e-01 3.18313897e-01 -3.78027260e-01 4.08123553e-01
5.95497787e-01 -8.86228204e-01 1.04446971e+00 1.19631898e+00
3.60613555e-01 2.35233098e-01 7.25400448e-01 -6.88005090e-01
-1.07562637e+00 -8.61069977e-01 7.93972313e-01 -4.53034073e-01
7.74698615e-01 5.11300890e-03 -7.13498950e-01 7.93324709e-01
2.67315507e-01 1.73347577e-01 3.25221956e-01 -2.81447530e-01
-2.63247788e-01 7.87082016e-02 -7.41418064e-01 6.69821322e-01
1.39373875e+00 -5.15129507e-01 -7.45251834e-01 4.28925782e-01
6.80443585e-01 -3.36050510e-01 -5.94437480e-01 3.33814740e-01
9.66433644e-01 -1.16802669e+00 9.47690725e-01 -1.08818984e+00
6.44780338e-01 -2.31554672e-01 1.32078201e-01 -6.85677528e-01
-5.88251710e-01 -9.07382250e-01 -4.25634086e-01 4.09040183e-01
8.16032961e-02 2.53623575e-02 1.02866960e+00 4.86911297e-01
-2.63440847e-01 -4.97262657e-01 -1.08916593e+00 -9.75996196e-01
-4.37460095e-01 -8.79509985e-01 1.47417426e-01 4.98173058e-01
3.32604975e-01 -1.85295418e-01 -6.58319652e-01 -3.21686625e-01
2.41524056e-01 -9.92746577e-02 7.92960703e-01 -6.14046693e-01
-4.34521228e-01 -7.00085759e-01 -7.22186148e-01 -1.38145423e+00
-3.92399309e-03 -1.30324692e-01 3.22709948e-01 -1.32043040e+00
9.84935611e-02 3.96594077e-01 -5.03031075e-01 7.50477016e-01
1.77445598e-02 4.39414680e-01 4.95783418e-01 1.20551109e-01
-1.25521970e+00 5.62803090e-01 1.02495396e+00 -5.85008748e-02
-2.81798512e-01 1.35848299e-01 2.15955347e-01 7.46382236e-01
5.54300725e-01 -4.52591151e-01 -4.03094918e-01 -2.21597254e-01
-2.98419803e-01 3.04542482e-01 2.42256165e-01 -1.47053921e+00
2.17742816e-01 -4.53835040e-01 3.80684942e-01 -8.91848266e-01
4.59397078e-01 -4.16351825e-01 -1.20354265e-01 5.92983842e-01
-6.37196720e-01 3.87495011e-01 2.43559197e-01 4.76008177e-01
-2.60034978e-01 2.12692156e-01 6.36311114e-01 -4.42112952e-01
-1.27909970e+00 3.28492582e-01 -1.09137273e+00 -6.59625679e-02
1.18683589e+00 -4.37851459e-01 -2.33614489e-01 -4.22305614e-01
-1.04127562e+00 8.23627785e-02 8.98084715e-02 8.14081550e-01
7.53971517e-01 -1.42622316e+00 -6.35755897e-01 -9.16428268e-02
-6.02460504e-02 -6.94822609e-01 2.87848562e-01 9.77420807e-01
-5.79583645e-01 6.53923213e-01 -5.74037313e-01 -6.03542328e-01
-1.70955420e+00 7.67385960e-01 3.64887208e-01 -2.49399781e-01
-1.04692757e+00 9.23536360e-01 -6.99770302e-02 3.24914992e-01
5.02674401e-01 -4.98525321e-01 -9.93787423e-02 -1.80056822e-02
1.02601874e+00 4.88921136e-01 -4.35129732e-01 -5.09433746e-01
-3.16980034e-01 1.03445731e-01 2.05222983e-03 -3.90687734e-01
1.17384231e+00 2.04577312e-01 2.63779849e-01 4.63337511e-01
8.02218556e-01 -7.26452470e-01 -1.82940555e+00 -7.45516866e-02
6.86881095e-02 -7.63239443e-01 -3.37498337e-01 -4.80460823e-01
-9.16232646e-01 7.62080848e-01 5.73217332e-01 3.18990387e-02
1.21230161e+00 -4.17153761e-02 1.01875341e+00 5.88847458e-01
4.95772600e-01 -1.45929694e+00 6.50236547e-01 6.63654506e-01
8.78731191e-01 -1.14794779e+00 1.29176095e-01 -1.76183656e-01
-7.80687392e-01 1.34024334e+00 7.96746016e-01 -2.15439409e-01
4.00289446e-01 1.91460341e-01 1.16816469e-01 -1.68718293e-01
-1.37239087e+00 -1.57549426e-01 2.40507692e-01 3.88566762e-01
5.13525605e-01 2.53542848e-02 -3.12394321e-01 9.17931795e-02
2.35851035e-01 3.88706684e-01 5.38152635e-01 1.25705099e+00
-5.67570865e-01 -7.93598294e-01 -1.42033651e-01 4.22257781e-01
-5.78927636e-01 -1.17516518e-02 -5.61749041e-01 7.39144802e-01
2.18691537e-03 7.87213683e-01 3.42377096e-01 -4.39716756e-01
2.79540598e-01 2.84097016e-01 5.82021832e-01 -4.21632618e-01
-7.44806767e-01 2.06667855e-01 4.09229875e-01 -1.39069581e+00
-7.85275638e-01 -1.09461105e+00 -1.20645928e+00 -1.21578053e-01
2.36861050e-01 -2.80050725e-01 -5.02510443e-02 1.08406711e+00
2.81782657e-01 3.24905038e-01 2.92733788e-01 -1.31566620e+00
-3.46876234e-01 -1.15337026e+00 -2.99042523e-01 4.84246910e-01
4.04048979e-01 -5.25784671e-01 -1.40775396e-02 4.22357857e-01] | [8.27669906616211, 0.4419221580028534] |
e82412c9-19b7-4a62-aaa9-b2362be62bf3 | a-weakly-supervised-surface-crack | 2109.00456 | null | https://arxiv.org/abs/2109.00456v3 | https://arxiv.org/pdf/2109.00456v3.pdf | Weakly-Supervised Surface Crack Segmentation by Generating Pseudo-Labels using Localization with a Classifier and Thresholding | Surface cracks are a common sight on public infrastructure nowadays. Recent work has been addressing this problem by supporting structural maintenance measures using machine learning methods. Those methods are used to segment surface cracks from their background, making them easier to localize. However, a common issue is that to create a well-functioning algorithm, the training data needs to have detailed annotations of pixels that belong to cracks. Our work proposes a weakly supervised approach that leverages a CNN classifier in a novel way to create surface crack pseudo labels. First, we use the classifier to create a rough crack localization map by using its class activation maps and a patch based classification approach and fuse this with a thresholding based approach to segment the mostly darker crack pixels. The classifier assists in suppressing noise from the background regions, which commonly are incorrectly highlighted as cracks by standard thresholding methods. Then, the pseudo labels can be used in an end-to-end approach when training a standard CNN for surface crack segmentation. Our method is shown to yield sufficiently accurate pseudo labels. Those labels, incorporated into segmentation CNN training using multiple recent crack segmentation architectures, achieve comparable performance to fully supervised methods on four popular crack segmentation datasets. | ['Gordon Morison', 'Peter Barrie', 'Mike Mannion', 'Mark Jenkins', 'Jacob König'] | 2021-09-01 | null | null | null | null | ['crack-segmentation'] | ['computer-vision'] | [ 7.44864821e-01 3.88230532e-01 -9.96319205e-03 -2.92793870e-01
-1.18583620e+00 -4.32627171e-01 6.58531487e-03 4.00265902e-01
-1.57590270e-01 3.91211659e-01 -2.06788450e-01 -1.94013238e-01
5.15557885e-01 -1.00223172e+00 -8.58326793e-01 -8.73697937e-01
4.29710597e-01 3.35084319e-01 7.63356864e-01 -2.01652393e-01
8.15991879e-01 3.51208210e-01 -1.67928553e+00 5.79247952e-01
7.10388660e-01 1.10969996e+00 -2.85686627e-02 7.36818373e-01
-1.19909076e-02 6.65901065e-01 -4.38872933e-01 2.44013101e-01
1.75418884e-01 -6.33556619e-02 -9.92304862e-01 1.41420811e-01
7.44657159e-01 -2.05579937e-01 3.08136135e-01 7.84817040e-01
3.69779080e-01 -2.56439328e-01 7.31520712e-01 -7.96741188e-01
-1.45363256e-01 4.26611274e-01 -7.11956501e-01 6.62675798e-02
3.42699081e-01 2.13559523e-01 1.18050575e+00 -1.18293726e+00
4.12155539e-01 8.60019624e-01 1.23962915e+00 3.43630284e-01
-1.31643546e+00 -5.48629999e-01 -1.92871377e-01 -5.32290637e-02
-1.21015012e+00 -3.33565474e-01 1.18449783e+00 -7.50931025e-01
8.30549955e-01 8.60851184e-02 5.07863164e-01 5.88189125e-01
-1.32066369e-01 5.71959257e-01 1.00461340e+00 -8.62439930e-01
4.55126077e-01 -1.06510893e-02 4.78827685e-01 9.07083750e-01
4.57219072e-02 -1.32036775e-01 -1.90213710e-01 4.71396968e-02
4.78191078e-01 -2.48936359e-02 -2.71596193e-01 -1.73312053e-01
-8.12210500e-01 1.03499663e+00 6.12814426e-01 3.97379756e-01
-3.27710986e-01 4.34493989e-01 2.10243776e-01 6.15899600e-02
6.50382161e-01 3.92464638e-01 -2.35327750e-01 3.37425113e-01
-1.36035204e+00 1.90707669e-01 6.21310949e-01 3.73522937e-01
1.47560000e+00 -2.50097483e-01 -7.17588887e-02 8.77378702e-01
4.82756644e-01 3.77283722e-01 -1.78654537e-01 -1.32877636e+00
2.98503399e-01 1.04333687e+00 1.12617565e-02 -9.80106056e-01
-5.64816833e-01 -3.11483562e-01 -3.57692748e-01 7.43837595e-01
7.28419363e-01 -3.87784801e-02 -1.17597783e+00 1.04349840e+00
3.54804993e-01 6.53945655e-02 -3.31732780e-01 5.84474206e-01
4.10906732e-01 4.04202461e-01 -1.10331312e-01 3.87012720e-01
1.20396256e+00 -1.05698407e+00 -3.11215580e-01 -3.55153829e-01
8.92100096e-01 -6.84963286e-01 1.32304561e+00 7.12699652e-01
-9.16124523e-01 -3.15686554e-01 -1.38467252e+00 1.52193485e-02
-4.76993889e-01 2.50782460e-01 -8.54285620e-03 9.36778247e-01
-1.21031547e+00 6.43488884e-01 -7.19615340e-01 -3.34823608e-01
8.85964036e-01 2.37545639e-01 8.37143809e-02 -1.26719028e-01
-7.81750143e-01 7.70303249e-01 3.15901667e-01 1.12467311e-01
-8.93929362e-01 -5.27892470e-01 -8.74641895e-01 -1.70733601e-01
3.25336695e-01 -7.30612734e-03 1.12421942e+00 -8.62580776e-01
-1.03778970e+00 9.30250764e-01 -1.14240581e-02 -2.41097331e-01
1.02346957e-01 -9.01339576e-02 2.14021340e-01 6.67147815e-01
2.99316287e-01 6.54040754e-01 1.29978538e+00 -1.85191154e+00
-7.44261622e-01 -9.60891172e-02 -4.41914201e-02 -2.27191418e-01
-1.07774079e-01 -1.99048758e-01 -1.21695928e-01 -5.10878325e-01
4.21412081e-01 -7.46966660e-01 -2.52381474e-01 1.86492652e-01
-7.47230470e-01 -2.31354281e-01 1.30865812e+00 -7.21828520e-01
1.01532185e+00 -1.87641037e+00 -3.11047941e-01 7.14810073e-01
3.39388460e-01 -3.89197618e-02 2.34522179e-01 3.04155946e-01
-1.18395582e-01 2.36728936e-01 -1.19000804e+00 -6.21517777e-01
-3.24741215e-01 2.31326431e-01 -5.69129512e-02 6.83211625e-01
6.20822072e-01 5.85447192e-01 -7.26628959e-01 -5.85993886e-01
1.34252608e-01 2.94831485e-01 -5.49356461e-01 1.63838491e-02
-2.49952525e-01 4.42415088e-01 -7.40612149e-02 1.20490336e+00
5.90513170e-01 -2.13730056e-02 -5.09332895e-01 -2.70530343e-01
-2.59299964e-01 -8.78729224e-02 -1.04996312e+00 1.60953605e+00
-3.41350436e-01 7.30989397e-01 2.98189223e-01 -1.37794483e+00
8.12237203e-01 3.80931139e-01 4.15892571e-01 -2.13042825e-01
3.41478676e-01 5.21444440e-01 -6.33574009e-01 -7.44827509e-01
2.97665894e-01 -1.65584147e-01 6.50881827e-02 7.64239073e-01
-3.23915064e-01 -6.45682037e-01 8.05777907e-02 3.87967825e-02
1.29322529e+00 2.22961023e-01 -5.51950574e-01 -3.44255000e-01
4.83580559e-01 5.85473597e-01 2.23197311e-01 6.46364868e-01
-1.03802845e-01 1.27910531e+00 2.76554257e-01 -5.18242776e-01
-1.05555689e+00 -6.48406982e-01 -9.70973969e-02 9.87538993e-01
4.44964170e-02 2.95684189e-01 -1.31761575e+00 -8.82513642e-01
-1.43733680e-01 3.21305752e-01 -1.01017046e+00 -1.05506945e-02
-8.02781105e-01 -7.30394244e-01 8.00694525e-01 6.75274074e-01
3.16157103e-01 -1.15669250e+00 -9.54268873e-01 3.11169624e-01
-2.37540171e-01 -7.78786123e-01 -3.05676848e-01 6.21312499e-01
-7.85318136e-01 -1.44830322e+00 -7.83616662e-01 -1.05522370e+00
9.76963222e-01 2.88894892e-01 9.53330457e-01 9.99596596e-01
-5.96035659e-01 3.70497912e-01 -6.22187018e-01 -1.18508331e-01
-4.40587759e-01 2.01815233e-01 -5.47771215e-01 2.44885549e-01
-9.28831752e-03 -3.32957149e-01 -8.39362264e-01 1.47067115e-01
-9.54849362e-01 -2.03123555e-01 5.57011008e-01 7.95372427e-01
4.73258048e-01 2.43371502e-01 4.18476671e-01 -1.11471188e+00
3.46426427e-01 -4.67096031e-01 -1.26665741e-01 -7.81248212e-02
-8.61418486e-01 -2.40149304e-01 1.36154801e-01 -1.80584520e-01
-8.84173810e-01 4.50905800e-01 -4.66548890e-01 -6.87109306e-02
-5.32541454e-01 4.91351962e-01 3.22451800e-01 -3.60284597e-01
9.10190105e-01 -1.20497301e-01 8.91265124e-02 -3.56850117e-01
3.15009952e-01 8.22813570e-01 6.36866927e-01 -6.93209529e-01
1.06475222e+00 9.56750929e-01 -1.47205010e-01 -9.28093910e-01
-1.13860452e+00 -8.54523480e-01 -8.71561170e-01 -5.55323124e-01
1.17600024e+00 -6.18976951e-01 -7.01557249e-02 7.92403936e-01
-1.17647231e+00 -7.26329446e-01 -3.21260154e-01 -2.53384769e-01
-2.43437737e-01 5.93929470e-01 -6.24071181e-01 -9.47276175e-01
-3.12802106e-01 -1.24138939e+00 1.40365028e+00 3.25234205e-01
2.64974590e-03 -8.75391126e-01 1.46685883e-01 7.10243404e-01
4.82139498e-01 5.70505261e-01 9.43933606e-01 -3.13286632e-01
-3.55148911e-01 -4.52791005e-01 -4.30552840e-01 6.38604641e-01
-1.00250259e-01 3.30327570e-01 -1.41379833e+00 -1.09143525e-01
-1.45392299e-01 -3.75809968e-01 1.14401102e+00 2.89041609e-01
8.68970692e-01 1.97782889e-01 -4.96883601e-01 -1.40140690e-02
1.55813682e+00 -1.97922856e-01 7.44218707e-01 3.20369035e-01
8.66598725e-01 9.12063479e-01 4.94596303e-01 1.10925138e-02
4.02451217e-01 3.54219317e-01 8.81927013e-01 -5.53052068e-01
-2.50771940e-01 2.46060655e-01 1.06421091e-01 2.80383974e-01
1.92926843e-02 2.10144833e-01 -1.39243126e+00 8.23943973e-01
-1.58923972e+00 -8.24969411e-01 -8.42458487e-01 1.82938325e+00
8.89792562e-01 5.22201359e-01 1.04499839e-01 1.01564157e+00
7.89014876e-01 -2.14026541e-01 -2.81073540e-01 -1.87853366e-01
1.83605049e-02 4.92573082e-01 5.39252579e-01 3.81537825e-01
-1.32704699e+00 7.19441116e-01 6.01107883e+00 6.25352740e-01
-1.07572854e+00 3.89060676e-01 7.91241467e-01 5.64228117e-01
-1.30681247e-01 2.88211703e-01 -3.57866138e-01 1.36699662e-01
5.52971065e-01 1.27006721e+00 8.30171406e-02 7.55531967e-01
1.90752938e-01 -6.47663474e-01 -8.15231085e-01 5.84306359e-01
1.76978797e-01 -1.40728807e+00 -5.39435267e-01 -1.78925216e-01
8.39071691e-01 -6.31756848e-03 -1.96233571e-01 7.93986395e-02
-8.65691341e-03 -9.11395967e-01 1.06818557e+00 3.61677974e-01
6.64912105e-01 -4.63775039e-01 7.67404258e-01 3.01021099e-01
-1.19289601e+00 -3.10239166e-01 -4.91262823e-02 -1.06385313e-01
1.96776167e-01 8.70594501e-01 -7.35721827e-01 2.09932402e-01
1.08225226e+00 5.61824381e-01 -6.63753331e-01 1.08934426e+00
-2.32182980e-01 1.08742464e+00 -4.80501682e-01 5.97141743e-01
3.19802642e-01 1.03267133e-01 1.03086822e-01 1.38172483e+00
1.86438281e-02 -2.24317953e-01 3.88033986e-01 9.89037037e-01
1.16765469e-01 7.05121085e-02 -4.18576419e-01 3.82281870e-01
3.06730121e-01 1.32625222e+00 -1.48514187e+00 -2.91988283e-01
-3.84391516e-01 7.13082194e-01 4.87449504e-02 1.33521482e-01
-7.09295154e-01 -4.09815550e-01 -1.28007710e-01 5.73466361e-01
2.60199904e-01 -8.86564851e-02 -1.05722857e+00 -5.47823429e-01
-1.75950706e-01 -6.31493807e-01 1.91685930e-01 -7.41700530e-01
-1.13151312e+00 3.93733442e-01 -2.69629627e-01 -1.00409365e+00
5.11241674e-01 -4.84192580e-01 -1.16197014e+00 5.26212752e-01
-1.72950113e+00 -1.44806862e+00 -6.83271110e-01 3.60529453e-01
7.86238313e-01 4.83359039e-01 3.19904625e-01 4.58415180e-01
-7.10549235e-01 2.58153170e-01 -3.59349579e-01 3.44210118e-01
5.00739992e-01 -1.60985947e+00 6.19796328e-02 1.06230462e+00
-1.61622182e-01 2.07070127e-01 6.09110713e-01 -7.84933209e-01
-7.21053958e-01 -1.17535031e+00 5.06434619e-01 -5.24933457e-01
3.00503105e-01 -1.65550813e-01 -1.20808077e+00 3.00688326e-01
2.70324379e-01 7.25917220e-02 3.00286770e-01 -3.12269360e-01
-3.67018491e-01 2.56161034e-01 -1.26405394e+00 9.19940472e-02
4.09415305e-01 -5.86394072e-01 -4.75552768e-01 3.20687979e-01
4.16483343e-01 -2.71570802e-01 -5.73481023e-01 3.63262326e-01
4.01520699e-01 -9.95931923e-01 8.52287769e-01 2.00797126e-01
5.53689182e-01 -5.31925440e-01 -4.44283932e-02 -9.22890544e-01
2.11252078e-01 -3.46419603e-01 1.35650739e-01 1.05288231e+00
7.38995016e-01 -4.87553805e-01 1.15182519e+00 4.65915680e-01
-6.50694370e-01 -8.70262027e-01 -8.84606481e-01 -2.63800740e-01
1.93488657e-01 -6.63988054e-01 2.44060248e-01 8.45973194e-01
-3.85460377e-01 2.14359500e-02 1.30936876e-01 3.88102531e-01
8.20194542e-01 -1.80413369e-02 6.31312013e-01 -1.73118973e+00
2.26226404e-01 -5.67704022e-01 7.11598843e-02 -7.75146902e-01
-2.13695303e-01 -6.50142074e-01 7.94008970e-01 -1.64970124e+00
-1.15569890e-01 -1.04033792e+00 -7.06779882e-02 8.96847844e-01
-1.05927087e-01 1.04493999e+00 -3.80710542e-01 3.34165066e-01
-3.17451447e-01 -1.47420038e-02 8.72050107e-01 -4.71503288e-01
-1.13926351e-01 -1.19058285e-02 -3.73803556e-01 8.83002639e-01
9.93384182e-01 -9.33644235e-01 1.00221790e-01 -1.71089381e-01
3.53864133e-01 -3.57481539e-01 6.27607942e-01 -1.47043812e+00
2.09392503e-01 2.73015141e-01 -1.68032795e-02 -8.25130522e-01
2.04377726e-01 -9.05967057e-01 -2.98242003e-01 7.15913415e-01
-8.36165249e-02 -3.29504222e-01 -7.15692714e-02 6.04812086e-01
-5.74211366e-02 -7.48753726e-01 1.15023005e+00 -1.86737120e-01
-3.98348153e-01 -1.66561425e-01 -7.51626849e-01 -5.46446331e-02
1.06699955e+00 -8.51751029e-01 -1.04592651e-01 -1.93528816e-01
-8.14222157e-01 2.23732851e-02 7.78370917e-01 -6.80096224e-02
6.24960005e-01 -7.25983918e-01 -5.56515932e-01 1.85969934e-01
1.00140303e-01 4.74974453e-01 1.05356211e-02 1.10585332e+00
-1.10214591e+00 -3.17495644e-01 2.45210364e-01 -1.10750878e+00
-9.40293670e-01 1.22978926e-01 4.69469190e-01 -5.82343154e-02
-7.90634215e-01 9.45461690e-01 -3.54982466e-01 -2.95889050e-01
2.83979595e-01 -7.17861712e-01 -3.23096842e-01 3.31187516e-01
1.91289172e-01 4.10374939e-01 5.53620040e-01 -6.41611814e-01
-2.04807103e-01 1.00400448e+00 1.48975909e-01 9.34100896e-02
1.52328444e+00 -1.68866083e-01 -2.96918333e-01 2.59154469e-01
1.08348298e+00 -1.59530491e-02 -1.32648897e+00 -1.82353780e-01
4.38231468e-01 -5.19564115e-02 3.11836928e-01 -6.23769104e-01
-1.54392946e+00 1.02987957e+00 8.79269540e-01 5.85205078e-01
1.09534776e+00 8.72332081e-02 1.12057197e+00 2.23670408e-01
2.13604391e-01 -1.40614855e+00 4.98121858e-01 2.94637918e-01
5.07734895e-01 -1.43621981e+00 -1.51589304e-01 -5.17026901e-01
-2.93406963e-01 1.37314558e+00 4.09367383e-01 -4.49375272e-01
7.88871646e-01 4.51130599e-01 4.57569838e-01 -9.46649492e-01
-5.80247231e-02 -3.49445969e-01 -6.86976090e-02 5.78752220e-01
9.41638425e-02 -4.01103616e-01 4.35664989e-02 1.45388022e-01
4.58365619e-01 -4.01169091e-01 7.54168808e-01 1.40899396e+00
-1.11330116e+00 -1.09467399e+00 -8.60834479e-01 5.87489605e-01
-5.22008419e-01 1.63680241e-02 -3.14199716e-01 4.94201392e-01
5.84365189e-01 1.31985497e+00 -2.04831734e-01 -3.28540385e-01
1.52279004e-01 1.81734458e-01 1.26770794e-01 -1.07275140e+00
-9.02104855e-01 -1.20742559e-01 -1.08076617e-01 -5.04193783e-01
-5.50795615e-01 -6.15122557e-01 -1.55644035e+00 5.22348762e-01
-8.70497704e-01 -2.59197429e-02 8.18112671e-01 1.16216373e+00
-2.19384179e-01 4.83441085e-01 6.85387373e-01 -1.43342865e+00
-1.41445324e-01 -8.86768222e-01 -2.45493427e-01 3.70952129e-01
4.46992338e-01 -8.99774730e-01 -5.60505569e-01 4.76839483e-01] | [7.502861022949219, 1.554753065109253] |
e1a15c98-30f1-4a38-bda8-01972c8775bb | modality-invariant-visual-odometry-for | 2305.00348 | null | https://arxiv.org/abs/2305.00348v1 | https://arxiv.org/pdf/2305.00348v1.pdf | Modality-invariant Visual Odometry for Embodied Vision | Effectively localizing an agent in a realistic, noisy setting is crucial for many embodied vision tasks. Visual Odometry (VO) is a practical substitute for unreliable GPS and compass sensors, especially in indoor environments. While SLAM-based methods show a solid performance without large data requirements, they are less flexible and robust w.r.t. to noise and changes in the sensor suite compared to learning-based approaches. Recent deep VO models, however, limit themselves to a fixed set of input modalities, e.g., RGB and depth, while training on millions of samples. When sensors fail, sensor suites change, or modalities are intentionally looped out due to available resources, e.g., power consumption, the models fail catastrophically. Furthermore, training these models from scratch is even more expensive without simulator access or suitable existing models that can be fine-tuned. While such scenarios get mostly ignored in simulation, they commonly hinder a model's reusability in real-world applications. We propose a Transformer-based modality-invariant VO approach that can deal with diverse or changing sensor suites of navigation agents. Our model outperforms previous methods while training on only a fraction of the data. We hope this method opens the door to a broader range of real-world applications that can benefit from flexible and learned VO models. | ['Amir Zamir', 'Roman Bachmann', 'Marius Memmel'] | 2023-04-29 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Memmel_Modality-Invariant_Visual_Odometry_for_Embodied_Vision_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Memmel_Modality-Invariant_Visual_Odometry_for_Embodied_Vision_CVPR_2023_paper.pdf | cvpr-2023-1 | ['visual-odometry'] | ['robots'] | [-7.35032633e-02 -2.48835310e-01 5.77572100e-02 -3.07110369e-01
-5.61477244e-01 -7.99104929e-01 5.65922678e-01 -6.02145009e-02
-6.95805907e-01 8.97510946e-01 -2.04491228e-01 -2.21337989e-01
1.38338342e-01 -7.56870508e-01 -1.04436588e+00 -6.91240132e-01
-1.62548140e-01 5.99916399e-01 5.86526453e-01 -4.54691768e-01
-4.32207026e-02 4.39610690e-01 -1.75142479e+00 -4.65121001e-01
6.51102602e-01 8.51040125e-01 5.08288682e-01 7.17473745e-01
5.75100891e-02 6.04470253e-01 -6.56732976e-01 3.51277232e-01
4.30674434e-01 -1.17135659e-01 -3.62888724e-01 8.03329125e-02
2.95267761e-01 -4.11394030e-01 -5.34160197e-01 1.07045019e+00
6.23911142e-01 2.18021721e-01 1.26173899e-01 -1.59936106e+00
-2.09111691e-01 -7.25282356e-02 -2.84718275e-01 -7.78860301e-02
5.54545999e-01 6.92313254e-01 2.72523612e-01 -5.70059299e-01
7.76699185e-01 9.86474276e-01 8.63538563e-01 5.88685572e-01
-1.08455265e+00 -3.44096690e-01 4.13232625e-01 1.24862611e-01
-1.19399583e+00 -5.92204392e-01 5.77442944e-01 -1.53875560e-01
1.04303002e+00 1.66898593e-01 1.07142150e+00 1.36659002e+00
2.89090753e-01 5.94739497e-01 9.46049988e-01 1.47531312e-02
7.68356562e-01 -1.58613369e-01 -4.25042182e-01 8.12069476e-01
5.18283784e-01 1.47299305e-01 -9.06120420e-01 -1.10024899e-01
8.25983107e-01 1.79344326e-01 -3.87670159e-01 -1.10544217e+00
-1.48841238e+00 5.92912197e-01 6.81991875e-01 -1.95688665e-01
-3.85805577e-01 6.42681122e-01 2.70390302e-01 5.59404790e-01
-1.18139632e-01 5.08704305e-01 -5.74632168e-01 -5.75007081e-01
-7.16025889e-01 4.85988110e-01 7.07238674e-01 1.31580830e+00
9.43918407e-01 1.19774386e-01 5.91523111e-01 4.04643565e-01
4.08501774e-01 8.01342845e-01 4.44596410e-01 -1.21530128e+00
3.18736553e-01 4.30984586e-01 4.82520163e-01 -7.89783776e-01
-8.29714179e-01 -1.03791624e-01 -5.96088052e-01 5.98556340e-01
6.16189599e-01 -3.35975051e-01 -1.36746836e+00 1.87504351e+00
4.97443795e-01 2.35532477e-01 1.02644421e-01 1.14639902e+00
6.44983113e-01 3.37375045e-01 -3.57005119e-01 1.25586733e-01
1.06400812e+00 -8.26228857e-01 -5.35544038e-01 -9.13060725e-01
3.59690964e-01 -3.92308950e-01 1.15403318e+00 3.25807840e-01
-8.97252977e-01 4.73243967e-02 -1.35033453e+00 -1.18609965e-02
-5.18997312e-01 -6.98521912e-01 9.97818708e-01 5.81885695e-01
-1.39098144e+00 2.92985380e-01 -1.41004646e+00 -8.15373838e-01
2.32649729e-01 5.34101725e-01 -4.66616184e-01 -3.37863952e-01
-9.85517442e-01 1.11645496e+00 -2.77549066e-02 1.25690117e-01
-1.12772751e+00 -3.43073487e-01 -1.09461379e+00 -4.77359146e-01
4.71210837e-01 -1.14041221e+00 1.40552735e+00 -4.45373416e-01
-1.88268411e+00 2.87600964e-01 -1.68723375e-01 -4.37975973e-01
7.66039252e-01 2.28441488e-02 -3.83485295e-02 -1.60028502e-01
9.88871083e-02 6.67270005e-01 7.27435172e-01 -1.47500873e+00
-5.12966692e-01 -4.01504844e-01 5.26459336e-01 5.95145583e-01
1.07548594e-01 -7.46086419e-01 -6.10620320e-01 6.46440759e-02
6.42250061e-01 -1.12156773e+00 -6.45621598e-01 3.92508745e-01
-3.61480005e-02 4.39660728e-01 7.63220668e-01 -2.94128731e-02
3.55540901e-01 -1.74796593e+00 5.97410537e-02 2.16991097e-01
2.14873105e-01 -5.71097136e-02 -3.36584523e-02 5.33283532e-01
7.92257011e-01 -2.63418704e-01 -1.51342014e-02 -5.12905478e-01
1.49732515e-01 7.47673392e-01 -7.98814371e-03 7.50608325e-01
-3.14473033e-01 7.75727689e-01 -1.16662347e+00 -1.40177935e-01
5.14846563e-01 5.61477482e-01 -5.63438118e-01 -1.16397187e-01
-2.28883594e-01 7.93410361e-01 -3.59240502e-01 7.26597190e-01
5.80388069e-01 -1.41588777e-01 1.61479935e-01 1.44084543e-01
-4.04965552e-03 4.72699225e-01 -1.31103420e+00 2.21796370e+00
-7.23894238e-01 5.42909086e-01 2.26035103e-01 -7.41561711e-01
5.91831446e-01 -3.33731398e-02 2.92906284e-01 -9.30531085e-01
9.97076705e-02 3.04601967e-01 -1.73204869e-01 -4.21360165e-01
6.96047366e-01 1.16388850e-01 -3.81290913e-01 2.18503073e-01
-8.07160884e-02 -6.65959597e-01 3.51087586e-03 5.00817932e-02
1.54365706e+00 3.02379459e-01 2.60062814e-01 1.14963368e-01
-1.55198678e-01 2.97772855e-01 7.87471116e-01 1.08437848e+00
-4.04576093e-01 4.59014654e-01 2.25422867e-02 -6.04831994e-01
-1.04194522e+00 -1.05128598e+00 6.44769594e-02 8.64109755e-01
8.04254293e-01 -2.59963989e-01 -5.29119015e-01 -1.66984379e-01
2.01823488e-01 1.88862056e-01 -3.29261422e-01 -3.19616497e-01
-3.91093314e-01 -6.29214883e-01 4.73234951e-01 4.23890293e-01
6.92798734e-01 -6.75489783e-01 -1.47058177e+00 3.64231586e-01
-8.96273330e-02 -1.14210951e+00 1.35382954e-02 4.10497874e-01
-8.09255838e-01 -9.23098445e-01 -3.83587450e-01 -4.49214906e-01
7.19249487e-01 7.01872051e-01 9.84999061e-01 9.13378969e-03
-4.24632523e-03 7.00381160e-01 -2.59596080e-01 -4.83618081e-01
8.30514170e-03 6.64248690e-03 4.23419923e-01 -6.15554631e-01
1.02038346e-01 -6.47626996e-01 -7.23629415e-01 2.80913353e-01
-5.94739199e-01 -6.93673966e-03 2.85593867e-01 9.73411679e-01
6.01511896e-01 -1.93629771e-01 3.21762800e-01 -4.84433413e-01
4.10786331e-01 -5.09627819e-01 -8.31855774e-01 -1.32011697e-01
-4.79207247e-01 8.33886489e-02 4.56295937e-01 -7.10462987e-01
-5.55301607e-01 1.82725012e-01 1.47410735e-01 -2.99786866e-01
-1.50856003e-01 5.25895953e-01 -2.39954814e-01 -5.31927168e-01
8.04884613e-01 1.36330977e-01 8.03662091e-02 -1.25341728e-01
2.65276760e-01 3.96254450e-01 6.28494680e-01 -3.77850354e-01
8.63623142e-01 7.79614329e-01 -1.15838675e-02 -7.37283528e-01
-1.54281959e-01 -2.51868546e-01 -1.80316478e-01 -1.77196980e-01
3.03792387e-01 -1.00656128e+00 -8.30678940e-01 6.80126429e-01
-8.88080597e-01 -8.84588718e-01 -2.35323384e-01 5.59164464e-01
-6.93125427e-01 1.68708965e-01 -3.17406297e-01 -7.69419611e-01
1.41499549e-01 -1.35787320e+00 1.06416714e+00 4.68443245e-01
-1.18221767e-01 -8.78926754e-01 6.17225096e-02 1.45734912e-02
6.75512731e-01 3.51219952e-01 1.10598326e-01 -6.42454028e-02
-9.63200569e-01 -3.28846633e-01 2.60712981e-01 -4.20029551e-01
1.80617690e-01 -4.52907920e-01 -9.06006634e-01 -6.53079212e-01
-1.80884242e-01 -4.36344028e-01 5.81805706e-01 1.13370761e-01
6.19203269e-01 -1.06528796e-01 -5.88655531e-01 9.03985500e-01
1.32689226e+00 2.73201704e-01 3.93064260e-01 8.84765863e-01
6.24287188e-01 -4.42714849e-03 6.93970382e-01 5.16743004e-01
1.08819556e+00 8.18686545e-01 9.69088972e-01 -5.88282421e-02
1.02110133e-01 -2.78204948e-01 4.12257314e-01 4.14710820e-01
-2.66648438e-02 -3.23503882e-01 -1.04485762e+00 6.22566164e-01
-2.07433820e+00 -6.50243044e-01 2.85719931e-01 2.41504502e+00
6.16999865e-01 3.10478687e-01 -1.57543451e-01 -3.85340787e-02
2.83751935e-01 2.13251337e-01 -1.10239971e+00 -5.15392907e-02
3.88666592e-03 -2.02177554e-01 9.75578249e-01 4.94681895e-01
-7.98022628e-01 1.01757944e+00 6.07476759e+00 6.05017804e-02
-1.28942096e+00 1.52929485e-01 -1.35503650e-01 -4.63199586e-01
-3.41076463e-01 1.58215016e-01 -5.02216041e-01 5.43237746e-01
7.76660264e-01 1.32949516e-01 8.66755724e-01 9.06275094e-01
8.77030790e-02 -7.07300246e-01 -1.06456184e+00 1.31273139e+00
5.27396426e-02 -1.16802752e+00 -6.23253226e-01 1.71321064e-01
6.26692474e-01 7.22964168e-01 -2.00419165e-02 3.20541859e-01
8.13250601e-01 -8.26231182e-01 1.03542781e+00 4.03503299e-01
6.04180932e-01 -4.85530645e-01 7.89746940e-01 6.52036607e-01
-1.04953587e+00 5.63292392e-02 -5.42302489e-01 -4.25132811e-01
2.64826566e-01 3.15191984e-01 -8.33966017e-01 2.52111942e-01
9.43660021e-01 2.99762726e-01 -3.86028409e-01 1.25374734e+00
-2.32832760e-01 4.39969497e-03 -9.82990384e-01 -2.42424324e-01
3.29664052e-01 1.41944095e-01 6.21778846e-01 5.03459394e-01
4.07328844e-01 -2.14127839e-01 2.94972092e-01 4.17830110e-01
2.00949535e-01 -7.68201888e-01 -1.05197442e+00 4.28575546e-01
1.00734413e+00 9.30618644e-01 -6.46619797e-01 -1.43185005e-01
-3.72283816e-01 9.94788885e-01 3.05591047e-01 5.89173377e-01
-8.17114592e-01 -1.34983823e-01 1.07306659e+00 -1.14428796e-01
2.72362053e-01 -8.34413767e-01 -1.52780294e-01 -1.31537950e+00
1.43477559e-01 -8.28465998e-01 -4.73161265e-02 -7.19686627e-01
-9.35178161e-01 5.04724145e-01 -2.18184724e-01 -1.32815099e+00
-6.07104540e-01 -4.06341672e-01 -2.09282130e-01 5.39115012e-01
-1.62654114e+00 -1.06321943e+00 -7.31556058e-01 7.75012791e-01
3.14716101e-01 7.09933713e-02 9.24060464e-01 9.50814784e-02
-2.29681581e-01 3.85509014e-01 2.60174960e-01 -2.17705995e-01
6.34943783e-01 -1.19501960e+00 7.72799432e-01 8.88499916e-01
1.42586291e-01 7.16089308e-01 1.14049709e+00 -4.81440783e-01
-2.03693795e+00 -7.26744533e-01 2.93470830e-01 -6.83167100e-01
5.13045013e-01 -6.02500439e-01 -5.49753606e-01 9.19873297e-01
-2.25356415e-01 5.62951207e-01 2.12071612e-01 1.43144317e-02
-2.17642248e-01 -3.42212543e-02 -1.33666897e+00 8.40956092e-01
1.38800418e+00 -5.42805195e-01 -2.39222988e-01 2.20556855e-01
4.76431072e-01 -1.21223915e+00 -5.93973517e-01 2.85397857e-01
6.91128314e-01 -9.79673684e-01 1.04605472e+00 -2.61352807e-01
-5.22777259e-01 -7.92339385e-01 -2.44219109e-01 -1.68899298e+00
5.96111119e-02 -6.19188905e-01 -1.94026947e-01 6.73539996e-01
1.00751661e-01 -1.04678833e+00 8.35392535e-01 8.28894258e-01
-6.97184131e-02 -4.29963678e-01 -1.29746199e+00 -9.05810475e-01
-3.78888369e-01 -6.78612173e-01 9.74247038e-01 8.24064076e-01
8.41053501e-02 -1.47446115e-02 -4.13188308e-01 5.10870457e-01
5.57061732e-01 -4.01131548e-02 1.28633749e+00 -9.23313677e-01
-2.06544727e-01 -6.96518123e-02 -8.64280701e-01 -1.20920718e+00
-2.12189212e-01 -2.84692138e-01 5.81608534e-01 -1.82132494e+00
-4.40953106e-01 -9.20421898e-01 -1.82570536e-02 4.68534559e-01
1.29300535e-01 2.42026344e-01 2.39827782e-01 2.17250526e-01
-8.33235025e-01 5.57156920e-01 9.98286903e-01 -1.61597893e-01
-4.15955842e-01 7.59036653e-03 -2.74163693e-01 9.00063694e-01
8.64704490e-01 -2.86230087e-01 -7.59284854e-01 -8.26900482e-01
3.97141308e-01 5.01917377e-02 6.84075475e-01 -1.38735747e+00
6.62411630e-01 -4.02727157e-01 3.52016687e-01 -1.62069649e-01
8.42656672e-01 -1.00514507e+00 3.70675474e-01 4.19795960e-01
3.04290533e-01 2.70215243e-01 1.96638256e-01 7.06720948e-01
1.37915894e-01 7.11732060e-02 3.68979067e-01 -4.51587200e-01
-1.10180390e+00 3.26692641e-01 -6.12269878e-01 -5.37893958e-02
9.25972283e-01 -6.28801286e-01 -6.11625135e-01 -6.60366237e-01
-5.02670586e-01 4.34770316e-01 1.20404482e+00 4.04263467e-01
6.22311771e-01 -1.21503627e+00 1.27756134e-01 3.32640141e-01
2.97033817e-01 7.00536191e-01 8.09994191e-02 7.28520036e-01
-6.50798678e-01 5.27055189e-03 -3.31937373e-01 -7.63651311e-01
-7.70792723e-01 3.25844675e-01 5.44416130e-01 1.85952961e-01
-8.43716621e-01 7.74832487e-01 -1.80348083e-01 -7.09038913e-01
3.84463280e-01 -5.95643759e-01 3.79206181e-01 -3.40427160e-01
4.06188667e-01 1.21948443e-01 1.16700783e-01 -4.98151600e-01
-6.68407559e-01 4.88442570e-01 3.10144842e-01 -3.99193287e-01
1.27522516e+00 -5.56542814e-01 3.08433235e-01 6.47941351e-01
6.05004728e-01 -2.27837652e-01 -1.86883748e+00 -2.61068910e-01
-2.02448085e-01 -6.12681806e-01 7.30308369e-02 -5.41350842e-01
-8.35745990e-01 5.79072833e-01 7.07815647e-01 2.35021729e-02
7.68148005e-01 -1.01633959e-01 8.50351572e-01 9.21430349e-01
1.45364463e+00 -1.18624914e+00 9.03672166e-03 7.22550869e-01
4.49801236e-01 -1.35966325e+00 -1.45839935e-03 3.27800438e-02
-4.57654029e-01 6.90989673e-01 6.28639758e-01 3.64710949e-02
2.53532261e-01 7.02720582e-01 4.21546787e-01 -8.14674720e-02
-6.93769753e-01 -3.73215020e-01 -5.06168902e-01 1.21178317e+00
-3.38549733e-01 -2.30347756e-02 4.24501359e-01 5.80824390e-02
-4.72279191e-01 9.17753652e-02 6.56819880e-01 1.46046972e+00
-6.19052887e-01 -7.34491944e-01 -5.11075735e-01 2.34436482e-01
2.18294501e-01 1.85324028e-01 6.80917799e-02 8.28442216e-01
1.38717651e-01 8.74407887e-01 -2.32837386e-02 -4.40842450e-01
3.83489043e-01 -1.43795267e-01 6.55129254e-01 -4.84277785e-01
-9.52647328e-02 -3.46142858e-01 1.30734086e-01 -9.37416971e-01
-4.35243189e-01 -7.22784936e-01 -1.48151112e+00 -6.88761413e-01
-1.62867770e-01 -2.80785501e-01 9.26577806e-01 1.00666904e+00
4.82556641e-01 4.86989766e-01 1.24284424e-01 -1.39596927e+00
-3.71287018e-01 -6.55837476e-01 -2.91322500e-01 -2.75236107e-02
7.87547171e-01 -1.16868937e+00 -1.57959685e-01 -2.73940861e-01] | [4.694626331329346, 0.6704190969467163] |
ec761344-92f5-4663-9df4-51e9c5184a9b | seeing-through-clouds-in-satellite-images | 2106.08408 | null | https://arxiv.org/abs/2106.08408v1 | https://arxiv.org/pdf/2106.08408v1.pdf | Seeing Through Clouds in Satellite Images | This paper presents a neural-network-based solution to recover pixels occluded by clouds in satellite images. We leverage radio frequency (RF) signals in the ultra/super-high frequency band that penetrate clouds to help reconstruct the occluded regions in multispectral images. We introduce the first multi-modal multi-temporal cloud removal model. Our model uses publicly available satellite observations and produces daily cloud-free images. Experimental results show that our system significantly outperforms baselines by 8dB in PSNR. We also demonstrate use cases of our system in digital agriculture, flood monitoring, and wildfire detection. We will release the processed dataset to facilitate future research. | ['Ranveer Chandra', 'Peder A. Olsen', 'Mingmin Zhao'] | 2021-06-15 | null | null | null | null | ['cloud-removal'] | ['computer-vision'] | [ 5.58480084e-01 -5.93909562e-01 -1.67632371e-01 -2.46871904e-01
-8.23339880e-01 -7.79989898e-01 8.57313871e-02 -4.97463733e-01
-5.17351814e-02 7.32552528e-01 9.26745534e-02 -5.51202655e-01
-3.16222787e-01 -1.26745427e+00 -7.57990003e-01 -9.90486681e-01
-6.97721422e-01 -4.52807158e-01 -6.78288937e-02 -2.77447373e-01
-1.13538213e-01 8.39372158e-01 -1.58316612e+00 6.40917420e-01
1.15802801e+00 7.31586397e-01 7.06459463e-01 7.07527339e-01
2.82221526e-01 4.72155958e-01 -4.76104438e-01 5.09825110e-01
6.96941018e-01 1.91588819e-01 -5.37113488e-01 1.96702868e-01
6.95661902e-01 -8.79483342e-01 -3.87736768e-01 1.16239321e+00
6.47298753e-01 -1.80616692e-01 3.52515638e-01 -6.98155522e-01
-8.39736700e-01 4.41089690e-01 -1.25926399e+00 4.83319432e-01
-6.97356835e-02 -2.06781328e-01 3.92379761e-01 -9.74992514e-01
3.94124538e-01 9.99224663e-01 1.03669775e+00 -5.89614175e-02
-1.12991953e+00 -1.10762584e+00 -1.78235322e-02 2.56061368e-02
-1.67415953e+00 -3.71313900e-01 2.88453519e-01 -1.85263082e-01
9.09525692e-01 3.73715073e-01 7.27683961e-01 3.94147128e-01
3.16303253e-01 6.01144016e-01 1.30617368e+00 -4.15555507e-01
-1.33473858e-01 -5.47031283e-01 5.76708280e-02 6.93804622e-02
4.82163787e-01 6.50067270e-01 -4.21492279e-01 -2.85966575e-01
8.87635767e-01 1.86489105e-01 -7.37387955e-01 4.85041022e-01
-7.57176876e-01 8.84369791e-01 9.79488492e-01 3.03841114e-01
-8.10127914e-01 4.03824568e-01 -2.97922224e-01 3.99945766e-01
8.81501794e-01 6.68272153e-02 -5.74998975e-01 8.59599054e-01
-1.44290018e+00 3.61043394e-01 2.71237880e-01 9.75837708e-01
6.66361213e-01 7.08498776e-01 1.27517313e-01 6.66764975e-01
2.85494238e-01 1.52015102e+00 -3.64459194e-02 -1.31394768e+00
1.05160631e-01 -3.81910801e-01 5.83079636e-01 -1.12267649e+00
-2.45178297e-01 -6.98492944e-01 -1.24672520e+00 7.66163766e-02
-4.01659459e-01 -2.31702432e-01 -1.09377968e+00 1.03702497e+00
1.13860689e-01 6.70333982e-01 3.84375036e-01 1.13990796e+00
7.89767623e-01 9.89011109e-01 -5.05519286e-03 -3.89860839e-01
1.23075902e+00 -5.34024775e-01 -1.00933361e+00 -4.99883622e-01
4.88954745e-02 -6.10292435e-01 5.05307198e-01 3.00613999e-01
-5.47436476e-01 -3.23248595e-01 -9.20288563e-01 5.16094029e-01
-4.25585240e-01 2.86526769e-01 1.04594719e+00 3.67474824e-01
-1.31469154e+00 6.52683079e-01 -6.46382391e-01 -1.94354430e-01
4.79503512e-01 6.24556504e-02 3.96173075e-02 -2.27630958e-01
-1.34567142e+00 6.48559928e-01 2.18103498e-01 8.96187365e-01
-7.34687150e-01 -7.41006076e-01 -5.25431633e-01 -3.09767644e-03
-7.31682107e-02 -2.85922021e-01 9.20787215e-01 -1.03750432e+00
-7.90361464e-01 5.32574058e-01 -1.40594378e-01 -6.35944426e-01
-1.78190231e-01 -5.04175007e-01 -7.30394423e-01 6.22997105e-01
1.97249219e-01 5.68360209e-01 9.30595577e-01 -1.66632104e+00
-7.46286452e-01 -4.71295744e-01 -2.55955487e-01 -1.60196740e-02
-5.64594716e-02 5.04438281e-02 1.46432325e-01 -1.06045079e+00
6.72883391e-01 -1.07426035e+00 -3.12120944e-01 1.83745712e-01
-9.83356088e-02 7.04429686e-01 1.08168590e+00 -9.16450560e-01
6.52998686e-01 -2.05089688e+00 -5.92867613e-01 7.60600939e-02
-1.58516854e-01 4.93333817e-01 -5.27055442e-01 2.95848310e-01
-2.50486106e-01 3.14806938e-01 -4.83852893e-01 2.97064304e-01
-6.41246021e-01 3.36915523e-01 -1.01625407e+00 7.20409751e-01
1.58904791e-01 4.72616911e-01 -5.51421762e-01 2.19173040e-02
9.02287513e-02 8.53669286e-01 -7.21947150e-03 -2.08036393e-01
6.24449030e-02 2.34388739e-01 -4.12948608e-01 1.28777301e+00
2.01395106e+00 6.06316957e-04 1.02893628e-01 -3.30358863e-01
-4.16708291e-01 -3.12750071e-01 -9.55592573e-01 1.26964808e+00
-3.67857963e-01 7.66344428e-01 6.40492797e-01 -5.97343504e-01
6.69302166e-01 2.96162754e-01 3.32330465e-01 -6.90711558e-01
-3.66368324e-01 1.43339053e-01 -6.35830641e-01 -7.62265980e-01
6.87598109e-01 -1.63070023e-01 4.47618961e-01 2.50501573e-01
-3.35769027e-01 -3.62318695e-01 -4.40420568e-01 -3.44043933e-02
6.05749428e-01 2.36038208e-01 -1.94500446e-01 -3.15477192e-01
6.85312673e-02 3.11062276e-01 4.24737692e-01 1.02338433e+00
-2.73842394e-01 7.42903590e-01 -4.06544447e-01 -7.14501500e-01
-5.54953396e-01 -1.11977792e+00 -4.72302437e-01 1.00976932e+00
1.05822764e-01 3.76180738e-01 -9.99446437e-02 -1.81361903e-02
2.91210979e-01 4.48221356e-01 -3.76362085e-01 4.12080854e-01
-3.53458971e-01 -1.47860861e+00 7.39963949e-01 4.82591629e-01
9.00319278e-01 -9.29828644e-01 -3.73315364e-01 9.59876329e-02
-6.98547840e-01 -1.24635375e+00 2.98962235e-01 2.87669748e-01
-1.30434990e+00 -6.89642906e-01 -5.40665865e-01 -6.28542602e-01
2.59384394e-01 1.38801491e+00 1.02963471e+00 2.62946069e-01
-5.21081448e-01 1.69184543e-02 -4.21705455e-01 -5.30320108e-01
3.10686052e-01 -3.63364190e-01 -1.04195386e-01 -3.51610750e-01
3.45692843e-01 -9.27427053e-01 -7.35826194e-01 -1.92814827e-01
-1.09245300e+00 -3.89156193e-01 7.12138355e-01 4.59657848e-01
7.24420309e-01 4.23504353e-01 2.89871842e-01 -4.39220518e-01
1.32930517e-01 -5.15457749e-01 -9.72141564e-01 -3.59574109e-02
-2.13318914e-01 -5.24879634e-01 -7.02990294e-02 -2.64486551e-01
-1.03700745e+00 6.07472241e-01 1.88469350e-01 -3.59875679e-01
-1.62620336e-01 1.02404571e+00 4.38249856e-01 -6.19985878e-01
7.52605975e-01 4.92116302e-01 -4.58905667e-01 -3.52241457e-01
3.26966703e-01 1.04275632e+00 1.02552831e+00 2.19163951e-02
1.19420624e+00 1.21813583e+00 -1.72550872e-01 -1.38528800e+00
-8.06909800e-01 -5.67115963e-01 -5.21636367e-01 -4.70344782e-01
3.95055175e-01 -1.83312321e+00 -3.99519265e-01 5.88095844e-01
-1.22119796e+00 -2.19467968e-01 2.93437392e-01 5.70692778e-01
1.95388328e-02 2.99367577e-01 -5.76577842e-01 -1.19145417e+00
-8.54778111e-01 -3.48584950e-01 1.32474649e+00 2.64369547e-01
6.83053553e-01 -3.90000463e-01 -1.32216094e-02 2.61127084e-01
9.39901650e-01 4.75972533e-01 -1.64485294e-02 5.59421420e-01
-1.08821321e+00 -1.03315413e-01 -5.91577232e-01 1.68725833e-01
1.98360458e-01 2.52687961e-01 -1.53937566e+00 -5.40234566e-01
-7.14389235e-02 -4.02722172e-02 1.64335716e+00 1.10146832e+00
1.22227216e+00 -2.65255898e-01 -4.10615981e-01 1.19091773e+00
2.02977848e+00 -2.78113723e-01 9.46287692e-01 4.95436609e-01
3.18429112e-01 4.57091451e-01 7.81724393e-01 4.43501800e-01
3.80883180e-02 6.19503902e-03 9.51552808e-01 -5.12910903e-01
1.77557871e-01 5.65800130e-01 1.40640929e-01 3.26139301e-01
-3.91157687e-01 -3.21826011e-01 -7.43736029e-01 9.42854404e-01
-1.55912101e+00 -1.26480782e+00 -5.73864460e-01 1.71018553e+00
5.61111927e-01 -4.96335059e-01 -6.00853264e-01 -1.73829332e-01
8.24551225e-01 5.72625339e-01 -2.57389218e-01 1.85976252e-01
-6.28651321e-01 5.07498324e-01 1.42805302e+00 5.20490289e-01
-1.70620108e+00 1.08713365e+00 7.45560265e+00 4.37010914e-01
-1.35767007e+00 2.29245335e-01 1.69661865e-02 -1.02197798e-02
-2.23286152e-01 -1.23806238e-01 -3.77814531e-01 6.47136495e-02
9.21267569e-01 2.59293914e-01 4.44105297e-01 5.46300709e-01
7.19644129e-01 -4.81183261e-01 3.48947167e-01 6.09812737e-01
-2.14609861e-01 -1.31087923e+00 -7.73753151e-02 4.22469787e-02
6.93827569e-01 8.22376490e-01 8.60474855e-02 -1.61937431e-01
6.31336510e-01 -8.82789016e-01 3.23629230e-01 6.01168573e-01
9.18083727e-01 -7.23048508e-01 7.37847388e-01 1.77476928e-01
-1.29740822e+00 -6.05026260e-02 -8.07986856e-01 -1.66726962e-01
-1.55519888e-01 1.09048915e+00 -3.92154962e-01 9.23415363e-01
1.18420076e+00 7.80723512e-01 -1.67899027e-01 1.25962651e+00
-2.74302065e-01 8.09625864e-01 -5.98802149e-01 7.21174181e-01
1.30605832e-01 -2.41174117e-01 3.64711672e-01 1.33176982e+00
6.76358342e-01 7.02103853e-01 1.10188656e-01 6.87453508e-01
2.32978016e-02 -5.68418503e-01 -9.53969359e-01 2.28714660e-01
7.44890332e-01 1.18668115e+00 -4.21900153e-01 4.56092022e-02
-1.52417123e-01 9.36584473e-01 -6.13338768e-01 7.39778757e-01
-7.62166321e-01 -4.44471866e-01 8.18343103e-01 -1.60041690e-01
6.79063857e-01 -4.37360793e-01 -1.28634319e-01 -9.43849921e-01
-4.52969819e-02 -6.24302566e-01 1.72555707e-02 -1.49445331e+00
-9.75801826e-01 4.20813560e-01 -1.15464635e-01 -1.56212890e+00
1.03946224e-01 -2.89008319e-01 -2.84452856e-01 1.15622818e+00
-2.54599786e+00 -1.53320205e+00 -9.14676309e-01 3.56717378e-01
1.62386194e-01 1.82859510e-01 1.18389511e+00 3.15046191e-01
-3.74403298e-02 -4.52899784e-01 5.44787049e-01 -1.82914771e-02
6.14813507e-01 -7.64225960e-01 4.96362567e-01 1.32789111e+00
-2.82032222e-01 3.15778166e-01 8.18431079e-01 -8.86956990e-01
-1.40956497e+00 -1.86099315e+00 7.02160537e-01 3.98270905e-01
2.74099588e-01 1.50669128e-01 -9.13610041e-01 8.38595450e-01
2.92617887e-01 4.50062156e-01 5.52750349e-01 -3.16989005e-01
-2.52350509e-01 -4.82888967e-01 -1.32460880e+00 -1.21523077e-02
5.80272198e-01 -5.90343773e-01 -1.18414722e-01 8.72021437e-01
7.85125732e-01 -2.79342771e-01 -5.85593164e-01 6.98657393e-01
7.33535290e-01 -7.33808041e-01 1.31030083e+00 -1.04408801e-01
3.49400371e-01 -6.91897750e-01 -6.89471364e-01 -1.27043188e+00
-7.95994878e-01 -1.35829791e-01 2.16019422e-01 6.86424911e-01
9.48289633e-02 -4.52639997e-01 4.38900501e-01 -3.18475932e-01
1.83416471e-01 2.78407335e-01 -8.28216136e-01 -8.12187254e-01
9.60472785e-03 -3.47628742e-01 6.84644878e-01 1.21916270e+00
-9.07322228e-01 -4.97188836e-01 -6.47917032e-01 1.36505580e+00
8.65967453e-01 6.38812065e-01 3.36024374e-01 -1.34587610e+00
1.65949270e-01 3.89038265e-01 1.20923541e-01 -6.68608844e-01
1.07754312e-01 -4.48362321e-01 2.51595736e-01 -1.46345258e+00
1.60741270e-01 -1.04699232e-01 -1.28886417e-01 7.76552320e-01
5.11825159e-02 7.92989790e-01 1.36954570e-02 3.84253383e-01
2.22575083e-01 4.05880868e-01 8.85758519e-01 -5.21746993e-01
-9.99192521e-02 1.87424216e-02 -4.65777934e-01 6.96869850e-01
1.00048351e+00 -7.91647613e-01 8.45628604e-02 -1.26002419e+00
2.82991640e-02 2.72061974e-01 7.19434440e-01 -1.07030022e+00
-7.57695213e-02 -5.65075219e-01 7.56841719e-01 -1.20499957e+00
3.41904163e-01 -1.05235922e+00 2.86695451e-01 6.01800978e-01
4.24664281e-02 -3.01312029e-01 6.65876448e-01 6.96214676e-01
-1.74224287e-01 1.74711272e-01 9.51893270e-01 -2.00751886e-01
-7.43251443e-01 1.65768325e-01 -8.47958803e-01 -7.36712217e-01
4.79266793e-01 3.09291780e-02 -7.94929206e-01 -5.50302207e-01
-6.36655390e-01 1.88760102e-01 1.29313245e-01 1.83203861e-01
7.42870331e-01 -1.04119539e+00 -1.06994009e+00 2.73060173e-01
-1.86076462e-01 -2.06853762e-01 2.54218161e-01 4.36913729e-01
-9.59073424e-01 3.41114432e-01 -3.11480612e-01 -7.38403440e-01
-1.29762816e+00 1.37832463e-01 6.61588967e-01 2.82216936e-01
-6.16051376e-01 6.81762636e-01 -3.49945664e-01 -5.26989460e-01
-1.06952980e-01 -5.11872590e-01 -1.76648408e-01 -3.57123315e-02
9.28429306e-01 5.40252449e-03 2.54770279e-01 -5.38508654e-01
-4.88646448e-01 4.94297445e-01 3.47275645e-01 -6.50693327e-02
2.04321337e+00 -2.38013729e-01 -4.06390637e-01 3.50857265e-02
6.70683205e-01 -9.25066397e-02 -9.15218413e-01 -3.92287195e-01
-3.98600549e-01 -9.32087541e-01 9.66051280e-01 -8.82326126e-01
-1.71690428e+00 5.31953990e-01 1.28348982e+00 1.30366132e-01
1.73611116e+00 -5.86985946e-01 6.19654536e-01 7.29337454e-01
2.75836110e-01 -7.99124062e-01 -5.24490476e-01 5.27107775e-01
1.04276979e+00 -1.09707880e+00 5.40592313e-01 -6.95518851e-01
-1.36043936e-01 1.10587728e+00 1.22862734e-01 -3.15382779e-01
9.01291847e-01 7.71705031e-01 4.51483637e-01 -2.08288863e-01
-3.43331039e-01 -5.60918152e-01 -5.15171468e-01 1.03917861e+00
1.28558084e-01 4.07929242e-01 1.98822156e-01 9.14258733e-02
2.34778434e-01 5.05153656e-01 8.81148279e-01 1.31497169e+00
-8.23084116e-01 -3.53956848e-01 -1.08273351e+00 4.67490822e-01
-4.53126013e-01 -5.85524023e-01 -5.22307828e-02 5.97110987e-01
3.30416742e-03 1.20000136e+00 1.45045817e-01 -1.64633349e-01
9.84728988e-03 -4.88180816e-01 2.13015795e-01 -2.05298424e-01
-2.90303588e-01 5.73489308e-01 -1.20368548e-01 -6.09057069e-01
-1.16872597e+00 -5.83064079e-01 -9.12601650e-01 -5.62900603e-01
-8.94029975e-01 -2.88688809e-01 7.85041034e-01 4.55455989e-01
4.62417424e-01 4.36278880e-01 8.95046949e-01 -1.16526854e+00
-2.34465390e-01 -1.15227258e+00 -1.17326546e+00 -5.40409625e-01
1.17238295e+00 -4.18292791e-01 -6.10696375e-01 1.42837137e-01] | [9.809422492980957, -1.7688673734664917] |
69be2ef0-1f49-4fbc-b04a-8b2d9c70ad58 | structure-aware-language-model-pretraining | 2305.19912 | null | https://arxiv.org/abs/2305.19912v1 | https://arxiv.org/pdf/2305.19912v1.pdf | Structure-Aware Language Model Pretraining Improves Dense Retrieval on Structured Data | This paper presents Structure Aware Dense Retrieval (SANTA) model, which encodes user queries and structured data in one universal embedding space for retrieving structured data. SANTA proposes two pretraining methods to make language models structure-aware and learn effective representations for structured data: 1) Structured Data Alignment, which utilizes the natural alignment relations between structured data and unstructured data for structure-aware pretraining. It contrastively trains language models to represent multi-modal text data and teaches models to distinguish matched structured data for unstructured texts. 2) Masked Entity Prediction, which designs an entity-oriented mask strategy and asks language models to fill in the masked entities. Our experiments show that SANTA achieves state-of-the-art on code search and product search and conducts convincing results in the zero-shot setting. SANTA learns tailored representations for multi-modal text data by aligning structured and unstructured data pairs and capturing structural semantics by masking and predicting entities in the structured data. All codes are available at https://github.com/OpenMatch/OpenMatch. | ['Ge Yu', 'Zhiyuan Liu', 'Yu Gu', 'Shi Yu', 'Chenyan Xiong', 'Zhenghao Liu', 'Xinze Li'] | 2023-05-31 | null | null | null | null | ['code-search', 'code-search'] | ['computer-code', 'computer-vision'] | [ 1.89968962e-02 3.57939601e-01 -5.42631924e-01 -4.80381310e-01
-1.31735492e+00 -6.22007608e-01 4.55050081e-01 3.52849275e-01
-3.79338712e-01 1.01764001e-01 8.71134698e-01 -1.50625527e-01
-1.96570218e-01 -7.19048142e-01 -6.67736173e-01 -2.02098563e-01
-1.03839479e-01 1.04834509e+00 -1.60360485e-01 -2.93743581e-01
3.77296726e-03 1.24370217e-01 -1.70524693e+00 9.86231983e-01
8.42435718e-01 9.60485756e-01 6.16025388e-01 5.04895687e-01
-7.69392014e-01 1.22599626e+00 3.58763570e-03 -4.86964047e-01
3.00448835e-01 -1.09186336e-01 -1.14008391e+00 -1.76855773e-01
4.99587804e-01 -3.08834374e-01 -8.57595742e-01 8.37887168e-01
5.12533486e-01 1.94175154e-01 6.51785672e-01 -8.93616319e-01
-1.45912898e+00 1.05622685e+00 -3.67980480e-01 2.40539744e-01
5.40910363e-01 -3.73116434e-01 1.70920515e+00 -1.48897552e+00
7.67876863e-01 1.14065385e+00 3.00727665e-01 6.98579907e-01
-1.08169675e+00 -4.42364395e-01 1.40585741e-02 1.82016358e-01
-1.57079446e+00 -6.31653786e-01 6.46723330e-01 -5.77090323e-01
1.38467598e+00 1.82973012e-01 8.46862644e-02 1.07339323e+00
-3.45159531e-01 1.37371969e+00 2.37163201e-01 -4.56229329e-01
-1.89849347e-01 3.58462840e-01 6.75726235e-01 1.03400016e+00
2.22781181e-01 1.93565473e-01 -6.28183007e-01 -3.77046168e-01
2.38127798e-01 3.96502405e-01 -1.68512374e-01 -7.49056220e-01
-9.81610894e-01 9.98966694e-01 4.15322006e-01 4.07592386e-01
-3.52810711e-01 -5.54283969e-02 5.72302043e-01 5.74278712e-01
2.32003614e-01 7.56989539e-01 -5.39042890e-01 2.92585433e-01
-1.00975311e+00 1.20165385e-01 7.68655241e-01 1.51752770e+00
1.11421931e+00 -1.74591705e-01 -3.11091244e-01 1.08099985e+00
5.17732918e-01 5.94791174e-01 1.09766710e+00 -5.42426288e-01
8.49650085e-01 1.03278875e+00 -1.63976684e-01 -9.22434330e-01
-1.76766440e-01 -2.50736028e-01 -6.51931882e-01 -6.44909203e-01
-2.94541627e-01 1.73833027e-01 -9.25322473e-01 1.57833958e+00
-7.76906535e-02 -1.96482971e-01 4.46557194e-01 7.44702399e-01
1.55012417e+00 8.46156061e-01 1.04347751e-01 1.62126854e-01
1.29045713e+00 -1.20818114e+00 -5.66061914e-01 -4.57288116e-01
1.10724688e+00 -6.75130665e-01 1.31494284e+00 -2.71476835e-01
-1.20851302e+00 -5.85648239e-01 -7.86848783e-01 -8.32377851e-01
-7.80737519e-01 4.08057481e-01 4.09915477e-01 6.91435784e-02
-1.07366705e+00 2.29048818e-01 -6.10760450e-01 -3.44936162e-01
2.41681650e-01 2.16894880e-01 -4.44513679e-01 -2.82232612e-01
-1.41555488e+00 5.76761544e-01 5.06483197e-01 -3.67989451e-01
-1.01045632e+00 -8.52741063e-01 -1.34979033e+00 5.45066595e-01
3.71270120e-01 -6.00928962e-01 1.32631469e+00 -8.48094821e-01
-6.31061852e-01 1.34877872e+00 -4.69660193e-01 -4.66358513e-01
-2.09168628e-01 -3.02150816e-01 -5.15635610e-01 1.54076084e-01
1.98112845e-01 6.45910263e-01 7.06292510e-01 -1.10429144e+00
-3.40806901e-01 -5.08678854e-01 -1.23958662e-01 3.09679657e-01
-6.35110080e-01 1.38681963e-01 -7.41878688e-01 -7.25166440e-01
1.37781575e-02 -3.98892313e-01 -9.44522172e-02 -2.19181165e-01
-4.39066350e-01 -4.86344993e-01 4.08787787e-01 -8.62435520e-01
1.61895204e+00 -2.12863588e+00 4.26595181e-01 1.62140876e-01
5.32639384e-01 3.86394523e-02 -6.81138813e-01 8.83829474e-01
-4.08460572e-02 1.32358968e-01 7.18406290e-02 -5.08756638e-01
3.98463517e-01 8.09404999e-02 -8.34298134e-01 -1.24515854e-01
4.74919192e-02 1.45086873e+00 -8.16585898e-01 -5.65828145e-01
-2.10652128e-01 3.23208839e-01 -5.41597605e-01 5.72722435e-01
-1.37896016e-01 -5.57720006e-01 -6.96464419e-01 8.66703808e-01
2.12159738e-01 -8.21614027e-01 1.67334780e-01 -1.84187680e-01
3.30641389e-01 6.70584977e-01 -8.40197921e-01 2.00218034e+00
-5.40276647e-01 5.16801119e-01 3.72414626e-02 -8.91302705e-01
9.26432848e-01 3.98020923e-01 4.89719659e-01 -1.06699657e+00
-1.40461966e-01 1.64549991e-01 -7.70557940e-01 -7.22555399e-01
1.04819262e+00 2.86836863e-01 -5.42582393e-01 6.22894287e-01
4.26058859e-01 8.22724476e-02 -9.31851752e-03 6.94303274e-01
1.27764904e+00 -1.68315068e-01 1.78573370e-01 -2.76235431e-01
4.68903691e-01 6.64035417e-03 2.28896812e-01 8.15278947e-01
3.40459049e-01 3.68947774e-01 2.69236695e-03 -5.80619693e-01
-9.54846025e-01 -1.12554288e+00 3.42010349e-01 1.69606149e+00
3.47833723e-01 -7.61777222e-01 -1.31298661e-01 -7.75843799e-01
3.33399683e-01 5.33266425e-01 -6.71401680e-01 -4.24753785e-01
-3.89620274e-01 -3.70761901e-02 2.86432385e-01 6.46519363e-01
9.33589414e-02 -1.09931362e+00 -1.44417509e-01 4.64931922e-03
-2.15219885e-01 -8.40603709e-01 -8.74116480e-01 4.30386603e-01
-6.31296515e-01 -1.12942743e+00 -4.75533664e-01 -1.36013699e+00
6.82275355e-01 5.63605487e-01 1.62883317e+00 2.52827853e-01
-3.58393610e-01 5.94029188e-01 -5.98440170e-01 -7.25530367e-03
-3.46621662e-01 3.81767869e-01 -1.37591183e-01 -2.94277072e-01
9.51956689e-01 -2.04454705e-01 -5.08920431e-01 1.87748984e-01
-1.14240050e+00 1.96244121e-02 7.05590129e-01 7.90811956e-01
7.66942203e-01 -3.59243125e-01 3.19648981e-01 -1.11005926e+00
8.52936745e-01 -8.92430246e-01 -5.60497284e-01 8.24681044e-01
-6.95676506e-01 5.77475071e-01 5.16977429e-01 -3.55089784e-01
-7.20826209e-01 2.59663891e-02 7.17807189e-02 -7.64913440e-01
-3.84050459e-02 9.05875206e-01 -5.54120615e-02 4.52861518e-01
8.28131616e-01 4.67204273e-01 -3.26389015e-01 -9.11636591e-01
6.57438517e-01 9.61110413e-01 3.69165123e-01 -7.14888394e-01
7.42934883e-01 1.46788165e-01 -8.06051791e-01 -4.87951934e-01
-1.11819446e+00 -9.47825730e-01 -5.58149993e-01 3.83669764e-01
7.75033772e-01 -1.34846556e+00 -6.82715653e-03 -1.32688716e-01
-9.05298293e-01 -2.07386896e-01 -6.08009815e-01 1.33453742e-01
-4.16299582e-01 2.71813154e-01 -6.72204554e-01 -3.71280611e-01
-6.28095388e-01 -9.21995521e-01 1.42186606e+00 -2.28738002e-02
-2.39049241e-01 -8.84510219e-01 2.93406159e-01 4.96722221e-01
3.67740542e-01 -4.20384586e-01 1.21721137e+00 -1.30363989e+00
-6.72806442e-01 -3.55295062e-01 -2.85652667e-01 5.96581027e-02
6.93195919e-03 -4.61405575e-01 -7.69700110e-01 -5.07119715e-01
-3.45821232e-01 -1.00326526e+00 1.04479933e+00 -1.83242545e-01
1.36699927e+00 -5.45821548e-01 -4.25083578e-01 6.16567850e-01
1.29492533e+00 -3.56369615e-02 3.71334225e-01 1.04020134e-01
6.81172669e-01 6.84584916e-01 4.60966796e-01 3.81986439e-01
6.21488452e-01 5.92322290e-01 2.70699173e-01 6.45317361e-02
-1.18202753e-01 -8.49820375e-01 2.35269219e-01 1.16948974e+00
8.26638818e-01 -2.17313096e-01 -1.04238105e+00 9.29100633e-01
-1.79656148e+00 -1.16436398e+00 5.25061965e-01 1.94764555e+00
1.16981542e+00 -3.56820911e-01 -1.26621380e-01 -5.86988270e-01
4.89951491e-01 3.00100684e-01 -7.42797852e-01 -4.47307006e-02
-1.49567708e-01 3.18878144e-01 2.93101341e-01 6.11000180e-01
-1.05681658e+00 1.12680423e+00 5.76880646e+00 8.26063395e-01
-6.49303377e-01 1.24178261e-01 2.95571327e-01 -3.10123503e-01
-9.32518125e-01 -2.73658216e-01 -1.00298083e+00 3.30314964e-01
1.00685430e+00 -6.78703904e-01 4.28259194e-01 1.22803092e+00
-4.10996914e-01 7.41195083e-01 -1.40112817e+00 1.21300232e+00
3.47985148e-01 -1.75390935e+00 5.10148823e-01 9.23684519e-03
7.27033138e-01 4.38579679e-01 9.28374827e-02 9.66102183e-01
6.81611419e-01 -1.09086978e+00 4.39302176e-01 6.53967619e-01
7.32919693e-01 -2.83729851e-01 4.80834603e-01 3.15417647e-01
-1.30765641e+00 -2.84004480e-01 -5.69577813e-01 5.87126434e-01
-1.93629593e-01 1.38080150e-01 -6.30724669e-01 3.43879580e-01
5.91277301e-01 1.01258111e+00 -7.49652684e-01 7.89153457e-01
3.76375690e-02 1.50473550e-01 -2.98299305e-02 3.43534760e-02
2.77822554e-01 2.10794508e-01 2.97740996e-01 1.15546906e+00
2.94165820e-01 -1.98237315e-01 3.12520117e-01 1.10322011e+00
-6.70849562e-01 3.46232891e-01 -8.43406856e-01 -6.70448244e-01
8.27714562e-01 1.04331493e+00 1.22941792e-01 -4.93057996e-01
-6.75643444e-01 9.50743794e-01 6.85728967e-01 5.63043356e-01
-3.29288900e-01 -5.13796210e-01 7.74701536e-01 9.56428796e-02
5.26836753e-01 8.66571367e-02 1.06934197e-01 -1.70550799e+00
9.92903188e-02 -1.23421395e+00 9.55196083e-01 -9.62648511e-01
-1.42026353e+00 8.71237874e-01 -6.31736144e-02 -1.12536526e+00
-6.06023848e-01 -5.05257964e-01 -2.88247839e-02 8.11915815e-01
-1.57938778e+00 -1.04836869e+00 -6.46917447e-02 8.51315618e-01
9.96461868e-01 -7.22344756e-01 1.05511928e+00 3.35279197e-01
-3.06643575e-01 9.29231405e-01 3.23183954e-01 6.36833668e-01
5.73998809e-01 -1.13813984e+00 4.63493645e-01 6.32769227e-01
7.61509717e-01 1.13610327e+00 2.29637414e-01 -4.94183898e-01
-1.76051724e+00 -1.04878461e+00 1.36208034e+00 -5.88297606e-01
9.52822685e-01 -5.33549607e-01 -1.21536684e+00 8.18297029e-01
3.40595126e-01 8.90013296e-03 1.24676657e+00 2.40950972e-01
-1.05887794e+00 1.32188499e-01 -9.11731064e-01 3.87964129e-01
9.69601035e-01 -1.33028436e+00 -1.21220374e+00 4.45315659e-01
1.24545014e+00 -3.41281205e-01 -8.64726722e-01 2.52093226e-01
3.35357308e-01 -4.09307897e-01 1.01060176e+00 -1.32400179e+00
5.60553133e-01 7.74276182e-02 -6.59516335e-01 -9.09487724e-01
-4.05446947e-01 -4.73176688e-01 -8.56005311e-01 9.81691241e-01
7.67166078e-01 -2.24391237e-01 8.64326298e-01 8.13756227e-01
-1.74155802e-01 -9.75261986e-01 -5.46012461e-01 -6.13415480e-01
5.67266382e-02 -1.06996395e-01 7.78752446e-01 1.18101323e+00
1.66251779e-01 5.26126385e-01 -5.56783602e-02 2.77856648e-01
3.10068995e-01 6.90215588e-01 4.55126762e-01 -1.13226271e+00
-4.03969824e-01 -3.02351326e-01 -1.97077274e-01 -1.54942155e+00
6.12656116e-01 -1.45276713e+00 -2.67836660e-01 -1.63487279e+00
5.11725426e-01 -3.08520108e-01 -4.59505320e-01 7.68405259e-01
4.01851684e-02 -1.52713150e-01 -9.34698507e-02 5.40839255e-01
-1.09462714e+00 7.05028415e-01 5.11664510e-01 -5.51181614e-01
-9.03455094e-02 -3.88893902e-01 -1.09469986e+00 1.29282519e-01
5.46647549e-01 -3.28250796e-01 -7.96901822e-01 -1.01316571e+00
6.54183209e-01 3.64995338e-02 7.17543513e-02 -4.97462034e-01
3.60582381e-01 1.14407465e-01 1.68450937e-01 -7.35979795e-01
2.31247559e-01 -1.02991760e+00 -3.10683787e-01 -3.93539742e-02
-1.16753328e+00 2.03783423e-01 2.30979919e-01 5.78313291e-01
-5.69812596e-01 -3.90126377e-01 3.09003443e-01 -2.31348410e-01
-1.12451041e+00 6.35530710e-01 -8.70310217e-02 6.00971639e-01
4.15340394e-01 9.86676663e-02 -5.33517182e-01 -5.03233075e-01
-8.84224415e-01 7.61684537e-01 3.47076893e-01 7.98289001e-01
8.93923461e-01 -1.57984459e+00 -5.04251480e-01 4.09234822e-01
7.73932397e-01 -2.34663159e-01 2.08481759e-01 2.08564460e-01
-9.56037566e-02 7.78433621e-01 2.46637940e-01 -1.92502052e-01
-1.36907434e+00 9.42022920e-01 2.41860524e-01 -5.12525260e-01
-3.62741500e-01 1.11944199e+00 2.36911774e-01 -7.80220509e-01
6.01644158e-01 -2.89236903e-01 -1.96431547e-01 1.41600907e-01
7.83171594e-01 -1.59879714e-01 2.13517383e-01 -6.49537981e-01
-1.44961268e-01 2.84472853e-01 -7.26469815e-01 2.50065267e-01
1.45283449e+00 -3.98468792e-01 -1.99331895e-01 3.30213010e-01
1.75652039e+00 8.34471919e-03 -6.71324372e-01 -1.06315207e+00
4.68647301e-01 -3.99411738e-01 2.53663182e-01 -7.00172246e-01
-1.00308490e+00 8.34028959e-01 3.39227408e-01 1.09968327e-01
1.01000464e+00 6.53284013e-01 9.94880080e-01 1.13786948e+00
2.20258221e-01 -9.78222609e-01 3.16317707e-01 6.86881483e-01
9.58879232e-01 -1.23867607e+00 -3.59455228e-01 1.22059748e-01
-8.39036703e-01 6.93726361e-01 6.48089409e-01 1.15462795e-01
7.21910000e-01 6.20974861e-02 3.29050012e-02 -7.21831322e-01
-1.11954093e+00 -5.06190479e-01 7.88167179e-01 3.63615155e-01
6.52039587e-01 -2.10908815e-01 3.06748450e-01 8.35690618e-01
-7.02706873e-02 -3.47901613e-01 2.06280462e-02 9.82862830e-01
-6.23027623e-01 -1.18365359e+00 2.66455233e-01 7.38162398e-01
-1.94885060e-01 -7.02470422e-01 -7.81923532e-01 4.95463848e-01
-4.10002619e-01 5.55904269e-01 1.34445757e-01 -4.78698939e-01
3.96384478e-01 5.87627232e-01 -3.98441441e-02 -1.16366172e+00
-5.53048670e-01 -2.74033844e-01 -7.72625059e-02 -9.39432859e-01
4.93756309e-02 -5.04932821e-01 -1.14570749e+00 1.18976697e-01
-1.24921367e-01 6.35506094e-01 2.63810515e-01 5.29067636e-01
1.11029899e+00 9.59384218e-02 6.14816308e-01 -3.72425407e-01
-7.83487022e-01 -7.66752243e-01 -4.73442465e-01 6.75195396e-01
5.06066978e-01 -3.62451971e-01 -4.02702808e-01 -5.30552343e-02] | [11.252346992492676, 7.591898441314697] |
7e7b78c1-79a1-449d-abd8-ee6a2d6c639e | graph-feedback-via-reduction-to-regression | 2302.08631 | null | https://arxiv.org/abs/2302.08631v2 | https://arxiv.org/pdf/2302.08631v2.pdf | Practical Contextual Bandits with Feedback Graphs | While contextual bandit has a mature theory, effectively leveraging different feedback patterns to enhance the pace of learning remains unclear. Bandits with feedback graphs, which interpolates between the full information and bandit regimes, provides a promising framework to mitigate the statistical complexity of learning. In this paper, we propose and analyze an approach to contextual bandits with feedback graphs based upon reduction to regression. The resulting algorithms are computationally practical and achieve established minimax rates, thereby reducing the statistical complexity in real-world applications. | ['Paul Mineiro', 'Haipeng Luo', 'Olga Vrousgou', 'Yuheng Zhang', 'Mengxiao Zhang'] | 2023-02-17 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [ 2.90393561e-01 -5.52587695e-02 -1.20559883e+00 -1.42083749e-01
-1.06604683e+00 -5.90092838e-01 6.92451358e-01 1.06740043e-01
-2.21037373e-01 1.21935070e+00 1.16807625e-01 -9.27675843e-01
-7.59070814e-01 -6.38208270e-01 -1.06614864e+00 -8.28574181e-01
-8.85455236e-02 1.31101936e-01 -3.81956506e-03 9.75013077e-02
4.69988763e-01 2.77175903e-01 -1.35119677e+00 2.32272655e-01
1.07783401e+00 1.19011724e+00 -1.99299440e-01 7.38595307e-01
-1.83476418e-01 9.57793474e-01 -2.87296504e-01 -4.36423749e-01
5.36808729e-01 -5.29905498e-01 -5.61828315e-01 -6.32445365e-02
7.14085698e-01 -3.05804849e-01 -5.90319216e-01 1.04642081e+00
1.83175877e-01 2.00821996e-01 4.21382725e-01 -9.19941723e-01
-8.50600660e-01 1.04572308e+00 -6.02124214e-01 5.67323387e-01
1.90842345e-01 -1.32186219e-01 1.34817266e+00 -4.37665701e-01
9.31798294e-02 1.15773797e+00 7.08633065e-01 5.01568139e-01
-1.61109006e+00 -8.15887153e-01 4.52614337e-01 3.69478405e-01
-8.89320970e-01 -4.23533320e-01 8.47186327e-01 -4.41092163e-01
3.62500459e-01 6.25342488e-01 1.16949642e+00 9.58979487e-01
-3.62674743e-01 1.44919133e+00 1.47262013e+00 -7.90071785e-01
4.23267096e-01 -8.80928244e-03 5.28272450e-01 5.51455319e-01
5.82005560e-01 7.69696236e-01 -9.77217019e-01 -3.21490198e-01
9.35359776e-01 2.93898195e-01 -2.52791852e-01 -6.77999675e-01
-5.45177579e-01 1.16030240e+00 6.65673614e-01 -5.36784492e-02
-3.36653173e-01 6.56020522e-01 6.91113770e-02 5.44928849e-01
9.78000104e-01 2.09182411e-01 -1.99080631e-01 -6.77472800e-02
-1.11948156e+00 4.12680924e-01 5.95873952e-01 1.04642773e+00
9.67974544e-01 -3.36250179e-02 -3.50830436e-01 8.48487020e-01
4.17597979e-01 4.13841128e-01 2.37633869e-01 -6.99184239e-01
7.84221470e-01 5.91270208e-01 3.20341766e-01 -6.01151526e-01
-1.62184983e-01 -7.13366628e-01 -5.14706016e-01 -3.81872691e-02
6.38451278e-01 -2.49159142e-01 -9.15923655e-01 1.36896062e+00
4.58293945e-01 2.95838296e-01 -1.26545861e-01 7.50123978e-01
1.09654345e-01 5.20539284e-01 -1.09368362e-01 -5.94745517e-01
7.34487295e-01 -1.02632737e+00 -8.83705854e-01 -6.35144413e-01
5.67463934e-01 -2.87112117e-01 9.72376406e-01 3.85429233e-01
-1.16708362e+00 4.34894487e-02 -1.10347641e+00 2.07100913e-01
-2.55842268e-01 -4.99726504e-01 1.23610246e+00 1.17339361e+00
-9.91781235e-01 5.74251354e-01 -7.67788470e-01 1.26794288e-02
7.34647155e-01 4.77312267e-01 1.72650427e-01 -2.12619767e-01
-1.04706442e+00 6.05704486e-01 5.37650108e-01 1.69839010e-01
-3.71121526e-01 -9.44732666e-01 -7.10441351e-01 5.58061451e-02
7.31242776e-01 -5.54595590e-01 1.65224481e+00 -1.04095745e+00
-1.71553719e+00 2.63464242e-01 -7.12969601e-02 -1.04871953e+00
8.97501111e-01 -5.97787440e-01 2.50837713e-01 -3.10291320e-01
-2.79335350e-01 7.91703686e-02 9.26029980e-01 -9.09869313e-01
-8.01126361e-01 -5.26910603e-01 5.13847768e-02 1.19020611e-01
-3.15464348e-01 -3.89880478e-01 -2.96896905e-01 -7.65379667e-01
9.94298533e-02 -1.15359211e+00 -3.53311807e-01 -4.88322884e-01
-3.15999180e-01 -1.31409749e-01 6.80466712e-01 -2.38867253e-01
1.61935961e+00 -1.97426355e+00 -2.06410006e-01 4.45316613e-01
-1.51395842e-01 1.09737694e-01 1.96825117e-01 4.58278924e-01
1.08370893e-01 2.20612541e-01 -1.00533113e-01 9.26730260e-02
7.98704475e-02 3.04441631e-01 -5.28544843e-01 6.01375103e-01
-1.55823722e-01 1.08095002e+00 -8.93696010e-01 -1.54832527e-01
2.13503793e-01 -8.06620717e-02 -9.88345683e-01 3.61187826e-03
-3.73515606e-01 1.54272512e-01 -6.32803679e-01 5.64765871e-01
3.97179306e-01 -3.44992429e-01 2.68291891e-01 6.77099109e-01
-1.15419313e-01 4.22766089e-01 -1.06606281e+00 1.20672107e+00
-2.91787684e-01 7.89663196e-01 -1.92775447e-02 -1.53847122e+00
6.78370893e-01 1.43267959e-01 3.77856433e-01 -7.10061252e-01
-1.14016041e-01 1.41183630e-01 -2.51125306e-01 -1.78755268e-01
3.56421530e-01 -1.67816147e-01 7.42095113e-02 5.12724578e-01
-5.79870343e-02 2.15772986e-02 2.34671310e-01 -1.58037648e-01
8.91146362e-01 -7.53913745e-02 6.63273335e-01 -4.54680234e-01
-3.77579443e-02 -1.34095475e-01 2.04830647e-01 1.46108770e+00
-1.55595019e-01 8.35523903e-02 3.91062766e-01 -6.92463994e-01
-6.49961531e-01 -8.01444173e-01 -5.62763773e-02 1.53340685e+00
3.18171047e-02 -1.72463983e-01 -5.90114295e-01 -6.32942438e-01
4.34507161e-01 5.38307607e-01 -9.62183535e-01 -1.14412323e-01
-4.57385302e-01 -5.75942457e-01 1.21195674e-01 3.32275063e-01
2.92337745e-01 -7.22754478e-01 -4.18155342e-01 3.57665062e-01
8.64497677e-04 -6.09441578e-01 -5.59057415e-01 5.69505095e-01
-1.33796024e+00 -1.05972767e+00 -6.76776946e-01 -1.46116242e-01
2.01186463e-01 4.87297952e-01 9.11984742e-01 -2.26742644e-02
-5.41234352e-02 3.93127471e-01 -2.99245328e-01 -7.91478932e-01
-1.23099856e-01 1.99733600e-01 -3.65387708e-01 7.63016343e-02
2.41032466e-01 -3.90612036e-01 -9.13831115e-01 5.71472682e-02
-9.18358982e-01 3.10090370e-02 5.39442539e-01 1.16971540e+00
4.75431055e-01 -5.55791080e-01 6.27784371e-01 -1.31481075e+00
7.86286533e-01 -4.41927612e-01 -1.01678097e+00 2.88122416e-01
-9.17101681e-01 3.66923064e-01 2.08724380e-01 -4.40908670e-01
-9.99438763e-01 -2.11249180e-02 3.03546697e-01 -1.69615254e-01
3.65938574e-01 8.76488447e-01 6.77936137e-01 -2.69502461e-01
8.01958859e-01 2.47489899e-01 -3.33745837e-01 -2.90510714e-01
6.19796813e-01 4.62923884e-01 9.88703817e-02 -5.81314206e-01
3.78878951e-01 3.23256224e-01 -9.01013985e-02 -7.25083709e-01
-1.39656079e+00 -5.66245079e-01 -1.85733467e-01 -2.67787039e-01
9.95719954e-02 -7.00551748e-01 -6.52311742e-01 -8.53977203e-02
-7.47146189e-01 -6.38034701e-01 -3.64732087e-01 6.28441751e-01
-1.14162302e+00 -1.17050819e-01 -5.16096830e-01 -1.40504384e+00
-2.12454468e-01 -7.15788484e-01 5.07483125e-01 2.92860270e-01
-4.63281339e-03 -1.01321673e+00 2.51386762e-01 3.92595232e-01
1.97595999e-01 -2.58361478e-03 7.15910852e-01 -5.73811054e-01
-1.01321971e+00 -1.35664657e-01 -2.29541153e-01 -1.81515701e-03
2.39786711e-02 -2.44555354e-01 -8.91779184e-01 -3.10594559e-01
-2.35367045e-01 -2.75637209e-01 1.24676538e+00 1.08518815e+00
1.10893476e+00 -7.62056828e-01 -2.42774650e-01 6.23975515e-01
1.39281750e+00 3.18913162e-01 3.21112841e-01 4.69454646e-01
2.31785864e-01 4.08657968e-01 5.56806684e-01 6.29877210e-01
-8.04584920e-02 4.85790730e-01 3.96431446e-01 4.80749086e-02
1.21528700e-01 -6.44158006e-01 2.09733725e-01 6.09720886e-01
-2.16754019e-01 8.45447853e-02 -6.90277696e-01 5.13949931e-01
-2.38349462e+00 -1.13143480e+00 1.11757115e-01 2.25672507e+00
1.04607844e+00 1.63828328e-01 2.83157468e-01 2.30973110e-01
7.80527651e-01 2.41070196e-01 -8.07178617e-01 -4.23700154e-01
4.28897850e-02 2.08152011e-01 1.03555715e+00 5.46475589e-01
-1.13741720e+00 9.94672596e-01 8.41208172e+00 8.57680082e-01
-8.63019764e-01 -6.88644266e-03 7.43675828e-01 -3.39984626e-01
-1.48635283e-01 -3.69711928e-02 -8.08510184e-01 3.51864457e-01
1.02045906e+00 -2.78455645e-01 7.99060881e-01 9.31668043e-01
1.64248258e-01 -1.05932832e-01 -1.13501823e+00 1.00800705e+00
-5.50950289e-01 -1.86261773e+00 -1.46031812e-01 1.28240496e-01
1.16625571e+00 1.87226057e-01 3.67397189e-01 4.30909127e-01
9.74457383e-01 -7.27663040e-01 7.51106262e-01 3.70230764e-01
6.07148230e-01 -8.25033605e-01 2.79555947e-01 5.17262757e-01
-7.93581307e-01 -4.74747151e-01 -2.78087705e-01 -3.93445253e-01
-3.73977512e-01 4.44722652e-01 -7.05751419e-01 4.02281642e-01
6.42572522e-01 7.08603442e-01 -2.75038749e-01 1.31906068e+00
-1.78997368e-01 1.00666595e+00 -3.72478366e-01 -4.59669083e-01
4.78581041e-01 -4.53337967e-01 2.93030947e-01 1.27098560e+00
3.00109237e-01 -4.64116503e-03 2.93196917e-01 3.78820091e-01
-1.15524828e-01 2.52195090e-01 -8.55892122e-01 -1.77468918e-02
5.46961665e-01 4.10061747e-01 -4.26006883e-01 -3.33180189e-01
-6.06302857e-01 2.65924662e-01 7.18942761e-01 7.57878065e-01
-5.60491979e-01 1.81871548e-01 6.45261586e-01 -5.83251379e-02
5.04472673e-01 5.65735511e-02 -7.05802977e-01 -9.51687574e-01
-4.40594517e-02 -6.88375890e-01 8.36631715e-01 -1.60233393e-01
-1.20858824e+00 -3.02045733e-01 9.36903134e-02 -8.87077034e-01
-1.92782372e-01 -4.22951698e-01 -1.99493870e-01 3.84561896e-01
-1.73417437e+00 -8.17370653e-01 2.36736968e-01 6.86050832e-01
5.73786199e-01 8.14312398e-02 4.37215865e-01 -1.50164450e-02
-5.28232157e-01 6.77636385e-01 8.15502405e-01 -2.76772588e-01
3.57261688e-01 -1.40828145e+00 3.97225618e-02 6.47864103e-01
4.33990687e-01 7.29128122e-01 8.34608376e-01 -4.14886355e-01
-1.56424689e+00 -9.19837654e-01 2.29650185e-01 -1.09408936e-02
1.03686404e+00 -1.66633368e-01 -6.33948326e-01 8.23982298e-01
-1.52806835e-02 5.75406589e-02 8.80220294e-01 7.58400023e-01
-6.14281535e-01 -3.43004197e-01 -7.03624547e-01 7.42733538e-01
1.08424389e+00 -5.96291423e-01 -3.23129237e-01 3.30852389e-01
6.09954119e-01 -3.54381382e-01 -5.75687528e-01 1.99730814e-01
7.89753616e-01 -1.13598335e+00 8.86325002e-01 -7.55068719e-01
1.94295287e-01 3.80322933e-01 -7.34371990e-02 -1.44677818e+00
-2.54033804e-01 -1.25671244e+00 -6.64900720e-01 7.10849047e-01
5.11155903e-01 -8.82590353e-01 1.33706832e+00 4.52126294e-01
1.21243097e-01 -8.11208546e-01 -1.13128960e+00 -8.72585416e-01
1.56179398e-01 -7.10622668e-01 4.43447977e-01 6.30123079e-01
5.75079679e-01 1.34267285e-01 -7.04820275e-01 -3.50588053e-01
6.42323196e-01 3.97268772e-01 7.42403924e-01 -1.24909687e+00
-4.15855467e-01 -8.56918752e-01 -3.93771529e-02 -1.56562674e+00
2.79916413e-02 -7.29789853e-01 4.15422507e-02 -1.20519876e+00
3.84116918e-01 -5.78174651e-01 -8.70159566e-01 1.81683302e-01
-4.13402081e-01 1.13934480e-01 9.93849561e-02 1.21088512e-01
-4.61138517e-01 3.67199123e-01 1.20127988e+00 -3.14583272e-01
-4.66431946e-01 5.30760467e-01 -9.36244488e-01 4.48866129e-01
8.16919267e-01 -5.85833848e-01 -5.38807213e-01 -1.05125554e-01
3.75003189e-01 3.97569984e-01 1.41920269e-01 -4.68060613e-01
4.03297156e-01 -6.08305871e-01 1.00804381e-01 -6.27424657e-01
1.82855695e-01 -7.70314038e-01 -1.29634947e-01 6.47572875e-01
-8.93577576e-01 -2.25409493e-01 2.27794319e-01 1.26451159e+00
-1.86463192e-01 -1.80686891e-01 7.08406925e-01 2.94215847e-02
-1.36531770e-01 4.38000619e-01 -3.36899698e-01 8.62827823e-02
9.64183033e-01 -3.33197951e-01 -3.57950449e-01 -6.52467668e-01
-5.15474319e-01 9.21188220e-02 -9.82031003e-02 -4.60921694e-03
4.15107101e-01 -1.33581924e+00 -5.46600699e-01 1.45180345e-01
-2.58502867e-02 -3.08194458e-01 5.70224002e-02 7.41137624e-01
-2.44389221e-01 1.04659843e+00 2.23818034e-01 -6.28970563e-01
-1.05181015e+00 6.47055626e-01 1.26061007e-01 -5.48310876e-01
-4.03315216e-01 8.82319450e-01 2.19202891e-01 -1.11159720e-01
4.19275403e-01 -4.03663993e-01 5.68740703e-02 1.49592191e-01
6.05160892e-01 5.38170576e-01 1.63672164e-01 2.54782587e-01
1.20278597e-01 7.41350949e-02 -1.58909246e-01 -2.16184869e-01
1.32008064e+00 -7.71780536e-02 2.20174536e-01 6.66267335e-01
8.75888884e-01 -1.02272734e-01 -1.77302861e+00 -6.85873449e-01
8.17411721e-01 -7.54138589e-01 3.72829854e-01 -4.40953016e-01
-7.51993597e-01 4.67558652e-01 4.01118040e-01 1.00316155e+00
7.99795508e-01 -8.53607804e-02 3.50847214e-01 5.10996342e-01
3.11287552e-01 -1.07874084e+00 -9.54087973e-02 5.24531662e-01
5.97562015e-01 -1.24410105e+00 2.53758073e-01 -2.60867685e-01
-1.23077482e-01 9.98443902e-01 5.84413595e-02 -4.79353905e-01
7.13041127e-01 -9.02554672e-03 -9.35237184e-02 2.77031977e-02
-1.02122796e+00 -3.13274652e-01 4.81722444e-01 4.71428096e-01
3.82233918e-01 3.08644146e-01 -4.91127014e-01 1.95709705e-01
-2.66261727e-01 1.16079599e-02 1.85633108e-01 8.24069440e-01
-6.24865949e-01 -9.26427186e-01 -5.34955144e-01 6.40671372e-01
-6.97654486e-01 -1.63688198e-01 -4.48566616e-01 8.81563962e-01
-7.33245254e-01 1.09283698e+00 -1.10563733e-01 4.29882072e-02
8.50927681e-02 1.08112186e-01 7.13315785e-01 -4.28239137e-01
-4.13907200e-01 3.52960914e-01 6.77562207e-02 -3.90671372e-01
-5.83533704e-01 -6.72220409e-01 -3.82130295e-01 -3.55378091e-01
-8.10472906e-01 3.08897406e-01 5.63661754e-01 1.08800340e+00
3.70644517e-02 2.68738061e-01 6.09233379e-01 -7.19987094e-01
-8.77012432e-01 -9.58961904e-01 -6.44256234e-01 9.85570773e-02
9.47242856e-01 -6.61723375e-01 -1.22451738e-01 -1.59237966e-01] | [4.510188102722168, 3.2366297245025635] |
8c99bd3a-9094-4074-b87a-ef8799815d3a | self-supervised-learning-of-depth-and-ego | 1909.13163 | null | https://arxiv.org/abs/1909.13163v1 | https://arxiv.org/pdf/1909.13163v1.pdf | Self-Supervised Learning of Depth and Ego-motion with Differentiable Bundle Adjustment | Learning to predict scene depth and camera motion from RGB inputs only is a challenging task. Most existing learning based methods deal with this task in a supervised manner which require ground-truth data that is expensive to acquire. More recent approaches explore the possibility of estimating scene depth and camera pose in a self-supervised learning framework. Despite encouraging results are shown, current methods either learn from monocular videos for depth and pose and typically do so without enforcing multi-view geometry constraints between scene structure and camera motion, or require stereo sequences as input where the ground-truth between-frame motion parameters need to be known. In this paper we propose to jointly optimize the scene depth and camera motion via incorporating differentiable Bundle Adjustment (BA) layer by minimizing the feature-metric error, and then form the photometric consistency loss with view synthesis as the final supervisory signal. The proposed approach only needs unlabeled monocular videos as input, and extensive experiments on the KITTI and Cityscapes dataset show that our method achieves state-of-the-art results in self-supervised approaches using monocular videos as input, and even gains advantage to the line of methods that learns from calibrated stereo sequences (i.e. with pose supervision). | ['KuoChin Lien', 'Yi Fang', 'Jing Zhu', 'Junli Gu', 'Yunxiao Shi'] | 2019-09-28 | null | null | null | null | ['depth-and-camera-motion'] | ['computer-vision'] | [ 1.68854102e-01 3.91858853e-02 -2.90763289e-01 -6.48269951e-01
-8.35349083e-01 -8.40546787e-01 5.73606372e-01 -5.80937207e-01
-4.27919269e-01 5.92978179e-01 -6.78739622e-02 1.22558802e-01
3.36769044e-01 -4.55138922e-01 -1.14387310e+00 -8.64211619e-01
4.84420031e-01 4.86024827e-01 1.57301232e-01 1.88286364e-01
2.99382329e-01 3.13998282e-01 -1.56708658e+00 -2.95513310e-02
4.99416918e-01 8.37535799e-01 3.11209947e-01 8.08622539e-01
1.64498404e-01 1.05719590e+00 6.85961843e-02 -2.59894997e-01
8.12213242e-01 -4.84271169e-01 -8.53606522e-01 8.45231652e-01
1.13533413e+00 -7.29896367e-01 -5.90552032e-01 1.04376841e+00
3.66332561e-01 -2.32345164e-02 3.02519739e-01 -1.15050220e+00
1.60668101e-02 -1.27499804e-01 -5.12383044e-01 -1.31268576e-01
6.48213685e-01 2.87349850e-01 9.67184007e-01 -9.50186610e-01
9.18513060e-01 9.64857101e-01 4.75595891e-01 5.24058938e-01
-1.31439662e+00 -2.76876003e-01 2.46139631e-01 3.20453465e-01
-1.27405560e+00 -6.68599427e-01 1.07541883e+00 -6.41344845e-01
6.83940589e-01 -1.33923590e-01 6.68556809e-01 1.05135798e+00
-5.51249757e-02 8.49773645e-01 1.23064613e+00 -3.82034868e-01
2.09319949e-01 2.06543550e-01 -4.00418520e-01 8.50510120e-01
-4.31146994e-02 2.26573989e-01 -6.99637055e-01 2.98671097e-01
1.25761938e+00 1.37818024e-01 -5.93521595e-01 -1.27996480e+00
-1.42036378e+00 7.35430181e-01 4.04000849e-01 -2.91463267e-02
-6.89717159e-02 1.37424961e-01 4.89180833e-02 4.12147194e-01
4.00952905e-01 1.43893078e-01 -6.13424599e-01 3.44066545e-02
-9.19477820e-01 2.43144501e-02 6.37813985e-01 1.25961685e+00
1.33514011e+00 4.12373021e-02 6.26111567e-01 4.24513072e-01
2.95923620e-01 7.18517959e-01 4.41111028e-01 -1.49394095e+00
7.98329234e-01 4.55852866e-01 2.98973620e-01 -8.13758910e-01
-2.76596218e-01 -5.21892309e-02 -5.78029156e-01 3.03889573e-01
6.23830378e-01 -8.02309811e-02 -6.67828023e-01 1.55946195e+00
5.80874622e-01 3.02936971e-01 1.38443425e-01 1.17225707e+00
5.81364036e-01 3.06569844e-01 -7.84437776e-01 -2.34676853e-01
6.19439423e-01 -1.13132524e+00 -4.73966300e-01 -5.46511889e-01
6.08002603e-01 -8.16972852e-01 9.97717798e-01 5.30513167e-01
-1.12060142e+00 -6.00145638e-01 -9.90626574e-01 -3.32215548e-01
7.11311772e-02 2.33070567e-01 4.54921335e-01 5.76834679e-01
-1.09171104e+00 5.87966800e-01 -1.09478045e+00 -4.39369678e-01
-6.82041198e-02 3.95473331e-01 -7.01876640e-01 -3.26365411e-01
-6.38016164e-01 7.33734727e-01 3.23316991e-01 7.65684769e-02
-1.04017687e+00 -5.12608528e-01 -1.36000478e+00 -4.94451851e-01
5.14340401e-01 -9.11667466e-01 1.11152673e+00 -1.17761385e+00
-1.89834237e+00 1.12143445e+00 -1.68348372e-01 -1.70706272e-01
8.87658060e-01 -4.95422304e-01 3.37671906e-01 4.47005540e-01
-3.34775485e-02 9.34046447e-01 9.39323962e-01 -1.46229529e+00
-6.64977372e-01 -5.26074708e-01 4.56511647e-01 6.04882002e-01
-4.74641062e-02 -6.20586395e-01 -6.74640119e-01 -4.72784564e-02
6.19930208e-01 -1.18557477e+00 -1.51766226e-01 3.11897486e-01
-3.00141066e-01 4.10310328e-01 8.43594313e-01 -5.63054800e-01
4.59609032e-01 -1.74845743e+00 5.64785719e-01 -1.59362212e-01
-1.82259262e-01 -2.22169206e-01 6.84045628e-02 1.18881941e-01
2.26919428e-02 -6.40838027e-01 -2.15108410e-01 -6.93661034e-01
-3.68645340e-01 4.80589151e-01 -1.52369335e-01 1.00753045e+00
-2.15183757e-02 6.47182167e-01 -1.03440046e+00 -4.36651766e-01
9.08614993e-01 4.08952683e-01 -8.09438229e-01 6.07568979e-01
-2.23333150e-01 1.10323489e+00 -9.27915946e-02 5.14104068e-01
5.95235765e-01 -3.32413018e-01 1.26147702e-01 -3.52170974e-01
-1.82732284e-01 2.26694047e-01 -1.46538806e+00 2.36098313e+00
-6.09632373e-01 6.92978501e-01 2.30217859e-01 -9.41283345e-01
6.26086950e-01 9.79898348e-02 6.34823561e-01 -2.84851581e-01
7.90592507e-02 1.72278464e-01 -4.45182204e-01 -5.22873700e-01
2.29754433e-01 -1.69781223e-01 3.10279548e-01 1.05820149e-01
2.18173102e-01 -8.08680177e-01 -1.73804149e-01 -6.98564516e-04
8.08796883e-01 9.13393497e-01 3.33698243e-01 9.36892480e-02
8.14435065e-01 -3.40811424e-02 6.53838038e-01 2.18439594e-01
1.10748000e-01 9.94741142e-01 6.79095462e-02 -4.06812400e-01
-1.30284214e+00 -9.54597414e-01 -2.10999954e-03 3.77416253e-01
4.03121799e-01 -4.85754982e-02 -7.02784836e-01 -6.10264719e-01
-2.59881616e-01 3.30657095e-01 -3.76155376e-01 3.64848852e-01
-6.17816329e-01 -2.40302935e-01 -1.37630135e-01 4.47282106e-01
7.20843017e-01 -4.29862440e-01 -8.57198894e-01 -1.17424894e-02
-3.65949601e-01 -1.78860176e+00 -5.86107552e-01 2.07123294e-01
-1.05933011e+00 -1.16280317e+00 -7.33651042e-01 -6.33444607e-01
7.61698842e-01 5.77240169e-01 8.27822685e-01 -3.25337738e-01
-7.79373795e-02 7.59456933e-01 -7.78274611e-02 1.67936429e-01
-2.25982387e-02 -8.50781947e-02 1.14527911e-01 2.95819640e-01
-9.42015648e-03 -7.09469318e-01 -6.46935463e-01 5.89930594e-01
-7.43375301e-01 3.58529270e-01 3.56451422e-01 9.74306166e-01
8.09823394e-01 -3.81719321e-01 -1.70072958e-01 -8.81446183e-01
-8.06552708e-01 -1.43090174e-01 -1.10333514e+00 -1.23947285e-01
-4.33971852e-01 -4.11862582e-02 6.25180304e-01 -4.62923758e-02
-1.09299195e+00 9.75803316e-01 1.49226546e-01 -1.07396150e+00
-2.16565534e-01 -5.65369315e-02 -5.91624260e-01 -3.54771882e-01
4.85308647e-01 3.90985727e-01 6.64094314e-02 -3.45390350e-01
4.09464002e-01 4.04935628e-01 7.23851919e-01 -2.77740747e-01
1.12428570e+00 1.17699873e+00 1.98919252e-01 -8.60221326e-01
-1.19441068e+00 -9.62080836e-01 -1.47163033e+00 -2.82002479e-01
1.00403118e+00 -1.51465213e+00 -4.86433744e-01 5.01507878e-01
-1.13460076e+00 -3.75730127e-01 3.02018994e-03 7.35594928e-01
-1.21003568e+00 6.86044872e-01 -4.64724422e-01 -5.97665071e-01
1.65167674e-01 -1.14664137e+00 1.51345766e+00 -1.32090151e-01
1.72434762e-01 -1.25669718e+00 6.57933429e-02 6.70774102e-01
-2.38702461e-01 4.80771750e-01 2.18835354e-01 2.69913882e-01
-1.23750842e+00 9.56344903e-02 4.27703597e-02 6.23284519e-01
3.29814047e-01 -4.18969393e-01 -1.27567959e+00 -4.83166665e-01
4.13092732e-01 -3.88887316e-01 5.18288970e-01 3.63431662e-01
8.82942736e-01 -1.11089714e-01 7.44502023e-02 1.20698977e+00
1.81516182e+00 -7.66776130e-02 5.02651513e-01 5.08156598e-01
1.28100491e+00 8.44403088e-01 7.53668070e-01 3.41292769e-01
6.36103690e-01 9.23795462e-01 7.04529464e-01 -5.09585552e-02
-1.17025105e-02 -4.34420675e-01 5.50668836e-01 9.53257263e-01
9.70410183e-02 1.24185152e-01 -7.92595267e-01 5.21402717e-01
-1.93397868e+00 -8.93801332e-01 -1.27956510e-01 2.33537841e+00
7.54027188e-01 -4.86661792e-02 -3.10299303e-02 3.07378680e-01
3.50357533e-01 1.59813911e-01 -9.13923562e-01 1.67265296e-01
-3.95381898e-01 -4.47449714e-01 8.97328675e-01 8.67534876e-01
-1.18071461e+00 1.02412522e+00 5.42716646e+00 -2.94729113e-03
-1.29180074e+00 1.70305315e-02 4.77813244e-01 -3.15860659e-01
-2.05444768e-01 1.71983510e-01 -7.71675706e-01 6.66372553e-02
4.69272524e-01 2.58832872e-01 6.15172684e-01 8.36869717e-01
2.51509726e-01 -3.23291510e-01 -1.62919724e+00 1.64377487e+00
4.29790080e-01 -1.09853935e+00 -1.89538628e-01 6.84211850e-02
1.12358916e+00 1.82396486e-01 -9.79639068e-02 -3.67558628e-01
1.15942411e-01 -5.75998485e-01 8.00209165e-01 3.99010777e-01
7.57333934e-01 -4.36672151e-01 6.07656240e-01 5.79056382e-01
-9.81208682e-01 1.29292021e-02 -3.90569657e-01 -2.77913243e-01
3.48821849e-01 4.70508009e-01 -6.54700220e-01 6.94692075e-01
7.52038777e-01 1.39350641e+00 -4.64880109e-01 8.22569907e-01
-4.03097183e-01 2.17318565e-01 -3.69352579e-01 5.74509442e-01
1.53463796e-01 -5.47485828e-01 4.78705257e-01 6.72391117e-01
1.72226340e-01 5.63099980e-02 3.12894106e-01 5.52897573e-01
1.20375402e-01 -2.45654676e-02 -7.93770969e-01 4.40316767e-01
-4.40371484e-02 1.10160935e+00 -5.11666417e-01 -1.54189408e-01
-6.86291754e-01 1.40504086e+00 1.97668016e-01 2.54835695e-01
-5.30030429e-01 2.53111303e-01 5.96681595e-01 2.39665657e-01
3.95521820e-01 -5.31623244e-01 -3.39701176e-01 -1.68164504e+00
2.83534348e-01 -6.56332612e-01 1.40372917e-01 -1.05739450e+00
-9.62329626e-01 2.90858358e-01 -9.34372004e-03 -1.72985590e+00
-6.48603499e-01 -6.81092083e-01 -1.89252466e-01 5.46298325e-01
-1.86758184e+00 -1.13671696e+00 -6.21370018e-01 8.74462903e-01
8.03990602e-01 5.52655347e-02 3.90786171e-01 2.25648001e-01
-1.57432675e-01 2.64285594e-01 2.12376654e-01 -7.02212900e-02
8.56113732e-01 -1.38680267e+00 1.42738268e-01 7.28282690e-01
3.29976887e-01 1.90009788e-01 7.06559300e-01 -1.83436126e-01
-1.89160407e+00 -1.03896940e+00 6.46526873e-01 -9.19052243e-01
4.60106999e-01 -4.54744041e-01 -5.60439408e-01 9.00091052e-01
1.01478636e-01 2.03380257e-01 2.17219755e-01 -3.93110126e-01
-2.23515913e-01 -2.38987520e-01 -9.78558362e-01 3.48440230e-01
1.33484840e+00 -7.23523438e-01 -3.47841352e-01 4.32138473e-01
5.59646964e-01 -8.50387394e-01 -6.73944831e-01 1.22453935e-01
3.14481854e-01 -1.23150110e+00 9.43415403e-01 -2.12022290e-01
4.74205345e-01 -6.66729808e-01 -5.09927154e-01 -1.24488652e+00
2.39844248e-01 -7.52818644e-01 3.49107236e-02 1.02128422e+00
8.17168280e-02 -2.61512429e-01 1.16177285e+00 4.65300143e-01
-5.58690764e-02 -4.41177547e-01 -8.90976191e-01 -8.50290239e-01
-1.64280698e-01 -3.18116069e-01 1.47602737e-01 1.00565600e+00
-3.26513052e-01 4.97755200e-01 -7.05402493e-01 4.35706943e-01
1.07960331e+00 2.83485621e-01 1.33169496e+00 -9.12278473e-01
-5.40827155e-01 1.64420485e-01 -7.19027162e-01 -1.51256227e+00
3.78793865e-01 -5.87620735e-01 1.92352891e-01 -1.22134781e+00
5.35851344e-02 -4.62421216e-02 4.22706276e-01 1.08680308e-01
1.07783593e-01 1.92872196e-01 1.16931446e-01 3.34545732e-01
-5.06091535e-01 6.20325387e-01 1.20972407e+00 2.11736746e-02
-1.51351988e-01 -4.72105853e-02 2.28178017e-02 1.01921117e+00
3.85283977e-01 -9.40302983e-02 -7.30619729e-01 -8.20127785e-01
1.20411076e-01 4.77978915e-01 5.67479134e-01 -9.43872154e-01
3.72850627e-01 -2.93960005e-01 4.59327787e-01 -6.13661647e-01
7.36837804e-01 -1.07185113e+00 1.72124967e-01 2.71670640e-01
-9.01567116e-02 9.21759605e-02 -2.30340526e-01 7.32440889e-01
-2.66783863e-01 -1.09313298e-02 9.40887511e-01 -2.62996942e-01
-8.43929946e-01 4.31289196e-01 1.46588549e-01 2.31246185e-02
9.41339910e-01 -4.64433253e-01 1.97576836e-01 -7.00562298e-01
-6.78291798e-01 9.95067805e-02 1.08717394e+00 4.35068488e-01
6.49963140e-01 -1.26359308e+00 -4.65825409e-01 3.22986215e-01
1.87564716e-01 5.67279339e-01 4.36143056e-02 5.81016064e-01
-7.37445652e-01 5.30907691e-01 -1.58365056e-01 -1.27743471e+00
-1.24586916e+00 5.49472809e-01 4.77461696e-01 1.57828867e-01
-6.94949031e-01 7.23828733e-01 4.99235064e-01 -6.53127491e-01
3.32702875e-01 -2.86488861e-01 2.46472821e-01 -2.25669846e-01
2.46804625e-01 2.95367360e-01 6.23096786e-02 -9.98681366e-01
-1.85211539e-01 1.13160050e+00 2.07025945e-01 -3.39583486e-01
1.31639075e+00 -7.00059891e-01 1.66480333e-01 6.64048493e-01
1.59227645e+00 -1.86688632e-01 -2.06283879e+00 -4.78955388e-01
-1.35168836e-01 -9.06732321e-01 1.51681453e-01 -2.15337187e-01
-1.20302820e+00 1.03663063e+00 5.22470057e-01 -6.43914163e-01
1.03431118e+00 -5.99366762e-02 6.30845308e-01 5.65426469e-01
8.68771136e-01 -1.04352450e+00 2.32797027e-01 4.46562976e-01
6.69077158e-01 -1.69900727e+00 7.39222988e-02 -6.28260255e-01
-4.90935713e-01 1.08771729e+00 7.75451899e-01 -2.72401214e-01
5.61715364e-01 -5.15523404e-02 3.13511878e-01 1.93557695e-01
-5.68862438e-01 -1.62669793e-01 1.84795037e-01 5.23068666e-01
2.80161828e-01 -3.19289178e-01 1.97250634e-01 -3.17516774e-01
-2.94963419e-01 -1.22983605e-01 7.46760368e-01 9.57546532e-01
-2.62228340e-01 -1.13679600e+00 -4.61662233e-01 -5.51108904e-02
3.26086581e-03 1.05991706e-01 -3.51061910e-01 6.64091170e-01
-7.62803014e-03 9.03499484e-01 -7.74695352e-02 -3.00226688e-01
2.97378778e-01 -1.82068855e-01 9.54047740e-01 -7.44211376e-01
-1.43525913e-01 3.37503612e-01 -1.33933872e-01 -9.69790101e-01
-8.74930263e-01 -1.07287693e+00 -1.05628538e+00 1.77332778e-02
-2.54992962e-01 -2.93131500e-01 6.90795660e-01 1.04745197e+00
-9.11072567e-02 -9.34295803e-02 1.06303596e+00 -1.37245226e+00
-4.34300065e-01 -5.03817320e-01 -4.46803063e-01 5.72420180e-01
7.79633045e-01 -6.08453095e-01 -6.05601132e-01 5.96478879e-01] | [8.580086708068848, -2.419633388519287] |
90c04a6a-dcb2-4889-9e9a-82042404e85e | adaptive-adversarial-training-method-for | 2211.16791 | null | https://arxiv.org/abs/2211.16791v1 | https://arxiv.org/pdf/2211.16791v1.pdf | Adaptive adversarial training method for improving multi-scale GAN based on generalization bound theory | In recent years, multi-scale generative adversarial networks (GANs) have been proposed to build generalized image processing models based on single sample. Constraining on the sample size, multi-scale GANs have much difficulty converging to the global optimum, which ultimately leads to limitations in their capabilities. In this paper, we pioneered the introduction of PAC-Bayes generalized bound theory into the training analysis of specific models under different adversarial training methods, which can obtain a non-vacuous upper bound on the generalization error for the specified multi-scale GAN structure. Based on the drastic changes we found of the generalization error bound under different adversarial attacks and different training states, we proposed an adaptive training method which can greatly improve the image manipulation ability of multi-scale GANs. The final experimental results show that our adaptive training method in this paper has greatly contributed to the improvement of the quality of the images generated by multi-scale GANs on several image manipulation tasks. In particular, for the image super-resolution restoration task, the multi-scale GAN model trained by the proposed method achieves a 100% reduction in natural image quality evaluator (NIQE) and a 60% reduction in root mean squared error (RMSE), which is better than many models trained on large-scale datasets. | ['Zhouping Yin', 'Zeyu Gong', 'Bo Tao', 'Jing Tang'] | 2022-11-30 | null | null | null | null | ['image-manipulation'] | ['computer-vision'] | [ 4.72692549e-01 2.26792153e-02 1.16766535e-01 -7.24456012e-02
-1.19798005e+00 -3.81040961e-01 3.31027657e-01 -7.80548334e-01
-6.26425073e-02 8.33740413e-01 3.56467590e-02 1.00719601e-01
6.29202873e-02 -1.00238645e+00 -7.24048913e-01 -1.13085103e+00
3.01739305e-01 1.89789962e-02 7.27147013e-02 -2.43720695e-01
8.35802704e-02 5.23181617e-01 -1.24687278e+00 2.43955538e-01
1.14163637e+00 1.16869891e+00 1.98240399e-01 8.58654261e-01
2.80512720e-01 7.24219084e-01 -9.74078774e-01 -6.12657487e-01
4.12688971e-01 -9.41269696e-01 -4.76107895e-01 -6.20631799e-02
4.87476110e-01 -5.14780164e-01 -2.71953791e-01 1.37214065e+00
7.98890471e-01 -6.09996021e-02 6.36228800e-01 -1.26515579e+00
-1.12245977e+00 3.49220306e-01 -6.04677498e-01 1.95811361e-01
3.44565995e-02 2.49631152e-01 6.62474990e-01 -3.91824245e-01
3.37672323e-01 1.28717291e+00 5.70699036e-01 9.38665390e-01
-1.17856622e+00 -9.12423313e-01 -2.69648373e-01 1.28551617e-01
-1.27461052e+00 -2.26654589e-01 8.25548172e-01 -1.67178810e-01
4.07907307e-01 4.42198187e-01 4.72453028e-01 1.21535671e+00
5.83647847e-01 5.48029304e-01 1.55850697e+00 -3.95526677e-01
2.25430354e-01 8.11363105e-03 -7.54958212e-01 6.06295407e-01
1.12693897e-02 2.27199897e-01 -1.74142450e-01 5.84566891e-02
1.31424201e+00 -3.59656096e-01 -2.78930366e-01 3.61921154e-02
-9.44699466e-01 9.92916763e-01 6.41603589e-01 3.61774087e-01
-2.68637478e-01 2.25959957e-01 2.81120479e-01 3.92107755e-01
6.24641180e-01 4.30763364e-01 -2.20137998e-01 7.40533415e-03
-7.01937795e-01 -2.04980806e-01 1.87957630e-01 7.40144014e-01
4.25929099e-01 5.36390066e-01 -1.34149760e-01 8.99708569e-01
-2.43661515e-02 7.63556957e-01 6.48702085e-01 -1.03966272e+00
3.83178890e-01 1.91819727e-01 8.37045684e-02 -9.87602592e-01
1.16337925e-01 -5.98010838e-01 -1.36283052e+00 5.80068767e-01
1.71524867e-01 -1.67313248e-01 -1.02278924e+00 1.88967741e+00
3.09981376e-01 1.84833333e-01 2.22158328e-01 7.66017616e-01
5.62795937e-01 8.14342976e-01 4.20626625e-02 -4.47091699e-01
9.68351305e-01 -9.44129467e-01 -9.75329995e-01 -1.07665814e-01
1.27943695e-01 -7.80839145e-01 1.39164257e+00 5.82055092e-01
-1.12377298e+00 -9.04723525e-01 -1.29746544e+00 2.79359341e-01
9.16375816e-02 -5.76493815e-02 5.07256269e-01 1.17642677e+00
-9.64842916e-01 5.61738670e-01 -7.42850840e-01 -4.78553772e-02
7.07874477e-01 5.19774528e-03 -1.78531602e-01 -1.66720420e-01
-1.06988108e+00 8.02916229e-01 2.57358432e-01 2.60334641e-01
-1.05742216e+00 -5.86019814e-01 -4.99143392e-01 -1.23689875e-01
2.40953088e-01 -7.60876715e-01 6.83604479e-01 -1.21208048e+00
-1.91775060e+00 6.24136090e-01 1.77198559e-01 -2.87939727e-01
5.63740492e-01 -2.85114825e-01 -6.93804801e-01 3.09400618e-01
-9.22132749e-04 5.39115965e-01 1.23902655e+00 -1.42328918e+00
-2.69600809e-01 -2.20591933e-01 -3.49322297e-02 1.24031626e-01
-2.79283732e-01 1.91191658e-01 -5.29720895e-02 -1.03607881e+00
-3.85765284e-02 -8.59677732e-01 -2.83507586e-01 -5.24979606e-02
-1.12639256e-01 2.55063534e-01 7.18223572e-01 -8.53749335e-01
8.67363930e-01 -2.19296217e+00 3.99905264e-01 -1.88208655e-01
-2.22278789e-01 4.08421546e-01 -2.84477025e-01 1.57416180e-01
1.61258038e-02 4.19847101e-01 -3.06315035e-01 -1.16841070e-01
-3.55261564e-01 3.28896880e-01 -5.35803020e-01 3.73276025e-01
1.72513530e-01 9.60782111e-01 -6.87227190e-01 -3.79000098e-01
2.78660297e-01 5.27298152e-01 -4.62114871e-01 5.60664833e-01
1.05611287e-01 7.89881468e-01 -3.49204868e-01 5.51661372e-01
7.71832466e-01 -1.03461131e-01 -2.28040934e-01 -3.20664763e-01
5.26178837e-01 -4.23544347e-01 -1.11878514e+00 1.68900919e+00
-7.18127906e-01 4.02586967e-01 -1.68252010e-02 -8.94956172e-01
8.68278980e-01 3.63976240e-01 3.54006588e-01 -7.34891474e-01
9.40682292e-02 1.29447147e-01 -7.57818222e-02 -2.02762306e-01
1.69309884e-01 -5.24691999e-01 3.27686295e-02 1.06493540e-01
6.39363378e-02 -6.27960384e-01 -2.24493265e-01 1.69693995e-02
8.00082624e-01 2.48704087e-02 2.10665762e-01 -1.58530042e-01
6.14955902e-01 -4.72146988e-01 6.74153388e-01 7.51602590e-01
-1.45529002e-01 8.20625067e-01 2.61063337e-01 -2.92030990e-01
-1.38403344e+00 -1.14958799e+00 -1.12529203e-01 6.22809350e-01
1.54313505e-01 -1.16998948e-01 -9.53654945e-01 -5.89967966e-01
-6.21120930e-01 6.42631650e-01 -6.16839886e-01 -3.82585585e-01
-5.86075127e-01 -1.12092161e+00 6.73038661e-01 5.48472345e-01
1.07091939e+00 -1.21076024e+00 -2.41458893e-01 -1.95315778e-02
-2.40356043e-01 -1.40970159e+00 -4.01812315e-01 -1.83295891e-01
-8.89191151e-01 -8.70694578e-01 -8.08601379e-01 -5.32077491e-01
6.69748902e-01 -2.07085714e-01 8.17555010e-01 -7.00772777e-02
-1.27101138e-01 2.95218408e-01 -4.90328580e-01 -4.14783001e-01
-1.04226398e+00 -1.75687551e-01 1.04995016e-02 1.00922203e-02
-5.93753755e-01 -7.77177572e-01 -6.68739140e-01 5.74622750e-01
-1.13102150e+00 3.31592709e-02 7.68986404e-01 9.64828193e-01
6.68677032e-01 6.43180668e-01 1.01858699e+00 -6.31423175e-01
5.39471984e-01 -1.14507914e-01 -6.40375435e-01 3.22148532e-01
-7.49903440e-01 -2.35184818e-03 1.00192773e+00 -7.36042261e-01
-1.20403767e+00 -3.31024289e-01 -4.65180606e-01 -3.82268965e-01
-7.24468902e-02 1.07600257e-01 -5.38674355e-01 -4.37579095e-01
7.24509776e-01 4.87206846e-01 -2.28060372e-02 -1.71697184e-01
4.95815814e-01 4.85825688e-01 6.36922717e-01 -5.99031627e-01
1.05086899e+00 4.26659077e-01 2.50663072e-01 -6.34778202e-01
-8.55054259e-01 2.84408510e-01 -3.18030000e-01 -2.63071567e-01
1.15514767e+00 -9.94125545e-01 -4.37978119e-01 1.06978786e+00
-7.64099360e-01 -5.68718731e-01 -4.55407768e-01 2.14332402e-01
-8.07520330e-01 5.18177807e-01 -7.97275841e-01 -7.21594572e-01
-7.04840064e-01 -1.20969248e+00 9.69282269e-01 3.46767038e-01
4.47635025e-01 -9.71501291e-01 -2.09748834e-01 6.56867564e-01
8.12639117e-01 6.64281905e-01 8.02182674e-01 8.93154356e-04
-5.87630332e-01 -1.22318514e-01 -4.77188416e-02 1.03998768e+00
2.08187386e-01 -2.66712189e-01 -9.04154241e-01 -6.57008052e-01
5.93335748e-01 -4.76185620e-01 4.74468946e-01 4.78939474e-01
1.50942242e+00 -4.10311580e-01 1.87899560e-01 9.05162215e-01
1.66928041e+00 3.62700492e-01 1.21837556e+00 2.82920569e-01
6.41950369e-01 -1.27124071e-01 6.65441692e-01 1.97392002e-01
-3.50087672e-01 6.58148289e-01 7.17847705e-01 -1.74156189e-01
-4.41945791e-01 -2.77150959e-01 5.46793759e-01 7.28793502e-01
-2.00614527e-01 -4.12265450e-01 -2.48227119e-01 2.57404447e-01
-1.38949955e+00 -1.06695414e+00 2.59737819e-01 1.95316696e+00
9.85387325e-01 5.93315847e-02 -1.65971369e-01 3.03738378e-02
8.39287162e-01 3.00149649e-01 -7.63844609e-01 -3.03636044e-01
-3.51695627e-01 5.31830788e-01 4.84657526e-01 3.32301825e-01
-8.80955815e-01 9.22814369e-01 6.93222618e+00 1.42511833e+00
-1.17230177e+00 3.37515533e-01 8.68585229e-01 2.13440835e-01
-2.07345814e-01 -2.86187083e-01 -4.96200502e-01 6.19334042e-01
9.82892632e-01 -6.30895421e-02 7.23467529e-01 9.23793912e-01
-1.95541512e-02 -3.74584906e-02 -6.81434751e-01 1.00362432e+00
1.39589056e-01 -1.14210510e+00 2.66634405e-01 1.84841841e-01
1.11905015e+00 -2.83358455e-01 3.81856680e-01 6.53631166e-02
2.54449785e-01 -1.24017179e+00 5.96608698e-01 3.10829937e-01
1.21903145e+00 -9.26393688e-01 7.71737218e-01 3.97036374e-01
-8.91869247e-01 -1.63302183e-01 -5.65397441e-01 3.02554637e-01
3.24952096e-01 6.55304253e-01 -4.43266958e-01 7.25970745e-01
6.08000159e-01 2.36485362e-01 -5.19677758e-01 5.96257925e-01
-6.21052206e-01 8.03920209e-01 -7.94185773e-02 3.37641060e-01
-1.45167232e-01 -3.86294931e-01 5.89288354e-01 6.33635044e-01
5.14398634e-01 1.88798398e-01 -1.93527386e-01 8.86322320e-01
-8.77904668e-02 -3.08274310e-02 -3.32591206e-01 2.02698976e-01
2.23321140e-01 1.23979914e+00 -7.29540229e-01 -2.55033433e-01
-5.77807426e-02 1.21828759e+00 2.19934806e-02 2.95094162e-01
-1.32720554e+00 -1.13690078e-01 5.04691005e-01 -1.18184917e-01
3.83104712e-01 -2.21374750e-01 -3.41062427e-01 -1.15708947e+00
6.62139431e-02 -1.21813703e+00 2.28311568e-01 -1.04864120e+00
-1.33405173e+00 8.32313657e-01 -1.23946793e-01 -1.11510038e+00
-9.28408876e-02 -4.02821630e-01 -5.40054440e-01 8.98345590e-01
-1.19024491e+00 -1.33427572e+00 -3.41593653e-01 6.20706856e-01
5.08077323e-01 -4.28438783e-01 9.10627306e-01 3.24756086e-01
-4.58084881e-01 8.46828878e-01 1.44808993e-01 1.63637936e-01
5.65674901e-01 -1.04602253e+00 2.13136435e-01 1.31029212e+00
1.74944997e-01 2.46133119e-01 7.15536535e-01 -4.98256415e-01
-1.13643372e+00 -1.06099117e+00 -1.51341558e-01 -3.50885421e-01
4.68236059e-01 -3.61406766e-02 -7.17004418e-01 5.76501250e-01
2.82378495e-01 1.35391280e-01 4.90275681e-01 -3.63215744e-01
-1.65862218e-01 -3.02574307e-01 -1.66824365e+00 4.59262669e-01
8.99940789e-01 -5.98402798e-01 -2.63728350e-01 1.93561301e-01
8.26662481e-01 -4.37066019e-01 -1.24137449e+00 7.27340817e-01
3.17744106e-01 -9.40063298e-01 1.14078474e+00 -5.27266383e-01
7.87039816e-01 -2.40870804e-01 -3.97939354e-01 -1.59865367e+00
-2.79067397e-01 -5.03612041e-01 -1.06952183e-01 1.29708195e+00
-5.11860698e-02 -7.58310556e-01 5.26114821e-01 4.21355963e-01
-1.41400561e-01 -7.57803679e-01 -1.11313117e+00 -9.77410197e-01
3.05560350e-01 -2.25675851e-01 6.10702932e-01 5.95292926e-01
-9.58294034e-01 1.69220641e-02 -7.95411468e-01 5.02100348e-01
8.85228217e-01 -5.50542362e-02 8.37722957e-01 -6.41195416e-01
-4.55449671e-01 -2.01455280e-01 -5.86835742e-01 -9.88163650e-01
8.11721161e-02 -5.12651086e-01 -1.64066926e-01 -1.23518753e+00
2.94010401e-01 -4.18324023e-01 -2.93623507e-01 2.68298149e-01
-4.24331248e-01 6.33979321e-01 1.88761279e-01 1.50218666e-01
-1.42037690e-01 8.09089541e-01 1.72677302e+00 -1.77532151e-01
7.65042678e-02 3.21563631e-02 -7.82878816e-01 5.09104490e-01
8.70548368e-01 -5.17406583e-01 -5.12508512e-01 -3.13027203e-01
2.24528685e-01 1.85674220e-01 3.67989272e-01 -1.32019913e+00
-3.26592386e-01 -3.22491944e-01 4.66575712e-01 -2.16503292e-02
4.01825905e-01 -8.44411075e-01 5.13148844e-01 5.28832436e-01
-2.51104176e-01 -2.54715741e-01 1.23091467e-01 5.82786739e-01
-2.61296600e-01 -1.44360304e-01 1.30691099e+00 -1.55464858e-01
-4.42739725e-01 2.29290932e-01 4.28941585e-02 3.78260687e-02
1.00972831e+00 -1.16610706e-01 -2.38594308e-01 -4.98408765e-01
-6.55058563e-01 -4.79757339e-01 5.49632907e-01 3.36510599e-01
6.09961450e-01 -1.56523204e+00 -8.38272512e-01 2.34397963e-01
-3.58730823e-01 6.03075512e-02 4.88946557e-01 5.04365206e-01
-4.64803517e-01 -8.32524523e-02 -5.99236012e-01 -4.41181183e-01
-1.12470567e+00 6.90014064e-01 4.82626140e-01 -3.54544282e-01
-4.85171586e-01 8.12405288e-01 2.20131248e-01 -2.00366713e-02
-3.26935798e-01 -1.59952119e-01 1.39818430e-01 -3.93230408e-01
4.67729747e-01 3.40078712e-01 -5.19258715e-02 -4.92526829e-01
6.77839369e-02 7.52573431e-01 1.27851337e-01 -5.98096289e-02
1.41090620e+00 -7.13015124e-02 -1.62740096e-01 1.57642260e-01
1.02242494e+00 3.30935270e-01 -1.33280408e+00 1.67673424e-01
-8.20953250e-01 -8.09556603e-01 5.89544028e-02 -9.42656815e-01
-1.49493194e+00 5.21375000e-01 9.12988126e-01 3.11736286e-01
1.66476345e+00 -5.93677200e-02 1.04939890e+00 -1.99739814e-01
6.75368607e-01 -8.41857255e-01 4.29770648e-01 -5.66435494e-02
1.19385982e+00 -1.27551436e+00 4.20838147e-02 -4.55235779e-01
-6.59494281e-01 8.41012418e-01 6.60215855e-01 -2.73128361e-01
3.13002467e-01 2.47971699e-01 1.09115429e-01 2.14937970e-01
-2.69291252e-01 3.11756283e-01 2.16923505e-01 7.74323225e-01
-6.61902945e-04 4.13659364e-02 -2.57745981e-01 3.86103004e-01
-3.20066273e-01 -1.33787483e-01 5.90751112e-01 3.47639531e-01
-1.89878702e-01 -1.06165528e+00 -4.75806177e-01 1.27294719e-01
-6.81070864e-01 -3.62893231e-02 1.59652919e-01 6.76239550e-01
2.15056404e-01 9.98698294e-01 -3.57350320e-01 -4.48523015e-01
1.01126581e-01 -2.09347278e-01 8.41619372e-01 -1.99077979e-01
-2.96814412e-01 -2.05103368e-01 -2.57849246e-01 -5.51205933e-01
-6.65704131e-01 -3.32968891e-01 -8.57749522e-01 -2.29046777e-01
-4.52645123e-01 -2.24083057e-03 6.06657684e-01 8.57052624e-01
2.46313334e-01 7.62993217e-01 9.03551161e-01 -6.09357357e-01
-8.50685000e-01 -1.05018008e+00 -6.30529404e-01 5.90988040e-01
-7.81037984e-03 -4.16620135e-01 -6.29843593e-01 7.77340606e-02] | [11.715883255004883, -0.6600180864334106] |
d9024c4d-3853-436a-b0bb-850bcc8bc915 | lagan-deep-semi-supervised-linguistic | 2301.13853 | null | https://arxiv.org/abs/2301.13853v1 | https://arxiv.org/pdf/2301.13853v1.pdf | LAGAN: Deep Semi-Supervised Linguistic-Anthropology Classification with Conditional Generative Adversarial Neural Network | Education is a right of all, however, every individual is different than others. Teachers in post-communism era discover inherent individualism to equally train all towards job market of fourth industrial revolution. We can consider scenario of ethnic minority education in academic practices. Ethnic minority group has grown in their own culture and would prefer to be taught in their native way. We have formulated such linguistic anthropology(how people learn)based engagement as semi-supervised problem. Then, we have developed an conditional deep generative adversarial network algorithm namely LA-GAN to classify linguistic ethnographic features in student engagement. Theoretical justification proves the objective, regularization and loss function of our semi-supervised adversarial model. Survey questions are prepared to reach some form of assumptions about z-generation and ethnic minority group, whose learning style, learning approach and preference are our main area of interest. | ['Zuzana Kubincova', 'Rossi Kamal'] | 2023-01-26 | null | null | null | null | ['culture'] | ['speech'] | [ 1.50042146e-01 7.49887288e-01 -2.61963636e-01 -4.94582862e-01
-1.23737492e-01 -6.28389060e-01 8.36508453e-01 -7.28572726e-01
-2.71244943e-01 1.24116099e+00 5.39466202e-01 -6.28422201e-01
-2.40480542e-01 -1.25036430e+00 -4.80855525e-01 -6.16414249e-01
4.60208654e-01 5.78041852e-01 -6.51591659e-01 -5.15635610e-01
2.47196525e-01 4.70325321e-01 -1.19768393e+00 -3.26660462e-03
1.68182480e+00 3.98391336e-02 -2.27640271e-01 7.16707349e-01
-2.47162282e-01 1.30909455e+00 -7.73721397e-01 -7.29381502e-01
3.45636219e-01 -1.18108809e+00 -1.19823420e+00 1.38312429e-01
7.12396502e-01 -5.19472808e-02 -3.44507009e-01 1.38615775e+00
6.47330582e-01 4.63405997e-01 1.23428893e+00 -1.23075950e+00
-1.63479912e+00 1.15307355e+00 -3.37463230e-01 -2.27119736e-02
-4.17072810e-02 1.19799554e-01 4.26982462e-01 -5.65612614e-01
4.06338155e-01 1.34507906e+00 5.08586466e-01 9.65655565e-01
-1.11101973e+00 -9.54063356e-01 -1.40117049e-01 -2.16897905e-01
-1.33997953e+00 -7.84376264e-02 1.01924825e+00 -8.72384846e-01
1.11453347e-01 2.92471588e-01 8.43681216e-01 1.28074157e+00
1.82156265e-01 5.87171137e-01 1.76913297e+00 -5.41444004e-01
-1.42308295e-01 7.39236891e-01 -1.98444411e-01 8.84794533e-01
1.82182610e-01 2.88578451e-01 -3.64508986e-01 3.22356880e-01
9.24557388e-01 1.88325043e-03 2.15573564e-01 1.09335043e-01
-8.40110123e-01 1.35718358e+00 1.66870445e-01 4.04314816e-01
-2.10656971e-01 5.66298552e-02 -1.21475838e-01 9.40130353e-01
3.59417200e-01 5.15676856e-01 -3.81562114e-02 5.75749986e-02
-8.77949953e-01 1.48085430e-01 4.92725909e-01 1.10288489e+00
5.04046917e-01 7.09279001e-01 -9.15948302e-02 5.78434050e-01
2.96171188e-01 6.51359856e-01 7.16410756e-01 -1.03038716e+00
-7.09024444e-02 7.99138486e-01 -5.15083194e-01 -6.72132373e-01
-2.16321386e-02 -6.40435755e-01 -1.08285534e+00 6.94073081e-01
4.97497559e-01 -8.29061866e-01 -8.01111281e-01 1.70836961e+00
2.48601556e-01 6.92605693e-03 3.89131546e-01 6.12712801e-01
8.47279608e-01 5.87457120e-01 1.90487057e-01 -1.48647726e-01
1.17496467e+00 -8.80769134e-01 -6.99066401e-01 1.73594147e-01
2.87611067e-01 -9.10883009e-01 1.08818340e+00 3.27043086e-01
-1.23483348e+00 -8.61981094e-01 -6.20267570e-01 3.20525048e-03
-4.14737314e-01 -2.17543542e-01 9.72318351e-01 1.56703949e+00
-9.10757899e-01 5.21378875e-01 -3.27138960e-01 -2.53412575e-01
4.87755537e-01 4.89980876e-01 -1.32602930e-01 5.60926378e-01
-1.19571519e+00 7.73697615e-01 2.85873294e-01 -1.79330721e-01
-1.03223944e+00 -8.56543124e-01 -5.58959901e-01 -2.49461576e-01
-6.84026405e-02 -8.37555468e-01 8.44091952e-01 -1.87028372e+00
-1.79773355e+00 1.19693160e+00 4.10284400e-01 -1.57678530e-01
7.91522980e-01 -7.27517456e-02 -6.49070442e-01 -3.19276333e-01
1.64394162e-03 4.37911808e-01 5.97695231e-01 -1.27794480e+00
-6.54775977e-01 -3.91377300e-01 8.61640200e-02 2.66529024e-01
-2.59681582e-01 1.18680997e-02 9.12951171e-01 -8.31199348e-01
-1.16858676e-01 -7.50318766e-01 -2.51849234e-01 -5.76075494e-01
-2.50346243e-01 -3.70151758e-01 7.56586134e-01 -6.08582854e-01
1.09366000e+00 -1.74825907e+00 1.23687081e-01 3.02306920e-01
3.40088904e-01 1.83580205e-01 2.34439895e-01 3.63694370e-01
-2.29595825e-01 5.71386635e-01 5.72966188e-02 2.79914647e-01
2.39154667e-01 6.36844337e-02 -3.82657915e-01 4.65684652e-01
1.17138177e-01 8.64943504e-01 -9.99348342e-01 -7.82918394e-01
1.48444474e-01 5.15641332e-01 -9.66911018e-01 4.77803677e-01
5.77504002e-02 1.16423082e+00 -8.27032089e-01 5.80337167e-01
5.05437315e-01 2.08952397e-01 4.76028398e-02 3.59090418e-01
-2.28013039e-01 -2.06828192e-01 -1.05046570e+00 1.47321630e+00
-5.84002793e-01 7.89386034e-01 -2.04014108e-01 -1.14670491e+00
1.04460132e+00 3.57849926e-01 2.70148933e-01 -3.34285110e-01
3.07279766e-01 2.81909406e-01 4.71565694e-01 -6.57999694e-01
4.66809750e-01 -5.59948683e-01 -3.87201132e-03 4.68690991e-01
2.59510607e-01 -3.65485191e-01 -2.88625449e-01 -9.06071290e-02
4.78284359e-01 4.83485490e-01 3.24448735e-01 -8.43127668e-01
8.88221323e-01 -1.01468518e-01 4.92002010e-01 7.29074180e-01
-3.51867944e-01 3.56627136e-01 3.06286842e-01 -6.21781945e-01
-1.03906226e+00 -1.38573778e+00 -2.10333213e-01 1.37401414e+00
-2.86184460e-01 6.84726059e-01 -8.58681202e-01 -8.75283480e-01
-3.84626180e-01 9.46167350e-01 -7.07797706e-01 -2.72236764e-01
-8.66600692e-01 -3.80598485e-01 5.85159242e-01 2.18959451e-02
6.62089527e-01 -1.40743065e+00 -3.39936227e-01 -1.46113306e-01
4.23365682e-01 -2.47725695e-01 -4.85247314e-01 -3.34511474e-02
-6.22242332e-01 -7.45947719e-01 -8.98102880e-01 -1.23458922e+00
8.91290188e-01 -5.94480693e-01 1.26810932e+00 -1.45963272e-02
-3.48234177e-03 3.78390044e-01 -1.88779756e-01 -8.41723502e-01
-7.66676009e-01 2.96242297e-01 2.82335460e-01 -2.38044169e-02
6.53452158e-01 -1.14206040e+00 -8.09036195e-01 -9.44787189e-02
-6.29770815e-01 2.27131248e-02 3.98687541e-01 5.91298461e-01
6.80572838e-02 3.58697355e-01 8.48155975e-01 -1.46513927e+00
6.16405845e-01 -5.28671682e-01 -2.79285550e-01 2.49342367e-01
-6.17888331e-01 -3.18561159e-02 8.17416430e-01 -6.06846631e-01
-1.41545749e+00 -3.30005705e-01 -1.41004130e-01 2.21661832e-02
-2.87185401e-01 2.70150214e-01 -2.19667971e-01 -4.95339148e-02
7.15947270e-01 2.09477976e-01 -2.76911885e-01 -2.08827645e-01
3.86868060e-01 9.31995273e-01 5.38730979e-01 -1.01169503e+00
1.27409315e+00 1.93353757e-01 -1.67024378e-02 -8.98118198e-01
-8.84571075e-01 3.41333717e-01 -8.10412347e-01 -4.63343769e-01
1.02989984e+00 -9.57851768e-01 -1.02384412e+00 3.57255161e-01
-7.45911002e-01 -3.98737520e-01 -8.61128271e-01 7.66829491e-01
-5.70225179e-01 -2.54598647e-01 -1.85499072e-01 -1.17277169e+00
-2.16379687e-01 -9.13850546e-01 3.54651332e-01 8.19775879e-01
-1.27419993e-01 -1.53687406e+00 2.35767171e-01 7.69577265e-01
3.42091560e-01 7.41163671e-01 1.01625705e+00 -8.81707668e-01
-3.14448893e-01 4.84547168e-01 3.35722685e-01 4.12946492e-01
2.20016330e-01 6.74639791e-02 -8.93058896e-01 -1.49202183e-01
7.63991848e-02 -5.38970172e-01 4.61750805e-01 2.53941298e-01
1.08600616e+00 -6.75802529e-01 3.20611864e-01 7.79729247e-01
1.48252106e+00 4.05942261e-01 5.23336709e-01 1.15035154e-01
1.00539875e+00 8.78012419e-01 6.39706627e-02 5.46770589e-03
2.07288668e-01 -2.07917485e-02 -1.17681414e-01 -2.21863940e-01
-2.20437929e-01 -4.19448584e-01 6.41595125e-01 1.33770299e+00
-7.06685424e-01 -1.57575950e-01 -9.83824790e-01 5.88223875e-01
-1.28060055e+00 -1.42039478e+00 -1.29632458e-01 1.83474743e+00
1.31805933e+00 -1.41615212e-01 2.35414773e-01 -2.75278598e-01
6.43870473e-01 -2.62590311e-02 -1.15163758e-01 -9.24187303e-01
-1.60512850e-01 8.25336874e-01 4.68749255e-01 8.52454543e-01
-6.91472411e-01 1.17684162e+00 6.54042339e+00 1.02755773e+00
-1.15626812e+00 2.35214248e-01 9.70613956e-01 1.68724895e-01
-8.87950718e-01 1.02989515e-02 -8.25521410e-01 4.90996748e-01
8.93379271e-01 -4.83773023e-01 6.48563087e-01 6.29132211e-01
1.52578135e-03 3.39607626e-01 -9.64847088e-01 4.56224650e-01
1.87441725e-02 -9.89499986e-01 1.32692173e-01 3.22464257e-01
1.64216447e+00 -5.72759748e-01 6.76283538e-01 7.02819884e-01
1.06990123e+00 -1.82038665e+00 5.43438852e-01 6.63351655e-01
7.90844679e-01 -1.28808582e+00 3.85971159e-01 5.37883699e-01
-5.73944807e-01 1.92208998e-02 -4.07235593e-01 -4.18193132e-01
-4.15144116e-01 1.34681553e-01 -4.83762711e-01 3.41772676e-01
2.37006307e-01 1.85386240e-02 -4.26024854e-01 -8.59325603e-02
-2.37181872e-01 9.62028384e-01 2.00189054e-01 -8.18980038e-02
4.76615071e-01 -7.94189095e-01 2.35422954e-01 8.55355084e-01
5.06563544e-01 4.66567248e-01 1.91997573e-01 1.11704063e+00
-1.84315592e-01 2.91628808e-01 -9.55196857e-01 -2.92522222e-01
3.81716609e-01 8.89439583e-01 -3.65371585e-01 -1.40259229e-02
-4.31514204e-01 6.24448955e-01 1.74130455e-01 3.10095310e-01
-7.57732391e-01 -3.97889853e-01 6.02439642e-01 2.79665649e-01
-3.17124546e-01 1.79060072e-01 -2.93626487e-01 -1.05064523e+00
-7.77207553e-01 -1.39658892e+00 5.04952148e-02 -1.87320769e-01
-1.34183490e+00 3.56792718e-01 -3.34469557e-01 -7.94267416e-01
-2.42522970e-01 -5.63252807e-01 -9.36238647e-01 1.07982528e+00
-9.98079002e-01 -1.49532962e+00 -6.05635345e-02 7.25141644e-01
5.46747684e-01 -1.11063612e+00 6.88415766e-01 3.54195416e-01
-4.43701684e-01 7.89693296e-01 1.91826314e-01 6.11264706e-01
4.07001734e-01 -1.48897958e+00 -2.14232475e-01 8.57202947e-01
1.72249734e-01 7.67532945e-01 7.02097952e-01 -6.62207246e-01
-1.05937862e+00 -1.15078819e+00 1.08910537e+00 -7.47819304e-01
5.36537647e-01 -1.81928158e-01 -3.46805543e-01 1.08063591e+00
8.96822453e-01 -6.03884935e-01 1.03502572e+00 5.36391623e-02
2.28423372e-01 1.81078658e-01 -1.31346083e+00 9.31752563e-01
1.12926245e+00 -5.94267666e-01 -7.21280694e-01 5.55806398e-01
7.89814353e-01 -8.08215067e-02 -1.21791983e+00 1.98910922e-01
5.19016802e-01 -1.08991134e+00 9.30193245e-01 -1.09753847e+00
8.60658050e-01 1.23939335e-01 1.05193928e-01 -1.27441955e+00
-3.54279369e-01 -1.18855762e+00 3.00596386e-01 1.56251395e+00
2.22904399e-01 -6.90958023e-01 1.25762570e+00 2.59710044e-01
-9.99523774e-02 -5.19971192e-01 -5.25869370e-01 -5.82387984e-01
1.14661539e+00 3.22101146e-01 7.64776409e-01 1.87307775e+00
-4.36647087e-01 4.32698548e-01 -6.31399870e-01 1.76846085e-03
7.21753180e-01 1.80552825e-01 9.14687753e-01 -1.36135912e+00
-1.26581773e-01 -6.25250697e-01 -1.63907558e-01 -3.66392314e-01
6.35575414e-01 -1.02767658e+00 -5.30952573e-01 -8.89156878e-01
1.34892106e-01 -4.73478019e-01 -1.62753895e-01 -3.34254429e-02
-1.65623710e-01 3.58921856e-01 -1.88962817e-01 -3.66871685e-01
2.06910685e-01 2.89994478e-01 2.06993222e+00 -1.80411950e-01
-3.81716996e-01 9.92718190e-02 -1.07166219e+00 9.25465941e-01
9.54670310e-01 -4.34042424e-01 -7.72373259e-01 -1.63989365e-01
3.47576529e-01 -5.65100424e-02 1.30561113e-01 -8.73354793e-01
6.86044767e-02 -1.03755689e+00 6.58205271e-01 -1.11969039e-01
-2.71680146e-01 -9.54616666e-01 3.26490581e-01 6.67115748e-01
-5.77230394e-01 -1.71611369e-01 -4.74755079e-01 -2.50095665e-01
-1.41636118e-01 -5.10685980e-01 1.12653911e+00 -3.81770015e-01
-3.11270863e-01 3.52251291e-01 -4.20637459e-01 6.37232602e-01
1.08578253e+00 -4.89172757e-01 -1.99337423e-01 -3.72504622e-01
-8.17427337e-01 7.46387169e-02 3.25526565e-01 2.13956684e-01
2.18719423e-01 -1.53308988e+00 -1.28571177e+00 2.71014690e-01
-5.21237314e-01 -7.55340829e-02 5.30179180e-02 4.34995472e-01
-9.53735352e-01 1.19279817e-01 -9.12501514e-01 4.35598008e-02
-9.83424902e-01 4.60159838e-01 4.54291284e-01 -3.05478424e-01
-1.74094036e-01 1.00906491e+00 3.97502393e-01 -1.09144866e+00
-6.87095523e-02 3.18384439e-01 -4.93353218e-01 -1.24955080e-01
5.88505529e-02 6.34471476e-01 -8.16801012e-01 -7.92004704e-01
2.02628314e-01 5.53094506e-01 2.67610818e-01 -1.16396941e-01
1.18638039e+00 5.74028268e-02 -5.11690392e-04 3.05604070e-01
9.97933626e-01 6.62908018e-01 -7.27134645e-01 9.41731557e-02
-4.10840064e-01 -3.81584138e-01 -1.05992474e-01 -7.02543139e-01
-1.24640489e+00 7.60657787e-01 6.66829526e-01 1.27341658e-01
9.10568833e-01 -3.05642396e-01 3.83856356e-01 2.77171910e-01
-2.74933614e-02 -1.10659122e+00 1.11948088e-01 5.26576638e-01
6.19533598e-01 -1.46464348e+00 1.08002581e-01 -2.29314491e-01
-6.17281795e-01 7.88619578e-01 9.38386798e-01 -5.12225032e-01
7.13947117e-01 -2.56094690e-02 3.32301140e-01 -1.07093967e-01
-2.16435865e-01 -1.03694968e-01 5.72460353e-01 9.64052916e-01
1.05014908e+00 3.15679669e-01 -5.24403214e-01 4.70633656e-01
-1.11962736e+00 -3.13262105e-01 3.64438653e-01 5.49912214e-01
-5.55534840e-01 -1.29791212e+00 -3.72612000e-01 1.38163060e-01
-8.04514647e-01 2.87887603e-02 -5.12477279e-01 9.98536468e-01
8.96368206e-01 7.36899257e-01 1.44886672e-01 -1.99363798e-01
8.81960317e-02 2.25649446e-01 7.68086791e-01 -5.65295517e-01
-1.24377143e+00 -2.79299945e-01 -2.40227014e-01 2.08349854e-01
-5.86039543e-01 -5.39778054e-01 -9.16359425e-01 -9.69984651e-01
6.60968944e-02 6.03593171e-01 3.20625305e-01 6.79014027e-01
-4.48288232e-01 7.65874803e-01 8.85819077e-01 2.97617493e-03
-5.78554273e-01 -1.00164735e+00 -8.28254759e-01 3.39381546e-01
1.14436522e-01 -7.71389604e-02 -3.74159455e-01 4.20018315e-01] | [11.701675415039062, -0.1062721312046051] |
97554b25-9d20-4806-856e-a2e50c8f5480 | learning-hierarchy-aware-quaternion-knowledge | null | null | https://aclanthology.org/2022.coling-1.175 | https://aclanthology.org/2022.coling-1.175.pdf | Learning Hierarchy-Aware Quaternion Knowledge Graph Embeddings with Representing Relations as 3D Rotations | Knowledge graph embedding aims to represent entities and relations as low-dimensional vectors, which is an effective way for predicting missing links. It is crucial for knowledge graph embedding models to model and infer various relation patterns, such as symmetry/antisymmetry. However, many existing approaches fail to model semantic hierarchies, which are common in the real world. We propose a new model called HRQE, which represents entities as pure quaternions. The relational embedding consists of two parts: (a) Using unit quaternions to represent the rotation part in 3D space, where the head entities are rotated by the corresponding relations through Hamilton product. (b) Using scale parameters to constrain the modulus of entities to make them have hierarchical distributions. To the best of our knowledge, HRQE is the first model that can encode symmetry/antisymmetry, inversion, composition, multiple relation patterns and learn semantic hierarchies simultaneously. Experimental results demonstrate the effectiveness of HRQE against some of the SOTA methods on four well-established knowledge graph completion benchmarks. | ['Bowei Xing', 'Taiyan Chen', 'Ruibin Wang', 'Xin Tong', 'Yongjie Shi', 'Xianghua Ying', 'Jinfa Yang'] | null | null | null | null | coling-2022-10 | ['knowledge-graph-embedding', 'knowledge-graph-embeddings', 'knowledge-graph-embeddings'] | ['graphs', 'graphs', 'methodology'] | [-2.26718768e-01 3.19428593e-01 -2.58473307e-01 -6.23189732e-02
3.54083449e-01 -4.55915868e-01 6.45638049e-01 4.09375280e-01
-6.95533603e-02 5.47100127e-01 3.90164942e-01 -2.32486427e-01
-3.68023634e-01 -1.22605300e+00 -6.60839200e-01 -3.71950537e-01
-3.95575404e-01 7.82016933e-01 2.47800037e-01 -6.48804367e-01
-8.49891901e-02 4.93674070e-01 -1.27605188e+00 -5.79028092e-02
6.55158699e-01 6.05905235e-01 -5.20591557e-01 2.27447107e-01
-2.00712115e-01 9.76304412e-01 -3.65237713e-01 -8.44481468e-01
9.95605588e-02 -1.93894431e-01 -1.11504889e+00 -8.85588210e-03
4.02162783e-02 -2.40075707e-01 -8.74979675e-01 1.15517867e+00
1.80537209e-01 2.39087231e-02 6.57179058e-01 -1.68018937e+00
-1.24143088e+00 6.33836329e-01 -5.49407423e-01 -1.20177813e-01
7.53425479e-01 -4.69985336e-01 1.63437438e+00 -8.98793399e-01
9.30557549e-01 1.47641885e+00 4.95398223e-01 9.81636420e-02
-1.42851686e+00 -6.04074419e-01 1.16034463e-01 6.85448527e-01
-1.77751768e+00 1.06098257e-01 8.22191417e-01 -4.97112870e-01
1.01804662e+00 2.98313737e-01 9.35981810e-01 7.47580826e-01
-6.83524236e-02 4.78182882e-01 7.13013351e-01 -2.74625152e-01
1.00222781e-01 7.69138988e-03 3.07976902e-01 9.61606443e-01
6.62860394e-01 -3.48399669e-01 -6.47087455e-01 -2.17256770e-01
8.82925630e-01 8.10716972e-02 -4.36486542e-01 -1.07963061e+00
-1.58649313e+00 9.25808907e-01 9.34302628e-01 3.09826471e-02
-2.84715325e-01 7.55620096e-03 2.69476324e-01 2.04001248e-01
1.95517644e-01 4.73019540e-01 -2.89285213e-01 3.26366991e-01
-1.32285923e-01 2.12145567e-01 1.03060591e+00 1.33819377e+00
1.09214735e+00 -2.60396540e-01 1.03025518e-01 7.33875334e-01
4.34752136e-01 2.42828190e-01 3.83582473e-01 -6.81444943e-01
4.78796512e-01 1.18188643e+00 1.63735934e-02 -1.54763591e+00
-5.56194067e-01 -1.91156626e-01 -1.01056266e+00 -4.15638834e-01
-6.06366806e-02 3.43812287e-01 -6.20980799e-01 1.50028527e+00
4.14225340e-01 4.16118592e-01 8.97688419e-02 9.08940077e-01
1.02074242e+00 5.70518672e-01 -1.69812441e-01 4.03804407e-02
1.64247644e+00 -7.34198630e-01 -9.05373573e-01 1.37445092e-01
6.65000319e-01 -3.81684542e-01 7.38246739e-01 -5.75550795e-02
-6.20332599e-01 -2.26648688e-01 -1.14815819e+00 -4.13392305e-01
-8.15136790e-01 -1.50491074e-01 1.12316346e+00 4.70996141e-01
-8.55107784e-01 3.67087126e-01 -8.07047546e-01 -3.20219278e-01
5.12096435e-02 2.58810729e-01 -8.19077551e-01 -1.36836588e-01
-1.81340814e+00 8.27672362e-01 7.09189355e-01 2.15252325e-01
-2.05019668e-01 -5.47788560e-01 -1.35528862e+00 1.41969472e-01
4.97394741e-01 -8.98070276e-01 4.71456885e-01 -1.38107061e-01
-1.18666101e+00 7.26288140e-01 -2.36891080e-02 -1.80087104e-01
7.68950731e-02 -8.69908035e-02 -4.46373105e-01 4.50543640e-03
7.27241412e-02 4.47356522e-01 4.27520573e-01 -1.29553604e+00
-2.11956963e-01 -5.45378685e-01 4.97111499e-01 3.46809834e-01
-6.40611053e-01 -4.45288211e-01 -6.16496801e-01 -4.59198892e-01
6.33490562e-01 -1.01874733e+00 1.66772023e-01 -8.08405355e-02
-7.25569367e-01 -5.08995831e-01 6.74624979e-01 -7.43269265e-01
1.41036665e+00 -1.99327338e+00 6.91436350e-01 4.22263980e-01
5.88630617e-01 -3.57820392e-02 3.41980048e-02 8.22051406e-01
-2.92168677e-01 2.91875422e-01 -5.50222062e-02 3.32174078e-02
4.66759056e-01 6.52231455e-01 -3.17924082e-01 3.11335087e-01
3.91105354e-01 1.10752535e+00 -1.11008942e+00 -5.03986120e-01
1.56552583e-01 7.54549801e-01 -5.71309626e-01 1.14009706e-02
3.95480767e-02 -5.89352213e-02 -4.67732489e-01 5.06147504e-01
6.51022375e-01 -4.95883822e-01 7.12401748e-01 -7.66468704e-01
2.39693761e-01 3.15849900e-01 -1.64480114e+00 1.59775734e+00
-9.00423527e-02 3.22543323e-01 -3.72911662e-01 -1.00101066e+00
8.85277629e-01 2.35265195e-01 4.61252689e-01 -5.06839752e-01
-8.95794779e-02 -9.03099552e-02 -8.90876651e-02 -2.27884606e-01
6.75622761e-01 4.39449064e-02 -2.33130995e-02 4.27212790e-02
1.01607405e-01 -2.01362506e-01 4.98780221e-01 6.52917981e-01
9.84529197e-01 5.62521629e-02 2.81854212e-01 -2.88068205e-02
5.15881777e-01 -3.62381160e-01 8.57630253e-01 2.05200300e-01
8.40988606e-02 2.70901352e-01 1.09769189e+00 -4.67002422e-01
-8.90719712e-01 -1.15754342e+00 2.43257545e-03 6.10812008e-01
5.03948152e-01 -1.00069892e+00 -2.39172846e-01 -5.37053406e-01
2.28211522e-01 5.33314884e-01 -6.59373224e-01 -2.82345027e-01
-2.92303890e-01 -7.07148850e-01 2.83624083e-01 5.13913989e-01
4.66229409e-01 -6.25955224e-01 5.45423366e-02 1.52696505e-01
-3.32193166e-01 -1.36470664e+00 -2.19337836e-01 -7.71040469e-02
-6.46306813e-01 -1.37747860e+00 -3.41898978e-01 -8.14553499e-01
6.74766183e-01 2.22920582e-01 1.03185201e+00 1.94662854e-01
-2.12640315e-01 4.34957892e-01 -5.36942244e-01 -8.46024677e-02
2.00892892e-02 -2.28865817e-02 1.66535869e-01 1.38121210e-02
2.43068680e-01 -6.58602357e-01 -6.01501346e-01 2.95878321e-01
-1.08970702e+00 2.54869521e-01 4.59928721e-01 8.90414357e-01
7.23856390e-01 5.41172504e-01 1.84250996e-01 -1.03448105e+00
6.83530211e-01 -4.62804109e-01 -3.03946823e-01 4.64128166e-01
-6.15404367e-01 2.75464326e-01 3.98566157e-01 -7.96913430e-02
-5.56187034e-01 -1.53068900e-01 2.39028841e-01 -3.58488053e-01
2.39783198e-01 8.28286767e-01 -2.40089029e-01 -2.97650136e-02
2.82019496e-01 9.59281530e-03 -2.65867621e-01 -2.62737483e-01
7.52586901e-01 2.53815353e-01 9.30378959e-02 -6.06626809e-01
9.95132864e-01 5.41226447e-01 5.02365232e-01 -9.36167598e-01
-8.03799093e-01 -6.32809818e-01 -8.10803771e-01 3.28732967e-01
8.77265632e-01 -9.15572703e-01 -8.11196029e-01 6.46441355e-02
-1.13946187e+00 2.76624650e-01 -1.47535011e-01 5.60033143e-01
-2.06936061e-01 8.18046331e-01 -8.04151833e-01 -3.50920200e-01
-1.63377225e-01 -8.48582745e-01 9.96834993e-01 4.90004718e-02
-9.64095965e-02 -1.02635574e+00 3.39377970e-02 4.08472449e-01
2.26411410e-02 2.63993263e-01 1.34012628e+00 -4.66579467e-01
-7.59946942e-01 -1.22537449e-01 -4.06298816e-01 1.14646651e-01
1.55856758e-01 6.16029166e-02 -2.89006859e-01 -2.01361522e-01
-6.81373596e-01 -1.82150319e-01 7.79207826e-01 -4.62014049e-01
7.56118774e-01 -2.82861054e-01 -3.75002414e-01 5.70297599e-01
1.32079637e+00 -1.11532331e-01 7.57362366e-01 3.44123334e-01
1.19538534e+00 4.16647673e-01 2.62658685e-01 3.89051527e-01
1.02197206e+00 6.81059837e-01 5.52603841e-01 1.13512814e-01
2.03898549e-02 -5.76297939e-01 -1.00835063e-01 1.32487977e+00
-3.47640812e-01 6.42758459e-02 -9.70173359e-01 3.95084351e-01
-1.94595170e+00 -7.31443405e-01 -2.68204361e-01 2.05968499e+00
8.34691882e-01 9.59909111e-02 -2.14594796e-01 2.65866101e-01
5.61937511e-01 3.07323486e-01 -2.76753843e-01 4.01961170e-02
-1.89420924e-01 2.64522672e-01 3.42718750e-01 5.66738069e-01
-1.01489115e+00 9.80120122e-01 5.31425810e+00 3.32632869e-01
-5.66296697e-01 -1.00483984e-01 -2.28680089e-01 5.79099476e-01
-6.07579052e-01 3.73024642e-01 -5.98520458e-01 3.22959386e-02
2.35953808e-01 -2.40749493e-01 5.23576856e-01 4.85982955e-01
-5.32761812e-01 3.25128496e-01 -1.14143312e+00 1.00359333e+00
