Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files Community
1
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
6cab27ec-7f83-4930-9b7b-e6e7ccffa97d
where2comm-communication-efficient
2209.12836
null
https://arxiv.org/abs/2209.12836v1
https://arxiv.org/pdf/2209.12836v1.pdf
Where2comm: Communication-Efficient Collaborative Perception via Spatial Confidence Maps
Multi-agent collaborative perception could significantly upgrade the perception performance by enabling agents to share complementary information with each other through communication. It inevitably results in a fundamental trade-off between perception performance and communication bandwidth. To tackle this bottleneck issue, we propose a spatial confidence map, which reflects the spatial heterogeneity of perceptual information. It empowers agents to only share spatially sparse, yet perceptually critical information, contributing to where to communicate. Based on this novel spatial confidence map, we propose Where2comm, a communication-efficient collaborative perception framework. Where2comm has two distinct advantages: i) it considers pragmatic compression and uses less communication to achieve higher perception performance by focusing on perceptually critical areas; and ii) it can handle varying communication bandwidth by dynamically adjusting spatial areas involved in communication. To evaluate Where2comm, we consider 3D object detection in both real-world and simulation scenarios with two modalities (camera/LiDAR) and two agent types (cars/drones) on four datasets: OPV2V, V2X-Sim, DAIR-V2X, and our original CoPerception-UAVs. Where2comm consistently outperforms previous methods; for example, it achieves more than $100,000 \times$ lower communication volume and still outperforms DiscoNet and V2X-ViT on OPV2V. Our code is available at https://github.com/MediaBrain-SJTU/where2comm.
['Siheng Chen', 'Yiqi Zhong', 'Zixing Lei', 'Shaoheng Fang', 'Yue Hu']
2022-09-26
null
null
null
null
['monocular-3d-object-detection']
['computer-vision']
[-2.23582506e-01 -1.14234962e-01 2.70298034e-01 -3.00734714e-02 -3.62280667e-01 -8.89866412e-01 7.47525811e-01 6.63588047e-01 -6.82468593e-01 5.94332516e-01 1.31361544e-01 -2.32622966e-01 -3.37237120e-01 -1.16676092e+00 -5.36060929e-01 -5.82565188e-01 -5.21111667e-01 4.49833065e-01 7.70899236e-01 -3.13201696e-01 1.43776432e-01 3.56696814e-01 -1.85489058e+00 5.31488156e-04 8.60571861e-01 1.17989647e+00 9.04112935e-01 8.69903386e-01 2.87890464e-01 6.80803120e-01 -9.52183604e-01 -1.36263475e-01 4.95521843e-01 1.06379442e-01 -3.77857506e-01 -4.31136489e-02 -4.11445787e-03 -3.17726135e-01 -1.75795838e-01 9.08792436e-01 6.42624080e-01 2.81949759e-01 4.82896030e-01 -1.63571119e+00 -5.59225738e-01 5.19777834e-01 -7.52921045e-01 3.41538548e-01 4.41007257e-01 3.99865538e-01 6.09902203e-01 -5.75460255e-01 3.30899000e-01 1.52120292e+00 3.69844437e-01 -5.00234589e-02 -1.23558044e+00 -6.30897522e-01 3.98474425e-01 3.39927405e-01 -1.55005527e+00 -4.47062194e-01 3.98634434e-01 -2.63077497e-01 8.36204469e-01 4.54902649e-01 7.27240205e-01 7.12027848e-01 6.17827214e-02 4.14262354e-01 1.24496853e+00 1.21236749e-01 6.07424021e-01 -8.07025135e-02 -3.99058014e-01 5.81825435e-01 1.72884330e-01 3.99011523e-01 -7.71861076e-01 -2.20081285e-01 7.87889540e-01 -2.96761412e-02 -3.30584973e-01 -1.06731810e-01 -1.53582835e+00 7.40701854e-01 7.52682030e-01 -1.78220123e-01 -5.64856410e-01 3.94910604e-01 1.96768910e-01 5.68735957e-01 2.54902601e-01 2.09891394e-01 -2.49699131e-01 -1.93247616e-01 -4.36792485e-02 1.36000380e-01 6.10531211e-01 1.09169555e+00 7.79592991e-01 -2.75701821e-01 1.02370597e-01 8.36544216e-01 4.04946446e-01 1.11467063e+00 2.09518634e-02 -1.58229768e+00 3.54289711e-01 4.50091183e-01 2.51465410e-01 -1.31649709e+00 -4.61834460e-01 -1.67668909e-01 -9.34665143e-01 6.38971925e-01 5.15490919e-02 -3.66429985e-01 -4.87287819e-01 1.71033287e+00 4.89596963e-01 7.20903724e-02 5.37150621e-01 1.12190890e+00 1.00289655e+00 8.01005065e-01 -3.09090223e-02 -1.28834069e-01 1.43082130e+00 -1.00834131e+00 -2.83776045e-01 -2.71525770e-01 8.74009803e-02 -8.92510056e-01 8.41141284e-01 2.43170202e-01 -1.00034177e+00 -5.97919464e-01 -9.42824304e-01 2.99025029e-01 -4.40786630e-01 -2.22094297e-01 6.38299346e-01 3.47234368e-01 -1.41662872e+00 -1.06138460e-01 -5.88361621e-01 -3.99788201e-01 1.65509075e-01 2.75457740e-01 -5.03887720e-02 -5.08158393e-02 -7.97362983e-01 5.75221360e-01 1.03371151e-01 -4.18837786e-01 -1.29918420e+00 -5.11590898e-01 -5.32518625e-01 4.71403152e-02 8.59763384e-01 -8.80389869e-01 1.17943799e+00 -5.87893128e-01 -1.56858289e+00 9.30805877e-02 1.31919041e-01 -4.37961280e-01 4.75454539e-01 1.70642704e-01 -2.27409258e-01 2.60057479e-01 1.85906351e-01 1.15651274e+00 4.97492045e-01 -1.70615184e+00 -1.08223534e+00 -3.27388108e-01 4.93974686e-01 5.93948781e-01 -1.12644946e-02 -4.22824919e-01 -6.38851821e-01 -3.57526511e-01 3.24742347e-02 -9.76521552e-01 -2.45134637e-01 2.73248225e-01 -2.98630834e-01 -3.06337714e-01 8.31492066e-01 1.73008353e-01 2.88392365e-01 -2.33040023e+00 1.99378118e-01 1.95534632e-01 5.00247478e-01 5.44784069e-02 -4.13204312e-01 7.67515659e-01 1.02538633e+00 -1.35529175e-01 1.34628952e-01 -6.55355930e-01 4.33365107e-02 4.78690445e-01 -1.69971883e-01 3.87948602e-01 -4.13181365e-01 6.15516663e-01 -1.00680757e+00 -4.34388548e-01 4.66255695e-01 6.01300299e-01 -5.40876091e-01 2.18913257e-01 -2.63096392e-01 4.50328887e-01 -3.16492647e-01 6.68280125e-01 9.59696174e-01 -2.49402151e-01 2.98539400e-01 4.04061982e-03 -4.83240336e-01 -1.42810285e-01 -1.21570420e+00 1.54124570e+00 -5.30656219e-01 6.50552034e-01 6.87953055e-01 -5.62933028e-01 8.69722605e-01 -4.01042290e-02 3.71044576e-01 -9.11909997e-01 1.49591058e-01 -5.13339043e-02 -2.67606139e-01 -1.84006602e-01 6.17682695e-01 5.04634678e-01 -2.13068858e-01 2.18167171e-01 -4.24818963e-01 -3.31647396e-01 1.44952595e-01 3.63840193e-01 1.30287445e+00 -6.14573419e-01 -3.03511205e-03 -1.95027605e-01 1.09058343e-01 1.30871207e-01 6.19179726e-01 9.17326450e-01 -4.44191277e-01 -2.28832308e-02 9.12999436e-02 -1.77234352e-01 -5.48819900e-01 -1.48107553e+00 2.61038244e-01 1.13496733e+00 1.22705781e+00 -3.79393876e-01 -2.37081423e-01 -2.21562102e-01 3.80329609e-01 5.07410288e-01 -4.16009843e-01 2.04675019e-01 1.23270415e-02 -3.80775332e-01 3.16945106e-01 3.18494707e-01 1.05697262e+00 -7.47201741e-01 -1.22684896e+00 6.16648793e-02 -2.95151383e-01 -1.31898272e+00 -3.66597235e-01 -1.87028974e-01 -2.26153731e-01 -1.13069546e+00 -3.81365627e-01 -4.01934236e-01 4.75869477e-01 1.23943520e+00 1.01155961e+00 1.87491670e-01 4.64858785e-02 8.89489949e-01 -7.32041419e-01 -7.65049636e-01 -2.49372125e-01 -3.12307328e-01 2.81505078e-01 -2.87310988e-01 8.21857676e-02 -6.03420973e-01 -8.97941709e-01 6.60276890e-01 -5.08320570e-01 8.44200850e-02 5.81131935e-01 2.60906219e-01 5.81612229e-01 3.42308521e-01 4.66106117e-01 -9.29391757e-02 7.91443586e-01 -6.17867827e-01 -9.86681938e-01 -1.02074936e-01 -3.82185638e-01 -5.90656161e-01 5.15603542e-01 -3.62900406e-01 -7.07698226e-01 -2.35492080e-01 2.28130236e-01 -2.72424906e-01 -2.32856497e-01 3.49413544e-01 2.35545799e-01 -1.40943617e-01 3.86838138e-01 1.42588928e-01 1.39870897e-01 -1.31783187e-01 4.92059231e-01 6.16451800e-01 6.43383622e-01 -3.50402743e-01 6.34838402e-01 8.24354947e-01 -8.07248354e-02 -1.03714025e+00 3.92085612e-02 -3.40340793e-01 2.05118172e-02 -4.96952266e-01 6.92568004e-01 -1.07208276e+00 -1.62094307e+00 4.32945609e-01 -1.28840327e+00 -6.43716872e-01 -1.68361112e-01 6.62806153e-01 -4.17051971e-01 2.90970087e-01 -3.17966610e-01 -9.44602847e-01 3.70802023e-02 -1.33575726e+00 7.69029558e-01 2.65464127e-01 2.98652858e-01 -6.76667571e-01 1.24992547e-03 3.15175802e-01 5.24753273e-01 2.29720809e-02 4.45345759e-01 -1.62704900e-01 -1.06157231e+00 4.35118288e-01 -6.71650112e-01 -2.02132165e-01 1.93657931e-02 -2.71556705e-01 -7.32514024e-01 -7.01972961e-01 -5.78438938e-01 -3.25066030e-01 8.77263069e-01 2.78285235e-01 8.81019473e-01 -3.14967901e-01 -4.42194849e-01 4.58435982e-01 1.30947816e+00 4.42068726e-01 1.96363315e-01 1.00254044e-01 1.63338721e-01 5.59161305e-01 8.10828805e-01 8.14271748e-01 1.28640628e+00 8.26227784e-01 1.27949607e+00 -1.16394147e-01 -3.99284154e-01 -7.86275640e-02 3.96043718e-01 7.72872329e-01 -2.07536951e-01 -8.57010007e-01 -6.63127244e-01 5.63570619e-01 -2.01907754e+00 -7.01419234e-01 2.34687626e-01 2.18641829e+00 2.97550589e-01 -5.55948503e-02 3.05233300e-01 -1.18583910e-01 5.75489581e-01 1.63677394e-01 -7.23978102e-01 -6.71844259e-02 -1.92378357e-01 -7.44911134e-01 6.48479164e-01 7.26186097e-01 -7.05995560e-01 6.57231808e-01 5.43335295e+00 7.04520047e-01 -8.73584569e-01 4.38224494e-01 2.28040591e-01 -2.84977734e-01 -8.35468322e-02 -3.19016635e-01 -4.18567955e-01 6.80996716e-01 5.80369174e-01 -2.11061686e-02 7.80851007e-01 6.11951232e-01 3.18720639e-01 -6.28963947e-01 -7.20660269e-01 1.13725221e+00 -1.57781512e-01 -1.55725837e+00 -2.70755827e-01 3.09456080e-01 4.08012092e-01 5.60359836e-01 2.17319027e-01 5.81055842e-02 9.76843953e-01 -7.23342538e-01 9.07543063e-01 2.13060200e-01 6.67197168e-01 -6.95266008e-01 8.11606050e-01 4.39323336e-01 -1.62293184e+00 -4.31943893e-01 -5.42691708e-01 -2.44566679e-01 1.59978151e-01 3.65125179e-01 -7.05113649e-01 4.52696204e-01 1.06667030e+00 3.75981033e-01 -2.39116341e-01 1.00745022e+00 -9.01647471e-03 -6.40414432e-02 -5.68718135e-01 -3.88856977e-01 2.07883507e-01 -5.04152067e-02 8.99361670e-01 9.33342814e-01 4.67759728e-01 4.60170954e-01 