1.19785182e-01 -1.06363595e+00 5.06140530e-01 3.76046374e-02
7.61995912e-01 -4.60819870e-01 -3.52184772e-01 6.54385090e-01
3.22119623e-01 -9.57442105e-01 3.84066492e-01 6.54789805e-01
5.27898431e-01 -7.56387711e-01 8.23048294e-01 -1.40165100e-02
-1.49316669e+00 2.95905739e-01 -5.75658798e-01 1.51309827e-02
4.55777943e-02 4.73543108e-01 -8.73114645e-01 1.13942742e+00
5.03030777e-01 9.57456112e-01 -6.24796331e-01 6.56422019e-01
-7.69990802e-01 3.17283273e-01 -2.23585352e-01 -7.96501637e-02
-9.88105386e-02 -6.31350398e-01 4.20232743e-01 9.36732829e-01
7.36610144e-02 1.21150546e-01 1.68947384e-01 6.86376333e-01
-3.29962343e-01 1.76024169e-01 -5.07321060e-01 -3.30153376e-01
6.01015806e-01 1.10395849e+00 -5.55383325e-01 -1.28501222e-01
-5.85358322e-01 1.13709366e+00 4.90538597e-01 6.63243294e-01
-7.54733920e-01 -5.92264652e-01 7.68646002e-01 -4.50935811e-02
2.06228346e-01 -6.10959888e-01 2.57627964e-01 -1.56509817e+00
1.27943441e-01 -7.00064123e-01 5.50291777e-01 -8.51850808e-01
-1.25142479e+00 2.37800077e-01 3.34968604e-03 -8.47623348e-01
1.43472880e-01 -7.68441677e-01 -6.22576661e-02 6.73242748e-01
-1.59196055e+00 -1.39417243e+00 -3.97931516e-01 6.48543715e-01
-1.94914922e-01 9.15739387e-02 1.12463081e+00 3.28531623e-01
-7.63437092e-01 3.00494373e-01 -4.87322025e-02 4.47420210e-01
4.38910335e-01 -1.55869520e+00 3.13772708e-01 3.46146286e-01
6.12237930e-01 1.08712399e+00 5.18789887e-01 -6.29231453e-01
-1.71563804e+00 -8.46166313e-01 1.05973768e+00 -3.13640535e-01
1.10990226e+00 -4.27639633e-01 -1.09147787e+00 1.06500387e+00
-1.06825873e-01 1.91439182e-01 7.55061269e-01 5.38061917e-01
-8.22507083e-01 -2.33176589e-01 -5.31611323e-01 7.42841125e-01
1.20466101e+00 -7.66916394e-01 -8.00079048e-01 6.15350604e-01
8.93794179e-01 -3.67625564e-01 -1.43973029e+00 5.30937016e-01
4.24471885e-01 -7.88218796e-01 1.32287240e+00 -8.51468265e-01
3.95529836e-01 -5.43547034e-01 -2.38378078e-01 -1.37075996e+00
-6.03773594e-01 -2.32600272e-01 -5.19091070e-01 1.19497156e+00
2.21774817e-01 -8.56856287e-01 6.88419819e-01 3.30472231e-01
4.03285205e-01 -7.91845024e-01 -8.42573225e-01 -7.62349665e-01
-2.01169327e-01 2.09250618e-02 9.28476810e-01 1.51968408e+00
1.51310906e-01 7.74693191e-01 -3.58068734e-01 4.69103754e-01
5.84996521e-01 2.70539731e-01 7.17400849e-01 -1.47852540e+00
-6.48600236e-02 -3.37129503e-01 -1.18590331e+00 -8.01619470e-01
3.51794034e-01 -1.16494799e+00 -7.54218698e-01 -1.99981833e+00
1.79105312e-01 -2.97257513e-01 -3.74481559e-01 5.26247203e-01
-1.51170030e-01 -4.04766984e-02 8.56349769e-04 1.68811977e-01
-6.37865067e-01 9.99223888e-01 1.24041522e+00 -3.18545491e-01
1.72939524e-02 -4.59771246e-01 -6.31558597e-01 6.68732822e-01
4.46425825e-01 -1.15011185e-01 -6.61623299e-01 -4.18952644e-01
9.74916697e-01 -1.56556070e-02 5.01558542e-01 -5.65279126e-01
3.24825317e-01 -1.29844159e-01 1.93405688e-01 -3.53182167e-01
4.44231778e-01 -9.38860118e-01 3.97193640e-01 2.20130518e-01
5.16213290e-02 4.09501605e-03 -1.69041753e-01 9.32998240e-01
-4.60600764e-01 9.47775170e-02 8.54694322e-02 1.23757057e-01
-9.97579396e-01 4.99104768e-01 3.56048584e-01 1.45405112e-02
8.54210734e-01 1.06885344e-01 -4.66109425e-01 -3.94501090e-01
-7.23929048e-01 4.58975255e-01 3.11028183e-01 5.98455489e-01
7.49153972e-01 -1.76243162e+00 -5.49403369e-01 1.91010479e-02
4.19878870e-01 1.53646305e-01 9.35147479e-02 6.96205735e-01
-7.04733372e-01 4.55259830e-01 -5.86860962e-02 -2.57633328e-01
-1.04935467e+00 6.93897963e-01 1.51953653e-01 -4.33288634e-01
-5.93545318e-01 6.92198515e-01 1.63680717e-01 -8.56235445e-01
1.17044300e-01 -4.95207533e-02 -4.07800615e-01 2.56592870e-01
4.99424934e-02 4.26816106e-01 -1.83961038e-02 -8.43592286e-01
-5.76587439e-01 5.81400871e-01 -7.84520805e-02 9.02975500e-02
1.20233214e+00 2.38298215e-02 -6.19910181e-01 4.38811034e-01
1.17027855e+00 8.35086703e-02 -5.74808538e-01 -5.27832985e-01
-2.54112529e-04 -3.46271992e-01 -1.62401751e-01 -2.71444827e-01
-7.45192170e-01 7.75664389e-01 8.50215480e-02 4.65480626e-01
7.70960510e-01 2.09387802e-02 6.67407453e-01 6.74740076e-01
5.92889726e-01 -8.79612923e-01 1.44920826e-01 6.19091392e-01
1.09472191e+00 -9.76902068e-01 3.50428253e-01 -9.06383574e-01
-6.05380595e-01 1.06149638e+00 5.66186368e-01 -3.88877373e-03
8.41750562e-01 -4.88934845e-01 -3.94731164e-01 -5.43823779e-01
-4.05745566e-01 -3.96704406e-01 6.19405150e-01 3.36079508e-01
4.06906068e-01 4.27261919e-01 -4.07158345e-01 4.49124753e-01
-2.89905101e-01 -4.51379508e-01 3.48615915e-01 8.40098262e-01
-1.14540257e-01 -1.18626463e+00 -1.11396901e-01 2.43894577e-01
-2.70012505e-02 3.88173163e-02 -6.54264808e-01 1.07209563e+00
1.07679792e-01 8.45569491e-01 -1.31317005e-01 -4.74670351e-01
7.76023924e-01 6.64750859e-02 4.04934257e-01 -6.72481358e-01
9.90526676e-02 -3.55986327e-01 1.04919806e-01 -4.94091749e-01
-3.20309609e-01 -2.65828371e-01 -1.46535766e+00 -2.93640018e-01
-3.66151929e-01 2.37074256e-01 5.91747403e-01 5.52514017e-01
4.33231056e-01 7.09842205e-01 3.77178460e-01 -3.19483280e-01
-5.20273745e-01 -7.84528255e-01 -1.03738368e+00 1.00823677e+00
1.46856240e-03 -1.11117423e+00 -3.68614048e-01 -7.70661682e-02] | [8.639185905456543, 7.774474620819092] |
7a263d51-2b65-41e6-a22a-45150f528e3a | textgrad-advancing-robustness-evaluation-in | 2212.09254 | null | https://arxiv.org/abs/2212.09254v1 | https://arxiv.org/pdf/2212.09254v1.pdf | TextGrad: Advancing Robustness Evaluation in NLP by Gradient-Driven Optimization | Robustness evaluation against adversarial examples has become increasingly important to unveil the trustworthiness of the prevailing deep models in natural language processing (NLP). However, in contrast to the computer vision domain where the first-order projected gradient descent (PGD) is used as the benchmark approach to generate adversarial examples for robustness evaluation, there lacks a principled first-order gradient-based robustness evaluation framework in NLP. The emerging optimization challenges lie in 1) the discrete nature of textual inputs together with the strong coupling between the perturbation location and the actual content, and 2) the additional constraint that the perturbed text should be fluent and achieve a low perplexity under a language model. These challenges make the development of PGD-like NLP attacks difficult. To bridge the gap, we propose TextGrad, a new attack generator using gradient-driven optimization, supporting high-accuracy and high-quality assessment of adversarial robustness in NLP. Specifically, we address the aforementioned challenges in a unified optimization framework. And we develop an effective convex relaxation method to co-optimize the continuously-relaxed site selection and perturbation variables and leverage an effective sampling method to establish an accurate mapping from the continuous optimization variables to the discrete textual perturbations. Moreover, as a first-order attack generation method, TextGrad can be baked into adversarial training to further improve the robustness of NLP models. Extensive experiments are provided to demonstrate the effectiveness of TextGrad not only in attack generation for robustness evaluation but also in adversarial defense. | ['Shiyu Chang', 'Sijia Liu', 'Yang Zhang', 'Guanhua Zhang', 'Yihua Zhang', 'Jinghan Jia', 'Bairu Hou'] | 2022-12-19 | null | null | null | null | ['adversarial-defense'] | ['adversarial'] | [ 2.41325974e-01 -1.48320660e-01 1.12684118e-02 1.36834467e-02
-1.19080400e+00 -1.18296742e+00 6.84289157e-01 9.08105224e-02
-2.19777033e-01 7.09112108e-01 1.40004769e-01 -5.32708108e-01
-6.83522969e-02 -7.62206674e-01 -9.98017669e-01 -7.29201019e-01
-1.22869723e-02 2.79758513e-01 -1.64637119e-01 -5.02825618e-01
2.49481164e-02 6.79804027e-01 -7.19337404e-01 9.85718146e-02
1.07655752e+00 7.48676181e-01 -4.16021854e-01 6.61276162e-01
2.68820286e-01 4.54151064e-01 -8.00007641e-01 -9.75577950e-01
6.72476828e-01 -2.92563200e-01 -4.79031831e-01 -3.96634668e-01
3.59016627e-01 -3.99080843e-01 -6.29111052e-01 1.38002717e+00
9.57849324e-01 9.20588896e-02 5.80123127e-01 -1.38245285e+00
-6.35594368e-01 6.65611863e-01 -4.68774438e-01 9.43948179e-02
4.81331468e-01 6.53287351e-01 1.04047418e+00 -6.77288175e-01
5.42008460e-01 1.34991276e+00 5.58930218e-01 7.34509230e-01
-1.05290413e+00 -7.03461170e-01 2.48067468e-01 -5.95698552e-03
-1.14954412e+00 -2.90105283e-01 9.53799367e-01 -2.69950867e-01
5.10061920e-01 5.10813832e-01 1.57737389e-01 1.87484622e+00
2.06407666e-01 8.21079612e-01 9.13928270e-01 -2.16109499e-01
2.56341726e-01 2.10954249e-01 -2.83467352e-01 4.71597672e-01
1.03091568e-01 4.87880528e-01 -4.23171908e-01 -5.60148776e-01
3.28755885e-01 -2.35754550e-01 -4.43844855e-01 -2.30582468e-02
-9.44193006e-01 9.23324406e-01 3.88381511e-01 5.86718954e-02
-2.38979816e-01 1.40161827e-01 6.74946308e-01 2.96023309e-01
4.00991142e-01 8.40627134e-01 -4.24348414e-01 -1.41650559e-02
-6.87717080e-01 6.78438187e-01 8.24113548e-01 5.51944137e-01
1.85309038e-01 3.55839849e-01 -4.72911417e-01 5.97994626e-01
3.15496504e-01 6.87410116e-01 3.00884008e-01 -4.77882028e-01
1.02394617e+00 3.36217433e-01 6.56086430e-02 -1.42389119e+00
1.11739188e-01 -6.55918479e-01 -8.85277450e-01 2.48938963e-01
4.62877214e-01 -3.83219212e-01 -6.66556954e-01 1.95595121e+00
4.93239850e-01 7.09172562e-02 2.90630966e-01 9.13998246e-01
3.54850411e-01 7.23019063e-01 7.01907650e-02 -1.72568575e-01
1.06696403e+00 -6.97331607e-01 -4.94250029e-01 -2.53711790e-01
3.52073163e-01 -5.97412705e-01 1.27155542e+00 3.75675738e-01
-9.82438624e-01 -1.86792791e-01 -1.22039640e+00 2.91543365e-01
-3.00478160e-01 -4.00872380e-01 3.54417413e-01 9.84717250e-01
-5.28878391e-01 5.50850689e-01 -5.53729773e-01 2.68853158e-01
4.26915526e-01 1.57856792e-01 -3.77790362e-01 3.11341714e-02
-1.89255679e+00 8.81088376e-01 3.70767832e-01 2.68080205e-01
-1.20762849e+00 -9.47948754e-01 -7.93901503e-01 -7.09540918e-02
3.63810331e-01 -4.77823794e-01 9.33844268e-01 -7.33220935e-01
-1.78518200e+00 5.04318058e-01 3.78364176e-01 -5.72601259e-01
1.16451252e+00 -2.89621472e-01 -3.40964168e-01 9.69964042e-02
-2.52170026e-01 1.11564822e-01 1.14188755e+00 -1.17563105e+00
-8.21948275e-02 -2.39526689e-01 1.82402432e-01 3.52338552e-01
-5.81320465e-01 1.18741393e-01 -3.05193871e-01 -1.05112422e+00
-4.78117555e-01 -8.35726798e-01 -4.38146710e-01 -1.52607590e-01
-8.54558706e-01 2.34287739e-01 9.15305018e-01 -8.11585605e-01
1.28097415e+00 -2.24079347e+00 3.10175866e-01 4.20690805e-01
1.03626043e-01 6.20326936e-01 -2.31348380e-01 5.98535180e-01
-2.00011745e-01 3.76303077e-01 -4.34013128e-01 -2.35380083e-01
4.15522814e-01 -1.93704575e-01 -1.12024856e+00 5.90699434e-01
2.67760336e-01 1.05289733e+00 -9.57728565e-01 -1.95835188e-01
6.20370507e-02 3.95388633e-01 -6.22896791e-01 3.48347306e-01
-3.96845460e-01 4.98026729e-01 -5.63152313e-01 5.08848548e-01
6.99456632e-01 3.55005205e-01 -1.06670678e-01 -4.85146306e-02
3.89685094e-01 9.11175981e-02 -1.15457523e+00 1.39393306e+00
-3.63026768e-01 3.09677809e-01 1.42646968e-01 -7.99136877e-01
7.99674153e-01 4.83067095e-01 1.67367890e-01 -4.60971743e-01
1.19179100e-01 1.04959153e-01 -7.92247206e-02 -3.01319718e-01
3.37021858e-01 -2.08498865e-01 -4.44019079e-01 4.60122257e-01
-2.13815495e-01 -4.52330500e-01 9.55522060e-02 3.91956955e-01
1.15593100e+00 -1.47018880e-01 5.48494561e-03 8.38297307e-02
6.84211612e-01 -2.73938209e-01 6.23264492e-01 8.62093687e-01
-3.12305331e-01 6.24421775e-01 7.34521806e-01 -2.72022426e-01
-1.04228735e+00 -9.81038392e-01 -3.61974277e-02 8.00492048e-01
-1.69218689e-01 -2.67573327e-01 -9.50977981e-01 -1.14468884e+00
6.10809959e-02 7.63818204e-01 -4.25118536e-01 -4.50370342e-01
-5.18798530e-01 -9.95043397e-01 1.24012268e+00 1.87669367e-01
4.69815731e-01 -9.96914744e-01 9.25021321e-02 7.60400817e-02
-1.46137342e-01 -1.11766684e+00 -7.64854848e-01 -4.77504916e-02
-2.87232995e-01 -9.45927858e-01 -4.33102787e-01 -3.24352503e-01
6.27096832e-01 -3.35765868e-01 7.43318796e-01 -4.33873646e-02
-1.88155010e-01 1.93073735e-01 -3.65425199e-01 -4.20056403e-01
-8.51397336e-01 -7.85803869e-02 2.75703788e-01 7.13526383e-02
-1.97688639e-01 -6.06612980e-01 -5.16780615e-01 1.99400216e-01
-1.34230793e+00 -4.34533298e-01 4.94213879e-01 9.10612643e-01
2.89863437e-01 4.25724715e-01 5.55281043e-01 -1.01247513e+00
1.06090105e+00 -5.45962572e-01 -6.74171746e-01 3.33310395e-01
-3.06693971e-01 1.48803582e-02 1.28671253e+00 -7.49556959e-01
-8.27795446e-01 -1.80251807e-01 -5.33799887e-01 -5.62302232e-01
7.43160546e-02 7.08732903e-01 -6.46063209e-01 -3.09922159e-01
1.04462647e+00 3.40732366e-01 -2.56031156e-01 -1.20512471e-02
6.98075414e-01 3.10587674e-01 4.81602490e-01 -9.91604447e-01
1.55645990e+00 1.75049603e-01 -6.95503429e-02 -4.83481526e-01
-6.77144885e-01 1.17659427e-01 -1.40805468e-01 6.30662069e-02
4.59449083e-01 -5.93678594e-01 -7.83273518e-01 6.99365973e-01
-1.21870840e+00 -3.63714695e-01 -8.38545486e-02 1.48741409e-01
-4.33874905e-01 9.08065021e-01 -7.41131961e-01 -8.21102440e-01
-7.33528912e-01 -1.35487092e+00 7.71090925e-01 -1.70551166e-01
-8.20769295e-02 -9.81465459e-01 2.37890229e-01 4.50043529e-01
3.37039709e-01 6.57341778e-01 9.91442084e-01 -9.27083850e-01
-3.18528652e-01 -7.47081935e-01 6.12356514e-02 6.49759829e-01
-1.90591976e-01 1.60416633e-01 -8.21906149e-01 -5.35869002e-01
3.31191808e-01 -5.86530089e-01 4.02251512e-01 -6.26609996e-02
9.70018446e-01 -7.97330976e-01 7.08606094e-02 7.84184813e-01
1.26901889e+00 -2.14065522e-01 6.51707292e-01 2.72518992e-01
7.51015544e-01 5.51922560e-01 5.04622936e-01 4.39984083e-01
-1.84515014e-01 6.79723561e-01 6.12979650e-01 4.24098335e-02
2.92856872e-01 -6.55638278e-01 7.02584028e-01 4.94085371e-01
4.75874543e-01 -6.14962339e-01 -9.40723717e-01 1.50611356e-01
-1.63064063e+00 -1.06091666e+00 2.82561600e-01 2.24985933e+00
1.21064055e+00 4.32001263e-01 -1.90793797e-01 2.77811378e-01
6.92249596e-01 3.96148533e-01 -7.02323973e-01 -4.13807541e-01
-2.15463296e-01 9.12543163e-02 4.05062050e-01 6.70478582e-01
-1.22892630e+00 1.11810005e+00 6.00205946e+00 1.19536912e+00
-1.19811189e+00 -7.54688457e-02 6.58775985e-01 -3.75525467e-02
-4.90894884e-01 -1.99993879e-01 -7.93695688e-01 6.58010662e-01
8.27052176e-01 -2.24280357e-01 5.83128512e-01 8.04379761e-01
2.68779606e-01 6.85687423e-01 -9.67472076e-01 6.58292413e-01
-3.30761187e-02 -1.02086043e+00 3.03861916e-01 9.06299129e-02
7.56432414e-01 -3.52582246e-01 6.06460214e-01 2.85559952e-01
6.21245801e-01 -1.15458345e+00 7.91256070e-01 1.40194207e-01
7.01713443e-01 -1.03073359e+00 5.00321150e-01 5.16069829e-01
-7.61776447e-01 -1.40578344e-01 -3.02614391e-01 3.13558728e-01
2.00913027e-01 7.26206183e-01 -8.43885243e-01 6.33427858e-01
3.50861013e-01 2.99121708e-01 -1.81550026e-01 4.96107072e-01
-7.32887924e-01 7.11803198e-01 -3.22433263e-01 8.25798437e-02
3.34181547e-01 -1.22434132e-01 1.08410668e+00 1.09614396e+00
-2.97487229e-02 1.69382561e-02 1.88237682e-01 9.79610085e-01
-3.49562913e-01 1.19483858e-01 -7.09053159e-01 -4.13837045e-01
6.17812812e-01 1.27243114e+00 -2.42111608e-01 1.89360246e-01
6.34809658e-02 9.56127286e-01 4.14895713e-01 6.10650659e-01
-1.18032575e+00 -4.62831736e-01 7.99860597e-01 -1.61346853e-01
1.29589224e-02 -2.86926150e-01 -1.97577059e-01 -1.24606442e+00
4.02021110e-01 -1.66065419e+00 2.87102133e-01 -2.32169062e-01
-1.66987848e+00 7.74383128e-01 -1.86911628e-01 -1.12681627e+00
-2.73604870e-01 -6.24959648e-01 -8.52620542e-01 1.10532677e+00
-1.40121269e+00 -1.23324299e+00 3.20800930e-01 7.63870537e-01
2.37434909e-01 -3.78186285e-01 7.98202574e-01 1.44320324e-01
-9.34222043e-01 1.24413383e+00 6.13979697e-02 3.43256772e-01
7.14364350e-01 -1.13469994e+00 6.95359051e-01 1.34345067e+00
1.30505608e-02 8.20174694e-01 9.46582079e-01 -7.43626297e-01
-1.54780591e+00 -1.15066993e+00 3.47393870e-01 -7.17427254e-01
1.00066316e+00 -6.25017345e-01 -8.53922427e-01 3.14864874e-01
-6.87257722e-02 -2.65138727e-02 5.67738235e-01 -2.41478622e-01
-7.16157615e-01 4.35890742e-02 -1.32273233e+00 1.01325548e+00
6.94522262e-01 -7.90032983e-01 -3.62695724e-01 6.45812213e-01
1.02500427e+00 -7.35835850e-01 -7.16679752e-01 5.23947597e-01
1.48695871e-01 -6.08746469e-01 1.12827253e+00 -8.21532607e-01
5.25948822e-01 -2.40872800e-01 -2.05724195e-01 -1.25483489e+00
5.24517447e-02 -1.16669917e+00 -2.24155396e-01 1.40591574e+00
6.19890094e-01 -7.76503742e-01 8.04344714e-01 7.60171115e-01
1.67193040e-02 -9.40027893e-01 -9.48018909e-01 -6.80388629e-01
4.07271206e-01 -6.24221206e-01 4.49872285e-01 9.12056506e-01
2.84075867e-02 4.38323896e-03 -6.06270790e-01 5.41232467e-01
5.55600643e-01 -4.39182103e-01 8.83266270e-01 -5.86942315e-01
-6.70301378e-01 -4.03749198e-01 -3.29412818e-01 -7.03253984e-01
5.55744946e-01 -8.45113993e-01 1.33802354e-01 -8.51788878e-01
-1.05435155e-01 -3.54120642e-01 -1.69438079e-01 2.68937945e-01
-6.48675084e-01 3.44468802e-02 1.88162744e-01 1.10128611e-01
-1.55946612e-01 7.41663814e-01 1.00105131e+00 -2.86999315e-01
-3.52797695e-02 1.87801912e-01 -7.92652786e-01 4.78976130e-01
7.36232698e-01 -5.58190763e-01 -5.33107162e-01 -2.52079546e-01
5.85571587e-01 -4.46889624e-02 3.54082257e-01 -7.33540952e-01
1.78554907e-01 -2.98372418e-01 7.40311593e-02 -1.13347746e-01
1.90719232e-01 -7.41745412e-01 -1.76554710e-01 4.54936951e-01
-4.78137553e-01 9.65010300e-02 2.50805825e-01 7.70377994e-01
-2.17284665e-01 -2.47236162e-01 7.96084762e-01 -2.75073461e-02
-8.60898495e-02 6.36593938e-01 -4.57624206e-03 4.18880075e-01
9.84089792e-01 1.43103912e-01 -4.42064583e-01 -4.71102983e-01
-2.86401451e-01 1.91686824e-01 4.84992534e-01 3.16001534e-01
6.58731699e-01 -1.21378660e+00 -1.02373564e+00 1.81704223e-01
-1.90279543e-01 -1.24953821e-01 3.94643784e-01 3.94851416e-01
-5.80409884e-01 1.13543794e-01 1.18205652e-01 -4.53130677e-02
-8.84840012e-01 8.14281940e-01 4.05304462e-01 -7.52030313e-01
-3.08789283e-01 9.01148677e-01 1.53142542e-01 -7.32452989e-01
4.47959095e-01 4.11072880e-01 1.45982012e-01 -3.51278484e-01
5.64567924e-01 1.05394229e-01 1.35798484e-01 -4.38095987e-01
-2.21005037e-01 1.26273841e-01 -2.03981593e-01 -4.73545104e-01
1.05330026e+00 3.47967118e-01 -6.16878308e-02 -1.49863169e-01
1.11885118e+00 4.52160299e-01 -1.32408130e+00 -1.64190248e-01
-1.63237378e-01 -5.05039990e-01 -2.13394761e-01 -8.57182324e-01
-9.70768332e-01 7.76224494e-01 1.10159755e-01 2.42061496e-01
8.88411403e-01 -6.78042173e-01 9.80881810e-01 3.30474466e-01
3.82366963e-02 -8.62991929e-01 3.93911630e-01 5.90483189e-01
1.23318136e+00 -8.99687529e-01 1.59135703e-02 -2.92367011e-01
-8.47922146e-01 8.70440543e-01 4.32617396e-01 -8.36696401e-02
3.50479186e-01 3.48828167e-01 1.64631069e-01 1.45180300e-01
-5.95662475e-01 4.96121734e-01 4.52144802e-01 6.21484816e-01
1.37411281e-01 -2.33493835e-01 -1.59174949e-01 5.57016075e-01
-3.71316612e-01 -6.24484301e-01 1.60114095e-01 6.18280172e-01
9.76483449e-02 -1.40006554e+00 -4.87667203e-01 -1.16849162e-01
-8.02053809e-01 -3.15150976e-01 -5.86986244e-01 4.81013119e-01
-2.55595297e-01 1.03757632e+00 -8.29832077e-01 -4.89117295e-01
3.94832522e-01 -8.06585327e-03 9.83816311e-02 -4.87819940e-01
-9.55876589e-01 -3.89511585e-01 -1.77676696e-02 -4.87369001e-01
3.86399627e-01 -3.91413271e-01 -8.21290553e-01 -5.62040925e-01
-3.46627176e-01 3.33769768e-01 7.38874435e-01 9.32590663e-01
2.74746269e-01 2.08750725e-01 1.11676478e+00 -7.23350286e-01
-1.43580532e+00 -7.71255553e-01 -2.76329935e-01 6.87481761e-01
2.58789271e-01 -1.52389869e-01 -7.69565642e-01 -3.24702203e-01] | [6.005665302276611, 8.090240478515625] |
c5fba83d-b3dd-4ac5-84d0-f76b2c92e744 | deep-hierarchical-representation-of-point | null | null | https://ieeexplore.ieee.org/document/9650574 | https://ieeexplore.ieee.org/document/9650574 | Deep Hierarchical Representation of Point Cloud Videos via Spatio-Temporal Decomposition | In point cloud videos, point coordinates are irregular and unordered but point timestamps exhibit regularities and order. Grid-based networks for conventional video processing cannot be directly used to model raw point cloud videos. Therefore, in this work, we propose a point-based network that directly handles raw point cloud videos. First, to preserve the spatio-temporal local structure of point cloud videos, we design a point tube covering a local range along spatial and temporal dimensions. By progressively subsampling frames and points and enlarging the spatial radius as the point features are fed into higher-level layers, the point tube can capture video structure in a spatio-temporally hierarchical manner. Second, to reduce the impact of the spatial irregularity on temporal modeling, we decompose space and time when extracting point tube representations. Specifically, a spatial operation is employed to capture the local structure of each spatial region in a tube and a temporal operation is used to model the dynamics of the spatial regions along the tube. | ['Mohan', 'Yi; Kankanhalli', 'Xin; Yang', 'Hehe; Yu', 'Fan'] | 2021-12-14 | null | null | null | ieee-transactions-on-pattern-analysis-and-20 | ['3d-human-action-recognition'] | ['computer-vision'] | [-2.72784710e-01 -5.63398123e-01 -1.76275477e-01 -5.68402968e-02
2.22469464e-01 -5.93528152e-01 3.55927080e-01 3.50897849e-01
-2.46640623e-01 3.44580412e-01 -4.75783134e-03 -1.68120191e-01
-2.43341222e-01 -1.14658642e+00 -8.42176735e-01 -4.17447776e-01
-5.23942649e-01 5.90334088e-02 7.18213320e-01 1.81051373e-01
1.75535992e-01 1.26565492e+00 -1.53012872e+00 3.12981814e-01
5.96775055e-01 1.02419686e+00 2.36245960e-01 6.13303423e-01
-5.00535786e-01 6.49425089e-01 -3.85302037e-01 2.90046871e-01
2.65164763e-01 -1.10682622e-01 -5.94426036e-01 4.68954027e-01
2.32217878e-01 -5.33485115e-01 -8.10885847e-01 8.24413002e-01
-9.82845649e-02 4.38300401e-01 3.46476942e-01 -1.36152947e+00
-1.16630554e-01 5.12110777e-02 -7.04578400e-01 2.56469637e-01
1.52978688e-01 7.21308887e-02 5.40679038e-01 -8.50988090e-01
7.47349799e-01 1.14110541e+00 7.18314767e-01 8.09805170e-02
-8.75374913e-01 -6.34677887e-01 4.82162714e-01 1.99578732e-01
-1.57254767e+00 -2.13165823e-02 1.11515510e+00 -3.93259823e-01
8.05983365e-01 3.88823211e-01 1.15283477e+00 4.41666275e-01
9.41996127e-02 4.09873337e-01 1.78614125e-01 1.41900584e-01
1.95527360e-01 -6.94965780e-01 -2.18552366e-01 5.14290750e-01
-1.40201390e-01 2.16697436e-02 -5.69748938e-01 -3.47399563e-01
1.63077414e+00 8.80683720e-01 -1.93730876e-01 -4.55623567e-01
-1.55666685e+00 5.35734355e-01 5.49523354e-01 4.36860859e-01
-6.75129771e-01 4.77432400e-01 3.71081054e-01 1.30404696e-01
6.66033745e-01 -1.58954591e-01 -3.99660289e-01 -1.93640307e-01
-1.31015217e+00 1.34485736e-01 3.36157948e-01 1.32471418e+00
1.00516891e+00 -2.19070643e-01 -1.05475344e-01 5.35998881e-01
1.34651110e-01 1.14144467e-01 -2.84708478e-02 -1.35189259e+00
5.40627480e-01 8.65337074e-01 7.93143660e-02 -1.41050661e+00
-3.27813476e-01 3.75566259e-03 -1.29953790e+00 2.55907606e-02
1.96605444e-01 9.76159796e-02 -6.59515679e-01 1.15423656e+00
5.11792779e-01 9.82792735e-01 -4.29330200e-01 6.85988843e-01
4.97912407e-01 1.13691485e+00 3.89426872e-02 -3.87790233e-01
1.22951853e+00 -5.30168712e-01 -3.12082469e-01 4.76883024e-01
3.56298923e-01 -4.95706946e-01 8.40760648e-01 -9.63945910e-02
-1.25356805e+00 -4.38735723e-01 -7.09157586e-01 -1.91961497e-01
-1.45306572e-01 -1.56205818e-01 3.03264976e-01 -3.90571237e-01
-1.00657117e+00 8.39858055e-01 -1.28422558e+00 -3.27285565e-02
3.07556570e-01 2.11038694e-01 -2.57636219e-01 -1.83607921e-01
-8.96675169e-01 -2.61989292e-02 2.32687280e-01 1.18328579e-01
-3.60599548e-01 -8.88121963e-01 -7.79245913e-01 3.34098011e-01
1.80354446e-01 -6.52228653e-01 7.97465205e-01 -4.46681887e-01
-9.59569037e-01 3.13574612e-01 -6.62567377e-01 -3.60334069e-01
4.70176846e-01 2.38861158e-01 -3.45141143e-01 4.77700710e-01
1.11629039e-01 6.61349118e-01 7.99253047e-01 -1.25045490e+00
-9.21509385e-01 -2.67554611e-01 1.78157166e-01 2.09939867e-01
-2.03750670e-01 -1.74876302e-01 -9.29417849e-01 -8.40809882e-01
7.45710969e-01 -7.95622706e-01 -2.37276539e-01 4.82823789e-01
-1.30388752e-01 -3.38548541e-01 1.25831831e+00 -3.67180437e-01
1.47235620e+00 -2.57477260e+00 5.07644042e-02 3.71772408e-01
3.97743970e-01 -1.20992728e-01 2.44264491e-02 4.88800526e-01
5.61502054e-02 2.82635152e-01 -9.36448947e-02 -3.10947567e-01
-4.26122904e-01 5.73615909e-01 -3.31749529e-01 4.95540857e-01
1.53616488e-01 5.20673454e-01 -1.00839806e+00 -5.47094941e-01
5.71079016e-01 7.65800953e-01 -7.48850763e-01 -1.27838954e-01
-1.79127902e-01 4.26741391e-01 -7.17672408e-01 5.60161948e-01
8.53381693e-01 -3.43198895e-01 -2.48973981e-01 -4.53011900e-01
-6.70352042e-01 4.79228571e-02 -1.12323916e+00 1.62575841e+00
-3.12245786e-01 5.07726192e-01 7.25962967e-02 -4.63834822e-01
7.52158821e-01 4.50534195e-01 1.23268962e+00 -3.51188630e-01
-3.13767999e-01 -7.43064880e-02 -5.41385233e-01 -3.38854551e-01
7.25692809e-01 2.00269133e-01 1.14773892e-01 1.07801467e-01
-4.53993887e-01 -1.97967254e-02 3.87292132e-02 1.50226593e-01
1.12833381e+00 -7.08297864e-02 -1.46823674e-01 6.67353496e-02
5.46949029e-01 3.80230904e-01 6.82272375e-01 3.38855654e-01
-1.80980861e-02 9.00836587e-01 4.64211315e-01 -8.38824272e-01
-1.29419041e+00 -1.05974674e+00 9.44105349e-03 6.80198848e-01
5.76447129e-01 -5.79024196e-01 -3.89974028e-01 -1.99727133e-01
9.27044451e-02 2.53863037e-01 -3.90845776e-01 4.19184007e-02
-1.00135887e+00 -2.37165734e-01 1.46453738e-01 6.54247761e-01
5.70684433e-01 -9.66795921e-01 -5.29639900e-01 4.79451925e-01
-2.12297276e-01 -1.05822897e+00 -8.51488769e-01 -1.84719756e-01
-1.33824790e+00 -9.73373473e-01 -5.06351292e-01 -7.74344325e-01
7.60606706e-01 8.71981859e-01 9.55920279e-01 4.57414538e-01
3.25415909e-01 1.80254921e-01 -3.09544116e-01 -1.09052565e-02
1.14676498e-01 -3.20501551e-02 -4.40069884e-02 -9.39747989e-02
3.63649547e-01 -1.06588185e+00 -8.92546654e-01 4.07484919e-01
-1.13413513e+00 2.35743999e-01 -1.00007895e-02 3.64498287e-01
1.02485120e+00 5.06829739e-01 -1.76458448e-01 -1.31244570e-01
2.44938478e-01 -6.40471816e-01 -6.87986434e-01 -1.27260938e-01
2.29823202e-01 -4.23499972e-01 5.88571727e-01 -5.98839760e-01
-2.97770143e-01 2.50323176e-01 2.18390957e-01 -1.10920238e+00
-7.63561279e-02 4.54443067e-01 8.46114308e-02 2.51042210e-02
-3.99597511e-02 3.77640456e-01 -1.18458763e-01 -5.23565054e-01
1.12464592e-01 2.36950517e-01 5.56943774e-01 -5.64828813e-01
8.65039408e-01 1.00103688e+00 2.09128693e-01 -9.63210344e-01
-2.87677586e-01 -6.74398899e-01 -1.03881729e+00 -4.04987514e-01
1.01909721e+00 -9.03059304e-01 -8.63261878e-01 2.18001083e-01
-1.49792826e+00 -1.09982342e-01 -4.40030098e-01 3.88418019e-01
-3.65991920e-01 5.05908012e-01 -7.83122241e-01 -4.77715343e-01
1.27914801e-01 -8.77069473e-01 1.19445193e+00 9.00654644e-02
-1.36234537e-01 -8.39073002e-01 7.24511687e-03 -4.25395906e-01
-7.12669268e-02 3.80312085e-01 8.83478940e-01 1.06440045e-01
-1.10073090e+00 -4.49279323e-02 -3.11185867e-01 -7.64144734e-02
3.76617432e-01 4.24117297e-01 -2.61187494e-01 -1.52263522e-01
1.04415372e-01 6.84340954e-01 3.19114387e-01 5.70309162e-01
1.61744130e+00 -4.19731319e-01 -3.91993552e-01 9.95209217e-01
1.35241330e+00 3.02813798e-01 4.98388916e-01 4.63327229e-01
1.05235231e+00 4.14050728e-01 3.90761554e-01 5.36984622e-01
5.29372036e-01 4.33017433e-01 4.71144706e-01 -1.01650842e-02
1.53418913e-01 -5.06683290e-01 -6.23735115e-02 9.41752553e-01
-4.45543289e-01 -1.24951690e-01 -9.00130510e-01 7.03077853e-01
-1.94351554e+00 -1.21275663e+00 -2.79828072e-01 2.24755645e+00
3.27446401e-01 -1.13314644e-01 1.56358168e-01 1.19765110e-01
9.13411081e-01 4.20078039e-01 -4.73429888e-01 2.27480084e-01
-3.98471765e-02 -4.29220706e-01 6.92584217e-01 2.69282788e-01
-8.27827573e-01 6.88921571e-01 5.96092892e+00 8.02218735e-01
-1.17525434e+00 -8.24996755e-02 3.11355084e-01 -4.78773952e-01
-2.77508587e-01 4.49391715e-02 -3.60656649e-01 6.88055575e-01
5.20291388e-01 -4.09828089e-02 4.14409816e-01 6.71021581e-01
7.45321572e-01 2.20878720e-01 -9.25857365e-01 1.10970902e+00
-5.87486327e-01 -1.79014969e+00 3.09609771e-01 3.68166924e-01
6.11831427e-01 1.36862159e-01 -2.47224346e-01 -1.58132091e-01
-1.82610080e-02 -6.60687447e-01 7.39469707e-01 5.97604632e-01
7.73147285e-01 -9.35535252e-01 1.61108091e-01 6.04518473e-01
-1.80397284e+00 1.08783893e-01 -6.04168952e-01 -4.77806218e-02
5.30155540e-01 4.68489975e-01 -1.50211707e-01 5.28393328e-01
9.15571988e-01 1.15676069e+00 -1.33795664e-01 1.11070275e+00
3.52417380e-01 2.69070685e-01 -7.61159122e-01 4.70867991e-01
6.09766722e-01 -5.74769199e-01 6.55141771e-01 9.71899331e-01
7.14146495e-01 4.45886821e-01 2.83000946e-01 6.48752749e-01
2.09597480e-02 -3.58641177e-01 -7.93087363e-01 3.23860705e-01
9.28022981e-01 7.83888161e-01 -8.55443358e-01 -4.98512298e-01
-6.79121137e-01 7.33535767e-01 3.14510651e-02 5.66103458e-01
-7.61059165e-01 -1.24991685e-01 9.32966292e-01 6.11913443e-01
3.90225172e-01 -9.72719133e-01 -3.09668094e-01 -1.08017027e+00
1.47587478e-01 -3.47243249e-01 1.47165105e-01 -8.21719170e-01
-1.05674815e+00 6.03670001e-01 2.99609154e-01 -1.98127246e+00
-8.08350071e-02 -1.10099897e-01 -6.93311572e-01 8.60549212e-01
-1.31761611e+00 -8.44846606e-01 -5.46890020e-01 1.14429867e+00
4.98739362e-01 3.21202397e-01 1.67832270e-01 3.83943677e-01
-3.04056346e-01 -1.42039686e-01 1.93625093e-01 2.16709837e-01
8.24547652e-03 -6.76673234e-01 6.44639969e-01 8.11770380e-01
2.67399289e-02 7.81029820e-01 3.83514762e-01 -8.65174592e-01
-1.26502097e+00 -1.38434303e+00 9.65866804e-01 -1.30447388e-01
6.88626647e-01 -2.66135424e-01 -1.42102706e+00 6.69569373e-01
-4.25521851e-01 4.21860278e-01 2.37329707e-01 -4.22495276e-01
-1.42667532e-01 -7.06283227e-02 -1.08232665e+00 7.37573564e-01
1.32680392e+00 -6.39839232e-01 -3.19183081e-01 3.53939921e-01
9.70378041e-01 -5.26631594e-01 -1.04268217e+00 4.91610885e-01
2.65268147e-01 -8.20564389e-01 1.21394765e+00 -2.18572631e-01
2.91685313e-01 -8.02326202e-01 -2.21742876e-02 -9.65126932e-01
-6.00938916e-01 -5.42165816e-01 -3.12339216e-01 9.78815973e-01
-2.14520365e-01 -2.43456423e-01 1.04999650e+00 4.55991715e-01
-1.15148537e-01 -6.87152684e-01 -1.23506773e+00 -7.45150506e-01
-2.31521860e-01 -6.35275304e-01 1.18656898e+00 8.49964142e-01
-1.93009585e-01 -2.23496720e-01 -1.70269072e-01 4.90925282e-01
5.25830269e-01 4.40019965e-02 6.49437428e-01 -1.24464929e+00
4.69294131e-01 -4.68179852e-01 -5.44738472e-01 -1.53235483e+00
-1.65969789e-01 -3.14654708e-01 -2.53149420e-01 -1.52407730e+00
-2.17676312e-01 -5.66225171e-01 -9.94185731e-02 5.18775024e-02
2.24849448e-01 1.61418200e-01 3.22329223e-01 8.79068971e-01
-2.74751991e-01 4.91244555e-01 1.40226102e+00 9.25627649e-02
-5.10150671e-01 4.71045636e-02 1.62771463e-01 8.95162582e-01
6.01841271e-01 -4.46697950e-01 -5.06839395e-01 -8.58865321e-01
5.72541282e-02 3.93209785e-01 6.86651647e-01 -1.14667380e+00
5.72970688e-01 -4.88058925e-01 3.80805880e-01 -1.37305069e+00
5.66783488e-01 -1.26589394e+00 6.31402016e-01 1.79842621e-01
9.28868204e-02 5.93126178e-01 8.24704245e-02 7.06489980e-01
-5.14601827e-01 1.19979814e-01 4.14845854e-01 -3.00479025e-01
-5.04055440e-01 1.12431729e+00 -3.94847006e-01 -6.34341002e-01
1.07105064e+00 -6.64900601e-01 1.50323808e-01 -2.32528433e-01
-6.32334888e-01 4.03882265e-01 9.73127544e-01 4.44455564e-01
7.90233970e-01 -1.60866010e+00 -1.59731150e-01 3.20135623e-01
-5.55379614e-02 7.88393378e-01 6.54716551e-01 5.43107152e-01
-1.03249919e+00 1.82196409e-01 -8.97732079e-02 -9.90462542e-01
-1.21938431e+00 7.18531668e-01 2.03106046e-01 2.83189639e-02
-1.20261645e+00 3.84026855e-01 2.72993028e-01 1.68642607e-02
2.31071740e-01 -8.78398478e-01 -2.05137715e-01 -1.28668100e-01
5.12817144e-01 4.61917460e-01 -2.69660741e-01 -9.02541399e-01
-1.89678252e-01 9.46542323e-01 3.46707016e-01 9.12085474e-02
1.39028156e+00 -4.12032843e-01 -5.28360486e-01 6.73114300e-01
1.27124906e+00 -9.55404267e-02 -1.66201711e+00 -1.44080549e-01
-1.67682245e-01 -6.28726304e-01 -2.09977850e-02 2.81436503e-01
-1.22357678e+00 6.99473560e-01 3.24387290e-02 5.02363980e-01
1.11648667e+00 -4.40862626e-02 1.04735935e+00 -1.84589345e-02
5.30373931e-01 -7.07757294e-01 -3.87581289e-01 7.93329954e-01
6.10376298e-01 -4.92756188e-01 -6.97034001e-02 -8.38769853e-01
-1.76191345e-01 1.23236251e+00 4.99734014e-01 -3.33363265e-01
9.19825196e-01 1.39300629e-01 -4.35524672e-01 -4.13897008e-01
-6.37828529e-01 1.67846248e-01 1.79771990e-01 5.70398748e-01
1.55277878e-01 -1.48434833e-01 3.83972265e-02 -1.69641599e-02
-1.42991289e-01 1.95028767e-01 2.13569835e-01 1.03116739e+00
-4.78433341e-01 -7.71913946e-01 -5.17030537e-01 3.06017995e-01
-1.81770727e-01 8.74499977e-02 1.16218761e-01 7.79543042e-01
3.00436378e-01 6.29901767e-01 8.78852248e-01 -4.09002393e-01
4.94315624e-01 -4.36161250e-01 5.20789884e-02 -4.71827537e-01
-4.09151465e-01 2.49648228e-01 -5.53245127e-01 -7.28028893e-01
-5.51975369e-01 -6.17385745e-01 -1.41694272e+00 -7.54788756e-01
2.69865781e-01 1.92092896e-01 4.88872588e-01 6.74159646e-01
6.36264324e-01 4.89008874e-01 6.57369375e-01 -1.30840254e+00
2.93790817e-01 -6.42477214e-01 -6.64397120e-01 2.84224063e-01
6.70706034e-01 -4.97379988e-01 -3.16170722e-01 1.09728821e-01] | [8.45756721496582, -2.0457632541656494] |
d2ecda2f-a869-4b5c-a261-f3edb09f7022 | when-search-meets-recommendation-learning | 2305.10822 | null | https://arxiv.org/abs/2305.10822v1 | https://arxiv.org/pdf/2305.10822v1.pdf | When Search Meets Recommendation: Learning Disentangled Search Representation for Recommendation | Modern online service providers such as online shopping platforms often provide both search and recommendation (S&R) services to meet different user needs. Rarely has there been any effective means of incorporating user behavior data from both S&R services. Most existing approaches either simply treat S&R behaviors separately, or jointly optimize them by aggregating data from both services, ignoring the fact that user intents in S&R can be distinctively different. In our paper, we propose a Search-Enhanced framework for the Sequential Recommendation (SESRec) that leverages users' search interests for recommendation, by disentangling similar and dissimilar representations within S&R behaviors. Specifically, SESRec first aligns query and item embeddings based on users' query-item interactions for the computations of their similarities. Two transformer encoders are used to learn the contextual representations of S&R behaviors independently. Then a contrastive learning task is designed to supervise the disentanglement of similar and dissimilar representations from behavior sequences of S&R. Finally, we extract user interests by the attention mechanism from three perspectives, i.e., the contextual representations, the two separated behaviors containing similar and dissimilar interests. Extensive experiments on both industrial and public datasets demonstrate that SESRec consistently outperforms state-of-the-art models. Empirical studies further validate that SESRec successfully disentangle similar and dissimilar user interests from their S&R behaviors. | ['Ji-Rong Wen', 'Kun Gai', 'Yang song', 'Xiaoxue Zang', 'Jun Xu', 'Xiao Zhang', 'Zhongxiang Sun', 'Zihua Si'] | 2023-05-18 | null | null | null | null | ['disentanglement', 'sequential-recommendation'] | ['methodology', 'miscellaneous'] | [-1.25394300e-01 -5.59587240e-01 -7.94866860e-01 -9.00216401e-01
-5.46636283e-01 -5.68321407e-01 7.07141995e-01 -7.37393573e-02
-3.11944157e-01 -1.74275398e-01 7.18598187e-01 -2.91051239e-01
-2.90841252e-01 -5.68543434e-01 -3.93421918e-01 -4.30012792e-01
3.84979732e-02 4.41404492e-01 -2.04653576e-01 -5.03328383e-01
3.31717700e-01 3.16080779e-01 -1.64103663e+00 5.06301820e-01
7.80808687e-01 1.32103539e+00 4.01439995e-01 3.62488508e-01
-2.45288059e-01 6.45877659e-01 -1.83129758e-01 -4.29860175e-01
4.82449502e-01 -2.48619244e-01 -3.80102187e-01 2.06929639e-01
2.34882906e-01 -5.51952243e-01 -8.58772099e-01 7.30305314e-01
2.13286310e-01 6.16627455e-01 5.91421187e-01 -1.29980147e+00
-1.66796494e+00 6.60088301e-01 -4.29577321e-01 3.50670785e-01
4.54125077e-01 -5.13148308e-02 1.72276938e+00 -8.45591247e-01
1.57885984e-01 1.11537421e+00 2.36234829e-01 4.62301999e-01
-1.31743932e+00 -7.81801164e-01 7.47676611e-01 2.96047121e-01
-1.20226943e+00 -3.96472484e-01 9.18112576e-01 -2.25453153e-01
1.19846344e+00 3.46409023e-01 4.70655799e-01 1.43784761e+00
-1.31450981e-01 1.50058925e+00 4.26931620e-01 2.61733294e-01
5.48837706e-03 5.92207909e-01 7.03826368e-01 1.60098627e-01
2.20224708e-01 1.89216137e-01 -3.60371888e-01 -2.04175189e-01
6.73283756e-01 1.00863385e+00 -1.61971733e-01 -3.02859753e-01
-9.25715506e-01 1.01475000e+00 4.01957214e-01 4.21403378e-01
-4.15694505e-01 -1.81696355e-01 3.46004128e-01 5.65715075e-01
2.00489536e-01 4.64136273e-01 -7.16411710e-01 1.53178096e-01
-5.03748119e-01 3.57964605e-01 8.14169943e-01 1.47972524e+00
5.12269557e-01 -3.06141544e-02 -2.34953582e-01 9.20705914e-01
5.58196723e-01 3.49780202e-01 9.84545231e-01 -4.25213128e-01
2.75990695e-01 6.31031573e-01 4.16397959e-01 -1.03810990e+00
3.31811719e-02 -5.01210451e-01 -4.83323842e-01 -8.09619486e-01
6.82704598e-02 1.68500796e-01 -7.39153624e-01 1.66569936e+00
-2.46965989e-01 3.13740343e-01 1.56975701e-01 9.33095396e-01
9.65874910e-01 6.96206391e-01 1.15846068e-01 8.75915587e-02
1.29790211e+00 -1.22805738e+00 -7.29995131e-01 -4.72190470e-01
4.82303232e-01 -4.36904043e-01 1.14449310e+00 2.16156617e-01
-1.05382061e+00 -7.21409500e-01 -8.03253472e-01 -3.27820122e-01
-4.72614706e-01 3.43563259e-01 7.60007799e-01 3.93503636e-01
-6.38012826e-01 4.61786538e-01 -5.90901017e-01 -2.52149284e-01
2.38843784e-01 4.72788185e-01 -8.53673369e-02 -3.64879817e-02
-1.40464151e+00 5.44529617e-01 -7.92075843e-02 -5.05030230e-02
-7.23392010e-01 -6.36349857e-01 -1.03628922e+00 5.89066923e-01
4.28330153e-01 -3.28531384e-01 1.41109478e+00 -1.00746465e+00
-1.28459620e+00 6.53260589e-01 -3.61708254e-01 -2.21056983e-01
-7.45777190e-02 -4.32733268e-01 -1.08516049e+00 -4.44090605e-01
8.70967731e-02 1.21802986e-02 6.97753310e-01 -1.16289508e+00
-9.45634604e-01 -5.67904651e-01 1.80109739e-01 2.05296755e-01
-5.74616134e-01 1.91290915e-01 -9.22214329e-01 -7.40580261e-01
6.29415512e-02 -6.57469094e-01 -2.14989275e-01 -4.69581097e-01
-1.32450715e-01 -5.75283349e-01 6.48528099e-01 -5.51387548e-01
1.30413043e+00 -2.32810068e+00 1.46795228e-01 1.78292438e-01
1.31928802e-01 1.51368067e-01 -5.34588635e-01 5.25629103e-01
-6.59126788e-02 5.99776432e-02 4.33238775e-01 -4.60274845e-01
3.48931849e-01 4.27509725e-01 -6.46786153e-01 3.55230540e-01
-3.79170924e-02 1.03712142e+00 -1.13090241e+00 1.31949231e-01
8.79761055e-02 3.06785196e-01 -9.74139273e-01 8.04577768e-01
-4.48121578e-02 1.06189460e-01 -7.90996194e-01 5.95378518e-01
4.82428879e-01 -6.28866017e-01 4.56585437e-01 -3.21678936e-01
9.59189981e-02 7.97504365e-01 -8.93402040e-01 1.39035094e+00
-7.58550465e-01 6.19079135e-02 -8.35508704e-02 -1.22468197e+00
9.47474658e-01 1.29161611e-01 7.41006494e-01 -1.21475434e+00
1.04558200e-01 1.62372261e-01 -1.28648862e-01 -6.31210089e-01
7.36147463e-01 -3.52587290e-02 -4.15990323e-01 6.66064620e-01
1.94428980e-01 4.91657943e-01 -1.52175903e-01 9.86566991e-02
8.43340933e-01 7.09015690e-03 3.69223624e-01 -1.18383832e-01
5.11867702e-01 -5.59627891e-01 7.26891339e-01 9.15486872e-01
-2.42657691e-01 2.73977757e-01 2.93292757e-02 -3.78237069e-01
-6.40473068e-01 -1.13551116e+00 9.79272723e-02 1.67827988e+00
7.07220376e-01 -4.75033611e-01 9.01614577e-02 -1.17071021e+00
4.65698719e-01 1.04141891e+00 -6.88396096e-01 -4.82285023e-01
-6.72042251e-01 -4.11916167e-01 2.48047654e-02 6.80239201e-01
1.42739927e-02 -1.08668888e+00 7.22477958e-02 1.31632537e-01
7.18928501e-02 -1.04574406e+00 -1.19636476e+00 2.06218496e-01
-6.39786780e-01 -8.14057589e-01 -5.55667222e-01 -8.12904894e-01
6.09493315e-01 9.92556930e-01 1.31191254e+00 3.22831571e-02
1.66539505e-01 4.83836710e-01 -5.90479791e-01 9.29265991e-02
3.85544635e-02 -7.46537838e-03 2.53081918e-01 3.85282665e-01
1.20233572e+00 -6.37480617e-01 -9.08344567e-01 9.52272356e-01
-9.94686067e-01 -3.50957602e-01 6.49725556e-01 6.91017687e-01
4.28251058e-01 9.16487910e-03 7.07702458e-01 -1.10849893e+00
7.37067580e-01 -1.33337772e+00 -6.82171062e-02 2.54603297e-01
-8.93717885e-01 3.84474024e-02 9.66705620e-01 -6.46402776e-01
-9.30882931e-01 -5.34285188e-01 -3.30500275e-01 -5.74456453e-01
-2.57773221e-01 3.70211273e-01 -3.39425564e-01 6.56813443e-01
1.82367325e-01 5.93590379e-01 -2.74895221e-01 -8.33668828e-01
5.09220898e-01 1.11112452e+00 1.00660823e-01 -3.19264978e-01
4.42020565e-01 3.47727776e-01 -1.08178473e+00 -5.86179078e-01
-8.80410790e-01 -1.15896297e+00 -1.89456064e-02 3.50840837e-01
4.96886283e-01 -7.20675588e-01 -7.56827295e-01 1.55373320e-01
-6.22890413e-01 1.56396329e-01 -2.82047153e-01 5.49881220e-01
-4.82862681e-01 2.82914072e-01 -6.27211452e-01 -6.49183631e-01
-1.21621065e-01 -1.28567457e+00 1.13819087e+00 3.38796288e-01
-2.53943622e-01 -9.63127375e-01 -2.19239593e-01 4.48669553e-01
5.55987000e-01 -7.43533194e-01 9.14025128e-01 -1.55967295e+00
-5.12877226e-01 -4.93838876e-01 -3.67314935e-01 3.70657295e-01
6.73058569e-01 -5.57343841e-01 -6.04597390e-01 -4.72131789e-01
1.44439578e-01 -5.09441793e-02 7.58117259e-01 1.47982150e-01
1.44363070e+00 -6.78155422e-01 -3.09820622e-01 5.72164893e-01
1.21246302e+00 7.41657794e-01 4.84944075e-01 -2.11972129e-02
6.41643107e-01 6.78417563e-01 6.25108838e-01 4.06291574e-01
6.17587745e-01 9.29393351e-01 3.65480304e-01 2.39349529e-01
3.44248444e-01 -5.70547938e-01 6.20222330e-01 8.65005136e-01
1.82948396e-01 -4.57503796e-01 -1.82849184e-01 5.29336393e-01
-1.97092581e+00 -9.92737055e-01 3.83237511e-01 2.07009912e+00
4.14115787e-01 -1.27525270e-01 1.80190444e-01 -4.55006480e-01
5.81911504e-01 4.34124202e-01 -1.06314695e+00 -3.76966059e-01
4.46671933e-01 -7.24727064e-02 4.70227271e-01 2.19032615e-01
-7.69946933e-01 6.93813622e-01 5.65392780e+00 5.85690141e-01
-8.15983534e-01 1.24314763e-02 4.08519924e-01 -4.84257698e-01
-1.03495514e+00 -2.30015457e-01 -1.03492069e+00 7.66772509e-01
8.20296407e-01 -2.54279375e-01 7.21298933e-01 1.20946050e+00
1.43024325e-01 7.28097916e-01 -1.55560350e+00 1.14357412e+00
4.13647771e-01 -1.07768774e+00 2.59108275e-01 1.28685609e-01
5.04731953e-01 8.85530040e-02 3.91556829e-01 1.00697863e+00
6.35094583e-01 -7.66900182e-01 5.21892965e-01 4.85037386e-01
2.60498255e-01 -5.40981710e-01 6.21063352e-01 3.95676047e-01
-1.19107687e+00 -5.72732091e-01 -1.09782547e-01 2.31576651e-01
3.27559203e-01 -3.74370292e-02 -2.06945270e-01 5.10140479e-01
7.79291391e-01 1.33849490e+00 -2.02467889e-01 4.41316724e-01
1.89422935e-01 3.32823098e-01 1.91966280e-01 -2.70669520e-01
3.34805787e-01 -6.81661963e-01 4.49589610e-01 1.09441245e+00
3.98247331e-01 6.33775145e-02 1.81924164e-01 9.24452603e-01
-2.54165530e-01 1.98343128e-01 -4.54756469e-01 -4.58061516e-01
3.92746240e-01 1.03843844e+00 -9.99277905e-02 -4.13484126e-01
-1.03129542e+00 1.18958116e+00 3.40363115e-01 5.27131081e-01
-9.05129850e-01 -1.97798222e-01 1.44485724e+00 3.25645477e-01
6.80202365e-01 1.77335329e-02 1.77596092e-01 -1.65002394e+00
-1.86614946e-01 -1.04411578e+00 4.15046185e-01 -5.31069338e-01
-1.93089747e+00 4.91651475e-01 -1.42841533e-01 -1.25339925e+00
-2.80204684e-01 -5.14271915e-01 -5.46214223e-01 9.50329602e-01
-1.61228883e+00 -1.09136593e+00 2.69465279e-02 5.82040012e-01
9.47407186e-01 -3.93977165e-01 6.57873631e-01 5.45605779e-01
-7.15223551e-01 9.10737038e-01 2.94590741e-01 -3.13387997e-02
4.88527000e-01 -1.09801280e+00 6.66310251e-01 3.73721689e-01
2.79600590e-01 1.30327344e+00 3.96476179e-01 -4.05044168e-01
-1.78830969e+00 -9.49027479e-01 1.00840294e+00 -4.69631582e-01
7.32916713e-01 -4.60943699e-01 -8.95736098e-01 1.10800052e+00
-1.17829246e-02 -3.37819569e-02 1.30772233e+00 4.90185112e-01
-6.59717977e-01 -1.81778491e-01 -9.95396852e-01 8.71411443e-01
1.24658835e+00 -7.89309561e-01 -8.36646616e-01 4.17518131e-02
1.00318253e+00 1.90815955e-01 -7.79043496e-01 1.89421460e-01
7.44670987e-01 -1.00285876e+00 1.29135656e+00 -1.22659874e+00
3.94545376e-01 1.92343667e-01 -6.76305830e-01 -1.14108992e+00
-6.65478110e-01 -3.60857666e-01 -4.81476426e-01 1.04418516e+00
2.37095803e-01 -7.46153414e-01 7.72469044e-01 7.81245947e-01
-1.28985271e-01 -8.96066606e-01 -2.67342240e-01 -5.83763421e-01
-3.09190303e-01 -4.06717300e-01 1.09270942e+00 1.05245817e+00
-7.14816712e-03 4.59335804e-01 -6.11343563e-01 1.86120525e-01
2.06782609e-01 7.40354061e-01 5.88711977e-01 -9.96522903e-01
-7.38022447e-01 -6.06390715e-01 -2.77610589e-02 -1.98694730e+00
1.48138136e-01 -1.05495739e+00 -2.99811751e-01 -1.32047200e+00
2.05021366e-01 -5.02692521e-01 -1.03635979e+00 9.17740837e-02
-1.03298269e-01 -1.75669461e-01 -1.35478586e-01 4.16102678e-01
-7.19742894e-01 6.89784884e-01 1.26860225e+00 -5.54804131e-02
-4.47611302e-01 4.59501892e-01 -1.51445055e+00 5.86378694e-01
5.87309420e-01 -7.92255625e-02 -7.75625050e-01 -3.88142049e-01
2.77042538e-01 3.49593349e-02 -1.28036693e-01 -1.38667256e-01
-7.04773515e-02 -4.35163915e-01 7.25224242e-02 -4.91559595e-01
2.21099272e-01 -1.16077197e+00 -2.83782691e-01 -5.25905527e-02
-9.63675916e-01 -6.05182815e-03 -2.69142926e-01 9.68268812e-01
-1.71232253e-01 -2.12482333e-01 2.95369506e-01 -1.54496759e-01
-8.53016615e-01 7.60221362e-01 -2.72236466e-01 -1.70108035e-01
7.73018181e-01 -6.36592433e-02 1.53953517e-02 -6.16987526e-01
-7.33586133e-01 4.31148261e-01 7.61519372e-02 1.22169495e+00
7.32823730e-01 -1.63120413e+00 -3.03298414e-01 6.81891859e-01
5.18407583e-01 -5.61191976e-01 4.00325596e-01 3.29930186e-01
4.44821447e-01 6.09728038e-01 2.83681303e-02 -9.60821286e-02
-1.09744871e+00 1.00649178e+00 1.45672873e-01 -4.10952449e-01
-3.72975439e-01 7.83497274e-01 4.60669905e-01 -7.53697276e-01
2.65741140e-01 -4.30333197e-01 -6.75725937e-01 1.61800310e-01
6.93412244e-01 2.70423859e-01 -1.87460348e-01 -8.48731160e-01
-2.11938992e-01 1.40685573e-01 -8.55908632e-01 3.13904613e-01
1.47624516e+00 -4.15449589e-01 3.48417819e-01 2.75202274e-01
1.73526907e+00 1.04869574e-01 -9.65067327e-01 -7.06143677e-01
-1.82257578e-01 -9.60260451e-01 1.12220585e-01 -4.81484383e-01
-1.23020446e+00 5.56733072e-01 2.38447517e-01 7.16046035e-01
1.17523575e+00 2.81776935e-01 1.22596002e+00 2.78535306e-01
3.92566413e-01 -9.40720081e-01 2.34134153e-01 3.16017926e-01
8.34266603e-01 -1.35615396e+00 -4.38077867e-01 -1.36903808e-01
-9.30270851e-01 5.93154907e-01 6.37370348e-01 -4.33661789e-01
8.76880407e-01 -1.29464000e-01 -1.58559769e-01 -1.99049339e-01
-8.92263234e-01 -2.80680180e-01 6.31469965e-01 2.12621018e-01
6.24681652e-01 8.24447945e-02 -1.35553747e-01 1.34079468e+00
1.70506209e-01 -7.06781223e-02 -1.00674786e-01 8.74048412e-01
-1.81427136e-01 -1.25721419e+00 4.17683214e-01 8.33981335e-01
-3.95537168e-01 -1.47097796e-01 -4.58941981e-02 4.73790914e-01
-7.24606887e-02 9.22659159e-01 3.36475164e-01 -8.83758068e-01
6.87637568e-01 -1.46805927e-01 5.37588708e-02 -7.82283783e-01
-7.90287197e-01 9.92961973e-02 -1.93365663e-02 -8.83834898e-01
-2.65093949e-02 -6.23467684e-01 -9.14210379e-01 -1.52113572e-01
-3.18041742e-01 3.33771616e-01 4.54395920e-01 8.85688007e-01
6.70159459e-01 5.18475115e-01 1.12949371e+00 -7.14730740e-01
-1.06824410e+00 -6.79451346e-01 -9.29084241e-01 1.09265816e+00
4.00696993e-01 -6.80379868e-01 -5.71845829e-01 -3.95327449e-01] | [10.169536590576172, 5.623807430267334] |
712ffdad-8d91-4460-aca5-fe25d451d88f | towards-object-re-identification-from-point | 2305.10210 | null | https://arxiv.org/abs/2305.10210v2 | https://arxiv.org/pdf/2305.10210v2.pdf | Object Re-Identification from Point Clouds | In this work, we study the problem of object re-identification (ReID) in a 3D multi-object tracking (MOT) context, by learning to match pairs of objects from cropped (e.g., using their predicted 3D bounding boxes) point cloud observations. We are not concerned with SOTA performance for 3D MOT, however. Instead, we seek to answer the following question: In a realistic tracking by-detection context, how does object ReID from point clouds perform relative to ReID from images? To enable such a study, we propose a lightweight matching head that can be concatenated to any set or sequence processing backbone (e.g., PointNet or ViT), creating a family of comparable object ReID networks for both modalities. Run in siamese style, our proposed point-cloud ReID networks can make thousands of pairwise comparisons in real-time (10 hz). Our findings demonstrate that their performance increases with higher sensor resolution and approaches that of image ReID when observations are sufficiently dense. Additionally, we investigate our network's ability to enhance 3D multi-object tracking (MOT), showing that our point-cloud ReID networks can successfully re-identify objects which led a strong motion-based tracker into error. To our knowledge, we are the first to study real-time object re-identification from point clouds in a 3D multi-object tracking context. | ['Krzysztof Czarnecki', 'Adrian Chow', 'Chengjie Huang', 'Benjamin Thérien'] | 2023-05-17 | null | null | null | null | ['multi-object-tracking', '3d-multi-object-tracking'] | ['computer-vision', 'computer-vision'] | [ 1.35774747e-01 -3.85831296e-01 1.36094302e-01 -3.55110168e-02
-7.31405973e-01 -9.98028100e-01 3.71726662e-01 1.51410967e-01
-6.25093639e-01 2.48268589e-01 -5.33348322e-01 -2.00908512e-01
-6.99638948e-02 -4.68900263e-01 -1.30275440e+00 -4.12101060e-01
-2.83390999e-01 9.28031087e-01 6.81847930e-01 4.36602309e-02
1.76870272e-01 1.18310177e+00 -1.78943193e+00 -3.53353053e-01
2.15298146e-01 7.49728680e-01 3.01755637e-01 9.93237913e-01
-2.96465441e-04 1.04024373e-01 -6.62425220e-01 -3.04478168e-01
8.60174716e-01 2.65832901e-01 -5.32471120e-01 -1.20091148e-01
1.50104129e+00 -4.68694299e-01 -1.34454727e-01 1.03395319e+00
5.36649466e-01 -1.58098601e-02 4.47367102e-01 -1.82802963e+00
-3.82104069e-01 3.05630982e-01 -7.66506016e-01 4.42708045e-01
2.85934925e-01 3.65577549e-01 5.55881262e-01 -7.59949505e-01
7.69820631e-01 1.54574072e+00 1.25067532e+00 5.15273392e-01
-1.34455431e+00 -1.22789490e+00 9.72231328e-02 -1.01554073e-01
-1.55796385e+00 -3.37648153e-01 3.20401520e-01 -4.10903573e-01
8.33297372e-01 2.65709549e-01 6.15516663e-01 8.77392709e-01
8.07319060e-02 4.51006055e-01 6.07138753e-01 -4.27360348e-02
-6.78027496e-02 -4.96643819e-02 -7.65308216e-02 3.92580420e-01
7.71767497e-01 3.15490454e-01 -5.39293170e-01 -4.09891874e-01
8.68214786e-01 1.54894635e-01 7.39315078e-02 -9.35224771e-01
-1.75835717e+00 3.74731541e-01 6.03048384e-01 1.27037555e-01
-2.12172493e-01 7.70012558e-01 2.31691301e-01 3.55010271e-01
2.77951181e-01 2.74601221e-01 -3.37933451e-01 4.74393571e-04
-9.44950640e-01 5.92046022e-01 6.52514935e-01 1.67443347e+00
6.17638052e-01 -1.23212129e-01 2.92670071e-01 1.21689335e-01
4.13507700e-01 1.16966486e+00 -2.58197218e-01 -1.09478450e+00
3.77920687e-01 3.14376354e-01 3.44290316e-01 -1.06430721e+00
-4.39719260e-01 -1.71488494e-01 -6.52376115e-01 5.77549934e-01
5.76549709e-01 -4.36662249e-02 -7.07029879e-01 1.70203376e+00
9.02394712e-01 6.74240828e-01 -6.59291148e-02 9.94651198e-01
7.79835939e-01 1.21443398e-01 3.50326113e-02 2.02194646e-01
1.47624457e+00 -4.10333097e-01 -1.61687642e-01 1.51846316e-02
6.18533432e-01 -7.45870113e-01 1.61666438e-01 -5.56514040e-03
-1.15281963e+00 -8.83966625e-01 -9.70257401e-01 2.41095543e-01
-3.55145007e-01 -3.03275704e-01 2.10432008e-01 6.13794863e-01
-1.39281416e+00 6.97641730e-01 -1.05176866e+00 -1.04267359e+00
4.15989876e-01 8.57383609e-01 -4.14749384e-01 6.96323812e-02
-5.04047930e-01 1.07353151e+00 3.92668217e-01 -1.96164593e-01
-7.01256573e-01 -9.45514381e-01 -5.34600735e-01 -2.64457166e-01
2.82521248e-01 -1.14959764e+00 1.14635623e+00 -5.25318205e-01
-7.53644049e-01 9.08191562e-01 -2.73489356e-01 -5.67820907e-01
6.14223301e-01 -2.28565291e-01 -2.97355115e-01 1.46605611e-01
2.41193235e-01 1.20859051e+00 7.49772131e-01 -1.55070019e+00
-9.82262552e-01 -6.25712693e-01 -1.07862122e-01 1.59752548e-01
7.99440965e-02 2.41895869e-01 -4.66396898e-01 -3.30111146e-01
2.86076427e-01 -1.52010572e+00 -8.98469519e-03 7.59807289e-01
-1.08606301e-01 -3.35273087e-01 1.25482845e+00 -5.97648211e-02
2.40401134e-01 -2.18847179e+00 -2.27092281e-01 1.40542954e-01
5.16251683e-01 2.38836482e-01 -3.28669459e-01 3.45927924e-01
2.85829119e-02 1.53745532e-01 1.54658228e-01 -5.47596574e-01
4.24026065e-02 2.83416539e-01 -1.25688165e-01 9.87632453e-01
1.69479072e-01 9.32759345e-01 -1.02651334e+00 -7.20501721e-01
2.54870087e-01 3.35104346e-01 -3.03411365e-01 -1.37278110e-01
4.93562547e-03 4.70663428e-01 -4.34883773e-01 8.57145607e-01
1.18899453e+00 -4.39613134e-01 -1.20891392e-01 -4.72988248e-01
-4.00687277e-01 -3.50619704e-01 -1.49769735e+00 1.66824520e+00
1.37585863e-01 6.70954049e-01 2.79488295e-01 -4.76317167e-01
7.68169641e-01 9.62461159e-02 8.81024480e-01 -2.11531371e-01
7.20880330e-02 1.55639976e-01 -1.08222991e-01 1.96145792e-02
9.93589878e-01 2.03111723e-01 8.73478949e-02 4.12377357e-01
-1.34775311e-01 -2.13087827e-01 3.63431834e-02 2.58889347e-01
1.24412501e+00 2.13557079e-01 -4.54388000e-02 -6.27895668e-02
6.26973733e-02 5.98600209e-01 3.64406437e-01 1.20803106e+00
-5.10933578e-01 5.63018143e-01 -4.27205175e-01 -2.43784189e-01
-1.40740550e+00 -1.30051911e+00 -3.91060382e-01 8.25426400e-01
8.35707128e-01 -5.83051294e-02 -2.01457500e-01 -4.80064631e-01
8.84636998e-01 1.19452588e-01 -1.99129447e-01 2.08193123e-01
-8.14329982e-01 -1.19984798e-01 8.82836282e-01 6.07865930e-01
2.55476832e-01 -5.00066936e-01 -8.20345104e-01 2.38653138e-01
1.46151379e-01 -1.31065822e+00 -5.62893152e-01 -1.79562829e-02
-1.08038116e+00 -1.21415830e+00 -7.36507714e-01 -7.03490317e-01
5.09173334e-01 1.03003311e+00 1.11721718e+00 3.09775293e-01
-5.85003160e-02 9.99567747e-01 -3.67659062e-01 -7.11754322e-01
-3.80215585e-01 -9.01055783e-02 7.86839008e-01 -5.38636863e-01
6.23195767e-01 -6.47125065e-01 -5.33320069e-01 7.94765830e-01
-6.34082735e-01 -3.33031625e-01 5.49977064e-01 1.98889539e-01
6.52733922e-01 -3.68014693e-01 1.50479510e-01 -1.32768229e-01
-2.50454489e-02 -4.21157271e-01 -1.05924034e+00 1.30272314e-01
-2.78927505e-01 -3.03623497e-01 -2.82152463e-02 -8.83216262e-01
-3.18398297e-01 5.13767779e-01 3.34696800e-01 -1.07542539e+00
-2.71178067e-01 -2.93414205e-01 3.64478856e-01 -1.05542922e+00
5.52729368e-01 1.84390019e-03 1.72553316e-01 -3.86339694e-01
5.78126132e-01 4.77033406e-01 9.50142980e-01 -5.74154735e-01
1.54192293e+00 8.04715931e-01 3.91388535e-01 -5.79054773e-01
-2.16370851e-01 -9.64924455e-01 -9.44221258e-01 -5.08237600e-01
9.46025074e-01 -1.22638583e+00 -1.48457527e+00 2.19057322e-01
-1.50458610e+00 1.56221211e-01 -2.73584217e-01 6.68795884e-01
-5.86709321e-01 4.65219796e-01 -3.27673167e-01 -7.77966678e-01
-2.03488693e-01 -7.98087358e-01 1.43056405e+00 1.69266775e-01
-1.13374226e-01 -5.36087096e-01 1.42139211e-01 -3.64470780e-02
2.55616814e-01 4.33824539e-01 3.09484769e-02 -8.08130324e-01
-1.20285666e+00 -1.00795582e-01 -5.45709848e-01 -2.53267825e-01
-7.22285062e-02 7.00947922e-03 -8.10904860e-01 -7.38435328e-01
-3.24959397e-01 2.36608777e-02 3.14836144e-01 2.75602072e-01
6.02580428e-01 7.51286149e-02 -9.51317489e-01 5.82653761e-01
1.37249386e+00 -1.48244109e-02 4.71581221e-02 3.71088803e-01
8.33617270e-01 2.31916919e-01 7.63712287e-01 2.22428292e-01
4.91648436e-01 8.34495544e-01 7.94770241e-01 -2.87193526e-02
-2.62625009e-01 -1.39067948e-01 4.51428443e-02 4.76384610e-01
-6.25960156e-02 -2.99961835e-01 -1.06102931e+00 6.63179755e-01
-1.84755552e+00 -9.99667883e-01 -5.57650805e-01 2.15570259e+00
1.68425322e-01 1.59852542e-02 4.30189580e-01 -5.26975036e-01
1.16875386e+00 -1.47205770e-01 -8.54924202e-01 4.78278875e-01
-1.17896028e-01 -2.19740704e-01 1.30032647e+00 -9.41839963e-02
-1.13777637e+00 6.42602980e-01 6.31630135e+00 3.65643144e-01
-6.00648463e-01 2.37766251e-01 -3.66422623e-01 -3.21225941e-01
2.69138962e-01 5.19545563e-02 -1.25281477e+00 3.22627783e-01
9.40489650e-01 -1.69859082e-01 7.94260949e-02 7.61852562e-01
-1.07919060e-01 3.96644324e-02 -1.48230541e+00 1.23957932e+00
5.31051541e-03 -1.37038267e+00 -1.08081773e-01 3.78269285e-01
4.17602360e-01 6.67681456e-01 -1.71523824e-01 1.06561361e-02
7.49637783e-01 -4.90034223e-01 9.82391357e-01 3.75329196e-01
5.96230090e-01 -4.22060490e-01 5.30969918e-01 4.46358442e-01
-1.69947314e+00 2.28235945e-01 -5.55721939e-01 2.98866928e-01
4.68038648e-01 1.25051454e-01 -1.08006191e+00 8.14359069e-01
1.16574442e+00 7.60905981e-01 -5.60516357e-01 1.60770118e+00
7.49319136e-01 4.72190715e-02 -9.99715149e-01 -8.23744237e-02
1.15984380e-01 2.71682680e-01 1.12617981e+00 1.08040881e+00
5.94693005e-01 1.63772389e-01 3.78062010e-01 9.08934236e-01
-1.48576081e-01 -4.75276023e-01 -8.32986712e-01 3.35845739e-01
8.67448032e-01 1.13254714e+00 -7.66660213e-01 -4.84504521e-01
-2.29929864e-01 5.99050581e-01 5.87216988e-02 -7.20212683e-02
-8.19723725e-01 6.85128272e-02 1.10099304e+00 7.45837092e-02
6.89202249e-01 -6.42881215e-01 1.99655183e-02 -9.00458097e-01
-2.88339090e-02 -4.69067514e-01 1.88537762e-01 -1.07984400e+00
-1.57962584e+00 2.01343954e-01 3.48092079e-01 -1.74970567e+00
-6.42743288e-03 -4.58411157e-01 -2.94390917e-01 7.75962472e-01
-1.42813885e+00 -1.21820593e+00 -3.99971128e-01 6.23907983e-01
3.55686732e-02 2.84892112e-01 1.89437538e-01 5.50915122e-01
2.95019317e-02 4.39712554e-01 2.66736329e-01 1.07399158e-01
8.43384147e-01 -9.96046066e-01 9.68210876e-01 8.40694726e-01
2.48279259e-01 5.91352284e-01 6.99308217e-01 -1.06207693e+00
-1.95378232e+00 -1.21562433e+00 4.37065214e-01 -1.17677498e+00
6.54871762e-01 -4.27529454e-01 -8.39823544e-01 8.48991036e-01
-1.84459120e-01 2.77615249e-01 2.08699360e-01 -1.37951881e-01
-2.45881572e-01 5.17830849e-02 -1.28279352e+00 2.46187851e-01
1.65814269e+00 -3.31365258e-01 -5.15371621e-01 5.21602809e-01
9.74363267e-01 -8.48346233e-01 -1.17578530e+00 4.13953394e-01
6.30064964e-01 -4.94990379e-01 1.51060450e+00 -4.44109112e-01
-4.52902645e-01 -9.29962695e-01 -1.96661830e-01 -8.05245757e-01
-3.28482091e-01 -3.36041421e-01 -3.82609181e-02 1.25743699e+00
-2.09122166e-01 -5.34519851e-01 1.00379121e+00 4.06500340e-01
-1.35761797e-01 2.11828575e-01 -1.25759101e+00 -1.36883116e+00
-1.50507584e-01 -4.42419827e-01 8.52105558e-01 8.75989258e-01
-6.42711937e-01 -3.45773458e-01 -1.09856397e-01 9.59359050e-01
1.19632626e+00 1.69438943e-01 1.47885704e+00 -1.54885006e+00
6.43411726e-02 -2.51734376e-01 -9.15368438e-01 -1.28267944e+00
-6.22407757e-02 -9.41384554e-01 1.89704329e-01 -9.89145935e-01
1.49179101e-01 -9.22947109e-01 5.02751246e-02 3.49933833e-01
1.06580399e-01 4.34957862e-01 9.14234102e-01 7.83927500e-01
-8.59178126e-01 1.53720766e-01 9.53747749e-01 -2.46898174e-01
1.52221650e-01 -1.05717666e-02 -1.81959584e-01 3.22776735e-01
4.12496477e-01 -9.56463814e-01 2.38434061e-01 -5.95510483e-01
-1.24894522e-01 3.21098045e-02 1.15428579e+00 -1.45667613e+00
8.46311927e-01 1.39287338e-02 4.61497039e-01 -1.31503999e+00
5.00044882e-01 -1.32775557e+00 7.46336102e-01 4.88769203e-01
1.50404006e-01 6.36404991e-01 5.67561388e-01 7.66626716e-01
2.80945659e-01 1.04949316e-02 4.42680776e-01 -1.61796540e-01
-8.00651014e-01 5.02530336e-01 -7.79735344e-03 -3.26177925e-01
1.14721143e+00 -7.84052610e-01 -4.32977259e-01 9.48749483e-02
-4.07914877e-01 5.48955619e-01 1.01771998e+00 5.21677017e-01
3.77992719e-01 -1.50659525e+00 -7.77331114e-01 -1.48270294e-01
2.39754260e-01 2.58459002e-01 8.50261524e-02 8.92032981e-01
-2.91735947e-01 3.07886362e-01 -1.56116337e-01 -1.48679507e+00
-1.68263805e+00 6.49278343e-01 3.10101300e-01 3.05611402e-01
-8.74886453e-01 5.86496055e-01 -7.26615712e-02 -6.29889548e-01
2.81956196e-01 -3.72188151e-01 4.07991886e-01 -2.54092127e-01
3.53523672e-01 3.97220939e-01 -1.72803309e-02 -8.10234666e-01
-7.32773602e-01 1.04213214e+00 5.52054234e-02 8.49992335e-02
1.14668965e+00 -2.32359260e-01 1.95869505e-01 2.07359180e-01
9.65947807e-01 -3.17269176e-01 -1.41013110e+00 -3.08071226e-01
-3.24717537e-02 -6.79746568e-01 -3.30612391e-01 -2.44638920e-01
-7.72440851e-01 2.73700684e-01 1.25010812e+00 2.23624796e-01
4.71913517e-01 4.22263116e-01 6.90744460e-01 7.89098024e-01
8.20860505e-01 -5.14505386e-01 -2.25937396e-01 3.65308940e-01
4.37906235e-01 -1.25607467e+00 1.26622349e-01 -2.29411334e-01
4.19167522e-03 9.34656441e-01 6.13102257e-01 -2.86792427e-01
3.36811930e-01 4.62381005e-01 -3.67271788e-02 -4.41813499e-01
-3.11440289e-01 -3.44592750e-01 1.25971586e-01 9.68178868e-01
-2.90987432e-01 -3.26232821e-01 5.47485948e-01 -5.46375453e-01
-1.63751453e-01 7.44784474e-02 2.69467324e-01 1.26569641e+00
-5.00191748e-01 -8.03287268e-01 -1.09161901e+00 3.76227438e-01
8.82615075e-02 2.45131582e-01 -2.62095213e-01 1.18008125e+00
8.97968858e-02 8.73386264e-01 5.55881679e-01 -3.85681450e-01
5.02606809e-01 -2.89873928e-01 6.38857126e-01 -3.53075415e-01
-8.19486380e-01 -3.24807882e-01 -1.46552578e-01 -6.73474967e-01
-1.07245505e+00 -9.98133957e-01 -1.10138190e+00 -8.59898448e-01
-6.27536297e-01 -1.94337398e-01 1.05326128e+00 7.05113947e-01
6.16521239e-01 1.98988914e-01 1.82087228e-01 -1.50976408e+00
-5.11347175e-01 -6.19552314e-01 -3.72109890e-01 3.99041891e-01
6.00132406e-01 -7.96558022e-01 -2.69837171e-01 -1.15685090e-01] | [6.596281051635742, -2.2074360847473145] |
ccd3e87b-3146-41d1-9c47-29d4d928fc61 | gene-sgan-a-method-for-discovering-disease | 2301.10772 | null | https://arxiv.org/abs/2301.10772v1 | https://arxiv.org/pdf/2301.10772v1.pdf | Gene-SGAN: a method for discovering disease subtypes with imaging and genetic signatures via multi-view weakly-supervised deep clustering | Disease heterogeneity has been a critical challenge for precision diagnosis and treatment, especially in neurologic and neuropsychiatric diseases. Many diseases can display multiple distinct brain phenotypes across individuals, potentially reflecting disease subtypes that can be captured using MRI and machine learning methods. However, biological interpretability and treatment relevance are limited if the derived subtypes are not associated with genetic drivers or susceptibility factors. Herein, we describe Gene-SGAN - a multi-view, weakly-supervised deep clustering method - which dissects disease heterogeneity by jointly considering phenotypic and genetic data, thereby conferring genetic correlations to the disease subtypes and associated endophenotypic signatures. We first validate the generalizability, interpretability, and robustness of Gene-SGAN in semi-synthetic experiments. We then demonstrate its application to real multi-site datasets from 28,858 individuals, deriving subtypes of Alzheimer's disease and brain endophenotypes associated with hypertension, from MRI and SNP data. Derived brain phenotypes displayed significant differences in neuroanatomical patterns, genetic determinants, biological and clinical biomarkers, indicating potentially distinct underlying neuropathologic processes, genetic drivers, and susceptibility factors. Overall, Gene-SGAN is broadly applicable to disease subtyping and endophenotype discovery, and is herein tested on disease-related, genetically-driven neuroimaging phenotypes. | ['Christos Davatzikos', 'Ilya M. Nasrallah', 'Haochang Shou', 'Susan M. Resnick', 'Jingxuan Bao', 'Li Shen', 'David A. Wolk', 'Tammie L. S. Benzinger', 'Daniel S. Marcus', 'Pamela Lamontagne', 'John C. Morris', 'Sterling C. Johnson', 'Marilyn S. Albert', 'Luigi Ferrucci', 'Jurgen Fripp', 'Paul Maruff', 'Colin L. Masters', 'Mohamad Habes', 'Jiong Chen', 'Sindhuja T. Govindarajan', 'Dhivya Srinivasan', 'Randa Melhem', 'Elizabeth Mamourian', 'Guray Erus', 'Yuhan Cui', 'Ahmed Abdulkadir', 'Junhao Wen', 'Zhijian Yang'] | 2023-01-25 | null | null | null | null | ['deep-clustering', 'deep-clustering'] | ['miscellaneous', 'natural-language-processing'] | [ 2.29820564e-01 -2.64422864e-01 -5.92286438e-02 -7.78131366e-01
-5.61383307e-01 -4.88300711e-01 4.27009255e-01 -8.97318646e-02
-1.62229128e-02 7.68611312e-01 4.12527382e-01 -9.58605483e-02
-8.10525954e-01 -2.68459231e-01 -2.30054513e-01 -7.86512196e-01
-7.32760787e-01 8.83160055e-01 -4.76580024e-01 3.19967479e-01
-1.34973407e-01 1.15152173e-01 -1.06699383e+00 1.56436488e-01
1.21550119e+00 5.98323226e-01 8.24607834e-02 2.87684292e-01
3.43127161e-01 -1.90076664e-01 -4.03559715e-01 -2.09201664e-01
4.17090580e-02 -5.55819035e-01 -3.49399954e-01 -8.72108992e-03
3.29779059e-01 -3.41014825e-02 -6.00121804e-02 1.25242162e+00
9.17474866e-01 -6.49178028e-01 9.91087198e-01 -1.02706444e+00
-1.01327217e+00 5.85511744e-01 -7.54174948e-01 2.37328753e-01
-2.25665510e-01 3.93734515e-01 8.97094071e-01 -4.62311983e-01
9.01261985e-01 1.30711377e+00 6.65563703e-01 7.09678411e-01
-1.89570189e+00 -3.77022266e-01 4.48536798e-02 3.03245217e-01
-1.32738161e+00 -3.16354781e-01 2.59472519e-01 -1.14381588e+00
4.62776572e-01 2.78680354e-01 6.85745597e-01 1.27458012e+00
3.45333904e-01 1.83212548e-01 1.38299465e+00 2.65004009e-01
3.59198719e-01 -6.20400012e-01 3.94979894e-01 5.85760593e-01
4.58122075e-01 5.56008108e-02 -3.12456060e-02 -9.23598349e-01
4.24333036e-01 1.07106708e-01 -3.27514619e-01 -1.48232788e-01
-1.52166283e+00 6.42152309e-01 -8.71546343e-02 4.95772716e-03
-6.19110107e-01 -1.81866124e-01 5.02432883e-01 8.78587663e-02
3.79877418e-01 2.10169688e-01 -6.04300618e-01 1.51760593e-01
-9.13727343e-01 5.38773164e-02 1.03045084e-01 5.52925110e-01
3.26747060e-01 5.34369275e-02 -2.05209643e-01 1.34538722e+00
3.96923661e-01 6.57031953e-01 9.99869168e-01 -9.76095080e-01
-1.87560637e-02 7.00376213e-01 -4.58249211e-01 -8.28330815e-01
-1.15640867e+00 -5.36278725e-01 -1.24162340e+00 7.04570711e-02
8.95974711e-02 -4.90618169e-01 -7.76852131e-01 2.31616545e+00
4.69206959e-01 6.15003891e-02 -3.44928771e-01 1.06327748e+00
5.51633120e-01 -1.95274621e-01 2.83922732e-01 -9.64583009e-02
1.78947103e+00 -2.93808103e-01 -8.28281045e-02 -1.95072994e-01
5.63558877e-01 8.44858289e-02 6.86525941e-01 2.48021290e-01
-8.17470729e-01 -1.84310749e-01 -4.81248587e-01 2.81285048e-01
4.01711166e-02 3.00857723e-01 6.89219117e-01 6.69251740e-01
-1.16347182e+00 2.69074202e-01 -9.46586668e-01 -3.34750801e-01
7.54962683e-01 3.86426836e-01 -5.82388222e-01 -5.62487096e-02
-1.04020655e+00 6.30147338e-01 5.01854904e-02 3.05305839e-01
-1.05791473e+00 -1.17948472e+00 -2.32078671e-01 -7.72684664e-02
-2.44043246e-02 -1.20572436e+00 3.75608981e-01 -2.98529088e-01
-1.00016534e+00 8.83638620e-01 -2.02320978e-01 -1.17493771e-01
2.36288123e-02 1.58947453e-01 -8.90692353e-01 2.08724946e-01
3.88006330e-01 5.20511150e-01 4.11773384e-01 -6.65307105e-01
-1.85487896e-01 -1.33994472e+00 -7.54319489e-01 -1.25593901e-01
2.25142211e-01 1.85054049e-01 3.88122976e-01 -7.74454296e-01
1.55511528e-01 -9.95975554e-01 -3.86889070e-01 3.09625030e-01
-7.47180879e-01 2.30257690e-01 3.30335498e-01 -8.73255074e-01
6.90769017e-01 -2.00449467e+00 4.48936313e-01 2.36015767e-01
8.54696214e-01 -1.42886757e-03 -4.31855381e-01 1.35610485e-02
-4.45226461e-01 4.67722476e-01 -4.57670510e-01 3.83668393e-01
1.48251593e-01 -3.23586434e-01 2.94742554e-01 9.04824793e-01
5.56652486e-01 1.22458637e+00 -5.63791335e-01 2.22678870e-01
-2.60943115e-01 5.35010576e-01 -5.83926678e-01 -2.62651443e-01
-2.50628114e-01 6.25397086e-01 -7.07337737e-01 6.36027336e-01
6.45440817e-01 -3.66002470e-01 7.92507172e-01 -6.51057214e-02
3.47913861e-01 1.58787146e-01 -5.13271093e-01 1.38433957e+00
1.26393124e-01 3.92766714e-01 5.04519463e-01 -1.24277246e+00
9.14482176e-01 1.07642107e-01 4.90889043e-01 -5.13495862e-01
3.89007553e-02 8.63372087e-02 9.24185932e-01 -7.84011424e-01
-6.69342458e-01 -8.40308368e-02 1.15784757e-01 6.86179042e-01
-1.64604694e-01 7.71034837e-01 4.86104116e-02 -1.06259391e-01
1.47520185e+00 -2.66501337e-01 2.85992861e-01 -5.32807887e-01
2.69633263e-01 -1.53536079e-02 8.60856116e-01 3.79611075e-01
-4.08632904e-01 7.16234267e-01 8.78950596e-01 -8.69135559e-02
-1.07805419e+00 -1.28123987e+00 -7.25520015e-01 7.73478448e-01
-5.36192060e-01 -1.61467567e-01 -5.21829784e-01 -4.59721595e-01
1.73285663e-01 1.65168226e-01 -7.10776269e-01 -4.95414108e-01
1.07761636e-01 -1.89383137e+00 1.22443843e+00 3.99008304e-01
1.11673288e-01 -2.55290031e-01 -2.91552506e-02 8.63005817e-02
1.43919706e-01 -7.13599682e-01 -2.49634027e-01 3.36316414e-02
-9.10046160e-01 -1.43643379e+00 -8.41028392e-01 -3.59743685e-01
6.44266307e-01 -1.58410624e-01 8.10787916e-01 -1.39182195e-01
-9.14471805e-01 1.22683354e-01 5.61020747e-02 1.07714720e-01
-4.04037476e-01 -3.73213768e-01 4.43999827e-01 1.07949004e-01
4.87023443e-01 -1.02068973e+00 -9.76726592e-01 2.62526006e-01
-5.61978459e-01 -1.65509999e-01 1.03791714e+00 7.83566475e-01
6.33966208e-01 -1.38128176e-01 9.96697426e-01 -7.80118942e-01
5.20733833e-01 -1.13686824e+00 -4.30091411e-01 3.13300163e-01
-7.26556361e-01 2.38295831e-02 5.17506540e-01 -2.96291411e-01
-9.72404182e-01 -1.66114658e-01 7.77308345e-02 2.29608908e-01
-7.13391840e-01 7.10746884e-01 -7.04537928e-01 3.89593095e-01
5.19942522e-01 3.18023801e-01 5.14366746e-01 -6.53575122e-01
3.35922241e-01 6.01789415e-01 3.18209618e-01 -6.45429909e-01
1.29509130e-02 5.97764075e-01 3.03366840e-01 -8.29943001e-01
-1.93201497e-01 -1.46439999e-01 -8.16894650e-01 3.30125749e-01
1.13088346e+00 -1.07588100e+00 -6.29887760e-01 5.97172499e-01
-6.95569932e-01 -4.88609970e-01 5.23887098e-01 6.80698812e-01
-1.46825090e-01 2.97110885e-01 -5.59050083e-01 -1.51667178e-01
-3.55419576e-01 -1.19931304e+00 1.10380745e+00 -3.03243369e-01
-5.48354089e-01 -1.16054821e+00 5.41423380e-01 5.07013679e-01
2.01568797e-01 3.63968670e-01 1.80231559e+00 -9.97098446e-01
4.73008677e-02 2.23099172e-01 -4.39651012e-01 6.90057948e-02
3.51140231e-01 -7.54195750e-02 -5.62119544e-01 9.22328234e-02
-3.37750256e-01 -9.92147326e-02 6.55324399e-01 6.67352915e-01
9.53078508e-01 -2.11194947e-01 -3.52084070e-01 5.50116658e-01
1.01938808e+00 1.73253685e-01 4.04364437e-01 -1.21740922e-02
7.41138279e-01 8.25421691e-01 -2.77032822e-01 3.44892502e-01
3.88011307e-01 8.43932688e-01 -4.65082563e-02 7.17003047e-02
-1.02192864e-01 3.66994888e-01 4.62113589e-01 5.51667213e-01
-1.02157094e-01 3.22044604e-02 -1.04628289e+00 3.42627138e-01
-1.51456416e+00 -1.02666509e+00 -7.90190101e-01 2.01623964e+00
8.93326104e-01 -6.13285661e-01 5.38030982e-01 -7.20862746e-01
1.15957630e+00 -1.53794378e-01 -1.09084976e+00 4.80834022e-02
-6.75721824e-01 3.25949048e-03 3.09660863e-02 -1.68503881e-01
-4.72384572e-01 2.31616631e-01 7.24668455e+00 1.00626580e-01
-1.12924159e+00 3.01145911e-01 1.06387126e+00 -5.44854641e-01
-4.08658236e-01 -4.02422309e-01 -4.14055437e-01 6.96544707e-01
1.07955420e+00 -8.98020566e-02 4.96305227e-01 5.62301099e-01
6.66039228e-01 2.57792890e-01 -1.15767133e+00 5.62545359e-01
-1.38604462e-01 -1.20992243e+00 -7.81631842e-02 4.42473054e-01
6.54141426e-01 4.64645684e-01 1.55960798e-01 -2.63828844e-01
3.99098068e-01 -1.05108559e+00 3.31922144e-01 7.33682752e-01
1.05086982e+00 -3.52985442e-01 3.57187688e-01 -2.07059979e-01
-6.13820493e-01 4.00461070e-02 -5.02848089e-01 1.96657255e-01
2.25729167e-01 1.32003605e+00 -8.01597834e-01 3.92101645e-01
4.30717766e-01 4.86105591e-01 -6.52239442e-01 8.27819407e-01
7.03267902e-02 6.39005363e-01 6.00270107e-02 4.10486728e-01
-2.83065140e-01 -6.23513401e-01 5.53365648e-01 9.63427186e-01
3.53771448e-01 1.68274388e-01 -9.53703225e-02 1.72088432e+00
1.97417587e-01 -1.79654211e-01 -1.02046616e-01 -4.34112400e-01
5.71287274e-01 1.49691081e+00 -7.55742908e-01 -2.89511353e-01
-3.87330979e-01 7.22747087e-01 3.84211391e-01 4.73835140e-01
-6.20485902e-01 -3.95902954e-02 1.25062084e+00 -9.72295031e-02
3.86593565e-02 -1.74608409e-01 -3.98398340e-01 -1.17541838e+00
-2.46000662e-03 -1.01316655e+00 2.51857817e-01 -4.93901521e-01
-1.85540891e+00 3.62160951e-01 -7.35855252e-02 -6.45288348e-01
-3.03892583e-01 -6.33122146e-01 -7.79494762e-01 1.05883884e+00
-8.86953890e-01 -8.48565698e-01 -1.11678399e-01 2.49959216e-01
-2.30993256e-01 -4.10203904e-01 8.48088682e-01 2.93111175e-01
-1.18111885e+00 1.67520911e-01 6.03180766e-01 -1.32135928e-01
6.56671643e-01 -1.09867930e+00 2.30644062e-01 3.79572988e-01
-4.03101295e-01 8.64578664e-01 2.32769758e-01 -1.06601512e+00
-1.31408286e+00 -1.56200647e+00 2.64600158e-01 -2.25177810e-01
1.03883743e+00 -5.54327846e-01 -9.24750507e-01 6.47308350e-01
-4.77714330e-01 -8.70926753e-02 1.26359916e+00 3.76807004e-01
-6.00919127e-01 1.09875157e-01 -1.36554933e+00 6.60288751e-01
1.42164063e+00 -5.70361257e-01 -4.41253722e-01 4.50634122e-01
2.32927993e-01 5.52478731e-01 -1.35223186e+00 1.60342649e-01
6.71907604e-01 -8.73080373e-01 9.96671438e-01 -1.01617992e+00
4.64019924e-01 -3.25569123e-01 -1.14484593e-01 -1.68318689e+00
-8.85190845e-01 -1.33422270e-01 2.19793499e-01 1.22080600e+00
5.73848784e-01 -7.62124717e-01 4.64655071e-01 9.49509799e-01
-9.05739740e-02 -6.34303331e-01 -8.70580673e-01 -8.00959706e-01
2.99772859e-01 -3.17056388e-01 7.70691156e-01 9.84786987e-01
8.10567662e-02 2.73235589e-01 1.07164778e-01 3.05318952e-01
7.68093824e-01 1.62085801e-01 2.50600010e-01 -1.47961485e+00
-4.30557191e-01 -8.50504577e-01 -7.09793746e-01 -1.14061460e-01
3.48205835e-01 -1.25864851e+00 -2.88840085e-01 -1.30055988e+00
8.32237124e-01 -5.51909208e-01 -7.42226094e-02 4.35895264e-01
-2.86000162e-01 4.74447682e-02 -4.77912813e-01 5.71050262e-03
-1.65194735e-01 5.51069438e-01 9.34348047e-01 -2.96168804e-01
-6.08466640e-02 -4.27876741e-01 -1.05929387e+00 5.70114195e-01
9.54948962e-01 -2.01784760e-01 -5.83712220e-01 -3.87024820e-01
-1.44344270e-01 -1.69184223e-01 9.62002933e-01 -6.02517068e-01
-4.55076575e-01 -2.27750406e-01 6.96129024e-01 7.05259815e-02
6.12385161e-02 -3.43422472e-01 6.49324775e-01 6.77313626e-01
-4.69344318e-01 3.12046316e-02 -2.11810142e-01 7.61417270e-01
4.26468074e-01 2.12539092e-01 5.41601241e-01 3.35613936e-02
-3.90167594e-01 5.32763064e-01 -7.06300259e-01 3.02220225e-01
8.53691697e-01 3.97316590e-02 -8.22764814e-01 1.62108719e-01
-1.00821340e+00 -8.87057856e-02 2.67597079e-01 2.67641306e-01
4.23586547e-01 -1.28899097e+00 -1.11039305e+00 1.80115134e-01
4.15592231e-02 -5.54383993e-01 8.58803093e-01 1.48313582e+00
-7.05308421e-03 3.53694499e-01 -4.32320565e-01 -6.90156579e-01
-1.07999909e+00 5.01151323e-01 3.76264542e-01 3.30186725e-01
-7.42554605e-01 3.82792264e-01 7.72498429e-01 -4.07710284e-01
-2.68697888e-01 -1.19217694e-01 -4.24331725e-02 6.14477396e-02
6.70613110e-01 5.81998587e-01 -4.08659391e-02 -5.62963367e-01
-4.95486110e-01 3.67139131e-01 -8.71634707e-02 3.04124746e-02
1.62138605e+00 -2.52302110e-01 -6.23570681e-01 2.26769894e-01
1.18699837e+00 -1.93263099e-01 -7.84451902e-01 1.67455480e-01
2.28776500e-01 1.16041444e-01 -3.09956092e-02 -1.18235886e+00
-1.46097517e+00 6.79144025e-01 8.56197119e-01 -1.17043525e-01
8.15895736e-01 2.91656911e-01 4.31410730e-01 -2.72253901e-02
4.10550892e-01 -6.93741620e-01 -8.46852958e-01 2.63926089e-02
8.14170420e-01 -8.36246610e-01 -1.53526962e-01 -1.88155949e-01
-3.76107752e-01 9.98318791e-01 5.96842349e-01 6.19481839e-02
6.49503291e-01 1.20739922e-01 -1.05769739e-01 -7.27328598e-01
-9.38816130e-01 2.68141955e-01 4.00124580e-01 1.21504760e+00
5.27237296e-01 4.72352147e-01 -5.66413581e-01 1.43928385e+00
-7.11223260e-02 -1.35270447e-01 2.65054643e-01 3.03361535e-01
-1.70025617e-01 -1.10050225e+00 -3.98017496e-01 1.12536979e+00
-4.17023957e-01 -3.10132295e-01 -7.92290866e-01 5.13573885e-01
2.57095635e-01 5.50558746e-01 2.79035687e-01 -2.03701943e-01
-1.46894321e-01 2.18415782e-01 2.90336758e-01 -7.33157516e-01
-9.28297732e-03 1.62982091e-01 3.82217824e-01 -5.61287880e-01
-3.24888647e-01 -1.06195986e+00 -1.30438340e+00 -6.51939884e-02
1.79414079e-01 -4.27372545e-01 4.63827997e-01 1.01320899e+00
1.45520735e+00 6.27326250e-01 3.31271946e-01 -5.31300783e-01
-4.27520484e-01 -8.61107349e-01 -1.00724041e+00 2.20837086e-01
-6.45356402e-02 -8.10224175e-01 1.62351504e-02 1.39421225e-01] | [6.583165168762207, 5.504162788391113] |
28082418-cfcf-4b8f-8fea-142dc9357d8b | signal-decomposition-using-masked-proximal | 2202.09338 | null | https://arxiv.org/abs/2202.09338v6 | https://arxiv.org/pdf/2202.09338v6.pdf | Signal Decomposition Using Masked Proximal Operators | We consider the well-studied problem of decomposing a vector time series signal into components with different characteristics, such as smooth, periodic, nonnegative, or sparse. We describe a simple and general framework in which the components are defined by loss functions (which include constraints), and the signal decomposition is carried out by minimizing the sum of losses of the components (subject to the constraints). When each loss function is the negative log-likelihood of a density for the signal component, this framework coincides with maximum a posteriori probability (MAP) estimation; but it also includes many other interesting cases. Summarizing and clarifying prior results, we give two distributed optimization methods for computing the decomposition, which find the optimal decomposition when the component class loss functions are convex, and are good heuristics when they are not. Both methods require only the masked proximal operator of each of the component loss functions, a generalization of the well-known proximal operator that handles missing entries in its argument. Both methods are distributed, i.e., handle each component separately. We derive tractable methods for evaluating the masked proximal operators of some loss functions that, to our knowledge, have not appeared in the literature. | ['Stephen P. Boyd', 'Bennet E. Meyers'] | 2022-02-18 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [ 3.48201603e-01 1.15519442e-01 -7.69878253e-02 -4.44126837e-02
-1.14642930e+00 -7.64843583e-01 9.27049518e-02 4.81156074e-03
-1.68665081e-01 6.28283799e-01 6.74523488e-02 -7.43100569e-02
-4.01301473e-01 -3.42711896e-01 -7.88956523e-01 -1.17045903e+00
-5.79301834e-01 3.62709403e-01 -1.70577839e-01 -1.14896763e-02
-8.00334886e-02 4.10881966e-01 -1.21727788e+00 -1.19033389e-01
8.26975584e-01 1.37149930e+00 -6.79611638e-02 6.84438586e-01
2.57662922e-01 5.37537932e-01 -5.75312495e-01 -2.43962631e-01
4.84175533e-01 -3.85287970e-01 -5.12237489e-01 3.09678346e-01
8.47793836e-03 6.21168911e-02 -9.04622972e-02 1.44778872e+00
3.61266971e-01 6.37831911e-02 7.41617918e-01 -1.48709571e+00
-3.79242063e-01 4.90343124e-01 -8.26597035e-01 5.61940297e-02
3.51938725e-01 -4.87625510e-01 1.21241641e+00 -8.75531554e-01
2.26881117e-01 9.89101112e-01 9.63026702e-01 6.02214821e-02
-1.57205236e+00 -1.93022057e-01 2.69332916e-01 8.22021812e-02
-1.41473019e+00 -6.00481451e-01 8.91836345e-01 -6.82982087e-01
5.77667415e-01 4.99915123e-01 2.81094581e-01 7.37558484e-01
-1.88526101e-02 9.49623883e-01 6.96075857e-01 -4.55776483e-01
3.67467344e-01 5.89032993e-02 2.19195917e-01 5.82041860e-01
1.88040569e-01 -2.25093469e-01 -2.90850639e-01 -7.08772719e-01
3.54955345e-01 -4.94354181e-02 -7.66939104e-01 -4.48646843e-01
-1.24123979e+00 7.68451333e-01 -2.45647788e-01 2.87724376e-01
-6.63669944e-01 5.38757481e-02 3.09134930e-01 3.53070855e-01
7.27444351e-01 1.49773046e-01 -5.10813475e-01 3.86883803e-02
-1.07297194e+00 2.12773845e-01 1.11261475e+00 8.92571330e-01
5.87205708e-01 1.09704889e-01 -1.45864740e-01 8.26125085e-01
1.82893351e-01 5.61873555e-01 2.23446682e-01 -1.04903424e+00
4.93323803e-01 -5.20630479e-02 3.30747545e-01 -1.01918328e+00
-3.64486963e-01 -5.57188928e-01 -8.69535744e-01 6.75579719e-03
6.23598099e-01 -4.11620289e-01 -5.19947529e-01 2.10209417e+00
6.28897250e-02 3.65728736e-01 -1.73209757e-01 8.66786599e-01
1.38589472e-01 9.08843100e-01 -4.34979111e-01 -8.62777531e-01
1.05269527e+00 -6.18327498e-01 -9.59469676e-01 3.80928107e-02
1.04684979e-01 -8.90624881e-01 6.44449174e-01 7.64673352e-01
-1.51265442e+00 -8.63582566e-02 -1.15813088e+00 2.26364940e-01
9.35473889e-02 2.79130787e-01 5.02102017e-01 4.35038120e-01
-1.16090417e+00 6.79653883e-01 -9.53416526e-01 1.17370136e-01
-7.66013861e-02 1.53691053e-01 -1.64809018e-01 1.62014902e-01
-1.06137240e+00 4.59959954e-01 -1.56777471e-01 3.01348954e-01
-9.10315096e-01 -7.71939397e-01 -8.18297565e-01 2.11798295e-01
2.21629664e-01 -5.23822427e-01 1.19827032e+00 -1.08737767e+00
-1.39215577e+00 6.80751443e-01 -3.47427458e-01 -3.66551697e-01
5.12795389e-01 -4.24660416e-03 -2.68160462e-01 1.47983924e-01
2.76057154e-01 -4.68341596e-02 1.33610201e+00 -9.28003252e-01
-2.33648837e-01 -2.98000365e-01 -1.26321316e-01 6.68181777e-02
-2.07240224e-01 5.98563515e-02 -4.13733453e-01 -1.01482260e+00
4.08447832e-01 -7.64481604e-01 -3.53972882e-01 9.28506255e-02
-5.17379820e-01 -1.25034615e-01 4.51305985e-01 -1.05398798e+00
1.29065943e+00 -2.45418811e+00 4.87894297e-01 4.50562567e-01
1.30284559e-02 -2.86621720e-01 -1.47205532e-01 5.87591767e-01
-4.26136643e-01 -4.64451350e-02 -6.98485553e-01 -5.73546112e-01
2.69010395e-01 -1.41782332e-02 -5.87399423e-01 1.00670063e+00
-2.98395734e-02 3.35076898e-01 -6.58463001e-01 -1.27427354e-01
-1.70065269e-01 4.55869257e-01 -4.82154489e-01 7.10304677e-02
-4.83047403e-02 1.60328135e-01 -3.12390924e-01 4.35626537e-01
8.32711875e-01 -2.11084172e-01 7.72216618e-02 -3.48490685e-01
-7.57826939e-02 7.65264481e-02 -1.68846667e+00 1.37106717e+00
-3.73378783e-01 5.11934578e-01 9.12358880e-01 -1.48221791e+00
5.91565788e-01 5.62142730e-01 9.82579589e-01 4.63426188e-02
-1.02924742e-01 4.26119685e-01 -2.98224837e-01 -4.86019373e-01
1.04206257e-01 -5.03262758e-01 -8.01686943e-02 3.51979852e-01
4.63139936e-02 3.01149562e-02 3.38739216e-01 7.68141896e-02
9.69428241e-01 -1.51273206e-01 4.42199200e-01 -4.36004549e-01
4.74836707e-01 -3.36933583e-01 8.45346808e-01 5.93025744e-01
-2.45994553e-01 7.01325655e-01 8.51138115e-01 1.55526415e-01
-7.80152678e-01 -1.02605951e+00 -3.23295236e-01 9.16043997e-01
-7.68829361e-02 -3.46133441e-01 -7.08059371e-01 -2.44000241e-01
-2.83279835e-04 5.00140071e-01 -3.23159069e-01 3.85303795e-02
-4.28939790e-01 -1.06778264e+00 2.58070052e-01 3.73842210e-01
1.06377862e-01 -3.62611085e-01 -1.59729660e-01 2.67987609e-01
-4.68331814e-01 -8.55405748e-01 -8.84588242e-01 4.73010510e-01
-8.55996311e-01 -8.66039276e-01 -1.12795520e+00 -7.33845711e-01
6.33687913e-01 2.14043871e-01 1.06569278e+00 -3.87006074e-01
-1.19055681e-01 5.58555484e-01 -9.81188044e-02 -1.80455506e-01
1.55193642e-01 -4.91682231e-01 1.17863581e-01 6.02994859e-01
-2.37704366e-02 -7.40477860e-01 -1.96872175e-01 3.28081727e-01
-7.85412133e-01 -5.13574600e-01 2.58190125e-01 8.18095624e-01
9.61066723e-01 4.47008520e-01 6.03038132e-01 -5.50899208e-01
7.59059906e-01 -7.84342527e-01 -6.40155911e-01 2.44524986e-01
-2.75303692e-01 1.32029787e-01 7.31981993e-01 -4.83544528e-01
-7.66381979e-01 1.17097698e-01 5.93235763e-03 -5.77573359e-01
2.02890843e-01 8.22765350e-01 -3.02433521e-01 -5.68279102e-02
4.78019565e-01 1.11455776e-01 -1.49968132e-01 -7.21068203e-01
3.05810034e-01 4.86535460e-01 5.29835463e-01 -7.98713744e-01
5.34826636e-01 6.44951880e-01 1.01175413e-01 -1.14154136e+00
-9.18148339e-01 -8.09216499e-01 -9.57632810e-02 -8.32775980e-03
6.22597635e-01 -7.71945059e-01 -6.58687055e-01 4.06769246e-01
-1.36009872e+00 -4.55743298e-02 -5.62385738e-01 7.51530707e-01
-8.43570948e-01 5.54976761e-01 -5.76176047e-01 -1.13741791e+00
-2.74662767e-02 -9.58007395e-01 1.12958109e+00 -2.47924700e-01
-2.65764408e-02 -1.16304207e+00 2.61231124e-01 -2.73384035e-01
2.81732619e-01 3.57475340e-01 7.71114945e-01 -4.46386874e-01
-2.36109376e-01 -4.12333906e-01 4.77457903e-02 7.16227293e-01
-8.74944180e-02 -1.86360449e-01 -8.51528585e-01 -3.85875642e-01
8.57748628e-01 8.64121020e-02 9.78055656e-01 1.00730979e+00
1.23192215e+00 -6.51530683e-01 -3.55686307e-01 8.85214984e-01
1.41621614e+00 1.35054946e-01 3.26693654e-01 -1.89007699e-01
2.87643433e-01 6.26815736e-01 2.26300076e-01 7.12419212e-01
-2.31186822e-02 6.92321360e-01 3.26610327e-01 9.49983746e-02
2.27905408e-01 9.83672142e-02 6.59868658e-01 1.00314057e+00
-6.68230429e-02 -2.26908550e-01 -4.19958621e-01 6.73950851e-01
-1.99638891e+00 -1.08568227e+00 -1.35596976e-01 2.49227738e+00
8.12616587e-01 -1.13908932e-01 2.55275428e-01 2.70731717e-01
9.49652612e-01 2.75605023e-01 -6.57804251e-01 -2.19260424e-01
-3.70993108e-01 1.08747311e-01 7.42734075e-01 9.63255286e-01
-1.30919707e+00 -2.59530800e-03 7.38125515e+00 8.81952643e-01
-7.48384237e-01 1.76360175e-01 2.83806682e-01 -7.88016766e-02
-4.27052200e-01 1.69596430e-02 -4.57103491e-01 5.41420937e-01
7.29519963e-01 -4.92763549e-01 8.08034718e-01 8.31273615e-01
2.78691500e-01 1.57541245e-01 -1.22454274e+00 1.11705637e+00
1.04802914e-01 -7.53871202e-01 -5.99282801e-01 -1.04621043e-02
7.19256878e-01 -2.11067244e-01 3.46540250e-02 -2.06760075e-02
-1.19075188e-02 -7.32274413e-01 9.65342879e-01 6.74713254e-01
5.01620173e-01 -6.25798047e-01 5.35158992e-01 3.84694993e-01
-1.38845766e+00 -2.63042841e-02 -2.42933363e-01 -9.45732743e-02
5.51844180e-01 1.31132233e+00 -1.03540279e-01 4.85096872e-01
6.33262157e-01 8.99865270e-01 1.73230618e-01 1.34759986e+00
-6.79901838e-02 7.61171877e-01 -7.64673233e-01 2.66508192e-01
-1.48745075e-01 -6.44244194e-01 1.22239006e+00 1.12807846e+00
5.10417283e-01 -2.01009624e-02 5.27170658e-01 8.79083097e-01
1.16418362e-01 -6.36899248e-02 -3.86204779e-01 -2.32407022e-02
3.41275036e-01 1.06070590e+00 -4.59526092e-01 -8.70730802e-02
-5.70712209e-01 8.67108822e-01 -7.98503309e-02 8.64546239e-01
-6.58479333e-01 -6.64702177e-01 8.11133742e-01 -4.35492620e-02
3.70796114e-01 -2.18435988e-01 -5.76940477e-01 -1.31995642e+00
5.79425812e-01 -7.22262561e-01 6.86997354e-01 -4.91283000e-01
-1.68218350e+00 3.45277101e-01 2.12741345e-01 -1.31874788e+00
-3.00493598e-01 -6.15714014e-01 -5.70061922e-01 1.03409600e+00
-1.27587521e+00 -5.05236804e-01 2.63654888e-01 7.46925235e-01
1.82903230e-01 1.14325084e-01 5.49083114e-01 5.12536108e-01
-6.26293242e-01 4.20146823e-01 2.67066151e-01 -2.34592795e-01
4.02042001e-01 -1.24255323e+00 -2.66438365e-01 1.05108476e+00
-8.55945945e-02 6.12021446e-01 1.07841814e+00 -3.04962993e-01
-1.28991544e+00 -7.17662156e-01 9.56862271e-01 6.14682734e-02
9.03954327e-01 -2.88510650e-01 -8.59862089e-01 8.20241988e-01
5.72236329e-02 -1.30972281e-01 6.91019773e-01 1.31389752e-01
-1.77026093e-01 -1.70769587e-01 -1.12164044e+00 2.27142513e-01
7.59880483e-01 -3.15571517e-01 -4.75280166e-01 6.55712724e-01
4.98921424e-01 -1.68526769e-01 -7.03320563e-01 3.23009938e-01
3.01309943e-01 -6.92910194e-01 1.10279965e+00 -6.27140522e-01
-2.78242212e-02 -5.77608883e-01 -6.16047561e-01 -1.26582789e+00
-3.65736991e-01 -1.13888454e+00 -4.25424248e-01 1.01249409e+00
4.65567201e-01 -8.74497414e-01 3.51364851e-01 4.12971556e-01
-3.71726513e-01 -7.58189499e-01 -1.14750910e+00 -1.17587185e+00
-1.71458393e-01 -5.31141341e-01 3.68577480e-01 8.87905240e-01
2.75498003e-01 2.03545839e-01 -5.20445287e-01 2.84195721e-01
8.58451188e-01 2.35258788e-01 5.08682504e-02 -1.21661961e+00
-7.86743939e-01 -6.87213063e-01 -3.83706391e-01 -1.34377575e+00
2.19112203e-01 -1.00003159e+00 2.12415680e-01 -1.32945895e+00
5.40058352e-02 -2.89765716e-01 -1.80758744e-01 4.05883729e-01
1.31044509e-02 -1.51459172e-01 6.51562884e-02 3.87552321e-01
-3.15689206e-01 6.23212695e-01 9.11600232e-01 -2.35180214e-01
-2.54059404e-01 4.50238407e-01 -8.25623870e-01 1.03427935e+00
4.36060667e-01 -4.87705171e-01 -3.30606133e-01 -2.95630634e-01
3.54292005e-01 3.32134813e-01 3.89620781e-01 -6.88501418e-01
2.54127413e-01 -3.70804891e-02 -1.21129952e-01 -5.68482101e-01
4.50568914e-01 -8.02688301e-01 2.92323142e-01 1.57721102e-01
-2.98792928e-01 -9.34299529e-02 -7.71886557e-02 7.09589124e-01
-5.37833333e-01 -5.16397953e-01 9.39089537e-01 1.54984251e-01
-1.61217883e-01 4.05809820e-01 -1.92654848e-01 2.19144553e-01
8.95822346e-01 7.94103593e-02 3.37641388e-02 -8.41042161e-01
-1.04705322e+00 2.37993211e-01 5.11286631e-02 -8.88256803e-02
4.16274965e-01 -1.57860160e+00 -8.50935280e-01 2.29833871e-01
-3.10035765e-01 -1.31063104e-01 1.39113858e-01 1.46155536e+00
-5.92081808e-02 3.27313095e-01 3.69606167e-01 -5.38219690e-01
-7.39066780e-01 7.56325841e-01 4.19066876e-01 -2.60359675e-01
-4.25682694e-01 9.23349202e-01 4.15277719e-01 -1.67877853e-01
5.33910632e-01 -3.35575312e-01 1.52215082e-02 2.68667459e-01
6.13837600e-01 8.03642511e-01 1.60109863e-01 -5.03758192e-01
-5.42124987e-01 7.20988929e-01 4.70862091e-01 -4.80327129e-01
1.34860849e+00 -1.37864292e-01 -5.94118595e-01 5.59855044e-01
1.58927369e+00 3.31993431e-01 -1.17916548e+00 -2.43280932e-01
-1.25660568e-01 -2.36542240e-01 -6.36253953e-02 -3.84896338e-01
-1.14881432e+00 7.08581030e-01 1.84203252e-01 7.52835214e-01
1.58661091e+00 4.03486714e-02 5.94015419e-01 6.22521639e-02
2.88065344e-01 -9.58184540e-01 -3.71918947e-01 4.57816064e-01
1.22119486e+00 -6.79239333e-01 -1.00669473e-01 -5.72280943e-01
-2.98691988e-01 1.02841270e+00 -1.92231715e-01 -3.91199499e-01
1.14713442e+00 4.01349664e-01 -4.31302696e-01 8.49170610e-02
-5.58636367e-01 -1.14432171e-01 4.69608068e-01 5.51339865e-01
4.25952226e-01 3.36327851e-01 -6.54580951e-01 1.01747549e+00
-3.75787131e-02 -2.40758657e-01 3.58212173e-01 5.59628367e-01
-3.23720008e-01 -8.08146954e-01 -6.30762160e-01 4.08906251e-01
-6.63124204e-01 -1.57763451e-01 -1.42201751e-01 1.24243215e-01
-1.12410866e-01 1.08140743e+00 -2.61729717e-01 -3.80497705e-03
4.71274972e-01 3.88236460e-03 3.44620854e-01 -3.93225759e-01
1.06886618e-01 5.26955724e-01 9.82574150e-02 -5.28951406e-01
-4.28284526e-01 -1.01422083e+00 -8.05910349e-01 -1.60776362e-01
-3.13364178e-01 3.29858661e-01 5.52317381e-01 8.38625312e-01
6.04097582e-02 3.66454482e-01 8.03222477e-01 -1.03050554e+00
-9.50694382e-01 -6.88343823e-01 -1.15202558e+00 1.82641119e-01
8.08600128e-01 -5.67972958e-01 -8.54720354e-01 2.16703698e-01] | [7.100869655609131, 4.327635288238525] |
6af65cea-d543-4b1c-a2e1-e43130bab8c4 | or-nerf-object-removing-from-3d-scenes-guided | 2305.10503 | null | https://arxiv.org/abs/2305.10503v2 | https://arxiv.org/pdf/2305.10503v2.pdf | OR-NeRF: Object Removing from 3D Scenes Guided by Multiview Segmentation with Neural Radiance Fields | The emergence of Neural Radiance Fields (NeRF) for novel view synthesis has led to increased interest in 3D scene editing. One important task in editing is removing objects from a scene while ensuring visual reasonability and multiview consistency. However, current methods face challenges such as time-consuming object labelling, limited capability to remove specific targets, and compromised rendering quality after removal. This paper proposes a novel object-removing pipeline, named OR-NeRF, that can remove objects from 3D scenes with either point or text prompts on a single view, achieving better performance in less time than previous works. Our method uses a points projection strategy to rapidly spread user annotations to all views, significantly reducing the processing burden. This algorithm allows us to leverage the recent 2D segmentation model Segment-Anything (SAM) to predict masks with improved precision and efficiency. Additionally, we obtain colour and depth priors through 2D inpainting methods. Finally, our algorithm employs depth supervision and perceptual loss for scene reconstruction to maintain consistency in geometry and appearance after object removal. Experimental results demonstrate that our method achieves better editing quality with less time than previous works, considering both quality and quantity. | ['Guosheng Lin', 'Fan Yang', 'Zhoujie Fu', 'Youtan Yin'] | 2023-05-17 | null | null | null | null | ['novel-view-synthesis'] | ['computer-vision'] | [ 7.03298748e-01 5.54136224e-02 2.90868551e-01 -3.77364188e-01
-5.88064790e-01 -7.68286109e-01 4.24508393e-01 8.33543316e-02
-1.12932488e-01 3.19262683e-01 -6.50809892e-03 -4.89959531e-02
1.48207143e-01 -6.87485576e-01 -7.93991566e-01 -2.08879262e-01
6.54146135e-01 1.20610103e-01 5.00219345e-01 -5.59149384e-02
3.10589910e-01 9.92641687e-01 -1.66935158e+00 2.07502589e-01
1.04105151e+00 8.79491806e-01 6.34698033e-01 7.12678850e-01
-1.44373102e-04 8.10096085e-01 -2.85453558e-01 -4.49063838e-01
6.50194108e-01 -2.79609919e-01 -5.18827856e-01 5.94116509e-01
8.85943890e-01 -8.09034228e-01 -1.14809029e-01 8.94585073e-01
5.65195560e-01 2.12983891e-01 3.89266670e-01 -9.84760702e-01
-3.86014432e-01 -2.23912120e-01 -9.65564668e-01 -3.28617603e-01
4.59668070e-01 2.33848825e-01 7.05542684e-01 -1.03874397e+00
7.60006130e-01 1.14989483e+00 5.73955119e-01 4.37420070e-01
-1.41379166e+00 -4.63806808e-01 3.55042249e-01 -3.23072746e-02
-1.20683563e+00 -6.50486708e-01 9.15290356e-01 -4.79493380e-01
1.06955028e+00 6.91836238e-01 8.56464326e-01 6.39002502e-01
9.67461318e-02 6.35045171e-01 1.01498902e+00 -4.57021832e-01
1.48971781e-01 1.59604594e-01 -3.83976072e-01 7.12591827e-01
-1.33917481e-01 -2.30640825e-02 -5.08016527e-01 1.33668840e-01
1.09414196e+00 6.91474825e-02 -4.81146961e-01 -6.17096603e-01
-1.31047225e+00 3.03339213e-01 2.56450981e-01 -3.62457156e-01
-4.71398324e-01 8.56250376e-02 1.36707306e-01 3.48451100e-02
9.47369099e-01 3.24777991e-01 -4.79050785e-01 2.24938571e-01
-9.63874996e-01 3.00523937e-01 3.03358614e-01 1.22128868e+00
7.46797860e-01 5.59783243e-02 -1.02363974e-01 1.04118311e+00
1.46343738e-01 5.47368169e-01 -3.78127992e-01 -1.49290264e+00
3.37962687e-01 6.16925120e-01 1.35158375e-01 -1.10939729e+00
-3.26248437e-01 -2.49650925e-01 -7.56227255e-01 8.38009357e-01
-5.27977347e-02 3.38670194e-01 -1.12262821e+00 1.26240230e+00
9.07488286e-01 -6.50831014e-02 -3.29003334e-01 1.02635610e+00
6.62864983e-01 5.66671312e-01 -2.62796909e-01 -6.10870197e-02
1.22781754e+00 -9.88789737e-01 -6.18215561e-01 -3.12922776e-01
2.19028041e-01 -1.13025343e+00 9.84243453e-01 7.40618229e-01
-1.50083005e+00 -4.71738130e-01 -9.01322007e-01 -7.09331095e-01
1.79491639e-02 -3.69704105e-02 7.39449084e-01 4.43077058e-01
-9.38698232e-01 5.02219200e-01 -7.56766260e-01 -1.70560807e-01
6.23639226e-01 2.42184967e-01 -3.41564476e-01 -3.06451708e-01
-5.10587096e-01 8.12138140e-01 1.82816029e-01 6.25563413e-02
-7.71898687e-01 -1.11868834e+00 -9.75349247e-01 -1.60777479e-01
7.07267344e-01 -1.05782402e+00 1.20727336e+00 -9.17010128e-01
-1.71501946e+00 8.86362314e-01 -2.50886202e-01 -6.10693265e-03
7.12477267e-01 -4.62732434e-01 -1.12989256e-02 2.40028873e-01
1.06772862e-01 9.56990242e-01 1.03767395e+00 -1.66034830e+00
-7.66192019e-01 -2.32573554e-01 1.77216247e-01 7.15283871e-01
2.91822590e-02 -6.73732907e-02 -1.09819961e+00 -7.80630291e-01
4.92731035e-01 -5.59229672e-01 -3.47345561e-01 5.86017609e-01
-4.35137004e-01 2.71473050e-01 8.48587632e-01 -9.66535449e-01
7.05825210e-01 -2.26370645e+00 2.50215977e-01 9.35936645e-02
5.05806446e-01 8.39932561e-02 -7.91842714e-02 1.21875390e-01
1.00832641e-01 -2.29688346e-01 -4.56276506e-01 -8.05249929e-01
-2.06296131e-01 1.54299602e-01 -2.86100745e-01 4.36754346e-01
2.20437333e-01 7.38676488e-01 -7.18330562e-01 -4.90077049e-01
9.25780475e-01 6.63946986e-01 -9.41506743e-01 2.84028769e-01
-5.97915351e-01 7.34270096e-01 5.56843951e-02 7.57805347e-01
1.08295655e+00 -1.78338662e-01 -5.90848736e-02 -4.51974034e-01
-2.83400804e-01 1.60750389e-01 -1.45162094e+00 2.07306886e+00
-7.53927112e-01 5.72811425e-01 3.63157481e-01 -3.91050130e-01
6.96968317e-01 -2.42741574e-02 5.00473440e-01 -8.52317274e-01
-2.59251595e-02 -5.73769305e-03 -5.38737476e-01 -2.43441284e-01
8.10986400e-01 1.31371021e-01 3.20845008e-01 3.33382785e-01
-2.54255533e-01 -8.37235391e-01 -6.46784902e-02 3.39248538e-01
6.83639586e-01 6.49467230e-01 1.43524498e-01 2.11166933e-01
2.16192290e-01 -1.39870949e-03 6.87905431e-01 5.09340405e-01
2.60562241e-01 1.14925468e+00 4.64415224e-03 -1.76948324e-01
-1.10192597e+00 -1.07707429e+00 1.38005316e-02 7.80134082e-01
4.22691762e-01 -3.12637120e-01 -5.33342302e-01 -4.28078175e-01
-1.59326613e-01 9.04065073e-01 -2.91798562e-01 1.47508293e-01
-6.09713793e-01 -1.43985599e-01 -4.75252345e-02 3.16359252e-01
3.63063395e-01 -8.07993412e-01 -1.02103257e+00 7.06903338e-02
-1.62447631e-01 -1.26060581e+00 -5.78250408e-01 1.11209927e-02
-8.98175478e-01 -9.46802318e-01 -6.31973982e-01 -5.49698591e-01
8.89396489e-01 6.69479847e-01 1.07627594e+00 3.41209732e-02
-4.38875943e-01 6.31302059e-01 -2.16264278e-01 -4.13013995e-01
-3.18869919e-01 -2.67220795e-01 -3.15406770e-01 -1.42103638e-02
-3.62942755e-01 -5.95952511e-01 -8.25808465e-01 3.31274688e-01
-1.16189611e+00 9.17933047e-01 4.62933421e-01 3.58656138e-01
8.11941385e-01 -7.00771436e-02 8.00474081e-03 -8.55440855e-01
-7.55609423e-02 5.14569134e-02 -8.93492639e-01 6.56464994e-02
-5.68682194e-01 -3.49969745e-01 4.60545570e-01 -1.21430986e-01
-1.59054947e+00 3.28966320e-01 -1.86624117e-02 -6.49754405e-01
-2.54399508e-01 -1.00646093e-01 -4.05586123e-01 -8.62412304e-02
4.04834896e-01 1.86462685e-01 -7.67118931e-02 -5.51413298e-01
7.08833694e-01 2.91769356e-01 6.24737442e-01 -1.50510937e-01
8.24368894e-01 8.89360011e-01 -1.33809124e-04 -8.67497623e-01
-9.62897003e-01 -4.73118037e-01 -8.27965319e-01 -5.25945306e-01
8.48066509e-01 -1.12337959e+00 -4.83354241e-01 5.37498772e-01
-1.35099316e+00 -3.54964167e-01 -3.28904122e-01 3.41128230e-01
-5.28405249e-01 5.87341964e-01 -4.83583421e-01 -6.91670239e-01
-2.42142558e-01 -1.01432383e+00 1.44712603e+00 3.44626158e-02
-1.63666710e-01 -7.20591962e-01 -2.85278261e-01 6.26842737e-01
2.27734938e-01 4.44576770e-01 7.88328886e-01 4.95698601e-01
-1.17019880e+00 2.21433505e-01 -4.76184011e-01 4.69978660e-01
2.36831665e-01 -3.27949002e-02 -1.13990533e+00 -1.72347978e-01
7.15381885e-03 -3.63989733e-02 5.57369649e-01 4.03389424e-01
1.29318988e+00 -8.05945545e-02 -2.41781130e-01 1.07889187e+00
1.46830034e+00 2.26033047e-01 6.01067722e-01 3.01277965e-01
1.26881766e+00 8.04109216e-01 8.18410039e-01 4.83510762e-01
3.63236457e-01 8.56391251e-01 6.54005587e-01 -5.26016295e-01
-5.34483254e-01 -1.59261897e-01 7.73126185e-02 7.76480794e-01
4.96580033e-03 -2.86056370e-01 -7.37140298e-01 5.11144102e-01
-1.79755223e+00 -6.24569952e-01 -4.43433851e-01 2.11983848e+00
8.85358632e-01 -8.07772726e-02 -2.70396680e-01 6.14423454e-02
4.52955544e-01 1.85533136e-01 -6.14560366e-01 -2.72714823e-01
-8.19993541e-02 1.35327846e-01 6.69769406e-01 7.70999730e-01
-8.03767502e-01 1.01630318e+00 5.43920231e+00 6.44440353e-01
-1.07282424e+00 1.08166501e-01 6.07189655e-01 -4.53055233e-01
-5.15742540e-01 7.62911364e-02 -4.99886751e-01 7.32296854e-02
1.37484699e-01 3.16536188e-01 5.60018897e-01 7.32120514e-01
4.63644177e-01 -4.81499046e-01 -1.03213620e+00 1.30387509e+00
4.53692108e-01 -1.35039330e+00 4.63276878e-02 -6.95439875e-02
8.92671525e-01 -1.57471552e-01 -1.07309647e-01 -3.29928696e-01
1.62684321e-01 -7.43832707e-01 1.13254452e+00 7.83407271e-01
9.69979286e-01 -8.63655746e-01 9.13220420e-02 2.94657081e-01
-1.12970340e+00 3.05742472e-01 -2.62749940e-01 9.40589905e-02
6.27576649e-01 9.13855553e-01 -8.80637825e-01 8.73744428e-01
8.34529936e-01 8.13787282e-01 -5.56303561e-01 9.71882522e-01
-1.63669989e-01 2.56529227e-02 -4.02393043e-01 5.69929302e-01
-2.28896767e-01 -3.40058208e-01 7.01341987e-01 8.71418118e-01
1.60960659e-01 2.74784774e-01 1.58285648e-01 9.82360780e-01
9.21745822e-02 -5.72323464e-02 -5.18459916e-01 5.62121749e-01
2.65371770e-01 1.33002102e+00 -9.25019324e-01 -2.22618237e-01
-3.08349252e-01 1.70154452e+00 2.22940758e-01 3.20480078e-01
-9.03724134e-01 -2.91524351e-01 5.01453996e-01 3.95405680e-01
3.77816647e-01 -4.12307858e-01 -6.92181528e-01 -1.05324900e+00
2.39663154e-01 -7.71254539e-01 -1.10883169e-01 -1.10685492e+00
-9.98607993e-01 3.33536416e-01 4.10920009e-03 -1.34861052e+00
1.64368898e-01 -3.47869664e-01 -9.68868807e-02 8.52238059e-01
-1.56566048e+00 -1.35267663e+00 -5.79511285e-01 5.36564767e-01
8.72796118e-01 3.80919486e-01 4.43707407e-01 4.95182931e-01
-2.81041414e-01 1.61017239e-01 -9.65720788e-02 -4.68640864e-01
7.99156606e-01 -1.13699687e+00 5.06753266e-01 1.00042987e+00
-2.40178499e-02 3.32034588e-01 4.25865024e-01 -8.12896132e-01
-1.45087945e+00 -1.22586381e+00 5.61034620e-01 -6.85402930e-01
-2.71991313e-01 -6.93264723e-01 -7.68636882e-01 6.13040984e-01
1.36186138e-01 -8.38409290e-02 3.36831540e-01 -3.18741351e-01
-1.51713267e-01 -4.42361534e-02 -1.15250099e+00 9.48030412e-01
1.33716977e+00 -4.36323702e-01 -1.74156368e-01 2.32359409e-01
8.41128111e-01 -9.19862807e-01 -7.04503417e-01 5.14209509e-01
4.97938305e-01 -1.33841574e+00 1.17687404e+00 1.14779696e-01
4.17108357e-01 -7.74954915e-01 -1.28355622e-01 -1.19786668e+00
-2.62852222e-01 -6.37826085e-01 -5.94806932e-02 1.26573610e+00
-2.80799996e-02 -3.06909025e-01 4.91491705e-01 9.07560825e-01
-3.55817318e-01 -5.09558141e-01 -6.57278359e-01 -3.57909858e-01
-6.05790675e-01 -8.49720299e-01 4.78345692e-01 1.08665240e+00
-8.44629765e-01 2.52030373e-01 -5.83450794e-01 4.61204767e-01
8.24792087e-01 2.73038656e-01 1.14111590e+00 -1.12837875e+00
-3.21135074e-01 -1.87266380e-01 -1.65275604e-01 -1.32310605e+00
-2.32115760e-01 -8.59782219e-01 1.09464206e-01 -1.92844760e+00
3.64042185e-02 -4.22266632e-01 3.25870007e-01 3.56296510e-01
-7.73205310e-02 6.12553418e-01 3.08925927e-01 4.35091332e-02
-5.73592663e-01 6.36589110e-01 1.47127533e+00 1.60298973e-01
-3.57686281e-01 -3.27202171e-01 -5.45485079e-01 1.02107668e+00
5.54904282e-01 -4.12987798e-01 -5.22321641e-01 -8.78767908e-01
3.79901111e-01 -8.80951807e-02 7.01129854e-01 -8.45693767e-01
4.74829599e-02 -2.06966221e-01 7.19728649e-01 -1.10374665e+00
6.03837013e-01 -1.08611751e+00 4.44746763e-01 2.92579941e-02
2.81453077e-02 -7.95398653e-02 3.52510959e-01 6.15223229e-01
2.10462153e-01 -5.32269776e-02 9.04474080e-01 -1.64479285e-01
-8.13718259e-01 3.43469113e-01 -2.26298999e-02 -2.65035510e-01
1.14465463e+00 -5.00741720e-01 1.16267398e-01 -2.46545836e-01
-5.89140177e-01 1.20446578e-01 1.06280220e+00 4.47691292e-01
9.39341545e-01 -1.12707150e+00 -5.22256911e-01 3.50662649e-01
7.51392022e-02 7.24865317e-01 6.66832089e-01 8.01763177e-01
-9.25838649e-01 7.28654116e-02 -7.44578913e-02 -9.72705841e-01
-1.53894126e+00 2.68568963e-01 2.64856577e-01 2.08124876e-01
-1.09701335e+00 8.19114506e-01 5.79229474e-01 -5.98144114e-01
2.33818963e-01 -3.53165984e-01 2.64491737e-01 -1.36936650e-01
4.26550746e-01 5.19098401e-01 2.32459858e-01 -3.75844419e-01
-2.53641635e-01 7.50234902e-01 -2.66758412e-01 -3.28588486e-01
1.40414464e+00 -5.34412265e-01 -1.58809230e-01 3.68775874e-01
8.79812121e-01 3.84869814e-01 -1.77478600e+00 -8.10293481e-02
-5.35246491e-01 -1.08961463e+00 3.59200001e-01 -9.18632448e-01
-1.03461075e+00 6.53040707e-01 6.04250968e-01 -1.34160727e-01
1.28793037e+00 -1.35473818e-01 8.62218082e-01 -1.10413112e-01
3.45268935e-01 -1.14412820e+00 -8.66174996e-02 3.24801505e-01
1.11486137e+00 -1.17917156e+00 3.09954047e-01 -8.37555349e-01
-6.60242915e-01 9.70050037e-01 7.96399176e-01 1.64064467e-01
3.20814639e-01 2.33431652e-01 5.50809428e-02 -2.78150767e-01
-4.66600895e-01 -2.49707680e-02 4.70826000e-01 7.29817212e-01
2.05161855e-01 -2.89276689e-01 1.69374391e-01 -1.11601345e-01
-5.44607267e-02 -2.62814254e-01 5.23714602e-01 9.53756571e-01
-8.82754847e-02 -9.59423184e-01 -4.03737158e-01 4.27475452e-01
-2.26866752e-01 -2.39154443e-01 -2.71388441e-01 5.25695562e-01
2.02545762e-01 8.48318696e-01 9.82515067e-02 -1.63943209e-02
6.72132015e-01 -3.27500492e-01 7.44431496e-01 -9.01982486e-01
-4.71083909e-01 3.40306520e-01 1.26572400e-02 -9.54717040e-01
-5.72075903e-01 -4.88247603e-01 -1.12220061e+00 -1.79842189e-01
-4.77171779e-01 -6.65544033e-01 7.85203278e-01 5.42646825e-01
6.41974986e-01 7.13759124e-01 6.78906143e-01 -1.32173789e+00
9.71732363e-02 -4.82854277e-01 -4.51910466e-01 3.44685405e-01
3.04096550e-01 -5.66014349e-01 -1.68416768e-01 3.29117984e-01] | [9.1834077835083, -3.038806438446045] |
b70c0071-be5c-4d4c-a8cf-212dbe98548f | covidctnet-an-open-source-deep-learning | 2005.03059 | null | https://arxiv.org/abs/2005.03059v3 | https://arxiv.org/pdf/2005.03059v3.pdf | CovidCTNet: An Open-Source Deep Learning Approach to Identify Covid-19 Using CT Image | Coronavirus disease 2019 (Covid-19) is highly contagious with limited treatment options. Early and accurate diagnosis of Covid-19 is crucial in reducing the spread of the disease and its accompanied mortality. Currently, detection by reverse transcriptase polymerase chain reaction (RT-PCR) is the gold standard of outpatient and inpatient detection of Covid-19. RT-PCR is a rapid method, however, its accuracy in detection is only ~70-75%. Another approved strategy is computed tomography (CT) imaging. CT imaging has a much higher sensitivity of ~80-98%, but similar accuracy of 70%. To enhance the accuracy of CT imaging detection, we developed an open-source set of algorithms called CovidCTNet that successfully differentiates Covid-19 from community-acquired pneumonia (CAP) and other lung diseases. CovidCTNet increases the accuracy of CT imaging detection to 90% compared to radiologists (70%). The model is designed to work with heterogeneous and small sample sizes independent of the CT imaging hardware. In order to facilitate the detection of Covid-19 globally and assist radiologists and physicians in the screening process, we are releasing all algorithms and parametric details in an open-source format. Open-source sharing of our CovidCTNet enables developers to rapidly improve and optimize services, while preserving user privacy and data ownership. | ['Reza Rawassizadeh', 'Benjamin Haibe-Kains', 'L. T. Chitkushev', 'Mannudeep K. Kalra', 'Rana Jahanban-Esfahlan', 'Mohammad Hadi Gharib', 'Amir Reza Radmard', 'Seyed Ali Javad Mousavi', 'Rosa Babaei', 'Reza Reiazi', 'Omid Ghaemi', 'Mehdi Hosseinzadeh', 'Hadi Karimi Mobin', 'Azadeh Laali', 'Fatemeh Homayounieh', 'Engy Abbas', 'Zohreh Amoozgar', 'Reza Malekzadeh', 'Tahereh Javaheri', 'Morteza Homayounfar', 'Khaled Seidi', 'Guanglan Zhang'] | 2020-05-06 | null | null | null | null | ['covid-19-image-segmentation'] | ['computer-vision'] | [ 2.11262647e-02 -6.57536924e-01 -1.45090684e-01 2.23512635e-01
-6.62289619e-01 -9.06610191e-01 -1.11531183e-01 4.93900269e-01
-5.73759258e-01 4.43300754e-01 -1.21182106e-01 -9.26759541e-01
8.83405283e-02 -5.06239414e-01 -2.42765799e-01 -5.84450603e-01
-1.16751663e-01 1.03805137e+00 5.04344344e-01 7.25052416e-01
-3.89330268e-01 8.20080340e-01 -9.25808668e-01 2.06574142e-01
7.39600062e-01 4.64192241e-01 5.76253414e-01 1.21318460e+00
1.76531047e-01 6.42786682e-01 -2.14656070e-01 8.61489847e-02
1.66659087e-01 -3.22416961e-01 -5.71078420e-01 -4.72964376e-01
-9.97826681e-02 -8.86553347e-01 1.65110692e-01 3.36513937e-01
5.43847620e-01 -5.69040656e-01 1.00528991e+00 -1.04286253e+00
-2.70919688e-02 -2.17262238e-01 -4.54970568e-01 5.61885357e-01
8.81579816e-02 5.78892350e-01 4.70561415e-01 -6.15311444e-01
6.83281362e-01 7.35481322e-01 9.87848103e-01 6.78574383e-01
-7.73687482e-01 -7.20879197e-01 -5.78237057e-01 -8.59063715e-02
-1.51678932e+00 3.01387608e-01 -2.90638626e-01 -9.22779977e-01
1.09254074e+00 4.56626445e-01 1.13087189e+00 9.71592426e-01
4.41347510e-01 3.20582420e-01 1.00526953e+00 1.33297578e-01
9.07854661e-02 7.34800100e-02 8.67093634e-03 7.59375930e-01
9.95637298e-01 -1.67964160e-01 2.82067508e-01 -8.08711886e-01
1.04697287e+00 8.60226631e-01 -3.87215883e-01 -3.26849043e-01
-1.30235958e+00 9.00183260e-01 -3.83709781e-02 1.88254252e-01
-4.51738358e-01 -2.30109155e-01 7.09667683e-01 -3.84884030e-01
-2.26503108e-02 1.53478712e-01 -5.43272913e-01 -9.65207741e-02
-5.26214182e-01 7.99583867e-02 4.83793199e-01 7.17939973e-01
6.47122459e-03 -4.27945465e-01 -4.52492565e-01 6.42827153e-01
3.56562138e-01 1.39282298e+00 2.54271239e-01 -6.11156464e-01
2.25593239e-01 4.49199200e-01 3.57520610e-01 -5.14225781e-01
-6.24430835e-01 -3.09594512e-01 -8.83991599e-01 -4.05203521e-01
9.46294367e-02 -4.38271523e-01 -9.14025307e-01 1.33404160e+00
3.82159114e-01 2.79298723e-01 -3.20497304e-01 8.68277967e-01
4.24036562e-01 4.56488907e-01 4.29948628e-01 -4.80677426e-01
1.97002161e+00 -5.55738807e-01 -2.84439743e-01 7.70855555e-03
9.62619245e-01 -6.50657594e-01 6.28103316e-01 1.46333911e-02
-4.95181948e-01 2.45865852e-01 -7.57102907e-01 5.02820790e-01
-3.97148840e-02 -6.65554181e-02 2.06838846e-02 8.50667059e-01
-9.08188224e-01 -1.52037635e-01 -1.17795432e+00 -9.80817616e-01
5.64622700e-01 2.41509140e-01 -1.79729208e-01 -3.77035797e-01
-7.73889422e-01 9.33436990e-01 -5.63155711e-02 -4.67418700e-01
-9.77240503e-01 -8.50933671e-01 -3.25136453e-01 -2.33423844e-01
1.36832535e-01 -1.52460706e+00 1.32165444e+00 1.52354300e-01
-7.14399934e-01 9.82380271e-01 -3.68696213e-01 -2.68736035e-01
5.92463076e-01 -2.17351481e-01 -1.76368520e-01 7.26101339e-01
2.80844361e-01 1.70410246e-01 5.06127894e-01 -8.64313900e-01
-8.49147201e-01 -3.46249878e-01 -8.27174067e-01 -1.67170659e-01
8.21952000e-02 5.18109322e-01 -2.89877445e-01 -3.50241035e-01
-3.80312234e-01 -1.30559421e+00 -3.95133793e-01 4.79361527e-02
4.18908596e-02 -3.82798687e-02 1.22338402e+00 -5.88504791e-01
1.02289724e+00 -1.96642542e+00 -8.86455655e-01 1.09056018e-01
6.80842638e-01 9.94698048e-01 1.29665509e-01 3.14254701e-01
3.43814820e-01 4.66921419e-01 -3.62087220e-01 2.01013446e-01
-5.48902988e-01 1.43440157e-01 1.14837006e-01 6.39324844e-01
2.75420278e-01 9.99366164e-01 -1.05336690e+00 -8.85244012e-01
3.10989112e-01 7.47581184e-01 -5.83588183e-01 5.02778828e-01
7.52312317e-02 7.43573308e-01 -7.62924910e-01 7.36562014e-01
7.27515817e-01 -9.78129983e-01 5.32008648e-01 2.75779158e-01
-4.99589033e-02 -8.79189745e-02 -5.87713301e-01 5.19717336e-01
-1.46493137e-01 2.40191117e-01 2.88158238e-01 -2.07786441e-01
3.80181521e-01 7.94622958e-01 6.90914452e-01 2.90425289e-02
3.66758049e-01 2.57862270e-01 -3.63259688e-02 -9.38536823e-01
-2.90925622e-01 -2.47651517e-01 5.38651109e-01 7.84514368e-01
-8.80845308e-01 -1.37826547e-01 -3.67123708e-02 1.69191003e-01
1.52077770e+00 -6.52469635e-01 6.47354186e-01 -3.95237952e-02
2.25075155e-01 3.97184283e-01 5.08275092e-01 8.37969363e-01
-5.16092062e-01 6.16350114e-01 2.92489510e-02 -1.24858856e-01
-1.02318561e+00 -1.31708968e+00 -3.64797622e-01 4.69272792e-01
-4.63133335e-01 -1.59604520e-01 -7.15263724e-01 -5.97853601e-01
4.88127097e-02 3.72881144e-01 -3.19805741e-01 4.07724500e-01
-6.73194468e-01 -7.51692176e-01 6.58447981e-01 6.96390390e-01
2.90765762e-01 -6.28294706e-01 -1.16307938e+00 2.81357259e-01
-5.56971490e-01 -1.11235452e+00 -6.05899155e-01 -5.25768548e-02
-1.02796495e+00 -1.56149781e+00 -1.00166380e+00 -8.85466337e-01
6.58406079e-01 7.12525487e-01 7.14347959e-01 5.12105942e-01
-9.36163127e-01 6.16144717e-01 -2.93354094e-01 -6.89978957e-01
-5.53374827e-01 -2.07370296e-01 1.48661630e-02 -6.36579454e-01
4.38550323e-01 -6.20629415e-02 -1.05163443e+00 4.35206503e-01
-6.77296996e-01 -1.99264839e-01 6.00903094e-01 5.24608254e-01
8.33819389e-01 -4.25279617e-01 1.89490482e-01 -9.27902758e-01
4.02096152e-01 -8.61682177e-01 -3.12178135e-01 8.58090594e-02
-8.69344890e-01 -6.08699858e-01 4.92753386e-01 -2.00654730e-01
-7.49327064e-01 -4.76775356e-02 2.94858124e-02 -4.66337204e-01
-2.03726277e-01 1.26531318e-01 7.52590358e-01 3.65505129e-01
4.41229403e-01 3.71541688e-03 3.57963383e-01 -3.09226930e-01
-3.95147413e-01 8.78690898e-01 2.70070761e-01 3.29850428e-02
4.63998467e-01 6.87181354e-01 4.22470532e-02 -8.37978363e-01
-4.68534291e-01 -1.22866619e+00 -2.21087232e-01 -5.53734116e-02
1.34572911e+00 -1.11501849e+00 -8.29255581e-01 5.11958838e-01
-1.07714653e+00 -1.20793760e-01 2.59082764e-01 9.15353358e-01
-1.35038301e-01 4.19488370e-01 -1.03524899e+00 -6.86154306e-01
-9.05835152e-01 -1.19616866e+00 9.35646772e-01 -1.04725845e-01
-4.84182835e-01 -8.01051199e-01 5.60000300e-01 3.15569311e-01
4.52728778e-01 2.75828481e-01 1.15842843e+00 -7.56949902e-01
-5.59432864e-01 -2.77110040e-01 -6.96040988e-01 1.04595527e-01
3.30955684e-01 3.88726085e-01 -6.05900049e-01 -5.28352916e-01
8.87262821e-02 6.58032969e-02 4.38196361e-01 7.08142877e-01
6.85713112e-01 -7.85170421e-02 -1.00325263e+00 4.84278768e-01
1.54516149e+00 6.05045080e-01 2.98109084e-01 1.46074966e-01
6.18359327e-01 2.79800713e-01 5.69056630e-01 6.40339673e-01
3.17002922e-01 4.26439159e-02 4.34165359e-01 -1.57686055e-01
2.35702783e-01 8.10425431e-02 -1.19815663e-01 8.72117639e-01
-4.02611703e-01 -4.05699283e-01 -1.52425313e+00 4.84530836e-01
-1.37773407e+00 -9.56057310e-01 -8.95882905e-01 1.96579742e+00
4.53081518e-01 -3.32669795e-01 3.71881217e-01 -4.31161046e-01
9.49484825e-01 -4.53944415e-01 -4.53297645e-01 -4.05570745e-01
2.15840980e-01 1.86924815e-01 5.91394305e-01 -2.97321212e-02
-8.28191400e-01 1.49540231e-01 7.55412006e+00 1.58741355e-01
-1.22380483e+00 4.82530266e-01 3.96694690e-01 6.45389780e-02
7.77017698e-02 -3.92051578e-01 -5.83191335e-01 3.47668052e-01
8.63113165e-01 -3.83074246e-02 -6.28339946e-02 9.57345665e-01
5.21553338e-01 -1.44349456e-01 -7.09225118e-01 8.44412267e-01
-7.59333447e-02 -1.41820192e+00 -2.00999036e-01 1.73476651e-01
5.20786405e-01 7.12807357e-01 -2.38381535e-01 -6.31386191e-02
3.06811273e-01 -7.94827700e-01 7.08928406e-02 8.94977823e-02
1.25821364e+00 -5.03152668e-01 1.14886630e+00 2.84066677e-01
-1.46589196e+00 1.16617821e-01 -8.51756483e-02 3.22065502e-01
3.53036284e-01 4.29200679e-01 -1.81587899e+00 -6.09364137e-02
9.46422338e-01 3.12219530e-01 -3.30173701e-01 1.19711196e+00
-8.15992653e-02 8.90573919e-01 -4.08249468e-01 -3.15362394e-01
-3.09934616e-02 1.06341615e-02 5.20743370e-01 1.71968853e+00
3.81084979e-01 5.59621930e-01 1.35650396e-01 3.92697513e-01
3.67476195e-01 3.26368600e-01 -7.24160969e-01 -1.36047706e-01
6.40645027e-01 1.06838226e+00 -9.45918322e-01 -5.42388976e-01
-4.67605591e-01 6.55177295e-01 -6.87546656e-02 1.47886306e-01
-9.46951449e-01 -4.77387458e-02 6.85355902e-01 5.86587906e-01
5.86103380e-01 -3.38199735e-02 -1.77306518e-01 -5.93119860e-01
-4.60727721e-01 -8.69857728e-01 7.63323426e-01 -7.17481077e-01
-1.12861884e+00 5.40197492e-01 -1.19043432e-01 -1.26353943e+00
-3.00060302e-01 -6.33270562e-01 -5.40346205e-01 7.60241747e-01
-1.12293422e+00 -7.55373538e-01 -3.91355067e-01 4.22499388e-01
5.30174188e-02 2.93824971e-01 1.10696268e+00 -1.15412913e-01
-5.83679855e-01 1.91465884e-01 4.12270993e-01 8.41733515e-02
6.86815202e-01 -7.56642818e-01 8.63441154e-02 5.01702070e-01
-1.08883429e+00 1.04473329e+00 3.90180737e-01 -1.07080567e+00
-1.07877040e+00 -1.57104075e+00 7.73829460e-01 -6.47590995e-01
5.08976042e-01 -2.41734255e-02 -7.03512430e-01 7.64364600e-01
4.34683859e-02 -2.09012963e-02 1.00284994e+00 -6.85345650e-01
-3.99960011e-01 4.25929010e-01 -1.46625245e+00 3.89718682e-01
7.32934535e-01 -5.96669853e-01 -4.29199487e-01 4.68915552e-01
6.31210864e-01 -7.04660788e-02 -9.61753249e-01 6.02465630e-01
7.60342062e-01 -8.03920925e-01 9.23524022e-01 -3.98958862e-01
-5.04847132e-02 -2.99339235e-01 2.08682820e-01 -6.65051043e-01
-5.82792222e-01 -1.85320467e-01 3.81310731e-01 5.10508418e-01
1.91673934e-01 -1.10683203e+00 6.94873095e-01 3.22038084e-01
3.32555324e-01 -7.68602133e-01 -4.20121551e-01 -7.85789907e-01
-1.75913289e-01 -2.76081830e-01 4.87652063e-01 9.66112077e-01
-1.90211356e-01 1.79896485e-02 3.31116058e-02 3.13131690e-01
4.49175149e-01 -2.84600332e-02 4.83723372e-01 -1.20420635e+00
-2.42746666e-01 -9.00769904e-02 -6.64627925e-02 -4.30318475e-01
-6.88126266e-01 -6.99701428e-01 -1.98428020e-01 -1.85921204e+00
9.31073606e-01 -4.50988829e-01 -7.74687231e-02 2.79486179e-01
-1.84339404e-01 3.12768936e-01 5.22036776e-02 5.53155124e-01
-2.46037543e-01 -2.46363327e-01 1.35478365e+00 1.62214786e-01
-5.63306324e-02 -3.79367471e-02 -1.22140132e-01 6.22193933e-01
9.92454708e-01 -1.01046956e+00 -3.45835418e-01 -1.31834030e-01
-9.79310349e-02 1.19154714e-01 4.89398986e-01 -6.23164773e-01
-1.07831806e-01 -4.64792639e-01 2.31492236e-01 -1.10762298e+00
-1.12090454e-01 -7.69683182e-01 6.49481237e-01 1.55657804e+00
5.40516853e-01 4.17534024e-01 2.17604980e-01 5.22578061e-01
3.58167410e-01 -1.52647033e-01 8.49453032e-01 -2.43994951e-01
-8.61451551e-02 4.55024064e-01 -1.42908597e+00 3.45747858e-01
1.31637645e+00 -1.37534916e-01 -7.39415646e-01 -1.33505270e-01
-1.59361124e-01 2.12443560e-01 7.68137097e-01 -1.37443602e-01
4.86355960e-01 -6.39611006e-01 -8.69119346e-01 4.66432795e-02
3.81085336e-01 9.11437571e-02 3.44148278e-01 1.49099648e+00
-1.49240577e+00 7.82926381e-01 1.93594933e-01 -1.09241164e+00
-1.71873534e+00 8.41458738e-01 2.74133682e-01 -4.10060495e-01
-6.95957899e-01 5.47739148e-01 4.02272284e-01 -9.59252939e-03
-8.87272805e-02 -2.72697300e-01 1.07724145e-01 -3.15372169e-01
7.11307466e-01 5.47084808e-01 -1.98941045e-02 -4.07276064e-01
-8.82829785e-01 4.38298464e-01 -2.28238270e-01 2.07587048e-01
1.07900882e+00 -6.06423654e-02 -3.03166896e-01 1.18963629e-01
1.13957143e+00 1.54302359e-01 -4.87225771e-01 2.12250173e-01
-4.41763967e-01 -2.69983202e-01 -3.99265438e-01 -6.08757794e-01
-5.41874707e-01 6.26382947e-01 8.80491436e-01 -8.89169276e-02
8.14573467e-01 2.40772516e-01 1.02738476e+00 1.74238712e-01
3.55088085e-01 -5.20657420e-01 -1.42550603e-01 2.19857350e-01
3.27996552e-01 -1.03432608e+00 -5.01350909e-02 -6.19780779e-01
-5.15737712e-01 6.56104982e-01 4.12678897e-01 1.92602530e-01
6.50790572e-01 6.09870434e-01 5.17516136e-01 -4.58485931e-01
-8.86988282e-01 1.24108620e-01 -2.93363124e-01 9.60068107e-01
3.86516809e-01 5.79590857e-01 -3.78188908e-01 4.51258868e-01
2.74133831e-01 1.31451130e-01 2.38897309e-01 1.37961721e+00
-5.20321548e-01 -8.27144563e-01 -8.41979742e-01 1.03966081e+00
-7.05012739e-01 -1.46734491e-01 -3.22184294e-01 9.11926687e-01
5.54613359e-02 1.03137887e+00 5.81654534e-02 -3.15124542e-02
1.06023692e-01 7.14181215e-02 2.12905556e-01 -6.81446552e-01
-4.76024598e-01 2.05730826e-01 1.01265021e-01 -2.40271777e-01
-3.19479316e-01 -9.47530866e-01 -1.58533943e+00 -3.97024363e-01
-2.33819976e-01 -2.40078196e-03 5.72442472e-01 7.17208326e-01
4.49266493e-01 2.77114093e-01 4.20487404e-01 1.64722037e-02
-3.49859595e-01 -4.84256566e-01 -6.14450634e-01 6.46600500e-02
3.60470206e-01 -2.80516148e-01 -4.45745051e-01 1.22379206e-01] | [15.563488960266113, -1.6926823854446411] |
46ea4b50-be61-42e5-9977-1f1cdf7ba149 | a-large-scale-dataset-for-empathetic-response | null | null | https://aclanthology.org/2021.emnlp-main.96 | https://aclanthology.org/2021.emnlp-main.96.pdf | A Large-Scale Dataset for Empathetic Response Generation | Recent development in NLP shows a strong trend towards refining pre-trained models with a domain-specific dataset. This is especially the case for response generation where emotion plays an important role. However, existing empathetic datasets remain small, delaying research efforts in this area, for example, the development of emotion-aware chatbots. One main technical challenge has been the cost of manually annotating dialogues with the right emotion labels. In this paper, we describe a large-scale silver dataset consisting of 1M dialogues annotated with 32 fine-grained emotions, eight empathetic response intents, and the Neutral category. To achieve this goal, we have developed a novel data curation pipeline starting with a small seed of manually annotated data and eventually scaling it to a satisfactory size. We compare its quality against a state-of-the-art gold dataset using both offline experiments and visual validation methods. The resultant procedure can be used to create similar datasets in the same domain as well as in other domains. | ['Pearl Pu', 'Yubo Xie', 'Anuradha Welivita'] | null | null | null | null | emnlp-2021-11 | ['empathetic-response-generation'] | ['natural-language-processing'] | [ 1.08360060e-01 5.40521324e-01 3.35170925e-01 -6.97503805e-01
-7.83101618e-01 -7.95723081e-01 7.22310901e-01 2.18553722e-01
-5.98332763e-01 1.14000285e+00 5.37315011e-01 1.13750309e-01
1.62451029e-01 -3.58499259e-01 4.28146832e-02 -4.01060492e-01
3.98556620e-01 1.15772069e+00 4.16116975e-02 -7.05666542e-01
4.63116407e-01 2.30846241e-01 -1.20839667e+00 6.60272598e-01
6.24729335e-01 7.89526522e-01 -1.16447665e-01 5.58756053e-01
-2.41434202e-01 1.07675123e+00 -9.77975905e-01 -9.12246764e-01
-4.27327007e-02 -7.34062016e-01 -1.51919258e+00 -1.61642790e-01
-2.26195469e-01 2.88686645e-03 4.19064641e-01 6.04216039e-01
6.89613879e-01 3.84084195e-01 3.45256597e-01 -1.35719681e+00
-3.44519049e-01 8.16239536e-01 -1.04473561e-01 -2.88475901e-01
5.71481764e-01 1.08263023e-01 1.09861791e+00 -5.28275669e-01
1.16110313e+00 1.09868860e+00 8.22166085e-01 1.05964994e+00
-1.21103632e+00 -2.99665868e-01 -3.29330653e-01 1.81221172e-01
-8.06309044e-01 -4.55514371e-01 9.03003395e-01 -4.37770516e-01
1.07523715e+00 1.76783264e-01 4.23871368e-01 1.58443928e+00
-4.98024046e-01 6.51110828e-01 1.50373709e+00 -6.68050945e-01
3.12552541e-01 7.14471161e-01 -1.05005614e-01 1.31999999e-01
-5.24408400e-01 -5.54221749e-01 -5.49022853e-01 -2.49771029e-01
1.73756033e-01 -6.92556560e-01 -1.20993212e-01 -1.05008416e-01
-1.07925498e+00 1.14906228e+00 1.05866894e-01 6.57504201e-01
-3.93031508e-01 -3.73763025e-01 8.49145234e-01 5.32287955e-01
6.46458268e-01 1.18769252e+00 -6.27357244e-01 -9.45425332e-01
-7.29362667e-01 3.54044348e-01 1.41235650e+00 5.92917144e-01
5.19384921e-01 -4.55027521e-01 -8.42291862e-02 1.25184822e+00
-7.42876306e-02 -2.76553094e-01 3.48025143e-01 -1.26480782e+00
3.67002308e-01 8.61844122e-01 5.02714992e-01 -7.03184545e-01
-6.45681322e-01 5.11936732e-02 -5.20736754e-01 2.39229366e-01
7.99597442e-01 -5.54403305e-01 -6.05750009e-02 1.78945136e+00
5.33803761e-01 -5.41148484e-01 4.88802582e-01 1.02983999e+00
8.23661268e-01 6.99782670e-01 1.21857144e-01 -2.44135886e-01
1.27114844e+00 -1.07076502e+00 -6.16815448e-01 -1.27022266e-01
7.95142949e-01 -9.64784026e-01 1.35820854e+00 6.70324862e-01
-1.01318574e+00 -2.19311669e-01 -6.25422955e-01 -2.08891630e-01
-4.52530712e-01 2.75794029e-01 7.08077610e-01 5.06858468e-01
-1.00692272e+00 4.46196318e-01 -2.78106332e-01 -8.38566303e-01
-6.92472979e-02 1.83656365e-01 -6.54153824e-01 1.86732814e-01
-1.38276780e+00 1.44869149e+00 2.40022540e-01 2.19160877e-03
-3.76504779e-01 -5.63751221e-01 -6.00269020e-01 -1.65163100e-01
1.61286756e-01 -1.57406226e-01 1.60312200e+00 -1.32691205e+00
-1.94855332e+00 1.16156709e+00 2.13457048e-01 -4.97252166e-01
7.55927563e-01 -1.55483216e-01 -1.60143480e-01 4.08580489e-02
1.06472909e-01 9.01927352e-01 3.88832867e-01 -1.13387561e+00
-5.59016407e-01 -4.58740778e-02 2.41699472e-01 2.14319408e-01
-4.63946879e-01 6.27876222e-01 5.67723885e-02 -2.16614529e-01
-6.29182935e-01 -9.73701298e-01 -3.02427351e-01 -3.99851292e-01
-8.14848468e-02 -5.16206622e-01 5.92745602e-01 -6.20069802e-01
8.87830615e-01 -1.94852746e+00 1.54839680e-01 -8.84001777e-02
-3.06155235e-02 2.07784787e-01 -9.61606503e-02 8.35600495e-01
9.40784439e-02 -3.30873728e-02 -9.30378959e-02 -6.24609947e-01
2.06276238e-01 1.25623077e-01 -8.81204456e-02 9.76329893e-02
4.64345336e-01 6.07107520e-01 -1.05756462e+00 -6.75124884e-01
-2.39214860e-02 3.78359914e-01 -5.86972952e-01 5.29401243e-01
-2.91308373e-01 8.16124439e-01 -2.59295493e-01 2.16337457e-01
1.66412145e-01 -4.31771800e-02 3.04560155e-01 6.30187914e-02
-2.34540775e-01 5.19622505e-01 -7.93917298e-01 1.79733706e+00
-7.40018606e-01 7.69629359e-01 3.41562659e-01 -8.40242922e-01
1.41962409e+00 6.06392086e-01 6.79174483e-01 -4.03546929e-01
3.81355703e-01 3.59456807e-01 1.89937428e-01 -6.47414863e-01
8.09766650e-01 -4.87089515e-01 -5.10233283e-01 8.79679382e-01
4.20021832e-01 -3.48113954e-01 3.19334596e-01 2.29254976e-01
1.07400680e+00 2.42995977e-01 2.99324155e-01 -1.72453314e-01
5.92931092e-01 5.62941074e-01 4.01845157e-01 4.26830828e-01
-4.38401610e-01 5.31481802e-01 9.20739770e-01 -4.97196615e-01
-1.07664895e+00 -1.85645178e-01 5.25531694e-02 1.41974139e+00
-4.06540394e-01 -4.04903203e-01 -9.11118805e-01 -7.69048035e-01
-6.46048665e-01 7.71030605e-01 -8.19929838e-01 1.06720656e-01
-4.23035473e-01 -6.01404786e-01 8.85288000e-01 1.44217297e-01
3.23474646e-01 -1.70117211e+00 -9.46775556e-01 5.22745192e-01
-6.72211409e-01 -1.10155308e+00 2.24450871e-01 3.85334969e-01
-1.99360028e-01 -8.96625340e-01 -5.53685844e-01 -6.20939910e-01
2.69553035e-01 -4.36908513e-01 1.41370475e+00 -9.90133509e-02
-3.89240906e-02 1.95970073e-01 -9.50404406e-01 -5.35468519e-01
-8.96192551e-01 4.35491353e-01 -2.87714660e-01 -9.41616073e-02
5.26992738e-01 -4.34077352e-01 -3.11519116e-01 3.47600043e-01
-6.91897154e-01 7.59392604e-02 2.83216417e-01 9.53922451e-01
-1.26251876e-01 -4.52288181e-01 1.00919378e+00 -1.02528882e+00
1.29483461e+00 -4.61092770e-01 -5.11030108e-02 1.68234676e-01
-2.68221349e-01 -1.33573294e-01 7.68161774e-01 -4.22649920e-01
-1.40500188e+00 2.23204300e-01 -4.79958177e-01 1.76045686e-01
-4.63607937e-01 4.72595364e-01 1.42968483e-02 2.36518234e-01
1.01929569e+00 -4.43697512e-01 1.28857359e-01 -4.87011433e-01
4.93192941e-01 1.06425822e+00 5.01882493e-01 -8.45653713e-01
1.11800775e-01 9.30773839e-02 -3.65258873e-01 -5.62638938e-01
-9.21618938e-01 -5.95769644e-01 -7.14817405e-01 -4.67963666e-01
7.10981965e-01 -5.59316993e-01 -5.75320005e-01 2.05640525e-01
-1.38701844e+00 -8.10448706e-01 -3.46053720e-01 2.63200611e-01
-6.74170434e-01 1.32909194e-01 -7.77825415e-01 -8.41496825e-01
-6.11705542e-01 -7.91201949e-01 8.28977704e-01 2.42902875e-01
-1.14683986e+00 -9.60121870e-01 6.45946503e-01 6.91999614e-01
4.53839511e-01 3.23847711e-01 6.77595198e-01 -1.15660763e+00
4.41008180e-01 -3.81679803e-01 -6.67211190e-02 2.27011070e-01
-2.73458868e-01 1.18306965e-01 -9.87807691e-01 2.76710004e-01
6.19091280e-02 -1.33589911e+00 1.97740093e-01 -2.85588354e-01
3.69058967e-01 -2.38774925e-01 1.74632624e-01 -3.01820546e-01
8.02977562e-01 -4.87486497e-02 5.23556113e-01 5.99246442e-01
1.33732215e-01 1.25241733e+00 1.13503468e+00 6.75962865e-01
4.54550594e-01 7.58463740e-01 1.61379278e-01 -8.06143805e-02
3.06958824e-01 2.14186870e-02 2.42420629e-01 4.29380983e-01
-1.11630790e-01 -1.69723257e-01 -1.05159748e+00 7.84849286e-01
-2.05389357e+00 -9.28360820e-01 -2.22068414e-01 1.63813746e+00
1.36581445e+00 -1.30706429e-01 3.34687918e-01 1.54190227e-01
4.65035886e-01 -7.40616322e-02 -7.44645894e-02 -1.05104673e+00
-6.78214356e-02 3.08478653e-01 -4.53285217e-01 5.19219220e-01
-8.94437790e-01 1.23453271e+00 5.47871494e+00 5.01909256e-01
-1.06403589e+00 2.58618861e-01 6.75519824e-01 -7.21539855e-02
4.54487558e-03 1.02851607e-01 -4.72729951e-01 1.46118417e-01
1.15236354e+00 2.05491453e-01 3.16664845e-01 1.14974928e+00
3.40499073e-01 -1.62430987e-01 -9.96714592e-01 6.79499209e-01
1.03143647e-01 -1.12904143e+00 -6.27679527e-01 -3.03388804e-01
4.88534540e-01 -3.35781053e-02 -6.72952771e-01 7.06348121e-01
4.88529414e-01 -8.98958623e-01 5.57589650e-01 1.67640209e-01
5.28647840e-01 -5.10053039e-01 1.12035823e+00 4.79124635e-01
-3.67581636e-01 2.42416665e-01 -1.60502031e-01 -3.85426700e-01
4.04764473e-01 2.10063577e-01 -1.18185031e+00 9.32380930e-02
7.79254615e-01 4.87706006e-01 -5.17203510e-01 5.60344517e-01
-3.40356112e-01 5.40083289e-01 -3.28748614e-01 -4.03281301e-01
4.59079385e-01 -2.18301877e-01 2.27476835e-01 1.51161444e+00
-7.65816867e-02 1.27315134e-01 6.31599426e-02 7.72662818e-01
-1.64499298e-01 4.34129596e-01 -3.67465734e-01 -8.28903392e-02
3.80788952e-01 2.05578399e+00 -7.17513382e-01 -1.85009167e-01
-2.49854460e-01 1.07463384e+00 7.34590411e-01 -1.01861291e-01
-7.26000369e-01 -4.41109270e-01 3.52206856e-01 -2.55261958e-01
-2.09873736e-01 2.06167012e-01 -2.64044195e-01 -8.70881379e-01
-2.01009035e-01 -1.21969235e+00 4.77056623e-01 -8.48814368e-01
-1.50117505e+00 1.02664185e+00 -2.20659748e-01 -8.42482686e-01
-7.09735692e-01 -4.94966418e-01 -6.21866167e-01 8.73163939e-01
-9.95286942e-01 -1.26136553e+00 -4.88271713e-01 5.31820834e-01
5.14292181e-01 2.78644599e-02 1.33170807e+00 2.16114268e-01
-3.93368691e-01 3.18360031e-01 -3.78105581e-01 1.61465675e-01
1.34088433e+00 -1.25469959e+00 2.92681367e-03 3.42535019e-01
3.36189270e-02 3.47092450e-01 1.03069663e+00 -2.60729492e-01
-7.06326962e-01 -6.56478047e-01 1.19960213e+00 -7.80971110e-01
8.49731684e-01 -5.98706007e-01 -9.03212965e-01 4.58892465e-01
6.39975667e-01 -3.83582741e-01 9.18534458e-01 4.53258753e-01
-1.91542536e-01 3.57279867e-01 -1.38375449e+00 4.01068628e-01
6.07060134e-01 -3.75110537e-01 -7.07114816e-01 3.44419628e-01
2.44659871e-01 -4.01092976e-01 -1.12660015e+00 1.02089718e-01
5.35060167e-01 -1.10538816e+00 2.89074421e-01 -8.92732620e-01
8.58822584e-01 1.08581915e-01 1.72747537e-01 -1.54527712e+00
1.84395194e-01 -9.32360709e-01 3.05462360e-01 1.82950377e+00
6.77877128e-01 -2.79594839e-01 7.13920951e-01 1.02375031e+00
-8.22115988e-02 -6.17264748e-01 -8.41423988e-01 -2.27145776e-01
1.56457841e-01 -3.04420203e-01 3.24672759e-01 1.17660451e+00
8.78560066e-01 1.02290547e+00 -6.09578073e-01 -6.69565737e-01
-1.62568852e-01 1.86834767e-01 1.19841814e+00 -1.36062014e+00
-9.26558226e-02 -4.95338798e-01 -3.00720166e-02 -4.32743043e-01
4.39115822e-01 -5.01282692e-01 4.06997979e-01 -1.52386236e+00
1.03406467e-01 -5.24866760e-01 1.93122834e-01 7.21869707e-01
-6.68657646e-02 4.01252478e-01 1.74622666e-02 5.85189201e-02
-7.87427127e-01 5.18747926e-01 8.06513667e-01 3.43705475e-01
-5.07409453e-01 -2.65515685e-01 -8.00294399e-01 7.61086345e-01
1.30393362e+00 -5.28588235e-01 -2.88318425e-01 3.99840809e-02
4.60622042e-01 -1.25248685e-01 1.51731819e-01 -7.42977858e-01
-6.59716204e-02 -2.12685972e-01 -6.62912801e-02 -1.41057953e-01
5.67212045e-01 -6.34599030e-01 -4.33042571e-02 -4.79315408e-02
-8.34318757e-01 -1.13278516e-01 8.35582241e-02 -1.47847440e-02
-3.87123078e-01 -6.10701501e-01 8.45230043e-01 -3.37409675e-01
-6.03702843e-01 -4.52755958e-01 -5.18830240e-01 3.67842227e-01
9.94692743e-01 7.07034320e-02 -4.66799080e-01 -8.32813680e-01
-6.17233157e-01 1.10426627e-01 5.84291756e-01 6.74540699e-01
1.35157071e-02 -8.85717690e-01 -8.74620974e-01 -4.01604861e-01
3.11114579e-01 -2.22763792e-01 9.03896391e-02 9.83862519e-01
-4.17561203e-01 1.31353498e-01 -5.78899026e-01 -2.40142718e-01
-1.33243859e+00 2.99664855e-01 2.52035946e-01 -6.63711548e-01
-3.76905143e-01 6.71510100e-01 -4.73758906e-01 -8.16891730e-01
3.11528668e-02 1.23863645e-01 -6.00974083e-01 4.06454772e-01
4.82557356e-01 1.43343046e-01 2.66095456e-02 -7.39494681e-01
-2.50349551e-01 -1.70438215e-01 1.23484589e-01 -5.97031295e-01
1.58662462e+00 -8.37860852e-02 -3.65670323e-01 5.76317310e-01
9.07718122e-01 1.50460660e-01 -1.06686461e+00 7.60651156e-02
3.58296275e-01 -1.20226607e-01 -4.37812924e-01 -1.12896264e+00
-4.48990732e-01 8.89007688e-01 -1.98124535e-02 5.58475137e-01
8.33167076e-01 1.51995569e-01 6.22977674e-01 4.14458960e-01
4.02768046e-01 -1.49653935e+00 3.08570564e-01 8.16856742e-01
1.05315971e+00 -1.44050586e+00 -2.61075974e-01 -1.63855687e-01
-1.39965284e+00 1.12075186e+00 8.12667489e-01 9.10502896e-02
4.01583686e-02 1.59676403e-01 6.76203907e-01 -3.34640861e-01
-9.90145802e-01 -1.60892174e-01 -8.70919824e-02 5.93248904e-01
9.18936312e-01 -6.97146803e-02 -6.10688150e-01 1.04405105e+00
-5.16833961e-01 6.66009262e-02 7.14746535e-01 8.10405910e-01
-1.03991568e-01 -1.40217257e+00 -1.68491796e-01 8.20327774e-02
-6.04805112e-01 2.84750685e-02 -1.32974303e+00 8.66361916e-01
-1.44970387e-01 1.35750639e+00 -1.76179782e-01 -3.16523343e-01
3.36883664e-01 4.68694210e-01 1.78255558e-01 -7.11283326e-01
-1.26887381e+00 -2.43810847e-01 1.00143790e+00 -3.31106275e-01
-8.26579928e-01 -7.97222435e-01 -1.08259928e+00 -2.20602840e-01
-1.43912792e-01 6.99886560e-01 6.79150105e-01 9.67055917e-01
3.50104123e-01 1.26272887e-02 5.07964075e-01 -8.82694304e-01
-4.37016547e-01 -1.58133543e+00 -1.03181027e-01 8.64414275e-01
-3.87626827e-01 -4.20458615e-01 -1.75409302e-01 1.18980519e-01] | [12.8829927444458, 6.925634860992432] |
b864868d-4802-4449-8c9c-3bacc3a9a4c3 | mixed-spectrum-signals-discrete | 2106.14696 | null | https://arxiv.org/abs/2106.14696v2 | https://arxiv.org/pdf/2106.14696v2.pdf | Mixed-Spectrum Signals -- Discrete Approximations and Variance Expressions for Covariance Estimates | The estimation of the covariance function of a stochastic process, or signal, is of integral importance for a multitude of signal processing applications. In this work, we derive closed-form expressions for the variance of covariance estimates for mixed-spectrum signals, i.e., spectra containing both absolutely continuous and singular parts. The results cover both finite-sample and asymptotic regimes, allowing for assessing the exact speed of convergence of estimates to their expectations, as well as their limiting behavior. As is shown, such covariance estimates may converge even for non-ergodic processes. Furthermore, we consider approximating signals with arbitrary spectral densities by sequences of singular spectrum, i.e., sinusoidal, processes, and derive the limiting behavior of covariance estimates as both the sample size and the number of sinusoidal components tend to infinity. We show that the asymptotic regime variance can be described by a time-frequency resolution product, with dramatically different behavior depending on how the sinusoidal approximation is constructed. In a few numerical examples we illustrate the theory and the corresponding implications for direction of arrival estimation. | ['Johan Karlsson', 'Filip Elvander'] | 2021-06-28 | null | null | null | null | ['direction-of-arrival-estimation'] | ['audio'] | [ 3.79931957e-01 -3.21557075e-01 2.87842512e-01 2.77238023e-02
-9.59147573e-01 -7.92208135e-01 4.57078218e-01 -6.68285191e-02
-1.90394834e-01 9.09575701e-01 1.93464980e-02 -1.60909459e-01
-3.99156481e-01 -3.42854857e-01 -3.35326999e-01 -1.00853074e+00
-4.17470217e-01 4.37384658e-02 -2.48785652e-02 5.82547672e-02
-3.52691486e-02 5.00548422e-01 -1.32435632e+00 -6.06770813e-01
6.28295124e-01 1.19690847e+00 -1.30440682e-01 1.03927815e+00
1.26661375e-01 1.83719769e-01 -9.71695542e-01 -1.80473566e-01
1.69773072e-01 -5.40230572e-01 -9.00818780e-02 1.79761901e-01
-2.41888642e-01 1.06008366e-01 1.20489746e-02 1.16183496e+00
5.98989129e-01 1.07807867e-01 1.18154764e+00 -1.10025167e+00
-3.40354919e-01 3.49628121e-01 -6.43288314e-01 5.94906032e-01
2.53802627e-01 -1.80651203e-01 5.52743733e-01 -7.49526262e-01
7.47849094e-03 9.67246711e-01 1.04028678e+00 8.44479501e-02
-1.42309380e+00 -4.16137874e-01 -4.93655324e-01 -2.16931313e-01
-1.54101539e+00 -6.96754217e-01 6.22228861e-01 -6.41853333e-01
2.42788285e-01 1.97031334e-01 3.90501469e-01 8.38066161e-01
2.81503528e-01 2.53106654e-01 9.20700371e-01 -6.46976292e-01
3.53145301e-01 -1.28200427e-02 3.04842740e-01 3.16040337e-01
4.87351835e-01 -8.86356160e-02 -1.66114002e-01 -5.96043348e-01
7.67104745e-01 -3.73525500e-01 -6.51345551e-01 -2.69702911e-01
-1.07874668e+00 5.35346389e-01 -3.95697653e-01 4.56980228e-01
-6.68365777e-01 4.18042421e-01 1.69646993e-01 2.61394501e-01
5.85451663e-01 1.78967208e-01 -8.30843076e-02 -2.85129547e-01
-7.69090056e-01 1.13110572e-01 1.00261700e+00 9.97877061e-01
2.54857421e-01 4.45873618e-01 -1.27907321e-01 9.56870377e-01
3.04114316e-02 1.12854612e+00 4.57550362e-02 -8.24805140e-01
1.87902421e-01 -5.06503344e-01 6.79106593e-01 -4.91005957e-01
-5.27771950e-01 -7.84713030e-01 -8.77935886e-01 -1.44470111e-01
1.12340653e+00 -5.77998698e-01 -2.75730759e-01 1.88743174e+00
1.27462164e-01 8.04821551e-02 2.48923600e-01 6.50371015e-01
-1.17731683e-01 6.81767642e-01 -1.19116850e-01 -8.46854508e-01
1.42122078e+00 -5.57433031e-02 -1.09571230e+00 1.70779675e-01
4.02222127e-02 -1.10552526e+00 7.33764350e-01 4.10124183e-01
-1.26339281e+00 -2.25430995e-01 -8.80713165e-01 5.79153538e-01
2.41316870e-01 2.29977677e-03 -6.43798858e-02 8.70538354e-01
-8.94256949e-01 6.84207737e-01 -5.92291176e-01 -1.67105168e-01
-1.08619541e-01 -2.12149471e-01 2.29111403e-01 2.63646930e-01
-1.07000983e+00 5.21933258e-01 -1.31663665e-01 1.64972648e-01
-4.14627045e-01 -8.82978380e-01 -5.00768006e-01 2.92882472e-01
-1.08318731e-01 -4.93171245e-01 1.62896705e+00 -1.06533349e+00
-1.18587828e+00 2.21594781e-01 -3.56982023e-01 -2.97046483e-01
4.69501466e-01 2.92960797e-02 -9.49826121e-01 4.42193627e-01
1.33218870e-01 -5.41361332e-01 1.44303119e+00 -8.80852282e-01
-2.84596384e-01 -9.62397978e-02 -4.12283629e-01 3.85404229e-02
-1.75681394e-02 4.47139284e-03 2.33109206e-01 -6.80516005e-01
1.36418983e-01 -8.44887018e-01 -2.82360673e-01 -7.16155991e-02
-2.64139920e-01 2.16098949e-01 3.56163681e-01 -5.17404020e-01
1.22624075e+00 -2.51193213e+00 -3.44439685e-01 4.47499514e-01
-1.63400337e-01 -2.27894634e-01 7.46654645e-02 7.68342793e-01
-2.24554211e-01 1.35355545e-02 -3.92816126e-01 -5.84668666e-03
1.89219527e-02 -2.80769140e-01 -4.78451788e-01 9.51314449e-01
3.65266234e-01 1.90085381e-01 -8.40611875e-01 -6.44778982e-02
-9.39237773e-02 5.32951474e-01 -1.35016769e-01 -1.71533063e-01
3.22373062e-01 4.02305126e-01 -4.46179032e-01 2.10658222e-01
5.96709669e-01 -4.06657308e-01 3.34724113e-02 -1.04314178e-01
-4.13903773e-01 -1.54684752e-01 -1.32543337e+00 8.27157378e-01
-7.35638261e-01 9.95495260e-01 4.85759854e-01 -8.32151532e-01
6.99536145e-01 6.77218974e-01 5.36471725e-01 3.86336111e-02
1.20294727e-01 5.89413404e-01 2.29471400e-01 -2.38796830e-01
2.73665518e-01 -8.77176762e-01 -2.76171020e-03 4.15200382e-01
-1.46025270e-01 -2.76793204e-02 3.19813490e-01 -7.84891695e-02
9.21261132e-01 -4.16853696e-01 8.11554492e-01 -7.86847591e-01
6.54801488e-01 -4.39509958e-01 5.07594049e-02 6.35532439e-01
-1.88058034e-01 4.61718321e-01 7.14167416e-01 2.00233296e-01
-1.16132212e+00 -1.31269324e+00 -7.18229175e-01 3.68452877e-01
3.20940763e-02 8.81604701e-02 -6.83878899e-01 3.77420306e-01
1.17942825e-01 9.19210255e-01 -2.03891248e-01 -1.77975923e-01
-4.44701135e-01 -9.56667304e-01 4.78648484e-01 2.52114028e-01
7.67797902e-02 -4.70800817e-01 -4.82217848e-01 3.99018437e-01
-1.95257422e-02 -1.38458836e+00 -6.82536662e-01 1.89472333e-01
-7.04343259e-01 -9.84180570e-01 -1.10485530e+00 -2.36446619e-01
3.55286241e-01 2.39491850e-01 7.32246816e-01 -5.77460527e-01
-3.20939124e-02 1.08961868e+00 1.15792319e-01 -5.41449368e-01
-5.56939363e-01 -5.40657997e-01 3.20140213e-01 3.92985463e-01
-1.27152324e-01 -7.87278831e-01 -5.10005355e-01 4.98825014e-01
-7.48306870e-01 -6.79635465e-01 3.28280896e-01 6.09538496e-01
2.97119349e-01 2.89157510e-01 8.19091678e-01 -9.26261023e-02
1.19617522e+00 -5.78277111e-01 -6.96921051e-01 -2.98860371e-02
-1.85840458e-01 1.94260374e-01 8.57726634e-01 -6.63026154e-01
-1.05201054e+00 -4.28480357e-01 1.91327736e-01 -4.79699165e-01
1.59413312e-02 4.01515692e-01 2.13299558e-01 1.25142513e-04
7.51933098e-01 4.96549696e-01 7.59778246e-02 -3.93190354e-01
2.27012664e-01 6.62709475e-01 7.03812182e-01 -6.58502936e-01
8.64176750e-01 5.03320456e-01 4.76603627e-01 -1.60982096e+00
-7.48372734e-01 -7.03470051e-01 -2.28918999e-01 -2.17194915e-01
5.31686306e-01 -5.70273280e-01 -8.50265801e-01 4.13083792e-01
-1.23498881e+00 -7.43188560e-02 -5.50053298e-01 9.27052379e-01
-9.79996741e-01 3.37980002e-01 -6.15798593e-01 -1.70298517e+00
2.47776858e-03 -6.23613060e-01 8.70735705e-01 2.25154757e-02
-2.37009034e-01 -1.22400510e+00 1.76301271e-01 -4.73034173e-01
4.62201804e-01 1.62606686e-01 8.18378508e-01 -5.38148165e-01
-1.53564677e-01 -4.15102363e-01 -5.96246086e-02 2.70784795e-01
1.32413507e-01 -8.42137709e-02 -8.02308738e-01 -3.19840014e-02
5.58550894e-01 4.01837051e-01 2.21983224e-01 1.08652806e+00
4.30876046e-01 -3.05581152e-01 -3.06822270e-01 3.74068916e-01
1.51844513e+00 2.88579166e-01 3.37519705e-01 -1.25903443e-01
5.83254509e-02 5.56759417e-01 3.12475353e-01 6.92117572e-01
-5.20944178e-01 4.18224156e-01 -2.36240610e-01 4.07865465e-01
1.13179587e-01 1.17460027e-01 2.46823519e-01 8.75351071e-01
-1.02966279e-01 -2.08834007e-01 -7.67327309e-01 4.41592008e-01
-1.28998208e+00 -1.25580549e+00 -4.21228111e-01 2.61219454e+00
4.90654200e-01 1.71706025e-02 2.16756254e-01 2.89130867e-01
1.12289584e+00 3.49805644e-03 -4.68310326e-01 -2.27132604e-01
-1.76398203e-01 1.60057887e-01 8.41541708e-01 4.97493356e-01
-8.45652163e-01 -1.53910100e-01 7.57556868e+00 9.01783764e-01
-8.80691230e-01 1.98569849e-01 2.91931123e-01 1.00827515e-01
-1.50554821e-01 -2.14881003e-01 -4.88699883e-01 6.38130844e-01
1.31606102e+00 -7.72963285e-01 3.25888634e-01 4.58738416e-01
3.51379335e-01 -2.19553128e-01 -6.72353685e-01 1.03304970e+00
-3.47113013e-01 -6.60428166e-01 -6.09521210e-01 2.12647796e-01
5.01429260e-01 -4.34765220e-01 3.47564891e-02 -2.70979311e-02
-1.26399666e-01 -6.18483424e-01 8.82894874e-01 7.08045959e-01
7.69250512e-01 -7.88209081e-01 5.67875683e-01 4.81367797e-01
-1.37264204e+00 -1.43338084e-01 -1.74749315e-01 -1.10030882e-01
6.88163996e-01 1.07608283e+00 -4.76272136e-01 5.52497447e-01
1.44323230e-01 2.50431061e-01 2.27070242e-01 1.34345770e+00
2.25131884e-01 8.23160648e-01 -7.75861740e-01 -3.33691120e-01
6.72623888e-02 -7.05733538e-01 1.04395342e+00 1.17959428e+00
1.01552427e+00 4.04013485e-01 -3.08140129e-01 6.44839048e-01
4.06921893e-01 7.44419023e-02 -4.59019244e-01 -2.84768939e-01
5.77255309e-01 1.06103468e+00 -8.87676716e-01 -9.63378549e-02
-5.14244676e-01 4.25687671e-01 -3.16803455e-01 7.79424489e-01
-6.06372952e-01 -6.36233211e-01 7.40290642e-01 2.84486294e-01
3.89068455e-01 -5.66098690e-01 -5.93780354e-02 -8.64416778e-01
1.82795063e-01 -3.28235000e-01 3.56616154e-02 -6.26025677e-01
-1.30447531e+00 3.18817914e-01 3.58391762e-01 -1.51377916e+00
-5.12926459e-01 -4.27940160e-01 -6.20737731e-01 1.08383250e+00
-1.02323699e+00 -2.47231379e-01 3.35276127e-01 3.36215794e-01
1.45920739e-01 4.65757726e-03 4.95414376e-01 1.55562058e-01
-2.82090634e-01 5.86039498e-02 7.95925319e-01 -2.19826669e-01
1.69391051e-01 -1.10646605e+00 2.18455330e-01 6.54068649e-01
-1.01768732e-01 6.57549322e-01 1.45219243e+00 -3.24669033e-01
-1.25220275e+00 -5.71495414e-01 4.48452681e-01 -1.09737158e-01
1.20108342e+00 -1.79176584e-01 -7.80196548e-01 3.14671040e-01
-7.77655169e-02 3.44382077e-02 7.53986776e-01 -6.28084540e-02
1.40036151e-01 1.27729237e-01 -9.82020378e-01 6.81263924e-01
7.24854827e-01 -3.47113281e-01 -2.97975212e-01 2.48331368e-01
2.43931457e-01 -4.80304137e-02 -8.20260346e-01 2.23873109e-01
6.61846757e-01 -9.74672139e-01 8.47004890e-01 -2.63560653e-01
-1.39084354e-01 -3.32485735e-01 -4.48655456e-01 -1.45787036e+00
-3.06120843e-01 -1.07253814e+00 -5.78986481e-02 9.39112127e-01
3.50434631e-01 -8.79382193e-01 1.90578446e-01 6.87815174e-02
9.03964192e-02 -6.07328892e-01 -1.04973149e+00 -1.33793044e+00
-1.75621510e-01 -6.34243727e-01 1.18488654e-01 5.04560411e-01
1.71548828e-01 3.83470446e-01 -1.37384713e-01 4.67482924e-01
1.05565441e+00 -3.37035283e-02 3.33664149e-01 -1.25974500e+00
-3.98331583e-01 -5.28041780e-01 -4.66366082e-01 -1.27773273e+00
-1.11420505e-01 -4.17783976e-01 1.52788103e-01 -9.60869610e-01
-2.82828093e-01 -2.81505555e-01 1.67743154e-02 -6.73407197e-01
3.32824774e-02 -5.70980348e-02 -1.81535482e-02 3.44594538e-01
3.84588167e-02 5.80967367e-01 9.81229782e-01 3.80145222e-01
-1.06660262e-01 7.26912796e-01 -3.72838765e-01 9.95976031e-01
7.27229536e-01 -4.42285448e-01 -5.76308787e-01 3.36405098e-01
2.13065431e-01 6.67206824e-01 3.71375382e-01 -1.21207583e+00
5.28729856e-02 1.02376118e-01 2.53472805e-01 -4.14034367e-01
4.36483830e-01 -6.27906561e-01 4.56501067e-01 3.32525373e-01
-3.56160104e-01 -1.01049423e-01 1.83694810e-01 1.11907005e+00
-1.39798477e-01 -6.60029888e-01 9.32294846e-01 2.50101715e-01
6.06373027e-02 -2.40542199e-02 -8.05583775e-01 2.36163840e-01
9.08489525e-01 -1.25357389e-01 -4.15734425e-02 -1.07330596e+00
-8.07216704e-01 -2.25065753e-01 1.55922091e-02 -2.53366500e-01
2.71770149e-01 -1.44691420e+00 -5.95830202e-01 4.63871993e-02
-2.41021588e-01 -6.32646441e-01 3.33863348e-01 1.24227417e+00
-1.78490952e-01 4.53619540e-01 2.99278617e-01 -6.91478431e-01
-8.83828759e-01 5.70321500e-01 4.78259921e-01 3.46797667e-02
-2.74603903e-01 2.08835945e-01 2.16585577e-01 2.20603272e-01
-8.95701423e-02 -3.81621718e-01 9.74039733e-02 2.41547376e-02
7.75387406e-01 5.52408636e-01 -1.58187509e-01 -4.45490450e-01
-1.10803194e-01 8.24349403e-01 6.35445595e-01 -6.46249115e-01
5.08275807e-01 -3.97380710e-01 -1.06575392e-01 1.08612561e+00
1.34896493e+00 4.99484867e-01 -1.21570873e+00 -1.90722838e-01
-4.23470736e-02 -1.08739339e-01 -3.39993656e-01 -1.03510350e-01
-5.23837209e-01 6.47762716e-01 4.84497219e-01 9.79284883e-01
1.11959898e+00 2.61269715e-02 4.04483974e-01 1.27947971e-01
3.45277429e-01 -5.38014710e-01 -2.07704067e-01 3.89536798e-01
8.28105986e-01 -5.27640045e-01 -4.47941646e-02 -6.25265062e-01
-1.95154428e-01 1.34057999e+00 -3.93483549e-01 -2.53215015e-01
1.06434655e+00 4.41031218e-01 -1.48547634e-01 2.06002846e-01
-4.94537354e-01 -1.00724675e-01 3.83734286e-01 7.55384803e-01
5.49008310e-01 2.13256657e-01 -5.13566256e-01 5.24015069e-01
-3.23393464e-01 -3.09321731e-01 8.42313766e-01 5.82589567e-01
-6.78654432e-01 -6.03979409e-01 -8.67158413e-01 5.10089874e-01
-7.37154007e-01 -2.47783497e-01 2.60293037e-01 7.39584565e-01
-7.10561037e-01 9.69433725e-01 2.93057293e-01 3.23331624e-01
5.10322988e-01 3.68640691e-01 5.07445395e-01 -1.09037228e-01
1.10046893e-01 4.47855860e-01 2.20850825e-01 2.63959039e-02
-2.40560636e-01 -1.01350045e+00 -8.07322741e-01 -2.14572325e-01
-3.61555576e-01 3.51316988e-01 6.38051808e-01 8.66278052e-01
-1.93595648e-01 4.39983398e-01 6.52107775e-01 -6.73804700e-01
-1.09268486e+00 -1.01282656e+00 -1.12078786e+00 6.51057437e-02
7.57231355e-01 -4.56388384e-01 -7.90045023e-01 -7.92352334e-02] | [6.703155994415283, 3.831418514251709] |
87d604d3-9dcf-4606-a1b8-37ab31bf383f | h2fa-r-cnn-holistic-and-hierarchical-feature | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Xu_H2FA_R-CNN_Holistic_and_Hierarchical_Feature_Alignment_for_Cross-Domain_Weakly_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Xu_H2FA_R-CNN_Holistic_and_Hierarchical_Feature_Alignment_for_Cross-Domain_Weakly_CVPR_2022_paper.pdf | H2FA R-CNN: Holistic and Hierarchical Feature Alignment for Cross-Domain Weakly Supervised Object Detection | Cross-domain weakly supervised object detection (CDWSOD) aims to adapt the detection model to a novel target domain with easily acquired image-level annotations. How to align the source and target domains is critical to the CDWSOD accuracy. Existing methods usually focus on partial detection components for domain alignment. In contrast, this paper considers that all the detection components are important and proposes a Holistic and Hierarchical Feature Alignment (H^2FA) R-CNN. H^2FA R-CNN enforces two image-level alignments for the backbone features, as well as two instance-level alignments for the RPN and detection head. This coarse-to-fine aligning hierarchy is in pace with the detection pipeline, i.e., processing the image-level feature and the instance-level features from bottom to top. Importantly, we devise a novel hybrid supervision method for learning two instance-level alignments. It enables the RPN and detection head to simultaneously receive weak/full supervision from the target/source domains. Combining all these feature alignments, H^2FA R-CNN effectively mitigates the gap between the source and target domains. Experimental results show that H^2FA R-CNN significantly improves cross-domain object detection accuracy and sets new state of the art on popular benchmarks. Code and pre-trained models are available at https://github.com/XuYunqiu/H2FA_R-CNN. | ['Yi Yang', 'Jiaxu Miao', 'Zongxin Yang', 'Yifan Sun', 'Yunqiu Xu'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['weakly-supervised-object-detection'] | ['computer-vision'] | [ 2.92206228e-01 6.03763983e-02 -5.04004657e-01 -2.44833678e-01
-1.09114385e+00 -6.11693203e-01 5.69242954e-01 -5.19094095e-02
-3.82287502e-01 3.31748188e-01 -1.47122860e-01 1.43614247e-01
7.08158836e-02 -6.10365033e-01 -7.43953288e-01 -6.09832227e-01
1.32175058e-01 4.27648276e-01 8.39095414e-01 -1.00563504e-01
-5.18355146e-02 5.36974072e-01 -1.63126278e+00 5.50898790e-01
3.47133875e-01 1.19459867e+00 4.84871924e-01 5.94643414e-01
2.74800248e-02 5.63383639e-01 -3.97152722e-01 -2.41916716e-01
8.10132742e-01 -2.55050570e-01 -6.95355833e-01 1.34268805e-01
7.80110061e-01 -4.19362068e-01 -2.79377878e-01 1.15393853e+00
5.40187836e-01 -2.35613450e-01 4.15348053e-01 -1.35445654e+00
-7.19151258e-01 2.99639553e-01 -1.03714228e+00 3.98322016e-01
5.76240867e-02 5.62734246e-01 1.17352664e+00 -1.15936553e+00
7.11916268e-01 1.01837707e+00 6.95819736e-01 5.43932378e-01
-1.34911692e+00 -8.29626441e-01 3.86174291e-01 4.99696471e-02
-1.47941208e+00 -2.45081246e-01 6.35328412e-01 -5.87194800e-01
9.47141588e-01 -1.30000919e-01 4.09275025e-01 1.08501899e+00
-2.35696927e-01 9.80522931e-01 1.00770032e+00 -3.62156004e-01
-1.38722464e-01 1.43375307e-01 1.66370615e-01 5.47700226e-01
4.69691485e-01 2.78765738e-01 -6.54724419e-01 9.70231816e-02
9.27531600e-01 -1.10798515e-01 4.74283174e-02 -6.59330845e-01
-1.35070777e+00 5.63279986e-01 5.76455712e-01 1.91938370e-01
-1.93520352e-01 -9.66384262e-02 6.28701806e-01 1.35661006e-01
3.24866146e-01 3.08310091e-01 -7.19494522e-01 3.39081109e-01
-8.22378874e-01 2.17584223e-01 3.26532423e-01 1.37162185e+00
8.22437584e-01 -2.42775112e-01 -5.03263891e-01 8.45961928e-01
1.38259411e-01 4.88337517e-01 2.14874268e-01 -8.64693105e-01
4.96439576e-01 8.42617452e-01 1.00950301e-01 -6.00598574e-01
-2.90373474e-01 -6.73591137e-01 -5.27670979e-01 3.01803023e-01
6.41963840e-01 1.11841418e-01 -8.67770851e-01 1.69325221e+00
5.84711909e-01 1.01990774e-01 -1.81714430e-01 9.71432447e-01
9.44293499e-01 2.56848902e-01 2.21978575e-02 1.81620643e-01
1.72916687e+00 -1.27707994e+00 -2.09504187e-01 -4.91631210e-01
5.56348622e-01 -7.59646058e-01 1.32362759e+00 1.51505709e-01
-1.07618451e+00 -8.72446358e-01 -9.86075521e-01 -2.34647870e-01
-4.50115800e-01 7.68540621e-01 1.11408971e-01 1.95583299e-01
-8.82638812e-01 1.30731285e-01 -6.00821674e-01 -5.20668387e-01
7.06795335e-01 4.52101052e-01 -6.12774134e-01 -1.19382180e-01
-1.04126322e+00 8.87463987e-01 4.80804861e-01 -1.74510896e-01
-1.13738728e+00 -8.10826480e-01 -7.90912092e-01 -1.25468671e-01
4.72428709e-01 -6.70801222e-01 1.42748773e+00 -1.05543220e+00
-1.18416464e+00 1.47496903e+00 5.25652431e-02 -4.83806342e-01
6.21165574e-01 -3.02958041e-01 -2.37324655e-01 1.78871214e-01
6.84921324e-01 9.42444146e-01 9.62062597e-01 -1.24017584e+00
-1.26981080e+00 -4.98438329e-01 6.67585656e-02 1.79584056e-01
-2.20623776e-01 2.79622108e-01 -7.98844695e-01 -6.99349284e-01
-5.01723140e-02 -5.40861785e-01 -6.53012842e-02 3.60873967e-01
-3.39682132e-01 -4.01299953e-01 8.99612248e-01 -3.63467962e-01
9.26148891e-01 -2.22511721e+00 -6.55386373e-02 -1.54186547e-01
3.74416947e-01 4.59466755e-01 -4.97597963e-01 2.81481855e-02
-2.91211009e-01 -3.76543194e-01 -8.30272213e-02 -4.65943366e-01
-3.33409049e-02 9.61054042e-02 -2.83957720e-01 6.52771413e-01
7.26365209e-01 9.09474850e-01 -9.48866069e-01 -4.92059439e-01
1.35982305e-01 2.53791183e-01 -4.39960033e-01 3.80766213e-01
-1.89352870e-01 3.17466408e-01 -2.92754829e-01 8.52429688e-01
8.42554927e-01 -4.88992512e-01 5.58010489e-02 -4.26675767e-01
-2.69723475e-01 3.43097597e-01 -1.21159518e+00 1.50568986e+00
-1.38967410e-01 4.64990139e-01 2.75264740e-01 -1.15787649e+00
9.61918592e-01 -8.23723972e-02 4.58474517e-01 -9.96244252e-01
8.12532082e-02 3.16956669e-01 -1.10208429e-01 -1.24358937e-01
2.55180180e-01 1.91724792e-01 -8.47852156e-02 1.38302848e-01
4.02563363e-01 1.14281140e-01 2.39381790e-01 2.76721925e-01
1.14690030e+00 4.21060950e-01 6.26738608e-01 -2.26492360e-01
6.06625021e-01 2.51063019e-01 8.97718966e-01 9.85938311e-01
-5.36991775e-01 9.07800496e-01 4.31442976e-01 -4.27824318e-01
-1.15804625e+00 -1.07158291e+00 -2.00206399e-01 1.58985353e+00
3.68011773e-01 -1.69338033e-01 -7.05129564e-01 -9.50540423e-01
2.26559222e-01 2.26925835e-01 -9.62659419e-01 -1.80155888e-01
-6.15631163e-01 -3.53926927e-01 7.57581651e-01 8.26719522e-01
5.75865626e-01 -9.69752431e-01 -5.74815869e-01 1.53129146e-01
-5.33379801e-02 -1.43531024e+00 -6.67830527e-01 4.93503571e-01
-5.25533557e-01 -1.16763091e+00 -6.42562211e-01 -8.53216469e-01
6.37171566e-01 6.81214154e-01 1.27956557e+00 1.38946921e-02
-5.67214072e-01 3.82737547e-01 -4.18105394e-01 -4.32374805e-01
-1.73633292e-01 1.48378730e-01 8.76261666e-02 -1.08977675e-01
5.53297997e-01 -3.19689333e-01 -6.11208022e-01 6.86297536e-01
-7.38179505e-01 -5.36272861e-02 8.17668140e-01 8.73653710e-01
9.83671546e-01 -3.66427422e-01 4.25501257e-01 -8.01165640e-01
-6.36845501e-03 -2.39685893e-01 -1.02170610e+00 1.73077211e-01
-3.85083497e-01 -2.61488080e-01 4.72962677e-01 -5.77835262e-01
-8.60294580e-01 5.02018571e-01 3.90817337e-02 -5.25496364e-01
-3.48575205e-01 -1.41745284e-01 -5.29844940e-01 2.07633153e-03
9.47803795e-01 2.07333878e-01 -1.71857998e-01 -4.85990882e-01
4.18636143e-01 5.29147506e-01 7.92898834e-01 -6.06955945e-01
1.05158877e+00 6.62504256e-01 -3.72299284e-01 -5.50565481e-01
-1.37570548e+00 -1.05985129e+00 -9.99431014e-01 -4.23838273e-02
8.23440194e-01 -1.21097589e+00 -3.14482570e-01 4.81711328e-01
-1.12218881e+00 -4.05755728e-01 -6.77653611e-01 1.33806229e-01
-5.53101540e-01 2.62666702e-01 -3.74061823e-01 -3.47574502e-01
-3.13889831e-01 -1.15657866e+00 1.51514053e+00 9.92763191e-02
-2.31501684e-02 -5.40410638e-01 4.64854389e-02 3.63342345e-01
1.03577167e-01 1.08156249e-01 3.22444052e-01 -8.82648408e-01
-6.01954639e-01 -1.57778129e-01 -8.07335496e-01 4.55012083e-01
6.10154942e-02 -1.88064247e-01 -1.05717587e+00 -2.87872016e-01
-2.68393397e-01 -2.87836999e-01 1.01849508e+00 3.11604917e-01
9.52769339e-01 -1.73932821e-01 -4.87451494e-01 5.43465912e-01
1.21899474e+00 -3.16115260e-01 4.36726809e-01 7.11256564e-01
7.30177522e-01 4.57024306e-01 1.10277367e+00 5.03078461e-01
4.25553828e-01 1.02105737e+00 5.90042770e-01 -3.66343021e-01
-6.05900347e-01 -1.64339170e-01 6.89475775e-01 1.10606126e-01
2.53711313e-01 2.17073962e-01 -9.60312665e-01 8.26538861e-01
-1.92328858e+00 -8.21736455e-01 -1.80216670e-01 2.00668716e+00
9.36235547e-01 3.83771032e-01 5.93272984e-01 -1.78764775e-01
1.03106344e+00 -1.44740954e-01 -7.67722249e-01 1.99152112e-01
-1.85637474e-01 1.67266995e-01 6.84817851e-01 2.67293394e-01
-1.35919809e+00 1.00442743e+00 5.31324482e+00 1.04242158e+00
-1.00152707e+00 3.08881998e-01 2.23756850e-01 -1.94630682e-01
3.00301880e-01 -1.10110760e-01 -1.44682002e+00 3.49195331e-01
3.09749663e-01 2.08448768e-01 7.98279643e-02 1.25339711e+00
-3.22837336e-03 1.01905420e-01 -1.23091829e+00 7.54980385e-01
-1.17765717e-01 -1.32863891e+00 2.09930427e-02 -3.61267179e-02
5.70691764e-01 4.90728050e-01 1.03378549e-01 3.90161008e-01
3.92763436e-01 -6.48987174e-01 9.71461952e-01 -2.96636354e-02
1.01284301e+00 -3.94585967e-01 6.06408060e-01 4.01308686e-01
-1.55026805e+00 -2.17256144e-01 -3.62854362e-01 2.79205024e-01
-2.38744915e-01 5.64106643e-01 -8.54647577e-01 4.99704033e-01
1.08882475e+00 9.72343028e-01 -6.69895530e-01 7.73006380e-01
-4.12164718e-01 3.37315977e-01 -2.53636897e-01 5.42215824e-01
1.08942956e-01 2.65325993e-01 7.47004867e-01 1.66383171e+00
-2.19900487e-03 -1.58412635e-01 3.61628860e-01 9.39598799e-01
-1.28619581e-01 -2.23677814e-01 -4.52903003e-01 2.46780187e-01
4.23614979e-01 1.56682634e+00 -7.08591759e-01 -3.88013095e-01
-5.66633344e-01 7.95767248e-01 6.24445796e-01 7.38212094e-02
-1.01235390e+00 -2.31003448e-01 9.04141188e-01 4.60260570e-01
7.11644292e-01 -8.56074542e-02 -2.31151462e-01 -1.23021221e+00
1.17597699e-01 -1.00414193e+00 7.58966804e-01 -4.90564495e-01
-1.61605322e+00 4.19092149e-01 -1.18562818e-01 -1.58350396e+00
2.81340420e-01 -9.14389193e-01 -4.40181255e-01 7.54415631e-01
-2.01974177e+00 -1.71742928e+00 -4.88816500e-01 8.50856900e-01
6.14678442e-01 -1.09918237e-01 5.40287256e-01 4.35710222e-01
-7.76510477e-01 8.43317449e-01 -1.46613434e-01 4.83613908e-01
1.03159440e+00 -1.27350950e+00 4.80058730e-01 1.02208996e+00
5.41126579e-02 5.83504021e-01 4.53564048e-01 -4.90801781e-01
-1.11219358e+00 -1.50182855e+00 4.82563853e-01 -6.33109629e-01
7.30482101e-01 -5.06006062e-01 -9.63319302e-01 6.99136376e-01
-2.04105861e-02 7.76170731e-01 5.25145233e-01 -7.61485472e-02
-8.75115216e-01 -3.39903355e-01 -1.15055227e+00 1.95722193e-01
1.28542602e+00 -6.14015758e-01 -6.04607642e-01 3.64400715e-01
7.17098355e-01 -5.19782186e-01 -7.23953307e-01 6.58251822e-01
4.82610404e-01 -9.63366270e-01 1.24093616e+00 -4.68375504e-01
2.88239658e-01 -8.41009855e-01 -1.77948982e-01 -6.45920277e-01
-5.84446013e-01 -1.61613569e-01 -2.85884827e-01 1.32891560e+00
3.02852362e-01 -5.54667771e-01 6.88178003e-01 -1.41660953e-02
-2.11185828e-01 -6.57653689e-01 -9.11709845e-01 -1.05003762e+00
1.76149956e-03 -3.28640938e-01 3.04965615e-01 9.53856051e-01
-4.41247553e-01 3.37491691e-01 -6.89768791e-02 5.69848239e-01
7.61424422e-01 3.89259249e-01 1.08084059e+00 -1.16319036e+00
-4.74924266e-01 -5.90953231e-01 -2.87745059e-01 -1.25035930e+00
-6.86794370e-02 -9.00684416e-01 1.58050179e-01 -1.35580671e+00
4.65057284e-01 -4.00424570e-01 -4.98148113e-01 8.80930424e-01
-2.51046509e-01 5.78951359e-01 1.79938957e-01 4.84553248e-01
-1.07535005e+00 3.68013710e-01 1.15071893e+00 -1.18850976e-01
-1.12135641e-01 -2.62594447e-02 -8.75687957e-01 7.00707495e-01
6.55567944e-01 -6.37371898e-01 8.03371742e-02 -3.21518213e-01
-2.56955363e-02 -5.64099371e-01 6.96118772e-01 -9.24808800e-01
3.04574102e-01 -4.56091464e-02 5.48352957e-01 -6.50604784e-01
2.63737813e-02 -7.05188334e-01 -3.02749634e-01 4.09138918e-01
-2.50180781e-01 -3.15736383e-01 3.82452607e-01 4.57153797e-01
-1.72985449e-01 3.68351899e-02 1.27388644e+00 -1.66778788e-01
-1.01378584e+00 3.54162842e-01 1.28795609e-01 1.60709053e-01
1.16780174e+00 -4.64396656e-01 -4.49583411e-01 1.46341503e-01
-7.55254388e-01 4.65108484e-01 4.65129167e-01 6.10903442e-01
3.85208338e-01 -1.26089740e+00 -7.88918316e-01 2.35669404e-01
6.59813464e-01 3.71118367e-01 5.47592491e-02 1.01719761e+00
-1.17712997e-01 2.24954098e-01 -2.92193085e-01 -1.00428963e+00
-1.36556542e+00 4.95688498e-01 5.11584878e-01 -6.07697666e-01
-4.80639428e-01 1.15259862e+00 6.46350086e-01 -6.65667713e-01
2.92210579e-01 -1.79673538e-01 -6.73353076e-02 1.25086099e-01
6.38065457e-01 2.72871524e-01 4.92083952e-02 -6.72454596e-01
-5.70204139e-01 5.91016531e-01 -5.06938756e-01 3.27582240e-01
1.35785007e+00 -1.41761294e-02 -9.81942285e-04 2.13657636e-02
9.45796967e-01 -2.17686206e-01 -1.72559583e+00 -7.30445683e-01
1.14444420e-01 -3.95653784e-01 -1.40457794e-01 -7.80520618e-01
-9.47616100e-01 8.85736585e-01 6.92802846e-01 -7.82356784e-02
1.30108738e+00 4.14095193e-01 5.56242645e-01 3.94366309e-02
2.31968403e-01 -1.15835392e+00 5.20585179e-01 6.34620249e-01
7.79946685e-01 -1.36697972e+00 3.24184597e-02 -4.17844087e-01
-5.33220410e-01 9.38844025e-01 1.30407929e+00 -2.39121825e-01
2.56000131e-01 4.60855663e-01 -2.97384653e-02 -1.02688789e-01
-4.92545784e-01 -6.30752504e-01 4.14280951e-01 7.03208148e-01
2.18260482e-01 -1.44255564e-01 1.85445905e-01 7.38772154e-01
2.74052411e-01 -4.75638583e-02 9.31852013e-02 8.68141294e-01
-5.21546781e-01 -1.22592211e+00 -4.67760086e-01 2.80828714e-01
-2.48869434e-01 1.54964579e-02 -5.14125943e-01 8.78261685e-01
6.44344091e-01 5.45845151e-01 4.74410132e-03 -1.68303207e-01
7.26347923e-01 -2.78429389e-01 3.99036050e-01 -9.71889198e-01
-7.40480661e-01 -2.58714780e-02 -2.08911508e-01 -8.29105198e-01
-4.89046633e-01 -7.21429527e-01 -1.40097010e+00 1.91291362e-01
-5.77453256e-01 -3.14031184e-01 2.08762199e-01 7.88658321e-01
7.19785869e-01 4.66486365e-01 4.64762241e-01 -9.67011631e-01
-6.18453324e-01 -8.36769104e-01 -4.42845643e-01 2.83836663e-01
6.13785088e-01 -8.65634501e-01 -1.64455995e-01 9.12469178e-02] | [9.460602760314941, 1.3896303176879883] |
2bcc451f-e372-4cd2-ac4e-f348b3e7979b | federated-neural-topic-models | 2212.02269 | null | https://arxiv.org/abs/2212.02269v2 | https://arxiv.org/pdf/2212.02269v2.pdf | Federated Neural Topic Models | Over the last years, topic modeling has emerged as a powerful technique for organizing and summarizing big collections of documents or searching for particular patterns in them. However, privacy concerns may arise when cross-analyzing data from different sources. Federated topic modeling solves this issue by allowing multiple parties to jointly train a topic model without sharing their data. While several federated approximations of classical topic models do exist, no research has been conducted on their application for neural topic models. To fill this gap, we propose and analyze a federated implementation based on state-of-the-art neural topic modeling implementations, showing its benefits when there is a diversity of topics across the nodes' documents and the need to build a joint model. In practice, our approach is equivalent to a centralized model training, but preserves the privacy of the nodes. Advantages of this federated scenario are illustrated by means of experiments using both synthetic and real data scenarios. | ['Jerónimo Arenas-García', 'Lorena Calvo-Bartolomé'] | 2022-12-05 | null | null | null | null | ['topic-models'] | ['natural-language-processing'] | [-2.86294311e-01 6.11484766e-01 -1.02559246e-01 -6.35093987e-01
-6.23951018e-01 -5.17411351e-01 9.59290564e-01 4.79762316e-01
-2.80035198e-01 6.63442671e-01 9.20120031e-02 -8.44094455e-02
-1.72841668e-01 -1.04018104e+00 -6.34724081e-01 -6.46402657e-01
-2.79911011e-01 7.86346793e-01 7.99839720e-02 2.42448792e-01
1.26419039e-02 1.76556066e-01 -1.59773910e+00 2.16124415e-01
7.44032383e-01 8.37824166e-01 -4.10864949e-02 1.23179451e-01
-7.45702624e-01 2.42677882e-01 -9.46813524e-01 -1.05760086e+00
4.74511176e-01 2.31142864e-02 -7.98105299e-01 -4.73864786e-02
4.18802619e-01 -3.05215359e-01 -1.35563597e-01 1.17182469e+00
1.84610769e-01 2.56279022e-01 3.25615376e-01 -1.91133773e+00
-3.64673257e-01 9.62355137e-01 -3.48628849e-01 -4.47042406e-01
-2.21379951e-01 -6.18813694e-01 9.25875604e-01 -3.77045989e-01
8.87705445e-01 1.06049979e+00 5.52515090e-01 6.41096711e-01
-1.28509915e+00 -8.40404630e-01 3.85263652e-01 4.32384424e-02
-1.15584159e+00 -1.51950210e-01 8.06193769e-01 -2.80977398e-01
5.22978008e-01 3.73350203e-01 5.35986006e-01 1.12315547e+00
8.56785774e-02 1.06328881e+00 9.02293921e-01 -2.42651030e-01
8.74484897e-01 9.54115987e-01 6.26079977e-01 1.37848675e-01
6.44024670e-01 -3.34389269e-01 -8.26908886e-01 -1.14499366e+00
4.68196757e-02 2.63135135e-01 -1.45086259e-01 -1.03502655e+00
-6.99171901e-01 1.07818735e+00 1.22570820e-01 2.51697510e-01
-4.42706883e-01 -4.28747833e-02 5.83033144e-01 3.19723070e-01
1.16437221e+00 1.78833321e-01 -4.97632593e-01 1.88200891e-01
-1.50319719e+00 6.37608409e-01 1.35804522e+00 1.10463071e+00
8.69336367e-01 -5.30564010e-01 1.52059689e-01 5.66915929e-01
4.47748244e-01 -8.28728825e-02 4.93965864e-01 -8.64452124e-01
3.90954316e-01 6.70349479e-01 -4.12899628e-02 -1.04469979e+00
4.08015698e-02 -2.41790354e-01 -6.83276355e-01 1.37453392e-01
3.07255894e-01 -3.57366651e-01 -5.18738270e-01 1.84030628e+00
5.84367096e-01 1.22853637e-01 1.44568413e-01 5.53547859e-01
2.64300197e-01 7.07863629e-01 9.27412733e-02 -1.45298867e-02
1.42646861e+00 -8.72761309e-01 -9.35552061e-01 1.00727908e-01
2.68132627e-01 -1.79282516e-01 2.81042218e-01 5.51495969e-01
-1.13809502e+00 2.06314087e-01 -7.67867804e-01 -1.98983684e-01
-1.04029775e+00 -3.67528141e-01 1.03433394e+00 1.24213803e+00
-1.27690542e+00 4.20193821e-01 -1.11573219e+00 -6.17510796e-01
7.71969199e-01 5.37085354e-01 -5.74693024e-01 -6.62836060e-02
-1.07145858e+00 4.68655139e-01 3.40593547e-01 -4.90341187e-01
-8.04730535e-01 -1.07101834e+00 -3.73120874e-01 5.65797567e-01
3.70337486e-01 -8.51256430e-01 1.27277744e+00 -5.48484087e-01
-1.06560361e+00 6.85060263e-01 -1.58068627e-01 -9.41953003e-01
6.60917103e-01 -1.55687496e-01 -1.58141166e-01 -4.53698635e-02
-1.55586554e-02 4.99049902e-01 6.52362823e-01 -1.27206087e+00
-9.13652480e-01 -8.29561532e-01 -1.43221796e-01 -2.03203499e-01
-9.39754665e-01 2.95801908e-01 -3.18742990e-01 -5.37356257e-01
-6.57786280e-02 -6.24317050e-01 -3.23122025e-01 3.91515851e-01
-5.32662988e-01 -3.37291092e-01 1.30297053e+00 -4.30784971e-01
9.09298480e-01 -1.96060133e+00 -2.42023215e-01 3.80717427e-01
4.09624130e-01 8.79515782e-02 3.59121531e-01 7.71995425e-01
2.30444685e-01 4.41541463e-01 -2.08440661e-01 -1.16996908e+00
2.79010147e-01 7.64902905e-02 -6.95380151e-01 4.22386616e-01
-5.00098228e-01 3.51978600e-01 -3.18577617e-01 -6.30242109e-01
-1.53253555e-01 7.13814378e-01 -5.09274423e-01 9.59727913e-02
-5.68120003e-01 -2.67896384e-01 -6.37810707e-01 4.53200579e-01
9.18420732e-01 -1.97042584e-01 4.50254291e-01 3.50347579e-01
-1.20522268e-03 2.60514140e-01 -1.24615121e+00 1.91302741e+00
-5.64489607e-03 6.38883948e-01 7.12515354e-01 -9.46786761e-01
1.01267600e+00 8.78859758e-01 7.60687411e-01 -3.18227932e-02
2.22057983e-01 -9.64112803e-02 -7.42396414e-01 1.41626373e-01
8.79074931e-01 -7.92968199e-02 -5.08928346e-03 1.20549107e+00
3.81839722e-01 3.25814873e-01 -1.17960453e-01 4.31242168e-01
8.63278449e-01 -3.75122637e-01 -8.07167683e-03 -3.49961102e-01
-2.33477615e-02 1.82417735e-01 3.63556951e-01 8.75843585e-01
-3.26906368e-02 5.50538182e-01 4.27977353e-01 -5.12483120e-01
-8.41480315e-01 -7.58662701e-01 -1.78399622e-01 1.05482948e+00
-1.05875313e-01 -7.61448979e-01 -1.11998737e+00 -7.22889602e-01
1.45301446e-01 9.31793809e-01 -8.22573304e-01 1.82445705e-01
8.45791027e-02 -5.55593014e-01 5.33893168e-01 1.20732695e-01
3.77949387e-01 -5.90151668e-01 -1.01481867e+00 2.15551838e-01
-1.89348891e-01 -6.41599119e-01 4.89332862e-02 1.92258522e-01
-1.09667873e+00 -8.26129675e-01 -6.96727514e-01 -2.28837088e-01
5.06016970e-01 5.15992105e-01 9.48556662e-01 -3.23244631e-01
-2.32734084e-01 5.42002141e-01 2.37114936e-01 -9.98004973e-01
-1.47995502e-01 3.90688926e-01 -7.98331872e-02 1.79077893e-01
6.77861571e-01 -8.18204284e-01 -3.45310003e-01 2.26006210e-01
-1.16498721e+00 -1.33097559e-01 -6.35691807e-02 6.62875950e-01
2.26538688e-01 2.84297079e-01 4.65012491e-01 -1.38198793e+00
8.69866192e-01 -9.73568141e-01 -7.82019198e-01 4.81758088e-01
-9.54731703e-01 -1.27241373e-01 2.82026619e-01 -2.60963529e-01
-1.29211223e+00 -1.09535508e-01 6.02872550e-01 -5.07376730e-01
-5.71829200e-01 5.53773582e-01 -3.01688373e-01 2.42572621e-01
3.93548578e-01 -3.17131616e-02 2.82941014e-01 -8.83224249e-01
5.48855424e-01 7.91381836e-01 3.19625437e-01 -6.59469128e-01
4.79872167e-01 1.00860333e+00 -2.93525398e-01 -5.50423622e-01
-2.52549917e-01 -6.26060963e-01 -2.74839878e-01 1.96517855e-01
4.75403905e-01 -9.68544304e-01 -6.62662506e-01 2.72650421e-01
-1.30749393e+00 2.49173760e-01 -5.94751000e-01 3.19949687e-01
-5.65101922e-01 8.73429403e-02 -3.33612919e-01 -9.00584280e-01
-6.88538253e-01 -7.03738391e-01 6.41178906e-01 1.87503770e-01
-2.12199479e-01 -1.14138162e+00 2.87075192e-01 3.31170075e-02
8.83790612e-01 2.41953619e-02 9.59892154e-01 -1.51447940e+00
-6.68774605e-01 -7.28272498e-01 1.04185991e-01 -1.39646932e-01
3.37050296e-02 -1.89135849e-01 -1.39171040e+00 -2.51879275e-01
3.55980039e-01 -3.46525982e-02 5.33574462e-01 2.75061697e-01
8.69043648e-01 -6.00197017e-01 -7.51560330e-01 4.78707552e-01
1.25798428e+00 -1.16366394e-01 3.21510941e-01 4.77574110e-01
1.51289389e-01 1.08854151e+00 2.85397142e-01 5.75684726e-01
5.33769429e-01 4.63951200e-01 5.13301909e-01 1.00222677e-01
5.31628430e-01 -3.13497931e-01 -7.66924918e-02 5.09055369e-02
3.55174333e-01 -4.94074404e-01 -7.88071930e-01 8.75679314e-01
-1.95284832e+00 -9.92463648e-01 2.37885669e-01 2.20893288e+00
5.40841997e-01 -4.51770961e-01 1.45722359e-01 -2.09363371e-01
8.41076314e-01 2.22383365e-01 -4.19973493e-01 -2.85265118e-01
-8.21992084e-02 -8.27510953e-02 4.20965672e-01 -5.96811250e-02
-1.17974865e+00 6.46057010e-01 6.43768454e+00 5.91004431e-01
-1.08450961e+00 5.55928886e-01 6.48759246e-01 -5.13366222e-01
-5.04318595e-01 3.14874709e-01 -7.48121262e-01 3.53922129e-01
1.07347381e+00 -8.80289435e-01 -1.64610278e-02 1.56769049e+00
-3.57838392e-01 1.62646890e-01 -1.07642591e+00 5.70163727e-01
1.48885205e-01 -1.52735674e+00 2.86044516e-02 6.87568724e-01
6.16690159e-01 2.74883240e-01 2.93512605e-02 -1.22706443e-02
7.32636333e-01 -7.60978401e-01 6.18618131e-01 3.58608544e-01
-4.81895171e-02 -9.14366186e-01 4.91535246e-01 6.82696700e-01
-5.92520475e-01 -5.19483425e-02 -6.86908185e-01 4.06635433e-01
1.14775948e-01 6.40066922e-01 -8.74282181e-01 7.34300017e-01
8.86839986e-01 1.73281223e-01 -3.98064822e-01 1.41853964e+00
4.59970534e-01 7.55699337e-01 -8.02179694e-01 4.09519672e-02
1.68311387e-01 -2.87977874e-01 6.34169042e-01 9.78672981e-01
5.51772594e-01 -2.78274924e-01 -1.05436109e-01 1.08828127e+00
-1.63362235e-01 2.84585357e-01 -8.14469397e-01 7.58000314e-02
5.67779243e-01 1.28760505e+00 -6.22788072e-01 -3.35984826e-01
-4.24572080e-01 5.27718067e-01 2.13574573e-01 2.76464492e-01
-2.84272820e-01 -1.49550840e-01 1.05345941e+00 1.06572829e-01
2.96897441e-01 -3.67845632e-02 -2.44418040e-01 -1.09959054e+00
4.81408387e-02 -7.55190074e-01 8.32484245e-01 -3.55105340e-01
-1.31580484e+00 7.35622942e-01 3.49409312e-01 -7.18794286e-01
-5.05649567e-01 -8.91920477e-02 -9.39417005e-01 8.54862034e-01
-1.19581485e+00 -1.43329680e+00 -4.11349721e-02 8.43894899e-01
1.67269275e-01 -3.83316457e-01 1.04937923e+00 -4.35725078e-02
-3.40796769e-01 5.53112328e-01 4.66529548e-01 -2.08796695e-01
6.95629895e-01 -1.07947636e+00 5.53279996e-01 8.09362710e-01
3.21116984e-01 1.13144040e+00 7.86862612e-01 -6.56771660e-01
-1.22809100e+00 -1.04375267e+00 1.15838850e+00 -2.03234613e-01
4.30942923e-01 -9.08224881e-01 -1.21611440e+00 9.52427566e-01
5.46486378e-01 -1.51596293e-01 1.22033441e+00 6.24486744e-01
-5.32018721e-01 -4.96874526e-02 -1.52409756e+00 3.18164319e-01
4.54043746e-01 -3.50506306e-01 -3.72166812e-01 4.95226383e-01
8.58725250e-01 2.19080478e-01 -8.25196147e-01 -3.59400451e-01
4.29183871e-01 -1.04186356e+00 6.83930337e-01 -7.92579710e-01
-1.06107235e-01 4.52038012e-02 -2.18498647e-01 -1.16802132e+00
1.22255504e-01 -9.77762580e-01 -3.12479615e-01 1.75685322e+00
3.52559149e-01 -9.20423687e-01 1.49841440e+00 1.67825186e+00
3.72751594e-01 -2.87514567e-01 -1.26625097e+00 -5.93265533e-01
2.69036591e-01 -5.33774197e-01 1.25633919e+00 1.18419707e+00
4.05584097e-01 -4.17385250e-01 -2.72599757e-01 2.12750465e-01
9.59446251e-01 2.70812780e-01 1.05267620e+00 -1.77529216e+00
-6.42890856e-02 -2.32718468e-01 -3.92987460e-01 -3.97421092e-01
3.36609066e-01 -7.38642514e-01 -3.64590287e-01 -1.42627943e+00
9.22403708e-02 -7.04809725e-01 -2.07278281e-01 7.46484876e-01
3.32363367e-01 -5.92514396e-01 2.89751381e-01 3.27800781e-01
-5.09069800e-01 6.12088680e-01 1.11416936e-01 -1.37910858e-01
-1.45433480e-02 4.12034661e-01 -1.11652482e+00 4.75112826e-01
8.24192345e-01 -8.87699783e-01 -5.74858069e-01 -4.33480740e-01
-9.04553160e-02 5.65509945e-02 4.21931833e-01 -9.98391151e-01
1.06372547e+00 8.12735707e-02 -5.97927757e-02 -1.00062776e+00
5.49227536e-01 -1.29858446e+00 5.46243727e-01 5.93663827e-02
-5.75496614e-01 -2.77820826e-01 1.26774013e-01 9.52960312e-01
-3.35456312e-01 -2.76034713e-01 1.99214548e-01 -1.96287170e-01
-7.57376850e-02 4.23085958e-01 -3.28414470e-01 -4.51254427e-01
1.22877181e+00 -6.73367977e-02 -5.18354714e-01 -6.10818624e-01
-4.76512283e-01 3.96165669e-01 7.68981755e-01 3.42256159e-01
2.86730140e-01 -7.03399897e-01 -4.58266348e-01 2.60418445e-01
-1.22507932e-02 1.69784322e-01 3.77379715e-01 1.71389773e-01
2.03206465e-01 7.84920990e-01 -9.44806263e-02 -3.79145056e-01
-1.37797153e+00 8.34732294e-01 -5.70144691e-02 -3.49918753e-01
-7.36686230e-01 6.46007061e-01 2.43912190e-01 -6.36561632e-01
7.34615743e-01 -1.45245967e-02 1.40349150e-01 3.05906683e-01
6.95668638e-01 4.46125418e-01 3.27624828e-01 -1.79775015e-01
-1.20263189e-01 -2.19462201e-01 -4.61838543e-01 -5.17110944e-01
1.57117414e+00 -2.03985721e-01 -3.59209716e-01 3.97548646e-01
1.07523382e+00 -1.82139754e-01 -9.80607986e-01 -5.25180042e-01
6.17976971e-02 -4.56999481e-01 1.17881082e-01 -6.39236510e-01
-1.23095953e+00 9.45613563e-01 4.20552850e-01 5.39778531e-01
8.77548158e-01 7.90386945e-02 4.92153764e-01 6.05276883e-01
6.36075079e-01 -8.65167260e-01 -7.75935233e-01 -7.56694824e-02
4.91190940e-01 -8.74350488e-01 7.63324350e-02 -4.48979139e-01
-4.97792840e-01 9.03789699e-01 2.87427366e-01 1.11300915e-01
1.10843182e+00 2.62915581e-01 1.42469347e-01 -4.20214295e-01
-1.21945739e+00 6.87170565e-01 -1.46297470e-01 8.65940571e-01
-1.64775290e-02 -4.07223403e-02 -6.72394857e-02 1.00832629e+00
-3.02781194e-01 4.63758931e-02 4.50551212e-01 1.29965615e+00
-9.44339037e-02 -1.40833509e+00 -4.06704545e-01 5.46687841e-01
-8.66620123e-01 1.43373936e-01 -5.11410117e-01 7.10863948e-01
-3.06442022e-01 9.31737304e-01 3.25084299e-01 -1.15494743e-01
-2.50845164e-01 5.92067599e-01 -4.62838471e-01 -5.60341477e-01
-1.10082197e+00 -3.33391130e-02 -2.05345854e-01 -4.99076813e-01
-2.55166620e-01 -9.08704460e-01 -8.04263711e-01 -3.69982541e-01
-6.30404174e-01 7.94532061e-01 1.56906486e+00 5.14964402e-01
9.83621657e-01 -1.30954444e-01 3.92987818e-01 -4.72221136e-01
-6.16469145e-01 -6.56464398e-01 -9.10084844e-01 1.35777280e-01
3.03780325e-02 -3.98194015e-01 -2.31308252e-01 -4.46928814e-02] | [5.889130115509033, 6.49974250793457] |
17246ea5-4944-4c5d-ae00-f359c633e232 | a-dual-scale-lead-seperated-transformer-with | 2211.12777 | null | https://arxiv.org/abs/2211.12777v1 | https://arxiv.org/pdf/2211.12777v1.pdf | A Dual-scale Lead-seperated Transformer With Lead-orthogonal Attention And Meta-information For Ecg Classification | Auxiliary diagnosis of cardiac electrophysiological status can be obtained through the analysis of 12-lead electrocardiograms (ECGs). This work proposes a dual-scale lead-separated transformer with lead-orthogonal attention and meta-information (DLTM-ECG) as a novel approach to address this challenge. ECG segments of each lead are interpreted as independent patches, and together with the reduced dimension signal, they form a dual-scale representation. As a method to reduce interference from segments with low correlation, two group attention mechanisms perform both lead-internal and cross-lead attention. Our method allows for the addition of previously discarded meta-information, further improving the utilization of clinical information. Experimental results show that our DLTM-ECG yields significantly better classification scores than other transformer-based models,matching or performing better than state-of-the-art (SOTA) deep learning methods on two benchmark datasets. Our work has the potential for similar multichannel bioelectrical signal processing and physiological multimodal tasks. | ['Li Sun', 'Wenming Yang', 'Zhourui Xia', 'Guijin Wang', 'Yang Li'] | 2022-11-23 | null | null | null | null | ['ecg-classification'] | ['medical'] | [ 5.10398328e-01 -6.87809065e-02 1.45940304e-01 -8.22381005e-02
-1.14927733e+00 -3.68379384e-01 -1.35855913e-01 1.58532023e-01
-2.83512890e-01 7.25104928e-01 1.47966295e-01 -3.20483506e-01
-4.03264433e-01 -2.59055287e-01 -4.51025426e-01 -8.13636303e-01
-2.90720046e-01 1.45427123e-01 -3.32752287e-01 -6.85494244e-02
-5.79950819e-03 2.41446912e-01 -8.86908293e-01 7.85257220e-01
1.00497663e+00 1.39170456e+00 -9.15738791e-02 5.37564993e-01
3.31229419e-01 4.98225391e-01 -7.55290627e-01 -1.98839068e-01
2.69451905e-02 -8.36955070e-01 -6.09446764e-01 -3.45887452e-01
1.17539965e-01 -3.51827941e-03 -1.01443864e-01 7.64157355e-01
1.29781997e+00 -2.80771285e-01 3.47081423e-01 -6.79693878e-01
-3.84566128e-01 7.06405282e-01 -5.77326000e-01 6.41799867e-01
1.07273251e-01 3.83375399e-02 7.67799199e-01 -9.43314612e-01
2.43328869e-01 5.39248347e-01 1.19443202e+00 2.44821161e-01
-1.54202390e+00 -4.65695798e-01 -1.36745855e-01 4.70431209e-01
-1.35493851e+00 -1.81420088e-01 1.12086308e+00 -1.89127326e-01
1.23268080e+00 3.66701275e-01 9.77481127e-01 1.26652730e+00
5.56099176e-01 4.78716642e-01 1.07257235e+00 -2.73375183e-01
7.25704385e-03 -7.81328306e-02 1.95463225e-01 2.38166958e-01
2.15788428e-02 -2.23748490e-01 -5.78197479e-01 -3.78188640e-01
5.92869282e-01 2.92054005e-02 -6.42760634e-01 -1.49841055e-01
-1.71754241e+00 3.03962976e-01 4.02608931e-01 6.65882230e-01
-9.94588375e-01 -1.68992567e-03 7.57773697e-01 4.66868460e-01
6.19859397e-01 7.88736463e-01 -7.21340477e-01 -1.81840703e-01
-9.16137516e-01 -1.24348767e-01 3.88500392e-01 2.84425408e-01
2.91431934e-01 2.65016735e-01 -7.13411570e-01 7.96765685e-01
-4.32031937e-02 4.82036531e-01 7.09906340e-01 -9.36695457e-01
4.99883085e-01 5.07942557e-01 -1.34582460e-01 -8.91071737e-01
-7.88261890e-01 -1.07726455e+00 -1.28475058e+00 -3.02482933e-01
1.20338380e-01 -3.14359009e-01 -7.00845599e-01 1.70360601e+00
-6.89985752e-02 4.19590592e-01 -9.07966271e-02 8.90952587e-01
8.70458066e-01 4.17741239e-01 -4.59289365e-02 -5.50654650e-01
1.49016333e+00 -3.82502705e-01 -9.12459135e-01 -5.65932244e-02
4.37841833e-01 -1.28149867e-01 5.78633785e-01 6.13677382e-01
-1.14311743e+00 -7.36180484e-01 -9.25883591e-01 1.24482095e-01
7.09376633e-02 2.33184457e-01 3.78991276e-01 7.60457993e-01
-1.16400981e+00 8.44844282e-01 -8.51038814e-01 9.43216383e-02
5.92818141e-01 4.40613925e-01 -3.54547769e-01 3.79101813e-01
-1.39232230e+00 7.59080291e-01 -1.20270595e-01 5.59928060e-01
-5.41220725e-01 -1.01604688e+00 -6.12833679e-01 2.64772147e-01
-8.54869857e-02 -8.46433699e-01 5.05548120e-01 -7.39267170e-01
-1.13815415e+00 8.24914515e-01 -2.44016245e-01 -6.00638270e-01
2.73835570e-01 -1.76016852e-01 -2.95088917e-01 2.97387332e-01
1.08722620e-01 3.24677050e-01 9.03394043e-01 -8.61331940e-01
-6.93519413e-03 -8.12995493e-01 -5.66826761e-01 1.21430807e-01
-3.61840099e-01 -1.29059136e-01 -1.92302123e-01 -8.80074739e-01
3.32946241e-01 -6.84952915e-01 -2.86319524e-01 -5.29942811e-01
-3.40189815e-01 9.09790769e-02 2.43930772e-01 -9.91097212e-01
1.46792924e+00 -2.25471020e+00 6.02692902e-01 3.55058730e-01
7.44394004e-01 4.18261349e-01 -2.49207914e-01 2.04150379e-01
-4.35076892e-01 2.85809427e-01 -2.88763106e-01 -4.28383887e-01
-3.80244285e-01 -9.99868661e-02 -1.36052072e-01 4.69831467e-01
2.44535983e-01 1.35460305e+00 -5.57816386e-01 -2.67612755e-01
1.57855824e-01 8.28953207e-01 -1.68148518e-01 -1.34828836e-02
5.20328820e-01 9.20138240e-01 -2.34997585e-01 5.00120342e-01
5.16809404e-01 -4.09945905e-01 5.32385051e-01 -7.38318980e-01
2.52926081e-01 3.17643851e-01 -7.10745811e-01 2.05316591e+00
-1.53994918e-01 5.84542871e-01 -8.85644779e-02 -1.32409906e+00
8.26372564e-01 7.89984941e-01 9.91561234e-01 -1.00513124e+00
4.14382398e-01 1.58520132e-01 4.60303545e-01 -5.27786732e-01
-1.71932012e-01 -1.53244108e-01 1.03750512e-01 3.72552514e-01
1.62436068e-01 3.98770452e-01 -1.98791727e-01 -6.17567711e-02
1.21369994e+00 1.11923032e-01 1.89698100e-01 -3.61715585e-01
4.35896456e-01 -6.53156698e-01 7.86698759e-01 9.70174909e-01
-3.08009148e-01 7.15600669e-01 5.18651009e-01 -6.86223269e-01
-5.97983420e-01 -9.03940856e-01 -3.28913063e-01 5.17675877e-01
-2.29510024e-01 -5.38900375e-01 -6.41731441e-01 -3.62967640e-01
-6.29410222e-02 1.20504193e-01 -9.63750124e-01 -3.70221257e-01
-6.79963231e-01 -1.06956172e+00 8.36566687e-01 9.64140952e-01
4.54884917e-02 -1.08820391e+00 -1.08060956e+00 6.87266648e-01
-8.14201415e-01 -9.10578787e-01 -2.09061787e-01 5.70260167e-01
-1.28372121e+00 -8.92150283e-01 -1.09516919e+00 -3.41811031e-01
1.34109914e-01 -2.65777528e-01 1.12582886e+00 -1.78569809e-01
-4.67999101e-01 2.74767548e-01 -2.54270494e-01 -5.77044070e-01
8.37318450e-02 -4.65806983e-02 -1.60937905e-01 5.52427173e-01
3.25315118e-01 -9.04602230e-01 -8.46173108e-01 -3.01905740e-02
-3.50591868e-01 -2.24839807e-01 6.94019675e-01 1.16309381e+00
7.45088518e-01 -5.57615757e-01 1.12375593e+00 -6.11705244e-01
9.01041806e-01 -2.64520437e-01 1.11058816e-01 2.06169426e-01
-6.61518693e-01 -2.23442435e-01 6.77770376e-01 -3.46694618e-01
-5.57949185e-01 -1.51878834e-01 -5.22687435e-02 -7.31470644e-01
4.58968617e-02 6.00087345e-01 -7.57561391e-03 -4.69817407e-02
5.92561901e-01 3.73701870e-01 6.87277615e-02 -4.62962449e-01
-2.48433039e-01 4.46179360e-01 5.57145596e-01 -2.86660820e-01
-8.86166915e-02 3.61977726e-01 1.93230569e-01 -6.59198105e-01
-4.92486745e-01 -3.61199915e-01 -8.27824533e-01 -1.11950532e-01
1.01509643e+00 -7.81741261e-01 -7.09144711e-01 5.31528592e-01
-1.19081688e+00 -1.23904571e-01 -2.96482623e-01 5.84452152e-01
-2.45332047e-01 5.07717311e-01 -6.72480822e-01 -9.07554507e-01
-1.04095531e+00 -8.77612233e-01 1.15225196e+00 -3.90387475e-01
-4.54114795e-01 -7.88622975e-01 1.43378332e-01 2.04503998e-01
4.89325523e-01 6.12986028e-01 8.67017746e-01 -6.51580870e-01
-6.47923350e-02 -1.77181527e-01 1.31964758e-01 4.69292492e-01
6.46000132e-02 -6.28854871e-01 -1.23554349e+00 -3.20592999e-01
3.27222139e-01 3.90225835e-02 1.04658270e+00 8.42594147e-01
1.18923986e+00 1.76004171e-01 -3.93310159e-01 8.81173968e-01
9.25840318e-01 3.08353305e-01 9.74360228e-01 -1.12929463e-01
8.27354372e-01 2.79953897e-01 -1.92449670e-02 5.49729466e-01
4.61584359e-01 5.09762883e-01 8.37547332e-02 -8.00281584e-01
1.00766465e-01 3.73778015e-01 4.32169959e-02 9.07245755e-01
-4.45215493e-01 5.46638034e-02 -9.43388283e-01 6.36470675e-01
-1.77340627e+00 -1.01643443e+00 -4.44310188e-01 2.26289392e+00
5.68484724e-01 -1.16118208e-01 1.25018939e-01 6.90380514e-01
5.78616858e-01 -1.15171574e-01 -7.73055792e-01 -2.50495225e-01
-3.92223686e-01 6.49512768e-01 5.29539734e-02 -1.06688887e-01
-1.08581996e+00 -6.44180784e-03 6.36542368e+00 3.20268631e-01
-1.45422602e+00 3.80251288e-01 7.85288393e-01 -1.39001906e-01
-5.35779260e-02 -5.52214563e-01 -6.31026775e-02 4.94093567e-01
1.12075639e+00 1.68265268e-01 2.26398602e-01 3.66128609e-02
2.28642404e-01 2.40965769e-01 -1.24091911e+00 1.43910694e+00
2.09983468e-01 -1.05773020e+00 -2.52552032e-01 2.12635454e-02
3.64980429e-01 2.59102732e-01 3.64163518e-02 1.98476389e-01
-9.88649189e-01 -9.59278762e-01 6.23301446e-01 7.73296475e-01
1.19600129e+00 -5.49742162e-01 9.66691792e-01 8.32994357e-02
-1.26677942e+00 -3.72070968e-01 -1.92834157e-02 1.61123782e-01
1.07059665e-01 7.66839564e-01 -2.25324020e-01 1.03738928e+00
1.00633502e+00 1.14168370e+00 -5.62498212e-01 9.47085381e-01
3.12859118e-01 8.22848558e-01 -2.12536827e-01 3.62450629e-01
-2.75876760e-01 2.81356927e-02 7.21864820e-01 1.12631166e+00
3.14048469e-01 2.17666492e-01 -1.17721595e-02 1.04818845e+00
6.81504235e-02 2.24477127e-02 -4.71733749e-01 2.28762522e-01
8.09424669e-02 1.20629060e+00 -5.69941521e-01 -5.67801714e-01
-2.47740954e-01 9.18628037e-01 2.10759081e-02 4.99803960e-01
-8.08973849e-01 -6.49117231e-01 2.12248832e-01 8.64929799e-03
1.16682097e-01 3.46255660e-01 -7.70981610e-01 -1.27622497e+00
3.82506162e-01 -1.01371598e+00 4.22904044e-01 -5.89352787e-01
-1.25236464e+00 1.04389250e+00 -4.76324975e-01 -1.25400043e+00
-1.51519507e-01 -3.35458785e-01 -6.12621844e-01 1.09483314e+00
-1.38282931e+00 -1.00091314e+00 -1.71589524e-01 6.39442623e-01
1.40402801e-02 1.48655921e-02 1.27743399e+00 7.40479052e-01
-4.14587706e-01 7.28482127e-01 -2.84051776e-01 1.14482217e-01
7.09785402e-01 -1.27383709e+00 2.56090730e-01 6.21572316e-01
7.60384202e-02 6.52501643e-01 1.71899274e-01 -3.53269845e-01
-1.36220777e+00 -9.53371286e-01 7.65456676e-01 -5.31617939e-01
1.00653945e-02 -2.67670900e-01 -1.11208868e+00 3.48018140e-01
2.41004795e-01 1.17363088e-01 9.45324779e-01 2.32306466e-01
-9.01043639e-02 -6.60086930e-01 -7.79853165e-01 9.14346129e-02
7.93970585e-01 -7.07092285e-01 -7.23149717e-01 -1.86224446e-01
7.09272176e-02 -1.69057369e-01 -1.28391707e+00 8.24108243e-01
8.46920669e-01 -1.03069472e+00 9.50839579e-01 -4.32662427e-01
4.55417670e-02 -7.19972774e-02 3.43954742e-01 -1.43225801e+00
-6.58325434e-01 -9.06971037e-01 -1.42008573e-01 7.76232302e-01
4.84545648e-01 -9.23047423e-01 3.30508560e-01 2.63563544e-01
-4.63190645e-01 -1.09616303e+00 -1.11845350e+00 -3.86317760e-01
-3.80404554e-02 -5.19203842e-01 3.62138271e-01 8.94156456e-01
3.39897186e-01 4.49504733e-01 -5.34876645e-01 -1.23697363e-01
6.77395523e-01 1.05307899e-01 2.59555411e-02 -1.49436057e+00
-4.15435255e-01 -4.65226740e-01 -4.67401236e-01 -5.82639754e-01
-6.86511695e-02 -1.08496141e+00 -3.46736610e-01 -1.51761770e+00
2.50809431e-01 -1.14362083e-01 -1.18762600e+00 7.16880202e-01
-3.74398917e-01 7.80398548e-01 4.49342690e-02 -2.96442211e-02
-5.20637572e-01 3.95601481e-01 1.02966332e+00 -2.15216115e-01
-4.28178400e-01 -2.03126609e-01 -7.52990901e-01 3.55796963e-01
6.64510071e-01 -3.91452163e-01 -2.91761667e-01 -2.78036118e-01
1.35110825e-01 5.28661311e-01 5.09218276e-01 -1.15895474e+00
-1.45292049e-02 8.27653170e-01 8.35940182e-01 -5.92610359e-01
4.49919552e-01 -5.34274697e-01 2.78044641e-01 5.38374603e-01
-3.36798817e-01 2.15073898e-01 3.41242969e-01 3.68611425e-01
-3.33670855e-01 5.89716494e-01 5.88601947e-01 -1.48942515e-01
6.31107623e-03 1.77800044e-01 -6.22322083e-01 1.65131837e-01
6.41929388e-01 -3.02710235e-01 9.88191888e-02 -1.12343483e-01
-1.13689768e+00 7.13151395e-02 -3.65613580e-01 3.11212718e-01
8.09316218e-01 -1.24297416e+00 -1.03099501e+00 4.88806725e-01
2.69328915e-02 -4.98032361e-01 7.64941633e-01 1.77675831e+00
1.04009390e-01 4.69602138e-01 -5.15680432e-01 -1.04537928e+00
-1.10602343e+00 3.56748402e-01 8.69648218e-01 -3.45258504e-01
-9.35188115e-01 7.02522993e-01 1.90297231e-01 8.13029259e-02
1.43210649e-01 -6.50706410e-01 -3.09360892e-01 2.70521730e-01
5.87672472e-01 3.62779468e-01 5.99142492e-01 -3.97067010e-01
-6.76904500e-01 7.25918949e-01 3.31325233e-01 1.18978709e-01
1.41601193e+00 -9.46570784e-02 -3.24013680e-01 6.77809656e-01
9.94574845e-01 -2.65320748e-01 -7.24569917e-01 -8.70715752e-02
-2.57305890e-01 -5.91982566e-02 1.92640752e-01 -1.06984210e+00
-1.27119434e+00 1.43538082e+00 9.72159624e-01 1.67428240e-01
1.59835494e+00 -4.60853904e-01 8.07790041e-01 1.32059425e-01
3.33292723e-01 -8.34253132e-01 -5.80003895e-02 7.72559196e-02
8.37451458e-01 -8.94218624e-01 -2.89588630e-01 1.05544776e-01
-6.83716476e-01 9.94673371e-01 1.47050470e-01 1.01156525e-01
7.33211398e-01 3.01415890e-01 2.44641811e-01 -3.50276977e-01
-8.03977072e-01 -5.46458352e-04 6.18613422e-01 7.03460217e-01
6.85130656e-01 -7.99815357e-02 -3.90827298e-01 1.19351304e+00
4.69922334e-01 1.10676341e-01 1.86118647e-01 5.50337672e-01
1.33519351e-01 -9.05804396e-01 -3.10102165e-01 7.52757370e-01
-1.01044989e+00 -4.42243218e-01 -2.74252892e-01 1.73377842e-01
2.14542449e-01 9.03622329e-01 -4.99416664e-02 -3.67197752e-01
4.16200250e-01 4.79876935e-01 6.27536535e-01 -4.37610865e-01
-1.17652071e+00 6.11006081e-01 -1.14853434e-01 -7.04704165e-01
-3.24719787e-01 -4.94242489e-01 -1.03040230e+00 4.73728836e-01
-2.46404842e-01 1.03341803e-01 3.23794365e-01 7.87754655e-01
1.02710485e+00 1.27572131e+00 4.75267261e-01 -9.73925889e-01
-4.79030699e-01 -1.19970047e+00 -6.50546372e-01 4.49491292e-01
6.24014556e-01 -3.39661866e-01 4.09340411e-02 2.26228721e-02] | [14.27277660369873, 3.2588775157928467] |
dc428348-0a33-410a-872d-cc6f889e04d5 | formal-guarantees-for-heuristic-optimization | 2208.00502 | null | https://arxiv.org/abs/2208.00502v1 | https://arxiv.org/pdf/2208.00502v1.pdf | Formal guarantees for heuristic optimization algorithms used in machine learning | Recently, Stochastic Gradient Descent (SGD) and its variants have become the dominant methods in the large-scale optimization of machine learning (ML) problems. A variety of strategies have been proposed for tuning the step sizes, ranging from adaptive step sizes to heuristic methods to change the step size in each iteration. Also, momentum has been widely employed in ML tasks to accelerate the training process. Yet, there is a gap in our theoretical understanding of them. In this work, we start to close this gap by providing formal guarantees to a few heuristic optimization methods and proposing improved algorithms. First, we analyze a generalized version of the AdaGrad (Delayed AdaGrad) step sizes in both convex and non-convex settings, showing that these step sizes allow the algorithms to automatically adapt to the level of noise of the stochastic gradients. We show for the first time sufficient conditions for Delayed AdaGrad to achieve almost sure convergence of the gradients to zero. Moreover, we present a high probability analysis for Delayed AdaGrad and its momentum variant in the non-convex setting. Second, we analyze SGD with exponential and cosine step sizes, which are empirically successful but lack theoretical support. We provide the very first convergence guarantees for them in the smooth and non-convex setting, with and without the Polyak-{\L}ojasiewicz (PL) condition. We also show their good property of adaptivity to noise under the PL condition. Third, we study the last iterate of momentum methods. We prove the first lower bound in the convex setting for the last iterate of SGD with constant momentum. Moreover, we investigate a class of Follow-The-Regularized-Leader-based momentum algorithms with increasing momentum and shrinking updates. We show that their last iterate has optimal convergence for unconstrained convex stochastic optimization problems. | ['Xiaoyu Li'] | 2022-07-31 | null | null | null | null | ['machine-learning', 'machine-learning'] | ['methodology', 'miscellaneous'] | [-2.99456328e-01 -1.75438691e-02 -1.59063831e-01 -1.88768104e-01
-8.17835331e-01 -5.49523950e-01 1.45885438e-01 2.10732087e-01
-6.95162356e-01 9.41027164e-01 -1.76922813e-01 -3.34851950e-01
-3.76106024e-01 -5.63512266e-01 -9.25635755e-01 -1.17718184e+00
-1.40642643e-01 4.85929310e-01 1.43261701e-01 -3.08316410e-01
1.84211120e-01 4.74674851e-01 -1.24010730e+00 -3.36934865e-01
9.57329929e-01 8.89335215e-01 1.88867286e-01 8.93285036e-01
-5.73071837e-02 2.63901204e-01 -2.81843305e-01 -2.73739725e-01
5.30352712e-01 -4.98180807e-01 -6.19179428e-01 1.80507794e-01
3.25617909e-01 9.46772844e-02 2.22442955e-01 1.37354672e+00
7.53493607e-01 3.08302373e-01 4.32993382e-01 -1.27649093e+00
-1.68175071e-01 5.68503439e-01 -8.05810690e-01 2.37935945e-01
-7.52028525e-02 -2.07885996e-01 1.06070971e+00 -9.84901905e-01
5.86921990e-01 9.02063131e-01 9.41463411e-01 5.87742686e-01
-1.02639484e+00 -2.01362506e-01 2.99716711e-01 8.28517154e-02
-1.09025168e+00 5.95190423e-03 6.07498348e-01 -3.98611069e-01
5.53922594e-01 4.70539331e-01 6.55493081e-01 6.18188739e-01
-2.00175270e-01 9.86382782e-01 1.22119009e+00 -7.80357718e-01
3.51302862e-01 4.09882963e-01 3.34755331e-01 9.80370402e-01
4.60492820e-01 -2.12122649e-01 -3.37483585e-01 -3.81436855e-01
5.48104703e-01 -2.82391191e-01 -4.33785051e-01 -5.34363627e-01
-9.19332325e-01 1.03452873e+00 1.35906085e-01 4.27986175e-01
-3.90831769e-01 1.52905723e-02 2.16994107e-01 3.35691422e-01
7.43465304e-01 3.54480952e-01 -5.85805953e-01 -2.72836983e-01
-8.52624178e-01 3.91216487e-01 1.17914200e+00 6.96446478e-01
5.23954391e-01 -1.16067864e-02 -1.97823867e-01 7.93099105e-01
4.99780849e-02 6.07356429e-01 4.67584312e-01 -7.19554842e-01
7.25260377e-01 8.50371346e-02 3.85386020e-01 -9.57999647e-01
-6.13316119e-01 -6.88914359e-01 -8.37854445e-01 3.47415268e-01
9.41563964e-01 -2.47184023e-01 -3.17830890e-01 1.85434508e+00
6.11289144e-01 7.97110870e-02 -2.53594726e-01 9.19240117e-01
-1.61870539e-01 6.56930208e-01 -5.06113827e-01 -7.88138449e-01
7.91376472e-01 -9.75313544e-01 -4.88166690e-01 1.34545401e-01
8.82507086e-01 -5.49839497e-01 1.37356937e+00 4.92103428e-01
-1.20870900e+00 -1.48662999e-01 -9.55496669e-01 1.69316784e-01
9.34082270e-02 3.53659481e-01 3.83910209e-01 7.59760618e-01
-9.69481230e-01 9.38792944e-01 -1.03505754e+00 -1.08158708e-01
1.36640668e-01 2.87954211e-01 -5.97213693e-02 2.97678947e-01
-6.95569992e-01 5.35127878e-01 1.90628216e-01 4.37987447e-01
-4.17920291e-01 -6.27732873e-01 -4.47284430e-01 -7.54367113e-02
4.49340910e-01 -5.81387043e-01 1.09678936e+00 -1.02520406e+00
-1.76130903e+00 8.30786288e-01 -2.48653293e-01 -5.49681187e-01
1.22083831e+00 -4.25616950e-01 4.22856748e-01 -7.97292739e-02
-1.61254793e-01 -2.75045425e-01 8.96184444e-01 -9.99992132e-01
-6.91101909e-01 -4.92379457e-01 8.94913673e-02 4.16074932e-01
-6.04055822e-01 -1.91743113e-02 -2.80196071e-01 -4.68038052e-01
6.15364090e-02 -1.03258312e+00 -4.69037980e-01 -3.19488123e-02
-4.00039077e-01 -2.33678445e-01 2.44211733e-01 -4.40574795e-01
1.16801727e+00 -2.01442647e+00 4.04184490e-01 3.15291643e-01
6.33102134e-02 2.52050817e-01 2.44963188e-02 2.17735663e-01
1.88877136e-01 7.14289024e-02 -4.30486470e-01 -6.62213027e-01
1.54634714e-01 1.87492102e-01 -4.02102023e-01 9.26349640e-01
-2.30387226e-01 5.95941186e-01 -9.72920060e-01 -2.79474974e-01
-9.81886759e-02 1.54010549e-01 -7.37596631e-01 -2.37814151e-02
-4.63701785e-02 2.87213683e-01 -5.70323110e-01 2.06316844e-01
7.63434112e-01 -2.59691060e-01 4.75506708e-02 8.81018937e-02
-2.51939714e-01 -3.98955457e-02 -1.59787834e+00 1.03560698e+00
-6.36840463e-01 5.10764480e-01 7.15841889e-01 -1.40657580e+00
5.11321902e-01 3.36911082e-02 5.71574867e-01 2.48106215e-02
5.82827516e-02 5.45124292e-01 -3.35386187e-01 -3.34581465e-01
4.45256889e-01 -3.76684040e-01 2.28766769e-01 3.90185684e-01
-2.63425946e-01 -1.52811334e-01 4.50968295e-01 -3.59201282e-02
8.39235067e-01 -1.40350729e-01 2.03493446e-01 -4.55339402e-01
7.21835732e-01 -2.42564335e-01 5.66910505e-01 1.03016663e+00
1.00786164e-02 7.06730604e-01 5.95968962e-01 -2.24560291e-01
-9.04590189e-01 -7.62133956e-01 -1.72781140e-01 1.20341182e+00
-2.16446854e-02 -2.85684884e-01 -6.79827869e-01 -6.84978068e-01
7.88730383e-02 4.62073833e-01 -6.03420198e-01 4.79873382e-02
-7.46981204e-01 -1.43687165e+00 1.59284353e-01 2.20802918e-01
4.39558476e-01 -5.26713550e-01 -3.94857913e-01 2.71062315e-01
7.65237957e-02 -9.82796252e-01 -5.24121702e-01 2.59743243e-01
-1.13798463e+00 -1.11935198e+00 -1.07675612e+00 -5.85991085e-01
8.89318168e-01 4.96052243e-02 8.29631507e-01 1.14025138e-02
8.90790150e-02 5.70930302e-01 -2.00421914e-01 -2.82524258e-01
-4.24366772e-01 1.97358325e-01 3.37699085e-01 6.91214353e-02
-3.50548059e-01 -5.25726497e-01 -3.40443403e-01 4.97024655e-01
-6.68162823e-01 -4.95590568e-02 2.07887858e-01 9.94026601e-01
8.89461875e-01 -1.09264450e-02 1.19072683e-01 -7.91881144e-01
8.88155997e-01 -3.59729350e-01 -1.08318794e+00 2.31240943e-01
-8.16346526e-01 2.71271676e-01 1.00438666e+00 -5.64101517e-01
-7.37203181e-01 -1.98649522e-02 -1.76378012e-01 -2.55834341e-01
5.05698800e-01 5.45270920e-01 1.60546556e-01 -2.57902175e-01
6.09038234e-01 2.50188738e-01 -1.25224262e-01 -6.26630247e-01
3.24263930e-01 3.37838799e-01 3.08801889e-01 -9.10152256e-01
9.52441931e-01 7.98036933e-01 3.24408948e-01 -9.32620108e-01
-1.22424018e+00 -4.80977863e-01 -4.14698422e-01 -1.19023606e-01
3.48676860e-01 -3.56225699e-01 -7.67906070e-01 5.94843268e-01
-8.69579554e-01 -6.20765030e-01 -4.96408969e-01 6.00020170e-01
-7.82655656e-01 5.04955888e-01 -7.06954837e-01 -1.08307707e+00
-3.28277379e-01 -8.57594430e-01 5.86286545e-01 2.27532640e-01
3.55756491e-01 -1.25724816e+00 2.25198612e-01 -3.42260264e-02
4.34978664e-01 1.58183470e-01 5.09613454e-01 -6.57717586e-01
-4.10431735e-02 -2.34740466e-01 8.70855674e-02 6.75518334e-01
-1.46819249e-01 -9.30593610e-02 -3.56184274e-01 -5.12632430e-01
5.67018867e-01 -8.59037191e-02 8.96998882e-01 7.19809353e-01
1.06402874e+00 -5.80896556e-01 -2.96347171e-01 9.91034210e-01
1.37177646e+00 -4.08777535e-01 -4.02548648e-02 5.86500585e-01
5.71844459e-01 3.07836860e-01 6.16105616e-01 7.42213845e-01
-2.21532118e-02 7.82078147e-01 2.78235555e-01 -8.63907021e-03
3.03184301e-01 -3.34365778e-02 4.97205973e-01 1.02525151e+00
-3.57536584e-01 -1.17480353e-01 -5.05329728e-01 3.65813464e-01
-2.02838492e+00 -8.13700736e-01 -2.75492936e-01 2.63666201e+00
1.06985199e+00 2.00658530e-01 2.79071212e-01 1.77633315e-01
8.44018042e-01 -7.50524253e-02 -4.40235198e-01 -4.39505219e-01
-3.70469213e-01 2.25769907e-01 8.47725153e-01 9.53281343e-01
-1.02961969e+00 7.51156807e-01 5.80715561e+00 9.89455640e-01
-1.17452633e+00 2.43018672e-01 4.69602495e-01 -3.07327718e-01
-6.32970259e-02 -1.28363431e-01 -1.10668111e+00 5.12011468e-01
5.12242854e-01 -3.44828516e-01 5.08062422e-01 1.18260419e+00
3.97556841e-01 -1.26301989e-01 -7.63006091e-01 8.45171452e-01
-1.45007819e-01 -1.10068607e+00 -5.16070843e-01 1.38705149e-01
1.13250625e+00 1.05462700e-01 6.69249892e-02 5.95759563e-02
2.50456333e-01 -5.76882720e-01 6.60949111e-01 2.12908775e-01
2.65097797e-01 -7.02930272e-01 7.18330801e-01 4.86269027e-01
-9.78615880e-01 -1.67795151e-01 -5.35139203e-01 -1.04398541e-01
3.68813932e-01 1.16410267e+00 -4.78118390e-01 4.94472593e-01
3.49656701e-01 3.87019783e-01 -1.28675938e-01 1.35562682e+00
-2.44195744e-01 9.02542412e-01 -9.86249506e-01 -3.78588110e-01
3.11219335e-01 -7.30786979e-01 1.06473172e+00 1.17195070e+00
4.74148184e-01 -1.57502085e-01 1.09593742e-01 5.21341741e-01
-1.10900491e-01 4.78153348e-01 -2.36628205e-02 1.52475193e-01
9.19916406e-02 1.12151587e+00 -7.82104313e-01 -2.03454509e-01
-1.59028262e-01 9.73337829e-01 4.67353076e-01 4.22846287e-01
-7.51013756e-01 -3.30559611e-01 6.11556470e-01 3.73681895e-02
4.95542049e-01 -4.34731096e-01 -2.88935423e-01 -1.25886679e+00
6.51723385e-01 -5.45053661e-01 5.24956882e-01 -8.08983892e-02
-1.24652183e+00 3.32677394e-01 3.94147784e-02 -8.02567899e-01
-2.46461526e-01 -7.41600215e-01 -5.64628243e-01 6.49565876e-01
-1.44103575e+00 -5.94004333e-01 7.78482184e-02 5.28789103e-01
3.67397994e-01 1.12543248e-01 3.35318774e-01 2.13881195e-01
-5.42090595e-01 6.19659007e-01 8.43444049e-01 -9.51321945e-02
4.46294695e-01 -1.45898592e+00 -1.19296826e-01 9.79090869e-01
1.80962473e-01 4.95297521e-01 1.15680587e+00 -3.36115330e-01
-1.43769336e+00 -6.78235173e-01 7.99579263e-01 -7.92438239e-02
8.61048102e-01 -3.58509988e-01 -7.92225480e-01 3.82986277e-01
-5.21374583e-01 2.19217688e-01 1.37399331e-01 2.99740970e-01
1.53149992e-01 -1.85071871e-01 -1.03274179e+00 4.07529145e-01
8.41700077e-01 -6.59593865e-02 -2.56205469e-01 6.89491391e-01
2.17976063e-01 -6.92748308e-01 -4.21777606e-01 2.69422859e-01
3.71708781e-01 -9.84003961e-01 8.66269946e-01 -7.04595923e-01
1.03928469e-01 -2.29883119e-01 2.21695565e-02 -1.31588280e+00
1.32059947e-01 -1.07273328e+00 -2.98972279e-01 9.41473484e-01
3.16844583e-01 -9.88905966e-01 8.66610587e-01 3.29856813e-01
-7.78096840e-02 -1.31344390e+00 -1.21033514e+00 -1.18231499e+00
3.44231308e-01 -6.02685392e-01 -1.46200657e-01 7.41580009e-01
-8.00671279e-02 -1.61692604e-01 -3.85931224e-01 8.19070041e-02
5.31432211e-01 1.32125929e-01 7.33624876e-01 -9.24648106e-01
-8.39034975e-01 -7.52706945e-01 -1.85216799e-01 -1.32560313e+00
9.35635269e-02 -9.08995271e-01 3.63438129e-02 -1.29423964e+00
2.37874568e-01 -6.26656711e-01 -3.04103792e-01 1.18336365e-01
-3.94496977e-01 -4.98183928e-02 2.02853933e-01 2.61987448e-01
-5.19163132e-01 6.01928890e-01 1.01727057e+00 1.61293045e-01
-6.85619891e-01 7.47634113e-01 -4.29960430e-01 8.93031895e-01
7.72665381e-01 -6.05444610e-01 -1.37831077e-01 -2.57668316e-01
8.44906807e-01 -3.30664307e-01 1.68342918e-01 -8.92317235e-01
1.60633013e-01 -1.39524385e-01 -2.37959057e-01 -2.25564867e-01
8.71103704e-02 -4.09062505e-01 -3.99767816e-01 5.81291258e-01
-4.47534680e-01 4.11604084e-02 -3.89784314e-02 6.34976685e-01
2.03832369e-02 -7.34057963e-01 9.43368375e-01 1.34551316e-01
1.02379158e-01 3.46037656e-01 -2.32450783e-01 5.16594529e-01
7.54925191e-01 1.23511724e-01 2.09974810e-01 -5.51579714e-01
-8.35028172e-01 3.04365695e-01 3.33867580e-01 -1.25300646e-01
2.30786219e-01 -1.00740266e+00 -8.10706258e-01 -1.74432755e-01
-4.94225085e-01 -5.76076731e-02 2.99171545e-04 1.55302596e+00
-4.81963366e-01 1.40241206e-01 3.99692744e-01 -4.90020424e-01
-1.34270501e+00 3.66506904e-01 5.25587201e-01 -6.94855988e-01
-5.43746889e-01 1.13776505e+00 -7.31927827e-02 -3.23195070e-01
3.92079741e-01 -4.64993507e-01 2.63426602e-01 -1.53397312e-02
4.79084432e-01 6.26380801e-01 1.05472766e-01 -1.57618448e-01
-1.65195227e-01 7.32644439e-01 1.11126073e-01 -4.31669474e-01
1.49214935e+00 -2.57220298e-01 -2.01763839e-01 5.90211928e-01
1.30619264e+00 5.73000312e-01 -1.27592850e+00 -2.00418368e-01
1.11752667e-01 -3.11413199e-01 -1.20462663e-01 -3.34506094e-01
-1.04613352e+00 7.37461448e-01 4.34453875e-01 3.83665562e-01
1.03087437e+00 -5.16888537e-02 7.10325420e-01 6.08482778e-01
3.04019839e-01 -1.45798206e+00 -2.07895234e-01 6.69488192e-01
7.17813492e-01 -1.08759451e+00 1.39089897e-01 -2.85923421e-01
-4.06985343e-01 1.40501797e+00 6.82014376e-02 -1.60183921e-01
6.47962451e-01 2.38790646e-01 -2.23805755e-01 2.34111771e-01
-4.40019071e-01 -2.32882008e-01 1.61107600e-01 1.89054772e-01
2.48107165e-01 -5.46510108e-02 -9.39140618e-01 4.80856478e-01
-2.78422028e-01 -7.32674897e-02 4.41110343e-01 7.08884776e-01
-5.20066261e-01 -1.06028914e+00 -3.96764338e-01 2.78605372e-01
-6.76045895e-01 -9.90676880e-02 -6.73915297e-02 7.92869091e-01
-2.89978117e-01 6.39405966e-01 -4.28025097e-01 1.71235263e-01
1.15532435e-01 1.05342017e-02 7.58307874e-01 -1.83148533e-01
-4.48103637e-01 -7.55644217e-02 -1.75476909e-01 -4.07494664e-01
-1.86656833e-01 -7.52409697e-01 -1.16419184e+00 -1.45556435e-01
-5.68417966e-01 5.31935871e-01 8.08722079e-01 1.10449648e+00
6.27800375e-02 -5.52665144e-02 8.48429978e-01 -6.75240815e-01
-1.21460164e+00 -7.29975283e-01 -7.12348759e-01 1.62678406e-01
2.58547127e-01 -3.84662539e-01 -1.06256604e+00 -3.12781632e-01] | [6.870411396026611, 4.292517185211182] |
e44e2698-6dea-4930-a67e-0f3c5b22d6b2 | discrete-time-convolution-for-fast-event | null | null | http://openaccess.thecvf.com//content/CVPR2022/html/Zhang_Discrete_Time_Convolution_for_Fast_Event-Based_Stereo_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Zhang_Discrete_Time_Convolution_for_Fast_Event-Based_Stereo_CVPR_2022_paper.pdf | Discrete Time Convolution for Fast Event-Based Stereo | Inspired by biological retina, dynamical vision sensor transmits events of instantaneous changes of pixel intensity, giving it a series of advantages over traditional frame-based camera, such as high dynamical range, high temporal resolution and low power consumption. However, extracting information from highly asynchronous event data is a challenging task. Inspired by continuous dynamics of biological neuron models, we propose a novel encoding method for sparse events - continuous time convolution (CTC) - which learns to model the spatial feature of the data with intrinsic dynamics. Adopting channel-wise parameterization, temporal dynamics of the model is synchronized on the same feature map and diverges across different ones, enabling it to embed data in a variety of temporal scales. Abstracted from CTC, we further develop discrete time convolution (DTC) which accelerates the process with lower computational cost. We apply these methods to event-based multi-view stereo matching where they surpass state-of-the-art methods on benchmark criteria of the MVSEC dataset. Spatially sparse event data often leads to inaccurate estimation of edges and local contours. To address this problem, we propose a dual-path architecture in which the feature map is complemented by underlying edge information from original events extracted with spatially-adaptive denormalization. We demonstrate the superiority of our model in terms of speed (up to 110 FPS), accuracy and robustness, showing a great potential for real-time fast depth estimation. Finally, we perform experiments on the recent DSEC dataset to demonstrate the general usage of our model. | ['Luziwei Leng', 'Qinghai Guo', 'Ziyang Zhang', 'Jie Cheng', 'JianGuo Zhang', 'Kaiwei Che', 'Kaixuan Zhang'] | 2022-01-01 | null | null | null | cvpr-2022-1 | ['stereo-matching-1'] | ['computer-vision'] | [ 3.53299588e-01 -4.31231171e-01 3.89766991e-01 -1.59696817e-01
-3.94533157e-01 -4.53496218e-01 5.60815394e-01 -1.24415914e-02
-7.13847399e-01 7.32738376e-01 -5.03526302e-03 3.23897392e-01
-1.37279049e-01 -7.76326954e-01 -8.21382999e-01 -9.64043438e-01
-4.47475612e-02 -1.48628145e-01 7.14604437e-01 8.83469582e-02
2.98300743e-01 6.58033788e-01 -1.87973046e+00 2.22625434e-01
6.03543103e-01 1.21389031e+00 4.10156161e-01 6.33908153e-01
5.92616014e-02 7.81565905e-01 -2.46620283e-01 -2.46697459e-02
1.08221821e-01 -2.48808473e-01 -2.67015874e-01 7.73952156e-02
4.23923403e-01 -4.96267051e-01 -8.33156586e-01 8.33723545e-01
5.51251352e-01 1.99266616e-02 3.70409131e-01 -1.00857377e+00
-1.94600135e-01 -1.14483111e-01 -5.79617500e-01 6.33813918e-01
2.05771446e-01 2.27271497e-01 5.62227428e-01 -8.16039562e-01
1.07518983e+00 9.35934126e-01 6.68075442e-01 5.96651316e-01
-1.29807627e+00 -3.95157307e-01 2.52488911e-01 2.70234108e-01
-1.16753125e+00 -4.95347828e-01 7.67023861e-01 -3.69843066e-01
1.04013848e+00 -6.43052459e-02 1.06739902e+00 1.08880389e+00
6.31506324e-01 6.86403871e-01 1.09675372e+00 7.84838852e-03
3.55054736e-01 -5.51027715e-01 -1.57668784e-01 6.00543499e-01
6.58266097e-02 4.57040846e-01 -1.02211428e+00 1.96764711e-02
1.22996473e+00 2.84161031e-01 -7.16204166e-01 -2.90527374e-01
-1.51379251e+00 3.93970609e-01 3.38340163e-01 1.66147292e-01
-5.82048059e-01 5.65548301e-01 2.66283214e-01 8.12873617e-02
3.54448140e-01 -2.15366244e-01 -2.98910022e-01 -2.79056847e-01
-8.35375190e-01 1.15815230e-01 5.41318893e-01 6.95093095e-01
6.34994805e-01 5.47737963e-02 -7.17984438e-02 3.35004747e-01
9.00982097e-02 4.48247850e-01 5.88487148e-01 -1.11078286e+00
-9.47738718e-03 4.03676331e-01 1.97092980e-01 -8.20366502e-01
-4.68930364e-01 -2.15070605e-01 -1.14110696e+00 5.15312016e-01
4.91725087e-01 9.01625231e-02 -8.42114091e-01 1.82410884e+00
4.26342547e-01 8.74625206e-01 1.15416341e-01 1.02477658e+00
5.07504404e-01 6.91983879e-01 -2.23614872e-01 -4.87957031e-01
1.31189001e+00 -3.27165574e-01 -7.22145557e-01 -7.60284439e-02
5.13025224e-02 -3.77616167e-01 5.69780171e-01 5.24134278e-01
-1.12135482e+00 -4.43680763e-01 -8.51328790e-01 -7.27374852e-02
-1.48025140e-01 -1.08793400e-01 7.47046590e-01 2.58907694e-02
-1.15707672e+00 7.06573308e-01 -1.24837029e+00 -3.15900803e-01
4.14942443e-01 3.79270673e-01 -4.70956564e-01 -1.09883295e-02
-8.28733981e-01 4.29748774e-01 2.53325254e-01 9.85403880e-02
-7.54158795e-01 -8.09954822e-01 -8.46063197e-01 -5.23676835e-02
-3.51508670e-02 -9.51933682e-01 9.57424939e-01 -8.18606675e-01
-1.60426891e+00 7.49139607e-01 -4.84081089e-01 -6.56029403e-01
5.16967714e-01 -4.67035249e-02 -1.98877141e-01 4.54256713e-01
-4.69089746e-02 7.55486906e-01 8.46609950e-01 -7.40867794e-01
-7.42669523e-01 -3.69193465e-01 -1.39264196e-01 1.43451512e-01
-4.59019333e-01 -1.64297804e-01 -5.86375237e-01 -6.40443802e-01
2.50311702e-01 -6.38942182e-01 -1.98362187e-01 5.27168751e-01
2.40171477e-01 3.40947136e-02 8.14918101e-01 -3.41976285e-01
1.14414060e+00 -2.36976433e+00 3.31241995e-01 -3.38182330e-01
4.05880809e-01 -8.98770094e-02 6.98640049e-02 2.83876419e-01
1.25283778e-01 -4.86417651e-01 -3.66013914e-01 -3.77370209e-01
-4.69406366e-01 3.10417235e-01 -4.28192556e-01 7.11187601e-01
4.58565205e-01 9.57440734e-01 -1.04724419e+00 -4.07298774e-01
5.45597732e-01 7.73714662e-01 -3.91792864e-01 7.11801052e-02
-4.69661579e-02 6.71017289e-01 -3.44920695e-01 5.59911549e-01
7.15752900e-01 -5.49986959e-01 -1.95224255e-01 -3.96306008e-01
-5.22931516e-01 -1.57704383e-01 -1.30126894e+00 2.06834102e+00
-3.18006665e-01 9.46275711e-01 1.04123168e-02 -8.03167403e-01
7.09525585e-01 2.19565690e-01 6.65991187e-01 -8.47203612e-01
1.26762405e-01 1.94216639e-01 -2.78867900e-01 -3.94851476e-01
3.33451629e-01 5.40733635e-02 1.81765392e-01 1.23224258e-01
8.51932466e-02 -7.59616345e-02 3.91879603e-02 6.81499615e-02
1.41100907e+00 2.32412785e-01 2.75344491e-01 -7.92366713e-02
3.91507298e-01 -2.91375577e-01 7.09395647e-01 5.05308926e-01
-2.30217859e-01 8.22383642e-01 1.44217387e-01 -5.65280497e-01
-8.79077911e-01 -1.13913226e+00 -3.61397475e-01 2.87585676e-01
5.98421693e-01 -2.49227211e-01 -5.32787681e-01 -1.08082838e-01
-3.77958175e-03 8.78181607e-02 -6.34130895e-01 7.15790838e-02
-7.01252341e-01 -7.23289728e-01 1.78040087e-01 5.31964123e-01
6.12433374e-01 -8.52496803e-01 -1.41262496e+00 6.32198691e-01
-1.88804224e-01 -1.55180776e+00 -3.75348330e-01 1.47234172e-01
-1.11273956e+00 -9.62706923e-01 -7.60109663e-01 -5.40482044e-01
4.93312001e-01 4.20482934e-01 9.55294967e-01 -2.48352602e-01
-6.09940767e-01 6.30113184e-01 -6.31828904e-02 -1.28709078e-01
1.54156417e-01 -4.98893082e-01 -1.02356598e-01 3.85045439e-01
6.05870821e-02 -1.01445222e+00 -1.09532499e+00 2.38082126e-01
-1.24611521e+00 3.72488499e-01 4.43487257e-01 9.56227899e-01
9.10403728e-01 -1.50936604e-01 3.51152420e-01 -3.50625008e-01
7.24645797e-03 -3.57374638e-01 -9.30988550e-01 -2.35478077e-02
-2.87182540e-01 -1.98137593e-02 5.37016153e-01 -7.79839814e-01
-1.06188333e+00 4.20105517e-01 2.54429013e-01 -7.50032425e-01
1.16221178e-02 2.07150787e-01 2.77012885e-01 -3.82918209e-01
4.03218716e-01 5.97289085e-01 7.59647191e-02 -3.07667330e-02
7.99988657e-02 2.88294494e-01 9.63862658e-01 -4.00671661e-01
4.23782617e-01 1.32874930e+00 1.88653886e-01 -7.08547056e-01
-3.89231145e-01 -5.08226573e-01 -4.79386955e-01 -5.23272157e-01
8.26543212e-01 -1.17735016e+00 -9.81093645e-01 9.94567692e-01
-1.51819956e+00 -3.61948043e-01 -3.89505684e-01 6.37152612e-01
-8.17568660e-01 3.73308569e-01 -8.46892774e-01 -7.08243310e-01
-1.78591743e-01 -8.07689905e-01 1.37051857e+00 5.30441582e-01
1.94281369e-01 -9.76316631e-01 7.82843828e-02 -2.68338978e-01
3.46288115e-01 6.26679182e-01 4.01643187e-01 1.88413456e-01
-1.14795816e+00 7.87463412e-02 -2.61058569e-01 1.84718724e-02
-4.12829556e-02 -3.56945605e-03 -1.08812642e+00 -2.21593440e-01
3.25580686e-01 -4.72226851e-02 1.10480368e+00 8.14130366e-01
9.99213696e-01 2.80497581e-01 -5.52964866e-01 9.79174495e-01
1.85872662e+00 9.02721360e-02 8.42901289e-01 1.76966891e-01
3.66538912e-01 4.74210322e-01 4.75724638e-01 7.52594292e-01
3.39008391e-01 6.85465753e-01 6.23034835e-01 -3.16271260e-02
-2.53038317e-01 2.38146051e-03 5.23377955e-01 5.93286991e-01
-2.49545619e-01 -2.92711914e-01 -6.23407304e-01 7.84261107e-01
-2.07407880e+00 -1.02858543e+00 -1.59538627e-01 2.25611472e+00
6.60915017e-01 4.76744063e-02 -2.11258337e-01 2.52890140e-02
6.67946100e-01 1.67592034e-01 -8.57582569e-01 2.24003971e-01
-4.73493308e-01 2.47915581e-01 4.94036853e-01 2.29993120e-01
-9.52483833e-01 7.53733635e-01 5.56693411e+00 6.68354332e-01
-1.36959231e+00 1.20007761e-01 4.55292255e-01 -3.10746312e-01
-9.06068981e-02 -2.28607897e-02 -8.00177872e-01 6.48747206e-01
9.04507875e-01 -1.25246465e-01 3.72268498e-01 2.45555654e-01
3.65726709e-01 -4.12514746e-01 -1.15768862e+00 1.37263131e+00
-1.69419497e-02 -1.73568261e+00 -1.65303439e-01 1.71734914e-01
6.92235470e-01 2.57788062e-01 -2.28580296e-01 -3.48366767e-01
7.14358389e-02 -7.23160625e-01 8.18740249e-01 8.18657637e-01
7.87356019e-01 -4.23975646e-01 3.66831541e-01 3.06412250e-01
-1.50061357e+00 -1.15253247e-01 -3.97765636e-01 -1.07755676e-01
5.81706345e-01 8.04534316e-01 -3.88972796e-02 5.81349730e-01
8.69374573e-01 1.30850708e+00 -1.84140697e-01 1.09523177e+00
2.34984219e-01 2.08767429e-01 -6.57253444e-01 2.17526972e-01
-6.72652125e-02 -2.30655193e-01 6.22865260e-01 1.07725024e+00
5.81685841e-01 4.80333149e-01 -3.03259343e-01 9.32138920e-01
2.26409987e-01 -4.73972887e-01 -7.53407359e-01 3.35353643e-01
4.20003563e-01 1.24274755e+00 -8.73016357e-01 -2.10689753e-01
-7.15411663e-01 1.36456835e+00 3.57119888e-01 4.56415504e-01
-8.55566800e-01 -2.22701728e-01 7.83752918e-01 5.77273406e-02
5.64121068e-01 -3.94025803e-01 1.17827738e-02 -1.40224719e+00
2.81951547e-01 -1.98254809e-01 1.81725740e-01 -8.86566997e-01
-1.10893774e+00 5.63594759e-01 -2.59426951e-01 -1.62031388e+00
-1.64863929e-01 -5.14258564e-01 -4.62398201e-01 6.09531760e-01
-1.88914132e+00 -9.26824391e-01 -7.07466066e-01 9.35567915e-01
5.13196826e-01 2.77402788e-01 5.39383650e-01 3.45058709e-01
-2.27353305e-01 4.58778180e-02 1.49333954e-01 -6.32567052e-03
5.76520622e-01 -9.27902520e-01 4.26145226e-01 9.91605222e-01
1.60894379e-01 1.15624838e-01 4.03373569e-01 -5.21253049e-01
-1.64004350e+00 -1.00253487e+00 5.60105443e-01 -2.61781365e-01
7.73594379e-01 -3.16959798e-01 -9.41344798e-01 3.53619039e-01
-1.09091438e-02 6.97962821e-01 1.33463949e-01 -8.45983565e-01
-8.67099762e-02 -2.83059090e-01 -9.98883963e-01 3.82801920e-01
1.33470702e+00 -5.63188136e-01 -2.61454970e-01 -3.25931460e-02
6.81427717e-01 -6.50702596e-01 -8.43002439e-01 3.58791113e-01
6.75244689e-01 -1.33477795e+00 9.92480576e-01 -4.32861736e-03
3.47921520e-01 -4.95864779e-01 -1.12995021e-01 -8.55751872e-01
-1.74457923e-01 -8.55665267e-01 -4.12688702e-01 1.14611471e+00
-3.21226925e-01 -8.89095664e-01 6.08138144e-01 3.10994744e-01
-1.31552458e-01 -7.19411910e-01 -1.41149056e+00 -6.96568191e-01
-4.72713649e-01 -4.08479899e-01 1.14423715e-01 6.49106562e-01
-3.66962999e-01 -1.74767986e-01 -1.39448509e-01 3.75406981e-01
8.81032705e-01 1.73959285e-01 3.73838693e-01 -1.27854204e+00
-2.57860661e-01 -1.74104840e-01 -9.90643203e-01 -1.28062570e+00
3.90556082e-02 -4.66954172e-01 6.62268884e-03 -1.20246756e+00
1.37490913e-01 -5.94026633e-02 -3.51282001e-01 6.37672096e-02
-2.20203381e-02 2.63512075e-01 4.97050993e-02 3.02038521e-01
-3.47618222e-01 6.26487434e-01 1.19114637e+00 1.98324248e-01
-1.00546569e-01 -3.77929986e-01 3.51866111e-02 6.42585635e-01
2.07923532e-01 -4.18910950e-01 -3.53314072e-01 -5.24002731e-01
1.21717326e-01 4.82320786e-01 8.73964012e-01 -1.29051304e+00
8.17582667e-01 -4.03406918e-02 3.64241183e-01 -4.72917885e-01
6.78399801e-01 -9.38890040e-01 4.44555938e-01 4.83169764e-01
8.69078562e-02 1.43665150e-01 3.23020071e-01 1.07749546e+00
-5.14620721e-01 2.03830689e-01 8.62740815e-01 3.52325626e-02
-1.00746429e+00 5.75710237e-01 -3.78502041e-01 3.78639698e-02
1.16567290e+00 -5.96773982e-01 -2.93411165e-01 -1.28122196e-01
-4.98118788e-01 -6.05751052e-02 6.98783338e-01 6.34887591e-02
8.97939086e-01 -1.25602877e+00 -6.28883541e-01 3.81571531e-01
6.97519556e-02 2.64721394e-01 5.86915910e-01 1.08150721e+00
-4.14680541e-01 1.65906265e-01 -3.38566095e-01 -1.17730546e+00
-9.99304950e-01 3.14440191e-01 4.33672786e-01 3.80845666e-02
-1.10007930e+00 6.76339030e-01 5.29040277e-01 3.74038488e-01
5.67916892e-02 -6.31235003e-01 -5.81591949e-03 -1.98713224e-02
6.67062640e-01 1.74276784e-01 -1.49063878e-02 -3.05225044e-01
-3.03053498e-01 1.17207611e+00 1.78607821e-01 -2.29798064e-01
1.51598895e+00 -3.44109595e-01 -2.69617047e-02 5.28267622e-01
1.04956484e+00 -2.49271065e-01 -2.08195066e+00 -3.24831903e-01
-2.43749961e-01 -4.81663167e-01 2.23597139e-01 -2.75240988e-01
-1.09353614e+00 9.05036092e-01 9.14695263e-01 -7.86993131e-02
1.53505719e+00 -2.45280471e-02 9.25851703e-01 -2.49451790e-02
5.94805837e-01 -8.12957048e-01 7.14160576e-02 3.45421761e-01
6.06447995e-01 -1.21381712e+00 -2.50724941e-01 -4.19974148e-01
-2.09202647e-01 1.38808823e+00 3.88393819e-01 -3.25127751e-01
5.84304214e-01 6.68036222e-01 -2.58015573e-01 -1.90189064e-01
-1.09693539e+00 -2.11125106e-01 -9.36332643e-02 5.62274873e-01
-3.03687733e-02 -4.07040715e-01 -8.52033198e-02 2.30618134e-01
3.86146039e-01 4.97747660e-01 6.29651010e-01 9.59740162e-01
-1.62141547e-01 -7.14595139e-01 -8.18340108e-02 1.28943205e-01
-2.37939849e-01 -1.17326498e-01 2.92454660e-01 7.08170533e-01
-4.91901338e-02 5.79911888e-01 4.61469740e-01 -1.36260763e-01
2.13857591e-01 -1.71695471e-01 7.19375312e-01 -1.44983575e-01
-3.44224036e-01 2.14767814e-01 -3.52878243e-01 -1.04225993e+00
-8.94280851e-01 -8.97311509e-01 -1.50520229e+00 -1.93394452e-01
-1.09077364e-01 -3.85057062e-01 5.70896566e-01 7.66961217e-01
7.25052774e-01 5.50896287e-01 4.67898756e-01 -1.15360224e+00
-1.43532380e-01 -4.73754972e-01 -4.66478258e-01 4.16732818e-01
6.55255973e-01 -6.78102612e-01 -5.13369501e-01 4.28627104e-01] | [8.722952842712402, -1.2630970478057861] |
887b19cd-91d2-4922-8434-bbf06c5e0aef | real-time-and-efficient-method-for-accuracy | 1407.6498 | null | http://arxiv.org/abs/1407.6498v1 | http://arxiv.org/pdf/1407.6498v1.pdf | Real-Time and Efficient Method for Accuracy Enhancement of Edge Based License Plate Recognition System | License Plate Recognition plays an important role on the traffic monitoring
and parking management. Administration and restriction of those transportation
tools for their better service becomes very essential. In this paper, a fast
and real time method has an appropriate application to find plates that the
plat has tilt and the picture quality is poor. In the proposed method, at the
beginning, the image is converted into binary mode with use of adaptive
threshold. And with use of edge detection and morphology operation, plate
number location has been specified and if the plat has tilt; its tilt is
removed away. Then its characters are distinguished using image processing
techniques. Finally, K Nearest Neighbour (KNN) classifier was used for
character recognition. This method has been tested on available data set that
has different images of the background, considering distance, and angel of view
so that the correct extraction rate of plate reached at 98% and character
recognition rate achieved at 99.12%. Further we tested our character
recognition stage on Persian vehicle data set and we achieved 99% correct
recognition rate. | ['Hamid Reza Shayegh', 'Reza Azad', 'Babak Azad'] | 2014-07-24 | null | null | null | null | ['license-plate-recognition'] | ['computer-vision'] | [ 2.09669173e-01 -4.53135312e-01 -2.96839271e-02 -1.60377175e-01
-7.39024356e-02 -5.64832687e-01 4.61489767e-01 8.94367099e-02
-6.92518055e-01 8.27065289e-01 -4.67988312e-01 -4.81532782e-01
2.79723736e-03 -8.88141990e-01 -1.77304804e-01 -7.73267090e-01
3.82774293e-01 7.88928211e-01 6.76358998e-01 -1.49298698e-01
9.60380793e-01 1.05546868e+00 -1.52776086e+00 -1.28610194e-01
6.67066514e-01 6.75242722e-01 3.59165251e-01 8.64546597e-01
-1.44075975e-01 5.47876537e-01 -4.19474334e-01 -1.27156287e-01
4.80300337e-01 -9.33802500e-02 -5.28145850e-01 4.77284312e-01
-2.19419107e-01 -3.97503585e-01 9.45349708e-02 1.06058252e+00
2.64422953e-01 1.93394184e-01 1.25684726e+00 -1.24282587e+00
-3.40963840e-01 -2.81199127e-01 -7.21455574e-01 4.54001397e-01
-1.04282610e-01 1.37434095e-01 2.75420308e-01 -1.13227856e+00
4.34257716e-01 5.53842485e-01 4.82372463e-01 1.51438475e-01
-7.55561411e-01 -6.43473029e-01 -6.32495403e-01 6.96510792e-01
-1.60118771e+00 -4.79556471e-01 6.26969337e-01 -4.76616532e-01
7.39756703e-01 2.01914817e-01 4.63873357e-01 -1.50975972e-01
2.75892407e-01 2.05091432e-01 1.21354163e+00 -7.21855104e-01
4.53875065e-02 5.70959866e-01 3.08566779e-01 5.36741853e-01
4.02900875e-01 -3.36180836e-01 2.48624161e-01 3.95836383e-01
4.89094317e-01 1.60758272e-01 1.17891900e-01 2.07843572e-01
-6.95772171e-01 5.88244140e-01 -1.43570051e-01 6.05926991e-01
-5.43306231e-01 -5.26151597e-01 2.16104224e-01 -1.88140109e-01
-3.05116355e-01 -4.33798879e-02 -7.62218889e-03 -3.38266283e-01
-1.05248034e+00 -2.01049909e-01 5.61786652e-01 8.34477007e-01
6.61833227e-01 8.35199952e-02 4.00849223e-01 7.97682762e-01
2.07037702e-01 8.31999421e-01 4.87956911e-01 -3.54740262e-01
4.58222002e-01 8.10738087e-01 3.58620673e-01 -1.41089749e+00
-1.32816598e-01 3.88643108e-02 -4.78091389e-01 7.97943175e-01
3.91369373e-01 -1.35308817e-01 -1.17538524e+00 6.40581310e-01
1.06316611e-01 2.41749406e-01 3.49623591e-01 7.31302619e-01
4.69347119e-01 1.08946371e+00 6.10290654e-02 -1.25108406e-01
1.73905373e+00 -5.16440153e-01 -6.63904309e-01 -2.48262033e-01
1.28973439e-01 -1.36309433e+00 7.18945026e-01 5.04220724e-01
-8.35452020e-01 -4.28496063e-01 -1.27980316e+00 4.62141126e-01
-6.09692812e-01 7.37290621e-01 -1.37939993e-02 7.60085225e-01
-6.98075175e-01 1.20625168e-01 -5.46432078e-01 -7.22387850e-01
3.20119262e-02 5.71187437e-01 -4.99638140e-01 -1.82754621e-01
-6.73081398e-01 1.22134781e+00 3.37032676e-01 4.09931391e-01
-1.11096956e-01 2.99463898e-01 -4.14355367e-01 -2.16568589e-01
7.65817836e-02 7.66318291e-02 6.07075393e-01 -8.70389581e-01
-1.26274085e+00 9.02459860e-01 -1.13857858e-01 -1.44456908e-01
4.51705486e-01 4.89318699e-01 -7.58284569e-01 2.95110673e-01
7.44815171e-02 7.20928609e-02 6.99647546e-01 -7.92683542e-01
-1.04414093e+00 -4.98825371e-01 -6.63631618e-01 6.48718476e-01
9.41495690e-03 3.29644918e-01 -5.33284664e-01 1.24617405e-01
1.72212899e-01 -1.02536201e+00 1.64533317e-01 -4.96775359e-01
-3.08892012e-01 -1.16411239e-01 1.13294291e+00 -9.58298087e-01
1.00664771e+00 -2.19291544e+00 -6.81113482e-01 8.12879860e-01
-2.87613004e-01 7.91719019e-01 6.97206855e-01 3.88409525e-01
1.01814888e-01 6.73048198e-02 -2.40097925e-01 2.23999158e-01
-2.75594950e-01 -1.93443510e-03 3.26674819e-01 5.89136600e-01
2.15450689e-01 -1.03724331e-01 1.66338142e-02 -9.60375667e-01
5.64382255e-01 3.36576968e-01 2.05692202e-02 1.68410186e-02
5.38027585e-01 -5.98016456e-02 -5.45962155e-01 7.86809146e-01
1.22574115e+00 2.76996642e-01 -1.22581691e-01 -3.17494720e-01
-6.13069415e-01 -5.41416049e-01 -1.51751101e+00 1.55096069e-01
-2.29906097e-01 9.15894866e-01 -5.57877030e-03 -9.31886852e-01
1.35837901e+00 2.81509966e-01 2.60740906e-01 -7.44944870e-01
3.01693082e-01 4.10872310e-01 6.17641769e-02 -8.41302633e-01
8.09520900e-01 9.67314094e-02 3.34218562e-01 -7.94015154e-02
-6.79406941e-01 1.45782784e-01 3.01434278e-01 -2.11270943e-01
5.34781396e-01 -2.72610277e-01 3.41145784e-01 -7.70294219e-02
1.06292284e+00 5.50945461e-01 1.36695638e-01 2.49634132e-01
-4.27435845e-01 3.26477885e-01 9.58916843e-02 2.22663283e-02
-1.34181452e+00 -6.35599017e-01 -4.65645820e-01 2.87001431e-01
2.41068944e-01 5.65001070e-01 -6.07253134e-01 7.14309067e-02
-1.45964056e-01 4.89656597e-01 -1.72539786e-01 2.35475361e-01
-7.77220964e-01 -5.24681568e-01 3.96943539e-01 1.51416091e-02
9.24674928e-01 -1.16328859e+00 -6.90510452e-01 1.01278074e-01
3.17045569e-01 -9.00810182e-01 -2.73179784e-02 -2.24555373e-01
-7.30415881e-01 -1.35320067e+00 -8.05282056e-01 -1.21037769e+00
8.64446819e-01 3.41915965e-01 2.79107749e-01 2.79207200e-01
-2.80876219e-01 -2.39319518e-01 -2.19370693e-01 -2.82357365e-01
-5.00063479e-01 -2.88915962e-01 -2.48417303e-01 1.17918119e-01
6.54361904e-01 -2.70738691e-01 -6.66697800e-01 5.33858716e-01
-6.16373897e-01 -2.26351783e-01 8.29942584e-01 2.36464977e-01
2.36444205e-01 7.62874126e-01 3.62719536e-01 -6.13402188e-01
5.63518345e-01 -4.71193463e-01 -8.71337354e-01 -1.12909991e-02
-6.75442457e-01 -3.11097115e-01 7.55957842e-01 1.43854246e-01
-9.34101760e-01 7.81965815e-03 -1.57613397e-01 1.12678148e-01
-5.31461358e-01 2.73616135e-01 -2.04209149e-01 -1.11459002e-01
2.17098251e-01 6.57084823e-01 3.73747647e-01 -1.46888256e-01
-4.72984433e-01 1.21457171e+00 5.98161697e-01 2.05261990e-01
7.41344154e-01 2.46752918e-01 3.15321892e-01 -1.21113420e+00
6.93269134e-01 -7.92928219e-01 -4.64426666e-01 -6.14159882e-01
1.20346153e+00 -3.45339864e-01 -8.14706624e-01 8.67101610e-01
-9.66595829e-01 5.31528950e-01 8.98051560e-01 6.97524965e-01
-1.15554854e-01 4.72384542e-01 -1.97538972e-01 -1.30812931e+00
-2.20010728e-01 -1.14085686e+00 1.62760571e-01 6.23083591e-01
1.62848637e-01 -7.75706768e-01 -3.25571477e-01 4.08472240e-01
3.69193047e-01 4.93847840e-02 8.96918237e-01 -9.20369744e-01
-4.75999057e-01 -8.77435684e-01 -4.27628398e-01 3.06197494e-01
3.37180257e-01 4.42621797e-01 -5.44695139e-01 3.52617830e-01
-2.41223291e-01 3.30882221e-01 3.79460514e-01 2.98044175e-01
4.31520760e-01 -2.04084679e-01 -3.60657930e-01 1.91255048e-01
1.83378470e+00 1.08380687e+00 1.17990124e+00 7.72293746e-01
4.21104491e-01 3.92252386e-01 9.60579753e-01 2.90743113e-01
1.30477995e-01 5.58707118e-01 1.95784852e-01 -8.98891687e-02
9.33740512e-02 1.49486512e-01 3.07483584e-01 4.43258435e-01
-3.07543576e-01 -3.42150182e-01 -1.05921757e+00 4.17910784e-01
-1.28356004e+00 -1.01912403e+00 -7.10278630e-01 2.28599358e+00
3.75105560e-01 2.84637123e-01 2.90442854e-01 7.25202799e-01
1.37000513e+00 -3.61805350e-01 -1.61011726e-01 -8.46646130e-01
-3.66402138e-03 -6.85942397e-02 1.05099428e+00 6.37699008e-01
-7.61222959e-01 7.34921515e-01 4.71831560e+00 8.34190726e-01
-1.47737455e+00 -4.14293736e-01 5.22885144e-01 5.51079810e-01
2.33917832e-01 -2.33844575e-02 -1.02938056e+00 8.47586513e-01
7.24007487e-01 9.20706689e-02 1.62956730e-01 6.05257273e-01
5.50958872e-01 -8.49351227e-01 -1.89168990e-01 9.79217231e-01
4.50756252e-02 -1.04225183e+00 -3.29000354e-01 2.70148814e-01
1.89655766e-01 -3.60464871e-01 8.24862123e-02 -1.00381136e-01
-3.03021222e-01 -8.46265316e-01 4.58001256e-01 6.32583618e-01
6.11392558e-01 -1.09776068e+00 1.03752685e+00 4.76553202e-01
-1.07411361e+00 6.96253181e-02 -3.75654876e-01 6.42381832e-02
1.00423038e-01 1.20616928e-01 -1.64891398e+00 8.91343206e-02
1.80526704e-01 2.55140394e-01 -5.03466666e-01 1.27614415e+00
4.30114269e-01 6.52177870e-01 -5.62199175e-01 -5.40894270e-01
4.73944992e-01 -7.36904740e-01 5.65208554e-01 1.24902463e+00
6.31863534e-01 2.06995681e-01 -3.20156366e-01 4.17565525e-01
5.71534336e-01 6.09281778e-01 -7.74765968e-01 1.64686307e-01
7.95361519e-01 9.95567262e-01 -1.20434391e+00 -2.29759991e-01
-2.38121018e-01 7.64116526e-01 -4.48691905e-01 3.54868583e-02
-7.96336949e-01 -8.41343999e-01 -3.89966369e-02 4.99529749e-01
3.68349612e-01 -2.05318779e-01 -3.24476123e-01 -3.63590270e-01
-7.20141008e-02 -5.40532708e-01 2.74588853e-01 -7.91437685e-01
-4.53854471e-01 4.27900076e-01 1.73236132e-02 -1.33310926e+00
-1.08590566e-01 -8.88556123e-01 -9.26862955e-01 1.11204886e+00
-1.05926716e+00 -8.50142777e-01 -6.36135936e-02 5.46118081e-01
6.12068594e-01 -5.55550814e-01 4.13243175e-01 4.38453615e-01
-8.33721042e-01 3.41121435e-01 4.62894171e-01 2.37801373e-01
4.78653282e-01 -8.49295497e-01 -3.25387180e-01 1.17215145e+00
-5.92538595e-01 4.01749760e-01 9.02576745e-01 -8.71857941e-01
-1.11840904e+00 -5.84885955e-01 1.03184116e+00 9.35638919e-02
3.64209831e-01 9.58290026e-02 -5.63728213e-01 2.90335149e-01
2.53371149e-01 -4.54177052e-01 3.73623759e-01 -6.64800763e-01
6.60795510e-01 -1.08598500e-01 -1.48421431e+00 5.60252905e-01
-6.05446734e-02 -1.60372928e-02 -2.75743246e-01 1.38539046e-01
-5.04349053e-01 -4.16889749e-02 -4.70295340e-01 -1.16810806e-01
5.85948646e-01 -1.11343670e+00 7.70375550e-01 -1.36668667e-01
2.42686681e-02 -7.85125732e-01 -5.56029230e-02 -5.79841375e-01
-2.97263891e-01 -1.03604890e-01 1.10543656e+00 1.51617348e+00
7.43466854e-01 -6.72369182e-01 9.66851592e-01 8.23662341e-01
-1.51397467e-01 -3.87852103e-01 -7.47148395e-01 -2.99243003e-01
-3.88486117e-01 1.36322239e-02 2.19169691e-01 6.60688162e-01
-2.97642916e-01 6.20720983e-02 -4.78445202e-01 3.71028066e-01
4.37580645e-01 -3.32526118e-01 6.29531741e-01 -9.32443500e-01
2.39477143e-01 -2.55633622e-01 -9.60417390e-01 -2.32731849e-01
-3.70987743e-01 -5.25620818e-01 4.10677046e-02 -1.61977899e+00
2.71554105e-02 -5.04473031e-01 8.00850429e-03 1.19899824e-01
5.26807681e-02 5.04298806e-01 1.14716977e-01 5.43270528e-01
-4.48978990e-02 -2.26040021e-01 8.52580905e-01 1.57093197e-01
-1.50208384e-01 5.25221765e-01 -1.20223179e-01 5.49337804e-01
1.01265669e+00 -2.59251803e-01 -3.18464339e-01 5.40618524e-02
1.11366399e-02 1.17800102e-01 1.42541274e-01 -1.13081336e+00
4.97670561e-01 -3.63167703e-01 6.11933470e-01 -1.04129672e+00
4.29728121e-01 -1.28727329e+00 2.58684874e-01 4.04001415e-01
3.73151332e-01 4.86757874e-01 1.99443921e-01 2.82181740e-01
-6.85167387e-02 -8.29269052e-01 8.09856832e-01 -2.89846808e-02
-1.21386349e+00 -1.50027305e-01 -1.07429612e+00 -5.08967876e-01
1.36724126e+00 -1.41683209e+00 -1.06325112e-01 -2.44967952e-01
-3.94318372e-01 -3.13944928e-02 6.46740258e-01 -2.84324467e-01
7.29133546e-01 -1.09369206e+00 -6.40284598e-01 1.34619042e-01
1.34834766e-01 -5.66643000e-01 2.05291569e-01 1.04795039e+00
-1.42338729e+00 4.70009327e-01 -6.95867538e-01 -4.41682339e-01
-1.66477787e+00 1.65450677e-01 1.79654032e-01 2.70171762e-01
-4.28609461e-01 2.33745545e-01 -6.82497263e-01 1.77340969e-01
-4.16246533e-01 1.84445996e-02 -9.85427022e-01 -1.10857390e-01
4.11073804e-01 7.26617038e-01 2.50179052e-01 -1.12737024e+00
-6.00239515e-01 1.17032635e+00 1.55250970e-02 -4.49311107e-01
1.05337095e+00 -1.43441617e-01 -5.14138825e-02 1.36438563e-01
1.08547187e+00 7.22390234e-01 -8.98722053e-01 3.29572707e-01
-4.10236567e-02 -7.32705295e-01 -1.29621103e-01 -7.37451613e-01
-6.10983014e-01 5.70224106e-01 7.50620544e-01 2.50405043e-01
1.06064332e+00 -6.42452002e-01 5.19972205e-01 3.02495658e-01
-5.39672412e-02 -1.44283950e+00 -6.52908266e-01 4.25300777e-01
2.68728763e-01 -1.31809783e+00 4.11168821e-02 -3.24267745e-01
-1.00141144e+00 1.36988223e+00 4.86070126e-01 -3.49341601e-01
6.10985518e-01 3.32819998e-01 -2.77269315e-02 -6.02418408e-02
-1.85154483e-01 -5.25909737e-02 -5.17658098e-03 7.08075047e-01
1.53083608e-01 8.37866366e-02 -6.39576495e-01 2.64753431e-01
-2.08585382e-01 1.15270957e-01 8.90402675e-01 1.08017528e+00
-1.16562653e+00 -6.84641540e-01 -9.32082295e-01 4.37388331e-01
-6.17661953e-01 -5.01819588e-02 -2.11269841e-01 1.24201357e+00
2.00610965e-01 1.08685482e+00 3.21990162e-01 -2.41171166e-01
1.62211835e-01 1.56804994e-01 2.22257406e-01 -1.14424266e-01
-4.92212698e-02 -7.05788210e-02 4.48347330e-01 3.03670615e-01
-1.62726045e-01 -8.52569699e-01 -1.71029711e+00 -5.36781192e-01
-2.91882724e-01 5.98131716e-01 1.25633025e+00 1.00655091e+00
-1.73642114e-01 -1.57667920e-01 7.77723074e-01 -3.19639653e-01
-1.44545361e-01 -7.69443095e-01 -6.11975014e-01 2.99706817e-01
1.29561946e-01 -6.26593113e-01 -3.87154907e-01 2.18777373e-01] | [9.794673919677734, -4.97521448135376] |
108a81cf-c358-4d1e-97d5-ccb8f4be1131 | direction-of-voice-dov-estimation-for | null | null | https://karan-ahuja.com/dov.html | https://karan-ahuja.com/assets/docs/paper/dov.pdf | Direction-of-Voice (DoV) Estimation for Intuitive Speech Interaction with Smart Devices Ecosystems | Future homes and offices will feature increasingly dense ecosystems of IoT devices, such as smart lighting, speakers, and domestic appliances. Voice input is a natural candidate for interacting with out-of-reach and often small devices that lack full-sized physical interfaces. However, at present, voice agents generally require wake-words and device names in order to specify the target of a spoken command (e.g., “Hey Alexa, kitchen lights to full brightness”). In this research, we explore whether speech alone can be used as a directional communication channel, in much the same way visual gaze specifies a focus. Instead of a device’s microphones simply receiving and processing spoken commands, we suggest they also infer the Direction of Voice (DoV). Our approach innately enables voice commands with addressability (i.e., devices know if a command was directed at them) in a natural and rapid manner. We quantify the accuracy of our implementation across users, rooms, spoken phrases, and other key factors that affect performance and usability. Taken together, we believe our DoV approach demonstrates feasibility and the promise of making distributed voice interactions much more intuitive and fluid. | ['and Chris Harrison.', 'Mayank Goel', 'Andy Kong', 'Karan Ahuja'] | 2020-11-15 | null | null | null | null | ['speaker-orientation'] | ['audio'] | [ 1.22799911e-01 1.78656757e-01 3.40510048e-02 -5.22096813e-01
-1.15953319e-01 -1.02824187e+00 5.41065633e-01 -5.55405378e-01
-1.06025971e-01 4.90252852e-01 6.84871376e-01 -7.83618569e-01
2.63335049e-01 -4.99705464e-01 -2.06165202e-02 -3.76008928e-01
5.38671374e-01 1.49002776e-01 -1.83185805e-02 -3.00286226e-02
4.78093922e-02 5.62612176e-01 -1.77528787e+00 1.46695405e-01
2.27377117e-01 8.73807609e-01 3.45490485e-01 1.19865882e+00
-4.17146385e-01 7.38435507e-01 -8.43798339e-01 2.43701547e-01
-2.15559512e-01 -1.09686717e-01 -7.43477464e-01 -1.01086982e-01
5.14870942e-01 -8.12422156e-01 7.85670057e-02 6.25567794e-01
9.55914676e-01 4.47429763e-03 1.17134593e-01 -1.33418930e+00
-5.45096159e-01 6.09275639e-01 3.31155490e-03 -9.63302255e-02
1.16152954e+00 8.01742196e-01 1.02493322e+00 -5.33481598e-01
3.57318968e-01 1.39978564e+00 3.66140872e-01 8.41325164e-01
-1.19985425e+00 -7.83785462e-01 3.25765491e-01 -5.31645000e-01
-1.15196776e+00 -1.29105175e+00 4.45813745e-01 -3.83882999e-01
1.17395389e+00 9.11376417e-01 4.25784588e-01 1.37913859e+00
-2.74440169e-01 4.09546882e-01 8.99647892e-01 -3.66842955e-01
3.51488233e-01 5.46174586e-01 1.93525359e-01 3.35789919e-01
-2.59553879e-01 -3.36072862e-01 -7.90271044e-01 -4.41924572e-01
7.25414634e-01 3.99649553e-02 -6.51602149e-01 1.04996681e-01
-1.48953474e+00 1.30620822e-01 1.81306470e-02 4.56724942e-01
-3.87848854e-01 1.33803189e-01 4.89338720e-03 1.54289454e-01
-1.32760644e-01 5.30716836e-01 -5.09672642e-01 -9.70715344e-01
-2.85239607e-01 -1.39892027e-01 1.33133221e+00 1.33338487e+00
1.60931319e-01 -3.29481751e-01 -2.45266929e-01 7.00435221e-01
5.58943629e-01 9.17679310e-01 1.36277810e-01 -1.22203183e+00
1.98406696e-01 2.53062546e-01 6.36028945e-01 -5.07692039e-01
-5.54763019e-01 3.10410112e-01 -3.28156173e-01 3.09367478e-01
7.09587693e-01 -6.37229562e-01 -3.68953943e-01 1.63432312e+00
3.89387518e-01 -3.11742306e-01 -2.45205551e-01 9.36390579e-01
7.68051028e-01 6.11099780e-01 6.26873598e-02 -1.73738599e-01
1.64815843e+00 -4.88052309e-01 -1.20695364e+00 -3.23372245e-01
2.07808837e-01 -9.45482135e-01 1.88505590e+00 3.17743331e-01
-7.47014225e-01 -3.50415468e-01 -6.51880085e-01 -2.46380538e-01
-1.98710665e-01 -2.21478090e-01 6.82386518e-01 9.19026792e-01
-1.33317244e+00 2.43247245e-02 -7.90592134e-01 -9.89378870e-01
6.51269257e-02 5.10910153e-01 -5.10049472e-03 3.10724467e-01
-4.77243066e-01 5.38740516e-01 -5.31747460e-01 -1.17683411e-01
-3.23523134e-02 -4.15834755e-01 -4.29190814e-01 1.25105381e-01
4.40626144e-02 -6.05582237e-01 1.81908965e+00 -7.30551958e-01
-2.16675091e+00 5.94541848e-01 -6.34456336e-01 4.03264076e-01
3.08702022e-01 -3.30222428e-01 -5.19721389e-01 -5.14197052e-02
-7.36657977e-02 7.51724243e-01 8.52608621e-01 -1.12674952e+00
-7.28348434e-01 -4.41666543e-01 2.35502511e-01 1.49574980e-01
-5.68686306e-01 2.69092113e-01 -1.21193364e-01 2.20268771e-01
-2.14347199e-01 -8.05665493e-01 3.73698294e-01 4.31712121e-01
-7.31021345e-01 -3.82345080e-01 1.13564754e+00 -2.14093566e-01
1.42365575e+00 -2.41072297e+00 -3.55599791e-01 2.41898432e-01
4.38907564e-01 -3.26167978e-02 1.55210301e-01 3.65787029e-01
3.56780589e-01 4.74694520e-01 5.25534749e-01 -4.15452987e-01
4.20124590e-01 3.80940773e-02 -4.56124902e-01 1.58433616e-01
-3.69138449e-01 7.35494554e-01 -7.99667001e-01 -1.85031384e-01
4.21655118e-01 8.26174438e-01 -4.87201780e-01 4.53173310e-01
1.42123215e-02 5.67388356e-01 -4.50762987e-01 5.14290929e-01
3.14426184e-01 -4.24733460e-01 2.30249062e-01 -3.52832153e-02
-7.51033425e-01 9.43373501e-01 -9.99640346e-01 1.05782914e+00
-1.03123677e+00 9.45094645e-01 7.45206952e-01 2.72675037e-01
6.33698583e-01 4.17307496e-01 1.24822021e-01 -4.97918397e-01
1.94868237e-01 1.19588487e-01 -1.26677111e-01 -7.21589506e-01
2.58151114e-01 3.18002522e-01 8.89378637e-02 7.50694573e-01
-6.81278646e-01 -3.91905189e-01 -5.07120132e-01 -1.89168319e-01
1.36489356e+00 -1.58233762e-01 2.77282655e-01 -7.75905326e-02
8.27490315e-02 -5.56727350e-01 -1.77690387e-01 8.68810773e-01
-4.44357604e-01 4.20059115e-01 1.39015913e-01 -1.86058003e-02
-6.25319541e-01 -9.84550714e-01 1.20089930e-02 1.76940501e+00
-3.84037346e-02 -4.61709410e-01 -7.67229915e-01 -2.37830147e-01
-7.90094659e-02 1.20636010e+00 2.76492774e-01 3.79908264e-01
-3.84082824e-01 4.37123954e-01 4.16037917e-01 2.83587217e-01
2.86073208e-01 -1.27079916e+00 -1.08222091e+00 -4.51040367e-04
-2.54575610e-01 -1.18126822e+00 -8.83434117e-01 1.43463746e-01
-5.16102016e-01 -5.43116450e-01 -4.10634518e-01 -5.46489298e-01
4.79188770e-01 6.93318129e-01 9.48891521e-01 3.59062552e-02
3.29504386e-02 1.03736269e+00 -9.21517760e-02 -4.20740366e-01
-1.99654758e-01 5.12149967e-02 2.64333993e-01 -9.40771997e-02
6.25548184e-01 -8.05922747e-01 -7.89534867e-01 6.30564988e-01
-1.86747849e-01 -4.14092094e-02 1.69487879e-01 1.36245698e-01
-1.19095013e-01 -3.33877951e-01 1.68411404e-01 -4.39868063e-01
1.09361482e+00 -1.73483461e-01 -2.15161979e-01 -3.64861526e-02
-3.72720450e-01 -1.75710857e-01 6.21890903e-01 -5.86752355e-01
-7.51369536e-01 3.17887776e-02 -1.62693366e-01 7.41560236e-02
-7.27111697e-01 -5.78267157e-01 -4.60068673e-01 2.19075143e-01
5.09718120e-01 -2.16731131e-02 -1.17579594e-01 -3.29188108e-01
5.38610816e-01 1.73649609e+00 3.54935348e-01 -4.86847371e-01
4.88639861e-01 5.43888271e-01 -9.15800989e-01 -1.28354037e+00
-1.36259452e-01 -5.92532098e-01 -1.94436952e-01 -2.44487122e-01
8.88617933e-01 -5.31432331e-01 -1.83694637e+00 4.73740131e-01
-1.57739425e+00 -7.11757243e-01 -7.13371858e-02 2.50705838e-01
-2.27400646e-01 -1.79494038e-01 -3.05200458e-01 -1.30217242e+00
-2.65468001e-01 -1.24067271e+00 1.50303233e+00 5.08123994e-01
-1.19242561e+00 -5.45140743e-01 -4.54099685e-01 2.80350655e-01
9.97005522e-01 -4.66822594e-01 5.96510351e-01 -3.99965793e-01
-4.51888382e-01 1.32286161e-01 -6.82425871e-02 -1.28775537e-01
9.23396409e-01 1.12822868e-01 -1.53214240e+00 -1.22374490e-01
1.16485078e-02 8.45259354e-02 -3.03235501e-01 3.22147012e-01
1.04809415e+00 -7.07198560e-01 -6.41336560e-01 2.95649797e-01
6.66082442e-01 4.56164926e-01 4.71459746e-01 -4.88970689e-02
7.22360671e-01 5.54199755e-01 -5.51123433e-02 5.86622834e-01
5.58603644e-01 8.01109135e-01 1.80252582e-01 6.53322041e-02
-1.06585912e-01 -1.78882122e-01 4.25671220e-01 4.19376850e-01
1.63296789e-01 -6.24200642e-01 -7.55811095e-01 4.01602425e-02
-1.26751268e+00 -9.04136598e-01 8.22593272e-02 2.18579674e+00
8.51922691e-01 1.51308298e-01 1.12133279e-01 -1.95317948e-03
7.36217618e-01 1.72614157e-01 -8.29259038e-01 -5.87554812e-01
4.05959576e-01 1.01677820e-01 4.02241886e-01 8.10467720e-01
-6.82839870e-01 6.87873542e-01 6.91032505e+00 -1.49952590e-01
-1.42523909e+00 4.15075719e-02 4.36243445e-01 -3.37456077e-01
-6.17102742e-01 -4.84210938e-01 -7.08518624e-01 7.27998614e-01
8.84345889e-01 4.63304996e-01 1.18635345e+00 7.90888965e-01
8.91474605e-01 -1.27851829e-01 -1.64266634e+00 1.43126595e+00
-2.71600932e-01 -7.72401571e-01 -4.52300847e-01 3.22330654e-01
-8.93032700e-02 -1.23343393e-01 1.15603507e-01 -5.15686274e-02
3.78153920e-01 -1.06388271e+00 8.82555604e-01 3.37734789e-01
1.04993713e+00 -1.00834452e-01 -4.72373515e-03 1.54121161e-01
-1.20373273e+00 -1.73510849e-01 3.58469784e-01 -6.64381504e-01
4.16514218e-01 2.26673067e-01 -1.01824343e+00 -7.33381271e-01
8.73030007e-01 -6.12872131e-02 -7.81183466e-02 4.45339501e-01
-3.44070703e-01 6.69414699e-01 -7.02263057e-01 -7.87556946e-01
-3.11773419e-01 1.55977264e-01 5.15412450e-01 8.78445446e-01
4.16230619e-01 4.67798769e-01 -5.86622879e-02 8.53938520e-01
1.35477865e-02 -1.93794698e-01 -9.80496168e-01 -2.20507488e-01
1.31842530e+00 1.17695248e+00 -5.88673592e-01 -1.29755676e-01
-4.59100425e-01 1.23997533e+00 -3.08545798e-01 6.64585292e-01
-4.80376303e-01 -4.49676543e-01 1.45177436e+00 2.32716754e-01
-2.52477229e-02 -4.43397373e-01 -3.30446720e-01 -6.41290128e-01
1.64833069e-01 -1.04909408e+00 -3.69782150e-01 -1.20009768e+00
-1.04739714e+00 3.83949935e-01 -5.40003657e-01 -8.23866665e-01
-3.63562703e-01 -5.53202808e-01 -7.04871655e-01 7.97687590e-01
-9.12097216e-01 -8.80672276e-01 -6.33351147e-01 5.30039608e-01
5.74002683e-01 3.56136799e-01 1.15426838e+00 1.90251842e-01
-2.46509850e-01 3.87854129e-01 -2.00898781e-01 2.52256710e-02
9.66187656e-01 -1.23550725e+00 5.96397698e-01 1.88172668e-01
-1.40740022e-01 1.27520227e+00 8.43475401e-01 -2.29583308e-01
-1.93870330e+00 -1.87487110e-01 8.79833758e-01 -8.99034798e-01
6.18422925e-01 -7.44385600e-01 -2.96985537e-01 6.75126433e-01
3.74014407e-01 -2.98587561e-01 7.55065978e-01 3.35610569e-01
-2.15683460e-01 -3.19549829e-01 -1.16473663e+00 1.01183867e+00
1.32366920e+00 -9.98967350e-01 -3.16706598e-01 1.30344942e-01
1.03821731e+00 -2.13783875e-01 -3.77176970e-01 -4.57722247e-01
1.07658350e+00 -1.08535409e+00 7.52408326e-01 -1.06704153e-01
-2.71814704e-01 -3.15708250e-01 -2.90809304e-01 -9.78975356e-01
-1.24720261e-01 -1.34057355e+00 8.91541541e-02 1.66536736e+00
2.53077745e-01 -1.15135539e+00 2.41704702e-01 1.43970954e+00
8.06803629e-02 -2.31328502e-01 -7.60752261e-01 -1.28830105e-01
-7.69504070e-01 -7.28331983e-01 1.08104360e+00 6.55964196e-01
5.25781989e-01 6.75610125e-01 -8.89339596e-02 3.36681396e-01
6.39395118e-02 -4.36420858e-01 9.60304558e-01 -1.12567484e+00
-7.40821436e-02 -6.07815266e-01 -8.31021834e-03 -1.66704118e+00
-1.64528385e-01 -2.85438985e-01 2.72990584e-01 -1.54106712e+00
-5.16521513e-01 -5.42985320e-01 2.94088870e-01 6.66347206e-01
1.64220408e-01 -2.36986727e-01 2.66359478e-01 1.96824387e-01
-3.68732303e-01 -2.59805080e-02 8.91784787e-01 -1.95991084e-01
-8.23759854e-01 2.25476876e-01 -9.28657830e-01 7.40004182e-01
8.09687197e-01 2.35601455e-01 -5.22213757e-01 -5.63990891e-01
1.89760327e-01 -2.21287608e-01 4.54706907e-01 -8.78700316e-01
5.66699862e-01 -2.81152278e-01 1.84423909e-01 3.54068093e-02
4.15654242e-01 -1.30699277e+00 8.42882693e-02 1.09328724e-01
-3.03966075e-01 -1.98332276e-02 4.88418303e-02 1.88334122e-01
4.75179464e-01 2.98553526e-01 1.75505370e-01 3.15413550e-02
-4.47136015e-01 -2.70335764e-01 -7.93129563e-01 -2.63759732e-01
5.36557198e-01 -4.84390587e-01 -6.47480428e-01 -9.07659233e-01
-4.50472474e-01 2.05264017e-01 6.71592772e-01 7.34269857e-01
3.75035256e-01 -1.00537252e+00 3.24871957e-01 3.59067589e-01
-6.65517151e-02 -4.88353707e-02 -3.72099996e-01 5.62004030e-01
-2.21627340e-01 4.59202230e-01 2.62210071e-01 -8.92882645e-01
-1.39109993e+00 6.11888692e-02 3.79245907e-01 7.83192873e-01
-6.06467605e-01 7.46370614e-01 4.41612303e-02 -5.16765714e-01
7.84135342e-01 -8.40568483e-01 -2.80438997e-02 1.88273787e-02
8.75038385e-01 2.43525878e-01 -2.57649511e-01 -2.78183639e-01
-8.04125249e-01 5.10173857e-01 5.12432158e-01 -4.02470529e-01
9.54799891e-01 -7.00755417e-01 -2.89190054e-01 7.74213910e-01
9.95619178e-01 5.75023472e-01 -1.10884058e+00 1.58679500e-01
-3.75419497e-01 -3.55444849e-01 -1.64768085e-01 -8.71584117e-01
-4.60320950e-01 5.97997963e-01 9.08358037e-01 7.31635094e-01
8.80575538e-01 3.82762223e-01 7.59008408e-01 6.57570660e-01
6.74800456e-01 -9.29332376e-01 -9.65941101e-02 2.68915117e-01
6.22541010e-01 -1.03080022e+00 -5.28855264e-01 -3.81923318e-01
-4.27041203e-01 1.16383779e+00 5.77181518e-01 9.65355754e-01
5.89988470e-01 7.81354785e-01 3.72812361e-01 -2.53639102e-01
-6.05968773e-01 -1.90368399e-01 -2.83955365e-01 9.48851883e-01
7.15387285e-01 4.18155968e-01 1.64425820e-01 3.66345584e-01
-7.32898653e-01 -1.69655643e-02 2.68511772e-01 8.40163112e-01
-6.95814908e-01 -7.30874479e-01 -6.03285551e-01 3.25105816e-01
-1.03430301e-01 -1.91366658e-01 -6.14523709e-01 2.59098917e-01
-1.44408509e-01 1.85577416e+00 2.73141891e-01 -5.42351425e-01
4.58587319e-01 2.79547691e-01 -2.48644613e-02 -7.14717448e-01
-4.91132081e-01 -1.89133197e-01 2.03700095e-01 -8.91880691e-01
-1.51906237e-01 -3.47829551e-01 -1.37922394e+00 -4.57324505e-01
-1.56963319e-01 -1.42872721e-01 8.30652833e-01 7.17307568e-01
7.34243095e-01 5.67913473e-01 5.77126145e-01 -1.05833185e+00
-4.88662392e-01 -8.88104737e-01 -1.38120130e-01 8.30478743e-02
7.62135804e-01 -2.28378460e-01 -5.78055322e-01 -1.99201275e-02] | [13.782565116882324, 6.128457546234131] |
ca8dc07c-6b3a-43ec-81d4-aa3ed8757176 | deep-parametric-indoor-lighting-estimation-1 | null | null | http://openaccess.thecvf.com/content_ICCV_2019/html/Gardner_Deep_Parametric_Indoor_Lighting_Estimation_ICCV_2019_paper.html | http://openaccess.thecvf.com/content_ICCV_2019/papers/Gardner_Deep_Parametric_Indoor_Lighting_Estimation_ICCV_2019_paper.pdf | Deep Parametric Indoor Lighting Estimation | We present a method to estimate lighting from a single image of an indoor scene. Previous work has used an environment map representation that does not account for the localized nature of indoor lighting. Instead, we represent lighting as a set of discrete 3D lights with geometric and photometric parameters. We train a deep neural network to regress these parameters from a single image, on a dataset of environment maps annotated with depth. We propose a differentiable layer to convert these parameters to an environment map to compute our loss; this bypasses the challenge of establishing correspondences between estimated and ground truth lights. We demonstrate, via quantitative and qualitative evaluations, that our representation and training scheme lead to more accurate results compared to previous work, while allowing for more realistic 3D object compositing with spatially-varying lighting.