7.01130927e-01 6.05296552e-01 5.39801307e-02 -2.58326024e-01 -5.84969401e-01 4.65604275e-01 1.46897578e+00 1.02075481e+00 -8.02293181e-01 -3.49053442e-01 -3.79350364e-01 7.73920417e-01 2.28759170e-01 3.58926415e-01 -8.30087960e-01 -2.32355937e-01 1.11301851e+00 -2.31840029e-01 4.21613067e-01 -7.15498745e-01 9.18702781e-02 -5.77108979e-01 -3.26816082e-01 -5.22679687e-01 1.58708796e-01 -7.24796832e-01 -1.07181668e+00 8.35318923e-01 2.91879755e-02 -1.11856294e+00 2.58596808e-01 -1.88438296e-01 -4.05518055e-01 3.31765831e-01 -1.65205979e+00 -1.13078475e+00 -9.02787209e-01 7.96327770e-01 6.82368338e-01 -3.66336614e-01 6.52437031e-01 1.74493656e-01 -2.66174942e-01 1.97072700e-01 -9.68126729e-02 -4.16157126e-01 6.16286099e-01 -1.02335274e+00 7.70800859e-02 5.74161828e-01 -6.84438273e-03 9.74313021e-02 6.61536157e-01 -5.27807951e-01 -1.74512172e+00 -1.07129490e+00 1.87501952e-01 -2.36384079e-01 3.61125290e-01 -1.65984824e-01 -2.12745994e-01 2.34895527e-01 5.58012486e-01 -1.70517281e-01 6.16235495e-01 -1.11245468e-01 -3.08464259e-01 -3.28021139e-01 -1.29730928e+00 6.47558808e-01 1.15598881e+00 -9.66356769e-02 -1.62528470e-01 2.53416181e-01 1.09429562e+00 -1.56630099e-01 -9.03958499e-01 1.75026372e-01 3.43018085e-01 -1.36124098e+00 1.11409426e+00 4.66571510e-01 -1.23533301e-01 -5.92114687e-01 -7.65448689e-01 -1.83463228e+00 -2.19768003e-01 -6.88120008e-01 1.20542631e-01 8.84824276e-01 2.80743420e-01 -1.21616089e+00 4.31946516e-01 -4.99540567e-02 -2.33274713e-01 -4.60642219e-01 -1.09696448e+00 -8.73114526e-01 -3.01974982e-01 -4.03792828e-01 1.09662211e+00 6.12292111e-01 -9.20489579e-02 1.63954049e-01 1.63592363e-03 7.87064075e-01 9.17735338e-01 2.25157559e-01 1.25164020e+00 -1.07826960e+00 -3.27945173e-01 -4.47013766e-01 -3.30643952e-01 -1.62597179e+00 -3.90510470e-01 -4.29540843e-01 4.31597605e-02 -1.65193462e+00 -4.02123258e-02 -1.03401065e+00 1.09904110e-01 5.56232929e-01 4.32872474e-01 4.09370273e-01 6.06141388e-01 4.46672380e-01 -9.81617689e-01 6.37924552e-01 1.24346960e+00 -2.45519280e-01 -3.23347986e-01 -1.04088075e-01 -6.11541033e-01 7.52163589e-01 9.39686120e-01 5.42905591e-02 -7.85438597e-01 -9.26976204e-01 1.04944356e-01 3.01360786e-01 8.18449914e-01 -1.20239258e+00 7.33213186e-01 -4.36763197e-01 3.76352891e-02 -7.45037615e-01 9.71787035e-01 -9.88044322e-01 1.00496694e-01 5.37335694e-01 -1.42116221e-02 4.93506044e-01 2.87298858e-01 8.63966882e-01 -3.05016488e-02 3.69643211e-01 5.13496757e-01 -1.67808294e-01 -1.08701682e+00 2.26024345e-01 -7.52975285e-01 -2.07655549e-01 1.37600291e+00 -1.22723863e-01 -1.11888528e+00 -7.01381087e-01 -3.34501594e-01 7.36273587e-01 7.27306247e-01 3.65134150e-01 1.01146054e+00 -1.06774390e+00 -7.86181569e-01 1.77422598e-01 2.00303137e-01 7.65071586e-02 3.64044368e-01 8.18753898e-01 -3.75993729e-01 2.55094975e-01 -2.92216331e-01 -8.70907128e-01 -1.42308998e+00 6.15364611e-01 -8.12977105e-02 4.67981607e-01 -3.57420385e-01 1.05523562e+00 2.98389286e-01 -3.59739929e-01 4.23922509e-01 -2.31938154e-01 -5.87386191e-02 -1.90106675e-01 5.15589535e-01 7.17257738e-01 -4.77203995e-01 -6.91049397e-01 -5.00432909e-01 5.72063446e-01 4.46040273e-01 -2.14156821e-01 1.13978171e+00 -8.60251367e-01 -4.43765409e-02 -5.08260094e-02 6.27161443e-01 2.26133496e-01 -1.62473166e+00 -2.12681621e-01 -6.40409112e-01 -8.63336563e-01 3.77344131e-01 -8.89307618e-01 -1.17893863e+00 3.13511372e-01 7.20678091e-01 8.49023223e-01 1.13867402e+00 4.36021537e-01 6.33220971e-01 3.43714833e-01 1.03210676e+00 -8.94812286e-01 4.80307102e-01 2.92179793e-01 1.05838633e+00 -1.31753516e+00 6.34819120e-02 -7.77752101e-01 -7.47258723e-01 5.81338406e-01 6.51319265e-01 2.33689487e-01 6.00746334e-01 4.82830703e-01 2.75862515e-01 -4.10419285e-01 -1.19992912e+00 -5.24420738e-01 -3.70292485e-01 1.27963424e+00 -4.86068457e-01 5.25483191e-01 2.55665064e-01 -1.23109603e-02 -3.03071558e-01 -4.59549993e-01 5.38746238e-01 8.12632918e-01 -7.24135041e-01 -7.94079244e-01 -4.70704406e-01 9.13855806e-02 3.46232355e-01 1.92995250e-01 -3.48537534e-01 6.60773218e-01 3.62333357e-01 1.79927707e+00 4.02911305e-01 -7.67019868e-01 2.67735392e-01 -1.22552741e+00 2.53847271e-01 -2.72162974e-01 -3.34506214e-01 7.40791932e-02 2.65538990e-01 -7.54167020e-01 -5.37232161e-01 -4.85499710e-01 -1.20168567e+00 -9.89797235e-01 -1.04726858e-01 2.04703733e-01 7.97863185e-01 2.39495605e-01 9.47129130e-01 6.50749087e-01 6.99915826e-01 -1.16811430e+00 -2.14729041e-01 -6.88915908e-01 -3.16338420e-01 -4.82690446e-02 4.23223108e-01 -1.13245082e+00 -3.58755976e-01 -3.62914741e-01]
[7.147497653961182, -1.9383258819580078]
e8da2d09-d511-4997-87ad-f01f37b1de50
using-snomed-to-recognize-and-index-chemical
null
null
https://aclanthology.org/D19-5718
https://aclanthology.org/D19-5718.pdf
Using Snomed to recognize and index chemical and drug mentions.
In this paper we describe a new named entity extraction system. Our work proposes a system for the identification and annotation of drug names in Spanish biomedical texts based on machine learning and deep learning models. Subsequently, a standardized code using Snomed is assigned to these drugs, for this purpose, Natural Language Processing tools and techniques have been used, and a dictionary of different sources of information has been built. The results are promising, we obtain 78{\%} in F1 score on the first sub-track and in the second task we map with Snomed correctly 72{\%} of the found entities.
['L. Alfonso Urena Lopez', "Manuel Carlos D{\\'\\i}az Galiano", "Pilar L{\\'o}pez {\\'U}beda", 'Maite Martin']
2019-11-01
null
null
null
ws-2019-11
['entity-extraction']
['natural-language-processing']
[-2.41922334e-01 4.76073027e-01 -4.77039188e-01 -2.53373176e-01 -6.94394767e-01 -4.77727681e-01 5.50122142e-01 9.84207630e-01 -7.99610078e-01 1.24328792e+00 5.01517914e-02 -7.42449239e-02 -3.83610427e-01 -7.60990739e-01 -4.95224327e-01 -5.62536240e-01 -1.23661563e-01 9.76300418e-01 -4.85026799e-02 2.07502004e-02 9.98342931e-02 6.16847754e-01 -8.92611444e-01 5.38594306e-01 8.57044041e-01 6.58805668e-01 1.19249150e-01 9.07155126e-02 -4.49950308e-01 1.00147951e+00 -6.88399613e-01 -9.04626131e-01 -2.28120103e-01 -3.77398282e-01 -1.21207523e+00 -6.05375946e-01 4.97504510e-02 5.16766347e-02 -1.20042086e-01 1.35047650e+00 6.14415109e-01 -1.87101349e-01 9.71537471e-01 -6.24761581e-01 -3.63976419e-01 1.16063535e+00 -4.15550061e-02 2.10925899e-02 6.26333833e-01 -3.41976315e-01 9.31262791e-01 -1.08522415e+00 1.32993853e+00 5.41401982e-01 9.40128386e-01 5.88517129e-01 -8.47900331e-01 -8.03349555e-01 -6.72132015e-01 8.84839669e-02 -1.89660180e+00 -2.53764898e-01 8.25650468e-02 -9.64601576e-01 1.38004506e+00 -3.04002576e-02 3.83937210e-01 9.81867433e-01 5.10250330e-01 3.61787826e-01 6.29527152e-01 -4.35107648e-01 4.21647191e-01 5.51376045e-01 1.06551275e-01 7.90997148e-01 6.55462921e-01 -2.43063718e-01 -2.53062516e-01 -3.48766923e-01 2.99581230e-01 -2.51477718e-01 4.42596041e-02 -1.67282104e-01 -9.30240571e-01 8.57574224e-01 4.49087262e-01 1.07070053e+00 -7.32969105e-01 -4.83427167e-01 7.27474809e-01 -2.62746871e-01 3.19634736e-01 1.06539655e+00 -6.63365722e-01 2.10467890e-01 -1.04268622e+00 5.59903905e-02 1.15638852e+00 1.11959755e+00 2.83996642e-01 -4.89456713e-01 -1.32033423e-01 7.44949341e-01 1.84900641e-01 2.90129781e-01 6.50231838e-01 -4.44819212e-01 3.94426614e-01 8.36946964e-01 8.30027908e-02 -1.06424904e+00 -1.02022135e+00 -4.00022417e-01 -9.48399007e-01 -3.24832141e-01 3.46629769e-01 -1.48548722e-01 -9.08891618e-01 1.40522373e+00 1.73651174e-01 -6.51814416e-02 3.57283682e-01 3.03660125e-01 1.34567773e+00 3.39999616e-01 8.19257975e-01 -2.64976591e-01 1.58010161e+00 -4.25979316e-01 -1.24827683e+00 2.28603110e-01 8.82438302e-01 -7.72654712e-01 1.19408913e-01 4.64872599e-01 -7.21525371e-01 -3.86755854e-01 -9.71185207e-01 1.55993202e-03 -1.09519839e+00 2.82788783e-01 5.65611660e-01 5.70972323e-01 -7.94553876e-01 9.78081286e-01 -6.26102507e-01 -5.39563656e-01 8.06316197e-01 6.48732901e-01 -8.00900817e-01 7.31367292e-03 -1.50998676e+00 1.38299704e+00 1.05117822e+00 -1.94375858e-01 -7.49164701e-01 -7.01804936e-01 -8.62388313e-01 -5.45149529e-03 2.70875767e-02 -6.69430614e-01 9.60914075e-01 -4.52324837e-01 -8.12961698e-01 1.29032576e+00 8.96467566e-02 -6.49672925e-01 2.61911988e-01 -9.94632021e-02 -8.89843881e-01 2.29583055e-01 6.45354331e-01 3.64703804e-01 -2.22631887e-01 -7.89845765e-01 -7.27036059e-01 -3.53190899e-01 -2.76467919e-01 -2.30734929e-01 -3.30363452e-01 3.17157000e-01 -3.62259209e-01 -5.47976375e-01 -3.26153219e-01 -6.01168990e-01 -2.69645125e-01 -5.51075637e-01 -4.91839200e-01 -4.39119309e-01 -1.33702382e-01 -1.02130044e+00 1.67210269e+00 -2.02914834e+00 4.11895588e-02 4.18234438e-01 7.10857987e-01 4.11496699e-01 2.56019801e-01 4.99861777e-01 -3.38538855e-01 3.22051167e-01 -2.84581810e-01 2.06968680e-01 -1.22659199e-01 6.75155744e-02 8.98927599e-02 2.51603574e-01 -1.39441527e-03 5.37274480e-01 -1.12665594e+00 -6.55370831e-01 3.91516574e-02 4.80559617e-01 -4.27762479e-01 -1.95760615e-02 3.97786219e-03 3.59326184e-01 -6.32310688e-01 5.80898166e-01 5.49704552e-01 -3.69085640e-01 5.89377403e-01 -2.34260783e-01 6.77852333e-03 3.79608333e-01 -8.00083339e-01 1.86879718e+00 -2.93838441e-01 2.33862087e-01 -3.87958854e-01 -7.61756063e-01 8.01386178e-01 8.07654023e-01 9.48058605e-01 -5.11204123e-01 3.64862084e-01 5.96470356e-01 -1.00904122e-01 -1.04410696e+00 2.88986146e-01 -9.86357257e-02 -3.96588333e-02 -3.25774789e-01 6.78557336e-01 3.73744160e-01 4.48219419e-01 4.48890403e-02 1.28186107e+00 