| [' Jean-Francois Lalonde', ' Christian Gagne', ' Kalyan Sunkavalli', ' Yannick Hold-Geoffroy', 'Marc-Andre Gardner'] | 2019-10-01 | null | null | null | iccv-2019-10 | ['lighting-estimation'] | ['computer-vision'] | [ 2.32580036e-01 -1.29155591e-01 5.54082930e-01 -9.12053943e-01
-6.25095785e-01 -7.21485794e-01 6.83518112e-01 -1.00970015e-01
-3.67441922e-01 6.53470457e-01 2.07376525e-01 -1.68429002e-01
3.87749791e-01 -8.40690792e-01 -1.04945779e+00 -3.52155834e-01
1.44458875e-01 2.88442165e-01 -1.11536965e-01 9.28084776e-02
-6.54132515e-02 7.04104722e-01 -1.67995369e+00 6.03459543e-03
5.83286405e-01 1.03696537e+00 1.68970600e-01 1.04321098e+00
-1.81393567e-02 6.11515701e-01 -5.99561274e-01 -1.72092482e-01
6.92904830e-01 -2.36988217e-01 -4.28870469e-01 4.24427032e-01
1.06818962e+00 -6.80720866e-01 -2.20187113e-01 8.76354694e-01
3.86465967e-01 2.27569342e-01 6.19457841e-01 -1.02060020e+00
-7.23694742e-01 -3.93213034e-01 -3.78548563e-01 -3.00923347e-01
6.67552948e-01 2.70545125e-01 7.64029801e-01 -7.00999737e-01
4.56680983e-01 1.14972115e+00 9.89546537e-01 2.31164292e-01
-1.52813089e+00 -3.09623629e-01 2.89446414e-01 -4.07432824e-01
-1.41926968e+00 -4.55730826e-01 9.17461276e-01 -5.12733936e-01
9.14011419e-01 1.23517282e-01 8.04348469e-01 9.59106028e-01
-3.33291814e-02 3.24339092e-01 1.35671759e+00 -5.42284667e-01
3.65541935e-01 2.21880957e-01 -3.05488169e-01 9.28875268e-01
-1.01092450e-01 1.50949568e-01 -2.67416507e-01 8.85618851e-02
9.97314870e-01 -4.03853580e-02 -3.27124268e-01 -6.68527663e-01
-1.02443182e+00 5.16115844e-01 8.09949398e-01 -3.04671079e-01
-3.14494878e-01 5.10757864e-01 -1.38033986e-01 8.03652331e-02
6.72134221e-01 4.73222822e-01 -4.63873386e-01 1.27197266e-01
-7.85327733e-01 1.85470477e-01 6.51042283e-01 9.03587759e-01
1.35118246e+00 -1.26555607e-01 2.27830380e-01 5.78693986e-01
5.24360061e-01 8.19975615e-01 -1.36925370e-01 -1.57931554e+00
1.48772135e-01 3.12119693e-01 4.24314559e-01 -8.52689087e-01
-3.56158257e-01 -1.63078040e-01 -3.80946726e-01 6.78323209e-01
4.64910507e-01 -1.38735846e-01 -1.13533878e+00 1.73189950e+00
3.80896509e-01 3.18995774e-01 -2.38288537e-01 9.32186246e-01
5.27331591e-01 5.45651674e-01 -1.98755413e-01 3.13884795e-01
8.50466728e-01 -7.71755099e-01 -4.03036028e-01 -4.13773745e-01
1.38265431e-01 -7.93921173e-01 1.40610564e+00 4.01582509e-01
-1.20365906e+00 -4.58797932e-01 -9.95556176e-01 -5.23298979e-01
-4.95789260e-01 1.51233330e-01 6.03910565e-01 6.20902121e-01
-1.56444061e+00 3.28545779e-01 -8.69279146e-01 -4.59204942e-01
1.63191766e-01 2.57014394e-01 -2.44464844e-01 -1.50857419e-01
-7.79737651e-01 9.68809664e-01 -4.47170474e-02 1.78038806e-01
-1.05839896e+00 -7.14823484e-01 -1.13730228e+00 -2.53457338e-01
-8.22385997e-02 -8.83021653e-01 1.23788977e+00 -9.21170175e-01
-1.60478532e+00 8.94989967e-01 -4.41655040e-01 -1.87234268e-01
5.11408746e-01 -2.83544838e-01 8.06645751e-02 -1.05265297e-01
3.51921394e-02 8.53153527e-01 5.48642695e-01 -1.90453494e+00
-5.26079476e-01 -2.52330929e-01 6.10888779e-01 3.22658151e-01
-5.95324151e-02 -2.70009845e-01 -5.67922354e-01 -1.27808034e-01
1.33497387e-01 -8.11525583e-01 -3.33348006e-01 5.99895597e-01
-5.03553808e-01 4.44485724e-01 4.62592691e-01 -6.77321672e-01
5.01200378e-01 -1.97248745e+00 -6.57641366e-02 2.50698566e-01
2.54570395e-02 -4.04321462e-01 -8.53962377e-02 5.61859868e-02
1.68556660e-01 -5.76927960e-02 -3.74042511e-01 -1.12913275e+00
1.81817248e-01 3.21604937e-01 -4.12211597e-01 7.53527820e-01
1.00398719e-01 7.13607013e-01 -1.05311549e+00 -1.66378215e-01
7.92267323e-01 1.14093363e+00 -6.83339238e-01 3.70504051e-01
-4.40949321e-01 7.36941636e-01 -3.95545177e-02 4.76375878e-01
7.28274763e-01 -6.78748041e-02 -9.11063924e-02 -1.67554751e-01
-2.66237497e-01 3.39902192e-01 -1.19494224e+00 2.11034775e+00
-1.12064576e+00 9.15967524e-01 2.10969687e-01 -2.49832198e-01
8.83294344e-01 2.64528599e-02 4.28050101e-01 -6.48638308e-01
1.86360687e-01 4.65926751e-02 -7.38490343e-01 -2.41845623e-01
5.90770185e-01 -1.69125050e-01 9.91392285e-02 3.36091250e-01
-1.36158079e-01 -7.63506889e-01 -3.14234465e-01 -2.02405289e-01
9.48090255e-01 7.07863510e-01 -9.98844430e-02 4.29112911e-02
3.01738262e-01 -2.06735417e-01 3.32016498e-01 5.25655389e-01
-1.03151752e-02 9.62804377e-01 -8.90444592e-02 -6.29331350e-01
-1.34961843e+00 -1.36754179e+00 -2.17379540e-01 6.22441292e-01
2.40721405e-01 -1.15671277e-01 -7.88306832e-01 -3.51647526e-01
1.57420456e-01 7.45001972e-01 -6.29414678e-01 2.32035413e-01
-6.10359013e-01 -4.11491930e-01 1.06365204e-01 4.21758950e-01
6.36310041e-01 -5.44136822e-01 -8.07604611e-01 -1.93987876e-01
-1.90565646e-01 -1.46155000e+00 -2.90390193e-01 2.89600462e-01
-3.84178430e-01 -9.74999845e-01 -2.86769897e-01 -5.76444685e-01
9.79869962e-01 2.97999084e-01 1.51838470e+00 -8.41239020e-02
-5.25573611e-01 8.12578976e-01 8.79493952e-02 -5.42693973e-01
-2.91555613e-01 -3.86443615e-01 2.01472118e-02 -8.54896456e-02
1.20704062e-01 -6.91431820e-01 -8.08879256e-01 1.31085262e-01
-6.43228412e-01 2.47547105e-01 2.15872731e-02 3.43834460e-01
8.17067266e-01 -3.46452072e-02 -3.21627975e-01 -5.44410944e-01
2.45008364e-01 -1.37114286e-01 -1.09150720e+00 3.63689195e-03
-3.73701155e-01 -2.69638207e-02 6.75233245e-01 1.50860567e-02
-1.20855629e+00 5.32435238e-01 -1.21533677e-01 -4.82365042e-01
-5.68226874e-01 -2.07908019e-01 -1.95081055e-01 -3.25894892e-01
7.50846386e-01 -1.00315504e-01 -2.99795538e-01 -3.47412854e-01
6.28446221e-01 3.56741279e-01 7.62268543e-01 -8.33849311e-01
1.16257846e+00 9.52190280e-01 2.61472702e-01 -5.87854326e-01
-1.04694557e+00 -3.08173776e-01 -8.78630996e-01 -1.68054074e-01
1.01051939e+00 -1.14242887e+00 -7.52271175e-01 2.51592904e-01
-1.24244773e+00 -9.17839825e-01 -4.13264543e-01 4.08565462e-01
-8.17885101e-01 -8.62098951e-03 -5.06233752e-01 -9.86226201e-01
3.20620865e-01 -1.13672149e+00 1.64512062e+00 5.13792410e-02
-9.46667045e-02 -1.24393702e+00 2.77058661e-01 1.45301372e-01
3.38066339e-01 7.27048755e-01 6.20390892e-01 4.49096471e-01
-1.02542722e+00 -9.63869095e-02 -2.91790724e-01 4.95015860e-01
4.58880097e-01 1.93886310e-01 -1.57688332e+00 -1.14199959e-01
3.72248963e-02 -3.14919800e-01 6.29544735e-01 4.78884488e-01
1.24912095e+00 -3.58171165e-02 2.12986078e-02 1.25378156e+00
1.91854215e+00 -1.31421462e-01 3.56058985e-01 5.59071660e-01
9.85663414e-01 5.53094149e-01 2.27734342e-01 3.20010841e-01
8.13144565e-01 5.76643169e-01 8.03424060e-01 -5.99934876e-01
-3.96121830e-01 -4.49757308e-01 5.89034371e-02 3.44372511e-01
-1.76400747e-02 -1.74071923e-01 -8.72551918e-01 4.75026935e-01
-1.48005331e+00 -4.60148245e-01 6.76099733e-02 2.39939618e+00
7.69791782e-01 -3.45327109e-02 -1.27666786e-01 -1.72910318e-01
3.10453504e-01 1.38299435e-01 -6.14970386e-01 -4.28743690e-01
-1.49058118e-01 9.66924429e-02 8.65287006e-01 1.09164822e+00
-1.11134732e+00 7.71674216e-01 7.33929968e+00 -2.30803519e-01
-1.12578714e+00 -2.94612683e-02 6.79442942e-01 -2.31726274e-01
-7.27769613e-01 -1.06816314e-01 -5.59445322e-01 3.01733911e-01
8.32373619e-01 4.40717399e-01 1.02983093e+00 7.27961898e-01
4.67547894e-01 -1.19951688e-01 -1.29766881e+00 9.03216720e-01
1.28090352e-01 -1.12435007e+00 -3.49950194e-01 2.26484597e-01
1.03769994e+00 3.22057337e-01 1.12701304e-01 -8.39266032e-02
9.19850647e-01 -1.01148653e+00 1.01615655e+00 6.28442347e-01
8.21951568e-01 -4.40262914e-01 2.24516749e-01 1.83482185e-01
-1.11827874e+00 5.60501553e-02 -2.61738092e-01 -2.77051866e-01
2.40191698e-01 5.69357753e-01 -1.01021206e+00 2.41607741e-01
8.33528757e-01 5.28569579e-01 -5.50059080e-01 9.27618563e-01
-5.88878810e-01 1.73973441e-01 -5.62702537e-01 3.74021024e-01
-1.99755244e-02 -2.98418522e-01 1.72620371e-01 1.07048368e+00
2.63851613e-01 -3.69524688e-01 3.51870745e-01 1.17074788e+00
-1.76506057e-01 -3.17594707e-01 -8.15539956e-01 6.88295245e-01
4.52742755e-01 1.29552162e+00 -6.37737155e-01 -9.67964306e-02
-3.50049883e-01 1.10266304e+00 5.04639149e-01 7.33430982e-01
-8.08926582e-01 -2.03030899e-01 8.96399081e-01 2.16752127e-01
1.06235489e-01 -6.33504808e-01 -5.67124307e-01 -1.18219447e+00
1.56350762e-01 -3.03896666e-01 -4.30617511e-01 -1.24406242e+00
-1.12961173e+00 3.64769608e-01 -4.96018045e-02 -8.79583776e-01
-2.60988921e-01 -6.23753667e-01 -4.78795409e-01 1.04042768e+00
-2.04738021e+00 -1.24614620e+00 -8.03961039e-01 3.87158573e-01
3.02812666e-01 5.72784483e-01 9.77598906e-01 2.53926605e-01
-2.50495702e-01 2.06894070e-01 6.97252601e-02 9.04165674e-03
6.85727119e-01 -1.87450194e+00 7.96716332e-01 9.08436000e-01
9.94929895e-02 6.13157690e-01 9.57894742e-01 -1.72879934e-01
-1.33840215e+00 -1.22832358e+00 4.41719502e-01 -9.86904442e-01
2.99160779e-01 -8.02846372e-01 -5.59733927e-01 8.92344177e-01
1.11585461e-01 3.72354209e-01 3.41819853e-01 9.13636163e-02
-4.87822592e-01 -2.13088840e-01 -1.42204809e+00 5.88250220e-01
1.16693246e+00 -8.43061209e-01 -1.14079081e-01 4.16629851e-01
9.05819237e-01 -7.83996403e-01 -7.60820329e-01 1.41136274e-01
4.86601859e-01 -1.22304630e+00 1.31779802e+00 1.01937674e-01
4.71577078e-01 -6.57304347e-01 -5.60717642e-01 -1.50941432e+00
-3.41406390e-02 -3.05017143e-01 -1.66266486e-02 1.00320899e+00
1.61926270e-01 -5.07444620e-01 8.34775388e-01 1.02847111e+00
-2.21817806e-01 -5.50363719e-01 -6.13232970e-01 -4.98080581e-01
-1.09183090e-02 -4.83075351e-01 9.70478237e-01 7.29087412e-01
-9.82364953e-01 2.88548297e-03 -1.67499810e-01 5.47941029e-01
1.00538003e+00 5.00213467e-02 1.07301998e+00 -1.13898599e+00
-2.79842764e-01 -1.54260367e-01 -3.46730798e-01 -1.12674534e+00
3.90273154e-01 -5.02429068e-01 5.82618713e-01 -1.94566834e+00
-1.27382383e-01 -6.83803856e-01 -8.14709514e-02 3.91631752e-01
6.37495816e-02 6.98393822e-01 -4.77468893e-02 -1.06030136e-01
-4.85093445e-01 4.60100085e-01 1.10063004e+00 -2.38638092e-02
-1.70852467e-01 -3.66140753e-01 -3.69478285e-01 8.92446637e-01
6.29293740e-01 -2.93255057e-02 -6.03637040e-01 -1.07191837e+00
3.33287030e-01 -5.99883735e-01 6.02097631e-01 -1.03116882e+00
-1.45554845e-03 -2.29889110e-01 9.06776905e-01 -3.86974216e-01
8.75947833e-01 -1.09329534e+00 1.45316020e-01 -2.54203118e-02
-2.83341587e-01 -4.51833606e-02 1.96564361e-01 3.97225171e-01
1.87914729e-01 1.47543505e-01 6.02897346e-01 -2.99526125e-01
-6.88313186e-01 2.66851038e-01 1.73025936e-01 -2.74088591e-01
7.89516747e-01 -4.30683285e-01 -1.44883871e-01 -4.50215399e-01
-4.58407521e-01 6.53611347e-02 1.33806276e+00 1.45699501e-01
6.32055581e-01 -1.36914527e+00 -3.10662895e-01 4.60468680e-01
1.91226527e-02 5.12595832e-01 -1.56608045e-01 1.77347213e-01
-9.12554622e-01 1.62214324e-01 1.84448466e-01 -7.76758909e-01
-8.86818886e-01 3.40051442e-01 8.84109676e-01 2.72625566e-01
-5.70331335e-01 8.75400245e-01 4.00178432e-01 -1.03491676e+00
3.23169589e-01 -7.08889902e-01 4.25177753e-01 -5.59533000e-01
2.16991588e-01 4.24405485e-02 8.31489936e-02 -6.23972595e-01
-2.45653197e-01 8.40319335e-01 6.60143852e-01 -2.63009787e-01
1.36246276e+00 -5.81088066e-01 -1.46432236e-01 7.38655508e-01
1.53891289e+00 2.80743152e-01 -1.97337794e+00 -1.85232535e-01
-5.26843369e-01 -8.76300812e-01 3.49265605e-01 -8.95043612e-01
-8.30553591e-01 7.80444086e-01 7.74420321e-01 2.16675713e-03
1.04569733e+00 -2.62233794e-01 6.73853457e-01 3.69252086e-01
6.07567430e-01 -8.91242981e-01 -1.10447392e-01 4.91341323e-01
5.42941511e-01 -1.53264678e+00 2.14724112e-02 -2.12560162e-01
-2.92102471e-02 9.76900160e-01 5.11319697e-01 -1.46096155e-01
6.78389549e-01 5.16146958e-01 4.05320376e-01 -1.14154831e-01
-3.23630333e-01 -2.47750655e-01 2.14412987e-01 7.41400659e-01
4.37437296e-01 3.16943265e-02 7.67275333e-01 -4.25879061e-01
-5.01583755e-01 -5.60486317e-02 4.22856748e-01 8.02349746e-01
-4.69992429e-01 -9.03926253e-01 -5.96019149e-01 -1.45704467e-02
-1.17384017e-01 -5.97440731e-03 -2.35032991e-01 4.99608934e-01
3.65815997e-01 8.73771846e-01 3.65523964e-01 -2.68730044e-01
4.47748423e-01 -2.44018748e-01 7.83633292e-01 -6.46036386e-01
-1.50246933e-01 -5.87771498e-02 -3.71246561e-02 -8.06315839e-01
-5.14160872e-01 -4.94510412e-01 -1.10079014e+00 -1.99919552e-01
2.34047882e-02 -3.06376010e-01 1.36012900e+00 6.63656354e-01
1.37355492e-01 6.19280040e-01 9.77412283e-01 -1.37499118e+00
1.94417350e-02 -3.54903698e-01 -4.27951425e-01 4.20905143e-01
9.81676161e-01 -6.10607564e-01 -5.89625418e-01 1.72795862e-01] | [9.606744766235352, -3.0209615230560303] |
6ef5e974-49df-4c60-93f5-bff86ad49ad8 | more-multi-order-relation-mining-for-dense | 2203.05203 | null | https://arxiv.org/abs/2203.05203v2 | https://arxiv.org/pdf/2203.05203v2.pdf | MORE: Multi-Order RElation Mining for Dense Captioning in 3D Scenes | 3D dense captioning is a recently-proposed novel task, where point clouds contain more geometric information than the 2D counterpart. However, it is also more challenging due to the higher complexity and wider variety of inter-object relations contained in point clouds. Existing methods only treat such relations as by-products of object feature learning in graphs without specifically encoding them, which leads to sub-optimal results. In this paper, aiming at improving 3D dense captioning via capturing and utilizing the complex relations in the 3D scene, we propose MORE, a Multi-Order RElation mining model, to support generating more descriptive and comprehensive captions. Technically, our MORE encodes object relations in a progressive manner since complex relations can be deduced from a limited number of basic ones. We first devise a novel Spatial Layout Graph Convolution (SLGC), which semantically encodes several first-order relations as edges of a graph constructed over 3D object proposals. Next, from the resulting graph, we further extract multiple triplets which encapsulate basic first-order relations as the basic unit, and construct several Object-centric Triplet Attention Graphs (OTAG) to infer multi-order relations for every target object. The updated node features from OTAG are aggregated and fed into the caption decoder to provide abundant relational cues, so that captions including diverse relations with context objects can be generated. Extensive experiments on the Scan2Cap dataset prove the effectiveness of our proposed MORE and its components, and we also outperform the current state-of-the-art method. Our code is available at https://github.com/SxJyJay/MORE. | ['Yu-Gang Jiang', 'Lin Ma', 'Jingjing Chen', 'Zequn Jie', 'Shaoxiang Chen', 'Yang Jiao'] | 2022-03-10 | null | null | null | null | ['dense-captioning', '3d-dense-captioning'] | ['computer-vision', 'computer-vision'] | [ 2.95249373e-02 4.47593182e-01 -1.63071826e-01 -4.39675301e-01
-5.77413380e-01 -5.58589995e-01 4.38412160e-01 1.58980504e-01
3.28471571e-01 4.88437265e-01 2.62733698e-01 -2.47432187e-01
-2.50750601e-01 -9.48726058e-01 -1.14710546e+00 -3.96132976e-01
-4.35803272e-02 7.60108173e-01 2.71662146e-01 -1.61161438e-01
-1.42127216e-01 6.48411572e-01 -1.45368421e+00 2.95122415e-01
9.26270902e-01 1.17903161e+00 4.51668650e-01 1.92394763e-01
-4.95966285e-01 4.35580671e-01 -2.74181902e-01 -6.12949193e-01
1.14721753e-01 -5.95478453e-02 -6.11054301e-01 3.51337582e-01
5.55066049e-01 -1.09841309e-01 -5.97270191e-01 1.04304552e+00
1.81397080e-01 -4.73244675e-03 4.92548913e-01 -1.33903027e+00
-1.11100757e+00 5.67970991e-01 -6.24159813e-01 6.68860897e-02
3.46943974e-01 1.14757791e-01 1.30195129e+00 -1.05077863e+00
5.13839900e-01 1.49389887e+00 3.11237276e-01 2.63160825e-01
-1.07080960e+00 -8.44024837e-01 6.28277540e-01 2.26225778e-01
-1.48362267e+00 -9.34229195e-02 1.09578228e+00 -2.85527110e-01
1.09816074e+00 3.74150217e-01 9.02283251e-01 9.10328388e-01
-2.56199896e-01 7.53851593e-01 5.69672704e-01 -1.79422423e-02
-2.90697187e-01 -6.08817562e-02 1.68265313e-01 8.21819007e-01
5.07181704e-01 -3.15314859e-01 -3.38094801e-01 1.20155767e-01
9.07220006e-01 1.12104453e-01 -4.23367947e-01 -5.52066684e-01
-1.26105559e+00 6.19975507e-01 1.17796457e+00 1.81115925e-01
-4.95451957e-01 1.61878899e-01 -2.14822628e-02 -1.59939662e-01
6.18295610e-01 3.43989700e-01 -3.96689832e-01 3.64999294e-01
-3.54590595e-01 2.53014594e-01 4.16458815e-01 1.70263970e+00
1.11827421e+00 -4.26515192e-01 -3.53433222e-01 7.31928945e-01
5.96303821e-01 4.83590096e-01 -1.77710429e-01 -5.39760649e-01
1.09942698e+00 1.19292367e+00 -2.30317906e-01 -1.39123738e+00
-3.93079519e-01 -6.80994630e-01 -9.10967588e-01 -4.48552668e-01
-2.88581580e-01 3.30263227e-01 -1.02128518e+00 1.62643671e+00
5.77704668e-01 4.19949740e-01 -1.59317359e-01 1.12260687e+00
1.22243977e+00 7.95092762e-01 5.50473034e-02 5.74942902e-02
1.51574981e+00 -1.05356681e+00 -6.36806190e-01 -3.85861874e-01
4.38744098e-01 -4.03419316e-01 9.58925545e-01 -1.13745533e-01
-8.41563046e-01 -5.77236772e-01 -8.76576841e-01 -4.35885578e-01
-3.57909471e-01 -1.16149992e-01 1.00051880e+00 2.90188324e-02
-8.47395837e-01 2.36411199e-01 -6.12713397e-01 -3.04589689e-01
8.69871199e-01 3.46912444e-01 -5.04738152e-01 -2.72908002e-01
-1.03604376e+00 6.16741538e-01 6.25858068e-01 4.15504277e-01
-4.78376001e-01 -7.92912841e-01 -1.22905028e+00 1.75729871e-01
6.72102273e-01 -1.05022264e+00 8.95488918e-01 -2.20630392e-01
-8.12223494e-01 8.94106388e-01 -1.65364653e-01 2.48120334e-02
8.27525407e-02 -1.64574102e-01 -1.92821845e-01 1.89373270e-01
2.32161522e-01 8.64451349e-01 3.80977064e-01 -1.71185303e+00
-4.16628212e-01 -4.98449951e-01 3.84764522e-01 5.34260571e-01
-8.35150182e-02 -2.63988167e-01 -1.14882767e+00 -4.83355045e-01
4.66589868e-01 -8.41217697e-01 -8.33597779e-02 8.73813033e-02
-8.52479517e-01 -4.42469805e-01 7.09391952e-01 -3.51236165e-01
1.02977085e+00 -2.14092684e+00 3.01445514e-01 2.43989915e-01
6.30303919e-01 1.84588626e-01 -2.31774524e-01 2.33464226e-01
-1.90410346e-01 3.08366477e-01 -4.11472112e-01 -5.18038750e-01
9.08652917e-02 5.01449764e-01 -3.32756668e-01 8.62560123e-02
8.24253023e-01 1.44005597e+00 -1.10134208e+00 -7.14376390e-01
3.06079537e-01 5.04406691e-01 -4.34909284e-01 2.99771428e-01
-5.92281342e-01 3.99738550e-01 -9.46494102e-01 8.34105790e-01
9.85070229e-01 -7.90086031e-01 -1.39889002e-01 -5.93298554e-01
1.23964503e-01 3.60784143e-01 -8.12832415e-01 1.93930399e+00
-4.61110562e-01 3.29647034e-01 -2.64327496e-01 -9.94705856e-01
1.14641225e+00 -6.49470985e-02 4.74838704e-01 -6.03283882e-01
9.80089828e-02 -6.02057949e-03 -2.41135165e-01 -7.08071709e-01
3.69545490e-01 7.76248500e-02 -4.59470451e-02 -1.89314615e-02
1.70731679e-01 -3.30278873e-01 1.03653014e-01 4.40998137e-01
9.56994891e-01 3.27100098e-01 3.62160392e-02 1.23312987e-01
4.11550820e-01 3.56204547e-02 3.82514805e-01 4.04714853e-01
2.06358537e-01 8.84467006e-01 7.05666482e-01 -2.56906211e-01
-9.03654337e-01 -9.99137282e-01 3.49906385e-02 2.21288800e-01
6.97547734e-01 -5.92031896e-01 -2.02137902e-01 -8.81882191e-01
1.91129774e-01 6.91539347e-01 -6.82261825e-01 -1.13105662e-01
-6.67249203e-01 -6.03016734e-01 9.01825875e-02 4.94211972e-01
4.00173306e-01 -9.71079767e-01 -1.18564881e-01 2.62876954e-02
-2.93511182e-01 -1.58755004e+00 -3.42375249e-01 9.78151802e-03
-7.72651434e-01 -9.78925347e-01 -4.55957502e-01 -8.73006463e-01
8.57724786e-01 5.98715663e-01 1.37146473e+00 3.23081762e-01
5.42364046e-02 1.92936510e-01 -6.00013494e-01 -3.44695270e-01
-9.21180379e-03 1.19868979e-01 -3.68330777e-01 5.11186458e-02
3.92202169e-01 -7.92824388e-01 -3.71204615e-01 9.25286859e-02
-7.86062479e-01 5.65318465e-01 9.84122276e-01 5.67590952e-01
9.97292161e-01 -1.49307638e-01 2.25084677e-01 -1.00655365e+00
1.39583901e-01 -7.01076865e-01 -5.29338360e-01 3.43822330e-01
-2.24393189e-01 1.53186247e-01 3.74290198e-01 -2.07953215e-01
-8.66771519e-01 7.11786598e-02 -1.65046051e-01 -1.05132449e+00
-1.40652627e-01 7.23044217e-01 -6.46963418e-01 1.17832340e-01
2.48590246e-01 1.04736172e-01 -2.21260443e-01 -5.24535060e-01
6.05052888e-01 3.31059724e-01 3.77178997e-01 -6.35534763e-01
1.21341801e+00 3.07176232e-01 3.03530335e-01 -5.53950131e-01
-1.11025512e+00 -4.68512267e-01 -5.81414640e-01 -1.60338074e-01
8.97961557e-01 -9.73173559e-01 -6.58611357e-01 1.04131676e-01
-1.62594879e+00 6.54354990e-02 -1.59784675e-01 3.12682271e-01
-3.01725835e-01 3.12849998e-01 -2.95634985e-01 -6.23251855e-01
-1.43210068e-01 -1.04448152e+00 1.60391331e+00 2.80612946e-01
3.28292310e-01 -6.51656032e-01 -2.50177175e-01 3.93239975e-01
-1.28047377e-01 4.37202126e-01 1.10834539e+00 -5.97951472e-01
-1.20772147e+00 2.08594743e-02 -8.99582982e-01 1.32663632e-02
1.53158724e-01 -2.43892983e-01 -7.66267538e-01 1.22354835e-01
-3.40183616e-01 -3.25307772e-02 7.82380998e-01 -5.25026396e-02
1.37812924e+00 -3.04444313e-01 -6.68894291e-01 7.17641056e-01
1.43819547e+00 -1.43710315e-01 4.80599493e-01 8.01335052e-02
1.42151225e+00 7.41421461e-01 7.04765439e-01 1.58523500e-01
7.35501766e-01 6.74731433e-01 9.67559636e-01 -2.11450726e-01
-2.82649159e-01 -5.74212193e-01 -1.96961448e-01 8.92633498e-01
-9.91177484e-02 -5.33633769e-01 -9.63904560e-01 5.09968340e-01
-1.90155673e+00 -6.17497087e-01 -5.10052323e-01 1.73805225e+00
5.44248939e-01 1.57871857e-01 -2.10261360e-01 -2.67965913e-01
8.84320855e-01 2.42168471e-01 -5.06419182e-01 3.12837839e-01
-2.01207787e-01 1.89676449e-01 3.43792260e-01 2.95901895e-01
-9.97929454e-01 1.04030430e+00 4.05285263e+00 6.71034276e-01
-8.11117053e-01 -1.15862362e-01 4.55672681e-01 1.52939381e-02
-8.31287861e-01 5.75974286e-02 -7.37226546e-01 3.02270681e-01
3.21243852e-01 -7.31003061e-02 1.67253241e-01 7.32640564e-01
-1.37256935e-01 3.25932860e-01 -1.14755118e+00 1.26896465e+00
1.48268104e-01 -1.40008295e+00 5.25475204e-01 1.55434817e-01
4.94580835e-01 4.53046970e-02 -1.13072969e-01 3.35062176e-01
1.93717759e-02 -7.51284182e-01 7.82423437e-01 4.86347437e-01
8.06829512e-01 -5.09182930e-01 6.67183816e-01 1.84365466e-01
-1.51422226e+00 1.69125631e-01 -4.28091645e-01 1.36457205e-01
3.37886184e-01 8.49871039e-01 -7.06729770e-01 1.30598819e+00
7.82905519e-01 1.08017480e+00 -6.00300491e-01 1.15187097e+00
-3.73782605e-01 2.48263016e-01 -5.73253810e-01 -1.12850808e-01
3.58132213e-01 -3.47775161e-01 7.08201706e-01 9.24976528e-01
4.91546005e-01 4.89829957e-01 4.85226735e-02 1.20533526e+00
-3.10831904e-01 9.91309136e-02 -8.09704602e-01 -6.44903556e-02
4.87524420e-01 1.36776233e+00 -7.01545119e-01 -2.28903279e-01
-6.19027436e-01 8.04346979e-01 6.96212173e-01 3.92655373e-01
-1.02941370e+00 -1.58743978e-01 6.54075801e-01 5.81257716e-02
5.04694104e-01 -3.03925335e-01 -2.83928663e-01 -1.31448138e+00
5.46236157e-01 -3.85434628e-01 3.57686311e-01 -1.16455936e+00
-1.49576926e+00 8.52564573e-01 2.40982190e-01 -1.46264720e+00
3.03155303e-01 -5.76853335e-01 -2.53463924e-01 8.73521805e-01
-1.80372334e+00 -1.76391232e+00 -7.88301826e-01 3.69505882e-01
4.50365812e-01 3.11974555e-01 4.35114056e-01 5.36545515e-01
-6.05482638e-01 3.21608186e-01 -7.23370492e-01 3.21208462e-02
1.91315532e-01 -1.07080221e+00 5.97488761e-01 7.10617602e-01
4.46555048e-01 4.75020528e-01 3.00995916e-01 -8.91657948e-01
-1.41696858e+00 -1.48992002e+00 9.07601297e-01 -5.86076438e-01
6.21326149e-01 -7.36858070e-01 -1.12698996e+00 8.13459396e-01
-1.93258688e-01 2.82856822e-01 3.71539146e-01 8.68487433e-02
-4.31201369e-01 4.06193035e-03 -6.50437951e-01 5.93456089e-01
1.76634467e+00 -4.10508037e-01 -6.04949176e-01 5.20652652e-01
1.45756960e+00 -7.56434917e-01 -7.38919199e-01 7.21239388e-01
6.34196848e-02 -7.02588022e-01 1.04519987e+00 -5.09520292e-01
6.40225828e-01 -6.02797568e-01 -1.93851948e-01 -1.06101108e+00
-5.32962203e-01 -3.03778976e-01 -3.62834007e-01 1.40915751e+00
6.69693649e-01 -4.63128835e-01 7.88623750e-01 3.35548401e-01
-7.48000026e-01 -1.03520834e+00 -7.36153901e-01 -6.63916886e-01
-3.67787063e-01 -6.15164459e-01 1.03088284e+00 9.80881274e-01
-4.74828839e-01 6.79137468e-01 -8.13079402e-02 7.33077109e-01
6.40940249e-01 5.48255861e-01 7.06099212e-01 -1.22822869e+00
-1.29359588e-01 -3.30548733e-01 -7.02877998e-01 -1.51650751e+00
2.43350759e-01 -1.25956416e+00 8.03412218e-03 -1.98712087e+00
9.66111571e-02 -8.98089647e-01 -2.30429098e-01 6.39881194e-01
-4.12651002e-01 2.85482317e-01 2.64807731e-01 1.94685951e-01
-6.88138127e-01 1.02020943e+00 1.66915250e+00 -4.08277780e-01
-5.47009446e-02 -1.37086898e-01 -8.58914971e-01 3.46417159e-01
3.99653137e-01 -4.06295002e-01 -5.42694688e-01 -8.03707123e-01
3.54951322e-01 1.62789747e-02 6.61888421e-01 -7.84804404e-01
1.43972665e-01 -9.81691554e-02 3.09783548e-01 -9.96833861e-01
5.49010515e-01 -1.07916069e+00 3.08768779e-01 -1.80004880e-01
1.95167679e-02 -8.74076784e-02 2.66418993e-01 6.97624683e-01
-2.27388740e-01 1.07252531e-01 2.04366669e-01 -8.58589485e-02
-7.34376669e-01 9.45243597e-01 6.82168305e-01 -1.60788998e-01
1.02101779e+00 -1.58477753e-01 -4.32061076e-01 -4.61690538e-02
-6.53239489e-01 6.05591118e-01 4.13604796e-01 7.54400194e-01
7.49708772e-01 -1.50893414e+00 -7.26316273e-01 1.77679598e-01
6.54501915e-01 9.62739587e-01 4.60365623e-01 6.89526021e-01
-4.57514971e-01 4.12715018e-01 -6.52497783e-02 -8.83382440e-01
-8.39786947e-01 9.59234476e-01 5.53074107e-02 -9.07152221e-02
-9.48220074e-01 1.07743895e+00 7.19707072e-01 -3.50027502e-01
-9.52898934e-02 -5.47840059e-01 -3.15585285e-01 -8.32470208e-02
1.39635459e-01 -2.90189058e-01 1.19220927e-01 -8.69538605e-01
-4.34669733e-01 8.32463443e-01 -1.41665954e-02 3.79925877e-01
1.41608942e+00 -1.35126084e-01 -2.84513742e-01 2.83745736e-01
1.20993221e+00 -1.75800845e-01 -1.18226850e+00 -4.25116897e-01
-2.07728475e-01 -6.77729964e-01 -1.10497594e-01 -4.04670447e-01
-1.26570296e+00 7.73017108e-01 2.05418728e-02 3.56525153e-01
1.15116978e+00 7.48070478e-01 7.63495445e-01 2.07372814e-01
3.62463623e-01 -2.04863995e-01 -1.52612943e-02 2.44893745e-01
1.21753705e+00 -1.07705140e+00 1.91584532e-03 -1.30413377e+00
-4.84969646e-01 7.99581170e-01 9.66107368e-01 -4.98667955e-02
5.11699677e-01 -9.58364010e-02 -3.56765091e-01 -6.57905936e-01
-5.99049985e-01 -5.55988669e-01 5.34716606e-01 7.76778698e-01
-2.45178137e-02 2.82273386e-02 1.79699168e-01 6.90000892e-01
-1.94027722e-01 -3.21398079e-01 8.24474841e-02 5.26466966e-01
-1.52456224e-01 -9.29026365e-01 -1.88879311e-01 6.48865938e-01
1.85470954e-01 -2.26978824e-01 -3.73448133e-01 7.41898477e-01
3.34577173e-01 7.03905463e-01 2.86925942e-01 -5.32565653e-01
7.34381735e-01 -2.92332709e-01 4.78374004e-01 -8.83401811e-01
-1.17090680e-01 -6.57976195e-02 1.38539135e-01 -6.42545402e-01
-5.17117321e-01 -5.23159564e-01 -1.34579229e+00 9.72860083e-02
-4.17238712e-01 1.49829596e-01 4.97526437e-01 8.57313514e-01
6.99619293e-01 8.02991331e-01 5.48934460e-01 -9.90474522e-01
7.07319230e-02 -7.93520808e-01 -3.09161395e-01 5.92865705e-01
3.02782536e-01 -1.08103371e+00 -1.51727349e-01 -1.62292004e-01] | [10.262702941894531, 1.5676100254058838] |
19f5492a-89ba-4ec6-a589-cad12612685b | speech-inpainting-context-based-speech | 2306.00489 | null | https://arxiv.org/abs/2306.00489v1 | https://arxiv.org/pdf/2306.00489v1.pdf | Speech inpainting: Context-based speech synthesis guided by video | Audio and visual modalities are inherently connected in speech signals: lip movements and facial expressions are correlated with speech sounds. This motivates studies that incorporate the visual modality to enhance an acoustic speech signal or even restore missing audio information. Specifically, this paper focuses on the problem of audio-visual speech inpainting, which is the task of synthesizing the speech in a corrupted audio segment in a way that it is consistent with the corresponding visual content and the uncorrupted audio context. We present an audio-visual transformer-based deep learning model that leverages visual cues that provide information about the content of the corrupted audio. It outperforms the previous state-of-the-art audio-visual model and audio-only baselines. We also show how visual features extracted with AV-HuBERT, a large audio-visual transformer for speech recognition, are suitable for synthesizing speech. | ['Jesper Jensen', 'Zheng-Hua Tan', 'Gloria Haro', 'Daniel Michelsanti', 'Juan F. Montesinos'] | 2023-06-01 | null | null | null | null | ['speech-synthesis'] | ['speech'] | [ 2.85623997e-01 1.56694815e-01 -5.35927899e-02 -5.44172935e-02
-1.39763045e+00 -4.53987509e-01 4.59723443e-01 -1.99036241e-01
1.51173517e-01 3.98453057e-01 8.57367039e-01 2.49749515e-02
5.42011678e-01 -9.78998542e-02 -9.23451900e-01 -7.29546010e-01
3.90781462e-01 -2.80395865e-01 -1.59240782e-01 -7.89776742e-02
-1.05066493e-03 4.85881567e-01 -2.11796498e+00 9.67276692e-01
2.42888406e-01 1.20907974e+00 1.43648267e-01 1.03255928e+00
-1.21992186e-01 9.14336801e-01 -8.60787332e-01 -5.88476956e-02
-5.53259887e-02 -5.22669017e-01 -4.80363548e-01 2.72613883e-01
8.54413509e-01 -3.93443048e-01 -7.26666272e-01 1.06981313e+00
7.63237476e-01 -2.73518097e-02 6.42426372e-01 -1.64627087e+00
-7.87399530e-01 3.97122920e-01 -4.62288707e-01 7.15818582e-03
6.54523611e-01 2.62682587e-01 1.01927662e+00 -1.28645670e+00
5.99556029e-01 1.50276792e+00 4.65340316e-01 8.31373155e-01
-1.29469979e+00 -5.28217673e-01 1.03342369e-01 4.34698761e-01
-1.16191578e+00 -1.32611191e+00 1.14156044e+00 -3.97526920e-01
8.24343204e-01 5.35636187e-01 7.71791816e-01 1.54181385e+00
4.30492498e-02 1.11757374e+00 9.30529416e-01 -4.60954994e-01
5.60680665e-02 2.79569793e-02 -6.48720562e-01 1.29064441e-01
-5.75935602e-01 5.48446596e-01 -1.16364837e+00 -1.49923610e-02
2.87953705e-01 -3.50384265e-01 -6.18213415e-01 -2.39790648e-01
-8.96458685e-01 4.98296112e-01 1.77339092e-01 3.10701907e-01
-4.00720507e-01 4.09358948e-01 4.80854690e-01 3.67561221e-01
6.08535945e-01 4.47519757e-02 -1.09742641e-01 -9.62507799e-02
-1.42751026e+00 1.71225548e-01 1.51615769e-01 5.92483401e-01
3.07102323e-01 9.54702854e-01 -3.71457458e-01 1.04953778e+00
5.73210120e-01 6.58868313e-01 3.91424239e-01 -1.21743119e+00
2.69721627e-01 -2.06399992e-01 -1.54091835e-01 -4.91252750e-01
-5.62440865e-02 5.69968484e-03 -5.45204401e-01 7.29757965e-01
-2.71427892e-02 2.38239571e-01 -1.00417411e+00 1.86478400e+00
3.47647309e-01 4.93243277e-01 6.93136156e-02 9.20094490e-01
1.43058491e+00 8.69475305e-01 -3.13757546e-02 -3.40016693e-01
1.20758092e+00 -9.54546750e-01 -1.31985617e+00 -3.00609380e-01
-1.72480017e-01 -1.15683305e+00 1.22126997e+00 3.19856197e-01
-1.48160720e+00 -7.62083054e-01 -9.95362341e-01 -4.72160310e-01
-2.51859874e-02 1.36224806e-01 -2.54360959e-03 4.20767546e-01
-1.42240083e+00 2.48513088e-01 -4.25114214e-01 -2.02087667e-02
2.72896945e-01 -2.08231136e-01 -6.42163634e-01 1.58717737e-01
-1.03360331e+00 5.68496168e-01 -1.80005774e-01 -6.11295877e-03
-1.49201059e+00 -9.77015615e-01 -1.23719192e+00 1.13970093e-01
4.55592312e-02 -5.36872208e-01 1.68672824e+00 -1.25528455e+00
-1.62773883e+00 8.32039595e-01 -5.25295198e-01 -4.12409544e-01
4.04110968e-01 8.93794894e-02 -6.10671043e-01 6.17054582e-01
-1.64615847e-02 9.14212346e-01 1.82282937e+00 -1.69992638e+00
-2.67022163e-01 -2.16338292e-01 -4.79093999e-01 2.24135920e-01
1.11070342e-01 4.18452248e-02 -4.90155578e-01 -1.33179057e+00
-6.33445457e-02 -4.96770650e-01 5.35125077e-01 3.85818481e-01
-4.76917535e-01 2.18179137e-01 1.18618274e+00 -1.21474612e+00
7.51589417e-01 -2.57403874e+00 2.11152285e-01 -1.71991020e-01
1.71969578e-01 3.12745422e-01 -6.03926897e-01 4.84735519e-01
-3.95186692e-01 -5.29473014e-02 -8.46726671e-02 -8.97403181e-01
7.45087191e-02 -1.53260738e-01 -8.49880755e-01 5.73041081e-01
2.56824911e-01 9.01861012e-01 -5.73495030e-01 -4.58033234e-01
4.17280257e-01 1.11390340e+00 -4.18830454e-01 3.53705287e-01
-1.07471734e-01 6.08938038e-01 3.83425176e-01 1.07062697e+00
7.08715081e-01 5.33489764e-01 -2.88754225e-01 -5.03190339e-01
1.30431652e-01 4.43826884e-01 -7.55827904e-01 1.68184030e+00
-5.76349616e-01 1.23059118e+00 9.81874943e-01 -5.86281300e-01
7.61614978e-01 8.65910411e-01 4.68315989e-01 -9.69105065e-01
9.59169939e-02 9.33976397e-02 -5.34239829e-01 -6.25460625e-01
3.30665946e-01 -3.31861377e-01 3.58754843e-01 1.69187084e-01
2.97894359e-01 -6.63217306e-01 -4.40140843e-01 1.55770898e-01
7.95187593e-01 -1.11005753e-01 -1.19026648e-02 4.65861678e-01
4.37905014e-01 -8.20231199e-01 1.72813043e-01 4.51201439e-01
-5.46353877e-01 1.06466413e+00 2.40324080e-01 4.11252171e-01
-1.14743149e+00 -1.47874320e+00 2.22789049e-02 1.21358776e+00
-2.69670039e-01 -5.36103666e-01 -8.43356192e-01 -3.16999197e-01
3.27798948e-02 7.56096423e-01 -6.69664264e-01 -3.95148426e-01
-1.12187475e-01 4.23983723e-01 7.65800178e-01 4.60147619e-01
-6.81739822e-02 -1.30979419e+00 7.90064186e-02 -3.99204306e-02
-5.83024085e-01 -1.14008975e+00 -7.49931574e-01 -9.55111310e-02
-3.60606790e-01 -7.74773180e-01 -1.04690135e+00 -9.80730414e-01
2.07238838e-01 3.11783582e-01 7.97927797e-01 -2.65304595e-01
-3.71236473e-01 7.50659049e-01 -2.12731525e-01 -3.99823904e-01
-9.07533467e-01 -6.72350109e-01 1.28216982e-01 4.02080745e-01
-1.93329722e-01 -8.12111020e-01 -3.16435486e-01 -2.51083493e-01
-9.71172929e-01 1.26217315e-02 3.24858457e-01 8.98629665e-01
7.01631665e-01 -3.36241215e-01 7.60186851e-01 8.07949826e-02
7.20222831e-01 -1.32302538e-01 -9.52058807e-02 1.48859667e-02
2.78300587e-02 -2.91350365e-01 4.40786332e-01 -6.93786383e-01
-9.97735739e-01 8.71063396e-02 -5.85064888e-01 -1.18735027e+00
-2.62039393e-01 1.44846767e-01 -4.83384132e-01 -5.18771894e-02
5.33449411e-01 3.59897405e-01 3.16972852e-01 -6.56782985e-01
5.85867345e-01 1.04467881e+00 1.09925723e+00 -9.62904841e-02
5.82686305e-01 5.95742226e-01 -8.70639905e-02 -1.26372087e+00
-3.64466608e-01 -4.05220419e-01 -2.44471192e-01 -4.53366309e-01
6.50608540e-01 -1.09755468e+00 -7.04133153e-01 5.68210065e-01
-1.43201137e+00 -3.49861920e-01 -5.25256336e-01 3.86126161e-01
-9.05501723e-01 6.25404239e-01 -8.09178054e-01 -1.04955959e+00
-2.30009452e-01 -1.39490724e+00 1.70891452e+00 -2.63767302e-01
-1.78575113e-01 -5.58723569e-01 1.00968949e-01 5.88081419e-01
2.24727780e-01 3.73212993e-02 5.71661592e-01 -1.24998920e-01
-2.86127448e-01 9.71737038e-03 -6.68383911e-02 6.30895019e-01
2.34722242e-01 2.71738082e-01 -1.88995790e+00 -3.53286535e-01
-9.70620215e-02 -4.86172616e-01 1.21272039e+00 8.28973651e-01
1.06552684e+00 -5.14587700e-01 -2.82610077e-02 6.28494859e-01
6.35371983e-01 1.60367519e-01 7.87206471e-01 -2.38472998e-01
5.26098013e-01 8.23202550e-01 5.17175853e-01 2.81410456e-01
1.77503079e-02 9.72454011e-01 7.02318251e-01 -4.38459158e-01
-1.10941625e+00 -7.70669341e-01 7.77176917e-01 8.35422695e-01
5.89482427e-01 -1.55205920e-01 -3.88737231e-01 7.97658086e-01
-1.49562275e+00 -1.07839203e+00 3.19678605e-01 1.86811507e+00
1.13134634e+00 -3.04514438e-01 1.64035305e-01 5.31919003e-01
7.78317988e-01 5.38684249e-01 -4.49592113e-01 -4.70193952e-01
-3.77416074e-01 3.90230298e-01 -2.44300708e-01 9.61077869e-01
-1.04890335e+00 9.03572381e-01 6.80527592e+00 1.11723840e+00
-1.45637906e+00 1.79843917e-01 2.81673968e-01 -4.40316319e-01
-4.86741453e-01 -4.60189223e-01 -1.33436203e-01 2.25244984e-01
1.02723944e+00 1.09520942e-01 6.63080752e-01 5.82470238e-01
6.58336997e-01 1.63785055e-01 -1.15418136e+00 1.34438038e+00
4.31683898e-01 -1.40295660e+00 2.66858518e-01 -1.46108881e-01
3.89582306e-01 -2.42311150e-01 7.24080384e-01 -4.04577777e-02
-2.42566854e-01 -1.30181885e+00 1.28927457e+00 4.72825140e-01
1.34189272e+00 -6.93605661e-01 1.21847354e-01 -1.07758082e-01
-1.37738538e+00 3.08158528e-02 2.47254878e-01 3.59830707e-01
2.85696387e-01 1.51019230e-01 -9.77523863e-01 1.31369233e-01
1.12176657e+00 7.67719269e-01 -2.96882212e-01 1.04207420e+00
-3.91623288e-01 5.80743968e-01 1.21510342e-01 6.95678294e-01
-2.49477431e-01 4.59345967e-01 1.06349027e+00 1.15423822e+00
3.50823969e-01 -5.18626928e-01 -2.56101698e-01 8.83648455e-01
-1.67112142e-01 1.67603418e-01 -9.83060837e-01 -2.81573057e-01
5.98292649e-01 9.23550904e-01 1.21116880e-02 -1.72076851e-01
-3.98098052e-01 8.94794643e-01 -2.20410064e-01 5.89366078e-01
-7.42638350e-01 -2.67712086e-01 1.15686858e+00 8.88441503e-02
3.90917063e-01 2.15177611e-01 -9.87911969e-02 -9.21970785e-01
1.83244199e-01 -1.15114582e+00 -1.71602190e-01 -1.52512681e+00
-1.11593926e+00 6.35655224e-01 -4.31412607e-01 -1.36703348e+00
-4.85558212e-01 -3.82078826e-01 -5.36192596e-01 7.90469348e-01
-1.66724670e+00 -1.43208492e+00 -1.23347633e-01 9.16850626e-01
7.39877462e-01 -1.97487608e-01 6.90173686e-01 2.41683915e-01
-6.76700622e-02 8.59922707e-01 -2.96105370e-02 6.49683028e-02
1.07126999e+00 -7.87848175e-01 4.42438543e-01 7.62492180e-01
4.10511643e-01 9.68483612e-02 1.01315629e+00 -4.74301100e-01
-1.27925098e+00 -1.13659108e+00 9.38024223e-01 -4.17043604e-02
4.82373983e-01 -4.48838532e-01 -9.55258369e-01 6.38919830e-01
4.18987036e-01 1.07529745e-01 5.94284236e-01 -6.72232568e-01
-7.45869100e-01 -2.80667335e-01 -1.12433410e+00 5.94495773e-01
6.25834346e-01 -1.51081991e+00 -6.67169094e-01 6.22783601e-03
1.08579350e+00 -3.17035228e-01 -4.01915044e-01 2.04940885e-01
4.75328088e-01 -8.58624101e-01 1.41722274e+00 -5.79507411e-01
2.25007758e-01 -4.68326211e-01 -3.41512412e-01 -1.61613095e+00
1.94744349e-01 -9.26759124e-01 -2.51912624e-01 1.48335159e+00
1.41738951e-01 -6.48729652e-02 3.68433297e-01 -1.26315147e-01
-3.56314242e-01 -5.44383265e-02 -1.42444694e+00 -5.37442684e-01
-1.00872941e-01 -7.53335714e-01 3.63753676e-01 9.11013484e-01
9.35448483e-02 2.08154172e-01 -7.55157828e-01 9.96065959e-02
5.07110476e-01 -2.26663366e-01 7.04055727e-01 -9.91342127e-01
-2.14770213e-01 -2.80926913e-01 -4.17268664e-01 -6.80828214e-01
6.80600166e-01 -7.17101097e-01 2.67768472e-01 -1.50082386e+00
-2.55938560e-01 4.43820536e-01 2.08379105e-02 6.13604248e-01
2.92252839e-01 5.12961268e-01 3.80937129e-01 -6.58591613e-02
-4.90019806e-02 9.75445688e-01 1.40767288e+00 -6.23304665e-01
-2.48204783e-01 -3.55227594e-03 -5.49542904e-01 5.28157830e-01
4.78145450e-01 -2.50894248e-01 -3.21333796e-01 -1.19767100e-01
-2.17153147e-01 6.98945880e-01 7.08013833e-01 -6.73944533e-01
2.23742947e-01 7.93477744e-02 2.50654936e-01 -6.19379997e-01
1.31635559e+00 -7.80123413e-01 -1.60665736e-02 5.86656593e-02
-6.74325645e-01 -2.48718783e-01 5.82630873e-01 6.41336560e-01
-8.21390092e-01 1.61925584e-01 1.08616579e+00 1.28767937e-01
-3.64282310e-01 1.48149669e-01 -7.50887275e-01 1.04864888e-01
6.74563706e-01 -1.17693380e-01 -2.96810776e-01 -1.11109936e+00
-1.09657490e+00 -4.83141333e-01 2.77831823e-01 5.79687595e-01
1.07159030e+00 -1.65142274e+00 -8.50687981e-01 4.58114982e-01
1.83194980e-01 -6.52981520e-01 5.81204414e-01 8.14855993e-01
-1.13968998e-01 2.03496620e-01 -2.83645213e-01 -6.41776979e-01
-1.84746969e+00 7.33935356e-01 5.28764844e-01 5.86244404e-01
-6.45119131e-01 9.39058721e-01 4.27167892e-01 3.16690058e-02
8.98003697e-01 -4.26474094e-01 -1.04914911e-01 2.07390934e-01
7.92662263e-01 1.99772373e-01 2.22158924e-01 -1.08567369e+00
-3.10548514e-01 3.82607669e-01 2.93043286e-01 -7.18605816e-01
1.18042231e+00 -3.41105282e-01 -1.43645182e-02 6.41360283e-01
1.41211140e+00 4.63207781e-01 -1.40725613e+00 -2.40742683e-01
-6.82487011e-01 -5.11264801e-01 4.70756561e-01 -7.93029070e-01
-1.20197940e+00 1.43708956e+00 5.35712957e-01 9.23201367e-02
1.28050447e+00 1.93930700e-01 6.62043869e-01 -3.03620130e-01
-3.06326449e-01 -1.01712239e+00 6.12788498e-01 2.12265491e-01
1.64501607e+00 -9.86564159e-01 -4.77485180e-01 -3.83618772e-01
-8.40583265e-01 1.16842890e+00 2.54978806e-01 4.14289147e-01
6.96680248e-01 6.03248119e-01 4.69174176e-01 1.86721623e-01
-8.70762885e-01 -4.21319306e-01 6.48665369e-01 1.22818005e+00
2.62535870e-01 -1.16022140e-01 7.28713632e-01 4.89602506e-01
-4.34513688e-01 -3.15841913e-01 2.87731439e-01 6.24217808e-01
-3.44373673e-01 -7.41388023e-01 -7.73911655e-01 -3.75695020e-01
-4.14077342e-01 -3.54683191e-01 -1.01576912e+00 3.20809722e-01
-2.00282216e-01 1.40072596e+00 4.26269323e-02 -5.15469968e-01
2.20707536e-01 3.38610679e-01 5.17867029e-01 -3.26256782e-01
-4.09368813e-01 8.02234530e-01 -1.30840880e-03 -7.15909362e-01
-1.71297595e-01 -6.36212528e-01 -1.04227197e+00 -2.33108819e-01
1.31025678e-02 -2.73471951e-01 6.97095811e-01 5.84887326e-01
3.70823175e-01 6.61744118e-01 5.91193259e-01 -1.38030505e+00
-2.26981476e-01 -9.50267017e-01 -8.31876159e-01 4.10355419e-01
1.34827936e+00 -5.69216490e-01 -9.11619008e-01 1.36241227e-01] | [14.433122634887695, 5.148718357086182] |
ccfdf37e-be64-4171-98fb-bc101ce43370 | relevant-entity-selection-knowledge-graph | 2306.16296 | null | https://arxiv.org/abs/2306.16296v1 | https://arxiv.org/pdf/2306.16296v1.pdf | Relevant Entity Selection: Knowledge Graph Bootstrapping via Zero-Shot Analogical Pruning | Knowledge Graph Construction (KGC) can be seen as an iterative process starting from a high quality nucleus that is refined by knowledge extraction approaches in a virtuous loop. Such a nucleus can be obtained from knowledge existing in an open KG like Wikidata. However, due to the size of such generic KGs, integrating them as a whole may entail irrelevant content and scalability issues. We propose an analogy-based approach that starts from seed entities of interest in a generic KG, and keeps or prunes their neighboring entities. We evaluate our approach on Wikidata through two manually labeled datasets that contain either domain-homogeneous or -heterogeneous seed entities. We empirically show that our analogy-based approach outperforms LSTM, Random Forest, SVM, and MLP, with a drastically lower number of parameters. We also evaluate its generalization potential in a transfer learning setting. These results advocate for the further integration of analogy-based inference in tasks related to the KG lifecycle. | ['Pierre Monnin', 'Miguel Couceiro', 'Lucas Jarnac'] | 2023-06-28 | null | null | null | null | ['graph-construction', 'transfer-learning'] | ['graphs', 'miscellaneous'] | [-1.31147519e-01 8.91714156e-01 -3.72457832e-01 -8.38929042e-02
-6.90569401e-01 -4.96982396e-01 8.66744518e-01 5.59143603e-01
-4.09663528e-01 1.14238691e+00 6.75558969e-02 -3.05659682e-01
-5.02805591e-01 -1.10405266e+00 -1.19083869e+00 -3.74353766e-01
-3.95001262e-01 9.77914035e-01 3.79816204e-01 -4.95683923e-02
3.01739238e-02 3.12756002e-01 -1.55167997e+00 3.56236547e-01
1.07721627e+00 9.23468709e-01 1.31109864e-01 3.71672958e-01
-2.60530055e-01 1.05481553e+00 -4.73126024e-01 -9.06452417e-01
9.09362547e-03 1.85511768e-01 -1.54673076e+00 -3.28567237e-01
5.70999444e-01 1.74213946e-01 1.04128465e-01 9.80143487e-01
1.44730076e-01 -1.82520337e-02 6.91865027e-01 -1.35692787e+00
-7.46633530e-01 1.40632117e+00 -6.60665929e-02 -1.89701065e-01
1.41358405e-01 -2.34976739e-01 1.27159619e+00 -6.30385697e-01
1.04430532e+00 9.50422466e-01 1.04472601e+00 1.61541924e-01
-1.23246121e+00 -2.75309265e-01 2.81146675e-01 5.37808299e-01
-1.28399706e+00 -1.22738495e-01 5.04840851e-01 -3.88391525e-01
1.22249782e+00 -7.15082213e-02 7.92003870e-01 1.13065898e+00
-2.98735201e-01 8.66252363e-01 9.27118659e-01 -8.90684366e-01
3.66228342e-01 3.65825742e-01 2.74730831e-01 9.72371638e-01
6.62436783e-01 -1.77118406e-01 -5.86354911e-01 -1.49042591e-01
3.43345284e-01 -7.12954521e-01 -1.56968549e-01 -7.14190543e-01
-1.04986656e+00 7.01822162e-01 4.49669629e-01 2.99098402e-01
-4.66632426e-01 2.92899489e-01 3.27778369e-01 3.02638918e-01
4.92550641e-01 7.49011815e-01 -9.83926117e-01 1.06489193e-02
-1.02746928e+00 2.39423618e-01 1.41206789e+00 1.27267563e+00
9.29069400e-01 -2.52603859e-01 1.65935621e-01 4.80986744e-01
1.59293085e-01 5.24369292e-02 1.86577812e-01 -7.88571537e-01
3.94809693e-01 7.17522264e-01 2.05257073e-01 -7.07594633e-01
-4.78539616e-01 -6.67036533e-01 -3.65859598e-01 -1.98870935e-02
5.69252133e-01 -2.17712760e-01 -8.81129920e-01 1.67047155e+00
6.11095428e-01 3.21581990e-01 2.40232676e-01 3.83262396e-01
9.87277329e-01 1.60615385e-01 3.01664054e-01 2.02084512e-01
1.29402685e+00 -1.19151282e+00 -4.02292550e-01 -1.52716577e-01
8.30617011e-01 -1.94410682e-02 7.08010197e-01 6.05140984e-01
-8.65078449e-01 -2.72704065e-01 -1.02335131e+00 -1.88426599e-01
-1.13595200e+00 1.59853995e-01 8.74780357e-01 5.41332841e-01
-1.12227178e+00 1.00239551e+00 -6.38027191e-01 -5.41463375e-01
4.29171950e-01 3.25820476e-01 -4.62665021e-01 1.77220538e-01
-1.56209517e+00 1.33423293e+00 1.08602417e+00 3.39925885e-02
-5.89740992e-01 -8.71935129e-01 -9.01337683e-01 2.11816773e-01
8.49658549e-01 -9.55966234e-01 1.04219031e+00 -5.96905470e-01
-1.10703838e+00 7.66245484e-01 2.30780095e-01 -9.48552907e-01
3.25304061e-01 -4.55780506e-01 -3.97961974e-01 -1.57769814e-01
1.76913172e-01 7.57310152e-01 6.34967744e-01 -1.45248687e+00
-9.12120223e-01 -2.15834036e-01 4.14556086e-01 1.15801170e-01
-1.09335057e-01 -3.48074734e-01 -5.08314908e-01 -2.28221282e-01
-2.53416836e-01 -8.06595027e-01 -2.00375736e-01 -5.63277304e-01
-6.07963979e-01 -3.47206146e-01 2.67774284e-01 -8.38414013e-01
1.19690239e+00 -1.41824543e+00 1.71030268e-01 2.99403876e-01
2.01744631e-01 2.94023335e-01 7.75541589e-02 5.55410981e-01
1.24448374e-01 3.26034486e-01 8.80862847e-02 -4.34069969e-02
2.12309539e-01 2.48077735e-01 -1.84145510e-01 -1.16861351e-01
2.38944337e-01 1.27322567e+00 -1.17916465e+00 -7.56780446e-01
-1.47801787e-01 2.10397214e-01 -3.84632438e-01 -3.02915424e-01
-8.46687436e-01 -1.12433203e-01 -3.92824411e-01 7.11169124e-01
2.27656424e-01 -6.95875108e-01 6.73473477e-01 -3.53529304e-01
1.40121207e-01 3.90754700e-01 -1.10601842e+00 1.74359477e+00
-5.68132222e-01 3.31717581e-01 -2.70573378e-01 -1.20745611e+00
7.20823765e-01 1.91787109e-01 1.12553395e-01 -1.95098132e-01
-1.30035594e-01 2.29279697e-01 -1.58217877e-01 -4.24004525e-01
8.35277796e-01 4.69799042e-02 -6.45435527e-02 2.69421488e-01
6.12716258e-01 1.88698217e-01 3.60220402e-01 3.48432302e-01
1.31376112e+00 5.99600852e-01 5.58002770e-01 -2.59081036e-01
3.04561377e-01 3.26158673e-01 2.42793977e-01 9.04660225e-01
9.66345221e-02 2.24210806e-02 5.23477137e-01 -3.29977661e-01
-7.81897604e-01 -9.42104459e-01 -1.34934232e-01 9.07486856e-01
-1.06425025e-01 -5.89648783e-01 -6.38682246e-01 -1.42982912e+00
3.25320333e-01 1.00303543e+00 -7.98431873e-01 6.08082302e-02
-3.93245012e-01 -5.22967756e-01 7.45405972e-01 4.99053836e-01
3.39596123e-01 -1.35650778e+00 -3.41485560e-01 3.66445541e-01
-1.22042418e-01 -1.27144933e+00 3.64372283e-01 5.22657037e-01
-6.68195367e-01 -1.19828057e+00 -2.62450188e-01 -7.38592565e-01
4.71536338e-01 -4.10496294e-01 1.60077727e+00 -1.72991231e-01
4.53289412e-02 5.11073470e-01 -3.59904379e-01 -3.49361271e-01
-4.99827266e-01 6.98413491e-01 -1.38232753e-01 -2.16610149e-01
5.33864319e-01 -7.76724458e-01 -5.75301014e-02 -2.24483266e-01
-4.47841942e-01 1.58600926e-01 6.80646360e-01 7.46509552e-01
3.12359601e-01 3.94448698e-01 9.65318441e-01 -1.40521836e+00
6.32996500e-01 -8.37643385e-01 -5.45038044e-01 7.82679021e-01
-9.79951322e-01 5.18242180e-01 5.45482337e-01 -3.00960004e-01
-1.19907379e+00 -7.10538849e-02 3.05081189e-01 -2.56393194e-01
-2.49319300e-01 1.22062361e+00 -2.47605354e-01 6.01529740e-02
9.11557555e-01 1.06340297e-01 -3.58180791e-01 -6.09731853e-01
9.32490528e-01 3.76464695e-01 5.90749264e-01 -1.00117028e+00
1.00263000e+00 1.79902717e-01 -2.76776049e-02 -5.37358999e-01
-9.52747226e-01 -1.94136500e-01 -8.63832176e-01 -1.38279572e-01
3.28160822e-01 -1.00826406e+00 -6.19965672e-01 7.28188306e-02
-1.07339942e+00 -4.95310575e-01 -5.58586895e-01 2.67107159e-01
-3.11759323e-01 2.97055811e-01 -5.84118068e-01 -4.80684102e-01
-4.41443175e-01 -4.63331461e-01 8.60353410e-01 1.48962796e-01
-4.30887014e-01 -1.37024271e+00 1.19567275e-01 4.70529824e-01
2.01597676e-01 2.41085410e-01 1.20481861e+00 -1.15176976e+00
-8.97752821e-01 -2.47750714e-01 -1.97101086e-01 9.88319889e-02
1.31889999e-01 5.50357327e-02 -9.16812956e-01 1.08695813e-01
-7.33542085e-01 -5.52021861e-01 8.12794030e-01 -1.11617222e-02
6.63468778e-01 -4.54861820e-01 -6.48761511e-01 2.10864961e-01
1.46028101e+00 -1.16234891e-01 3.90951395e-01 7.73823977e-01
7.65244484e-01 6.94864690e-01 5.39942980e-01 -1.87934469e-02
8.35229695e-01 4.35045063e-01 2.48229429e-01 8.05871040e-02
-9.46057960e-02 -6.26893044e-01 1.04480319e-01 5.84838390e-01
-3.06115299e-01 -4.79822122e-02 -1.01346672e+00 1.07009554e+00
-1.94767630e+00 -8.79594505e-01 -6.09944165e-02 1.97655272e+00
1.34657395e+00 2.89169908e-01 2.38878410e-02 8.89012441e-02
6.49076998e-01 -1.60158634e-01 -2.62238860e-01 -3.29964072e-01
-4.41640057e-02 4.28801239e-01 6.16866767e-01 3.95776212e-01
-1.15266049e+00 1.35492849e+00 5.88637590e+00 1.13778973e+00
-7.98939586e-01 3.01030022e-03 2.38082767e-01 1.50980994e-01
-2.83745497e-01 4.12207603e-01 -1.13049531e+00 3.67124647e-01
1.03745294e+00 -1.85898811e-01 3.80101889e-01 1.20708489e+00
-4.15537775e-01 -2.43026689e-01 -1.33004522e+00 2.93507040e-01
-2.02092230e-01 -1.79653490e+00 1.61564827e-01 1.60963565e-01
7.12496877e-01 2.69273400e-01 -4.59674805e-01 6.99113607e-01
1.02385080e+00 -9.16843891e-01 6.87053680e-01 5.07314265e-01
4.26877379e-01 -4.47344989e-01 5.47155380e-01 4.14665908e-01
-9.78463769e-01 2.48636752e-01 -1.88488513e-01 1.67299584e-01
-3.79366241e-02 7.15931177e-01 -1.26511741e+00 1.29746604e+00
7.29290128e-01 5.33491015e-01 -8.54483724e-01 9.52310801e-01
-6.58738136e-01 6.59556031e-01 -3.49227458e-01 -8.94881561e-02
1.69491410e-01 -1.98678136e-01 1.85217962e-01 1.30464578e+00
7.26587102e-02 -3.60746950e-01 -7.77704865e-02 1.00583732e+00
-2.13596165e-01 5.23630641e-02 -6.32779419e-01 -1.85782164e-02
6.04491711e-01 1.54358506e+00 -7.19139457e-01 -5.04446805e-01
-4.33153093e-01 6.37276232e-01 9.07189786e-01 4.14427340e-01
-8.63966405e-01 -3.94889444e-01 2.74674781e-02 6.08770829e-03
7.46166110e-01 2.13637203e-01 -2.85242200e-02 -1.04941630e+00
2.79422015e-01 -6.61966920e-01 5.31058192e-01 -8.41081202e-01
-1.32478034e+00 5.71722806e-01 2.21610039e-01 -7.28619695e-01
-4.79103476e-01 -5.07646501e-01 -3.85225058e-01 5.32771230e-01
-1.62447977e+00 -1.58062410e+00 -9.93912742e-02 2.44800344e-01
-7.77895600e-02 1.98101681e-02 8.07805657e-01 1.48111433e-01
-3.98557127e-01 3.73980075e-01 -4.04634513e-02 1.27682686e-01
3.47185910e-01 -1.72976017e+00 6.47507250e-01 6.79771304e-01
5.42732239e-01 6.09009206e-01 4.51187223e-01 -1.01364708e+00
-1.06488621e+00 -1.29795945e+00 1.23739278e+00 -7.12755263e-01
1.03003883e+00 -3.85859162e-01 -9.77445781e-01 1.12017453e+00
2.83540636e-01 -5.33266850e-02 4.01645958e-01 8.76286626e-01
-4.68744457e-01 -1.34526053e-02 -9.62900162e-01 6.22096539e-01
1.34677947e+00 -4.22435850e-01 -7.94179499e-01 3.05146396e-01
7.60060906e-01 -8.37034881e-02 -1.39986181e+00 5.36109865e-01
5.24408519e-01 -6.86393619e-01 9.59005177e-01 -9.36218023e-01
3.84685010e-01 -1.48539409e-01 1.91615880e-01 -1.52027535e+00
-1.57902852e-01 -6.59207344e-01 -6.31620467e-01 1.43672168e+00
9.14540529e-01 -6.02546394e-01 1.05793166e+00 5.92306793e-01
-2.43039373e-02 -7.69600511e-01 -6.57131553e-01 -1.10441935e+00
2.31416062e-01 -4.76522326e-01 5.78952551e-01 1.21412110e+00
5.06490767e-01 6.99278533e-01 -5.45869544e-02 4.14295942e-01
6.33903325e-01 2.44629085e-01 7.83006608e-01 -1.65483570e+00
-3.05082440e-01 -1.89519733e-01 -1.25028625e-01 -5.58750927e-01
4.89484787e-01 -1.21539104e+00 -2.04644576e-01 -1.82969344e+00
-7.65192434e-02 -9.30710554e-01 -2.06594020e-01 9.30248380e-01
-1.56421602e-01 -3.44705462e-01 -2.51283884e-01 -3.16355377e-02
-8.02510142e-01 3.58817458e-01 7.43563771e-01 -1.73970580e-01
-1.19817406e-01 -3.10058951e-01 -7.92946160e-01 8.18352878e-01
7.17418194e-01 -5.11673272e-01 -4.10416126e-01 -5.42835779e-02
6.83414042e-01 -1.00380458e-01 4.46382582e-01 -9.65286970e-01
5.17524362e-01 8.31962153e-02 1.88622549e-01 -3.97687286e-01
2.82140732e-01 -5.64070165e-01 4.57443804e-01 1.72453806e-01
-1.25570223e-01 -5.45687079e-01 2.08801553e-01 6.18300736e-01
-9.78723913e-02 -4.96422529e-01 2.18518347e-01 -2.43775517e-01
-7.12125778e-01 -8.36652219e-02 -1.36125550e-01 2.51395315e-01
7.21866727e-01 -3.42107892e-01 -5.28454781e-01 -1.20600134e-01
-1.02170384e+00 3.31539661e-01 3.35578322e-01 2.95821667e-01
1.16616815e-01 -1.10370874e+00 -5.97304344e-01 -1.14385933e-01
1.61796242e-01 1.60462379e-01 -1.48688182e-01 6.99864626e-01
-3.71072173e-01 6.03913009e-01 6.33502230e-02 -2.10378453e-01
-1.00722861e+00 7.57399797e-01 2.33523950e-01 -8.54652822e-01
-6.34966671e-01 6.60157859e-01 -2.42250070e-01 -7.25893617e-01
2.63825595e-01 -5.80496609e-01 -3.92090172e-01 1.97644353e-01
-1.81083784e-01 2.78149158e-01 3.77985984e-01 -1.26136318e-01
-1.71290889e-01 1.36549279e-01 -1.30851582e-01 3.68969999e-02
1.37486053e+00 1.34705484e-01 -8.99601653e-02 2.08779007e-01
4.17204231e-01 2.63561383e-02 -8.28793406e-01 -4.63262379e-01
9.45677817e-01 1.36733964e-01 -1.22841395e-01 -1.22440195e+00
-9.41098869e-01 2.70713061e-01 -2.04581246e-01 4.40119892e-01
6.99787378e-01 2.39087433e-01 3.48528683e-01 8.99361193e-01
8.18327010e-01 -1.01135504e+00 -4.52145755e-01 3.28427315e-01
6.97378516e-01 -9.85165119e-01 2.03688443e-01 -6.04583025e-01
-4.49468434e-01 8.58122766e-01 5.31054616e-01 5.03670983e-02
6.60040259e-01 2.28458032e-01 -4.10558432e-01 -4.63411868e-01
-1.02965474e+00 -5.63514292e-01 3.11886758e-01 8.28141868e-01
1.21717267e-01 1.74732357e-02 -9.73798335e-02 8.56804490e-01
-3.27379644e-01 2.51951188e-01 4.09013391e-01 8.14293087e-01
-2.44040608e-01 -1.29085588e+00 6.68754503e-02 5.47454655e-01
-3.04730833e-01 -5.21744072e-01 -4.84689415e-01 1.30543625e+00
3.25206965e-01 6.19995296e-01 -3.10191542e-01 -1.92809388e-01
1.45040765e-01 5.32356441e-01 7.45869219e-01 -7.33223379e-01
-7.30733275e-01 -3.94143999e-01 9.69847977e-01 -2.82813102e-01
-4.75528687e-01 -6.30722582e-01 -1.16949654e+00 -1.74807962e-02
-6.64780915e-01 3.83592427e-01 4.85451579e-01 1.07019961e+00
5.95988572e-01 3.37619215e-01 -1.26595810e-01 -3.64522576e-01
-2.70490199e-01 -1.09102130e+00 -5.82898557e-01 1.91667154e-01
-2.57320493e-01 -7.90658891e-01 -2.21263662e-01 3.17733049e-01] | [9.281012535095215, 8.260896682739258] |
d034da15-cfc6-41b1-9cb2-f0199a958bc1 | zero-shot-everything-sketch-based-image | 2303.14348 | null | https://arxiv.org/abs/2303.14348v1 | https://arxiv.org/pdf/2303.14348v1.pdf | Zero-Shot Everything Sketch-Based Image Retrieval, and in Explainable Style | This paper studies the problem of zero-short sketch-based image retrieval (ZS-SBIR), however with two significant differentiators to prior art (i) we tackle all variants (inter-category, intra-category, and cross datasets) of ZS-SBIR with just one network (``everything''), and (ii) we would really like to understand how this sketch-photo matching operates (``explainable''). Our key innovation lies with the realization that such a cross-modal matching problem could be reduced to comparisons of groups of key local patches -- akin to the seasoned ``bag-of-words'' paradigm. Just with this change, we are able to achieve both of the aforementioned goals, with the added benefit of no longer requiring external semantic knowledge. Technically, ours is a transformer-based cross-modal network, with three novel components (i) a self-attention module with a learnable tokenizer to produce visual tokens that correspond to the most informative local regions, (ii) a cross-attention module to compute local correspondences between the visual tokens across two modalities, and finally (iii) a kernel-based relation network to assemble local putative matches and produce an overall similarity metric for a sketch-photo pair. Experiments show ours indeed delivers superior performances across all ZS-SBIR settings. The all important explainable goal is elegantly achieved by visualizing cross-modal token correspondences, and for the first time, via sketch to photo synthesis by universal replacement of all matched photo patches. Code and model are available at \url{https://github.com/buptLinfy/ZSE-SBIR}. | ['Yonggang Qi', 'Yi-Zhe Song', 'Timothy Hospedales', 'Da Li', 'Mingkang Li', 'Fengyin Lin'] | 2023-03-25 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Lin_Zero-Shot_Everything_Sketch-Based_Image_Retrieval_and_in_Explainable_Style_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Lin_Zero-Shot_Everything_Sketch-Based_Image_Retrieval_and_in_Explainable_Style_CVPR_2023_paper.pdf | cvpr-2023-1 | ['sketch-based-image-retrieval'] | ['computer-vision'] | [ 1.86471805e-01 7.91479126e-02 -5.05040511e-02 -1.89075962e-01
-1.17392874e+00 -7.98651159e-01 9.26671386e-01 -1.14890737e-02
-1.01179276e-02 2.63522536e-01 2.40065768e-01 -8.04745853e-02
-5.42573571e-01 -7.57320762e-01 -8.55596066e-01 -6.03782475e-01
1.22919224e-01 3.53086352e-01 4.15587425e-03 -3.99451762e-01
2.51995564e-01 5.64320385e-01 -1.74262607e+00 5.34342766e-01
4.36052322e-01 1.21306801e+00 1.22744203e-01 5.47141969e-01
-3.07960153e-01 3.40478301e-01 -1.22745179e-01 -8.29055250e-01
3.98154229e-01 -2.53829598e-01 -9.06647980e-01 -4.16846573e-02
9.64056849e-01 -1.24524504e-01 -2.90284902e-01 8.70718300e-01
5.16920388e-01 4.96077538e-02 7.20520973e-01 -1.36249495e+00
-1.15634787e+00 3.96350145e-01 -5.15042305e-01 -3.30262929e-01
4.26354706e-01 9.45882127e-02 1.49020422e+00 -1.18620729e+00
8.06209803e-01 1.27657819e+00 6.60679281e-01 4.71797645e-01
-1.36710823e+00 -4.66024607e-01 3.10709444e-03 1.35837823e-01
-1.59755898e+00 -4.50987518e-01 8.37290764e-01 -5.10828495e-01
6.92646265e-01 4.77793664e-01 5.85209846e-01 1.04102063e+00
-3.80786628e-01 8.12600255e-01 9.08753514e-01 -5.05181015e-01
-1.09823391e-01 1.66834116e-01 -5.37893511e-02 6.90187931e-01
-2.14657798e-01 1.17389299e-01 -5.99949121e-01 -8.99362043e-02
1.05202198e+00 4.51879472e-01 -2.05051512e-01 -6.65177703e-01
-1.53883433e+00 6.73376918e-01 7.98626661e-01 7.04294145e-01
-2.30721325e-01 4.50824469e-01 2.82769352e-01 3.96761417e-01
4.03566152e-01 4.86927569e-01 -3.24631661e-01 3.34300876e-01
-8.01412523e-01 1.25513703e-01 4.74986583e-01 9.75591779e-01
1.00590050e+00 -1.58175826e-01 -2.23292246e-01 1.09762132e+00
3.92134599e-02 4.98937905e-01 2.78689981e-01 -7.96798706e-01
2.07455486e-01 4.65573698e-01 -6.47159293e-02 -1.02028894e+00
9.15563926e-02 -3.39052230e-01 -9.06421602e-01 2.76408732e-01
2.46673003e-01 4.71274525e-01 -7.39484370e-01 1.95742214e+00
4.88382392e-02 9.47853997e-02 -1.44816339e-01 9.42158520e-01
1.05400038e+00 4.59943146e-01 7.09796250e-02 3.60491216e-01
1.68371618e+00 -8.71867418e-01 -2.80392945e-01 -1.25281583e-03
1.20499492e-01 -8.81282628e-01 1.38318717e+00 8.44358653e-02
-1.32341993e+00 -7.34374940e-01 -8.78356218e-01 -4.24370170e-01
-8.91607344e-01 1.69955790e-01 6.36533201e-01 1.30999073e-01
-1.56390500e+00 5.60916841e-01 -1.00635596e-01 -5.75754642e-01
4.22910661e-01 1.71671450e-01 -8.45608890e-01 -1.37003958e-01
-1.08829749e+00 7.24729419e-01 5.62795699e-02 -3.91813554e-02
-5.88214040e-01 -9.63803291e-01 -8.48487496e-01 2.85406291e-01
2.99534947e-01 -1.14329374e+00 8.09660017e-01 -1.39620602e+00
-1.05578184e+00 1.29418230e+00 -2.46108115e-01 1.64381228e-02
4.91349369e-01 -1.50985438e-02 -6.68943971e-02 1.52633771e-01
2.32573912e-01 9.90049124e-01 9.87163961e-01 -1.66115725e+00
-3.77040327e-01 -3.22616130e-01 1.48101062e-01 1.49731725e-01
-1.90537035e-01 -1.49450108e-01 -7.29080081e-01 -9.36105430e-01
1.11907884e-01 -8.09255719e-01 1.67570725e-01 5.76211333e-01
-3.10122728e-01 -4.37963784e-01 5.30823171e-01 -6.34387553e-01
7.62895823e-01 -2.24143314e+00 2.31491446e-01 2.70513922e-01
2.71201998e-01 2.05785111e-01 -6.18438005e-01 9.02126372e-01
-5.10849714e-01 2.31730044e-01 -2.21238598e-01 -5.45279503e-01
2.97114402e-01 -7.71039575e-02 -6.09399438e-01 2.38015771e-01
4.04719591e-01 1.34000516e+00 -9.61496592e-01 -3.24894071e-01
4.33679044e-01 5.99308729e-01 -4.26637620e-01 1.86326757e-01
-9.33330134e-02 1.88809991e-01 -2.44906425e-01 6.77322984e-01
5.72968841e-01 -2.31092125e-01 -3.68681103e-02 -6.10229850e-01
-2.20105820e-03 -7.18282238e-02 -1.05602658e+00 2.09388876e+00
-6.00035250e-01 5.35805523e-01 1.12745263e-01 -1.13772058e+00
9.22605813e-01 2.43536681e-01 5.42083979e-01 -7.84834564e-01
-1.69219717e-01 2.00184762e-01 -5.79315722e-01 -3.68263036e-01
4.51133996e-01 -3.47378761e-01 -8.46156776e-02 4.56704378e-01
2.90507823e-01 -1.18468218e-01 -1.36379614e-01 3.58578861e-01
8.50927293e-01 2.60950983e-01 9.04886872e-02 -1.33963093e-01
4.64906722e-01 -1.96819216e-01 3.51252705e-02 7.05785692e-01
1.14454582e-01 1.02139568e+00 3.94501805e-01 -5.41255474e-01
-1.15894198e+00 -1.52346718e+00 -3.13009396e-02 1.09171915e+00
3.24364573e-01 -2.55062997e-01 -3.48631799e-01 -3.79583448e-01
2.74446279e-01 3.68674397e-01 -8.34690094e-01 1.03368104e-01
-2.68784195e-01 1.04557067e-01 4.23934966e-01 4.49753106e-01
2.96986282e-01 -1.20779145e+00 -9.62206125e-02 -2.46499285e-01
-1.89736262e-01 -6.98148131e-01 -7.04869390e-01 -7.91777149e-02
-4.39980447e-01 -1.00341427e+00 -9.01723921e-01 -8.94007504e-01
6.01447463e-01 6.91596210e-01 1.29293656e+00 4.01910484e-01
-5.69546640e-01 6.73092067e-01 -3.24204683e-01 -2.84683760e-02
5.63828982e-02 -2.52169639e-01 -2.85543084e-01 2.46531427e-01
-2.77501214e-02 -8.23304296e-01 -9.24383998e-01 2.96623439e-01
-1.16127527e+00 1.77132547e-01 7.85668492e-01 9.39195931e-01
6.39627159e-01 -4.34245169e-01 5.09991288e-01 -6.32111371e-01
4.31697488e-01 -4.50943649e-01 -2.74693608e-01 6.65176332e-01
-3.23378384e-01 -5.13545843e-03 3.70472699e-01 -4.12585497e-01
-7.81505823e-01 -1.16187274e-01 -4.98190187e-02 -8.72999668e-01
-2.39219755e-01 2.31931999e-01 -1.94030404e-01 3.78329195e-02
5.14512956e-01 3.65317315e-01 4.26093750e-02 -4.82399285e-01
1.04923034e+00 5.14493287e-01 7.64319777e-01 -7.75121450e-01
1.07430708e+00 8.21399689e-01 -1.00400500e-01 -6.18951619e-01
-6.11636937e-01 -6.65836751e-01 -7.44098485e-01 -1.35991111e-01
9.32308793e-01 -9.31296051e-01 -9.27516580e-01 1.21305622e-01
-1.16297066e+00 -2.05673859e-01 -6.32701457e-01 2.63256505e-02
-6.86394036e-01 3.76518786e-01 -4.75596756e-01 -6.38842165e-01
-4.99356955e-01 -8.86234105e-01 1.53437448e+00 1.01864457e-01
-1.21007763e-01 -1.01165760e+00 4.19153720e-02 4.21826571e-01
4.91841257e-01 2.20053911e-01 8.66039097e-01 -4.44269478e-01
-8.70276868e-01 -1.29780784e-01 -8.30519199e-01 1.56405270e-01
1.75930753e-01 -2.76747104e-02 -1.17774642e+00 -3.13871175e-01
-5.88098407e-01 -3.21416080e-01 8.27817857e-01 2.08801582e-01
1.21189559e+00 -3.48344773e-01 -1.90200806e-01 6.73331022e-01
1.68227088e+00 -1.66954458e-01 7.68773973e-01 2.80512869e-02
7.59767830e-01 8.48202348e-01 3.53728175e-01 1.02997206e-01
3.85623574e-01 1.03429914e+00 3.64837438e-01 -4.24100041e-01
-5.29550731e-01 -5.90361595e-01 1.20262213e-01 6.43820226e-01
-6.29264722e-03 4.28913236e-02 -5.28761744e-01 8.53318870e-01
-2.03021097e+00 -1.28767729e+00 1.38787165e-01 2.22331691e+00
5.61226308e-01 -6.32558584e-01 8.54555517e-02 -2.17101872e-01
6.97646737e-01 2.69222170e-01 -3.54057491e-01 -3.53387654e-01
-1.45347953e-01 4.64229405e-01 2.96518393e-02 4.20137286e-01
-1.01689076e+00 9.99826550e-01 4.96635532e+00 1.17020762e+00
-1.07955360e+00 1.39349729e-01 4.99013901e-01 -7.57616339e-03
-6.53062224e-01 1.57399148e-01 -1.70187548e-01 2.24359378e-01
1.80447966e-01 1.64828435e-01 7.04806149e-01 7.12971568e-01
-1.64384604e-01 1.96270540e-01 -1.11147451e+00 1.26998067e+00
2.93851227e-01 -1.53099489e+00 3.85253906e-01 -1.15507215e-01
5.24477661e-01 -1.72181144e-01 1.36556432e-01 7.14799166e-02
1.81005269e-01 -1.03338492e+00 1.00424588e+00 8.27372193e-01
1.28355050e+00 -4.60655928e-01 3.30249578e-01 -8.93271267e-02
-1.49784136e+00 4.21004854e-02 -4.24952388e-01 2.98586428e-01
-1.23360790e-02 4.49296832e-01 -3.66344005e-01 9.22629356e-01
7.93978930e-01 8.14977765e-01 -4.63919878e-01 8.71685565e-01
-7.73541108e-02 -2.00761482e-01 -6.34403005e-02 3.22560906e-01
1.02469310e-01 -3.39281470e-01 4.86856431e-01 1.21698844e+00
4.33334261e-01 -4.28424068e-02 -1.35956690e-01 1.23230505e+00
-1.06967948e-01 9.13669865e-05 -8.98689449e-01 1.40845627e-01
4.55266058e-01 1.55819988e+00 -5.67110717e-01 -3.78995687e-01
-2.64715254e-01 1.30516207e+00 5.42545080e-01 3.86804283e-01
-5.67937076e-01 -3.92756432e-01 8.22093725e-01 1.16041712e-01
4.74099666e-01 4.94609699e-02 -3.13993722e-01 -1.16340280e+00
1.94498375e-01 -7.03974843e-01 4.45154309e-01 -1.36361051e+00
-1.78919911e+00 5.16550124e-01 -1.83205888e-01 -1.33164024e+00
1.02594063e-01 -4.89350975e-01 -6.43214285e-01 1.08860421e+00
-1.59221983e+00 -1.95009315e+00 -3.61608326e-01 8.48806679e-01
4.71091062e-01 8.86361003e-02 8.60074043e-01 4.60908145e-01
-1.17625535e-01 7.41076112e-01 -7.42624328e-02 1.86071530e-01
9.45072532e-01 -1.23874378e+00 3.75091732e-01 5.38075745e-01
5.43132544e-01 9.04652238e-01 3.16383332e-01 -3.01373005e-01
-1.50877631e+00 -9.89381850e-01 9.91799355e-01 -6.50283635e-01
7.90066719e-01 -4.92184043e-01 -7.84276724e-01 4.93221074e-01
2.46112645e-01 1.65755883e-01 3.48454148e-01 1.75566897e-01
-9.01893079e-01 -3.28967661e-01 -9.62799072e-01 6.98833585e-01
1.39328957e+00 -1.01050162e+00 -4.37624902e-01 4.08240825e-01
6.89165354e-01 -1.63968295e-01 -1.05604339e+00 1.56238228e-01
9.25314963e-01 -1.14303672e+00 1.59764779e+00 -6.19302452e-01
6.73532963e-01 -3.99006069e-01 -1.98326200e-01 -1.02143323e+00
-3.69536370e-01 -4.10160184e-01 5.58805883e-01 1.51908576e+00
2.69842416e-01 -6.19475782e-01 4.45077419e-01 3.78449023e-01
-1.80962756e-01 -7.98872471e-01 -8.78352523e-01 -6.87588871e-01
1.81098527e-03 -3.66428256e-01 6.82203174e-01 1.16656590e+00
-1.79643724e-02 2.45278850e-01 -4.28448439e-01 2.93187294e-02
4.88822132e-01 4.10405904e-01 8.64972651e-01 -9.62733150e-01
-4.05593127e-01 -7.73164511e-01 -3.33902031e-01 -9.28962708e-01
8.28681737e-02 -1.16113126e+00 2.42194626e-02 -1.55223405e+00
4.98578489e-01 -7.80743241e-01 -2.85199910e-01 7.08469331e-01
-1.58417657e-01 6.52049899e-01 4.60953861e-01 4.62618262e-01
-5.10684013e-01 4.82274711e-01 1.20315635e+00 -2.32910424e-01
1.93071052e-01 -3.82780880e-01 -7.58185625e-01 2.41216853e-01
4.12519723e-01 -1.53547928e-01 -3.93103927e-01 -3.83215576e-01
2.26466507e-01 1.77513927e-01 1.06311238e+00 -5.80565631e-01
1.57491699e-01 -3.39247361e-02 2.42577106e-01 -3.88582230e-01
5.03003240e-01 -8.42794836e-01 5.08849144e-01 1.36081830e-01
-4.32750702e-01 1.47627011e-01 3.90603170e-02 5.47108233e-01
-2.62440652e-01 1.73569806e-02 6.99493647e-01 -2.06587523e-01
-7.14326680e-01 4.55785364e-01 2.69686460e-01 -2.17216983e-01
8.14687431e-01 -3.17878067e-01 -5.87783217e-01 -4.62451041e-01
-6.89608693e-01 -6.65982515e-02 7.55563438e-01 6.05831206e-01
5.86047590e-01 -1.57030439e+00 -5.28379083e-01 2.58666337e-01
5.62623560e-01 -1.41696110e-01 5.66288054e-01 6.91147447e-01
-2.03121051e-01 4.78270322e-01 -1.46545812e-01 -6.53095007e-01
-1.16988826e+00 7.26628602e-01 2.46154472e-01 -1.81807801e-01
-6.13997817e-01 9.25990522e-01 5.51144361e-01 -6.19175553e-01
8.15237090e-02 3.65985483e-02 1.81513801e-01 1.66326061e-01
3.24802250e-01 -8.10728818e-02 4.61881980e-02 -6.18049383e-01
-3.35952789e-01 1.01637387e+00 3.29311609e-01 -1.65218294e-01
1.36214185e+00 -6.18915111e-02 -2.86370039e-01 3.08603793e-01
1.37636125e+00 -5.76622859e-02 -1.05525124e+00 -4.42728609e-01
-1.75246194e-01 -5.56298733e-01 -2.94588834e-01 -9.12506163e-01
-1.14622498e+00 9.35832262e-01 5.30814052e-01 1.30437568e-01
1.05013609e+00 3.77657026e-01 7.50353336e-01 2.86407918e-02
3.64513874e-01 -5.32187521e-01 3.20583403e-01 5.86577840e-02
1.31423199e+00 -1.21764159e+00 -1.78723797e-01 -1.64821982e-01
-5.58251023e-01 9.52661395e-01 1.56341225e-01 -2.15963230e-01
5.07531106e-01 -2.01683447e-01 4.61666584e-02 -4.84063834e-01
-5.89534879e-01 -4.66985971e-01 7.16620326e-01 6.11467421e-01
3.46822798e-01 -1.27117075e-02 -2.33806726e-02 3.06586295e-01
1.39720701e-02 -1.53830916e-01 -1.99883416e-01 5.06950676e-01
-1.15468800e-01 -1.32623398e+00 -2.44011775e-01 3.09345722e-01
1.49438977e-02 -2.27776676e-01 -5.78001976e-01 9.22164917e-01
1.49176687e-01 6.03674710e-01 1.49413094e-01 -2.95310318e-01
4.30298567e-01 4.63355444e-02 4.31761354e-01 -4.80762392e-01
-6.97834730e-01 -1.06703937e-01 -2.23590359e-01 -8.01484108e-01
-4.42347229e-01 -4.31362063e-01 -7.16300786e-01 -4.96587217e-01
-1.56392038e-01 -8.56805742e-02 7.28749394e-01 6.94013357e-01
6.32550776e-01 2.41107628e-01 6.09707713e-01 -1.17076135e+00
-1.84483543e-01 -6.70718372e-01 -5.17748117e-01 9.88282323e-01
2.78652817e-01 -5.97974181e-01 -3.69981289e-01 1.08280383e-01] | [11.628899574279785, 0.590360164642334] |
390cf461-14aa-4b8c-af3a-3370741b3108 | what-happens-before-and-after-multi-event | 2302.09715 | null | https://arxiv.org/abs/2302.09715v2 | https://arxiv.org/pdf/2302.09715v2.pdf | What happens before and after: Multi-Event Commonsense in Event Coreference Resolution | Event coreference models cluster event mentions pertaining to the same real-world event. Recent models rely on contextualized representations to recognize coreference among lexically or contextually similar mentions. However, models typically fail to leverage commonsense inferences, which is particularly limiting for resolving lexically-divergent mentions. We propose a model that extends event mentions with temporal commonsense inferences. Given a complex sentence with multiple events, e.g., "The man killed his wife and got arrested", with the target event "arrested", our model generates plausible events that happen before the target event - such as "the police arrived", and after it, such as "he was sentenced". We show that incorporating such inferences into an existing event coreference model improves its performance, and we analyze the coreferences in which such temporal knowledge is required. | ['Vered Shwartz', 'Raymond Ng', 'Chris Tanner', 'Sahithya Ravi'] | 2023-02-20 | null | null | null | null | ['coreference-resolution'] | ['natural-language-processing'] | [ 3.70581359e-01 3.19695234e-01 -3.46343696e-01 -4.97528076e-01
-9.30136204e-01 -7.37610936e-01 9.68828321e-01 9.75121975e-01
-4.98616606e-01 1.07486188e+00 9.86735463e-01 -5.66224635e-01
-3.82880755e-02 -9.14065659e-01 -5.28831899e-01 -2.39623308e-01
-6.06642962e-02 5.89881837e-01 4.68977362e-01 -5.33972323e-01
2.72022098e-01 4.72655028e-01 -1.26243365e+00 4.37030345e-01
4.20657188e-01 2.14159474e-01 3.75166029e-01 2.11665586e-01
-2.09420562e-01 1.16476536e+00 -7.54861355e-01 -7.15400934e-01
-3.05026174e-01 -6.67155087e-01 -1.36114037e+00 -5.77446818e-01
2.48387419e-02 -1.42423939e-02 -3.24815273e-01 1.06322086e+00
2.34643683e-01 7.21263051e-01 5.31098902e-01 -1.13236403e+00
-3.70266467e-01 1.35346007e+00 -2.68459439e-01 8.67297113e-01
9.19357240e-01 -1.77337587e-01 1.21131134e+00 -4.22057420e-01
1.11429560e+00 1.42649209e+00 6.53487682e-01 5.31456113e-01
-1.43454516e+00 -5.48596382e-01 3.26630652e-01 5.71496904e-01
-1.16408479e+00 -7.38967299e-01 7.72090197e-01 -8.74425173e-02
1.54807305e+00 4.71693516e-01 3.13456535e-01 1.38281548e+00
4.33814168e-01 3.97098035e-01 8.28861773e-01 -4.33475256e-01
1.65165618e-01 -4.10478294e-01 6.59646392e-01 -3.90743352e-02
3.73108357e-01 3.22449744e-01 -5.59321821e-01 -4.10060287e-01
1.98149607e-01 -2.09672064e-01 -2.48280037e-02 4.64215696e-01
-1.29796243e+00 8.69777560e-01 2.22189158e-01 5.78406572e-01
-5.20709336e-01 -1.90164568e-03 6.17945492e-01 1.49776280e-01
2.52136618e-01 4.93861288e-01 -4.15963233e-01 -2.60268778e-01
-7.83559144e-01 5.27332127e-01 8.96325707e-01 7.75263250e-01
5.60111165e-01 -4.50941086e-01 2.63300892e-02 4.73096132e-01
2.12936457e-02 1.45618379e-01 2.61860132e-01 -1.26041710e+00
4.56903070e-01 3.10324848e-01 5.61119497e-01 -1.05314195e+00
-7.27638125e-01 6.11597933e-02 -2.35891491e-01 -4.03069586e-01
3.05314362e-01 -1.78369537e-01 -4.28650856e-01 2.30870914e+00
2.57301390e-01 6.23469710e-01 5.03241122e-01 7.22094715e-01
1.03085780e+00 5.71259856e-01 8.22771370e-01 -1.00553882e+00
1.70565641e+00 -3.89309041e-02 -9.08512473e-01 -5.66960812e-01
5.62421024e-01 -7.13172674e-01 6.43004298e-01 -3.99731874e-01
-1.34993231e+00 1.84861589e-02 -7.34900773e-01 -3.00496399e-01
4.26019123e-03 -6.77885473e-01 6.71006620e-01 -9.66759585e-03
-3.76407266e-01 5.20349801e-01 -9.54704404e-01 -8.17272484e-01
-2.94508845e-01 1.53783606e-02 -4.42098469e-01 2.15790674e-01
-1.77531743e+00 1.40768659e+00 8.99435878e-01 -3.45911197e-02
-6.49620652e-01 -5.94021857e-01 -1.10495949e+00 1.39380068e-01
7.20395625e-01 -8.48341048e-01 1.53065431e+00 -6.80069029e-01
-8.38385522e-01 1.22759187e+00 -7.80724466e-01 -7.89558530e-01
-3.24426070e-02 -2.37492785e-01 -1.04084015e+00 2.33473122e-01
8.55850458e-01 3.63146245e-01 1.76320583e-01 -1.05952334e+00
-7.77812481e-01 -4.70278829e-01 6.14447951e-01 3.99962842e-01
3.13676268e-01 1.07411444e+00 1.36172861e-01 -3.96331012e-01
2.85816997e-01 -7.45659709e-01 -1.41238987e-01 -7.73699939e-01
-7.02272356e-01 -4.21144545e-01 4.78971720e-01 -7.15066642e-02
1.52892232e+00 -2.02590775e+00 -9.15049165e-02 -1.01027399e-01
-2.09449545e-01 -3.26407701e-01 3.39828551e-01 7.54614592e-01
-6.62815928e-01 1.57464117e-01 -8.65495354e-02 -1.08855806e-01
4.93321605e-02 6.56521857e-01 -7.44515359e-01 3.11490148e-01
2.79742390e-01 7.20247090e-01 -1.16597557e+00 -7.42812216e-01
6.51156381e-02 2.38067247e-02 -3.29326510e-01 -2.53018495e-02
-2.71002769e-01 7.30411112e-02 -4.05452192e-01 1.45223588e-01
2.04203248e-01 -1.35717943e-01 5.96319854e-01 -4.69783336e-01
-4.34541464e-01 1.08688962e+00 -1.24616873e+00 1.38723874e+00
-8.22413564e-02 3.14001173e-01 -5.96650578e-02 -7.34717607e-01
4.96639013e-01 6.87769532e-01 1.33930484e-03 -4.40126777e-01
2.09060282e-01 7.40287639e-03 -8.50792155e-02 -4.44725305e-01
6.90750837e-01 -1.01712072e+00 -7.10216939e-01 5.52773952e-01
-3.25573504e-01 3.36666815e-02 4.84921664e-01 6.32814229e-01
1.19618130e+00 8.10955837e-02 1.01643014e+00 -3.08099002e-01
3.51890266e-01 5.19939899e-01 1.12862861e+00 7.32959747e-01
-4.61118340e-01 2.42673993e-01 5.76542556e-01 -2.46345386e-01
-7.39249945e-01 -1.48025930e+00 -2.33547106e-01 1.17429805e+00
6.09078169e-01 -7.19216824e-01 -2.06302181e-01 -5.03338158e-01
-3.78666461e-01 1.62116420e+00 -6.67914748e-01 -2.02367559e-01
-1.03499055e+00 -6.44357979e-01 7.64299691e-01 7.87811875e-01
9.60564762e-02 -1.26818264e+00 -1.07841098e+00 7.22545326e-01
-8.65772605e-01 -1.09896445e+00 -3.28415453e-01 4.24340129e-01
-4.56992447e-01 -1.15547037e+00 3.62945884e-01 -5.02738178e-01
5.82034178e-02 -1.49585381e-01 1.29575348e+00 -1.32337034e-01
2.87310947e-02 1.01889953e-01 -3.65554184e-01 -4.27898079e-01
-4.40212965e-01 -4.12698179e-01 1.92703903e-01 -4.28541511e-01
8.16757143e-01 -6.55638099e-01 -1.06558263e-01 -1.45982802e-01
-7.98678935e-01 -2.11067662e-01 -2.71935195e-01 7.26214051e-01
3.45930904e-01 -7.84244090e-02 7.21519291e-01 -1.11179876e+00
7.69442856e-01 -8.72023225e-01 -1.35453828e-02 2.46207163e-01
-4.65542115e-02 3.56003121e-02 4.25834060e-01 -4.09387559e-01
-1.54997194e+00 -4.58522588e-01 -8.26948583e-02 7.89483339e-02
-4.61465627e-01 8.82431090e-01 -2.20642328e-01 1.16676700e+00
8.74553382e-01 -3.04388162e-02 -5.97666025e-01 1.12396814e-02
2.05656916e-01 2.48868335e-02 1.34621894e+00 -1.18431556e+00
4.04853165e-01 5.10765970e-01 2.14438024e-03 -3.92525852e-01
-1.21880972e+00 -5.14541090e-01 -5.13289988e-01 1.46088019e-01
1.01495552e+00 -9.48717892e-01 -8.89461815e-01 -1.81245685e-01
-1.73100412e+00 8.82988516e-03 -4.32291001e-01 7.64020503e-01
-7.35334277e-01 4.90801007e-01 -8.49031806e-01 -6.94102943e-01
-1.01663686e-01 -5.25648654e-01 7.54189968e-01 2.50245720e-01
-1.12938249e+00 -1.12712193e+00 9.96580273e-02 -1.44895449e-01
-4.55566719e-02 4.16568249e-01 1.17457187e+00 -1.09571123e+00
7.24316761e-02 2.86307275e-01 -4.19726968e-02 -6.31961823e-01
2.69058883e-01 1.17078856e-01 -4.68224823e-01 1.24615893e-01
5.79370111e-02 -9.48498100e-02 3.72914344e-01 2.84109145e-01
2.23842278e-01 -4.18703318e-01 -8.09478581e-01 6.36352077e-02
1.21712077e+00 3.64937097e-01 6.45136714e-01 4.83121783e-01
-4.11463231e-02 5.40626407e-01 7.57742286e-01 3.75391722e-01
5.97666860e-01 6.41713321e-01 9.03810188e-03 4.30565089e-01
2.27840971e-02 -3.37748617e-01 3.53967726e-01 2.59837389e-01
-3.46588977e-02 -3.06294262e-01 -8.63754511e-01 1.01339960e+00
-2.01719284e+00 -1.92480505e+00 -2.72023350e-01 1.78606308e+00
1.23808408e+00 2.92708308e-01 -2.59551316e-01 5.36978059e-03
1.10311890e+00 4.01760697e-01 -3.22400689e-01 -4.54574317e-01
-3.77362490e-01 2.83954740e-01 -1.45364597e-01 9.73995864e-01
-9.70745504e-01 1.31481969e+00 6.05467081e+00 1.97406054e-01
-6.40031278e-01 2.55953550e-01 -1.66168496e-01 -7.33400062e-02
-6.22695565e-01 6.74395323e-01 -9.39194322e-01 2.67716467e-01
1.17010343e+00 -8.23116004e-01 -1.44932657e-01 2.98618585e-01
3.94421399e-01 -2.84400821e-01 -1.43507767e+00 6.21311367e-01
1.26097232e-01 -1.45681798e+00 1.85102105e-01 -4.37135220e-01
3.16564173e-01 -3.73034984e-01 -5.64838290e-01 3.27874571e-01
8.98050785e-01 -6.69688523e-01 9.78366077e-01 3.22782069e-01
4.40104991e-01 -8.44017267e-01 5.43389559e-01 3.70809674e-01
-1.28981912e+00 6.47660811e-03 -1.41716525e-01 -3.01383674e-01
9.88584578e-01 4.44247574e-01 -6.36501729e-01 6.63046718e-01
5.08177638e-01 5.36313236e-01 7.06137866e-02 4.92088914e-01
-7.33127773e-01 6.14570856e-01 -5.31042397e-01 1.91727027e-01
3.35381143e-02 1.22345842e-01 1.11453950e+00 1.39895129e+00
9.90011394e-02 1.10726941e+00 3.02543491e-01 7.98207998e-01
1.39181495e-01 -1.44839391e-01 -8.00442040e-01 3.82523835e-01
1.05636930e+00 7.10085690e-01 -7.24258900e-01 -6.18856430e-01
-2.54193902e-01 7.90382266e-01 4.26355928e-01 2.65047580e-01
-9.49532628e-01 -2.59987593e-01 6.05289042e-01 8.70276019e-02
-6.34522066e-02 8.67391452e-02 1.06735706e-01 -1.10510719e+00
-1.64881334e-01 -5.15707731e-01 1.01326668e+00 -1.11676967e+00
-1.32884681e+00 3.74352038e-01 6.30318165e-01 -5.55102706e-01
-6.85695529e-01 -8.89553800e-02 -1.12769508e+00 7.24039495e-01
-1.18263602e+00 -1.01934063e+00 3.68257791e-01 8.53148341e-01
4.57426161e-01 5.50347030e-01 1.18086338e+00 2.01875065e-02
-3.68530750e-01 2.73422420e-01 -8.92488599e-01 2.01509178e-01
9.42144156e-01 -1.08238840e+00 1.25240624e-01 1.32923245e+00
-6.79148128e-03 1.21886480e+00 1.34007692e+00 -1.04265273e+00
-8.43637407e-01 -9.88909900e-01 1.89763153e+00 -5.20386815e-01
9.92727458e-01 2.66663432e-01 -1.06101060e+00 1.66663754e+00
3.54541153e-01 -2.04373524e-01 8.35941613e-01 6.21352792e-01
-6.10461533e-01 4.07928914e-01 -1.04803383e+00 9.83538091e-01
1.25801003e+00 -8.19728911e-01 -2.13777876e+00 2.33877867e-01
9.75983083e-01 -5.56076348e-01 -7.20318019e-01 4.20645446e-01
1.23981640e-01 -6.91960514e-01 8.91618490e-01 -1.39545834e+00
2.21774995e-01 -4.66089785e-01 -4.39843237e-01 -1.14996040e+00
-4.46202904e-01 -6.02579236e-01 1.93665605e-02 1.37489772e+00
5.43005526e-01 -5.15055120e-01 -6.92275725e-03 1.01947975e+00
-4.11045045e-01 2.62327760e-01 -1.27117515e+00 -6.07467592e-01
-3.47645804e-02 -7.18324780e-01 6.98644340e-01 1.28378260e+00
9.54400718e-01 8.84252071e-01 8.09828714e-02 2.93521851e-01
4.41918582e-01 5.76864421e-01 4.34426814e-02 -1.42538536e+00
-9.83177647e-02 -1.63061708e-01 -1.46041304e-01 -5.92239201e-01
7.74941921e-01 -9.25788283e-01 1.78093180e-01 -1.31584346e+00
5.95965028e-01 -2.97315329e-01 -6.69954792e-02 7.39558518e-01
-5.22224426e-01 -1.28415629e-01 1.64034054e-01 3.82093228e-02
-6.19960189e-01 1.61256850e-01 6.29546642e-01 -4.51453030e-02
-2.04451382e-01 -2.19378859e-01 -9.50375557e-01 1.12662077e+00
6.03548050e-01 -7.01322615e-01 -2.81281531e-01 -2.02440575e-01
4.91295427e-01 7.46983051e-01 5.60292721e-01 -3.67953748e-01
5.67126095e-01 -6.36348724e-01 -3.46904900e-03 -3.55585426e-01
3.47576112e-01 -4.98962432e-01 6.52874827e-01 3.46697778e-01
-6.18135452e-01 3.35074276e-01 4.66652930e-01 4.76465702e-01
-2.47949943e-01 -4.07885313e-01 4.18959618e-01 -1.73942789e-01
-9.80027795e-01 -2.18568772e-01 -4.40373152e-01 4.91866022e-01
1.17153502e+00 4.69632000e-02 -7.27262616e-01 -1.67323425e-01
-1.40318465e+00 -1.17788590e-01 3.45111251e-01 3.05346847e-01
5.53286910e-01 -1.06649065e+00 -9.05328214e-01 -6.23319268e-01
1.19322881e-01 -2.54075021e-01 9.23019275e-02 8.18956494e-01
1.19165704e-01 1.51346937e-01 -2.93546095e-02 -4.92408201e-02
-1.32204497e+00 9.68352854e-01 9.83827114e-02 -2.23795995e-01
-6.89245582e-01 5.52349746e-01 3.88460606e-02 -1.19064875e-01
-2.50351071e-01 -3.99355918e-01 -3.36822838e-01 2.13377342e-01
7.01931357e-01 5.44735491e-02 -2.50021279e-01 -1.01361871e+00
-9.72603440e-01 1.10806756e-01 -1.97052568e-01 -3.85904551e-01
1.07717705e+00 -4.73734230e-01 -4.29714978e-01 8.01820576e-01
6.76086009e-01 2.97813118e-01 -6.21376932e-01 -1.55924112e-01
1.65427268e-01 -1.22347400e-01 -4.30562913e-01 -7.93450415e-01
-1.66559294e-01 4.41406697e-01 -4.36725825e-01 1.46830261e-01
8.24644089e-01 5.68985403e-01 6.48148358e-01 2.99854994e-01
6.45268738e-01 -7.85669267e-01 -4.67413932e-01 8.61050129e-01
7.90528893e-01 -6.84929967e-01 6.13756198e-03 -5.82173526e-01
-5.54323018e-01 8.77138674e-01 4.51399058e-01 1.37077048e-01
2.76267916e-01 5.70682287e-01 -8.97526219e-02 -3.97116691e-01
-1.10773051e+00 -3.31100285e-01 -3.15360039e-01 3.64281625e-01
5.74252784e-01 4.62726891e-01 -8.79391491e-01 8.64728510e-01
-4.06342596e-01 -2.66603798e-01 6.86185420e-01 8.66226971e-01
-3.62503588e-01 -8.45292628e-01 -4.56221312e-01 1.62330493e-01
-8.42869639e-01 -5.40632129e-01 -3.45284849e-01 8.20145547e-01
2.01366186e-01 1.27014375e+00 3.90021920e-01 1.55456215e-01
4.21435893e-01 2.72064477e-01 4.41099644e-01 -8.93650234e-01
-8.43755424e-01 -2.60799855e-01 7.76146829e-01 -5.20574808e-01
-8.30026150e-01 -1.15583444e+00 -2.08043313e+00 -4.51832533e-01
-2.42973700e-01 4.86062437e-01 -2.26232961e-01 1.42020309e+00
3.53682339e-02 2.02005088e-01 1.41712308e-01 -2.52966821e-01
-9.49212089e-02 -6.74808443e-01 -3.93856108e-01 8.99889410e-01
2.38259912e-01 -7.86815524e-01 -4.09212530e-01 2.93400258e-01] | [9.893674850463867, 9.253043174743652] |
39cf4124-564c-4a87-bd4e-ded62f9df53c | one-wug-two-wug-s-transformer-inflection | null | null | https://aclanthology.org/2022.computel-1.5 | https://aclanthology.org/2022.computel-1.5.pdf | One Wug, Two Wug+s Transformer Inflection Models Hallucinate Affixes | Data augmentation strategies are increasingly important in NLP pipelines for low-resourced and endangered languages, and in neural morphological inflection, augmentation by so called data hallucination is a popular technique. This paper presents a detailed analysis of inflection models trained with and without data hallucination for the low-resourced Canadian Indigenous language Gitksan. Our analysis reveals evidence for a concatenative inductive bias in augmented models—in contrast to models trained without hallucination, they strongly prefer affixing inflection patterns over suppletive ones. We find that preference for affixation in general improves inflection performance in “wug test” like settings, where the model is asked to inflect lexemes missing from the training set. However, data hallucination dramatically reduces prediction accuracy for reduplicative forms due to a misanalysis of reduplication as affixation. While the overall impact of data hallucination for unseen lexemes remains positive, our findings call for greater qualitative analysis and more varied evaluation conditions in testing automatic inflection systems. Our results indicate that further innovations in data augmentation for computational morphology are desirable. | ['Miikka Silfverberg', 'Farhan Samir'] | null | null | null | null | computel-acl-2022-5 | ['morphological-inflection'] | ['natural-language-processing'] | [ 3.27534556e-01 3.25916529e-01 -1.56341910e-01 -3.82849663e-01
-5.33811033e-01 -9.46840584e-01 6.00755394e-01 6.32682979e-01
-8.48875761e-01 6.34444773e-01 9.69161153e-01 -6.04823947e-01
2.87395656e-01 -8.11382532e-01 -6.03982210e-01 -3.88960660e-01
1.71854839e-01 7.33772337e-01 -7.08121896e-01 -4.01086241e-01
1.66711628e-01 1.28629893e-01 -1.23407423e+00 1.88389182e-01
1.19471991e+00 1.13066444e-02 -1.98215786e-02 2.79994279e-01
-3.55803251e-01 5.57650387e-01 -7.90814817e-01 -8.66372585e-01
2.36621648e-01 -1.58251822e-01 -9.16941643e-01 -5.38717210e-02
9.64188099e-01 -2.23481432e-01 -6.83528231e-03 1.09902096e+00
6.96949422e-01 1.20130964e-02 4.90735799e-01 -7.22861111e-01
-1.47163594e+00 1.44282877e+00 -2.91274786e-01 4.34095830e-01
1.76961288e-01 7.08698988e-01 1.01149881e+00 -1.28507340e+00
6.41139686e-01 1.44000876e+00 9.00193572e-01 4.96966779e-01
-1.74952102e+00 -8.26095521e-01 1.69522632e-02 -1.94571689e-01
-1.12840843e+00 -7.37646401e-01 4.90661263e-01 -4.83185023e-01
1.41534257e+00 4.00472492e-01 8.38813603e-01 8.89659584e-01
-3.35729837e-01 7.40665197e-01 1.13756025e+00 -3.40151489e-01
-7.37056956e-02 6.10261746e-02 4.71970141e-01 3.80814254e-01
4.78569299e-01 2.97365397e-01 -3.82701635e-01 -2.41393715e-01
5.70405781e-01 -5.77953160e-01 -2.01726571e-01 4.93779302e-01
-1.36154544e+00 8.37867856e-01 4.55401212e-01 2.13623658e-01
-5.71800649e-01 -1.88466668e-01 5.04447401e-01 3.05288225e-01
8.17264736e-01 1.23133337e+00 -6.15857303e-01 -2.67637134e-01
-7.75191903e-01 2.60300279e-01 4.66945738e-01 7.30777204e-01
5.28541327e-01 6.42991364e-01 -1.65850431e-01 1.43472719e+00
-5.96718565e-02 4.13136631e-01 6.40018344e-01 -6.48020208e-01
5.57611823e-01 7.76715100e-01 -3.08038771e-01 -6.27386212e-01
-4.95955557e-01 -7.05917895e-01 -6.75587773e-01 1.06956951e-01
4.12955105e-01 -2.49657646e-01 -1.04716992e+00 2.16677999e+00
-1.19798016e-02 -5.58086574e-01 3.42992157e-01 7.69295990e-01
8.35977733e-01 5.79060137e-01 7.72538006e-01 -1.39549211e-01
1.40097904e+00 -3.70661885e-01 -8.04790139e-01 -7.44629443e-01
8.99073303e-01 -8.20189953e-01 1.80614984e+00 2.94258326e-01
-1.46761334e+00 -3.48786443e-01 -9.63789880e-01 -5.80685437e-01
-5.37665188e-01 -3.00230086e-01 9.61909413e-01 8.11755061e-01
-8.79060626e-01 3.23453814e-01 -6.33380949e-01 -4.48589802e-01
6.36372566e-01 3.65151912e-01 -5.14140248e-01 1.86387170e-02
-1.04376578e+00 1.47666252e+00 7.67753005e-01 3.43359224e-02
-2.68488288e-01 -1.19802964e+00 -1.07799113e+00 -1.77402064e-01
-2.11913466e-01 -3.29352379e-01 1.02455962e+00 -1.07116008e+00
-6.23734474e-01 1.28644025e+00 1.30672734e-02 -4.70288366e-01
5.65086603e-02 -3.70951891e-01 -6.39432490e-01 -5.42241275e-01
1.11686938e-01 1.03331077e+00 2.89161325e-01 -1.24290538e+00
-2.02023253e-01 -5.81669867e-01 -3.37669313e-01 4.33468908e-01
-5.86918175e-01 3.25513512e-01 3.41510206e-01 -1.15913403e+00
3.43332708e-01 -5.47889471e-01 -6.13624640e-02 -3.56490552e-01
-5.25076687e-01 -1.50507122e-01 3.35725218e-01 -8.80225658e-01
1.33840597e+00 -1.95435762e+00 -1.65896378e-02 2.26459168e-02
1.40510678e-01 4.52251196e-01 -1.80964202e-01 2.58622557e-01
-3.66558641e-01 6.88352227e-01 -7.13642955e-01 -3.73792320e-01
-3.80831584e-02 4.08413708e-01 -3.69869173e-01 3.22786987e-01
6.77329838e-01 1.16275144e+00 -1.07975340e+00 -2.37704694e-01
9.00707394e-02 2.71346122e-01 -8.47687304e-01 -7.02402666e-02
-1.76596999e-01 2.30496317e-01 6.33465528e-01 7.77032554e-01
9.95231569e-01 4.65597123e-01 -5.37852235e-02 -9.60726216e-02
-6.97037399e-01 7.59135783e-01 -6.57009900e-01 1.56843889e+00
-3.50291699e-01 5.39879858e-01 -7.51402229e-02 -8.92421156e-02
9.15814996e-01 6.46914542e-02 -2.82284766e-01 -5.14117062e-01
4.49917279e-02 4.46301460e-01 7.20090270e-01 -4.48798239e-01
1.04914951e+00 -7.92632818e-01 -6.93989219e-03 5.36924422e-01
1.22024193e-02 -5.16122520e-01 1.66404665e-01 1.17775276e-01
8.58703196e-01 -1.78248867e-01 4.07856762e-01 -6.63339496e-01
-1.03306197e-01 4.57818657e-01 5.70263922e-01 5.01709163e-01
6.80005625e-02 5.68156362e-01 7.93087482e-02 -3.29873979e-01
-1.16150165e+00 -1.25778389e+00 -6.42416775e-01 1.17642534e+00
-7.63232946e-01 -3.47645313e-01 -5.96148968e-01 -2.97448963e-01
-2.41299316e-01 1.64841437e+00 -6.74832106e-01 -5.23507893e-01
-7.69217014e-01 -1.37151325e+00 1.15656614e+00 5.91620028e-01
3.06008726e-01 -1.54040921e+00 -2.67608523e-01 2.79763401e-01
-3.14626217e-01 -4.47762877e-01 -5.13375580e-01 4.22215015e-01
-1.01546192e+00 -4.50231135e-01 -4.14795399e-01 -1.03126490e+00
5.95029056e-01 -2.85644144e-01 1.23786306e+00 3.54341537e-01
1.52193516e-01 -1.46399915e-01 -2.40158081e-01 -7.73367286e-01
-8.55713427e-01 3.06328744e-01 1.13738708e-01 -5.63091636e-01
7.47758031e-01 -5.18397152e-01 -1.97131187e-01 -3.17087799e-01
-1.17308557e+00 9.63990763e-02 5.23800731e-01 6.69322908e-01
4.21420217e-01 -5.17271578e-01 7.45424509e-01 -1.23553514e+00
9.11691666e-01 -7.63191640e-01 3.15029137e-02 4.01866296e-03
-4.98652905e-01 1.97492316e-01 4.46596384e-01 -8.27463448e-01
-1.06115222e+00 -2.88831025e-01 -5.57487309e-01 1.06697112e-01
-3.01429749e-01 7.22908676e-01 -2.63159871e-01 4.71481442e-01
1.20236623e+00 -8.11035857e-02 1.17630608e-01 -5.75776339e-01
4.83660847e-01 6.11771107e-01 1.12041318e+00 -5.95624030e-01
7.39813566e-01 1.50811717e-01 -6.27410471e-01 -1.20250380e+00
-6.42816544e-01 1.14187337e-01 -6.75631344e-01 2.09149048e-01
8.09427619e-01 -6.87998056e-01 -2.95562714e-01 4.83286172e-01
-1.06820130e+00 -4.99181598e-01 -8.39342892e-01 4.36617583e-01
-1.84101641e-01 5.48571646e-02 -6.35988593e-01 -6.65079474e-01
-6.78786099e-01 -9.37748909e-01 5.81545949e-01 -1.39666826e-01
-9.55424845e-01 -9.59480822e-01 3.26521367e-01 1.65849954e-01
5.79214275e-01 2.41423130e-01 1.82893121e+00 -1.01727259e+00
1.31761700e-01 -7.00419024e-02 -3.33256274e-02 2.44547218e-01
4.15255278e-01 -6.68328106e-02 -1.05010831e+00 1.55640068e-02
9.07117967e-03 -5.21957040e-01 7.49146938e-01 5.49627319e-02
8.04340661e-01 -7.96460688e-01 4.01817530e-01 6.77211583e-01
9.84788299e-01 -1.39482677e-01 6.54229581e-01 2.27099106e-01
6.64415300e-01 7.42826104e-01 2.31008098e-01 5.14515489e-02
4.99394149e-01 2.39151970e-01 2.53831029e-01 -2.84221947e-01
-5.36540210e-01 -6.29015028e-01 3.95704806e-01 7.89789557e-01
3.01690668e-01 -3.39598209e-01 -1.17001092e+00 9.41880882e-01
-1.19651484e+00 -8.43770683e-01 -6.77885175e-01 2.37427998e+00
1.30510020e+00 -6.51457831e-02 -2.51779128e-02 3.60422522e-01
6.02456331e-01 -2.43188534e-02 -6.01803005e-01 -7.83134639e-01
-8.54483962e-01 4.81016248e-01 2.74233788e-01 7.76337326e-01
-6.48468792e-01 1.53534424e+00 6.65295219e+00 4.87123400e-01
-8.38570595e-01 2.35279631e-02 5.40768206e-01 -1.22898281e-01
-8.61151218e-01 -6.49684519e-02 -3.70467097e-01 1.92586333e-01
8.58720303e-01 -8.98357406e-02 5.60328364e-01 1.74512178e-01
1.00384042e-01 7.40295127e-02 -1.07819819e+00 5.98764241e-01
1.82077065e-01 -1.15339065e+00 4.13366646e-01 -1.16552837e-01
3.43881696e-01 1.92689940e-01 3.78023714e-01 3.48769933e-01
5.96432090e-01 -1.34722245e+00 1.01098013e+00 9.34562832e-02
9.92488742e-01 -8.30336094e-01 4.78598207e-01 2.53492802e-01
-4.65012968e-01 1.23491660e-01 -5.41705787e-01 -4.51849073e-01
3.65350038e-01 2.48932764e-01 -1.12774527e+00 -3.11486989e-01
1.73841357e-01 3.17105979e-01 -1.06386781e+00 9.45750296e-01
-4.16528255e-01 1.10858524e+00 -4.63058084e-01 2.61231273e-01
4.56020422e-03 -1.04509696e-01 6.66088223e-01 1.49105549e+00
5.10758758e-02 3.00690919e-01 -2.53940493e-01 1.01057696e+00
-1.44211888e-01 3.69578809e-01 -6.94033921e-01 -3.27868372e-01
6.47592008e-01 9.96970594e-01 -4.65428621e-01 -3.44385773e-01
-2.37886250e-01 9.00515735e-01 7.67990887e-01 1.92394152e-01
-3.28865707e-01 7.79199600e-02 7.78857291e-01 2.66609073e-01
-5.86756647e-01 -2.49008149e-01 -1.05811715e+00 -9.13099468e-01
-3.81462723e-01 -1.00240970e+00 5.58909714e-01 -9.36980009e-01
-1.28399897e+00 5.58623493e-01 -7.39521347e-03 -6.60920501e-01
-9.90646426e-03 -3.81261736e-01 -5.41088462e-01 1.11588955e+00
-8.83687556e-01 -1.46566308e+00 1.77292407e-01 1.65412933e-01
6.71731055e-01 6.59539700e-02 1.15567625e+00 1.59928128e-01
-5.54074168e-01 8.05287778e-01 -5.55671096e-01 3.33830923e-01
6.61823511e-01 -1.34143221e+00 9.57012236e-01 1.05287111e+00
3.40597004e-01 1.08669734e+00 9.31712389e-01 -1.22182274e+00
-9.54180419e-01 -1.22033036e+00 1.18450642e+00 -7.26876676e-01
4.73687112e-01 -4.68765527e-01 -1.51150346e+00 1.06538177e+00
4.99836653e-01 -5.14089346e-01 1.12133813e+00 3.69700402e-01
-4.60212290e-01 7.07745135e-01 -1.35720193e+00 1.10408592e+00
1.26897681e+00 -3.72919321e-01 -1.10552692e+00 1.61399156e-01
9.36915636e-01 -1.32389992e-01 -7.73338914e-01 3.21558625e-01
2.58752316e-01 -2.98231751e-01 8.36593568e-01 -1.02414060e+00
5.06155252e-01 -1.59019232e-01 -2.49063969e-01 -1.86401522e+00
-4.58390832e-01 -4.19408232e-01 1.34057805e-01 1.57037032e+00
9.80403364e-01 -5.56737304e-01 5.07040977e-01 8.82683516e-01
-7.16532707e-01 -1.47253454e-01 -5.95711231e-01 -5.37068069e-01
8.10956538e-01 -8.25986385e-01 7.99431086e-01 1.35536933e+00
2.85276353e-01 5.70355117e-01 -3.46217826e-02 1.37201563e-01
4.63535994e-01 -5.21241248e-01 1.77371845e-01 -9.92710531e-01
2.23940406e-02 -4.51211572e-01 -5.35536222e-02 -3.64395797e-01
3.45136762e-01 -1.64940500e+00 -6.11124784e-02 -1.35109496e+00
3.61895934e-02 -6.09007001e-01 2.97073364e-01 9.59002018e-01
-3.25909048e-01 5.84702194e-01 2.12435797e-01 3.72904651e-02
3.29570740e-01 5.31255424e-01 9.27927017e-01 4.17857990e-02
-6.11624956e-01 -3.35104704e-01 -1.18634403e+00 8.89219046e-01
1.13318539e+00 -4.16089803e-01 -1.68543607e-01 -1.29385281e+00
5.11076152e-01 -4.80970263e-01 2.63522983e-01 -5.88404298e-01
-1.04014296e-02 -1.17178753e-01 4.64042962e-01 -5.23966670e-01
3.42692167e-01 -4.27449167e-01 -2.73291348e-03 4.74555850e-01
-5.67111492e-01 5.96335411e-01 7.55906940e-01 -2.61210620e-01
2.75520533e-01 -2.97455370e-01 7.72996187e-01 -2.47348249e-01
-3.68101567e-01 4.66105901e-02 -8.00914586e-01 5.07642567e-01
2.28879243e-01 -3.28941971e-01 -6.16441011e-01 -1.48315681e-02
-6.99120581e-01 7.99028501e-02 6.80923820e-01 5.50470173e-01
6.57364845e-01 -1.26696229e+00 -1.21913075e+00 6.41650200e-01
2.79507726e-01 -1.04508556e-01 -1.38258621e-01 4.73741680e-01
-2.37897828e-01 -1.79640099e-01 -2.35347852e-01 6.75986111e-02
-9.21732843e-01 5.43168664e-01 9.47372913e-02 3.31797510e-01
-3.57554287e-01 9.22135711e-01 1.36413723e-01 -1.06892681e+00
-1.31934792e-01 -4.98425961e-01 -1.59769103e-01 2.77377665e-01
4.98631924e-01 5.63403428e-01 2.12905824e-01 -9.49813962e-01
-1.78562701e-02 -1.39090240e-01 -3.14060330e-01 -5.93882561e-01
1.34881675e+00 5.59549257e-02 -2.71388382e-01 6.19053781e-01
7.75513113e-01 3.48974407e-01 -7.17310488e-01 -4.33602631e-02
1.48034573e-01 -2.78866708e-01 5.93142547e-02 -1.40240502e+00
-5.47792912e-01 8.61877084e-01 3.36691380e-01 -1.78787082e-01
8.36638629e-01 3.82943116e-02 4.24802154e-01 2.75397360e-01
-1.33921474e-01 -8.27877462e-01 -5.51961064e-01 8.74091506e-01
1.38708150e+00 -9.95767891e-01 -2.16849387e-01 -2.79091865e-01
-7.34473169e-01 6.08240306e-01 8.60580981e-01 1.09488845e-01
2.17546418e-01 3.99726301e-01 1.38066918e-01 -4.54321563e-01
-5.75365961e-01 -2.80349106e-01 -6.32788166e-02 9.21645582e-01
9.92740154e-01 2.63426006e-01 -5.52828312e-01 6.51368797e-01
-1.31725240e+00 -6.68841660e-01 7.05237091e-01 4.95736152e-01
-2.77851820e-01 -9.62973416e-01 -5.04840136e-01 8.87865305e-01
-4.16936517e-01 -9.72050786e-01 -7.98837662e-01 9.04501200e-01
3.23903084e-01 9.05272007e-01 6.26316488e-01 -1.75576091e-01
3.57464343e-01 5.93640685e-01 3.36104006e-01 -1.13508344e+00
-1.36275446e+00 -2.91662544e-01 4.27771986e-01 1.45522341e-01
1.89691350e-01 -9.77741003e-01 -1.54810107e+00 -3.55708659e-01
-2.36743495e-01 -1.17145173e-01 4.57235396e-01 7.24817455e-01
4.69552688e-02 8.97459760e-02 -3.23516577e-01 -3.94438356e-01
-3.12836289e-01 -1.47893131e+00 -2.53111660e-01 5.10457456e-01
3.36436361e-01 -1.54302269e-01 -1.77275687e-01 -1.07605778e-01] | [10.739154815673828, 9.771591186523438] |
d48440d0-1c8c-4ba2-a2e4-ef37c37674f9 | chatgpt-or-grammarly-evaluating-chatgpt-on | 2303.13648 | null | https://arxiv.org/abs/2303.13648v1 | https://arxiv.org/pdf/2303.13648v1.pdf | ChatGPT or Grammarly? Evaluating ChatGPT on Grammatical Error Correction Benchmark | ChatGPT is a cutting-edge artificial intelligence language model developed by OpenAI, which has attracted a lot of attention due to its surprisingly strong ability in answering follow-up questions. In this report, we aim to evaluate ChatGPT on the Grammatical Error Correction(GEC) task, and compare it with commercial GEC product (e.g., Grammarly) and state-of-the-art models (e.g., GECToR). By testing on the CoNLL2014 benchmark dataset, we find that ChatGPT performs not as well as those baselines in terms of the automatic evaluation metrics (e.g., $F_{0.5}$ score), particularly on long sentences. We inspect the outputs and find that ChatGPT goes beyond one-by-one corrections. Specifically, it prefers to change the surface expression of certain phrases or sentence structure while maintaining grammatical correctness. Human evaluation quantitatively confirms this and suggests that ChatGPT produces less under-correction or mis-correction issues but more over-corrections. These results demonstrate that ChatGPT is severely under-estimated by the automatic evaluation metrics and could be a promising tool for GEC. | ['Michael Lyu', 'Wenxiang Jiao', 'Yuxuan Wan', 'Wenxuan Wang', 'Haoran Wu'] | 2023-03-15 | null | null | null | null | ['grammatical-error-correction'] | ['natural-language-processing'] | [-1.21025786e-01 5.25614083e-01 3.44674855e-01 -5.15886009e-01
-1.06289077e+00 -4.94171739e-01 3.00395548e-01 4.18290913e-01
-4.77416396e-01 7.42330372e-01 2.41385028e-01 -5.23562014e-01
1.19083658e-01 -6.30547881e-01 -7.86758661e-01 -2.37043172e-01
2.78992623e-01 6.22602046e-01 1.57338575e-01 -6.37335300e-01
6.37732327e-01 -5.12623310e-01 -1.31388402e+00 4.71739054e-01
1.63652337e+00 3.27693999e-01 2.57375032e-01 7.24487960e-01
-2.85106152e-01 6.88625574e-01 -7.76491344e-01 -1.24650574e+00
-4.56901729e-01 -5.72421432e-01 -1.22445023e+00 -5.96488714e-01
7.24917173e-01 1.02624044e-01 2.54600793e-01 1.23888791e+00
3.89040589e-01 -8.97067934e-02 5.18476546e-01 -7.63963521e-01
-1.23202515e+00 8.44958901e-01 -1.24054462e-01 2.76163459e-01
4.86189067e-01 2.29320571e-01 1.41356993e+00 -9.41733122e-01
6.13612890e-01 1.31884837e+00 9.23618019e-01 7.29720235e-01
-8.60509336e-01 -2.04196230e-01 2.35651910e-01 2.94211835e-01
-1.19072127e+00 -1.07477203e-01 2.99885482e-01 -1.05623506e-01
1.56337023e+00 6.12943888e-01 -1.19251562e-02 9.45352376e-01
5.49219668e-01 7.26706743e-01 8.92682791e-01 -7.86793947e-01
-1.41577691e-01 -1.43659711e-02 3.62034231e-01 8.25387180e-01
1.02618538e-01 -9.99455601e-02 -3.84200126e-01 2.03451663e-01
-1.10420369e-01 -5.16395926e-01 -5.59371233e-01 7.25214303e-01
-9.97771382e-01 7.23655939e-01 2.58096516e-01 4.76828545e-01
-7.86098987e-02 2.62624443e-01 4.20557827e-01 6.03580475e-01
7.61105597e-01 8.74986053e-01 -8.40180814e-01 -6.33233190e-01
-4.28928733e-01 4.51769769e-01 8.48074257e-01 1.17299545e+00
5.46435237e-01 -2.10261598e-01 -6.59472287e-01 1.02745903e+00
3.17759901e-01 4.25010443e-01 4.91834283e-01 -9.10072803e-01
8.89969647e-01 7.01279163e-01 -4.49245982e-02 -9.80200231e-01
-3.93633127e-01 -5.54892361e-01 -6.44060135e-01 -5.00914872e-01
3.53373736e-01 -7.68899471e-02 -5.19016743e-01 1.85344994e+00
-1.65976912e-01 -4.49965209e-01 -4.68234196e-02 6.69401228e-01
1.15060627e+00 7.03853786e-01 3.70243430e-01 5.75624742e-02
1.21955752e+00 -1.28494000e+00 -9.26052272e-01 -3.52667540e-01
1.32463372e+00 -8.75291646e-01 1.70204866e+00 2.07259327e-01
-1.11907244e+00 -5.65137565e-01 -7.97090650e-01 -3.41383338e-01
-2.28226721e-01 2.16013551e-01 4.15235341e-01 7.19335496e-01
-1.16188931e+00 8.68172050e-01 -6.42203212e-01 -5.73150754e-01
-2.77876198e-01 -1.86641619e-01 -1.26196712e-01 -7.22815618e-02
-1.27495623e+00 1.22218573e+00 3.38519216e-01 -8.83577839e-02
-2.84406811e-01 -7.68464625e-01 -7.79126585e-01 2.12985680e-01
2.08082080e-01 -6.66902661e-01 1.70763624e+00 -4.09689754e-01
-1.45743871e+00 9.72512901e-01 -2.05676645e-01 -3.58237654e-01
3.51887614e-01 -3.91772121e-01 -4.37355250e-01 -3.28822345e-01
2.77581930e-01 6.42820477e-01 1.82420716e-01 -6.90244615e-01
-7.90435195e-01 -2.22193703e-01 1.36823922e-01 1.36281431e-01
-1.45367324e-01 3.78926992e-01 -4.13896978e-01 -6.83984578e-01
-5.53124491e-03 -1.03784549e+00 3.17824283e-03 -5.72165072e-01
-5.32380283e-01 -9.24727559e-01 3.55006307e-01 -9.23063874e-01
1.87169600e+00 -1.68332076e+00 7.36400764e-03 -2.89739966e-01
5.97482212e-02 3.40648204e-01 -2.80536085e-01 5.80089629e-01
3.51572409e-02 7.39366412e-01 -4.39556807e-01 -5.74454367e-01
1.12393379e-01 1.93079054e-01 1.09029248e-01 -2.23916415e-02
2.82525212e-01 1.16269290e+00 -9.75625575e-01 -3.30971271e-01
-2.63132751e-01 4.60096262e-02 -9.67158139e-01 8.07738379e-02
-4.61163491e-01 3.82386535e-01 -6.36766925e-02 5.80795407e-01
4.86266524e-01 -2.35341355e-01 -4.17467467e-02 4.04766142e-01
-1.83168471e-01 9.59153831e-01 -4.49656874e-01 1.57013595e+00
-5.14944196e-01 3.98199320e-01 -4.05011177e-01 -5.89988708e-01
9.17421818e-01 3.34889740e-01 -4.03841108e-01 -9.04981256e-01
1.43890306e-01 5.29115260e-01 2.39401430e-01 -5.82160830e-01
8.43936443e-01 1.53369695e-01 -3.49351943e-01 5.38956821e-01
2.29243249e-01 -2.69105554e-01 4.98037755e-01 2.89713174e-01
1.25782311e+00 -1.24754198e-01 2.80579954e-01 -5.07118523e-01
5.92393994e-01 4.78166826e-02 5.79100311e-01 1.01522088e+00
-2.28554621e-01 7.71274626e-01 3.51391256e-01 -3.35708335e-02
-7.86851585e-01 -6.26238406e-01 9.01267305e-02 1.23696530e+00
-1.99953765e-01 -8.14848840e-01 -1.29564416e+00 -9.92073894e-01
-2.91264355e-01 1.32844841e+00 -4.43727195e-01 -2.54996389e-01
-6.82287395e-01 -6.63592935e-01 7.76519299e-01 3.38009387e-01
6.00349903e-01 -1.29645789e+00 -9.48794261e-02 4.73093241e-01
-7.34808862e-01 -9.63593662e-01 -7.66970098e-01 -1.37626126e-01
-9.06620145e-01 -8.98645699e-01 -2.92375267e-01 -1.02056789e+00
5.37980318e-01 4.54137586e-02 1.73244572e+00 7.43199170e-01
3.93565983e-01 2.48354390e-01 -9.99657691e-01 -6.86647654e-01
-8.88676763e-01 4.48442876e-01 -2.81408131e-01 -6.33000731e-01
5.98312616e-01 -3.29236239e-01 -2.81711996e-01 2.29159445e-01
-4.52876538e-01 1.19291134e-01 4.60849762e-01 7.63984501e-01
3.78636181e-01 -3.09765399e-01 7.24727392e-01 -1.34034860e+00
1.05886197e+00 -3.02270114e-01 -2.33444825e-01 5.76342642e-01
-9.77245688e-01 1.04603045e-01 6.30922258e-01 8.16664696e-02
-1.06692338e+00 -8.44917357e-01 -7.42234766e-01 3.66972238e-01
-3.14740986e-02 9.12319243e-01 -2.14718014e-01 9.39260200e-02
7.11587608e-01 2.60838177e-02 -3.70144933e-01 -7.27470040e-01
2.09484920e-01 7.15804398e-01 6.66438997e-01 -6.53988242e-01
4.27979857e-01 -4.14924920e-01 -7.39472330e-01 -4.42986518e-01
-1.29959512e+00 -2.59341151e-01 -1.72776207e-01 -1.47072867e-01
6.46124303e-01 -6.12449050e-01 -5.67140162e-01 4.80836064e-01
-1.73218644e+00 -1.20068558e-01 7.92750716e-02 2.57305712e-01
-2.19417691e-01 2.92739213e-01 -9.17708158e-01 -4.30307597e-01
-8.20314825e-01 -9.22107935e-01 9.48963761e-01 1.90512687e-01
-5.74298620e-01 -1.11193061e+00 1.91799194e-01 6.49311662e-01
5.61380684e-01 -1.30195275e-01 1.19228268e+00 -6.31182492e-01
-2.79148281e-01 -3.56732123e-02 -1.00600064e-01 5.89851499e-01
-1.72573298e-01 4.47757952e-02 -6.36282384e-01 -1.65627494e-01
5.28709404e-03 -5.08974828e-02 8.14981461e-01 2.51670349e-02
1.24213338e+00 -5.64853668e-01 9.42794979e-02 2.71567345e-01
1.19993246e+00 1.18279040e-01 8.87403488e-01 3.68191093e-01
5.16731143e-01 5.04593670e-01 6.54784799e-01 -1.92151349e-02
8.72642040e-01 5.63492060e-01 4.18130815e-01 4.01024073e-01
-2.34836981e-01 -4.91309762e-01 4.36397731e-01 1.58274174e+00
-2.45358478e-02 -7.61697292e-01 -1.02507830e+00 5.95345676e-01
-1.80127251e+00 -5.72809994e-01 -8.32711279e-01 1.89844310e+00
1.05787957e+00 2.76809394e-01 -5.50384104e-01 -7.27139413e-03
7.09240377e-01 -2.26757929e-01 -6.45735636e-02 -1.10429668e+00
-2.66910464e-01 2.98290431e-01 -9.91843175e-03 6.10878348e-01
-7.25173235e-01 1.22181416e+00 6.15779686e+00 8.09732020e-01
-6.72811627e-01 3.87620628e-01 5.93716085e-01 4.71712977e-01
-5.07850170e-01 7.67174140e-02 -9.81899679e-01 6.97564006e-01
1.24967015e+00 -2.08717898e-01 2.24786848e-01 6.55794084e-01
1.07788041e-01 -1.13486975e-01 -1.06068134e+00 5.41499496e-01
3.10606301e-01 -1.23316550e+00 1.12940788e-01 -3.94521803e-01
8.11921895e-01 1.36529177e-01 -1.46183640e-01 1.04997003e+00
3.63616765e-01 -8.41819286e-01 7.78567374e-01 4.09329861e-01
4.73087549e-01 -5.96790552e-01 1.13894784e+00 6.23000622e-01
-6.37787342e-01 2.89124519e-01 -5.05821586e-01 -4.91638452e-01
1.30097076e-01 6.34047091e-01 -7.12329030e-01 5.57891607e-01
9.99081314e-01 5.14941096e-01 -1.10893846e+00 9.85217452e-01
-7.95336545e-01 1.18723142e+00 -5.92284985e-02 -5.20843148e-01
2.75903881e-01 -1.67699412e-01 5.52461326e-01 1.41424108e+00
5.62081039e-01 2.33970746e-01 -1.22487815e-02 9.19627309e-01
-4.43880260e-01 3.25127155e-01 -1.85899064e-01 -5.28693479e-03
5.41351199e-01 8.89520168e-01 -7.29149804e-02 -3.66269886e-01
-5.15925884e-01 1.05270624e+00 7.87433624e-01 6.64535835e-02
-7.80292988e-01 -5.92020392e-01 3.44465673e-01 -3.42308693e-02
5.05638197e-02 1.40558988e-01 -5.16828299e-01 -1.03859293e+00
3.68649930e-01 -9.73101079e-01 3.39297533e-01 -8.77699971e-01
-1.34769058e+00 8.20497811e-01 -4.31625009e-01 -9.72941041e-01
-1.75538018e-01 -3.40767115e-01 -8.69224727e-01 9.18393195e-01
-1.57972300e+00 -9.26508009e-01 -9.86570269e-02 1.47441044e-01
5.84227264e-01 1.15308523e-01 9.07510579e-01 4.42120939e-01
-7.33672082e-01 1.13532269e+00 2.74081044e-02 1.79085702e-01
7.01144874e-01 -1.46027684e+00 9.83584344e-01 1.01673138e+00
4.08355184e-02 7.80771017e-01 9.76364434e-01 -7.00424314e-01
-8.23623300e-01 -1.38529849e+00 2.00075865e+00 -8.05885553e-01
6.07658803e-01 -1.02006383e-01 -1.27250135e+00 6.50951147e-01
4.75424170e-01 -3.05336982e-01 3.76852065e-01 4.82982397e-01
-2.50432074e-01 2.59833127e-01 -1.07204592e+00 5.01101971e-01
1.12395465e+00 -4.43477929e-01 -8.58189106e-01 4.72669363e-01
1.18508863e+00 -6.60914123e-01 -7.15198517e-01 5.05034864e-01
-1.54541600e-02 -9.69723880e-01 2.72387654e-01 -7.58996665e-01
7.57429361e-01 -7.39720613e-02 -2.62628738e-02 -1.82866323e+00
-6.07761741e-01 -6.07195079e-01 1.69955358e-01 1.64130628e+00
9.94904399e-01 -6.73059762e-01 5.06412983e-01 5.91680110e-01
-9.44521606e-01 -7.21766829e-01 -9.74300027e-01 -9.76953089e-01
4.91621196e-01 -5.59707284e-01 4.61433589e-01 8.33188534e-01
1.65056169e-01 5.25692463e-01 -4.60704155e-02 5.58227375e-02
1.46232486e-01 -2.02261522e-01 5.18766761e-01 -1.09982967e+00
-2.58665681e-01 -3.64441752e-01 -4.12874334e-02 -9.56259549e-01
1.83468565e-01 -1.12768137e+00 3.50393116e-01 -1.46451366e+00
1.64527833e-01 -4.10301089e-01 4.22601542e-03 6.13137543e-01
-8.31547678e-01 6.34876341e-02 1.63038164e-01 2.35988833e-02
-8.17223430e-01 6.31682754e-01 1.20944953e+00 -1.57479882e-01
1.01860873e-02 -1.41744524e-01 -1.13923764e+00 6.23773158e-01
9.88985240e-01 -6.47489846e-01 1.24315351e-01 -1.03107142e+00
5.87708831e-01 -1.85216933e-01 4.90171723e-02 -8.27239990e-01
4.78402488e-02 1.17186278e-01 -4.65822965e-01 -4.05032605e-01
-4.87561077e-01 -7.35384747e-02 -3.15247267e-01 2.44471103e-01
-4.00234967e-01 5.57951450e-01 1.70465842e-01 2.01511875e-01
-3.77816319e-01 -6.56172156e-01 6.72564626e-01 -2.80471563e-01
-5.06770790e-01 3.32205072e-02 -4.24265295e-01 5.94906330e-01
3.02965730e-01 1.04015574e-01 -8.53760242e-01 -2.83850223e-01
-2.41506636e-01 4.77811277e-01 1.25347063e-01 6.76516116e-01
3.27895969e-01 -1.07549345e+00 -1.23653662e+00 -1.69956550e-01
6.66478574e-01 -1.55254215e-01 2.54045695e-01 7.86589742e-01
-5.93344271e-01 7.86486983e-01 4.28764611e-01 -5.44953704e-01
-1.28274238e+00 -1.76444873e-02 3.04329544e-01 -6.78001106e-01
-5.47445178e-01 1.23523951e+00 -8.67212042e-02 -1.13609862e+00
1.39264613e-01 -6.40135348e-01 -2.05052808e-01 -3.94935846e-01
6.18779242e-01 2.61341512e-01 8.44110429e-01 -2.25964546e-01
-2.38650933e-01 2.68402278e-01 -4.17378455e-01 2.29914054e-01
1.16483450e+00 -3.62805784e-01 -3.71486098e-01 6.97119460e-02
1.05214572e+00 -9.02018398e-02 -5.49560666e-01 -2.26286635e-01
4.11529869e-01 -2.74779350e-01 -8.26654360e-02 -1.40226650e+00
-8.02885532e-01 8.88978183e-01 1.97029948e-01 7.91194513e-02
7.76853383e-01 9.68200937e-02 8.64233851e-01 6.80118144e-01
4.92635101e-01 -1.25309920e+00 -3.82418931e-02 1.34213817e+00
1.25032187e+00 -1.31497562e+00 -5.14985681e-01 -4.64879334e-01
-7.36751199e-01 7.96401680e-01 9.97206628e-01 1.40963554e-01
3.47249329e-01 -1.85748175e-01 1.14927106e-01 -2.91154593e-01
-1.14571047e+00 -2.74238344e-02 2.72335321e-01 3.28826815e-01
1.02244520e+00 2.18868822e-01 -9.60960865e-01 8.29652727e-01
-8.24783564e-01 -3.46918166e-01 7.38031209e-01 6.13139272e-01
-5.41797161e-01 -1.10191751e+00 -4.37949263e-02 4.10310388e-01
-5.43229878e-01 -4.74963844e-01 -4.94153798e-01 7.01784432e-01
1.11737914e-01 1.36462772e+00 -7.08437338e-02 -3.55327904e-01
6.86905444e-01 1.29484385e-01 3.85647029e-01 -1.00678372e+00
-1.16117954e+00 -5.61670363e-01 4.39469248e-01 -4.40075189e-01
-6.22561835e-02 -6.13590479e-01 -1.20663798e+00 -5.59112012e-01
-5.01620591e-01 5.40483415e-01 5.80015719e-01 1.02137733e+00
4.38910335e-01 6.70860946e-01 2.35083789e-01 -1.85734760e-02
-5.50664008e-01 -1.60446262e+00 -1.85530752e-01 4.45318788e-01
-1.35935694e-01 -2.28237644e-01 -5.01021683e-01 -2.65099853e-01] | [11.086725234985352, 10.70658016204834] |
391ddaa0-c432-4605-8f17-2a65cba5c89d | a-pseudo-multi-exposure-fusion-method-using | 1808.00195 | null | http://arxiv.org/abs/1808.00195v1 | http://arxiv.org/pdf/1808.00195v1.pdf | A Pseudo Multi-Exposure Fusion Method Using Single Image | This paper proposes a novel pseudo multi-exposure image fusion method based
on a single image. Multi-exposure image fusion is used to produce images
without saturation regions, by using photos with different exposures. However,
it is difficult to take photos suited for the multi-exposure image fusion when
we take a photo of dynamic scenes or record a video. In addition, the
multi-exposure image fusion cannot be applied to existing images with a single
exposure or videos. The proposed method enables us to produce pseudo
multi-exposure images from a single image. To produce multi-exposure images,
the proposed method utilizes the relationship between the exposure values and
pixel values, which is obtained by assuming that a digital camera has a linear
response function. Moreover, it is shown that the use of a local contrast
enhancement method allows us to produce pseudo multi-exposure images with
higher quality. Most of conventional multi-exposure image fusion methods are
also applicable to the proposed multi-exposure images. Experimental results
show the effectiveness of the proposed method by comparing the proposed one
with conventional ones. | ['Sayaka Shiota', 'Hitoshi Kiya', 'Yuma Kinoshita'] | 2018-08-01 | null | null | null | null | ['multi-exposure-image-fusion'] | ['computer-vision'] | [ 9.97268677e-01 -6.01057827e-01 3.82577807e-01 -1.32548213e-01
-5.02696693e-01 -3.97582948e-01 5.18375814e-01 1.02919288e-01
-6.54560626e-01 6.01956785e-01 -3.04275334e-01 -4.73554060e-02
-3.72915953e-01 -8.56001198e-01 -6.00812793e-01 -8.09048295e-01
2.88302749e-01 -3.51670921e-01 3.15881848e-01 -1.69037208e-01
2.71842510e-01 4.42447871e-01 -1.92354858e+00 8.89070779e-02
9.29396272e-01 8.28232527e-01 7.80760288e-01 7.31266320e-01
2.08897799e-01 2.21197158e-01 -6.17512345e-01 -1.63461044e-01
5.19370377e-01 -9.01372254e-01 -3.12677145e-01 5.59919178e-01
6.89308763e-01 -6.06105626e-01 -9.71405581e-02 1.46684670e+00
2.28197947e-01 3.93994600e-01 5.84392309e-01 -1.07288313e+00
-6.46451950e-01 1.36090681e-01 -8.89341712e-01 4.29490149e-01
5.00512064e-01 1.61395539e-02 1.94475465e-02 -5.25849998e-01
3.67353410e-01 1.05077469e+00 3.93014878e-01 4.40497845e-02
-1.15899110e+00 -4.93018389e-01 -1.89091474e-01 2.48985112e-01
-1.44333148e+00 -3.18932056e-01 9.25177276e-01 -1.48884892e-01
2.62112707e-01 4.37965900e-01 6.88689113e-01 3.44432950e-01
6.48638904e-01 5.16488068e-02 1.92205930e+00 -7.61587977e-01
-1.43707678e-01 2.71321505e-01 1.12567350e-01 4.73128676e-01
4.59871262e-01 1.85370982e-01 -1.45751880e-02 1.06935211e-01
6.03855908e-01 5.78813493e-01 -7.99847543e-01 1.68639809e-01
-1.04164183e+00 3.19064260e-01 2.07074448e-01 8.59710932e-01
-5.45215011e-01 -1.97741225e-01 -7.18460381e-02 2.96931237e-01
1.33068115e-01 1.09100856e-01 2.97102302e-01 1.94603264e-01
-1.27321804e+00 1.28506019e-03 4.40948665e-01 6.53035402e-01
9.79150832e-01 1.95764173e-02 2.01320097e-01 4.39266026e-01
2.87564844e-01 8.98259401e-01 4.80286509e-01 -7.64631391e-01
2.07740486e-01 1.74856111e-01 1.94751099e-01 -1.13314199e+00
-1.65371284e-01 5.20403683e-02 -8.07658613e-01 7.89313197e-01
4.64502394e-01 -7.81654194e-02 -7.86019385e-01 1.52069497e+00
2.85792828e-01 2.15820462e-01 6.75935864e-01 9.09855783e-01
5.15337110e-01 1.09984684e+00 -1.70390010e-01 -1.12002420e+00
1.43899214e+00 -7.83835828e-01 -1.25601947e+00 1.20443832e-02
-2.12690219e-01 -1.24397850e+00 8.76269698e-01 8.09434712e-01
-1.39378250e+00 -9.53469515e-01 -1.49027300e+00 1.64614946e-01
-3.39844406e-01 -4.87704948e-02 -1.24337897e-01 8.46309721e-01
-8.80774021e-01 3.82396579e-01 -3.93112689e-01 -4.55007762e-01
-3.43044639e-01 2.72707701e-01 -5.05965769e-01 -3.77137989e-01
-1.17530274e+00 1.07826126e+00 5.59270501e-01 1.27892062e-01
-4.18950409e-01 -4.16438192e-01 -6.91642821e-01 7.59784132e-02
2.28219584e-01 -3.51011574e-01 7.85300612e-01 -1.30660403e+00
-1.49329591e+00 4.18460846e-01 -1.61223039e-01 -2.26945609e-01
3.99732172e-01 -9.11897570e-02 -7.41216362e-01 7.87965417e-01
-2.68351406e-01 9.45531577e-02 1.06076264e+00 -1.59590042e+00
-5.72575212e-01 -1.13138698e-01 2.41513461e-01 4.11119878e-01
-2.48927549e-01 -1.71430752e-01 -3.48600447e-01 -2.00758576e-01
3.53816748e-02 -6.49483085e-01 1.50704226e-02 -1.70677990e-01
-8.38746876e-02 4.87166077e-01 1.33593011e+00 -6.91940308e-01
1.05945480e+00 -2.30721378e+00 -2.23695815e-01 1.92494571e-01
-1.82350650e-01 3.80778104e-01 -5.39001171e-03 5.89476109e-01
-1.29064217e-01 -1.77289948e-01 -3.00814360e-01 -3.07114691e-01
-4.90844041e-01 -9.60220322e-02 -1.38069531e-02 7.01318085e-01
-3.68989140e-01 1.96719244e-01 -5.84099412e-01 -8.88569474e-01
9.09729421e-01 7.89920866e-01 2.38537848e-01 2.16792673e-01
3.17592293e-01 3.93039584e-01 4.66963276e-02 2.34897390e-01
1.14245284e+00 1.71028808e-01 2.99830705e-01 -6.02745056e-01
-4.64869320e-01 -8.16999614e-01 -1.37242579e+00 1.29880738e+00
-6.43379271e-01 5.26748538e-01 7.64437020e-03 -6.21527672e-01
1.04742980e+00 5.45729995e-01 5.26887000e-01 -7.38650978e-01
3.31987619e-01 1.87795445e-01 -1.87372237e-01 -5.77485740e-01
5.99604487e-01 -4.57157254e-01 3.00074428e-01 5.00777900e-01
-2.73838222e-01 -2.26812959e-01 7.33988881e-01 -2.18867753e-02
3.14505011e-01 -8.04620087e-02 5.66776097e-01 -1.74943656e-01
1.00757968e+00 -5.37078202e-01 2.89017826e-01 5.01330554e-01
9.80321243e-02 6.46780968e-01 -2.24104270e-01 -9.32094231e-02
-1.21727693e+00 -9.40456092e-01 -2.98206776e-01 9.48394611e-02
8.74767482e-01 -2.62130629e-02 -6.81874216e-01 -8.99579190e-03
-4.11589712e-01 3.35902005e-01 -2.66515762e-01 -5.86856119e-02
-4.57935452e-01 -7.36049235e-01 1.06389433e-01 -1.22945532e-01
1.07979643e+00 -6.95946276e-01 -1.01027465e+00 1.36303544e-01
-2.45914280e-01 -1.06135678e+00 -4.59537178e-01 -2.91987658e-01
-7.49304473e-01 -1.12297297e+00 -9.34920490e-01 -8.76246333e-01
7.76077867e-01 8.31754565e-01 3.73506069e-01 4.47327346e-01
-1.55202955e-01 3.82200480e-01 -4.03265744e-01 -2.45406665e-02
-6.68891788e-01 -5.05591273e-01 7.93741643e-02 4.82969046e-01
-5.07517979e-02 -6.27486587e-01 -7.94123888e-01 2.13747919e-01
-1.65723133e+00 1.85085207e-01 7.19894290e-01 5.44828057e-01
5.84493697e-01 6.33413970e-01 5.91385543e-01 -6.12568438e-01
5.02369404e-01 -1.11326583e-01 -8.42185915e-01 7.01728702e-01
-8.64531696e-01 -3.08850616e-01 1.00048387e+00 -5.34243762e-01
-1.69025731e+00 6.43077120e-02 1.46649346e-01 -3.45902383e-01
-2.74938792e-01 1.25059262e-01 -2.46735841e-01 -4.43730593e-01
3.41305614e-01 4.90428895e-01 2.17947856e-01 -3.21080446e-01
4.50246751e-01 8.00806522e-01 7.87201941e-01 -6.79375231e-03
8.14537644e-01 5.21414638e-01 3.24631423e-01 -1.03525603e+00
-1.13052770e-01 -1.32145911e-01 -4.70970213e-01 -5.50807178e-01
1.08028591e+00 -7.62814462e-01 -8.83540690e-01 6.70241416e-01
-8.51616561e-01 1.37604117e-01 1.26906022e-01 8.08491468e-01
-4.30261731e-01 1.09999239e+00 -5.41093886e-01 -9.44480836e-01
-3.38153869e-01 -1.33017182e+00 6.42811537e-01 8.96449327e-01
3.17637384e-01 -1.09314072e+00 -1.97856724e-01 1.63163945e-01
4.24061358e-01 3.76369804e-01 5.04627585e-01 1.05207898e-01
-6.18513286e-01 -4.20694910e-02 -9.31349769e-02 3.96415204e-01
5.61187923e-01 2.10932806e-01 -7.69805193e-01 -5.18213332e-01
4.07348365e-01 -3.71812433e-02 5.51309705e-01 4.99021053e-01
8.47857535e-01 3.78308035e-02 -2.77705252e-01 6.54450834e-01
2.26018000e+00 8.32669497e-01 9.96259272e-01 4.53618944e-01
1.58621475e-01 2.73644090e-01 9.71459806e-01 5.51086143e-02
-2.42982544e-02 5.04814506e-01 2.59422272e-01 -6.35561049e-01
-7.08778203e-02 7.46817887e-02 2.69331664e-01 5.37519932e-01
-2.35949494e-02 -6.95630729e-01 -3.09429884e-01 3.52591455e-01
-1.50291991e+00 -1.29175556e+00 -5.48289537e-01 2.38005805e+00
5.05032182e-01 -1.15351543e-01 -3.51328820e-01 5.49835920e-01
1.06167388e+00 1.67079702e-01 -1.10479392e-01 -4.49189454e-01
-3.41146767e-01 2.51852959e-01 5.94746411e-01 7.92680502e-01
-9.56137061e-01 1.59880906e-01 6.45636606e+00 7.28252113e-01
-1.42901540e+00 2.16706559e-01 2.22736284e-01 2.68417299e-01
-3.41980159e-01 1.34351268e-01 -5.35821080e-01 7.76621819e-01
6.57770932e-01 -5.90856552e-01 2.37285063e-01 9.30145532e-02
3.32916260e-01 -9.55565035e-01 -6.10689819e-01 1.22317648e+00
5.27700484e-01 -7.16866791e-01 7.54933953e-02 7.29284957e-02
1.01882720e+00 -1.05961263e+00 3.15544814e-01 -4.85380471e-01
-6.05813384e-01 -5.58270097e-01 4.08817470e-01 8.40427279e-01
7.00925291e-01 -7.69734681e-01 7.22526789e-01 2.10126817e-01
-1.33041143e+00 -4.28383239e-02 -1.96220264e-01 -2.69620214e-02
7.76000559e-01 5.56114495e-01 -3.92351836e-01 1.14246297e+00
3.49773437e-01 3.51850063e-01 -6.22245431e-01 1.16216016e+00
-9.35754180e-03 9.15083513e-02 -3.74850899e-01 2.69091457e-01
1.83387697e-02 -7.15936303e-01 3.66024524e-01 9.46892023e-01
7.96053231e-01 3.59005660e-01 7.77189732e-02 6.77735686e-01
3.95182878e-01 2.05801427e-01 -7.87826538e-01 1.61600053e-01
2.40157142e-01 1.27646220e+00 -9.01538253e-01 -6.93246484e-01
-5.43666005e-01 1.15900326e+00 -6.90155745e-01 5.70500970e-01
-7.94672787e-01 -6.60116673e-01 -1.34776592e-01 -1.68509930e-02
1.96806803e-01 -2.33400941e-01 1.16704270e-01 -9.37433004e-01
-1.43637344e-01 -7.28206873e-01 1.55351728e-01 -1.21043503e+00
-7.74208307e-01 8.58983934e-01 5.65806448e-01 -1.43292427e+00
-1.03275351e-01 -1.25146106e-01 -6.08707905e-01 1.11608136e+00
-1.46713996e+00 -1.09953308e+00 -5.82325757e-01 6.75904214e-01
4.68574077e-01 1.85475290e-01 5.16475201e-01 5.01846552e-01
-2.52349138e-01 2.25890204e-01 3.73221546e-01 -6.21327817e-01
7.80603051e-01 -9.74322796e-01 -3.51606250e-01 1.48883462e+00
-1.85894310e-01 4.45171058e-01 8.82147372e-01 -7.11147487e-01
-1.04294956e+00 -4.88637149e-01 4.44227904e-01 3.24670553e-01
-2.16141164e-01 1.92532554e-01 -1.00282288e+00 5.21334171e-01
1.01424444e+00 -3.16287607e-01 5.91721177e-01 -8.08592856e-01
3.31786513e-01 -4.70751882e-01 -1.43552876e+00 2.30766028e-01
3.23501438e-01 -3.72583479e-01 -6.03424788e-01 -3.85012664e-02
4.70689237e-01 -2.13634849e-01 -1.07158852e+00 3.82110238e-01
6.18960083e-01 -1.14846253e+00 8.55108500e-01 6.43025696e-01
2.63611466e-01 -8.95682096e-01 -9.76659507e-02 -1.18572009e+00
-6.57498837e-02 -4.43269491e-01 5.18259883e-01 1.38214815e+00
5.81590012e-02 -7.97319770e-01 -5.85557744e-02 4.88050520e-01
1.54987693e-01 -1.79155096e-01 -4.03447986e-01 -7.24620998e-01
-4.75537241e-01 3.78754348e-01 5.80375791e-01 7.91345775e-01
-9.45383906e-02 -2.11480117e-04 -7.25960851e-01 4.95801359e-01
9.64483500e-01 3.37202370e-01 7.20770419e-01 -8.51018369e-01
-3.26671720e-01 8.47925469e-02 -3.63392353e-01 -6.93468809e-01
-2.05046192e-01 -2.89298147e-01 -6.20398484e-02 -1.55966163e+00
2.75592268e-01 -7.68452212e-02 -1.87855661e-01 -2.37525612e-01
-3.10789257e-01 8.13159168e-01 5.22059321e-01 1.80410221e-01
-1.49625674e-01 1.46612555e-01 1.40567219e+00 2.63127387e-02
-1.51506677e-01 -2.95998633e-01 -4.11176950e-01 5.86624324e-01
6.61418974e-01 -1.62863344e-01 -7.73491561e-01 -1.70548663e-01
-1.42832816e-01 3.93859833e-01 2.69633651e-01 -1.37691903e+00
3.39315474e-01 -1.96448594e-01 5.06985247e-01 -7.32398748e-01
5.49754798e-01 -1.13917601e+00 9.50509012e-01 5.53684175e-01
1.73540682e-01 2.94109643e-01 1.52748942e-01 4.88392979e-01
-3.68814558e-01 -6.36149228e-01 1.06843019e+00 -2.02046186e-01
-9.20856953e-01 -2.05052406e-01 -5.00064969e-01 -8.45657051e-01
1.50939739e+00 -6.54731750e-01 -4.67219472e-01 -4.54194784e-01
-6.50866032e-01 -1.74598500e-01 9.47937667e-01 2.42361650e-02
7.85958290e-01 -1.26072240e+00 -2.68636286e-01 2.51319319e-01
-2.60782957e-01 -4.60144103e-01 8.93361628e-01 9.99527574e-01
-8.56969059e-01 -1.54199198e-01 -7.34194219e-01 -5.25768995e-01
-1.88935030e+00 9.84014034e-01 2.10030839e-01 9.06143486e-02
-6.12173915e-01 -1.02838643e-01 2.33101919e-01 6.16020262e-01
-3.18097949e-01 -2.16845661e-01 -3.98388177e-01 -2.43093133e-01
9.16140020e-01 4.07705218e-01 -2.13970095e-01 -7.83328056e-01
-4.48035374e-02 1.14153183e+00 1.37299836e-01 -5.02058387e-01
8.61092806e-01 -6.91712797e-01 -1.30628735e-01 5.11194348e-01
1.31961405e+00 1.70400709e-01 -1.01681912e+00 7.55321234e-02
-8.99815321e-01 -1.06876600e+00 2.04402264e-02 -5.44168830e-01
-8.11250687e-01 8.36346447e-01 1.15804470e+00 4.00750518e-01
1.87368572e+00 -6.46585107e-01 7.58554518e-01 -9.28827934e-03
3.10072750e-01 -9.10150230e-01 1.31782237e-03 -4.35735255e-01
6.59574032e-01 -1.07888734e+00 3.94002050e-01 -5.36334813e-01
-2.60029435e-01 1.45145643e+00 5.67062020e-01 -1.31930895e-02
4.95017290e-01 1.27355337e-01 1.07648849e-01 7.73705244e-02
-3.27177823e-01 -4.06946301e-01 6.29711896e-02 4.86981779e-01
1.78044662e-01 -2.54594594e-01 -1.08709276e+00 -1.50312200e-01
2.59512752e-01 2.14692220e-01 1.16402352e+00 9.23997283e-01
-6.67944908e-01 -1.17953110e+00 -1.04359925e+00 1.00521632e-01
-5.45836449e-01 2.81546563e-01 1.33865550e-01 9.81344581e-01
5.03290832e-01 1.28147113e+00 9.36098844e-02 -2.62541622e-01
3.00539047e-01 -9.47267413e-02 7.65423059e-01 -7.54422471e-02
-3.41624439e-01 2.40793064e-01 -4.58607376e-01 -2.75610704e-02
-1.19136405e+00 -5.07491350e-01 -1.02609253e+00 -3.88342291e-01
-5.39520144e-01 2.80917168e-01 8.24936330e-01 7.94713974e-01
-2.60512590e-01 5.24106860e-01 8.60656738e-01 -7.64593780e-01
7.44143501e-02 -8.86787176e-01 -8.36346745e-01 6.36692464e-01
4.60352033e-01 -4.95650440e-01 -4.90946293e-01 5.62825382e-01] | [10.967155456542969, -2.5254645347595215] |
a0f6b174-f64f-4cd6-89c4-75c770d66a98 | multi-task-learning-for-aggregated-data-using | 1906.09412 | null | https://arxiv.org/abs/1906.09412v4 | https://arxiv.org/pdf/1906.09412v4.pdf | Multi-task Learning for Aggregated Data using Gaussian Processes | Aggregated data is commonplace in areas such as epidemiology and demography. For example, census data for a population is usually given as averages defined over time periods or spatial resolutions (cities, regions or countries). In this paper, we present a novel multi-task learning model based on Gaussian processes for joint learning of variables that have been aggregated at different input scales. Our model represents each task as the linear combination of the realizations of latent processes that are integrated at a different scale per task. We are then able to compute the cross-covariance between the different tasks either analytically or numerically. We also allow each task to have a potentially different likelihood model and provide a variational lower bound that can be optimised in a stochastic fashion making our model suitable for larger datasets. We show examples of the model in a synthetic example, a fertility dataset, and an air pollution prediction application. | ['Mauricio A. Álvarez', 'Michael Thomas Smith', 'Fariba Yousefi'] | 2019-06-22 | multi-task-learning-for-aggregated-data-using-1 | http://papers.nips.cc/paper/9644-multi-task-learning-for-aggregated-data-using-gaussian-processes | http://papers.nips.cc/paper/9644-multi-task-learning-for-aggregated-data-using-gaussian-processes.pdf | neurips-2019-12 | ['air-pollution-prediction'] | ['miscellaneous'] | [ 1.13062285e-01 1.92328021e-01 2.04601139e-01 -3.99491489e-01
-1.22359824e+00 -5.08904696e-01 8.99560034e-01 3.23528558e-01
-6.26132727e-01 1.26131296e+00 2.89561361e-01 -1.34782806e-01
-3.95216346e-01 -9.28797305e-01 -8.81154001e-01 -8.71972680e-01
-1.51756823e-01 8.52602959e-01 -3.37988511e-02 4.85188842e-01
-3.26227516e-01 2.58537591e-01 -1.29296005e+00 -9.12233070e-02
1.14522827e+00 4.89741057e-01 3.91202360e-01 7.93180645e-01
-4.01763357e-02 -5.19096432e-03 -5.09632468e-01 -4.48889345e-01
4.17822637e-02 3.05732656e-02 -4.14587706e-01 1.75197441e-02
4.47878003e-01 3.83398563e-01 3.88620853e-01 1.08362794e+00
4.58042234e-01 1.66252375e-01 1.35760653e+00 -1.19304466e+00
-2.40282193e-01 3.43753278e-01 -5.42584479e-01 -1.10181659e-01
-1.50193647e-01 -5.56484014e-02 8.74932110e-01 -4.39506829e-01
3.33111942e-01 1.68551743e+00 6.06575966e-01 1.14353471e-01
-2.10953903e+00 -4.48683530e-01 3.47526133e-01 -4.41721529e-01
-1.42068553e+00 -2.04948246e-01 2.51700282e-01 -7.85465777e-01
6.60159051e-01 1.03430249e-01 3.58467311e-01 1.37900531e+00
7.42962480e-01 5.98547220e-01 1.20323181e+00 -2.84166411e-02
6.26889706e-01 1.05311722e-01 1.89557746e-02 1.84640780e-01
5.32355368e-01 -1.95874438e-01 -7.73579255e-02 -6.24843955e-01
8.45264673e-01 1.59170315e-01 2.23030582e-01 -5.03438056e-01
-1.09038174e+00 1.01305354e+00 2.75701694e-02 7.50066265e-02
-6.61885798e-01 4.71022964e-01 2.59091496e-01 1.10603623e-01
1.12155378e+00 3.92054915e-02 -6.25392318e-01 5.05285412e-02
-1.13233829e+00 6.05529845e-01 8.60185206e-01 6.02766573e-01
8.66922975e-01 -2.33296871e-01 -5.46776533e-01 7.59831011e-01
4.08662707e-01 9.76776183e-01 -5.11195287e-02 -9.73144352e-01
5.59629798e-01 8.85427147e-02 5.36281765e-01 -5.48306108e-01
-6.99224472e-01 -3.10408622e-01 -1.13082325e+00 2.84105480e-01
7.34445214e-01 -6.61819875e-01 -1.01600969e+00 2.15845656e+00
4.19942766e-01 4.82960612e-01 2.28314623e-02 3.49712402e-01
1.19318530e-01 8.21819127e-01 7.96754837e-01 -2.47388110e-01
1.51352549e+00 -3.78275037e-01 -7.87881494e-01 -2.29489386e-01
5.24047434e-01 -4.16565895e-01 9.15906906e-01 5.38644373e-01
-1.13491905e+00 -4.70778584e-01 -3.58805865e-01 5.85399047e-02
-5.49778819e-01 1.46970451e-01 3.83623332e-01 7.51604617e-01
-1.26207972e+00 5.71581066e-01 -1.18808031e+00 -5.17424405e-01
4.03893024e-01 2.08554834e-01 3.03328577e-02 2.83911943e-01
-1.14268160e+00 8.01328897e-01 2.48745874e-01 -1.97272912e-01
-8.10588062e-01 -8.39287221e-01 -7.59672225e-01 3.88410002e-01
2.24162638e-01 -1.11235809e+00 1.00442207e+00 -5.81409037e-01
-1.02430391e+00 5.20296991e-01 -3.03905845e-01 -5.70806503e-01
5.63281119e-01 -1.04913518e-01 -1.82391942e-01 -3.99918109e-01
4.15402412e-01 4.91615504e-01 9.46548104e-01 -9.97237980e-01
-6.69428051e-01 -4.35681999e-01 -2.54156381e-01 1.12516358e-01
-8.53162557e-02 5.30000813e-02 -2.66069531e-01 -6.68491304e-01
-4.27378207e-01 -1.01337647e+00 -7.11717308e-01 -1.90014929e-01
-1.46162376e-01 -4.17001039e-01 1.29244566e-01 -7.51747072e-01
1.09551406e+00 -2.08731198e+00 5.98286748e-01 3.20894159e-02
8.17832351e-03 -2.43870974e-01 -9.46945623e-02 3.03244203e-01
1.69037491e-01 3.25389951e-01 -5.18924952e-01 -9.27299857e-01
1.63615465e-01 4.56486791e-01 -4.27358523e-02 5.29625654e-01
3.19527090e-01 8.09129000e-01 -1.05005455e+00 -5.38753688e-01
-2.66404636e-02 6.80631280e-01 -3.58034253e-01 -2.78102756e-01
-2.73427516e-01 5.85309267e-01 -6.79525018e-01 9.04117897e-02
6.00410640e-01 -3.50104719e-01 6.45482242e-02 3.42774898e-01
-4.80098687e-02 -4.87135313e-02 -1.55309296e+00 1.58289444e+00
-8.70834172e-01 2.72823930e-01 2.88953364e-01 -8.54128659e-01
5.56556106e-01 4.99864131e-01 4.10396934e-01 -7.23534301e-02
-2.21470073e-01 1.31501615e-01 -2.85440236e-01 -2.15041026e-01
2.80953974e-01 -4.19704527e-01 -4.66293126e-01 4.23258752e-01
1.08117364e-01 -1.88557088e-01 2.16749802e-01 -1.75890148e-01
9.09492075e-01 1.12664014e-01 3.80307853e-01 -5.93983471e-01
2.85299301e-01 -3.18670005e-01 4.55591708e-01 1.21085179e+00
1.03681512e-01 5.20643771e-01 8.08518589e-01 -2.32966289e-01
-1.27577710e+00 -1.53730822e+00 -5.95321596e-01 1.05834568e+00
-5.55759370e-01 -2.44404133e-02 -5.88789582e-01 -3.61765206e-01
2.57480353e-01 7.56355226e-01 -6.07867420e-01 3.07831556e-01
-2.64556229e-01 -1.28468287e+00 2.95574307e-01 3.84329975e-01
-3.36108841e-02 -1.00389862e+00 -4.76757646e-01 2.41926149e-01
8.54757894e-03 -7.69579947e-01 -1.69257700e-01 2.76878715e-01
-9.73865867e-01 -6.00595593e-01 -1.13857055e+00 -2.81379491e-01
4.82438177e-01 -3.94837826e-01 1.18854749e+00 -8.13410997e-01
5.59793487e-02 4.79956150e-01 2.73815662e-01 -7.16550410e-01
-3.58940423e-01 4.85887021e-01 2.65684247e-01 3.40478987e-01
8.08176547e-02 -7.50115156e-01 -3.61090511e-01 -2.51913726e-01
-1.00765431e+00 -1.10226847e-01 4.40436572e-01 6.05017722e-01
5.62983572e-01 -2.22235434e-02 6.86080396e-01 -8.67152214e-01
9.27416801e-01 -9.11801755e-01 -8.91897142e-01 4.73018348e-01
-3.13987762e-01 4.12076175e-01 2.91806787e-01 -6.02675200e-01
-1.16021097e+00 5.09641059e-02 3.63914222e-01 -2.05873206e-01
-5.45849025e-01 6.69996321e-01 -1.64119765e-01 6.11260176e-01
4.25805181e-01 -2.00969493e-03 -1.69045031e-01 -6.91493094e-01
5.45199454e-01 5.28030634e-01 2.23369852e-01 -8.28062236e-01
6.24777079e-01 5.01431644e-01 3.72267365e-01 -8.16340864e-01
-6.72716737e-01 -3.42642099e-01 -6.59251928e-01 5.14729396e-02
1.18674552e+00 -1.25915289e+00 -6.49289131e-01 3.21391612e-01
-1.31748474e+00 -5.51806211e-01 -3.51194024e-01 6.44350111e-01
-6.83468401e-01 -7.38182962e-02 -3.24954718e-01 -1.27211845e+00
-5.55038564e-02 -1.17809308e+00 1.53993571e+00 7.63467774e-02
-2.65660197e-01 -1.69381511e+00 4.33470458e-01 -1.97689548e-01
3.30677688e-01 4.43593293e-01 9.12014544e-01 -5.80699146e-01
-4.33298022e-01 9.62918252e-02 -2.54008561e-01 2.00375855e-01
2.49182835e-01 -9.23177898e-02 -8.92062008e-01 -3.59373659e-01
-1.52671576e-01 -9.00158584e-02 1.16939819e+00 1.05819404e+00
1.06462657e+00 -2.83976823e-01 -6.39075637e-01 3.80825549e-01
1.47327220e+00 -1.68653443e-01 2.53017426e-01 -2.18316093e-02
6.04079306e-01 8.67860198e-01 2.98998564e-01 2.85763234e-01
3.78713012e-01 5.67923963e-01 2.97259334e-02 -6.02867864e-02
4.63746846e-01 7.65111716e-03 2.18604758e-01 2.04320759e-01
-1.18731402e-01 -2.86453426e-01 -1.16037130e+00 8.61583531e-01
-2.32494354e+00 -1.04326534e+00 -3.29081088e-01 2.48177052e+00
6.60961568e-01 -4.63662781e-02 2.78102368e-01 -6.22808039e-01
6.83982909e-01 2.30883405e-01 -4.31593925e-01 -3.44490916e-01
-4.14266847e-02 3.08489680e-01 8.36389363e-01 8.70834768e-01
-1.37828565e+00 5.66360056e-01 7.12797546e+00 8.95753264e-01
-4.99270707e-01 4.60775375e-01 7.30388284e-01 -3.74299049e-01
-2.84488738e-01 -7.85000846e-02 -9.03794527e-01 6.15967035e-01
1.28349864e+00 -3.26946706e-01 8.51615593e-02 4.70270455e-01
4.81898874e-01 -2.69564629e-01 -1.04825139e+00 6.26704514e-01
-4.01211619e-01 -8.59345675e-01 -8.18707868e-02 5.45773745e-01
9.19656396e-01 1.87950268e-01 2.57818788e-01 3.51942509e-01
8.62339914e-01 -9.69570696e-01 4.47702318e-01 1.03484106e+00
4.47607636e-01 -7.43200302e-01 5.01606762e-01 6.40574157e-01
-9.39109027e-01 -4.32662591e-02 -2.28034958e-01 -3.80242616e-02
6.17615163e-01 7.00338781e-01 -5.80013633e-01 5.47248542e-01
5.40891111e-01 4.10662681e-01 -4.06183749e-01 1.07613552e+00
4.83538508e-02 6.31582439e-01 -6.39587700e-01 1.04365628e-02
2.39775032e-01 -6.78039730e-01 6.11055374e-01 1.23569560e+00
7.33846307e-01 -3.99464697e-01 1.70238778e-01 9.55752611e-01
2.23525420e-01 1.89123750e-02 -6.29412830e-01 2.65630513e-01
1.62137061e-01 1.21474016e+00 -7.29536414e-01 -3.87899905e-01
-6.83041930e-01 7.72273242e-01 4.31707293e-01 9.54725087e-01
-8.28036487e-01 1.06293753e-01 1.07709038e+00 -2.94918399e-02
4.64721709e-01 -1.96466878e-01 -2.15665951e-01 -1.08248401e+00
-1.30504757e-01 -2.53516376e-01 2.78165609e-01 -5.37121892e-01
-1.59286010e+00 1.56638637e-01 6.20698750e-01 -7.48593152e-01
-3.54456782e-01 -5.84465027e-01 -5.03063262e-01 1.51088953e+00
-1.26364100e+00 -9.91353631e-01 3.92080903e-01 2.01645151e-01
3.34388554e-01 3.35826427e-02 6.87819660e-01 1.40461966e-01
-5.45246184e-01 -1.16190933e-01 5.75096369e-01 -3.12960923e-01
6.42163813e-01 -1.70710766e+00 6.99370801e-01 5.49386740e-01
1.70205123e-04 4.24342155e-01 9.10546482e-01 -8.92335176e-01
-4.33628112e-01 -1.30091405e+00 1.08921361e+00 -7.51008809e-01
8.37121964e-01 -6.54882491e-01 -9.56379652e-01 8.10213268e-01
4.78549227e-02 -3.19094986e-01 6.49919331e-01 5.64258814e-01
8.56224373e-02 -2.23279912e-02 -1.07752371e+00 6.00615561e-01
6.90806627e-01 -4.41488653e-01 -2.64077187e-01 4.53805238e-01
6.72523916e-01 -1.68879833e-02 -8.89771700e-01 2.56673515e-01
5.07744074e-01 -6.14039600e-01 1.01900971e+00 -8.02293718e-01
1.54108226e-01 -9.85200703e-02 -4.63177711e-02 -1.68418801e+00
-3.59237760e-01 -4.55172569e-01 -8.70611966e-02 1.23543203e+00
6.37860358e-01 -8.06056619e-01 4.00372148e-01 8.14308703e-01
4.35592473e-01 -3.90068054e-01 -1.32705152e+00 -7.85639346e-01
5.16952932e-01 -5.48516214e-01 6.55516624e-01 5.79261601e-01
-5.82766116e-01 3.22749525e-01 -4.54508007e-01 5.65859973e-01
9.73978102e-01 -1.80128187e-01 5.17670870e-01 -1.81462431e+00
-2.68726379e-01 -4.31231380e-01 -1.66884791e-02 -7.97265589e-01
3.20027739e-01 -7.34022737e-01 -7.13655353e-02 -1.58295679e+00
4.08556938e-01 -4.92098212e-01 -1.61791041e-01 2.24387154e-01
-4.88784343e-01 -2.84404069e-01 2.63579600e-02 -5.81295602e-02
-3.99619669e-01 4.41192180e-01 9.05464768e-01 -1.08194850e-01
-4.29429442e-01 2.72829890e-01 -4.67608482e-01 6.97956860e-01
8.13377559e-01 -7.94551313e-01 -4.04615760e-01 -3.66979957e-01
3.99752349e-01 2.49866635e-01 5.57572722e-01 -8.00984144e-01
1.54434606e-01 -3.84656936e-01 3.66160393e-01 -4.55218881e-01
4.83991534e-01 -7.05047369e-01 6.03594482e-01 1.91104501e-01
-3.56632352e-01 -5.13078943e-02 1.10291630e-01 9.09551024e-01
1.63509265e-01 -9.72882807e-02 4.40455884e-01 -1.71700805e-01
1.27212211e-01 3.52335721e-01 -6.52786136e-01 -1.17789119e-01
8.56401265e-01 2.25067616e-01 9.20408219e-02 -4.17601138e-01
-1.13513410e+00 6.95653379e-01 2.53189564e-01 1.86536029e-01
8.32589790e-02 -1.24274015e+00 -1.07257903e+00 -9.30117618e-04
-1.43720016e-01 1.53939471e-01 4.15608138e-01 8.19458187e-01
9.48291048e-02 4.89387006e-01 6.52280301e-02 -6.86880171e-01
-1.07853520e+00 6.70073509e-01 3.59323621e-01 -7.92910993e-01
-4.28049155e-02 2.14179412e-01 5.21655262e-01 -5.34402072e-01
1.78996772e-02 -5.88549018e-01 -2.08868831e-01 5.03654838e-01
3.01451564e-01 5.24949968e-01 -3.94144803e-01 -4.75762278e-01
-6.12764992e-02 5.07761955e-01 2.96106249e-01 -6.29992008e-01
1.33338046e+00 -4.15668666e-01 -3.68628348e-03 1.15568554e+00
9.03510511e-01 -2.85234749e-01 -1.55357325e+00 -2.35210761e-01
9.21405777e-02 -2.51758993e-02 -8.29849690e-02 -5.08426487e-01
-5.70056558e-01 8.68287027e-01 5.53267777e-01 5.43914020e-01
8.34199011e-01 2.49424890e-01 8.83885995e-02 5.60377091e-02
3.93180579e-01 -9.59764838e-01 -4.48380440e-01 2.61412919e-01
6.98756218e-01 -1.23339617e+00 -5.62406331e-02 -6.41511753e-02
-4.33802336e-01 6.11250520e-01 -1.20079778e-01 -1.20158389e-01
9.41849470e-01 3.18605483e-01 -2.14459077e-01 -5.78525029e-02
-8.32824349e-01 -5.44520319e-01 4.84493643e-01 6.36683166e-01
4.39716637e-01 3.58145356e-01 -4.33336705e-01 5.45536995e-01
3.82334851e-02 -1.95537526e-02 2.65112042e-01 4.82413024e-01
-3.20004433e-01 -1.14849973e+00 -5.45047700e-01 6.21083915e-01
-6.67662621e-01 -8.19410831e-02 1.90769687e-01 6.33044362e-01
1.62497342e-01 6.24335051e-01 4.84049439e-01 5.46713293e-01
1.81646407e-01 5.22823334e-01 3.44639122e-01 -7.93486834e-01
-2.39570841e-01 3.28679055e-01 -2.55789645e-02 -2.04225674e-01
-5.80466330e-01 -1.16898322e+00 -7.09918439e-01 2.60511283e-02
1.83593452e-01 1.23341866e-01 5.42447984e-01 9.20548856e-01
2.09296137e-01 5.73501825e-01 1.21558622e-01 -9.63300169e-01
-4.31681275e-01 -1.11202466e+00 -9.70121384e-01 2.91834980e-01
5.96794963e-01 -6.76603734e-01 -4.52333093e-01 8.95860717e-02] | [6.94701623916626, 3.918076753616333] |
e21fe60f-67cb-48ba-b7a0-ae597a9dc2ae | invisible-steganography-via-generative | 1807.08571 | null | http://arxiv.org/abs/1807.08571v3 | http://arxiv.org/pdf/1807.08571v3.pdf | Invisible Steganography via Generative Adversarial Networks | Nowadays, there are plenty of works introducing convolutional neural networks
(CNNs) to the steganalysis and exceeding conventional steganalysis algorithms.
These works have shown the improving potential of deep learning in information
hiding domain. There are also several works based on deep learning to do image
steganography, but these works still have problems in capacity, invisibility
and security. In this paper, we propose a novel CNN architecture named as
\isgan to conceal a secret gray image into a color cover image on the sender
side and exactly extract the secret image out on the receiver side. There are
three contributions in our work: (i) we improve the invisibility by hiding the
secret image only in the Y channel of the cover image; (ii) We introduce the
generative adversarial networks to strengthen the security by minimizing the
divergence between the empirical probability distributions of stego images and
natural images. (iii) In order to associate with the human visual system
better, we construct a mixed loss function which is more appropriate for
steganography to generate more realistic stego images and reveal out more
better secret images. Experiment results show that ISGAN can achieve
start-of-art performances on LFW, Pascal VOC2012 and ImageNet datasets. | ['Shiqi Dong', 'Ru Zhang', 'Jianyi Liu'] | 2018-07-23 | null | null | null | null | ['steganalysis', 'image-steganography'] | ['computer-vision', 'computer-vision'] | [ 6.50908411e-01 2.23826781e-01 3.64544898e-01 1.43972024e-01
-1.66205823e-01 1.02571985e-02 2.93440700e-01 -9.40925539e-01
-3.43736112e-01 6.38456881e-01 -1.35300770e-01 -4.39502448e-01
2.43075565e-01 -1.10332406e+00 -9.66987371e-01 -1.17455924e+00
-1.85806721e-01 -3.26741844e-01 2.34354004e-01 -4.91609961e-01
1.74411252e-01 8.30988660e-02 -1.36151481e+00 4.70623821e-01
6.93793356e-01 9.95037913e-01 9.02819932e-02 7.37728357e-01
1.87824577e-01 1.05248690e+00 -6.55751765e-01 -2.97980726e-01
4.48013783e-01 -1.05554593e+00 -6.92811728e-01 3.16761702e-01
-1.90596148e-01 -3.79511088e-01 -5.82528889e-01 1.47991335e+00
5.64602971e-01 -4.00763899e-01 3.90505344e-01 -1.30342805e+00
-1.14377427e+00 6.98190868e-01 -5.63776672e-01 4.39215414e-02
-1.21033713e-01 6.61839604e-01 3.83174270e-01 -2.91682839e-01
3.63025278e-01 1.42578173e+00 5.21543622e-01 7.86936462e-01
-6.93841696e-01 -1.16538203e+00 -2.63390332e-01 5.23382604e-01
-1.17708004e+00 -1.62280127e-01 9.46637511e-01 -1.92576703e-02
5.50732493e-01 2.69161046e-01 7.50008702e-01 1.02586806e+00
5.65719366e-01 6.84683263e-01 1.50349295e+00 -4.73838568e-01
-2.17740431e-01 3.20660412e-01 -5.42056382e-01 7.35095084e-01
2.24269509e-01 6.75171137e-01 1.11808643e-01 3.57347190e-01
7.00465083e-01 1.32097274e-01 -5.76844811e-01 -1.60468757e-01
-1.09259927e+00 1.09354973e+00 7.46874928e-01 5.86326540e-01
-9.72272828e-04 3.95755559e-01 1.96486890e-01 7.20725656e-01
8.49315971e-02 1.73988879e-01 1.12599440e-01 5.94387531e-01
-5.26409388e-01 -5.58007844e-02 9.27917480e-01 7.37313211e-01
6.72612250e-01 4.85745609e-01 1.84799641e-01 1.65787891e-01
5.77993453e-01 8.17196965e-01 7.54341066e-01 -4.54914540e-01
4.78360683e-01 3.01440686e-01 -3.00408751e-01 -1.49666286e+00
1.11297682e-01 -3.37296844e-01 -1.27293622e+00 6.32554650e-01
-2.92406064e-02 -2.18863294e-01 -1.08621049e+00 1.28177845e+00
-1.46010295e-01 3.68523031e-01 5.58553815e-01 8.21520209e-01
8.65005076e-01 1.00846279e+00 -1.06569365e-01 9.30962786e-02
1.18008900e+00 -1.01503801e+00 -8.70262980e-01 -3.50949377e-01
3.41503471e-01 -6.84196949e-01 4.84268278e-01 2.86942661e-01
-8.99079919e-01 -7.60445237e-01 -1.51487052e+00 3.34060907e-01
-5.84116638e-01 -3.25391024e-01 3.19291353e-01 1.15888000e+00
-1.07599187e+00 4.07174170e-01 -5.26654840e-01 2.37637252e-01
6.14350855e-01 5.72202802e-01 -3.00868005e-01 -1.32976189e-01
-1.88552105e+00 8.11465502e-01 1.10489488e+00 2.64509976e-01
-1.20968783e+00 -7.95133039e-02 -1.11745059e+00 1.31948873e-01
1.99175045e-01 -3.30188602e-01 4.62335199e-01 -1.59677899e+00
-1.35450613e+00 8.18991125e-01 6.72476232e-01 -7.29157805e-01
5.59689283e-01 4.56548989e-01 -7.08918989e-01 9.48336571e-02
-4.36695307e-01 6.81046426e-01 1.17969751e+00 -1.40667951e+00
-5.34529626e-01 -1.21969298e-01 -2.08213761e-01 -7.27815703e-02
-1.74415261e-01 -3.43979627e-01 -2.20272601e-01 -5.50935090e-01
-9.67927650e-02 -9.64499414e-01 -5.98568358e-02 -1.84355870e-01
-5.25542319e-01 2.97302008e-01 1.50445580e+00 -9.16871607e-01
8.29719603e-01 -2.31360006e+00 3.88914272e-02 2.54329145e-01
2.98744917e-01 1.07144570e+00 -1.98764563e-01 2.78206885e-01
-3.16218704e-01 3.24206322e-01 -4.61455643e-01 2.56900210e-02
-1.90123096e-01 2.16836825e-01 -2.42855221e-01 7.07944334e-01
4.50352803e-02 1.27387869e+00 -5.89564264e-01 -4.63551372e-01
4.72750634e-01 9.02246773e-01 -2.83373922e-01 1.09185323e-01
3.24400724e-03 5.70303202e-01 -2.48919755e-01 1.93922833e-01
1.16569448e+00 -1.40061453e-01 2.97953226e-02 1.11274131e-01
2.77176559e-01 -3.45054448e-01 -1.04874861e+00 9.34391081e-01
-1.96030319e-01 7.60149837e-01 -3.86281274e-02 -1.17785287e+00
9.49015558e-01 3.38931561e-01 5.88230081e-02 -7.39370883e-01
7.30804324e-01 2.45057330e-01 2.60461152e-01 -8.22048604e-01
9.90545657e-03 -3.00626099e-01 1.01148561e-01 3.30625802e-01
-5.36802709e-02 3.49858403e-02 -3.60966533e-01 -2.62812704e-01
6.24574423e-01 -1.00677818e-01 1.49547830e-02 4.47035804e-02
9.78172600e-01 -4.96839195e-01 3.16586941e-01 6.05036199e-01
-8.27528611e-02 4.15519089e-01 4.41622466e-01 -4.62505132e-01
-1.27965009e+00 -4.63409811e-01 2.65891850e-01 4.00052190e-01
4.96601373e-01 3.73056740e-01 -9.56013858e-01 -5.48985958e-01
-3.23506624e-01 3.80489469e-01 -6.57920539e-01 -6.45280242e-01
-6.97616577e-01 -7.59836257e-01 8.94231558e-01 4.67408113e-02
1.62984872e+00 -1.60232246e+00 -4.49418485e-01 5.19334227e-02
-3.64131659e-01 -1.06531894e+00 -3.09477329e-01 -1.50694415e-01
-5.68363547e-01 -1.08198583e+00 -9.76846397e-01 -1.10719478e+00
6.70160592e-01 2.14412302e-01 4.28853780e-01 5.34983814e-01
-1.91659644e-01 -2.25551397e-01 -5.29470742e-01 -5.01672745e-01
-9.46970701e-01 -8.88046622e-02 -5.59926093e-01 3.14657927e-01
1.68829411e-01 -2.60552138e-01 -6.12417340e-01 2.76380897e-01
-1.47196054e+00 1.52902901e-01 1.02772570e+00 9.02970791e-01
1.95031598e-01 7.97313631e-01 4.12821248e-02 -1.07133949e+00
3.41060132e-01 -4.65172350e-01 -6.74362063e-01 4.30578887e-02
-8.18727732e-01 1.40982375e-01 6.89682364e-01 -3.60230982e-01
-7.86191344e-01 -3.53770107e-01 -3.77930939e-01 -3.24122608e-01
1.47867929e-02 1.73137859e-01 -4.25239146e-01 -5.72621882e-01
3.04558158e-01 9.12476420e-01 5.34274340e-01 -9.65046957e-02
-2.11105607e-02 7.34545946e-01 4.45826203e-01 2.95170426e-01
1.23791587e+00 6.85159743e-01 8.68317261e-02 -6.42713785e-01
-4.17621791e-01 2.42283961e-04 -9.01326314e-02 2.56545432e-02
1.14154315e+00 -6.90362811e-01 -1.06401968e+00 1.16564107e+00
-1.04589283e+00 -1.55415982e-01 1.62481010e-01 2.46666804e-01
-4.91592914e-01 7.73913860e-01 -4.22000527e-01 -6.96035087e-01
-4.93608296e-01 -1.40903378e+00 6.26803696e-01 2.85800606e-01
7.35241950e-01 -1.19555187e+00 -2.83665627e-01 2.94158936e-01
6.30565524e-01 8.31903756e-01 7.07193732e-01 -3.66700023e-01
-9.03611779e-01 -3.38204145e-01 -4.38991338e-01 9.24084187e-01
-1.20420253e-03 -5.83532631e-01 -8.12581539e-01 -6.56363964e-01
4.23832119e-01 -5.65162785e-02 1.31579351e+00 7.76069686e-02
1.31345081e+00 -8.01009774e-01 -2.00950921e-01 1.24436319e+00
1.76587605e+00 6.41604304e-01 1.68482721e+00 5.59251010e-01
8.09444010e-01 5.36389828e-01 5.49563169e-02 -1.08218342e-01
1.10109121e-01 1.71370804e-01 9.15024102e-01 -4.78168517e-01
-4.25910592e-01 -1.81062728e-01 5.25271237e-01 6.45637035e-01
-1.19397923e-01 -8.41145217e-01 -4.88063484e-01 3.48162562e-01
-1.52258611e+00 -1.20879698e+00 -3.85635048e-02 1.74624681e+00
6.48646414e-01 1.14418030e-01 -3.31284523e-01 5.28128266e-01
9.16971564e-01 5.07496715e-01 -2.82678545e-01 -1.96235195e-01
-3.54426205e-01 1.61060393e-01 9.41120207e-01 5.07669270e-01
-1.20174587e+00 9.36205149e-01 5.27790546e+00 9.44802105e-01
-1.48941684e+00 1.22931227e-01 7.52124608e-01 6.01780355e-01
-2.31801942e-01 -7.49662817e-02 -6.94112182e-01 8.33393753e-01
7.34317482e-01 2.01482370e-01 5.89326859e-01 4.33354586e-01
-3.50485295e-01 1.96066245e-01 -4.21132118e-01 9.45173204e-01
5.49367368e-01 -1.14936411e+00 3.48427147e-02 4.60400969e-01
8.90751302e-01 -3.49289805e-01 6.21383786e-01 1.58290088e-01
1.56557441e-01 -1.36902988e+00 3.27904224e-01 4.54741269e-01
8.93269002e-01 -9.62220430e-01 1.28688347e+00 3.69580179e-01
-7.04188287e-01 -1.79902792e-01 -6.14697576e-01 2.19468713e-01
-1.09072767e-01 1.52826905e-01 -7.07047462e-01 6.43661261e-01
6.83031797e-01 6.49288118e-01 -4.79449481e-01 7.90798306e-01
-4.62244600e-01 5.91420949e-01 2.01961994e-01 -1.36124134e-01
6.32576406e-01 1.47383869e-01 5.92391610e-01 1.03988934e+00
2.92008787e-01 9.34234075e-03 -2.16984242e-01 9.70141053e-01
-8.57508108e-02 -2.36001447e-01 -7.66908109e-01 -8.35112855e-02
-6.52347505e-02 7.47242391e-01 -7.26830423e-01 -1.42748222e-01
9.06536356e-03 1.19716072e+00 -5.53179383e-01 3.78340602e-01
-9.49299335e-01 -8.66850376e-01 1.40721753e-01 3.97532471e-02
8.21240664e-01 4.86808605e-02 -6.13422832e-03 -9.58543301e-01
-3.00554991e-01 -1.15516877e+00 1.25902191e-01 -5.90941191e-01
-7.64588356e-01 6.34763896e-01 -2.14599311e-01 -1.32153726e+00
-4.01901044e-02 -6.79011047e-01 -5.36271453e-01 9.15717304e-01
-2.14971828e+00 -1.40406597e+00 -4.10656363e-01 8.07679951e-01
2.09757820e-01 -5.94982862e-01 6.10370398e-01 2.10681736e-01
-7.52786398e-02 5.90666115e-01 1.94548905e-01 6.75895393e-01
1.80428058e-01 -7.07805932e-01 5.38075328e-01 9.76119876e-01
-3.40767056e-01 1.71523541e-01 6.42934978e-01 -6.31875932e-01
-1.20557213e+00 -1.21394479e+00 5.30324459e-01 2.72498578e-02
1.58444539e-01 -2.00516403e-01 -8.02220404e-01 6.84155285e-01
5.26214719e-01 -1.41598687e-01 2.67266095e-01 -9.95967805e-01
-2.70810276e-01 2.25555971e-02 -1.54914486e+00 3.19199026e-01
5.06614089e-01 -2.15111837e-01 -1.99378267e-01 8.83363839e-03
8.06944609e-01 -4.30749238e-01 -3.11535805e-01 2.25693464e-01
4.34935927e-01 -1.11252284e+00 1.11733496e+00 -1.60683528e-01
6.94702148e-01 -3.80647838e-01 1.22353867e-01 -1.07401013e+00
-1.59435630e-01 -6.76266551e-01 1.83241710e-01 7.62934983e-01
5.96515881e-03 -9.79680419e-01 6.55197442e-01 -1.86766371e-01
2.25389928e-01 -5.22770941e-01 -6.97224081e-01 -8.26296508e-01
1.38190940e-01 -4.24521156e-02 8.90052378e-01 7.74734378e-01
-6.44262612e-01 -2.29355469e-01 -1.15175712e+00 2.44726434e-01
9.57534552e-01 -2.56479770e-01 5.56219995e-01 -8.40389788e-01
-2.41341665e-01 -2.09245324e-01 -7.93334603e-01 -6.43250823e-01
8.67244527e-02 -7.76317716e-01 1.08302332e-01 -1.19497800e+00
-2.26304755e-02 -2.61502355e-01 -2.84475863e-01 4.54799145e-01
-3.11657917e-02 7.80939519e-01 3.14619392e-01 1.92598403e-01
-3.77948165e-01 4.02084351e-01 1.85444832e+00 -4.94734943e-01
2.31119394e-01 -9.06923693e-03 -8.60860884e-01 5.65463662e-01
8.48358393e-01 -7.46561587e-01 -2.89753675e-01 -4.17173415e-01
2.84004975e-02 1.35011628e-01 7.33595967e-01 -1.23418963e+00
-5.16454652e-02 3.34732309e-02 4.95060176e-01 -2.65688598e-01
1.10159323e-01 -1.11001849e+00 3.28424513e-01 1.26980662e+00
-1.82078689e-01 -4.79830682e-01 -4.58553955e-02 6.81236863e-01
-2.31186703e-01 -3.25744361e-01 1.28889692e+00 -5.08145213e-01
-8.82470787e-01 2.54540920e-01 -2.27264076e-01 -2.27240592e-01
1.25470042e+00 -4.79117006e-01 -2.26629183e-01 -6.18086696e-01
-4.36809033e-01 -1.88388806e-02 2.15617970e-01 4.33377087e-01
1.04450750e+00 -1.21916139e+00 -7.52420962e-01 7.68116176e-01
-2.26179197e-01 -3.66895586e-01 2.87918866e-01 4.66884166e-01
-9.81519699e-01 4.95956451e-01 -6.16938055e-01 -2.60349870e-01
-1.20470166e+00 8.32298279e-01 5.78855395e-01 -3.48874420e-01
-6.80855513e-01 9.38706040e-01 2.77097762e-01 -9.39575806e-02
1.37372658e-01 8.83264765e-02 -4.96274173e-01 -5.02411664e-01
6.91637456e-01 2.52449572e-01 -3.68560731e-01 -8.72427881e-01
-1.76267081e-03 5.05865037e-01 -1.38752088e-01 1.13496572e-01
1.25224006e+00 -2.77491540e-01 -3.45804393e-01 -2.77963400e-01
1.63864017e+00 -3.70822698e-01 -1.22063386e+00 -2.08786190e-01
-5.26278377e-01 -5.78179300e-01 1.69778809e-01 -6.10211492e-01
-1.68259132e+00 9.77080703e-01 1.15033197e+00 4.75778222e-01
1.21554887e+00 -3.07799280e-01 1.43882346e+00 1.30982637e-01
2.28075922e-01 -7.13580370e-01 2.88065374e-01 3.31416070e-01
6.67529583e-01 -1.26151860e+00 -2.43693322e-01 -1.89534292e-01
-6.67418063e-01 1.08059978e+00 2.88023889e-01 -5.01876175e-01
7.57727325e-01 1.73643336e-01 6.07116781e-02 -3.01172942e-01
-1.09785542e-01 -1.13522567e-01 1.05109468e-01 8.96461904e-01
-1.67344972e-01 -1.74862653e-01 -2.45858714e-01 1.50742963e-01
-2.01658145e-01 7.60977566e-02 5.95971406e-01 7.59930491e-01
-6.27983749e-01 -1.04908419e+00 -7.36196756e-01 -1.09929815e-01
-7.82462418e-01 -1.72804430e-01 -1.34731069e-01 8.33889306e-01
6.36678278e-01 9.46571529e-01 -2.17355579e-01 -6.88551188e-01
-1.97916240e-01 -2.68883169e-01 2.81420052e-01 -2.87131250e-01
-3.52213413e-01 -1.10768974e-01 -5.91223419e-01 -1.48554131e-01
-4.87140745e-01 3.25794481e-02 -9.49219763e-01 -6.18151903e-01
-4.11460459e-01 5.50774038e-02 7.98763752e-01 9.54758048e-01
-9.55202803e-02 8.17312777e-01 8.51719320e-01 -5.68439782e-01
-3.83229345e-01 -8.15590084e-01 -6.41605556e-01 3.75659615e-01
7.13217914e-01 -8.88427868e-02 -6.40910923e-01 3.05587232e-01] | [4.303144931793213, 8.056011199951172] |
b923a75d-2d31-499e-af67-12b8278bb164 | measured-albedo-in-the-wild-filling-the-gap | 2306.15662 | null | https://arxiv.org/abs/2306.15662v2 | https://arxiv.org/pdf/2306.15662v2.pdf | Measured Albedo in the Wild: Filling the Gap in Intrinsics Evaluation | Intrinsic image decomposition and inverse rendering are long-standing problems in computer vision. To evaluate albedo recovery, most algorithms report their quantitative performance with a mean Weighted Human Disagreement Rate (WHDR) metric on the IIW dataset. However, WHDR focuses only on relative albedo values and often fails to capture overall quality of the albedo. In order to comprehensively evaluate albedo, we collect a new dataset, Measured Albedo in the Wild (MAW), and propose three new metrics that complement WHDR: intensity, chromaticity and texture metrics. We show that existing algorithms often improve WHDR metric but perform poorly on other metrics. We then finetune different algorithms on our MAW dataset to significantly improve the quality of the reconstructed albedo both quantitatively and qualitatively. Since the proposed intensity, chromaticity, and texture metrics and the WHDR are all complementary we further introduce a relative performance measure that captures average performance. By analysing existing algorithms we show that there is significant room for improvement. Our dataset and evaluation metrics will enable researchers to develop algorithms that improve albedo reconstruction. Code and Data available at: https://measuredalbedo.github.io/ | ['Soumyadip Sengupta', 'David Jacobs', 'Hariharmano Shanmugaraja', 'Sanjoy Chowdhury', 'Jiaye Wu'] | 2023-06-27 | null | null | null | null | ['intrinsic-image-decomposition', 'inverse-rendering'] | ['computer-vision', 'computer-vision'] | [ 2.19301075e-01 -3.48721981e-01 2.47568101e-01 -3.79649162e-01
-5.93693137e-01 -5.58702111e-01 5.27931511e-01 -9.18384269e-02
-1.29851550e-01 5.41804492e-01 3.67754996e-01 -8.73592496e-02
-6.67067850e-03 -7.65883803e-01 -2.75667250e-01 -9.36489940e-01
7.63034150e-02 -5.73266000e-02 1.90576583e-01 -4.68625844e-01
1.18844911e-01 4.56008047e-01 -1.81455946e+00 -1.02676779e-01
1.18958414e+00 9.79803443e-01 -1.17644891e-01 8.54632676e-01
2.08366677e-01 1.02313709e+00 -6.58169389e-01 -1.66821092e-01
8.03298652e-01 -7.46544182e-01 -6.69521451e-01 3.10341287e-02
1.03992045e+00 -4.31920707e-01 -5.06248288e-02 8.60479772e-01
5.86423039e-01 1.27290100e-01 5.17101169e-01 -1.36780131e+00
-4.04189020e-01 -1.82428971e-01 -5.63742220e-01 -4.42653447e-02
5.38499117e-01 4.66428161e-01 1.02222455e+00 -8.42431843e-01
5.33078372e-01 8.43667865e-01 7.65934110e-01 2.24037126e-01
-1.17802989e+00 -2.86331952e-01 -1.84276775e-01 1.12002537e-01
-1.30275989e+00 -4.64147776e-01 8.72395217e-01 -5.04736543e-01
5.08986294e-01 6.14205778e-01 6.40137792e-01 4.34425205e-01
-1.05994023e-01 5.14558256e-01 1.68935668e+00 -7.07624674e-01
5.18996492e-02 5.11301076e-03 -1.53873548e-01 7.93416619e-01
3.38660747e-01 1.20389186e-01 -7.36843228e-01 8.75662789e-02
4.12452310e-01 -1.31644934e-01 -6.73166096e-01 -2.88905054e-01
-1.44339538e+00 5.99882245e-01 6.24768972e-01 -2.68612623e-01
-2.29770124e-01 9.50933322e-02 -2.00104848e-01 4.26895440e-01
8.21666420e-01 5.46177566e-01 -3.14870477e-01 -1.11886617e-02
-8.51363659e-01 3.74106556e-01 8.09615076e-01 5.88854790e-01
1.28800619e+00 2.83313747e-02 1.25056148e-01 1.03369021e+00
3.58436614e-01 1.17384243e+00 -1.20241001e-01 -1.55792475e+00
1.45028368e-01 4.57993597e-01 4.76282209e-01 -1.14370489e+00
-2.49345541e-01 -1.26233324e-01 -5.97726047e-01 7.52555311e-01
5.93460083e-01 -2.08579209e-02 -8.60633016e-01 1.40355325e+00
4.74759191e-01 -3.83572400e-01 1.73336193e-02 1.30984306e+00
9.92708504e-01 5.28176129e-01 -4.12417591e-01 -1.31163597e-01
1.03493190e+00 -1.28220510e+00 -7.15588629e-01 -2.35123336e-01
4.94642794e-01 -1.29793561e+00 1.29126918e+00 2.88676798e-01
-9.25575614e-01 -2.81388015e-01 -1.09883583e+00 -9.13365260e-02
-8.14932063e-02 -8.10096040e-02 6.06464982e-01 6.81070566e-01
-1.06862521e+00 5.45131981e-01 -4.91004705e-01 -3.84468913e-01
-3.93591076e-02 -2.03253135e-01 -1.88740313e-01 -1.91680014e-01
-8.83361042e-01 9.01979983e-01 -3.08295906e-01 1.31952330e-01
-6.83752656e-01 -7.09665000e-01 -7.15796709e-01 -4.56671178e-01
2.86365092e-01 -6.59275591e-01 8.81731808e-01 -1.13570261e+00
-1.54865861e+00 9.49826002e-01 -1.39307961e-01 -6.95632920e-02
5.56954324e-01 -9.60058942e-02 -3.61835986e-01 3.93767864e-01
-1.11931041e-01 5.90319157e-01 7.20925689e-01 -1.71804273e+00
-4.80594337e-01 -2.56630719e-01 2.18076736e-01 4.40791756e-01
2.84526944e-02 -1.98192924e-01 -2.26986930e-01 -4.18968230e-01
4.06876951e-01 -1.00032485e+00 7.84862190e-02 3.65696758e-01
-2.98382845e-02 2.08258301e-01 5.61242878e-01 -6.43362045e-01
1.10563707e+00 -1.99013305e+00 -1.25437513e-01 -1.96382813e-02
4.72644687e-01 1.65718570e-01 -3.17570090e-01 3.29443693e-01
1.94956601e-01 -4.76082563e-02 -6.04540646e-01 -1.31110072e-01
-1.93322524e-01 1.84919521e-01 -3.50236706e-02 6.88033462e-01
1.28160536e-01 5.24346709e-01 -9.60886240e-01 -4.75010812e-01
4.70101506e-01 6.16658509e-01 -4.24629360e-01 2.76110560e-01
-9.51023772e-02 3.88383865e-01 6.94661140e-02 7.88045704e-01
1.12014389e+00 1.63060546e-01 -1.83231622e-01 -3.54673803e-01
-2.61117876e-01 8.06911960e-02 -1.21877623e+00 1.17186666e+00
-5.84479213e-01 1.10110831e+00 1.25507712e-01 -3.79587561e-01
1.06191063e+00 -2.01850280e-01 6.23168290e-01 -8.58786106e-01
-1.01641499e-01 2.94748753e-01 -2.12900564e-01 -5.42602777e-01
6.18898451e-01 -2.22851411e-01 5.05138040e-01 4.16467309e-01
-3.28135639e-01 -7.43085980e-01 2.19143890e-02 2.49688756e-02
9.51240599e-01 4.92457241e-01 2.66909778e-01 -3.99618357e-01
5.52155495e-01 3.01984072e-01 4.24154490e-01 6.00624263e-01
-5.62982023e-01 1.30476308e+00 3.31305861e-01 -7.02076018e-01
-1.11331356e+00 -1.18180728e+00 -1.89705655e-01 1.11612403e+00
2.96005070e-01 -2.71541804e-01 -7.59429455e-01 -3.87250662e-01
-9.05287638e-02 2.83351630e-01 -7.10354686e-01 4.45778489e-01
-2.29231611e-01 -9.00974452e-01 3.66412550e-01 -1.29192099e-02
7.93834746e-01 -7.22617447e-01 -8.96501780e-01 -3.91720414e-01
-5.79296470e-01 -1.07240117e+00 -3.47392470e-01 -2.99432904e-01
-6.64871454e-01 -1.15846527e+00 -7.81403601e-01 -2.37059236e-01
4.50680494e-01 8.44851315e-01 1.52411783e+00 3.80308837e-01
-2.86490053e-01 5.58954298e-01 -7.41442442e-01 -5.77564538e-01
-3.28110188e-01 -2.18235150e-01 -1.46425068e-01 1.05229795e-01
2.52876058e-02 -4.43215162e-01 -1.05084229e+00 7.63696492e-01
-1.01785541e+00 2.38578364e-01 2.07063183e-01 4.60992783e-01
4.80428845e-01 -4.78405654e-01 -2.36022130e-01 -7.55273342e-01
2.13918969e-01 -1.37396991e-01 -7.59959459e-01 8.15753639e-02
-1.13181579e+00 -8.14564005e-02 2.25546315e-01 5.56361163e-03
-1.03105903e+00 -4.74238992e-02 9.39810723e-02 -1.56578958e-01
-6.94547594e-02 2.63868868e-01 1.60153627e-01 -3.85351062e-01
1.11415231e+00 -1.08199660e-02 2.07056671e-01 -2.94744402e-01
2.87374198e-01 6.18510544e-01 3.79996896e-01 -3.28376651e-01
1.00339711e+00 9.11606252e-01 2.52914310e-01 -1.14867842e+00
-1.13636732e+00 -7.23055720e-01 -4.76136625e-01 -5.18883884e-01
5.62574327e-01 -1.06122959e+00 -6.75498307e-01 6.74077153e-01
-7.71124601e-01 -6.78196311e-01 -3.21883291e-01 4.23151433e-01
-4.05595541e-01 3.87931198e-01 -1.62065830e-02 -9.30180252e-01
-1.27321139e-01 -1.02029681e+00 1.27860701e+00 8.18402022e-02
1.93326145e-01 -1.16971505e+00 4.60808843e-01 6.75126433e-01
7.67407119e-01 7.75160670e-01 3.10748816e-01 6.80695176e-01
-5.17039120e-01 5.17337583e-02 -5.75685799e-01 5.36122322e-01
4.11289223e-02 3.10618967e-01 -1.22880292e+00 -7.36950859e-02
-3.21414955e-02 -1.47547200e-01 1.10311508e+00 3.80508721e-01
7.97699690e-01 -2.64875799e-01 3.17863137e-01 1.02630174e+00
1.83275259e+00 -4.15401489e-01 1.04145575e+00 6.74792826e-01
8.39249909e-01 9.19943869e-01 6.53139174e-01 4.24241632e-01
6.09210670e-01 6.23896837e-01 7.85853207e-01 -4.18497384e-01
-7.70860970e-01 2.20798150e-01 3.77075553e-01 7.17807055e-01
-8.64690959e-01 -2.12662324e-01 -1.05076468e+00 3.48780692e-01
-1.61371493e+00 -9.26297307e-01 -6.31424606e-01 2.45762277e+00
6.05657995e-01 -4.42731351e-01 6.25362620e-02 2.03472093e-01
3.04655939e-01 7.02186823e-01 -1.07461102e-01 -3.53235543e-01
-5.50157964e-01 9.73428413e-02 8.67618024e-01 1.06479383e+00
-8.32250714e-01 7.02082157e-01 6.66151237e+00 2.52967745e-01
-1.20689213e+00 1.73820466e-01 4.41517502e-01 -1.19535580e-01
-6.43536985e-01 1.38076961e-01 -2.37535849e-01 8.38626772e-02
7.22731650e-01 1.69614658e-01 7.90908694e-01 5.35370350e-01
3.37800056e-01 -7.44011343e-01 -5.70502222e-01 9.94788289e-01
3.50189656e-01 -9.93727446e-01 -2.63036698e-01 1.69306397e-01
1.04432130e+00 3.23775440e-01 -3.02283950e-02 -1.84354618e-01
1.64775297e-01 -9.76782441e-01 6.81909502e-01 8.05448830e-01
9.09463406e-01 -2.80328661e-01 7.24963546e-01 -2.41268203e-01
-1.04628062e+00 3.53892386e-01 -3.45680743e-01 -2.54507542e-01
-2.05832943e-01 8.17618251e-01 -5.59951544e-01 6.12002552e-01
8.68861496e-01 6.47061586e-01 -9.16315496e-01 9.49367225e-01
-4.26856160e-01 5.52357793e-01 -4.40203130e-01 3.35571021e-02
1.36450976e-01 -5.70190310e-01 5.00761569e-01 1.00461411e+00
1.61285207e-01 -1.28143923e-02 -6.79330081e-02 6.51443481e-01
2.08934061e-02 2.24906489e-01 -5.22840202e-01 4.62782025e-01
1.11893326e-01 1.39051425e+00 -3.57900351e-01 -1.35779187e-01
-4.21813965e-01 9.63125646e-01 -1.37164928e-02 5.97926080e-01
-8.09244156e-01 -4.23522770e-01 9.24519420e-01 3.20701271e-01
-2.11216167e-01 -4.31537688e-01 -2.81429023e-01 -1.13738942e+00
1.58700183e-01 -8.72721255e-01 7.66760856e-02 -1.16593385e+00
-1.00052023e+00 4.99441594e-01 1.73732042e-02 -1.57329142e+00
1.74330860e-01 -7.02644646e-01 -4.77529734e-01 9.62765694e-01
-1.83205497e+00 -1.00778306e+00 -1.34441423e+00 4.24381465e-01
3.51673335e-01 1.67973951e-01 7.30153680e-01 1.00261997e-02
-3.99151504e-01 1.38816088e-01 3.10450315e-01 -6.60550967e-02
1.10523856e+00 -1.53259623e+00 9.89732966e-02 9.46616232e-01
7.51231238e-02 2.70139128e-01 1.09456027e+00 -1.13157988e-01
-1.54442012e+00 -9.13310289e-01 5.12406349e-01 -7.06954658e-01
3.71279061e-01 -2.16981377e-02 -6.68360949e-01 3.50230008e-01
3.25147122e-01 4.96309251e-01 4.08380121e-01 6.15979880e-02
-5.50693214e-01 -5.30725777e-01 -1.09507549e+00 5.70481420e-01
8.22348297e-01 -4.04434979e-01 -1.40513927e-01 4.82800275e-01
5.59622467e-01 -3.66895586e-01 -1.03190517e+00 4.89461958e-01
8.31239223e-01 -1.79406333e+00 9.29107428e-01 3.41059640e-02
4.78241831e-01 -6.66843951e-01 -5.86770892e-01 -1.31014001e+00
-9.40074995e-02 -7.11045742e-01 2.42522880e-02 8.77487957e-01
4.90155905e-01 -8.42303514e-01 5.37206173e-01 3.64164025e-01
1.19315699e-01 -5.46624184e-01 -4.98039514e-01 -7.74273932e-01
-1.09102823e-01 -5.93585312e-01 6.10076010e-01 9.65196609e-01
-6.20280087e-01 2.13514701e-01 -5.43087125e-01 2.68816411e-01
9.67742622e-01 3.84877503e-01 1.27128005e+00 -1.09876359e+00
1.73474662e-02 -3.73028010e-01 -2.78793037e-01 -5.84225297e-01
-4.12270069e-01 -6.48372471e-01 2.51785606e-01 -1.71727824e+00
1.58805683e-01 -3.66781890e-01 -1.08277477e-01 3.40527356e-01
-2.84114033e-01 9.97386992e-01 2.32001394e-01 5.62929153e-01
-4.52765971e-01 4.56504285e-01 1.34368408e+00 -5.89499362e-02
-2.91595519e-01 -3.77312541e-01 -3.22479934e-01 7.41263926e-01
9.16224122e-01 -3.15978318e-01 -9.82568264e-02 -8.19681048e-01
4.05250043e-01 -2.75183320e-01 4.68341231e-01 -1.28789568e+00
-2.18002200e-01 -4.96267647e-01 3.78923118e-01 -2.91279852e-01
2.10679278e-01 -5.77770114e-01 6.37090728e-02 3.25080365e-01
1.04010545e-01 -1.92676067e-01 -3.25231016e-01 1.78685024e-01
-2.07571790e-01 -2.75788084e-02 9.98878598e-01 -1.50403669e-02
-7.28965044e-01 2.11934939e-01 -5.87529801e-02 1.38379812e-01
4.93370384e-01 -3.22399288e-01 -5.88946700e-01 -6.51518762e-01
-1.78940967e-01 -5.88564724e-02 1.11108196e+00 7.54051954e-02
6.15657151e-01 -1.14722621e+00 -1.03149879e+00 3.97770619e-03
6.85360968e-01 -2.01763451e-01 -4.20494154e-02 1.25990152e+00
-1.37545383e+00 -1.66167021e-01 -1.52732953e-01 -7.32440114e-01
-1.37546635e+00 5.85089214e-02 6.38628483e-01 1.59459874e-01
-3.76614034e-01 5.67097187e-01 7.73921758e-02 -4.39828306e-01
-1.25757843e-01 -2.45090052e-01 1.71031758e-01 2.49690823e-02
4.41184074e-01 8.22621942e-01 6.89997822e-02 -7.71507680e-01
-3.78738731e-01 1.00243473e+00 8.54405224e-01 -1.73524767e-01
1.25861156e+00 -6.28452122e-01 -3.93029660e-01 4.74464417e-01
9.91098106e-01 4.90980357e-01 -1.46858788e+00 -1.32944480e-01
-4.56446022e-01 -1.08864307e+00 3.62444490e-01 -7.86568940e-01
-1.14398170e+00 8.16096544e-01 8.65614295e-01 3.86559933e-01
1.61877596e+00 -2.32900485e-01 6.14329457e-01 3.33178133e-01
1.74002022e-01 -1.02300572e+00 1.18670709e-01 5.46525002e-01
1.04630685e+00 -1.70106769e+00 6.69532537e-01 -5.33511639e-01
-7.30735719e-01 1.14973903e+00 3.98777336e-01 6.94143819e-03
4.84376401e-01 9.31845382e-02 8.10625851e-01 -1.82231084e-01
-2.82150179e-01 -7.55325019e-01 4.75113839e-01 6.42719507e-01
6.96569920e-01 1.64626390e-01 -2.06882715e-01 -4.59589034e-01
-3.88826251e-01 -3.32412541e-01 7.84878016e-01 6.35498941e-01
-7.26503789e-01 -6.86438322e-01 -6.89279139e-01 2.28727788e-01
-5.90173863e-02 -1.45151354e-02 -3.95940185e-01 7.49750793e-01
1.60243198e-01 1.11079562e+00 -1.14836797e-01 -4.76694256e-01
2.86564440e-01 -2.40868390e-01 6.71238601e-01 -8.41134861e-02
-2.06054285e-01 8.50355078e-04 1.30192488e-01 -7.74888396e-01
-1.03601921e+00 -3.08478504e-01 -6.73012078e-01 -6.92717016e-01
-2.31249481e-01 -1.02892622e-01 8.49685669e-01 4.56286043e-01
5.38455173e-02 2.10187808e-01 1.09275174e+00 -9.96761441e-01
1.29044667e-01 -1.09459436e+00 -5.37376940e-01 5.94054163e-01
7.42783904e-01 -5.98377883e-01 -7.10300326e-01 1.17654070e-01] | [9.9053955078125, -2.861752510070801] |
9d7193c8-015f-4c00-950c-b48e0015b040 | a-comparison-of-supervised-and-unsupervised | 2107.09204 | null | https://arxiv.org/abs/2107.09204v1 | https://arxiv.org/pdf/2107.09204v1.pdf | A Comparison of Supervised and Unsupervised Deep Learning Methods for Anomaly Detection in Images | Anomaly detection in images plays a significant role for many applications across all industries, such as disease diagnosis in healthcare or quality assurance in manufacturing. Manual inspection of images, when extended over a monotonously repetitive period of time is very time consuming and can lead to anomalies being overlooked.Artificial neural networks have proven themselves very successful on simple, repetitive tasks, in some cases even outperforming humans. Therefore, in this paper we investigate different methods of deep learning, including supervised and unsupervised learning, for anomaly detection applied to a quality assurance use case. We utilize the MVTec anomaly dataset and develop three different models, a CNN for supervised anomaly detection, KD-CAE for autoencoder anomaly detection, NI-CAE for noise induced anomaly detection and a DCGAN for generating reconstructed images. By experiments, we found that KD-CAE performs better on the anomaly datasets compared to CNN and NI-CAE, with NI-CAE performing the best on the Transistor dataset. We also implemented a DCGAN for the creation of new training data but due to computational limitation and lack of extrapolating the mechanics of AnoGAN, we restricted ourselves just to the generation of GAN based images. We conclude that unsupervised methods are more powerful for anomaly detection in images, especially in a setting where only a small amount of anomalous data is available, or the data is unlabeled. | ['Zhenning Li', 'Håkon Sandaker', 'Tabea Redl', 'Sauraj Verma', 'Vincent Wilmet'] | 2021-07-20 | null | null | null | null | ['supervised-anomaly-detection'] | ['computer-vision'] | [ 3.84444058e-01 1.21885896e-01 5.67083836e-01 -7.78759867e-02
-2.44022846e-01 -3.61329734e-01 5.42236090e-01 1.24102138e-01
-2.45852888e-01 4.75363612e-01 -3.67643982e-01 -6.82703078e-01
-4.42613550e-02 -9.01603281e-01 -6.02945685e-01 -7.71257162e-01
-5.30509949e-02 5.07726610e-01 -3.95178758e-02 -1.82723343e-01
2.30484143e-01 7.46173143e-01 -1.63853848e+00 1.80590615e-01
6.99926734e-01 1.12285745e+00 -3.45018268e-01 8.29347372e-01
-1.07341446e-01 8.70945752e-01 -8.71405303e-01 -4.26293492e-01
4.56303924e-01 -7.06692278e-01 -6.22652590e-01 4.46973443e-01
2.15649292e-01 -3.43776375e-01 -1.00047506e-01 1.03298962e+00
4.72415835e-01 8.29103962e-02 6.93128586e-01 -1.39905095e+00
-5.05422890e-01 8.35129768e-02 -4.48478073e-01 4.11261201e-01
1.91359594e-01 4.60149854e-01 5.04413128e-01 -6.56690598e-01
3.93302500e-01 7.77295113e-01 7.42065728e-01 6.92916989e-01
-1.18885374e+00 -4.99907106e-01 -2.62264609e-01 1.48106694e-01
-9.84983802e-01 -1.84909567e-01 9.63411987e-01 -5.14866233e-01
1.07286835e+00 1.97176933e-01 5.48572838e-01 1.40266228e+00
4.56899941e-01 5.78548253e-01 1.04896998e+00 -5.81911623e-01
5.19086599e-01 -7.55363107e-02 -4.00500804e-01 5.20958364e-01
3.05549115e-01 3.85467231e-01 -2.75569614e-02 -1.38910517e-01
8.30702066e-01 1.66672453e-01 9.06888843e-02 -1.06911585e-01
-8.92091751e-01 1.00138950e+00 1.51792288e-01 6.68773592e-01
-7.17272580e-01 7.08198454e-03 6.33584857e-01 9.54333901e-01
4.15953308e-01 9.03793752e-01 -4.47752774e-01 -2.83235580e-01
-7.33925819e-01 -5.02784252e-02 6.05072916e-01 6.07654333e-01
4.71118242e-01 7.63754845e-01 1.50903270e-01 6.72647536e-01
-1.44604161e-01 2.40069702e-01 7.99121857e-01 -7.42194355e-01
-6.07767925e-02 5.84827960e-01 -2.05157503e-01 -1.10421538e+00
-3.99249554e-01 -3.61746818e-01 -1.14756942e+00 6.60093725e-01
4.37595218e-01 -2.37640068e-01 -1.32168889e+00 1.33993399e+00
-5.35673387e-02 1.44167738e-02 2.27114230e-01 4.47313637e-01
2.79608339e-01 5.58791578e-01 -2.64080107e-01 -5.93080521e-02
8.21009040e-01 -5.94604790e-01 -7.68245280e-01 -1.07273467e-01
8.87621522e-01 -6.81076944e-01 1.06188989e+00 9.59142983e-01
-8.18769455e-01 -5.41041255e-01 -1.05566156e+00 5.43861091e-01
-7.54005909e-01 -1.86630413e-01 5.31013131e-01 7.79245734e-01
-9.24290776e-01 8.53005409e-01 -8.23227942e-01 -5.18100917e-01
6.75485849e-01 4.77335185e-01 -5.67911565e-01 -7.26731718e-02
-8.08873415e-01 8.40107918e-01 3.84356111e-01 1.81666583e-01
-7.26920784e-01 -1.47108302e-01 -9.76129889e-01 -1.88310862e-01
2.67514557e-01 -2.29584098e-01 1.05385149e+00 -1.48180997e+00
-1.36404681e+00 8.02496791e-01 3.42870891e-01 -9.09946024e-01
5.47425568e-01 -1.51976958e-01 -1.02624083e+00 3.37971970e-02
-1.72156483e-01 3.45203608e-01 1.07857096e+00 -1.17108047e+00
-3.84518296e-01 -2.84555018e-01 -2.29054198e-01 -3.93526703e-01
-5.86566105e-02 -2.76125282e-01 1.92077443e-01 -8.09971809e-01
1.41435429e-01 -7.21898913e-01 -3.80162627e-01 -3.57644200e-01
-5.30039668e-01 9.37440693e-02 1.28739047e+00 -7.73963749e-01
8.14248502e-01 -2.21950603e+00 -3.07138652e-01 7.61019945e-01
-2.91013531e-02 5.48378050e-01 3.71416621e-02 2.68089592e-01
-4.71005619e-01 4.08304445e-02 -6.90145135e-01 -3.31821479e-02
-3.44050676e-01 6.84552848e-01 -1.32453948e-01 3.72336119e-01
6.87444627e-01 8.69418442e-01 -8.47145617e-01 -1.99433044e-01
4.70579505e-01 1.12450749e-01 -3.68744522e-01 2.87537277e-01
-1.01248056e-01 8.97570729e-01 -1.75130919e-01 8.88473332e-01
4.14231896e-01 -2.18173042e-02 -2.74238765e-01 1.62199780e-01
2.71373779e-01 -2.37335369e-01 -9.21946645e-01 1.49352574e+00
-4.63861525e-01 8.19607794e-01 -3.13646197e-01 -1.35381258e+00
9.74350750e-01 4.32188153e-01 4.98970598e-01 -1.12906837e+00
2.50651121e-01 3.73900414e-01 4.80229497e-01 -6.63935781e-01
2.15417594e-01 -2.28340432e-01 1.49032041e-01 4.99934733e-01
2.31876150e-01 -8.39255676e-02 1.55041546e-01 1.54536311e-02
1.50290847e+00 -1.34593129e-01 2.19653919e-01 -3.73609830e-03
5.61956644e-01 1.06990568e-01 2.60198176e-01 7.54327834e-01
-8.01669881e-02 9.43479002e-01 5.70395052e-01 -6.86015904e-01
-1.43709457e+00 -9.12361205e-01 3.23218852e-02 3.35103363e-01
-3.71056169e-01 2.34202687e-02 -8.46510589e-01 -1.11555266e+00
-1.81847721e-01 9.72876132e-01 -5.68416297e-01 -5.44126570e-01
-4.78366494e-01 -6.62498534e-01 5.69404244e-01 6.61230683e-01
5.86210012e-01 -1.58558285e+00 -6.03330195e-01 2.01082006e-01
3.99395287e-01 -1.03822017e+00 6.92113340e-02 3.42299968e-01
-9.84837651e-01 -1.00765705e+00 -3.14431459e-01 -5.54032624e-01
9.06886280e-01 -4.63229090e-01 1.22578585e+00 3.28460127e-01
-6.34466410e-01 6.26297116e-01 -4.68156815e-01 -8.34389031e-01
-8.93148124e-01 -3.29443574e-01 5.75922951e-02 4.68531437e-02
5.57148993e-01 -7.71039546e-01 -4.81187373e-01 3.82120967e-01
-1.35380709e+00 -6.29637182e-01 7.77459443e-01 1.04207635e+00
5.46841502e-01 4.59508896e-01 6.47524416e-01 -1.25859416e+00
4.77182090e-01 -4.65899408e-01 -4.89158064e-01 -3.90414029e-01
-7.24094391e-01 -7.68831521e-02 7.52169311e-01 -4.04542893e-01
-7.62866914e-01 5.23814410e-02 -7.92259932e-01 -6.22575879e-01
-7.32363164e-01 3.37067068e-01 -1.37311086e-01 -6.17114007e-02
9.10230160e-01 1.05555318e-01 3.20510954e-01 -3.76826346e-01
-7.08047524e-02 5.10091364e-01 8.08971167e-01 -1.13800772e-01
1.01320410e+00 4.30057108e-01 9.65565890e-02 -9.59912956e-01
-3.75949234e-01 -3.32603395e-01 -6.14956737e-01 -1.69690579e-01
8.66515636e-01 -4.86933231e-01 -1.66888878e-01 7.64115155e-01
-7.76072323e-01 -3.55856121e-01 -1.01883531e+00 3.35936159e-01
-5.47111332e-01 3.63657653e-01 -4.01305586e-01 -7.70321608e-01
-2.16380730e-01 -9.08907175e-01 6.71221435e-01 -5.01536913e-02
-2.44038880e-01 -1.18258202e+00 3.44885588e-02 -3.68879102e-02
6.21638834e-01 7.65820265e-01 1.03589821e+00 -1.16086125e+00
-2.21810669e-01 -6.46049678e-01 2.38778159e-01 1.04469371e+00
4.85752016e-01 -7.86320791e-02 -1.15437126e+00 -2.28642017e-01
3.79101634e-01 -2.40421325e-01 6.03941083e-01 1.38829991e-01
1.64884019e+00 -2.19884723e-01 6.94104061e-02 3.69255155e-01
1.25872338e+00 6.68554366e-01 1.24217319e+00 4.64817971e-01
7.41418660e-01 3.38398218e-01 5.56569099e-01 6.41971454e-02
-4.92827028e-01 2.62011766e-01 7.93590665e-01 -4.05856043e-01
1.71774581e-01 5.54035865e-02 2.60848343e-01 7.47608602e-01
-2.35960886e-01 -4.50805098e-01 -8.43450904e-01 6.48199499e-01
-1.54457724e+00 -9.01981890e-01 -7.89151266e-02 2.19922137e+00
1.87093720e-01 4.67131674e-01 1.73451900e-01 8.00997794e-01
5.49422622e-01 -3.54350567e-01 -5.94934464e-01 -8.79688323e-01
-1.10205673e-01 5.65891147e-01 3.84374201e-01 -1.12198487e-01
-1.00910807e+00 4.22499537e-01 6.31531954e+00 5.66780508e-01
-1.08300805e+00 3.98379862e-02 8.35480094e-01 2.82924414e-01
1.21991271e-02 -3.76201004e-01 1.67006969e-01 6.17136538e-01
1.07559323e+00 5.70083082e-01 7.46999234e-02 1.02308774e+00
-1.34172454e-01 -2.18319416e-01 -1.14425135e+00 9.41880524e-01
5.67743927e-02 -9.49644685e-01 1.62877247e-01 2.04396605e-01
8.12858164e-01 -1.74227744e-01 6.82245344e-02 9.86198559e-02
7.56485164e-02 -1.22553110e+00 8.53999630e-02 3.52012575e-01
4.74441260e-01 -9.81151521e-01 1.38018465e+00 2.73642212e-01
-4.20230925e-01 -1.60438567e-01 -2.20081687e-01 -6.19206689e-02
-8.55478197e-02 7.16207385e-01 -7.17460334e-01 5.79654872e-01
6.79616153e-01 3.86959314e-01 -5.86792231e-01 9.05460060e-01
-1.13179505e-01 7.78904200e-01 -3.00079912e-01 5.91142058e-01
3.85753691e-01 -2.33176664e-01 6.66199088e-01 9.49590981e-01
6.17481411e-01 -2.81308711e-01 -1.91356555e-01 7.38816738e-01
1.21266469e-01 -1.38681516e-01 -1.12157512e+00 -1.38292685e-01
-1.34938836e-01 1.16584146e+00 -9.26859736e-01 -1.55407652e-01
-4.51254547e-01 1.26028049e+00 -3.09052408e-01 2.16071531e-01
-6.59471631e-01 -4.30927426e-01 3.72209191e-01 1.21592142e-01
4.56094742e-01 1.54833253e-02 -9.30001438e-02 -9.01819706e-01
9.51085519e-03 -1.19486725e+00 5.59611499e-01 -7.53520191e-01
-1.50221312e+00 7.59156287e-01 -1.88481987e-01 -1.35530007e+00
-7.84572721e-01 -9.90767002e-01 -1.03280067e+00 5.42028606e-01
-1.05767918e+00 -8.20401490e-01 -3.14863861e-01 7.04455495e-01
6.04540169e-01 -6.22684300e-01 9.33402658e-01 2.85375595e-01
-5.10492086e-01 4.61919039e-01 5.24611548e-02 4.49645370e-01
3.67030442e-01 -1.45628786e+00 6.45857751e-01 1.22649610e+00
4.67658162e-01 9.56119224e-02 7.86841154e-01 -5.86409092e-01
-9.42607880e-01 -1.15123749e+00 2.13663802e-01 -4.99459594e-01
3.93500537e-01 -2.61276871e-01 -1.29410100e+00 7.17754364e-01
2.67665774e-01 1.50244623e-01 5.78465879e-01 -3.13749760e-01
9.18250531e-02 5.75864948e-02 -1.60604012e+00 3.66674036e-01
7.68513203e-01 -4.99439865e-01 -4.91173059e-01 3.00568253e-01
3.72865170e-01 -4.04008627e-01 -7.89127409e-01 4.02075678e-01
8.14533532e-02 -1.29080200e+00 7.08029449e-01 -6.06936574e-01
6.22188509e-01 -3.00331354e-01 6.93158209e-02 -1.52577806e+00
-8.03207532e-02 -3.83629858e-01 -2.84339100e-01 1.18287110e+00
4.96973306e-01 -9.25205767e-01 8.96168828e-01 3.33547056e-01
-1.98146805e-01 -7.18210757e-01 -6.78922653e-01 -8.85842323e-01
-1.58050433e-01 -6.75282598e-01 5.76864719e-01 1.05710638e+00
-5.78611076e-01 -7.07193860e-04 -3.28539878e-01 2.19118595e-01
2.68155783e-01 -2.79943228e-01 7.95202553e-01 -1.22520840e+00
-3.36492956e-01 -2.29147807e-01 -1.12646914e+00 -1.44609854e-01
-1.00747503e-01 -3.71989518e-01 4.81111631e-02 -1.10570908e+00
-4.16381210e-01 -1.33744448e-01 -4.22626257e-01 4.51110184e-01
1.62459657e-01 6.37007415e-01 -3.29119265e-01 3.57546695e-02
-4.70924787e-02 3.76205705e-02 8.78427923e-01 -2.65237242e-01
-1.60290375e-01 2.63964474e-01 -3.84288043e-01 9.70925987e-01
1.00234556e+00 -4.46135402e-01 -4.04760092e-01 -5.43826930e-02
2.08773866e-01 -5.71232498e-01 4.74310040e-01 -1.39148569e+00
-1.13003992e-01 3.98285300e-01 7.32303739e-01 -4.51496214e-01
-1.10979483e-01 -1.20137107e+00 1.61225751e-01 5.81532121e-01
5.88863902e-02 5.40740550e-01 3.10109168e-01 5.64716280e-01
-5.06419182e-01 -5.02458692e-01 6.36408269e-01 -3.06513965e-01
-9.37023997e-01 1.45310223e-01 -6.41906381e-01 -1.40996501e-01
1.01371706e+00 -5.73055923e-01 4.31865044e-02 -6.66398346e-01
-8.96562994e-01 -1.38448402e-01 5.99148512e-01 2.55664140e-01
7.14030921e-01 -1.27820742e+00 -3.01325828e-01 7.14290619e-01
1.83720887e-01 3.80558789e-01 2.29068294e-01 7.23025501e-01
-7.24410772e-01 -6.57623336e-02 -5.26167512e-01 -8.25338185e-01
-9.25913632e-01 7.55944490e-01 2.65764177e-01 -3.27221930e-01
-9.26876247e-01 3.95966917e-01 -7.30106235e-02 -3.59728575e-01
5.49617335e-02 -3.68925422e-01 -8.14717263e-02 -3.04239810e-01
3.07949454e-01 4.01962340e-01 5.99641740e-01 -5.09486735e-01
6.25892207e-02 4.08951044e-01 -1.32359207e-01 2.13982403e-01
1.16407657e+00 2.25765392e-01 1.52329393e-02 5.89107394e-01
9.02190387e-01 -1.77393496e-01 -9.20228541e-01 6.84623420e-02
2.32778221e-01 -3.79536927e-01 -3.19524892e-02 -8.02013278e-01
-1.40987813e+00 8.26349974e-01 9.58049476e-01 6.65312350e-01
1.60825670e+00 -1.57751620e-01 6.33435071e-01 6.60380781e-01
1.06549077e-01 -1.07204044e+00 3.22875261e-01 3.78189832e-01
6.16784692e-01 -1.48352361e+00 -1.18926518e-01 -5.64212166e-02
-6.99411750e-01 1.11893225e+00 7.38376498e-01 -2.60320067e-01
4.03547853e-01 2.73824006e-01 2.33591393e-01 -4.63471174e-01
-2.02374667e-01 -1.15197837e-01 1.73587516e-01 9.17369843e-01
4.00302000e-02 -3.18809569e-01 3.73447508e-01 3.39648724e-02
-1.65408343e-01 -2.12307781e-01 6.01337612e-01 1.21833408e+00
5.41420840e-02 -1.10749626e+00 -4.26661819e-01 9.41620886e-01
-8.04801702e-01 1.01510584e-01 -4.49593544e-01 1.22864044e+00
3.69641840e-01 6.92316890e-01 6.16449952e-01 -3.49066406e-01
5.01952887e-01 3.96517068e-01 2.90561587e-01 -3.71828139e-01
-5.95555425e-01 -2.28778735e-01 3.62395719e-02 -6.66224241e-01
-4.93456602e-01 -5.01250386e-01 -1.02955747e+00 -1.94128826e-01
-3.70601445e-01 2.69554835e-02 6.95792854e-01 1.19888902e+00
2.29222536e-01 7.05471933e-01 6.83988929e-01 -7.11516619e-01
-2.55053043e-01 -1.05619311e+00 -6.91222310e-01 8.89090776e-01
3.73483837e-01 -5.33956051e-01 -5.78069091e-01 -8.46419763e-03] | [7.489923000335693, 2.161741256713867] |
06e77897-6f61-471a-9e4f-cecf1a14163c | cat-controllable-attribute-translation-for | 2209.06850 | null | https://arxiv.org/abs/2209.06850v1 | https://arxiv.org/pdf/2209.06850v1.pdf | CAT: Controllable Attribute Translation for Fair Facial Attribute Classification | As the social impact of visual recognition has been under scrutiny, several protected-attribute balanced datasets emerged to address dataset bias in imbalanced datasets. However, in facial attribute classification, dataset bias stems from both protected attribute level and facial attribute level, which makes it challenging to construct a multi-attribute-level balanced real dataset. To bridge the gap, we propose an effective pipeline to generate high-quality and sufficient facial images with desired facial attributes and supplement the original dataset to be a balanced dataset at both levels, which theoretically satisfies several fairness criteria. The effectiveness of our method is verified on sex classification and facial attribute classification by yielding comparable task performance as the original dataset and further improving fairness in a comprehensive fairness evaluation with a wide range of metrics. Furthermore, our method outperforms both resampling and balanced dataset construction to address dataset bias, and debiasing models to address task bias. | ['Wael Abd-Almageed', 'Jiazhi Li'] | 2022-09-14 | null | null | null | null | ['facial-attribute-classification'] | ['computer-vision'] | [ 3.85206699e-01 2.16482818e-01 -5.32033443e-01 -9.48523998e-01
-5.45603156e-01 -4.05575424e-01 4.30219233e-01 2.24049419e-01
-2.59494424e-01 8.81521523e-01 3.26407641e-01 6.28200546e-02
-8.52244422e-02 -8.13782573e-01 -3.28335702e-01 -4.87503737e-01
4.61590618e-01 9.89119932e-02 -5.13481379e-01 1.04818903e-01
3.69399041e-01 4.53244179e-01 -1.86169040e+00 2.97969431e-01
1.14148784e+00 1.09949315e+00 -7.99293578e-01 4.44481857e-02
7.59294210e-03 5.94968319e-01 -6.76150501e-01 -1.28140092e+00
3.96409929e-01 -4.63452876e-01 -7.25428462e-01 -7.26306066e-02
1.09698331e+00 -6.70511127e-01 4.55893241e-02 1.17168057e+00
1.03868103e+00 -4.18671936e-01 5.53495526e-01 -1.86366570e+00
-8.22704792e-01 4.51638162e-01 -9.63292003e-01 -2.22227111e-01
2.94739515e-01 3.02426577e-01 9.23229694e-01 -6.98584676e-01
7.38917828e-01 1.51457882e+00 5.82639813e-01 6.74120426e-01
-1.28469586e+00 -1.60461628e+00 4.79467995e-02 -6.02920987e-02
-1.25956619e+00 -7.36220181e-01 7.15179980e-01 -3.65408510e-01
7.57861808e-02 2.72968024e-01 6.72872365e-01 1.27698088e+00
9.90484431e-02 5.63127518e-01 1.58760464e+00 -7.11784810e-02
7.12760985e-02 9.06938165e-02 1.21225379e-01 2.71074027e-01
7.47364819e-01 3.03859383e-01 -9.17020798e-01 -5.24876297e-01
3.29827875e-01 -1.58549190e-01 1.39172167e-01 -2.45066151e-01
-1.08840394e+00 8.59497786e-01 2.41213992e-01 -4.23573047e-01
-3.82174343e-01 -8.76637548e-02 6.92095935e-01 2.57668614e-01
6.79292142e-01 2.46093959e-01 -1.82253435e-01 1.21277742e-01
-9.53396201e-01 6.56767786e-01 3.51006210e-01 8.21999252e-01
6.69682860e-01 1.87880006e-02 -6.47535682e-01 9.29832041e-01
2.95041110e-02 8.82471144e-01 7.57717341e-02 -1.32398462e+00
2.39446864e-01 8.19286883e-01 -1.35710165e-01 -1.32906044e+00
-3.03136736e-01 -3.89069021e-01 -1.00758803e+00 4.86055911e-01
5.51502407e-01 2.66641080e-01 -7.02730596e-01 2.28954911e+00
6.03418648e-01 -4.09121484e-01 -2.32772484e-01 9.24199998e-01
9.04074728e-01 -1.17990293e-01 6.48429096e-01 -2.58128762e-01
1.62133217e+00 -1.90676585e-01 -9.21250343e-01 3.56540605e-02
4.04999733e-01 -6.82770610e-01 1.16503596e+00 3.62225950e-01
-9.98885155e-01 -4.57139492e-01 -9.17625487e-01 -1.70301497e-01
-4.31137867e-02 2.94534471e-02 7.60615349e-01 1.31834388e+00
-6.35481477e-01 2.07272246e-01 2.14469463e-01 -1.66785061e-01
1.45342815e+00 2.36241654e-01 -7.76571989e-01 -1.61500588e-01
-1.19378328e+00 6.88520074e-01 -8.03107172e-02 -2.45467573e-01
-8.43366206e-01 -1.22506392e+00 -7.36173272e-01 -1.48306191e-01
6.57504424e-02 -6.68338954e-01 6.39105916e-01 -1.30889571e+00
-9.05361950e-01 1.57984746e+00 -1.82909891e-01 -1.28646210e-01
8.57273936e-01 1.29014663e-02 -4.41291600e-01 -4.59753573e-02
2.38582432e-01 1.05350578e+00 9.37099218e-01 -1.32152700e+00
-5.57691813e-01 -9.30236638e-01 -1.82845190e-01 5.65383434e-02
-4.92093533e-01 2.86447614e-01 4.04217392e-01 -7.61161506e-01
-1.88709781e-01 -7.33250856e-01 9.59580839e-02 3.82518530e-01
-2.79038996e-01 -4.23287600e-02 6.94596648e-01 -6.75584972e-01
1.09004259e+00 -2.15503216e+00 -4.57618684e-01 1.27409160e-01
5.42058408e-01 2.10186020e-01 -2.59990066e-01 -2.14716017e-01
-3.45327049e-01 4.57247555e-01 -1.07254483e-01 -3.00662905e-01
7.93557167e-02 -1.47077277e-01 -3.60313743e-01 6.79389417e-01
3.49173993e-01 8.76975536e-01 -6.71063960e-01 -8.66989911e-01
-3.08659971e-01 2.62983710e-01 -7.73247480e-01 8.20872560e-02
4.60509837e-01 3.61506939e-01 -1.17702872e-01 1.16090584e+00
1.49035478e+00 3.30277950e-01 4.82651517e-02 -1.74161211e-01
3.66303444e-01 -2.06643306e-02 -9.42160249e-01 1.26671314e+00
-1.88347846e-02 3.63882691e-01 1.38044626e-01 -4.60682839e-01
1.34624445e+00 -2.36715470e-02 4.84663844e-01 -1.12904096e+00
1.41698912e-01 2.25772709e-01 -5.05516194e-02 -2.20295250e-01
6.66572094e-01 -5.20333052e-01 -2.67520577e-01 6.32513881e-01
-2.15455309e-01 -6.16607890e-02 -2.32092351e-01 2.29427084e-01
3.84272099e-01 6.67331368e-02 2.73751467e-01 -3.94964695e-01
3.97761285e-01 -1.57723546e-01 1.09157741e+00 5.82691252e-01
-9.11587656e-01 7.42874861e-01 9.99374330e-01 -5.68304360e-01
-1.14007747e+00 -9.18644547e-01 -5.25965333e-01 1.09789848e+00
4.54814062e-02 -1.68838859e-01 -9.58181441e-01 -9.57996666e-01
5.44735193e-01 4.66153800e-01 -9.96489882e-01 -4.68546122e-01
-3.70526081e-03 -1.01895809e+00 1.16102326e+00 3.10423434e-01
5.93940139e-01 -9.28110242e-01 -3.92205089e-01 -4.29154426e-01
-4.49107885e-01 -9.17194963e-01 -5.02172172e-01 -5.59306264e-01
-5.00361979e-01 -1.36293209e+00 -5.55404127e-01 -1.29723608e-01
7.32321024e-01 5.63599467e-02 1.32499099e+00 2.04435661e-01
-4.38392758e-01 -1.20183311e-01 -8.77972916e-02 -9.03351426e-01
-3.32702667e-01 -8.21702927e-02 2.67064840e-01 3.74606758e-01
6.37022495e-01 -3.14793915e-01 -7.35680521e-01 5.65100074e-01
-9.98703241e-01 -1.12863906e-01 3.97061288e-01 7.80452728e-01
3.47838014e-01 -1.89430624e-01 1.11301088e+00 -9.34649467e-01
6.91750884e-01 -2.94984132e-01 -2.70626187e-01 1.35545984e-01
-1.07973289e+00 -4.53792423e-01 4.60033864e-02 -3.66841882e-01
-1.08061194e+00 -2.77645439e-01 1.29022688e-01 -7.20316991e-02
-8.22869018e-02 -9.63773206e-02 -6.40729010e-01 -4.42482568e-02
7.01048017e-01 -2.01554462e-01 7.67472625e-01 -9.20860916e-02
1.72231495e-01 9.52127934e-01 3.17944765e-01 -7.78912723e-01
5.88334680e-01 7.10596144e-01 1.47721499e-01 -1.61640242e-01
-9.35010493e-01 2.40199134e-01 -5.82355320e-01 -4.20164376e-01
5.34013212e-01 -1.04633975e+00 -1.07290494e+00 8.79584193e-01
-8.53395700e-01 1.77246720e-01 -3.22882444e-01 2.73897409e-01
-3.11118364e-01 2.50717849e-01 -1.79382712e-01 -8.82421792e-01
-4.55972612e-01 -1.12653387e+00 1.07188809e+00 9.31665674e-02
-5.44739962e-01 -2.29541332e-01 -1.85942680e-01 8.24449956e-01
4.63884085e-01 6.27925873e-01 8.55625331e-01 -3.83475959e-01
-1.49264738e-01 -2.42434487e-01 -7.50545025e-01 3.30332547e-01
2.20654130e-01 2.15663925e-01 -1.29117048e+00 -3.42384845e-01
-2.01676324e-01 -8.32521319e-01 6.32249713e-01 2.96622574e-01
1.40048635e+00 -1.79499745e-01 2.32933369e-02 6.46448076e-01
1.12970769e+00 -2.04585359e-01 9.74400580e-01 2.39217296e-01
4.69292819e-01 1.31225550e+00 8.35211873e-01 6.98526263e-01
5.71561873e-01 5.73704720e-01 5.39224327e-01 -3.23223650e-01
-2.18028218e-01 -3.53473008e-01 -1.35394617e-03 -2.03003407e-01
4.95451391e-02 2.23263025e-01 -7.70230234e-01 4.72702205e-01
-1.43502355e+00 -1.12622726e+00 -2.16086566e-01 2.36157060e+00
9.64955449e-01 -2.40371689e-01 4.54895437e-01 2.37243608e-01
1.00320137e+00 1.90318614e-01 -7.75570214e-01 -5.36994696e-01
-5.67196250e-01 -1.57570451e-01 3.18966448e-01 8.79857913e-02
-8.18228364e-01 6.80945516e-01 7.38401604e+00 8.38172019e-01
-9.63618219e-01 9.30990744e-03 1.33477199e+00 -4.20885414e-01
-6.49296999e-01 -1.05588749e-01 -7.06953645e-01 6.69628441e-01
5.22819281e-01 -4.56073225e-01 7.17131868e-02 7.21874356e-01
2.61342674e-01 -6.23165481e-02 -7.72482991e-01 1.07991898e+00
1.37417227e-01 -9.54305291e-01 4.64845270e-01 5.37595272e-01
5.44797122e-01 -6.94196880e-01 4.59880322e-01 6.16872311e-02
1.98744088e-01 -1.47764802e+00 8.15777600e-01 2.84600258e-01
1.39024615e+00 -1.10001028e+00 8.18322301e-01 -1.69296920e-01
-5.49901068e-01 -1.25755683e-01 -3.35096627e-01 -1.81642175e-01
-2.65337080e-01 9.64781880e-01 -3.46831858e-01 5.46184719e-01
8.22897792e-01 5.26456952e-01 -8.24489355e-01 5.75798094e-01
1.04220867e-01 2.94497073e-01 1.26441240e-01 4.09235626e-01
-3.59604269e-01 -6.09878376e-02 1.08580165e-01 8.31825316e-01
8.39369744e-02 -5.68399318e-02 -2.75864810e-01 8.92000258e-01
-4.26432759e-01 2.58783460e-01 -5.38653195e-01 1.18595324e-01
6.64211333e-01 1.34198070e+00 -2.63337195e-01 -1.34748876e-01
-1.51986063e-01 4.95373994e-01 2.98911840e-01 1.93051361e-02
-9.82422829e-01 -2.04619139e-01 1.21969938e+00 1.99289113e-01
-3.94626677e-01 4.69971865e-01 -8.50326359e-01 -9.68286276e-01
1.31585330e-01 -1.38826907e+00 6.62982047e-01 -5.29965043e-01
-1.43190932e+00 3.43435287e-01 -1.56987101e-01 -1.33717644e+00
3.75636905e-01 -1.27000928e-01 -3.25048268e-02 1.01014185e+00
-1.67418087e+00 -1.59040201e+00 -7.69795656e-01 5.69670737e-01
-3.03256184e-01 -2.24042386e-01 6.67013645e-01 4.82739002e-01
-5.81324100e-01 1.24378765e+00 -5.88163435e-01 7.23434985e-02
1.43754065e+00 -7.72787809e-01 1.03498898e-01 5.38248718e-01
-6.14780307e-01 6.48871839e-01 3.51274580e-01 -7.34212220e-01
-9.03660834e-01 -1.02125025e+00 9.08723712e-01 -6.12710714e-01
-4.62250412e-02 -3.99403453e-01 -7.97633648e-01 3.78900737e-01
-1.31787270e-01 1.69016987e-01 1.19042373e+00 1.50453225e-01
-1.02967799e+00 -5.82794428e-01 -1.85534286e+00 3.26300293e-01
1.36569870e+00 -5.06428182e-01 -2.14417666e-01 -1.06905587e-01
3.10227662e-01 -2.88860202e-01 -1.03628588e+00 5.53375959e-01
1.06991184e+00 -1.21858728e+00 8.93061638e-01 -9.04446363e-01
9.11866248e-01 -1.19026616e-01 -1.48195252e-01 -9.98968244e-01
-4.02635902e-01 -4.20366108e-01 4.29711640e-01 1.81794417e+00
2.33077317e-01 -7.91206300e-01 9.26295161e-01 9.90164399e-01
5.21859467e-01 -5.21567225e-01 -9.55826819e-01 -5.60248315e-01
2.79641956e-01 -2.34723240e-01 1.43133461e+00 1.24426818e+00
-2.34867826e-01 -1.92861050e-01 -6.94955528e-01 -1.45432860e-01
1.10734832e+00 2.90505469e-01 1.03207707e+00 -1.32374394e+00
6.87505543e-01 -6.15750849e-01 -5.25790215e-01 -7.45453835e-02
6.39210105e-01 -9.57832634e-01 -5.17423868e-01 -9.24040437e-01
8.45967114e-01 -7.93583691e-01 -1.90137565e-01 5.93129516e-01
-5.85579753e-01 7.96086550e-01 1.20451607e-01 2.08962515e-01
-2.03538775e-01 6.15138531e-01 1.31032777e+00 -1.92624941e-01
2.87074327e-01 -2.71668166e-01 -1.61000514e+00 2.71934301e-01
5.47556996e-01 -6.45568848e-01 -2.28268683e-01 -1.76029608e-01
3.46616238e-01 -2.88036138e-01 5.05092025e-01 -5.72918415e-01
-2.77666241e-01 -5.14705837e-01 7.68113017e-01 -3.51845831e-01
2.10256204e-02 -6.77101016e-01 2.12677598e-01 4.29433346e-01
-5.97246408e-01 -1.23971231e-01 9.27584693e-02 2.79516965e-01
-1.62325725e-01 3.37357789e-01 1.18439853e+00 1.51142418e-01
-2.76029259e-01 7.19775259e-01 2.88389206e-01 2.50934362e-01
1.14122677e+00 -4.96497929e-01 -7.91384757e-01 -3.27090621e-01
-1.85577720e-01 3.10627818e-01 9.15694654e-01 5.31317353e-01
4.42773074e-01 -1.71150386e+00 -1.29145980e+00 4.42131013e-01
6.07876718e-01 -4.23000127e-01 4.39570874e-01 8.14748645e-01
-1.58328339e-01 -3.40215228e-02 -9.46910560e-01 -4.30500865e-01
-1.52732480e+00 5.35796940e-01 2.45356113e-01 1.54634178e-01
4.65735868e-02 6.14674389e-01 3.43002886e-01 -7.39536166e-01
5.11314906e-02 3.07563126e-01 -1.18943512e-01 6.36759937e-01
8.41590822e-01 6.33509815e-01 8.12036321e-02 -8.52125525e-01
-5.64946830e-01 4.42278504e-01 1.23283558e-01 1.53106868e-01
1.09182632e+00 -2.97395229e-01 -4.77433443e-01 -1.70911267e-01
1.04408264e+00 2.94152141e-01 -1.16314280e+00 3.99977863e-02
-3.71059448e-01 -1.23742139e+00 -1.10140227e-01 -8.95640075e-01
-1.21433759e+00 5.89701176e-01 6.46853328e-01 -8.15341994e-02
1.30685890e+00 -4.78740633e-01 5.09649992e-01 -4.21351194e-01
4.15757507e-01 -1.09287298e+00 -1.03636511e-01 -1.35079101e-01
8.04854214e-01 -1.53662145e+00 4.44603771e-01 -5.98013341e-01
-9.46713388e-01 6.88748002e-01 7.91489065e-01 2.87469089e-01
3.74194860e-01 1.85572907e-01 3.76819104e-01 -1.68646231e-01
-5.31679571e-01 1.22239411e-01 1.67857274e-01 9.08541083e-01
4.52150822e-01 9.18437839e-02 -7.45332003e-01 6.22216880e-01
-4.89930153e-01 3.28893721e-01 3.88052702e-01 4.85283703e-01
1.12219051e-01 -1.18090820e+00 -5.27239919e-01 6.84265971e-01
-7.65647054e-01 1.31253034e-01 -6.62285864e-01 6.80258274e-01
3.02635401e-01 8.32834721e-01 9.48046893e-02 -4.13916737e-01
3.55901510e-01 1.87337045e-02 3.45648468e-01 -1.08635746e-01
-7.39170551e-01 -6.56938553e-01 2.14039952e-01 -8.23198020e-01
-5.32611847e-01 -7.12208629e-01 -6.95487678e-01 -9.89184022e-01
-1.47358745e-01 -2.10491285e-01 4.88598496e-01 5.22347331e-01
6.80321276e-01 -1.72022674e-02 9.33390141e-01 -3.53079855e-01
-6.01848185e-01 -7.30581582e-01 -7.36233473e-01 1.06794846e+00
2.36909598e-01 -7.15167522e-01 -3.28311443e-01 -1.88896000e-01] | [13.02203369140625, 1.262813687324524] |
c20d145f-a70b-45b6-a94a-cdadc357bc76 | characterizing-political-bias-in-automatic | 2305.02321 | null | https://arxiv.org/abs/2305.02321v1 | https://arxiv.org/pdf/2305.02321v1.pdf | Characterizing Political Bias in Automatic Summaries: A Case Study of Trump and Biden | Growing literature has shown that powerful NLP systems may encode social biases; however, the political bias of summarization models remains relatively unknown. In this work, we use an entity replacement method to investigate the portrayal of politicians in automatically generated summaries of news articles. We develop a computational framework based on political entities and lexical resources, and use it to assess biases about Donald Trump and Joe Biden in both extractive and abstractive summarization models. We find consistent differences, such as stronger associations of a collective US government (i.e., administration) with Biden than with Trump. These summary dissimilarities are most prominent when the entity is heavily featured in the source article. Our systematic characterization provides a framework for future studies of bias in summarization. | ['Chenhao Tan', 'Karen Zhou'] | 2023-05-03 | null | null | null | null | ['abstractive-text-summarization'] | ['natural-language-processing'] | [ 5.36999181e-02 4.84578133e-01 -8.91340613e-01 -2.48176962e-01
-7.28260338e-01 -8.20800126e-01 1.36062455e+00 7.23139822e-01
-7.03596413e-01 1.02084923e+00 1.62881887e+00 -4.05477524e-01
2.71720588e-02 -8.54514718e-01 -7.21642733e-01 -2.99913943e-01
3.15396011e-01 2.33264878e-01 -3.32858026e-01 -3.07534128e-01
7.85043836e-01 -4.81425598e-03 -8.16608906e-01 3.88825506e-01
1.15768564e+00 1.58463806e-01 -3.94172400e-01 3.71329576e-01
-1.00615904e-01 1.02242148e+00 -9.95605290e-01 -8.36121738e-01
-3.05125210e-02 -4.90792215e-01 -7.12028623e-01 -3.47350121e-01
1.00250089e+00 -1.47960216e-01 -6.35989368e-01 1.07671976e+00
8.94223213e-01 -2.00201079e-01 1.10070896e+00 -7.21252680e-01
-9.14575398e-01 1.50479102e+00 -8.19549739e-01 9.57417786e-01
5.20328820e-01 1.18571535e-01 1.59685898e+00 -6.79894805e-01
1.31235027e+00 1.45701826e+00 7.66010106e-01 1.64845049e-01
-1.40968454e+00 -7.19623089e-01 2.64769763e-01 -1.71613500e-01
-8.70897293e-01 -7.96085835e-01 6.69707477e-01 -5.68925440e-01
6.72462821e-01 4.28535163e-01 7.90408492e-01 1.55342770e+00
4.85838503e-01 6.52711093e-01 1.26271343e+00 -1.65192083e-01
-1.42039776e-01 3.75414044e-02 6.72696769e-01 3.86590809e-01
1.29437387e+00 -4.84386444e-01 -6.62869096e-01 -8.26120496e-01
1.85058981e-01 -4.57995266e-01 -2.82579333e-01 4.27751541e-01
-1.36928713e+00 1.20988095e+00 3.14742684e-01 4.43709463e-01
-8.14904988e-01 2.28649661e-01 6.31176174e-01 3.24681103e-02
1.16726410e+00 9.70702171e-01 -2.54799068e-01 -5.59039824e-02
-1.30351484e+00 6.35825753e-01 1.17865694e+00 5.80350101e-01
3.14032793e-01 -5.78933470e-02 -6.45789504e-01 7.40919232e-01
-1.32123053e-01 6.51255012e-01 4.56716239e-01 -9.19004440e-01
7.14634836e-01 3.75050604e-01 2.65062332e-01 -1.89987159e+00
-4.49084580e-01 -5.18458247e-01 -6.95955396e-01 -6.48455918e-01
1.13904096e-01 -5.12041450e-01 -3.27095985e-01 1.70115089e+00
8.01582411e-02 -3.13265622e-01 1.69397950e-01 6.23127401e-01
1.27878416e+00 7.46303439e-01 3.41730297e-01 -4.84786034e-01
1.68034685e+00 -5.48861384e-01 -7.80489326e-01 -4.96144861e-01
6.05927885e-01 -7.15503573e-01 6.00770056e-01 -2.59239346e-01
-1.01644802e+00 2.07865499e-02 -8.55723083e-01 -1.51295334e-01
-1.06672525e-01 8.15650374e-02 5.48140168e-01 2.95341462e-01
-6.34205639e-01 5.62820077e-01 -2.44862169e-01 -6.79188073e-01
5.72084546e-01 -3.68448496e-01 -1.79103300e-01 2.54854679e-01
-1.33367026e+00 1.25007558e+00 2.37555310e-01 -3.34924281e-01
-1.08457938e-01 -6.29631877e-01 -8.60335171e-01 -2.51373321e-01
3.92761350e-01 -1.04262793e+00 1.02113378e+00 -8.69951129e-01
-7.32841194e-01 1.12836242e+00 -3.97621244e-01 -6.30071998e-01
4.45154190e-01 -3.22002977e-01 -6.29090965e-01 7.77213201e-02
8.00514042e-01 1.80640370e-01 4.78141963e-01 -1.01525605e+00
-5.57307601e-01 -3.57590914e-01 4.91359308e-02 3.10500801e-01
-1.55449003e-01 5.46816945e-01 3.83121938e-01 -9.12228763e-01
5.14693968e-02 -6.81227565e-01 -1.13493703e-01 -8.35499585e-01
-8.09433341e-01 -5.09692311e-01 1.09768070e-01 -9.62351203e-01
1.70723343e+00 -1.76027751e+00 4.34877500e-02 -5.69551531e-03
6.33338094e-01 -2.42915362e-01 2.37439826e-01 7.89457798e-01
1.16921820e-01 8.90649617e-01 -5.18633239e-02 1.26422480e-01
1.35642916e-01 7.84584209e-02 -7.14348435e-01 7.47884393e-01
1.45979211e-01 9.93563473e-01 -1.02828848e+00 -7.32730389e-01
-6.22278690e-01 -8.33778158e-02 -6.59470201e-01 -6.97175026e-01
7.53787160e-02 -2.64578369e-02 -5.13769448e-01 5.74684381e-01
4.00192738e-01 -1.39947012e-01 1.98949695e-01 -4.90565211e-01
-8.67942631e-01 1.02707863e+00 -4.18870538e-01 1.05586267e+00
1.50216296e-01 1.22832227e+00 2.71375299e-01 -5.54390073e-01
5.39899468e-01 8.88826028e-02 -1.71392690e-02 -3.88157070e-01
1.87847674e-01 2.22566634e-01 1.73544332e-01 -3.80637407e-01
1.29218125e+00 -2.23804191e-01 -7.30431199e-01 5.41662514e-01
-4.01107132e-01 -4.16872114e-01 4.21112835e-01 7.69521594e-01
9.65982795e-01 -4.69885349e-01 7.55627275e-01 -7.06033111e-01
-1.39587045e-01 5.36986709e-01 8.75164151e-01 1.05062211e+00
-9.77442367e-04 3.27901512e-01 9.21886623e-01 -3.13495189e-01
-1.30130804e+00 -3.60028327e-01 -4.07558829e-01 8.69492531e-01
-1.33303881e-01 -6.55402064e-01 -4.43905890e-01 -6.17109120e-01
3.54228020e-01 1.40704036e+00 -7.85079777e-01 6.98848143e-02
-7.13376224e-01 -1.02795744e+00 7.86146820e-01 1.35726975e-02
2.68800110e-01 -7.15025604e-01 -6.53564453e-01 9.16537642e-02
-5.53661227e-01 -1.02249181e+00 -3.38313341e-01 -3.80076975e-01
-5.64023077e-01 -8.45761418e-01 -5.72459996e-01 -2.35823199e-01
4.17079270e-01 3.29822674e-02 1.29186559e+00 -3.57510522e-02
2.82334566e-01 4.31663811e-01 -1.53856754e-01 -8.72551322e-01
-7.53757119e-01 4.19769585e-01 2.06915647e-01 -4.48753357e-01
2.42554218e-01 -2.80550867e-01 -4.21298206e-01 -3.09928864e-01
-7.07657933e-01 7.59388357e-02 4.36503172e-01 6.88385367e-01
-4.38248273e-03 -3.85736704e-01 8.48686337e-01 -1.35199201e+00
1.29615748e+00 -9.68941867e-01 -2.58688331e-02 1.13321720e-02
-7.30847239e-01 -2.61377450e-02 1.97261214e-01 -1.13109350e-01
-9.81359482e-01 -1.05383778e+00 2.00802878e-01 7.25910902e-01
2.96479166e-01 1.12872100e+00 5.48173010e-01 4.48433876e-01
9.54685211e-01 -1.98441610e-01 -1.06203295e-01 -2.02220529e-01
3.55034262e-01 8.27733636e-01 3.27365935e-01 -6.50884807e-01
6.73846066e-01 5.31597137e-01 -5.59543788e-01 -1.21645653e+00
-1.17411101e+00 -2.04876274e-01 -9.68323797e-02 -1.61967307e-01
7.98455834e-01 -1.23541927e+00 -2.13947862e-01 4.76221144e-02
-1.42651165e+00 1.33370638e-01 -3.21460336e-01 6.16559267e-01
-6.90149218e-02 4.67741787e-01 -6.58959627e-01 -4.81952369e-01
-5.56556344e-01 -5.23251891e-01 6.47990108e-01 1.64074212e-01
-1.04987884e+00 -7.40188718e-01 5.38442791e-01 2.75036752e-01
2.46520713e-01 7.16250837e-01 1.08699286e+00 -1.07333171e+00
9.81384367e-02 -1.58966705e-01 -2.78342903e-01 -1.86445564e-01
5.42856827e-02 6.63789809e-01 -4.03123856e-01 5.91344759e-02
-1.91926695e-02 -1.18458986e-01 1.12362707e+00 6.14013851e-01
2.92549133e-01 -1.08629858e+00 -5.04977345e-01 -6.95864633e-02
1.11532795e+00 -4.16050345e-01 3.31432879e-01 5.80121458e-01
4.03939426e-01 7.49988914e-01 3.48087847e-01 6.64758086e-01
6.78998470e-01 2.81924397e-01 -3.36334616e-01 1.79622889e-01
-2.32423451e-02 -5.26944160e-01 3.94864827e-01 9.53448117e-01
-1.90928817e-01 -4.27925348e-01 -1.05560184e+00 9.09272015e-01
-1.61117494e+00 -1.41971147e+00 -4.32225138e-01 1.44388855e+00
9.69422936e-01 1.06159523e-01 -8.38876739e-02 -3.73453677e-01
8.49838257e-01 1.01593459e+00 -1.04490593e-01 -4.27701324e-01
-7.39903569e-01 -1.83128029e-01 7.26054966e-01 4.06993449e-01
-8.41101766e-01 9.90910649e-01 6.94335318e+00 3.86590689e-01
-8.17051411e-01 -9.91229191e-02 6.91519260e-01 -2.21824095e-01
-7.90440738e-01 1.52047008e-01 -7.34827399e-01 4.94526714e-01
8.10332775e-01 -1.00583875e+00 -2.53506750e-01 6.20093405e-01
4.34453815e-01 -3.36695671e-01 -8.45101655e-01 3.60464990e-01
3.45270693e-01 -1.61786044e+00 1.06443241e-01 2.45105550e-01
1.00377727e+00 1.51193932e-01 -9.72871631e-02 1.82003245e-01
7.16167986e-01 -5.58690667e-01 1.17992115e+00 3.41873497e-01
6.17156804e-01 -4.33004081e-01 6.57015204e-01 3.49256694e-01
-2.28287309e-01 -1.60400905e-02 -5.60007870e-01 -3.22337210e-01
3.97544593e-01 9.27322209e-01 -8.21300507e-01 4.73524272e-01
4.33226824e-01 7.64075994e-01 -4.88141596e-01 4.91573989e-01
-4.50418562e-01 1.07192814e+00 -5.27720563e-02 -2.94908822e-01
3.66780370e-01 -1.03620157e-01 1.29142165e+00 1.52019465e+00
1.90943018e-01 5.28043568e-01 -2.39463136e-01 5.74639142e-01
-5.38094342e-01 4.43291962e-01 -1.00726342e+00 -6.33238137e-01
6.75576687e-01 1.22595954e+00 -6.86818242e-01 -7.24462867e-01
-2.72682846e-01 4.95955288e-01 4.34737206e-01 2.35538408e-01
-6.86072469e-01 7.87345693e-03 4.63459492e-01 2.91157901e-01
-9.41910297e-02 -1.71016365e-01 -5.02928495e-01 -1.37876201e+00
-3.10195863e-01 -1.03069603e+00 4.30434763e-01 -6.56837106e-01
-1.32706237e+00 3.24090302e-01 3.45515311e-01 -4.28433418e-01
3.90676931e-02 8.77484307e-03 -8.08201313e-01 6.73074484e-01
-1.00316930e+00 -6.37066185e-01 4.62876074e-02 -2.09745646e-01
3.07219803e-01 -3.07461079e-02 4.29833561e-01 -9.24276784e-02
-8.36638749e-01 1.94791853e-01 -2.19323151e-02 1.86362147e-01
1.07579255e+00 -1.03282118e+00 6.00756466e-01 8.19473267e-01
-4.90640244e-03 7.68477559e-01 1.27100396e+00 -1.20643783e+00
-9.34391975e-01 -7.47870564e-01 1.55900609e+00 -7.23683000e-01
9.92956460e-01 1.02942720e-01 -6.00170314e-01 8.20691943e-01
9.60377872e-01 -8.55030715e-01 8.70014191e-01 4.50553477e-01
-3.94208729e-01 5.33539891e-01 -1.05446589e+00 9.73591030e-01
1.13166022e+00 -3.15026462e-01 -1.35459292e+00 4.43186134e-01
7.82870710e-01 -6.37556463e-02 -6.92933083e-01 1.46739095e-01
5.34718692e-01 -5.28141797e-01 6.48614049e-01 -9.10999298e-01
1.06537879e+00 1.11619994e-01 -3.72165442e-03 -1.71052325e+00
-8.70648503e-01 -4.61158931e-01 3.51771355e-01 1.47593713e+00
8.62115383e-01 -9.59965348e-01 1.98774502e-01 7.24615157e-01
-1.26346990e-01 -2.33378738e-01 -7.53942192e-01 -1.54935777e-01
1.73458457e-01 1.71158582e-01 2.24993780e-01 1.62316442e+00
2.26633742e-01 7.37179577e-01 -2.46697754e-01 -1.04586780e-01
4.43352401e-01 1.92301914e-01 8.28510106e-01 -1.41589046e+00
2.55587935e-01 -6.45128071e-01 -4.00405750e-02 -5.97447276e-01
5.15097916e-01 -9.79701340e-01 -4.23090875e-01 -2.03686476e+00
6.40016437e-01 -3.59892622e-02 3.70763242e-01 7.22693652e-02
-3.23764980e-01 -1.29782826e-01 7.29239434e-02 4.84855592e-01
-3.56862128e-01 4.09156382e-01 1.04915321e+00 -1.96909040e-01
-1.42728552e-01 -4.62064564e-01 -1.58958936e+00 1.05462086e+00
8.40077519e-01 -7.42179692e-01 4.12570387e-01 -4.91241634e-01
8.86003673e-01 -2.48824403e-01 2.61731625e-01 -3.93638194e-01
2.03503460e-01 -2.13158727e-01 1.88063458e-01 -5.80322742e-01
-2.30752140e-01 -1.06303573e-01 8.85257572e-02 4.45424378e-01
-6.82240367e-01 4.42065150e-01 1.42021134e-01 6.12778425e-01
-1.17320336e-01 -3.13776672e-01 2.96061039e-01 -2.12732077e-01
-2.21740715e-02 -3.05615664e-01 -8.15771699e-01 6.26690030e-01
4.33172643e-01 6.55138195e-02 -9.91254628e-01 -6.45878971e-01
-1.62747294e-01 -2.41364427e-02 6.66553795e-01 1.02113329e-01
6.85884878e-02 -1.26203895e+00 -1.65684056e+00 -6.83844745e-01
2.39411201e-02 -5.42769969e-01 7.52364546e-02 1.15750539e+00
-5.02671123e-01 3.36917698e-01 4.88134511e-02 2.34694093e-01
-1.13419747e+00 2.21785516e-01 -2.97212511e-01 -2.21305162e-01
-5.31288624e-01 6.37364388e-01 9.19143409e-02 2.72513032e-02
-5.06721556e-01 -3.78213495e-01 -4.89744514e-01 1.07538760e+00
3.30604970e-01 7.22977996e-01 -5.79738021e-01 -1.05688846e+00
-2.96365559e-01 -8.91997479e-03 -1.80417284e-01 -2.51798332e-01
1.46137869e+00 -1.10854588e-01 -6.41935050e-01 6.57299340e-01
7.76786149e-01 9.84947383e-01 -2.77745306e-01 -2.77982652e-01
3.10165256e-01 -2.64407635e-01 9.55941007e-02 -6.19337857e-01
-3.98595691e-01 -8.48054588e-02 -5.39422154e-01 6.20974183e-01
1.94616392e-01 3.23477983e-01 3.73059750e-01 1.77642435e-01
8.99563134e-02 -1.33301938e+00 -5.61305642e-01 5.29941678e-01
1.44537902e+00 -1.03042269e+00 8.00914168e-01 -3.03638518e-01
-9.39595819e-01 5.80576420e-01 1.29335687e-01 -2.73848265e-01
1.78273737e-01 -1.62718624e-01 -2.24483877e-01 -7.89216995e-01
-8.40980649e-01 2.20788956e-01 3.19585711e-01 -1.94187015e-01
7.83745646e-01 3.61710340e-01 -1.63677955e+00 9.32935536e-01
-7.17303634e-01 -6.39789224e-01 1.11396062e+00 6.61281705e-01
-4.57642555e-01 -4.20232385e-01 -4.42361593e-01 9.33409214e-01
-1.08323359e+00 -6.34481549e-01 -1.25164151e+00 9.01197255e-01
-2.17768744e-01 7.58417726e-01 1.86807811e-01 -1.22903183e-01
3.19200844e-01 1.36689261e-01 1.42691314e-01 -6.85290515e-01
-9.78555799e-01 -2.16637086e-02 1.17254937e+00 -1.00272536e-01
-7.57690430e-01 -1.09337306e+00 -6.63096428e-01 -7.49784410e-01
-3.46627891e-01 3.85928363e-01 4.41759348e-01 5.76343238e-01
6.18461847e-01 1.99321821e-01 2.89505512e-01 -4.11363125e-01
-5.37484407e-01 -1.33070314e+00 -4.60375845e-01 2.63387620e-01
1.51771992e-01 -4.83848661e-01 -7.02989221e-01 -2.25548074e-01] | [8.870532989501953, 9.97246265411377] |
b392ff62-9c32-49f6-aa52-6edf914ae75d | retrieve-rerank-and-rewrite-soft-template | null | null | https://aclanthology.org/P18-1015 | https://aclanthology.org/P18-1015.pdf | Retrieve, Rerank and Rewrite: Soft Template Based Neural Summarization | Most previous seq2seq summarization systems purely depend on the source text to generate summaries, which tends to work unstably. Inspired by the traditional template-based summarization approaches, this paper proposes to use existing summaries as soft templates to guide the seq2seq model. To this end, we use a popular IR platform to Retrieve proper summaries as candidate templates. Then, we extend the seq2seq framework to jointly conduct template Reranking and template-aware summary generation (Rewriting). Experiments show that, in terms of informativeness, our model significantly outperforms the state-of-the-art methods, and even soft templates themselves demonstrate high competitiveness. In addition, the import of high-quality external summaries improves the stability and readability of generated summaries. | ['Furu Wei', 'Ziqiang Cao', 'Wenjie Li', 'Sujian Li'] | 2018-07-01 | null | null | null | acl-2018-7 | ['summarization', 'abstractive-sentence-summarization'] | ['natural-language-processing', 'natural-language-processing'] | [ 4.31811929e-01 2.57599771e-01 -4.15175796e-01 -1.79516122e-01
-1.26090193e+00 -7.55292356e-01 8.43543768e-01 2.82200426e-01
-1.36891589e-01 1.02422512e+00 1.08093190e+00 -3.16978805e-02
1.82608366e-02 -7.10200906e-01 -3.52523059e-01 -3.09432030e-01
4.71725076e-01 3.43422711e-01 3.19381207e-01 -5.15436411e-01
7.67319918e-01 -6.34007622e-03 -1.58207655e+00 3.81491810e-01
1.70268834e+00 4.29450095e-01 2.84664541e-01 6.79923892e-01
-2.93755442e-01 4.32120085e-01 -1.07071197e+00 -5.77544451e-01
-7.18694255e-02 -7.36977875e-01 -6.22180641e-01 -2.80871987e-01
4.28087592e-01 -3.42642963e-01 -4.11229193e-01 9.33718801e-01
7.62213707e-01 2.27498844e-01 6.61694169e-01 -7.72661567e-01
-6.32141829e-01 1.12213647e+00 -2.40393505e-01 2.85645574e-02
8.45768869e-01 1.09362893e-01 1.19643235e+00 -7.55472660e-01
7.90954173e-01 1.09449518e+00 2.48896644e-01 5.67875981e-01
-7.76376069e-01 -3.79336447e-01 2.40607619e-01 -5.96689545e-02
-1.14973533e+00 -8.44605863e-01 7.14291453e-01 1.08166948e-01
9.39833522e-01 8.56362045e-01 5.31037211e-01 1.14482772e+00
2.51424015e-01 1.03475094e+00 4.27754939e-01 -3.48195493e-01
1.77392825e-01 -2.84543872e-01 2.15273216e-01 2.82649189e-01
6.25935316e-01 -2.88122326e-01 -6.83819354e-01 -4.19061661e-01
1.51545525e-01 -2.06492171e-01 -4.34919089e-01 3.33215564e-01
-1.51112413e+00 5.37405789e-01 -1.95879444e-01 1.32396281e-01
-5.55142939e-01 -9.30153430e-02 7.54219055e-01 9.81256440e-02
5.27084947e-01 9.29140449e-01 -8.91864449e-02 -3.62727642e-01
-1.35530400e+00 5.37048459e-01 8.97869468e-01 1.31813228e+00
4.38077092e-01 -3.49811353e-02 -1.25113821e+00 7.75289893e-01
-1.20844930e-01 5.49087703e-01 7.12895572e-01 -9.82455730e-01
7.40651608e-01 7.30712175e-01 1.72913343e-01 -1.07055795e+00
-1.91484729e-03 -6.18619978e-01 -9.37895954e-01 -8.13170195e-01
-2.84656376e-01 -1.45680606e-01 -5.85817516e-01 1.33294654e+00
-4.17687185e-02 -1.72248393e-01 2.52602518e-01 5.97900510e-01
1.12964010e+00 8.72994423e-01 -3.72386903e-01 -6.97671294e-01
1.11800361e+00 -1.29376853e+00 -1.02024961e+00 -2.04430982e-01
5.00217259e-01 -8.42032552e-01 1.03777814e+00 3.58928025e-01
-1.35772562e+00 -4.23332423e-01 -1.07174528e+00 -1.05327658e-01
6.19181842e-02 3.44756365e-01 2.41848484e-01 4.01474983e-01
-9.77831423e-01 6.48620009e-01 -5.37326276e-01 -2.87617803e-01
1.49841741e-01 6.92099426e-03 -1.19167836e-02 -1.31715955e-02
-1.18751276e+00 6.22039914e-01 7.01861203e-01 -1.10892244e-01
-4.92711514e-01 -6.35126412e-01 -7.74434805e-01 3.08679074e-01
5.91953814e-01 -1.12650132e+00 1.46114480e+00 -2.97697514e-01
-1.73540020e+00 1.75464317e-01 -5.68486691e-01 -1.89493552e-01
5.71475923e-01 -3.98974270e-01 -2.41451561e-01 1.49443150e-01
2.10980222e-01 3.46729994e-01 4.82155383e-01 -1.07628095e+00
-6.34253800e-01 8.51021186e-02 -7.47347325e-02 3.77343714e-01
-5.78649998e-01 2.62393709e-02 -7.73979545e-01 -9.01707411e-01
-2.28428155e-01 -5.40732324e-01 -4.40032035e-01 -8.44266415e-01
-1.00427496e+00 -5.15796959e-01 3.55648249e-01 -5.51965356e-01
2.22504067e+00 -1.75047052e+00 1.58577234e-01 -1.45040993e-02
1.41305178e-01 4.25336927e-01 -3.96053940e-01 1.05011153e+00
4.62724537e-01 3.41068894e-01 -2.83156157e-01 -3.48724991e-01
2.46423513e-01 -6.44612610e-02 -6.41595483e-01 -1.90081254e-01
-1.39766801e-02 1.08729637e+00 -1.22188008e+00 -8.59655261e-01
-2.53034145e-01 -9.96492356e-02 -4.76097465e-01 3.11014563e-01
-5.17631590e-01 1.39695719e-01 -7.11687028e-01 3.73350948e-01
4.41172838e-01 5.52475546e-03 2.26515695e-01 -8.93664733e-02
-3.02549779e-01 7.71143794e-01 -8.27320814e-01 1.82320559e+00
-1.95682406e-01 2.38662228e-01 -4.93846029e-01 -4.66438353e-01
1.18938136e+00 1.37670204e-01 1.84022397e-01 -5.61646819e-01
-2.17936262e-01 4.37120199e-01 -4.12570775e-01 -3.91224623e-01
1.40390241e+00 3.92618030e-01 -3.52468133e-01 5.74430227e-01
-2.62886941e-01 -2.00978577e-01 8.04938376e-01 6.79354489e-01
1.25599909e+00 1.58607122e-02 5.51805735e-01 -7.91823491e-02
5.64051330e-01 1.84714183e-01 8.27732205e-01 1.01837993e+00
3.86579365e-01 9.71080780e-01 6.50016069e-01 1.16119087e-01
-9.50365305e-01 -8.00480783e-01 3.47881287e-01 9.08953965e-01
1.68230459e-01 -1.13033593e+00 -7.38087714e-01 -7.71882474e-01
-3.75986606e-01 1.18493235e+00 -1.94181979e-01 -3.24948549e-01
-8.69389594e-01 -4.82786030e-01 7.14220583e-01 3.72111052e-01
2.95123547e-01 -1.09645426e+00 -4.56395358e-01 5.27019799e-01
-6.25319123e-01 -7.36178815e-01 -1.17873144e+00 -4.66910094e-01
-7.94569671e-01 -7.90420115e-01 -8.34796727e-01 -4.10597444e-01
4.18648452e-01 3.96512002e-01 1.22942066e+00 1.74850315e-01
3.14887643e-01 -1.94922574e-02 -6.75480366e-01 -4.01070714e-01
-6.69839799e-01 7.96209216e-01 -1.54364899e-01 -3.50020468e-01
-8.56284499e-02 -5.35267413e-01 -6.53178811e-01 -3.37461345e-02
-1.18162596e+00 6.96443990e-02 8.08282912e-01 7.43252218e-01
4.53137636e-01 -3.27500105e-01 1.06641245e+00 -1.00705206e+00
1.33274519e+00 -2.68291980e-01 -3.18892002e-01 7.21932888e-01
-8.12278867e-01 1.22546434e-01 1.02533710e+00 -9.54843163e-02
-1.30964077e+00 -2.98312366e-01 -3.86589542e-02 8.34216643e-03
1.84361532e-01 8.37922037e-01 -2.23571330e-01 6.78227484e-01
5.34450710e-01 7.39924669e-01 -1.13971241e-01 -4.90397453e-01
3.95851284e-01 8.20119917e-01 6.97190285e-01 -3.76648694e-01
7.39543498e-01 -4.00857441e-03 -3.49332154e-01 -5.70171237e-01
-9.17505741e-01 -4.28459316e-01 -3.06069016e-01 -9.42116305e-02
4.04263049e-01 -7.73350596e-01 -3.50746781e-01 9.65980813e-02
-1.50464940e+00 2.68546697e-02 -2.63773024e-01 1.53132364e-01
-4.66828525e-01 8.09551299e-01 -3.52652311e-01 -6.32587016e-01
-1.09661746e+00 -9.76399899e-01 1.24912250e+00 4.00886357e-01
-5.42786360e-01 -5.94638348e-01 3.04618746e-01 1.89453647e-01
5.75709164e-01 1.04313537e-01 7.88913071e-01 -1.01819849e+00
-4.79227483e-01 -2.13012159e-01 -2.99929194e-02 6.86700046e-02
2.14009270e-01 2.94429988e-01 -5.13917506e-01 -5.27131110e-02
-3.37626219e-01 1.65845472e-02 1.14811277e+00 2.29487047e-01
1.04905188e+00 -9.07559693e-01 -3.97309124e-01 2.96621740e-01
1.00589621e+00 1.91589668e-02 7.08144546e-01 2.28484020e-01
5.45987189e-01 4.72383469e-01 1.02655351e+00 8.24099123e-01
6.99086726e-01 5.60625553e-01 3.82188410e-02 3.26699704e-01
-1.58029005e-01 -6.39463782e-01 6.48566961e-01 1.52753508e+00
1.99362665e-01 -7.20378280e-01 -5.85740685e-01 5.65876484e-01
-2.16116834e+00 -1.09347165e+00 -1.07101351e-01 2.05795455e+00
1.11564708e+00 9.58860517e-02 7.29462355e-02 -1.81005254e-01
7.08991587e-01 5.86160481e-01 -4.77277994e-01 -5.15954196e-01
-3.65698129e-01 -2.91908514e-02 1.69505998e-01 3.24301332e-01
-6.10544205e-01 9.96030152e-01 6.63622236e+00 1.17816639e+00
-8.51802707e-01 -3.00290495e-01 2.94729143e-01 -3.22926551e-01
-1.05156624e+00 -1.84418187e-02 -8.80931258e-01 6.96331203e-01
7.74166167e-01 -1.05400157e+00 2.21129879e-01 5.77108145e-01
4.09746349e-01 4.76098526e-03 -1.03506386e+00 5.39624810e-01
3.49948555e-01 -1.70521009e+00 5.92524230e-01 -1.28933176e-01
8.70476127e-01 -3.34805697e-01 -3.11176568e-01 4.72221911e-01
3.10691148e-01 -7.57182956e-01 6.03087604e-01 8.07709992e-01
7.37441540e-01 -8.42415392e-01 8.12425911e-01 5.84050894e-01
-9.97505724e-01 1.40364841e-01 -5.26537657e-01 2.14021087e-01
5.06387174e-01 8.68189156e-01 -7.30561852e-01 1.20364738e+00
6.43157586e-02 7.86787033e-01 -6.87841475e-01 1.14666247e+00
-4.47939485e-01 5.60437202e-01 -5.71448281e-02 -4.42904651e-01
6.07227869e-02 -2.53796399e-01 9.21718895e-01 1.37431538e+00
6.91142499e-01 2.43915528e-01 2.96356738e-01 6.06159210e-01
-3.45972806e-01 2.68004268e-01 -5.44282794e-01 1.75632909e-02
9.45775568e-01 1.21097100e+00 -1.77872032e-01 -7.65516937e-01
1.24782018e-01 8.58684897e-01 9.80299413e-02 2.22432777e-01
-6.38820231e-01 -8.15169632e-01 2.91654021e-01 -2.51357138e-01
3.36571008e-01 -3.59150693e-02 -4.01477307e-01 -1.38145030e+00
3.81394625e-01 -1.16406226e+00 3.61434162e-01 -5.83192468e-01
-1.08347225e+00 7.67512143e-01 1.22040510e-01 -1.40153897e+00
-5.27305365e-01 4.22234863e-01 -1.11969197e+00 6.12120867e-01
-1.48029220e+00 -7.39251137e-01 -1.67808354e-01 -5.31318411e-02
8.50850284e-01 -5.08475453e-02 5.70828021e-01 6.38034288e-03
-6.90231740e-01 8.38774920e-01 2.75824279e-01 -2.67787129e-01
1.04671085e+00 -1.06780624e+00 7.33406484e-01 1.06122363e+00
-1.75095603e-01 9.22235310e-01 8.78996313e-01 -1.03213477e+00
-1.58993030e+00 -1.30106652e+00 1.29552436e+00 -1.86240822e-01
3.90334725e-01 -1.20346650e-01 -8.47403765e-01 2.14634731e-01
6.25688732e-01 -7.14241147e-01 5.11684656e-01 -1.03966989e-01
-1.21870413e-01 -2.33639076e-01 -6.68999732e-01 1.05258334e+00
1.31055868e+00 -4.33835760e-02 -8.83659422e-01 3.14884692e-01
1.13014054e+00 -3.87767226e-01 -7.08060861e-01 5.26174545e-01
5.86459637e-01 -8.17238450e-01 6.18281305e-01 -4.09910351e-01
7.65547514e-01 -2.92261124e-01 9.44419727e-02 -1.58794630e+00
-2.40078434e-01 -1.18508875e+00 -3.25778753e-01 1.78848016e+00
5.21973312e-01 -5.40252149e-01 5.72926939e-01 3.32175881e-01
-7.12440193e-01 -7.64865637e-01 -6.79305792e-01 -9.63705361e-01
-1.72315732e-01 1.27691552e-01 1.06735504e+00 4.41415697e-01
3.45980883e-01 6.15882039e-01 -3.10359210e-01 -2.58932948e-01
4.18813258e-01 4.58532602e-01 9.69734132e-01 -1.02600682e+00
-1.03791364e-01 -8.06806028e-01 2.89449275e-01 -1.41317439e+00
1.66458234e-01 -8.46863389e-01 3.00548881e-01 -1.96148801e+00
4.03451055e-01 -1.94916934e-01 4.35258495e-03 2.51379579e-01
-8.47861826e-01 -1.98719710e-01 2.51603693e-01 3.09313029e-01
-1.14321256e+00 1.09130704e+00 1.29855311e+00 -1.77363262e-01
-4.88909215e-01 1.41430059e-02 -1.17691970e+00 2.12936699e-01
9.05449152e-01 -4.61445928e-01 -3.73405010e-01 -4.30123121e-01
3.18677455e-01 2.49899447e-01 -3.41914296e-01 -7.05158055e-01
5.67509532e-01 -2.81214476e-01 -2.37880319e-01 -1.06253481e+00
-1.33236125e-01 6.11412758e-03 6.45477697e-02 1.94172159e-01
-7.55997658e-01 4.02143031e-01 -1.26821712e-01 5.90732574e-01
-3.85883927e-01 -3.98068249e-01 4.94465604e-02 -6.80076107e-02
-6.36368692e-02 2.30956048e-01 -4.72334236e-01 3.72399062e-01
6.22241318e-01 -1.70722589e-01 -8.32009971e-01 -4.89687860e-01
2.36648604e-01 5.24578571e-01 6.29403710e-01 3.48125458e-01
6.95522070e-01 -1.22899544e+00 -1.09721577e+00 -3.33024561e-01
2.67088085e-01 1.81469426e-01 6.53987080e-02 8.15871537e-01
-2.10134596e-01 6.59978211e-01 1.33959353e-01 -2.90823519e-01
-1.16740108e+00 3.50909323e-01 -3.61963838e-01 -6.81917131e-01
-4.93020535e-01 4.01747078e-01 -7.17945546e-02 -3.10531229e-01
7.99278170e-02 -1.76095128e-01 -3.96136433e-01 2.66534418e-01
7.34419703e-01 6.24145925e-01 2.77497977e-01 -6.63957372e-02
-2.18736783e-01 1.65809974e-01 -3.51880997e-01 -1.53705671e-01
1.24672222e+00 -1.36459947e-01 -3.25912803e-01 1.78171545e-01
8.61592352e-01 5.82087636e-01 -6.95148289e-01 -1.29278943e-01
2.66034573e-01 -2.97107458e-01 -3.11758131e-01 -7.82736123e-01
-7.46590912e-01 3.82611811e-01 -5.83635211e-01 2.98870444e-01
1.15264261e+00 -1.76269963e-01 1.16298616e+00 5.67387640e-01
8.97175446e-02 -1.03933227e+00 9.89408791e-03 7.80759573e-01
1.10757864e+00 -7.53208339e-01 3.10822576e-01 -4.63046253e-01
-8.20537627e-01 1.15789056e+00 5.31751275e-01 6.45125210e-02
-1.97813928e-01 -1.24951666e-02 -1.14078574e-01 8.05124864e-02
-1.31379557e+00 2.70149354e-02 5.20175278e-01 3.71014833e-01
5.14550865e-01 -5.66039048e-02 -9.22333181e-01 8.80081475e-01
-5.00179350e-01 -9.01338458e-02 8.76024485e-01 7.46659398e-01
-5.81744015e-01 -1.34792364e+00 -1.87822372e-01 8.15715373e-01
-3.15743566e-01 -4.01101381e-01 -7.03591883e-01 2.98379779e-01
-5.82050502e-01 1.21621108e+00 -2.64410198e-01 -3.62966269e-01
6.23095334e-01 -1.72702432e-01 1.03319757e-01 -8.12855899e-01
-1.04844987e+00 1.26374200e-01 5.71321070e-01 -3.82031769e-01
-5.11030070e-02 -7.25169122e-01 -9.82533693e-01 -2.85187393e-01
-3.81424695e-01 4.79078710e-01 3.01475525e-01 6.93904698e-01
9.01490331e-01 6.46041155e-01 8.80507708e-01 -6.18839562e-01
-9.92916405e-01 -1.06894672e+00 -3.17928605e-02 4.31069657e-02
1.89937592e-01 -1.36632711e-01 -2.11148858e-01 -1.77358136e-01] | [12.4541654586792, 9.416108131408691] |
9d42bc53-0f58-4811-8ded-e6a65e7c2e6c | visual-and-language-navigation-a-survey-and | 2108.11544 | null | https://arxiv.org/abs/2108.11544v3 | https://arxiv.org/pdf/2108.11544v3.pdf | Vision-Language Navigation: A Survey and Taxonomy | Vision-Language Navigation (VLN) tasks require an agent to follow human language instructions to navigate in previously unseen environments. This challenging field involving problems in natural language processing, computer vision, robotics, etc., has spawn many excellent works focusing on various VLN tasks. This paper provides a comprehensive survey and an insightful taxonomy of these tasks based on the different characteristics of language instructions in these tasks. Depending on whether the navigation instructions are given for once or multiple times, this paper divides the tasks into two categories, i.e., single-turn and multi-turn tasks. For single-turn tasks, we further subdivide them into goal-oriented and route-oriented based on whether the instructions designate a single goal location or specify a sequence of multiple locations. For multi-turn tasks, we subdivide them into passive and interactive tasks based on whether the agent is allowed to question the instruction or not. These tasks require different capabilities of the agent and entail various model designs. We identify progress made on the tasks and look into the limitations of existing VLN models and task settings. Finally, we discuss several open issues of VLN and point out some opportunities in the future, i.e., incorporating knowledge with VLN models and implementing them in the real physical world. | ['Xinmeng Li', 'Tao Chang', 'Wansen Wu'] | 2021-08-26 | null | null | null | null | ['vision-language-navigation'] | ['computer-vision'] | [ 3.65540572e-02 -1.64389029e-01 3.22572491e-03 -4.19400960e-01
-1.50837407e-01 -1.04107070e+00 9.63726938e-01 -1.15491524e-01
-8.30070198e-01 6.75581872e-01 -1.41636670e-01 -7.79289842e-01
-2.51616955e-01 -6.03310049e-01 -4.41831589e-01 -6.31972909e-01
-2.10794389e-01 6.61010027e-01 5.36843240e-01 -6.66559160e-01
6.56341851e-01 4.79411781e-01 -1.74669182e+00 -9.01011527e-02
5.28327823e-01 4.19543684e-01 9.24296379e-01 9.11709368e-01
-3.10996860e-01 7.83050656e-01 -3.13676417e-01 3.37727994e-01
3.45648348e-01 -2.18787268e-01 -9.81564283e-01 2.49423273e-02
3.09728265e-01 -1.67319268e-01 -1.31206825e-01 1.06295478e+00
1.66511998e-01 6.24596596e-01 7.43856251e-01 -1.54272687e+00
-4.79642183e-01 7.50744641e-02 -4.73885462e-02 1.76132992e-01
8.83363545e-01 2.52205998e-01 8.28632057e-01 -8.88623774e-01
7.11608291e-01 1.32019711e+00 3.66887569e-01 6.61797822e-01
-9.53756928e-01 -8.39400962e-02 7.47848749e-01 5.18126845e-01
-9.75655735e-01 -4.37542468e-01 4.81927276e-01 -5.97878635e-01
1.25185788e+00 8.81738588e-02 4.63958681e-01 9.22573507e-01
5.44207394e-01 7.75956154e-01 1.09530604e+00 -6.24352515e-01
3.89932275e-01 -2.38157973e-01 3.00529301e-01 8.93044710e-01
2.08067775e-01 3.32018137e-01 -3.61583292e-01 1.00080617e-01
8.66387427e-01 -2.38728613e-01 -2.52678275e-01 -1.02491593e+00
-1.54886830e+00 9.38285649e-01 3.28451961e-01 3.52223486e-01
-4.06073600e-01 9.82597396e-02 3.87559265e-01 3.13802451e-01
-4.49217349e-01 6.35615110e-01 -3.84924442e-01 -7.61349425e-02
-3.74827415e-01 5.85477412e-01 8.74004245e-01 1.03310931e+00
1.06685007e+00 3.22443843e-02 2.51500815e-01 6.92013204e-01
3.94125491e-01 4.47643787e-01 4.57439184e-01 -1.32262194e+00
5.77043653e-01 1.51399866e-01 5.59280396e-01 -9.29492950e-01
-9.18488204e-01 2.09137261e-01 -2.96605289e-01 7.26230145e-01
6.16546750e-01 -3.19104880e-01 -1.01433897e+00 1.71569920e+00
2.79664338e-01 -3.78134549e-01 2.83342510e-01 1.09119892e+00
9.25562739e-01 5.96940815e-01 1.08399242e-01 -3.31744663e-02
1.63319576e+00 -1.53173411e+00 -8.62807989e-01 -7.44646609e-01
8.09037209e-01 -6.92948699e-01 1.04337084e+00 4.19601291e-01
-6.79584622e-01 -7.13540316e-01 -9.62690175e-01 -2.58352757e-01
-9.63704765e-01 -9.61704925e-03 8.29998374e-01 2.70383865e-01
-1.49403059e+00 7.68999616e-03 -7.89353728e-01 -8.72834504e-01
-3.29981148e-01 3.47241491e-01 -4.18504596e-01 -1.22972116e-01
-1.24088991e+00 1.22982752e+00 3.36315066e-01 2.38617629e-01
-1.05900967e+00 2.35796019e-01 -1.41295362e+00 -4.69320804e-01
7.01820135e-01 -7.31141269e-01 1.81770992e+00 -4.21898246e-01
-1.40957916e+00 9.28972304e-01 -4.28517789e-01 -3.71251941e-01
3.89145195e-01 -1.38536096e-02 -1.39167443e-01 -1.15504615e-01
5.94634056e-01 9.58657205e-01 1.69275999e-01 -1.35554373e+00
-1.07295632e+00 -2.68752545e-01 8.49805295e-01 7.41405725e-01
5.16540110e-01 -1.23650275e-01 -4.23723340e-01 -1.57306835e-01
2.48211369e-01 -1.15705526e+00 -5.95062613e-01 5.11550605e-02
-2.43926451e-01 -4.51032132e-01 1.01266205e+00 2.29124110e-02
7.27677107e-01 -1.66534996e+00 1.26475453e-01 -2.48112276e-01
1.94490612e-01 8.51576701e-02 -2.78531551e-01 7.77421653e-01
1.73542887e-01 -2.65698910e-01 -6.43716753e-02 -2.92879343e-01
1.13106512e-01 7.50593364e-01 -2.46150389e-01 2.71960765e-01
-5.01979470e-01 9.82434332e-01 -1.14573085e+00 -2.71086037e-01
5.87694824e-01 2.15030074e-01 -2.60515213e-01 -4.22754511e-03
-3.49763632e-01 6.45115793e-01 -6.78080916e-01 3.44535708e-01
3.14733177e-01 1.12118706e-01 -2.67751440e-02 2.49253765e-01
-6.91764474e-01 4.23846751e-01 -1.15529060e+00 1.81374562e+00
-7.37470031e-01 8.41372013e-01 4.97594655e-01 -8.81270647e-01
5.78785062e-01 1.80873230e-01 -8.07358790e-03 -6.95383906e-01
-1.02227584e-01 2.20928431e-01 1.56710595e-01 -8.29330862e-01
6.12039387e-01 -1.59663782e-02 -2.55676687e-01 4.59270269e-01
-4.54032600e-01 -1.74258530e-01 5.31725645e-01 -6.00973368e-02
1.13013363e+00 4.07375276e-01 7.80865192e-01 -3.92982870e-01
7.92376995e-01 4.59233642e-01 1.05635643e-01 1.38975441e+00
-7.06252217e-01 2.03770399e-01 7.40655512e-02 -8.35011363e-01
-4.89568204e-01 -9.20042396e-01 3.05126935e-01 1.56181228e+00
6.26428425e-01 -2.15208456e-01 -6.15465760e-01 -6.76156819e-01
-3.60602528e-01 9.49007988e-01 -6.93875194e-01 3.42787683e-01
-7.56971538e-01 -3.50271463e-01 1.92675054e-01 4.44688112e-01
6.99529231e-01 -1.68679559e+00 -1.46376729e+00 1.57307349e-02
-4.65271562e-01 -1.24138558e+00 -4.10184950e-01 4.76655722e-01
-5.83928883e-01 -1.23494995e+00 -3.25414181e-01 -1.45727658e+00
7.78106093e-01 8.07124496e-01 1.05913103e+00 -1.77678186e-02
9.17527080e-02 1.01142311e+00 -4.88075376e-01 -3.49995345e-01
-4.78229411e-02 -6.41157925e-02 2.60245025e-01 -4.26075786e-01
4.98447657e-01 -2.73455024e-01 -3.93356264e-01 4.25569355e-01
-6.11895502e-01 1.33256674e-01 4.01504606e-01 5.72578549e-01
5.21587133e-01 1.45485010e-02 9.08323079e-02 -7.83455908e-01
9.18858528e-01 -1.94488272e-01 -6.95423722e-01 1.91820860e-01
-3.45130265e-01 9.80917886e-02 5.83231211e-01 -2.61991978e-01
-1.04041827e+00 3.10309738e-01 -5.19962832e-02 2.52682537e-01
-6.91528499e-01 5.19700646e-01 -1.40194416e-01 -3.02167684e-01
5.95314980e-01 5.40253401e-01 -2.00224847e-01 -3.07292435e-02
5.27635276e-01 2.96034902e-01 4.83268410e-01 -6.78490043e-01
5.29254675e-01 4.84992921e-01 -7.48480260e-02 -1.09808993e+00
-4.57017273e-01 -5.18713474e-01 -7.50554681e-01 -3.08770388e-01
1.14069164e+00 -5.62546730e-01 -8.88120174e-01 4.77934748e-01
-1.36229503e+00 -9.17516410e-01 4.48370352e-02 4.72059608e-01
-1.13285339e+00 3.77844542e-01 -5.00823736e-01 -7.53474176e-01
3.10733736e-01 -1.69818509e+00 8.64012182e-01 3.30404699e-01
-3.07986736e-01 -1.29561973e+00 8.95770043e-02 2.08187342e-01
4.28725421e-01 -1.50975898e-01 8.10525179e-01 -6.30107582e-01
-6.11854196e-01 2.53556389e-02 -8.79620239e-02 -4.70292419e-01
3.65597427e-01 -6.30791068e-01 -4.73849624e-01 -1.80530041e-01
1.66762784e-01 -2.55557567e-01 6.34997368e-01 5.24269462e-01
6.31757140e-01 -2.59331148e-03 -6.69117928e-01 3.33675623e-01
1.37112343e+00 9.77543235e-01 2.62004644e-01 9.88482416e-01
5.55262029e-01 1.03136218e+00 8.60235274e-01 -1.53722569e-01
9.30004835e-01 7.24575400e-01 7.46613026e-01 1.94219738e-01
9.60132852e-02 4.60560881e-02 3.42476189e-01 4.87949371e-01
-8.48374367e-02 -4.02449340e-01 -1.11927772e+00 5.06336153e-01
-2.07989430e+00 -7.13823140e-01 -2.55476058e-01 1.83498478e+00
1.99851826e-01 9.01839063e-02 -1.46718726e-01 -1.98089242e-01
6.35108113e-01 3.54919612e-01 -7.51123250e-01 -6.84277296e-01
1.09326378e-01 -4.38760579e-01 3.23878706e-01 1.19558597e+00
-1.17382324e+00 1.25924444e+00 6.55518866e+00 2.80502617e-01
-8.42093587e-01 -7.16076195e-02 -8.10529590e-02 3.96381795e-01
6.26854300e-02 7.80597478e-02 -9.89233673e-01 9.99215096e-02
4.93045926e-01 8.32258314e-02 6.46874428e-01 8.51643443e-01
4.38079774e-01 -8.00644159e-01 -1.30541039e+00 8.83862793e-01
2.20737398e-01 -8.31283152e-01 6.73565418e-02 8.08837637e-02
1.98315889e-01 1.45370603e-01 -3.51846628e-02 6.28852010e-01
8.09833765e-01 -8.58700275e-01 8.22415173e-01 2.00127289e-01
2.27124095e-01 -2.62344360e-01 5.56993306e-01 8.52632105e-01
-1.37565362e+00 -1.72618508e-01 -1.56827644e-01 -7.95710802e-01
4.45714653e-01 -1.77162290e-01 -5.13419986e-01 5.78762829e-01
6.71749473e-01 3.98486227e-01 -3.09031397e-01 1.08128810e+00
-5.61769068e-01 -2.72378087e-01 -6.88237324e-02 -4.93733943e-01
7.88456440e-01 -5.52979410e-01 5.91862321e-01 1.06123102e+00
1.40897512e-01 1.75318569e-01 7.41025090e-01 4.54082698e-01
8.27388763e-01 -2.06763059e-01 -1.00603878e+00 5.20490408e-01
2.96422452e-01 1.03213704e+00 -1.05458105e+00 -1.40202433e-01
-6.15209401e-01 9.42667425e-01 2.64977872e-01 8.20973754e-01
-6.16104662e-01 -7.20945120e-01 7.59121895e-01 -1.82181951e-02
1.83854535e-01 -8.98621559e-01 -9.87210646e-02 -8.14932764e-01
-1.02325447e-01 -6.55082166e-01 1.85958177e-01 -1.10799015e+00
-9.12589431e-01 7.79748559e-01 3.30982089e-01 -1.10445011e+00
-4.84866798e-01 -1.02468133e+00 -4.89694774e-01 7.45166242e-01
-1.78824568e+00 -7.97500610e-01 -4.36538011e-01 5.18375456e-01
1.17159557e+00 6.18038028e-02 9.49774146e-01 -1.49982423e-01
-2.93012321e-01 -1.55164212e-01 -2.72345275e-01 1.55275792e-01
5.60169876e-01 -1.22798872e+00 4.02129173e-01 7.66047776e-01
2.97571812e-02 8.54521453e-01 9.38746154e-01 -4.41136628e-01
-1.05088794e+00 -6.98854864e-01 1.09496260e+00 -6.72514260e-01
4.80160415e-01 -5.40500998e-01 -5.43933153e-01 1.24673676e+00
4.01418209e-01 -1.63625062e-01 3.90754491e-01 -1.17910206e-01
-4.91072983e-02 3.28287661e-01 -9.59821820e-01 1.21333075e+00
1.31622410e+00 -3.07757467e-01 -1.01376963e+00 4.28800583e-01
5.38496137e-01 -5.98644078e-01 1.66814715e-01 1.14070714e-01
4.27979141e-01 -1.25011635e+00 9.09570873e-01 -5.58634996e-01
3.92197035e-02 -7.92311847e-01 -2.60729939e-01 -1.35959482e+00
-4.96903330e-01 -2.84678847e-01 3.59086096e-01 5.32209516e-01
4.08655822e-01 -1.03216112e+00 7.16933846e-01 5.49779475e-01
-3.17196280e-01 -5.70160866e-01 -8.63654792e-01 -7.30999827e-01
-1.32006451e-01 -4.58244175e-01 1.63162693e-01 6.31973863e-01
1.17981704e-02 5.27245402e-01 -1.51070908e-01 3.34453106e-01
4.58167672e-01 1.93186790e-01 7.53036797e-01 -1.24614835e+00
2.31411204e-01 -5.08786380e-01 -2.54370987e-01 -1.82713485e+00
3.43301415e-01 -6.59837604e-01 6.95636213e-01 -2.25083542e+00
-3.05889875e-01 -3.37313563e-01 1.90969139e-01 6.42458260e-01
1.99591547e-01 -1.94260687e-01 2.24214673e-01 1.98389575e-01
-9.97946322e-01 3.54578197e-01 1.27670908e+00 5.73367998e-02
-3.66867095e-01 1.81021497e-01 -5.27286828e-01 1.05332017e+00
8.89562666e-01 -1.31415799e-01 -8.83485734e-01 -8.29928935e-01
1.92020267e-01 6.32687584e-02 2.31781930e-01 -1.08471727e+00
8.29467416e-01 -4.63949889e-01 -2.01336443e-02 -6.43087864e-01
5.47825933e-01 -9.60853815e-01 -3.75685632e-01 6.60304129e-01
-2.48846769e-01 5.69979906e-01 1.89175889e-01 6.16834104e-01
-1.34809494e-01 -4.96342748e-01 4.27943677e-01 -7.45944142e-01
-1.56619608e+00 -1.38052016e-01 -1.41368198e+00 -1.32301077e-01
1.28638232e+00 -7.12150574e-01 -2.91084200e-01 -6.00373566e-01
-1.06958485e+00 8.92908812e-01 3.03292751e-01 6.45864666e-01
6.62406206e-01 -1.02654958e+00 -1.15206040e-01 1.63196325e-01
1.48010626e-01 1.55960292e-01 1.53874248e-01 6.80462539e-01
-7.29415059e-01 1.13055706e+00 -3.49524975e-01 -6.07079029e-01
-1.13089895e+00 6.99724495e-01 4.18706745e-01 -3.32170337e-01
-2.88825065e-01 7.22357690e-01 7.51744449e-01 -1.16293359e+00
4.46344256e-01 -3.28636855e-01 -8.39509130e-01 -1.18923791e-01
3.79631370e-01 2.01096624e-01 -1.26257926e-01 -7.68343449e-01
-5.19376397e-01 9.41793978e-01 -8.38447213e-02 -2.77799726e-01
8.42020452e-01 -6.15796387e-01 -1.82513416e-01 7.65636325e-01
4.93020266e-01 -3.45903367e-01 -1.15395629e+00 -1.25482157e-01
1.47539467e-01 4.47451957e-02 -3.26068372e-01 -8.18154395e-01
-4.86711800e-01 8.39083135e-01 3.61658663e-01 3.37944359e-01
8.34906518e-01 -6.65959567e-02 3.30707937e-01 9.91076589e-01
1.08600092e+00 -9.13203180e-01 6.74206093e-02 1.35129619e+00
1.00930882e+00 -1.29179859e+00 -2.33500540e-01 -4.18074578e-01
-5.16789854e-01 1.16283143e+00 1.00692081e+00 -6.96355570e-03
6.79645419e-01 9.71290618e-02 5.73768914e-01 -3.31626147e-01
-5.56465328e-01 -4.17064309e-01 -1.77584335e-01 1.20824146e+00
9.22484696e-02 1.43031385e-02 -3.43289584e-01 1.88601553e-01
-5.00417531e-01 -2.49744847e-01 6.79000139e-01 1.26665258e+00
-8.75269592e-01 -1.01883972e+00 -6.07446730e-01 8.04463327e-02
2.56452560e-01 9.65468735e-02 -3.98035914e-01 1.03494751e+00
2.89536178e-01 1.39673030e+00 -9.17279646e-02 -2.04017192e-01
6.60135567e-01 -3.27627920e-02 3.60132307e-01 -9.35785711e-01
-3.41769814e-01 -3.64642918e-01 1.09404609e-01 -5.85645616e-01
-5.34563839e-01 -7.56272733e-01 -1.56373870e+00 1.25489250e-01
1.11982800e-01 1.75251558e-01 6.10842168e-01 1.10021889e+00
-4.58247624e-02 4.03741837e-01 -9.52208564e-02 -1.45392931e+00
-2.35551298e-01 -5.40280998e-01 -2.11269289e-01 -3.19242716e-01
7.27379143e-01 -9.87407804e-01 -3.84194523e-01 -1.14515260e-01] | [4.505753517150879, 0.5893887281417847] |
bd57b146-b386-4ed3-891f-34532370b0b2 | cross-lingual-text-independent-speaker | 1908.01447 | null | https://arxiv.org/abs/1908.01447v1 | https://arxiv.org/pdf/1908.01447v1.pdf | Cross-lingual Text-independent Speaker Verification using Unsupervised Adversarial Discriminative Domain Adaptation | Speaker verification systems often degrade significantly when there is a language mismatch between training and testing data. Being able to improve cross-lingual speaker verification system using unlabeled data can greatly increase the robustness of the system and reduce human labeling costs. In this study, we introduce an unsupervised Adversarial Discriminative Domain Adaptation (ADDA) method to effectively learn an asymmetric mapping that adapts the target domain encoder to the source domain, where the target domain and source domain are speech data from different languages. ADDA, together with a popular Domain Adversarial Training (DAT) approach, are evaluated on a cross-lingual speaker verification task: the training data is in English from NIST SRE04-08, Mixer 6 and Switchboard, and the test data is in Chinese from AISHELL-I. We show that with the ADDA adaptation, Equal Error Rate (EER) of the x-vector system decreases from 9.331\% to 7.645\%, relatively 18.07\% reduction of EER, and 6.32\% reduction from DAT as well. Further data analysis of ADDA adapted speaker embedding shows that the learned speaker embeddings can perform well on speaker classification for the target domain data, and are less dependent with respect to the shift in language. | ['John H. L. Hansen', 'Jing Huang', 'Wei Xia'] | 2019-08-05 | null | null | null | null | ['text-independent-speaker-verification'] | ['speech'] | [ 9.26464722e-02 -1.47939473e-01 -3.84004135e-03 -7.23022163e-01
-1.27748239e+00 -9.72577274e-01 6.07155859e-01 -3.73480618e-01
-4.36745524e-01 6.57698035e-01 3.47806931e-01 -6.06432080e-01
4.16080266e-01 -2.45440796e-01 -5.69648147e-01 -8.07992637e-01
1.34713128e-01 2.59963185e-01 -1.26985282e-01 -3.58072490e-01
-3.32256049e-01 5.76691866e-01 -1.14004529e+00 5.06219193e-02
7.33426690e-01 8.27330589e-01 -1.97296113e-01 6.18871987e-01
-4.31799553e-02 2.40673661e-01 -9.61619914e-01 -4.57586467e-01
4.09795821e-01 -4.80939716e-01 -6.48719549e-01 -1.50310963e-01
6.31015062e-01 -6.57491386e-02 -4.71098453e-01 1.19429100e+00
9.31800902e-01 9.70522463e-02 6.76464319e-01 -1.22705579e+00
-8.77426386e-01 6.96646452e-01 -2.95412153e-01 2.45916694e-01
2.43321300e-01 9.03030485e-02 5.61165273e-01 -1.16652405e+00
4.17139620e-01 1.30736291e+00 6.76987886e-01 1.13144016e+00
-1.31325471e+00 -1.32599342e+00 1.31726757e-01 1.47710249e-01
-1.78865623e+00 -9.45029736e-01 9.62029874e-01 -3.43189299e-01
6.67381585e-01 2.14830369e-01 -1.71663448e-01 1.33434737e+00
-1.57317057e-01 5.24990618e-01 1.14561713e+00 -6.06237352e-01
3.21457684e-01 7.10186720e-01 2.50083834e-01 3.10854733e-01
-1.29330248e-01 4.90052372e-01 -7.30191708e-01 -1.66011393e-01
-1.65053625e-02 -4.91766125e-01 -3.34208399e-01 -1.73267871e-01
-8.23294759e-01 9.21434045e-01 2.83038586e-01 3.95490497e-01
-8.09480902e-03 -4.78476167e-01 6.24322534e-01 7.29186833e-01
4.86073375e-01 6.82491586e-02 -5.61654150e-01 1.72318831e-01
-7.98690915e-01 -9.59690735e-02 7.44422078e-01 9.52829659e-01
4.81074542e-01 8.02639782e-01 1.23860762e-01 1.13897061e+00
5.73654056e-01 1.06560004e+00 1.04604733e+00 -3.25691283e-01
5.01204252e-01 1.71190575e-01 -2.42403403e-01 -3.44216228e-01
7.30870739e-02 -3.67730290e-01 -6.32763028e-01 3.20798129e-01
4.62802798e-01 -3.39348972e-01 -1.03554749e+00 2.16342783e+00
2.15629920e-01 5.80147803e-02 6.39001012e-01 5.37429929e-01
8.83026958e-01 7.69626737e-01 1.34583846e-01 -6.63833767e-02
1.22699773e+00 -6.65552199e-01 -8.90981376e-01 -4.44456488e-01
6.59694016e-01 -8.98434758e-01 1.28974152e+00 1.64564982e-01
-6.77460372e-01 -8.39455545e-01 -1.47119021e+00 1.22844420e-01
-5.79337299e-01 5.49093895e-02 -2.23043993e-01 1.40876186e+00
-1.04955637e+00 -2.45094951e-02 -4.01496708e-01 -2.82676965e-01
7.79353604e-02 5.54400861e-01 -6.24648750e-01 -2.59389639e-01
-1.49405265e+00 1.02931738e+00 3.08893740e-01 -3.39482427e-01
-8.90480161e-01 -8.28027725e-01 -1.21670198e+00 -1.29568413e-01
-2.36892536e-01 1.60755247e-01 1.26442313e+00 -9.96577561e-01
-1.81367481e+00 9.72737610e-01 -3.29931468e-01 -6.29817784e-01
1.55525520e-01 7.58234486e-02 -1.26378047e+00 -4.00285482e-01
-4.30350788e-02 5.14324009e-01 9.93615091e-01 -1.15686536e+00
-3.31430435e-01 -6.13266230e-01 -6.65767372e-01 4.41671088e-02
-5.14447927e-01 2.95616955e-01 1.20744392e-01 -7.37369537e-01
-1.25546679e-01 -1.02528834e+00 3.70992750e-01 -2.95937300e-01
-2.09072560e-01 -2.81784654e-01 1.40352666e+00 -1.14003599e+00
1.09691417e+00 -2.64762878e+00 -9.97645035e-02 3.28302830e-01
-5.60208201e-01 5.94774544e-01 -2.11752757e-01 9.29071829e-02
-4.76319641e-01 -1.51763372e-02 -5.01684606e-01 -4.22463834e-01
2.42330596e-01 2.06239730e-01 -4.49007958e-01 6.15526080e-01
1.72784671e-01 4.34937149e-01 -4.95140672e-01 -2.21233726e-01
1.59052342e-01 6.10253751e-01 -4.03733671e-01 1.96968734e-01
4.40070748e-01 5.37495077e-01 1.23442546e-01 5.91401935e-01
1.00675845e+00 6.20677412e-01 6.72539547e-02 2.12188922e-02
1.21653356e-01 5.96248627e-01 -1.15117741e+00 1.60231125e+00
-5.37916422e-01 8.73186827e-01 3.32825512e-01 -8.37994397e-01
1.23849630e+00 6.57067716e-01 -4.44005504e-02 -6.97226048e-01
1.02126323e-01 3.43484461e-01 3.11330765e-01 -9.99725517e-03
6.08427152e-02 -6.19477451e-01 -4.55225766e-01 3.55701298e-01
3.61205250e-01 -3.00276488e-01 -4.04617637e-01 -6.73542172e-02
6.71846747e-01 -4.40528333e-01 2.17165455e-01 -4.90338862e-01
1.01905644e+00 -4.04015064e-01 6.10030413e-01 3.35990548e-01
-6.68712080e-01 1.86592937e-01 -9.13928971e-02 1.06762491e-01
-9.15480256e-01 -1.33569717e+00 -5.05288184e-01 1.17670548e+00
-2.23011732e-01 3.97719294e-02 -9.30334032e-01 -9.70262527e-01
1.64773598e-01 1.18271434e+00 -3.09879005e-01 -5.24207175e-01
-7.08473861e-01 -3.36716175e-01 1.11725378e+00 5.42217076e-01
4.66637999e-01 -7.23921657e-01 5.89368522e-01 4.98614870e-02
2.45484747e-02 -1.11950064e+00 -1.11517906e+00 3.01770926e-01
-5.12109458e-01 -5.88968039e-01 -7.12441623e-01 -1.30685687e+00
5.24036407e-01 -1.53510064e-01 7.86960363e-01 -6.41372323e-01
1.89558759e-01 3.26048106e-01 -1.02753766e-01 -5.72388887e-01
-1.18449152e+00 -6.77634627e-02 8.37675154e-01 2.03648940e-01
8.96167040e-01 -2.45401993e-01 -4.19116542e-02 7.34634399e-01
-6.38842642e-01 -7.38925576e-01 1.85126439e-01 1.13930178e+00
2.90210813e-01 -1.02608616e-03 1.01174402e+00 -4.85980958e-01
5.48156977e-01 -2.37656593e-01 -5.48056543e-01 6.31103888e-02
-7.45468736e-01 1.72281131e-01 6.69324815e-01 -6.11570954e-01
-1.20880353e+00 1.94797829e-01 -5.33803999e-01 -3.82269144e-01
-2.78604448e-01 1.06931157e-01 -7.74377108e-01 -1.01663470e-01
9.42804694e-01 4.08571571e-01 1.14296533e-01 -5.20640075e-01
2.42330179e-01 1.32139778e+00 7.10535467e-01 -2.55729914e-01
1.15693438e+00 7.64364079e-02 -6.93679452e-01 -9.88704860e-01
-1.66833237e-01 -3.87360930e-01 -6.10767007e-01 1.12457260e-01
6.16821647e-01 -1.14510942e+00 -2.81077296e-01 6.47128403e-01
-8.53365183e-01 -2.85019010e-01 -4.09338057e-01 7.40813673e-01
-1.29518256e-01 3.11167717e-01 -4.18396950e-01 -6.78422332e-01
-3.33333850e-01 -1.24095690e+00 7.91785479e-01 1.05405170e-02
-2.68410265e-01 -1.08452666e+00 1.24071524e-01 4.26047295e-01
6.01451874e-01 -3.77016276e-01 7.27770865e-01 -1.30507362e+00
1.84524566e-01 -4.83344734e-01 2.53635496e-01 1.23148751e+00
4.44077760e-01 -4.56902981e-01 -1.50472927e+00 -7.36974955e-01
4.37016875e-01 -4.37697545e-02 3.61529738e-01 9.37134698e-02
7.28083909e-01 -3.29390258e-01 -1.17866874e-01 4.35187459e-01
1.02128112e+00 5.56180596e-01 4.34241384e-01 -7.33460560e-02
4.63897049e-01 5.11000395e-01 2.70592451e-01 -1.41606078e-01
2.42705233e-02 9.72475529e-01 -1.52789176e-01 5.54112233e-02
-4.99452114e-01 -3.62642795e-01 1.11879814e+00 1.23438096e+00
5.86914659e-01 -1.12112626e-01 -9.45292890e-01 5.50586879e-01
-8.01824808e-01 -8.53382111e-01 3.05303663e-01 2.45231438e+00
1.09914517e+00 -2.11609751e-02 1.73794419e-01 4.82370198e-01
9.45971906e-01 -8.82970169e-02 -6.85368299e-01 -7.05621958e-01
-2.41238981e-01 4.33972985e-01 5.69498718e-01 9.85084832e-01
-1.05418837e+00 8.76901209e-01 6.11375809e+00 8.35885584e-01
-1.56373715e+00 4.75559562e-01 3.38433504e-01 7.33830035e-02
-8.45324025e-02 -5.67176163e-01 -9.74660933e-01 6.66192830e-01
1.59392178e+00 -4.20791090e-01 3.07894647e-01 1.06423926e+00
-7.48737603e-02 6.18641973e-01 -1.40936494e+00 1.16028774e+00
4.91494477e-01 -6.58967972e-01 -1.86466500e-01 1.32369056e-01
6.55289531e-01 9.75369513e-02 4.09244210e-01 7.93166816e-01
3.39373618e-01 -9.89657879e-01 5.72452426e-01 -2.27067366e-01
1.10438275e+00 -8.32353652e-01 9.54350770e-01 2.26116747e-01
-1.07613444e+00 1.05713554e-01 -1.10431388e-01 4.29902732e-01
1.12145327e-01 6.17711209e-02 -1.31526053e+00 1.68733314e-01
6.47784472e-01 2.30502754e-01 -3.30157518e-01 2.96468318e-01
-1.65767759e-01 9.27715898e-01 -2.50437528e-01 2.05060229e-01
-1.64193183e-01 1.42952740e-01 6.99272513e-01 1.34255314e+00
2.77240962e-01 -2.45886564e-01 -2.11043790e-01 5.10376573e-01
-4.28559005e-01 1.67132065e-01 -8.75137150e-01 1.50422782e-01
7.27440655e-01 4.67125118e-01 1.75191417e-01 -3.64899963e-01
-3.26348901e-01 1.09075427e+00 -8.80628452e-02 5.19679248e-01
-9.04853165e-01 -4.44122374e-01 1.04657769e+00 -9.10039470e-02
1.94360301e-01 -1.31458789e-01 -2.67096758e-01 -9.91470754e-01
8.93489495e-02 -1.24873912e+00 2.39541724e-01 -3.81041020e-01
-1.53044319e+00 7.98418999e-01 -2.65921265e-01 -1.39488614e+00
-4.52363551e-01 -6.60520852e-01 -4.85100031e-01 1.41110122e+00
-1.48479724e+00 -8.98003936e-01 2.13961601e-01 9.49302852e-01
7.10197270e-01 -8.18407953e-01 1.14828181e+00 6.63006067e-01
-5.67908823e-01 1.52277792e+00 4.06922400e-01 5.22824049e-01
1.02907336e+00 -1.09448171e+00 5.98856270e-01 1.09035254e+00
1.54863134e-01 6.49204195e-01 5.23718536e-01 -1.18896231e-01
-1.20669687e+00 -1.14499044e+00 1.14502823e+00 -4.96445298e-01
4.57231134e-01 -6.75434649e-01 -1.06807435e+00 7.33436167e-01
3.45271885e-01 6.35901690e-02 1.15509105e+00 -1.44874565e-02
-8.43195319e-01 -4.64094549e-01 -1.68595946e+00 4.00374174e-01
5.90645611e-01 -1.28927147e+00 -9.11397099e-01 9.72310230e-02
6.88812733e-01 -1.99268579e-01 -9.10509408e-01 1.88542798e-01
3.20960760e-01 -4.00489479e-01 1.03404927e+00 -5.99832952e-01
-4.08062071e-01 -4.77763802e-01 -6.54669464e-01 -1.55992329e+00
-4.30278964e-02 -4.13806647e-01 1.39275834e-01 1.59182739e+00
6.89405441e-01 -1.05637634e+00 5.22669852e-01 5.63877642e-01
-3.74169230e-01 2.27500513e-01 -1.50890541e+00 -1.28003788e+00
4.90947515e-01 -5.53871930e-01 6.97655201e-01 1.21416783e+00
-3.30551416e-02 4.37082469e-01 -1.46437645e-01 6.05050683e-01
5.19974649e-01 -5.09994626e-01 7.05325842e-01 -9.26074028e-01
-1.43703043e-01 -2.96265960e-01 -5.57000339e-01 -9.14707482e-01
5.45768797e-01 -1.21220624e+00 1.05639689e-01 -6.43363178e-01
-4.37081575e-01 -3.46443683e-01 -5.96863806e-01 4.18739408e-01
4.96079661e-02 2.76379079e-01 4.78474759e-02 -1.29178807e-01
3.15245092e-01 6.27085268e-01 6.32478714e-01 -6.75036132e-01
-2.58306742e-01 8.66018683e-02 -5.46630621e-01 4.81071234e-01
8.55257928e-01 -5.97523808e-01 -3.15031409e-01 -2.34613284e-01
-8.74137461e-01 -2.73894101e-01 -5.73218428e-02 -9.47374642e-01
3.28918211e-02 3.49546731e-01 2.65169472e-01 -2.54862189e-01
4.76523340e-01 -9.32321608e-01 -1.14871927e-01 5.19474804e-01
-4.34052110e-01 -1.65966153e-01 6.96587622e-01 3.82664293e-01
-5.43531835e-01 -1.25307053e-01 1.21805787e+00 5.64234436e-01
-6.78568244e-01 6.09568022e-02 -2.57743329e-01 1.18218502e-02
9.47631001e-01 -2.18596861e-01 -6.37500510e-02 -2.97597677e-01
-8.86771500e-01 -1.28518507e-01 2.21455395e-01 7.28967905e-01
3.26195747e-01 -1.69604945e+00 -1.06329727e+00 8.36770117e-01
2.61207819e-01 -5.03901780e-01 2.77864337e-01 3.40615690e-01
-7.29150996e-02 5.04294455e-01 -2.78152749e-02 -6.70899212e-01
-1.61527598e+00 5.56057572e-01 5.22922933e-01 1.50723785e-01
-9.74325463e-02 1.05250680e+00 3.00491542e-01 -8.91627789e-01
4.01676029e-01 -2.81522423e-01 1.60282150e-01 -8.94994512e-02
6.70319676e-01 2.00034827e-01 4.15750444e-01 -1.06344795e+00
-7.37099588e-01 3.98236245e-01 -1.55332521e-01 -4.05886710e-01
8.96494150e-01 -1.07934378e-01 4.76714879e-01 3.69645894e-01
1.52486801e+00 5.38392007e-01 -8.74621987e-01 -5.51608026e-01
-1.47258252e-01 -2.37267792e-01 3.79835181e-02 -9.20347452e-01
-9.01910722e-01 9.60303366e-01 1.27138937e+00 1.89170271e-01
1.01977921e+00 9.83828902e-02 5.97247005e-01 3.38243209e-02
1.38178572e-01 -1.01700783e+00 -3.45392972e-01 4.99890596e-01
1.10207379e+00 -1.36171114e+00 -4.80950236e-01 -9.32467431e-02
-8.34398508e-01 6.41473234e-01 4.69821125e-01 1.79210842e-01
8.77202988e-01 2.67668128e-01 5.42456329e-01 3.67384911e-01
-2.34257564e-01 2.37742081e-01 3.18700790e-01 9.29555774e-01
4.55836117e-01 2.69374043e-01 1.59696832e-01 6.84597731e-01
-5.64327776e-01 -5.87977171e-01 8.86433572e-02 6.13251746e-01
-1.37500083e-02 -1.44912744e+00 -8.24598670e-01 -1.41653076e-01
-3.94417524e-01 -1.48111284e-01 -3.62948298e-01 7.17896700e-01
8.68308842e-02 1.11599302e+00 1.03859361e-02 -5.06815851e-01
5.53425670e-01 8.94837856e-01 1.13199644e-01 -5.21960974e-01
-5.17934740e-01 -1.87566280e-01 -3.46068107e-02 4.34826910e-02
-1.19917989e-01 -8.27220023e-01 -1.10406458e+00 -3.25076371e-01
-4.12794352e-01 3.00084978e-01 1.12862849e+00 7.55419850e-01
3.45406711e-01 4.25513357e-01 1.04438651e+00 -3.47003460e-01
-9.47587013e-01 -1.26710844e+00 -6.50946200e-01 4.49005485e-01
6.16557896e-01 -4.31696922e-01 -7.22693801e-01 3.47334653e-01] | [14.325096130371094, 6.222207069396973] |
12c76d5e-ebfd-46ff-9750-d9a6fc31924c | data-driven-perception-of-neuron-point | 1811.00688 | null | http://arxiv.org/abs/1811.00688v2 | http://arxiv.org/pdf/1811.00688v2.pdf | Data-driven Perception of Neuron Point Process with Unknown Unknowns | Identification of patterns from discrete data time-series for statistical
inference, threat detection, social opinion dynamics, brain activity prediction
has received recent momentum. In addition to the huge data size, the associated
challenges are, for example, (i) missing data to construct a closed
time-varying complex network, and (ii) contribution of unknown sources which
are not probed. Towards this end, the current work focuses on statistical
neuron system model with multi-covariates and unknown inputs. Previous research
of neuron activity analysis is mainly limited with effects from the spiking
history of target neuron and the interaction with other neurons in the system
while ignoring the influence of unknown stimuli. We propose to use unknown
unknowns, which describes the effect of unknown stimuli, undetected neuron
activities and all other hidden sources of error. The maximum likelihood
estimation with the fixed-point iteration method is implemented. The
fixed-point iterations converge fast, and the proposed methods can be
efficiently parallelized and offer computational advantage especially when the
input spiking trains are over long time-horizon. The developed framework
provides an intuition into the meaning of having extra degrees-of-freedom in
the data to support the need for unknowns. The proposed algorithm is applied to
simulated spike trains and on real-world experimental data of mouse
somatosensory, mouse retina and cat retina. The model shows a successful
increasing of system likelihood with respect to the conditional intensity
function, and it also reveals the convergence with iterations. Results suggest
that the neural connection model with unknown unknowns can efficiently estimate
the statistical properties of the process by increasing the network likelihood. | ['Paul Bogdan', 'Gaurav Gupta', 'Ruochen Yang'] | 2018-11-02 | null | null | null | null | ['activity-prediction', 'activity-prediction'] | ['computer-vision', 'time-series'] | [ 4.98167753e-01 -2.59751946e-01 2.68339306e-01 -3.91384140e-02
4.49958555e-02 -2.67172456e-01 4.34065044e-01 -1.58077925e-01
-6.15107238e-01 1.13518500e+00 -1.95957944e-01 -9.85077396e-03
-4.93028849e-01 -3.10936630e-01 -7.76513159e-01 -1.29699802e+00
-3.04351836e-01 2.10050374e-01 1.79256201e-01 1.38255730e-01
3.76932442e-01 3.37062418e-01 -1.58728361e+00 -3.47901672e-01
6.49272561e-01 9.27478075e-01 4.07978088e-01 4.66792971e-01
2.30843633e-01 6.03153467e-01 -3.94773126e-01 2.47529984e-01
1.87261671e-01 -3.56213510e-01 1.39184952e-01 9.49239358e-02
-3.26153547e-01 5.38631044e-02 -2.49869540e-01 1.07261705e+00
5.79696715e-01 -1.58834308e-01 8.46338391e-01 -1.62822223e+00
-2.09007543e-02 4.49351579e-01 -5.98380089e-01 3.96522373e-01
-4.76313680e-01 4.12301630e-01 3.55066419e-01 -6.26204014e-01
4.07057226e-01 1.09153080e+00 5.48745453e-01 3.74046057e-01
-1.47577989e+00 -1.01402605e+00 2.86893070e-01 2.89330967e-02
-1.49931681e+00 -2.26598740e-01 6.62930787e-01 -8.19303393e-01
7.21789241e-01 -1.16327502e-01 7.06077695e-01 1.35222483e+00
6.69111371e-01 2.71782666e-01 1.40637589e+00 9.50118750e-02
6.17198884e-01 4.23030630e-02 6.13797307e-01 2.50188977e-01
3.84501308e-01 3.48923951e-01 -4.36266154e-01 -4.71819550e-01
9.53668773e-01 2.17850968e-01 -2.21289277e-01 4.44811359e-02
-1.23947132e+00 4.51990217e-01 -1.73042834e-01 1.83155164e-01
-6.89416409e-01 2.62594551e-01 7.79982731e-02 2.77486682e-01
5.03722787e-01 1.91923659e-02 -8.26733410e-01 -5.02432808e-02
-7.23817647e-01 1.46516249e-01 8.87171447e-01 6.38483465e-01
7.42867887e-01 2.99230814e-01 -1.39450788e-01 6.74258113e-01
2.91989684e-01 8.58226478e-01 1.91548347e-01 -8.66733909e-01
2.59365551e-02 4.78956759e-01 7.82055929e-02 -8.59309793e-01
-6.31583571e-01 -7.98225701e-01 -1.39583540e+00 3.81946802e-01
5.66142797e-01 -5.50040364e-01 -8.72144401e-01 2.08860874e+00
1.20217949e-01 9.27081168e-01 -5.00828177e-02 6.29040718e-01
2.23661557e-01 8.22013676e-01 -8.47340152e-02 -9.37956154e-01
1.24263501e+00 -1.86724365e-01 -1.00361431e+00 -9.70077142e-02
2.64589161e-01 -3.42781425e-01 2.85762787e-01 5.61558485e-01
-8.27001870e-01 -2.72055924e-01 -6.54297352e-01 7.20075667e-01
-1.45333529e-01 1.92561671e-01 4.59525406e-01 1.17014669e-01
-9.44545627e-01 4.59815651e-01 -1.19701755e+00 -3.44536722e-01
4.19577986e-01 6.97243571e-01 -8.32408518e-02 2.37181142e-01
-8.23079526e-01 6.44013345e-01 2.47711703e-01 4.27943856e-01
-9.53024864e-01 -6.85707092e-01 -8.89516342e-03 -5.22869304e-02
2.78739303e-01 -6.33345306e-01 5.39911449e-01 -1.03514862e+00
-1.18251204e+00 2.95243204e-01 -5.79017401e-01 -4.46527362e-01
2.35782221e-01 3.59093130e-01 -2.25939408e-01 -3.92570138e-01
-1.07421175e-01 2.39749625e-01 7.48400867e-01 -9.54539478e-01
-3.80938441e-01 -5.83282769e-01 -6.73277974e-01 1.57720894e-02
-7.44711608e-02 -1.93790898e-01 3.50409821e-02 -5.78736305e-01
2.52438337e-01 -1.07068956e+00 -4.48853195e-01 2.33889092e-03
-1.53750643e-01 -1.98097795e-01 6.53048337e-01 -5.68030119e-01
9.74907041e-01 -2.19769502e+00 1.13339864e-01 3.43150109e-01
2.63551384e-01 -2.02109963e-01 -1.34321794e-01 4.92453665e-01
-5.75377084e-02 -1.71896622e-01 -3.68343502e-01 -1.80579439e-01
-4.30732161e-01 4.04250175e-01 -2.17573613e-01 5.32122850e-01
3.77188176e-01 5.29136896e-01 -5.50055623e-01 -9.97225270e-02
-7.44191855e-02 5.71121454e-01 -2.45228663e-01 1.58812106e-02
8.28300491e-02 7.49474704e-01 -4.78931069e-01 4.84052300e-01
7.44548619e-01 -4.55044508e-01 -1.38441756e-01 -1.92993015e-01
-5.56866467e-01 -2.90862113e-01 -1.55875254e+00 8.93288612e-01
8.05238858e-02 6.09926999e-01 1.59156829e-01 -1.25762713e+00
8.86442602e-01 4.02432233e-01 5.81791162e-01 -3.26930672e-01
4.10021156e-01 3.04856539e-01 5.60008705e-01 -4.50176090e-01
-3.55552733e-01 1.32142007e-01 4.25043404e-01 3.83988470e-01
2.37622574e-01 3.69251966e-01 8.90875012e-02 7.45278150e-02
1.35344732e+00 -1.76232368e-01 3.28914911e-01 -4.93236065e-01
2.00242326e-01 -4.01416361e-01 9.47295249e-01 8.56687665e-01
-1.52152792e-01 8.52701664e-02 6.59318328e-01 7.66251311e-02
-1.06924367e+00 -8.83415043e-01 -4.26216632e-01 4.84784871e-01
1.83156915e-02 3.54712725e-01 -6.20558262e-01 3.27861756e-01
-8.92261043e-02 2.15921551e-01 -7.35603452e-01 -5.32079414e-02
-3.61170769e-01 -1.32343912e+00 1.65182039e-01 2.29289323e-01
4.10483293e-02 -1.15199447e+00 -5.66537261e-01 4.91684943e-01
-1.28109738e-01 -9.77187455e-01 1.09406799e-01 6.47274792e-01
-1.09464467e+00 -8.72858644e-01 -6.73236251e-01 -5.09027958e-01
9.18219984e-01 -1.44159555e-01 4.54066247e-01 -1.62700996e-01
-3.94877255e-01 1.08196110e-01 1.65836051e-01 -6.87363505e-01
-4.08255458e-02 -4.38412726e-01 4.07907665e-01 2.51837313e-01
3.99213791e-01 -9.14354861e-01 -5.18955350e-01 4.07231539e-01
-8.73773992e-01 -5.88674005e-03 6.95493340e-01 1.07901859e+00
6.58076406e-01 2.29503810e-01 7.96160817e-01 -5.86031735e-01
4.79903340e-01 -8.51526380e-01 -7.99414217e-01 8.62935409e-02
-6.21805370e-01 1.81909904e-01 2.98433781e-01 -1.06889749e+00
-8.15850496e-01 1.14545584e-01 3.51723671e-01 -2.72605777e-01
-1.32240519e-01 5.59777617e-01 1.89023092e-01 -2.42435429e-02
3.65714967e-01 4.22851354e-01 2.23998085e-01 -2.54350752e-01
-3.91055912e-01 3.15861076e-01 1.24921165e-01 -2.76179820e-01
4.65205610e-01 5.63976169e-01 4.41127479e-01 -9.49906051e-01
-2.62889236e-01 -4.51325148e-01 -3.53012472e-01 -4.51225609e-01
4.54470634e-01 -1.05156112e+00 -1.12524629e+00 1.12193584e+00
-1.24003005e+00 -1.95491567e-01 1.26304582e-01 9.23822999e-01
-3.96154255e-01 -1.75413825e-02 -6.39228225e-01 -1.55034018e+00
-3.26647945e-02 -7.93520093e-01 5.34277380e-01 2.94008613e-01
6.28801957e-02 -9.21180725e-01 1.22395590e-01 -1.75984532e-01
3.11351061e-01 2.87535965e-01 8.76293480e-01 -5.68661392e-01
-7.61185288e-01 -1.06908754e-01 -5.28256297e-02 2.35062137e-01
-3.68878804e-02 3.65253478e-01 -9.09164846e-01 -3.40080053e-01
5.21789432e-01 8.84259269e-02 7.27606535e-01 1.09282029e+00
7.37244666e-01 -3.49862099e-01 -5.69321394e-01 3.14040333e-01
1.49958503e+00 5.07230520e-01 4.28185433e-01 -2.20638737e-01
2.40805030e-01 8.05985510e-01 2.64527410e-01 8.62465262e-01
3.81538714e-03 1.14825942e-01 5.75464427e-01 5.53525798e-02
5.18222630e-01 8.47373530e-02 3.71986002e-01 1.25922239e+00
-1.43914551e-01 -4.71302897e-01 -7.93612659e-01 6.24119639e-01
-2.16916776e+00 -1.05811143e+00 -5.21857977e-01 2.28237987e+00
4.90473866e-01 1.38942271e-01 1.34006394e-02 2.02084586e-01
8.55038822e-01 -4.64122683e-01 -1.07329297e+00 3.08932036e-01
-7.17783332e-01 -1.73539538e-02 6.95080936e-01 3.65656942e-01
-4.55292434e-01 3.66190970e-01 6.32474899e+00 6.76501513e-01
-1.08515644e+00 -4.38187048e-02 6.07004225e-01 -6.89873397e-02
3.94036761e-03 2.47222498e-01 -1.04764295e+00 6.98062301e-01
9.03228045e-01 -2.59377807e-01 6.52703822e-01 2.49731503e-02
7.18843520e-01 -3.86845320e-01 -8.62692952e-01 1.04398084e+00
-1.99688569e-01 -9.14136171e-01 -4.26128805e-01 4.48594779e-01
6.90762222e-01 4.64696199e-01 -1.95913389e-02 4.01306190e-02
1.28813982e-01 -7.07691848e-01 3.75843465e-01 1.19140792e+00
7.16997534e-02 -5.01973867e-01 7.82084048e-01 9.91065741e-01
-8.14214408e-01 -2.89434493e-01 -4.34541672e-01 -6.51764214e-01
1.87997758e-01 1.00317991e+00 -6.85655534e-01 -1.26299933e-01
4.99673784e-01 8.41901183e-01 -2.52143323e-01 1.30119789e+00
2.32818827e-01 1.10626328e+00 -8.78203034e-01 -4.75894630e-01
-7.11367205e-02 -5.37348211e-01 8.00063193e-01 7.60304391e-01
6.58291519e-01 3.92578959e-01 -1.98212489e-01 1.12261033e+00
6.21331513e-01 -1.09596178e-01 -6.10795796e-01 -2.43275687e-01
5.64559817e-01 1.20352292e+00 -8.97345185e-01 -7.84844626e-03
-3.21722180e-01 2.97083735e-01 1.74479544e-01 9.46411014e-01
-5.82709908e-01 4.90861535e-02 5.60135305e-01 1.85462415e-01
2.43413165e-01 -2.93189138e-01 -3.11460316e-01 -1.04639757e+00
8.50959197e-02 -5.43796301e-01 -3.04619595e-02 -7.68884838e-01
-1.56221640e+00 3.52044761e-01 -9.80174169e-02 -1.05116999e+00
-1.69473514e-01 -7.49154627e-01 -5.34183562e-01 1.01835382e+00
-1.14896190e+00 -5.57782888e-01 4.38278094e-02 6.56638980e-01
1.75244719e-01 -9.93211269e-02 5.75041056e-01 2.36911565e-01
-7.25070775e-01 -6.75570294e-02 4.86442894e-01 -1.89005837e-01
3.32617551e-01 -6.04901731e-01 -5.45253940e-02 7.65545666e-01
-2.37715289e-01 6.33653283e-01 1.10752404e+00 -8.43687177e-01
-1.06437230e+00 -6.49324179e-01 6.47573650e-01 -2.08241925e-01
9.15076673e-01 -5.41228890e-01 -1.05484235e+00 1.87596560e-01
-5.69063686e-02 2.39461418e-02 5.35540402e-01 -1.90212429e-01
2.11503550e-01 -9.59960297e-02 -8.82488489e-01 5.58938205e-01
7.46778965e-01 -2.30217800e-01 -1.36649877e-01 1.71657745e-03
2.95546830e-01 2.42862493e-01 -5.71578681e-01 5.62358677e-01
5.65775335e-01 -4.89327133e-01 4.28842306e-01 -4.79425520e-01
-1.64225638e-01 -4.54225123e-01 4.88494374e-02 -1.16136980e+00
-4.22186941e-01 -6.19897068e-01 -5.26764989e-02 1.19952154e+00
3.99865717e-01 -1.05144536e+00 5.51550150e-01 6.62566185e-01
3.57044458e-01 -6.11415744e-01 -1.18288088e+00 -6.10363126e-01
-1.97898179e-01 -3.33365738e-01 -1.09623931e-01 8.31421971e-01
-3.87808532e-02 2.80368000e-01 -5.59187293e-01 3.91018569e-01
1.03407753e+00 -4.11479026e-01 5.22517502e-01 -1.72841847e+00
-3.11715782e-01 -1.33715481e-01 -6.67333186e-01 -1.00555587e+00
7.84106553e-02 -3.81574333e-01 1.70666143e-01 -1.04483056e+00
5.34531236e-01 -3.02612960e-01 -6.57480478e-01 2.40412369e-01
-1.51875854e-01 -2.99947530e-01 -2.86380917e-01 3.06262374e-01
-7.34741762e-02 5.21349967e-01 1.02214539e+00 2.12328717e-01
8.69017933e-03 4.33466136e-01 -1.83083937e-01 6.86088920e-01
7.18472958e-01 -1.01209104e+00 -4.78573114e-01 -1.48590982e-01
3.63731325e-01 4.63337481e-01 7.80194044e-01 -9.31075931e-01
6.68490589e-01 -3.45431417e-01 1.66856915e-01 -8.24926972e-01
5.73676467e-01 -9.71960962e-01 5.84179401e-01 5.98608911e-01
-7.52030089e-02 -1.08139031e-03 1.60887361e-01 1.10604191e+00
-1.76425248e-01 -1.39123842e-01 5.46282411e-01 -5.53965895e-03
-1.99583679e-01 4.00138795e-01 -9.33781803e-01 -2.56942119e-02
8.53376746e-01 -3.31903249e-01 -2.10748419e-01 -3.46532196e-01
-8.96757185e-01 2.94366151e-01 4.52791294e-03 8.72715116e-02
4.33558792e-01 -9.58061934e-01 -7.64610410e-01 3.44516337e-01
-8.77124369e-02 -4.51418966e-01 4.08089042e-01 1.29080582e+00
1.91931561e-01 1.82805657e-01 -3.23032856e-01 -8.82571697e-01
-1.18413579e+00 3.47070158e-01 2.69522786e-01 1.74593955e-01
7.20206797e-02 6.80839777e-01 4.52685505e-01 -1.16388746e-01
2.92699665e-01 -5.11894464e-01 -3.26133162e-01 8.47291574e-02
2.51276135e-01 4.24081743e-01 -3.56188506e-01 -3.93214792e-01
-1.60256281e-01 6.25983894e-01 2.33848527e-01 -2.18649209e-01
1.36075628e+00 -3.84216756e-01 -3.56313020e-01 1.20580006e+00
9.13352370e-01 -5.49710214e-01 -1.56892622e+00 -4.87770945e-01
-5.91134764e-02 7.47428536e-02 1.15521364e-01 -7.03060389e-01
-7.79352546e-01 9.32879925e-01 9.35160339e-01 1.47074133e-01
9.21987414e-01 -2.69872725e-01 1.61670148e-02 3.55085671e-01
5.58508873e-01 -9.39621627e-01 -1.42187029e-01 6.14766121e-01
5.38754046e-01 -1.04021633e+00 -2.21822664e-01 -2.53346473e-01
-4.01725799e-01 9.04997766e-01 5.67131042e-01 -4.32200700e-01
1.22893000e+00 6.82585895e-01 -6.54026419e-02 -6.21753819e-02
-1.45972693e+00 1.07829429e-01 1.66648664e-02 4.50966746e-01
1.92462683e-01 -3.37141082e-02 -5.11337221e-01 7.21616209e-01
3.75902355e-01 4.39334303e-01 5.28884828e-01 6.99099660e-01
-4.64868695e-01 -6.19608462e-01 -2.27370173e-01 7.18860030e-01
-5.32301843e-01 -3.02291214e-01 9.74005368e-03 4.77755547e-01
1.38301671e-01 8.22987676e-01 2.21128911e-01 1.60474360e-01
1.67777181e-01 -7.68176690e-02 1.78150281e-01 -2.84743905e-01
-3.51569355e-01 4.44527030e-01 -4.30085152e-01 -1.56853512e-01
-5.99166393e-01 -1.01101601e+00 -1.08825839e+00 -1.97192915e-02
-8.23872089e-01 6.26403019e-02 7.93190658e-01 1.10441458e+00
4.18421417e-01 4.81643319e-01 5.92281103e-01 -6.52267337e-01
-5.56822956e-01 -1.12771356e+00 -8.39560866e-01 -2.44663712e-02
4.78295922e-01 -7.34027147e-01 -1.06013012e+00 2.04914108e-01] | [6.880707263946533, 3.6932973861694336] |
a0435090-35f6-400c-858b-78ac25598f03 | learning-large-euclidean-margin-for-sketch | 1812.04275 | null | https://arxiv.org/abs/1812.04275v2 | https://arxiv.org/pdf/1812.04275v2.pdf | Domain-Aware SE Network for Sketch-based Image Retrieval with Multiplicative Euclidean Margin Softmax | This paper proposes a novel approach for Sketch-Based Image Retrieval (SBIR), for which the key is to bridge the gap between sketches and photos in terms of the data representation. Inspired by channel-wise attention explored in recent years, we present a Domain-Aware Squeeze-and-Excitation (DASE) network, which seamlessly incorporates the prior knowledge of sample sketch or photo into SE module and make the SE module capable of emphasizing appropriate channels according to domain signal. Accordingly, the proposed network can switch its mode to achieve a better domain feature with lower intra-class discrepancy. Moreover, while previous works simply focus on minimizing intra-class distance and maximizing inter-class distance, we introduce a loss function, named Multiplicative Euclidean Margin Softmax (MEMS), which introduces multiplicative Euclidean margin into feature space and ensure that the maximum intra-class distance is smaller than the minimum inter-class distance. This facilitates learning a highly discriminative feature space and ensures a more accurate image retrieval result. Extensive experiments are conducted on two widely used SBIR benchmark datasets. Our approach achieves better results on both datasets, surpassing the state-of-the-art methods by a large margin. | ['Yanwei Fu', 'Guodong Guo', 'Wenming Yang', 'Hangyu Lin', 'Peng Lu', 'Gao Huang'] | 2018-12-11 | null | null | null | null | ['sketch-based-image-retrieval'] | ['computer-vision'] | [ 3.02347243e-01 -5.99979162e-01 -4.19730693e-01 -4.18164462e-01
-8.64519119e-01 -4.42965060e-01 6.69339418e-01 -1.56117946e-01
-4.04436499e-01 4.31925774e-01 1.50676221e-01 1.07549362e-01
-4.31740403e-01 -7.28325069e-01 -5.70757866e-01 -7.21530259e-01
3.11113745e-01 -2.15379804e-01 9.82924178e-02 -1.23065719e-02
4.96195108e-01 4.66260970e-01 -1.33371329e+00 2.88807839e-01
7.54508078e-01 1.40024543e+00 3.40317160e-01 7.40704313e-02
-3.01664799e-01 3.80339712e-01 -4.97386247e-01 -2.54223973e-01
3.79219443e-01 -4.20385748e-01 -3.51380169e-01 2.00832337e-01
5.41484177e-01 -3.32313418e-01 -6.44679248e-01 1.01941907e+00
6.60781741e-01 2.35273555e-01 8.22245896e-01 -1.33020401e+00
-1.13667750e+00 2.75797844e-01 -6.66709006e-01 -9.28030163e-02
1.02435956e-02 -9.33565572e-02 1.20104265e+00 -1.13920176e+00
5.66593826e-01 1.21568358e+00 1.74387291e-01 3.68667871e-01
-1.19814050e+00 -1.02558637e+00 3.47314477e-01 3.02961528e-01
-1.82113731e+00 -3.05362582e-01 1.34589744e+00 -1.63471952e-01
3.67082983e-01 1.51015952e-01 3.24422419e-01 9.63940918e-01
-1.96564451e-01 1.06492591e+00 8.10267031e-01 -3.50020975e-01
8.77600014e-02 2.10682586e-01 -1.99285179e-01 4.72017467e-01
-2.34013319e-01 -1.02712102e-01 -6.28472388e-01 -1.97078455e-02
1.04141402e+00 4.25989538e-01 -4.65967566e-01 -6.50056243e-01
-1.33040798e+00 7.94618309e-01 7.48779058e-01 4.75717247e-01
-3.26144904e-01 5.41343763e-02 2.85195470e-01 3.03581625e-01
1.26900584e-01 5.76474816e-02 -5.66996261e-02 1.85132429e-01
-9.06750083e-01 2.13017743e-02 2.40474880e-01 8.69838357e-01
7.04907179e-01 -1.25872925e-01 -4.33388770e-01 1.19178343e+00
3.98982048e-01 4.79360700e-01 3.43201816e-01 -6.51506007e-01
4.99474972e-01 6.67768717e-01 -4.02818695e-02 -1.21006405e+00
2.06867427e-01 -4.65438932e-01 -9.27418113e-01 4.00474481e-02
1.78090364e-01 3.25635642e-01 -8.02941859e-01 1.72013879e+00
1.63014773e-02 2.77517051e-01 -1.20797537e-01 1.32506919e+00
8.40897143e-01 8.02159369e-01 4.05912958e-02 4.30490412e-02
1.05049849e+00 -9.26041663e-01 -6.22718751e-01 -1.05781920e-01
-2.95917373e-02 -9.14205015e-01 1.28490114e+00 4.26133066e-01
-9.29182053e-01 -8.15966725e-01 -1.26116288e+00 -2.13784531e-01
-3.66395563e-01 6.56902075e-01 3.92064720e-01 2.28305399e-01
-6.39935732e-01 4.36760783e-01 -3.38341862e-01 -7.94333369e-02
4.67969567e-01 2.27223486e-01 -4.72234219e-01 -2.05650717e-01
-1.11801839e+00 3.87141794e-01 1.28149569e-01 1.72377050e-01
-5.36532819e-01 -5.72002769e-01 -6.63237572e-01 2.58503020e-01
3.77299756e-01 -2.95091957e-01 6.58559680e-01 -1.00415957e+00
-1.54457653e+00 7.46558070e-01 -1.38328469e-03 9.38377455e-02
4.18699533e-01 2.11545490e-02 -5.49382031e-01 1.82613462e-01
-1.35234207e-01 9.05126810e-01 1.13749588e+00 -1.28887224e+00
-3.55234325e-01 -3.76576453e-01 -8.76869410e-02 1.92153364e-01
-8.82070661e-01 -2.41337687e-01 -9.26859796e-01 -9.11134839e-01
1.38027251e-01 -6.50669158e-01 1.76683769e-01 6.72955692e-01
-9.61428434e-02 -4.26233023e-01 9.82315600e-01 -5.71914017e-01
1.40604401e+00 -2.38436389e+00 2.19860524e-01 2.73985028e-01
6.79483265e-02 4.21172678e-01 -5.34336150e-01 5.22426188e-01
2.72837859e-02 -1.96611106e-01 -3.08722049e-01 -2.52615422e-01
1.23228483e-01 6.83781728e-02 -4.97086525e-01 4.21864182e-01
4.35987979e-01 8.29833806e-01 -6.07315004e-01 -5.04251838e-01
3.73023868e-01 7.53660381e-01 -4.14826334e-01 1.76742360e-01
2.19534356e-02 2.35204130e-01 -5.60842812e-01 6.16294980e-01
1.03630102e+00 -1.89984515e-01 -2.45613810e-02 -3.91798407e-01
-5.75489067e-02 8.49727914e-02 -1.36573374e+00 2.20799398e+00
-5.59264362e-01 6.09466672e-01 1.61173001e-01 -1.13871145e+00
1.45237577e+00 -7.87612423e-02 2.95556158e-01 -1.04945803e+00
4.20649461e-02 2.08942279e-01 -3.34854424e-01 -1.52244791e-01
3.82465035e-01 2.85522658e-02 2.61514224e-02 1.31123289e-01
-9.94525403e-02 1.67550221e-01 -1.54079095e-01 2.21222684e-01
6.19780779e-01 1.89686697e-02 -3.57265398e-02 -2.05854356e-01
8.75286102e-01 -5.86545169e-01 5.38458645e-01 4.21904743e-01
-1.35235995e-01 7.36949801e-01 2.96419591e-01 -2.60447979e-01
-9.29350257e-01 -1.20096707e+00 -1.86781809e-01 9.07878697e-01
8.33283246e-01 -1.77789956e-01 -2.77145416e-01 -6.63394570e-01
1.85271740e-01 2.61761039e-01 -5.95967531e-01 -2.76829541e-01
-3.83951128e-01 -2.10419923e-01 2.64822394e-01 5.76853454e-01
7.61984229e-01 -9.86723304e-01 -1.25442505e-01 -4.30200621e-02
-4.69878353e-02 -8.66839349e-01 -9.16816294e-01 -2.85656244e-01
-5.66244781e-01 -6.95111334e-01 -1.16743875e+00 -1.09721518e+00
6.20697558e-01 6.41808271e-01 6.70186818e-01 1.07929297e-01
-4.39025879e-01 2.78125644e-01 -4.26548511e-01 4.67961840e-02
3.42856586e-01 3.86592746e-02 -4.78393435e-01 3.31635863e-01
4.98596311e-01 -5.75442493e-01 -1.01729250e+00 3.84672850e-01
-1.09858143e+00 -1.30407587e-01 8.69813263e-01 1.08580136e+00
6.65084064e-01 -6.11406192e-02 6.96024537e-01 -1.87891155e-01
6.69826150e-01 -3.55961829e-01 -5.63068807e-01 5.67541003e-01
-6.62304819e-01 1.54333875e-01 6.85985208e-01 -5.88537931e-01
-8.15901041e-01 -1.47058487e-01 7.06851408e-02 -7.30392218e-01
1.40317068e-01 2.76090145e-01 -4.67574179e-01 -2.29119554e-01
2.35371649e-01 6.51722014e-01 -7.97319263e-02 -4.95832890e-01
3.87644768e-01 9.79154587e-01 4.13393080e-01 -4.97291058e-01
7.00037122e-01 5.80095887e-01 -3.47235277e-02 -7.03052342e-01
-3.11991662e-01 -6.28116608e-01 -4.62876469e-01 -6.23289160e-02
5.65376461e-01 -9.96446609e-01 -8.28075528e-01 4.49199229e-01
-9.47349548e-01 3.69524918e-02 1.19137615e-01 4.23978359e-01
-1.11160643e-01 6.05985403e-01 -3.50081563e-01 -7.78370261e-01
-4.93279278e-01 -1.14246118e+00 1.34838045e+00 3.82410914e-01
2.68865258e-01 -6.80359423e-01 -1.25012532e-01 3.43055241e-02
5.70725143e-01 -1.01634003e-01 8.05789649e-01 -3.73297065e-01
-7.32037485e-01 -1.75419986e-01 -7.76573539e-01 4.67827499e-01
1.23580210e-01 -9.10997093e-02 -8.16343665e-01 -3.96186054e-01
-4.35089886e-01 -3.31828058e-01 1.19216156e+00 9.13179740e-02
1.51448715e+00 -1.47804663e-01 -2.94281363e-01 5.57424247e-01
1.61801302e+00 1.79339677e-01 7.90783703e-01 2.64179856e-01
4.37219948e-01 3.85161847e-01 7.27836072e-01 5.00052452e-01
4.09832567e-01 9.60319400e-01 2.98121482e-01 -9.15606841e-02
-1.26578003e-01 -5.58125436e-01 6.72887117e-02 4.64121997e-01
4.11994517e-01 -3.10870707e-01 -3.16364884e-01 6.09667540e-01
-1.80777800e+00 -8.23938072e-01 4.53051835e-01 2.34985471e+00
7.03343451e-01 -1.41567603e-01 2.27482654e-02 1.52313530e-01
8.18631530e-01 5.34549057e-01 -5.91163278e-01 -9.58521068e-02
-1.07654415e-01 2.40266830e-01 1.98782653e-01 3.06419224e-01
-1.17330527e+00 8.00556183e-01 4.67002392e+00 1.33225894e+00
-1.47957861e+00 -2.41014168e-01 3.88029039e-01 -3.65981571e-02
-4.66837496e-01 -1.20533019e-01 -4.49430287e-01 7.63186157e-01
4.89534698e-02 1.45229965e-01 6.83285594e-01 9.00447845e-01
-1.60468921e-01 6.85945824e-02 -9.19007659e-01 1.36976850e+00
2.17878908e-01 -1.29763603e+00 1.87617868e-01 -8.48413929e-02
6.41798317e-01 -3.35753083e-01 2.68410236e-01 2.50436515e-01
-3.59226108e-01 -7.79229343e-01 6.43407166e-01 5.70944190e-01
9.95278299e-01 -9.13506210e-01 4.12614971e-01 1.65312320e-01
-1.43091774e+00 -8.35790262e-02 -6.15137994e-01 2.53144234e-01
-2.04646349e-01 5.08001864e-01 -1.40075371e-01 6.01656020e-01
5.56802869e-01 9.89093542e-01 -2.53722757e-01 1.05633163e+00
-1.59788430e-01 1.66657448e-01 -6.41854703e-02 -8.98573473e-02
2.78498769e-01 -4.58488017e-01 4.54809606e-01 1.16217601e+00
2.26001263e-01 2.15217248e-02 2.37234816e-01 1.03051925e+00
-2.80719399e-01 1.69161856e-01 -3.30231160e-01 -2.39729770e-02
7.55457044e-01 1.34520793e+00 -4.37411994e-01 -1.93789035e-01
-3.59039456e-01 1.36237335e+00 1.23082571e-01 4.53446597e-01
-6.62592113e-01 -9.14066732e-01 6.42452300e-01 -1.63902283e-01
6.76893055e-01 -9.80614945e-02 -1.51507765e-01 -1.12510300e+00
3.27004880e-01 -6.27169430e-01 2.95119345e-01 -7.59524405e-01
-1.51309383e+00 5.04162967e-01 -4.65876609e-01 -1.57034636e+00
2.24940240e-01 -6.05225503e-01 -6.11177981e-01 9.94747400e-01
-1.98899615e+00 -1.43301630e+00 -4.50732172e-01 6.73272491e-01
5.26952207e-01 -8.59152600e-02 5.12656569e-01 7.00269580e-01
-4.34195638e-01 9.89922464e-01 3.22530746e-01 2.96589583e-01
1.17636943e+00 -7.85846472e-01 -2.89437678e-02 5.34791827e-01
2.07821414e-01 7.49257445e-01 8.91972110e-02 -3.37137580e-01
-1.50997615e+00 -8.99731457e-01 6.61881804e-01 6.44873008e-02
4.21565831e-01 -4.79775697e-01 -9.01497006e-01 1.45670101e-01
-8.52655154e-03 2.29105040e-01 3.63245100e-01 -1.80347607e-01
-8.15324247e-01 -4.94930327e-01 -1.12890732e+00 4.44237649e-01
1.06813812e+00 -9.30230737e-01 -2.56280005e-01 -7.12542832e-02
4.52769607e-01 -1.81099415e-01 -7.90829837e-01 4.48026031e-01
1.00007617e+00 -7.63764918e-01 1.16500580e+00 -3.50487322e-01
6.32360280e-01 -5.41780829e-01 -3.09489191e-01 -1.07569993e+00
-3.93636912e-01 -2.42626384e-01 1.83668807e-01 1.46447492e+00
3.50158708e-03 -4.21150744e-01 5.76403260e-01 3.57786566e-01
1.43315405e-01 -8.69761646e-01 -8.01332295e-01 -7.54290819e-01
-1.31895803e-02 -1.18898183e-01 6.43494248e-01 9.34231222e-01
-1.67575613e-01 1.19914643e-01 -5.73566735e-01 1.00383662e-01
5.38153827e-01 5.75162351e-01 6.58669293e-01 -1.04491341e+00
-8.65458623e-02 -7.25267351e-01 -5.43527842e-01 -1.37883234e+00
1.71176463e-01 -8.24578345e-01 -2.39550573e-04 -1.41583300e+00
4.85316306e-01 -6.70274854e-01 -6.71866477e-01 3.59682411e-01
-1.88542113e-01 5.32932341e-01 3.87753099e-01 2.67006487e-01
-7.23640621e-01 9.49589312e-01 1.27965903e+00 -3.97379756e-01
-1.18002027e-01 -3.01484942e-01 -7.22617030e-01 2.34015688e-01
3.76023680e-01 -2.73172766e-01 -4.59049106e-01 -5.76053739e-01
-8.36444423e-02 4.56748791e-02 5.85371673e-01 -7.26515472e-01
4.13767934e-01 -1.06564566e-01 5.17586946e-01 -6.19216681e-01
4.25543427e-01 -1.04252112e+00 -1.26062319e-01 1.10123947e-01
-6.82087958e-01 -3.93413603e-01 3.88832018e-02 7.09253311e-01
-5.99493563e-01 2.26303004e-02 7.66071975e-01 2.45843932e-01
-9.22806025e-01 4.33041006e-01 2.03614756e-01 -1.91548392e-01
8.80125701e-01 -1.99699059e-01 -3.21183741e-01 -4.34022009e-01
-2.36355156e-01 2.73534954e-01 3.81856620e-01 6.66758537e-01
8.10246885e-01 -1.68635488e+00 -5.61702669e-01 4.00969326e-01
4.63686496e-01 -3.28555942e-01 5.71598887e-01 4.92528170e-01
4.00867575e-04 5.27995050e-01 -2.53422618e-01 -5.14111638e-01
-1.10236692e+00 4.64995950e-01 7.50409290e-02 -4.67997640e-02
-4.07808036e-01 1.07190859e+00 3.36230904e-01 -3.78174931e-01
4.61212337e-01 6.15511574e-02 3.91015522e-02 1.96977500e-02
7.24183798e-01 3.08412731e-01 -9.56156254e-02 -4.49319065e-01
-5.24037123e-01 9.11486804e-01 -1.10758945e-01 -1.07619157e-02
1.27691162e+00 -1.54963464e-01 1.65193656e-03 6.65939301e-02
1.57445943e+00 2.82503385e-02 -1.49017131e+00 -5.38969576e-01
-4.39871490e-01 -9.48993325e-01 3.42617363e-01 -9.10137057e-01
-1.12618363e+00 1.10168517e+00 7.90623069e-01 -1.88408136e-01
1.42606759e+00 -4.63511571e-02 9.47714090e-01 2.08041951e-01
2.31888846e-01 -1.18359232e+00 4.68811542e-01 1.02598362e-01
1.08359647e+00 -1.35667288e+00 -4.59591784e-02 -2.78713048e-01
-6.30162477e-01 1.01553452e+00 5.88824034e-01 -4.58848596e-01
5.94777405e-01 -1.86297759e-01 -1.66431487e-01 7.63750896e-02
-4.74768758e-01 2.37451028e-02 6.78902984e-01 3.22233707e-01
2.40100890e-01 -4.64569144e-02 -4.73169714e-01 6.85274124e-01
4.67943996e-01 1.90879121e-01 -2.64792979e-01 7.22447991e-01
-3.19730192e-01 -1.28567767e+00 -2.29656309e-01 1.91130325e-01
-1.02250658e-01 -9.77202728e-02 -4.78023112e-01 7.00975299e-01
7.84975216e-02 7.32718408e-01 1.62352949e-01 -4.24436122e-01
2.62043506e-01 -2.13587299e-01 4.84541267e-01 -6.18330427e-02
-1.90948322e-01 2.23688513e-01 -4.66243476e-01 -3.69961441e-01
-3.86584312e-01 -2.35548154e-01 -1.02545846e+00 -1.70609534e-01
-5.01539350e-01 3.98572274e-02 6.96722031e-01 6.89442515e-01
6.07215345e-01 2.41793469e-01 1.05213320e+00 -7.00580657e-01
-5.65933883e-01 -7.40591228e-01 -6.97491407e-01 3.95247728e-01
2.62812674e-01 -6.46458268e-01 -2.47713670e-01 -2.14840382e-01] | [11.495368003845215, 0.7529169321060181] |