1.77685022e-01 1.17206037e+00 -3.35705698e-01 8.30462813e-01 4.28223759e-01 4.33848321e-01 5.35058320e-01 -1.39252737e-01 1.57777950e-01 4.20767784e-01 -8.08454573e-01 -1.06686819e+00 -6.13643885e-01 -7.01390445e-01 5.67429483e-01 -4.11517441e-01 -5.26526213e-01 -1.06603861e+00 -9.34956968e-01 5.70214493e-03 7.23773599e-01 -7.13482916e-01 3.56524847e-02 -4.46226507e-01 -9.13403511e-01 1.06433952e+00 2.47308969e-01 1.43974483e-01 -1.33101177e+00 -3.32535475e-01 4.55026060e-01 4.11641225e-02 -1.30481207e+00 4.08093669e-02 4.78377491e-01 -5.94422638e-01 -1.55155110e+00 -6.33986950e-01 -9.82410848e-01 4.34938371e-01 -9.08710778e-01 1.26025331e+00 -3.40473086e-01 -2.77233213e-01 -4.48167086e-01 -2.53512472e-01 -5.19374192e-01 -6.30786598e-01 2.00081274e-01 -2.86127515e-02 -3.19891959e-01 1.13753450e+00 -7.22810179e-02 -3.48763525e-01 -7.51656964e-02 -7.63657749e-01 -3.45009625e-01 8.02188277e-01 5.70021510e-01 6.59965456e-01 -1.75205201e-01 6.98644638e-01 -1.41016412e+00 5.22731364e-01 -7.63988376e-01 -5.37629187e-01 1.50642648e-01 -7.93116271e-01 1.28250420e-01 6.21012986e-01 -1.10708877e-01 -6.13676608e-01 5.82870960e-01 -8.65330756e-01 4.31553787e-03 -5.09834826e-01 8.44747603e-01 -2.88960397e-01 2.07388565e-01 1.09070992e+00 -7.94270337e-02 -5.40970862e-01 -6.15352750e-01 3.04490238e-01 9.63012457e-01 2.86858469e-01 2.01987233e-02 9.42374617e-02 9.79204550e-02 -3.49914543e-02 -5.19986033e-01 -7.90136218e-01 -6.09796405e-01 -7.54772246e-01 2.95524418e-01 1.49646878e+00 -8.72363091e-01 -7.24524856e-01 1.45463124e-01 -1.41282964e+00 2.47738346e-01 -1.55550435e-01 8.87246311e-01 -3.53337735e-01 1.21592917e-01 -6.14668310e-01 -3.16536009e-01 -6.71360672e-01 -1.03592503e+00 8.36245954e-01 2.40226299e-01 -7.93910563e-01 -9.48020637e-01 5.03567100e-01 3.90507802e-02 1.56808466e-01 4.34068322e-01 9.46116388e-01 -1.70827508e+00 2.40573600e-01 -4.21898931e-01 -1.23831369e-01 1.33082522e-02 3.32888365e-01 -4.78303552e-01 -8.72313797e-01 7.43694603e-02 -9.87745449e-02 2.37534985e-01 5.42802215e-01 3.83988976e-01 9.67578530e-01 -2.91148961e-01 -9.27705824e-01 4.97474968e-01 1.40346289e+00 8.49461436e-01 5.81697643e-01 4.82170731e-01 6.63957894e-01 3.92563105e-01 2.78369635e-01 3.36240709e-01 1.42200440e-01 6.67752683e-01 3.29942927e-02 -1.80738732e-01 5.36266640e-02 -9.26332772e-02 -7.26879910e-02 4.94784236e-01 -6.98303282e-02 -5.56058176e-02 -1.21940529e+00 5.33066034e-01 -1.41438985e+00 -8.15062642e-01 -2.19564453e-01 1.90315413e+00 1.27975643e+00 3.85978632e-02 -5.72423786e-02 -2.18094081e-01 7.73888350e-01 -5.27177989e-01 -2.83779114e-01 -1.98979288e-01 -1.48014143e-01 8.69170129e-01 7.42450118e-01 2.39840358e-01 -1.46332800e+00 1.05428505e+00 6.70819044e+00 9.17735398e-01 -8.73426795e-01 2.36150995e-01 3.41135591e-01 4.06768560e-01 2.43754461e-01 -4.40810949e-01 -1.03151357e+00 5.10769248e-01 1.31006873e+00 -2.07307935e-01 -2.20020726e-01 8.58137429e-01 -5.38435206e-03 1.62492335e-01 -1.03914785e+00 9.80520487e-01 2.62213886e-01 -1.61702478e+00 6.90972060e-02 1.20085977e-01 4.58844811e-01 1.77350253e-01 -5.10315597e-01 2.78480649e-01 2.86416650e-01 -1.19957590e+00 2.66462952e-01 8.44560683e-01 1.04039466e+00 -7.36514866e-01 1.37990308e+00 1.36627322e-02 -1.00441873e+00 1.91010870e-02 -4.03490156e-01 6.62608564e-01 -1.83612078e-01 7.66183078e-01 -1.31964171e+00 9.83124018e-01 5.41811168e-01 8.35942745e-01 -6.70725107e-01 1.49989259e+00 -4.70055401e-01 2.84265697e-01 9.65729542e-03 -1.70035690e-01 3.72930281e-02 2.12851077e-01 2.42869034e-01 1.60133231e+00 3.30874234e-01 -1.13405772e-01 7.14755207e-02 8.28791142e-01 -4.19381857e-01 8.77517879e-01 -7.93781757e-01 -2.02757329e-01 4.09549415e-01 1.24290073e+00 -8.00532103e-01 -7.33167827e-01 -2.03492075e-01 7.81513393e-01 4.27848399e-02 -8.02380890e-02 -8.51014674e-01 -8.54808033e-01 2.42899925e-01 -8.56012404e-02 9.43864435e-02 3.84342879e-01 -3.57892901e-01 -1.11512923e+00 -4.46793020e-01 -8.80929232e-01 9.21728730e-01 -6.45677388e-01 -1.22950888e+00 1.05993211e+00 -1.18435264e-01 -1.24174559e+00 -3.39775324e-01 -9.14207280e-01 2.66999125e-01 8.04443002e-01 -9.48717713e-01 -9.04398918e-01 2.44154304e-01 3.91004413e-01 2.25594342e-01 -6.31440938e-01 1.53565145e+00 1.06269550e+00 -4.94594157e-01 4.50437844e-01 2.36411378e-01 6.45611048e-01 8.38065624e-01 -1.31292164e+00 6.94287717e-02 3.49472791e-01 1.63028926e-01 8.51448357e-01 5.52672148e-01 -9.66465831e-01 -6.40300453e-01 -1.16432297e+00 1.74677336e+00 -5.74556112e-01 7.07164645e-01 -1.12884544e-01 -7.91923583e-01 5.73991179e-01 3.34366828e-01 -2.76053578e-01 1.35865223e+00 -2.26837024e-02 -3.54887724e-01 2.01364353e-01 -1.47002208e+00 -3.28694619e-02 6.22645080e-01 -4.27828461e-01 -7.83251643e-01 8.33745420e-01 4.67518926e-01 -4.46083874e-01 -1.45837033e+00 3.27752143e-01 3.95012677e-01 -2.02913985e-01 8.81695926e-01 -1.09568751e+00 3.85279715e-01 -3.38465184e-01 7.10339099e-02 -9.84752536e-01 -2.83812732e-01 -5.70617355e-02 4.68588322e-02 1.11605728e+00 1.05238640e+00 -3.50512475e-01 4.64164436e-01 3.78280729e-01 -2.00847030e-01 -6.04565859e-01 -1.01908755e+00 -4.67912793e-01 1.60582080e-01 -1.10929057e-01 5.52455664e-01 1.54291046e+00 3.08490932e-01 4.79572475e-01 -1.53814360e-01 1.53391019e-01 1.00263357e-01 -3.78442854e-01 7.65516385e-02 -1.36573982e+00 7.37379491e-02 -5.67905962e-01 -5.40492654e-01 -5.60571700e-02 9.44428146e-02 -1.26522410e+00 -2.19885379e-01 -1.74547601e+00 5.20867050e-01 -1.21734805e-01 -5.93168080e-01 8.55123520e-01 9.22684520e-02 3.45695727e-02 -3.08681905e-01 9.11667272e-02 -6.57607019e-01 -1.76924139e-01 4.92833257e-01 -2.19858706e-01 -1.08095773e-01 -1.95642054e-01 -6.57292724e-01 8.34874213e-01 6.58132434e-01 -1.02682412e+00 3.23624492e-01 -1.22079402e-01 5.37926316e-01 -1.88361660e-01 -2.43709385e-01 -9.78841066e-01 2.64111042e-01 1.28085747e-01 5.55997908e-01 -5.01273155e-01 -2.33834684e-01 -9.01666284e-01 5.56094170e-01 9.45646286e-01 -4.19688761e-01 -9.22474563e-02 4.32584316e-01 1.90514535e-01 -3.49273831e-01 -6.28001928e-01 7.65900373e-01 -2.15823293e-01 -5.25561333e-01 3.31146903e-02 -4.27839756e-01 -9.19659659e-02 1.05584359e+00 3.27060401e-01 -4.96834470e-03 3.74557465e-01 -1.35826325e+00 -8.31420049e-02 -6.33841427e-03 2.79454887e-01 3.84917617e-01 -1.26745427e+00 -6.58032238e-01 -1.70951977e-01 4.81282622e-01 -4.05546993e-01 8.78100321e-02 8.32145751e-01 -1.15199685e+00 7.89499760e-01 -4.68707472e-01 -2.10269421e-01 -1.05128360e+00 8.95986557e-01 5.30547500e-01 -6.98266268e-01 -2.33321235e-01 7.53379524e-01 -7.96326697e-02 -5.57909071e-01 2.38179773e-01 -1.88063577e-01 -1.16667986e+00 5.68236709e-01 6.63975060e-01 1.18189484e-01 5.17981529e-01 -8.82157266e-01 -7.24956036e-01 3.75728875e-01 9.37006809e-03 3.20647925e-01 1.55040419e+00 4.66282159e-01 -4.93250072e-01 1.01653792e-01 1.14309335e+00 2.24877194e-01 1.32316366e-01 -3.00321616e-02 7.31244087e-01 1.35392338e-01 -3.03346157e-01 -1.22936368e+00 -9.44522917e-01 5.75568736e-01 9.84552443e-01 9.40306038e-02 8.94101799e-01 1.03708632e-01 4.85170990e-01 4.99487042e-01 2.54120559e-01 -1.01696587e+00 -8.05931807e-01 5.86164951e-01 5.30245006e-01 -1.05452955e+00 6.02371991e-02 -4.17328149e-01 -4.36490536e-01 1.16485715e+00 9.67731476e-02 1.78108901e-01 7.05527544e-01 2.27129996e-01 -1.01404518e-01 -7.40476429e-01 -1.25765949e-01 -1.78265885e-01 4.34812576e-01 4.92732793e-01 9.13655341e-01 1.74881041e-01 -1.07649982e+00 1.16285300e+00 -5.28948158e-02 3.13590109e-01 1.39525652e-01 5.25575042e-01 -3.06271613e-01 -1.20051765e+00 -7.24793747e-02 3.78392279e-01 -1.19284391e+00 -4.96626258e-01 -6.67890728e-01 9.31572795e-01 6.93213344e-01 6.43379033e-01 -4.69600588e-01 -3.57940733e-01 5.95839798e-01 3.79948258e-01 8.76842737e-02 -8.94244313e-01 -9.58184481e-01 6.28535002e-02 3.20983946e-01 -4.20757473e-01 -4.97241616e-01 -4.29537535e-01 -1.58775592e+00 2.34213546e-01 -2.36328840e-01 4.98148531e-01 6.68387055e-01 1.00285602e+00 5.06705284e-01 6.31464183e-01 1.44016102e-01 -7.25840544e-03 4.34805229e-02 -1.19189847e+00 -4.83924329e-01 5.22973776e-01 -2.98147917e-01 -5.24425149e-01 1.49664953e-01 2.19012231e-01]
[8.495266914367676, 8.758156776428223]
e5954286-0ed3-4b86-96a8-7498ebe5ac59
warm-start-alphazero-self-play-search
2004.12357
null
https://arxiv.org/abs/2004.12357v1
https://arxiv.org/pdf/2004.12357v1.pdf
Warm-Start AlphaZero Self-Play Search Enhancements
Recently, AlphaZero has achieved landmark results in deep reinforcement learning, by providing a single self-play architecture that learned three different games at super human level. AlphaZero is a large and complicated system with many parameters, and success requires much compute power and fine-tuning. Reproducing results in other games is a challenge, and many researchers are looking for ways to improve results while reducing computational demands. AlphaZero's design is purely based on self-play and makes no use of labeled expert data ordomain specific enhancements; it is designed to learn from scratch. We propose a novel approach to deal with this cold-start problem by employing simple search enhancements at the beginning phase of self-play training, namely Rollout, Rapid Action Value Estimate (RAVE) and dynamically weighted combinations of these with the neural network, and Rolling Horizon Evolutionary Algorithms (RHEA). Our experiments indicate that most of these enhancements improve the performance of their baseline player in three different (small) board games, with especially RAVE based variants playing strongly.
['Hui Wang', 'Aske Plaat', 'Mike Preuss']
2020-04-26
null
null
null
null
['board-games']
['playing-games']
[-3.42913479e-01 -1.14105850e-01 -1.20701902e-01 -9.04837996e-03 -4.43925112e-01 -6.38540149e-01 5.05648255e-01 -1.79725781e-01 -8.45550895e-01 1.03011286e+00 -1.58477992e-01 -3.83218586e-01 -3.57942760e-01 -9.67102051e-01 -5.66537857e-01 -8.31567764e-01 -2.50811338e-01 7.36027300e-01 6.23105407e-01 -1.24122882e+00 3.92912567e-01 9.28489491e-02 -1.51167619e+00 5.53160906e-02 8.00299406e-01 9.14527297e-01 1.44366607e-01 8.80346179e-01 2.02191904e-01 1.13743901e+00 -8.07408690e-01 -5.04830897e-01 6.81067586e-01 -6.32479429e-01 -6.97032332e-01 -2.97922820e-01 -2.61773288e-01 -4.74898040e-01 -2.22776785e-01 7.40167022e-01 9.01305795e-01 5.55961549e-01 4.02317643e-02 -1.24120951e+00 -5.78173669e-03 7.67191589e-01 -5.65316916e-01 4.83675748e-01 8.16247314e-02 6.15406632e-01 8.75606537e-01 -1.95464611e-01 3.09747249e-01 9.23174977e-01 9.82130051e-01 8.28881681e-01 -9.53841150e-01 -6.77353024e-01 -1.55670181e-01 3.05084437e-01 -1.08958864e+00 -2.92992413e-01 4.25820082e-01 -4.40802649e-02 1.34426558e+00 2.58312553e-01 1.21138453e+00 9.89462137e-01 3.13530654e-01 7.29230583e-01 1.08445477e+00 -4.05139267e-01 6.47451103e-01 -5.01383618e-02 -5.95240116e-01 8.84932101e-01 -1.48756117e-01 7.85793364e-01 -4.01439220e-01 -1.31508157e-01 9.41117704e-01 -3.62672687e-01 -3.08283344e-02 -5.31486988e-01 -7.76621759e-01 1.02323627e+00 3.41172069e-01 7.54825026e-02 -4.54247326e-01 5.33603787e-01 5.07397711e-01 7.39854693e-01 2.47180358e-01 1.14234567e+00 -7.49867082e-01 -8.51864457e-01 -8.65926623e-01 7.17292070e-01 9.40014005e-01 4.72186387e-01 6.34702146e-01 3.55456501e-01 2.04340443e-02 8.72876227e-01 -1.94290251e-01 1.05371717e-02 9.82050657e-01 -1.15898943e+00 2.57499725e-01 2.30817899e-01 1.26791537e-01 -7.04800546e-01 -6.65258288e-01 -5.37071884e-01 -3.68720502e-01 7.89106965e-01 4.74376202e-01 -7.90841460e-01 -6.89472198e-01 1.71135890e+00 3.18612307e-01 2.61457115e-01 1.36756286e-01 8.58286440e-01 3.92765522e-01 5.86847723e-01 -7.14265257e-02 6.30453378e-02 8.32453966e-01 -1.21050775e+00 -1.63527593e-01 -3.66961360e-01 6.02844656e-01 -2.97807574e-01 9.28834558e-01 7.51404822e-01 -1.51625395e+00 -4.04989690e-01 -1.22609746e+00 5.22724032e-01 -2.80192316e-01 -3.76597404e-01 9.45304990e-01 8.00306857e-01 -1.31713378e+00 1.19451225e+00 -8.35283339e-01 -7.03511685e-02 1.43738016e-01 7.18636870e-01 -3.87216397e-02 4.24149513e-01 -1.43164146e+00 1.28940260e+00 8.13957572e-01 -1.27703130e-01 -1.04038250e+00 -6.28561318e-01 -5.10539412e-01 2.57284015e-01 9.99013960e-01 -7.50267446e-01 1.81455529e+00 -1.22386491e+00 -2.29155731e+00 3.52583200e-01 7.28040814e-01 -7.26382911e-01 6.47099733e-01 -3.96665260e-02 -7.11008236e-02 -2.31180772e-01 -2.23804072e-01 7.20834374e-01 9.20272708e-01 -8.22870791e-01 -9.29502904e-01 -7.25611970e-02 4.88644421e-01 6.07636690e-01 -1.92095831e-01 9.63612366e-03 -2.85081089e-01 -5.91833532e-01 -4.07743961e-01 -9.36447144e-01 -6.57049417e-01 -4.89053518e-01 3.02120417e-01 -3.21978062e-01 1.79312050e-01 -2.74212122e-01 1.15033066e+00 -1.85595393e+00 9.68307331e-02 2.67359704e-01 2.10404664e-01 5.18539965e-01 -2.71525085e-01 4.95730281e-01 -6.51751012e-02 -1.49266854e-01 9.03774127e-02 3.03852290e-01 1.33513302e-01 3.21227372e-01 5.35144545e-02 6.46452904e-02 7.34783486e-02 1.08153844e+00 -1.27221406e+00 -2.17372864e-01 5.54246716e-02 -2.01831341e-01 -1.08929181e+00 2.29440942e-01 -4.32238460e-01 -8.72003585e-02 -2.95287728e-01 2.97768950e-01 1.44450843e-01 -8.28643516e-02 2.65220106e-01 4.37096655e-01 -2.02251926e-01 1.03784956e-01 -1.20840847e+00 1.57886457e+00 -5.60909271e-01 3.09373707e-01 -1.07145466e-01 -1.14409232e+00 8.33111525e-01 2.21510485e-01 6.61617577e-01 -1.02087998e+00 4.33935374e-01 1.39879659e-01 4.83189076e-01 -4.95797664e-01 5.58578730e-01 -1.40818283e-01 4.77356985e-02 5.09347558e-01 2.07452625e-01 -4.32616770e-01 4.66987967e-01 -5.82013391e-02 1.36096239e+00 4.24045533e-01 4.04020637e-01 -2.40919769e-01 1.53510317e-01 2.15223521e-01 6.45849586e-01 1.12184918e+00 -2.69301593e-01 2.64025301e-01 5.79961658e-01 -7.32072413e-01 -1.13362038e+00 -7.94386506e-01 4.16905612e-01 1.66956019e+00 5.15850596e-02 -5.43230534e-01 -7.45990157e-01 -5.05396724e-01 8.70430470e-03 4.85097140e-01 -7.98580170e-01 -4.49797124e-01 -7.01185226e-01 -9.28568721e-01 4.76638436e-01 5.69108903e-01 7.09241629e-01 -1.42458987e+00 -1.24911058e+00 7.08929360e-01 1.69334814e-01 -3.72640640e-01 -2.65169650e-01 7.17075229e-01 -7.30662048e-01 -9.83644783e-01 -7.79742837e-01 -5.78671992e-01 -1.02538732e-03 -8.83741118e-03 1.37763572e+00 3.21851581e-01 -2.07603753e-01 6.71321601e-02 -4.85245377e-01 -5.37286997e-01 -3.92029166e-01 1.75426513e-01 -4.11495380e-02 -7.09181249e-01 8.94325897e-02 -6.93978846e-01 -5.56188226e-01 3.35734308e-01 -6.36630714e-01 3.23208347e-02 6.22453272e-01 1.33969879e+00 -1.49879642e-02 3.28000963e-01 6.74584746e-01 -6.94213152e-01 1.22333801e+00 -4.60590482e-01 -7.69474328e-01 1.87710360e-01 -7.85958648e-01 2.74671078e-01 7.66687393e-01 -5.68511963e-01 -9.14531708e-01 -2.00544536e-01 -4.34125215e-01 -3.17189217e-01 2.75822461e-01 3.69631171e-01 3.18092972e-01 -2.75319278e-01 1.14672375e+00 1.63449392e-01 2.41033405e-01 -1.82414129e-01 2.67356038e-01 2.62804687e-01 4.68875200e-01 -8.68898511e-01 4.81463701e-01 -2.38545522e-01 -3.35860312e-01 -2.85274148e-01 -4.93014276e-01 -1.11934066e-01 -9.15275142e-02 -3.54334980e-01 4.09721017e-01 -7.45789468e-01 -1.14971054e+00 6.34352744e-01 -4.48480546e-01 -1.02289653e+00 -6.60289884e-01 2.70557523e-01 -8.20557892e-01 5.37179932e-02 -5.46723545e-01 -4.65280324e-01 -3.27694386e-01 -1.01414311e+00 3.81421894e-01 6.33088350e-01 -2.46767011e-02 -7.87075460e-01 5.34853220e-01 8.71139318e-02 7.32036054e-01 7.94350430e-02 6.58183455e-01 -5.82414031e-01 -5.14986888e-02 1.19944490e-01 1.56084806e-01 3.20941567e-01 -1.18420117e-01 -1.15632571e-01 -4.71702725e-01 -4.51361299e-01 -5.62646650e-02 -8.88051033e-01 6.21512771e-01 3.15205634e-01 1.10758448e+00 -1.64096609e-01 1.39951959e-01 6.89274549e-01 1.37901366e+00 5.87072551e-01 5.57012618e-01 1.04942870e+00 1.42549708e-01 1.07065365e-01 6.22489333e-01 9.10636187e-01 3.15459847e-01 6.57878757e-01 6.69921279e-01 9.40624326e-02 2.30586827e-01 -7.71170035e-02 4.58605558e-01 3.89674485e-01 -5.92352867e-01 -1.15777902e-01 -8.36255491e-01 4.28032577e-01 -2.15686083e+00 -1.22128379e+00 5.22614419e-01 1.92365980e+00 1.10492718e+00 4.94698197e-01 6.74135923e-01 -2.39859987e-02 3.26267570e-01 -7.28032887e-02 -9.23033655e-01 -8.33331943e-01 1.90196913e-02 8.07549477e-01 5.01367867e-01 2.79228330e-01 -8.12734187e-01 1.33161855e+00 7.19891453e+00 9.23798323e-01 -1.13047397e+00 -1.19253516e-01 5.85249007e-01 -4.37759817e-01 3.97415496e-02 -9.65259150e-02 -3.47072005e-01 4.59451705e-01 9.37131286e-01 -4.34811085e-01 9.90582228e-01 1.07208741e+00 2.03602090e-02 -3.58938694e-01 -5.84734201e-01 7.36474395e-01 -2.57704049e-01 -1.54524565e+00 -5.70718527e-01 -1.09038286e-01 9.92046058e-01 2.18346536e-01 -5.21254130e-02 8.90457451e-01 1.17068207e+00 -8.61932635e-01 7.13492990e-01 2.20800877e-01 3.34099084e-01 -1.20513821e+00 7.19637990e-01 4.76389706e-01 -7.46512413e-01 -5.25888979e-01 -4.79034841e-01 -4.32028621e-01 -3.10360134e-01 -1.96785033e-01 -7.72784650e-01 2.98297346e-01 8.20528328e-01 2.21386135e-01 -5.04470527e-01 1.26858962e+00 -2.71893948e-01 5.44372499e-01 -2.56003976e-01 -5.38002610e-01 7.11064041e-01 -1.22119948e-01 2.75958776e-01 7.39221871e-01 1.73281416e-01 3.53393912e-01 1.78855389e-01 3.75303566e-01 1.99671775e-01 2.14889739e-02 -2.04727098e-01 9.04751346e-02 2.24423885e-01 1.09643722e+00 -7.52146065e-01 -2.47267455e-01 -9.33465660e-02 7.49587417e-01 4.87797141e-01 -5.42016365e-02 -1.05289674e+00 -5.34332216e-01 6.74472511e-01 -1.25709578e-01 5.38701355e-01 -1.72328681e-01 -3.28593962e-02 -1.02895689e+00 -4.03268814e-01 -1.37818956e+00 3.58609647e-01 -7.55014539e-01 -9.53025520e-01 6.27728403e-01 -4.40068506e-02 -9.93921697e-01 -9.04533386e-01 -4.50946778e-01 -7.70197928e-01 5.83497763e-01 -1.09684134e+00 -3.35293472e-01 2.78520118e-03 8.30158114e-01 5.62085688e-01 -6.10255599e-01 6.39344215e-01 6.47503063e-02 -4.82943982e-01 7.85330474e-01 4.12215352e-01 -1.55949533e-01 4.64239419e-01 -1.51014948e+00 5.42550683e-01 4.19814706e-01 -2.65521128e-02 2.23046675e-01 8.82069290e-01 -3.11208129e-01 -1.26820576e+00 -3.74383390e-01 1.40306652e-01 -1.24213569e-01 8.77350211e-01 -1.11618772e-01 -4.65612113e-01 2.44484037e-01 3.75860751e-01 -2.89705575e-01 3.13424617e-01 3.93619537e-01 1.88286632e-01 -2.31795102e-01 -9.87051606e-01 7.82701373e-01 8.94255698e-01 3.37393619e-02 -5.77451348e-01 1.11819111e-01 3.78622055e-01 -7.60344565e-01 -6.22532785e-01 1.86484233e-01 6.07485235e-01 -1.25747502e+00 9.10547793e-01 -8.79630744e-01 5.18042207e-01 4.24541086e-02 3.43948215e-01 -1.98364532e+00 -4.75087017e-01 -9.41288888e-01 -9.22582150e-02 4.92911667e-01 2.55695045e-01 -5.40640771e-01 1.38191783e+00 5.84678292e-01 -1.91462144e-01 -1.06838596e+00 -7.30280519e-01 -9.49094057e-01 1.44191563e-01 -8.87629688e-02 