a2b9fde5-3f1b-47d3-9daa-15b9ed837676 | multi-dimensional-evaluation-of-text | 2306.01200 | null | https://arxiv.org/abs/2306.01200v1 | https://arxiv.org/pdf/2306.01200v1.pdf | Multi-Dimensional Evaluation of Text Summarization with In-Context Learning | Evaluation of natural language generation (NLG) is complex and multi-dimensional. Generated text can be evaluated for fluency, coherence, factuality, or any other dimensions of interest. Most frameworks that perform such multi-dimensional evaluation require training on large manually or synthetically generated datasets. In this paper, we study the efficacy of large language models as multi-dimensional evaluators using in-context learning, obviating the need for large training datasets. Our experiments show that in-context learning-based evaluators are competitive with learned evaluation frameworks for the task of text summarization, establishing state-of-the-art on dimensions such as relevance and factual consistency. We then analyze the effects of factors such as the selection and number of in-context examples on performance. Finally, we study the efficacy of in-context learning based evaluators in evaluating zero-shot summaries written by large language models such as GPT-3. | ['Chunting Zhou', 'Graham Neubig', 'PengFei Liu', 'Patrick Fernandes', 'Swarnashree Mysore Sathyendra', 'Vaishakh Keshava', 'Sameer Jain'] | 2023-06-01 | null | null | null | null | ['text-summarization'] | ['natural-language-processing'] | [ 6.74074590e-02 2.13855177e-01 -3.51920426e-01 -2.31636450e-01
-1.35855615e+00 -5.38384140e-01 1.03765130e+00 6.11395180e-01
-5.02489328e-01 1.00481439e+00 1.01714826e+00 -8.08300748e-02
-3.98612954e-02 -6.06196523e-01 -3.42880815e-01 -1.18540391e-01
1.25930950e-01 8.92081082e-01 -5.17829098e-02 -2.52251953e-01
4.87847388e-01 -7.94558004e-02 -1.53458571e+00 5.62152803e-01
1.36099005e+00 4.59771663e-01 -8.10206309e-03 1.15444314e+00
-2.25333661e-01 9.60372508e-01 -1.16344666e+00 -5.24915814e-01
-3.46963257e-01 -6.89724147e-01 -8.05308521e-01 -2.99235191e-02
7.80047834e-01 -4.21087205e-01 1.51894689e-01 4.44600523e-01
1.07306409e+00 2.72671312e-01 1.14131439e+00 -1.03339946e+00
-6.95296586e-01 8.92761409e-01 2.39791367e-02 2.97759622e-01
6.43538475e-01 3.52729321e-01 1.22946918e+00 -6.37216210e-01
9.07552958e-01 1.45524728e+00 5.10232747e-01 5.03465414e-01
-1.36206055e+00 -1.98473111e-01 -1.20227113e-01 7.50780106e-02
-6.33987665e-01 -7.92139411e-01 4.21943277e-01 -5.13470411e-01
1.16108000e+00 2.12369859e-01 4.80271041e-01 1.31769168e+00
2.22847387e-01 9.27905440e-01 8.89202535e-01 -7.11435556e-01
4.53477532e-01 1.25514194e-01 3.29276979e-01 4.06737328e-01
4.00786459e-01 -6.41572997e-02 -6.48098886e-01 -3.84409130e-01
8.51141065e-02 -7.87364900e-01 -2.03494564e-01 1.86334401e-02
-1.26871789e+00 1.04341674e+00 -2.55150795e-01 1.45464405e-01
-2.78006762e-01 1.00012690e-01 8.23696554e-01 1.84979290e-01
8.10298860e-01 1.33446705e+00 -3.33652437e-01 -5.99988103e-01
-1.27976131e+00 5.43042362e-01 9.90922809e-01 8.68390381e-01
4.09652919e-01 1.51544407e-01 -1.22734892e+00 1.04625463e+00
-3.23181719e-01 3.76138538e-01 9.90118742e-01 -1.22349870e+00
6.31853342e-01 4.91250575e-01 1.99185327e-01 -4.90630567e-01
-3.87279302e-01 -3.42183799e-01 -4.85933453e-01 -2.28529826e-01
2.21614842e-03 -6.67390823e-01 -4.54669744e-01 1.75176871e+00
-5.75326532e-02 -3.48539770e-01 3.90697241e-01 3.21482897e-01
8.96766126e-01 6.84917331e-01 1.58542857e-01 -5.44933796e-01
1.06858385e+00 -1.05364192e+00 -8.22944820e-01 -1.62994117e-01
1.11234534e+00 -8.47836792e-01 1.59661579e+00 8.96308124e-02
-1.35894251e+00 -4.02908564e-01 -8.41043234e-01 -2.24639490e-01
-1.70556232e-01 3.89580160e-01 2.05468372e-01 4.37151462e-01
-1.10029078e+00 6.78905606e-01 -3.92323315e-01 -3.75831008e-01
1.47650376e-01 -2.77902395e-01 2.77930982e-02 1.81781575e-01
-1.21109843e+00 1.16902435e+00 3.88945997e-01 -7.07560062e-01
-8.18012297e-01 -6.90131068e-01 -8.70131791e-01 1.52375996e-01
1.28627717e-01 -8.75625432e-01 1.68345332e+00 -4.59349006e-01
-1.18931544e+00 5.76434076e-01 -1.81472197e-01 -4.77015376e-01
5.70554495e-01 -3.43530983e-01 -2.79248118e-01 9.92522314e-02
5.01334906e-01 7.23693788e-01 3.81807595e-01 -9.67943549e-01
-3.71699512e-01 -1.78414151e-01 4.81276624e-02 6.28093839e-01
-5.83368063e-01 -2.22553089e-01 -6.82165520e-03 -5.80511570e-01
-6.21603727e-01 -7.27469206e-01 3.09527554e-02 -7.42763102e-01
-6.31919444e-01 -6.04237676e-01 5.60025334e-01 -6.90150023e-01
1.64270627e+00 -1.52802849e+00 -1.13759264e-02 -4.27890331e-01
-1.02434240e-01 2.78932482e-01 -5.15239179e-01 6.59777999e-01
3.78624976e-01 5.47597051e-01 1.01119339e-01 -5.77009559e-01
2.23856241e-01 -2.35746115e-01 -1.57254428e-01 -2.58867085e-01
3.02060455e-01 9.76456642e-01 -1.14054024e+00 -9.38821495e-01
-6.92091584e-02 1.51807219e-01 -5.77979267e-01 2.75166690e-01
-7.14901090e-01 6.80198148e-02 -3.52679282e-01 1.99620545e-01
-2.86592215e-01 -3.08749765e-01 6.07136190e-02 2.38297284e-02
-3.32128331e-02 7.26226985e-01 -6.47482634e-01 1.62912142e+00
-8.28701079e-01 9.06854510e-01 -6.16961360e-01 -4.03506398e-01
7.95822561e-01 3.74266893e-01 6.89416900e-02 -8.73049796e-01
-7.58910691e-03 4.67182696e-02 -4.72608469e-02 -5.52128136e-01
1.12731087e+00 -6.92298114e-02 -2.74122834e-01 1.10808086e+00
1.85703874e-01 -5.40359437e-01 1.08426714e+00 6.75651133e-01
1.09134734e+00 -1.38771638e-01 1.43994004e-01 -2.94815898e-01
7.29153156e-02 2.62114018e-01 9.92831811e-02 8.97896111e-01
6.94553256e-02 4.75135565e-01 7.94795811e-01 7.28162229e-02
-1.17540526e+00 -6.57101274e-01 -4.86539789e-02 1.23136532e+00
-3.72191191e-01 -7.71627486e-01 -9.97921050e-01 -5.79318523e-01
-2.23127186e-01 1.58721316e+00 -4.72186595e-01 -3.84736925e-01
-2.95267284e-01 -6.34392738e-01 6.62534714e-01 4.21268940e-01
1.40730843e-01 -1.27752459e+00 -7.74905384e-01 2.66295999e-01
-6.74057484e-01 -1.06451869e+00 -6.18316352e-01 -1.86658353e-01
-9.74376619e-01 -8.76633823e-01 -6.54148638e-01 -4.40194815e-01
2.01204434e-01 -8.97206590e-02 1.81062627e+00 -3.45282316e-01
-1.15200244e-01 6.83527470e-01 -3.77010584e-01 -6.06641531e-01
-9.10826325e-01 5.54780364e-01 -5.38112149e-02 -7.36884654e-01
1.19651996e-01 -4.39202964e-01 -3.48354101e-01 -3.48457992e-02
-7.94696152e-01 1.70743316e-01 5.18354475e-01 8.80191863e-01
4.14755046e-01 -2.70162731e-01 1.11232686e+00 -9.58529174e-01
1.93030477e+00 -1.75530270e-01 -6.60850331e-02 6.97376907e-01
-8.06798458e-01 5.01604080e-01 4.07729328e-01 -5.08816898e-01
-1.15654731e+00 -7.57262647e-01 3.67249489e-01 1.76723599e-02
1.99348718e-01 7.45575845e-01 2.01145649e-01 6.70433700e-01
1.25886142e+00 -4.10137996e-02 -1.78160533e-01 -7.76049197e-02
6.86211407e-01 6.27025247e-01 2.83908606e-01 -8.52302551e-01
9.91607681e-02 -2.58840173e-01 -3.44738930e-01 -7.72071183e-01
-1.02312601e+00 -1.85726061e-01 -2.80669183e-01 -2.25842044e-01
7.41281271e-01 -9.80190575e-01 -5.69116659e-02 4.20382321e-02
-1.21565962e+00 -6.52569711e-01 -7.14053690e-01 3.81454080e-01
-8.18511784e-01 2.07937419e-01 -5.67063212e-01 -6.17217839e-01
-9.07065392e-01 -9.65426624e-01 1.21655977e+00 1.61105394e-01
-8.51529419e-01 -1.13037002e+00 6.18952096e-01 4.57921237e-01
4.42003638e-01 2.85146326e-01 1.25182915e+00 -8.19425702e-01
-1.52409822e-01 -2.97652543e-01 7.16718510e-02 3.11514169e-01
-1.73514208e-03 1.06443226e-01 -7.46734321e-01 -3.86796147e-02
-1.83366477e-01 -9.97997940e-01 7.09100842e-01 4.78916943e-01
7.85480082e-01 -6.45345628e-01 -6.64108247e-02 -7.23827928e-02
1.05482960e+00 -3.50774050e-01 4.02448863e-01 1.19059615e-01
4.48933810e-01 7.74221778e-01 7.20358372e-01 7.21616268e-01
4.70656037e-01 4.77071911e-01 -3.05705518e-01 3.05064946e-01
-3.56737643e-01 -4.15848374e-01 4.17490780e-01 7.65664935e-01
2.29984224e-01 -5.92329443e-01 -9.16221142e-01 7.11557567e-01
-1.65173900e+00 -1.08499479e+00 1.84369072e-01 2.11056828e+00
1.25037909e+00 1.04263850e-01 1.81485593e-01 -2.05047011e-01
7.00996161e-01 2.77292162e-01 -4.08173561e-01 -7.79340148e-01
-3.26392740e-01 -7.67659582e-03 -2.07353309e-02 5.32380700e-01
-6.55081034e-01 8.34284425e-01 6.43302059e+00 8.88102710e-01
-8.14218163e-01 1.53994160e-02 9.32546020e-01 -5.81283808e-01
-6.55813456e-01 -1.30086005e-01 -7.78119206e-01 4.62401778e-01
1.22021592e+00 -7.89568841e-01 1.64226428e-01 7.18549550e-01
5.13504565e-01 -3.01099598e-01 -1.43163550e+00 6.36531413e-01
3.79713446e-01 -1.35592365e+00 3.39544952e-01 -1.74590964e-02
1.31789529e+00 -1.58161502e-02 -1.72689050e-01 7.63159513e-01
5.01278758e-01 -8.92686784e-01 6.44833982e-01 3.22064042e-01
8.74774039e-01 -6.34324610e-01 5.76043069e-01 5.70969343e-01
-4.36473072e-01 7.55403489e-02 -2.50309318e-01 1.29826561e-01
3.73530805e-01 6.88508928e-01 -1.16856086e+00 1.48141697e-01
-1.57174021e-02 2.74815440e-01 -9.26195323e-01 6.62318587e-01
-3.68214846e-01 5.69582820e-01 2.34858543e-02 -4.28241879e-01
1.68371812e-01 6.86672404e-02 5.58824301e-01 1.46182954e+00
3.78805995e-01 -1.04467981e-01 9.97995883e-02 7.28433073e-01
-3.39913934e-01 3.58823836e-01 -6.66386962e-01 -4.32586998e-01
5.54521501e-01 1.13313305e+00 -1.97182551e-01 -7.77882457e-01
1.96093112e-01 5.61577082e-01 4.78991389e-01 2.93998420e-01
-3.96306157e-01 -2.18272582e-01 1.66483924e-01 1.90706640e-01
-6.29464015e-02 5.91117982e-03 -4.55086470e-01 -1.08210111e+00
9.37382281e-02 -1.08538556e+00 3.29976529e-01 -8.47837746e-01
-1.20147419e+00 5.89265466e-01 2.45823994e-01 -9.71754432e-01
-9.84854937e-01 -4.53000404e-02 -9.50724065e-01 5.82912743e-01
-1.17529750e+00 -6.97577894e-01 -1.71626166e-01 7.75781795e-02
8.82635951e-01 -3.40243518e-01 8.34350407e-01 -3.08534116e-01
-4.63291198e-01 5.39278746e-01 2.18737170e-01 -1.93294749e-01
1.05586839e+00 -1.33278799e+00 4.56610143e-01 6.14621222e-01
6.66531920e-02 3.15269202e-01 6.65283382e-01 -7.83597469e-01
-8.35986674e-01 -1.07829702e+00 1.26027513e+00 -6.72496855e-01
3.69166881e-01 -9.84256566e-02 -4.42159235e-01 2.38536611e-01
4.13399220e-01 -6.41283453e-01 8.72620344e-01 5.60310662e-01
-1.97315454e-01 1.87919199e-01 -9.42339420e-01 8.58148992e-01
9.00153875e-01 -5.64424038e-01 -6.66779101e-01 7.23780572e-01
8.98205340e-01 -2.41973519e-01 -7.59285688e-01 3.12461168e-01
3.18987429e-01 -7.80631781e-01 6.92602754e-01 -8.29753697e-01
1.15409124e+00 3.29619080e-01 1.05427718e-02 -1.76975191e+00
-1.51457131e-01 -5.35056770e-01 -2.75896072e-01 1.35089099e+00
7.51440763e-01 -8.78369361e-02 4.65955466e-01 7.44532108e-01
-2.63021618e-01 -7.98614323e-01 -5.07484972e-01 -7.48758435e-01
3.27376783e-01 -1.36424020e-01 4.93059814e-01 7.35903621e-01
3.30575794e-01 1.34169900e+00 -7.27087036e-02 -6.65323615e-01
2.91787922e-01 4.15374227e-02 7.89119780e-01 -1.08225930e+00
-1.92586154e-01 -5.80455482e-01 1.43413097e-01 -4.41930830e-01
3.68826449e-01 -7.64844954e-01 1.43912947e-02 -2.03064775e+00
2.50888228e-01 -2.07596004e-01 2.37862319e-01 8.86921808e-02
-5.47186613e-01 -4.79258716e-01 1.63693368e-01 1.35203555e-01
-1.07095098e+00 7.78452396e-01 1.32203019e+00 -2.16664299e-01
-4.93821025e-01 -2.19284460e-01 -8.61665010e-01 2.45242193e-01
9.60890949e-01 -1.69747397e-01 -8.13193560e-01 -6.68057859e-01
9.19109955e-02 2.50832736e-01 -1.45714968e-01 -1.03573239e+00
1.30530834e-01 -8.25481787e-02 3.46682727e-01 -4.38883096e-01
2.82495707e-01 2.05534026e-01 -3.24354798e-01 -9.44683254e-02
-1.04878962e+00 4.78648812e-01 1.93418041e-01 3.92722189e-01
-1.41683936e-01 -4.82841820e-01 4.02629495e-01 -3.56054038e-01
-2.71504462e-01 -1.39555380e-01 -2.44686767e-01 9.26004648e-01
6.11972034e-01 3.78483976e-03 -7.85161972e-01 -6.55317903e-01
-2.44541809e-01 3.87145996e-01 5.02623558e-01 3.51560771e-01
3.83763522e-01 -1.10904717e+00 -1.20941484e+00 -4.76624429e-01
3.61303777e-01 -1.40150294e-01 1.72956452e-01 4.26364303e-01
-2.70465314e-01 6.27477229e-01 4.45100293e-02 -4.10486251e-01
-1.12068284e+00 1.62730262e-01 3.96485776e-02 -9.00731027e-01
-6.25707507e-02 4.34540123e-01 -1.34730324e-01 -4.68110144e-01
1.67709798e-01 -9.24765915e-02 -2.13855907e-01 5.05107820e-01
5.23240745e-01 6.19376659e-01 2.05879137e-01 -1.99820548e-01
2.37346008e-01 1.47877140e-02 -1.80209875e-01 -7.52129674e-01
8.91253650e-01 2.19385102e-01 1.21493205e-01 4.67818648e-01
1.11822569e+00 -1.88883424e-01 -8.91287684e-01 -5.59272580e-02
1.74444303e-01 -4.62721214e-02 7.75965527e-02 -1.18734550e+00
-1.96166530e-01 8.53751957e-01 1.76698223e-01 6.53309003e-02
7.35963345e-01 -8.02686810e-02 5.85073471e-01 7.69306958e-01
4.01863568e-02 -1.54829562e+00 6.48055673e-01 7.79483020e-01
1.10616541e+00 -1.24641752e+00 1.56847015e-01 2.69678473e-01
-1.02210021e+00 8.20446551e-01 7.34525800e-01 3.35913271e-01
1.19460316e-03 -1.15824275e-01 -1.17447479e-02 -1.78657860e-01
-1.41019452e+00 3.92997786e-02 3.69843662e-01 4.41864669e-01
9.41302657e-01 3.59302551e-01 -6.29951715e-01 2.92929798e-01
-7.03763604e-01 -1.13290176e-01 7.42486715e-01 8.40614080e-01
-4.51025754e-01 -1.01720631e+00 -2.47186217e-02 8.76509070e-01
-1.54993176e-01 -3.97457719e-01 -6.54555380e-01 5.09368181e-01
-4.34028059e-01 1.16744149e+00 1.29640982e-01 -1.20334171e-01
3.98826838e-01 3.60102922e-01 4.68972743e-01 -1.04829681e+00
-6.90866172e-01 -1.24352828e-01 8.28213155e-01 -1.77421421e-02
-7.65421987e-02 -8.30091834e-01 -7.97179937e-01 -1.99234530e-01
-4.48003829e-01 3.73498082e-01 6.75481558e-01 8.50936413e-01
5.59643090e-01 5.51364541e-01 3.53681475e-01 -4.85199928e-01
-9.49412704e-01 -1.49417174e+00 -1.12735573e-02 4.02267784e-01
1.09135941e-01 -1.96528360e-01 -3.48730862e-01 -2.43113283e-03] | [11.888023376464844, 9.109991073608398] |
06b52ec3-098a-4633-8a9b-32e655f0383e | show-me-your-nft-and-i-tell-you-how-it-will | 2302.01676 | null | https://arxiv.org/abs/2302.01676v2 | https://arxiv.org/pdf/2302.01676v2.pdf | Show me your NFT and I tell you how it will perform: Multimodal representation learning for NFT selling price prediction | Non-Fungible Tokens (NFTs) represent deeds of ownership, based on blockchain technologies and smart contracts, of unique crypto assets on digital art forms (e.g., artworks or collectibles). In the spotlight after skyrocketing in 2021, NFTs have attracted the attention of crypto enthusiasts and investors intent on placing promising investments in this profitable market. However, the NFT financial performance prediction has not been widely explored to date. In this work, we address the above problem based on the hypothesis that NFT images and their textual descriptions are essential proxies to predict the NFT selling prices. To this purpose, we propose MERLIN, a novel multimodal deep learning framework designed to train Transformer-based language and visual models, along with graph neural network models, on collections of NFTs' images and texts. A key aspect in MERLIN is its independence on financial features, as it exploits only the primary data a user interested in NFT trading would like to deal with, i.e., NFT images and textual descriptions. By learning dense representations of such data, a price-category classification task is performed by MERLIN models, which can also be tuned according to user preferences in the inference phase to mimic different risk-return investment profiles. Experimental evaluation on a publicly available dataset has shown that MERLIN models achieve significant performances according to several financial assessment criteria, fostering profitable investments, and also beating baseline machine-learning classifiers based on financial features. | ['Andrea Tagarelli', 'Lucio La Cava', 'Davide Costa'] | 2023-02-03 | null | null | null | null | ['multimodal-deep-learning'] | ['natural-language-processing'] | [-1.94720477e-01 4.16193828e-02 -6.13934100e-01 -2.20298812e-01
-5.36742687e-01 -8.43685031e-01 1.17880332e+00 -1.52874570e-02
-1.29939467e-01 1.41571045e-01 3.16513151e-01 -7.39995062e-01
7.40511566e-02 -1.05665064e+00 -6.93866611e-01 -3.86415243e-01
5.42024300e-02 7.86231518e-01 -2.53441125e-01 -1.65017322e-01
4.38545942e-01 3.44923168e-01 -9.57923114e-01 6.02152646e-01
5.30331075e-01 1.62629795e+00 -2.22217485e-01 1.82033747e-01
-3.76073420e-01 1.36873984e+00 -4.28831756e-01 -1.18784618e+00
6.26178265e-01 -5.29757291e-02 -4.75298315e-01 -2.92430133e-01
3.18828642e-01 -5.39698839e-01 -5.35048127e-01 1.17612815e+00
4.45680022e-02 -6.42755330e-01 8.47838581e-01 -1.14070976e+00
-1.16782475e+00 1.32214332e+00 -5.77337086e-01 1.02385193e-01
-4.58941907e-02 4.85840052e-01 1.71103752e+00 -7.32941806e-01
7.92214453e-01 1.26520503e+00 5.02800465e-01 2.37120353e-02
-1.25331914e+00 -7.92946339e-01 -1.82298228e-01 2.34862179e-01
-7.45681882e-01 -6.70943931e-02 1.05024660e+00 -6.10127389e-01
8.00189853e-01 1.92884400e-01 9.75152373e-01 1.60214472e+00
6.87539518e-01 1.12717330e+00 1.42380023e+00 -2.12389693e-01
3.01607903e-02 1.57625273e-01 -1.80218890e-01 4.44129229e-01
3.68682712e-01 1.88729882e-01 -4.84884560e-01 -5.41283265e-02
7.49924481e-01 8.46258085e-03 8.27733334e-03 -5.44795454e-01
-1.26182294e+00 1.35205317e+00 6.40460134e-01 6.56365931e-01
-3.24851185e-01 4.02043939e-01 5.78274310e-01 6.37368560e-01
5.93610942e-01 5.14412522e-01 -1.63178638e-01 -2.51416713e-02
-1.06386149e+00 1.84938788e-01 8.05545688e-01 6.49085820e-01
4.06863898e-01 2.98046824e-02 -2.36546919e-01 3.94351423e-01
4.54079390e-01 5.40221751e-01 4.83378470e-01 -4.59971189e-01
7.43877351e-01 8.77025425e-01 -2.70469189e-02 -8.80683601e-01
-2.11159557e-01 -7.00534999e-01 -7.56052077e-01 2.25457236e-01
4.78223562e-01 2.52676010e-01 -7.92089403e-01 1.18876112e+00
-2.14180902e-01 -2.81867623e-01 -5.04552200e-02 8.88046265e-01
4.84570563e-01 7.13615835e-01 -2.10241914e-01 1.79441825e-01
1.39307892e+00 -7.14410007e-01 -4.40459311e-01 2.43103374e-02
6.12440884e-01 -7.23980129e-01 8.14062595e-01 5.21633267e-01
-8.21766376e-01 -4.34367657e-02 -8.46514046e-01 1.85780115e-02
-5.95884144e-01 -1.47875249e-02 7.96669424e-01 6.53208911e-01
-7.70111859e-01 7.53485441e-01 -3.79496932e-01 1.14629418e-01
9.28378046e-01 1.10571004e-01 -1.43194005e-01 1.65101185e-01
-1.31544149e+00 1.05335379e+00 4.13214147e-01 1.80548891e-01
-1.03601813e+00 -4.27598745e-01 -5.65209746e-01 3.79428864e-01
4.11869377e-01 -4.90325123e-01 1.11106408e+00 -1.33457470e+00
-1.24188912e+00 1.01132464e+00 9.24423397e-01 -1.04676437e+00
1.14471841e+00 -7.66423494e-02 -3.19175720e-01 6.31129974e-03
1.08251795e-02 4.12139326e-01 1.14886951e+00 -8.81655633e-01
-3.29857916e-01 -2.91793585e-01 2.26497829e-01 -3.14329296e-01
-4.02485937e-01 4.55316603e-02 -3.51008603e-05 -1.03826928e+00
-3.59876484e-01 -9.28838432e-01 1.68935731e-01 -2.19633460e-01
-6.27822399e-01 -3.03124994e-01 8.01646233e-01 -7.68303216e-01
7.40917802e-01 -2.02847481e+00 -1.10093445e-01 5.73333740e-01
3.50646019e-01 1.53071836e-01 -1.00145154e-01 5.19133091e-01
1.46262839e-01 3.53917807e-01 7.09856451e-02 -2.18916927e-02
7.17550039e-01 -2.70592868e-01 -6.00588977e-01 5.62712431e-01
7.40029290e-02 1.49607372e+00 -7.19815016e-01 -3.09491485e-01
1.00965388e-01 3.18665296e-01 -2.15707600e-01 3.24337259e-02
-6.64529443e-01 2.66947806e-01 -6.68531239e-01 9.35742557e-01
5.22258937e-01 -4.94901568e-01 4.70460296e-01 -2.61000544e-01
-8.02953690e-02 3.54637533e-01 -4.45764929e-01 1.24975300e+00
-3.54521602e-01 6.56414807e-01 -1.75809786e-01 -1.05605233e+00
8.98945868e-01 9.41551998e-02 3.92640501e-01 -1.10692132e+00
5.88817477e-01 6.36487961e-01 -3.98253575e-02 -1.22747846e-01
3.04363340e-01 -3.01123172e-01 -2.64661938e-01 7.08385050e-01
-1.29372805e-01 6.91803917e-02 -7.53195211e-02 1.20994508e-01
1.00345123e+00 1.99127421e-01 1.25420853e-01 -2.18727365e-01
1.66808411e-01 -3.49400975e-02 2.28165120e-01 6.35532975e-01
4.79137599e-02 2.69141257e-01 9.78310943e-01 -8.12288463e-01
-1.27369511e+00 -9.59420264e-01 -6.18890263e-02 6.78990602e-01
-1.76777437e-01 -7.24050626e-02 -5.75130641e-01 -9.96560574e-01
4.56837595e-01 7.11587012e-01 -7.19259262e-01 8.42686892e-02
-4.79474217e-01 -4.44314361e-01 3.96436989e-01 3.92883539e-01
5.48288643e-01 -1.16095901e+00 -8.85878742e-01 1.78864196e-01
1.82720482e-01 -1.01348841e+00 -4.04602557e-01 7.64482766e-02
-6.92056477e-01 -1.04231298e+00 -9.93828595e-01 -3.04802448e-01
2.56548822e-01 -2.06605345e-01 1.05778706e+00 -8.72082338e-02
-1.05900452e-01 2.37336308e-01 -3.13796550e-01 -3.11663568e-01
-6.10092461e-01 -1.73143744e-02 -2.65684694e-01 5.00426471e-01
1.96557596e-01 -4.08861458e-01 -6.98587894e-01 -1.20085040e-02
-1.00356519e+00 1.43966824e-01 1.02487803e+00 8.62849593e-01
3.08686852e-01 -2.00240508e-01 4.01776761e-01 -1.03776538e+00
6.53980434e-01 -4.68326658e-01 -1.12528896e+00 4.67208117e-01
-9.14949000e-01 1.51372582e-01 5.33202767e-01 -3.81316394e-01
-8.24515164e-01 -6.98347509e-01 3.86049420e-01 -9.25125241e-01
6.17502630e-01 8.14968884e-01 1.45632057e-02 -1.20650843e-01
1.39311582e-01 6.06789216e-02 -1.28844455e-01 -3.95585477e-01
5.72891414e-01 5.76454461e-01 2.00910315e-01 -3.56571794e-01
9.32777226e-01 3.64823967e-01 1.32423252e-01 -1.72559261e-01
-5.29829085e-01 -4.60415473e-03 -2.68313944e-01 -3.03222924e-01
6.63171053e-01 -7.47593701e-01 -8.37630212e-01 6.52009487e-01
-1.00067115e+00 -1.83158353e-01 -1.90724656e-01 4.93992448e-01
-5.15644670e-01 3.32617879e-01 -8.31734776e-01 -8.02357495e-01
-5.17965436e-01 -1.04624808e+00 8.15269887e-01 -1.92770600e-01
-6.05911016e-02 -1.03964508e+00 3.25777046e-02 7.54644930e-01
6.26941204e-01 4.89339888e-01 1.26619518e+00 -1.10677218e+00
-1.15053713e+00 -3.70275944e-01 -5.15791833e-01 4.30063158e-01
-1.12279654e-01 -2.66664863e-01 -8.83945942e-01 -4.55965638e-01
1.05767965e-01 -4.55230713e-01 1.10962009e+00 2.80522138e-01
8.10575485e-01 -7.53466785e-01 -2.42546387e-02 5.17709017e-01
1.48113465e+00 1.11301161e-01 5.57905257e-01 7.41794169e-01
6.87267840e-01 6.68119609e-01 2.28384212e-01 5.50032139e-01
4.11845416e-01 5.87193012e-01 7.99288213e-01 2.05916539e-01
1.59512475e-01 -5.16182661e-01 5.10490596e-01 4.05044466e-01
-1.11804761e-01 -2.65858978e-01 -9.94078815e-01 3.36691290e-01
-1.72244799e+00 -1.05458546e+00 2.07737163e-01 1.93641949e+00
3.90991092e-01 4.69116271e-01 9.28653926e-02 -1.38699085e-01
6.24069393e-01 4.51453358e-01 -6.51563823e-01 -4.70725626e-01
-4.64097261e-01 1.04154430e-01 8.35272312e-01 -2.07419798e-01
-1.10209274e+00 6.97017193e-01 4.91119623e+00 9.16465044e-01
-1.46004331e+00 3.12220771e-02 1.12712181e+00 -1.82136819e-01
-8.66866648e-01 1.27372712e-01 -2.42687672e-01 6.92476392e-01
8.22207928e-01 -7.94280171e-02 6.20502830e-01 6.69613481e-01
-2.28735767e-02 3.58405799e-01 -1.14806330e+00 9.81554151e-01
-2.48421915e-02 -1.88288593e+00 3.57648939e-01 5.48731148e-01
6.06260598e-01 1.40908912e-01 5.82672894e-01 2.63982803e-01
4.50069815e-01 -1.16186726e+00 1.47618842e+00 5.60660422e-01
7.07371056e-01 -6.48414135e-01 8.94678414e-01 6.90713525e-02
-8.58937740e-01 -4.74809706e-01 -2.28079185e-02 2.82474905e-01
-7.11723268e-02 5.76370955e-01 -7.25852847e-01 6.39129817e-01
7.08776712e-01 7.37387002e-01 -5.03606200e-01 5.32680571e-01
-1.94641680e-01 5.16415238e-01 -8.54774415e-02 -1.81346402e-01
7.28107333e-01 -4.67337668e-01 3.74978304e-01 1.04549348e+00
6.51579261e-01 -3.60901207e-01 -2.17050552e-01 1.35605204e+00
-5.77512920e-01 2.04482988e-01 -8.41222286e-01 -7.78558731e-01
1.55243501e-01 1.23115814e+00 -8.36671293e-01 -1.74520120e-01
-6.85660422e-01 5.78715265e-01 2.70652622e-01 1.33718312e-01
-8.41682315e-01 -1.12430349e-01 1.50917783e-01 3.51781756e-01
7.39184976e-01 9.57926735e-02 -2.86872715e-01 -1.42051029e+00
1.23848692e-01 -9.79968607e-01 3.98124933e-01 -6.56220675e-01
-1.56358719e+00 4.11654502e-01 -1.87303439e-01 -1.31931949e+00
-1.89607784e-01 -8.94833624e-01 -7.82382846e-01 4.96564567e-01
-1.67473209e+00 -1.75770128e+00 4.34898287e-01 4.25184339e-01
5.72403856e-02 -6.35948360e-01 2.55227655e-01 -8.94476026e-02
-2.81269014e-01 4.24192488e-01 3.99832815e-01 5.97362638e-01
2.50771284e-01 -1.13533378e+00 5.90537548e-01 4.21523243e-01
6.63443387e-01 4.13104326e-01 5.07781088e-01 -6.96804047e-01
-1.68976235e+00 -8.11339378e-01 8.57551873e-01 -2.78592497e-01
1.35322249e+00 -5.38677812e-01 -5.97353816e-01 7.62129366e-01
4.11714762e-01 -1.64219379e-01 4.65521663e-01 -1.89155385e-01
-8.34968388e-01 -1.14346050e-01 -1.10565937e+00 2.97097147e-01
6.27352893e-01 -7.68821597e-01 -4.96572345e-01 2.83581048e-01
2.61167765e-01 -3.73768322e-02 -8.12712967e-01 8.94737095e-02
7.38522053e-01 -1.17902672e+00 7.95567691e-01 -4.18844789e-01
7.34077036e-01 3.19792628e-01 -1.42249763e-01 -9.08652663e-01
-1.31039694e-01 -7.05113232e-01 -1.49265915e-01 1.15812707e+00
3.87488693e-01 -7.98372567e-01 9.37456131e-01 3.80040228e-01
3.04631174e-01 -7.76662767e-01 -1.35535920e+00 -8.76125813e-01
2.72674650e-01 -4.37880278e-01 8.96554947e-01 1.02999473e+00
2.74162926e-02 1.53386548e-01 -5.00858784e-01 -4.21794653e-01
8.53765488e-01 8.22642505e-01 4.82453912e-01 -1.24846363e+00
-2.52670377e-01 -9.54391658e-01 -4.46457207e-01 -6.22972190e-01
3.59865129e-01 -1.32510471e+00 -7.61704147e-01 -1.36739349e+00
2.84481138e-01 -4.10819292e-01 -5.35584867e-01 5.12553275e-01
5.37059903e-01 1.36488006e-01 5.91868639e-01 6.31251037e-01
-3.41598481e-01 6.00216031e-01 1.22481942e+00 -9.08768475e-01
3.69914711e-01 -1.32156253e-01 -4.76966470e-01 5.24169743e-01
5.41120350e-01 -2.33649552e-01 -6.97486894e-03 -4.14448857e-01
6.45052671e-01 2.38196477e-01 7.09769011e-01 -5.08462429e-01
-7.13563189e-02 -1.24380440e-01 3.60096693e-01 -5.35292983e-01
1.23691835e-01 -9.01623785e-01 3.49034607e-01 6.17524683e-01
-3.37938756e-01 1.03906929e-01 -3.49200487e-01 5.78353703e-01
-2.72757918e-01 -1.58346534e-01 5.31974792e-01 -3.17440063e-01
-3.58452082e-01 4.16754812e-01 -2.18420684e-01 -1.39412567e-01
8.34053993e-01 -1.00474991e-01 -5.31849682e-01 -4.64978367e-01
-2.87033230e-01 1.35099739e-01 5.47782362e-01 5.63693404e-01
6.05845511e-01 -1.49525535e+00 -9.29495752e-01 2.15543844e-02
1.09924920e-01 -5.37270486e-01 -8.21530968e-02 8.40085268e-01
-7.09263682e-01 7.20768273e-01 -3.54597360e-01 -3.24879259e-01
-7.10217953e-01 7.39746869e-01 1.24384381e-01 -8.63743603e-01
-7.34800100e-01 4.88559812e-01 2.19379023e-01 -2.91737262e-02
-1.85946104e-04 -5.53660572e-01 -3.62711340e-01 4.59618926e-01
1.11600377e-01 1.32112250e-01 -6.61485046e-02 -6.86716497e-01
-1.39880434e-01 3.19389939e-01 -1.53042957e-01 -1.31072365e-02
1.60117960e+00 4.82112437e-01 -2.03697696e-01 2.76396781e-01
1.17746532e+00 -2.11094655e-02 -1.10544682e+00 -3.08154225e-01
5.33814251e-01 -6.38771653e-01 7.31414780e-02 -8.18376720e-01
-1.57074356e+00 8.39248955e-01 4.66806918e-01 5.33550262e-01
6.92654669e-01 1.42566994e-01 9.99306023e-01 3.35115999e-01
6.53038383e-01 -9.70083117e-01 1.26643375e-01 2.64913976e-01
9.74480987e-01 -1.09573460e+00 -1.76581189e-01 3.12267512e-01
-5.85242510e-01 1.28303611e+00 -1.83996156e-01 -1.23306133e-01
4.27464932e-01 -8.19544792e-02 4.40795608e-02 -4.70542103e-01
-6.32794499e-01 3.41088951e-01 2.82936782e-01 3.47625464e-01
1.64446875e-01 3.55710357e-01 -1.83059890e-02 7.72749782e-01
-2.15684161e-01 -3.09511833e-02 3.64570379e-01 5.16636372e-01
-2.05442756e-01 -1.25105870e+00 -4.00798023e-01 7.53091872e-01
-8.18404078e-01 -1.97125107e-01 -5.03302693e-01 8.00102293e-01
1.32595580e-02 4.63601679e-01 -5.42029627e-02 -2.69490540e-01
9.69268307e-02 -6.88174739e-02 3.52000952e-01 -1.84932902e-01
-9.41899061e-01 9.17218849e-02 1.60433576e-01 -4.00515854e-01
-2.72370785e-01 -6.99874878e-01 -7.34259129e-01 -4.57695246e-01
-3.17226648e-01 1.03141896e-01 5.81000865e-01 7.87974894e-01
-1.04271993e-01 3.52702569e-03 8.22985590e-01 -7.72339880e-01
-1.06563568e+00 -7.93536067e-01 -8.74530613e-01 5.10741532e-01
2.05577910e-01 -4.91455644e-01 -3.31005156e-01 -3.07583183e-01] | [6.725698471069336, 6.129147052764893] |
4591b1b3-8a97-4bf9-8a21-0c4ce2679ef1 | revisiting-self-supervised-contrastive | 2210.03853 | null | https://arxiv.org/abs/2210.03853v1 | https://arxiv.org/pdf/2210.03853v1.pdf | Revisiting Self-Supervised Contrastive Learning for Facial Expression Recognition | The success of most advanced facial expression recognition works relies heavily on large-scale annotated datasets. However, it poses great challenges in acquiring clean and consistent annotations for facial expression datasets. On the other hand, self-supervised contrastive learning has gained great popularity due to its simple yet effective instance discrimination training strategy, which can potentially circumvent the annotation issue. Nevertheless, there remain inherent disadvantages of instance-level discrimination, which are even more challenging when faced with complicated facial representations. In this paper, we revisit the use of self-supervised contrastive learning and explore three core strategies to enforce expression-specific representations and to minimize the interference from other facial attributes, such as identity and face styling. Experimental results show that our proposed method outperforms the current state-of-the-art self-supervised learning methods, in terms of both categorical and dimensional facial expression recognition tasks. | ['Benny Lo', 'Guang-Zhong Yang', 'Xiao Gu', 'Yuxuan Shu'] | 2022-10-08 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [ 3.48706424e-01 -2.75039133e-02 -3.23724657e-01 -8.98599863e-01
-6.11819565e-01 -2.41103083e-01 5.50159454e-01 -1.63909003e-01
-3.17654550e-01 8.46089065e-01 -1.43787369e-01 4.42489088e-01
-5.92518002e-02 -3.65280718e-01 -1.66325539e-01 -1.02859724e+00
-4.37517418e-03 2.04016730e-01 -5.43467641e-01 -2.60596544e-01
-3.36335935e-02 6.09527528e-01 -1.98075235e+00 5.54805882e-02
8.89902234e-01 1.49412858e+00 -4.77233708e-01 -5.73334172e-02
-3.35282326e-01 9.39845324e-01 -5.11549354e-01 -7.56593823e-01
4.71406197e-03 -5.14415920e-01 -6.02772295e-01 4.20036018e-01
7.49318123e-01 -1.87616646e-01 7.10557997e-02 1.17155349e+00
5.40711939e-01 -1.74331833e-02 7.56853402e-01 -1.62430418e+00
-4.88942534e-01 -4.71573621e-02 -8.96574199e-01 -7.53849894e-02
2.33005583e-01 -5.70036098e-02 9.11047339e-01 -9.62561309e-01
6.26547515e-01 1.05614781e+00 6.09957576e-01 8.41224909e-01
-1.31708002e+00 -1.18278289e+00 1.75408855e-01 1.09679297e-01
-1.42084169e+00 -9.63165700e-01 1.09753370e+00 -4.11899328e-01
5.19373894e-01 4.94225249e-02 3.67906421e-01 1.16932106e+00
-4.65989977e-01 9.07322645e-01 1.58584380e+00 -3.90781850e-01
1.75348148e-01 1.29209042e-01 -1.21691346e-01 1.02937758e+00
-2.91063637e-01 -1.50240630e-01 -6.79203153e-01 -3.32509041e-01
5.55063188e-01 -1.79458618e-01 -1.15421433e-02 -4.64890271e-01
-5.39370418e-01 6.95946932e-01 2.12328453e-02 3.16675752e-01
-3.68088424e-01 -1.61423847e-01 6.81358933e-01 2.67769545e-01
8.64584267e-01 1.97958529e-01 -4.10090476e-01 -3.36012244e-01
-9.88148689e-01 -7.54641555e-03 5.10526240e-01 7.54924476e-01
1.12284601e+00 3.40691775e-01 4.29682098e-02 1.43080997e+00
1.34656683e-01 2.77829260e-01 4.16551292e-01 -9.75744963e-01
9.09711272e-02 5.67390382e-01 -2.47803003e-01 -1.07273686e+00
-2.12183401e-01 -2.56228626e-01 -1.09475160e+00 3.80382508e-01
4.05603379e-01 -1.25662550e-01 -6.31076634e-01 2.08707881e+00
2.19012126e-01 2.78638989e-01 -9.10005532e-03 8.05750370e-01
7.81860352e-01 1.90421209e-01 5.50783575e-01 -6.14078999e-01
1.14570713e+00 -8.94564748e-01 -1.06392348e+00 -5.96637949e-02
5.08547008e-01 -4.87681895e-01 8.21516395e-01 2.44377658e-01
-8.34066868e-01 -5.11584938e-01 -7.56090283e-01 1.05990142e-01
-2.10814342e-01 2.78134376e-01 1.19195151e+00 8.50206554e-01
-7.95826495e-01 3.95088583e-01 -4.89196211e-01 -2.15587765e-01
9.63836312e-01 4.93741006e-01 -9.41644311e-01 2.26004049e-01
-1.01302302e+00 6.91724420e-01 -1.49280444e-01 2.11276501e-01
-5.00258446e-01 -6.63392305e-01 -1.04007411e+00 -1.31157354e-01
3.36915374e-01 -9.16353911e-02 1.10687268e+00 -1.77283216e+00
-1.81993079e+00 1.51031148e+00 -4.18551624e-01 4.09353599e-02
4.43991125e-01 -1.26652434e-01 -4.30044830e-01 5.72930239e-02
-4.32763621e-02 6.34570062e-01 1.05290151e+00 -1.12989497e+00
-3.27419609e-01 -6.77416623e-01 -1.88871443e-01 9.54914987e-02
-7.14964271e-01 2.78297275e-01 -4.26224113e-01 -4.70727503e-01
-7.42676109e-02 -8.43882382e-01 -8.02468508e-02 5.19373894e-01
7.36698229e-03 -4.86332476e-01 8.80840361e-01 -2.80364782e-01
9.27709103e-01 -2.38147211e+00 -9.32733645e-04 2.15276092e-01
1.12628773e-01 3.74646872e-01 -1.18230745e-01 1.35428607e-02
-2.38975301e-01 -1.21507078e-01 -3.47153485e-01 -7.59793162e-01
-5.30553721e-02 2.75246143e-01 -8.68085399e-02 5.25357246e-01
5.79406977e-01 7.22078264e-01 -8.67388308e-01 -8.32878947e-01
3.81992571e-02 4.53839630e-01 -3.55129808e-01 3.92042577e-01
-6.13847300e-02 6.53666496e-01 -4.09360796e-01 9.57192183e-01
7.81392634e-01 -3.95211354e-02 3.37514400e-01 -1.24740608e-01
3.54193822e-02 -1.12961963e-01 -9.55532074e-01 1.62361956e+00
-4.94762212e-01 6.46331072e-01 4.00002182e-01 -1.34808326e+00
1.29374909e+00 2.30965033e-01 8.34112108e-01 -7.37874568e-01
2.15600669e-01 2.90593654e-01 -2.86459714e-01 -5.62738121e-01
2.04871938e-01 -4.60322440e-01 -4.49917931e-03 2.93831319e-01
3.04303944e-01 1.32846445e-01 6.20172322e-02 -1.85135588e-01
5.19643009e-01 4.76762623e-01 5.00634909e-01 -2.37617314e-01
8.47852349e-01 -5.17692685e-01 9.00176287e-01 1.03594318e-01
-5.91968060e-01 3.27940077e-01 6.04033887e-01 -3.76260549e-01
-5.59524298e-01 -6.99650288e-01 -3.04639161e-01 1.40714526e+00
-2.47097820e-01 -4.16721582e-01 -8.22123766e-01 -8.25158000e-01
3.22541147e-02 8.13587084e-02 -7.76543856e-01 -3.75343040e-02
-2.87739426e-01 -7.95237064e-01 8.12102199e-01 5.42449415e-01
6.54734790e-01 -9.81417418e-01 -1.79154471e-01 -4.12005186e-02
-1.09886870e-01 -1.16787910e+00 -1.49056807e-01 7.55934939e-02
-7.15384066e-01 -9.13596034e-01 -7.05928147e-01 -7.44280815e-01
9.21813071e-01 -3.93988639e-02 9.71359313e-01 1.36639088e-01
-3.69552433e-01 3.24062705e-01 -1.30152106e-01 -4.96458828e-01
-2.40400434e-01 -9.87852514e-02 1.33794531e-01 6.33381009e-01
6.66660905e-01 -6.08737171e-01 -3.31840038e-01 2.79164612e-01
-8.02970231e-01 -2.90179223e-01 4.46692407e-01 1.03300595e+00
7.11956322e-01 -1.55713767e-01 7.89160132e-01 -1.04531229e+00
4.74531263e-01 -3.46457779e-01 -3.70271027e-01 2.29771331e-01
-5.63054025e-01 -1.02896422e-01 5.16319394e-01 -3.84721458e-01
-1.31853330e+00 3.78645092e-01 -2.75655210e-01 -4.63387787e-01
-3.29968661e-01 3.67894292e-01 -3.13830972e-01 -3.63053709e-01
5.12214243e-01 1.15565784e-01 5.99086046e-01 -2.71538734e-01
1.71742886e-01 7.75948286e-01 4.20699954e-01 -8.23448837e-01
5.29592812e-01 5.57790518e-01 2.38828599e-01 -1.02148664e+00
-1.05683172e+00 -3.17610770e-01 -8.89045119e-01 -3.79252940e-01
5.94007730e-01 -1.00215769e+00 -7.38611460e-01 7.88045228e-01
-7.53750443e-01 -1.14457592e-01 -1.15633406e-01 -3.28179100e-04
-7.53393233e-01 5.04902959e-01 -4.71624494e-01 -1.11816299e+00
-2.38082930e-01 -8.75369251e-01 1.15701973e+00 2.76480079e-01
-4.23494965e-01 -8.67625058e-01 -8.38974565e-02 4.52875286e-01
3.52670729e-01 4.68438447e-01 7.16469705e-01 -3.31558138e-01
-1.28702804e-01 -9.98999774e-02 -2.00985849e-01 6.47596240e-01
5.44615805e-01 2.14077935e-01 -1.39198887e+00 -3.34781334e-02
-2.34846622e-01 -1.03122926e+00 5.85556209e-01 -7.42218122e-02
1.42830110e+00 -1.04163468e-01 -3.79406363e-02 7.10320473e-01
1.14042544e+00 -1.18527874e-01 6.31086528e-01 1.23790383e-01
4.55808014e-01 1.06691444e+00 8.84042084e-01 5.85575342e-01
1.28254235e-01 8.20787013e-01 2.29073137e-01 -4.47824180e-01
1.68105453e-01 -2.19758436e-01 3.90080996e-02 4.11945969e-01
-4.53912616e-01 3.74953359e-01 -5.20756185e-01 1.87075630e-01
-1.83345532e+00 -1.01998782e+00 2.62857497e-01 1.92640376e+00
1.22673392e+00 -3.83722097e-01 2.38010973e-01 2.80408531e-01
4.30451393e-01 3.58688742e-01 -4.60062087e-01 -4.04957443e-01
-4.78273630e-01 5.48110902e-01 -1.57593340e-01 9.91730317e-02
-1.36181188e+00 1.09390497e+00 6.27241230e+00 8.35529149e-01
-1.32362187e+00 1.04510337e-01 9.57089841e-01 3.39860171e-02
1.44321591e-01 -4.69418824e-01 -6.46357656e-01 3.50324661e-01
5.02353370e-01 1.17036611e-01 8.79279226e-02 1.23752356e+00
-3.16007473e-02 1.30075485e-01 -1.06340647e+00 1.43685865e+00
3.10299844e-01 -9.69883144e-01 -1.51555508e-01 -1.30527928e-01
7.17001438e-01 -5.07598698e-01 1.92342877e-01 3.91868800e-01
-3.08707416e-01 -1.18660927e+00 4.17217940e-01 3.44447672e-01
1.11312461e+00 -7.96069324e-01 6.76454842e-01 7.51708299e-02
-8.88331831e-01 5.58544584e-02 -1.90379292e-01 -2.19438523e-01
-7.59949312e-02 4.43256915e-01 -2.23718435e-01 2.42969662e-01
6.29196763e-01 8.27873528e-01 -3.04179132e-01 4.14251894e-01
-3.53324503e-01 3.98817152e-01 -1.00713305e-01 5.54745756e-02
1.02276742e-01 -4.47931260e-01 -9.60577801e-02 1.25454497e+00
7.86642805e-02 1.82489216e-01 8.96858945e-02 6.77620828e-01
-3.09637874e-01 5.02052188e-01 -5.23046792e-01 -1.04975335e-01
2.36167192e-01 1.54600060e+00 -2.04066202e-01 -2.15094522e-01
-5.93933165e-01 1.13059437e+00 8.08629692e-01 2.29952529e-01
-6.62767410e-01 1.14705540e-01 1.20162249e+00 -4.10086475e-02
2.95249280e-02 -1.23280577e-01 -8.42162594e-03 -1.08868361e+00
1.58301517e-01 -1.04908907e+00 4.23518598e-01 -4.79011148e-01
-1.55144465e+00 5.88014901e-01 -2.63404727e-01 -1.08869731e+00
-4.14124131e-01 -7.09315777e-01 -3.66081327e-01 5.95551133e-01
-1.84948659e+00 -1.40521622e+00 -5.53924620e-01 8.39138269e-01
2.39851952e-01 -3.74319166e-01 1.35431457e+00 5.00361085e-01
-6.37835801e-01 1.08860636e+00 -1.21579565e-01 3.24303359e-01
9.68386114e-01 -8.26416910e-01 -4.13555205e-01 1.60745189e-01
1.78766057e-01 6.55687332e-01 3.98022115e-01 -1.45289242e-01
-1.27661169e+00 -8.48003089e-01 8.14948559e-01 -9.12625864e-02
6.10394955e-01 -4.43309665e-01 -9.88230705e-01 3.90426427e-01
-2.19149776e-02 4.70555514e-01 1.26929379e+00 4.19081062e-01
-7.99937904e-01 -4.32559371e-01 -1.22584331e+00 4.25484955e-01
1.27739847e+00 -8.84369612e-01 -1.28442332e-01 2.11698309e-01
-3.34446311e-01 -1.20523483e-01 -9.31947887e-01 6.66379988e-01
7.66945899e-01 -1.13276446e+00 6.20771587e-01 -8.24909508e-01
4.68742073e-01 -2.27526668e-02 -6.99436292e-02 -1.05864775e+00
-2.08903104e-01 -7.30376601e-01 -1.39162578e-02 1.65149033e+00
2.00236246e-01 -5.76352298e-01 1.08988869e+00 7.52595186e-01
3.29982877e-01 -8.68966401e-01 -1.03272963e+00 -7.34192073e-01
-8.02989155e-02 -1.79859728e-01 4.98195231e-01 1.33749890e+00
1.22604139e-01 3.30303878e-01 -6.37961864e-01 -2.58329809e-01
6.85004354e-01 2.72066861e-01 8.69899392e-01 -1.37509370e+00
-4.53152619e-02 -4.69142944e-01 -7.51093209e-01 -7.65950739e-01
1.02848220e+00 -6.74111128e-01 3.94473225e-03 -7.38196671e-01
2.72771269e-01 -5.90713620e-01 -4.26644981e-01 6.94154859e-01
-2.85091996e-01 7.02342689e-01 -6.46863878e-02 1.96774438e-01
-6.57583833e-01 8.13218296e-01 9.60695982e-01 -1.64917618e-01
9.75128040e-02 -1.77886382e-01 -7.11467564e-01 8.22635829e-01
8.09090137e-01 -7.90220201e-02 -4.20958161e-01 -2.02802822e-01
-9.42533314e-02 -1.67013317e-01 1.26481414e-01 -6.22546554e-01
-9.63903591e-02 -2.09892035e-01 4.09218937e-01 4.31471802e-02
6.79404259e-01 -7.45258629e-01 -1.11197643e-01 -1.94146391e-02
-3.39561641e-01 -1.90265477e-01 2.00672209e-01 3.33324760e-01
-5.95948935e-01 -3.92962769e-02 1.10803711e+00 -3.07760779e-02
-9.52921867e-01 5.65246522e-01 -2.51572758e-01 -9.28161293e-02
1.01389861e+00 -2.58724928e-01 -2.69843955e-02 -5.76095939e-01
-6.36646688e-01 -3.68765630e-02 4.80482310e-01 4.82265145e-01
4.01148677e-01 -1.45315957e+00 -6.06072664e-01 4.36026037e-01
6.08409226e-01 -4.14243072e-01 2.56548941e-01 8.04885864e-01
9.85102877e-02 1.82579446e-03 -6.04624987e-01 -5.06616294e-01
-1.67324877e+00 2.82190144e-01 3.47315490e-01 1.91254932e-02
-9.73948240e-02 8.93225789e-01 2.21649691e-01 -2.95204908e-01
2.74359077e-01 4.88520682e-01 -3.40139657e-01 3.56735736e-01
5.86953938e-01 3.94625627e-02 -2.93687638e-02 -1.02481759e+00
-4.41378951e-01 6.69622838e-01 -7.36620426e-02 2.83204287e-01
1.31098461e+00 7.29648471e-02 -4.42456365e-01 2.78425545e-01
1.32485044e+00 -5.69025166e-02 -1.24359214e+00 -3.56894255e-01
9.08704400e-02 -6.58894241e-01 -1.11021511e-01 -5.19540846e-01
-1.20686126e+00 7.91471481e-01 5.52303791e-01 -2.42420197e-01
1.42178845e+00 -2.05279496e-02 4.40188497e-01 2.99940139e-01
5.26637852e-01 -1.28554201e+00 2.16400445e-01 2.42457300e-01
7.40287304e-01 -1.49644220e+00 4.05652262e-02 -7.79606402e-01
-6.79121017e-01 1.05958474e+00 8.72251689e-01 1.09114334e-01
6.14287972e-01 2.54504353e-01 5.17854512e-01 -1.37254462e-01
-4.62500036e-01 -2.48953462e-01 1.31618202e-01 7.32133627e-01
8.87975574e-01 -5.82702272e-02 -2.98756182e-01 5.48056483e-01
1.16007067e-02 1.33857504e-01 -1.82086393e-01 8.23918343e-01
7.06171393e-02 -1.34431422e+00 -7.16758333e-03 3.12446535e-01
-6.73141956e-01 1.69366568e-01 -6.42155170e-01 7.52866626e-01
1.80335209e-01 7.33588338e-01 7.21155331e-02 -1.79582641e-01
2.06878796e-01 3.93214077e-01 7.56215453e-01 -4.70932186e-01
-3.37201059e-01 -1.02248996e-01 3.44959527e-01 -6.18869305e-01
-1.07646310e+00 -8.80724132e-01 -1.06216085e+00 -6.71948865e-02
-1.87532350e-01 5.00645600e-02 5.98672390e-01 9.02861655e-01
4.50952411e-01 -8.32500011e-02 7.83046603e-01 -7.65636504e-01
-6.26567423e-01 -9.54970419e-01 -7.17427611e-01 9.66592491e-01
1.87791571e-01 -1.00260782e+00 -2.52308697e-01 1.18407473e-01] | [13.582799911499023, 1.6349762678146362] |
96ae015e-4050-417d-82e3-c1824533292f | global-inference-with-explicit-syntactic-and | null | null | https://www.ijcai.org/proceedings/2022/0570 | https://www.ijcai.org/proceedings/2022/0570.pdf | Global inference with explicit syntactic and discourse structures for dialogue-level relation extraction | Recent research attention for relation extraction has been paid to the dialogue scenario, ie, dialoguelevel relation extraction (DiaRE). Existing DiaRE methods either simply concatenate the utterances in a dialogue into a long piece of text, or employ naive words, sentences or entities to build dialogue graphs, while the structural characteristics in dialogues have not been fully utilized. In this work, we investigate a novel dialogue-level mixed dependency graph (D2G) and an argument reasoning graph (ARG) for DiaRE with a global relation reasoning mechanism. First, we model the entire dialogue into a unified and coherent D2G by explicitly integrating both syntactic and discourse structures, which enables richer semantic and feature learning for relation extraction. Second, we stack an ARG graph on top of D2G to further focus on argument inter-dependency learning and argument representation refinement, for sufficient argument relation inference. In our global reasoning framework, D2G and ARG work collaboratively, iteratively performing lexical, syntactic and semantic information exchange and representation learning over the entire dialogue context. On two DiaRE benchmarks, our framework shows considerable improvements over the current best-performing baselines. Further analyses show that the model effectively solves the long-range dependence issue, and meanwhile gives explainable predictions. | ['Fei Li', 'Donghong Ji', 'Chenliang Li', 'Shengqiong Wu', 'Jingye Li', 'Hao Fei'] | 2022-07-30 | null | null | null | conference-2022-7 | ['dialog-relation-extraction'] | ['natural-language-processing'] | [ 1.92490444e-01 1.00163817e+00 -5.40487885e-01 -3.73335093e-01
-7.91757405e-01 -7.58701026e-01 1.08421779e+00 5.45070708e-01
-6.27034083e-02 8.34104419e-01 8.72697711e-01 -5.44974327e-01
-1.03195878e-02 -8.91614497e-01 -1.30459696e-01 -1.04390867e-01
2.06554994e-01 6.95466995e-01 3.50743771e-01 -7.10168660e-01
-1.09775841e-01 5.64409159e-02 -9.06231761e-01 4.78442758e-01
1.06794131e+00 7.78397977e-01 -2.69550800e-01 7.32938588e-01
-5.75818896e-01 1.36235583e+00 -8.39941800e-01 -8.13093603e-01
-2.84881294e-01 -7.36746132e-01 -1.76998305e+00 -5.31596765e-02
-1.58079505e-01 -2.67270952e-01 -1.77191481e-01 7.61925161e-01
2.98018277e-01 1.36734053e-01 5.99348605e-01 -1.16262162e+00
-1.84496403e-01 1.27835357e+00 -2.66576678e-01 -2.23111231e-02
8.86317909e-01 -3.17216307e-01 1.68173432e+00 -6.17280900e-01
7.54743874e-01 1.78895223e+00 5.25621295e-01 4.77955997e-01
-1.26140177e+00 -2.89638311e-01 5.27653635e-01 1.58333704e-01
-6.36968911e-01 -4.52522725e-01 9.79903340e-01 -2.48542354e-01
1.52627909e+00 4.25593495e-01 5.83133519e-01 1.00323939e+00
-1.08553380e-01 9.59357142e-01 8.21273625e-01 -5.50528049e-01
-9.60035548e-02 -9.83280614e-02 8.90885353e-01 6.09096706e-01
-9.50541124e-02 -5.19281268e-01 -4.38968003e-01 -3.21212858e-01
4.92518395e-01 -5.26886046e-01 -1.58348188e-01 2.71360166e-02
-9.37442601e-01 1.14880979e+00 2.87978292e-01 2.31193021e-01
-2.15468392e-01 -4.03656781e-01 7.38856852e-01 5.14285147e-01
6.12122893e-01 6.36298716e-01 -7.54948199e-01 -1.62786782e-01
-7.97438473e-02 4.64996755e-01 1.58458543e+00 9.65299249e-01
5.64224362e-01 -6.19335592e-01 -4.63692814e-01 1.03539121e+00
3.04294020e-01 1.32515924e-02 1.86030328e-01 -9.69499290e-01
9.49076176e-01 1.16897023e+00 -9.32991430e-02 -9.66502666e-01
-7.46161520e-01 1.70294508e-01 -6.44089282e-01 -5.11246324e-01
5.42576075e-01 -5.37158072e-01 -2.54015118e-01 1.78058243e+00
6.71247959e-01 -3.35678935e-01 6.86695337e-01 6.46397293e-01
1.55867100e+00 4.91066247e-01 2.19439581e-01 -4.23517644e-01
1.89032650e+00 -1.31888306e+00 -1.13172150e+00 -3.65537643e-01
1.11053860e+00 -5.68250895e-01 8.00487041e-01 -7.34752715e-02
-1.07781601e+00 -2.28116721e-01 -9.29965794e-01 -4.23062921e-01
-2.40926012e-01 -1.61174908e-01 9.37085927e-01 2.47125581e-01
-5.80304742e-01 3.04379433e-01 -4.90524411e-01 -2.52096713e-01
1.45715326e-01 2.20390052e-01 -3.13172102e-01 1.91627711e-01
-1.78040636e+00 1.20243740e+00 7.13998020e-01 8.67413580e-02
-1.27347514e-01 -2.65497237e-01 -1.36799228e+00 -1.60878077e-02
9.97091770e-01 -6.73385262e-01 1.56399155e+00 -2.44381383e-01
-1.81165111e+00 7.94377983e-01 -2.80418217e-01 -5.29620409e-01
1.46049887e-01 -4.94323015e-01 -1.61986738e-01 -7.57781975e-03
-2.73798164e-02 2.04916820e-01 3.36797833e-01 -1.14743602e+00
-5.77102959e-01 -2.82510847e-01 7.80527055e-01 6.38495147e-01
1.02438681e-01 3.10791850e-01 -4.20811176e-01 -4.05265421e-01
2.42050797e-01 -7.75608361e-01 -1.03788383e-01 -8.16076994e-01
-9.00637746e-01 -1.03989327e+00 6.77106917e-01 -8.44080687e-01
1.71801150e+00 -1.60203505e+00 5.48221409e-01 -4.52080369e-02
5.56445479e-01 3.90060902e-01 3.48217413e-02 7.40656078e-01
8.44641849e-02 5.61245009e-02 -1.08430013e-01 -2.81601191e-01
1.92465082e-01 5.89970410e-01 -1.21587969e-01 -1.22245952e-01
3.46953273e-01 1.29750049e+00 -1.00205910e+00 -7.87120104e-01
2.91150436e-02 3.97822559e-02 -6.19658411e-01 5.03646910e-01
-5.99635541e-01 4.56117600e-01 -8.31153214e-01 3.32975239e-01
2.95907974e-01 -4.72210139e-01 8.91247869e-01 -4.14195478e-01
2.19606489e-01 1.21207786e+00 -8.25079620e-01 1.64058638e+00
-6.31564498e-01 3.87781382e-01 1.99260026e-01 -1.02973998e+00
9.67949688e-01 4.54468310e-01 2.06726611e-01 -5.74059188e-01
2.43112445e-01 1.29372880e-01 1.57669276e-01 -5.05546093e-01
5.05681217e-01 8.25023055e-02 -5.03576696e-01 5.65529227e-01
1.89227194e-01 -2.77710229e-01 2.99955398e-01 5.60595155e-01
1.03030086e+00 1.81162804e-01 8.43945384e-01 1.01774633e-01
8.16946030e-01 6.66723773e-02 4.53271955e-01 4.07585472e-01
6.10803626e-02 1.48613304e-01 1.19385397e+00 -1.80884346e-01
-5.59117377e-01 -5.93821526e-01 3.89075987e-02 1.32106590e+00
2.93449491e-01 -9.31999505e-01 -8.57573390e-01 -1.33015168e+00
-7.02084824e-02 6.66148901e-01 -3.76987249e-01 4.71344553e-02
-1.03050816e+00 -6.87365890e-01 7.08735287e-01 4.66578871e-01
7.43923604e-01 -1.12374043e+00 -1.95396245e-01 4.07267958e-01
-7.12239206e-01 -1.42017102e+00 -6.83574453e-02 3.77037913e-01
-3.58820528e-01 -1.39349043e+00 9.29621980e-02 -7.81785905e-01
1.15641236e-01 -2.37430260e-01 1.27746451e+00 1.95994556e-01
2.79276460e-01 5.44428676e-02 -6.39970303e-01 -4.28795777e-02
-7.09161460e-01 3.93809050e-01 -4.90377396e-01 -4.18916672e-01
3.87201846e-01 -3.64807576e-01 -1.73070416e-01 2.36651093e-01
-3.83475095e-01 3.14257771e-01 2.34704956e-01 1.11118531e+00
7.27334172e-02 -1.51648387e-01 6.79885745e-01 -1.52000213e+00
1.18224931e+00 -4.48209882e-01 -1.05820492e-01 4.76210445e-01
-6.14921093e-01 3.26673567e-01 4.10964787e-01 -5.98356649e-02
-1.55974615e+00 -4.69445378e-01 -4.64258790e-01 4.21474040e-01
-1.90543588e-02 8.49893570e-01 -6.42540276e-01 5.68626523e-01
4.33345318e-01 -3.22735995e-01 7.69936144e-02 -3.73648256e-01
8.18893313e-01 7.78498530e-01 4.25468564e-01 -9.65148687e-01
5.53156078e-01 -1.90638781e-01 -1.56584769e-01 -6.75817907e-01
-1.42564559e+00 -4.25463349e-01 -9.67885733e-01 2.58642316e-01
9.69501197e-01 -6.79814756e-01 -8.58528078e-01 2.38359302e-01
-1.59048331e+00 -5.14232159e-01 -3.05475771e-01 2.85199970e-01
-4.20136511e-01 5.24377882e-01 -1.04903877e+00 -7.59667695e-01
-3.86683702e-01 -9.17068303e-01 1.05434096e+00 1.42764121e-01
-8.09081256e-01 -1.38516533e+00 1.03188813e-01 7.37213194e-01
1.14324912e-02 2.48526290e-01 1.31632519e+00 -1.39324999e+00
-8.42271745e-02 2.25866344e-02 -5.45337737e-01 1.07337542e-01
4.89876121e-01 -3.50676030e-01 -8.52390885e-01 1.79742634e-01
-1.62192201e-03 -5.29167891e-01 5.82782865e-01 -8.59928429e-02
4.13990200e-01 -7.50697672e-01 -3.31075221e-01 1.36454612e-01
5.34573257e-01 1.65985256e-01 3.76431555e-01 3.61160964e-01
7.79358745e-01 1.22572875e+00 8.62806618e-01 3.04401368e-01
1.12834132e+00 8.29354823e-01 1.62988186e-01 -1.49978921e-01
-3.13397318e-01 -3.39785486e-01 5.71512021e-02 1.05178261e+00
-5.05120158e-02 -3.49327356e-01 -8.09534490e-01 2.54770815e-01
-2.26035261e+00 -7.60417223e-01 -3.76758456e-01 1.51580501e+00
1.45965910e+00 1.77141041e-01 2.66463995e-01 1.51446983e-01
6.15017593e-01 4.85354573e-01 -2.64004320e-01 -5.26977897e-01
-1.84031233e-01 9.20795873e-02 -7.61310011e-02 9.29444849e-01
-1.22458386e+00 1.20614767e+00 5.57926941e+00 6.78622425e-01
-4.74998593e-01 -9.67066288e-02 3.59767795e-01 5.44995189e-01
-4.40916389e-01 3.91600043e-01 -9.88898337e-01 -1.27249673e-01
7.81000555e-01 -2.60682851e-01 1.80650100e-01 4.97493684e-01
-1.88533723e-01 -2.25899536e-02 -1.25945842e+00 6.07615590e-01
-1.00350402e-01 -1.32133043e+00 1.38667217e-02 -9.88356546e-02
2.86238223e-01 -5.37671328e-01 -5.76161623e-01 6.83559239e-01
7.57834077e-01 -1.02555954e+00 1.62593439e-01 4.07044999e-02
4.37537760e-01 -6.03640854e-01 1.08431029e+00 3.49299550e-01
-1.40023649e+00 5.80200888e-02 1.30768821e-01 -3.09822142e-01
2.11253434e-01 1.89018592e-01 -9.09084320e-01 1.15377438e+00
8.31408426e-02 7.58225858e-01 -5.76700330e-01 1.26969010e-01
-8.12182307e-01 4.87942308e-01 -2.21526816e-01 -1.43874034e-01
2.55540878e-01 -2.62385428e-01 4.98172373e-01 1.26875293e+00
-5.42846262e-01 4.96586204e-01 4.80768710e-01 5.23409843e-01
-1.85324937e-01 2.34443977e-01 -2.87382901e-01 -9.38043520e-02
6.31411314e-01 1.33838499e+00 -5.27837932e-01 -5.30972064e-01
-6.29352927e-01 7.37912178e-01 6.33883059e-01 1.71045870e-01
-5.32300115e-01 -4.03611571e-01 5.41904032e-01 -3.61756593e-01
-6.89213276e-02 1.01900123e-01 -1.15967371e-01 -1.18302345e+00
2.50315852e-02 -1.03722560e+00 7.84874320e-01 -2.74039477e-01
-1.30088449e+00 7.90280521e-01 3.22120577e-01 -6.03088737e-01
-8.35004628e-01 -4.18386579e-01 -5.83299756e-01 8.53480875e-01
-1.53517580e+00 -1.39907420e+00 -1.09404720e-01 5.34303665e-01
8.12227845e-01 8.62452313e-02 1.10988808e+00 -5.24555370e-02
-8.32097173e-01 5.40818632e-01 -6.92907870e-01 5.31075597e-01
5.08321047e-01 -1.49946856e+00 4.89721388e-01 3.86429369e-01
9.18212906e-02 7.41498232e-01 4.76461589e-01 -7.79874086e-01
-1.26279175e+00 -8.11118603e-01 1.30787265e+00 -6.75666511e-01
9.60881293e-01 -3.91840756e-01 -1.26115227e+00 7.81135380e-01
4.70368773e-01 -3.18117410e-01 6.34582579e-01 7.89748251e-01
-4.51998591e-01 5.07771134e-01 -8.55822325e-01 6.23404086e-01
1.16358924e+00 -8.09912086e-01 -1.39296079e+00 3.11001390e-01
1.16232383e+00 -8.24440479e-01 -1.15193892e+00 5.02433240e-01
3.60254645e-01 -7.28988826e-01 9.08585131e-01 -8.31177056e-01
5.17511249e-01 3.02684736e-02 5.13783377e-03 -1.14038885e+00
1.12041242e-01 -1.03723645e+00 -6.00573301e-01 1.63862216e+00
7.71020710e-01 -6.17736936e-01 4.47598368e-01 7.76347578e-01
-1.34517148e-01 -8.86400759e-01 -5.59307516e-01 -3.78623396e-01
7.17846975e-02 -2.55055726e-01 7.40053773e-01 1.00445330e+00
7.93904185e-01 1.49293196e+00 -1.41459838e-01 -7.96579197e-02
2.43994772e-01 5.69265008e-01 9.17318225e-01 -1.51514482e+00
-3.15965325e-01 -4.99331534e-01 1.97886109e-01 -1.40945280e+00
6.06744587e-01 -9.40171719e-01 2.49649701e-03 -2.02555466e+00
2.08420698e-02 -3.34565729e-01 3.05579513e-01 5.08132219e-01
-6.69220567e-01 -4.15263414e-01 -1.05527192e-02 2.37037808e-01
-7.78433502e-01 6.36450589e-01 1.39175081e+00 -6.65589571e-02
-5.48045158e-01 1.70893461e-01 -7.65473604e-01 9.13747847e-01
5.89656293e-01 -5.57633974e-02 -5.45589030e-01 -3.71421650e-02
1.07250474e-01 5.15715480e-01 3.29597592e-02 -4.99210209e-02
2.96629161e-01 -1.99699134e-01 -2.20435485e-01 -5.23622513e-01
3.14919889e-01 -4.49579060e-01 -4.17024404e-01 2.18409728e-02
-6.43513918e-01 -3.04719359e-01 -7.49721080e-02 4.14008230e-01
-5.68351924e-01 -1.97399795e-01 1.76145986e-01 6.23720624e-02
-4.67043519e-01 -8.86322409e-02 -8.40314776e-02 6.22256577e-01
7.52825201e-01 3.76477987e-02 -6.73322499e-01 -4.57756221e-01
-9.80807304e-01 6.21372283e-01 4.34365049e-02 4.26957697e-01
3.34836543e-01 -1.09645462e+00 -7.11197913e-01 -5.93735799e-02
7.09429830e-02 4.66430694e-01 -1.64754391e-01 8.11750054e-01
-4.24196571e-02 5.27459204e-01 3.75942111e-01 -3.18606645e-01
-1.50691497e+00 3.28156412e-01 1.33232504e-01 -1.05609298e+00
-8.37832034e-01 8.31855416e-01 2.22632661e-01 -8.05737138e-01
2.34226584e-01 -3.60717475e-01 -8.84429693e-01 5.40499032e-01
1.24334872e-01 3.35331038e-02 -8.79775137e-02 -7.06010938e-01
-3.18158954e-01 3.22707176e-01 -2.11501360e-01 -3.00678127e-02
1.14125824e+00 -4.42714453e-01 -3.67992848e-01 3.52191776e-01
8.81446660e-01 2.33736455e-01 -1.01465249e+00 -6.05513275e-01
5.65607309e-01 5.48148640e-02 -4.11537915e-01 -6.35304809e-01
-5.18845916e-01 5.05057752e-01 -6.62601233e-01 7.94681787e-01
7.08901465e-01 5.34102976e-01 8.75334859e-01 5.85434675e-01
7.56637976e-02 -9.13022399e-01 1.26480788e-01 1.12783849e+00
9.11008477e-01 -1.18457830e+00 1.20790049e-01 -1.12286830e+00
-1.02617264e+00 1.10735345e+00 6.30876303e-01 1.89880550e-01
3.79985392e-01 3.65437895e-01 6.97897300e-02 -5.22013426e-01
-9.54666853e-01 -4.55241352e-01 3.86106580e-01 5.02097130e-01
7.75087476e-01 -6.80392161e-02 -5.28782427e-01 1.03166640e+00
-5.10183334e-01 -6.41097367e-01 1.67233407e-01 9.32350397e-01
-1.84918433e-01 -1.54357493e+00 8.93185958e-02 3.19597781e-01
-3.21008801e-01 -2.06391990e-01 -9.46661353e-01 1.02512789e+00
-3.55532616e-01 1.34838498e+00 -1.55500442e-01 -3.57670099e-01
5.53993046e-01 3.68764371e-01 4.51351047e-01 -9.54972088e-01
-9.82818544e-01 5.21641448e-02 1.24970305e+00 -4.68739778e-01
-6.14567041e-01 -4.19289023e-01 -1.65981233e+00 -1.88669875e-01
-6.97954476e-01 6.49318516e-01 2.09349051e-01 1.43568826e+00
9.39398184e-02 8.28241289e-01 5.82227409e-01 -3.43298554e-01
-3.66158664e-01 -1.14502978e+00 -1.28803000e-01 4.34384316e-01
2.84733355e-01 -6.79116249e-01 -2.48413786e-01 -2.68294692e-01] | [12.38904094696045, 8.077890396118164] |
d08f4100-f65b-43f5-ac3e-cf03dc6a2fdf | seal-simultaneous-exploration-and | 2306.12623 | null | https://arxiv.org/abs/2306.12623v1 | https://arxiv.org/pdf/2306.12623v1.pdf | SEAL: Simultaneous Exploration and Localization in Multi-Robot Systems | The availability of accurate localization is critical for multi-robot exploration strategies; noisy or inconsistent localization causes failure in meeting exploration objectives. We aim to achieve high localization accuracy with contemporary exploration map belief and vice versa without needing global localization information. This paper proposes a novel simultaneous exploration and localization (SEAL) approach, which uses Gaussian Processes (GP)-based information fusion for maximum exploration while performing communication graph optimization for relative localization. Both these cross-dependent objectives were integrated through the Rao-Blackwellization technique. Distributed linearized convex hull optimization is used to select the next-best unexplored region for distributed exploration. SEAL outperformed cutting-edge methods on exploration and localization performance in extensive ROS-Gazebo simulations, illustrating the practicality of the approach in real-world applications. | ['Ramviyas Parasuraman', 'Ehsan Latif'] | 2023-06-22 | null | null | null | null | ['gaussian-processes'] | ['methodology'] | [-3.95015925e-01 7.77161717e-02 -1.24484189e-01 -1.54143244e-01
-1.15669203e+00 -5.61639726e-01 4.74491596e-01 5.56344569e-01
-7.80427396e-01 1.33917773e+00 -1.48305133e-01 -1.74493536e-01
-6.68447495e-01 -7.78074324e-01 -8.50430608e-01 -8.39261353e-01
-9.61426377e-01 7.14567840e-01 1.46858469e-01 -2.12908939e-01
5.32551050e-01 7.80277967e-01 -1.42242706e+00 -6.75039947e-01
1.09463513e+00 9.29363787e-01 7.88554966e-01 7.60832846e-01
2.79840804e-03 4.20478135e-01 -5.24965346e-01 3.74309093e-01
3.01069826e-01 9.47137177e-03 -4.36609089e-01 -3.14021349e-01
-3.17367017e-01 9.57556441e-02 1.70866489e-01 9.76245344e-01
5.01679480e-01 6.79693103e-01 3.24165404e-01 -1.30334449e+00
-2.17096508e-01 4.82680202e-01 -8.81246269e-01 2.49153659e-01
4.66002822e-01 -1.92888021e-01 5.92218816e-01 -7.24878907e-01
4.30299520e-01 1.35287237e+00 6.38497412e-01 -1.86579108e-01
-1.37075222e+00 -5.08946955e-01 7.39830732e-02 -1.16648227e-01
-1.75063753e+00 -2.65451789e-01 3.74779910e-01 -5.68713397e-02
7.08041131e-01 5.13546681e-03 5.07994056e-01 7.26058960e-01
1.15200722e+00 2.88157731e-01 1.18890953e+00 -2.02314481e-01
7.58150578e-01 1.18835941e-01 -2.59672254e-01 8.31323147e-01
9.25778270e-01 2.67329246e-01 -1.12367511e+00 -4.49568659e-01
7.58004487e-01 -2.04301119e-01 -2.45493323e-01 -8.93414974e-01
-1.37779212e+00 9.85885918e-01 7.42881358e-01 -8.14655274e-02
-7.10587263e-01 5.69638669e-01 -1.67169925e-02 1.51544675e-01
1.57408878e-01 7.05641687e-01 -6.82878345e-02 -3.94281358e-01
-7.94946790e-01 2.92488307e-01 1.01190102e+00 1.29011750e+00
1.15999854e+00 -3.65406983e-02 3.33743960e-01 2.22374007e-01
5.87004542e-01 7.24349976e-01 3.94508928e-01 -1.07985604e+00
5.16711235e-01 2.26907700e-01 6.26821041e-01 -1.21116841e+00
-7.08490312e-01 -6.61684036e-01 -5.30594170e-01 5.35614192e-01
1.41763855e-02 -5.76125085e-01 -8.41383696e-01 1.33857954e+00
4.36666042e-01 -9.24602598e-02 5.56483090e-01 7.69135535e-01
-5.78491762e-03 5.79246163e-01 -1.95257276e-01 4.14368883e-03
8.63273740e-01 -7.79545665e-01 -8.11440527e-01 -5.67053318e-01
4.74240214e-01 -3.54618311e-01 2.95338631e-01 5.38077414e-01
-6.99106157e-01 -3.59642535e-01 -1.33638728e+00 2.69407213e-01
-4.79735106e-01 -5.17959222e-02 7.86007524e-01 5.82659483e-01
-1.33788967e+00 3.25879335e-01 -1.27707255e+00 -4.19471681e-01
3.17050546e-01 5.42395234e-01 -4.50952351e-01 -2.16066949e-02
-6.65937424e-01 1.35071492e+00 5.69061458e-01 1.14755742e-01
-1.15545857e+00 -2.09476992e-01 -1.00658321e+00 -2.40171224e-01
2.86283642e-01 -3.62566113e-01 8.28801692e-01 -5.07082976e-02
-1.61983526e+00 -1.42416835e-01 -4.45056498e-01 -9.17801201e-01
4.86779451e-01 -6.37293518e-01 9.87274647e-02 5.13394885e-02
5.95281363e-01 6.52476132e-01 3.55919749e-01 -1.70011818e+00
-9.55287635e-01 -4.68491971e-01 -4.03317958e-01 8.27902198e-01
2.14962319e-01 -8.07417572e-01 -1.90506622e-01 1.09341271e-01
9.46911216e-01 -1.09270871e+00 -7.98077762e-01 -1.96122870e-01
-3.49537507e-02 1.13794856e-01 7.49958575e-01 -4.06851470e-01
6.46902382e-01 -1.93675673e+00 4.06680584e-01 7.06569314e-01
-7.54195303e-02 -8.50057304e-01 1.71564981e-01 8.06155086e-01
8.71522009e-01 -1.15008831e-01 -1.49061847e-02 -8.52383137e-01
-5.03002405e-02 6.45643115e-01 -2.87982553e-01 1.23788333e+00
-3.63052428e-01 6.51092827e-01 -1.20594990e+00 -5.84629357e-01
3.13796788e-01 2.90851384e-01 -1.76710084e-01 -7.14231953e-02
1.36320427e-01 7.27845311e-01 -6.35371804e-01 9.74880993e-01
6.16777778e-01 1.55713111e-01 -5.37001379e-02 5.07699013e-01
-6.28140032e-01 -2.03074262e-01 -1.38076127e+00 2.26936078e+00
-6.73796892e-01 6.31603420e-01 7.52336323e-01 -4.06262636e-01
1.27549851e+00 -3.37456793e-01 4.27001864e-01 -3.97090822e-01
3.67917232e-02 4.95785713e-01 -3.29799324e-01 1.16141364e-02
1.19753253e+00 1.73396870e-01 -3.44849974e-01 -2.49432474e-02
-6.18336536e-02 -5.87296546e-01 -2.35094130e-01 9.52089354e-02
1.10113347e+00 3.35790664e-01 3.98730427e-01 -7.47721612e-01
1.11000076e-01 5.62001407e-01 5.10277331e-01 1.14663887e+00
-2.81937957e-01 2.24211544e-01 -7.20675066e-02 -2.53859665e-02
-2.11991727e-01 -1.20518970e+00 -1.23966366e-01 8.94523382e-01
1.21913445e+00 -1.68763444e-01 1.39296278e-01 -3.16037774e-01
4.17596638e-01 9.79184926e-01 -6.76996648e-01 2.46438056e-01
-3.94039065e-01 -7.62504160e-01 3.45801175e-01 7.90710896e-02
4.02680188e-01 -4.63468105e-01 -1.37456930e+00 3.63526881e-01
-9.08040926e-02 -4.20871943e-01 3.50235373e-01 9.67763603e-01
-6.66791439e-01 -8.30280125e-01 -5.34446299e-01 -3.30781788e-01
8.05540025e-01 5.71370363e-01 3.20686102e-01 -4.10107851e-01
-1.24190236e-03 5.87809682e-01 -4.07117277e-01 -5.70382416e-01
1.31551474e-01 1.70940422e-02 2.79875517e-01 -4.48506325e-01
-1.81062847e-01 -5.10079563e-01 -3.59261632e-01 2.84445941e-01
-1.74465224e-01 -5.13202965e-01 8.65342081e-01 8.24649513e-01
8.86774957e-01 3.35832953e-01 4.13455099e-01 2.88883019e-02
8.43270898e-01 -9.40393090e-01 -1.04284036e+00 -8.83227661e-02
-5.88463545e-01 1.16193242e-01 -3.20509821e-02 -2.54072517e-01
-9.35185969e-01 2.37111166e-01 4.82891053e-01 -3.09158027e-01
2.17076451e-01 8.29817176e-01 1.91991627e-01 -9.30666625e-01
6.27614379e-01 2.62800992e-01 1.91791639e-01 -2.64222592e-01
4.86864448e-01 3.44745576e-01 6.15447819e-01 -5.16650796e-01
4.80219692e-01 1.03183293e+00 5.47239125e-01 -8.56576025e-01
-2.42533475e-01 -6.91412985e-01 -7.29475498e-01 -8.29223171e-02
4.48104769e-01 -1.01754606e+00 -7.64800489e-01 -2.80906469e-01
-1.06217420e+00 -2.13051096e-01 -2.25354970e-01 7.95466185e-01
-8.05434823e-01 3.00648957e-01 1.55092180e-01 -1.44657981e+00
4.50438112e-02 -1.19720304e+00 1.01232374e+00 4.30769295e-01
-1.52719282e-02 -1.06265473e+00 4.82428879e-01 -2.33227298e-01
4.18459743e-01 5.49924612e-01 -2.75526762e-01 -5.34092188e-01
-9.77792263e-01 -3.77064526e-01 -8.43756869e-02 -5.43859005e-01
-7.49057345e-03 -7.33803511e-01 -4.66294467e-01 -6.47107303e-01
-1.89117603e-02 -7.36972094e-02 7.80927598e-01 3.98956895e-01
3.80174875e-01 4.60541947e-03 -1.04806805e+00 5.83943844e-01
1.73056459e+00 2.21289158e-01 1.27839088e-01 8.78594160e-01
1.85199916e-01 5.27048588e-01 1.41188741e+00 6.83007300e-01
7.35719621e-01 4.10228014e-01 1.02427924e+00 2.66523719e-01
4.07472014e-01 -4.27177519e-01 3.10528368e-01 1.27957791e-01
2.99921155e-01 -4.99693692e-01 -1.05528545e+00 9.58364785e-01
-2.24150872e+00 -4.16493088e-01 -1.12540573e-01 2.29001355e+00
2.16487914e-01 -2.21602935e-02 -5.97648859e-01 -6.36042655e-01
4.15954143e-01 -1.18622884e-01 -4.30174261e-01 -5.58099672e-02
-8.34838152e-02 -2.45626464e-01 1.69319713e+00 9.21680450e-01
-9.67310429e-01 7.69471407e-01 6.46022987e+00 7.38521457e-01
-6.26074016e-01 3.47742885e-01 -1.67181805e-01 -7.21651688e-02
-1.16538532e-01 4.98869479e-01 -1.03266430e+00 1.45842031e-01
8.64204764e-01 -8.44009519e-02 3.49820495e-01 1.08206177e+00
2.42613584e-01 -1.36878920e+00 -4.72178906e-01 8.69699180e-01
1.04408242e-01 -1.31043983e+00 -5.51749945e-01 4.91880745e-01
8.62303913e-01 6.59945071e-01 -2.84164436e-02 9.81248394e-02
9.69301879e-01 -8.47704589e-01 8.80660236e-01 7.78342366e-01
1.65372416e-01 -9.63065386e-01 1.01334488e+00 5.76002419e-01
-1.25044692e+00 -3.18675429e-01 -4.68560457e-01 -7.42266774e-02
7.59762406e-01 3.42523783e-01 -1.13969040e+00 1.07127202e+00
7.63766885e-01 2.24751085e-01 -3.46783698e-01 1.45916498e+00
-2.21578836e-01 -2.29593873e-01 -9.27781999e-01 -3.94316703e-01
6.48002565e-01 -2.60068625e-01 1.05584264e+00 9.83682930e-01
8.06301296e-01 -3.39983463e-01 6.21455312e-01 8.50799739e-01
6.06586397e-01 -1.86496839e-01 -7.36180305e-01 3.67386699e-01
9.29686308e-01 1.08884096e+00 -1.01251793e+00 -9.20360833e-02
2.55975071e-02 8.09883654e-01 3.58304322e-01 2.79935986e-01
-5.06387770e-01 -6.64584935e-01 4.25490826e-01 -3.88004720e-01
3.13542277e-01 -1.04992735e+00 -5.22453666e-01 -5.02688527e-01
-4.09068853e-01 6.18448704e-02 1.32283583e-01 -4.11924720e-01
-6.32957816e-01 4.42891657e-01 2.24504530e-01 -1.08898890e+00
-2.24870011e-01 -1.00923009e-01 -3.83708864e-01 1.07396519e+00
-1.81321454e+00 -1.19396341e+00 -3.82901996e-01 1.11389711e-01
4.00177926e-01 -1.58728153e-01 5.79871953e-01 -4.17075515e-01
-5.46363965e-02 -2.49661505e-01 5.71600914e-01 -7.91347861e-01
5.15886545e-01 -1.49317002e+00 -1.15834745e-02 9.82044637e-01
5.45214638e-02 7.54347026e-01 1.16604185e+00 -1.36463237e+00
-1.85861957e+00 -9.35579658e-01 3.68648171e-01 -3.44537944e-01
7.52171695e-01 -1.91810146e-01 -4.77191508e-01 7.38132775e-01
1.16645373e-01 -2.17300370e-01 2.01386601e-01 9.01225060e-02
5.38645208e-01 1.09858669e-01 -1.18268919e+00 4.46953446e-01
5.19445121e-01 1.44740835e-01 -4.55587178e-01 1.52377248e-01
4.76957172e-01 -7.32532442e-01 -7.83219755e-01 3.80983084e-01
3.98464978e-01 -8.53053331e-01 8.03045750e-01 6.89316928e-01
-7.75394559e-01 -5.84379971e-01 -5.32452226e-01 -1.57218313e+00
8.13739523e-02 -8.91690671e-01 -3.21999528e-02 8.25601280e-01
2.76783496e-01 -9.63750899e-01 8.02006364e-01 2.05230832e-01
-3.16409141e-01 -5.14323294e-01 -1.54201567e+00 -1.09544992e+00
-3.21252435e-01 -2.12582797e-01 3.16543341e-01 3.00957471e-01
3.05463940e-01 -3.22690398e-01 -3.25768858e-01 9.59468901e-01
1.12113380e+00 4.10837500e-04 1.15640628e+00 -9.73957062e-01
7.86040202e-02 -3.45422924e-02 -4.49962795e-01 -1.12080300e+00
1.20086744e-01 -5.17667294e-01 7.88095713e-01 -1.84562314e+00
-5.21675885e-01 -9.58923340e-01 8.81838333e-03 1.08902715e-01
3.63626212e-01 -1.23566329e-01 -1.45103242e-02 4.90370423e-01
-8.77730429e-01 9.93820071e-01 7.35614538e-01 2.45590284e-01
-8.03182900e-01 6.52600266e-03 -2.73162752e-01 5.03954589e-01
7.53059447e-01 -5.42987525e-01 -4.04324114e-01 -3.44943583e-01
2.21381426e-01 3.46064389e-01 3.93275142e-01 -1.58953583e+00
1.06945908e+00 -3.38528812e-01 2.02615917e-01 -1.14871979e+00
9.42948580e-01 -8.87343645e-01 3.00422311e-01 6.88382745e-01
-4.98304293e-02 2.81489551e-01 4.55903679e-01 1.51306689e+00
-8.29114765e-02 -3.47409487e-01 4.67277467e-01 -1.45040721e-01
-1.19450176e+00 -6.95246086e-02 -4.80954856e-01 -6.56066597e-01
1.48355770e+00 -3.35945636e-01 -1.31429613e-01 -5.12151837e-01
-5.90084851e-01 9.62350667e-01 6.69169366e-01 2.01015428e-01
8.73434961e-01 -7.55456567e-01 -4.09724832e-01 1.29856795e-01
3.42868418e-02 2.29928225e-01 -1.33181795e-01 1.21625733e+00
-7.23070025e-01 3.45695436e-01 -1.31479248e-01 -6.58111095e-01
-8.90038848e-01 2.26556003e-01 1.96809947e-01 6.43858537e-02
-4.45906967e-01 1.03506804e+00 -3.27530682e-01 -3.51841837e-01
2.75840402e-01 -1.82300255e-01 6.61740303e-02 -1.44571275e-01
2.58636236e-01 7.91906834e-01 -3.71280342e-01 -5.16571164e-01
-5.03874242e-01 3.38616192e-01 4.39684480e-01 -8.10537577e-01
1.14997828e+00 -1.15219939e+00 -2.36670360e-01 7.43520796e-01
6.07333302e-01 3.45746785e-01 -1.32263815e+00 1.67430341e-01
4.05288488e-01 -6.37582481e-01 4.49131429e-01 -6.53251767e-01
-1.04333445e-01 3.37954372e-01 4.39823866e-01 -3.62285040e-02
2.51833111e-01 1.20310493e-01 7.77067393e-02 7.19270408e-01
1.34616566e+00 -9.64955807e-01 -2.45381013e-01 2.85271227e-01
9.82698083e-01 -1.24657679e+00 4.82794672e-01 -1.23315454e-01
-6.99032605e-01 1.01964879e+00 4.58232313e-01 -2.59599119e-01
5.46745658e-01 3.19010556e-01 -2.37981543e-01 -3.48807186e-01
-4.84010965e-01 -2.98152983e-01 -2.39235267e-01 7.74631143e-01
-4.88543600e-01 9.58015099e-02 -1.35782436e-01 1.39218777e-01
-2.22337335e-01 -4.03945178e-01 5.68091333e-01 1.48670673e+00
-1.11580455e+00 -6.04113102e-01 -9.77195323e-01 2.60341972e-01
1.24245733e-01 3.44190776e-01 2.45434180e-01 9.22176063e-01
-3.29936668e-02 1.17346394e+00 5.87967671e-02 -9.11092535e-02
-2.16240376e-01 -5.50521970e-01 2.00819448e-01 -5.04504561e-01
-8.62375200e-02 1.70820415e-01 1.56232283e-01 -9.80178714e-01
-1.85935810e-01 -8.22615206e-01 -1.50325930e+00 4.34692614e-02
-7.78607607e-01 8.45445812e-01 1.38755393e+00 7.48054922e-01
6.31963253e-01 3.23799044e-01 2.79138982e-01 -1.44830334e+00
-5.73071837e-01 -7.87193835e-01 -7.39238024e-01 -8.85953426e-01
6.58975899e-01 -1.27628326e+00 -4.36463028e-01 -6.72460258e-01] | [5.391463279724121, 0.9845951199531555] |
6f948086-8fae-4e97-9517-e5bda8604b2b | anomaly-detection-requires-better | 2210.10773 | null | https://arxiv.org/abs/2210.10773v1 | https://arxiv.org/pdf/2210.10773v1.pdf | Anomaly Detection Requires Better Representations | Anomaly detection seeks to identify unusual phenomena, a central task in science and industry. The task is inherently unsupervised as anomalies are unexpected and unknown during training. Recent advances in self-supervised representation learning have directly driven improvements in anomaly detection. In this position paper, we first explain how self-supervised representations can be easily used to achieve state-of-the-art performance in commonly reported anomaly detection benchmarks. We then argue that tackling the next generation of anomaly detection tasks requires new technical and conceptual improvements in representation learning. | ['Yedid Hoshen', 'Ron Abutbul', 'Eliahu Horwitz', 'Niv Cohen', 'Tal Reiss'] | 2022-10-19 | null | null | null | null | ['3d-anomaly-detection-and-segmentation'] | ['methodology'] | [ 4.21161830e-01 6.68854043e-02 -1.81866903e-02 -3.90877873e-01
-3.86772186e-01 -4.16522115e-01 7.17841029e-01 7.73899317e-01
1.01423554e-01 4.17824596e-01 -1.92253724e-01 -4.91874337e-01
-1.48399070e-01 -5.14788151e-01 -3.59154105e-01 -3.53790462e-01
-5.75859427e-01 4.42019373e-01 6.28697574e-02 -3.17421913e-01
6.04106784e-01 8.61792147e-01 -1.98278165e+00 3.35790932e-01
6.48621917e-01 1.22906291e+00 -1.02319276e+00 8.23414326e-01
-7.45724440e-01 1.02245796e+00 -6.76712275e-01 1.05240360e-01
1.02979302e-01 -5.29935062e-01 -8.51945162e-01 3.73837613e-02
5.99800885e-01 -4.20294292e-02 -2.12656602e-01 9.93305087e-01
1.60724279e-02 2.13797033e-01 9.73677874e-01 -1.60714102e+00
-5.79526782e-01 1.31242275e-01 -4.77690637e-01 9.26389039e-01
5.76295733e-01 1.62511319e-01 1.10473573e+00 -8.28155816e-01
3.02134186e-01 8.26475322e-01 6.15311980e-01 7.76578367e-01
-1.30936277e+00 -4.89516854e-01 4.90827918e-01 3.02164406e-01
-9.16457474e-01 -4.18053895e-01 7.96371579e-01 -5.45109272e-01
1.53455997e+00 3.86969626e-01 3.90988857e-01 1.17269444e+00
3.77094984e-01 6.70979798e-01 4.64671820e-01 -4.51821417e-01
4.50391829e-01 -2.88902074e-01 5.35130501e-01 5.85999429e-01
7.76661873e-01 2.26073489e-01 -3.39494407e-01 -5.30228913e-01
4.54987168e-01 4.18681204e-01 3.61603737e-01 -4.20082062e-01
-9.08704937e-01 8.61244500e-01 9.21499506e-02 5.58458030e-01
-3.49253714e-01 1.87484518e-01 1.04871845e+00 7.73350954e-01
5.26376247e-01 1.11471033e+00 -5.25415361e-01 -4.27188486e-01
-6.69452548e-01 1.54550955e-01 7.99819827e-01 6.96379364e-01
5.02519190e-01 9.03922021e-01 6.96941912e-02 6.35551333e-01
7.16327205e-02 2.92370617e-01 7.46914387e-01 -4.43533033e-01
-1.50156394e-02 9.26082075e-01 -2.11127013e-01 -8.48039746e-01
-4.43240136e-01 -3.32432687e-01 -8.75581443e-01 3.31115097e-01
2.68653244e-01 7.38288537e-02 -1.18072498e+00 9.80841994e-01
-1.08625188e-01 4.93956417e-01 1.90783262e-01 1.52918130e-01
5.95593393e-01 4.51518238e-01 6.02627061e-02 -1.82079077e-01
7.66189992e-01 -5.62357664e-01 -8.06978047e-01 -4.10645813e-01
7.79612243e-01 -6.52860820e-01 6.68271303e-01 6.38843477e-01
-7.00954199e-01 -2.27903113e-01 -1.11931920e+00 5.11041760e-01
-8.52498055e-01 -6.41960859e-01 8.57038677e-01 5.62648416e-01
-4.69402492e-01 7.32145488e-01 -9.82674718e-01 -6.32974982e-01
7.30458736e-01 2.25095361e-01 -3.74990940e-01 3.05102199e-01
-8.42962921e-01 6.99512243e-01 3.76083851e-01 -2.59288669e-01
-8.00758958e-01 -6.27192199e-01 -1.01371193e+00 -2.02242866e-01
3.26224983e-01 -6.30074665e-02 1.31285250e+00 -7.52197862e-01
-9.36868846e-01 9.42330062e-01 -3.60441655e-01 -7.70483255e-01
5.50792776e-02 -6.55710101e-01 -1.34912133e+00 -1.29908830e-01
-1.19955176e-02 -3.19910407e-01 1.15799963e+00 -1.00879109e+00
-9.62981582e-01 -2.77301997e-01 -7.06082761e-01 -4.57792163e-01
-2.73195267e-01 2.85022031e-03 5.18053234e-01 -7.59539783e-01
4.90551114e-01 -5.29669464e-01 -4.10249978e-01 -9.77046937e-02
-4.56384569e-01 -4.94295835e-01 1.34275663e+00 2.37473398e-02
1.19205856e+00 -2.22107530e+00 -4.61606562e-01 6.75538301e-01
3.05881679e-01 4.63250875e-01 4.50788997e-02 4.19180840e-01
-7.98927307e-01 1.17989734e-01 -4.63705361e-01 -1.98896274e-01
-1.80794746e-01 4.73056972e-01 -1.04140949e+00 4.86433625e-01
8.22018921e-01 7.20530927e-01 -1.12254786e+00 2.85488497e-02
3.40415895e-01 -1.60821855e-01 -3.87700975e-01 5.04798949e-01
-2.78514475e-01 5.26725233e-01 -4.86463696e-01 1.11660278e+00
2.24649400e-01 -1.72869474e-01 -3.05775821e-01 5.61903596e-01
-1.02501281e-01 3.98478121e-01 -1.04919207e+00 1.44583142e+00
-1.10122092e-01 7.46766925e-01 -5.72080910e-01 -1.65319037e+00
1.18733752e+00 2.09764823e-01 8.68726611e-01 -5.17979860e-01
-1.80055395e-01 3.70008200e-01 1.01846695e-01 -3.97701144e-01
2.84840852e-01 2.92807538e-03 -2.04759732e-01 6.25630200e-01
2.65016109e-01 -1.02437243e-01 2.01691806e-01 3.67335193e-02
1.61128175e+00 -3.43796134e-01 6.43198311e-01 -1.94902942e-01
7.19301045e-01 1.33815914e-01 4.21731919e-01 1.16395879e+00
-3.09867501e-01 5.15477180e-01 5.99401593e-01 -1.25471473e+00
-8.62155318e-01 -1.54526687e+00 -2.50551671e-01 1.15143096e+00
-3.56465816e-01 -5.06409407e-01 1.53854322e-02 -1.07801199e+00
3.75603735e-01 8.56834412e-01 -6.45842791e-01 -5.93344033e-01
-5.79151869e-01 -6.06241405e-01 6.78578258e-01 6.37925982e-01
-6.86452612e-02 -1.48995101e+00 -5.25467336e-01 3.16815376e-01
5.42830288e-01 -8.91428828e-01 3.15273732e-01 7.71953166e-01
-1.07600844e+00 -1.26979196e+00 2.41155997e-02 -4.82489407e-01
7.94965327e-01 -1.04937397e-01 1.44971430e+00 2.44007349e-01
-9.26311374e-01 6.23823643e-01 -4.74672675e-01 -9.94266033e-01
-4.44231391e-01 -5.43614030e-02 5.64999402e-01 -5.29176667e-02
9.82309401e-01 -8.01918805e-01 -1.19610377e-01 2.25503650e-02
-1.02078331e+00 -1.20954347e+00 3.41432273e-01 8.25451791e-01
5.83420396e-01 1.86685473e-01 8.56559515e-01 -1.28845143e+00
6.68485105e-01 -8.45120370e-01 -4.21538502e-01 -2.89356466e-02
-1.09742796e+00 2.14398116e-01 6.94863915e-01 -2.25385979e-01
-5.95378578e-01 3.23031507e-02 -3.35892946e-01 -4.39988852e-01
-8.63961518e-01 1.97701082e-01 3.29521060e-01 -1.04020193e-01
1.19450235e+00 4.30581182e-01 4.57563587e-02 -3.03028494e-01
2.46729791e-01 5.05818129e-01 5.91384947e-01 -6.23720586e-01
9.35094833e-01 5.55138648e-01 9.36166123e-02 -1.04975045e+00
-9.49995279e-01 -7.63363481e-01 -6.14570558e-01 1.75755292e-01
2.44382516e-01 -5.12819767e-01 -3.20663959e-01 3.37711662e-01
-8.75284672e-01 1.34476468e-01 -1.17813766e+00 -1.02187164e-01
-5.50192058e-01 4.35450315e-01 -1.69083118e-01 -1.05274904e+00
-4.05104369e-01 -4.25861388e-01 8.56844962e-01 -5.65047935e-02
-8.16072583e-01 -1.04709184e+00 4.97554749e-01 -4.18887973e-01
7.28665888e-01 4.51886892e-01 7.96796978e-01 -1.61677015e+00
-8.33989773e-03 -6.87854707e-01 2.80614644e-02 4.72199470e-01
7.16813922e-01 1.79192171e-01 -1.16954863e+00 -1.98719114e-01
-7.35651180e-02 -1.86279327e-01 9.51422393e-01 -9.64541920e-03
1.85862625e+00 -2.73057342e-01 -2.05606982e-01 5.36528111e-01
1.01719975e+00 2.08993316e-01 6.91170990e-01 3.81447941e-01
6.96568370e-01 4.33859557e-01 4.84943688e-01 5.65510392e-01
-2.19585493e-01 2.51419004e-02 4.26854640e-01 -7.30305398e-03
4.07935470e-01 1.31042609e-02 2.28703097e-01 4.58459556e-01
1.25083849e-01 1.23077542e-01 -1.29586303e+00 6.87389970e-01
-1.98532569e+00 -1.23347020e+00 -9.42136347e-02 2.12364674e+00
2.78728127e-01 6.52953386e-01 2.80088305e-01 6.63045049e-01
4.55454946e-01 9.30861905e-02 -9.41764474e-01 -9.93216872e-01
-1.68492757e-02 5.47629178e-01 2.21869588e-01 -8.22771043e-02
-1.49847996e+00 8.09105873e-01 8.33511543e+00 3.29226404e-01
-1.08655632e+00 -4.67723668e-01 3.53707045e-01 2.57833526e-02
-8.51567164e-02 -3.20988238e-01 -3.60773325e-01 3.29824567e-01
1.28316414e+00 -2.33563244e-01 -7.35826790e-02 1.23582983e+00
-2.70937264e-01 2.13048100e-01 -1.62183440e+00 1.10340083e+00
4.74438816e-02 -1.27245712e+00 3.15544486e-01 -1.58018127e-01
7.45245457e-01 2.50542313e-01 3.17022055e-01 3.90367866e-01
2.74692744e-01 -1.32765603e+00 -2.03059241e-01 5.43812990e-01
5.74831426e-01 -7.77525425e-01 8.76033843e-01 -1.72628924e-01
-1.07886839e+00 -4.05007392e-01 -1.35838881e-01 -2.84542441e-01
-9.59364995e-02 9.77123737e-01 -6.77373111e-01 3.93234402e-01
7.78690636e-01 1.19251525e+00 -5.84715128e-01 1.01959074e+00
-6.85277581e-02 9.00926352e-01 -3.16005319e-01 3.70635211e-01
2.70079851e-01 7.68813789e-02 9.00821388e-01 1.36091900e+00
1.52105719e-01 -3.43542129e-01 2.39642948e-01 6.30473137e-01
2.55547166e-01 -1.27512902e-01 -1.35873890e+00 -5.24629653e-01
2.00931028e-01 8.76946867e-01 -5.17436147e-01 -3.48034352e-01
-5.45774221e-01 8.49354029e-01 2.35322118e-01 1.56757131e-01
-3.24346274e-01 -5.24434745e-01 1.30518949e+00 2.11149901e-01
2.07978964e-01 -6.99826777e-02 -5.23294330e-01 -1.25133085e+00
-7.99605250e-02 -9.72676873e-01 9.11948502e-01 8.27434883e-02
-2.06866717e+00 4.22856390e-01 -2.24996150e-01 -1.50831342e+00
-7.90905118e-01 -8.92165244e-01 -1.24127710e+00 2.09859774e-01
-1.35990763e+00 -6.19185209e-01 -1.49835974e-01 6.54592752e-01
6.11151278e-01 -9.14876282e-01 1.47494698e+00 -3.47976871e-02
-7.40788221e-01 6.41545594e-01 2.23370761e-01 2.51643866e-01
6.09965801e-01 -1.57555759e+00 1.06878340e+00 8.83970082e-01
5.35535693e-01 5.80194712e-01 7.98261285e-01 -6.00910068e-01
-1.05636668e+00 -1.29899144e+00 4.85891074e-01 -7.97774374e-01
1.12841868e+00 -1.12240352e-01 -1.43215764e+00 8.24290812e-01
-3.24872404e-01 7.02591717e-01 1.11066389e+00 5.97850800e-01
-7.62775481e-01 2.67545134e-02 -1.36181760e+00 2.89096117e-01
1.09444487e+00 -7.20803857e-01 -1.07923532e+00 3.57783079e-01
2.47614875e-01 -1.75422519e-01 -5.26510239e-01 4.57486570e-01
3.15158248e-01 -8.28926325e-01 9.10158694e-01 -1.34041512e+00
2.52800524e-01 -2.35105574e-01 6.27259165e-02 -1.24928761e+00
-1.05914369e-01 -7.28613138e-01 -1.00622070e+00 8.33726943e-01
2.88902223e-01 -1.10328138e+00 9.29966211e-01 4.14597005e-01
-1.61016703e-01 -6.45830452e-01 -9.58856046e-01 -1.16846848e+00
1.12758018e-01 -6.42182589e-01 4.63895768e-01 1.11238968e+00
3.89901012e-01 -4.84399917e-03 -1.14660598e-01 1.23984562e-02
7.14390039e-01 -7.19645917e-02 6.13309503e-01 -1.95985615e+00
2.40155101e-01 -5.12236357e-01 -1.27187097e+00 -5.30694664e-01
2.46074826e-01 -7.16415405e-01 -4.56313081e-02 -9.24419582e-01
-4.55180764e-01 -1.01357386e-01 -9.34640408e-01 6.15699232e-01
-2.22690590e-02 1.50088251e-01 -5.44164300e-01 2.20925838e-01
-9.29555893e-01 3.73836786e-01 2.73442775e-01 -1.45550206e-01
-1.61230579e-01 1.80569947e-01 -7.68787205e-01 1.03964031e+00
1.19048142e+00 -5.36893010e-01 -2.36353189e-01 1.01757340e-01
1.70476019e-01 -8.41475904e-01 2.39110678e-01 -1.28427875e+00
1.40148476e-01 -1.16966620e-01 5.80033600e-01 -4.75191891e-01
-9.69346911e-02 -9.65675294e-01 -6.28503263e-01 4.69108731e-01
-3.16043019e-01 5.05410016e-01 5.76549470e-01 9.25798476e-01
-3.71296883e-01 -1.17257312e-01 6.10535622e-01 -7.36633092e-02
-1.04350960e+00 2.18991652e-01 -7.12805092e-01 2.84011841e-01
1.21650958e+00 -1.62805036e-01 -2.29930550e-01 -2.66008675e-01
-7.67819047e-01 2.28993475e-01 1.78698555e-01 6.93131506e-01
8.94301593e-01 -1.15713048e+00 -6.02742016e-01 7.44244397e-01
8.52994621e-01 8.26115906e-02 -1.10885002e-01 2.60091454e-01
-4.41952944e-01 3.22111845e-01 -4.67199773e-01 -7.49051392e-01
-9.08871651e-01 5.56534529e-01 2.92178988e-01 -2.76619822e-01
-9.11290705e-01 5.37109315e-01 -3.15804660e-01 -2.83654362e-01
3.93183351e-01 -9.08028409e-02 -1.39087841e-01 -3.05340409e-01
8.10744464e-01 4.49876308e-01 1.96650475e-01 -3.43378037e-01
-6.35094047e-01 2.61673301e-01 -6.66510582e-01 4.22850549e-01
1.32773316e+00 4.19143885e-01 -8.67461041e-02 1.06421709e+00
8.91795874e-01 -2.85690159e-01 -5.24776399e-01 -1.72830552e-01
8.30799997e-01 -4.69314575e-01 -2.72240579e-01 -4.76642251e-01
-8.10938060e-01 7.26035833e-01 7.78720915e-01 7.73012519e-01
8.99827003e-01 6.66067228e-02 5.57755709e-01 1.05897403e+00
9.99562740e-02 -1.17290354e+00 4.30026680e-01 8.77869368e-01
7.75153339e-01 -1.67736006e+00 1.78374693e-01 -1.62013352e-01
-3.89232606e-01 1.20738328e+00 8.87067139e-01 -7.27210402e-01
9.04748619e-01 1.83184892e-01 1.19236588e-01 -6.70799494e-01
-6.46573544e-01 -3.03701192e-01 5.26573122e-01 8.89809906e-01
6.41589344e-01 -1.93506852e-01 3.06040168e-01 2.72217125e-01
2.85245627e-02 -5.36224961e-01 4.03637409e-01 1.31065416e+00
-4.96975660e-01 -1.21368504e+00 -2.21891090e-01 1.19900751e+00
-7.06013262e-01 8.76188800e-02 -5.53149879e-01 4.04862791e-01
-1.38328031e-01 9.50022936e-01 6.65414155e-01 -2.12112829e-01
5.91117740e-01 5.26924670e-01 -8.84931684e-02 -1.05114686e+00
-3.51535171e-01 -6.53317213e-01 -6.96588457e-02 -9.17311430e-01
-4.13729772e-02 -7.13063657e-01 -1.36675704e+00 -9.56462100e-02
-9.21208337e-02 2.48142466e-01 3.48771274e-01 1.03767419e+00
5.24320006e-01 8.03583324e-01 6.66007876e-01 -4.55334008e-01
-5.04918754e-01 -7.88816631e-01 -6.99998975e-01 7.02137232e-01
9.00091410e-01 -7.10593879e-01 -8.70198905e-01 -2.07678333e-01] | [7.559846878051758, 2.4914379119873047] |
01715ad9-095f-4091-9813-24df515e453b | convolutional-neural-networks-for-image-spam | 2204.01710 | null | https://arxiv.org/abs/2204.01710v1 | https://arxiv.org/pdf/2204.01710v1.pdf | Convolutional Neural Networks for Image Spam Detection | Spam can be defined as unsolicited bulk email. In an effort to evade text-based filters, spammers sometimes embed spam text in an image, which is referred to as image spam. In this research, we consider the problem of image spam detection, based on image analysis. We apply convolutional neural networks (CNN) to this problem, we compare the results obtained using CNNs to other machine learning techniques, and we compare our results to previous related work. We consider both real-world image spam and challenging image spam-like datasets. Our results improve on previous work by employing CNNs based on a novel feature set consisting of a combination of the raw image and Canny edges. | ['Mark Stamp', 'Katerina Potika', 'Fabio Di Troia', 'Tazmina Sharmin'] | 2022-04-02 | null | null | null | null | ['spam-detection'] | ['natural-language-processing'] | [ 5.97391427e-01 -3.52021456e-01 2.33759046e-01 -2.76780516e-01
-1.23735413e-01 -4.86668795e-01 8.07152390e-01 4.57643643e-02
-5.69862545e-01 3.35522711e-01 -5.74048162e-02 -5.64317107e-01
3.24647307e-01 -9.07728732e-01 -6.71725333e-01 -4.90746796e-01
2.66365379e-01 1.11611485e-01 5.97267568e-01 -2.86953926e-01
6.74894512e-01 4.54658836e-01 -1.32111120e+00 6.86191082e-01
9.05551493e-01 1.14028752e+00 -5.14904037e-02 5.57710886e-01
-3.59740764e-01 8.45081091e-01 -8.38612854e-01 -5.42441607e-01
3.34345609e-01 -4.03476149e-01 -7.61386096e-01 4.37235534e-01
8.60389233e-01 -4.55392838e-01 -8.14874411e-01 1.55082512e+00
1.51070401e-01 3.47984931e-03 3.45297247e-01 -1.37651682e+00
-9.52666581e-01 1.09007858e-01 -3.05535704e-01 3.21716458e-01
8.98959637e-02 1.71123281e-01 4.09223020e-01 -6.58062696e-01
5.83293200e-01 1.52782750e+00 7.56768942e-01 4.51068074e-01
-1.19491756e+00 -4.64868307e-01 9.01517719e-02 4.87648398e-01
-7.04728246e-01 -1.88930966e-02 6.20592356e-01 -2.02992737e-01
5.04355550e-01 2.45061427e-01 4.38487202e-01 1.34100127e+00
3.91908079e-01 1.03762221e+00 1.25559032e+00 -3.67786825e-01
5.68038858e-02 9.63306874e-02 3.38349134e-01 7.54068136e-01
5.23538768e-01 1.91688761e-01 -2.33283699e-01 -3.93359691e-01
4.91449237e-01 3.53424191e-01 -2.75242001e-01 -2.05093712e-01
-8.98510575e-01 1.10903668e+00 8.30936551e-01 2.89175689e-01
-6.23554224e-04 3.51118922e-01 5.86337924e-01 7.17497826e-01
6.86067522e-01 2.33698577e-01 -1.17102414e-01 2.21800327e-01
-9.16539431e-01 1.95755512e-01 1.01531005e+00 7.24473357e-01
5.97120762e-01 1.37903810e-01 -1.21303692e-01 7.26666629e-01
8.41288716e-02 4.22206730e-01 8.05236399e-01 -6.35558724e-01
1.59958109e-01 9.45775032e-01 1.50891647e-01 -1.35292304e+00
-1.82646111e-01 -1.80973023e-01 -8.98710728e-01 3.37853670e-01
7.43288755e-01 2.71862537e-01 -1.32853353e+00 9.88950670e-01
-1.04227647e-01 1.95020676e-01 -4.12099749e-01 1.22521102e+00
8.37723613e-01 4.45571035e-01 -5.91164976e-02 4.06966098e-02
1.30667901e+00 -1.37657702e+00 -6.56220794e-01 -3.76102090e-01
3.00081640e-01 -8.13390553e-01 1.09807110e+00 4.92837518e-01
-5.84175050e-01 -3.90865505e-01 -7.74464071e-01 6.87035397e-02
-1.02909648e+00 1.52370960e-01 3.17061931e-01 8.35818112e-01
-9.26706314e-01 8.41462672e-01 -5.00042975e-01 -6.55669272e-01
8.61306727e-01 4.68742043e-01 -3.55915964e-01 2.95967758e-02
-1.02022314e+00 9.15722787e-01 2.28893474e-01 9.46373791e-02
-7.46027112e-01 3.25089842e-02 -5.15763819e-01 1.12819262e-02
4.23089385e-01 -4.44833428e-01 1.29865038e+00 -1.73922992e+00
-1.08571720e+00 1.09590137e+00 -4.49530706e-02 -9.30773377e-01
8.28277051e-01 -2.01678231e-01 -3.84336263e-01 4.99691367e-01
-5.21212891e-02 2.52144873e-01 1.52879572e+00 -1.47866058e+00
-5.24498820e-01 -3.55108857e-01 -2.40203723e-01 -5.29898643e-01
-6.64843380e-01 2.27193356e-01 -2.41233930e-01 -9.97085571e-01
1.55138761e-01 -9.79132771e-01 -2.24543914e-01 7.18995258e-02
-4.55533773e-01 -4.92956154e-02 1.75846565e+00 -6.04631662e-01
1.11253893e+00 -2.00781989e+00 -3.14152062e-01 1.83310136e-01
4.62240666e-01 8.55536163e-01 -2.61587143e-01 2.73979515e-01
-8.13193843e-02 2.51950681e-01 -3.88397008e-01 2.70388499e-02
-9.09443274e-02 8.34731758e-02 -5.82357407e-01 7.64183760e-01
2.08068177e-01 1.21329701e+00 -8.65699053e-01 -5.12111127e-01
3.65593821e-01 3.06581613e-02 1.10434055e-01 1.01922318e-01
-2.70309299e-01 4.06690501e-03 -4.85556364e-01 8.43343556e-01
9.82960999e-01 -3.04951727e-01 -2.40402948e-02 -1.31727859e-01
2.85475433e-01 -2.47735921e-02 -5.43831587e-01 9.53938782e-01
-3.64097506e-01 9.74493384e-01 2.06995875e-01 -1.49044585e+00
7.27078676e-01 8.12650472e-02 1.58914804e-01 -4.45502102e-01
5.67803800e-01 2.98468500e-01 -8.46986398e-02 -5.91381192e-01
5.14844179e-01 1.59260899e-01 1.87849492e-01 6.82853818e-01
-8.98121372e-02 -7.13492334e-02 1.55265152e-01 4.95168477e-01
1.43123329e+00 -2.89421290e-01 -2.30848859e-03 -3.08984369e-01
7.76740074e-01 2.10111126e-01 -1.82370290e-01 1.42993557e+00
-5.30792773e-01 6.85485542e-01 7.17512071e-01 -8.10502052e-01
-1.10767770e+00 -1.10545194e+00 -8.15688297e-02 8.83405447e-01
4.65099782e-01 -1.10148072e-01 -9.95243073e-01 -1.51011550e+00
4.97315854e-01 1.48531511e-01 -6.12538874e-01 -2.65498549e-01
-9.18192744e-01 -1.00637555e+00 5.12564480e-01 2.11014062e-01
7.91589558e-01 -1.31223762e+00 -2.55980045e-01 3.24739277e-01
7.59193301e-02 -1.15282905e+00 -5.60484171e-01 1.44392535e-01
-1.09210134e+00 -1.35839427e+00 -5.53647339e-01 -1.00522959e+00
1.04712451e+00 8.15631807e-01 1.17558944e+00 8.49042177e-01
-6.38161659e-01 3.46462309e-01 -6.42205298e-01 -2.29602307e-01
-8.05076301e-01 -1.20868273e-02 -2.28691101e-01 1.71365529e-01
7.08961785e-01 -1.82279333e-01 -6.60745800e-01 4.43698585e-01
-1.47209311e+00 -3.03771645e-01 6.28816724e-01 1.27662408e+00
-1.49100080e-01 6.61152527e-02 7.62481272e-01 -1.11270320e+00
9.06976998e-01 -2.86789954e-01 -6.32722259e-01 1.88072369e-01
-6.44919991e-01 -1.60774305e-01 9.66965020e-01 -5.92060745e-01
-8.91761482e-01 2.50187933e-01 5.91026731e-02 -2.11048856e-01
-2.61533260e-01 -1.21077128e-01 1.77456304e-01 -7.08884895e-01
7.08743989e-01 3.76952052e-01 3.10210437e-01 -3.60411078e-01
4.11766440e-01 1.35684299e+00 5.21028876e-01 -1.99277863e-01
8.74887049e-01 9.73337173e-01 -2.42627487e-01 -7.04649746e-01
-7.77167976e-01 -5.99917054e-01 -4.50332910e-01 -2.28797808e-01
3.79452914e-01 -3.25260520e-01 -7.95063615e-01 8.31930041e-01
-1.26076436e+00 1.19391017e-01 9.93985981e-02 -8.75209123e-02
-5.36876082e-01 1.23574805e+00 -1.07771528e+00 -7.84575582e-01
-4.51779842e-01 -1.08769166e+00 1.16530955e+00 1.14865182e-02
5.69755062e-02 -9.66353416e-01 -6.17995262e-01 4.09822881e-01
6.42863989e-01 9.57774930e-03 7.17624843e-01 -8.95343006e-01
-8.23928893e-01 -7.73819685e-01 -7.79817760e-01 7.72302091e-01
2.94939075e-02 -4.66360420e-01 -7.88522780e-01 -4.02699828e-01
2.13729352e-01 -1.68242827e-01 1.57496023e+00 1.24251768e-01
1.56665421e+00 -4.74842370e-01 -3.88662934e-01 2.85245478e-01
1.49191630e+00 1.06734363e-02 7.98570573e-01 8.85599256e-01
4.15860236e-01 4.77261156e-01 3.16372812e-01 8.76691192e-03
-5.24315357e-01 3.57781708e-01 7.05415308e-01 -1.02311015e-01
-1.74814872e-02 -7.56044462e-02 1.73032477e-01 2.61547953e-01
5.18767655e-01 -4.70897168e-01 -6.11267865e-01 5.26500523e-01
-2.27524161e+00 -8.59011292e-01 -6.53988898e-01 1.79583478e+00
5.51353157e-01 1.81023896e-01 3.61554176e-02 -1.34573430e-01
1.12029505e+00 3.18019949e-02 -3.57498646e-01 -3.53783399e-01
-3.02354217e-01 4.03630078e-01 7.72547960e-01 1.48649409e-01
-1.73849142e+00 1.37743294e+00 7.17214251e+00 9.81443822e-01
-1.19907653e+00 1.55169189e-01 5.08726060e-01 1.65799052e-01
2.60674447e-01 -1.82362068e-02 -2.95644522e-01 8.95858228e-01
7.43098736e-01 -5.95571287e-03 4.70212132e-01 1.01578999e+00
2.45484427e-01 -3.46363515e-01 -5.39562345e-01 7.24529088e-01
4.06459987e-01 -1.22404408e+00 1.75905630e-01 -1.58521309e-01
7.37369239e-01 -1.19413689e-01 1.45264104e-01 -1.39497414e-01
4.09873366e-01 -1.11128831e+00 4.68522698e-01 2.45684177e-01
2.36174911e-01 -2.42989838e-01 1.14188242e+00 1.01142019e-01
-5.57428360e-01 -3.54359567e-01 -3.90866280e-01 1.59348339e-01
7.30438605e-02 6.97221577e-01 -6.73353016e-01 2.68592477e-01
8.79621267e-01 7.43689775e-01 -1.04202902e+00 1.28244925e+00
-1.36304781e-01 6.75681710e-01 -7.53563941e-02 -2.82839805e-01
4.29112315e-01 -2.65662968e-01 5.54031253e-01 1.46309316e+00
3.64942476e-02 -5.65976977e-01 3.28920111e-02 9.97573972e-01
-2.51587510e-01 4.54128012e-02 -7.12133467e-01 -1.32739604e-01
-1.43808886e-01 1.23303139e+00 -1.11945939e+00 -7.04682469e-01
-4.49376792e-01 1.50960779e+00 -1.65235072e-01 3.91944766e-01
-5.40269792e-01 -6.62379622e-01 3.91592413e-01 1.44840330e-01
2.64618963e-01 -1.72998935e-01 -4.16149497e-01 -1.26427031e+00
-3.05530839e-02 -8.97319376e-01 1.27388671e-01 -6.77380800e-01
-1.84405339e+00 6.02530301e-01 -4.54323739e-01 -1.32951033e+00
2.09785402e-01 -1.18689597e+00 -7.90317476e-01 5.32606542e-01
-1.36158156e+00 -1.28553915e+00 -5.07979810e-01 2.37851053e-01
6.03803635e-01 -5.05635917e-01 3.88690770e-01 2.04277456e-01
-2.80362755e-01 1.79690361e-01 4.27294642e-01 2.85912514e-01
8.36737454e-01 -1.20537937e+00 8.61680150e-01 7.47243404e-01
-4.15358983e-04 4.58909005e-01 7.71371543e-01 -8.69571030e-01
-1.07493472e+00 -1.15695906e+00 6.62955821e-01 -5.14519870e-01
9.38467920e-01 -3.29230487e-01 -1.00330091e+00 4.93189067e-01
2.91446388e-01 8.61409232e-02 1.64374821e-02 -6.29858434e-01
-3.56355667e-01 1.99826613e-01 -1.24915051e+00 4.40847784e-01
1.00266993e+00 -4.02862221e-01 -6.34567022e-01 5.73897779e-01
3.44538540e-01 -8.45567137e-02 2.45605968e-03 1.52987674e-01
5.13833106e-01 -1.08768463e+00 1.04622328e+00 -9.29469764e-01
4.51288790e-01 -2.87324637e-01 3.04786831e-01 -1.19157040e+00
-1.24493524e-01 -2.18855545e-01 -7.18360841e-02 5.33785880e-01
-2.50965804e-01 -6.40293598e-01 1.17042720e+00 -1.90644518e-01
9.54588205e-02 -3.03913385e-01 -9.07901645e-01 -1.10307002e+00
8.42919946e-02 -6.57947287e-02 2.44972721e-01 6.25567794e-01
-1.90898836e-01 -1.79197431e-01 -5.46427250e-01 -2.09126219e-01
7.40080118e-01 2.01152176e-01 7.04649925e-01 -1.22585344e+00
4.05515730e-01 -6.50514841e-01 -6.35314822e-01 -1.16130555e+00
5.17594814e-01 -8.72205794e-01 1.29273623e-01 -1.32199955e+00
1.13305382e-01 2.67194092e-01 6.49408028e-02 1.24282293e-01
-2.09134996e-01 6.24043107e-01 2.30446830e-01 2.36910373e-01
-7.60936797e-01 2.94136792e-01 1.15749121e+00 -6.69093907e-01
2.47772217e-01 2.14578196e-01 -4.34351265e-01 8.25123668e-01
9.90589738e-01 -5.68929493e-01 3.24433029e-01 -1.40509501e-01
-1.78167954e-01 -5.28139830e-01 9.08684909e-01 -7.59125292e-01
2.43183509e-01 6.27102107e-02 3.64718229e-01 -4.47672278e-01
-2.96220910e-02 -1.04542208e+00 -5.67653835e-01 9.91082370e-01
-2.23633334e-01 -2.23375008e-01 1.05017126e-02 9.37677503e-01
-4.02156115e-01 -6.64444268e-01 1.14806008e+00 -5.83729267e-01
-8.14751565e-01 -2.80737597e-02 -8.92836869e-01 -3.08045834e-01
9.37667966e-01 -2.53451347e-01 -9.06593680e-01 -6.94324672e-01
-6.87684655e-01 1.70419943e-02 6.75622702e-01 7.68956423e-01
8.07748139e-01 -1.19952977e+00 -3.66500258e-01 3.67305249e-01
1.37660325e-01 -7.88927853e-01 -1.09145239e-01 8.55347097e-01
-1.08920467e+00 3.94467235e-01 -1.54719412e-01 -4.72239941e-01
-1.48078120e+00 7.32336581e-01 4.96613264e-01 1.93721473e-01
-6.99029565e-01 4.02341843e-01 4.79700044e-02 -3.63141805e-01
3.71028930e-01 8.81420672e-02 5.87979518e-02 -2.92332768e-01
5.21224558e-01 3.55552286e-01 1.83714032e-01 -5.46320260e-01
-1.68562546e-01 4.89217490e-02 -2.03259408e-01 4.44050878e-01
9.86106277e-01 -2.12258119e-02 -6.58226192e-01 -2.29702890e-01
1.41820550e+00 -4.23933506e-01 -7.01860130e-01 -2.95847803e-01
4.53368336e-01 -1.09918749e+00 7.20434785e-02 -6.87140763e-01
-1.07619452e+00 7.50094354e-01 9.35826540e-01 8.76334548e-01
9.37305093e-01 -1.08586170e-01 1.22251868e+00 6.39399469e-01
9.16721597e-02 -1.48491538e+00 5.46259940e-01 5.88868678e-01
7.12470174e-01 -1.74010015e+00 -4.91254553e-02 -6.06097698e-01
-2.49329373e-01 1.50100780e+00 7.24849641e-01 -6.13667130e-01
6.85772657e-01 3.93986627e-02 1.42955810e-01 -7.28395507e-02
-3.71484756e-01 -4.72237647e-01 5.68172783e-02 5.08387208e-01
1.75284043e-01 -2.47886166e-01 -7.16426790e-01 3.16617697e-01
1.54398888e-01 1.40037239e-01 6.99012399e-01 1.04552114e+00
-9.74429727e-01 -1.23473418e+00 -7.65019715e-01 9.98445570e-01
-6.28086090e-01 -1.11247987e-01 -9.60974932e-01 4.82171714e-01
8.09001401e-02 1.04007435e+00 1.70929372e-01 -3.29322934e-01
2.73866713e-01 -1.76063669e-03 2.22956225e-01 -3.83366376e-01
-9.57208276e-01 -2.06216037e-01 -2.57565975e-01 -2.96444833e-01
-3.42994809e-01 -1.87919810e-01 -6.90023303e-01 -5.63353598e-01
-6.44499660e-01 1.35352428e-04 6.58337831e-01 7.27430344e-01
6.20768666e-02 1.96707711e-01 5.79961836e-01 -6.72999978e-01
-8.53107572e-01 -9.67335999e-01 -6.56467438e-01 1.12998247e+00
3.89831007e-01 -3.21444631e-01 -8.21946502e-01 1.84823602e-01] | [7.788241386413574, 9.959676742553711] |
b1ee856e-380f-4430-b7b0-40e4c09bde4a | align-with-purpose-optimize-desired | 2307.01715 | null | https://arxiv.org/abs/2307.01715v2 | https://arxiv.org/pdf/2307.01715v2.pdf | Align With Purpose: Optimize Desired Properties in CTC Models with a General Plug-and-Play Framework | Connectionist Temporal Classification (CTC) is a widely used criterion for training supervised sequence-to-sequence (seq2seq) models. It enables learning the relations between input and output sequences, termed alignments, by marginalizing over perfect alignments (that yield the ground truth), at the expense of imperfect alignments. This binary differentiation of perfect and imperfect alignments falls short of capturing other essential alignment properties that hold significance in other real-world applications. Here we propose $\textit{Align With Purpose}$, a $\textbf{general Plug-and-Play framework}$ for enhancing a desired property in models trained with the CTC criterion. We do that by complementing the CTC with an additional loss term that prioritizes alignments according to a desired property. Our method does not require any intervention in the CTC loss function, enables easy optimization of a variety of properties, and allows differentiation between both perfect and imperfect alignments. We apply our framework in the domain of Automatic Speech Recognition (ASR) and show its generality in terms of property selection, architectural choice, and scale of training dataset (up to 280,000 hours). To demonstrate the effectiveness of our framework, we apply it to two unrelated properties: emission time and word error rate (WER). For the former, we report an improvement of up to 570ms in latency optimization with a minor reduction in WER, and for the latter, we report a relative improvement of 4.5% WER over the baseline models. To the best of our knowledge, these applications have never been demonstrated to work on a scale of data as large as ours. Notably, our method can be implemented using only a few lines of code, and can be extended to other alignment-free loss functions and to domains other than ASR. | ['Tal Rosenwein', 'Amnon Shashua', 'Jacob Bitterman', 'Oren Tadmor', 'David Zar', 'Yael Ben-Oren', 'Ayana Shenhav', 'Noam Wies', 'Ronen Katsir', 'Maya Alroy', 'Eliya Segev'] | 2023-07-04 | null | null | null | null | ['speech-recognition', 'automatic-speech-recognition'] | ['speech', 'speech'] | [ 7.84148037e-01 4.15737629e-02 -1.25200987e-01 -4.14355576e-01
-1.33861816e+00 -8.53727460e-01 6.43742561e-01 -7.32212234e-03
-5.73049247e-01 7.57099926e-01 -3.88383903e-02 -7.95751929e-01
-2.60914005e-02 -3.30006272e-01 -7.19404101e-01 -7.55127192e-01
-2.17854798e-01 3.80153686e-01 4.69467163e-01 -2.60540873e-01
-4.92565632e-02 3.88222218e-01 -1.47543228e+00 3.65238518e-01
5.95245302e-01 9.08825099e-01 3.25322956e-01 7.86112547e-01
1.83687523e-01 7.12898493e-01 -7.44792223e-01 -3.38000536e-01
1.30764782e-01 -4.01073694e-01 -1.03405523e+00 -1.63794965e-01
2.69659191e-01 -6.29722746e-03 -3.45998645e-01 7.22335041e-01
5.44030666e-01 1.79430872e-01 4.25243229e-01 -9.83888328e-01
-9.23176855e-02 4.65844899e-01 -2.66283363e-01 4.45324123e-01
2.85579681e-01 2.31751248e-01 1.44596338e+00 -6.14550054e-01
3.24358553e-01 9.22976851e-01 5.15971959e-01 6.12322271e-01
-1.34528661e+00 -4.75621492e-01 2.71873534e-01 5.91612197e-02
-1.39211059e+00 -8.45409214e-01 1.89926356e-01 -2.84432858e-01
1.59513354e+00 6.81178749e-01 2.19033614e-01 1.17842209e+00
-9.21216980e-02 8.76657844e-01 9.55886245e-01 -4.98087466e-01
1.72381625e-01 -1.45363435e-01 4.94496375e-02 5.43382883e-01
-3.93456191e-01 1.80498183e-01 -7.38436401e-01 -1.17194317e-01
3.28852594e-01 -5.67185879e-01 -2.24176466e-01 7.82018155e-03
-1.32545459e+00 4.48055685e-01 -6.58818036e-02 4.03814226e-01
1.00528337e-01 4.29029584e-01 5.19618392e-01 5.48424602e-01
3.14669400e-01 4.18887675e-01 -6.98803425e-01 -7.11340427e-01
-1.01163876e+00 -5.43464646e-02 7.29938447e-01 1.02369392e+00
6.15986943e-01 1.56822667e-01 -2.20784873e-01 9.37629819e-01
-1.67656347e-01 5.28752387e-01 5.47271669e-01 -9.05539751e-01
6.77040815e-01 -9.76997912e-02 2.35129688e-02 -2.49456644e-01
-3.19569468e-01 -4.70837653e-01 -4.54110473e-01 -1.43247932e-01
5.20920336e-01 -2.35902369e-02 -9.13201034e-01 2.04580712e+00
-9.71589983e-02 1.61958724e-01 -1.87115166e-02 7.10064709e-01
1.71818733e-01 7.13047564e-01 -1.69256702e-01 -3.40002328e-01
1.24267423e+00 -9.43210423e-01 -4.48580146e-01 -4.39876974e-01
8.80248249e-01 -9.47903931e-01 1.41584480e+00 4.60989416e-01
-1.08682299e+00 -2.97990978e-01 -1.13565218e+00 3.17472629e-02
-8.11021253e-02 6.77498654e-02 5.46533287e-01 8.24406266e-01
-1.28127801e+00 7.46513665e-01 -1.13366425e+00 -3.92002672e-01
-3.04570757e-02 6.58908904e-01 -1.86164990e-01 2.69746244e-01
-1.21569049e+00 9.29930627e-01 1.32933497e-01 -1.50003493e-01
-8.33411336e-01 -4.50146794e-01 -6.33499503e-01 1.87697597e-02
4.98477399e-01 -3.96890223e-01 1.63776243e+00 -1.09706938e+00
-1.78526485e+00 7.33011603e-01 -4.58164573e-01 -5.92895746e-01
2.96677262e-01 -2.19864830e-01 -4.61179197e-01 -1.48582369e-01
-1.38717756e-01 4.65841889e-01 4.80006725e-01 -5.57857275e-01
-6.45812154e-01 2.29854789e-02 1.70820042e-01 1.30725548e-01
-4.61863458e-01 2.09370658e-01 -6.48203313e-01 -7.69090176e-01
-1.14634722e-01 -1.12591207e+00 -9.31643173e-02 -3.44123363e-01
-2.86050111e-01 -5.11969626e-02 4.08089161e-01 -5.79287112e-01
1.40881360e+00 -2.21050310e+00 5.55595942e-02 2.24149793e-01
-3.18553686e-01 2.86471754e-01 -3.15863490e-01 6.31423235e-01
-9.24406499e-02 1.55461580e-01 -5.79273522e-01 -3.11543435e-01
2.93652229e-02 2.78942883e-01 -3.58943671e-01 4.20283347e-01
3.59695822e-01 7.91584730e-01 -9.41827178e-01 -2.37573385e-01
6.87388191e-03 2.30470836e-01 -5.02581894e-01 1.22362353e-01
-3.79119009e-01 3.04613501e-01 -2.71229863e-01 2.75732875e-01
1.10480487e-01 -1.69132218e-01 4.75737125e-01 1.33106008e-01
-1.61431789e-01 1.08273196e+00 -8.99533212e-01 1.72674394e+00
-8.87090147e-01 8.47174346e-01 -6.53455481e-02 -1.07518899e+00
7.98886240e-01 5.15457571e-01 3.58928651e-01 -7.79369414e-01
-1.24620214e-01 5.38043678e-01 3.22073877e-01 -2.65510231e-01
4.39939022e-01 -2.39939168e-01 -8.52769613e-02 6.21293843e-01
7.41656423e-02 -6.63464516e-02 1.20982610e-01 2.31356267e-02
1.41487348e+00 2.14889020e-01 2.08945677e-01 -1.80507764e-01
4.12379682e-01 -3.02245319e-01 3.94663632e-01 6.91750228e-01
-2.96846554e-02 6.08014226e-01 5.57150722e-01 -4.99359332e-02
-1.34460509e+00 -1.01419961e+00 -1.79584473e-02 1.25400186e+00
-1.31272092e-01 -4.45509255e-01 -6.24840617e-01 -5.51226199e-01
-2.85435617e-01 8.27694714e-01 -2.16738105e-01 -5.11477981e-03
-9.37806129e-01 -6.92551792e-01 1.02757061e+00 5.80462039e-01
1.27520159e-01 -9.53999937e-01 -4.00785387e-01 2.90798813e-01
-2.96568453e-01 -1.34974158e+00 -8.47693264e-01 5.71911097e-01
-7.94492900e-01 -7.28907645e-01 -3.40147823e-01 -7.38542438e-01
2.46581227e-01 -3.52411927e-03 1.03944910e+00 -4.49553132e-02
-4.19375040e-02 1.54731318e-01 -2.66009301e-01 -3.23892087e-02
-5.79327822e-01 3.93694460e-01 1.14908464e-01 2.55114939e-02
1.24691546e-01 -7.94245720e-01 -3.42555374e-01 5.91484666e-01
-7.39395916e-01 -1.53635268e-03 4.80948925e-01 9.97047961e-01
3.45380098e-01 -1.85599491e-01 5.34634531e-01 -8.07774782e-01
2.90794551e-01 -1.50237530e-01 -5.48887134e-01 2.60025859e-01
-5.74829340e-01 3.78854841e-01 6.97716594e-01 -4.38304722e-01
-7.24805951e-01 1.51596159e-01 -5.45386493e-01 -1.32298812e-01
-6.25178143e-02 4.57474768e-01 -1.39490321e-01 3.08406055e-01
5.58183014e-01 5.16114116e-01 -1.31477900e-02 -4.52574074e-01
3.57867479e-01 8.36931705e-01 5.45215607e-01 -7.29168415e-01
6.94268525e-01 3.23372811e-01 -1.69134274e-01 -7.49816537e-01
-6.88082218e-01 -5.42998254e-01 -3.70109379e-01 2.00380892e-01
5.05873084e-01 -8.57346714e-01 -8.11674178e-01 1.92631796e-01
-1.11803448e+00 -6.55600965e-01 -1.65222600e-01 4.56071913e-01
-9.32135701e-01 5.05908072e-01 -6.94243133e-01 -1.01142907e+00
-4.06972706e-01 -1.17126870e+00 1.08392155e+00 -3.86158258e-01
-4.73848611e-01 -7.84060776e-01 -2.42706552e-01 2.46114954e-01
3.39628965e-01 -1.80515721e-01 9.09836054e-01 -8.23444188e-01
-6.46973014e-01 7.24310661e-03 1.62955925e-01 5.67007542e-01
1.79727912e-01 -1.59413114e-01 -1.15533638e+00 -4.76789236e-01
-6.06935434e-02 -2.29236245e-01 8.58186662e-01 -1.27394395e-02
1.19463038e+00 -3.86341244e-01 -2.09128246e-01 3.41129720e-01
1.13506830e+00 4.97991651e-01 8.37085307e-01 3.02829266e-01
3.61300796e-01 5.01540720e-01 6.77771449e-01 1.94639713e-01
-6.67477474e-02 1.19994867e+00 2.32510418e-01 8.07343240e-05
-6.07994981e-02 -1.48602858e-01 7.78079093e-01 1.03098953e+00
1.09745815e-01 -4.78038937e-01 -9.09663677e-01 6.70859754e-01
-1.75667465e+00 -1.05233765e+00 1.83888122e-01 2.60858655e+00
1.16272974e+00 4.54235733e-01 3.08107466e-01 3.00862283e-01
6.14785671e-01 2.24547654e-01 -3.34847659e-01 -6.05585337e-01
-1.31777972e-01 6.99758947e-01 5.65171719e-01 8.27967286e-01
-9.75522459e-01 1.03541553e+00 6.14737892e+00 1.34011555e+00
-1.34670830e+00 2.37487152e-01 6.05943561e-01 -3.17372262e-01
-4.61077780e-01 1.65945262e-01 -7.65711784e-01 4.11625177e-01
1.40642631e+00 5.42704947e-02 8.08518410e-01 4.91844058e-01
3.64410907e-01 2.13392988e-01 -1.54034317e+00 8.12691391e-01
-2.47777849e-01 -1.09535229e+00 -2.71294534e-01 -2.49506743e-03
3.17407101e-01 1.35983884e-01 1.39087215e-01 2.05488861e-01
1.27514616e-01 -1.21186066e+00 1.07828403e+00 -6.80134743e-02
1.11611533e+00 -6.39156163e-01 5.68421543e-01 4.10841316e-01
-1.13699353e+00 4.34676483e-02 -7.37055689e-02 -1.26934618e-01
2.53614575e-01 3.48667353e-01 -9.88257706e-01 5.32365620e-01
3.28880489e-01 3.35723758e-01 -9.83079448e-02 8.14836562e-01
-1.79618537e-01 1.05181170e+00 -4.59618419e-01 -2.06915438e-01
3.85067463e-01 1.00014843e-01 5.76127827e-01 1.58418250e+00
3.56216103e-01 -9.86336097e-02 5.84450783e-03 3.98541510e-01
4.27142419e-02 1.81737289e-01 -3.80218923e-01 -1.61471888e-01
7.35388398e-01 7.54754424e-01 -4.57952619e-01 -1.26564026e-01
-4.92708683e-01 1.07572353e+00 4.02175933e-01 3.65750223e-01
-1.03312218e+00 -3.93302351e-01 7.85946786e-01 1.98906884e-01
5.20785749e-01 -4.79624927e-01 -3.00315082e-01 -9.73206937e-01
4.02846694e-01 -1.26460397e+00 2.09091917e-01 -4.15930897e-01
-9.77402151e-01 6.96078062e-01 -1.45319030e-01 -1.17030585e+00
-3.44434053e-01 -5.81005275e-01 -3.61820340e-01 1.03898215e+00
-1.41033542e+00 -7.62909114e-01 2.97441542e-01 2.69364595e-01
6.84160709e-01 -6.54715672e-02 9.14045811e-01 5.68083584e-01
-4.84138966e-01 9.93811488e-01 5.31300120e-02 9.50823538e-03
6.34637594e-01 -1.13785219e+00 9.40333545e-01 9.14026856e-01
4.67615992e-01 7.36072302e-01 8.41323078e-01 -1.49384931e-01
-1.20367670e+00 -1.05194044e+00 1.15950501e+00 -4.87662077e-01
8.66163552e-01 -6.71706200e-01 -6.33361220e-01 7.43709445e-01
5.67495525e-02 -2.94817656e-01 6.42309785e-01 2.95033932e-01
-6.42842114e-01 -1.82028025e-01 -5.61550617e-01 7.02628911e-01
1.32626450e+00 -8.60313356e-01 -2.20516309e-01 3.63000333e-01
1.00392652e+00 -5.01263440e-01 -7.41916955e-01 3.95346016e-01
5.89424491e-01 -7.98924267e-01 7.69297004e-01 -6.21873498e-01
1.73459083e-01 -3.30690742e-01 -5.52487195e-01 -1.03255737e+00
-6.70188069e-02 -1.10854590e+00 8.31726715e-02 1.21948850e+00
9.82593715e-01 -6.78764105e-01 7.24402905e-01 2.96425849e-01
-5.48809826e-01 -9.05528486e-01 -1.25695038e+00 -1.30761290e+00
1.31948873e-01 -9.09928977e-01 5.85470259e-01 6.87091351e-01
1.61275506e-01 4.46752369e-01 -5.12170553e-01 2.39312410e-01
3.42642069e-02 -1.05646528e-01 5.35948575e-01 -4.98512268e-01
-8.63870144e-01 -5.81573725e-01 -2.89739847e-01 -1.67843080e+00
6.34466186e-02 -9.79412079e-01 2.36397728e-01 -1.00602233e+00
3.84351052e-02 -7.57989228e-01 -3.88237119e-01 7.39293635e-01
-2.78651267e-02 1.38051257e-01 1.58133715e-01 1.83193356e-01
-4.69432354e-01 4.92807925e-01 7.29312003e-01 -1.01680286e-01
-1.37776852e-01 1.90027118e-01 -4.07931715e-01 2.56929308e-01
8.15497041e-01 -4.30629790e-01 -4.32226092e-01 -6.13860548e-01
8.08037296e-02 1.36272475e-01 1.72228858e-01 -8.45877349e-01
5.25914691e-02 -1.27522245e-01 -4.28914070e-01 -1.88253090e-01
5.31894624e-01 -5.01183867e-01 2.11482391e-01 3.19343567e-01
-5.05156696e-01 8.78309757e-02 3.44366670e-01 3.60554785e-01
-2.08745897e-01 -1.13099910e-01 7.25059152e-01 2.43744642e-01
-4.66364354e-01 3.76838669e-02 -4.88269389e-01 1.73524573e-01
7.18395412e-01 -2.65050888e-01 -2.94112861e-01 -5.86191297e-01
-6.26654685e-01 6.41604885e-02 3.68430525e-01 3.83736134e-01
2.86009431e-01 -1.14141738e+00 -6.60312772e-01 9.90774110e-03
1.85754493e-01 -3.98647547e-01 -2.06818908e-01 9.94182408e-01
-3.05940181e-01 8.33258450e-01 3.57697457e-01 -6.45789087e-01
-1.45163989e+00 4.37426001e-01 5.26571095e-01 -4.21266347e-01
-3.89066994e-01 8.42362463e-01 2.09547564e-01 -3.67881656e-01
3.95958573e-01 -2.72869706e-01 3.68377268e-01 -3.42881322e-01
3.01120281e-01 1.59770802e-01 5.11757135e-01 -5.06403446e-01
-6.14589751e-01 2.15665057e-01 3.21744457e-02 -5.34373939e-01
1.15963995e+00 -2.27063857e-02 1.44694179e-01 5.46361387e-01
1.32539856e+00 1.10471971e-01 -1.17411327e+00 -2.67494798e-01
2.66950846e-01 -1.76430404e-01 -3.26432645e-01 -7.68927217e-01
-8.09802175e-01 1.00262082e+00 3.95103663e-01 1.17716752e-01
1.18000448e+00 4.24747206e-02 8.51360977e-01 3.50034595e-01
5.07883787e-01 -8.55646074e-01 1.24447517e-01 7.55860031e-01
6.08234525e-01 -7.32541800e-01 -4.19646889e-01 -5.06998062e-01
-5.55711865e-01 9.18828607e-01 2.94540763e-01 3.26037347e-01
2.20788196e-01 5.64253569e-01 3.59914750e-02 1.90030873e-01
-1.11968160e+00 -3.47647756e-01 1.24509051e-01 4.39190239e-01
8.00458252e-01 1.01202808e-01 -4.80078161e-01 2.61045545e-01
-3.22988272e-01 -3.57559621e-01 3.33355188e-01 9.12138343e-01
-3.59887570e-01 -1.51436651e+00 3.64874932e-03 3.60215217e-01
-7.13624716e-01 -6.22920513e-01 -2.34352708e-01 5.84715188e-01
-2.02018335e-01 1.12715852e+00 1.58194974e-01 -5.96142709e-01
4.13307965e-01 2.69680470e-01 4.13489223e-01 -7.18900442e-01
-6.46184504e-01 1.57605618e-01 6.64454281e-01 -4.96670514e-01
-1.17871247e-01 -8.21221232e-01 -1.22065866e+00 -2.60266006e-01
-4.55964744e-01 2.40604341e-01 5.16785026e-01 1.02397859e+00
3.11300814e-01 4.21605200e-01 6.55505002e-01 -4.07818884e-01
-7.94286311e-01 -8.64317477e-01 -3.60031337e-01 2.30503574e-01
3.77514809e-01 -3.22923571e-01 -3.54756862e-01 5.05022779e-02] | [14.331936836242676, 6.868392467498779] |
7e2393a6-195c-44b1-9618-b015f9b064a9 | detect-reject-for-transferability-of-black | 2112.12095 | null | https://arxiv.org/abs/2112.12095v1 | https://arxiv.org/pdf/2112.12095v1.pdf | Detect & Reject for Transferability of Black-box Adversarial Attacks Against Network Intrusion Detection Systems | In the last decade, the use of Machine Learning techniques in anomaly-based intrusion detection systems has seen much success. However, recent studies have shown that Machine learning in general and deep learning specifically are vulnerable to adversarial attacks where the attacker attempts to fool models by supplying deceptive input. Research in computer vision, where this vulnerability was first discovered, has shown that adversarial images designed to fool a specific model can deceive other machine learning models. In this paper, we investigate the transferability of adversarial network traffic against multiple machine learning-based intrusion detection systems. Furthermore, we analyze the robustness of the ensemble intrusion detection system, which is notorious for its better accuracy compared to a single model, against the transferability of adversarial attacks. Finally, we examine Detect & Reject as a defensive mechanism to limit the effect of the transferability property of adversarial network traffic against machine learning-based intrusion detection systems. | ['Tayeb Kenaza', 'Wim Mees', 'Jean-Michel Dricot', 'Thibault Debatty', 'Islam Debicha'] | 2021-12-22 | null | null | null | null | ['network-intrusion-detection'] | ['miscellaneous'] | [ 4.08706665e-01 4.20698188e-02 3.14293057e-01 -1.89038128e-01
9.32139978e-02 -1.08355772e+00 8.36491466e-01 -3.46818343e-02
-3.47770989e-01 3.41592133e-01 -6.58042908e-01 -8.32676888e-01
3.57405618e-02 -1.03146350e+00 -7.36868322e-01 -6.45770311e-01
-2.94976920e-01 2.20399126e-01 2.50803947e-01 -3.61549586e-01
3.78752291e-01 1.27083194e+00 -1.32968915e+00 3.10522497e-01
4.22235191e-01 8.36555660e-01 -1.07298470e+00 1.07384956e+00
-3.83318886e-02 1.03962004e+00 -1.14312851e+00 -4.66066331e-01
6.25809014e-01 -4.66385633e-01 -6.36453509e-01 -3.53118002e-01
8.99494588e-01 -2.33304426e-01 -8.03053498e-01 1.22807074e+00
2.24694818e-01 -9.20755491e-02 5.08937657e-01 -2.00685573e+00
-5.86557746e-01 2.90007293e-01 -1.75370768e-01 6.71884954e-01
3.89747359e-02 5.31811237e-01 4.74900901e-01 -3.46326858e-01
1.08213708e-01 1.50059056e+00 3.04776400e-01 1.15058386e+00
-1.16690385e+00 -1.17965913e+00 5.84192015e-02 2.08968818e-02
-1.08577335e+00 -4.47915375e-01 7.32088387e-01 -2.40956038e-01
6.61547482e-01 5.41647911e-01 3.03870171e-01 1.36739182e+00
8.36930454e-01 4.05166209e-01 1.10074711e+00 -4.41644549e-01
2.92514861e-01 2.93299705e-01 2.44520232e-01 8.59193981e-01
4.54352856e-01 7.71381617e-01 8.45921412e-02 -6.04271173e-01
5.58715463e-01 -6.35948107e-02 -3.75399105e-02 3.19165923e-02
-3.46385390e-01 1.12840056e+00 6.07310772e-01 2.87000239e-01
-1.80292681e-01 1.95980504e-01 6.49477124e-01 8.43164623e-01
1.54827848e-01 9.36638355e-01 -5.19420803e-01 4.42440122e-01
-2.84533232e-01 -6.15820140e-02 1.15727055e+00 3.65342230e-01
2.91386098e-01 9.04488742e-01 3.33044052e-01 2.38168269e-01
1.68192521e-01 6.03089809e-01 3.98371845e-01 -6.07520819e-01
-1.88889489e-01 4.10910219e-01 -4.49181497e-01 -1.20691979e+00
-2.64791191e-01 -3.45737934e-01 -7.93336511e-01 9.29919481e-01
5.14279604e-01 -4.29863721e-01 -9.82644677e-01 1.56325579e+00
1.05767272e-01 3.93334538e-01 3.49736869e-01 3.88411820e-01
2.47315139e-01 3.72003645e-01 2.56523609e-01 1.60678476e-01
8.49319935e-01 -5.16752064e-01 -3.31352204e-01 -3.30443472e-01
4.44142699e-01 -4.27138150e-01 5.82744777e-01 4.26660120e-01
-6.82490885e-01 -2.56653607e-01 -1.28932655e+00 7.59939611e-01
-8.40147436e-01 -1.12158287e+00 6.07974410e-01 1.49553049e+00
-6.28061354e-01 4.86777216e-01 -6.39256001e-01 -2.89298326e-01
7.65251815e-01 6.61344945e-01 -3.94808948e-01 1.14836603e-01
-1.31762636e+00 1.18908322e+00 4.20245111e-01 -3.42286289e-01
-1.19448555e+00 -6.38549328e-01 -7.24119723e-01 -7.52370805e-02
1.40170947e-01 -4.32272494e-01 7.92438447e-01 -1.80090654e+00
-9.63693678e-01 8.52555156e-01 4.00292575e-01 -1.11301386e+00
5.43826401e-01 1.34855518e-02 -1.12942290e+00 1.70324892e-01
-4.91716176e-01 2.09157720e-01 1.38703561e+00 -1.17125428e+00
-5.69282234e-01 -4.80343223e-01 2.58883446e-01 -3.92480612e-01
-5.94242036e-01 2.21579537e-01 6.55861199e-01 -6.76195741e-01
-3.65269691e-01 -9.88046050e-01 -2.10724398e-01 -5.99754043e-02
-4.81766164e-01 1.78386644e-01 1.45949209e+00 2.03747265e-02
9.04991627e-01 -2.27333069e+00 -3.89791608e-01 7.11568534e-01
2.32964694e-01 8.77995014e-01 -2.74508417e-01 1.60865709e-01
-5.28330028e-01 6.40765011e-01 -2.16084018e-01 5.32368124e-01
-2.74776787e-01 5.02637744e-01 -7.80301690e-01 7.00936913e-01
4.40085113e-01 7.28825986e-01 -5.63624620e-01 5.14555685e-02
2.70342827e-01 2.72933602e-01 -3.86616230e-01 1.52997598e-01
5.86191528e-02 2.66013891e-01 -4.78904963e-01 7.48537421e-01
5.17674088e-01 3.16149056e-01 -6.49708882e-02 1.83675393e-01
4.07683641e-01 -3.40519160e-01 -6.42330408e-01 3.33543032e-01
-2.51165152e-01 8.17258060e-01 -8.11588094e-02 -1.01608276e+00
8.82191539e-01 2.27782875e-01 2.24964142e-01 -5.97744226e-01
4.66486096e-01 2.05752075e-01 8.18390369e-01 -2.68360674e-01
-5.58491983e-02 -1.72510087e-01 1.86448637e-02 6.44683123e-01
1.62884966e-02 1.80961370e-01 -3.36316913e-01 2.19183818e-01
1.61891496e+00 -7.33314931e-01 1.69938833e-01 -1.28105894e-01
8.96795571e-01 7.00597838e-02 3.02852422e-01 1.26149511e+00
-6.13810003e-01 -6.61579072e-02 2.84130692e-01 -8.05621862e-01
-9.90292311e-01 -1.31809092e+00 -1.19965583e-01 9.48264599e-01
-1.08670034e-01 2.15163946e-01 -8.85672867e-01 -1.36967659e+00
4.02218103e-01 9.46208239e-01 -7.18486190e-01 -1.06174660e+00
-7.01157928e-01 -6.89113855e-01 1.55304444e+00 4.56029356e-01
5.47468901e-01 -1.14878273e+00 -4.26015824e-01 7.76631087e-02
6.44588470e-01 -1.06026495e+00 -5.16191609e-02 1.56137645e-01
-6.97772563e-01 -1.52125609e+00 4.58486825e-02 -4.45035517e-01
7.45573163e-01 1.66653618e-02 1.03435731e+00 7.66140699e-01
-5.51047683e-01 8.82282197e-01 -3.42252374e-01 -1.16640174e+00
-1.22583354e+00 -2.10622102e-01 5.68928003e-01 1.88251421e-01
6.55189037e-01 -4.00837719e-01 -1.39225274e-01 3.75556529e-01
-1.34796786e+00 -9.84816849e-01 2.72909760e-01 6.45809829e-01
-2.21082792e-01 4.51142818e-01 9.07284737e-01 -1.06459165e+00
7.39882827e-01 -6.36625171e-01 -4.55612719e-01 4.04978991e-02
-7.42952466e-01 -2.46745497e-01 1.00272441e+00 -8.02157283e-01
-6.01970077e-01 -2.58740902e-01 -2.01937407e-01 -7.12851286e-01
-6.05679750e-01 -2.33911839e-03 -2.24854872e-01 -9.85523224e-01
1.15691090e+00 1.32054225e-01 4.50896651e-01 8.48466381e-02
7.29863867e-02 3.47472280e-01 3.86321664e-01 -4.57994938e-01
1.37513900e+00 5.02081513e-01 4.00461227e-01 -1.13582349e+00
-6.01465642e-01 1.81473359e-01 -4.42053825e-01 -4.21289742e-01
4.53241676e-01 -3.10738742e-01 -6.82855070e-01 8.23367178e-01
-9.68749404e-01 -4.42552157e-02 -9.06986594e-02 4.17998293e-03
-1.96170181e-01 2.32606933e-01 -5.01553178e-01 -8.33249450e-01
-3.05830330e-01 -8.21413398e-01 -1.63590267e-01 2.33623207e-01
-1.87689558e-01 -1.34418881e+00 -7.24540055e-02 1.00382559e-01
6.97814763e-01 7.25412667e-01 1.00277138e+00 -1.81411028e+00
-2.99895823e-01 -8.03531229e-01 1.34511739e-01 7.98699260e-01
1.08938545e-01 4.47998196e-01 -1.12875283e+00 -4.37117040e-01
2.65612513e-01 -1.33230552e-01 5.85139334e-01 -5.99104390e-02
1.04222345e+00 -7.22735643e-01 -1.59467235e-01 4.55402136e-01
1.24918807e+00 7.40313649e-01 7.76350558e-01 7.73760796e-01
5.81467748e-01 6.76679194e-01 1.83949202e-01 -1.21246092e-01
-5.00917375e-01 2.31529903e-02 9.56876695e-01 -2.05182478e-01
4.76438224e-01 2.57498324e-01 4.71123666e-01 -4.03453484e-02
2.28458524e-01 -3.90283734e-01 -1.00591362e+00 -1.08896330e-01
-1.29280305e+00 -1.26786649e+00 -6.50836378e-02 2.03310418e+00
1.71415374e-01 7.32198775e-01 2.40398109e-01 4.16516006e-01
6.56310380e-01 5.66224195e-02 -7.73369014e-01 -1.13072622e+00
-2.84984708e-01 5.85696876e-01 7.29428470e-01 4.01418090e-01
-1.37470913e+00 9.91160631e-01 6.99702120e+00 5.02564549e-01
-1.18719566e+00 -1.07410908e-01 6.35561407e-01 -6.48046732e-02
1.45789251e-01 -2.22237572e-01 -2.73976624e-01 4.73126590e-01
1.32114363e+00 -3.57599497e-01 3.02472144e-01 9.54625368e-01
-6.05434515e-02 3.28543246e-01 -1.16674209e+00 2.13774696e-01
2.32992187e-01 -1.18757081e+00 5.67139983e-01 2.23025143e-01
3.09836745e-01 -3.50091718e-02 4.13491964e-01 4.77994174e-01
4.40344065e-01 -1.37074471e+00 1.41514033e-01 3.48392725e-01
9.76248831e-02 -1.36322773e+00 7.17555761e-01 3.79947424e-01
-5.59416831e-01 -4.33703035e-01 -6.76604435e-02 -8.63882601e-02
-3.85532856e-01 2.40501259e-02 -7.00483859e-01 1.62924662e-01
4.82456446e-01 1.36345793e-02 -8.20531130e-01 6.25618815e-01
3.05251814e-02 8.69603574e-01 -1.26172721e-01 2.21943974e-01
5.06811917e-01 1.19490229e-01 9.55153942e-01 9.98846114e-01
-3.88985902e-01 6.51052669e-02 2.28676736e-01 6.02205992e-01
2.47313827e-02 -3.61783624e-01 -1.24915767e+00 -2.42271796e-01
4.47236151e-01 1.02768111e+00 -6.33036673e-01 -1.37709036e-01
-1.99747249e-01 8.91085207e-01 -9.08046588e-02 2.40910351e-01
-8.15867007e-01 -4.57921773e-01 1.12863147e+00 -1.41203254e-02
-2.67384887e-01 -6.14034198e-03 -2.14836240e-01 -8.26423585e-01
-4.52415138e-01 -1.42557359e+00 6.01840794e-01 -2.20803276e-01
-1.65096402e+00 7.87568450e-01 -4.11943138e-01 -9.01355028e-01
-8.68730322e-02 -8.33444417e-01 -1.12134039e+00 7.95707881e-01
-8.68668914e-01 -1.05842447e+00 2.24885881e-01 8.66277516e-01
3.26688319e-01 -9.39899206e-01 9.88734305e-01 7.10164458e-02
-8.02558184e-01 1.04598117e+00 -2.13799402e-01 8.12605858e-01
5.52611232e-01 -9.91650820e-01 6.87018037e-01 1.21432590e+00
1.52786046e-01 7.05205500e-01 7.60488331e-01 -5.91895461e-01
-1.20481873e+00 -1.30283070e+00 4.94969562e-02 -8.48009944e-01
9.08197939e-01 -6.25937060e-02 -1.35199368e+00 9.27835584e-01
4.37357798e-02 2.52258629e-01 8.40088725e-01 -3.09104472e-01
-8.15934718e-01 7.29933903e-02 -1.78851008e+00 6.40371859e-01
4.08957869e-01 -6.77925646e-01 -4.64931339e-01 3.90228303e-03
5.73013961e-01 2.00957293e-03 -6.38910055e-01 4.37316984e-01
6.44011736e-01 -9.27566588e-01 1.25974917e+00 -1.48053300e+00
1.35358036e-01 -1.65628582e-01 -1.24753930e-01 -1.15510774e+00
-2.46051714e-01 -4.01340157e-01 -4.01976317e-01 1.07088935e+00
2.50953883e-01 -1.29977179e+00 7.10463524e-01 6.01378977e-01
3.03849608e-01 -4.35328186e-01 -8.69953215e-01 -1.08841181e+00
5.58681369e-01 -2.54704267e-01 4.29501384e-01 1.36628401e+00
-5.76518714e-01 1.36687970e-02 -2.11751252e-01 6.35887802e-01
8.37986469e-01 -7.01694489e-01 8.75875771e-01 -1.43708229e+00
2.52887476e-02 -5.78331649e-01 -9.50854659e-01 1.87087014e-01
4.85862821e-01 -7.51346827e-01 -5.39902508e-01 -3.66890758e-01
-4.86623436e-01 -3.45313489e-01 -6.52746379e-01 4.92125779e-01
-3.64714973e-02 5.05806029e-01 5.11797726e-01 2.16964066e-01
6.91476613e-02 -1.29390791e-01 7.25976408e-01 -3.92215818e-01
1.10861152e-01 2.52639472e-01 -6.33658171e-01 1.16538680e+00
1.04184449e+00 -8.51362646e-01 -2.19189540e-01 5.49409352e-02
-1.18283823e-01 -4.70913053e-01 7.17624485e-01 -1.16974437e+00
1.49987623e-01 -2.26165891e-01 6.10746562e-01 1.11206584e-01
-2.22036336e-02 -1.27540469e+00 -1.33989707e-01 8.97875190e-01
-2.81512469e-01 7.02122450e-02 5.32187164e-01 6.71075642e-01
5.72251156e-02 -2.93363601e-01 1.28099835e+00 -2.04606280e-01
-6.78961277e-01 5.04470587e-01 -9.98410165e-01 6.51284456e-02
1.53302097e+00 -2.68792212e-01 -5.01769841e-01 -1.01942204e-01
-7.66290545e-01 4.32763956e-02 4.04265553e-01 7.03798354e-01
7.48848915e-01 -1.02205539e+00 -6.35280490e-01 5.03912389e-01
-8.85716826e-02 -8.06699395e-01 -2.06084754e-02 3.00288349e-01
-5.66833198e-01 -2.47083250e-02 -7.58981884e-01 -2.46159256e-01
-1.59376204e+00 9.75718856e-01 9.09343481e-01 9.86448675e-02
-4.20480281e-01 6.08540773e-01 4.31372449e-02 -2.55682409e-01
1.41324550e-01 7.10894942e-01 2.94915028e-02 -2.62841970e-01
7.39553034e-01 4.75836098e-01 -2.51421314e-02 -6.33651793e-01
-6.27407014e-01 -1.10896245e-01 -5.85285783e-01 1.88314930e-01
7.34831452e-01 4.33367550e-01 -1.07775718e-01 1.47048995e-01
1.07653999e+00 -3.03638369e-01 -5.10230958e-01 -5.33926897e-02
1.39729038e-01 -4.53944355e-01 -6.21345565e-02 -9.99200046e-01
-1.08174574e+00 7.26691782e-01 7.59801984e-01 9.44460213e-01
1.05217493e+00 -6.51143074e-01 5.04136682e-01 5.98380983e-01
1.82464644e-01 -6.32676661e-01 1.56683862e-01 7.72382379e-01
5.76612949e-01 -1.36173058e+00 -2.51480699e-01 -6.07375242e-02
-3.29721332e-01 1.45370257e+00 1.02596104e+00 -6.92283332e-01
8.36902857e-01 4.64040190e-01 3.77188742e-01 -2.57707745e-01
-5.18317461e-01 3.75057578e-01 1.83417380e-01 9.30286646e-01
-1.13409743e-01 -5.59892654e-02 3.77193153e-01 5.81881180e-02
-5.01169302e-02 -5.67897081e-01 8.77646744e-01 1.15433884e+00
-5.56199849e-01 -8.83145690e-01 -8.35526347e-01 6.07134581e-01
-8.34994376e-01 -8.98869336e-03 -1.08958125e+00 1.07297659e+00
-7.81910717e-02 1.13550603e+00 3.43129814e-01 -6.26746416e-01
4.51926529e-01 2.72712409e-01 2.78322667e-01 -4.17841882e-01
-1.24809861e+00 -7.23380864e-01 -1.21837474e-01 -5.79653978e-01
-1.61603004e-01 -2.57185906e-01 -1.02229249e+00 -9.25919890e-01
-2.99609840e-01 -2.15975679e-02 4.41442728e-01 1.14477682e+00
2.16925070e-01 7.48468995e-01 1.04998219e+00 -4.12410051e-01
-7.72710085e-01 -5.58882117e-01 -5.07069111e-01 3.27117562e-01
5.79812467e-01 -4.85135019e-01 -9.63898838e-01 -3.01291436e-01] | [5.506717205047607, 7.558452606201172] |
8f57f228-0ba6-4abe-9857-04c385d2e5b0 | coldnas-search-to-modulate-for-user-cold | 2306.03387 | null | https://arxiv.org/abs/2306.03387v1 | https://arxiv.org/pdf/2306.03387v1.pdf | ColdNAS: Search to Modulate for User Cold-Start Recommendation | Making personalized recommendation for cold-start users, who only have a few interaction histories, is a challenging problem in recommendation systems. Recent works leverage hypernetworks to directly map user interaction histories to user-specific parameters, which are then used to modulate predictor by feature-wise linear modulation function. These works obtain the state-of-the-art performance. However, the physical meaning of scaling and shifting in recommendation data is unclear. Instead of using a fixed modulation function and deciding modulation position by expertise, we propose a modulation framework called ColdNAS for user cold-start problem, where we look for proper modulation structure, including function and position, via neural architecture search. We design a search space which covers broad models and theoretically prove that this search space can be transformed to a much smaller space, enabling an efficient and robust one-shot search algorithm. Extensive experimental results on benchmark datasets show that ColdNAS consistently performs the best. We observe that different modulation functions lead to the best performance on different datasets, which validates the necessity of designing a searching-based method. | ['Quanming Yao', 'Dejing Dou', 'daxiang dong', 'Qinghe Jing', 'Yaqing Wang', 'Shiguang Wu'] | 2023-06-06 | null | null | null | null | ['architecture-search'] | ['methodology'] | [ 1.21640980e-01 -3.86886358e-01 -8.38430524e-01 -4.19600517e-01
-3.99349183e-01 -7.16921866e-01 5.17282665e-01 -5.60006440e-01
-1.69246182e-01 4.20779020e-01 2.37205416e-01 -2.86584377e-01
-6.58829689e-01 -6.59109056e-01 -4.13806438e-01 -8.79542530e-01
-9.16347280e-03 3.27809602e-01 1.03440933e-01 -6.62856102e-01
3.50830227e-01 -1.66011885e-01 -1.63126612e+00 2.41296053e-01
9.96441782e-01 8.19207191e-01 2.46884912e-01 3.81848127e-01
7.34669939e-02 3.77450511e-02 -4.50023890e-01 -1.10747308e-01
5.43624759e-01 -6.59307718e-01 -5.74634135e-01 -4.71676826e-01
1.49106726e-01 -4.98630479e-02 -2.82384813e-01 7.30113208e-01
6.15902603e-01 5.95988631e-01 5.47054768e-01 -8.13729465e-01
-7.88094878e-01 9.72477257e-01 -7.30107054e-02 2.77599633e-01
3.83084118e-01 1.04169920e-01 1.34015679e+00 -4.48307753e-01
4.18065637e-01 8.44825506e-01 4.50943232e-01 6.99242830e-01
-1.16256368e+00 -7.32511222e-01 5.65590918e-01 4.40914571e-01
-1.49190032e+00 -3.01964313e-01 6.23184919e-01 -1.79319039e-01
7.78187394e-01 6.72258675e-01 6.63542747e-01 1.27405071e+00
-7.68642575e-02 4.59045827e-01 7.41445601e-01 -4.64305162e-01
3.10889095e-01 1.16104655e-01 4.67360914e-01 3.94222796e-01
1.84561625e-01 2.30399966e-01 -5.83179891e-01 -2.98769951e-01
4.63353664e-01 3.46537083e-01 -7.85302877e-01 -2.23205075e-01
-9.53065872e-01 8.59091997e-01 4.20610130e-01 4.39948827e-01
-8.37757438e-02 -6.37356788e-02 -1.16976589e-01 7.33738422e-01
1.73683092e-01 1.03097475e+00 -6.67287529e-01 -7.15752542e-02
-6.96732283e-01 1.98987946e-01 1.04473531e+00 8.57214451e-01
5.37407577e-01 -2.87159383e-01 -7.48554885e-01 9.11728024e-01
1.83466688e-01 3.58859420e-01 6.81284726e-01 -7.26727724e-01
2.64075190e-01 3.41125995e-01 2.78090924e-01 -8.05409431e-01
-5.76517761e-01 -8.99016678e-01 -6.97666466e-01 -4.16682541e-01
3.23602110e-01 -3.02240342e-01 -6.59728706e-01 1.88449550e+00
3.11786622e-01 5.11058390e-01 -3.17405581e-01 1.26345146e+00
4.72708523e-01 5.83037853e-01 -2.45005772e-01 -6.07067883e-01
1.26593530e+00 -1.20212650e+00 -5.41341484e-01 1.32290140e-01
6.47001445e-01 -3.63576591e-01 1.42310345e+00 6.78572476e-01
-8.35900843e-01 -5.28771758e-01 -1.30158472e+00 3.78792316e-01
-2.26799801e-01 -1.37818038e-01 7.22904325e-01 9.61843908e-01
-7.99825311e-01 9.92445409e-01 -4.15088177e-01 -3.54894340e-01
-3.70868742e-02 5.93908548e-01 3.02177191e-01 1.61982313e-01
-1.58298647e+00 6.55472040e-01 1.50304779e-01 1.11823760e-01
-3.55621189e-01 -8.02963555e-01 -1.32607803e-01 3.56909990e-01
7.19227612e-01 -9.19470191e-01 1.22184491e+00 -7.43806005e-01
-2.05463529e+00 1.18155954e-02 -1.01950923e-02 -2.33208403e-01
2.19450658e-03 -2.58666873e-01 -7.97724903e-01 -3.27651054e-01
-5.76212287e-01 -1.35095656e-01 1.08375621e+00 -1.03185046e+00
-6.20880365e-01 -1.32275924e-01 2.84472287e-01 3.03989649e-01
-9.88004446e-01 -1.50000706e-01 -8.47730219e-01 -4.73115444e-01
-3.67165022e-02 -1.09238410e+00 -5.14156401e-01 -6.98033512e-01
-2.97163665e-01 -3.41637015e-01 3.68769288e-01 -8.60204771e-02
2.18932462e+00 -1.89014602e+00 2.34742463e-01 7.52958119e-01
2.66854495e-01 3.92727584e-01 -3.55646044e-01 3.91428977e-01
2.56933719e-01 2.26525247e-01 3.92799318e-01 -2.00231955e-03
2.26939902e-01 1.32307038e-01 -3.62712145e-01 4.58532393e-01
-5.05419314e-01 8.77193272e-01 -8.82318258e-01 -3.51168332e-03
-1.34652462e-02 3.38406831e-01 -9.95379746e-01 2.99705207e-01
-4.16541129e-01 3.98526520e-01 -7.66337812e-01 4.80372429e-01
3.91770899e-01 -6.97030902e-01 5.51320016e-01 -2.00820774e-01
-7.86476582e-02 3.85923147e-01 -1.04913759e+00 1.58435977e+00
-6.13904595e-01 1.65376306e-01 -1.77834556e-01 -8.70569527e-01
9.18795586e-01 1.05993696e-01 3.75518978e-01 -9.72399652e-01
1.97345525e-01 -5.52347489e-02 1.86471730e-01 -5.28570175e-01
5.30068219e-01 2.21044332e-01 -5.40086702e-02 5.50539076e-01
-1.21984512e-01 5.40936172e-01 7.87555948e-02 -1.91312447e-01
1.19118178e+00 -8.32896456e-02 5.66596277e-02 -2.68545240e-01
3.92002285e-01 -5.53395927e-01 3.95501226e-01 1.29582047e+00
1.52788252e-01 5.41359365e-01 1.40371293e-01 -2.89269537e-01
-6.58294797e-01 -5.91872036e-01 -2.14694709e-01 1.72604251e+00
4.75393325e-01 -8.83092284e-01 -5.89866042e-01 -6.10118687e-01
-1.36894241e-01 7.89823472e-01 -6.33185983e-01 -4.50524062e-01
-5.42056203e-01 -9.31411743e-01 8.31100717e-02 -6.92267716e-03
-6.96135163e-02 -6.71020806e-01 -3.60847533e-01 1.55191034e-01
-3.42120640e-02 -4.90900427e-01 -8.56980503e-01 2.83911765e-01
-7.14445174e-01 -7.99087107e-01 -7.11312413e-01 -5.40027618e-01
4.07288522e-01 5.58099926e-01 9.74153578e-01 4.25230175e-01
2.55749077e-01 1.65460557e-01 -7.41485894e-01 2.20274210e-01
9.96399447e-02 7.47618914e-01 2.52085000e-01 3.90661135e-02
2.54654765e-01 -9.58297729e-01 -1.14733839e+00 7.38966048e-01
-4.99972254e-01 -2.99389005e-01 5.20872831e-01 9.14454818e-01
3.66256773e-01 -1.87461570e-01 5.93772471e-01 -1.13465250e+00
9.26799476e-01 -7.62080967e-01 -5.55323243e-01 5.15121877e-01
-1.28629541e+00 2.68477172e-01 1.09490681e+00 -6.55421913e-01
-6.99089289e-01 -2.60570526e-01 -6.99125379e-02 -3.76102507e-01
1.09655775e-01 4.85215127e-01 -7.65248165e-02 -2.32243329e-01
1.03323853e+00 4.70051691e-02 -4.00780439e-01 -7.64671803e-01
6.27996862e-01 7.51324177e-01 1.84193507e-01 -3.67103010e-01
6.84863627e-01 9.38877240e-02 -3.25783759e-01 -5.50284266e-01
-1.05466592e+00 -6.74098432e-01 -3.01636457e-01 -4.28364910e-02
4.28045243e-01 -5.21761715e-01 -9.74180281e-01 -2.36415789e-01
-6.83471620e-01 -3.29150766e-01 6.48617595e-02 2.75872469e-01
-3.34779620e-01 1.84612453e-01 -3.27790380e-01 -6.83302283e-01
-6.92751586e-01 -8.72704983e-01 5.56821942e-01 4.21416730e-01
-2.41163522e-01 -5.85369527e-01 3.28452229e-01 1.41507566e-01
6.48549616e-01 -3.75602961e-01 8.92320871e-01 -8.35936368e-01
-6.42202616e-01 -1.10589713e-01 2.85444736e-01 -2.31221095e-01
-1.72475483e-02 -2.17053175e-01 -7.77652919e-01 -3.81816626e-01
-5.80037460e-02 2.55027652e-01 8.09634387e-01 3.28837812e-01
1.51802123e+00 -4.17282850e-01 -6.79115295e-01 1.16736102e+00
1.25553501e+00 3.89095336e-01 5.49015462e-01 3.00666124e-01
6.48547888e-01 1.48719950e-02 5.88656843e-01 6.03148282e-01
2.35401049e-01 1.21806872e+00 1.56013548e-01 3.70745212e-01
2.44878069e-01 -2.46743947e-01 2.22220123e-01 9.55102801e-01
-2.52158582e-01 -6.40756190e-01 -3.84877682e-01 2.63448134e-02
-2.17351985e+00 -1.00743032e+00 1.99266419e-01 2.55596232e+00
9.76600230e-01 8.61440822e-02 2.75607169e-01 -1.75757885e-01
4.89360482e-01 4.91549931e-02 -6.40388489e-01 -2.78203458e-01
1.87433392e-01 3.77922475e-01 4.69781220e-01 2.85433322e-01
-1.03442454e+00 7.72365391e-01 6.70719671e+00 9.91907537e-01
-1.26494145e+00 1.40247136e-01 4.21081483e-01 -5.28707683e-01
-4.71240014e-01 -3.69520374e-02 -9.61504877e-01 8.95513356e-01
1.28643990e+00 -2.52756536e-01 1.01985896e+00 7.24466503e-01
3.55404079e-01 3.91632795e-01 -1.03860271e+00 9.59987402e-01
-6.47604316e-02 -1.46045268e+00 -1.53870896e-01 -7.05537423e-02
1.00308251e+00 -5.84330522e-02 1.03590116e-01 6.42568529e-01
9.85197797e-02 -9.34543490e-01 1.88225359e-01 8.29403758e-01
5.89353144e-01 -7.37536252e-01 2.99456805e-01 3.52321655e-01
-1.04733956e+00 -4.33254719e-01 -4.19650853e-01 7.04902783e-02
1.15061678e-01 4.06517863e-01 -4.25540537e-01 4.39994752e-01
7.02723265e-01 5.93650460e-01 -3.28535050e-01 1.40880716e+00
-3.80731523e-02 9.71928120e-01 -3.25590193e-01 -6.60450578e-01
-6.08187094e-02 -3.62915963e-01 4.08171535e-01 1.04896390e+00
4.71639782e-01 3.78023058e-01 2.60457844e-01 6.05648696e-01
-7.62086660e-02 2.98366934e-01 -2.04089448e-01 1.53729217e-02
6.77916348e-01 1.21033883e+00 -4.56477404e-01 -8.57510343e-02
-2.44763985e-01 9.25232947e-01 2.91851252e-01 5.50342798e-01
-8.19515407e-01 -5.12599647e-01 7.59236574e-01 3.09686773e-02
6.58020139e-01 2.50418615e-02 -3.24153304e-01 -1.12353742e+00
-2.00796664e-01 -1.01259160e+00 4.28987741e-01 -2.23435417e-01
-1.31650603e+00 6.12847745e-01 -7.97169209e-02 -1.56301820e+00
-3.04178655e-01 -1.82489917e-01 -6.54046297e-01 7.21490443e-01
-1.10117817e+00 -5.77516258e-01 -3.04002017e-01 4.93994266e-01
4.02635515e-01 -2.42734388e-01 9.33173358e-01 4.45063144e-01
-6.95208907e-01 1.28998530e+00 5.47598779e-01 -6.67911291e-01
7.68625796e-01 -9.40683007e-01 2.52699107e-01 3.91463935e-01
2.29127049e-01 1.10934305e+00 7.99827814e-01 -3.35056663e-01
-1.86762726e+00 -7.88049698e-01 5.82278073e-01 -4.85462576e-01
4.70949143e-01 -4.32153732e-01 -8.17588568e-01 4.86990400e-02
6.69123307e-02 -2.38027513e-01 9.66150939e-01 7.35195816e-01
-2.48845562e-01 -2.13220477e-01 -7.90913403e-01 8.98076355e-01
1.49606752e+00 -2.19712600e-01 -1.82454333e-01 3.94026995e-01
1.10860193e+00 -2.97614992e-01 -7.77638972e-01 3.17045599e-01
9.04400289e-01 -7.92457998e-01 9.72983956e-01 -7.80681849e-01
2.41364595e-02 -3.03176522e-01 5.21160252e-02 -1.51959085e+00
-8.53778720e-01 -1.18244910e+00 -7.19368935e-01 7.61955678e-01
6.81689918e-01 -5.86662054e-01 6.67775393e-01 5.90829670e-01
-1.69867322e-01 -1.19921958e+00 -5.19848406e-01 -7.91954398e-01
-1.96095005e-01 -1.49890542e-01 8.60658884e-01 9.98798072e-01
3.50873142e-01 6.10883057e-01 -9.14905310e-01 1.66548863e-02
2.47657850e-01 5.97850025e-01 6.22036040e-01 -1.22973430e+00
-8.49190950e-01 -5.77808857e-01 1.89228877e-01 -1.62412524e+00
-9.83847529e-02 -8.80940795e-01 1.01183169e-01 -1.15358722e+00
6.28386065e-02 -6.90934956e-01 -7.52522707e-01 3.26667130e-01
-3.22606593e-01 8.83924440e-02 6.29123673e-02 3.25168937e-01
-9.13881123e-01 5.82634330e-01 1.30489397e+00 1.61902949e-01
-7.25134432e-01 4.37510103e-01 -1.21322191e+00 2.15943828e-01
7.47215211e-01 -3.04404616e-01 -7.04876602e-01 -1.82967559e-01
4.60528642e-01 -1.12425834e-01 -3.74860495e-01 -8.13665211e-01
3.49477023e-01 -5.37724376e-01 9.08802152e-02 -2.80501753e-01
2.13923588e-01 -7.30741858e-01 1.43076256e-01 2.22490400e-01
-6.56431019e-01 -3.21598738e-01 -3.46053183e-01 7.74791777e-01
4.44496095e-01 -2.75289595e-01 4.21781659e-01 2.66480863e-01
-6.58628583e-01 5.56870997e-01 -1.50328977e-02 -1.40924290e-01
6.65126860e-01 1.71721336e-02 -3.55274558e-01 -5.13150632e-01
-6.23723388e-01 3.79590541e-01 1.32039994e-01 6.52881742e-01
2.28314117e-01 -1.35442197e+00 -1.79966003e-01 2.98685551e-01
1.45442337e-01 -6.53561413e-01 3.07798237e-01 7.47096717e-01
2.50225127e-01 5.10005534e-01 3.19932818e-01 -3.50182444e-01
-1.00489616e+00 7.40666747e-01 3.43013912e-01 -8.50036219e-02
-5.29160261e-01 9.14402783e-01 -1.80219397e-01 -2.20692679e-01
5.52129865e-01 -2.92254567e-01 -4.76980805e-01 -7.22322613e-02
7.04705775e-01 2.55315393e-01 1.10885225e-01 -5.96578699e-03
-2.00957328e-01 5.06172895e-01 -2.13075265e-01 3.14520746e-01
1.14041066e+00 -2.18758956e-01 3.22783083e-01 1.88330889e-01
1.11435926e+00 -1.34103090e-01 -9.43076611e-01 -4.12172347e-01
-7.29325041e-02 -7.10768342e-01 1.68641344e-01 -8.57655346e-01
-1.08715355e+00 2.28512108e-01 7.63782024e-01 5.37416637e-01
1.12310851e+00 -6.34499192e-02 9.04842079e-01 8.14759016e-01
5.28151393e-01 -1.29981172e+00 -5.71776479e-02 6.62046134e-01
5.49464345e-01 -9.65231299e-01 -8.36895853e-02 -1.76878855e-01
-3.91796112e-01 9.05171990e-01 6.85520768e-01 -1.09091196e-02
1.06211329e+00 -1.41473308e-01 -1.78899750e-01 -7.16108233e-02
-1.10792744e+00 -3.19886535e-01 6.62544012e-01 3.30946743e-01
4.42697108e-01 2.07797214e-01 -7.44941294e-01 1.35291588e+00
-3.88953030e-01 4.28670198e-02 1.79029927e-02 5.49601197e-01
-5.59053659e-01 -1.45320404e+00 -6.03836067e-02 9.20557261e-01
-9.16028693e-02 -2.92176127e-01 -9.16042253e-02 1.57822505e-01
5.29907495e-02 1.10518348e+00 -1.90028697e-01 -1.06151116e+00
4.34464157e-01 -2.32581384e-02 3.36229265e-01 -6.20161235e-01
-9.01025653e-01 6.08502403e-02 1.78259447e-01 -6.57129884e-01
-8.09008852e-02 -3.23383629e-01 -8.81833673e-01 -3.61074179e-01
-6.48738801e-01 3.23722988e-01 3.56821120e-01 1.08127046e+00
6.14223003e-01 4.47396666e-01 1.08897114e+00 -8.07909608e-01
-9.50276732e-01 -7.76142716e-01 -4.31035697e-01 2.29820177e-01
1.87661156e-01 -8.29266727e-01 -2.89246708e-01 -5.44193089e-01] | [10.078405380249023, 5.620729923248291] |
62ded650-1cfe-4e2b-a11e-c3bd39d27f4c | class-balanced-pixelnet-for-neurological | 2204.11048 | null | https://arxiv.org/abs/2204.11048v1 | https://arxiv.org/pdf/2204.11048v1.pdf | Class Balanced PixelNet for Neurological Image Segmentation | In this paper, we propose an automatic brain tumor segmentation approach (e.g., PixelNet) using a pixel-level convolutional neural network (CNN). The model extracts feature from multiple convolutional layers and concatenate them to form a hyper-column where samples a modest number of pixels for optimization. Hyper-column ensures both local and global contextual information for pixel-wise predictors. The model confirms the statistical efficiency by sampling a few pixels in the training phase where spatial redundancy limits the information learning among the neighboring pixels in conventional pixel-level semantic segmentation approaches. Besides, label skewness in training data leads the convolutional model often converge to certain classes which is a common problem in the medical dataset. We deal with this problem by selecting an equal number of pixels for all the classes in sampling time. The proposed model has achieved promising results in brain tumor and ischemic stroke lesion segmentation datasets. | ['Hongliang Ren', 'Mobarakol Islam'] | 2022-04-23 | null | null | null | null | ['ischemic-stroke-lesion-segmentation', 'brain-tumor-segmentation'] | ['medical', 'medical'] | [ 5.05458713e-01 2.97312945e-01 -4.37800378e-01 -4.74283248e-01
-5.06216526e-01 1.11776084e-01 2.19067007e-01 1.90772414e-01
-8.10346246e-01 9.04160678e-01 -8.54491293e-02 -1.59877867e-01
3.02421544e-02 -8.57886195e-01 -6.13984346e-01 -1.01962399e+00
1.32865325e-01 2.55920649e-01 3.84786874e-01 4.11372602e-01
3.20913881e-01 6.58802807e-01 -1.30460560e+00 5.89147389e-01
1.02895272e+00 1.24157846e+00 5.19345641e-01 4.00605202e-01
-6.71305537e-01 7.00364053e-01 -4.92812514e-01 3.69160533e-01
3.49116325e-01 -2.76654154e-01 -8.81953716e-01 3.03247124e-01
5.55546582e-01 -1.16172962e-01 -1.24770179e-02 1.36527193e+00
3.27989012e-01 -1.43001109e-01 5.89713573e-01 -9.66975689e-01
1.30600810e-01 7.34007001e-01 -7.05435157e-01 1.85361072e-01
-6.28254890e-01 8.75857919e-02 6.83873773e-01 -7.02211559e-01
6.85177982e-01 9.24135864e-01 7.83000946e-01 2.34484240e-01
-1.01163018e+00 -5.28517246e-01 1.76815733e-01 1.01955995e-01
-1.36693180e+00 7.71028921e-02 7.63941884e-01 -5.24475455e-01
6.40129328e-01 2.35015482e-01 1.06541359e+00 5.16695380e-01
4.21258003e-01 1.00226748e+00 1.19927251e+00 -3.20670754e-01
3.56693476e-01 6.51384071e-02 4.14613545e-01 7.01596081e-01
3.02370816e-01 -1.16362482e-01 -2.04443395e-01 2.32509762e-01
9.97333825e-01 2.54425853e-01 -5.73004708e-02 -2.67644584e-01
-1.20053554e+00 7.75591910e-01 8.49620402e-01 5.98250926e-01
-6.15960360e-01 2.69768864e-01 3.84948641e-01 -1.07874900e-01
2.22591028e-01 2.14803904e-01 -6.11560106e-01 4.43371862e-01
-1.45459151e+00 -3.17344032e-02 3.40943277e-01 7.13560820e-01
8.58605504e-01 9.09470618e-02 -4.37025785e-01 9.08479154e-01
-4.30963226e-02 -3.29542533e-02 7.51269698e-01 -6.16900325e-01
3.37301224e-01 8.32928002e-01 -3.41462642e-01 -6.96735799e-01
-8.15608323e-01 -9.76333857e-01 -1.06943548e+00 5.68930507e-01
6.99505031e-01 -2.83780813e-01 -1.36470044e+00 1.29496253e+00
2.06049517e-01 2.75990129e-01 -7.70396739e-02 7.72565603e-01
9.09044743e-01 4.12434906e-01 3.61067235e-01 -2.22032040e-01
1.33506227e+00 -1.14035702e+00 -5.85979044e-01 -3.48958731e-01
6.63359404e-01 -5.17202973e-01 8.46533477e-01 3.33015025e-01
-9.52782154e-01 -5.10123312e-01 -1.04352951e+00 1.68859452e-01
-5.11848569e-01 5.35356224e-01 5.88292420e-01 4.04340148e-01
-1.04408026e+00 6.07659221e-01 -8.04577351e-01 -2.77767062e-01
9.62724268e-01 6.33228600e-01 -1.53005183e-01 1.17331438e-01
-8.00060034e-01 6.03164852e-01 6.61232233e-01 9.29940119e-02
-7.72182465e-01 -7.82930374e-01 -4.74858552e-01 -3.01719247e-03
1.12728208e-01 -3.10070634e-01 8.27982128e-01 -1.55493927e+00
-1.23870993e+00 8.65020335e-01 -1.63078785e-01 -5.64988196e-01
7.02830315e-01 1.16832614e-01 -3.97964939e-02 1.60737455e-01
1.36893913e-02 1.20006692e+00 7.44985998e-01 -1.15760410e+00
-9.18158054e-01 -3.17912221e-01 -3.72663945e-01 2.32997328e-01
-2.40419373e-01 -2.09953874e-01 -5.48668146e-01 -7.78650761e-01
7.20716119e-01 -7.28881299e-01 -7.33838618e-01 5.05674221e-02
-7.44359434e-01 1.66832343e-01 8.85007679e-01 -4.42771822e-01
9.65755939e-01 -1.91274869e+00 -1.89155757e-01 5.33403456e-01
2.16224149e-01 -3.59514612e-03 1.26216248e-01 -2.87413269e-01
-3.88205737e-01 8.29557702e-02 -3.97859037e-01 -2.94868611e-02
-5.97505987e-01 1.56634763e-01 4.09287840e-01 4.87851948e-01
2.54534245e-01 8.79509866e-01 -5.63407838e-01 -1.00157714e+00
3.38492841e-01 2.74717599e-01 -5.45186818e-01 -2.66490579e-01
-1.49780720e-01 6.87618315e-01 -7.18091965e-01 8.73885691e-01
8.50734711e-01 -1.97898299e-01 -1.62672877e-01 -3.65656316e-01
-2.79999524e-01 -2.21643480e-03 -1.15448856e+00 1.49489844e+00
-1.52124912e-01 7.19813645e-01 1.06129751e-01 -1.30889916e+00
7.30781257e-01 2.36789033e-01 9.98301685e-01 -4.74164963e-01
5.69361091e-01 2.52183586e-01 1.92274645e-01 -5.24898052e-01
3.19196582e-01 1.22741044e-01 2.38150835e-01 3.12440563e-02
-5.72932884e-02 -8.52883514e-03 8.98845047e-02 -2.52163917e-01
8.44819486e-01 -1.71670005e-01 2.63613284e-01 -6.55195355e-01
5.72993040e-01 4.06484276e-01 7.83682942e-01 9.21348095e-01
-4.48293388e-01 7.85786629e-01 6.68261766e-01 -6.16536856e-01
-1.05433643e+00 -5.87337196e-01 -5.61896861e-01 4.74869817e-01
-3.75545509e-02 1.51866734e-01 -1.14522672e+00 -6.85612202e-01
-2.54896492e-01 3.16495419e-01 -8.00104499e-01 9.86903906e-02
-7.13261604e-01 -1.06041169e+00 2.65304178e-01 6.26545787e-01
8.85430694e-01 -1.20013106e+00 -7.82278180e-01 3.76457244e-01
1.32933781e-01 -1.00292110e+00 -1.31523028e-01 6.89492464e-01
-1.20668542e+00 -1.02457583e+00 -8.29559147e-01 -1.19029629e+00
1.25646281e+00 -1.88610762e-01 8.30469429e-01 7.36134946e-02
-7.11966455e-01 -4.14187253e-01 -2.01523602e-01 -4.16724652e-01
-4.24760468e-02 2.90994436e-01 -5.64910948e-01 5.65780103e-02
4.04488683e-01 -3.18161398e-01 -9.22692537e-01 6.70573339e-02
-7.68264353e-01 2.46747464e-01 8.34536195e-01 1.08397305e+00
1.04467380e+00 2.40649775e-01 3.66302907e-01 -1.07445872e+00
3.79023403e-01 -2.41959214e-01 -5.79368949e-01 1.55150592e-01
-4.15747136e-01 -1.52383670e-01 5.38313508e-01 -4.41862732e-01
-5.86282611e-01 5.48651814e-01 -2.10860209e-03 -2.41734654e-01
-4.38153028e-01 5.30096889e-01 -1.35091627e-02 -1.53881788e-01
4.85982269e-01 2.35631689e-01 6.40859082e-02 -1.23985030e-01
1.12398706e-01 4.78199244e-01 4.54010874e-01 -2.62794912e-01
2.73395568e-01 5.47967672e-01 8.24877098e-02 -7.59102702e-01
-4.14178789e-01 -3.75896305e-01 -8.73824358e-01 -3.62952888e-01
1.25307178e+00 -7.55125344e-01 -3.11033547e-01 5.82044244e-01
-8.74388933e-01 -3.45931053e-01 -5.51491022e-01 6.22734487e-01
-3.91667932e-01 -2.71121096e-02 -4.75741416e-01 -5.35446227e-01
-2.77703762e-01 -1.73576677e+00 7.55486071e-01 5.38166821e-01
-8.23841244e-02 -8.24900627e-01 -4.86166954e-01 7.33589157e-02
2.75139600e-01 3.47534120e-01 9.67033267e-01 -6.62443578e-01
-7.46143043e-01 -2.85713196e-01 -4.11596775e-01 2.70344377e-01
6.12000003e-02 8.95129051e-03 -9.82465148e-01 -3.64289805e-02
-2.44428515e-01 -1.01482220e-01 1.21984017e+00 1.23312318e+00
1.77850235e+00 3.68861016e-03 -6.22787058e-01 7.48929322e-01
1.67095995e+00 4.25282747e-01 5.60861111e-01 5.50839245e-01
6.60505712e-01 5.06901264e-01 3.40884268e-01 3.31175894e-01
-1.53697833e-01 1.91578791e-01 5.36283135e-01 -7.53747284e-01
-3.36791098e-01 2.80823350e-01 -3.78476202e-01 3.44452858e-01
3.58528644e-01 6.19233027e-02 -9.31839764e-01 7.03810811e-01
-1.68491292e+00 -5.77805817e-01 -2.74876922e-01 2.03374195e+00
9.51064467e-01 1.62816897e-01 -2.04973429e-01 1.94192708e-01
7.25137651e-01 7.67632574e-02 -5.64801455e-01 -1.46069795e-01
-2.89755970e-01 4.04636532e-01 1.15961242e+00 4.00728732e-01
-1.34844387e+00 1.10207486e+00 6.28097534e+00 1.05555618e+00
-1.45592487e+00 8.71838406e-02 1.24688411e+00 -1.71592087e-02
3.89300776e-03 -1.55815646e-01 -7.74176300e-01 4.71116483e-01
3.23580652e-01 5.40138543e-01 -5.67751937e-02 7.70577967e-01
4.56736714e-01 -5.71448028e-01 -7.39212990e-01 8.53276670e-01
-2.46413887e-01 -1.49830759e+00 -3.03832293e-02 -1.98449176e-02
1.14722192e+00 1.69981450e-01 9.14688036e-02 -2.08243296e-01
1.62599891e-01 -1.33772874e+00 6.20268643e-01 5.10634661e-01
6.15485132e-01 -7.72140443e-01 8.40429246e-01 2.30010509e-01
-8.96970868e-01 -2.74995714e-01 -4.56968844e-01 4.42420840e-01
-1.43866345e-01 8.94019186e-01 -1.09158087e+00 1.01413622e-01
6.99390650e-01 5.02890706e-01 -4.34404135e-01 1.39523041e+00
9.28251818e-02 6.27738118e-01 -4.72298175e-01 3.93088609e-02
7.03595161e-01 -3.45560312e-01 2.99622834e-01 1.19296587e+00
2.17453718e-01 -9.55731794e-02 3.28091234e-01 8.50479722e-01
2.05626488e-02 3.90240729e-01 -2.19192758e-01 1.87407240e-01
1.15398571e-01 1.24180937e+00 -1.29392982e+00 -4.69805866e-01
-4.12027299e-01 6.70261621e-01 1.91610515e-01 4.74923611e-01
-4.94953662e-01 -1.88424021e-01 3.54630321e-01 3.74027379e-02
2.43488595e-01 -1.25831319e-02 -1.29322374e+00 -6.18246734e-01
-1.45636231e-01 -4.83942747e-01 3.32093835e-01 -3.73751312e-01
-6.83012486e-01 6.21821702e-01 -2.14426249e-01 -1.33017445e+00
2.73724735e-01 -6.93705678e-01 -8.15149844e-01 8.53844941e-01
-1.69231176e+00 -9.82604802e-01 -4.55010533e-01 6.05168343e-01
8.14167261e-01 -3.12314063e-01 3.27245921e-01 1.48243785e-01
-7.87418187e-01 4.71202910e-01 2.42434919e-01 2.80926615e-01
4.12575275e-01 -1.11533761e+00 -1.71189800e-01 7.55059183e-01
-2.75278151e-01 1.51987880e-01 4.96553540e-01 -7.18792677e-01
-5.00075817e-01 -1.22005868e+00 7.09535658e-01 4.13151115e-01
2.59609342e-01 1.00416712e-01 -7.25192308e-01 4.25645739e-01
2.44324416e-01 2.41962910e-01 4.19073075e-01 -3.47668111e-01
3.31319809e-01 -1.58277065e-01 -1.39079893e+00 6.24130785e-01
5.03967881e-01 -4.13908111e-03 -7.76508749e-02 5.16610742e-01
3.58910352e-01 -4.33477372e-01 -5.67897022e-01 5.05491555e-01
2.91861087e-01 -7.86998689e-01 9.29137647e-01 -4.01038617e-01
5.09432375e-01 -2.52029181e-01 2.25328177e-01 -1.18337476e+00
-2.08233461e-01 -5.85354902e-02 5.81005454e-01 6.90471947e-01
7.06247628e-01 -5.65500975e-01 1.35965610e+00 5.12312472e-01
-3.42815965e-01 -1.09566438e+00 -9.01262701e-01 -4.47619766e-01
3.11451107e-01 -3.25916916e-01 3.83082598e-01 8.12033415e-01
-3.66763026e-01 -4.95801449e-01 -9.12758987e-03 2.10978270e-01
6.55482173e-01 -6.72388375e-02 1.68818787e-01 -1.18387818e+00
1.54513434e-01 -9.16323364e-01 -4.80998337e-01 -9.35935318e-01
6.23265170e-02 -9.64272857e-01 1.44276634e-01 -1.53317583e+00
1.99500069e-01 -8.51083457e-01 -5.94855785e-01 5.81630111e-01
-6.97813109e-02 1.81460276e-01 -1.56345323e-01 2.06798300e-01
-4.10147533e-02 1.42419701e-02 1.68356025e+00 -4.30673152e-01
-3.32716703e-01 1.64815769e-01 -3.47556621e-01 8.22804928e-01
1.02754033e+00 -4.80943233e-01 -3.33652616e-01 -2.28606373e-01
-3.97797108e-01 -6.50854185e-02 2.83859849e-01 -1.23202145e+00
4.25176799e-01 -2.69292355e-01 1.01772714e+00 -9.75223362e-01
6.57013431e-02 -1.11309028e+00 -1.06694408e-01 6.57499552e-01
-6.30470216e-01 -3.02063763e-01 1.46870419e-01 1.21598393e-01
-3.79310220e-01 -4.93979543e-01 1.20214069e+00 -5.09809315e-01
-7.13089228e-01 4.79561746e-01 -4.76710081e-01 -2.90473133e-01
1.20795751e+00 -7.54436255e-01 3.09017263e-02 1.55741975e-01
-1.04953766e+00 2.64491647e-01 1.04418006e-02 -3.69915590e-02
4.61141586e-01 -1.14090967e+00 -5.02175629e-01 4.28513318e-01
-8.27923864e-02 3.19475085e-01 2.25890636e-01 1.22505593e+00
-9.69396293e-01 3.35118979e-01 -4.25843537e-01 -9.00652766e-01
-1.19965196e+00 1.05622455e-01 7.00034857e-01 -1.86638355e-01
-7.99097598e-01 1.02570760e+00 2.56967574e-01 -3.04166656e-02
4.91670072e-01 -8.35450411e-01 -4.59028333e-01 3.61397304e-02
2.57920772e-01 -8.15437734e-03 7.80336708e-02 -5.38165450e-01
-2.25834802e-01 5.55479348e-01 -1.86571240e-01 4.23348844e-02
1.21977448e+00 1.23145021e-01 -2.41622970e-01 1.61715433e-01
1.36676717e+00 -3.91294658e-01 -1.42377400e+00 -2.60474205e-01
2.34147999e-02 -2.34843493e-01 5.94296038e-01 -8.19460869e-01
-1.70304227e+00 8.30110729e-01 1.05451024e+00 -1.78375974e-01
1.11623299e+00 -1.86168730e-01 7.21443594e-01 1.88174054e-01
-8.13072249e-02 -1.56815493e+00 -2.09966168e-01 4.00262684e-01
5.44946074e-01 -1.27039909e+00 3.00331526e-02 -4.75657612e-01
-5.48461378e-01 1.33349729e+00 7.52409160e-01 -2.88493395e-01
9.24228966e-01 5.15760064e-01 2.55906999e-01 -2.49338940e-01
-2.60256499e-01 -2.95308143e-01 3.90896238e-02 4.88787889e-01
2.67542183e-01 3.12020630e-01 -5.89548230e-01 3.29753339e-01
-8.13869610e-02 -7.10192993e-02 2.25209743e-01 8.97816122e-01
-8.83732677e-01 -9.48266387e-01 -3.50363821e-01 7.08999038e-01
-5.54109097e-01 -2.82259613e-01 -1.13923296e-01 8.45918059e-01
5.73758125e-01 4.57350552e-01 3.26956034e-01 -8.94680340e-03
-3.45664774e-03 1.00873157e-01 8.16755220e-02 -2.45287776e-01
-8.38380635e-01 4.63827372e-01 -1.94130346e-01 -5.18393576e-01
-2.50698209e-01 -7.71904707e-01 -1.62045300e+00 3.47442985e-01
-2.54116416e-01 -2.60427564e-01 8.21641326e-01 9.25734460e-01
-1.08346850e-01 1.01251912e+00 3.73304844e-01 -5.93301117e-01
-1.58663198e-01 -8.83039415e-01 -6.34527504e-01 1.62264034e-01
3.40770155e-01 -3.80837560e-01 -1.97467152e-02 1.03961296e-01] | [14.519079208374023, -2.4648501873016357] |
ccfba1ff-d448-4810-902d-4406ebcdd678 | a-review-of-panoptic-segmentation-for-mobile | 2304.13980 | null | https://arxiv.org/abs/2304.13980v1 | https://arxiv.org/pdf/2304.13980v1.pdf | A Review of Panoptic Segmentation for Mobile Mapping Point Clouds | 3D point cloud panoptic segmentation is the combined task to (i) assign each point to a semantic class and (ii) separate the points in each class into object instances. Recently there has been an increased interest in such comprehensive 3D scene understanding, building on the rapid advances of semantic segmentation due to the advent of deep 3D neural networks. Yet, to date there is very little work about panoptic segmentation of outdoor mobile-mapping data, and no systematic comparisons. The present paper tries to close that gap. It reviews the building blocks needed to assemble a panoptic segmentation pipeline and the related literature. Moreover, a modular pipeline is set up to perform comprehensive, systematic experiments to assess the state of panoptic segmentation in the context of street mapping. As a byproduct, we also provide the first public dataset for that task, by extending the NPM3D dataset to include instance labels. | ['Konrad Schindler', 'Torben Peters', 'Yuanwen Yue', 'Binbin Xiang'] | 2023-04-27 | null | null | null | null | ['panoptic-segmentation'] | ['computer-vision'] | [ 2.95563906e-01 -7.29820356e-02 -1.83316201e-01 -5.60979724e-01
-4.03289795e-01 -7.27414370e-01 7.01027393e-01 1.98750362e-01
-1.37923524e-01 2.26911232e-01 -1.46114960e-01 -5.12828410e-01
-2.04636961e-01 -1.09977615e+00 -5.69530964e-01 -2.95935780e-01
-4.00110930e-02 9.38285828e-01 4.70678985e-01 -5.52240536e-02
2.00543210e-01 9.07715738e-01 -1.74743366e+00 8.00385922e-02
9.17343616e-01 1.07368183e+00 4.29998636e-01 4.13927644e-01
-4.59702224e-01 6.09337986e-02 -3.79802585e-01 -5.32216914e-02
6.19997025e-01 -1.20892636e-01 -9.96300340e-01 5.62071264e-01
8.29863489e-01 -1.40356019e-01 7.74964690e-02 9.29555058e-01
1.61992326e-01 1.40331835e-01 4.71427232e-01 -1.15965688e+00
-9.68456045e-02 1.20362975e-01 -4.74193692e-01 1.99396953e-01
-2.04615369e-01 4.29925844e-02 1.15913010e+00 -4.10253674e-01
6.15881264e-01 1.00778949e+00 7.40842044e-01 5.18689640e-02
-1.03257704e+00 -4.12979722e-01 3.67045552e-01 -2.54585952e-01
-1.12084079e+00 -2.09921226e-02 5.34379959e-01 -5.73666811e-01
1.09544468e+00 3.61870527e-01 1.00827324e+00 4.42110866e-01
-4.27434266e-01 8.11063588e-01 1.14985621e+00 -2.31726840e-01
3.40797663e-01 -4.67308797e-02 3.36884856e-01 3.17739487e-01
2.11872101e-01 -5.91891408e-02 1.26944399e-02 3.11493605e-01
6.97081447e-01 1.67964362e-02 2.00936034e-01 -6.22457922e-01
-8.70261729e-01 6.92627430e-01 7.30805814e-01 3.56196821e-01
-2.43860021e-01 8.51753578e-02 2.52323955e-01 7.23983422e-02
9.24099267e-01 4.75052834e-01 -6.65839255e-01 4.22708057e-02
-1.33867204e+00 4.56619769e-01 7.33860493e-01 8.33214700e-01
1.03979421e+00 -2.61874437e-01 4.25192207e-01 8.82348537e-01
3.80427778e-01 4.50245827e-01 -1.43825740e-01 -1.17424273e+00
5.81298172e-01 8.41435552e-01 9.17787999e-02 -9.03802633e-01
-6.62993371e-01 -2.46378407e-01 -3.18600327e-01 3.07178587e-01
3.11891168e-01 -2.45713852e-02 -1.19909394e+00 1.14865232e+00
4.86851126e-01 9.04531777e-02 -1.77431509e-01 8.95058751e-01
8.56723130e-01 7.31295645e-01 6.47018999e-02 3.79026383e-01
1.22894537e+00 -7.24612236e-01 -2.65294343e-01 -3.77727479e-01
3.81179899e-01 -6.43736124e-01 8.77571523e-01 1.48737326e-01
-6.28683329e-01 -6.60627604e-01 -1.07527661e+00 -4.41830382e-02
-9.32162762e-01 -1.59786955e-01 1.00763595e+00 1.05217195e+00
-1.10421968e+00 4.36429620e-01 -9.50627744e-01 -7.32985258e-01
7.50766873e-01 2.94701427e-01 -1.39427900e-01 8.94379020e-02
-1.10589159e+00 7.14755297e-01 5.29614687e-01 3.98071520e-02
-5.87704599e-01 -7.90265262e-01 -7.86978245e-01 -2.12642983e-01
3.21690738e-01 -6.16924226e-01 1.17794847e+00 -7.82376826e-01
-1.20103514e+00 1.33973420e+00 8.76002610e-02 -5.89454055e-01
3.10314596e-01 -2.29979455e-01 -1.82463676e-01 1.62322998e-01
3.89121920e-01 1.33076847e+00 2.14670122e-01 -1.45111585e+00
-1.12127900e+00 -5.99787772e-01 4.04789567e-01 3.35242987e-01
1.38279855e-01 -8.45323280e-02 -7.05455005e-01 -3.81879628e-01
4.19578731e-01 -9.74338174e-01 -5.07951677e-01 -1.88572541e-01
-3.10621381e-01 -8.19397196e-02 8.47447336e-01 -4.85735685e-01
8.74920845e-01 -1.98674929e+00 -2.28821427e-01 1.48312286e-01
7.10821450e-02 3.22806984e-01 1.78231671e-01 2.97442496e-01
-2.69223023e-02 2.98788041e-01 -9.28202212e-01 -3.18766356e-01
7.86050409e-02 3.89981985e-01 -5.87156892e-01 2.87071496e-01
1.17183477e-01 7.85922229e-01 -6.82828069e-01 -1.94795951e-01
7.32389510e-01 1.80138886e-01 -5.02883255e-01 -2.55300999e-01
-7.01442420e-01 4.47097868e-01 -4.56325561e-01 7.95432270e-01
1.06574595e+00 -7.64402822e-02 -7.14960918e-02 2.07105383e-01
-6.75166011e-01 4.50722992e-01 -1.18859148e+00 1.71295822e+00
-3.44603688e-01 9.03596163e-01 3.36286500e-02 -1.06465876e+00
9.52606857e-01 4.16815318e-02 7.84127295e-01 -6.67964816e-01
-1.59178898e-02 3.50987107e-01 -4.12388682e-01 -3.38235736e-01
8.09432805e-01 -1.00465186e-01 -2.18913689e-01 2.41609782e-01
-1.80464715e-01 -8.43033195e-01 2.39955068e-01 -1.03842847e-01
5.15370905e-01 3.39800268e-01 -1.69766188e-01 -2.83663213e-01
2.54752308e-01 6.82335496e-01 1.82038367e-01 6.38052106e-01
-6.27318919e-02 8.93477023e-01 3.47671837e-01 -7.27788925e-01
-9.08350766e-01 -9.03655171e-01 -5.47920644e-01 6.75261676e-01
4.37257320e-01 -7.59274065e-02 -9.36606288e-01 -5.43052673e-01
1.00282766e-01 6.80918932e-01 -4.07736719e-01 8.23964238e-01
-4.85546708e-01 -1.19638693e+00 4.98499364e-01 4.52279061e-01
8.50086153e-01 -1.01775169e+00 -8.32782090e-01 7.40300938e-02
-1.15033634e-01 -1.26816332e+00 2.58236438e-01 2.55534858e-01
-1.21123016e+00 -1.17091858e+00 -4.52726185e-01 -7.35444784e-01
3.56543958e-01 4.33303297e-01 1.17785859e+00 -1.29480705e-01
-1.46188006e-01 2.18999550e-01 -2.92167276e-01 -4.51086849e-01
-7.66167715e-02 5.44737518e-01 -4.17794287e-01 -3.10112149e-01
5.60974181e-01 -5.53435802e-01 -3.91843975e-01 4.84838039e-01
-8.91629457e-01 1.97662339e-01 2.85544991e-01 -1.82872608e-01
8.61065388e-01 1.21233717e-01 1.51379779e-01 -9.07272577e-01
8.88760760e-02 -5.24170160e-01 -1.05500162e+00 -6.03439771e-02
-4.00894701e-01 -4.39919889e-01 -2.88536157e-02 3.59218597e-01
-9.61327076e-01 2.77558953e-01 -7.00122416e-01 6.63201138e-02
-8.66198540e-01 5.50162673e-01 -2.32933775e-01 1.15762755e-01
3.91420215e-01 -1.05810210e-01 -3.88116688e-01 -8.56888294e-01
6.43261731e-01 8.93512428e-01 3.48008841e-01 -4.50900733e-01
6.26690626e-01 9.90681767e-01 8.96113366e-02 -1.25321531e+00
-9.61840510e-01 -8.39917719e-01 -1.18821537e+00 -1.42779365e-01
1.12613821e+00 -8.05989504e-01 -5.58195747e-02 5.68290949e-01
-9.21609581e-01 -7.09481776e-01 -3.19257349e-01 1.54150695e-01
-6.43088996e-01 3.41429770e-01 -2.58581489e-01 -5.98068595e-01
-1.75593756e-02 -1.23433650e+00 1.18690050e+00 2.14408517e-01
-4.22350988e-02 -9.13239539e-01 2.13708714e-01 6.39802992e-01
4.95775323e-03 3.13711405e-01 7.22710669e-01 -3.75742406e-01
-9.22474563e-01 -2.16014042e-01 -4.49283391e-01 2.50634462e-01
8.90610144e-02 1.83876336e-01 -1.11814022e+00 3.41225207e-01
-1.40666857e-01 1.32654458e-01 1.09207523e+00 7.33766854e-01
1.04369926e+00 3.21949452e-01 -4.93269295e-01 8.74690950e-01
1.51461041e+00 1.77464664e-01 7.04269767e-01 6.72812939e-01
7.69056380e-01 8.46818864e-01 6.47602320e-01 -1.36738792e-01
6.19777083e-01 6.54979467e-01 7.43979156e-01 -2.62674361e-01
-2.68690139e-01 -1.08845599e-01 -3.22722167e-01 4.19591874e-01
-9.71194580e-02 -4.28445280e-01 -1.34948325e+00 7.29763687e-01
-1.56778395e+00 -6.13243282e-01 -6.28915906e-01 1.98353732e+00
2.38015860e-01 2.54287630e-01 3.39967795e-02 1.45133749e-01
6.53982222e-01 4.96518701e-01 -1.67456016e-01 -2.47297198e-01
-1.93772659e-01 2.60741562e-01 7.29704797e-01 6.07541740e-01
-1.72326481e+00 1.35480118e+00 6.43943548e+00 6.86647177e-01
-1.07977057e+00 -6.57138377e-02 6.73235297e-01 2.56426901e-01
-1.16127945e-01 2.66094804e-01 -9.99940336e-01 3.14288825e-01
7.96003401e-01 6.55233264e-01 8.75973105e-02 7.68885791e-01
2.69329518e-01 -4.58950430e-01 -5.72281122e-01 5.69149792e-01
-3.16579103e-01 -1.46951997e+00 8.49947631e-02 2.17896268e-01
9.15068507e-01 7.28903055e-01 -5.84184676e-02 7.23151341e-02
2.88357228e-01 -8.70338500e-01 5.07379651e-01 2.41557777e-01
5.47040164e-01 -5.68250835e-01 4.53052342e-01 3.75506729e-01
-1.15778732e+00 6.91675544e-02 -2.90621668e-01 -2.71001071e-01
3.19051355e-01 6.11556947e-01 -6.61780000e-01 7.32290089e-01
9.29355502e-01 8.19117785e-01 -4.38766986e-01 1.40067184e+00
-3.30621302e-01 6.36780679e-01 -6.12340093e-01 3.83748233e-01
7.53476918e-01 -4.85559702e-01 4.13841456e-01 1.10775042e+00
1.47432789e-01 -1.20516516e-01 3.61434013e-01 7.90183127e-01
9.12097767e-02 3.14807780e-02 -4.57156330e-01 -4.47411016e-02
1.34253636e-01 1.15483487e+00 -1.36665201e+00 -2.20785454e-01
-5.37123978e-01 6.47248924e-01 -9.77813900e-02 1.40940085e-01
-5.17314017e-01 -2.41019741e-01 9.08536494e-01 3.50811630e-01
3.34921032e-01 -6.52529776e-01 -9.43845332e-01 -7.20149636e-01
-2.53426522e-01 -2.76898324e-01 4.01744634e-01 -9.21475649e-01
-9.65559304e-01 2.26503849e-01 3.24573398e-01 -9.16121185e-01
2.06114426e-01 -7.13049650e-01 -3.46728653e-01 8.01813066e-01
-1.73882103e+00 -1.18568480e+00 -5.04470587e-01 -7.47030675e-02
7.50149071e-01 3.38998914e-01 6.77335918e-01 3.60157311e-01
-3.22690159e-01 -3.61150920e-01 8.81311223e-02 1.17683016e-01
1.63483992e-01 -1.29971838e+00 1.02975988e+00 7.18022764e-01
2.24827990e-01 1.80348158e-01 3.73633653e-01 -8.56389642e-01
-5.11666834e-01 -1.20447981e+00 8.77187192e-01 -8.15849364e-01
5.99499524e-01 -3.84851992e-01 -8.18993390e-01 6.85264826e-01
-1.09418742e-01 -3.71582329e-01 4.78995770e-01 1.17069162e-01
1.23354375e-01 -9.15988013e-02 -9.77859497e-01 3.46245497e-01
1.06690311e+00 -3.08907181e-01 -7.39135921e-01 3.65117252e-01
7.56103992e-01 -5.12192190e-01 -6.31702065e-01 6.86143637e-01
2.87962139e-01 -1.13874531e+00 1.14515352e+00 -2.24038661e-02
3.70142430e-01 -4.67988521e-01 -3.13374877e-01 -9.70234275e-01
3.81005071e-02 -1.62312955e-01 4.29863662e-01 1.01471496e+00
3.86492968e-01 -5.23448169e-01 1.04873896e+00 3.52144688e-01
-5.94579577e-01 -5.34659922e-01 -8.95119309e-01 -6.50720060e-01
1.74573600e-01 -8.95131946e-01 8.50529909e-01 9.09540176e-01
-6.75021887e-01 1.25081927e-01 1.57453731e-01 4.16316032e-01
2.43431211e-01 6.91363990e-01 8.76760960e-01 -1.64515138e+00
2.63896942e-01 -6.81126535e-01 -3.94907683e-01 -1.46428525e+00
-9.79634896e-02 -1.07959449e+00 1.91818047e-02 -2.07272315e+00
-2.15791300e-01 -1.01078808e+00 7.46568814e-02 2.98789352e-01
1.26114875e-01 6.02808654e-01 -5.72926272e-03 4.71710950e-01
-4.78955835e-01 2.81593084e-01 1.08211374e+00 -2.25689393e-02
-5.43226004e-01 3.86537671e-01 -5.49297214e-01 8.60437334e-01
1.14680040e+00 -4.04580534e-01 -4.07688946e-01 -7.12018669e-01
3.32134396e-01 -4.66062427e-01 3.83943528e-01 -1.27978742e+00
-1.56575590e-01 -1.16548747e-01 1.62573591e-01 -1.27859771e+00
3.74082595e-01 -9.21949387e-01 1.08600073e-01 1.92316771e-01
1.77499965e-01 -3.89406472e-01 4.53393549e-01 3.14654201e-01
-1.64383218e-01 -2.63050109e-01 6.41059041e-01 -4.07324314e-01
-1.19443357e+00 5.18018007e-01 -3.78347546e-01 -2.28741765e-02
8.51702332e-01 -6.97591603e-01 -9.79160797e-03 1.69572741e-01
-7.32639015e-01 3.84169370e-01 6.45831287e-01 4.30765510e-01
7.45834410e-02 -7.30392814e-01 -3.45139503e-01 2.78048217e-01
7.28916749e-02 5.42368591e-01 2.84447312e-01 5.45894861e-01
-1.07223618e+00 7.30971634e-01 -2.13601440e-01 -8.61487091e-01
-8.78778636e-01 3.52577955e-01 5.37303388e-01 -2.65111458e-02
-7.42284834e-01 6.10047817e-01 2.63424248e-01 -8.68011057e-01
-6.58390597e-02 -6.95163548e-01 -4.08697486e-01 4.16956961e-01
1.39166061e-02 3.63590658e-01 2.48810366e-01 -6.97953701e-01
-2.93827862e-01 7.75673151e-01 5.84499121e-01 -2.06580505e-01
1.50131261e+00 -2.16699719e-01 -2.03681856e-01 5.97947419e-01
8.48630965e-01 -1.88873082e-01 -1.43442893e+00 8.98059607e-02
1.50494367e-01 -4.46746737e-01 2.13754803e-01 -7.06784666e-01
-9.45074320e-01 1.12884676e+00 5.96794486e-01 6.00306273e-01
9.37425911e-01 2.08810598e-01 7.62198329e-01 2.12864295e-01
2.90270597e-01 -1.20869780e+00 -5.23724377e-01 7.84627140e-01
5.04741430e-01 -1.14447296e+00 6.44767955e-02 -7.60056436e-01
-3.81929845e-01 9.31534171e-01 2.46370718e-01 -7.64793381e-02
8.17660153e-01 3.09548136e-02 1.50634363e-01 -6.47508085e-01
2.45400548e-01 -6.26896501e-01 2.80665994e-01 6.72200978e-01
1.05394594e-01 2.83504039e-01 1.19329788e-01 2.26884052e-01
-3.69383842e-01 1.15915172e-01 3.38509530e-02 6.75677359e-01
-8.57163668e-01 -9.93878186e-01 -3.38106811e-01 3.75676155e-01
-1.79544538e-01 -1.63962200e-01 -4.09066767e-01 1.03704524e+00
4.33607340e-01 8.68730128e-01 5.33356130e-01 -4.41547185e-02
4.08089727e-01 -5.41527607e-02 3.17719042e-01 -7.41639376e-01
-4.50233132e-01 -6.67328686e-02 1.17819846e-01 -2.01305300e-01
-8.00468087e-01 -7.62964547e-01 -1.20604527e+00 -1.74849212e-01
-1.89557318e-02 -6.50459528e-02 1.02554309e+00 9.17177558e-01
9.66384560e-02 2.83312291e-01 3.12220365e-01 -1.18072605e+00
4.00904298e-01 -7.01279104e-01 -5.15767097e-01 4.52392399e-02
-3.82437417e-03 -5.12304783e-01 -2.20351875e-01 -3.99561711e-02] | [8.653562545776367, -2.2076261043548584] |
e48c06c6-e7f7-434e-bf5c-51b6577ffc73 | masked-autoencoders-as-image-processors | 2303.17316 | null | https://arxiv.org/abs/2303.17316v1 | https://arxiv.org/pdf/2303.17316v1.pdf | Masked Autoencoders as Image Processors | Transformers have shown significant effectiveness for various vision tasks including both high-level vision and low-level vision. Recently, masked autoencoders (MAE) for feature pre-training have further unleashed the potential of Transformers, leading to state-of-the-art performances on various high-level vision tasks. However, the significance of MAE pre-training on low-level vision tasks has not been sufficiently explored. In this paper, we show that masked autoencoders are also scalable self-supervised learners for image processing tasks. We first present an efficient Transformer model considering both channel attention and shifted-window-based self-attention termed CSformer. Then we develop an effective MAE architecture for image processing (MAEIP) tasks. Extensive experimental results show that with the help of MAEIP pre-training, our proposed CSformer achieves state-of-the-art performance on various image processing tasks, including Gaussian denoising, real image denoising, single-image motion deblurring, defocus deblurring, and image deraining. | ['Guangtao Zhai', 'Jia Wang', 'Long Teng', 'Danyang Tu', 'Xiongkuo Min', 'Wei Shen', 'Huiyu Duan'] | 2023-03-30 | null | null | null | null | ['deblurring'] | ['computer-vision'] | [ 3.12012345e-01 -1.66129515e-01 4.33431238e-01 -3.41246992e-01
-6.47387266e-01 5.14224023e-02 6.83632612e-01 -3.86909515e-01
-6.73973918e-01 2.90049165e-01 3.76086712e-01 -1.00920610e-01
4.59052995e-03 -3.26694012e-01 -9.15714741e-01 -9.84135866e-01
5.42741343e-02 -4.18921232e-01 2.98391700e-01 -1.43423319e-01
2.17357874e-01 2.31356829e-01 -1.85217094e+00 3.27942461e-01
1.27317119e+00 1.05605030e+00 6.17648482e-01 8.80374193e-01
1.84233680e-01 8.96103621e-01 -5.97184479e-01 -4.02822584e-01
7.03932568e-02 -2.65840977e-01 -3.84119898e-01 2.50802845e-01
8.29485595e-01 -5.58808386e-01 -6.33502245e-01 1.32980943e+00
7.60032594e-01 6.14161938e-02 5.51280499e-01 -9.18227792e-01
-1.20287347e+00 2.47800961e-01 -5.73139548e-01 5.46783507e-01
-1.79669052e-01 3.69793624e-01 5.85672855e-01 -1.15836406e+00
1.49006948e-01 1.17406917e+00 7.87048876e-01 5.92844903e-01
-9.59014535e-01 -5.49154520e-01 1.13276895e-02 7.59773970e-01
-8.17350090e-01 -5.96770942e-01 6.20054007e-01 -2.46138215e-01
1.33058167e+00 -2.28668496e-01 4.64628935e-01 9.47119951e-01
7.24154413e-01 8.99523497e-01 1.47657168e+00 -4.15535390e-01
8.19222108e-02 -1.81683555e-01 3.07611972e-01 5.68834066e-01
1.36884600e-01 3.60035837e-01 -7.36618996e-01 5.16643167e-01
8.41536999e-01 -1.34866536e-01 -6.69179380e-01 9.66345146e-02
-1.16165936e+00 5.89205682e-01 6.41365170e-01 2.89147109e-01
-7.23992348e-01 1.81186065e-01 2.63827175e-01 2.33675927e-01
5.74446499e-01 1.91027507e-01 -3.68959010e-01 1.52365059e-01
-8.41518223e-01 -2.95720965e-01 1.17642239e-01 5.64677298e-01
6.83626533e-01 6.22832954e-01 -4.64385271e-01 9.13492382e-01
9.07472596e-02 7.15085685e-01 8.52690279e-01 -9.32709515e-01
1.27312005e-01 -3.80900763e-02 1.35338172e-01 -7.09845960e-01
-1.87490746e-01 -6.40132725e-01 -1.51377332e+00 4.93500739e-01
-2.17162333e-02 -4.27010804e-02 -1.37538624e+00 1.54705286e+00
9.49916616e-03 5.70773482e-01 4.02155787e-01 1.05635512e+00
1.05230486e+00 8.56764913e-01 -1.07660882e-01 -2.58429796e-01
1.42768466e+00 -1.28819084e+00 -1.15733016e+00 -4.46470171e-01
-1.68147638e-01 -6.77745044e-01 9.92587686e-01 7.19034731e-01
-1.15603888e+00 -1.07690883e+00 -1.07487714e+00 -3.65974128e-01
-8.05884898e-02 1.01039529e-01 6.78962767e-01 5.79035640e-01
-1.41736174e+00 6.54828072e-01 -1.19914329e+00 -1.30465791e-01
6.63897455e-01 3.06473196e-01 -3.98293436e-01 -3.86734635e-01
-9.62571919e-01 1.06377375e+00 3.16502340e-02 4.99723345e-01
-1.30302703e+00 -6.48228884e-01 -8.73046875e-01 1.57724246e-01
1.29035860e-01 -8.61976981e-01 1.08291912e+00 -7.37203896e-01
-1.65457380e+00 6.66558027e-01 -4.31249052e-01 -8.41803730e-01
1.53325005e-02 -6.53353453e-01 -6.16282940e-01 2.94352591e-01
2.29971223e-02 7.77904630e-01 1.69413793e+00 -1.02473891e+00
-4.59756911e-01 -3.25955153e-01 -4.06067163e-01 2.58188814e-01
-5.60042083e-01 1.70363244e-02 -4.04544890e-01 -7.95810223e-01
-4.63131117e-03 -4.58563834e-01 -8.82842690e-02 -1.75071731e-01
-1.33977026e-01 -1.02350928e-01 9.83923018e-01 -9.85869765e-01
9.23789501e-01 -2.25773120e+00 1.64476648e-01 -5.59191167e-01
3.37212235e-01 7.00932503e-01 -3.96832705e-01 -5.74522056e-02
-2.08529904e-01 -3.48807842e-01 -2.66744226e-01 -5.23431659e-01
-2.75211602e-01 2.15640828e-01 -2.57254779e-01 4.93430704e-01
5.20632148e-01 1.07491076e+00 -5.82168579e-01 -1.53965101e-01
6.42274618e-01 9.89464223e-01 -5.28605938e-01 4.62775767e-01
7.86432326e-02 5.19254863e-01 3.21984291e-03 5.30671954e-01
8.73180568e-01 -2.50400782e-01 -6.10026836e-01 -8.06116402e-01
-2.84302711e-01 -2.99268246e-01 -6.82284117e-01 1.55358613e+00
-5.70452332e-01 9.86071825e-01 3.70231241e-01 -9.14633453e-01
6.06490135e-01 1.75174788e-01 1.89919248e-01 -8.60152483e-01
1.80585995e-01 7.52722919e-02 -4.15258147e-02 -7.48799562e-01
4.26074564e-01 -7.52583146e-02 6.30451381e-01 6.15572147e-02
4.85654324e-01 -2.60830730e-01 -1.05594411e-01 5.11143804e-02
1.02362132e+00 -1.08031802e-01 5.40775731e-02 -1.25346959e-01
7.26545691e-01 -3.56311113e-01 2.88950771e-01 9.08608377e-01
-4.24895614e-01 8.04047346e-01 -1.58023059e-01 -2.08741799e-01
-9.74786520e-01 -8.97777498e-01 -1.16148345e-01 9.64313805e-01
9.85291600e-02 -1.19103290e-01 -1.01786411e+00 -1.64421275e-01
-3.39107871e-01 5.78234851e-01 -5.92518449e-01 -3.06326926e-01
-3.43759954e-01 -1.04669607e+00 2.86706626e-01 4.22490150e-01
1.18419969e+00 -1.15284812e+00 -5.95716834e-01 1.66389823e-01
-2.62137145e-01 -1.41158640e+00 -6.17534161e-01 2.38881126e-01
-8.86910260e-01 -7.40573704e-01 -1.08827817e+00 -1.00745404e+00
6.40923798e-01 6.73096418e-01 8.20908010e-01 -3.39635462e-01
-3.63078803e-01 6.69476449e-01 -2.08719984e-01 -3.08970928e-01
-1.69143379e-01 -5.11121154e-01 1.17288679e-01 2.62644321e-01
1.63338915e-01 -7.03047335e-01 -9.81935620e-01 1.92712635e-01
-1.05718684e+00 -1.70952678e-01 1.03233278e+00 1.19626546e+00
5.83216429e-01 3.79321992e-01 4.88328487e-01 -3.68977636e-01
7.91571438e-01 7.23373517e-02 -4.68995363e-01 1.76664948e-01
-6.92944050e-01 2.01433644e-01 5.67698896e-01 -6.22553527e-01
-1.38286650e+00 -2.95090556e-01 -3.46165687e-01 -8.46697628e-01
-1.65944010e-01 3.43718916e-01 -2.19696704e-02 -3.40410739e-01
6.39588118e-01 8.12901795e-01 2.85410166e-01 -4.46294338e-01
4.19667393e-01 7.06342101e-01 1.35178304e+00 -6.71612248e-02
7.48027086e-01 6.05709434e-01 -9.64531153e-02 -1.14846396e+00
-9.49041307e-01 -3.72376502e-01 -3.59274268e-01 -1.03658766e-01
1.31429696e+00 -1.30464745e+00 -7.40924478e-01 1.22236145e+00
-1.04736185e+00 -3.42574865e-01 -1.83740824e-01 6.28094196e-01
-4.27013248e-01 6.13553524e-01 -7.14643419e-01 -5.63155890e-01
-7.28583276e-01 -1.38492298e+00 7.95055270e-01 4.03100669e-01
5.42239249e-01 -6.99493289e-01 -2.66456634e-01 5.53633869e-01
8.53729367e-01 -3.22166264e-01 6.18045509e-01 -6.89777434e-02
-7.21250474e-01 6.14482798e-02 -3.40915531e-01 1.03693736e+00
1.08275535e-02 -5.15797436e-01 -1.39568472e+00 -5.10431111e-01
4.76444691e-01 -3.98797601e-01 1.40072405e+00 1.03387165e+00
1.37255943e+00 -8.47091824e-02 1.76633924e-01 1.12423086e+00
1.24087214e+00 -1.12088464e-01 1.09549713e+00 2.82736927e-01
7.07451820e-01 3.73083621e-01 3.06696743e-01 1.86209038e-01
3.24328870e-01 2.41444781e-01 5.67675352e-01 -3.92086953e-01
-6.58774078e-01 1.19678669e-01 5.78719795e-01 9.58946764e-01
-9.73939523e-02 -7.27690086e-02 -5.07179856e-01 6.53534353e-01
-1.55915689e+00 -8.94247174e-01 -1.96093857e-01 1.96038640e+00
6.36195362e-01 -8.00096840e-02 -4.46687877e-01 6.40469939e-02
7.04282522e-01 3.57990205e-01 -9.44489777e-01 -3.61718573e-02
-4.43558842e-01 4.00377512e-01 3.68413031e-01 5.11141717e-01
-1.29819310e+00 1.02758479e+00 5.64974213e+00 8.90739918e-01
-1.19544733e+00 4.22216564e-01 6.14537060e-01 2.78958797e-01
1.84424907e-01 -4.68983501e-01 -6.30395412e-01 3.91683072e-01
8.17490458e-01 1.54543683e-01 5.81713617e-01 5.63179433e-01
2.53890127e-01 -1.46882355e-01 -8.03770542e-01 1.36039913e+00
3.83647144e-01 -1.13453245e+00 -1.87599942e-01 -2.01052517e-01
8.19575012e-01 4.38273698e-01 5.30395329e-01 3.10545206e-01
6.53424254e-03 -1.09149134e+00 4.26800430e-01 4.02901143e-01
7.96119988e-01 -4.42565560e-01 7.56636977e-01 2.18327165e-01
-8.22445095e-01 -1.45581961e-01 -6.11365736e-01 2.03356519e-01
5.23520783e-02 7.59677708e-01 -3.04205090e-01 4.43350673e-01
1.06501007e+00 1.03484070e+00 -5.91726661e-01 1.05424857e+00
-4.19822752e-01 7.10566521e-01 -9.77848396e-02 2.91123092e-01
1.59484655e-01 -8.83474797e-02 5.04475832e-01 1.15246022e+00
3.10939968e-01 2.97244161e-01 -4.14454371e-01 5.64291716e-01
2.38506943e-02 -4.07703131e-01 -1.02006920e-01 7.51098990e-02
1.08584546e-01 1.21329010e+00 -3.28354925e-01 -3.26638192e-01
-6.13453925e-01 1.44445693e+00 -4.94952016e-02 7.46356070e-01
-6.07312024e-01 -4.65062171e-01 7.72475898e-01 -1.39352426e-01
7.82232106e-01 -2.23878518e-01 -8.12303182e-03 -1.51381338e+00
-3.58352363e-02 -9.30211425e-01 1.22313216e-01 -1.23295200e+00
-1.33661234e+00 8.33043575e-01 -3.21147084e-01 -1.00641954e+00
3.03469747e-02 -8.55308235e-01 -7.28118360e-01 9.45357263e-01
-2.08915377e+00 -1.04128289e+00 -6.89852536e-01 1.01511383e+00
6.90081358e-01 -3.25032115e-01 5.15052736e-01 2.25004539e-01
-6.56889200e-01 3.89212728e-01 3.28338027e-01 1.51805682e-02
7.29726970e-01 -1.05924535e+00 3.77268344e-01 1.37812972e+00
-8.12205300e-02 4.94720757e-01 7.85127699e-01 -4.03661460e-01
-1.65271723e+00 -1.15961361e+00 3.03124994e-01 -8.09891373e-02
5.84345222e-01 -1.83683991e-01 -1.19698310e+00 4.85871881e-01
6.05608940e-01 3.20853323e-01 1.35936692e-01 -2.75775760e-01
-3.74233097e-01 -3.76859426e-01 -1.00104499e+00 3.48103970e-01
7.61153758e-01 -5.80890536e-01 -6.63008928e-01 2.28054330e-01
7.00881660e-01 -3.30521911e-01 -7.09767997e-01 4.38816011e-01
3.09006840e-01 -1.26259851e+00 1.27857816e+00 -1.00067295e-01
6.32646859e-01 -2.72278130e-01 -4.78113554e-02 -1.61089897e+00
-4.30376917e-01 -7.38780975e-01 -3.88165563e-01 1.06280911e+00
-1.38067544e-01 -7.32042193e-01 5.56058824e-01 9.66638327e-02
-5.93576610e-01 -5.62273920e-01 -7.43314862e-01 -6.27160370e-01
-6.54605627e-02 -4.36308175e-01 5.25530465e-02 6.48100495e-01
-5.59158921e-01 4.91623908e-01 -7.09112823e-01 5.49370766e-01
1.15529871e+00 -1.76599517e-01 5.67358673e-01 -8.97118807e-01
-3.76653165e-01 -2.40821600e-01 -3.47116083e-01 -1.50783253e+00
1.86649576e-01 -5.61384439e-01 2.55138814e-01 -1.67901838e+00
9.80059728e-02 3.91800076e-01 -3.02532822e-01 3.73361975e-01
-3.81612062e-01 2.61274308e-01 -3.27138416e-02 2.35670581e-01
-4.07508135e-01 9.90003407e-01 1.40165830e+00 -2.53163248e-01
6.92176074e-02 -1.98280722e-01 -6.70016229e-01 6.08148217e-01
5.12801647e-01 -1.48692518e-01 -4.47736442e-01 -8.89293611e-01
-3.52565706e-01 6.01753443e-02 5.90646505e-01 -1.21302509e+00
6.12553775e-01 3.26548427e-01 6.09146416e-01 -6.66630805e-01
5.08970559e-01 -6.88719153e-01 -3.81010771e-01 4.62538272e-01
-5.77370115e-02 -1.41484708e-01 3.74462426e-01 7.62844980e-01
-6.12838626e-01 5.59377659e-04 9.82315481e-01 -1.60084516e-02
-1.09678197e+00 3.45318884e-01 -5.33196330e-01 -8.27938914e-02
5.07178366e-01 -2.63915002e-01 -4.97304499e-01 -5.62730074e-01
-6.98318005e-01 1.20148763e-01 -8.70323703e-02 1.10360004e-01
1.14111066e+00 -7.64452159e-01 -8.80815268e-01 5.54288208e-01
-2.36546770e-01 -5.14807962e-02 6.61069155e-01 9.32346165e-01
-1.68288082e-01 3.28383327e-01 -4.83969122e-01 -8.16470802e-01
-1.16662204e+00 6.85559452e-01 3.69653493e-01 9.41288248e-02
-8.88360858e-01 1.15460145e+00 3.79825234e-01 2.97237001e-02
4.44904536e-01 -5.97860396e-01 -1.36668116e-01 -2.77369201e-01
8.49648356e-01 2.76860923e-01 2.75014162e-01 -4.61854279e-01
-1.43437833e-01 8.03189754e-01 -3.37143093e-01 1.74678620e-02
1.55674160e+00 -3.35795045e-01 -2.01501250e-01 -2.30522975e-02
1.09703875e+00 -3.38605911e-01 -1.70277524e+00 -3.55226099e-01
-4.44459528e-01 -3.76563758e-01 8.24940264e-01 -8.24891686e-01
-1.28567815e+00 1.32825732e+00 1.07219601e+00 -1.23345584e-01
1.79997730e+00 -2.64218450e-01 7.93216228e-01 5.04701793e-01
1.50862381e-01 -7.06477940e-01 5.81633687e-01 5.29388607e-01
1.01573682e+00 -1.55861247e+00 -1.22774370e-01 -2.81377971e-01
-7.10775614e-01 8.81246686e-01 5.35271406e-01 -1.84613019e-01
6.88270986e-01 3.34506512e-01 2.09788129e-01 -2.18290463e-02
-6.49309039e-01 -4.39751714e-01 3.78752738e-01 1.00169909e+00
2.11925596e-01 -3.76556128e-01 2.57179230e-01 4.92259979e-01
-2.93572489e-02 5.06259017e-02 6.24002576e-01 4.68304724e-01
-6.61217749e-01 -6.12122357e-01 -6.05110228e-01 3.00053269e-01
-5.84601521e-01 -5.32307684e-01 1.15492441e-01 3.88580978e-01
-1.24028318e-01 1.11272478e+00 1.09930351e-01 -3.30900878e-01
1.76028267e-01 -1.90058932e-01 6.95295036e-01 -2.17415452e-01
-5.03564775e-01 2.89608926e-01 -3.42023492e-01 -4.52627987e-01
-6.89071357e-01 -1.94239184e-01 -7.87307262e-01 -2.44682096e-02
-3.81008416e-01 -6.81260079e-02 6.51571572e-01 6.94022834e-01
5.07509768e-01 7.79889464e-01 3.90415281e-01 -1.11720419e+00
-6.89835608e-01 -1.26444924e+00 -4.34791088e-01 2.82428592e-01
8.73810768e-01 -4.49700713e-01 -4.45068121e-01 2.55079687e-01] | [11.274456977844238, -2.3479089736938477] |
7fd2ca02-ea4b-44c0-bd28-ac86e3824ef4 | video-sparse-transformer-with-attention | null | null | https://ieeexplore.ieee.org/document/9798833 | https://ieeexplore.ieee.org/document/9798833 | Video Sparse Transformer With Attention-Guided Memory for Video Object Detection | Detecting objects in a video, known as Video Object Detection (VOD), is challenging since appearance changes of objects over time may bring detection errors. Recent research has focused on aggregating features from adjacent frames to compensate for the deteriorated appearances of a frame. Moreover, using distant frames is also proposed to deal with deteriorated appearances over several frames. Since an object’s position may change significantly at a distant frame, they only use features of object candidate regions, which do not depend on their position. However, such methods rely on object candidate regions’ detection performance and are not practical for deteriorated appearances. In this paper, we enhance features element-wisely before the object candidate region detection, proposing Video Sparse Transformer with Attention-guided Memory (VSTAM). Furthermore, we propose aggregating element-wise features sparsely to reduce processing time and memory cost. In addition, we introduce an external memory update strategy based on the utilization of the aggregation to hold long-term information effectively. Our method achieved 8.3% and 11.1% accuracy gain from the baseline on ImageNet VID and UA-DETRAC datasets. Our method demonstrates superior performance against state-of-the-art results on widely used VOD datasets. | ['Akihiro Sugimoto', 'Masato Fujitake'] | 2022-06-17 | null | null | null | ieee-access-2022-6 | ['video-object-detection', 'video-instance-segmentation'] | ['computer-vision', 'computer-vision'] | [ 1.31372288e-01 -6.67053759e-01 -1.10867143e-01 -1.35288998e-01
-3.15974593e-01 -2.05062120e-03 6.55350834e-02 1.07621841e-01
-4.64766860e-01 5.30510724e-01 9.33618024e-02 3.76368940e-01
1.98466390e-01 -5.91095090e-01 -6.23335898e-01 -6.84531569e-01
-1.01405509e-01 -3.02774608e-01 6.93758607e-01 1.90500066e-01
2.79392362e-01 3.66817802e-01 -1.73675120e+00 4.27244842e-01
8.64241898e-01 1.29662466e+00 6.67604566e-01 3.03520948e-01
-1.22724108e-01 9.21126306e-01 -6.30366385e-01 1.58173949e-01
3.28329206e-01 -3.04097176e-01 -3.47833931e-01 5.51804841e-01
7.72062778e-01 -7.40040064e-01 -6.05428576e-01 1.17490792e+00
4.08583909e-01 1.86004400e-01 4.07305151e-01 -1.11056888e+00
-5.77423871e-01 6.79643825e-02 -1.09445024e+00 8.71794939e-01
7.92185590e-02 1.44226313e-01 6.16141796e-01 -1.09910738e+00
4.18363363e-01 1.04162931e+00 5.08449018e-01 4.68112618e-01
-7.86874294e-01 -6.69867456e-01 5.06233931e-01 6.95508838e-01
-1.51619196e+00 -6.32358193e-01 8.44902456e-01 -4.93208945e-01
7.97892570e-01 2.56090611e-01 6.50526583e-01 5.41473925e-01
3.57289612e-02 1.08082187e+00 5.51734030e-01 -9.28003415e-02
-1.81176178e-02 -8.66873264e-02 5.61477244e-02 6.40905082e-01
4.38902676e-01 -2.66069114e-01 -5.83931804e-01 6.13504797e-02
9.16383505e-01 4.73498911e-01 -5.56019664e-01 -2.48119563e-01
-1.19712019e+00 6.25533164e-01 4.18641299e-01 3.67898196e-01
-7.57515430e-01 4.39920612e-02 4.76773739e-01 5.54745086e-02
4.90927845e-01 -9.47806537e-02 -3.84302497e-01 -5.28917126e-02
-1.12958694e+00 -8.06878135e-02 7.53853396e-02 9.06990290e-01
6.75041735e-01 3.08379591e-01 -4.48772788e-01 1.08131266e+00
2.27653190e-01 3.83125156e-01 5.64199269e-01 -8.70297313e-01
5.19249856e-01 6.00497186e-01 5.20635843e-02 -1.52576816e+00
-2.72441775e-01 -6.26167119e-01 -1.09326434e+00 -2.40072254e-02
2.45762810e-01 2.20576078e-02 -9.37041283e-01 1.54273498e+00
4.12452787e-01 6.52689159e-01 -9.24288630e-02 1.28708887e+00
9.55609381e-01 9.90568936e-01 -9.20715481e-02 -7.61562645e-01
1.27592194e+00 -9.85253930e-01 -8.46526027e-01 -1.87531039e-01
2.67618567e-01 -7.12625563e-01 6.78070366e-01 4.25489306e-01
-1.18152797e+00 -8.61754298e-01 -8.71798456e-01 1.23893872e-01
1.22086383e-01 2.33194739e-01 5.00942945e-01 3.63517731e-01
-8.40494752e-01 3.77433240e-01 -7.86659181e-01 -7.94630125e-02
5.92966080e-01 2.43056804e-01 -1.01152457e-01 -2.13622779e-01
-9.35621798e-01 4.51828957e-01 2.02527225e-01 3.46672684e-01
-9.05078650e-01 -6.91395164e-01 -7.76824415e-01 7.21592382e-02
5.49976110e-01 -3.52364779e-01 8.76485825e-01 -1.12014306e+00
-8.80345404e-01 4.74692672e-01 -4.12327111e-01 -4.83761221e-01
3.88672531e-01 -2.08741263e-01 -6.08668506e-01 3.88006032e-01
6.24065697e-02 5.77057958e-01 1.16685212e+00 -8.97143424e-01
-8.32773209e-01 -3.32527369e-01 -1.59773692e-01 3.64561528e-01
-5.88965893e-01 -4.73634154e-02 -9.66016471e-01 -8.92887473e-01
5.41084349e-01 -5.89171767e-01 -1.09758370e-01 2.54030704e-01
-1.47259990e-02 -2.14662611e-01 1.20855820e+00 -8.72759938e-01
1.54032862e+00 -2.40374160e+00 -1.24448061e-01 -3.75184119e-02
3.99212092e-01 5.68138540e-01 -1.20189160e-01 -2.55093127e-01
1.50847822e-01 -2.06284195e-01 -1.25130117e-01 -2.86535993e-02
-4.75549459e-01 7.45403320e-02 -4.46669757e-02 6.21080935e-01
1.89225346e-01 5.15012741e-01 -7.95538425e-01 -7.62165844e-01
3.73587072e-01 5.53836346e-01 -6.20755017e-01 2.70801894e-02
1.96498066e-01 2.14445397e-01 -4.48215574e-01 8.01833093e-01
9.62890267e-01 -3.87766510e-01 -1.45612955e-01 -7.41114795e-01
-1.90271139e-01 -9.56920460e-02 -1.43783212e+00 1.51185632e+00
-2.46135846e-01 5.73968709e-01 1.17168710e-01 -9.90254402e-01
7.42265046e-01 1.97290674e-01 7.35937536e-01 -8.09133410e-01
-5.45853116e-02 1.68849394e-01 9.18072313e-02 -6.09872580e-01
5.94386876e-01 3.58738363e-01 4.97787803e-01 -6.02086559e-02
-1.90110102e-01 5.66979468e-01 2.22657546e-01 2.11186230e-01
1.02599740e+00 -1.83806852e-01 2.46800870e-01 -6.77097142e-02
7.93556452e-01 -2.76399046e-01 1.04807520e+00 5.37458956e-01
-5.56670070e-01 7.68535495e-01 4.27151248e-02 -5.56247413e-01
-7.32021511e-01 -6.56995416e-01 -2.05939636e-01 8.34631085e-01
5.35462737e-01 -4.85386789e-01 -3.90215129e-01 -5.75482368e-01
1.49114253e-02 2.14366913e-01 -4.52946603e-01 -1.41398400e-01
-7.36896873e-01 -7.72596776e-01 6.83843717e-02 5.18785238e-01
7.54930139e-01 -9.50105429e-01 -7.87102461e-01 4.98130172e-01
-3.26129675e-01 -1.02765906e+00 -9.12986159e-01 -2.03959867e-01
-9.96418595e-01 -1.01891470e+00 -1.01485372e+00 -8.36792529e-01
8.57895017e-01 8.97795260e-01 8.26461554e-01 3.01840127e-01
-2.86718547e-01 2.95539945e-01 -4.18441653e-01 2.88022496e-02
5.34722880e-02 -2.80662149e-01 1.41680270e-01 4.60721403e-01
2.20531687e-01 -3.67363632e-01 -1.09950364e+00 4.73534375e-01
-9.35141861e-01 5.91139980e-02 6.67122126e-01 8.03463042e-01
7.66195416e-01 1.51884601e-01 5.19920528e-01 -4.46001858e-01
1.17567204e-01 -4.93174940e-01 -4.68297660e-01 8.64074528e-02
-4.00987685e-01 -2.61708915e-01 4.47180331e-01 -6.00503325e-01
-9.55387950e-01 7.69957528e-02 1.24304332e-01 -9.64811862e-01
1.17698729e-01 2.29960844e-01 -1.47599623e-01 -6.44058734e-02
1.44111022e-01 5.46608746e-01 -3.97223420e-02 -5.02831757e-01
-1.76652089e-01 7.00230181e-01 4.42778647e-01 -1.70345053e-01
4.83727366e-01 5.11904657e-01 -1.24635570e-01 -7.89247096e-01
-5.95927477e-01 -6.83839440e-01 -3.07589561e-01 -3.57448816e-01
7.19387293e-01 -1.32424045e+00 -4.58979547e-01 6.38548315e-01
-1.11966109e+00 1.10090405e-01 8.46470445e-02 5.74645638e-01
3.73006240e-02 6.67182088e-01 -6.52074099e-01 -7.61811078e-01
-4.97798175e-01 -1.17251480e+00 8.46060634e-01 3.80392551e-01
2.17130065e-01 -5.25469899e-01 -4.34824109e-01 1.19427755e-01
4.68034297e-01 2.60786340e-02 3.03311855e-01 -1.87135547e-01
-7.72656441e-01 -5.81704527e-02 -3.86495173e-01 4.56969231e-01
3.50641608e-01 -1.30443290e-01 -6.50267184e-01 -4.37690049e-01
9.82375443e-02 8.47828537e-02 1.06757390e+00 6.32807076e-01
1.32084799e+00 -3.80401284e-01 -3.58315200e-01 5.07219195e-01
1.18400896e+00 4.12548035e-01 5.96271276e-01 2.42198393e-01
1.00758827e+00 2.66122222e-01 9.30468976e-01 7.32501507e-01
2.94815242e-01 8.79825771e-01 4.83268946e-01 -2.43058596e-02
-4.22467619e-01 1.22734219e-01 5.57787240e-01 8.07605505e-01
-9.45452303e-02 -1.07217737e-01 -4.69592512e-01 7.97894061e-01
-1.80964363e+00 -1.15329742e+00 -4.47262883e-01 2.27427030e+00
6.19830906e-01 -6.99893534e-02 7.69018680e-02 5.67417629e-02
1.05206728e+00 3.33395422e-01 -7.21686423e-01 1.18138082e-01
-3.18323612e-01 -2.65786767e-01 5.26310205e-01 4.89014797e-02
-1.14602065e+00 5.58969438e-01 5.01499844e+00 9.28042769e-01
-1.16683805e+00 4.13769275e-01 8.47648144e-01 -4.32028174e-01
1.08045496e-01 -2.73698777e-01 -8.83966267e-01 8.42424512e-01
4.22093272e-01 8.20655748e-02 1.92097500e-01 7.91166067e-01
4.16296571e-01 -3.19126308e-01 -7.18403935e-01 1.45620930e+00
3.94946307e-01 -1.24108267e+00 8.50033686e-02 -1.27623379e-01
8.02851856e-01 -1.54196084e-01 2.41533853e-03 1.85336858e-01
-6.47409201e-01 -4.98917043e-01 7.05612361e-01 4.59887803e-01
5.17526865e-01 -7.12658465e-01 6.32322311e-01 1.20767646e-01
-1.49992883e+00 -8.18814412e-02 -6.27236903e-01 -3.26043554e-02
1.29843220e-01 8.18764925e-01 -4.60442185e-01 2.71505445e-01
1.00167799e+00 1.06418240e+00 -6.00585163e-01 1.37464106e+00
2.14399040e-01 5.73405623e-01 -3.22311282e-01 2.49545619e-01
1.64311856e-01 -7.03324527e-02 7.39513874e-01 9.65421557e-01
5.34487784e-01 4.02681470e-01 2.94050425e-01 5.12453616e-01
5.41183762e-02 1.75083533e-01 -2.67108321e-01 4.15452987e-01
3.86702746e-01 1.06614470e+00 -5.37060380e-01 -5.84058046e-01
-7.52931416e-01 1.19047797e+00 -2.74658985e-02 2.08404645e-01
-1.02377856e+00 -1.54121161e-01 7.17075229e-01 3.32721770e-01
7.62151301e-01 -2.03094989e-01 1.50301114e-01 -1.40799570e+00
3.67264926e-01 -7.46606171e-01 5.89022577e-01 -5.91166496e-01
-1.18648458e+00 5.17011464e-01 -3.60356510e-01 -1.71337140e+00
2.86132425e-01 -1.25323489e-01 -4.73408699e-01 5.24458945e-01
-1.55539107e+00 -6.57088935e-01 -6.05947971e-01 8.12932193e-01
9.12355840e-01 3.56684215e-02 1.14485793e-01 8.77715111e-01
-8.35702479e-01 6.21000707e-01 1.37688676e-02 1.92367077e-01
6.90200150e-01 -6.00663483e-01 -4.67148013e-02 1.09651113e+00
1.48682863e-01 4.27282333e-01 4.78456914e-01 -8.34292948e-01
-1.52480257e+00 -1.30721402e+00 4.64070648e-01 1.90003112e-01
1.98431566e-01 -1.36638538e-03 -1.22114575e+00 3.48023474e-01
-2.35422924e-02 6.59495831e-01 2.47980312e-01 -3.04977119e-01
-2.12408993e-02 -4.30448741e-01 -1.03927910e+00 3.25702518e-01
1.17298305e+00 -1.42002031e-01 -3.78691494e-01 2.41183043e-01
4.04072225e-01 -5.22827625e-01 -6.14564836e-01 6.16864979e-01
4.27876413e-01 -9.55785692e-01 1.01687896e+00 -2.00315997e-01
6.19138032e-02 -9.20460105e-01 -2.00962409e-01 -8.08408022e-01
-5.65187454e-01 -3.86834413e-01 -6.75362587e-01 1.21596670e+00
-4.18082923e-02 -3.09445202e-01 6.48260295e-01 5.13090014e-01
-2.04818800e-01 -6.47126317e-01 -9.28514659e-01 -7.32052028e-01
-7.12924421e-01 -3.92723888e-01 2.97755599e-01 8.40454221e-01
-3.13997835e-01 9.09109786e-02 -9.01890337e-01 4.72080588e-01
6.58227146e-01 2.18370765e-01 6.01741731e-01 -9.72008765e-01
-2.55019903e-01 -1.98840320e-01 -7.88469374e-01 -1.31437695e+00
-2.01501831e-01 -6.02520943e-01 -5.90952998e-03 -1.31540692e+00
5.28428972e-01 -3.41507018e-01 -7.30765700e-01 4.37348723e-01
-5.14721692e-01 6.67866826e-01 4.10119593e-01 4.47942436e-01
-1.01666760e+00 5.89110494e-01 1.29890645e+00 -1.73210621e-01
-3.79407495e-01 -5.06502613e-02 -3.02575856e-01 7.21442759e-01
6.17143869e-01 -4.19523895e-01 -3.52013260e-01 -5.49489260e-01
-2.57669449e-01 6.28072321e-02 4.88980800e-01 -1.28547943e+00
4.42588270e-01 4.51905392e-02 7.42260456e-01 -9.91930485e-01
4.47730511e-01 -8.35212767e-01 1.91839606e-01 5.83916545e-01
5.62757626e-02 1.38307527e-01 2.42986307e-01 7.91521251e-01
-4.07367617e-01 -1.25198245e-01 8.40138078e-01 -2.08388288e-02
-1.25485575e+00 5.77805221e-01 -3.61595184e-01 -8.97371620e-02
1.27061117e+00 -4.08118427e-01 -2.24359427e-02 -2.21376553e-01
-6.18214369e-01 3.23280007e-01 3.21561188e-01 5.02921522e-01
1.04718149e+00 -1.38875186e+00 -8.30630124e-01 2.32799470e-01
-8.03956389e-03 -3.90932746e-02 8.76795590e-01 1.13560784e+00
-3.14281404e-01 1.08647414e-01 -1.89426512e-01 -8.52147996e-01
-1.66915739e+00 7.34609962e-01 1.65584132e-01 1.26874149e-01
-7.84319341e-01 8.58382821e-01 4.71330732e-01 5.38789392e-01
2.14785412e-01 -3.15207690e-01 -3.86924654e-01 2.16347218e-01
9.47743714e-01 3.67758721e-01 -6.14422299e-02 -8.05231869e-01
-5.77562630e-01 6.62336111e-01 -2.38233238e-01 5.03175616e-01
1.20096219e+00 -5.40091157e-01 3.44875306e-02 3.44809383e-01
1.15102756e+00 1.45965172e-02 -1.55402422e+00 -5.52040040e-01
-3.48061740e-01 -1.02104127e+00 4.68527317e-01 -4.20395851e-01
-1.71545672e+00 6.19724572e-01 9.12558675e-01 -1.98853895e-01
1.38134575e+00 -2.29240760e-01 9.70151603e-01 1.38260853e-02
3.16973388e-01 -9.58693862e-01 1.60284579e-01 2.72850305e-01
6.95372581e-01 -1.41071033e+00 2.37434000e-01 -4.50552106e-01
-6.77669346e-01 1.05693007e+00 7.99636543e-01 -3.19418386e-02
4.08191293e-01 -4.93573062e-02 -1.95393711e-01 -4.86636050e-02
-6.04347706e-01 -2.77037948e-01 5.03294170e-01 3.09458703e-01
1.78643361e-01 -2.73860902e-01 -3.39186370e-01 3.90891939e-01
6.63662612e-01 -8.69942531e-02 4.19279486e-01 8.72732997e-01
-5.76129913e-01 -6.54293895e-01 -5.95225275e-01 6.81458473e-01
-5.66817820e-01 -2.10485771e-01 2.91251302e-01 4.01316524e-01
3.32353741e-01 9.92280722e-01 2.97803432e-01 -2.75734782e-01
1.75761372e-01 -2.34941587e-01 3.85257572e-01 -3.83060098e-01
-2.31097311e-01 4.36517149e-01 -1.54997587e-01 -5.81155300e-01
-6.62757218e-01 -9.72845078e-01 -1.22020543e+00 -1.96092755e-01
-4.58475232e-01 -2.01207157e-02 2.71927893e-01 6.72316909e-01
6.75842881e-01 6.78878009e-01 6.99453771e-01 -8.50221038e-01
-2.19487220e-01 -8.28195632e-01 -6.58205032e-01 4.45052385e-01
4.07813817e-01 -7.62211919e-01 -2.62808591e-01 1.24627754e-01] | [9.119589805603027, -0.2580391466617584] |
8e4bd2dd-05ab-4d0e-af4e-fa939351be15 | explainable-image-classification-with | 2004.07511 | null | https://arxiv.org/abs/2004.07511v1 | https://arxiv.org/pdf/2004.07511v1.pdf | Explainable Image Classification with Evidence Counterfactual | The complexity of state-of-the-art modeling techniques for image classification impedes the ability to explain model predictions in an interpretable way. Existing explanation methods generally create importance rankings in terms of pixels or pixel groups. However, the resulting explanations lack an optimal size, do not consider feature dependence and are only related to one class. Counterfactual explanation methods are considered promising to explain complex model decisions, since they are associated with a high degree of human interpretability. In this paper, SEDC is introduced as a model-agnostic instance-level explanation method for image classification to obtain visual counterfactual explanations. For a given image, SEDC searches a small set of segments that, in case of removal, alters the classification. As image classification tasks are typically multiclass problems, SEDC-T is proposed as an alternative method that allows specifying a target counterfactual class. We compare SEDC(-T) with popular feature importance methods such as LRP, LIME and SHAP, and we describe how the mentioned importance ranking issues are addressed. Moreover, concrete examples and experiments illustrate the potential of our approach (1) to obtain trust and insight, and (2) to obtain input for model improvement by explaining misclassifications. | ['Tom Vermeire', 'David Martens'] | 2020-04-16 | null | null | null | null | ['counterfactual-explanation'] | ['miscellaneous'] | [ 6.64756417e-01 5.70662200e-01 -4.75370079e-01 -5.66308856e-01
-2.21863404e-01 -3.26170057e-01 7.58326411e-01 3.54889929e-01
2.78444141e-02 8.55833828e-01 5.51786534e-02 -6.94844425e-01
-5.23503423e-01 -6.58587337e-01 -8.45877767e-01 -5.28335631e-01
8.24755207e-02 3.35706741e-01 -6.50856122e-02 9.02626514e-02
6.81503832e-01 5.93735278e-01 -1.94829905e+00 7.04727948e-01
1.13616621e+00 7.75989950e-01 1.17933057e-01 3.50879163e-01
-3.10658991e-01 5.55022836e-01 -5.18037260e-01 -6.29443645e-01
6.19240329e-02 -5.97795486e-01 -8.67951870e-01 5.44813573e-01
3.03783774e-01 3.81893590e-02 4.59945351e-01 9.57543135e-01
-1.34928137e-01 -1.68152615e-01 9.89344239e-01 -1.78289223e+00
-8.96354198e-01 6.39759958e-01 -4.92403448e-01 2.96450779e-02
1.28420427e-01 2.66365916e-01 9.15778637e-01 -8.26952279e-01
4.57416624e-01 1.37430358e+00 4.56007212e-01 6.64448857e-01
-1.60904276e+00 -3.73589039e-01 7.22992778e-01 6.06680632e-01
-8.18134427e-01 4.31130901e-02 7.32051671e-01 -4.46548432e-01
7.98366129e-01 9.82487798e-01 8.37948561e-01 9.21194732e-01
2.21932799e-01 7.83742130e-01 1.56199992e+00 -6.70289576e-01
4.71752286e-01 4.50192094e-01 1.96150512e-01 4.77053881e-01
6.15355492e-01 2.22632930e-01 -4.42641169e-01 -2.11426735e-01
5.02489507e-01 6.96085468e-02 -3.68727446e-01 -5.60870051e-01
-1.08803964e+00 1.06380606e+00 7.63116300e-01 5.41433096e-02
-4.62238282e-01 3.10963303e-01 8.18592981e-02 1.47298619e-01
6.96900845e-01 8.94330263e-01 -5.20610213e-01 4.37933087e-01
-6.36138618e-01 3.96805286e-01 4.92477566e-01 5.43122590e-01
5.67776620e-01 7.13113993e-02 -3.35472912e-01 5.51156998e-01
3.64688158e-01 2.27303669e-01 4.20838982e-01 -9.21819925e-01
-7.08035566e-03 8.42424691e-01 2.67667696e-02 -1.02386773e+00
-2.53538102e-01 -5.96365154e-01 -6.43233776e-01 7.92820036e-01
1.28195509e-01 4.27584022e-01 -1.17321181e+00 1.52639413e+00
1.04300559e-01 1.30538389e-01 1.99327543e-02 8.95259440e-01
4.00399685e-01 3.96543533e-01 2.77530789e-01 -5.19413054e-01
1.40222538e+00 -9.58282590e-01 -6.12660527e-01 -4.52128917e-01
4.20484006e-01 -5.35063922e-01 1.17597640e+00 4.52245086e-01
-8.74438941e-01 -4.90596622e-01 -7.98444867e-01 3.37228596e-01
-6.14018917e-01 -7.58425146e-02 1.00739050e+00 6.28154635e-01
-7.78113008e-01 8.45084190e-01 -5.50711751e-01 -2.89961547e-01
5.28897107e-01 3.34907204e-01 -2.31084272e-01 7.25179538e-02
-8.22304606e-01 1.02398121e+00 4.42920387e-01 4.89447005e-02
-5.82194686e-01 -6.29369199e-01 -7.30098307e-01 2.86772639e-01
3.94379556e-01 -8.41163695e-01 9.67566431e-01 -1.58082116e+00
-9.66340184e-01 7.15107560e-01 -1.92322269e-01 -8.05421889e-01
6.70988262e-01 1.08463287e-01 -2.43381962e-01 -2.96957456e-02
1.74318567e-01 1.01668453e+00 1.07513249e+00 -1.82205427e+00
-6.31751895e-01 -4.00844872e-01 2.61528105e-01 1.93295449e-01
-2.35409308e-02 -3.81963819e-01 -6.42993003e-02 -7.76785553e-01
3.37796390e-01 -8.83729100e-01 -6.03584051e-01 3.42139989e-01
-5.82849562e-01 5.07893898e-02 6.52073145e-01 -2.69547850e-01
1.01894271e+00 -1.83335364e+00 -1.16099380e-01 3.29035431e-01
2.59586573e-01 1.02101132e-01 -6.44120201e-02 1.10147290e-01
-5.00781775e-01 7.12156236e-01 -5.35527229e-01 -1.98647156e-01
-1.07065298e-01 3.59253287e-01 -3.40780139e-01 1.69279695e-01
3.64234895e-01 7.42880583e-01 -8.01092625e-01 -4.56655562e-01
5.14340818e-01 2.87461936e-01 -6.32674456e-01 -9.67446715e-02
-3.52306962e-01 2.41427690e-01 -2.71296978e-01 3.88759971e-01
6.65649652e-01 -3.33798468e-01 2.15203032e-01 -1.39815480e-01
-1.77371383e-01 5.58909886e-02 -9.60877180e-01 9.46140826e-01
-4.04665768e-01 7.48099566e-01 -7.01839387e-01 -9.86843288e-01
7.49979079e-01 1.03176735e-01 8.34571868e-02 -3.51713926e-01
-8.55228156e-02 2.62347519e-01 2.71181405e-01 -4.00778860e-01
2.73640752e-01 -2.97123224e-01 2.41579130e-01 4.07431722e-01
-4.09423709e-01 -2.30583161e-01 4.14740667e-02 -5.05344979e-02
6.46097541e-01 3.00655421e-02 9.20199156e-01 -3.19065481e-01
5.42853773e-01 2.86390334e-01 4.99666393e-01 9.64984238e-01
-5.81733249e-02 8.28548253e-01 6.62414968e-01 -6.93304420e-01
-9.23597336e-01 -7.80133545e-01 -2.62606144e-01 4.51500326e-01
2.89034307e-01 7.46465176e-02 -8.09022188e-01 -9.16971564e-01
1.62184879e-01 1.36528206e+00 -1.03394330e+00 -3.30735505e-01
-1.16634220e-01 -9.18617964e-01 -1.81406215e-01 4.49088693e-01
3.23213577e-01 -1.27307010e+00 -9.69774008e-01 2.53491383e-02
-1.24387853e-01 -4.87845868e-01 4.51838896e-02 3.06153089e-01
-1.11848748e+00 -1.24047220e+00 -4.88482416e-01 -1.18054122e-01
1.18336701e+00 5.20130754e-01 1.13571227e+00 6.05293632e-01
-2.94817686e-01 2.27792546e-01 -4.90074873e-01 -8.48821163e-01
-4.03480858e-01 -3.19163591e-01 -1.63794339e-01 1.83528200e-01
2.97596782e-01 -3.18824887e-01 -6.49410009e-01 3.01938176e-01
-9.22548294e-01 5.37134767e-01 7.83596456e-01 1.02136016e+00
6.46017373e-01 -2.79005338e-02 5.11843383e-01 -1.21868932e+00
7.00848401e-01 -2.72128761e-01 -5.41107357e-01 5.03840566e-01
-1.36395335e+00 3.62580389e-01 5.42193174e-01 -4.92100745e-01
-1.21523666e+00 1.72795206e-01 3.41232479e-01 -3.02758008e-01
-4.17752266e-01 4.93122131e-01 -1.57032907e-02 -2.88491603e-02
7.81929851e-01 7.87985176e-02 -2.09304076e-02 -2.80590713e-01
4.29469794e-01 3.35952401e-01 2.46524498e-01 -1.37378082e-01
7.38416433e-01 4.91886288e-01 1.57815218e-01 -3.48712325e-01
-7.97603190e-01 -5.08384742e-02 -6.73496425e-01 -3.85917276e-01
7.42241621e-01 -3.85033697e-01 -6.99740767e-01 -1.25069499e-01
-1.27814424e+00 1.79600809e-02 -5.42031884e-01 5.05147398e-01
-6.53815150e-01 1.12800129e-01 1.52504385e-01 -1.17953050e+00
1.02124222e-01 -1.31074846e+00 9.42482531e-01 1.65065348e-01
-4.03709859e-01 -9.14838672e-01 -4.67199594e-01 5.26631713e-01
2.32231349e-01 4.57168728e-01 1.28684890e+00 -5.93947709e-01
-7.48958111e-01 -1.63744673e-01 -3.28993052e-01 6.59419447e-02
8.13475922e-02 8.12002867e-02 -1.11224866e+00 1.25312209e-01
-1.08327627e-01 1.60742193e-01 9.08386230e-01 7.36489356e-01
1.67856550e+00 -6.40007615e-01 -5.59126675e-01 1.67012453e-01
1.46637678e+00 3.36107820e-01 6.95600212e-01 5.84051073e-01
2.97360271e-01 9.51091051e-01 1.02395284e+00 3.58580053e-01
5.71130626e-02 7.45058596e-01 1.13038623e+00 -5.82059443e-01
1.01876925e-04 -8.37038532e-02 -9.10417885e-02 -2.76146848e-02
-2.88885355e-01 -3.17224711e-01 -6.30254388e-01 6.38588905e-01
-2.19090271e+00 -9.62782919e-01 -7.45696187e-01 2.30696392e+00
2.26396680e-01 2.75319129e-01 -1.27082407e-01 2.76089191e-01
7.79799461e-01 -4.82647009e-02 -6.70916259e-01 -7.31403768e-01
-4.60429899e-02 -2.91329712e-01 5.20502508e-01 5.29336512e-01
-8.38688135e-01 5.36919355e-01 6.46820688e+00 5.75151682e-01
-8.44483435e-01 -1.53350532e-01 1.11123133e+00 6.27097040e-02
-8.12619150e-01 4.58661944e-01 -2.96288967e-01 3.84908766e-01
5.97355843e-01 -1.16421491e-01 1.66827530e-01 9.76474404e-01
4.57244277e-01 -1.78470001e-01 -1.14965844e+00 7.46042550e-01
1.36032458e-02 -1.44905448e+00 6.27258003e-01 3.13567579e-01
7.24483967e-01 -7.80440629e-01 1.76860467e-01 -1.32350743e-01
6.30170554e-02 -1.21886277e+00 9.50191557e-01 6.68871999e-01
3.33337486e-01 -7.91302443e-01 9.69937503e-01 2.55680293e-01
-6.72346711e-01 -3.21872413e-01 -5.04207015e-01 -3.57228309e-01
3.31790149e-02 5.77002823e-01 -9.04337287e-01 4.46201622e-01
5.71603000e-01 4.00822490e-01 -8.61891031e-01 1.02407169e+00
-5.54480016e-01 4.76944506e-01 1.75835386e-01 -2.00415596e-01
1.24177627e-01 2.55491976e-02 5.54943264e-01 8.39970291e-01
2.98390836e-01 1.50067508e-01 -2.02167705e-01 1.17800486e+00
3.31559271e-01 2.74471808e-02 -6.67608559e-01 3.45709145e-01
3.46741825e-01 9.70620453e-01 -1.15154862e+00 -4.64637011e-01
-2.18168408e-01 1.11521578e+00 3.65506820e-02 2.42173702e-01
-9.92435992e-01 1.61991313e-01 6.56634629e-01 2.32715562e-01
-1.79869495e-02 5.22548079e-01 -8.38434279e-01 -9.90942717e-01
-3.44188623e-02 -9.86847758e-01 3.44105244e-01 -1.11133134e+00
-1.13142800e+00 6.46186650e-01 2.95973748e-01 -1.49816334e+00
-2.52128601e-01 -7.50689447e-01 -6.09077215e-01 6.54635847e-01
-1.61811638e+00 -1.15754426e+00 -3.64744484e-01 2.27579728e-01
9.32140946e-01 7.55132213e-02 7.90776730e-01 -3.96883845e-01
-3.62745255e-01 1.80096045e-01 -4.98396717e-02 -6.60288811e-01
2.58925587e-01 -1.58581376e+00 2.66372085e-01 8.58366609e-01
2.62163490e-01 5.02629101e-01 1.21514738e+00 -5.69347799e-01
-5.66597879e-01 -9.52579796e-01 1.05784357e+00 -2.98352301e-01
1.97226286e-01 -7.38854110e-02 -9.58720684e-01 5.23076713e-01
1.25138655e-01 -7.85121396e-02 7.11818755e-01 1.42097294e-01
-1.69153705e-01 -5.63593060e-02 -1.26250780e+00 8.95474255e-01
9.80241001e-01 -2.42892541e-02 -6.63081825e-01 3.30975026e-01
5.86164355e-01 2.02560157e-01 -3.76195788e-01 4.18597907e-01
5.93382955e-01 -1.34546721e+00 1.06392109e+00 -8.09839189e-01
6.16776586e-01 -3.79063189e-01 8.21466744e-02 -1.46832561e+00
-4.92599338e-01 -6.90948814e-02 2.31850192e-01 9.90537047e-01
7.66442895e-01 -6.84155464e-01 7.69896388e-01 9.35497105e-01
-9.43224784e-03 -7.69747734e-01 -6.62083089e-01 -8.01567554e-01
-3.06685001e-01 -5.58822572e-01 9.25838590e-01 8.94289434e-01
-1.14492020e-02 1.49826899e-01 -4.90089595e-01 2.35869512e-01
6.79456592e-01 4.53216314e-01 7.24626780e-01 -1.55320919e+00
-2.68644154e-01 -5.41194260e-01 -4.89128321e-01 -3.14412743e-01
7.33226389e-02 -5.53452194e-01 3.20965424e-02 -1.62323928e+00
4.26253736e-01 -2.32339621e-01 -2.49292478e-01 6.37418866e-01
-4.94069070e-01 2.14372769e-01 4.68754262e-01 3.33011329e-01
-8.56878459e-02 3.51054579e-01 1.04063809e+00 -2.28240430e-01
-3.14149521e-02 1.75329566e-01 -1.03874385e+00 9.96108949e-01
8.05489480e-01 -7.39400566e-01 -6.38400733e-01 -5.17784096e-02
4.45776023e-02 -3.21513154e-02 7.48639703e-01 -5.95159113e-01
-3.47833335e-01 -6.52383387e-01 5.07825315e-01 -2.30292588e-01
2.81629711e-01 -1.13110304e+00 4.86925393e-01 9.08934355e-01
-6.74675465e-01 -1.35479227e-01 3.23680043e-02 6.55987620e-01
-2.38055080e-01 -5.65943658e-01 7.17375040e-01 -2.28364155e-01
-8.32726002e-01 -1.15423284e-01 -4.46564436e-01 -7.93989956e-01
1.11749518e+00 -6.20889425e-01 -3.51344734e-01 -5.13258994e-01
-7.86839962e-01 8.53051767e-02 4.49325979e-01 3.65157604e-01
7.25423634e-01 -1.20584726e+00 -6.08262777e-01 -2.63030455e-03
5.09790480e-01 -3.88647646e-01 2.09849462e-01 6.89933717e-01
-2.97266632e-01 4.80830520e-01 -3.48042995e-01 -5.46836555e-01
-1.65480876e+00 7.61151373e-01 2.70833939e-01 -3.94133419e-01
-4.33457941e-01 5.21614850e-01 7.74788678e-01 -3.10478240e-01
-2.29303405e-01 -4.87997174e-01 -4.72219706e-01 -1.16613731e-01
3.97230446e-01 2.36791000e-01 -2.01244161e-01 -3.49960238e-01
-4.26198065e-01 2.92560518e-01 1.12339646e-01 -3.98287289e-02
1.36829782e+00 -2.33668506e-01 7.45087415e-02 3.91122460e-01
6.24761224e-01 -4.20774609e-01 -1.36803854e+00 3.95735294e-01
1.39792666e-01 -8.91055167e-01 -7.92634040e-02 -1.16892970e+00
-9.03651893e-01 8.93173695e-01 7.77818441e-01 5.16191244e-01
1.44652200e+00 -1.95304572e-03 -1.98406845e-01 2.05696244e-02
1.05805360e-01 -7.13717341e-01 -5.06521128e-02 -2.76460916e-01
1.36643171e+00 -1.54567456e+00 1.87163427e-01 -7.67004251e-01
-8.16322148e-01 1.21655035e+00 6.59158647e-01 2.17149213e-01
2.50175267e-01 -1.76610261e-01 -2.56220438e-02 -3.71404737e-01
-9.32184100e-01 -2.92478651e-01 5.93190968e-01 7.61967123e-01
4.44414169e-01 2.78548181e-01 -7.05707192e-01 3.51191610e-01
-6.12833872e-02 -2.17539713e-01 6.46305025e-01 5.37601173e-01
-4.59683001e-01 -9.14488554e-01 -6.33971632e-01 7.00873911e-01
-2.68733203e-01 -1.21779360e-01 -6.27404332e-01 1.15348864e+00
3.44059110e-01 9.74614263e-01 1.15660066e-02 -1.04768567e-01
3.79936010e-01 -9.64132696e-02 1.20678268e-01 -5.89664757e-01
-3.97495091e-01 -4.92483042e-02 3.15828621e-02 -4.37261015e-01
-5.66144705e-01 -7.74359584e-01 -1.06069112e+00 -6.34526983e-02
-6.11186624e-01 1.49289414e-01 8.20432484e-01 1.09000969e+00
3.20767105e-01 6.50169253e-01 5.26020944e-01 -6.91400647e-01
-3.03820431e-01 -7.08338857e-01 -5.12350678e-01 6.15613878e-01
2.69712627e-01 -8.04604709e-01 -5.77084422e-01 2.94309855e-01] | [8.827472686767578, 5.549739360809326] |
22852020-2bd0-4152-92ee-c7c3a4c82462 | learning-global-aware-kernel-for-image | 2305.11676 | null | https://arxiv.org/abs/2305.11676v1 | https://arxiv.org/pdf/2305.11676v1.pdf | Learning Global-aware Kernel for Image Harmonization | Image harmonization aims to solve the visual inconsistency problem in composited images by adaptively adjusting the foreground pixels with the background as references. Existing methods employ local color transformation or region matching between foreground and background, which neglects powerful proximity prior and independently distinguishes fore-/back-ground as a whole part for harmonization. As a result, they still show a limited performance across varied foreground objects and scenes. To address this issue, we propose a novel Global-aware Kernel Network (GKNet) to harmonize local regions with comprehensive consideration of long-distance background references. Specifically, GKNet includes two parts, \ie, harmony kernel prediction and harmony kernel modulation branches. The former includes a Long-distance Reference Extractor (LRE) to obtain long-distance context and Kernel Prediction Blocks (KPB) to predict multi-level harmony kernels by fusing global information with local features. To achieve this goal, a novel Selective Correlation Fusion (SCF) module is proposed to better select relevant long-distance background references for local harmonization. The latter employs the predicted kernels to harmonize foreground regions with both local and global awareness. Abundant experiments demonstrate the superiority of our method for image harmonization over state-of-the-art methods, \eg, achieving 39.53dB PSNR that surpasses the best counterpart by +0.78dB $\uparrow$; decreasing fMSE/MSE by 11.5\%$\downarrow$/6.7\%$\downarrow$ compared with the SoTA method. Code will be available at \href{https://github.com/XintianShen/GKNet}{here}. | ['Yong liu', 'Chengjie Wang', 'Yabiao Wang', 'Yue Han', 'Shipeng Bai', 'Jun Chen', 'Jiangning Zhang', 'Xintian Shen'] | 2023-05-19 | null | null | null | null | ['image-harmonization'] | ['computer-vision'] | [ 2.35123083e-01 -5.06887853e-01 -1.97749972e-01 -3.37414779e-02
-7.85586655e-01 -2.54133195e-01 3.38699639e-01 -2.55806446e-01
-2.05729634e-01 6.36469603e-01 8.22577067e-03 5.00742868e-02
-1.30554974e-01 -8.19014490e-01 -5.82039833e-01 -1.19293356e+00
3.35008293e-01 -2.73139060e-01 6.47818506e-01 -3.12098116e-01
2.76117444e-01 4.08198535e-01 -1.56352961e+00 2.07057461e-01
1.20957053e+00 1.06872928e+00 3.61910492e-01 4.22574729e-01
-1.68886349e-01 6.97672904e-01 -5.09801447e-01 -3.70700389e-01
3.23264062e-01 -7.80378461e-01 -2.19686300e-01 -5.79223298e-02
4.41542327e-01 3.76699562e-03 -3.32455456e-01 1.36891723e+00
7.82394826e-01 1.46878526e-01 2.54186243e-01 -1.02494299e+00
-6.90589964e-01 3.65390688e-01 -1.15587890e+00 5.98392785e-01
-1.47588521e-01 3.40752721e-01 6.75045133e-01 -8.48786891e-01
4.39523011e-01 7.97712326e-01 5.12477398e-01 1.63946539e-01
-1.23832476e+00 -1.08689249e+00 6.57100929e-04 4.76820350e-01
-1.79385686e+00 -4.44656819e-01 1.08158374e+00 -1.22648805e-01
4.52339053e-01 6.26710951e-01 5.97772121e-01 4.85700876e-01
3.22941005e-01 5.65109670e-01 1.26379788e+00 -3.00904691e-01
-1.80515870e-01 2.63932776e-02 -6.25443310e-02 5.62146783e-01
1.26470596e-01 2.80726068e-02 -5.90334594e-01 1.40570685e-01
8.60525668e-01 -2.82092720e-01 -7.49420881e-01 -1.27801567e-01
-1.22624934e+00 4.77444857e-01 6.34819329e-01 5.37842691e-01
-2.80846983e-01 2.98540965e-02 1.28005669e-01 -7.53794611e-02
2.81404942e-01 1.55644536e-01 -9.98932794e-02 2.11894512e-01
-1.03014374e+00 1.76912174e-02 2.95768589e-01 8.24255168e-01
9.52042282e-01 3.56674314e-01 -5.47773540e-01 1.16221821e+00
1.24556501e-03 6.52517796e-01 3.04038256e-01 -8.99691701e-01
3.72477472e-01 5.62956333e-01 -2.20262771e-03 -1.39054108e+00
-1.81185067e-01 -8.61184299e-01 -1.20074499e+00 1.77694350e-01
2.19991311e-01 1.05613321e-01 -4.80160207e-01 1.65125346e+00
4.88372833e-01 4.65540826e-01 -6.61800951e-02 1.14577544e+00
8.57593596e-01 9.57415044e-01 3.90485413e-02 -6.02126956e-01
1.34177852e+00 -1.02778709e+00 -8.23690772e-01 -4.44301841e-04
-4.59458865e-03 -1.28586781e+00 1.03081620e+00 3.70986283e-01
-1.31258631e+00 -1.05272949e+00 -1.16314268e+00 7.67576769e-02
-1.63180083e-01 2.78577954e-01 2.58518487e-01 6.00506306e-01
-1.10525525e+00 1.38876140e-01 -4.54492688e-01 -9.06657428e-02
2.00242952e-01 2.59694099e-01 1.50424123e-01 1.81061402e-02
-1.23881257e+00 6.55730426e-01 4.91000086e-01 2.98803985e-01
-4.39225882e-01 -7.43228614e-01 -6.30882442e-01 -1.42703116e-01
5.42605996e-01 -5.67234576e-01 6.63590312e-01 -1.02994442e+00
-1.50190568e+00 7.56263077e-01 -1.03893287e-01 -1.76298231e-01
2.82239258e-01 7.64247254e-02 -8.14100564e-01 1.85605913e-01
3.96746118e-03 4.76024508e-01 8.76050174e-01 -1.47163761e+00
-8.67714763e-01 -2.11140007e-01 -2.79106259e-01 4.39894348e-01
-2.73405373e-01 8.73881429e-02 -1.11014080e+00 -1.15896142e+00
1.67223617e-01 -6.94390059e-01 1.09506980e-01 -2.21970230e-01
-3.82632226e-01 1.96294546e-01 9.56267059e-01 -1.07058358e+00
1.71153796e+00 -2.10281444e+00 4.54579331e-02 2.94142872e-01
1.91008642e-01 5.58424294e-01 -1.32687703e-01 -2.81965528e-02
-1.14737205e-01 -2.97108352e-01 -2.36005083e-01 2.82573372e-01
-1.07636690e-01 -1.35437936e-01 -2.68932600e-02 5.30033171e-01
-2.52309497e-02 9.56120312e-01 -6.39788985e-01 -6.60743415e-01
5.05666912e-01 6.89990282e-01 -1.40870422e-01 3.47833596e-02
2.84326434e-01 4.90747154e-01 -3.61410290e-01 8.06903780e-01
1.07322490e+00 -8.58510286e-02 -1.95447598e-02 -8.01751792e-01
-3.68897766e-01 -4.31858271e-01 -1.54981005e+00 1.50054896e+00
-3.18225801e-01 4.17999804e-01 3.62324983e-01 -7.91255891e-01
1.15904176e+00 2.42022965e-02 5.26238143e-01 -1.08259165e+00
1.20674111e-01 3.28709960e-01 -6.50216490e-02 -1.61511466e-01
3.95787418e-01 7.47047290e-02 -4.93695475e-02 -1.10647537e-01
-2.11025909e-01 -4.77731861e-02 7.40391389e-02 -1.58806160e-01
5.54744899e-01 2.10151300e-01 4.00165766e-01 -4.76704299e-01
1.12795901e+00 -1.09952293e-01 8.63651812e-01 4.98921663e-01
-2.76340693e-01 6.36703432e-01 1.30144358e-01 -2.26273283e-01
-7.64026046e-01 -1.15259051e+00 -1.28930718e-01 1.01322186e+00
9.40375209e-01 -2.24634454e-01 -7.73542583e-01 -3.44379712e-03
-2.84563333e-01 6.65454626e-01 -3.85618657e-01 -3.00094843e-01
-8.04431856e-01 -9.95131075e-01 3.84880275e-01 3.61456245e-01
1.13430977e+00 -8.17306578e-01 -5.60543597e-01 4.15625423e-02
-3.30015004e-01 -9.07351732e-01 -7.23885953e-01 -8.72507766e-02
-3.74623537e-01 -8.04505348e-01 -9.11186576e-01 -6.43503368e-01
3.39115351e-01 5.66640675e-01 9.26320314e-01 1.09245844e-01
-4.40803587e-01 7.52372444e-02 -2.64184862e-01 5.48663437e-02
-1.23054281e-01 -1.77905157e-01 -3.04733098e-01 2.85883099e-01
1.51165307e-01 -6.45762146e-01 -1.00306404e+00 6.75891340e-01
-9.60312307e-01 3.22656959e-01 7.48787463e-01 9.01062012e-01
9.32432652e-01 4.02643353e-01 1.97421625e-01 -3.19767714e-01
3.31005275e-01 -1.74508676e-01 -8.42513740e-01 4.63691413e-01
-3.66921842e-01 -4.39086646e-01 5.53493142e-01 -3.53506625e-01
-1.29663551e+00 -2.85939991e-01 1.55870661e-01 -5.45224845e-01
-4.76668030e-02 2.54842807e-02 -5.68178236e-01 -2.73300678e-01
3.59815925e-01 6.18358254e-01 -2.50166088e-01 -2.67410934e-01
4.41901684e-01 4.67423141e-01 9.89003658e-01 -5.52058637e-01
9.41231132e-01 4.56983566e-01 -1.01669997e-01 -6.21670604e-01
-4.48448032e-01 -3.02177012e-01 -4.48221624e-01 -4.31466907e-01
1.11427116e+00 -1.01718640e+00 -7.03574181e-01 5.08232832e-01
-6.90124094e-01 -4.03000593e-01 -1.72912776e-01 4.11645323e-01
-2.35327631e-01 3.55914235e-01 -7.00203776e-01 -6.61823630e-01
-4.90756840e-01 -1.19612503e+00 8.36105227e-01 7.79211700e-01
8.30688402e-02 -6.13340974e-01 -3.48141700e-01 4.68980074e-01
6.48174226e-01 3.69439662e-01 6.23842895e-01 -5.87650910e-02
-7.69631445e-01 2.05687746e-01 -6.01487637e-01 2.98251599e-01
2.61542678e-01 1.17169730e-01 -8.10569704e-01 -1.95114672e-01
-9.08127725e-02 2.62525469e-01 8.12962115e-01 5.45993090e-01
1.06740081e+00 -2.95265508e-03 -1.97411969e-01 9.08123314e-01
1.64788377e+00 5.79749703e-01 8.26704383e-01 4.42874551e-01
8.14566433e-01 4.84022409e-01 7.89422154e-01 5.75510681e-01
1.90195724e-01 1.09159684e+00 1.51075616e-01 -5.61093211e-01
-6.44650996e-01 1.37219265e-01 1.90023854e-01 7.95242071e-01
-1.97519317e-01 -1.19120747e-01 -7.57003486e-01 3.64915252e-01
-1.73651779e+00 -9.12001193e-01 -1.58798188e-01 2.20425224e+00
9.50328171e-01 -3.63484249e-02 -1.63560675e-03 -8.10607672e-02
1.18697512e+00 3.53109092e-01 -5.36452472e-01 -4.92359176e-02
-6.60450280e-01 3.09799761e-01 5.93188047e-01 4.10241246e-01
-1.21035314e+00 1.00132370e+00 3.83419633e+00 1.69045699e+00
-1.27310026e+00 1.62415311e-01 9.42100108e-01 1.56266671e-02
-1.04704417e-01 -1.02543794e-01 -6.95589244e-01 6.97196424e-01
3.42599243e-01 -1.88916754e-02 5.20466268e-01 3.32341939e-01
2.55095422e-01 -4.08234626e-01 -1.21438242e-01 1.23016691e+00
1.54407740e-01 -1.19337773e+00 -7.90798664e-02 -2.34176278e-01
8.71592462e-01 -5.14730275e-01 2.11598277e-01 -7.59072453e-02
-5.76431230e-02 -6.54767632e-01 7.50811338e-01 6.91831708e-01
8.96088958e-01 -9.68411386e-01 5.67105949e-01 9.40564368e-03
-1.70010912e+00 5.39447516e-02 -2.14618832e-01 4.21620786e-01
1.19488843e-01 6.00627899e-01 -1.03503548e-01 9.71862793e-01
1.08498335e+00 4.77091551e-01 -6.63984597e-01 9.15389240e-01
-1.67277366e-01 3.82753134e-01 -1.53657272e-01 4.36553776e-01
5.51088378e-02 -6.03606462e-01 5.72919428e-01 1.27237880e+00
4.87354100e-01 3.24912459e-01 1.69911399e-01 9.13022161e-01
2.69749790e-01 3.76614422e-01 9.91537701e-03 4.40316558e-01
4.31778163e-01 1.42279530e+00 -9.96135414e-01 -3.63827407e-01
-4.28005308e-01 1.13448787e+00 -1.80876330e-01 5.28231621e-01
-1.30126727e+00 -6.45714879e-01 5.15075803e-01 -5.04082516e-02
5.19916654e-01 -4.25545350e-02 -3.15624028e-01 -1.02803111e+00
5.21572493e-02 -8.83331537e-01 5.12965679e-01 -9.47879255e-01
-1.10362196e+00 5.40914416e-01 5.63886389e-02 -1.36290419e+00
4.28821474e-01 -1.87405065e-01 -5.49248993e-01 1.05431247e+00
-1.60994649e+00 -1.39137518e+00 -6.80463433e-01 8.75941455e-01
4.28285092e-01 2.43932963e-03 2.18906388e-01 4.48831141e-01
-9.04839814e-01 6.50817931e-01 1.14523031e-01 -1.59823000e-02
9.90667999e-01 -8.51236641e-01 -2.07373694e-01 1.16498387e+00
-6.52423054e-02 4.99636889e-01 5.46952486e-01 -6.67158902e-01
-1.25083506e+00 -1.05537105e+00 4.84551460e-01 9.82800499e-02
4.94095773e-01 -4.29384895e-02 -1.04916513e+00 6.19542673e-02
4.02315080e-01 1.93038404e-01 5.12773633e-01 -5.83553076e-01
-2.11998761e-01 -6.89860940e-01 -9.65364397e-01 8.02075028e-01
9.48585689e-01 -2.88089246e-01 -1.45925775e-01 -1.69416979e-01
5.97016335e-01 -5.12878001e-01 -1.01991582e+00 7.69146442e-01
4.15283740e-01 -1.38937938e+00 1.22927630e+00 5.00562966e-01
1.10217892e-01 -1.05782413e+00 -3.54732096e-01 -6.12897277e-01
-5.17309070e-01 -7.41043031e-01 -1.16221942e-02 1.58568859e+00
1.03080615e-01 -5.76917052e-01 3.88362706e-01 2.70074278e-01
-2.31448412e-01 -7.52467394e-01 -7.84358919e-01 -5.63502192e-01
-2.47687697e-01 -1.68616101e-01 5.82694054e-01 1.04654288e+00
-5.54926813e-01 9.28872153e-02 -5.61021209e-01 3.27542394e-01
6.93947792e-01 5.21425307e-01 6.96669996e-01 -5.44087887e-01
-2.98467278e-01 -8.17506671e-01 -2.41037786e-01 -7.49259591e-01
-1.22518189e-01 -6.24755442e-01 -1.47934154e-01 -1.24373007e+00
2.83419818e-01 -3.12973022e-01 -7.35199630e-01 2.75990665e-01
-5.32302976e-01 8.92133713e-01 4.07142431e-01 3.02401304e-01
-6.26955807e-01 5.67929745e-01 1.45014560e+00 -1.61501139e-01
-3.88275504e-01 -1.79125652e-01 -7.88520098e-01 5.06044626e-01
9.75694180e-01 2.54130810e-02 -3.74800831e-01 -1.77398622e-01
-2.36385316e-01 3.82092781e-02 5.58938265e-01 -1.34569252e+00
2.75348842e-01 -3.52899671e-01 7.74280131e-01 -7.02126086e-01
2.87853330e-01 -6.31695569e-01 7.16139019e-01 4.26320672e-01
1.84170771e-02 2.03768089e-02 3.06050837e-01 4.75026488e-01
-4.71671343e-01 2.76998311e-01 1.09668624e+00 5.79726249e-02
-1.11578000e+00 1.01121962e-01 -1.27120642e-02 -1.35049194e-01
1.25923467e+00 -5.41496515e-01 -4.31477338e-01 -1.69066757e-01
-4.19794440e-01 1.14739172e-01 6.81335926e-01 2.60440290e-01
4.99709070e-01 -1.32225752e+00 -7.04480171e-01 1.97350010e-01
-9.69581865e-03 -3.18798333e-01 8.96317720e-01 1.34120452e+00
-6.29325509e-01 3.62443589e-02 -3.06994051e-01 -5.55276990e-01
-1.36149168e+00 6.14915848e-01 4.12754893e-01 -9.78796482e-02
-4.71422195e-01 9.58057284e-01 5.20529866e-01 8.42500776e-02
-3.95044088e-02 -3.93203758e-02 -1.88202485e-01 7.93916211e-02
4.39579695e-01 5.38976431e-01 -1.14753880e-01 -1.05171061e+00
-3.16393107e-01 1.07031465e+00 2.74843752e-01 1.98232681e-01
9.20759737e-01 -4.79586393e-01 -4.60337758e-01 1.74714923e-01
1.09972775e+00 3.59411955e-01 -1.25594819e+00 -4.02988017e-01
-3.06677908e-01 -7.49902546e-01 -1.74591839e-02 -7.51023650e-01
-1.43274760e+00 7.09818184e-01 9.83190358e-01 -1.90003216e-01
1.91907585e+00 -6.71338439e-02 9.50644910e-01 -3.07478279e-01
8.80430788e-02 -1.14103556e+00 1.45075768e-01 7.13477656e-02
7.96409190e-01 -9.25992668e-01 2.67379254e-01 -5.92236221e-01
-7.19343245e-01 1.02382994e+00 7.96972811e-01 -4.46447730e-02
4.88966078e-01 1.83407769e-01 2.03963146e-01 1.50621645e-02
-1.89832851e-01 -3.25057715e-01 6.23576581e-01 3.67905647e-01
2.42573351e-01 -7.40911886e-02 -4.53056902e-01 4.92110044e-01
1.68570206e-02 -3.56735051e-01 -6.63303286e-02 6.57144606e-01
-4.19457197e-01 -9.42592978e-01 -7.68585145e-01 1.14781164e-01
-3.93786222e-01 -2.22956598e-01 6.13652617e-02 8.28115642e-01
7.07486928e-01 9.48777616e-01 -2.93932017e-02 -4.74652797e-01
2.20086664e-01 -2.58191019e-01 3.26570779e-01 1.01550326e-01
-6.35515273e-01 7.68440783e-01 -2.72792071e-01 -8.07231843e-01
-7.35103250e-01 -4.49211508e-01 -1.07826316e+00 -2.39776477e-01
-3.01031530e-01 1.65846515e-02 1.65791944e-01 3.28174949e-01
2.68084139e-01 7.07719862e-01 6.08170211e-01 -8.92491281e-01
1.55421540e-01 -5.14748812e-01 -7.68372893e-01 3.34326178e-01
8.02699775e-02 -6.03759050e-01 -1.60663456e-01 2.69394934e-01] | [11.175573348999023, -1.3394991159439087] |
634ad4f4-b7c9-41b6-925e-b368156de124 | prompt-agnostic-essay-scorer-a-domain | 2008.01441 | null | https://arxiv.org/abs/2008.01441v1 | https://arxiv.org/pdf/2008.01441v1.pdf | Prompt Agnostic Essay Scorer: A Domain Generalization Approach to Cross-prompt Automated Essay Scoring | Cross-prompt automated essay scoring (AES) requires the system to use non target-prompt essays to award scores to a target-prompt essay. Since obtaining a large quantity of pre-graded essays to a particular prompt is often difficult and unrealistic, the task of cross-prompt AES is vital for the development of real-world AES systems, yet it remains an under-explored area of research. Models designed for prompt-specific AES rely heavily on prompt-specific knowledge and perform poorly in the cross-prompt setting, whereas current approaches to cross-prompt AES either require a certain quantity of labelled target-prompt essays or require a large quantity of unlabelled target-prompt essays to perform transfer learning in a multi-step manner. To address these issues, we introduce Prompt Agnostic Essay Scorer (PAES) for cross-prompt AES. Our method requires no access to labelled or unlabelled target-prompt data during training and is a single-stage approach. PAES is easy to apply in practice and achieves state-of-the-art performance on the Automated Student Assessment Prize (ASAP) dataset. | ['Xin-yu Dai', 'Shu-Jian Huang', 'Jia-Jun Chen', 'Liang He', 'Robert Ridley'] | 2020-08-04 | null | null | null | null | ['automated-essay-scoring'] | ['natural-language-processing'] | [ 3.1389391e-01 -2.7601150e-01 -2.3985758e-02 -6.4796269e-01
-1.5271480e+00 -1.1185250e+00 3.9292565e-01 5.6337863e-01
-6.9409049e-01 7.6050943e-01 -7.6783612e-02 -5.9614748e-01
-1.9087727e-01 -6.6322500e-01 -3.5464194e-01 -1.1749011e-01
7.2638738e-01 7.7453989e-01 3.4230858e-01 -3.5276514e-01
6.0755199e-01 1.4828081e-01 -1.2827823e+00 3.8995942e-01
9.7903466e-01 7.1923977e-01 3.6158945e-02 1.3071809e+00
-2.9641241e-01 9.5144498e-01 -1.0127919e+00 -7.5099140e-01
2.8680071e-01 -6.5072381e-01 -1.0511719e+00 -5.4021454e-01
8.3491069e-01 -2.8121540e-01 1.6577888e-01 7.2371513e-01
5.4376805e-01 3.0400530e-01 9.0328926e-01 -1.1476451e+00
-8.3603507e-01 3.7142515e-01 -4.0698805e-01 4.9377987e-01
6.8800396e-01 1.8609636e-01 1.4333980e+00 -1.0448574e+00
3.8701610e-04 4.4712338e-01 7.4386555e-01 6.6769516e-01
-1.1179985e+00 -8.0486828e-01 -2.2062002e-01 3.4393409e-01
-7.7901912e-01 -2.4667974e-01 7.0323473e-01 -4.6858150e-01
4.8482195e-01 3.8830727e-01 5.6092626e-01 1.0179374e+00
1.3226256e-01 1.1429573e+00 1.6488926e+00 -6.1490363e-01
2.2230121e-01 -1.1639476e-02 8.0947208e-01 4.4772425e-01
-1.2676267e-01 2.4051699e-01 -1.0443451e+00 -1.8176870e-01
4.5153540e-01 -2.0744696e-01 -1.7181966e-01 -9.5890611e-02
-1.0139500e+00 8.8431180e-01 -2.0062128e-02 -1.2180487e-01
-3.3682212e-01 -2.7376264e-01 3.0168653e-01 1.0274858e+00
1.6100167e-01 1.0842274e+00 -7.8230387e-01 -7.7945882e-01
-1.3958815e+00 5.7967526e-01 1.1305151e+00 5.6998754e-01
4.8062757e-01 -1.3893569e-01 -5.3134960e-01 8.8798177e-01
-1.7768401e-01 3.1169492e-01 6.9685137e-01 -6.9476271e-01
5.4632229e-01 9.1744131e-01 2.9220286e-01 -4.5839095e-01
-2.9712945e-01 -1.9787186e-01 -3.1401557e-01 3.9000207e-01
9.0089846e-01 -1.1070838e-01 -7.3867816e-01 1.6133425e+00
1.2393997e-01 -1.9896738e-02 -1.3896519e-02 8.7495965e-01
8.7151599e-01 5.4676044e-01 3.6957961e-01 3.4003280e-02
1.4398385e+00 -1.1094064e+00 -2.9902729e-01 -2.6148355e-01
6.3603973e-01 -1.1108067e+00 1.7841352e+00 6.1351246e-01
-1.4140213e+00 -4.6291605e-01 -9.9706274e-01 -2.7196920e-01
-1.7117290e-01 1.9942942e-01 3.2254893e-01 8.1097448e-01
-9.6646345e-01 3.3799481e-01 -2.9252264e-01 1.7308138e-02
2.0214166e-01 4.3346879e-01 -2.2426605e-01 -2.8429139e-02
-1.1176969e+00 1.0750035e+00 -5.7006620e-02 -5.2775031e-01
-4.9092337e-01 -1.3450339e+00 -5.8582532e-01 4.3249342e-01
8.1776321e-02 -4.8002842e-01 2.2918365e+00 -9.2558646e-01
-1.8882756e+00 9.5295316e-01 1.6638795e-01 -7.1783207e-02
5.1695722e-01 -4.2045888e-01 -5.7370840e-03 5.6312125e-02
-2.0505564e-02 3.3601761e-01 5.0281638e-01 -5.6008345e-01
-7.3217398e-01 -1.5939330e-01 9.3421444e-02 5.7231075e-01
-6.5266871e-01 3.1039026e-01 2.0488317e-01 -6.9094539e-01
-2.2480796e-01 -7.3295820e-01 9.0687506e-02 -2.7122006e-01
2.2511740e-01 -6.9181317e-01 4.7683737e-01 -5.4894930e-01
1.4363835e+00 -1.6475471e+00 -3.4121394e-01 4.5887060e-03
1.8983783e-01 4.3760756e-01 -3.8147745e-01 3.1069762e-01
-5.5392869e-02 -4.1222683e-01 -2.7223346e-01 -1.4824034e-01
2.2181937e-01 -2.7548248e-01 -2.9672757e-01 -3.3703998e-02
9.3070939e-02 8.7675363e-01 -1.2067220e+00 -5.2704674e-01
7.1403176e-02 -3.0821383e-01 -5.9628719e-01 6.5393621e-01
-5.6419387e-03 2.0398633e-01 -3.1727320e-01 6.2109709e-01
1.1063740e-01 -2.6844785e-01 -4.1874996e-01 4.5425171e-01
5.7013746e-02 8.1725621e-01 -1.0533589e+00 1.5779408e+00
-4.8424783e-01 4.2531726e-01 -2.4818040e-01 -7.9091817e-01
1.1889508e+00 4.3240929e-01 3.3199283e-01 -8.7939465e-01
6.9472380e-03 5.1877594e-01 6.1400618e-02 -1.5615343e-01
9.6676731e-01 -2.5523070e-01 -4.6804357e-01 1.3998935e+00
2.2459757e-01 -3.9163885e-01 2.9217362e-01 3.5247302e-01
1.5857605e+00 -8.2690865e-02 4.9020395e-01 -2.1566790e-01
5.5213100e-01 1.3609654e-01 1.8803950e-01 8.3775651e-01
-5.2382565e-01 7.3953795e-01 7.2351567e-02 -5.0918013e-01
-9.9131411e-01 -1.1610062e+00 1.8107207e-01 1.7313768e+00
-2.6294783e-01 -2.2003682e-01 -7.6749563e-01 -1.0584344e+00
-8.8588811e-02 8.1955940e-01 -4.6441042e-01 -2.1992901e-01
-5.0298345e-01 -5.0701964e-01 6.3229042e-01 7.7638417e-01
1.2060817e-01 -1.3766150e+00 -1.0035222e+00 5.4363310e-01
-1.9050112e-01 -5.6857890e-01 -9.4834262e-01 5.7425666e-01
-5.4923415e-01 -1.2996432e+00 -9.5926976e-01 -7.0189065e-01
6.2895679e-01 1.6283825e-01 1.7385443e+00 8.4138803e-02
5.2409437e-02 8.3335108e-01 -3.4405157e-01 -7.4589431e-01
-2.6572156e-01 5.3286111e-01 -1.0089126e-01 -3.7578577e-01
1.1991442e+00 -3.2216159e-01 -4.7270042e-01 2.1547173e-01
-6.9935793e-01 -1.7754020e-02 6.2622684e-01 1.2005243e+00
1.5149756e-01 -7.9094565e-01 9.8296809e-01 -9.3896055e-01
1.4264901e+00 -2.7006575e-01 -5.8392012e-01 4.5192498e-01
-8.6812979e-01 -3.5654634e-02 8.1800467e-01 -7.7580965e-01
-8.8451701e-01 8.7130837e-02 -8.3615705e-02 3.7007080e-03
-1.6486283e-01 6.9838166e-01 4.3528929e-01 -1.8602549e-01
1.0537328e+00 2.1809633e-01 -2.8003708e-01 -2.6487467e-01
-2.6728133e-02 7.5628376e-01 8.8268697e-01 -1.1421117e+00
7.2363603e-01 -5.3266817e-01 -2.4668916e-01 -4.6177447e-02
-1.3815975e+00 -8.1207168e-01 -6.2825537e-01 -3.5642821e-01
2.5227389e-01 -7.0872271e-01 -7.6432341e-01 5.1637572e-01
-1.0170977e+00 -8.1656623e-01 -4.7709012e-01 2.7152717e-01
-5.3968918e-01 7.5667225e-02 -6.7029077e-01 -5.8924931e-01
-7.8145427e-01 -1.0799828e+00 6.4453101e-01 5.8308464e-01
-1.0305907e+00 -1.0148972e+00 6.0819119e-01 5.9946287e-01
3.7503672e-01 -5.3752619e-01 7.6875347e-01 -1.2480116e+00
-2.3738702e-01 -5.0315660e-01 -4.9385563e-02 2.9595220e-01
-1.2162152e-01 -2.3725055e-01 -9.7736830e-01 -4.5303848e-02
-2.2002133e-02 -1.1594254e+00 4.8779130e-01 6.6995241e-02
1.0913916e+00 -7.0034862e-02 5.2043915e-01 1.4830530e-01
8.2971919e-01 -9.2814811e-02 8.2803823e-02 5.9350544e-01
2.8885481e-01 4.9869800e-01 7.7163935e-01 3.0211416e-01
8.0718833e-01 5.5639249e-01 -3.8415629e-01 2.3596624e-01
-5.0846562e-02 -2.5323611e-01 4.9903372e-01 1.1943804e+00
2.9524332e-01 -9.1954721e-03 -1.1859946e+00 9.5045781e-01
-1.4402093e+00 -7.3984265e-01 -4.0468946e-01 2.2081683e+00
1.5882561e+00 -7.5997852e-02 9.9547170e-02 4.9496755e-01
2.2966409e-01 -1.5633988e-01 -3.2674128e-01 -8.9071774e-01
1.4414297e-01 9.9973857e-01 1.4059137e-01 4.3018857e-01
-7.9767704e-01 7.3614556e-01 6.4623394e+00 6.5708441e-01
-9.5910257e-01 7.8376494e-02 3.9570647e-01 -1.6350378e-01
-2.2847807e-01 -9.5381953e-02 -8.3715338e-01 5.5918866e-01
1.3800859e+00 -5.2306223e-01 1.9263534e-01 9.0089577e-01
-3.5165316e-01 -3.2788894e-01 -1.3461757e+00 8.6375856e-01
4.1551154e-02 -8.9542186e-01 -2.3995376e-01 -4.6333721e-01
9.5552123e-01 -3.4834927e-01 1.4119337e-01 8.7916082e-01
8.8403332e-01 -1.0319571e+00 6.3833982e-01 2.4466337e-01
9.2622244e-01 -8.0117679e-01 5.6813264e-01 4.5022112e-01
-7.7620214e-01 -9.2711583e-02 -2.6154903e-01 -4.8008201e-01
-1.5300125e-01 4.0976161e-01 -8.9960802e-01 2.3132995e-01
5.0846100e-01 1.1542852e-01 -7.7713871e-01 1.2774230e+00
-6.6289812e-01 1.1285665e+00 -1.8039285e-01 -4.1734320e-01
1.8296018e-01 1.3202897e-01 2.1368168e-02 1.0185516e+00
4.1756359e-01 6.3220590e-01 2.1656772e-01 4.8670706e-01
-3.0546126e-01 1.8374705e-01 1.9229177e-02 7.1776256e-02
6.5936238e-01 1.3724347e+00 -2.1043469e-01 -2.8990594e-01
-4.4731444e-01 9.1210628e-01 5.5476564e-01 7.9201907e-02
-3.5999623e-01 -4.3749782e-01 1.2151854e-01 -9.3158066e-02
-2.4574594e-01 1.0320298e-01 -1.0116276e+00 -9.0012306e-01
2.7196700e-02 -1.2804490e+00 8.2863134e-01 -8.4392780e-01
-1.7252187e+00 2.4394363e-01 -3.9899260e-01 -1.1442107e+00
-3.1336331e-01 -6.0958093e-01 -1.0923206e+00 1.1961840e+00
-1.8218783e+00 -7.4524069e-01 -6.6304672e-01 7.1536791e-01
7.3071140e-01 -3.5254535e-01 1.2080894e+00 2.5488573e-01
-3.8761467e-02 1.2273594e+00 -2.3183893e-01 9.6307918e-02
1.5738846e+00 -1.7705545e+00 3.2756925e-01 6.5062451e-01
3.0280632e-01 3.3916464e-01 6.3031209e-01 -4.1979101e-01
-9.8349923e-01 -7.0170289e-01 1.3998539e+00 -1.1352208e+00
6.9126958e-01 1.0676906e-01 -1.1498717e+00 4.9997187e-01
1.9408989e-01 -1.0078942e-01 1.2932894e+00 5.0942838e-01
-4.2714149e-01 8.3974637e-02 -8.9015573e-01 6.3433295e-01
2.8775626e-01 -6.5415120e-01 -1.2797153e+00 1.2995459e-01
1.4606626e-02 -8.1375200e-01 -7.9756981e-01 1.9008854e-01
5.2649534e-01 -6.1162120e-01 6.1408716e-01 -6.2880021e-01
8.6108989e-01 2.8797151e-03 4.5679983e-01 -1.6443819e+00
-2.4648514e-01 -5.3888887e-01 -3.2222015e-01 1.0034388e+00
3.9731577e-01 -1.7367548e-01 1.0859984e+00 1.0466002e+00
-4.2156747e-01 -8.5920423e-01 -5.0457513e-01 -6.4094943e-01
4.2862368e-01 -2.5291851e-01 6.8138593e-01 1.1380284e+00
2.9974484e-01 5.1477283e-01 -5.3198379e-02 -2.1212123e-01
4.5356128e-01 2.3903984e-01 1.0202792e+00 -1.6536815e+00
-6.3440524e-02 -7.6937634e-01 1.8929069e-01 -1.0718254e+00
1.1355143e-01 -7.9113835e-01 2.8811845e-01 -1.1294632e+00
2.6873386e-01 -7.8412241e-01 -6.6136533e-01 6.9903123e-01
-9.4838202e-01 2.9510644e-01 -5.6226589e-03 1.0143476e-01
-7.4205196e-01 4.3958911e-01 1.1520460e+00 1.6784981e-02
-1.2454110e-01 2.2474974e-01 -9.1352630e-01 4.4809285e-01
7.4955601e-01 -7.1265948e-01 -4.5368412e-01 -2.2736871e-01
1.9837189e-01 1.6096574e-01 1.6513699e-01 -8.8849276e-01
8.9794159e-01 -4.6725017e-01 4.7452062e-01 -5.3961962e-01
5.3286042e-02 -4.6631542e-01 -6.5019333e-01 -1.8183207e-02
-7.7021736e-01 5.3619647e-01 9.1997959e-02 7.5758725e-02
-3.6964586e-01 -8.1391490e-01 7.3264021e-01 -2.6300704e-01
-4.9523646e-01 2.0932616e-01 -2.2440921e-01 4.7602239e-01
9.3902898e-01 -3.9465597e-01 -3.3680961e-01 -2.5135732e-01
-1.2249815e-01 3.5926157e-01 4.3253034e-01 3.9378014e-01
4.4090772e-01 -1.2023511e+00 -9.8106962e-01 8.9296967e-02
2.5950834e-01 8.3140694e-02 2.1378732e-01 6.4579225e-01
-3.6957932e-01 5.7883108e-01 -4.4484240e-01 -3.9158195e-01
-1.4088104e+00 -4.4993121e-02 -3.7350085e-02 -8.9085424e-01
-3.2902193e-01 1.2110837e+00 -1.2540257e-01 -1.0184662e+00
1.4735374e-01 8.0767930e-02 -2.2165205e-01 5.2724354e-02
7.8436041e-01 2.0334138e-01 4.9215016e-01 -1.7802235e-03
2.4971327e-01 4.8246086e-02 -4.5884791e-01 -3.1924933e-01
1.2262510e+00 3.8328013e-01 4.3980420e-01 5.5037701e-01
5.3774816e-01 7.5815536e-02 -1.2416313e+00 -3.2741782e-01
3.9281109e-01 -3.6277324e-01 -2.9968647e-02 -1.5423348e+00
-5.3683186e-01 9.1719180e-01 1.0510276e-02 -2.0853542e-01
1.0630099e+00 -6.3182002e-01 9.5335114e-01 5.1310253e-01
2.6049125e-01 -1.2639570e+00 6.4270526e-01 9.1438258e-01
6.7545897e-01 -1.2667903e+00 -8.1113979e-02 1.6021265e-01
-6.5026736e-01 1.2730371e+00 1.2154945e+00 -5.4606836e-02
1.5323514e-01 2.8429112e-01 5.7346797e-01 -1.1061095e-01
-9.3806392e-01 4.7801316e-01 7.0366520e-01 2.3663022e-01
7.7641302e-01 1.5871838e-01 -4.4972149e-01 1.2156144e+00
-7.4586380e-01 1.2462607e-01 9.0381610e-01 1.2132275e+00
-3.8023052e-01 -1.4422163e+00 -5.0732499e-01 8.5964388e-01
-5.5185145e-01 -3.2408035e-01 -5.6545264e-01 3.6969939e-01
-6.6063803e-01 7.4312288e-01 -2.6206052e-01 -7.4970655e-02
3.7470144e-01 8.1315506e-01 7.2890025e-01 -9.7121853e-01
-1.3677076e+00 -7.5007606e-01 -2.5625506e-01 -9.1985501e-02
-1.7700767e-02 -7.8268540e-01 -9.6482807e-01 -4.2473796e-01
-3.7557089e-01 4.4136885e-01 5.0992054e-01 1.1228606e+00
2.0056151e-02 5.6675845e-01 4.5541486e-01 -3.2448444e-01
-1.2395282e+00 -1.0636975e+00 -3.9055002e-01 4.8780242e-01
2.2738503e-01 -5.6437731e-01 -2.0492062e-01 2.3667393e-02] | [11.298749923706055, 9.336723327636719] |
c2771711-8a39-48b5-b324-a726efb8e394 | improving-code-switching-dependency-parsing | null | null | https://aclanthology.org/2022.findings-naacl.87 | https://aclanthology.org/2022.findings-naacl.87.pdf | Improving Code-Switching Dependency Parsing with Semi-Supervised Auxiliary Tasks | Code-switching dependency parsing stands as a challenging task due to both the scarcity of necessary resources and the structural difficulties embedded in code-switched languages. In this study, we introduce novel sequence labeling models to be used as auxiliary tasks for dependency parsing of code-switched text in a semi-supervised scheme. We show that using auxiliary tasks enhances the performance of an LSTM-based dependency parsing model and leads to better results compared to an XLM-R-based model with significantly less computational and time complexity. As the first study that focuses on multiple code-switching language pairs for dependency parsing, we acquire state-of-the-art scores on all of the studied languages. Our best models outperform the previous work by 7.4 LAS points on average. | ['Özlem Çetinoğlu', 'Tunga Gungor', 'Arzucan Özgür', 'Şaziye Özateş'] | null | null | null | null | findings-naacl-2022-7 | ['xlm-r'] | ['natural-language-processing'] | [ 1.00077912e-01 1.77907571e-02 -3.55903208e-01 -4.03624296e-01
-1.19010222e+00 -6.08896077e-01 1.39076903e-01 2.39813268e-01
-4.19250131e-01 6.86782837e-01 6.62628263e-02 -9.83233690e-01
2.88267970e-01 -2.08091781e-01 -7.50335813e-01 -2.56589890e-01
-2.53643274e-01 4.36800092e-01 3.71569604e-01 -4.75969940e-01
1.57580927e-01 1.24956429e-01 -1.13103950e+00 4.50439453e-01
1.04844069e+00 3.23490858e-01 7.11214542e-01 6.63787127e-01
-4.84869152e-01 9.43308532e-01 -5.22763848e-01 -2.87711620e-01
-1.26526073e-01 -5.54897726e-01 -1.01510715e+00 -2.16360211e-01
2.73351014e-01 2.98156291e-01 3.90310511e-02 1.02868974e+00
1.88480243e-01 -1.48323238e-01 1.21557973e-01 -7.49554694e-01
-5.25837839e-01 1.18339288e+00 -4.35674489e-01 4.36790764e-01
4.85528171e-01 -3.59766603e-01 9.72968280e-01 -5.46096981e-01
6.09984815e-01 8.97763193e-01 6.07924223e-01 6.65666103e-01
-1.23882842e+00 -7.54606724e-01 2.86507249e-01 -1.75869521e-02
-9.05047655e-01 -5.62853575e-01 6.24673963e-01 -5.23922980e-01
1.63609219e+00 -2.01417208e-01 2.45328709e-01 1.13596249e+00
4.56481397e-01 5.30432701e-01 1.04930604e+00 -8.99725616e-01
-7.24682733e-02 1.04202449e-01 5.21802545e-01 1.17736566e+00
1.62036762e-01 -1.74736187e-01 -3.73646349e-01 2.61379480e-01
4.25033510e-01 -4.17127371e-01 8.95436928e-02 -6.01258390e-02
-1.13528347e+00 8.25058579e-01 2.90670305e-01 7.88470626e-01
1.85456976e-01 3.55946422e-01 6.70037925e-01 6.97884977e-01
5.74979544e-01 4.63251799e-01 -9.00558770e-01 -3.82509232e-01
-9.07706857e-01 -4.95108515e-01 9.58584964e-01 1.50279212e+00
5.33426225e-01 1.37576923e-01 -4.25365083e-02 8.66420984e-01
3.26555640e-01 2.68329859e-01 4.74831462e-01 -5.89090586e-01
9.50333357e-01 5.46128929e-01 -4.34563786e-01 -3.18386197e-01
-7.36721992e-01 -6.28200829e-01 -5.78104556e-01 -1.37770055e-02
4.42511529e-01 -3.28246146e-01 -8.39302540e-01 1.83180392e+00
-3.75940830e-01 -2.00208649e-01 1.77509353e-01 9.04877260e-02
8.32379699e-01 4.59269702e-01 1.75726399e-01 -1.82330266e-01
1.44484806e+00 -1.44106126e+00 -5.59419334e-01 -6.35790408e-01
1.28844869e+00 -9.43836868e-01 1.04877710e+00 1.38436437e-01
-9.51629758e-01 -5.78535020e-01 -1.01048565e+00 7.26808282e-03
-1.34124488e-01 3.20354581e-01 7.31241286e-01 9.76233363e-01
-1.51990199e+00 4.44729567e-01 -9.81718183e-01 -5.22202730e-01
1.55905634e-01 5.40095389e-01 -6.54861093e-01 -1.90048106e-02
-8.67519200e-01 8.91469538e-01 3.32031518e-01 -2.03619570e-01
-6.84318483e-01 -3.56574178e-01 -1.16048193e+00 7.26363510e-02
2.99314499e-01 -1.71084747e-01 1.33851135e+00 -9.82026517e-01
-1.55854201e+00 1.15674353e+00 -2.82060951e-01 -3.60919744e-01
1.84983999e-01 -2.63105363e-01 -4.47142243e-01 -3.06801081e-01
1.80789292e-01 3.92494589e-01 5.60278237e-01 -1.02939844e+00
-5.90135634e-01 -5.18807136e-02 2.76211500e-01 -3.29093993e-01
-1.12427682e-01 5.47117889e-01 -7.16973662e-01 -5.68310201e-01
-1.12661511e-01 -1.38697016e+00 -3.63686353e-01 -7.89267242e-01
-3.36894274e-01 -1.21448018e-01 5.02884090e-01 -1.02834868e+00
1.32205415e+00 -2.16521502e+00 5.00927210e-01 -3.39045018e-01
-9.00457054e-02 3.63159418e-01 -5.67114413e-01 4.44614887e-01
-3.65379989e-01 3.09485048e-01 -7.03176558e-01 -8.15736175e-01
-4.48792100e-01 4.06227022e-01 6.70175254e-02 1.90421894e-01
4.65896904e-01 8.15000713e-01 -8.81416857e-01 -5.38758218e-01
-6.51894584e-02 1.75361216e-01 -6.89238250e-01 3.01367551e-01
-3.12386245e-01 8.25139761e-01 -6.97612390e-02 5.88712096e-01
3.61530185e-01 -5.53449802e-02 6.11333489e-01 5.37669480e-01
-3.82293999e-01 5.82316399e-01 -4.36210692e-01 2.43503666e+00
-1.31075048e+00 8.09381545e-01 -1.70808546e-02 -1.05492389e+00
9.58077848e-01 3.82665902e-01 -1.75051749e-01 -8.56291413e-01
3.64581384e-02 4.56012607e-01 4.73186523e-01 -5.34293532e-01
2.44906932e-01 5.16918786e-02 -7.13401020e-01 2.53747791e-01
3.78131688e-01 4.26055007e-02 4.26108211e-01 1.16154194e-01
1.29021537e+00 3.63684237e-01 3.74089986e-01 -7.10566759e-01
1.03320432e+00 -6.96649775e-02 4.24152941e-01 6.26720250e-01
1.21380553e-01 1.73234373e-01 8.05453539e-01 -2.32842386e-01
-8.51971984e-01 -6.31212831e-01 -1.76550969e-02 1.47649562e+00
-2.85768747e-01 -3.99288565e-01 -8.05685937e-01 -1.06096578e+00
-5.35812855e-01 8.67644668e-01 -5.23701966e-01 6.35215938e-02
-1.21529508e+00 -7.35737920e-01 6.77705824e-01 7.14123487e-01
2.49290124e-01 -1.09483469e+00 -4.56028908e-01 3.94809008e-01
-2.33199805e-01 -1.46566200e+00 -2.60864824e-01 9.29894328e-01
-1.07995248e+00 -8.67870033e-01 -3.65567565e-01 -1.49385118e+00
6.73961937e-01 -1.78559661e-01 1.53410006e+00 5.44215560e-01
7.91734159e-02 -2.31217876e-01 -7.69432068e-01 -1.32751390e-02
-1.09902310e+00 7.18451500e-01 -4.67536092e-01 -5.91001630e-01
1.14799313e-01 -5.05310833e-01 1.58189952e-01 -3.38284008e-04
-6.43027544e-01 -5.36984466e-02 8.02924037e-01 8.44559431e-01
-1.89427361e-02 -4.02016133e-01 2.96234131e-01 -1.38780737e+00
4.95663911e-01 -5.40568352e-01 -8.29183042e-01 2.93509126e-01
-4.93944407e-01 6.26964688e-01 1.05954254e+00 -1.57620043e-01
-1.27713406e+00 3.52943778e-01 -4.87385958e-01 3.21217388e-01
-3.03903818e-01 5.43271542e-01 8.33026171e-02 -2.38218620e-01
3.88245404e-01 1.04520880e-01 -4.54329133e-01 -6.80763602e-01
2.31592610e-01 4.93594289e-01 3.40639472e-01 -8.47809017e-01
6.20185316e-01 2.92853601e-02 5.24746254e-02 -7.68162787e-01
-8.57533753e-01 -4.17127103e-01 -1.24126184e+00 2.17827991e-01
1.04445314e+00 -9.34339166e-01 -1.16467468e-01 4.60374027e-01
-1.49498534e+00 -6.37450218e-01 2.64736593e-01 2.49584273e-01
-3.62121999e-01 6.79607213e-01 -9.40092087e-01 -4.80778456e-01
-7.99365044e-02 -1.56586289e+00 9.24691975e-01 -2.01409459e-01
-1.21719189e-01 -1.15228236e+00 3.18266064e-01 3.76192182e-01
4.79631454e-01 -6.21097051e-02 1.41238189e+00 -6.88715696e-01
-5.31561434e-01 2.68641207e-02 -1.42206594e-01 3.98599476e-01
-9.56878904e-03 -8.86971653e-02 -7.38687932e-01 -4.16275918e-01
-3.13452780e-02 -2.84993023e-01 9.46966410e-01 9.49664637e-02
6.96900010e-01 1.97052374e-01 -2.15124547e-01 5.62200189e-01
1.56394255e+00 4.24961954e-01 3.88802260e-01 3.83481801e-01
7.05930233e-01 7.20883310e-01 3.28662276e-01 1.51682675e-01
4.23297018e-01 5.95961630e-01 2.21159771e-01 -1.19716398e-01
-3.24278951e-01 7.79565051e-02 6.60858333e-01 1.75086451e+00
9.16941911e-02 -3.16159129e-01 -1.41839981e+00 7.18479037e-01
-1.78477728e+00 -2.62440562e-01 -6.68207645e-01 2.02462697e+00
7.61693835e-01 4.40498561e-01 -1.37467921e-01 -5.22360206e-02
7.09944069e-01 2.02068947e-02 3.49938460e-02 -8.76493156e-01
1.63665101e-01 6.47174299e-01 5.01189172e-01 7.23180413e-01
-1.09704196e+00 1.36222148e+00 6.74446917e+00 4.76835459e-01
-1.02601504e+00 6.12656116e-01 2.84367561e-01 3.88218760e-01
-4.13452461e-02 3.64998847e-01 -7.82090545e-01 3.49483728e-01
1.43481052e+00 3.12357306e-01 4.12071854e-01 7.93722808e-01
-3.54157180e-01 -1.66980460e-01 -1.39372587e+00 7.22057045e-01
2.15448111e-01 -8.49049866e-01 -3.97709697e-01 -2.34704807e-01
6.68194115e-01 5.50251067e-01 -2.52792686e-01 5.79269350e-01
5.31240284e-01 -9.58018780e-01 7.21330345e-01 1.60527229e-02
7.81488776e-01 -6.82515144e-01 9.76190031e-01 1.93089083e-01
-1.35573161e+00 -1.36752665e-01 -1.00434311e-01 -4.52988863e-01
2.04205528e-01 5.05103350e-01 -4.20159429e-01 5.43855727e-01
4.27189648e-01 7.21616447e-01 -8.87294412e-01 7.59269893e-01
-6.71374679e-01 7.93591440e-01 5.35693876e-02 -2.97786277e-02
4.03026879e-01 -1.52238652e-01 2.17420071e-01 1.90886056e+00
2.23493338e-01 -4.88657594e-01 2.98202723e-01 5.45092106e-01
-1.31144360e-01 1.91858336e-01 -5.98237872e-01 -9.95390583e-03
1.61269173e-01 1.12313390e+00 -1.14902115e+00 -3.80813509e-01
-9.33476865e-01 1.04668128e+00 7.98963904e-01 5.11369072e-02
-7.98115432e-01 -4.88121390e-01 3.17510962e-01 -4.18368757e-01
4.66888726e-01 -7.16823280e-01 -9.87344533e-02 -1.20461273e+00
6.27401005e-03 -6.47678077e-01 3.09406519e-01 -1.98085472e-01
-8.64153266e-01 1.22359121e+00 -8.65594968e-02 -1.06813967e+00
-3.50365788e-01 -6.79615974e-01 -5.54352343e-01 6.41629994e-01
-1.78542233e+00 -1.15951753e+00 3.81186977e-02 3.03469926e-01
8.31677675e-01 -3.63020569e-01 1.08332825e+00 5.54505765e-01
-7.99243867e-01 6.23441219e-01 2.25956231e-01 3.43415678e-01
4.03391063e-01 -1.25730884e+00 1.12203610e+00 1.33412075e+00
2.72928119e-01 5.47682941e-01 4.37638015e-01 -4.67576861e-01
-1.18866611e+00 -8.59895766e-01 1.24577427e+00 -6.26255333e-01
8.72373104e-01 -8.21865797e-01 -8.22553992e-01 9.15407777e-01
5.45126796e-01 -4.00823094e-02 8.28270316e-01 3.89269024e-01
-5.24020791e-01 3.25008303e-01 -6.84602857e-01 2.50614554e-01
1.35171521e+00 -7.84570932e-01 -7.48340487e-01 2.14638755e-01
1.10746610e+00 -3.58100981e-01 -9.71031368e-01 1.20250940e-01
1.90018803e-01 -9.85478401e-01 5.20798385e-01 -5.33020258e-01
7.15379179e-01 -2.04055570e-02 -2.13889480e-01 -1.23851824e+00
-4.87679422e-01 -3.48236889e-01 3.99657190e-01 1.36553037e+00
9.06038284e-01 -5.16882718e-01 4.69225466e-01 2.40904197e-01
-4.53352123e-01 -3.73902500e-01 -9.75555241e-01 -9.72464085e-01
4.27295804e-01 -5.30047297e-01 4.85039316e-02 9.77752864e-01
1.68258697e-01 6.93059742e-01 -2.12365538e-01 -7.86425546e-02
2.82510698e-01 2.77708799e-01 4.41986293e-01 -1.23112524e+00
-4.87863332e-01 -3.54896754e-01 -1.00003980e-01 -9.15153563e-01
9.16295290e-01 -1.24637043e+00 3.30405116e-01 -1.37672675e+00
2.02765465e-01 -6.61125422e-01 -4.08473730e-01 6.04367614e-01
1.82206109e-02 1.51834607e-01 3.74299377e-01 2.66657192e-02
-6.58776939e-01 1.27096936e-01 6.16536677e-01 -4.69796546e-03
-1.28354043e-01 -1.22633204e-01 -5.56020319e-01 7.41423011e-01
8.39632213e-01 -9.87909853e-01 -1.55706450e-01 -8.53959799e-01
4.80704427e-01 5.30958295e-01 -4.78510052e-01 -1.10785115e+00
7.25061670e-02 3.43140602e-01 -1.72344357e-01 -1.71271950e-01
-1.23456195e-01 -6.93788290e-01 -2.03577876e-01 9.05498743e-01
-3.39181602e-01 4.65980381e-01 6.27162099e-01 2.28941232e-01
-3.07633698e-01 -6.84942305e-01 7.71098971e-01 -4.07225251e-01
-7.11330652e-01 -2.56446332e-01 -8.56810927e-01 3.78756255e-01
6.73481643e-01 2.95391500e-01 -2.10665837e-01 1.08578064e-01
-4.91676629e-01 -6.82907477e-02 4.70902473e-01 8.42804670e-01
5.64725548e-02 -6.53015018e-01 -6.59961104e-01 2.26348504e-01
1.87264964e-01 -1.84288070e-01 -1.20735891e-01 7.15350807e-01
-7.24244058e-01 7.36828089e-01 -3.88297200e-01 -5.66339731e-01
-1.41562879e+00 6.98045611e-01 -1.59188956e-01 -7.28821814e-01
-5.17061591e-01 1.01021886e+00 -5.18429168e-02 -8.16659093e-01
4.52680029e-02 -7.04622328e-01 -2.68095136e-01 -2.77577825e-02
-3.06253582e-02 9.24476422e-03 4.51696157e-01 -7.72657931e-01
-6.26429975e-01 6.61555290e-01 -1.47193834e-01 1.14267714e-01
1.39633942e+00 1.10868782e-01 -5.78352094e-01 5.21740377e-01
1.17458248e+00 2.72390872e-01 -7.77625144e-01 -1.31271526e-01
6.44747674e-01 -1.39600113e-01 -1.39688879e-01 -7.68163681e-01
-1.08692646e+00 9.12307739e-01 3.56512159e-01 -1.42782340e-02
9.47812557e-01 2.46970028e-01 6.41174197e-01 5.74253261e-01
6.21952891e-01 -7.71982551e-01 -3.11121810e-02 1.00766754e+00
5.39025366e-01 -1.26394796e+00 -4.29936588e-01 -5.53705156e-01
-4.01613832e-01 1.16100740e+00 6.56650424e-01 -7.99658298e-02
5.66778839e-01 5.96006632e-01 1.54780015e-01 4.60045263e-02
-7.62696326e-01 -2.70192772e-01 1.09175839e-01 5.30399501e-01
1.09851015e+00 -7.78101012e-02 -5.39267838e-01 4.23512191e-01
-1.57356188e-01 -3.01985234e-01 7.36988783e-01 1.05066752e+00
-6.78869113e-02 -1.87137139e+00 -7.79261738e-02 3.44543129e-01
-6.56944573e-01 -6.48812413e-01 -1.61884829e-01 8.27281892e-01
-4.73333225e-02 1.04952049e+00 -1.45235872e-02 -2.91537046e-01
2.97107071e-01 3.28159094e-01 6.31923139e-01 -1.07857394e+00
-1.03756261e+00 -1.88200951e-01 3.91887546e-01 -5.15925288e-01
-6.43815100e-01 -4.63560373e-01 -1.39601862e+00 1.38485372e-01
-3.82850230e-01 1.11391656e-01 8.50793779e-01 1.02243042e+00
2.77970672e-01 9.46939528e-01 3.15081030e-01 -6.25501752e-01
-3.70706208e-02 -8.63858402e-01 -1.66531205e-01 2.29268163e-01
3.69141161e-01 -2.92801082e-01 -8.98908600e-02 2.61032768e-02] | [10.404730796813965, 9.822115898132324] |
c2b202ff-aef0-404c-b7d5-3f47b1a70d9c | an-adaptive-music-generation-architecture-for | 2207.01698 | null | https://arxiv.org/abs/2207.01698v2 | https://arxiv.org/pdf/2207.01698v2.pdf | An adaptive music generation architecture for games based on the deep learning Transformer mode | This paper presents an architecture for generating music for video games based on the Transformer deep learning model. Our motivation is to be able to customize the generation according to the taste of the player, who can select a corpus of training examples, corresponding to his preferred musical style. The system generates various musical layers, following the standard layering strategy currently used by composers designing video game music. To adapt the music generated to the game play and to the player(s) situation, we are using an arousal-valence model of emotions, in order to control the selection of musical layers. We discuss current limitations and prospects for the future, such as collaborative and interactive control of the musical components. | ['Antonio Luz Furtado', 'Bruno Feijó', 'Jean-Pierre Briot', 'Augusto Baffa', 'Gustavo Amaral Costa dos Santos'] | 2022-07-04 | null | null | null | null | ['music-generation', 'music-generation'] | ['audio', 'music'] | [-5.85206673e-02 3.80382761e-02 2.32745603e-01 -1.23670071e-01
4.30668630e-02 -7.95305073e-01 3.31388712e-01 -2.12442741e-01
-2.10788742e-01 1.95465103e-01 2.91239738e-01 2.33451217e-01
-2.73361892e-01 -1.14354551e+00 -7.44909346e-02 -4.77396905e-01
7.38134608e-02 6.37230277e-01 2.18661249e-01 -7.77658045e-01
3.10791850e-01 4.29147035e-01 -2.12961817e+00 1.04737234e+00
3.17420959e-01 8.89831185e-01 5.78665078e-01 9.13221657e-01
-2.14340433e-01 7.12244987e-01 -9.57610667e-01 -4.18043792e-01
2.42881164e-01 -8.65150392e-01 -5.59008241e-01 -1.74778894e-01
-2.29355618e-01 1.64041430e-01 1.57268912e-01 8.82477164e-01
6.67504311e-01 3.33235383e-01 7.08207369e-01 -9.28389966e-01
-1.89226240e-01 1.22364473e+00 4.38853115e-01 -2.60357648e-01
3.31019193e-01 8.49696100e-02 1.06281865e+00 -4.32530344e-01
7.85201788e-01 8.19160819e-01 4.01901037e-01 1.24124968e+00
-1.04339325e+00 -8.76294792e-01 2.89991051e-02 3.54904771e-01
-1.29850376e+00 -3.14714491e-01 1.03146541e+00 -4.88069713e-01
7.72453487e-01 4.84678209e-01 1.29236436e+00 1.11648679e+00
-1.48783913e-02 5.32498896e-01 8.04106295e-01 -6.11101329e-01
6.32683814e-01 3.30542475e-01 -7.05613494e-01 8.22447240e-02
-3.89142156e-01 9.30723548e-02 -8.77567649e-01 2.73027848e-02
1.14946699e+00 -8.98675740e-01 -1.06312275e-01 -4.43262935e-01
-9.22023773e-01 6.53144181e-01 6.91249371e-02 6.74781978e-01
-5.85888684e-01 1.63780004e-01 7.17847288e-01 5.07147729e-01
-1.43964551e-02 1.18016326e+00 -5.16930819e-01 -4.08678591e-01
-1.01405323e+00 6.51284516e-01 9.99909997e-01 7.10203409e-01
2.95232236e-01 3.03303272e-01 -3.89718801e-01 1.06598389e+00
4.17403400e-01 -1.75362676e-01 9.01346445e-01 -1.11753333e+00
-1.68356642e-01 3.96873534e-01 -4.66243401e-02 -5.17043531e-01
-2.93051541e-01 -5.58907509e-01 -4.97102648e-01 9.29548323e-01
1.74660504e-01 -2.37457767e-01 -5.07511497e-01 1.75757027e+00
-7.97895268e-02 -3.13903764e-02 1.38247728e-01 8.74935627e-01
8.19746196e-01 4.62919056e-01 2.88578898e-01 -1.52382284e-01
1.26934028e+00 -3.42681855e-01 -6.30992889e-01 6.59557059e-02
1.99912176e-01 -6.74485266e-01 1.30965281e+00 9.13428903e-01
-1.32984519e+00 -8.03239763e-01 -1.16161346e+00 2.84242928e-01
-1.12845585e-01 1.69646084e-01 6.74195290e-01 8.60630035e-01
-1.07766831e+00 1.04577446e+00 -3.71372521e-01 -1.62524194e-01
-5.37323765e-02 6.93736672e-01 1.61756605e-01 9.30574238e-01
-1.51661384e+00 7.51338959e-01 9.05317366e-01 4.88669574e-02
-7.24185228e-01 -5.65889955e-01 -3.44687164e-01 3.01132590e-01
-2.46612981e-01 -1.06446159e+00 1.35632789e+00 -1.64883423e+00
-2.34055018e+00 1.00279462e+00 6.94108069e-01 -2.22726226e-01
3.81676435e-01 2.56252110e-01 -3.18689018e-01 -1.26886666e-01
-2.92842269e-01 7.86726177e-01 5.77611506e-01 -1.13918936e+00
-6.45245910e-01 2.68987678e-02 3.24849099e-01 6.14959240e-01
-2.80265301e-01 4.79047485e-02 -6.24499083e-01 -8.70403290e-01
-2.26632938e-01 -7.63100922e-01 -4.19385523e-01 -5.42434573e-01
4.50674780e-02 -1.83972538e-01 1.87060446e-01 -1.65377468e-01
1.36549723e+00 -2.19958997e+00 6.17155015e-01 4.30543423e-01
-2.56534815e-01 1.31190464e-01 -1.68278918e-01 4.51096028e-01
-1.92682847e-01 -2.11911619e-01 4.79420722e-01 -1.31885275e-01
2.46180475e-01 1.23476073e-01 -2.92680770e-01 -3.94156545e-01
-1.63903594e-01 4.91240531e-01 -6.17521226e-01 -2.77968705e-01
1.44984126e-01 4.20558810e-01 -1.02751350e+00 4.82853889e-01
-6.45876348e-01 4.00723934e-01 -5.67393780e-01 3.06228362e-02
7.85930082e-02 3.82286429e-01 4.23241884e-01 -1.83448344e-01
-2.31357083e-01 5.77375054e-01 -1.52308309e+00 1.85567641e+00
-7.62621760e-01 3.20151359e-01 1.01032220e-01 -4.10677522e-01
1.44753373e+00 6.74886584e-01 4.27330464e-01 -5.54106474e-01
2.90340990e-01 1.49531782e-01 2.76792407e-01 -4.81257379e-01
7.30612397e-01 -6.34733021e-01 -1.47790998e-01 5.77832520e-01
2.30721951e-01 -4.89859313e-01 5.45400500e-01 -2.22653762e-01
6.64690614e-01 7.03339815e-01 9.32279751e-02 -1.83719113e-01
3.88787985e-01 -1.86998412e-01 3.86050880e-01 3.51706356e-01
1.77887842e-01 5.09248257e-01 7.48924613e-01 -5.14426827e-01
-1.02015102e+00 -8.43244135e-01 3.46655935e-01 1.63460124e+00
-2.91166008e-01 -6.92339003e-01 -8.99184406e-01 7.35245124e-02
-7.35009074e-01 8.81070137e-01 -3.77560884e-01 -1.75543278e-01
-6.06023967e-01 -5.22345960e-01 4.95008945e-01 2.53653795e-01
-3.18664052e-02 -2.07570720e+00 -9.22783256e-01 5.87238312e-01
-5.73938563e-02 -4.68121201e-01 -6.92542642e-02 4.48066890e-01
-8.04423749e-01 -7.49852777e-01 -1.70762062e-01 -8.95375550e-01
-8.19178522e-02 -7.80376077e-01 1.32033682e+00 -1.37000844e-01
9.48522314e-02 3.56863648e-01 -5.48925221e-01 -5.16083539e-01
-7.62939095e-01 4.69105065e-01 -1.43217221e-01 -1.04777113e-01
1.20603040e-01 -8.83746028e-01 -4.90961641e-01 -8.87290314e-02
-1.13383889e+00 4.39704508e-01 3.87144476e-01 2.85126776e-01
6.62779391e-01 3.32381934e-01 4.48103130e-01 -8.03035736e-01
1.16865468e+00 -3.24105844e-02 -3.47265303e-01 -7.88367353e-03
-3.82754982e-01 -1.32378433e-02 6.61712348e-01 -6.93195403e-01
-1.06276786e+00 2.83217609e-01 -6.64038599e-01 -2.65044630e-01
-2.38534197e-01 2.87792623e-01 -4.33741719e-01 1.42336607e-01
8.16962361e-01 -4.22584005e-02 -3.16198051e-01 -1.93939433e-01
4.26553965e-01 5.95011234e-01 5.63240707e-01 -7.05817997e-01
4.53558803e-01 -2.16420010e-01 -2.42732048e-01 -4.67039257e-01
-3.17981780e-01 -2.37529236e-03 -6.98001981e-01 -6.89390838e-01
9.31920528e-01 -6.80885077e-01 -8.92967224e-01 2.90573835e-01
-1.04688144e+00 -7.83354938e-01 -7.49929428e-01 3.25708240e-01
-1.10621369e+00 -1.99446395e-01 -6.74340308e-01 -8.01904917e-01
-5.34009218e-01 -1.13205862e+00 6.74945593e-01 3.66498828e-01
-8.38342249e-01 -7.61505783e-01 4.14502203e-01 5.89402355e-02
3.99220645e-01 1.59852833e-01 9.48557258e-01 -3.92001420e-01
-3.37123573e-01 -2.43211333e-02 5.70824623e-01 3.66749167e-01
-2.17769191e-01 2.79825509e-01 -9.79745209e-01 2.85030723e-01
-1.31012909e-02 -2.95781851e-01 4.47868764e-01 3.18288356e-01
1.19824803e+00 1.16174437e-01 2.28056654e-01 6.71507120e-01
1.20436537e+00 4.48321432e-01 1.04992187e+00 5.70093751e-01
3.75953227e-01 6.07349455e-01 4.37236369e-01 6.06284738e-01
8.26510936e-02 1.00302911e+00 4.49103117e-01 2.55115032e-01
-6.16107211e-02 -3.08643699e-01 5.20248353e-01 5.79101205e-01
-6.10310674e-01 3.68658565e-02 -3.57460499e-01 1.49352392e-02
-1.72920263e+00 -1.44688451e+00 1.43218204e-01 2.17929769e+00
9.28774834e-01 2.70186931e-01 4.16995376e-01 4.34922367e-01
5.11115730e-01 5.45430183e-02 -1.13357566e-01 -1.04409897e+00
4.74221967e-02 9.31447327e-01 -5.41799366e-02 3.94806832e-01
-8.41844320e-01 1.14210188e+00 6.95131350e+00 8.11215341e-01
-1.40103889e+00 -2.77273178e-01 2.56953537e-01 -5.25297284e-01
-4.50235099e-01 -3.19086522e-01 -4.96067852e-01 2.63443828e-01
9.18230832e-01 -2.40668535e-01 9.80787337e-01 7.61817336e-01
3.88343304e-01 4.73395646e-01 -1.08376527e+00 8.53074789e-01
5.73179424e-02 -1.42033911e+00 1.70799479e-01 -2.98945367e-01
4.32466090e-01 -2.52717435e-01 8.84661451e-02 5.16870975e-01
5.36077879e-02 -1.01296782e+00 1.15272593e+00 5.90438604e-01
7.16340959e-01 -1.00485647e+00 5.32965481e-01 1.23364016e-01
-1.15847957e+00 -6.64748773e-02 -7.93451536e-03 -4.34450001e-01
-1.90222859e-02 -1.40643880e-01 -6.78708315e-01 7.47587606e-02
8.24543238e-01 1.40616655e-01 -3.05871546e-01 9.48073149e-01
-4.26410407e-01 2.85147458e-01 5.26208337e-03 -4.22788203e-01
-1.16487980e-01 -3.14419299e-01 5.42355299e-01 1.28818512e+00
5.97653091e-01 2.13038594e-01 -1.74803615e-01 1.07060206e+00
3.35812896e-01 6.26422763e-01 -2.73516208e-01 -1.58662662e-01
3.26061845e-01 1.33469117e+00 -7.45319247e-01 -1.99591711e-01
2.35233992e-01 1.15386343e+00 -3.45839411e-02 1.55590639e-01
-5.07898986e-01 -4.72018003e-01 7.38611281e-01 2.57228464e-01
3.68055850e-01 -2.75973082e-02 -4.63083714e-01 -9.41820979e-01
-3.54273528e-01 -1.13395345e+00 3.11134219e-01 -1.07844639e+00
-9.17913139e-01 9.61522579e-01 -2.30804771e-01 -1.25028324e+00
-5.98454773e-01 -7.89736152e-01 -9.16748703e-01 8.89543474e-01
-5.72595954e-01 -1.08463073e+00 -1.41637595e-02 5.75339437e-01
3.96366239e-01 -6.21405184e-01 1.30253613e+00 2.59382039e-01
2.42088046e-02 2.02000633e-01 -5.61759055e-01 -1.18481778e-01
6.57470942e-01 -1.36749506e+00 1.96932983e-02 1.61967725e-01
5.52618921e-01 3.35512400e-01 1.11692691e+00 -1.64622009e-01
-8.70649397e-01 -6.34892941e-01 9.46206748e-01 -6.86572716e-02
5.79826534e-01 -3.80901486e-01 -6.30013406e-01 1.54098690e-01
3.22448850e-01 -8.60219955e-01 1.20106149e+00 3.41552198e-01
-1.12389833e-01 -3.69483441e-01 -8.87308478e-01 8.62857103e-01
8.46557736e-01 -4.86328065e-01 -4.31928784e-01 -1.04384907e-01
1.37278661e-01 -5.25813639e-01 -7.97739506e-01 2.17051506e-01
8.13410342e-01 -1.37837720e+00 7.25056589e-01 -4.97448057e-01
4.12392467e-01 -5.04404604e-01 7.11035803e-02 -1.57643223e+00
-7.13965476e-01 -7.59233773e-01 3.47424060e-01 1.25061798e+00
5.70645750e-01 1.77778244e-01 1.14192855e+00 4.17487442e-01
-2.08440393e-01 -3.93388271e-01 -7.73620963e-01 1.14395386e-02
-2.15898268e-02 -8.55585515e-01 8.91473055e-01 6.47738874e-01
4.72528696e-01 5.39102256e-01 -3.88130218e-01 -1.13042876e-01
-3.32997702e-02 6.85811639e-02 7.60652959e-01 -1.48124409e+00
-1.00982392e+00 -1.16415715e+00 -4.11421895e-01 -3.08806211e-01
7.44857490e-02 -8.50014985e-01 2.97372490e-02 -1.37328160e+00
-1.70053110e-01 -3.62560838e-01 -6.65634036e-01 2.48839781e-01
1.31190389e-01 7.40388751e-01 5.38853049e-01 -2.21820235e-01
-4.96245235e-01 3.40303868e-01 1.25435209e+00 8.33787546e-02
-8.43814075e-01 3.61059934e-01 -8.61464024e-01 6.74591482e-01
1.01027298e+00 -1.27881527e-01 -5.56433499e-01 -9.07977819e-02
1.01168430e+00 -8.66080355e-03 -1.36902258e-01 -1.21670449e+00
9.70360488e-02 -2.99934715e-01 5.34263611e-01 -1.67646348e-01
3.35739523e-01 -6.64171636e-01 7.34407663e-01 4.64269191e-01
-6.91298544e-01 -1.42449260e-01 2.45514408e-01 -1.93287432e-01
-2.84214616e-01 -6.08168721e-01 7.62356102e-01 -3.30971301e-01
-7.43715823e-01 4.60143201e-02 -8.20689678e-01 -4.12645400e-01
8.03407967e-01 -3.25966060e-01 4.46274757e-01 -5.44558704e-01
-1.34778583e+00 -3.19719553e-01 3.23052585e-01 5.86495221e-01
3.57232064e-01 -1.55061507e+00 -5.78790784e-01 3.66766453e-01
2.37717986e-01 -6.17171228e-01 3.08295727e-01 -9.94420946e-02
-7.30529726e-01 -1.37900352e-01 -8.30964684e-01 7.78973103e-02
-1.36045218e+00 4.16272849e-01 6.85843706e-01 -3.08368951e-01
-3.10698360e-01 6.97487891e-01 6.45496184e-03 -2.43866235e-01
2.06504658e-01 -1.97321072e-01 -8.96174729e-01 3.47352028e-01
5.66936612e-01 1.09430008e-01 1.08152129e-01 -2.61286914e-01
9.87572595e-02 4.01584387e-01 4.40116465e-01 -5.01274705e-01
1.25048852e+00 3.82012695e-01 -1.97235540e-01 7.06327736e-01
2.18384966e-01 1.49048910e-01 -8.86538148e-01 4.99278396e-01
-1.55825913e-01 3.11358809e-03 -8.20509121e-02 -9.82786536e-01
-1.21136773e+00 4.79906708e-01 6.01797402e-01 3.21347833e-01
1.45320487e+00 -2.10555419e-01 3.34222108e-01 2.44769111e-01
4.94087726e-01 -1.47297144e+00 3.00552901e-02 6.33783638e-01
1.09004128e+00 -2.41601571e-01 -3.99729401e-01 6.00750893e-02
-1.03960478e+00 1.52598059e+00 5.55376470e-01 -3.54812950e-01
5.66336334e-01 4.66623664e-01 5.43617189e-01 -1.57319739e-01
-1.08786988e+00 -1.97547883e-01 3.96264553e-01 8.42543662e-01
9.79679048e-01 3.49147946e-01 -6.42232001e-01 9.21760499e-01
-9.65462029e-01 3.16893429e-01 5.17944753e-01 6.70348555e-02
-4.06524211e-01 -1.85447633e+00 -2.38185033e-01 1.52970105e-01
-5.05103648e-01 1.62949964e-01 -6.56317115e-01 3.85052472e-01
1.03920829e+00 7.18219101e-01 2.22079426e-01 -5.53909123e-01
6.58206701e-01 2.13054657e-01 7.85886288e-01 -9.59317207e-01
-1.33387256e+00 4.18333322e-01 2.49685884e-01 -3.81796479e-01
-2.73057938e-01 -5.76967776e-01 -1.31825006e+00 7.23516569e-02
2.40518004e-01 3.65677655e-01 5.79622090e-01 6.10522926e-01
-3.54556404e-02 8.79688621e-01 3.52561474e-01 -1.31803024e+00
8.10585544e-02 -9.63632822e-01 -9.21071112e-01 4.15909737e-01
-5.65151930e-01 -2.72363305e-01 8.77587944e-02 2.11950660e-01] | [16.00882339477539, 5.504275321960449] |
dcf3d31b-a172-46c6-a36a-5591d7fcafe0 | linear-video-transformer-with-feature | 2210.08164 | null | https://arxiv.org/abs/2210.08164v1 | https://arxiv.org/pdf/2210.08164v1.pdf | Linear Video Transformer with Feature Fixation | Vision Transformers have achieved impressive performance in video classification, while suffering from the quadratic complexity caused by the Softmax attention mechanism. Some studies alleviate the computational costs by reducing the number of tokens in attention calculation, but the complexity is still quadratic. Another promising way is to replace Softmax attention with linear attention, which owns linear complexity but presents a clear performance drop. We find that such a drop in linear attention results from the lack of attention concentration on critical features. Therefore, we propose a feature fixation module to reweight the feature importance of the query and key before computing linear attention. Specifically, we regard the query, key, and value as various latent representations of the input token, and learn the feature fixation ratio by aggregating Query-Key-Value information. This is beneficial for measuring the feature importance comprehensively. Furthermore, we enhance the feature fixation by neighborhood association, which leverages additional guidance from spatial and temporal neighbouring tokens. The proposed method significantly improves the linear attention baseline and achieves state-of-the-art performance among linear video Transformers on three popular video classification benchmarks. With fewer parameters and higher efficiency, our performance is even comparable to some Softmax-based quadratic Transformers. | ['Yiran Zhong', 'Yuchao Dai', 'Xiaodong Han', 'Hui Deng', 'Xuyang Shen', 'Dong Li', 'Zhen Qin', 'Weixuan Sun', 'Jianyuan Wang', 'Zexiang Liu', 'Kaiyue Lu'] | 2022-10-15 | null | null | null | null | ['video-classification'] | ['computer-vision'] | [ 6.84244484e-02 -2.72714943e-01 -4.16960686e-01 -2.68586040e-01
-7.57971227e-01 -2.32848257e-01 5.81746459e-01 2.23935336e-01
-7.89864361e-01 3.33512783e-01 4.77377892e-01 -3.58267650e-02
-8.27643499e-02 -6.10820651e-01 -7.93292940e-01 -9.00674999e-01
8.63933191e-02 -2.15630233e-01 4.40427393e-01 7.34684691e-02
4.63678956e-01 6.32761642e-02 -1.55712903e+00 2.99506426e-01
7.87438929e-01 1.45661247e+00 4.57774788e-01 4.54826444e-01
-8.94804224e-02 1.00021887e+00 -6.33942604e-01 -3.72874975e-01
1.16081975e-01 -5.59789091e-02 -7.15109110e-01 -1.30639642e-01
8.20564806e-01 -5.28760850e-01 -6.90259337e-01 1.03433239e+00
4.92633522e-01 2.44724259e-01 4.70044166e-01 -1.27053344e+00
-1.00965822e+00 6.42118812e-01 -7.77068198e-01 5.88543832e-01
4.17598821e-02 2.35082164e-01 1.45833647e+00 -1.02139556e+00
4.92143370e-02 1.15204787e+00 5.30991912e-01 1.71305344e-01
-8.40257764e-01 -5.22624850e-01 6.37041867e-01 7.76469350e-01
-1.38978553e+00 -4.83002335e-01 5.40926516e-01 -3.60664815e-01
1.18258715e+00 3.25860322e-01 6.25693083e-01 8.11769247e-01
-3.93115729e-02 9.76881742e-01 5.43144107e-01 -2.70470172e-01
3.13739181e-02 -4.17575054e-02 3.09522092e-01 8.16910684e-01
-6.97927624e-02 -2.69752145e-01 -6.88248694e-01 1.56819299e-01
7.10077465e-01 4.28286314e-01 -4.80503470e-01 -3.91438566e-02
-1.24065316e+00 7.17453420e-01 7.87617683e-01 1.45283878e-01
-5.68298578e-01 4.64161575e-01 3.91863346e-01 2.04957455e-01
3.71189862e-01 2.72063941e-01 -5.41273832e-01 -4.30555552e-01
-9.46252406e-01 -5.19898273e-02 1.26042038e-01 8.87260139e-01
7.27971375e-01 3.64429876e-02 -9.83217716e-01 8.33433926e-01
2.77457982e-01 4.07484889e-01 5.33109486e-01 -8.50723088e-01
5.92305958e-01 5.81039727e-01 -1.40600994e-01 -9.35207546e-01
-1.63861930e-01 -5.71530759e-01 -6.21550500e-01 3.49645577e-02
3.76794666e-01 6.04298040e-02 -1.06742561e+00 1.82461953e+00
-3.23974267e-02 1.62584618e-01 -2.69484282e-01 1.06116724e+00
8.53539824e-01 6.08807564e-01 1.82839736e-01 -6.87067807e-02
1.51996875e+00 -1.37136602e+00 -6.38622642e-01 -1.63124993e-01
3.65582466e-01 -6.88871086e-01 1.32371020e+00 1.38780460e-01
-1.18625224e+00 -6.39133751e-01 -9.23507154e-01 -4.24770027e-01
-2.37673610e-01 1.57097787e-01 7.82971382e-01 3.40459555e-01
-9.50156629e-01 4.50380266e-01 -8.28434944e-01 2.44860835e-02
6.84081137e-01 4.53006119e-01 -7.58745670e-02 -2.79265475e-02
-1.16649091e+00 7.13967562e-01 8.62211436e-02 2.33213678e-02
-6.92510307e-01 -8.93593132e-01 -9.51238632e-01 6.50199115e-01
5.82878828e-01 -5.70206642e-01 1.12709570e+00 -8.43886137e-01
-1.27571702e+00 3.49822164e-01 -5.49110472e-01 -3.62368494e-01
3.18906456e-01 -3.93304408e-01 1.53309524e-01 1.06707990e-01
7.60020241e-02 8.75036836e-01 1.04886699e+00 -6.71288073e-01
-7.94141829e-01 -1.57623157e-01 5.14010072e-01 1.63929000e-01
-1.04416919e+00 -1.25888009e-02 -1.10357332e+00 -8.75586510e-01
-5.59422337e-02 -6.93365574e-01 -7.86701143e-02 1.34255901e-01
-1.79955304e-01 -5.17286897e-01 6.76518679e-01 -5.34030259e-01
1.56512487e+00 -2.28682590e+00 2.09463269e-01 -5.96302301e-02
4.60728794e-01 2.39346430e-01 -1.50112078e-01 3.68784182e-02
8.71728137e-02 8.48416314e-02 4.43672836e-02 -3.54667515e-01
6.92097470e-02 -1.31013840e-02 -1.99119970e-01 2.25483403e-01
4.67536718e-01 1.23862362e+00 -8.97208035e-01 -4.63843554e-01
1.63872957e-01 5.30361056e-01 -9.54563916e-01 4.18698415e-02
-6.58880472e-02 -1.51833400e-01 -3.49734396e-01 7.33028650e-01
4.78891790e-01 -5.42565703e-01 -2.07327932e-01 -5.61301172e-01
-1.71737760e-01 3.86822820e-01 -7.52239287e-01 1.41391921e+00
-3.78043234e-01 9.30106997e-01 -1.93673104e-01 -1.17135274e+00
3.93600196e-01 1.22037895e-01 4.57721055e-01 -9.51061547e-01
7.58156404e-02 -2.11309776e-01 1.27274748e-02 -7.43634343e-01
5.87214231e-01 3.21952283e-01 8.50776285e-02 2.68605918e-01
8.67245719e-02 5.28696299e-01 1.63898647e-01 2.25266591e-01
9.54103768e-01 -1.15719074e-02 8.48285779e-02 -1.35763347e-01
3.86339962e-01 -4.99740869e-01 5.50977290e-01 7.68901885e-01
-2.31934473e-01 4.66563076e-01 7.09872067e-01 -2.22761750e-01
-7.27621436e-01 -8.77606869e-01 1.36269599e-01 1.66658795e+00
3.07618648e-01 -6.70984805e-01 -7.23769546e-01 -7.25122571e-01
7.32360259e-02 2.92589188e-01 -9.17371571e-01 -4.23302174e-01
-6.23591602e-01 -6.88744962e-01 4.56770688e-01 8.99656951e-01
4.41824764e-01 -9.59428370e-01 -6.36607707e-01 7.15597812e-03
-3.60197425e-01 -1.04276264e+00 -1.04004061e+00 2.27280423e-01
-5.46893835e-01 -8.73429298e-01 -9.49403405e-01 -7.68670738e-01
6.88133776e-01 6.40284657e-01 9.59829390e-01 1.18127041e-01
-5.19325808e-02 2.34216824e-01 -3.83635074e-01 -3.11077744e-01
4.43147004e-01 2.40860656e-01 -6.23195246e-02 1.46267876e-01
3.31083387e-01 -2.23718584e-01 -8.42854023e-01 2.32670322e-01
-8.24733734e-01 -1.34827331e-01 7.11538553e-01 1.01500916e+00
3.57118040e-01 -5.91928959e-02 3.96591961e-01 -2.56296068e-01
5.24821281e-01 -4.31137919e-01 -3.49601209e-01 3.79072696e-01
-4.01392311e-01 3.43010545e-01 5.14462292e-01 -6.44068360e-01
-8.16508412e-01 -3.06322187e-01 -6.32404862e-03 -7.28426397e-01
3.09937298e-01 6.28110230e-01 -3.73263657e-02 -2.49576028e-02
1.94933087e-01 2.23258838e-01 -3.00486326e-01 -3.52433980e-01
2.23784342e-01 5.71338356e-01 3.82233709e-01 -4.20468211e-01
5.31298161e-01 3.17344099e-01 -2.70723373e-01 -8.53822708e-01
-9.19146538e-01 -4.31263447e-01 -3.02996904e-01 -1.30015135e-01
8.39411139e-01 -8.11937809e-01 -1.00040209e+00 5.53972125e-01
-1.08310413e+00 -1.91152394e-01 -1.74914405e-01 6.30325198e-01
-1.28784627e-01 4.39950705e-01 -6.62717760e-01 -5.70578098e-01
-3.81128609e-01 -1.43791080e+00 1.08698070e+00 2.87310839e-01
-7.97285326e-03 -6.19980574e-01 -5.62637627e-01 3.33479166e-01
7.72270322e-01 -2.59888917e-01 8.56786072e-01 -1.24519914e-01
-8.52597475e-01 -2.67806947e-02 -7.26666033e-01 2.93447614e-01
2.59710968e-01 -2.09909659e-02 -1.03153932e+00 -4.08062339e-01
-2.63021916e-01 -2.58912295e-01 1.43665600e+00 6.95878685e-01
1.42527831e+00 -4.04003024e-01 -1.90873444e-01 7.11336911e-01
1.00465679e+00 9.12872702e-02 4.14421439e-01 2.52399296e-01
9.77497220e-01 3.33904624e-01 5.33235550e-01 5.37043631e-01
6.18433714e-01 8.78295720e-01 6.01883888e-01 -3.08389276e-01
-1.65366769e-01 1.07997194e-01 5.11456132e-01 7.74465501e-01
-2.18574435e-01 -1.92099884e-01 -5.88764787e-01 6.92657471e-01
-1.91793716e+00 -1.08674359e+00 2.33200341e-01 2.21770239e+00
7.63399005e-01 4.57936943e-01 1.00365169e-01 1.08179577e-01
5.46037972e-01 2.44893968e-01 -6.14295125e-01 -1.75300464e-01
2.06124932e-02 1.89738929e-01 6.14858866e-01 4.63254273e-01
-1.12071931e+00 9.44794834e-01 6.24804163e+00 9.80808258e-01
-1.50950813e+00 6.51569813e-02 5.74024618e-01 -5.87547660e-01
-8.95987451e-02 -3.51629704e-01 -7.51670063e-01 6.30487859e-01
6.11138403e-01 -1.42679855e-01 4.66751665e-01 6.88098729e-01
2.75456607e-01 -6.59507960e-02 -1.11168146e+00 1.04412723e+00
2.28237584e-01 -1.28208053e+00 1.98767915e-01 1.12792768e-01
4.73647118e-01 1.69197261e-01 4.25623298e-01 4.86577958e-01
-2.42541671e-01 -9.07299221e-01 9.01359200e-01 5.45963228e-01
8.16448510e-01 -6.34457946e-01 6.80643201e-01 -3.79233994e-02
-1.46318245e+00 -4.56310451e-01 -2.63521761e-01 -1.31797835e-01
2.84386072e-02 3.75093907e-01 -3.53971928e-01 2.00492188e-01
9.83982861e-01 7.74062753e-01 -7.21907556e-01 1.21137667e+00
-6.49984181e-02 6.62722528e-01 -1.55214712e-01 -1.33466989e-01
5.46869755e-01 1.60602972e-01 2.74482429e-01 1.29740036e+00
1.95381194e-01 -9.21656936e-02 2.07404494e-01 6.26014769e-01
-3.75904799e-01 2.44817454e-02 -5.43241650e-02 -5.86603954e-02
3.89961839e-01 1.06193399e+00 -6.05735123e-01 -5.84841192e-01
-7.60660350e-01 9.37673628e-01 5.00412524e-01 3.72826487e-01
-1.17841709e+00 -6.85195506e-01 8.87160301e-01 1.42191537e-02
7.75076687e-01 -1.21104956e-01 -1.51650652e-01 -1.05190897e+00
4.43658829e-01 -6.19254708e-01 2.26101801e-01 -6.34441614e-01
-9.40170228e-01 5.77737093e-01 -4.20829505e-02 -1.20848334e+00
7.66339824e-02 -5.02263844e-01 -5.39119840e-01 7.12356627e-01
-1.63731050e+00 -1.08921659e+00 -5.08204937e-01 5.98232985e-01
8.15802276e-01 5.33025637e-02 3.27968031e-01 7.28723228e-01
-7.08639741e-01 1.20413554e+00 -1.51094824e-01 9.03682411e-02
7.94975102e-01 -1.24361825e+00 3.96221966e-01 7.40564048e-01
3.72891836e-02 7.77909517e-01 3.27793211e-01 -2.09607691e-01
-1.20313263e+00 -1.04326165e+00 9.08871591e-01 -3.17843318e-01
5.56880355e-01 -2.35743061e-01 -9.61634517e-01 4.13886815e-01
1.92986950e-01 1.71519950e-01 3.54732871e-01 9.46947709e-02
-4.63849634e-01 -2.16372088e-01 -6.89207673e-01 6.83445930e-01
1.07231462e+00 -6.32029891e-01 -5.13205767e-01 4.46860418e-02
1.09903431e+00 -1.30577087e-01 -6.97271347e-01 3.04208010e-01
8.23955834e-01 -7.54728675e-01 1.00262129e+00 -4.20872003e-01
4.91591781e-01 -3.21536839e-01 -1.02474324e-01 -1.08937800e+00
-7.57258356e-01 -3.59825015e-01 -4.27231610e-01 1.24633420e+00
3.38238567e-01 -4.91844147e-01 7.08354354e-01 4.99858022e-01
-2.20142290e-01 -1.08331299e+00 -8.64528239e-01 -5.76738417e-01
-2.60759115e-01 -3.28426033e-01 4.95690107e-01 8.61292779e-01
8.06262270e-02 4.01062399e-01 -5.19319713e-01 4.33367752e-02
3.83917004e-01 -5.80441542e-02 3.11202645e-01 -9.02778268e-01
-4.13146645e-01 -9.02924657e-01 -4.56340939e-01 -1.59431422e+00
-5.12361564e-02 -7.78192043e-01 -2.23812871e-02 -1.31784236e+00
4.44111437e-01 -2.61204273e-01 -8.96539629e-01 7.88415432e-01
-6.65362716e-01 4.35457498e-01 3.39673996e-01 1.57371268e-01
-9.08275127e-01 6.82420492e-01 1.23440492e+00 -3.84224027e-01
-1.79743752e-01 -2.26157561e-01 -8.80033255e-01 6.27066314e-01
5.47076285e-01 -2.92924613e-01 -3.39088738e-01 -9.36403513e-01
7.81011954e-02 -3.35856795e-01 3.19249809e-01 -8.98508251e-01
4.65875924e-01 -1.18923679e-01 4.03065771e-01 -4.93506789e-01
4.33870226e-01 -6.42936289e-01 -5.48370898e-01 2.70941764e-01
-5.77342570e-01 1.72220916e-01 1.26676455e-01 4.88390177e-01
-3.00875843e-01 -6.77897483e-02 6.35090172e-01 6.33006543e-02
-6.92052186e-01 5.03992617e-01 -4.10962462e-01 -6.31205887e-02
7.79575765e-01 -2.67354339e-01 -3.59483451e-01 -3.01679581e-01
-4.58811045e-01 4.08327490e-01 1.73309043e-01 7.00296819e-01
5.82625270e-01 -1.34385955e+00 -5.92968941e-01 2.64364451e-01
1.44905180e-01 -1.28257781e-01 2.41551578e-01 1.10274291e+00
-5.38034588e-02 5.63345313e-01 -1.08362012e-01 -6.75673306e-01
-1.41068828e+00 6.26609206e-01 8.92506987e-02 -1.58017084e-01
-4.05785412e-01 1.07300508e+00 4.73274648e-01 3.71535540e-01
6.75266266e-01 -6.82252824e-01 -3.38516235e-01 2.71742225e-01
6.74786866e-01 4.11421686e-01 -8.57287366e-03 -6.70693517e-01
-4.37281907e-01 7.29499698e-01 -3.45213234e-01 1.85375631e-01
1.23185146e+00 -6.53379336e-02 9.31226388e-02 9.51025113e-02
1.25502610e+00 1.85297318e-02 -1.49124181e+00 -4.58431095e-01
-1.26814246e-01 -5.62425017e-01 3.82578194e-01 -4.44430262e-01
-1.45912540e+00 8.88169110e-01 5.37164807e-01 1.69193223e-01
1.25004983e+00 2.37470604e-02 8.42038691e-01 4.24915135e-01
-7.31391162e-02 -9.01188552e-01 2.45490491e-01 5.88413894e-01
7.89811969e-01 -1.36518943e+00 -6.10712096e-02 -2.17378750e-01
-5.26451886e-01 8.18863094e-01 8.26944530e-01 9.97383967e-02
5.31876922e-01 1.71006456e-01 -1.36350304e-01 -1.38679564e-01
-8.93596053e-01 -3.25398266e-01 5.01006305e-01 2.23584712e-01
5.47541022e-01 -2.07471669e-01 -1.42277673e-01 6.07304335e-01
2.79507060e-02 -1.60177946e-01 9.02779400e-02 8.12532187e-01
-5.03556073e-01 -7.07237363e-01 -1.26260459e-01 6.91261768e-01
-7.36744106e-01 -5.64167917e-01 1.33246661e-03 3.77354831e-01
1.58144191e-01 8.92213941e-01 3.09386283e-01 -3.91804606e-01
3.65763068e-01 -2.73609072e-01 4.42552030e-01 -2.30534136e-01
-6.26214802e-01 6.18446097e-02 -2.53821582e-01 -6.21803939e-01
-2.98611552e-01 -4.36536521e-01 -1.02168643e+00 -2.00717852e-01
-5.21566212e-01 5.35085723e-02 3.30088764e-01 9.55956638e-01
4.34254259e-01 8.26177239e-01 5.78031361e-01 -1.02530956e+00
-5.09664714e-01 -1.03114748e+00 3.11331023e-02 3.72175127e-01
6.97243571e-01 -9.06814933e-01 -2.08408505e-01 2.47461293e-02] | [9.364638328552246, 0.7212293744087219] |
c8676bcd-7b5a-4df1-9b5a-6ed39ff7e685 | explicit-sentence-compression-for-neural | 1912.11980 | null | https://arxiv.org/abs/1912.11980v1 | https://arxiv.org/pdf/1912.11980v1.pdf | Explicit Sentence Compression for Neural Machine Translation | State-of-the-art Transformer-based neural machine translation (NMT) systems still follow a standard encoder-decoder framework, in which source sentence representation can be well done by an encoder with self-attention mechanism. Though Transformer-based encoder may effectively capture general information in its resulting source sentence representation, the backbone information, which stands for the gist of a sentence, is not specifically focused on. In this paper, we propose an explicit sentence compression method to enhance the source sentence representation for NMT. In practice, an explicit sentence compression goal used to learn the backbone information in a sentence. We propose three ways, including backbone source-side fusion, target-side fusion, and both-side fusion, to integrate the compressed sentence into NMT. Our empirical tests on the WMT English-to-French and English-to-German translation tasks show that the proposed sentence compression method significantly improves the translation performances over strong baselines. | ['Kehai Chen', 'Zuchao Li', 'Rui Wang', 'Masao Utiyama', 'Zhuosheng Zhang', 'Eiichiro Sumita', 'Hai Zhao'] | 2019-12-27 | null | null | null | null | ['sentence-compression'] | ['natural-language-processing'] | [ 6.4788306e-01 1.4404322e-01 -3.4541067e-01 -5.1377910e-01
-1.3103414e+00 -2.2079027e-01 6.5749103e-01 -9.3145706e-02
-3.9184532e-01 1.0132521e+00 7.2163057e-01 -6.9949001e-01
4.7400677e-01 -7.0804930e-01 -1.0396401e+00 -4.2842135e-01
6.1907965e-01 4.9867678e-01 -2.0457363e-01 -5.6998116e-01
1.5928791e-01 -2.7082920e-01 -8.0906862e-01 6.8233413e-01
1.1774069e+00 6.9536287e-01 6.9954681e-01 5.6723040e-01