7.52716780e-01 7.92978168e-01 3.08828622e-01 2.60793179e-01 -6.77038491e-01 -4.82777327e-01 2.17992514e-01 6.28697174e-03 8.88389885e-01 -9.19543207e-01 -9.03385460e-01 -7.19860911e-01 -1.99898452e-01 -8.21161985e-01 -1.55337781e-01 -4.59620535e-01 3.17144811e-01 -1.06744206e+00 -1.79800764e-01 -5.46433091e-01 -4.23077285e-01 8.26179266e-01 -1.77061349e-01 8.15432742e-02 3.58913869e-01 -1.48420766e-01 -6.81498587e-01 6.79300725e-01 1.34786248e+00 -6.72363862e-02 -4.51515943e-01 2.56495029e-01 -8.20595801e-01 6.34680867e-01 1.16637588e+00 -4.63030934e-01 -7.04453886e-01 -3.50538045e-01 6.16050959e-01 4.39469367e-01 3.81280221e-02 -1.47126591e+00 3.58004361e-01 -3.83429229e-01 3.30664277e-01 -4.69301455e-03 2.33675629e-01 -4.69460428e-01 3.96757498e-02 7.36096740e-01 -2.12964147e-01 6.29458785e-01 4.52921748e-01 1.97066203e-01 -4.48019318e-02 -5.64441860e-01 7.85094023e-01 -6.07935190e-01 -1.08837974e+00 1.34390339e-01 -5.93218625e-01 4.57995713e-01 1.17503881e+00 -3.64381194e-01 -1.44353122e-01 -5.75008571e-01 -7.93051720e-01 5.44316173e-01 2.42236689e-01 3.85440946e-01 2.96077132e-01 -9.94892716e-01 -6.11895978e-01 6.84353337e-02 -3.38024795e-01 -3.01232547e-01 1.65065318e-01 3.49599391e-01 -7.59290278e-01 3.68125588e-02 -8.32382143e-01 -1.51567861e-01 -1.03441095e+00 3.61660331e-01 8.13668907e-01 -7.21272767e-01 -4.47810173e-01 1.24028015e+00 -3.72441322e-01 -4.55648869e-01 2.23816946e-01 3.81887034e-02 -1.32121384e-01 -5.20295724e-02 3.19928169e-01 4.15022910e-01 1.95310771e-01 2.06426680e-01 -7.57077113e-02 4.95622978e-02 -2.48735428e-01 -3.67273837e-01 1.62262416e+00 4.37910140e-01 2.31930405e-01 -7.78074265e-02 5.44335783e-01 -4.84413445e-01 -1.67665935e+00 -1.02951765e-01 -2.20555793e-02 -3.92237931e-01 1.42846793e-01 -1.14773953e+00 -1.13086736e+00 4.76748616e-01 5.18167615e-01 2.04176813e-01 1.24953544e+00 -5.11681020e-01 7.51553953e-01 8.96473646e-01 6.18069589e-01 -1.59368253e+00 3.96237522e-01 8.55574310e-01 6.15650892e-01 -1.07609272e+00 3.80193330e-02 5.86727262e-01 -8.17465544e-01 1.10768294e+00 1.11804557e+00 -4.27655488e-01 3.96856010e-01 3.28834087e-01 5.46904877e-02 -1.63961485e-01 -1.34753168e+00 -3.18875581e-01 -2.99330026e-01 5.24065375e-01 -4.41789255e-02 -8.46265629e-02 -4.35585082e-01 4.07502174e-01 -6.51236176e-01 1.20293073e-01 4.44040805e-01 1.20672822e+00 -6.65205598e-01 -1.44268334e+00 -5.18008918e-02 4.62200224e-01 -2.85542548e-01 -9.78817865e-02 -1.35990322e-01 1.03293824e+00 2.30826229e-01 6.70124650e-01 -4.49694581e-02 -6.69495225e-01 3.07052076e-01 -2.44665399e-01 8.04003358e-01 -3.07184190e-01 -1.26658928e+00 -1.93018943e-01 1.61761925e-01 -7.83421576e-01 6.44233897e-02 -5.84151387e-01 -1.04091823e+00 -7.36498713e-01 -2.91692048e-01 5.50881803e-01 5.44761837e-01 9.49428797e-01 1.53418049e-01 5.55075228e-01 8.35005522e-01 -9.87072408e-01 -1.10710025e+00 -5.95167339e-01 -6.12963676e-01 4.18456756e-02 6.82259127e-02 -7.93100417e-01 8.51794854e-02 -5.74414670e-01]
[3.561708927154541, 1.514366865158081]
e2d2905b-8c6b-40e1-b541-5ba8bd09a725
joint-optimization-of-cascade-ranking-models
null
null
https://dl.acm.org/citation.cfm?id=3290986
http://culpepper.io/publications/gcbc19-wsdm.pdf
Joint Optimization of Cascade Ranking Models
Reducing excessive costs in feature acquisition and model evaluation has been a long-standing challenge in learning-to-rank systems. A cascaded ranking architecture turns ranking into a pipeline of multiple stages, and has been shown to be a powerful approach to balancing efficiency and effectiveness trade-offs in large-scale search systems. However, learning a cascade model is often complex, and usually performed stagewise independently across the entire ranking pipeline. In this work we show that learning a cascade ranking model in this manner is often suboptimal in terms of both effectiveness and efficiency. We present a new general framework for learning an end-to-end cascade of rankers using backpropagation. We show that stagewise objectives can be chained together and optimized jointly to achieve significantly better trade-offs globally. This novel approach is generalizable to not only differentiable models but also state-of-the-art tree-based algorithms such as LambdaMART and cost-efficient gradient boosted trees, and it opens up new opportunities for exploring additional efficiency-effectiveness trade-offs in large-scale search systems.
['Ruey-Chen Chen', 'Roi Blanco', 'J. Shane Culpepper', 'Luke Gallagher']
2019-02-11
null
null
null
wsdm-2019-2
['ad-hoc-information-retrieval']
['natural-language-processing']
[ 2.89018769e-02 -3.45727175e-01 -4.17149216e-01 -7.26551890e-01 -1.27060866e+00 -6.96198761e-01 4.73834038e-01 1.78875387e-01 -5.77952683e-01 4.60947871e-01 1.52871951e-01 -3.51964921e-01 -6.42795384e-01 -4.70352620e-01 -6.85255826e-01 -3.61676008e-01 -3.28214675e-01 9.97532189e-01 1.68408662e-01 -4.20231730e-01 4.02484268e-01 4.29840654e-01 -1.76422429e+00 4.49128181e-01 9.97804105e-01 1.01548541e+00 7.28800520e-02 9.82686460e-01 6.03660792e-02 7.79514134e-01 -3.85071784e-01 -5.76987922e-01 2.77538508e-01 -5.31801991e-02 -9.69339728e-01 -5.73658049e-01 9.03643608e-01 -3.00576389e-01 -1.57554343e-01 6.91146553e-01 6.93890870e-01 1.41479418e-01 3.81308496e-01 -9.95367825e-01 -3.57919514e-01 8.45938802e-01 -5.29185236e-01 2.25307029e-02 -5.46011664e-02 -7.15218261e-02 1.68173385e+00 -8.07142198e-01 3.49262625e-01 1.49882507e+00 5.72876215e-01 1.86551765e-01 -1.42764866e+00 -6.70541465e-01 4.28310573e-01 2.40569830e-01 -9.07925367e-01 -3.55211318e-01 6.10191286e-01 -2.13241771e-01 1.01573896e+00 3.08290124e-01 6.75680697e-01 5.87448657e-01 1.38310045e-01 1.28494632e+00 1.04491329e+00 -3.44486773e-01 -1.12924069e-01 3.89570519e-02 4.25171256e-01 1.11539233e+00 -5.30279577e-02 3.48565012e-01 -9.99670804e-01 -4.16325569e-01 4.93463427e-01 1.52600948e-02 2.69801348e-01 -6.42278075e-01 -1.08810604e+00 9.84564126e-01 8.94430339e-01 -1.23308092e-01 -2.64563300e-02 7.40021288e-01 4.29202586e-01 6.70396745e-01 5.72194457e-01 1.06517100e+00 -9.20978785e-01 -2.21402332e-01 -1.20809233e+00 3.60095739e-01 6.79813683e-01 5.68454921e-01 8.09276700e-01 -4.37482566e-01 -4.49547946e-01 1.20001805e+00 1.61098793e-01 8.89991820e-02 2.22897410e-01 -9.12711143e-01 3.88068497e-01 6.92950070e-01 9.19927731e-02 -5.74267864e-01 -5.49799025e-01 -1.09218383e+00 -4.10298347e-01 2.29454905e-01 2.45770678e-01 -2.78940462e-02 -9.45320547e-01 1.84101772e+00 2.04765350e-01 -1.41094804e-01 -4.03029740e-01 8.03076982e-01 3.67108881e-01 3.21796268e-01 -3.02188545e-02 6.34128153e-02 1.13491523e+00 -1.56964195e+00 -1.23250723e-01 -3.49409461e-01 6.92052186e-01 -7.65815794e-01 1.33903551e+00 5.10342062e-01 -1.36205924e+00 -3.33821952e-01 -1.01736486e+00 -4.18291509e-01 -3.05830240e-01 2.80199289e-01 1.21796787e+00 4.18033868e-01 -1.23880672e+00 1.01699841e+00 -8.63480449e-01 1.73781766e-03 5.68082154e-01 9.02978301e-01 1.17433146e-01 -4.37168293e-02 -1.08449697e+00 1.08788729e+00 7.22239017e-02 2.18709201e-01 -1.04595995e+00 -9.01178598e-01 -3.42852086e-01 2.15446532e-01 5.19558072e-01 -1.09572732e+00 1.75670695e+00 -7.25157142e-01 -1.60206199e+00 6.84972942e-01 -2.61713117e-01 -2.59716719e-01 7.80424774e-01 -6.79528594e-01 8.92131031e-02 -4.40272272e-01 -7.29543269e-02 5.91039002e-01 6.44957125e-01 -7.47565866e-01 -1.04414213e+00 -5.40523231e-01 1.82256669e-01 6.30854070e-01 -5.98867416e-01 2.50686914e-01 -6.07687831e-01 -4.39124912e-01 -1.65944546e-01 -9.51533914e-01 -4.69513357e-01 -2.76686270e-02 -1.49769813e-01 -4.00079280e-01 2.23741695e-01 -3.50213438e-01 1.48591661e+00 -1.56349981e+00 4.76405293e-01 3.11538756e-01 3.12768370e-01 7.49044344e-02 -3.95993471e-01 2.63078004e-01 3.26142401e-01 5.95291220e-02 -1.98924132e-02 -3.14886063e-01 1.94397479e-01 -1.65952802e-01 -1.03586987e-01 2.31453523e-01 2.78234005e-01 1.21666133e+00 -1.21252811e+00 -2.88964033e-01 -1.59687385e-01 3.91414195e-01 -7.75175035e-01 6.42269626e-02 -3.69092643e-01 9.28379223e-02 -5.46449542e-01 7.47726858e-01 2.47294098e-01 -4.30514634e-01 -5.57424724e-02 4.14486453e-02 -7.21387044e-02 4.72345501e-01 -9.44939613e-01 1.76354110e+00 -9.89129841e-01 5.61311364e-01 1.06363460e-01 -8.52698207e-01 5.75833201e-01 -3.18672657e-01 3.16779733e-01 -6.18344009e-01 -1.87445506e-01 5.93552947e-01 -1.81761030e-02 5.48953265e-02 7.70600617e-01 1.31366670e-03 -1.56840250e-01 4.40497160e-01 2.68048663e-02 -2.49757424e-01 4.74506348e-01 9.04239491e-02 1.31369328e+00 1.20607063e-01 -1.62161455e-01 -2.93825865e-01 3.51440907e-01 1.70731053e-01 1.43271893e-01 1.04011166e+00 3.44666213e-01 3.14675003e-01 4.14398462e-01 -4.98871535e-01 -8.23428512e-01 -7.80673385e-01 -1.24227606e-01 2.14234471e+00 -2.87630726e-02 -5.30294240e-01 -2.31419191e-01 -7.75768459e-01 3.56338769e-01 1.23678051e-01 -4.10137862e-01 -2.36189783e-01 -6.84457719e-01 -9.99845088e-01 4.34882343e-01 5.85167468e-01 2.11602435e-01 -7.87981391e-01 -6.25443220e-01 3.75057966e-01 1.36942104e-01 -4.25969362e-01 -6.61650240e-01 7.59271860e-01 -1.18795455e+00 -1.04527068e+00 -8.13283265e-01 -7.82519221e-01 6.93558633e-01 1.03511825e-01 1.76821184e+00 2.43006036e-01 -2.86887497e-01 -2.98912767e-02 6.76958403e-03 -3.29425663e-01 -5.29028662e-02 6.72830939e-01 -1.05390437e-01 -3.19071233e-01 -3.46844569e-02 -2.67815113e-01 -9.56213713e-01 5.36667168e-01 -6.22745335e-01 -8.29646066e-02 9.61670637e-01 1.16166675e+00 6.33987844e-01 -2.53956094e-02 4.07134950e-01 -1.25568378e+00 1.17894721e+00 -1.51603371e-01 -1.01539075e+00 5.94152451e-01 -1.21593595e+00 7.04604208e-01 5.94776392e-01 -3.33421618e-01 -7.19069064e-01 1.54944777e-01 -2.00181510e-02 -2.09965721e-01 5.35978556e-01 6.88798845e-01 2.49081880e-01 -2.68227339e-01 6.14520371e-01 -2.39004239e-01 -2.04044372e-01 -6.41440690e-01 7.10210681e-01 3.73058140e-01 3.41973215e-01 -6.60558105e-01 6.77444100e-01 1.68548957e-01 6.93697333e-02 -1.64309055e-01 -1.04326713e+00 -6.03411555e-01 -4.64588135e-01 6.51585460e-02 1.41993031e-01 -9.72992063e-01 -5.94322264e-01 3.73315305e-01 -7.59970069e-01 -5.30545652e-01 -1.32037073e-01 1.80094540e-01 -4.50595975e-01 -1.67649835e-01 -7.04541683e-01 -3.86598289e-01 -6.54247463e-01 -1.26982331e+00 1.28703690e+00 2.73962498e-01 -5.47448285e-02 -9.26998734e-01 2.60618746e-01 2.57577151e-01 6.58214450e-01 -2.23127633e-01 1.03729379e+00 -4.49364543e-01 -6.17654264e-01 -3.60238373e-01 -2.33006641e-01 1.83202311e-01 -2.95622170e-01 1.42926117e-02 -8.25596213e-01 -4.41057920e-01 -6.32626235e-01 -6.81695580e-01 1.51675689e+00 4.80862111e-01 1.19747341e+00 -2.56470293e-01 -4.12329704e-01 7.22512603e-01 1.30930376e+00 -2.74133563e-01 1.44294605e-01 4.28168803e-01 7.36289024e-01 5.02850652e-01 8.12942028e-01 6.87346533e-02 4.21447694e-01 9.24239576e-01 3.03173095e-01 -3.39287788e-01 -3.83724235e-02 -2.71470517e-01 4.50406879e-01 5.87655723e-01 9.85156521e-02 6.98576644e-02 -8.60777855e-01 3.50757837e-01 -2.14585137e+00 -5.75188398e-01 4.24191542e-02 2.42322636e+00 1.08488691e+00 2.70520478e-01 3.05814773e-01 -1.47124603e-01 5.12287259e-01 1.00797422e-01 -9.23640132e-01 -5.56644559e-01 1.12114109e-01 3.67495835e-01 6.28073990e-01 7.33575940e-01 -1.21818793e+00 1.21289313e+00 7.35149860e+00 9.57491279e-01 -1.12937737e+00 -5.10150827e-02 7.83673227e-01 -6.26014352e-01 -4.08752978e-01 1.06300309e-01 -8.91290843e-01 1.81760579e-01 8.55970919e-01 -1.41634151e-01 7.85520196e-01 1.09287119e+00 -1.22251652e-01 1.37619808e-01 -1.48525834e+00 8.50443006e-01 -3.19208324e-01 -1.26655209e+00 -1.95784047e-01 1.48257822e-01 8.67005944e-01 4.43295807e-01 3.79981160e-01 6.34471953e-01 8.54652643e-01 -1.23428798e+00 6.12358034e-01 1.99912459e-01 7.50764310e-01 -8.91677141e-01 6.30937219e-01 2.28170171e-01 -1.21576333e+00 -5.16127706e-01 -4.31792289e-01 7.57354796e-02 -4.16085906e-02 8.62134457e-01 -5.77097595e-01 2.31524646e-01 6.92237079e-01 6.18342221e-01 -8.92787099e-01 1.28500640e+00 -2.83454061e-01 3.36366951e-01 -4.64689285e-01 -3.17465812e-01 3.25995266e-01 -6.33318722e-02 1.89708218e-01 1.24995112e+00 5.94888907e-03 -4.70306247e-01 1.28228009e-01 5.23457050e-01 -4.15154934e-01 2.60646939e-02 -2.29353651e-01 -5.60532063e-02 3.11550826e-01 1.40917420e+00 -4.49563205e-01 -2.72247493e-01 -1.20230243e-01 6.99971259e-01 8.89686048e-01 1.30907267e-01 -7.24360168e-01 -3.36498410e-01 4.97498751e-01 7.91234896e-02 2.97366858e-01 -3.43704671e-02 -3.21265131e-01 -1.04624319e+00 5.83208762e-02 -9.80877876e-01 7.56313801e-01 -2.86621332e-01 -1.20850837e+00 5.76164722e-01 1.47436308e-02 -9.82629895e-01 -2.93363035e-01 -6.03676498e-01 -4.05515611e-01 7.71715522e-01 -1.97981775e+00 -1.12330282e+00 -2.10817158e-01 3.03837210e-01 6.33438587e-01 -9.54868793e-02 6.87246501e-01 2.98440814e-01 -4.35968578e-01 9.31842327e-01 5.38238704e-01 -4.54884410e-01 8.00935924e-01 -1.54710400e+00 4.25471812e-01 4.66972321e-01 3.40653747e-01 7.13586807e-01 4.70415473e-01 -2.02484667e-01 -1.68074131e+00 -8.78702521e-01 8.63694966e-01 -6.09986246e-01 7.49916971e-01 -2.79741138e-01 -5.64614415e-01 1.55172467e-01 -1.15217291e-01 -9.32014957e-02 4.56020027e-01 9.81560886e-01 -5.37201166e-01 -5.65222383e-01 -6.34896398e-01 4.96880710e-01 1.20134592e+00 -4.96449560e-01 -2.11178288e-01 4.75596428e-01 5.26847422e-01 -5.07968307e-01 -6.72442257e-01 6.28766537e-01 9.93222773e-01 -9.04948950e-01 1.15077937e+00 -8.07256222e-01 5.60802579e-01 -1.15909785e-01 1.55204087e-01 -1.54083896e+00 -4.80552495e-01 -7.80905485e-01 -4.72898960e-01 8.35811079e-01 8.76022100e-01 -4.90802020e-01 9.61194336e-01 5.08882046e-01 -2.36911867e-02 -1.34661114e+00 -6.39022827e-01 -6.07264400e-01 2.42140725e-01 -1.37094483e-01 4.04559672e-01 3.96496266e-01 -1.83213651e-01 7.53313005e-01 -3.32992077e-01 -1.60688207e-01 5.54813206e-01 3.67789686e-01 6.68536603e-01 -1.51900232e+00 -4.69715893e-01 -1.19901383e+00 1.19307511e-01 -1.40396667e+00 3.55238579e-02 -1.13894999e+00 1.00901194e-01 -1.65906072e+00 3.85322779e-01 -8.68785322e-01 -8.75021636e-01 4.77388412e-01 -5.33889532e-01 6.75022230e-02 1.54899508e-02 3.88669908e-01 -9.34225798e-01 5.27572155e-01 1.13957930e+00 -2.51418442e-01 -4.22417730e-01 3.38162273e-01 -1.03788567e+00 4.87427235e-01 5.46696007e-01 -5.93284130e-01 -5.70155144e-01 -7.87360370e-01 6.79393411e-01 -1.36810206e-02 -7.12815374e-02 -5.83957613e-01 4.99749243e-01 -8.55906755e-02 3.76872420e-01 -5.64255714e-01 2.09324300e-01 -6.25994921e-01 -3.08967948e-01 4.88322139e-01 -8.38500559e-01 4.88703310e-01 -2.54992694e-02 6.60383582e-01 -2.92654335e-01 3.17902453e-02 6.50418460e-01 -1.97059527e-01 -5.77061057e-01 3.61703008e-01 2.78038770e-01 1.82191492e-03 5.31546235e-01 2.48992592e-02 -3.61605018e-01 -2.64241964e-01 -3.05975109e-01 8.80339026e-01 3.50865811e-01 5.13330638e-01 3.79031479e-01 -1.11397171e+00 -8.18921387e-01 -1.52260642e-02 7.58020133e-02 5.42888045e-02 -2.28177235e-01 6.74354911e-01 -4.87548053e-01 5.30707657e-01 1.40059233e-01 -5.32489955e-01 -1.45639849e+00 1.53997898e-01 3.83582711e-01 -1.20981097e+00 -1.78625733e-02 1.37632596e+00 -5.38684390e-02 -6.97579741e-01 6.51842058e-01 -1.24369904e-01 1.31378416e-02 2.10568115e-01 2.57393360e-01 4.94959652e-01 3.70301306e-01 -1.19335586e-02 -5.28891206e-01 7.52102017e-01 -5.09428024e-01 -2.30108157e-01 1.53603256e+00 2.49395996e-01 -1.80467144e-01 1.18125722e-01 1.23121238e+00 -4.37037647e-02 -1.22842014e+00 -1.69882894e-01 4.29460287e-01 -5.07404685e-01 3.93068373e-01 -1.26035738e+00 -1.06294823e+00 7.21508205e-01 5.40704370e-01 -1.65798649e-01 1.18879569e+00 7.79807270e-02 5.80874562e-01 7.95544803e-01 3.96099746e-01 -1.34141219e+00 2.80092090e-01 6.87517703e-01 8.52113068e-01 -1.36807251e+00 3.31698745e-01 2.52213180e-02 -4.68127519e-01 9.83008027e-01 5.11595428e-01 -1.31028071e-01 6.21378541e-01 3.30390893e-02 -1.10619463e-01 -3.13570976e-01 -1.41155982e+00 -2.63099313e-01 6.93134964e-01 -3.28132287e-02 6.65580094e-01 1.37326300e-01 -3.32507491e-01 2.18840480e-01 -3.46485376e-01 -1.33542657e-01 -1.46603212e-01 1.08498478e+00 -3.82777482e-01 -1.53169441e+00 -5.19824773e-02 8.69624436e-01 -4.87399876e-01 -3.81540835e-01 -5.97584248e-01 4.62079674e-01 -2.73622274e-01 8.69479001e-01 -2.33055949e-01 -5.67416251e-01 4.97739315e-01 5.34702204e-02 5.89699447e-01 -6.46167815e-01 -1.06754827e+00 -6.67496026e-02 2.40056276e-01 -6.84711277e-01 -7.52020488e-03 -3.57975334e-01 -8.60388756e-01 -1.59266382e-01 -6.95703745e-01 2.45435879e-01 9.07734215e-01 5.30875266e-01 5.99484324e-01 5.54912686e-01 8.64218891e-01 -8.00020695e-01 -1.09014082e+00 -7.16363370e-01 -1.62557125e-01 6.52265251e-02 5.11692643e-01 -5.94500661e-01 -2.57708162e-01 -5.75588167e-01]
[10.179295539855957, 5.211532115936279]
7ec74284-fdd3-4352-918a-377cba89e0a5
towards-improving-faithfulness-in-abstractive
2210.01877
null
https://arxiv.org/abs/2210.01877v1
https://arxiv.org/pdf/2210.01877v1.pdf
Towards Improving Faithfulness in Abstractive Summarization
Despite the success achieved in neural abstractive summarization based on pre-trained language models, one unresolved issue is that the generated summaries are not always faithful to the input document. There are two possible causes of the unfaithfulness problem: (1) the summarization model fails to understand or capture the gist of the input text, and (2) the model over-relies on the language model to generate fluent but inadequate words. In this work, we propose a Faithfulness Enhanced Summarization model (FES), which is designed for addressing these two problems and improving faithfulness in abstractive summarization. For the first problem, we propose to use question-answering (QA) to examine whether the encoder fully grasps the input document and can answer the questions on the key information in the input. The QA attention on the proper input words can also be used to stipulate how the decoder should attend to the source. For the second problem, we introduce a max-margin loss defined on the difference between the language and the summarization model, aiming to prevent the overconfidence of the language model. Extensive experiments on two benchmark summarization datasets, CNN/DM and XSum, demonstrate that our model significantly outperforms strong baselines. The evaluation of factual consistency also shows that our model generates more faithful summaries than baselines.