-5.1740795e-01 7.3786545e-01 -8.0669379e-01 -8.1108516e-01
4.8980992e-02 -9.6280241e-01 -8.9868599e-01 -1.4414583e-01
2.2355825e-01 -3.5604843e-01 -4.2622137e-01 1.1646038e+00
5.6349832e-01 -2.2711073e-01 6.5560436e-01 -6.3126045e-01
-1.1061748e+00 1.1846186e+00 -6.4966470e-01 3.1594995e-01
2.5432506e-01 -8.1843354e-02 9.3112165e-01 -1.0746069e+00
5.9858668e-01 1.3568980e+00 2.4668948e-01 5.0842124e-01
-9.9643105e-01 -5.3515750e-01 6.8421140e-02 2.2316825e-01
-1.1393374e+00 -6.7605877e-01 6.1804837e-01 7.4161135e-02
1.4239669e+00 2.9861116e-01 5.5323893e-01 1.1254399e+00
7.5391412e-01 9.5407915e-01 9.5687419e-01 -6.6632038e-01
-1.2148134e-01 7.6339744e-02 -2.0148583e-01 4.9466944e-01
-2.2890016e-02 -8.0548115e-02 -6.6516501e-01 2.5721815e-01
6.7815852e-01 -1.6572292e-01 -2.5522640e-01 2.8244776e-01
-1.3357927e+00 7.6929855e-01 3.4605566e-01 3.2375255e-01
-4.5856175e-01 5.9203353e-02 6.8562490e-01 6.9756317e-01
8.6817449e-01 7.5147174e-02 -5.1590496e-01 -3.6996913e-01
-1.1483145e+00 1.0696032e-01 5.1675558e-01 1.2601506e+00
6.0202098e-01 5.9911758e-02 -6.8291664e-01 8.9863628e-01
1.5788056e-01 7.7782720e-01 9.3350393e-01 -5.3393501e-01
1.3190643e+00 4.1881800e-01 -3.0866334e-01 -4.4826880e-01
3.7095785e-01 -7.9877454e-01 -1.2374817e+00 -5.2969903e-01
-3.4070268e-01 -2.9234123e-01 -9.8082548e-01 1.8523078e+00
-1.2916075e-01 -1.3065998e-01 4.2327842e-01 7.7307862e-01
7.0367056e-01 1.1023611e+00 -3.8092139e-01 -5.2359849e-01
1.2051538e+00 -1.4247599e+00 -1.0255693e+00 -5.1104528e-01
7.6197672e-01 -9.7181463e-01 1.0193051e+00 -2.9497400e-01
-1.5267792e+00 -6.3290787e-01 -1.0534822e+00 -4.3874016e-01
-9.7655736e-02 2.8276527e-01 1.7597096e-01 1.6172962e-01
-1.0090903e+00 5.3938037e-01 -7.4122065e-01 -1.8545032e-01
2.7828744e-01 1.7078845e-01 -3.9682069e-01 -2.2304924e-01
-1.5391585e+00 1.2950181e+00 5.2296644e-01 6.7058183e-02
-8.4352320e-01 -5.1243681e-01 -9.4435912e-01 2.5135779e-01
-2.2717372e-02 -1.1958143e+00 1.5679411e+00 -9.5233560e-01
-1.6381675e+00 5.8902055e-01 -7.6800632e-01 -7.4518806e-01
1.7898126e-01 -1.8969387e-01 -1.3398366e-01 9.6710838e-02
3.3289754e-01 7.4299115e-01 8.9056253e-01 -6.2054664e-01
-5.4961097e-01 -1.5489787e-01 -7.0578188e-02 6.7403704e-01
-5.1037842e-01 2.6699916e-01 -3.9140072e-01 -8.1531858e-01
-4.4157665e-02 -5.9156162e-01 -8.7815456e-02 -4.1469732e-01
-5.0527281e-01 -1.3594657e-01 7.0378447e-01 -1.0085067e+00
1.4334384e+00 -1.8035349e+00 5.7763618e-01 -4.6745193e-01
-1.8753314e-01 3.0338711e-01 -3.6231372e-01 8.0951500e-01
-3.1371068e-02 2.5570953e-02 -4.5987082e-01 -7.3422146e-01
3.9239228e-03 2.4479397e-01 -6.3119733e-01 1.3885789e-02
3.5400358e-01 1.3473622e+00 -8.1205577e-01 -6.8353647e-01
7.3787391e-02 5.0768638e-01 -3.6795485e-01 2.5814342e-01
-1.8498240e-01 3.2495514e-01 -2.9708874e-01 3.3895722e-01
4.7800562e-01 -3.4641311e-02 -1.1780394e-01 -9.7485669e-02
2.2930974e-02 1.1012740e+00 -1.3989352e-01 2.3254812e+00
-6.4708799e-01 6.3853139e-01 -6.6524558e-02 -8.9020538e-01
9.8240924e-01 5.9985042e-01 -3.3748001e-03 -8.0229664e-01
1.5694122e-01 3.9324108e-01 8.2046062e-02 -2.8090200e-01
7.4169779e-01 -3.6481628e-01 2.0052304e-02 6.1597091e-01
2.7895632e-01 -1.3227087e-01 2.8023371e-01 2.2984727e-01
8.7966841e-01 2.7437752e-01 1.6481712e-01 -1.9105078e-01
4.8835886e-01 8.8980854e-02 4.6461549e-01 2.9101843e-01
2.2483949e-01 7.2765416e-01 2.0581347e-01 1.1891068e-02
-1.4250869e+00 -7.7652687e-01 1.8516771e-01 7.8861254e-01
-1.9619049e-01 -6.9750005e-01 -1.1500096e+00 -6.0962158e-01
-4.4664767e-01 9.1445357e-01 -2.9085675e-01 -5.7125854e-01
-8.3891225e-01 -6.4188188e-01 5.9345609e-01 4.6149614e-01
7.3237419e-01 -9.0604573e-01 -3.5431185e-01 4.1991660e-01
-9.2754775e-01 -9.9301100e-01 -1.1594019e+00 2.1358569e-01
-1.1476496e+00 -3.0162570e-01 -1.0850170e+00 -1.0456978e+00
6.0349584e-01 4.6982300e-01 1.1231135e+00 -2.2463624e-01
4.4722280e-01 -4.6633926e-01 -4.2256340e-01 -2.5568756e-01
-7.5220948e-01 4.4981050e-01 -1.7679794e-01 -2.0465481e-01
3.5650671e-01 -6.4582968e-01 -3.7044081e-01 -1.6014904e-01
-6.5240198e-01 6.9621420e-01 1.0611838e+00 9.9123406e-01
5.8051640e-01 -2.5951156e-01 5.1420975e-01 -6.0826427e-01
1.0531545e+00 -3.0364051e-01 -2.1956056e-02 4.4980288e-01
-5.4632568e-01 2.6653877e-01 7.8923303e-01 -2.1899219e-01
-9.3441504e-01 -2.4937417e-01 -4.9409589e-01 -5.3290683e-01
3.4512568e-01 9.3881482e-01 -1.7670931e-01 3.2492539e-01
3.6616352e-01 1.0252136e+00 2.9216027e-02 -6.6598207e-01
2.7727720e-01 1.0575956e+00 5.3214413e-01 -4.5627043e-01
6.1363006e-01 -2.9110062e-01 -2.2246023e-01 -2.5727391e-01
-8.7377763e-01 -1.7355273e-02 -6.1692715e-01 2.5594971e-01
6.2308294e-01 -1.1493682e+00 1.2593210e-01 1.6018160e-01
-1.7066841e+00 2.5377644e-02 -2.7585328e-01 4.7529975e-01
-6.0430551e-01 3.1535950e-01 -1.0354551e+00 -5.4634988e-01
-1.0777131e+00 -1.4031285e+00 1.3684711e+00 -2.0266955e-01
4.2787403e-02 -8.6066675e-01 -4.8432577e-02 3.6939800e-01
6.5001220e-01 -4.4644642e-01 1.0046549e+00 -3.4092233e-01
-3.3943570e-01 -4.2834561e-02 -2.1491937e-01 6.2366670e-01
2.2542070e-01 -5.8427376e-01 -6.0684556e-01 -4.7809377e-01
4.2538700e-01 -2.2107293e-01 1.0849688e+00 2.1843390e-01
7.8682417e-01 -6.8724281e-01 -6.5831684e-02 6.0891056e-01
1.2277521e+00 -1.5701713e-02 8.4656239e-01 1.4456195e-01
5.5840391e-01 2.3993768e-01 4.9237981e-01 -5.6806464e-02
6.0828453e-01 6.9630146e-01 1.7137200e-01 -1.3644230e-01
-4.7413859e-01 -6.7381048e-01 9.5564318e-01 1.7396384e+00
1.6618951e-01 -4.3125430e-01 -3.7107348e-01 4.3263340e-01
-1.8071251e+00 -8.7122518e-01 1.4664136e-01 1.9454161e+00
1.3511049e+00 2.5224158e-01 -2.1845599e-01 1.5838580e-02
8.4842569e-01 1.2206626e-01 -3.7755990e-01 -6.5283889e-01
-3.2739940e-01 1.5558355e-01 4.1681704e-01 5.9640616e-01
-6.3185263e-01 1.0378753e+00 6.3953762e+00 1.0869914e+00
-1.1577874e+00 4.6298528e-01 5.6705314e-01 -1.3949871e-02
-4.7824144e-01 9.8918639e-02 -8.7270498e-01 6.0064912e-01
1.3744202e+00 -6.7936510e-01 4.7559360e-01 4.4571260e-01
2.6458436e-01 3.7003875e-01 -1.0545670e+00 8.0123848e-01
2.5154030e-01 -1.4238442e+00 5.4240030e-01 1.1935689e-01
5.9856445e-01 8.5087620e-02 -6.4886741e-02 4.6758673e-01
-1.2876020e-01 -8.3373958e-01 8.7558383e-01 4.1813245e-01
1.1367296e+00 -7.4417764e-01 7.6101321e-01 7.2567254e-01
-1.0861200e+00 2.8141668e-01 -7.0201403e-01 -2.0534819e-01
4.4598973e-01 6.1036795e-01 -7.0863628e-01 9.3958610e-01
1.3200526e-01 8.9100981e-01 -4.2914709e-01 6.3060451e-01
-4.2729726e-01 7.6897025e-01 3.2461535e-03 -3.2071378e-02
4.0222344e-01 -2.1178453e-01 5.6939238e-01 1.3601360e+00
7.3877984e-01 -1.9343682e-01 -1.3763304e-01 8.9416414e-01
-3.3638644e-01 2.0478420e-01 -6.7143011e-01 -2.4236321e-01
4.1264665e-01 8.4001160e-01 -5.6782246e-02 -6.1041611e-01
-4.7044039e-01 1.4852270e+00 4.1089034e-01 3.1639585e-01
-7.3171616e-01 -5.5680758e-01 4.6437898e-01 -1.3201053e-01
3.6061013e-01 -1.8909664e-01 -3.0118552e-01 -1.6221672e+00
4.0174031e-01 -1.0066346e+00 -2.0581965e-01 -8.9658326e-01
-1.0259215e+00 9.2306310e-01 -7.1475856e-02 -1.3893661e+00
-6.2733960e-01 -1.5449011e-01 -6.5492082e-01 1.4004440e+00
-1.5825821e+00 -1.3321114e+00 5.0233316e-01 5.3198129e-01
1.0474153e+00 -3.2700568e-01 9.8749906e-01 4.0188852e-01
-5.2077454e-01 8.0817777e-01 1.9410153e-01 2.0665185e-01
5.4924911e-01 -1.0171887e+00 8.8550740e-01 1.1285306e+00
1.8970881e-01 8.0541468e-01 5.8750218e-01 -8.2003307e-01
-1.6383510e+00 -1.2743717e+00 1.6689180e+00 -1.5373884e-01
4.5791149e-01 -4.8521516e-01 -8.3742929e-01 7.5403333e-01
9.8685002e-01 -5.0733608e-01 3.0362296e-01 -2.0233460e-01
-1.6512485e-01 -1.1576831e-01 -8.2753736e-01 6.5910929e-01
9.9333829e-01 -7.1315104e-01 -9.8675627e-01 5.0873309e-01
1.3573853e+00 -5.6185019e-01 -7.6319855e-01 3.9431044e-01
1.3111989e-01 -5.4798442e-01 7.3429078e-01 -4.7593269e-01
1.0672294e+00 -1.6367042e-01 -3.0252540e-01 -1.7283525e+00
-4.1808128e-01 -5.9113449e-01 -4.8610911e-01 1.3023109e+00
6.6589612e-01 -3.8154638e-01 4.2178884e-01 -5.1587224e-03
-7.0207483e-01 -9.6530002e-01 -1.2081786e+00 -7.2666788e-01
5.0341946e-01 -2.3432009e-02 9.4205475e-01 5.0825232e-01
2.7554187e-01 1.1734785e+00 -6.9121933e-01 -4.0099737e-01
4.4781825e-01 1.7852084e-01 4.2427489e-01 -6.0819101e-01
-4.3456039e-01 -5.2451336e-01 7.2859399e-02 -1.6447887e+00
2.0219867e-01 -1.3335544e+00 6.0115147e-02 -1.8994030e+00
5.9557325e-01 2.9470006e-01 -1.1415402e-01 3.9779845e-01
-3.6264753e-01 1.9657016e-02 2.0360436e-01 3.0617365e-01
-2.8234330e-01 1.1439217e+00 1.7951221e+00 -4.6080801e-01
2.4566694e-01 -1.7062271e-01 -8.1013632e-01 -1.4229261e-02
7.3551804e-01 -4.1669562e-01 -5.3905326e-01 -1.2059436e+00
-5.3423539e-02 4.9668726e-01 -5.2104663e-02 -7.4986476e-01
3.7023792e-01 -6.4989947e-02 1.7810100e-01 -8.9344358e-01
4.5794192e-01 -5.4374641e-01 -3.6309373e-02 4.6256363e-01
-6.7174560e-01 5.5647767e-01 4.7546063e-02 2.1130364e-01
-6.1739612e-01 -2.4255149e-01 5.1631939e-01 -3.7048054e-01
6.8389259e-02 3.1547168e-01 -2.7562734e-01 -6.2845074e-02
4.2034602e-01 8.3434068e-02 -4.1525257e-01 -5.3648680e-01
-8.0935046e-02 2.0592666e-01 2.6252222e-01 5.6567365e-01
8.8596660e-01 -1.6217216e+00 -1.3120377e+00 3.2893860e-01
-1.2687157e-01 -1.8283141e-01 -8.8182092e-02 9.2972642e-01
-1.4518456e-01 9.0124208e-01 -1.1362643e-01 -3.2337174e-01
-1.0794702e+00 5.2626663e-01 2.2161289e-01 -4.1049483e-01
-6.7348671e-01 7.1932411e-01 8.8796616e-02 -3.2760802e-01
-1.4801895e-03 -3.2076403e-01 1.1092514e-01 -3.8766214e-01
6.2226087e-01 -9.5896482e-02 2.2762573e-01 -7.7102214e-01
-7.4093990e-02 4.1406846e-01 -4.4281912e-01 -4.3968478e-01
1.0326118e+00 -4.7018406e-01 -5.3863388e-01 3.3232078e-01
1.3967868e+00 -4.0737286e-01 -6.6850311e-01 -4.3926227e-01
-2.5254634e-01 -2.8862578e-01 2.0618005e-01 -8.5857868e-01
-9.6246243e-01 1.3390042e+00 7.7917524e-02 -2.8382954e-01
1.3921275e+00 -2.5944388e-01 1.4497452e+00 4.4283533e-01
4.8271522e-01 -9.2097110e-01 -3.5870290e-01 9.4738412e-01
1.0383855e+00 -9.8537749e-01 -3.1690335e-01 -2.6416039e-01
-5.2140075e-01 1.1506066e+00 4.9590731e-01 1.2415886e-01
2.1824916e-01 1.9188841e-01 -1.2610130e-01 3.0353224e-01
-1.3815833e+00 7.5785898e-02 2.9260960e-01 2.5714669e-01
8.0601639e-01 -4.2818859e-03 -6.8640333e-01 5.9038550e-01
-5.8724880e-01 1.1272053e-01 3.5831127e-01 9.1755056e-01
-5.6753349e-01 -1.4474770e+00 -1.6045569e-01 5.2318639e-01
-5.8305842e-01 -7.4406558e-01 -3.8015166e-01 1.2526442e-02
-1.7184390e-01 1.0226419e+00 -6.7061000e-02 -6.9471788e-01
2.4404959e-01 1.8489976e-01 6.4310384e-01 -8.3241892e-01
-7.3362643e-01 3.4967247e-01 1.4024968e-01 -1.6562767e-01
-2.7829072e-01 -5.2092522e-01 -1.0601363e+00 -3.8008663e-01
-3.3108494e-01 4.3264255e-01 7.4012393e-01 1.0749577e+00
6.1461836e-01 6.5827066e-01 7.1795064e-01 -6.5108484e-01
-7.5259089e-01 -1.6284519e+00 -1.5520884e-01 -2.7916601e-02
5.1589483e-01 -1.3659777e-03 -6.1855610e-02 4.4415694e-02] | [11.7619047164917, 9.922804832458496] |
3bd70d1e-e470-4b9d-a430-2f76460ed32c | localized-questions-in-medical-visual | 2307.01067 | null | https://arxiv.org/abs/2307.01067v1 | https://arxiv.org/pdf/2307.01067v1.pdf | Localized Questions in Medical Visual Question Answering | Visual Question Answering (VQA) models aim to answer natural language questions about given images. Due to its ability to ask questions that differ from those used when training the model, medical VQA has received substantial attention in recent years. However, existing medical VQA models typically focus on answering questions that refer to an entire image rather than where the relevant content may be located in the image. Consequently, VQA models are limited in their interpretability power and the possibility to probe the model about specific image regions. This paper proposes a novel approach for medical VQA that addresses this limitation by developing a model that can answer questions about image regions while considering the context necessary to answer the questions. Our experimental results demonstrate the effectiveness of our proposed model, outperforming existing methods on three datasets. Our code and data are available at https://github.com/sergiotasconmorales/locvqa. | ['Raphael Sznitman', 'Pablo Márquez-Neila', 'Sergio Tascon-Morales'] | 2023-07-03 | null | null | null | null | ['visual-question-answering', 'visual-question-answering-1', 'question-answering'] | ['computer-vision', 'computer-vision', 'natural-language-processing'] | [ 1.21412747e-01 3.37590486e-01 -1.80790275e-01 -4.35710788e-01
-9.67178345e-01 -5.61453402e-01 4.76467013e-01 6.22109950e-01
-2.67690569e-01 5.53891480e-01 3.25179368e-01 -5.45881629e-01
-8.30721483e-03 -8.03550065e-01 -4.78332192e-01 -3.16847324e-01
3.22461039e-01 3.54300886e-01 5.38338065e-01 -1.97311521e-01
2.29163408e-01 3.09978694e-01 -1.30241513e+00 5.73360980e-01
9.07329023e-01 7.73901165e-01 3.62986118e-01 7.91374803e-01
-3.68690163e-01 1.16804874e+00 -5.88559806e-01 -3.54538620e-01
-1.54106513e-01 -7.27869451e-01 -1.41124976e+00 1.01541407e-01
4.58697855e-01 -3.75804216e-01 -2.35056311e-01 1.05338395e+00
2.36126333e-01 -1.03433140e-01 5.85393786e-01 -1.16879058e+00
-8.90029907e-01 8.32074359e-02 -2.29342267e-01 6.52131915e-01
6.23224378e-01 1.82116985e-01 1.09233177e+00 -7.68536985e-01
7.40783215e-01 1.03541100e+00 1.46605998e-01 9.21310902e-01
-7.41401672e-01 -5.69818914e-02 2.85032898e-01 6.86533093e-01
-1.17469263e+00 -2.11013198e-01 8.01675916e-01 -3.97721529e-01
6.31035805e-01 6.48435056e-01 4.37795699e-01 7.83119559e-01
2.41191357e-01 1.19269967e+00 1.05634046e+00 -3.15791160e-01
3.26857656e-01 2.64405906e-01 3.34427685e-01 7.73974836e-01
-3.08265667e-02 -3.80492151e-01 -3.37294377e-02 -2.42961720e-01
6.35922968e-01 8.40114355e-02 -4.59411323e-01 -3.76395643e-01
-1.05891049e+00 9.74055529e-01 7.61876285e-01 4.71602172e-01
-4.17868793e-01 3.86008509e-02 3.88683766e-01 2.32396442e-02
2.02498123e-01 4.22891200e-01 -2.26220861e-01 2.11292088e-01
-6.25051737e-01 2.13308156e-01 5.65776825e-01 7.06268251e-01
6.78647220e-01 -4.97411162e-01 -5.80970585e-01 4.76578593e-01
3.61423522e-01 3.45215708e-01 3.06940556e-01 -9.05330122e-01
2.40277603e-01 9.17957425e-01 9.61624756e-02 -1.00237155e+00
-2.34229282e-01 -1.33639053e-01 -5.65751493e-01 -2.43668910e-04
5.30963540e-01 1.44037619e-01 -1.09385920e+00 1.40518010e+00
6.10277057e-01 -1.16770819e-01 1.11256227e-01 1.14615703e+00
1.53968215e+00 5.99895060e-01 3.61387283e-01 1.21806219e-01
1.88021159e+00 -9.04708862e-01 -8.61395538e-01 -3.82874072e-01
4.63179320e-01 -7.12982595e-01 1.35139930e+00 2.31349040e-02
-9.87565160e-01 -3.82275254e-01 -6.43384397e-01 -5.03665328e-01
-3.74347717e-01 5.98062016e-02 2.85856009e-01 3.95679504e-01
-1.17834246e+00 -1.60456851e-01 -6.25980854e-01 -6.05218709e-01
5.54676116e-01 5.27543575e-02 -1.04540110e-01 -3.26209158e-01
-1.09374142e+00 8.11602533e-01 7.32309893e-02 1.90501884e-01
-1.08069456e+00 -4.14273411e-01 -7.26552486e-01 4.83453907e-02
6.33841217e-01 -9.48446035e-01 1.40888619e+00 -1.05985653e+00
-8.57390225e-01 1.08613181e+00 -4.58370328e-01 -3.14881712e-01
5.52675366e-01 3.73421907e-02 -3.00473928e-01 8.62727404e-01
1.68844715e-01 6.89321041e-01 5.51584363e-01 -1.38560104e+00
-5.29722095e-01 -4.02675599e-01 7.87520230e-01 7.75720701e-02
-5.46283275e-02 -1.53572530e-01 -7.26692259e-01 -4.73274320e-01
-9.65137035e-02 -6.15419149e-01 -3.68145853e-01 3.84066582e-01
-3.69542181e-01 -3.57541084e-01 8.67738128e-01 -7.84218967e-01
1.14877176e+00 -1.84634566e+00 -1.91639110e-01 -1.87908355e-02
4.34654474e-01 3.83846104e-01 -1.27937689e-01 4.15829659e-01
2.33344257e-01 2.51282424e-01 -2.91751951e-01 1.75772429e-01
-3.01741213e-01 5.23767292e-01 -9.48022865e-03 2.02347115e-01
3.75360399e-01 1.25773954e+00 -1.04004431e+00 -1.03790200e+00
1.77843153e-01 4.50266123e-01 -4.58573252e-01 3.49036545e-01
-5.67790210e-01 5.79921007e-01 -8.40173721e-01 8.14549327e-01
5.48161149e-01 -7.48304069e-01 9.95513145e-03 -1.03682324e-01
3.41140062e-01 2.54900642e-02 -6.29061043e-01 1.41993678e+00
-3.16228926e-01 4.23506618e-01 1.12300485e-01 -1.04849410e+00
6.68236077e-01 4.41988736e-01 4.01716262e-01 -8.23399544e-01
1.75637752e-01 2.98063960e-02 -1.92117438e-01 -1.05412555e+00
2.65095055e-01 -1.75970301e-01 1.76883027e-01 1.85404465e-01
-2.06990764e-01 -1.01649523e-01 1.51306525e-01 4.56739008e-01
9.88284230e-01 -1.67240426e-02 5.12038052e-01 -1.21920183e-01
1.07839227e+00 3.61754477e-01 2.80460536e-01 6.95631266e-01
-6.33972645e-01 6.97325945e-01 4.76337343e-01 -3.99340481e-01
-7.20696986e-01 -1.19666791e+00 -4.73268405e-02 6.90158367e-01
4.26898539e-01 -3.00952703e-01 -7.22640753e-01 -1.04806733e+00
-3.24141055e-01 5.88824570e-01 -1.04281437e+00 1.47271797e-01
-4.83356446e-01 -4.68717478e-02 1.30475253e-01 4.70473945e-01
3.90993297e-01 -1.37320483e+00 -8.79017830e-01 -2.55737036e-01
-6.85380936e-01 -9.70671952e-01 -4.25257742e-01 -4.71636951e-01
-9.51726913e-01 -1.52553070e+00 -9.86092687e-01 -9.58805680e-01
9.79898572e-01 2.87376374e-01 1.38710463e+00 4.59188491e-01
-5.02752960e-01 9.39667583e-01 -6.18631005e-01 -6.03487074e-01
-4.83698457e-01 8.76346231e-03 -8.77117038e-01 -5.39544933e-02
4.94717926e-01 5.08059934e-02 -1.14948916e+00 2.66845673e-01
-1.27264762e+00 7.18805417e-02 6.14210367e-01 6.15926623e-01
7.93261707e-01 -6.00330353e-01 6.69897437e-01 -1.10801470e+00
6.86394095e-01 -7.06012011e-01 -2.27299824e-01 7.24560618e-01
-3.63905966e-01 4.75200564e-02 3.58352035e-01 -1.53695658e-01
-9.94505584e-01 -8.08875933e-02 -5.04260540e-01 -2.05363572e-01
-5.11877179e-01 4.59417403e-01 5.59230633e-02 1.69129327e-01
6.30781770e-01 2.52983719e-01 4.39265296e-02 -2.84679711e-01
4.25471634e-01 4.40707296e-01 4.24213618e-01 -3.49787235e-01
5.30790985e-01 7.53727317e-01 -2.55844116e-01 -7.37931609e-01
-8.31670880e-01 -7.15471506e-01 -3.73001307e-01 -4.59288031e-01
1.24036825e+00 -4.97478187e-01 -7.78703988e-01 -2.71888435e-01
-1.09377837e+00 9.97407958e-02 -4.04568255e-01 9.22506601e-02
-4.11987633e-01 6.65236175e-01 -3.65767777e-01 -8.19626272e-01
-5.55768490e-01 -1.08648562e+00 1.02529371e+00 2.67363966e-01
-2.77160615e-01 -1.21282947e+00 1.57634076e-02 8.64598095e-01
2.83711523e-01 3.50036711e-01 1.14055669e+00 -5.07668555e-01
-7.72408843e-01 2.84201764e-02 -4.54159558e-01 6.19668998e-02
3.62123132e-01 -3.86208415e-01 -8.07334065e-01 -1.31060213e-01
5.59313819e-02 -3.49375248e-01 7.18655825e-01 4.83430833e-01
1.41639435e+00 -3.92559737e-01 -2.81786323e-01 4.73239943e-02
1.50806093e+00 8.44796300e-02 6.30520463e-01 1.96457341e-01
4.63041693e-01 6.55744672e-01 8.96449745e-01 2.22966835e-01
6.95325077e-01 3.60142708e-01 9.61261570e-01 -4.38943952e-01
-1.89157218e-01 -1.61839768e-01 -2.35674813e-01 3.94224018e-01
9.13191587e-02 -3.14439178e-01 -1.13308620e+00 1.21825373e+00
-1.90315366e+00 -6.21718347e-01 -4.12230462e-01 1.72817433e+00
6.43028200e-01 -3.08956921e-01 2.96756122e-02 -1.43269211e-01
3.27290118e-01 1.53291151e-01 -6.62547767e-01 -4.95639235e-01
1.08160555e-01 1.06892616e-01 7.05343112e-03 6.04651570e-01
-9.45310473e-01 6.05530500e-01 6.14313936e+00 6.84813857e-01
-8.58678043e-01 1.60434365e-01 7.48300612e-01 2.24219367e-01
-7.95085311e-01 -1.23894535e-01 -2.98817992e-01 1.29345581e-01
4.44293886e-01 3.05085238e-02 -1.44079730e-01 5.84700882e-01
2.89951354e-01 -3.65585893e-01 -9.75225806e-01 8.45613539e-01
3.01898003e-01 -1.37098467e+00 5.18539429e-01 -1.69443727e-01
4.62350786e-01 -3.51591021e-01 8.53300020e-02 -5.81870088e-03
-7.80422194e-03 -1.27233982e+00 4.17487651e-01 7.28012562e-01
4.84882057e-01 -4.90519375e-01 9.54579234e-01 4.34022397e-01
-1.02665329e+00 5.45622408e-02 -2.08622739e-01 1.36156142e-01
-1.92648694e-02 1.97105616e-01 -1.27616715e+00 7.15621352e-01
8.12269747e-01 2.28952900e-01 -9.46478248e-01 1.05818701e+00
-3.76377463e-01 5.60971498e-01 1.50755331e-01 -3.19045186e-01
3.56290191e-01 1.15209520e-02 2.77697265e-01 9.72216547e-01
1.50181070e-01 4.60798532e-01 1.88361853e-02 7.66232967e-01
8.62761438e-02 5.21855295e-01 -4.90462691e-01 2.06267107e-02
-6.14893399e-02 1.11442471e+00 -6.66653275e-01 -3.45811665e-01
-7.28953838e-01 9.43963766e-01 2.07033649e-01 4.72469836e-01
-7.87607014e-01 -1.40434206e-01 3.03093195e-01 5.20480037e-01
2.55450845e-01 1.91405758e-01 -1.90310195e-01 -1.00840032e+00
8.61823037e-02 -9.80701268e-01 9.56193388e-01 -1.03568959e+00
-1.19039035e+00 7.03848779e-01 -5.20121306e-02 -1.20459414e+00
-1.77996784e-01 -6.34410858e-01 -3.61794174e-01 6.91907823e-01
-1.70424449e+00 -1.44165826e+00 -5.83864152e-01 8.66278946e-01
6.11684144e-01 3.32803577e-01 8.34746182e-01 7.76946694e-02
4.29047868e-02 2.19835714e-01 -2.66349018e-01 1.76693738e-01
5.33633649e-01 -1.23137283e+00 -9.74557102e-02 7.50511229e-01
1.07757859e-01 5.07199466e-01 7.63469994e-01 -4.03365195e-01
-1.10805130e+00 -8.71531188e-01 1.03049326e+00 -6.44383192e-01
2.72214323e-01 -5.05351499e-02 -1.17664981e+00 5.55565774e-01
5.43434560e-01 1.30511135e-01 8.94986331e-01 -2.32512265e-01
-1.65078461e-01 9.47204605e-02 -1.33951843e+00 6.00211203e-01
6.95685744e-01 -5.59070528e-01 -7.24294364e-01 2.96648741e-01
5.44019282e-01 -3.33131164e-01 -7.04235196e-01 3.75895113e-01
2.99312562e-01 -8.05784166e-01 9.83697236e-01 -7.82311320e-01
5.68569362e-01 -5.50185025e-01 2.98012723e-03 -7.88353682e-01
-1.06031887e-01 7.98395127e-02 -5.19540906e-02 8.20067048e-01
4.20136988e-01 -3.52627009e-01 8.94726098e-01 4.64840770e-01
2.21598104e-01 -9.10372555e-01 -8.69137645e-01 -1.94012135e-01
1.10815644e-01 -1.89568833e-01 4.45796192e-01 1.00336897e+00
-1.52601898e-01 2.33729944e-01 -8.96829665e-02 4.25175875e-01
2.97884583e-01 4.44855630e-01 5.50891399e-01 -1.01608264e+00
-2.41594657e-01 -1.73255667e-01 -3.29789162e-01 -9.80988145e-01
-2.21568465e-01 -8.58905613e-01 -1.20549258e-02 -2.36535096e+00
4.09672290e-01 -4.39973511e-02 -2.20961496e-01 5.82438946e-01
-5.13207316e-01 4.02848393e-01 1.72913834e-01 1.15572484e-02
-7.81430900e-01 2.78847784e-01 1.62221599e+00 -3.84384334e-01
1.72069103e-01 -2.95158606e-02 -8.44505966e-01 5.99265218e-01
9.59773183e-01 -2.06320152e-01 -6.87419891e-01 -5.83646655e-01
2.68894076e-01 2.84736633e-01 7.07772553e-01 -6.84431970e-01
2.41255671e-01 -3.13467979e-01 2.50316381e-01 -6.10488474e-01
1.72983974e-01 -8.54559958e-01 -1.19843014e-01 7.76832998e-01
-3.63706619e-01 5.39464131e-02 2.44949222e-01 4.73840296e-01
-4.74841982e-01 -4.34409201e-01 6.71860039e-01 -3.53537768e-01
-9.64267910e-01 3.93591493e-01 -5.19305587e-01 1.91384852e-01
1.25815558e+00 -1.19986050e-01 -4.59817946e-01 -7.81871855e-01
-9.59710896e-01 6.88690543e-01 4.64589268e-01 5.18226266e-01
1.12871277e+00 -1.07395983e+00 -8.33831370e-01 -2.79301167e-01
6.76459849e-01 -3.98776829e-02 6.48028851e-01 6.49832726e-01
-8.94909084e-01 5.43424845e-01 -3.73746008e-02 -5.95093071e-01
-1.58243263e+00 9.42815959e-01 4.25151616e-01 -2.94304788e-01
-5.10012865e-01 6.23578906e-01 6.67867362e-01 -3.13074619e-01
-7.60603417e-03 -4.71787155e-01 -6.68678761e-01 -2.28501201e-01
5.31856835e-01 -7.77788274e-03 -1.33070812e-01 -6.55488312e-01
-5.57463884e-01 4.89721388e-01 -5.61879985e-02 1.40802220e-01
8.45402896e-01 -3.18710566e-01 -2.05305871e-02 3.11880082e-01
1.01348913e+00 -4.31281999e-02 -8.24914575e-01 -3.59706253e-01
-4.21768911e-02 -4.78582084e-01 -1.38092250e-01 -9.88023162e-01
-9.45003033e-01 1.10729587e+00 8.13390195e-01 2.88452566e-01
1.35022616e+00 6.43613040e-01 6.34004354e-01 4.10966799e-02
3.43032666e-02 -7.35690594e-01 3.29777092e-01 9.23962221e-02
9.91616189e-01 -1.43900335e+00 -2.04330459e-02 -5.94518900e-01
-7.75201917e-01 8.80070508e-01 6.87386096e-01 1.44947007e-01
6.29058778e-01 -2.41548166e-01 5.95305681e-01 -5.79037189e-01
-5.95680475e-01 -4.31252509e-01 5.26155591e-01 6.65641546e-01
4.95059878e-01 -9.88871232e-02 -5.82229197e-01 1.94243446e-01
1.64876625e-01 7.12607056e-03 4.43020970e-01 9.93804753e-01
-4.11869824e-01 -1.07293963e+00 -4.38370973e-01 3.12951058e-01
-6.31280363e-01 4.03761566e-02 -6.45877361e-01 7.07224667e-01
2.70971488e-02 1.24225545e+00 -9.13472325e-02 2.10843399e-01
3.17124099e-01 -8.50459188e-02 3.63421053e-01 -5.76749146e-01
-5.23824275e-01 -1.49195045e-01 -7.30934069e-02 -4.99419928e-01
-5.54298043e-01 -4.28131998e-01 -1.43978846e+00 1.59105778e-01
1.91922281e-02 3.71524304e-01 3.32678348e-01 8.84909749e-01
2.94257492e-01 5.53052306e-01 2.88358331e-01 1.30690366e-01
-6.03119209e-02 -6.72473073e-01 -1.76381484e-01 6.34325266e-01
7.10536003e-01 -1.55507267e-01 1.54245980e-02 6.46056011e-02] | [10.955460548400879, 1.634764313697815] |
0a902810-d257-4eb9-bff5-e49bae85ed7f | affine-and-regional-dynamic-time-warpng | 1505.06531 | null | http://arxiv.org/abs/1505.06531v1 | http://arxiv.org/pdf/1505.06531v1.pdf | Affine and Regional Dynamic Time Warpng | Pointwise matches between two time series are of great importance in time
series analysis, and dynamic time warping (DTW) is known to provide generally
reasonable matches. There are situations where time series alignment should be
invariant to scaling and offset in amplitude or where local regions of the
considered time series should be strongly reflected in pointwise matches. Two
different variants of DTW, affine DTW (ADTW) and regional DTW (RDTW), are
proposed to handle scaling and offset in amplitude and provide regional
emphasis respectively. Furthermore, ADTW and RDTW can be combined in two
different ways to generate alignments that incorporate advantages from both
methods, where the affine model can be applied either globally to the entire
time series or locally to each region. The proposed alignment methods
outperform DTW on specific simulated datasets, and one-nearest-neighbor
classifiers using their associated difference measures are competitive with the
difference measures associated with state-of-the-art alignment methods on real
datasets. | ['Tsu-Wei Chen', 'Daniel Stashuk', 'Meena Abdelmaseeh'] | 2015-05-25 | null | null | null | null | ['time-series-alignment'] | ['time-series'] | [ 2.33666986e-01 -5.48934102e-01 -1.54416129e-01 -2.85202324e-01
-5.31089962e-01 -8.72152209e-01 8.48235309e-01 4.87908512e-01
-3.83602679e-01 6.03862703e-01 1.68932170e-01 -2.60796249e-01
-8.13185573e-01 -6.30960405e-01 -2.67248511e-01 -7.93906927e-01
-6.65149748e-01 3.10105622e-01 4.65014011e-01 -5.21066487e-01
4.35396969e-01 8.33556950e-01 -1.32126737e+00 -8.60608369e-02
7.12881148e-01 7.85979152e-01 -2.49716505e-01 5.11835694e-01
-5.48539273e-02 1.40290722e-01 -6.39324963e-01 8.97889063e-02
6.10913754e-01 -4.00292933e-01 -5.34759462e-01 -2.05587462e-01
-1.67038701e-02 1.94961607e-01 -3.99769768e-02 7.68047631e-01
5.58561146e-01 3.43204081e-01 5.98616064e-01 -1.42604232e+00
-1.48303077e-01 3.30807835e-01 -8.75846028e-01 6.34575486e-01
6.36640966e-01 -1.25621259e-01 6.33191049e-01 -4.61535156e-01
6.85660839e-01 8.04894090e-01 9.63133991e-01 -1.36102423e-01
-1.41698337e+00 -4.96254086e-01 4.86932397e-02 3.68223965e-01
-1.18749475e+00 -2.05148552e-02 1.06982887e+00 -4.01990235e-01
9.60709393e-01 5.34213066e-01 6.34888470e-01 7.90075719e-01
7.42991626e-01 2.65512228e-01 1.49460590e+00 -5.10992050e-01
2.53156424e-01 -6.12365901e-01 1.91697944e-02 -1.67416692e-01
-3.73698883e-02 3.56348932e-01 -2.55390286e-01 -4.67019498e-01
5.72986126e-01 1.65014833e-01 -3.07204574e-01 -5.35925984e-01
-1.64699101e+00 6.33040190e-01 2.40023091e-01 8.20674837e-01
-5.21540940e-01 -3.75513792e-01 7.76772559e-01 7.36613750e-01
7.08426952e-01 5.46951652e-01 -2.92926371e-01 -3.94571722e-01
-1.01670134e+00 5.16466618e-01 4.87917989e-01 6.67131543e-01
3.77098173e-01 7.23187104e-02 -1.13374107e-01 6.49595141e-01
-3.18616211e-01 2.22498402e-01 9.43601012e-01 -4.72479612e-01
5.04804730e-01 4.34732556e-01 8.98810998e-02 -1.05982220e+00
-6.88274801e-01 -3.05432916e-01 -8.41643095e-01 6.70178160e-02
6.13183856e-01 1.21773310e-01 -7.21245885e-01 1.74924660e+00
4.31449980e-01 5.17586291e-01 7.20270351e-02 6.77111089e-01
1.05404027e-01 8.39238763e-01 -2.44458765e-01 -8.14214349e-01
1.12386107e+00 -6.04720354e-01 -8.64658296e-01 7.93908760e-02
4.74194050e-01 -1.31582606e+00 5.78992903e-01 2.25573167e-01
-9.76114750e-01 -6.30866706e-01 -1.21878397e+00 4.29260403e-01
-4.71655011e-01 -4.65084851e-01 1.20584890e-02 2.81435758e-01
-8.65343213e-01 9.51954067e-01 -9.42280054e-01 -5.94942570e-01
-4.11137015e-01 2.35762000e-01 -4.94831085e-01 3.82461160e-01
-1.22414660e+00 1.05159855e+00 2.00760022e-01 1.04170643e-01
-2.05889061e-01 -7.16348350e-01 -5.13775229e-01 -2.69148231e-01
-9.34670717e-02 -1.10327147e-01 9.66847777e-01 -6.24403179e-01
-1.34119463e+00 6.92061961e-01 -1.51854083e-01 -6.71522021e-01
8.18282306e-01 -4.33245376e-02 -9.70254421e-01 -1.34887006e-02
-7.70783126e-02 6.06603250e-02 1.02767587e+00 -7.01489806e-01
-2.76715368e-01 -3.43365967e-01 -4.15130854e-01 -1.09245144e-01
-1.84508920e-01 2.48199612e-01 3.06800306e-01 -1.07307100e+00
5.10446668e-01 -9.96263921e-01 -2.42275387e-01 -1.51415497e-01
-5.81305772e-02 -1.47020832e-01 1.26117480e+00 -7.03492165e-01
1.44111931e+00 -2.02939296e+00 3.30403239e-01 5.61381340e-01
-2.01336905e-01 3.28328073e-01 -2.49116585e-01 1.12151647e+00
-6.86577916e-01 -4.20144171e-01 -3.14515233e-01 1.72674105e-01
-2.72830039e-01 3.55170339e-01 -6.40411556e-01 8.44005644e-01
2.34453946e-01 4.04820502e-01 -9.21136498e-01 -2.24842384e-01
3.28808486e-01 2.83377409e-01 -8.18727612e-02 1.40605688e-01
3.56432587e-01 4.98115838e-01 -1.39714673e-01 1.67599499e-01
5.83455145e-01 3.93487811e-01 2.27562755e-01 -2.87488729e-01
-5.93289852e-01 2.44460665e-02 -1.23174763e+00 1.27586150e+00
-3.71307582e-01 5.73316157e-01 -4.09313679e-01 -1.29973388e+00
1.61951363e+00 6.12679541e-01 9.54672933e-01 -7.60470152e-01
1.25605345e-01 4.41449583e-01 3.59424084e-01 -4.06116992e-01
3.54239404e-01 -1.00718319e-01 -3.18845958e-02 6.43050075e-01
-1.84813932e-01 -2.60952353e-01 2.66141295e-01 -5.56087315e-01
1.07851815e+00 1.72660977e-01 9.01434600e-01 -6.31404519e-01
6.80268109e-01 -1.66029334e-01 7.00666547e-01 2.65572578e-01
-3.27232450e-01 6.42501652e-01 3.78208309e-01 -8.45340788e-01
-1.53718317e+00 -9.09033775e-01 -2.75266439e-01 5.25366724e-01
9.32810083e-02 -2.21963868e-01 -3.55357587e-01 -1.51268274e-01
-1.03252262e-01 4.43848699e-01 -6.16043270e-01 -2.45798543e-01
-1.05997074e+00 -5.72080493e-01 3.49198490e-01 6.17155492e-01
3.83825451e-01 -7.93738365e-01 -1.05312622e+00 5.77637553e-01
-1.22238085e-01 -9.11833763e-01 -6.37357831e-01 3.94823313e-01
-1.35754561e+00 -8.64315510e-01 -9.63280022e-01 -6.00562036e-01
4.73516077e-01 3.15905809e-01 8.21188331e-01 -4.12394285e-01
-2.85787642e-01 6.99527040e-02 -6.71236157e-01 -3.46353889e-01
-4.37481582e-01 7.62020797e-03 2.63268620e-01 1.81126073e-01
2.26441875e-01 -1.07263863e+00 -4.83391434e-01 8.73731196e-01
-9.43242073e-01 -2.95100063e-01 2.59942953e-02 8.53671968e-01
4.92207766e-01 1.15376703e-01 5.06120741e-01 -4.50033218e-01
8.00412416e-01 -3.59103143e-01 -6.00433588e-01 1.64944604e-01
-6.45689070e-01 5.64379580e-02 8.61155152e-01 -9.06036377e-01
-5.49678326e-01 -1.61570802e-01 8.59052986e-02 -6.95576549e-01
2.90427990e-02 5.63027978e-01 4.80805933e-01 -1.35937676e-01
6.64056063e-01 2.95605868e-01 2.65742064e-01 -3.47073734e-01
6.28248826e-02 5.01007378e-01 5.94685555e-01 -6.50840223e-01
8.80785704e-01 4.80731696e-01 2.76661426e-01 -5.77407777e-01
-3.75622399e-02 -6.20996773e-01 -1.05500901e+00 -3.37248266e-01
5.34876943e-01 -3.54959905e-01 -1.79677725e-01 6.53862953e-01
-9.29371893e-01 1.89664721e-01 -4.08152938e-01 5.86030304e-01
-7.32502997e-01 3.49364758e-01 -2.42799357e-01 -6.81477427e-01
-2.28387401e-01 -9.16673183e-01 9.06659186e-01 4.81005460e-02
-7.00418770e-01 -1.24655867e+00 4.01542515e-01 -5.12863576e-01
6.01770997e-01 9.35292125e-01 1.03823233e+00 -1.01762128e+00
2.90996790e-01 -6.19420588e-01 3.42971504e-01 -3.51836672e-03
5.09644806e-01 3.76320183e-01 -5.29015303e-01 -6.00057662e-01
2.49084651e-01 4.16337699e-01 7.71122426e-02 4.88598019e-01
6.61848903e-01 -2.67401755e-01 -3.37279528e-01 1.87063485e-01
1.29105520e+00 7.92691112e-01 6.11644566e-01 6.09360933e-01
4.92575988e-02 7.00339317e-01 9.26955104e-01 5.48847020e-01
-1.27535060e-01 1.20862377e+00 2.00301886e-01 -1.14420354e-01
2.27831036e-01 2.04840042e-02 2.23118484e-01 1.11744142e+00
-5.34070671e-01 1.70067012e-01 -1.04397345e+00 7.37717748e-01
-1.96563923e+00 -1.29163754e+00 -2.08724037e-01 2.54668236e+00
5.82434654e-01 2.34874234e-01 4.81403977e-01 9.59479988e-01
9.96492684e-01 5.66467643e-01 -3.21036398e-01 -7.70208299e-01
-1.00018598e-01 4.18438494e-01 4.90751386e-01 1.55699804e-01
-1.07473958e+00 5.44444397e-02 6.86489296e+00 7.43763208e-01
-1.39199543e+00 -3.19427960e-02 4.96576950e-02 1.19348057e-01
-6.52710572e-02 7.14177266e-02 -1.80057228e-01 4.74680930e-01
9.32763875e-01 -7.42769659e-01 1.97922692e-01 5.93477547e-01
4.61817831e-01 2.62415349e-01 -1.08180630e+00 7.96742439e-01
-1.97886676e-01 -9.53995705e-01 -2.61423081e-01 -2.32019499e-02
8.17330003e-01 -3.76100808e-01 -4.70918305e-02 -1.11291893e-01
-9.35300738e-02 -5.29680669e-01 6.53765619e-01 5.19636512e-01
3.08713049e-01 -7.78734624e-01 8.95359814e-01 2.25238428e-01
-1.77601004e+00 -1.63598463e-01 -1.94652736e-01 -1.60081431e-01
5.09666026e-01 5.55432081e-01 -4.93778527e-01 1.09381819e+00
6.35001838e-01 8.54869902e-01 -2.26586491e-01 1.12700844e+00
2.82436907e-01 2.81641573e-01 -4.05153781e-01 3.86131629e-02
3.39608788e-01 -5.18784761e-01 9.30720568e-01 8.02088082e-01
8.08764398e-01 -7.67898932e-02 9.15548876e-02 2.69200742e-01
8.84649336e-01 1.66078463e-01 -7.38905787e-01 8.19387436e-02
7.35974371e-01 8.90949667e-01 -7.56349087e-01 -5.96012883e-02
-2.75922567e-01 6.97900712e-01 -2.80597955e-01 1.79677472e-01
-9.68338251e-01 -6.48329973e-01 7.48526573e-01 2.21098185e-01
1.97651461e-01 -4.97911423e-01 7.21504837e-02 -7.22432256e-01
1.28237337e-01 -8.73810291e-01 7.82819092e-01 -7.43609965e-01
-1.34221888e+00 9.22111034e-01 5.33238053e-01 -2.30904961e+00
-6.30238652e-01 -3.58831912e-01 -8.91950488e-01 7.93061376e-01
-1.15962160e+00 -8.62727880e-01 -2.66987801e-01 7.18196273e-01
5.46722114e-01 -3.46417241e-02 7.50925243e-01 4.15956914e-01
-1.90307066e-01 3.72289449e-01 2.33726859e-01 -2.25850627e-01
7.55209684e-01 -8.98341715e-01 4.78813916e-01 9.27264452e-01
4.54009175e-02 5.54673731e-01 1.23817325e+00 -4.03864890e-01
-1.16435075e+00 -9.54927564e-01 9.00821149e-01 5.81051037e-02
9.25188780e-01 -7.01790676e-03 -1.13415730e+00 5.22742808e-01
2.32029602e-01 5.53229153e-02 4.17947501e-01 -1.61332488e-01
-4.51717526e-01 -3.14208239e-01 -1.15635645e+00 5.79256475e-01
8.32759559e-01 -4.64659482e-01 -1.08305526e+00 2.14892030e-01
4.52983707e-01 -4.00232583e-01 -1.31767154e+00 6.29785776e-01
6.46683455e-01 -1.02556646e+00 1.04382825e+00 -5.50760448e-01
1.23066045e-01 -6.16416335e-01 -1.31484732e-01 -1.59321094e+00
-3.96817535e-01 -8.51468921e-01 2.77847141e-01 1.19879472e+00
9.89134833e-02 -1.09605157e+00 9.68554765e-02 -2.15353388e-02
-1.28567755e-01 -5.12922227e-01 -1.43647325e+00 -1.42362297e+00
4.23136055e-02 -1.53994411e-01 9.69943583e-01 1.27000284e+00
3.72620553e-01 -3.20232689e-01 -3.45124602e-01 3.26918773e-02
2.70407975e-01 4.76084650e-01 5.72840691e-01 -1.23996568e+00
1.92812476e-02 -5.60026526e-01 -1.03952456e+00 -3.39023441e-01
-6.14614300e-02 -4.74422723e-01 -2.78384030e-01 -1.02537274e+00
-3.93747240e-01 -1.95447773e-01 -4.05832708e-01 3.17129850e-01
5.04032113e-02 1.75932288e-01 7.27740377e-02 5.16640902e-01
4.02423471e-01 4.83879566e-01 9.46635008e-01 -3.52905178e-03
-3.98798496e-01 2.03379840e-01 3.57676059e-01 4.40793067e-01
6.95293903e-01 -3.58189404e-01 -4.15023416e-01 -2.42075045e-02
-2.01366499e-01 4.37317938e-01 1.51115265e-02 -1.23779666e+00
2.64250815e-01 -2.21314281e-01 -1.62263382e-02 -6.74726188e-01
1.51050657e-01 -1.09087873e+00 8.50964308e-01 6.49444222e-01
-2.60303915e-01 1.07162499e+00 2.63978302e-01 3.48958611e-01
-6.58170104e-01 -7.83405453e-03 9.26660359e-01 4.08195883e-01
-5.15963078e-01 -2.30644234e-02 -2.97593027e-01 -2.77032167e-01
1.36062181e+00 -6.93709373e-01 -3.48226935e-01 -5.01800001e-01
-6.19444966e-01 -5.37313893e-02 2.78759360e-01 6.88429713e-01
3.23390156e-01 -1.73295748e+00 -6.19176328e-01 2.29717389e-01
2.60678381e-01 -5.75926185e-01 2.64787406e-01 1.36612308e+00
-4.44093674e-01 3.70505184e-01 -8.14961910e-01 -7.99927473e-01
-1.42894244e+00 8.43016982e-01 2.38584340e-01 -3.66797686e-01
-7.50733018e-01 2.38528907e-01 -3.70127320e-01 -3.43585253e-01
-1.09043747e-01 -6.20660901e-01 -2.55895883e-01 3.83344412e-01
3.80352736e-01 5.53501546e-01 3.38308364e-01 -7.94315398e-01
-4.65208709e-01 1.31769145e+00 1.27802342e-01 -9.00769904e-02
1.41220343e+00 7.17796683e-02 -1.17903829e-01 6.76761091e-01
1.21374488e+00 -1.41430885e-01 -9.44538593e-01 -2.20523372e-01
2.61493564e-01 -5.75622559e-01 -4.26237822e-01 -1.67104259e-01
-8.19677532e-01 6.35653615e-01 6.39615595e-01 7.30598688e-01
1.59976816e+00 -5.15718877e-01 6.11892223e-01 -1.59216806e-01
5.59103072e-01 -8.92324030e-01 -1.46643281e-01 2.19880477e-01
9.97847974e-01 -5.31482637e-01 2.41889134e-01 -3.68072927e-01
-4.17637587e-01 1.49594080e+00 1.47746801e-01 -4.51680720e-01
6.10574007e-01 4.17791963e-01 1.61338508e-01 2.49423355e-01
-7.60471344e-01 1.04203867e-02 4.09709632e-01 5.32122135e-01
5.16533554e-01 -1.46912426e-01 -8.97263348e-01 -3.44021842e-02
-2.99670368e-01 -2.05410197e-01 2.88344592e-01 1.20487380e+00
-2.23991629e-02 -1.35606539e+00 -6.60693288e-01 1.73172906e-01
-1.14613205e-01 4.85337168e-01 -1.22883759e-01 1.22096193e+00
-2.36250937e-01 7.18979597e-01 4.16495144e-01 -5.53651571e-01
7.11587846e-01 2.21582025e-01 3.83989364e-01 1.69515610e-01
-8.43642592e-01 3.86607915e-01 -1.13343991e-01 -7.03530133e-01
-7.61455953e-01 -8.64096999e-01 -9.42853332e-01 -3.85884136e-01
-3.62256348e-01 1.62859619e-01 4.73860532e-01 1.01203883e+00
1.81375891e-01 4.86661255e-01 1.05698252e+00 -9.02982414e-01
-5.70924997e-01 -8.73754978e-01 -6.29792988e-01 5.48393667e-01
3.20718616e-01 -7.54086077e-01 -5.26864409e-01 4.17577773e-02] | [7.287787914276123, 3.3184478282928467] |
1205a039-07d5-44e4-b0e3-48e08fb55605 | paramixer-parameterizing-mixing-links-in | 2204.10670 | null | https://arxiv.org/abs/2204.10670v1 | https://arxiv.org/pdf/2204.10670v1.pdf | Paramixer: Parameterizing Mixing Links in Sparse Factors Works Better than Dot-Product Self-Attention | Self-Attention is a widely used building block in neural modeling to mix long-range data elements. Most self-attention neural networks employ pairwise dot-products to specify the attention coefficients. However, these methods require $O(N^2)$ computing cost for sequence length $N$. Even though some approximation methods have been introduced to relieve the quadratic cost, the performance of the dot-product approach is still bottlenecked by the low-rank constraint in the attention matrix factorization. In this paper, we propose a novel scalable and effective mixing building block called Paramixer. Our method factorizes the interaction matrix into several sparse matrices, where we parameterize the non-zero entries by MLPs with the data elements as input. The overall computing cost of the new building block is as low as $O(N \log N)$. Moreover, all factorizing matrices in Paramixer are full-rank, so it does not suffer from the low-rank bottleneck. We have tested the new method on both synthetic and various real-world long sequential data sets and compared it with several state-of-the-art attention networks. The experimental results show that Paramixer has better performance in most learning tasks. | ['Zhirong Yang', 'Lei Cheng', 'Ruslan Khalitov', 'Tong Yu'] | 2022-04-22 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Yu_Paramixer_Parameterizing_Mixing_Links_in_Sparse_Factors_Works_Better_Than_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Yu_Paramixer_Parameterizing_Mixing_Links_in_Sparse_Factors_Works_Better_Than_CVPR_2022_paper.pdf | cvpr-2022-1 | ['long-range-modeling'] | ['natural-language-processing'] | [-1.62702844e-01 -2.16475829e-01 8.47846866e-02 -3.47435385e-01
-7.32019782e-01 -2.44162247e-01 8.07802379e-02 -1.23310070e-02
-8.86992931e-01 6.27715409e-01 1.12963423e-01 -2.79273123e-01
-7.42618665e-02 -5.31573057e-01 -1.19070876e+00 -9.40081596e-01
-2.93995552e-02 4.59630400e-01 8.50254856e-03 -1.65117308e-01
1.94328770e-01 7.67843351e-02 -1.45583355e+00 1.89360544e-01
1.15114737e+00 1.03360939e+00 4.52874601e-01 5.54910183e-01
-2.94716388e-01 9.01152730e-01 -4.42444474e-01 -4.25716221e-01
2.92353302e-01 -1.97636619e-01 -6.20682061e-01 -3.65722388e-01
6.73912585e-01 -1.35925487e-01 -5.83120465e-01 1.25020576e+00
7.12291420e-01 4.35798764e-01 3.45691860e-01 -1.14666975e+00
-9.46452260e-01 1.26376259e+00 -8.00613403e-01 4.91220921e-01
-1.92778394e-01 -8.08269531e-03 1.35918283e+00 -1.29472733e+00
1.85396001e-01 1.21355784e+00 7.29795277e-01 2.89069355e-01
-1.16356468e+00 -1.18142021e+00 5.95417500e-01 5.92240930e-01
-1.70649385e+00 -4.75231737e-01 6.41698420e-01 -3.38771909e-01
1.37164319e+00 2.46618345e-01 5.91983676e-01 6.36885166e-01
-4.09749672e-02 9.73178983e-01 6.53719842e-01 -2.45912239e-01
-3.23554575e-02 -2.06772953e-01 7.11784422e-01 8.89511168e-01
2.73130864e-01 -4.19485182e-01 -5.61758518e-01 4.61404249e-02
7.29921877e-01 2.98474014e-01 -3.73788923e-01 -2.92921644e-02
-1.12579632e+00 7.75688529e-01 7.79376149e-01 3.92319083e-01
-4.46782649e-01 5.73646307e-01 3.72978002e-01 2.05475464e-01
3.89867127e-01 2.52442062e-01 -4.25214559e-01 -1.56008735e-01
-7.17355371e-01 9.91287604e-02 4.70426232e-01 1.04993832e+00
8.99987280e-01 1.31187454e-01 -1.44491658e-01 1.14808846e+00
2.06067681e-01 5.03443658e-01 6.72435939e-01 -7.16736674e-01
8.21374118e-01 4.96816427e-01 -5.21532781e-02 -1.21246636e+00
-3.64357203e-01 -8.29454124e-01 -1.38542330e+00 -3.66785586e-01
3.13820481e-01 -3.34466606e-01 -7.60789633e-01 2.07393956e+00
-7.39425793e-02 2.77051598e-01 -3.70379955e-01 8.59751999e-01
7.04292655e-01 1.02600980e+00 -1.34823561e-01 -2.38934606e-01
1.21956956e+00 -1.49056506e+00 -7.66501784e-01 -4.47028756e-01
5.52026749e-01 -5.21986902e-01 1.23936582e+00 3.95907730e-01
-1.37938869e+00 -7.34777987e-01 -1.07701564e+00 -3.82432997e-01
4.08432744e-02 1.60018638e-01 7.13245213e-01 3.32780510e-01
-1.03633964e+00 5.33959806e-01 -8.22176456e-01 1.21852964e-01
3.80017012e-01 6.72918618e-01 -2.95203596e-01 -1.30650014e-01
-1.30930758e+00 5.59257388e-01 3.59988421e-01 3.83612216e-01
-6.80658162e-01 -8.44215095e-01 -5.88454068e-01 6.09420121e-01
4.25773948e-01 -5.95727026e-01 9.81943548e-01 -8.34311485e-01
-1.43177319e+00 3.10713559e-01 -3.64589274e-01 -4.25151438e-01
7.55396560e-02 -7.52366543e-01 1.31672189e-01 -2.81891972e-01
-8.26161876e-02 5.67382872e-01 7.41844952e-01 -7.80971646e-01
-5.44608951e-01 -3.39991152e-01 5.12369201e-02 3.77465546e-01
-6.10611737e-01 -3.91397402e-02 -9.61832583e-01 -8.05631697e-01
1.25665590e-01 -9.91012275e-01 -3.97571653e-01 -3.51978719e-01
-2.84155220e-01 -3.63670051e-01 3.95739704e-01 -4.73365933e-01
1.56593871e+00 -2.21257472e+00 4.33303267e-01 7.69294873e-02
3.69130254e-01 4.23397720e-01 -2.33486533e-01 3.57412905e-01
-2.10866719e-01 6.82272390e-02 -2.15272188e-01 -5.86222470e-01
-1.95184741e-02 1.67137161e-01 -2.62808979e-01 3.15170228e-01
3.54148932e-02 7.42416561e-01 -7.27773130e-01 -3.21844161e-01
-2.57023841e-01 6.69857860e-01 -1.09323430e+00 2.66531467e-01
-5.79056814e-02 -5.64159788e-02 -8.56531262e-02 1.50739878e-01
6.56091511e-01 -5.49526036e-01 1.22761428e-01 -2.78226495e-01
6.29181787e-02 2.66431391e-01 -1.29169738e+00 1.79311574e+00
-5.35450995e-01 5.70786536e-01 1.06508031e-01 -1.03273761e+00
6.39430761e-01 -1.65055115e-02 3.80461127e-01 -3.13959837e-01
3.64223391e-01 2.20204368e-01 4.69754189e-01 -3.58655840e-01
4.08237696e-01 6.65306300e-02 1.57967314e-01 5.54501653e-01
1.94469467e-01 5.38592875e-01 4.91236925e-01 3.07048172e-01
1.11451030e+00 -2.99101800e-01 1.54957296e-02 -5.98196387e-01
5.50978124e-01 -4.84408379e-01 8.01303864e-01 7.54341185e-01
1.29027620e-01 4.39073503e-01 6.21329904e-01 -4.48423386e-01
-9.01651263e-01 -5.69493473e-01 3.30446243e-01 1.56248307e+00
-1.87478393e-01 -5.36207557e-01 -1.06817687e+00 -3.35314304e-01
-2.33159289e-01 2.91284084e-01 -7.57582724e-01 7.17220381e-02
-7.31351078e-01 -1.05778706e+00 3.40398610e-01 8.06811213e-01
4.81064498e-01 -1.02129436e+00 -3.47645223e-01 3.61111492e-01
-3.39441746e-01 -7.30215669e-01 -1.18220615e+00 5.12333512e-01
-6.73732519e-01 -6.76843584e-01 -8.17530870e-01 -8.97826195e-01
8.50540161e-01 4.03605074e-01 9.02342200e-01 1.60730943e-01
-6.51299879e-02 -3.94698739e-01 -2.19226152e-01 -2.00145736e-01
4.02819037e-01 3.81561697e-01 2.63979584e-01 2.83986211e-01
4.34184909e-01 -6.67987704e-01 -6.69345140e-01 1.84003398e-01
-8.31993401e-01 7.77204558e-02 8.45597446e-01 1.21614158e+00
6.01527750e-01 2.30297241e-02 4.40819651e-01 -8.90897930e-01
6.80906951e-01 -5.08394957e-01 -6.70679152e-01 -6.39487803e-02
-4.31087345e-01 5.39650738e-01 9.22144115e-01 -6.94473922e-01
-7.23634303e-01 4.95051779e-02 -2.73722559e-01 -7.10636497e-01
5.59289277e-01 9.11973000e-01 -3.08580577e-01 1.64853677e-01
4.77474302e-01 2.44792268e-01 -5.24928272e-02 -7.11883426e-01
3.22115481e-01 5.69999039e-01 2.92656720e-01 -5.43643713e-01
4.81349230e-01 5.81967905e-02 -2.10291117e-01 -8.11248302e-01
-9.42492485e-01 -3.46330941e-01 -2.97156155e-01 3.35180700e-01
5.30628502e-01 -9.21976209e-01 -1.03410733e+00 6.75063193e-01
-1.15525651e+00 -2.96309143e-01 1.88825931e-02 5.43600798e-01
-2.96161503e-01 2.32513368e-01 -8.40144157e-01 -6.73401535e-01
-7.22251594e-01 -1.43538928e+00 4.13046807e-01 4.87349816e-02
7.01760575e-02 -4.89200622e-01 7.29117729e-03 3.72442037e-01
5.83855093e-01 -5.47448575e-01 9.96705115e-01 -5.80160141e-01
-6.40858352e-01 -3.93665805e-02 -4.88600641e-01 3.67856055e-01
-1.89083993e-01 -2.32442737e-01 -8.02327335e-01 -3.57843727e-01
4.63892147e-02 -3.00004244e-01 1.08969617e+00 4.32163417e-01
1.38577402e+00 -4.73006010e-01 -1.14455000e-01 8.47870708e-01
1.32911754e+00 4.83024210e-01 4.19806987e-01 -1.23394355e-02
1.06375647e+00 2.39788890e-01 2.23775044e-01 3.50127697e-01
3.32725316e-01 4.07837600e-01 2.32122049e-01 -1.05754728e-03
1.68341473e-01 -1.30582690e-01 3.63357097e-01 1.58377755e+00
-1.14280745e-01 -3.94820005e-01 -9.85310316e-01 4.75883842e-01
-2.13758707e+00 -1.02551389e+00 3.98382992e-02 1.96355736e+00
1.00001347e+00 2.65951753e-01 -7.12181032e-02 2.58371502e-01
5.86762965e-01 1.49108917e-01 -7.12659478e-01 -2.65289307e-01
-8.04343377e-04 3.42115968e-01 5.36610901e-01 8.31581473e-01
-9.84674633e-01 9.41543460e-01 5.91886330e+00 9.63474393e-01
-1.06758618e+00 1.36943281e-01 6.51670396e-01 -4.90579307e-01
-7.79730082e-02 -3.15329731e-01 -1.09386730e+00 4.55254197e-01
8.77386808e-01 1.01141453e-01 6.25984430e-01 8.49355280e-01
-1.26735866e-01 2.08963141e-01 -1.02943122e+00 1.31898439e+00
2.33391270e-01 -1.29706287e+00 9.76520553e-02 -2.18549654e-01
5.47724605e-01 3.73855680e-01 1.13235787e-01 5.38242161e-01
3.87779713e-01 -9.73073840e-01 6.23087525e-01 2.69632995e-01
6.91537142e-01 -1.00548065e+00 7.27821648e-01 3.78625661e-01
-1.20840490e+00 -3.10874850e-01 -5.56932211e-01 -1.40587494e-01
6.10673986e-02 6.19769752e-01 -3.03429872e-01 8.58100802e-02
8.74502063e-01 6.20218873e-01 -4.63724256e-01 7.95445800e-01
7.73877054e-02 5.05194426e-01 -5.71622491e-01 -3.34845424e-01
4.28721994e-01 -2.89603412e-01 2.50986457e-01 1.11426985e+00
3.70700747e-01 2.44754165e-01 -4.98289093e-02 5.42836905e-01
-4.00605321e-01 3.93216223e-01 -9.24441740e-02 -2.93385834e-01
3.68448257e-01 1.04858565e+00 -5.19180238e-01 -4.52092826e-01
-4.62815255e-01 8.33743930e-01 9.71663892e-01 2.81012565e-01
-1.01611924e+00 -8.72742176e-01 8.64061713e-01 7.54301101e-02
5.84094703e-01 -3.58424604e-01 -1.50031760e-01 -1.36252856e+00
3.75904702e-02 -1.09140325e+00 1.76128715e-01 -6.89919353e-01
-1.27073562e+00 1.08339226e+00 -2.35921323e-01 -6.65577114e-01
7.85230473e-02 -5.30010521e-01 -3.23942125e-01 9.73700047e-01
-1.07730579e+00 -7.40643799e-01 -1.81859732e-01 5.84858894e-01
5.79379559e-01 -1.04678363e-01 5.94424546e-01 8.62653613e-01
-1.07940555e+00 1.03028369e+00 2.99206406e-01 1.50325671e-01
6.03384852e-01 -1.13965380e+00 5.48545122e-01 8.20819438e-01
2.86379933e-01 1.22954249e+00 6.42634630e-01 -4.30876225e-01
-1.51341593e+00 -8.86885583e-01 8.73642027e-01 4.97481637e-02
6.84001148e-01 -5.87121427e-01 -1.13449383e+00 8.91389191e-01
4.13998097e-01 1.98290378e-01 8.20322275e-01 4.88040835e-01
-5.45632005e-01 -3.49451751e-01 -4.19004023e-01 7.28006899e-01
1.26017261e+00 -3.03341538e-01 -3.90944093e-01 1.49759367e-01
9.93217945e-01 -3.33689600e-01 -6.33293688e-01 1.89278752e-01
5.18988431e-01 -8.43702257e-01 7.80843556e-01 -6.38173819e-01
3.87082338e-01 -3.17217231e-01 -2.51304477e-01 -1.40261388e+00
-8.93217742e-01 -8.03070247e-01 -2.84749031e-01 1.00227606e+00
6.78120494e-01 -5.55718005e-01 9.10142004e-01 5.40316403e-01
-2.51676172e-01 -1.09830379e+00 -7.34839618e-01 -4.23525006e-01
2.27923300e-02 -3.91488940e-01 6.13219559e-01 8.35838079e-01
-4.20364887e-02 9.41910565e-01 -6.89715624e-01 -1.23079354e-02
5.07977784e-01 5.37328906e-02 7.63351381e-01 -9.32551682e-01
-7.49624252e-01 -4.50558394e-01 -5.58943935e-02 -1.61142719e+00
-1.39759174e-02 -9.45557058e-01 1.91339791e-01 -1.28081787e+00
3.40461522e-01 -5.74328721e-01 -6.59315705e-01 5.41022897e-01
-6.64920151e-01 9.39795971e-02 2.66499221e-01 3.82606462e-02
-5.49814045e-01 7.82612860e-01 9.02662992e-01 -2.00221211e-01
-2.40557939e-01 -3.88376981e-01 -9.72147822e-01 7.67483652e-01
7.24625349e-01 -5.61315358e-01 -5.45421243e-01 -1.07895601e+00
4.69549090e-01 -1.47210270e-01 -4.43895578e-01 -1.16853380e+00
4.98120338e-01 -8.89298618e-02 6.67054728e-02 -7.02602506e-01
4.06579167e-01 -7.49366224e-01 1.18490323e-01 4.08629924e-01
-3.74039769e-01 3.93253118e-01 1.62986815e-01 3.95422131e-01
-1.33887619e-01 -1.98342472e-01 9.04891074e-01 -2.31267163e-03
-1.55273229e-01 7.22617209e-01 -1.70098901e-01 2.55478531e-01
6.21425390e-01 3.37179780e-01 -1.72045171e-01 -2.20404238e-01
-5.30103326e-01 2.21695289e-01 2.99074706e-02 3.34073842e-01
4.35356736e-01 -1.30287504e+00 -5.86376190e-01 4.69382972e-01
-3.01184297e-01 3.91693145e-01 6.11856937e-01 8.25670481e-01
-6.04838073e-01 5.99310100e-01 -2.32044384e-01 -4.80481178e-01
-1.19493675e+00 9.45966005e-01 1.10600419e-01 -4.46464211e-01
-3.02745461e-01 1.54865563e+00 4.50528413e-01 -6.44864812e-02
5.25538325e-01 -4.94198471e-01 -4.99699228e-02 2.05074906e-01
7.66820967e-01 2.40929514e-01 -1.57248229e-01 -6.80820584e-01
-2.62519240e-01 6.88490808e-01 -5.68609297e-01 -4.30227593e-02
1.40455651e+00 1.48825869e-02 -2.96025604e-01 4.34009612e-01
1.45945799e+00 -3.49129289e-02 -1.17283046e+00 -6.73853755e-01
-2.68066585e-01 -3.38575810e-01 1.52388051e-01 -4.11448367e-02
-1.50672567e+00 1.19871676e+00 3.29618275e-01 9.33978930e-02
1.14093399e+00 -4.54524696e-01 9.54262853e-01 7.25471020e-01
1.66048899e-01 -8.89675379e-01 6.06820248e-02 8.78945827e-01
9.79269087e-01 -9.98434961e-01 -1.27831846e-01 -1.89814553e-01
-5.60385466e-01 4.65544075e-01 9.52544928e-01 -4.11354393e-01
9.02486145e-01 2.72310376e-01 -1.25054970e-01 1.07065991e-01
-1.05511570e+00 9.44298729e-02 8.29463229e-02 -1.81069188e-02
5.68485856e-01 -2.38479421e-01 -3.85636210e-01 9.22733724e-01
-2.01806322e-01 -1.04816779e-02 1.97025865e-01 6.60705626e-01
-4.63189989e-01 -1.04079235e+00 -1.13157339e-01 5.69017708e-01
-7.93193996e-01 -5.89083135e-01 1.59580588e-01 2.97187328e-01
7.17654079e-02 5.84454894e-01 1.97713330e-01 -5.47966242e-01
1.91765979e-01 1.59828857e-01 2.17602059e-01 -5.09748101e-01
-8.52291822e-01 1.61878452e-01 -2.76969999e-01 -5.46217859e-01
-4.60048579e-02 -3.88824463e-01 -1.17684245e+00 -3.30560952e-01
-3.97645205e-01 4.03988600e-01 4.04333413e-01 6.38627291e-01
6.74183667e-01 6.78684235e-01 3.01755875e-01 -8.59059751e-01
-5.18239856e-01 -1.31200492e+00 -3.58470827e-01 3.38035852e-01
3.37539375e-01 -6.91811442e-01 -3.05983484e-01 -2.82499045e-02] | [8.748310089111328, 4.429534435272217] |
b677a3f2-2641-4328-9ab0-f0723ddf1542 | mri-based-classification-of-idh-mutation-and | 2210.03779 | null | https://arxiv.org/abs/2210.03779v1 | https://arxiv.org/pdf/2210.03779v1.pdf | MRI-based classification of IDH mutation and 1p/19q codeletion status of gliomas using a 2.5D hybrid multi-task convolutional neural network | Isocitrate dehydrogenase (IDH) mutation and 1p/19q codeletion status are important prognostic markers for glioma. Currently, they are determined using invasive procedures. Our goal was to develop artificial intelligence-based methods to non-invasively determine these molecular alterations from MRI. For this purpose, pre-operative MRI scans of 2648 patients with gliomas (grade II-IV) were collected from Washington University School of Medicine (WUSM; n = 835) and publicly available datasets viz. Brain Tumor Segmentation (BraTS; n = 378), LGG 1p/19q (n = 159), Ivy Glioblastoma Atlas Project (Ivy GAP; n = 41), The Cancer Genome Atlas (TCGA; n = 461), and the Erasmus Glioma Database (EGD; n = 774). A 2.5D hybrid convolutional neural network was proposed to simultaneously localize the tumor and classify its molecular status by leveraging imaging features from MR scans and prior knowledge features from clinical records and tumor location. The models were tested on one internal (TCGA) and two external (WUSM and EGD) test sets. For IDH, the best-performing model achieved areas under the receiver operating characteristic (AUROC) of 0.925, 0.874, 0.933 and areas under the precision-recall curves (AUPRC) of 0.899, 0.702, 0.853 on the internal, WUSM, and EGD test sets, respectively. For 1p/19q, the best model achieved AUROCs of 0.782, 0.754, 0.842, and AUPRCs of 0.588, 0.713, 0.782, on those three data-splits, respectively. The high accuracy of the model on unseen data showcases its generalization capabilities and suggests its potential to perform a 'virtual biopsy' for tailoring treatment planning and overall clinical management of gliomas. | ['Aristeidis Sotiras', 'Daniel S. Marcus', 'Joshua Shimony', 'Pamela Lamontagne', 'Satrajit Chakrabarty'] | 2022-10-07 | null | null | null | null | ['tumor-segmentation', 'brain-tumor-segmentation'] | ['computer-vision', 'medical'] | [ 1.13739505e-01 2.13708133e-01 -3.94940346e-01 -2.79788077e-01
-1.02887499e+00 -4.25932944e-01 4.31867659e-01 6.91957653e-01
-7.00612485e-01 8.32982779e-01 1.46664172e-01 -8.38437259e-01
-4.25427049e-01 -7.31974483e-01 -2.13253766e-01 -1.02906621e+00
-3.48719031e-01 5.48561215e-01 1.29303068e-01 1.51370570e-01
2.10002050e-01 5.53793252e-01 -1.11148393e+00 1.33678600e-01
1.07234979e+00 1.15159953e+00 4.33284968e-01 8.03140163e-01
9.84454677e-02 3.82601470e-01 -2.12914035e-01 -6.08454179e-03
-1.56810984e-01 -3.69694345e-02 -8.23085010e-01 -2.99961209e-01
-1.43551841e-01 -5.68786561e-02 -3.50458115e-01 1.17786455e+00
5.09056926e-01 -3.26285303e-01 8.52320910e-01 -9.77521420e-01
-1.72857210e-01 4.50647295e-01 -7.33680785e-01 3.29708755e-01
-1.25668883e-01 2.13433713e-01 7.64268577e-01 -4.96602923e-01
7.12831497e-01 2.31616974e-01 7.56695509e-01 4.57313746e-01
-7.94831038e-01 -5.96102953e-01 -1.10709950e-01 1.12406999e-01
-1.48979044e+00 -1.10684514e-01 -1.34728611e-01 -7.28178144e-01
9.40295935e-01 4.02225018e-01 9.77286637e-01 5.50559700e-01
7.99557149e-01 5.29610097e-01 1.28156602e+00 -2.38966495e-01
4.40636337e-01 -2.18168557e-01 3.64210486e-01 9.55467284e-01
1.66825384e-01 -2.96547152e-02 -2.57686049e-01 -1.81969240e-01
4.94137973e-01 6.35951832e-02 -5.50683379e-01 -2.98600569e-02
-1.02637494e+00 7.23118663e-01 5.77221155e-01 3.44075173e-01
-2.48236939e-01 -8.93087313e-02 5.87294936e-01 1.28818139e-01
3.87214303e-01 7.61954188e-02 -3.60111117e-01 4.22335677e-02
-4.99623984e-01 -1.51091337e-01 3.21950376e-01 5.94699860e-01
2.83537179e-01 -4.92641509e-01 8.13058764e-02 8.85666609e-01
3.84700149e-01 1.70991257e-01 1.35163164e+00 -2.40881592e-01
-6.12504501e-03 6.31185889e-01 -1.00921579e-01 -3.86428803e-01
-1.40247059e+00 -6.44766808e-01 -9.51604903e-01 -7.54225999e-02
4.42618757e-01 5.38454019e-03 -1.26254451e+00 1.69785202e+00
-5.73862791e-02 -1.55446067e-01 2.01462910e-01 8.40419292e-01
1.02851808e+00 -1.06629077e-02 1.30793124e-01 -8.57446566e-02
1.65508890e+00 -5.91616392e-01 -2.04433277e-01 -7.15447143e-02
1.39606392e+00 -2.09713727e-01 5.99103034e-01 3.06218982e-01
-6.97721303e-01 1.87217891e-01 -8.74404550e-01 2.94372499e-01
-3.94732475e-01 1.19219460e-01 7.81830430e-01 7.96118855e-01
-1.37477660e+00 8.96770135e-02 -1.23708761e+00 -6.14313722e-01
7.70957112e-01 6.85484052e-01 -6.98305309e-01 -2.96578199e-01
-1.06345999e+00 9.60069656e-01 5.59149981e-01 -2.33874649e-01
-1.17663264e+00 -8.56555104e-01 -3.93631548e-01 -1.80667013e-01
-1.87583163e-01 -7.57147789e-01 1.06740296e+00 -6.70115113e-01
-1.19169998e+00 1.16874456e+00 4.46599647e-02 -4.50195819e-01
3.01533401e-01 6.71854913e-01 -3.75464916e-01 3.85000795e-01
2.34436885e-01 6.92534745e-01 -7.77234286e-02 -3.86304766e-01
-8.86132836e-01 -1.09806156e+00 -3.18972379e-01 3.34350944e-01
-1.39090508e-01 -1.81396961e-01 -1.22971423e-01 -3.09780419e-01
4.46509242e-01 -9.41045880e-01 -3.93247873e-01 -3.24324548e-01
-3.55976969e-01 -2.00780947e-02 4.04307663e-01 -9.80277419e-01
5.74553907e-01 -1.91952014e+00 6.24194071e-02 4.26880151e-01
5.86824536e-01 1.19781695e-01 7.21379519e-02 -1.14434674e-01
-3.47082376e-01 1.08793207e-01 -9.86923277e-02 2.40145996e-01
-2.94550747e-01 -9.92964804e-02 7.35264897e-01 9.02403593e-01
-1.53271422e-01 1.12847960e+00 -8.50895286e-01 -3.89881688e-03
7.06274733e-02 2.37949625e-01 -3.68902564e-01 -1.16624780e-01
1.84171289e-01 3.83074284e-01 -2.64609754e-01 7.45109499e-01
1.88791990e-01 -1.32642567e-01 5.85473292e-02 1.21998355e-01
5.45988269e-02 -2.01602176e-01 -3.63301337e-01 1.80382037e+00
-1.39369234e-01 8.31585348e-01 5.90911247e-02 -9.29659069e-01
9.94231462e-01 4.11960453e-01 6.69834256e-01 -5.79401910e-01
2.82770485e-01 4.18500662e-01 5.01117468e-01 -5.75541079e-01
-2.11108014e-01 -3.82461905e-01 1.53652206e-01 1.57449111e-01
8.22504088e-02 -1.61618754e-01 9.88893062e-02 7.08354041e-02
1.77231264e+00 -6.72899723e-01 4.21781272e-01 -4.27017272e-01
4.97081578e-01 4.61190045e-02 4.22303796e-01 5.64150214e-01
-4.96902943e-01 6.68169677e-01 7.34965265e-01 -3.72867852e-01
-6.22626007e-01 -9.25386012e-01 -7.31565773e-01 4.50419307e-01
-2.64951169e-01 -9.24795046e-02 -7.46525168e-01 -5.18843055e-01
-1.46082580e-01 5.16712427e-01 -7.53765941e-01 -3.36731225e-01
1.59385204e-01 -1.31334949e+00 8.52109432e-01 4.78111118e-01
4.37785000e-01 -3.24790031e-01 -6.60213232e-01 2.74874866e-01
2.15485264e-02 -7.81918526e-01 6.44874498e-02 6.07283711e-01
-8.25076044e-01 -1.21847737e+00 -9.26441967e-01 -5.37610888e-01
8.18738699e-01 -2.31009588e-01 4.17881489e-01 1.40441367e-02
-4.27805841e-01 1.35938004e-01 -3.11498404e-01 -5.82549632e-01
-2.33932048e-01 2.41476983e-01 1.52914435e-01 -1.44466072e-01
5.76291084e-01 -6.81578100e-01 -5.10894299e-01 3.04636747e-01
-7.53159583e-01 2.90746570e-01 8.32229495e-01 9.77702916e-01
6.07438624e-01 8.91374722e-02 7.42785633e-01 -5.94469845e-01
3.49528342e-01 -5.32552242e-01 -6.96624041e-01 -9.71003156e-03
-7.65210629e-01 -5.65256894e-01 4.11110640e-01 -5.31384610e-02
-6.32085800e-01 5.18165268e-02 -2.68492758e-01 1.61476079e-02
-5.12751043e-01 8.71386766e-01 -3.48588340e-02 -1.49282485e-01
6.65487051e-01 -1.90835446e-02 2.57053375e-01 4.36627381e-02
-2.72386014e-01 8.65058064e-01 7.27094233e-01 -1.76974773e-01
6.23380616e-02 3.42318147e-01 2.38524526e-01 -6.86776936e-01
-5.15549362e-01 -6.43587291e-01 -5.00884593e-01 -8.98934901e-02
8.21695328e-01 -7.85920203e-01 -5.56556463e-01 6.82898462e-01
-5.10261238e-01 -3.81377548e-01 -4.20957357e-02 8.98817956e-01
-5.77201188e-01 -6.69989660e-02 -7.16805100e-01 -2.92746663e-01
-6.91662848e-01 -1.47381914e+00 8.49902570e-01 2.70041257e-01
-2.88594306e-01 -8.78484070e-01 4.55452316e-02 3.34105492e-01
4.35271472e-01 5.53151369e-01 1.33228624e+00 -1.06984520e+00
-4.82942611e-02 -8.03207994e-01 -3.96331072e-01 -1.24108896e-01
2.80604541e-01 -1.02369234e-01 -1.05359101e+00 -3.08295012e-01
-3.28045100e-01 -1.18196525e-01 6.82408512e-01 6.19307160e-01
1.12364972e+00 7.17886239e-02 -8.31317961e-01 8.02014232e-01
1.38922834e+00 8.14031780e-01 5.63457251e-01 5.39516211e-01
3.36560547e-01 3.19937706e-01 2.63921581e-02 1.87454000e-01
3.09877187e-01 3.20114732e-01 6.95448518e-01 1.98560655e-01
-2.18350701e-02 7.93904513e-02 -1.28262430e-01 2.64489502e-01
-1.67717412e-01 -2.31157884e-01 -1.61911154e+00 6.82812691e-01
-1.37905252e+00 -3.59058261e-01 -2.51467675e-01 1.98570144e+00
6.24047875e-01 2.66687661e-01 -2.20811546e-01 2.85909086e-01
6.13236964e-01 -3.90259683e-01 -6.47210717e-01 -2.43168190e-01
-9.39183682e-02 4.65698764e-02 5.84125698e-01 2.23402739e-01
-8.44170451e-01 3.87516290e-01 4.87499094e+00 5.35284400e-01
-1.02347338e+00 2.55095541e-01 1.20760453e+00 -1.27695024e-01
5.53463884e-02 -3.75119627e-01 -4.63611782e-01 3.66922706e-01
1.32262385e+00 -3.03207964e-01 1.60499498e-01 4.19682145e-01
1.68924049e-01 -4.35298264e-01 -7.87265360e-01 8.57978165e-01
-1.90145329e-01 -1.51760685e+00 -5.60643733e-01 4.08185482e-01
6.76077068e-01 7.46139407e-01 -4.81656007e-02 1.05550990e-01
2.94903517e-01 -1.29878724e+00 2.10374504e-01 6.21561527e-01
1.18284619e+00 -9.13302302e-01 1.35548711e+00 4.77042258e-01
-5.01325846e-01 -7.62488053e-05 -1.36204530e-02 1.03943720e-01
-1.04391322e-01 7.00240970e-01 -1.31228638e+00 5.13083577e-01
6.66581690e-01 3.82377565e-01 -5.44739246e-01 1.08795190e+00
-1.84124023e-01 5.80778718e-01 -1.31442204e-01 -1.28698379e-01
4.07970130e-01 2.17255786e-01 2.21715346e-01 9.44753468e-01
2.27428541e-01 6.07845664e-01 -2.56500453e-01 4.01907951e-01
3.41887102e-02 1.56385288e-01 -9.77078676e-02 -7.98629820e-02
1.44923985e-01 1.40592420e+00 -1.09491575e+00 1.73373939e-03
-4.42631394e-01 2.16135398e-01 3.38121951e-01 2.26087868e-01
-5.11772811e-01 -5.73544264e-01 4.41971689e-01 1.34518772e-01
-2.74416596e-01 2.11982340e-01 -4.32078093e-01 -9.18364048e-01
-5.61710894e-01 -5.99841356e-01 7.06346452e-01 -9.35144424e-01
-1.02702832e+00 5.16380250e-01 -1.61195740e-01 -6.47430539e-01
-1.18091917e-02 -8.92886400e-01 -6.16725743e-01 1.16904008e+00
-1.08676529e+00 -8.27754319e-01 -4.49358225e-01 1.52474031e-01
8.48574340e-02 -3.37613642e-01 9.97805536e-01 -1.52251229e-01
-6.59428895e-01 6.69895351e-01 1.97398171e-01 2.69791424e-01
1.46450281e-01 -1.09187162e+00 -1.44330367e-01 2.76754916e-01
-6.68469608e-01 2.78188765e-01 1.75389633e-01 -4.00944531e-01
-1.17350817e+00 -1.25464964e+00 5.85891366e-01 -5.53730987e-02
8.35611165e-01 1.83794692e-01 -8.25032949e-01 8.09068024e-01
-1.80234998e-01 8.87395814e-02 1.04005623e+00 -2.79656947e-01
-1.54239563e-02 1.23188309e-01 -1.62283063e+00 3.41037184e-01
6.41794324e-01 -2.47813731e-01 -2.04700336e-01 6.28853083e-01
4.29040432e-01 -7.15119839e-01 -1.52697670e+00 5.48635066e-01
6.65117264e-01 -6.36454225e-01 5.94327867e-01 -4.66470003e-01
2.58548617e-01 -1.57767132e-01 -2.76111126e-01 -1.53562760e+00
-5.66373646e-01 -4.50618789e-02 3.86169702e-01 6.47842348e-01
6.67387903e-01 -1.14239371e+00 8.60966563e-01 8.46780419e-01
-7.12710500e-01 -1.34731615e+00 -1.19563997e+00 -4.37996209e-01
3.13880175e-01 -4.00397599e-01 7.87605226e-01 7.99387097e-01
3.56045276e-01 -1.21082067e-01 6.87254012e-01 3.67952913e-01
2.04312339e-01 -4.13327545e-01 7.89464563e-02 -1.07895041e+00
1.06463648e-01 -8.47169459e-01 -1.11435282e+00 1.97846983e-02
1.79924384e-01 -1.40620971e+00 -5.42802155e-01 -1.69772208e+00
5.03560603e-01 -5.92801273e-01 -6.64346218e-01 7.06148088e-01
8.06505829e-02 6.18559904e-02 -5.64741433e-01 1.45436004e-01
9.58582386e-02 1.80735424e-01 7.38010585e-01 -3.04383069e-01
1.08343540e-02 -2.63595171e-02 -6.88342273e-01 8.93172503e-01
1.17753398e+00 -2.63873428e-01 -3.92028630e-01 -2.57231623e-01
-1.96188837e-01 3.70474130e-01 2.84768343e-01 -1.07296312e+00
1.86274350e-01 -1.76855072e-01 7.82653391e-01 -5.34750938e-01
9.26580355e-02 -3.27076077e-01 2.95000553e-01 9.12526846e-01
-2.95408279e-01 1.03362463e-01 2.48684332e-01 3.69631350e-01
-4.48316932e-02 -3.85671526e-01 8.41794789e-01 -2.27017641e-01
-5.80448270e-01 6.68103278e-01 -8.12412620e-01 -2.23929539e-01
1.21548283e+00 -3.22863281e-01 -6.12370670e-01 -2.05966473e-01
-8.51323426e-01 1.15398481e-01 4.31688547e-01 1.18761435e-01
5.82964182e-01 -1.02530587e+00 -7.65871346e-01 2.23059222e-01
4.24910575e-01 1.07839726e-01 2.75047600e-01 1.45016563e+00
-5.37609220e-01 5.90636075e-01 -2.08584994e-01 -7.65891194e-01
-1.01003003e+00 2.23901674e-01 7.33751714e-01 -1.68222681e-01
-5.97659111e-01 1.20954514e+00 2.29882509e-01 -5.00526547e-01
1.56864241e-01 -4.17276144e-01 -2.18411312e-01 -6.85108081e-02
6.57876611e-01 2.06610411e-01 7.55959451e-01 -5.35503030e-01
-4.30818915e-01 1.77922592e-01 -3.06447208e-01 -6.48496747e-02
1.35036075e+00 2.12944582e-01 -3.86386245e-01 1.65988371e-01
1.32247722e+00 -6.03914559e-01 -5.14522254e-01 -5.15450500e-02
6.26569092e-02 -1.79455280e-01 4.86822844e-01 -1.12200117e+00
-1.28831410e+00 4.94060218e-01 1.05158925e+00 -4.29051146e-02
1.40628254e+00 1.34721354e-01 3.53469551e-01 -3.19160335e-02
5.56539834e-01 -6.49692297e-01 -6.00093365e-01 3.15650284e-01
5.73694527e-01 -8.38980436e-01 -3.04814488e-01 -1.08650699e-01
-1.87592536e-01 1.20134747e+00 4.96116638e-01 4.03358817e-01
5.01755714e-01 2.81937301e-01 -8.64754338e-03 -3.58070403e-01
-1.00354517e+00 2.41885915e-01 -6.85084090e-02 6.67204499e-01
5.28355479e-01 6.83698356e-01 -5.34892976e-01 1.00637591e+00
-4.39060390e-01 1.26902506e-01 7.08494365e-01 9.22662437e-01
-6.24299705e-01 -6.41483784e-01 -1.29340008e-01 1.05395269e+00
-3.81828010e-01 2.53890548e-02 -3.64538729e-01 8.63906741e-01
-1.56946570e-01 5.95552564e-01 5.72784655e-02 -3.13754767e-01
4.24909592e-02 1.57135725e-01 1.86025932e-01 -3.26094151e-01
-3.47270846e-01 3.05290315e-02 2.36487329e-01 -1.43676633e-02
-6.29410073e-02 -6.91229761e-01 -1.58675635e+00 7.80363902e-02
-4.83450353e-01 2.80007422e-01 7.71671772e-01 8.69603038e-01
2.11737677e-01 6.81824088e-01 2.89843082e-01 -3.21072400e-01
-1.73963085e-01 -9.91120815e-01 -9.23837900e-01 -2.53558517e-01
1.47451311e-01 -6.46196902e-01 -1.30434483e-01 -4.25177753e-01] | [14.785070419311523, -2.568054437637329] |
8b866799-29e6-414e-8c84-d9df9bba8b3e | deep-variation-structured-reinforcement | 1703.03054 | null | http://arxiv.org/abs/1703.03054v1 | http://arxiv.org/pdf/1703.03054v1.pdf | Deep Variation-structured Reinforcement Learning for Visual Relationship and Attribute Detection | Despite progress in visual perception tasks such as image classification and
detection, computers still struggle to understand the interdependency of
objects in the scene as a whole, e.g., relations between objects or their
attributes. Existing methods often ignore global context cues capturing the
interactions among different object instances, and can only recognize a handful
of types by exhaustively training individual detectors for all possible
relationships. To capture such global interdependency, we propose a deep
Variation-structured Reinforcement Learning (VRL) framework to sequentially
discover object relationships and attributes in the whole image. First, a
directed semantic action graph is built using language priors to provide a rich
and compact representation of semantic correlations between object categories,
predicates, and attributes. Next, we use a variation-structured traversal over
the action graph to construct a small, adaptive action set for each step based
on the current state and historical actions. In particular, an ambiguity-aware
object mining scheme is used to resolve semantic ambiguity among object
categories that the object detector fails to distinguish. We then make
sequential predictions using a deep RL framework, incorporating global context
cues and semantic embeddings of previously extracted phrases in the state
vector. Our experiments on the Visual Relationship Detection (VRD) dataset and
the large-scale Visual Genome dataset validate the superiority of VRL, which
can achieve significantly better detection results on datasets involving
thousands of relationship and attribute types. We also demonstrate that VRL is
able to predict unseen types embedded in our action graph by learning
correlations on shared graph nodes. | ['Lisa Lee', 'Xiaodan Liang', 'Eric P. Xing'] | 2017-03-08 | deep-variation-structured-reinforcement-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Liang_Deep_Variation-Structured_Reinforcement_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Liang_Deep_Variation-Structured_Reinforcement_CVPR_2017_paper.pdf | cvpr-2017-7 | ['visual-relationship-detection'] | ['computer-vision'] | [ 2.46471867e-01 -1.21127807e-01 -3.14609140e-01 -6.17853224e-01
-3.47134799e-01 -7.19526350e-01 5.35977125e-01 5.43664157e-01
-3.14074606e-01 3.04015577e-01 4.55043167e-02 -1.31116346e-01
-2.58098423e-01 -8.14898074e-01 -8.29207480e-01 -4.32762563e-01
-3.33699405e-01 7.65713155e-01 4.59248930e-01 2.69013532e-02
1.61980480e-01 3.23574603e-01 -1.81342804e+00 3.89954865e-01
5.52926064e-01 1.08093643e+00 5.52869856e-01 5.90835214e-01
-2.03043953e-01 7.27630138e-01 -5.11124372e-01 -1.00683644e-01
1.99552521e-01 -2.94208884e-01 -7.29848802e-01 5.97551465e-01
5.64067185e-01 -3.35336000e-01 -3.50468725e-01 1.03898764e+00
-1.18209226e-02 1.64079636e-01 5.35570562e-01 -1.41541183e+00
-7.76762009e-01 5.03227890e-01 -5.54551899e-01 3.61540616e-01
4.32255119e-01 4.79792982e-01 1.58946121e+00 -7.39825547e-01
7.43916214e-01 1.47368288e+00 6.44835308e-02 4.51248258e-01
-1.57090080e+00 -4.93609041e-01 9.22631443e-01 6.71327651e-01
-1.20377982e+00 -1.19827446e-02 7.51426220e-01 -5.35970211e-01
1.30012417e+00 4.76806723e-02 8.60658765e-01 1.20649278e+00
3.68455797e-02 8.69934916e-01 8.37849498e-01 -1.17947228e-01
4.94164169e-01 -7.81188440e-03 3.92396361e-01 8.80364716e-01
2.87416637e-01 -1.23019904e-01 -6.20970726e-01 -1.44745246e-01
6.88120663e-01 2.30276026e-04 2.46430650e-01 -7.63660550e-01
-1.20562506e+00 6.23654425e-01 6.72800720e-01 5.65474443e-02
-4.36243027e-01 1.68505579e-01 1.68947846e-01 1.73187226e-01
-1.42398283e-01 5.22702694e-01 -6.76379681e-01 2.94089526e-01
-1.08115733e-01 2.78264284e-01 6.56592488e-01 9.92409229e-01
1.11917210e+00 -4.10012305e-01 -2.82528698e-01 7.21921682e-01
3.59593302e-01 4.46804613e-01 1.53079122e-01 -8.55189145e-01
4.26308632e-01 1.02703309e+00 -3.36232297e-02 -1.22337389e+00
-3.96642834e-01 -3.90829027e-01 -4.35832143e-01 6.30066171e-02
1.05718262e-01 2.61386335e-01 -1.20389950e+00 2.12068176e+00
4.38598424e-01 2.75618911e-01 5.09619676e-02 9.34433877e-01
6.92724288e-01 4.23615485e-01 4.46258307e-01 1.47903174e-01
1.51724625e+00 -5.37468553e-01 -3.46346080e-01 -8.54265451e-01
4.21806902e-01 -1.74705073e-01 8.22101235e-01 6.33665174e-02
-6.27391160e-01 -6.20609105e-01 -8.26376379e-01 -9.38618779e-02
-4.01013881e-01 -1.20242126e-01 9.25029755e-01 -1.14891604e-02
-8.32035482e-01 2.63633668e-01 -8.73520017e-01 -4.08702433e-01
7.83266962e-01 4.67297435e-01 -3.78679395e-01 -2.00484604e-01
-9.27364111e-01 7.13224292e-01 6.99937403e-01 -4.28847112e-02
-1.26230884e+00 -3.85627478e-01 -1.05604446e+00 1.95542008e-01
9.51412976e-01 -8.00661504e-01 9.84253645e-01 -1.10860133e+00
-7.81510592e-01 8.40535164e-01 -3.44080180e-01 -5.88242173e-01
-4.89921570e-02 -1.00569457e-01 -2.50125051e-01 1.94703996e-01
2.83052832e-01 9.18433607e-01 9.08683240e-01 -1.45990419e+00
-9.43925560e-01 -5.81785500e-01 1.87728003e-01 4.83167142e-01
6.68278933e-02 -1.95474312e-01 -7.02139854e-01 -3.18798393e-01
3.79635036e-01 -9.48262811e-01 -2.49505460e-01 1.73561752e-01
-5.29968679e-01 -5.29548407e-01 7.34532118e-01 -4.05398965e-01
7.74784505e-01 -2.18833637e+00 3.63214970e-01 2.36965761e-01
3.58631551e-01 1.25636421e-02 -5.16261697e-01 1.24514788e-01
6.24445789e-02 8.96801706e-03 -2.09115893e-01 -2.24884644e-01
-7.06216767e-02 8.99769545e-01 -4.75138485e-01 1.75285742e-01
6.62062466e-01 1.05707347e+00 -1.19837737e+00 -2.27261245e-01
1.15643926e-01 1.85338244e-01 -6.01932526e-01 2.27591634e-01
-8.23500991e-01 2.95043111e-01 -7.94763029e-01 7.36790419e-01
2.99305618e-01 -6.94442809e-01 5.00910044e-01 -2.16259629e-01
3.38440031e-01 3.24587733e-01 -1.18536639e+00 1.62484443e+00
-2.32783288e-01 3.89112025e-01 -2.37452924e-01 -1.20978808e+00
8.84420514e-01 -1.50344476e-01 4.17605817e-01 -6.97191954e-01
-1.16210662e-01 -8.00703615e-02 2.39044309e-01 -5.77256382e-01
1.35638490e-01 2.86059260e-01 -1.18418753e-01 1.74085453e-01
2.52154559e-01 1.71377629e-01 1.92881912e-01 5.96013248e-01
1.27626777e+00 1.12005182e-01 3.63263816e-01 1.35888338e-01
3.35694611e-01 6.91299215e-02 7.66006649e-01 1.03544819e+00
-1.84307560e-01 2.83314198e-01 8.06786716e-01 -5.84286392e-01
-6.55066669e-01 -1.28380966e+00 1.51471481e-01 1.10381508e+00
6.75331593e-01 -5.42552948e-01 -1.05120957e-01 -9.94838595e-01
2.59061694e-01 7.34946072e-01 -7.48507082e-01 -4.14160967e-01
-4.20173675e-01 -6.37348175e-01 -2.28947595e-01 6.64524853e-01
3.57893288e-01 -1.33809459e+00 -6.84921026e-01 1.90933868e-01
3.59238365e-05 -1.44508779e+00 -6.70434907e-02 3.43774676e-01
-6.66063309e-01 -1.40134633e+00 4.66017365e-01 -9.11455750e-01
7.68168569e-01 1.75892055e-01 1.16448390e+00 1.35599509e-01
-4.22948271e-01 6.94335341e-01 -1.64583564e-01 -1.95164517e-01
-2.87897110e-01 -3.54913622e-01 -7.71207884e-02 2.49074653e-01
5.26310980e-01 -5.07052481e-01 -5.59862494e-01 3.29450846e-01
-5.69679081e-01 2.01120134e-03 8.17609131e-01 8.12961638e-01
9.03609216e-01 1.68821424e-01 4.87334192e-01 -7.98386693e-01
1.43051103e-01 -6.08253598e-01 -7.65174091e-01 4.64045852e-01
-4.16287333e-01 3.50924164e-01 3.17563146e-01 -5.38945079e-01
-8.90977085e-01 3.12309444e-01 3.27299356e-01 -5.73415101e-01
-3.93132150e-01 3.22620481e-01 -4.02982235e-01 4.13663387e-01
3.53555858e-01 3.74666035e-01 -2.54185855e-01 -2.86089003e-01
4.84720141e-01 3.78892757e-02 6.41704917e-01 -5.97633183e-01
7.34572709e-01 5.17565906e-01 3.14977169e-02 -5.50387383e-01
-1.13556433e+00 -6.24976575e-01 -6.96456730e-01 3.59477326e-02
1.08156121e+00 -9.75211918e-01 -8.31295013e-01 1.76924169e-01
-1.13129485e+00 -2.95438409e-01 -3.34238052e-01 3.09646219e-01
-5.50692737e-01 5.01041338e-02 -3.70811522e-01 -5.33496797e-01
3.13285053e-01 -1.14003062e+00 1.21216619e+00 1.94296092e-01
-1.81870043e-01 -8.25301051e-01 -2.28854328e-01 3.40039611e-01
-2.01082766e-01 8.62820745e-02 1.35381448e+00 -8.19297612e-01
-1.14880753e+00 2.66493522e-02 -2.67605364e-01 1.49580851e-01
2.86463141e-01 -5.28721064e-02 -6.16618991e-01 -1.27447829e-01
-4.05695140e-01 -4.97597903e-01 1.11051548e+00 2.05597460e-01
1.48652720e+00 -2.70027727e-01 -7.07420170e-01 3.41541618e-01
1.36474502e+00 3.16474020e-01 2.00570434e-01 1.44549862e-01
8.97567034e-01 6.08551621e-01 6.67992473e-01 3.73079002e-01
5.66451609e-01 5.44688821e-01 8.89716923e-01 2.00915024e-01
-1.06637843e-01 -5.59531331e-01 1.29695624e-01 1.11969054e-01
1.79745257e-01 -2.84531146e-01 -8.20628107e-01 4.88442332e-01
-1.96536458e+00 -9.20521498e-01 3.13764542e-01 2.12868953e+00
6.87217653e-01 3.88243884e-01 -3.93515341e-02 -3.11332703e-01
5.71436584e-01 1.50337100e-01 -1.07849646e+00 -2.02343091e-02
-1.78121820e-01 -1.47839189e-01 2.71043479e-01 4.76680219e-01
-1.17341340e+00 1.27216995e+00 5.60991764e+00 3.85918438e-01
-8.28786194e-01 -2.96177536e-01 6.08383894e-01 1.22764952e-01
-4.00785834e-01 3.66462618e-01 -9.80082929e-01 1.05060749e-01
1.77493751e-01 1.37714282e-01 4.28106844e-01 7.34645128e-01
-1.60534769e-01 -2.67726362e-01 -1.47257745e+00 8.40482175e-01
1.34303063e-01 -1.24018240e+00 4.19841886e-01 3.60593051e-02
4.70993042e-01 2.90592425e-02 1.09249301e-01 3.85886490e-01
6.78046703e-01 -7.97381997e-01 4.92587656e-01 3.72882634e-01
2.97180384e-01 -4.57897961e-01 1.81171834e-01 4.10272270e-01
-1.15665460e+00 -5.34058034e-01 -4.75680619e-01 -3.46817113e-02
-2.30931595e-01 1.94588929e-01 -1.13443089e+00 3.47830623e-01
6.72388673e-01 1.06940198e+00 -7.79469550e-01 7.60908544e-01
-6.70196354e-01 4.19843763e-01 -2.11889997e-01 -1.61234513e-02
1.60579979e-01 -9.43519101e-02 8.07950854e-01 6.48002863e-01
-2.22512379e-01 2.88204283e-01 5.54046214e-01 9.87702310e-01
-1.01009503e-01 -3.66285712e-01 -5.30330002e-01 -6.19079694e-02
5.20688176e-01 1.26539028e+00 -8.85259390e-01 -4.37362105e-01
-5.90067506e-01 8.21160316e-01 7.10168064e-01 5.16788483e-01
-6.54641032e-01 2.58950531e-01 1.02389932e+00 -2.52253413e-01
7.77483225e-01 -4.68138039e-01 2.51278691e-02 -1.17297161e+00
7.32354894e-02 -8.22927892e-01 8.19616258e-01 -7.75522947e-01
-1.39052868e+00 4.03722107e-01 1.08826995e-01 -9.83468175e-01
-2.40363717e-01 -6.68312311e-01 -2.85508513e-01 5.57539344e-01
-1.38125110e+00 -1.04911351e+00 -1.95960358e-01 6.66688144e-01
7.83752799e-01 -8.42452869e-02 6.71659768e-01 -3.30622137e-01
-5.65044940e-01 1.44550353e-01 -6.13498092e-01 2.53278553e-01
3.18440646e-01 -1.25360680e+00 3.22633147e-01 8.09399724e-01
7.78224111e-01 5.60256898e-01 6.13116324e-01 -8.76220882e-01
-1.53388357e+00 -1.13711047e+00 6.29534066e-01 -6.80193186e-01
8.27899754e-01 -7.31668353e-01 -1.23738229e+00 9.62704480e-01
-1.37146875e-01 4.36298698e-01 4.14674312e-01 3.32479477e-01
-8.82692158e-01 -1.94088578e-01 -7.08430588e-01 6.36331916e-01
1.49907970e+00 -5.98733664e-01 -8.83710146e-01 4.32091713e-01
8.49420488e-01 -2.23202541e-01 -3.65521759e-01 4.03322726e-01
2.97470957e-01 -7.23154843e-01 1.06309462e+00 -9.59276140e-01
3.54309231e-01 -4.25597578e-01 -3.23200047e-01 -1.08664775e+00
-5.51106513e-01 -1.86829925e-01 -2.82302111e-01 9.97338474e-01
4.24565107e-01 -5.25524676e-01 7.59050667e-01 6.03860795e-01
1.72282815e-01 -7.39392161e-01 -7.45828331e-01 -6.27945125e-01
-5.87353170e-01 -4.90751415e-01 4.63058561e-01 8.32427204e-01
-4.50937033e-01 7.28377283e-01 -2.23540254e-02 8.33062828e-01
7.56656289e-01 7.72750616e-01 5.92952728e-01 -1.18999076e+00
-6.61331296e-01 -2.85252839e-01 -8.25698614e-01 -1.06045628e+00
4.51852471e-01 -9.32830691e-01 9.90295634e-02 -1.62892950e+00
4.09710824e-01 -5.90450466e-01 -5.98074317e-01 9.38129783e-01
-4.71380413e-01 -1.33328363e-01 1.99971914e-01 1.62811399e-01
-9.94308054e-01 4.47787553e-01 1.20174325e+00 -4.84187752e-01
-2.77806312e-01 -1.34273291e-01 -6.48649216e-01 8.04549158e-01
5.76269925e-01 -5.36024511e-01 -7.20771432e-01 -5.88129938e-01
3.50233227e-01 -3.80077586e-02 8.55819821e-01 -6.93374872e-01
2.54718691e-01 -3.39102924e-01 5.05758226e-01 -5.66355169e-01
3.80823851e-01 -8.63296390e-01 -1.93529472e-01 3.55491281e-01
-6.15951359e-01 -1.30950972e-01 1.31512105e-01 1.13239682e+00
-3.92090641e-02 2.82854319e-01 4.20054972e-01 -1.35168418e-01
-1.38180768e+00 6.56124413e-01 -1.36893108e-01 7.65906200e-02
1.10709584e+00 -1.06806040e-01 -2.53080726e-01 -1.06803201e-01
-9.92345154e-01 7.10603297e-01 2.57894635e-01 7.82733738e-01
8.74659240e-01 -1.04062879e+00 -4.07859892e-01 3.66449386e-01
4.92445499e-01 2.54997820e-01 2.60236919e-01 2.80593872e-01
2.30969414e-01 1.66084114e-02 -2.95750827e-01 -9.84396815e-01
-1.35231471e+00 9.11751568e-01 3.43308240e-01 -7.28050917e-02
-7.32536793e-01 9.37187672e-01 7.30599582e-01 -8.56984407e-02
1.86290920e-01 -4.22571748e-01 -3.89449060e-01 2.80578807e-03
2.93396890e-01 -2.46668980e-01 -3.01939547e-01 -5.29879332e-01
-5.92897654e-01 5.79504669e-01 -2.39726290e-01 1.88429803e-01
1.29423320e+00 -1.75983980e-01 -7.72561356e-02 5.57768643e-01
1.01333737e+00 -2.07189620e-01 -1.68397295e+00 -6.02330208e-01
1.90943807e-01 -5.55164814e-01 -2.36367062e-01 -8.28548968e-01
-8.57508957e-01 6.40661895e-01 2.39861086e-01 7.21255690e-02
1.19771397e+00 7.45101392e-01 2.46809289e-01 6.86103284e-01
2.90852487e-01 -7.56156564e-01 5.86848855e-01 3.86296064e-01
8.27917218e-01 -1.39812827e+00 -5.52624092e-02 -6.47678316e-01
-8.34101915e-01 8.97691131e-01 9.51417148e-01 -1.02898277e-01
4.37378436e-01 -1.46805225e-02 -2.15120599e-01 -3.36873442e-01
-1.23631501e+00 -5.01030028e-01 4.09602255e-01 6.49920642e-01
-3.11463058e-01 7.61867911e-02 2.52717912e-01 3.36779356e-01
2.35786483e-01 -5.57632983e-01 2.71390408e-01 8.30676675e-01
-5.17988086e-01 -1.04519105e+00 7.65722916e-02 5.61964512e-01
5.32957278e-02 -3.57034765e-02 -4.47705388e-01 6.68397844e-01
3.03287923e-01 7.41009831e-01 4.92850035e-01 -3.71888548e-01
2.31107652e-01 -1.49599820e-01 5.56813002e-01 -9.52871680e-01
-5.48800789e-02 -1.33225247e-01 -9.06092953e-03 -8.96640003e-01
-3.74318570e-01 -8.31553102e-01 -1.51980162e+00 4.96152252e-01
7.62301087e-02 -2.52957463e-01 4.13071096e-01 1.07895362e+00
5.48969269e-01 5.58638513e-01 5.17742515e-01 -3.45003337e-01
-4.41962749e-01 -2.56446630e-01 -4.40517157e-01 7.78238177e-01
5.91955304e-01 -9.52537715e-01 -2.32969671e-01 2.01208759e-02] | [10.318490982055664, 1.63418447971344] |
6afe09f5-277c-4cc8-a71e-b843d43f48a1 | bet-a-backtranslation-approach-for-easy-data | 2009.12452 | null | https://arxiv.org/abs/2009.12452v1 | https://arxiv.org/pdf/2009.12452v1.pdf | BET: A Backtranslation Approach for Easy Data Augmentation in Transformer-based Paraphrase Identification Context | Newly-introduced deep learning architectures, namely BERT, XLNet, RoBERTa and ALBERT, have been proved to be robust on several NLP tasks. However, the datasets trained on these architectures are fixed in terms of size and generalizability. To relieve this issue, we apply one of the most inexpensive solutions to update these datasets. We call this approach BET by which we analyze the backtranslation data augmentation on the transformer-based architectures. Using the Google Translate API with ten intermediary languages from ten different language families, we externally evaluate the results in the context of automatic paraphrase identification in a transformer-based framework. Our findings suggest that BET improves the paraphrase identification performance on the Microsoft Research Paraphrase Corpus (MRPC) to more than 3% on both accuracy and F1 score. We also analyze the augmentation in the low-data regime with downsampled versions of MRPC, Twitter Paraphrase Corpus (TPC) and Quora Question Pairs. In many low-data cases, we observe a switch from a failing model on the test set to reasonable performances. The results demonstrate that BET is a highly promising data augmentation technique: to push the current state-of-the-art of existing datasets and to bootstrap the utilization of deep learning architectures in the low-data regime of a hundred samples. | ['Jean-Philippe Corbeil', 'Hadi Abdi Ghadivel'] | 2020-09-25 | null | null | null | null | ['paraphrase-identification'] | ['natural-language-processing'] | [ 1.84832178e-02 -1.25203326e-01 -2.26097614e-01 -3.38587791e-01
-1.04376423e+00 -8.34230959e-01 8.88445795e-01 8.83896872e-02
-7.19418585e-01 7.66740084e-01 2.92405069e-01 -7.43635237e-01
1.81415351e-03 -6.19998574e-01 -7.77553082e-01 -2.02220857e-01
4.59194690e-01 8.25254679e-01 -1.00296915e-01 -7.00953662e-01
9.74326432e-02 4.03725654e-01 -1.08747399e+00 6.09020829e-01
1.10680699e+00 7.13224411e-01 -1.76538974e-01 5.10259151e-01
-2.93341696e-01 7.32335985e-01 -6.49502814e-01 -9.27388608e-01
4.38606352e-01 -2.78548181e-01 -1.18999839e+00 -4.24453914e-01
5.75929523e-01 -2.42694110e-01 -1.91093251e-01 6.88994586e-01
4.93178815e-01 -1.81533635e-01 3.47780704e-01 -1.20409751e+00
-1.02920425e+00 9.46458220e-01 -2.46342510e-01 4.38759655e-01
5.61611712e-01 2.79227972e-01 1.14426446e+00 -1.12896883e+00
6.67967021e-01 9.96493399e-01 1.00703359e+00 2.56219864e-01
-1.21378922e+00 -6.43945992e-01 -4.00841862e-01 4.38442022e-01
-1.04590082e+00 -4.69028801e-01 6.52884662e-01 -2.61938989e-01
1.27278841e+00 2.97327191e-01 5.68149924e-01 1.79276848e+00
-5.82393222e-02 7.31989861e-01 1.29858899e+00 -6.52737916e-01
3.66067439e-02 2.70581394e-01 3.71312022e-01 4.33423758e-01
1.67334601e-01 -1.62812665e-01 -6.60153508e-01 -2.03573003e-01
5.58669642e-02 -2.90341347e-01 -2.04186559e-01 9.32842027e-03
-1.27485859e+00 1.00646424e+00 4.86529708e-01 5.57862341e-01
-2.88154095e-01 -7.31741041e-02 6.68209434e-01 9.98427629e-01
6.67538047e-01 9.91484344e-01 -6.50247514e-01 -5.61137259e-01
-9.66827273e-01 3.35432768e-01 1.07253706e+00 9.87251937e-01
5.68368316e-01 -2.63488948e-01 4.31692041e-02 1.12115300e+00
-2.08625361e-01 3.98486614e-01 1.11684823e+00 -7.92616010e-01
1.01076281e+00 8.55741799e-01 -4.23286483e-03 -7.03207910e-01
-2.85232484e-01 -6.94572926e-01 -7.42612302e-01 -4.93400842e-01
5.52654624e-01 -5.40701225e-02 -6.84270024e-01 1.62628675e+00
-9.67421830e-02 -3.64080727e-01 6.24730140e-02 5.47024727e-01
7.71880865e-01 5.17917156e-01 -3.05082291e-01 1.44011468e-01
1.25941753e+00 -1.14391208e+00 -4.16617781e-01 -3.67648363e-01
9.94598031e-01 -9.24641192e-01 1.81644714e+00 2.63532966e-01
-9.93617654e-01 -6.36946142e-01 -1.13286972e+00 -4.76055682e-01
-7.04231262e-01 1.04988679e-01 4.24787343e-01 7.02480674e-01
-1.14727402e+00 7.19627261e-01 -5.55144131e-01 -7.37808406e-01
2.04822674e-01 2.81547338e-01 -5.25138080e-01 -1.68869108e-01
-1.44383132e+00 1.26091349e+00 3.04879457e-01 -1.32469609e-01
-4.33712095e-01 -8.69481027e-01 -4.83769625e-01 9.79692563e-02
-1.10066049e-01 -7.69512236e-01 1.31226861e+00 -1.04074156e+00
-1.55967331e+00 1.20736277e+00 9.19295922e-02 -9.37917471e-01
7.08042979e-01 -4.40484136e-01 -3.21253717e-01 -1.06782645e-01
2.70851910e-01 5.17314017e-01 4.64845598e-01 -5.55007994e-01
-2.95380950e-01 -1.24921411e-01 2.51038849e-01 4.14450541e-02
-6.67858362e-01 1.18392378e-01 -1.06362768e-01 -6.60314679e-01
-2.69083709e-01 -1.04607773e+00 2.12607488e-01 -4.33226883e-01
-2.86536515e-01 -3.00113052e-01 5.96024752e-01 -7.59071946e-01
1.07568204e+00 -1.90167987e+00 1.96628004e-01 -1.32703394e-01
-8.22347775e-02 4.15867746e-01 -4.82544631e-01 7.92061746e-01
-2.78767169e-01 2.85798013e-01 -1.78698257e-01 -7.57396460e-01
-5.64042740e-02 2.53191382e-01 -4.42487866e-01 3.46444398e-01
1.79022849e-01 1.19768357e+00 -7.35886335e-01 -2.73625374e-01
1.94195900e-02 -8.06310028e-02 -5.63492596e-01 7.28907883e-02
-3.21442559e-02 1.69675440e-01 -6.24066330e-02 5.30018151e-01
5.61125159e-01 -3.46415132e-01 -4.82944362e-02 1.76524688e-02
-8.61588791e-02 9.06963408e-01 -3.76070231e-01 1.97589135e+00
-7.88830161e-01 8.71833980e-01 -2.40337729e-01 -1.02315080e+00
8.20796907e-01 1.74010068e-01 3.19616109e-01 -9.18107510e-01
2.83782333e-02 6.57658458e-01 1.68198422e-01 -3.50885481e-01
7.04843223e-01 -2.10382849e-01 -1.42892435e-01 8.17870498e-01
2.72044450e-01 -1.91525385e-01 3.36563885e-01 2.57389158e-01
1.17859638e+00 -2.78209060e-01 2.86989182e-01 -3.72632831e-01
6.03777945e-01 1.62378997e-01 -1.05232233e-02 8.87781680e-01
-8.01386237e-02 7.02333748e-01 3.78260225e-01 -4.06605303e-01
-1.23778975e+00 -8.14103007e-01 -1.05551079e-01 1.27722347e+00
-4.62938905e-01 -5.00795126e-01 -8.61655116e-01 -9.26235378e-01
3.24367322e-02 9.27037776e-01 -5.68221688e-01 -2.68171638e-01
-8.68781924e-01 -7.62191236e-01 9.41109002e-01 4.03350770e-01
7.45039284e-01 -1.06956995e+00 -1.63678795e-01 1.69156119e-01
-6.71858251e-01 -1.32608974e+00 -2.84153581e-01 1.11118190e-01
-7.35142648e-01 -1.01114166e+00 -6.39018238e-01 -7.50085354e-01
6.63545281e-02 8.00034329e-02 1.41297340e+00 -1.75084602e-02
2.63168514e-01 2.20648900e-01 -6.82123780e-01 -1.97166830e-01
-9.81208205e-01 9.48638558e-01 -1.60476007e-02 -2.86419034e-01
5.25759399e-01 -8.24420631e-01 -5.12936950e-01 1.34317011e-01
-6.93984807e-01 1.17199183e-01 6.63172424e-01 8.88624907e-01
1.96866784e-02 -6.67893350e-01 9.47887957e-01 -8.57026756e-01
1.04410398e+00 -6.57646358e-01 -2.57306904e-01 2.20235929e-01
-6.44511342e-01 7.16985092e-02 9.98148322e-01 -4.57181633e-01
-7.06014991e-01 -4.19228733e-01 -3.75193149e-01 -2.40700990e-01
1.12740107e-01 7.25938737e-01 3.37152541e-01 -3.80943045e-02
1.14745009e+00 3.51466328e-01 5.29236980e-02 -7.13120043e-01
5.58659017e-01 1.06689143e+00 5.00042498e-01 -5.73991060e-01
7.53361464e-01 1.10175654e-01 -3.81370336e-01 -5.21121621e-01
-9.57718730e-01 -4.22741920e-01 -6.39963329e-01 2.12980419e-01
4.19611096e-01 -9.12034690e-01 -3.02770138e-01 4.97309119e-01
-1.17165148e+00 -4.29823071e-01 -4.47443336e-01 1.45392776e-01
-4.39736426e-01 5.59914351e-01 -7.96933532e-01 -7.66900778e-02
-8.06627572e-01 -1.05805898e+00 9.78132606e-01 -1.88981831e-01
-4.96167988e-01 -1.00826633e+00 3.47389519e-01 7.76664734e-01
7.45878339e-01 -7.27940276e-02 1.19854462e+00 -1.31818688e+00
-2.56755948e-01 -3.16230178e-01 -1.41111329e-01 6.95447624e-01
2.36039311e-01 -2.18774557e-01 -8.57950568e-01 -5.32640576e-01
2.63499051e-01 -6.60179377e-01 5.72668016e-01 -3.27198774e-01
1.01288724e+00 -4.44947362e-01 1.31349266e-01 5.40267885e-01
1.04804051e+00 -3.78776252e-01 4.14670169e-01 8.00444305e-01
4.10568416e-01 3.75622958e-01 2.70127207e-01 3.60242352e-02
4.21398014e-01 7.45518804e-01 1.50987655e-01 9.94047597e-02
-3.13147902e-01 -3.66323411e-01 4.86905903e-01 1.26977360e+00
3.14454079e-01 -3.13006729e-01 -1.06867909e+00 6.55106187e-01
-1.59741902e+00 -7.31461227e-01 -9.69742462e-02 2.03109455e+00
1.24246478e+00 1.84410557e-01 2.49277815e-01 1.23465925e-01
4.12335783e-01 -1.06159979e-02 -3.38953346e-01 -6.59308374e-01
-2.46582374e-01 5.08641660e-01 3.03457588e-01 3.62245530e-01
-7.97448933e-01 9.64548349e-01 6.55954218e+00 7.84446895e-01
-1.24781811e+00 4.26863551e-01 4.47074085e-01 -3.55400629e-02
-2.40503535e-01 -6.60500824e-02 -6.42358065e-01 8.68179500e-01
1.45033133e+00 -2.92940497e-01 6.67831540e-01 8.52699816e-01
-5.93288662e-03 2.06348971e-01 -1.50445068e+00 9.27523255e-01
2.06066087e-01 -1.43844414e+00 1.95160314e-01 -3.64612862e-02
6.58572197e-01 7.52783537e-01 6.05868511e-02 8.94493222e-01
2.94854879e-01 -9.72870648e-01 5.59004545e-01 5.70626520e-02
8.29738975e-01 -3.92633647e-01 9.01137054e-01 5.89188039e-01
-4.63538140e-01 -2.19478458e-01 -3.53661358e-01 -1.78690687e-01
-3.38232033e-02 4.18152630e-01 -1.30558872e+00 5.94129562e-01
5.77133119e-01 7.59478271e-01 -1.14057565e+00 6.79639041e-01
-1.26777992e-01 8.16749871e-01 -5.16700506e-01 -1.27550632e-01
4.30580556e-01 -2.02475712e-01 4.43343997e-01 1.35369802e+00
2.56574154e-01 -5.29283464e-01 -3.66152108e-01 9.48294580e-01
-5.63011527e-01 1.79223225e-01 -6.25191450e-01 -1.83067918e-02
6.25355303e-01 1.20464563e+00 -8.57871920e-02 -4.80003953e-01
-4.64262724e-01 1.01134324e+00 6.61982834e-01 2.33287454e-01
-8.69773626e-01 -3.05004686e-01 2.86780030e-01 1.22965649e-01
-1.14993662e-01 -8.30560923e-02 -4.03171808e-01 -1.44325280e+00
3.75373930e-01 -1.45561016e+00 3.42013985e-01 -7.58241534e-01
-1.62520146e+00 6.14095807e-01 3.79986241e-02 -1.05244434e+00
-5.55476487e-01 -6.63382053e-01 -5.64064085e-01 9.68359649e-01
-1.51484215e+00 -1.22619557e+00 -2.67003715e-01 5.99385858e-01
8.03272307e-01 -5.01482666e-01 8.05635989e-01 4.81250972e-01
-3.37109536e-01 9.79001760e-01 4.77239221e-01 2.30339259e-01
8.32202911e-01 -1.29065323e+00 9.69253480e-01 7.26435542e-01
3.86024833e-01 8.65477145e-01 5.56718469e-01 -1.45037308e-01
-1.23831570e+00 -9.77334261e-01 1.30298126e+00 -7.83749640e-01
1.14383352e+00 -6.46154821e-01 -9.54047918e-01 8.54916573e-01
4.43208277e-01 -3.20117325e-01 5.73258162e-01 3.48877609e-01
-5.12635052e-01 -1.76665485e-01 -1.16400397e+00 7.06738591e-01
9.63454664e-01 -9.09890234e-01 -1.11313057e+00 7.25964189e-01
8.60270679e-01 -3.16599458e-01 -8.71057928e-01 3.35815102e-01
3.64577621e-01 -1.06799090e+00 8.85919273e-01 -8.72307539e-01
7.22708762e-01 3.87669802e-01 -1.06570929e-01 -1.49237573e+00
-5.19810989e-02 -6.12960875e-01 1.28265381e-01 1.41845012e+00
5.21728754e-01 -9.49880064e-01 8.34722340e-01 2.61322796e-01
-2.17442095e-01 -7.91599751e-01 -1.36671638e+00 -8.74526560e-01
7.48161972e-01 -3.27117175e-01 6.22425199e-01 1.18346894e+00
1.43779457e-01 8.52979124e-01 -7.27683157e-02 -5.46721637e-01
5.21145426e-02 1.81479573e-01 1.01191306e+00 -8.87213588e-01
-3.50979626e-01 -4.83670831e-01 4.26036082e-02 -9.75364506e-01
4.59041893e-01 -1.34534121e+00 -4.35789704e-01 -1.28754461e+00
2.43947312e-01 -4.08374369e-01 -3.19786556e-02 4.38496709e-01
-3.54563631e-02 2.20647097e-01 2.05641374e-01 4.05297220e-01
-2.57255822e-01 6.02364779e-01 7.59856284e-01 -1.60575390e-01
-3.71947810e-02 8.74723420e-02 -6.78972900e-01 5.05504131e-01
9.76044357e-01 -3.74471873e-01 -2.62044728e-01 -8.44770014e-01
5.58433652e-01 -2.76885182e-01 3.55491400e-01 -7.81219959e-01
4.28621508e-02 3.43080223e-01 -6.71005473e-02 -4.38427389e-01
2.54153371e-01 -7.14474916e-01 -1.43324390e-01 4.76026058e-01
-6.58785760e-01 6.56716168e-01 2.05843270e-01 1.06780902e-01
-1.58825785e-01 -4.23309028e-01 7.34971404e-01 -1.19446464e-01
-1.43806264e-01 -1.48913607e-01 -3.13829750e-01 4.44937497e-01
3.78875613e-01 -1.00191519e-01 -6.88507318e-01 -4.81638789e-01
-2.28698373e-01 1.39496103e-01 4.30891484e-01 7.04313397e-01
1.29869983e-01 -1.25593996e+00 -9.27371562e-01 2.46735811e-01
2.36560583e-01 -3.80719662e-01 -2.14397669e-01 9.52800870e-01
-7.03873277e-01 6.60871089e-01 -1.95620969e-01 -5.79502821e-01
-1.05819273e+00 4.46394026e-01 4.93235320e-01 -6.86887026e-01
-4.29856300e-01 7.09238946e-01 -2.55933553e-01 -9.08386648e-01
-8.54001120e-02 -6.23710275e-01 7.49445781e-02 6.75761104e-02
-7.42760208e-03 3.99016798e-01 5.85367501e-01 -4.38097179e-01
-2.43777812e-01 1.70921177e-01 -4.16614890e-01 -1.16512284e-01
1.31378996e+00 -3.22457664e-02 -2.89040446e-01 3.00818592e-01
1.47221780e+00 1.33057937e-01 -3.56453210e-01 -3.34872037e-01
1.63188607e-01 -1.87207446e-01 -4.09175277e-01 -1.14253235e+00
-6.23459339e-01 8.24631691e-01 3.40408564e-01 3.41991276e-01
8.43849182e-01 -1.08401909e-01 9.35324609e-01 7.84829378e-01
2.20554963e-01 -8.72819841e-01 2.49576807e-01 9.09825385e-01
1.24982035e+00 -1.22700787e+00 -1.51002392e-01 9.66685116e-02
-4.02163148e-01 9.71178114e-01 4.74869788e-01 -1.12442493e-01
2.58951455e-01 -5.21380492e-02 7.32067451e-02 -6.22125864e-02
-7.76708186e-01 3.07263166e-01 1.18022822e-01 2.34710410e-01
4.71174538e-01 -2.59188056e-01 -4.23806190e-01 4.81809258e-01
-9.76695359e-01 7.89986085e-03 5.37972987e-01 4.79727477e-01
2.91216485e-02 -1.22568333e+00 -2.59814002e-02 3.40567142e-01
-6.47345066e-01 -4.68982339e-01 -8.39906573e-01 1.22616613e+00
-2.45266140e-01 9.42983568e-01 -4.21024859e-02 -3.37933183e-01
4.60484385e-01 3.61979365e-01 3.08509558e-01 -5.53133607e-01
-1.18098736e+00 -7.85846889e-01 3.65318537e-01 -3.79019529e-01
-2.65756190e-01 -5.11775792e-01 -5.47493875e-01 -6.29126549e-01
-2.70927727e-01 2.49112889e-01 5.13645113e-01 1.06788790e+00
5.84642828e-01 9.80193689e-02 5.69728255e-01 -5.22851884e-01
-1.13010871e+00 -1.46334577e+00 8.66369233e-02 6.86466277e-01
2.67611474e-01 -1.66627944e-01 -5.32912910e-01 -3.11913699e-01] | [11.228553771972656, 9.117602348327637] |
d3dabe53-6dad-40fb-8c36-5df9a992148b | pointconvformer-revenge-of-the-point-based | 2208.02879 | null | https://arxiv.org/abs/2208.02879v3 | https://arxiv.org/pdf/2208.02879v3.pdf | PointConvFormer: Revenge of the Point-based Convolution | We introduce PointConvFormer, a novel building block for point cloud based deep network architectures. Inspired by generalization theory, PointConvFormer combines ideas from point convolution, where filter weights are only based on relative position, and Transformers which utilize feature-based attention. In PointConvFormer, attention computed from feature difference between points in the neighborhood is used to modify the convolutional weights at each point. Hence, we preserved the invariances from point convolution, whereas attention helps to select relevant points in the neighborhood for convolution. PointConvFormer is suitable for multiple tasks that require details at the point level, such as segmentation and scene flow estimation tasks. We experiment on both tasks with multiple datasets including ScanNet, SemanticKitti, FlyingThings3D and KITTI. Our results show that PointConvFormer offers a better accuracy-speed tradeoff than classic convolutions, regular transformers, and voxelized sparse convolution approaches. Visualizations show that PointConvFormer performs similarly to convolution on flat areas, whereas the neighborhood selection effect is stronger on object boundaries, showing that it has got the best of both worlds. | ['Li Fuxin', 'Qi Shan', 'Wenxuan Wu'] | 2022-08-04 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Wu_PointConvFormer_Revenge_of_the_Point-Based_Convolution_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Wu_PointConvFormer_Revenge_of_the_Point-Based_Convolution_CVPR_2023_paper.pdf | cvpr-2023-1 | ['scene-flow-estimation'] | ['computer-vision'] | [-3.36743891e-01 -2.62907982e-01 1.53154120e-01 -4.52174902e-01
1.38037756e-01 -4.96658385e-01 6.72199368e-01 1.59277827e-01
-5.23425221e-01 2.60063767e-01 2.03887150e-01 -2.07832426e-01
-1.63766697e-01 -1.20561755e+00 -8.70383799e-01 -3.56971323e-01
-2.93761790e-01 2.77161121e-01 5.68553805e-01 -5.18371046e-01
3.90391469e-01 1.09178066e+00 -1.55742621e+00 4.00860637e-01
7.58760810e-01 1.12303638e+00 8.35842416e-02 5.05793214e-01
-5.56216002e-01 2.21084923e-01 -4.60241586e-01 -2.23071769e-01
5.51913083e-01 2.11325571e-01 -8.55730176e-01 -4.20463145e-01
9.65528190e-01 -2.08916828e-01 -5.05773187e-01 8.28020811e-01
2.71443605e-01 5.33943117e-01 5.12115419e-01 -1.14375234e+00
-7.59984672e-01 4.73420739e-01 -6.78180039e-01 5.11957169e-01
-7.51940981e-02 7.08454669e-01 9.56517875e-01 -1.12391925e+00
4.36209679e-01 1.59546185e+00 1.03475440e+00 1.65749878e-01
-1.20122933e+00 -6.36786520e-01 4.24622715e-01 7.94145390e-02
-1.23454440e+00 -6.36182874e-02 4.70714241e-01 -4.15899485e-01
1.35235500e+00 3.54539156e-01 8.59632373e-01 6.61940873e-01
2.23885402e-01 5.51715612e-01 5.47198653e-01 1.55821461e-02
1.98056757e-01 -3.99947584e-01 1.41638458e-01 7.23162353e-01
3.73339690e-02 3.05940449e-01 -5.64880610e-01 -7.40416273e-02
1.55466878e+00 4.51423466e-01 -3.84610593e-01 -3.88973206e-01
-1.65319264e+00 9.02007103e-01 1.31649292e+00 2.29340419e-01
-4.74779725e-01 7.83693314e-01 1.42874345e-01 2.89138228e-01
4.58545178e-01 7.22228944e-01 -6.43461823e-01 1.57895442e-02
-9.48144674e-01 4.72186923e-01 3.92713934e-01 1.05256951e+00
9.83519554e-01 1.09147884e-01 -5.72473586e-01 4.36347485e-01
1.40134528e-01 5.71964681e-01 2.48355120e-01 -1.15777862e+00
4.05389160e-01 7.20793068e-01 -9.01217759e-02 -1.06138396e+00
-5.34661651e-01 -5.22326767e-01 -6.30144238e-01 7.95658231e-01
4.11765099e-01 -2.83447150e-02 -1.42588603e+00 1.41133201e+00
1.52064979e-01 4.35646087e-01 -4.19988602e-01 1.23373795e+00
1.15926480e+00 6.79247797e-01 1.06421024e-01 7.74862885e-01
1.22602177e+00 -8.46072853e-01 -9.96537954e-02 -6.23119324e-02
3.95859450e-01 -4.19468313e-01 1.33995938e+00 6.70542121e-02
-1.13277209e+00 -6.47386730e-01 -9.61275578e-01 -3.97792876e-01
-6.86339915e-01 -2.33024538e-01 1.01792216e+00 9.03726071e-02
-1.30340838e+00 1.08287549e+00 -9.94909227e-01 -3.87046844e-01
7.94118643e-01 4.10638601e-01 -3.23947668e-01 7.83250332e-02
-7.31398940e-01 7.95991242e-01 3.95741425e-02 -3.93881537e-02
-6.99842513e-01 -1.45532644e+00 -7.92133808e-01 5.75254261e-01
-1.42115541e-03 -1.04447258e+00 1.19328880e+00 -5.84890127e-01
-1.20236957e+00 5.76682985e-01 -2.38453373e-02 -6.43872440e-01
4.36274409e-01 -3.39621603e-01 -2.99857510e-03 2.52717763e-01
1.33352503e-01 1.15341532e+00 7.26922154e-01 -8.46604884e-01
-6.33713186e-01 -3.70775968e-01 3.10984582e-01 1.55231804e-01
5.85797876e-02 -3.87920201e-01 -3.67370486e-01 -4.55540061e-01
3.52168083e-01 -4.04014945e-01 -4.24018770e-01 7.31946170e-01
-3.30693096e-01 -4.65189129e-01 1.27918327e+00 -8.99700150e-02
5.16824186e-01 -2.35388327e+00 -4.24185917e-02 3.29218030e-01
6.52312458e-01 2.02309445e-01 -2.19826192e-01 2.80914545e-01
-2.23317787e-01 2.94336945e-01 -2.55575389e-01 -4.97527607e-02
-7.23485798e-02 2.29833677e-01 -2.68398196e-01 4.82032537e-01
5.40249646e-01 1.15551138e+00 -9.39688683e-01 -2.75666505e-01
7.41566420e-01 7.45809436e-01 -9.04667437e-01 -2.40926102e-01
-3.15396637e-01 2.02012986e-01 -5.01838386e-01 5.70541441e-01
8.01227331e-01 -2.86628544e-01 -8.42376292e-01 -3.66597831e-01
-5.32717466e-01 2.17408881e-01 -8.31843257e-01 2.04111028e+00
-4.14821208e-01 8.16696405e-01 1.59380987e-01 -6.30766809e-01
8.85107934e-01 -3.63360308e-02 5.30845463e-01 -6.10461593e-01
9.60530341e-02 -9.17859152e-02 1.49644064e-02 -2.68093139e-01
6.25390589e-01 1.58846080e-01 2.80811846e-01 5.88338897e-02
2.28726104e-01 -5.79451025e-01 -1.08911462e-01 3.56551260e-01
1.35307252e+00 2.88046658e-01 -1.95847154e-01 -5.37061214e-01
7.94459358e-02 3.51239294e-01 1.23337761e-01 8.91844630e-01
-6.44671023e-02 9.13592339e-01 2.02905566e-01 -7.89900482e-01
-8.55561912e-01 -1.26807809e+00 -1.49124205e-01 9.29006577e-01
4.03819859e-01 -4.37083960e-01 -4.40975636e-01 -6.15474641e-01
4.02823269e-01 6.10745132e-01 -8.58176053e-01 -1.08489215e-01
-6.72994256e-01 -1.05423540e-01 3.33727032e-01 8.69127691e-01
8.54662299e-01 -1.10040140e+00 -1.04251122e+00 4.25992347e-02
2.99210459e-01 -6.56927347e-01 -5.10952413e-01 3.78539294e-01
-1.03224146e+00 -9.77818549e-01 -6.46599054e-01 -4.81206626e-01
4.74917889e-01 5.19677103e-01 1.43806148e+00 2.28485599e-01
-3.07219654e-01 4.53459352e-01 -3.41309547e-01 -5.65238893e-01
4.20524955e-01 1.67624369e-01 -3.46358985e-01 -4.34966236e-01
4.27411437e-01 -8.10586512e-01 -9.37301576e-01 2.01655418e-01
-6.73635781e-01 -1.61665797e-01 5.84668517e-01 5.65642715e-01
6.62673116e-01 -2.66929328e-01 -8.78721625e-02 -4.83046442e-01
3.73176575e-01 -4.84140277e-01 -4.72883344e-01 -2.24974006e-01
-6.26957882e-03 1.30400434e-01 5.32739043e-01 -3.91505927e-01
-7.07082391e-01 -2.03804467e-02 -3.26645881e-01 -9.00206983e-01
-2.62897760e-01 2.02249572e-01 2.42755249e-01 -3.45736206e-01
8.13367724e-01 -1.17767267e-01 5.33135375e-03 -5.85256100e-01
5.91405034e-01 -1.00965925e-01 6.31463647e-01 -4.64508772e-01
8.50278139e-01 9.79320765e-01 -8.73228759e-02 -9.15941119e-01
-3.92889947e-01 -2.58128434e-01 -7.89216399e-01 -1.19876549e-01
1.08442211e+00 -6.45544827e-01 -8.34489644e-01 2.73890868e-02
-1.27248955e+00 -4.58934963e-01 -1.02313864e+00 5.13609707e-01
-3.73277336e-01 -2.34711215e-01 -5.10123789e-01 -1.94805518e-01
-3.12815040e-01 -1.05543303e+00 1.21427131e+00 4.10367638e-01
-1.45115837e-01 -9.12504792e-01 -1.91949636e-01 -5.62310696e-01
7.82469273e-01 2.36454755e-01 8.60420644e-01 -2.78345883e-01
-8.80805194e-01 2.40114070e-02 -5.42274535e-01 -1.01115324e-01
-9.78948250e-02 1.29352465e-01 -9.06013668e-01 -3.06137875e-02
-2.51082510e-01 1.07801646e-01 1.25049233e+00 6.44050658e-01
1.64163888e+00 8.93178806e-02 -4.54566896e-01 1.22119725e+00
1.34090328e+00 -8.89272019e-02 8.12540472e-01 3.26274842e-01
9.54681933e-01 1.22236967e-01 1.73916325e-01 2.06057921e-01
1.80337206e-01 5.20857573e-01 1.09175873e+00 -5.91919184e-01
-4.53611493e-01 -2.03670934e-01 -1.43600136e-01 1.63689241e-01
-2.89629936e-01 1.31460829e-02 -7.16837049e-01 4.81851578e-01
-1.69854009e+00 -9.20509756e-01 -4.28412288e-01 1.96241057e+00
1.92420527e-01 -6.18199760e-04 -5.05567677e-02 -1.07284844e-01
4.28917348e-01 2.24069640e-01 -6.08086646e-01 -4.69827890e-01
-5.61902933e-02 7.88541317e-01 7.85960615e-01 2.05938667e-01
-1.02815819e+00 1.09546256e+00 6.53160954e+00 5.63978493e-01
-1.43207157e+00 1.14508435e-01 3.40068072e-01 -4.19194877e-01
-4.90993589e-01 -2.35766824e-02 -3.77371132e-01 2.17028663e-01
3.22907329e-01 1.48941338e-01 2.49580860e-01 8.13470066e-01
2.21368834e-01 1.44461095e-01 -1.10259569e+00 1.02117407e+00
-4.48926717e-01 -1.70868635e+00 7.85589665e-02 -1.88016184e-02
4.22788560e-01 7.16968894e-01 2.16699034e-01 3.02257299e-01
5.28243899e-01 -1.24093473e+00 8.52347255e-01 4.75023538e-01
6.66477203e-01 -7.44859457e-01 4.30146068e-01 -7.35176876e-02
-1.24173784e+00 1.43104717e-01 -6.63336277e-01 -1.97588474e-01
9.26131234e-02 5.96281350e-01 -6.40652478e-01 2.86313057e-01
1.26394665e+00 8.31403852e-01 -5.05488694e-01 1.44407463e+00
-1.02279365e-01 4.02539670e-01 -5.89299679e-01 5.01442552e-02
6.80947423e-01 -7.46011063e-02 6.59780145e-01 1.23908079e+00
4.15583372e-01 9.59879905e-02 8.99364427e-03 1.48938525e+00
2.71165688e-02 -1.39125168e-01 -7.24268436e-01 3.67544562e-01
4.51191843e-01 1.29852712e+00 -9.93633687e-01 -3.97513181e-01
-4.23338115e-01 8.72447312e-01 3.30640882e-01 5.13128936e-01
-8.60444188e-01 -6.94791079e-01 1.26253712e+00 5.80442846e-01
7.75415897e-01 -5.20117760e-01 -5.39856136e-01 -9.08046186e-01
-2.78427929e-01 -1.05473571e-01 9.20337364e-02 -9.66352761e-01
-1.08785558e+00 6.79334641e-01 -9.40947700e-03 -1.14771032e+00
4.16912138e-01 -7.64101386e-01 -1.16110671e+00 1.03512335e+00
-1.48559773e+00 -8.77854109e-01 -7.01712310e-01 8.95507693e-01
5.07041812e-01 2.13405952e-01 4.63847071e-01 2.02797368e-01
-1.09481625e-01 1.92837954e-01 -3.72449458e-01 1.37579784e-01
2.11601108e-01 -1.27821100e+00 9.83138621e-01 6.70250654e-01
1.63452208e-01 7.53492057e-01 3.38017821e-01 -5.36303341e-01
-1.32394528e+00 -1.31283104e+00 2.89987803e-01 -3.93062681e-01
3.38476658e-01 -2.23842204e-01 -9.69396532e-01 6.21043563e-01
2.27127180e-01 4.14667636e-01 3.39650214e-02 3.65411602e-02
-4.45945978e-01 1.46285696e-02 -1.33968830e+00 5.54224730e-01
1.46019292e+00 -2.27722034e-01 -3.98839176e-01 2.61524588e-01
1.14917159e+00 -6.09913886e-01 -7.66629696e-01 3.12867910e-01
2.80461043e-01 -1.15594435e+00 1.30752945e+00 -6.01371050e-01
4.57815319e-01 -3.47331792e-01 -3.75594273e-02 -1.71139538e+00
-8.09743762e-01 -2.78204054e-01 2.10495014e-02 7.34976470e-01
3.58847022e-01 -8.58303428e-01 9.07431722e-01 3.50043625e-01
-7.80475259e-01 -6.44597113e-01 -8.71886432e-01 -6.95191622e-01
2.55139053e-01 -4.69367057e-01 1.21745467e+00 8.78980100e-01
-4.81913954e-01 1.09085580e-03 3.63845587e-01 2.71150827e-01
4.44698006e-01 1.10243201e-01 6.59584761e-01 -1.39875197e+00
-1.70914847e-02 -8.04296613e-01 -6.48117125e-01 -1.33779538e+00
-2.64889777e-01 -1.00613081e+00 -9.78684872e-02 -1.82441747e+00
-4.50398237e-01 -7.47607529e-01 -1.48796946e-01 6.55214489e-01
-9.68582463e-03 4.09334600e-01 3.90000582e-01 9.78138894e-02
-1.87794939e-01 5.55580854e-01 1.64192176e+00 -2.74674624e-01
-4.33040202e-01 -9.74049270e-02 -6.88136518e-01 7.69000292e-01
8.01563978e-01 -7.25681782e-02 -3.50609571e-01 -1.09176242e+00
-1.45252705e-01 -3.69238853e-01 8.54469597e-01 -1.38707042e+00
1.45858139e-01 -1.84410200e-01 8.11731160e-01 -9.19306278e-01
5.96595526e-01 -7.90989757e-01 -6.20686226e-02 3.62497747e-01
-1.51180923e-01 4.51828569e-01 3.95656854e-01 3.03462446e-01
3.83048877e-02 1.59162894e-01 6.60301149e-01 -5.21729767e-01
-9.16707993e-01 8.40619624e-01 -1.72419116e-01 -7.52058849e-02
7.42217600e-01 -4.78409410e-01 -4.34441030e-01 -1.17275104e-01
-6.77727222e-01 2.34840363e-01 5.46983600e-01 4.43166018e-01
1.03256297e+00 -1.35633326e+00 -6.40056252e-01 5.17778039e-01
-9.44927260e-02 6.64914608e-01 1.00092664e-01 8.12962055e-01
-8.58621895e-01 2.08576709e-01 -2.80247599e-01 -1.10956204e+00
-7.76967049e-01 5.08753598e-01 4.31795537e-01 5.04750729e-01
-1.28674901e+00 1.36840916e+00 5.48050523e-01 -4.89728063e-01
1.09369449e-01 -1.15434599e+00 2.00176109e-02 -4.29707855e-01
4.79537398e-01 2.63548285e-01 2.58652627e-01 -3.61813635e-01
-3.56243461e-01 6.07736647e-01 1.04227863e-01 4.78036478e-02
1.39520848e+00 2.85011858e-01 2.29222607e-02 1.49179965e-01
9.85136867e-01 -1.76147193e-01 -1.49319255e+00 -2.97153313e-02
-4.90865469e-01 -8.00782263e-01 4.52683091e-01 -6.42751992e-01
-1.54515612e+00 1.15005457e+00 5.56634724e-01 2.58409351e-01
6.78769350e-01 1.19557090e-01 7.41249144e-01 1.64162338e-01
5.38807750e-01 -4.88249421e-01 -6.56997487e-02 7.31250525e-01
1.07338989e+00 -8.74210894e-01 -1.54119000e-01 -5.02592802e-01
-2.26607531e-01 1.04861677e+00 9.99714255e-01 -6.16386354e-01
9.13421035e-01 3.22173685e-01 -1.26054317e-01 -7.36051679e-01
-5.09070992e-01 -4.03294414e-01 1.79203749e-01 8.24550152e-01
2.41177112e-01 3.75495180e-02 2.41249070e-01 9.70011577e-02
-5.09634376e-01 -2.80254573e-01 1.18930236e-01 9.50042069e-01
-5.57453752e-01 -5.87602437e-01 -4.86362666e-01 7.94176996e-01
2.81586479e-02 -2.82776386e-01 -2.77190626e-01 8.82542133e-01
3.78928035e-01 4.56625462e-01 7.41766334e-01 -4.31598455e-01
6.98437035e-01 -3.85967314e-01 5.15399039e-01 -6.49030983e-01
-1.01923335e+00 -2.21309811e-01 -4.16540623e-01 -1.06842494e+00
-1.03593029e-01 -3.85852605e-01 -1.58252192e+00 -6.81645095e-01
-2.08373979e-01 1.28083095e-01 7.64740586e-01 5.87662876e-01
7.34725833e-01 9.42228496e-01 1.66663930e-01 -1.48428619e+00
-1.67018682e-01 -8.58432114e-01 -4.41157222e-01 3.06919634e-01
3.96381736e-01 -8.45901847e-01 -2.28007406e-01 -4.38182741e-01] | [7.924136161804199, -3.6539289951324463] |
8bd2d870-8dcb-4f76-a6b0-48b1ca977eaf | enhancing-mixup-based-graph-learning-for | 2210.03123 | null | https://arxiv.org/abs/2210.03123v2 | https://arxiv.org/pdf/2210.03123v2.pdf | On the Effectiveness of Hybrid Pooling in Mixup-Based Graph Learning for Language Processing | Graph neural network (GNN)-based graph learning has been popular in natural language and programming language processing, particularly in text and source code classification. Typically, GNNs are constructed by incorporating alternating layers which learn transformations of graph node features, along with graph pooling layers that use graph pooling operators (e.g., Max-pooling) to effectively reduce the number of nodes while preserving the semantic information of the graph. Recently, to enhance GNNs in graph learning tasks, Manifold-Mixup, a data augmentation technique that produces synthetic graph data by linearly mixing a pair of graph data and their labels, has been widely adopted. However, the performance of Manifold-Mixup can be highly affected by graph pooling operators, and there have not been many studies that are dedicated to uncovering such affection. To bridge this gap, we take an early step to explore how graph pooling operators affect the performance of Mixup-based graph learning. To that end, we conduct a comprehensive empirical study by applying Manifold-Mixup to a formal characterization of graph pooling based on 11 graph pooling operations (9 hybrid pooling operators, 2 non-hybrid pooling operators). The experimental results on both natural language datasets (Gossipcop, Politifact) and programming language datasets (JAVA250, Python800) demonstrate that hybrid pooling operators are more effective for Manifold-Mixup than the standard Max-pooling and the state-of-the-art graph multiset transformer (GMT) pooling, in terms of producing more accurate and robust models. | ['Jianjun Zhao', 'Mike Papadakis', 'Yuejun Guo', 'Zhenya Zhang', 'Yves Le Traon', 'Maxime Cordy', 'Qiang Hu', 'Zeming Dong'] | 2022-10-06 | null | null | null | null | ['code-classification'] | ['computer-code'] | [ 1.05807357e-01 2.39897028e-01 -2.98775494e-01 -1.45894051e-01
-3.54579184e-04 -4.52976555e-01 6.16027713e-01 4.88028437e-01
-1.44443691e-01 2.86284477e-01 7.57058803e-03 -3.68573636e-01
1.62662894e-01 -1.39456964e+00 -8.08466911e-01 -5.18878639e-01
-3.43895227e-01 -7.26742893e-02 1.65695116e-01 -3.40726256e-01
1.68197706e-01 3.08707058e-01 -1.18382800e+00 2.90517509e-01
8.75058472e-01 5.20978689e-01 -1.92880630e-04 4.87728387e-01
-5.04668355e-01 1.06626296e+00 -4.20460761e-01 -6.70528829e-01
3.17815214e-01 -3.37342739e-01 -6.96940064e-01 -1.57303527e-01
2.93789625e-01 1.57335028e-01 -6.72436655e-01 1.41499674e+00
3.57262105e-01 1.44824415e-01 2.95672506e-01 -1.77776277e+00
-1.20531917e+00 1.13602316e+00 -6.24220550e-01 8.14191252e-02
1.82595834e-01 1.62821636e-01 1.25936759e+00 -7.41210997e-01
5.88367760e-01 1.51394868e+00 7.79444456e-01 2.59409606e-01
-1.27949667e+00 -7.31098413e-01 8.26208144e-02 -8.64899997e-03
-1.48300540e+00 2.15522274e-01 1.11489165e+00 -4.67271477e-01
1.07399607e+00 1.04935244e-01 6.13092840e-01 8.31969917e-01
4.98901606e-01 5.98140001e-01 8.94228041e-01 -3.81557643e-01
-1.10424414e-01 -1.20844468e-01 4.93389100e-01 1.28813481e+00
3.43696296e-01 -2.63001829e-01 -4.62024510e-01 -1.93394229e-01
8.46937597e-01 2.58743674e-01 -1.66638702e-01 -3.97936136e-01
-1.25454354e+00 8.43407989e-01 1.11168730e+00 4.28749472e-01
-2.24085853e-01 3.93644780e-01 6.08244896e-01 6.02715492e-01
5.65136254e-01 7.08998561e-01 -1.53033897e-01 4.07036901e-01
-4.93256897e-01 2.19159454e-01 7.53585279e-01 1.14066219e+00
1.22241914e+00 1.53439239e-01 -3.27654302e-01 6.73249662e-01
2.40480438e-01 2.73557790e-02 5.31423450e-01 -2.42471218e-01
8.52368534e-01 1.54229593e+00 -7.83509851e-01 -1.43379319e+00
-5.19809544e-01 -6.18221276e-02 -1.19821227e+00 -8.75838697e-02
1.98863879e-01 -7.81058595e-02 -9.79818106e-01 1.68408811e+00
-1.03299491e-01 3.03109944e-01 9.27859917e-02 5.11715829e-01
1.35973585e+00 6.41061187e-01 8.55397284e-02 1.99221417e-01
1.21319151e+00 -1.17611003e+00 -5.09632945e-01 -2.75151938e-01
9.70117390e-01 -4.66269314e-01 1.42553377e+00 -1.96631067e-02
-5.95352411e-01 -5.24432659e-01 -1.16493499e+00 -1.35370299e-01
-1.02926874e+00 -2.15636328e-01 9.22438860e-01 6.20772123e-01
-1.15300357e+00 8.60836506e-01 -8.11608911e-01 -4.07036364e-01
4.39476758e-01 3.34649712e-01 -7.48348534e-01 -3.91572684e-01
-1.24432933e+00 4.91256416e-01 7.76373267e-01 -2.85601392e-02
-5.70786119e-01 -5.33006310e-01 -1.24938202e+00 3.35804939e-01
5.88340163e-01 -5.75342000e-01 6.05323315e-01 -7.90006995e-01
-1.23693180e+00 8.43374312e-01 4.04760450e-01 -4.67417598e-01
2.04758961e-02 2.18681335e-01 -2.71197140e-01 -8.06979463e-02
-1.38293698e-01 6.09502435e-01 8.12896371e-01 -8.96067858e-01
7.75996149e-02 -3.54780942e-01 3.61010700e-01 8.70261937e-02
-6.78398728e-01 7.72882178e-02 -4.32855606e-01 -9.18827772e-01
1.00368842e-01 -9.16544855e-01 -3.72769594e-01 -3.63579929e-01
-8.95698428e-01 -5.26741862e-01 8.50595593e-01 -5.03190279e-01
1.31413770e+00 -2.09103537e+00 2.39815205e-01 1.46088630e-01
7.50581920e-01 4.20362025e-01 -4.73206401e-01 7.70391822e-01
-1.99206918e-01 7.28753567e-01 -4.68167156e-01 -3.22837770e-01
7.50503018e-02 2.44076028e-01 -5.87581806e-02 3.15602958e-01
5.52347779e-01 1.34532011e+00 -9.99203026e-01 -4.94730711e-01
1.74884126e-01 3.59832823e-01 -7.63202667e-01 2.43579835e-01
-3.40841711e-01 1.10256515e-01 -4.72027302e-01 6.83047235e-01
6.65940285e-01 -4.04771090e-01 1.72224745e-01 5.85136469e-03
1.81944072e-01 6.85900375e-02 -1.02642810e+00 1.68097126e+00
-2.70822257e-01 4.93429869e-01 -2.17032179e-01 -1.11785650e+00
1.07235003e+00 6.99903741e-02 3.04958612e-01 -4.73603666e-01
2.01099828e-01 -4.99537662e-02 4.64543849e-02 -5.11830032e-01
6.35988295e-01 1.11295849e-01 -3.48939747e-01 4.26138639e-01
4.12121147e-01 -2.25570902e-01 4.35491055e-01 4.14964288e-01
1.33902955e+00 -1.85256287e-01 2.38049015e-01 -4.54193115e-01
5.07323742e-01 -2.33631447e-01 4.84644353e-01 7.15161026e-01
6.45128712e-02 6.69262052e-01 1.12138414e+00 -3.66359681e-01
-7.66388655e-01 -8.08432043e-01 4.56499159e-01 1.08560050e+00
6.45265281e-02 -9.17042971e-01 -7.78557241e-01 -9.97252643e-01
1.87033005e-02 3.28663379e-01 -6.20215356e-01 -5.25074780e-01
-6.40033603e-01 -1.01429212e+00 7.85396874e-01 4.42963839e-01
8.40024769e-01 -1.35677135e+00 1.18543923e-01 7.56023377e-02
1.60379231e-01 -1.19504142e+00 -6.72053158e-01 1.23180030e-02
-7.83539712e-01 -1.25931871e+00 -3.99996221e-01 -1.04312229e+00
9.64384198e-01 3.20671320e-01 1.16619062e+00 6.44242108e-01
-5.75853884e-02 1.43583998e-01 -4.98461813e-01 -2.00544357e-01
-4.95890230e-01 5.16247869e-01 -2.92275995e-01 1.07327849e-01
2.76090205e-01 -7.03282118e-01 -7.16988817e-02 5.41656129e-02
-1.24963653e+00 2.74120301e-01 5.90400755e-01 7.75674820e-01
2.29012564e-01 1.64900750e-01 4.57172722e-01 -1.23548651e+00
1.06814492e+00 -6.28654063e-01 -6.24348223e-01 2.94931710e-01
-6.03728533e-01 2.81832337e-01 9.94524956e-01 -3.65122139e-01
-5.28717816e-01 -2.96086848e-01 9.69094783e-02 -5.47242224e-01
4.08774227e-01 1.16413951e+00 -1.58402726e-01 -3.94014537e-01
7.51272798e-01 7.48484582e-02 9.18875933e-02 -2.80408561e-01
5.74246526e-01 3.53925407e-01 3.58856887e-01 -4.68584418e-01
9.56635237e-01 1.13510080e-01 1.83924422e-01 -7.59220660e-01
-4.25111979e-01 -2.83856601e-01 -5.02804279e-01 3.11841108e-02
9.12621081e-01 -6.77073002e-01 -2.93040276e-01 7.93270469e-01
-9.63154674e-01 -4.79342461e-01 3.91675495e-02 2.27658764e-01
-2.05142096e-01 4.62303102e-01 -7.71116257e-01 -2.58646578e-01
-3.61198872e-01 -1.33056843e+00 7.97711849e-01 3.38952184e-01
2.56918341e-01 -1.17870939e+00 -8.63751024e-02 4.69420291e-02
3.99199545e-01 6.16674423e-01 1.27303958e+00 -7.87797034e-01
-7.68046439e-01 -3.24658692e-01 -6.24281764e-01 4.40490514e-01
2.13454589e-01 8.25380832e-02 -7.53726244e-01 -4.68436569e-01
-5.53556681e-01 -2.24266201e-01 8.33677113e-01 -2.90524475e-02
1.17896688e+00 -3.89978528e-01 -1.70335218e-01 6.84226871e-01
1.60318601e+00 -3.44558924e-01 6.51548266e-01 7.59497508e-02
1.45096755e+00 4.24065530e-01 -5.07337153e-02 7.76880085e-02
4.61612344e-01 2.51738936e-01 6.64274871e-01 -2.43607700e-01
-2.33356625e-01 -5.62972307e-01 3.48091155e-01 9.01476085e-01
4.34029661e-03 -5.26372612e-01 -1.14072371e+00 4.59347963e-01
-1.88854682e+00 -4.09495115e-01 -1.63962066e-01 2.03371859e+00
4.89183187e-01 2.14666307e-01 1.83130503e-02 8.50026086e-02
9.69668090e-01 4.93623644e-01 -2.28179589e-01 -3.01922053e-01
-1.58440500e-01 3.94110054e-01 5.60193539e-01 2.55442828e-01
-1.05168176e+00 1.24034858e+00 5.33400631e+00 5.51971018e-01
-1.25140369e+00 -9.41214934e-02 2.91296095e-01 5.63015163e-01
-4.33337092e-01 1.33322909e-01 -2.47849748e-01 3.42474192e-01
6.36577666e-01 -3.60537350e-01 7.82502830e-01 7.71717370e-01
-3.06787431e-01 4.26212430e-01 -9.91629541e-01 9.10457015e-01
1.01318330e-01 -1.48738933e+00 1.95887506e-01 -6.18923791e-02
8.49748194e-01 2.70188153e-01 -2.66240180e-01 7.79179215e-01
5.66175997e-01 -1.08745337e+00 2.96187997e-01 2.75520682e-01
3.70378822e-01 -6.57833219e-01 8.70817721e-01 2.06216499e-01
-1.47486055e+00 5.53191453e-02 -4.07776266e-01 -2.56617010e-01
-2.60269284e-01 5.68781674e-01 -7.63804674e-01 9.02093112e-01
6.94792330e-01 9.23658192e-01 -9.96076703e-01 8.76145899e-01
-5.71600914e-01 5.00952482e-01 -1.01590343e-01 -1.70280352e-01
3.22278142e-01 -2.82470793e-01 5.63657522e-01 1.05891275e+00
7.97576159e-02 -1.73191711e-01 4.44111496e-01 1.38207996e+00
-8.05581987e-01 2.95311689e-01 -1.01189315e+00 -7.28074968e-01
2.53888190e-01 1.53729475e+00 -9.79383707e-01 -2.69508064e-01
-5.74557304e-01 6.08233929e-01 6.84686363e-01 4.54563648e-01
-6.57430410e-01 -7.64087081e-01 3.53502542e-01 4.48998734e-02
2.47156229e-02 -4.90241170e-01 -1.84086680e-01 -1.34845638e+00
4.84801903e-02 -7.24024296e-01 5.31693995e-01 -5.88575721e-01
-1.36267722e+00 6.37747467e-01 1.07297271e-01 -8.58040690e-01
9.71548036e-02 -6.76045120e-01 -1.10091507e+00 7.60739267e-01
-1.42328084e+00 -1.49567187e+00 -5.12071550e-01 5.13801217e-01
-5.50655127e-02 -2.23242626e-01 5.97368240e-01 3.33104193e-01
-9.49631035e-01 6.96083486e-01 -5.12176275e-01 5.37302494e-01
3.45518172e-01 -1.34343386e+00 8.21813941e-01 1.04362643e+00
2.72084236e-01 7.99452186e-01 2.74164468e-01 -5.90561092e-01
-1.65493083e+00 -1.36210620e+00 7.59477854e-01 -2.28208229e-01
8.58607531e-01 -8.02603543e-01 -1.28033519e+00 9.16964233e-01
2.40914211e-01 5.61790884e-01 3.10872972e-01 7.97422044e-03
-5.21515310e-01 5.04729077e-02 -1.11929297e+00 7.91244328e-01
1.19378912e+00 -6.83813453e-01 -4.49143529e-01 3.21648926e-01
1.18772638e+00 -3.70741189e-01 -1.07733631e+00 5.38735688e-01
5.76762035e-02 -8.57948065e-01 7.89014041e-01 -6.38186336e-01
4.51651484e-01 -2.72695869e-01 4.64592166e-02 -1.41476405e+00
-2.14530379e-01 -7.31893957e-01 1.75649822e-01 1.38797140e+00
4.30680543e-01 -1.12261462e+00 7.94569552e-01 2.91943818e-01
-3.49211901e-01 -7.20699131e-01 -4.92029011e-01 -7.40455568e-01
1.97740793e-01 -2.47964084e-01 8.93906057e-01 1.43881702e+00
1.79544732e-01 2.88314164e-01 -1.62237898e-01 1.47753045e-01
3.24300408e-01 -3.52332070e-02 1.00401616e+00 -1.13514996e+00
-8.92806053e-02 -6.59684420e-01 -7.38749325e-01 -4.78198498e-01
3.40408564e-01 -1.55733705e+00 -2.67292142e-01 -1.67870700e+00
1.08890727e-01 -2.55005181e-01 -4.46515888e-01 9.86151397e-01
-4.61302936e-01 1.21679045e-01 2.86056608e-01 -1.76760368e-02
-3.27369153e-01 6.10800743e-01 1.37083113e+00 -4.45753336e-01
-3.79674375e-01 -3.08607519e-01 -8.09655249e-01 4.98143196e-01
8.03193390e-01 -4.79690552e-01 -4.69465882e-01 -3.61563444e-01
5.47954977e-01 -2.75138319e-01 3.62727940e-01 -8.72511744e-01
1.91168502e-01 -4.50437255e-02 -3.82419616e-01 -5.18591926e-02
-2.56861836e-01 -4.10682619e-01 -8.24659467e-02 4.40477729e-01
-9.80661586e-02 4.59469110e-01 3.18670601e-01 4.36347812e-01
-4.34483379e-01 -9.82486084e-02 5.59376180e-01 -3.31157357e-01
-7.69790232e-01 7.12467313e-01 -4.30512019e-02 1.23378657e-01
8.45018327e-01 8.05658549e-02 -6.56858265e-01 -6.75363392e-02
-4.71977681e-01 2.83438891e-01 4.77864504e-01 7.80336976e-01
5.70403636e-01 -1.50416315e+00 -6.38305008e-01 5.70493698e-01
4.08092499e-01 2.72182941e-01 3.92408632e-02 7.55436242e-01
-6.64573073e-01 9.45482776e-02 -2.07955837e-01 -4.08604771e-01
-9.73462880e-01 7.05573857e-01 2.78837502e-01 -7.34333813e-01
-5.73465586e-01 8.02050412e-01 1.08834669e-01 -1.08652008e+00
6.33751750e-02 -8.33408654e-01 -1.85536101e-01 -1.71849772e-01
6.71215057e-02 2.68034697e-01 1.02291934e-01 -3.42402130e-01
-1.08589739e-01 4.19102907e-01 -1.93441659e-02 5.68218589e-01
1.21483386e+00 4.70860600e-01 -7.24008918e-01 2.45515466e-01
1.33642685e+00 -1.67192206e-01 -7.66375840e-01 -4.15190965e-01
-1.60992183e-02 -2.54038841e-01 -4.86530103e-02 -1.22255929e-01
-1.40849566e+00 8.45814586e-01 1.34778589e-01 8.50927532e-01
9.85398591e-01 -3.31870117e-03 5.82795739e-01 3.56371909e-01
2.87990332e-01 -6.09286129e-01 3.01976085e-01 4.98185635e-01
9.99404728e-01 -1.32796299e+00 -6.35997131e-02 -7.84459591e-01
-2.29860321e-01 1.00529242e+00 9.46225464e-01 -5.57757795e-01
8.36928308e-01 -5.98445758e-02 -3.83796453e-01 -6.23143315e-01
-3.89397383e-01 -7.72337094e-02 4.16096061e-01 3.41372877e-01
3.78695279e-01 1.99739322e-01 -1.33942336e-01 5.20679474e-01
-2.74159938e-01 -1.47578746e-01 5.31355560e-01 9.80962276e-01
4.71448041e-02 -1.21123326e+00 -5.82717359e-02 6.81547642e-01
-2.57300496e-01 -4.08674777e-01 -6.87405407e-01 1.06727850e+00
-4.15849388e-02 6.86603069e-01 -5.48453405e-02 -7.81959891e-01
3.87942851e-01 -3.66561897e-02 3.13714474e-01 -9.50886965e-01
-8.10222387e-01 -5.85750341e-01 -6.34471551e-02 -4.10287380e-01
-2.74798930e-01 -1.51150167e-01 -1.38433015e+00 -4.18649644e-01
-4.81840312e-01 -1.58667073e-01 3.35327744e-01 7.38368571e-01
4.11506653e-01 8.55368733e-01 5.04493833e-01 -7.02396691e-01
-1.61036447e-01 -1.15964961e+00 -5.84312558e-01 5.51624715e-01
6.83129728e-02 -5.32111824e-01 -4.07359213e-01 -3.50606769e-01] | [7.07392692565918, 6.305450439453125] |
a893faf5-0c3e-4f30-8e3e-4f423ddfd8dc | recommendation-as-instruction-following-a | 2305.07001 | null | https://arxiv.org/abs/2305.07001v1 | https://arxiv.org/pdf/2305.07001v1.pdf | Recommendation as Instruction Following: A Large Language Model Empowered Recommendation Approach | In the past decades, recommender systems have attracted much attention in both research and industry communities, and a large number of studies have been devoted to developing effective recommendation models. Basically speaking, these models mainly learn the underlying user preference from historical behavior data, and then estimate the user-item matching relationships for recommendations. Inspired by the recent progress on large language models (LLMs), we take a different approach to developing the recommendation models, considering recommendation as instruction following by LLMs. The key idea is that the preferences or needs of a user can be expressed in natural language descriptions (called instructions), so that LLMs can understand and further execute the instruction for fulfilling the recommendation task. Instead of using public APIs of LLMs, we instruction tune an open-source LLM (3B Flan-T5-XL), in order to better adapt LLMs to recommender systems. For this purpose, we first design a general instruction format for describing the preference, intention, task form and context of a user in natural language. Then we manually design 39 instruction templates and automatically generate a large amount of user-personalized instruction data (252K instructions) with varying types of preferences and intentions. To demonstrate the effectiveness of our approach, we instantiate the instruction templates into several widely-studied recommendation (or search) tasks, and conduct extensive experiments on these tasks with real-world datasets. Experiment results show that the proposed approach can outperform several competitive baselines, including the powerful GPT-3.5, on these evaluation tasks. Our approach sheds light on developing more user-friendly recommender systems, in which users can freely communicate with the system and obtain more accurate recommendations via natural language instructions. | ['Ji-Rong Wen', 'Leyu Lin', 'Wayne Xin Zhao', 'Yupeng Hou', 'Ruobing Xie', 'Junjie Zhang'] | 2023-05-11 | null | null | null | null | ['instruction-following'] | ['natural-language-processing'] | [ 5.20665348e-02 -3.47911209e-01 -4.72566307e-01 -8.33728254e-01
-4.00301665e-01 -5.69545805e-01 5.25880635e-01 -1.94452181e-01
-3.17729026e-01 1.11562453e-01 5.55143058e-01 -6.30768597e-01
-1.48207545e-01 -7.41748869e-01 -5.20278037e-01 -1.50010064e-01
1.72157705e-01 4.86263692e-01 1.75859734e-01 -6.13247335e-01
5.66972435e-01 5.00575814e-04 -1.96169674e+00 7.02473402e-01
1.27992475e+00 8.36142719e-01 5.78788519e-01 3.82624745e-01
-2.57178158e-01 6.02246404e-01 -1.51862025e-01 -5.52515566e-01
-9.06619355e-02 -2.44648352e-01 -7.63336241e-01 -1.36121571e-01
2.14301392e-01 -5.08502424e-01 -2.16096729e-01 6.31914496e-01
3.17029923e-01 7.35377192e-01 7.17943132e-01 -9.05686021e-01
-1.08763444e+00 1.06154454e+00 3.93909812e-02 -2.15608522e-01
8.09489846e-01 -1.76721543e-01 1.23056674e+00 -8.77466023e-01
2.61320919e-01 1.15568030e+00 3.14545840e-01 6.96420372e-01
-8.85206461e-01 -5.54464698e-01 6.38132036e-01 2.47458026e-01
-1.14486194e+00 -3.43973517e-01 4.84045833e-01 -2.96325624e-01
1.00110865e+00 7.54154921e-01 2.52461404e-01 1.38466644e+00
1.14313155e-01 1.05486047e+00 6.45838082e-01 -3.65740895e-01
2.05379203e-02 4.00387496e-01 8.08789015e-01 4.71130103e-01
-8.61293599e-02 -6.52457923e-02 -2.91694015e-01 -3.50570083e-01
3.79619122e-01 6.16741717e-01 -4.42781717e-01 -1.50746927e-01
-1.09748685e+00 8.07676435e-01 1.30785316e-01 3.45989019e-01
-1.93613932e-01 -3.71785790e-01 2.31845170e-01 4.47028637e-01
2.16212809e-01 5.51685929e-01 -6.99364424e-01 -9.43345204e-02
-3.71147126e-01 3.41964841e-01 1.00021720e+00 1.22104704e+00
5.86335659e-01 -4.10424739e-01 -4.92274672e-01 1.23985589e+00
6.60009384e-01 4.43793446e-01 1.12864912e+00 -5.52999556e-01
4.27825451e-01 5.69722295e-01 3.86273295e-01 -1.06103468e+00
-3.28755438e-01 -3.89849842e-01 -4.85912353e-01 -6.77760780e-01
1.18926488e-01 5.74353300e-02 -4.79618192e-01 1.54376554e+00
-7.32694492e-02 1.72912836e-01 1.86124370e-01 9.47209775e-01
1.01278138e+00 8.58040988e-01 -4.58366759e-02 -1.76974133e-01
1.41449201e+00 -1.25327325e+00 -4.02719378e-01 -3.54714185e-01
9.37037528e-01 -8.05612862e-01 1.79928660e+00 5.66622496e-01
-5.49361944e-01 -9.00538385e-01 -7.10302889e-01 4.01452444e-02
-2.29467332e-01 3.26162457e-01 7.44219780e-01 6.58028543e-01
-7.23914623e-01 6.12001240e-01 -6.47823453e-01 -5.86587727e-01
-3.20592999e-01 4.57703352e-01 1.29003718e-01 -3.78921628e-01
-1.30419040e+00 4.42570329e-01 6.54209107e-02 -1.43724913e-02
-4.23595279e-01 -5.68956375e-01 -9.25677717e-01 1.65275037e-01
4.44532841e-01 -5.73252618e-01 1.52241278e+00 -8.50375235e-01
-1.84777153e+00 2.42148682e-01 -1.07492112e-01 6.08913861e-02
-2.11605936e-01 -4.31562275e-01 -8.94032061e-01 -8.35928380e-01
-3.04537177e-01 1.49329633e-01 4.64255631e-01 -9.70649660e-01
-9.42434013e-01 -1.58828706e-01 4.44513202e-01 2.13607788e-01
-8.57547879e-01 3.02128226e-01 -1.01715076e+00 -5.82707644e-01
-3.48907024e-01 -1.15669358e+00 -4.74960715e-01 -9.10898864e-01
-1.49729162e-01 -5.81115544e-01 2.69896626e-01 -2.97409534e-01
1.79409289e+00 -2.00920534e+00 2.86429189e-03 2.77976245e-01
-1.92285150e-01 4.00559992e-01 -5.89789689e-01 5.07355273e-01
3.07633728e-01 -8.15668851e-02 1.90355316e-01 -2.90352404e-01
3.83191377e-01 3.78250331e-01 -6.91229224e-01 -1.32231936e-01
-6.73939466e-01 8.80174577e-01 -1.04534733e+00 -5.26939631e-02
1.02511778e-01 4.05577362e-01 -1.26901352e+00 7.92016268e-01
-4.43542778e-01 2.04198942e-01 -9.85016942e-01 3.72990072e-01
2.15508848e-01 -3.62977773e-01 4.14946884e-01 -1.74959823e-01
-9.59491059e-02 7.87965119e-01 -1.22686720e+00 1.80870140e+00
-1.02271068e+00 -1.42640427e-01 -3.59546900e-01 -6.60694599e-01
9.54169214e-01 8.09808299e-02 3.25987339e-01 -7.60226071e-01
1.81218274e-02 8.13641548e-02 -1.09425887e-01 -7.17603743e-01
7.76845396e-01 6.17389560e-01 -2.82834589e-01 8.48187983e-01
-1.39965802e-01 4.14935499e-01 2.44525820e-01 6.52410015e-02
8.64525259e-01 3.05215925e-01 2.51723409e-01 -2.93784082e-01
9.15558636e-01 -2.99577713e-01 4.78602260e-01 7.83804297e-01
4.53506351e-01 2.60687858e-01 -2.37291530e-01 -5.63380480e-01
-2.18417402e-02 -3.44254732e-01 1.32343471e-01 1.99816823e+00
1.81494474e-01 -1.08454275e+00 -5.11754632e-01 -9.63954985e-01
-1.07975431e-01 1.15036166e+00 -3.79870683e-01 -2.34638676e-01
-5.89138627e-01 -7.01096892e-01 -5.09447232e-02 4.02393460e-01
1.22732699e-01 -1.25088000e+00 -2.89019257e-01 3.84729207e-01
-3.45432907e-01 -8.37761939e-01 -1.25413394e+00 -2.24382460e-01
-6.74073398e-01 -8.59704971e-01 -2.74981350e-01 -7.35357046e-01
6.71765685e-01 7.56579101e-01 1.43502188e+00 4.39238608e-01
3.79436553e-01 4.97502655e-01 -9.09459233e-01 1.56500377e-02
-3.53982210e-01 3.92305076e-01 2.92451739e-01 1.31226227e-01
6.55220270e-01 -5.42316437e-01 -6.54452205e-01 1.03824747e+00
-9.65906262e-01 1.03748500e-01 5.55515051e-01 5.98082423e-01
4.97197032e-01 -1.93847656e-01 4.58633155e-01 -1.42406857e+00
9.93243039e-01 -6.01250529e-01 -5.58442295e-01 4.88370210e-01
-1.09074056e+00 2.62003839e-01 1.15663564e+00 -5.41371047e-01
-1.14685667e+00 -1.00002721e-01 -6.45052314e-01 1.18119176e-02
-3.47217202e-01 8.43147516e-01 -4.41554636e-01 1.13773257e-01
6.66694462e-01 3.09909999e-01 -4.78332072e-01 -8.59585762e-01
6.70737982e-01 1.23116612e+00 3.06655407e-01 -9.66808558e-01
4.57363814e-01 -1.87158644e-01 -9.72842574e-01 -3.85811776e-01
-1.02484763e+00 -6.89773142e-01 -2.36321256e-01 9.94646251e-02
3.10020089e-01 -5.44466853e-01 -7.67996967e-01 -7.48974159e-02
-7.41589665e-01 -4.48755711e-01 2.49066204e-01 6.01672709e-01
-3.78172636e-01 3.58624369e-01 -6.84341252e-01 -4.67720956e-01
-6.01713300e-01 -1.57790589e+00 8.67068112e-01 3.11145008e-01
-3.98741215e-01 -9.57072675e-01 2.39272982e-01 5.06435871e-01
6.32772088e-01 -8.19986165e-01 9.40958500e-01 -1.07386661e+00
-1.82547346e-01 -2.04414412e-01 9.76468101e-02 1.59144595e-01
3.30651045e-01 -6.81829080e-02 -7.32052445e-01 -3.08715761e-01
-1.76100016e-01 -9.66693684e-02 4.44401890e-01 4.09934521e-02
1.52379191e+00 -5.32290816e-01 -5.21777213e-01 6.12310469e-01
1.07942462e+00 3.75152677e-01 5.16086042e-01 1.42087728e-01
7.13216662e-01 5.50488830e-01 8.85170996e-01 3.75398248e-01
8.96491408e-01 1.05920005e+00 1.32744297e-01 3.68368119e-01
2.86168069e-01 -5.13876021e-01 8.22852671e-01 1.33468390e+00
-2.03570321e-01 -4.64862764e-01 -4.12307829e-01 4.32313792e-02
-2.08057451e+00 -7.12544918e-01 4.40678261e-02 2.34019756e+00
8.56693447e-01 -1.24377489e-01 2.13735119e-01 -3.94996434e-01
2.40456343e-01 -8.01195577e-02 -4.66406882e-01 -4.80073810e-01
5.28143108e-01 -5.32180592e-02 8.47575068e-02 5.21345258e-01
-7.68683076e-01 9.40185010e-01 5.79680061e+00 8.13268721e-01
-1.11324787e+00 -2.83670068e-01 3.97387952e-01 2.26551920e-01
-6.41958654e-01 -1.54126599e-01 -1.19918799e+00 5.63534141e-01
1.23734963e+00 -2.74140924e-01 8.70033503e-01 9.81045008e-01
2.29871586e-01 5.86991072e-01 -1.45432806e+00 9.03612018e-01
2.08957911e-01 -1.14177203e+00 2.84553140e-01 5.08462787e-02
5.63128531e-01 -7.14926943e-02 1.26869753e-01 1.06819546e+00
6.60186172e-01 -6.91776037e-01 2.80671418e-01 4.91996288e-01
4.91543651e-01 -5.12310565e-01 7.05745935e-01 7.92693377e-01
-1.24099231e+00 -2.16577426e-01 -6.55793309e-01 -1.78491980e-01
-1.07406385e-01 2.29978234e-01 -5.12229443e-01 6.85703099e-01
6.21376872e-01 9.77789819e-01 -5.37049532e-01 7.45857894e-01
-2.60146827e-01 7.40631163e-01 -2.30414849e-02 -2.53148139e-01
-4.09338921e-02 -5.46300292e-01 -8.08124095e-02 1.24339414e+00
5.24551988e-01 3.50869745e-01 4.53276068e-01 5.65739930e-01
-1.82105839e-01 7.96524644e-01 -3.19396138e-01 4.04532515e-02
4.56926525e-01 1.58592677e+00 -2.26223186e-01 -1.94829062e-01
-1.00657213e+00 9.12573218e-01 3.02573889e-01 3.48909885e-01
-8.00956964e-01 -1.53061986e-01 8.56192589e-01 1.53537780e-01
1.85956180e-01 1.36564925e-01 3.68708014e-01 -1.55720997e+00
-3.16780031e-01 -1.47414947e+00 6.45714998e-01 -5.67868173e-01
-1.40806711e+00 8.73915195e-01 -9.78747159e-02 -1.34627175e+00
-5.21440983e-01 -5.08400321e-01 -7.68642128e-01 6.66297793e-01
-1.30458975e+00 -9.37656522e-01 -2.97776282e-01 5.49314082e-01
9.00505662e-01 -2.26127401e-01 1.13156497e+00 5.75911701e-01
-8.50505650e-01 9.11213577e-01 1.03934236e-01 -2.34041095e-01
1.08042431e+00 -1.03276408e+00 5.65374255e-01 5.72771370e-01
4.97765422e-01 1.37771487e+00 6.57725811e-01 -3.64425331e-01
-1.85969949e+00 -1.24116147e+00 8.69831085e-01 -6.01558208e-01
6.02496028e-01 -3.92258763e-01 -9.82286036e-01 8.10990095e-01
1.18090354e-01 -2.40288809e-01 1.20408285e+00 4.81238425e-01
-3.48575979e-01 -2.13969141e-01 -6.42895997e-01 7.26944923e-01
1.24184346e+00 -2.24673197e-01 -6.84208691e-01 3.37530851e-01
8.61205101e-01 -3.16255152e-01 -9.50192630e-01 2.73453057e-01
8.21517050e-01 -8.17130983e-01 1.01936126e+00 -8.41952145e-01
1.30047470e-01 -2.28815958e-01 -2.87066668e-01 -1.52642488e+00
-6.82543755e-01 -7.37710595e-01 -1.49276242e-01 1.32019842e+00
5.99509299e-01 -5.68911791e-01 5.42599142e-01 9.77753401e-01
-4.76689160e-01 -8.76403272e-01 9.53475907e-02 -4.09324646e-01
-2.09553897e-01 -5.71326911e-01 1.14822066e+00 8.06868732e-01
4.05837387e-01 5.66531003e-01 -5.99152327e-01 1.23160712e-01
3.95599566e-02 8.51168275e-01 9.64279592e-01 -1.30575645e+00
-6.84045196e-01 -4.45394069e-01 4.70241576e-01 -1.99284911e+00
3.73397052e-01 -9.94412422e-01 1.06814794e-01 -1.36705339e+00
7.32803866e-02 -6.76881552e-01 -7.14646876e-01 5.14909983e-01
-3.96160960e-01 -1.47457272e-01 -1.07937776e-01 4.18293208e-01
-9.81827378e-01 4.65828478e-01 1.08615065e+00 -9.54473671e-03
-4.61211264e-01 6.23224437e-01 -1.27437055e+00 7.58959115e-01
4.71626282e-01 -2.07407236e-01 -8.20478737e-01 -5.52810609e-01
4.23866570e-01 4.00604047e-02 -4.89889413e-01 -5.45807719e-01
2.45303020e-01 -5.96335530e-01 -2.55584747e-01 -3.29640985e-01
-1.57779112e-01 -9.50529039e-01 9.49504673e-02 4.02356461e-02
-6.94668055e-01 1.97012410e-01 -9.52601656e-02 3.70525926e-01
3.89282927e-02 -2.46774152e-01 8.03883467e-03 1.68054640e-01
-9.30686533e-01 5.67556143e-01 -3.92563134e-01 -3.61777067e-01
5.86693406e-01 3.16793174e-01 -2.97080338e-01 -4.24256086e-01
-3.66882712e-01 4.24801379e-01 4.74114656e-01 1.00965118e+00
5.52255213e-01 -1.38036311e+00 -5.04559517e-01 4.35200393e-01
6.86273158e-01 -5.27082562e-01 1.18571497e-01 4.97776270e-01
1.21211290e-01 7.66527474e-01 1.26132950e-01 -1.30944878e-01
-1.15274489e+00 8.14457655e-01 1.89820692e-01 -3.35675567e-01
-4.53102857e-01 6.49816751e-01 3.72716486e-01 -9.36381936e-01
3.74759704e-01 -6.57932162e-01 -8.84968638e-01 -3.03284466e-01
9.82413828e-01 5.28356321e-02 1.46038264e-01 -4.79496419e-01
-1.98139921e-01 4.78404135e-01 -3.93697023e-01 4.57884759e-01
1.35759592e+00 -2.16577426e-01 1.12254784e-01 1.74610525e-01
9.48530674e-01 3.53190720e-01 -7.24559128e-01 -5.70107222e-01
4.58401963e-02 -5.13062119e-01 2.55184062e-02 -7.60259628e-01
-1.16365623e+00 4.00325209e-01 3.08459222e-01 2.06766456e-01
1.05950761e+00 -1.18446127e-01 1.15498018e+00 8.63834679e-01
5.99502742e-01 -7.95687675e-01 -1.45277619e-01 6.88197792e-01
7.53437221e-01 -1.15569985e+00 -3.15445751e-01 -2.99985975e-01
-7.13271976e-01 1.00638294e+00 8.06915045e-01 1.28666654e-01
6.90169692e-01 -9.94372368e-02 1.27033815e-01 3.46696153e-02
-1.06605303e+00 -7.52223702e-03 8.91767740e-01 2.79046863e-01
1.13124681e+00 1.68637723e-01 -4.84298617e-01 1.65735686e+00
-1.79682583e-01 1.63352951e-01 2.69091696e-01 5.79737008e-01
-4.20782626e-01 -1.82586551e+00 -9.96750593e-02 7.32946217e-01
-7.82079771e-02 -2.61423111e-01 -5.86923063e-02 2.54129887e-01
-1.25345036e-01 1.24659789e+00 -3.31687003e-01 -1.09788263e+00
8.43560994e-01 -5.79702184e-02 7.62430802e-02 -1.06603551e+00
-7.95515478e-01 -4.41174097e-02 2.05515504e-01 -9.28559959e-01
-1.58442095e-01 -6.19884253e-01 -1.15285730e+00 -1.98208407e-01
-4.26329494e-01 6.85603499e-01 3.07554662e-01 1.03346598e+00
5.38781822e-01 3.96692812e-01 7.84449518e-01 -7.16253102e-01
-6.82775617e-01 -1.01329768e+00 -3.24630618e-01 5.74798107e-01
-2.33874232e-01 -5.68985879e-01 -1.24521732e-01 -1.00735165e-02] | [10.236934661865234, 5.761397361755371] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.