['Xiangliang Zhang', 'Xin Gao', 'Mingzhe Li', 'Xiuying Chen']
2022-10-04
null
null
null
null
['abstractive-text-summarization']
['natural-language-processing']
[ 3.98224145e-01 6.59087598e-01 -1.45072207e-01 -1.94756433e-01 -9.38655853e-01 -4.54155952e-01 6.96235597e-01 3.03321272e-01 -1.59026667e-01 8.54101360e-01 9.82537031e-01 -2.58319288e-01 2.65013516e-01 -7.66421258e-01 -8.26415420e-01 -2.76756883e-01 6.37784839e-01 3.56880605e-01 2.29207858e-01 -3.69760185e-01 6.59041166e-01 -1.28760040e-01 -1.03066444e+00 5.66599607e-01 1.43118989e+00 6.15426779e-01 2.85382688e-01 7.10446656e-01 -2.97604412e-01 1.13487077e+00 -9.99501348e-01 -5.57646453e-01 -2.76524425e-01 -1.15255141e+00 -1.17048752e+00 -4.10021767e-02 7.14880943e-01 -5.86671948e-01 -3.71480823e-01 1.23405397e+00 3.94456655e-01 5.22640496e-02 7.76481092e-01 -8.07335913e-01 -7.75611937e-01 1.07972419e+00 -3.71965080e-01 3.61858457e-01 3.94936234e-01 2.23359853e-01 1.17821157e+00 -6.58803523e-01 4.78287011e-01 1.23843312e+00 3.75115037e-01 7.63705850e-01 -8.10536563e-01 -1.63711935e-01 1.28775641e-01 2.31467068e-01 -8.76055717e-01 -7.39738047e-01 6.02609813e-01 -1.25668287e-01 1.07180619e+00 3.83415133e-01 4.67284441e-01 1.13965952e+00 5.28379917e-01 1.05191958e+00 4.03143227e-01 -3.27548802e-01 2.38229945e-01 6.46591187e-02 4.84814554e-01 5.29725134e-01 4.38226938e-01 -2.76995033e-01 -5.76114297e-01 -1.71587113e-02 3.54767859e-01 -3.51798177e-01 -6.07241690e-01 3.12137336e-01 -1.07601333e+00 8.08036447e-01 3.97395253e-01 4.46701795e-01 -7.39836633e-01 4.98581715e-02 6.38548791e-01 5.11902198e-02 4.64308739e-01 7.84624338e-01 -8.47495794e-02 -6.89317062e-02 -1.42688298e+00 2.40107164e-01 8.40077162e-01 8.41235042e-01 5.02168119e-01 3.30018997e-01 -7.82707453e-01 4.93259430e-01 2.05261424e-01 4.12992507e-01 7.13175893e-01 -9.04852688e-01 9.32935297e-01 6.18812501e-01 -1.17857531e-02 -9.77437198e-01 3.80755141e-02 -4.37256545e-01 -1.08821976e+00 -5.43624997e-01 7.27369413e-02 -7.86319301e-02 -6.79514468e-01 1.76990342e+00 -3.25607300e-01 -8.14179257e-02 4.78578627e-01 7.74460077e-01 1.29747796e+00 1.10729170e+00 -6.63162172e-02 -4.30936515e-01 1.21524692e+00 -1.27931023e+00 -1.03878999e+00 -7.31405795e-01 4.15749013e-01 -6.02512717e-01 1.07536745e+00 1.33499065e-02 -1.58142936e+00 -5.91690838e-01 -1.26657057e+00 -3.31504077e-01 1.48789734e-01 2.11586028e-01 4.02235091e-02 1.64611340e-01 -1.14454162e+00 6.84219122e-01 -6.70161963e-01 -3.74026567e-01 3.23214471e-01 -1.57556638e-01 -8.66306424e-02 -4.09790613e-02 -1.37228370e+00 1.08363068e+00 6.88978732e-01 1.25733674e-01 -7.16527939e-01 -5.05599558e-01 -1.00283587e+00 5.37857294e-01 2.74503797e-01 -1.24983859e+00 1.42754138e+00 -1.11164916e+00 -1.52161849e+00 5.13660192e-01 -4.61288452e-01 -7.17358291e-01 3.91607970e-01 -4.22740489e-01 3.34798619e-02 4.31964040e-01 3.57012391e-01 5.77946603e-01 6.13349020e-01 -1.19217992e+00 -4.70825136e-01 -3.08781087e-01 -1.24373659e-01 2.56158739e-01 -1.99583814e-01 -2.50529706e-01 -4.61563408e-01 -7.54816353e-01 5.51460236e-02 -5.42825639e-01 6.60521835e-02 -6.90579116e-01 -1.14028311e+00 -4.51324582e-01 4.15012211e-01 -1.07004178e+00 1.63592100e+00 -1.85721993e+00 3.78513604e-01 -2.93898463e-01 4.82464246e-02 5.27839184e-01 -3.06116581e-01 7.13925958e-01 1.17637813e-01 2.89330721e-01 -5.37015200e-01 -4.77284819e-01 -1.02723539e-01 1.53883740e-01 -1.13219225e+00 -1.52696006e-03 4.13671076e-01 1.12759113e+00 -8.82155240e-01 -5.06181300e-01 -2.61014998e-01 9.42538902e-02 -4.86943841e-01 4.69450772e-01 -4.22635645e-01 2.62233287e-01 -3.18650573e-01 3.44230272e-02 4.34730977e-01 -1.80676028e-01 -4.18807864e-02 -3.03570151e-01 2.21252274e-02 1.00168121e+00 -7.02515900e-01 1.72574878e+00 -2.12996230e-01 6.12942755e-01 -3.44718516e-01 -9.00648952e-01 6.12586856e-01 4.17775095e-01 -2.16324136e-01 -8.32723856e-01 3.93751524e-02 2.22969726e-01 -9.04769450e-02 -5.88041842e-01 9.94399428e-01 -2.73379713e-01 -1.38356969e-01 7.91746378e-01 1.23645432e-01 -3.26504260e-01 3.70429188e-01 8.02227736e-01 7.87272453e-01 -1.26994491e-01 4.79422420e-01 -1.75845474e-01 6.15469337e-01 5.80364540e-02 5.43565631e-01 9.79508817e-01 5.06002754e-02 8.49957466e-01 8.81318569e-01 -3.02015767e-02 -8.24289739e-01 -9.35930252e-01 5.07866025e-01 7.35404670e-01 1.80522740e-01 -6.24809146e-01 -9.98839855e-01 -7.87828922e-01 -4.77354348e-01 1.65765047e+00 -5.11827409e-01 -6.47692680e-01 -7.87704349e-01 -3.03880304e-01 7.33356476e-01 6.82909846e-01 6.38173044e-01 -1.10818863e+00 -7.00565100e-01 2.38131627e-01 -8.43913615e-01 -9.67940807e-01 -7.77256250e-01 -2.15335384e-01 -1.07434881e+00 -8.62627208e-01 -6.04540288e-01 -5.72163761e-01 6.08558416e-01 2.66109586e-01 1.17775452e+00 2.65533835e-01 5.65743685e-01 2.05168426e-01 -3.24731499e-01 -4.50140208e-01 -9.54066694e-01 3.57508242e-01 -2.47340664e-01 -2.26182446e-01 1.75876357e-02 -2.18722671e-01 -4.75100785e-01 -2.10061789e-01 -1.17871368e+00 3.63311261e-01 6.17877245e-01 6.85964406e-01 2.85121918e-01 -1.82898581e-01 9.45765257e-01 -9.38965321e-01 1.08759034e+00 -4.25521672e-01 3.36240083e-02 4.73546058e-01 -4.66004759e-01 4.23891842e-01 7.62596607e-01 -1.30570024e-01 -1.05895412e+00 -5.76363325e-01 -4.01699007e-01 3.94864716e-02 1.54037088e-01 7.11217046e-01 -2.50659585e-01 8.77352059e-01 5.89699805e-01 6.97850823e-01 3.80899981e-02 -1.84265703e-01 3.99689704e-01 5.93720078e-01 8.32411110e-01 -4.22343701e-01 4.84311283e-01 2.55723178e-01 -5.27949035e-01 -7.75047541e-01 -1.24555671e+00 -1.43749386e-01 -2.22769856e-01 2.84129307e-02 7.84878850e-01 -6.99554205e-01 -1.70925692e-01 2.10428059e-01 -1.83424151e+00 -1.34606689e-01 -4.48930860e-01 1.64468095e-01 -6.15777850e-01 8.04180086e-01 -6.37326181e-01 -6.29937112e-01 -1.09799612e+00 -9.74234641e-01 9.56205130e-01 4.11540985e-01 -6.68588161e-01 -9.22806740e-01 1.36787415e-01 3.70573491e-01 5.38041651e-01 -1.94837339e-02 1.28424919e+00 -9.31277275e-01 -3.23638201e-01 -1.39514640e-01 -1.79701328e-01 6.08162463e-01 4.84807231e-02 1.76097989e-01 -7.94084370e-01 -1.21791996e-01 3.69901270e-01 -3.85395974e-01 1.26375353e+00 3.87692571e-01 8.77732038e-01 -9.80049014e-01 5.05686961e-02 4.16886136e-02 1.06363893e+00 -1.52497336e-01 1.09283137e+00 3.23783271e-02 4.78851050e-01 7.45075285e-01 2.96308756e-01 2.07999498e-01 7.02694297e-01 2.88982898e-01 3.83865476e-01 1.04767859e-01 -3.39511871e-01 -8.09072793e-01 6.46450877e-01 1.26976418e+00 2.58819580e-01 -6.80380762e-01 -6.18271172e-01 7.05890000e-01 -2.01288724e+00 -1.22626519e+00 -3.18655431e-01 2.06554508e+00 9.98769462e-01 2.72476882e-01 -1.27585262e-01 1.83352940e-02 5.42655766e-01 4.13384587e-01 -5.33230305e-01 -6.36404157e-01 -2.46205926e-01 -9.82503816e-02 -1.19966753e-01 7.20160425e-01 -5.66511214e-01 1.04634380e+00 5.64490795e+00 6.55178845e-01 -1.18160391e+00 -1.18169263e-01 4.80079859e-01 -1.32133523e-02 -5.86723328e-01 5.95996007e-02 -6.83540523e-01 5.19847214e-01 9.04501557e-01 -5.92455626e-01 4.48553031e-03 3.30249727e-01 2.67767429e-01 -2.35974148e-01 -1.29046369e+00 3.90603453e-01 5.60740173e-01 -1.41491067e+00 7.88768530e-01 -3.32823724e-01 6.89332843e-01 -1.46522656e-01 -5.63935749e-02 5.08244693e-01 -4.79077287e-02 -9.23943460e-01 1.05595648e+00 6.85278237e-01 4.59047705e-01 -6.34267569e-01 8.99884820e-01 9.10846293e-01 -5.60255527e-01 2.68817484e-01 -4.69095647e-01 -1.66834872e-02 3.06637794e-01 4.77633417e-01 -8.79978955e-01 6.48376644e-01 9.84314978e-02 6.43203497e-01 -6.49185896e-01 7.82300174e-01 -8.12954545e-01 7.81941950e-01 1.13837078e-01 -1.11980408e-01 4.45827037e-01 5.39653488e-02 8.34328711e-01 1.25053406e+00 3.48961294e-01 1.29607558e-01 -3.39314401e-01 1.23639667e+00 -2.69483745e-01 -2.70820558e-02 -4.45058286e-01 -3.17429096e-01 2.88215876e-01 8.64988089e-01 -2.67369658e-01 -5.47790289e-01 -1.35188457e-02 1.28069794e+00 3.42865229e-01 4.05728191e-01 -6.09930515e-01 -4.80676234e-01 2.42742561e-02 -1.49909584e-02 1.62967056e-01 3.99964228e-02 -4.93085265e-01 -1.34521961e+00 1.46060660e-01 -9.61041152e-01 3.82685721e-01 -9.05417323e-01 -9.93467450e-01 6.89025521e-01 -1.86616912e-01 -6.98804736e-01 -5.21541357e-01 3.48658301e-02 -1.16781747e+00 9.23475683e-01 -1.69196415e+00 -9.43022788e-01 2.99671013e-02 1.42333224e-01 1.00072646e+00 4.77348194e-02 6.42058551e-01 -1.90251559e-01 -7.00841427e-01 3.55818868e-01 -3.74370545e-01 -2.91056652e-02 6.18211329e-01 -1.22551501e+00 5.33929884e-01 1.26582754e+00 4.88805398e-02 8.75349164e-01 1.04824817e+00 -7.41052091e-01 -1.14188600e+00 -1.11358845e+00 1.54514575e+00 -4.63424116e-01 2.72513419e-01 1.42235145e-01 -1.37946177e+00 6.68287873e-01 8.35579813e-01 -8.96928668e-01 6.62324905e-01 -2.35439092e-01 -3.24095994e-01 7.70376697e-02 -7.68479764e-01 6.13680243e-01 4.81629997e-01 -3.55468988e-01 -1.40439284e+00 2.04804614e-01 9.10653234e-01 -2.05391258e-01 -1.76105320e-01 3.45693350e-01 6.00166731e-02 -9.34908569e-01 5.36581159e-01 -8.08178663e-01 1.23146379e+00 -1.28703952e-01 -7.61903077e-02 -1.61972463e+00 -2.37149000e-01 -4.81134892e-01 -3.34971488e-01 1.43957424e+00 4.01032925e-01 -3.24923962e-01 2.98240423e-01 4.68543351e-01 -4.56003100e-01 -5.69915891e-01 -9.08306241e-01 -4.59332049e-01 3.83719355e-01 7.00359745e-03 4.55037951e-01 5.78933120e-01 2.31550589e-01 1.11592352e+00 -2.78373927e-01 1.77360196e-02 3.10492933e-01 2.31155500e-01 6.50974691e-01 -8.23031366e-01 -1.35416701e-01 -7.92766094e-01 3.71281326e-01 -1.40342009e+00 4.36402172e-01 -9.17332351e-01 4.28894371e-01 -2.26101398e+00 5.18614471e-01 4.08908337e-01 1.60926610e-01 3.80378515e-01 -6.72171712e-01 -2.94650018e-01 2.63836443e-01 3.10148180e-01 -7.82777905e-01 9.60050821e-01 1.21919632e+00 -2.43093446e-01 -7.70435035e-02 9.95878354e-02 -1.33047223e+00 7.21382141e-01 7.00634420e-01 -3.37639004e-01 -4.19511646e-01 -8.77297521e-01 4.24954474e-01 3.69375944e-01 3.55039090e-01 -7.31911063e-01 5.45372844e-01 -3.55077460e-02 -1.58900693e-01 -8.33458304e-01 -1.10060416e-01 -2.33060837e-01 -3.72507900e-01 5.95604002e-01 -6.87063336e-01 6.49420917e-02 2.75428563e-01 4.96470302e-01 -4.14911896e-01 -5.74471772e-01 7.70480096e-01 -2.25320607e-01 -2.41811618e-01 -1.25560954e-01 -4.92559582e-01 5.31636715e-01 4.31135893e-01 3.04482225e-02 -7.49495566e-01 -6.76654398e-01 -9.68376547e-02 4.64249969e-01 2.97457337e-01 3.03884566e-01 8.25178266e-01 -1.06367600e+00 -1.18239379e+00 -5.46853803e-02 3.01947445e-03 5.77845499e-02 3.07995081e-01 6.21712744e-01 -3.90475482e-01 7.47395039e-01 7.94338435e-02 -2.02767536e-01 -9.20940697e-01 2.91743338e-01 3.76422644e-01 -5.12232721e-01 -5.11550069e-01 6.39532387e-01 2.14386180e-01 -1.41985431e-01 2.29142427e-01 -4.31849808e-01 -3.28854382e-01 1.47059873e-01 8.28771710e-01 3.98555785e-01 8.57221410e-02 -5.94102025e-01 -1.63411573e-01 1.56232893e-01 -3.59884590e-01 -1.38280928e-01 1.17023504e+00 -2.07601517e-01 -3.82323384e-01 4.55605090e-01 9.58461881e-01 1.65178534e-02 -9.96743917e-01 -2.82390922e-01 6.23142608e-02 6.83101788e-02 -3.88212800e-02 -8.79864275e-01 -6.44545197e-01 1.24075770e+00 -4.74138528e-01 3.93174380e-01 9.20277953e-01 1.35342954e-02 1.22973955e+00 4.09582347e-01 -2.86741883e-01 -1.01544261e+00 4.24578369e-01 9.61514533e-01 1.36108446e+00 -9.50447261e-01 -4.81953733e-02 -1.19853236e-01 -1.02068222e+00 1.23407149e+00 6.68448985e-01 -3.73489745e-02 -2.27451846e-01 -2.98302382e-01 -1.69512346e-01 -2.23651320e-01 -9.28187132e-01 1.32443264e-01 3.86733025e-01 1.60800025e-01 4.47304338e-01 -2.34176293e-01 -4.48308289e-01 1.01678085e+00 -5.29119790e-01 -1.09923139e-01 8.55825126e-01 4.98101652e-01 -8.28110456e-01 -4.76536125e-01 -1.25393689e-01 2.98028082e-01 -4.51952010e-01 -2.60923415e-01 -8.54673386e-01 2.76970893e-01 -3.82310867e-01 1.23392761e+00 4.56843376e-02 -4.58144806e-02 5.32962918e-01 1.24260269e-01 3.57217520e-01 -9.21106219e-01 -7.28345931e-01 -1.09228425e-01 1.18905425e-01 -2.74395436e-01 -1.88832432e-01 -3.71640950e-01 -1.38702190e+00 -2.65236109e-01 -2.32693344e-01 3.75659227e-01 3.89728636e-01 1.12608969e+00 4.37447518e-01 7.27467537e-01 4.11981106e-01 -4.12210852e-01 -1.05285013e+00 -1.22269356e+00 -1.51925221e-01 3.37906718e-01 4.75632429e-01 1.82848856e-01 -3.68361026e-01 -1.05065862e-02]
[12.392855644226074, 9.351301193237305]
6b32edac-c9c1-4df1-b17d-bc94a4f74c7b
pi-qt-opt-predictive-information-improves
2210.08217
null
https://arxiv.org/abs/2210.08217v2
https://arxiv.org/pdf/2210.08217v2.pdf
PI-QT-Opt: Predictive Information Improves Multi-Task Robotic Reinforcement Learning at Scale
The predictive information, the mutual information between the past and future, has been shown to be a useful representation learning auxiliary loss for training reinforcement learning agents, as the ability to model what will happen next is critical to success on many control tasks. While existing studies are largely restricted to training specialist agents on single-task settings in simulation, in this work, we study modeling the predictive information for robotic agents and its importance for general-purpose agents that are trained to master a large repertoire of diverse skills from large amounts of data. Specifically, we introduce Predictive Information QT-Opt (PI-QT-Opt), a QT-Opt agent augmented with an auxiliary loss that learns representations of the predictive information to solve up to 297 vision-based robot manipulation tasks in simulation and the real world with a single set of parameters. We demonstrate that modeling the predictive information significantly improves success rates on the training tasks and leads to better zero-shot transfer to unseen novel tasks. Finally, we evaluate PI-QT-Opt on real robots, achieving substantial and consistent improvement over QT-Opt in multiple experimental settings of varying environments, skills, and multi-task configurations.
['Yao Lu', 'Ian Fischer', 'Paul Wohlhart', 'Adrian Li', 'Ted Xiao', 'Kuang-Huei Lee']
2022-10-15
null
null
null
null
['robot-manipulation']
['robots']
[ 2.78850347e-01 -6.98044151e-02 -3.85574624e-02 -1.82111651e-01 -6.44462049e-01 -4.47599262e-01 5.87407649e-01 -1.79122351e-02 -8.22217941e-01 9.74816680e-01 -2.72578776e-01 -4.87218983e-02 -5.87168396e-01 -2.81020850e-01 -1.06748915e+00 -5.91522634e-01 -7.14542389e-01 1.01323771e+00 2.02090278e-01 -4.01174128e-01 1.32467449e-01 3.76095802e-01 -1.62401378e+00 -1.74863771e-01 8.53481412e-01 7.29892373e-01 1.04643488e+00 7.23211825e-01 7.16506600e-01 9.54397202e-01 -4.71085399e-01 7.41454288e-02 5.32054245e-01 -1.49004459e-01 -8.23315501e-01 -1.76569834e-01 9.30534154e-02 -4.82540309e-01 -6.18891537e-01 8.13074470e-01 4.15353954e-01 5.82424104e-01 8.13057065e-01 -1.58653438e+00 -4.50925052e-01 5.15861869e-01 -2.05818161e-01 3.02143365e-01 1.60728052e-01 8.34137142e-01 8.24270248e-01 -4.80236888e-01 5.52660584e-01 1.43913770e+00 6.31087542e-01 8.10214818e-01 -1.48070228e+00 -5.97324610e-01 2.81697005e-01 3.97650838e-01 -7.79431641e-01 -6.36974573e-02 2.42835388e-01 -5.19789040e-01 1.32688355e+00 -6.02059782e-01 6.80559933e-01 1.36256921e+00 6.40578747e-01 1.02445090e+00 9.63170290e-01 -6.23890795e-02 1.48745120e-01 -9.57877487e-02 -1.29367322e-01 8.81648600e-01 3.41037638e-04 9.39425826e-01 -5.79789340e-01 1.66677386e-01 7.72131562e-01 1.79036856e-01 -2.82601625e-01 -7.99959600e-01 -1.35610151e+00 7.35814452e-01 5.99967837e-01 -2.21180633e-01 -6.07620835e-01 4.88012612e-01 4.73575264e-01 8.92023683e-01 -6.90310523e-02 1.13840258e+00 -7.25684941e-01 -3.64381194e-01 1.36591911e-01 5.42617977e-01 9.23463345e-01 1.29483819e+00 7.85556495e-01 2.73755252e-01 -3.24713737e-01 7.88237333e-01 -3.28277126e-02 6.78444386e-01 5.50257742e-01 -1.45601368e+00 5.73502600e-01 1.84075162e-01 5.11409283e-01 -4.13685352e-01 -5.88322222e-01 -4.75564986e-01 -2.77233362e-01 6.99838340e-01 6.83144107e-02 -5.22551119e-01 -1.13796401e+00 2.17212939e+00 -5.22071831e-02 8.82408395e-02 5.52200317e-01 8.93319130e-01 1.19615711e-01 6.46783710e-01 2.82376230e-01 7.89345279e-02 8.23563755e-01 -1.40097880e+00 -3.55041176e-01 -8.19856644e-01 5.43818831e-01 -3.36745083e-01 1.21571755e+00 4.97112989e-01 -9.40497458e-01 -6.86876178e-01 -1.00377250e+00 2.36806810e-01 -1.94083080e-01 -1.23041175e-01 8.16163182e-01 -3.12246352e-01 -1.02488542e+00 1.08890176e+00 -1.07788515e+00 -4.00602669e-01 5.22685111e-01 4.87901628e-01 -3.34318161e-01 -3.50644857e-01 -9.72238123e-01 1.51355684e+00 6.17235124e-01 -4.10718173e-01 -1.97206700e+00 -6.61290526e-01 -8.45221221e-01 1.58456743e-01 6.31679237e-01 -8.91164780e-01 1.84533846e+00 -7.65617549e-01 -1.63782632e+00 3.12447876e-01 3.08186918e-01 -7.66708553e-01 3.05117786e-01 -5.95460176e-01 3.05697143e-01 3.63560254e-03 2.93275058e-01 9.35363114e-01 8.47697020e-01 -1.33336842e+00 -8.71563137e-01 -2.49985367e-01 2.59953678e-01 7.41875589e-01 7.67080858e-02 -6.40153170e-01 -1.23592027e-01 -4.12849724e-01 -2.81167477e-01 -1.24576139e+00 -4.25004274e-01 -1.00301519e-01 1.95975304e-01 -4.85126853e-01 7.33087063e-01 -3.48640621e-01 1.50829598e-01 -1.92276037e+00 6.03947759e-01 -2.57341981e-01 -1.10126659e-01 1.82569861e-01 -5.33998549e-01 5.38883984e-01 1.30867422e-01 -5.00223517e-01 -1.20839372e-01 -3.86679649e-01 2.85486996e-01 7.74177790e-01 -3.79897058e-01 4.71631847e-02 2.20087782e-01 9.67708290e-01 -1.28907549e+00 -7.60952681e-02 1.88648239e-01 1.56726107e-01 -5.18351376e-01 4.80635375e-01 -6.83706582e-01 6.13994241e-01 -6.74098670e-01 1.61782265e-01 -8.96079168e-02 -1.78299621e-01 -6.00599162e-02 3.70454580e-01 3.22206765e-01 5.26697859e-02 -4.55281883e-01 1.97870255e+00 -7.32643962e-01 6.61472917e-01 1.71752140e-01 -1.10697758e+00 7.91082323e-01 2.08848357e-01 4.89084482e-01 -9.07234967e-01 7.40319863e-02 -9.57021341e-02 2.74760306e-01 -5.48621833e-01 5.05013525e-01 -1.72461614e-01 -1.06143832e-01 4.08780932e-01 4.76939112e-01 -6.50306225e-01 3.33536088e-01 1.74777582e-01 1.31648874e+00 4.31608170e-01 1.08900834e-02 2.46535125e-03 -1.08727142e-01 5.47483265e-01 5.77852190e-01 1.09329271e+00 -3.34596455e-01 1.06957294e-01 2.32988909e-01 -2.73142755e-01 -1.13730133e+00 -1.20218885e+00 3.46497536e-01 1.45874858e+00 3.62352580e-01 -4.25695591e-02 -2.31940761e-01 -6.31397367e-01 3.06615591e-01 1.08992469e+00 -5.72776914e-01 -7.08573878e-01 -5.42044282e-01 -4.06570137e-01 9.19013247e-02 6.31438971e-01 3.76026571e-01 -1.60109854e+00 -1.15577626e+00 2.89449811e-01 8.62282887e-02 -8.87526214e-01 -4.78284508e-01 7.87320495e-01 -8.56421649e-01 -1.31329215e+00 -5.17754436e-01 -1.07438195e+00 5.04533708e-01 3.01596850e-01 1.29463017e+00 -2.55536973e-01 -4.46651250e-01 1.11031055e+00 -2.53247052e-01 -6.55978441e-01 -5.47527432e-01 -1.68560147e-01 4.52484012e-01 -7.76601136e-01 -2.41741706e-02 -6.25933945e-01 -4.47787613e-01 2.78644443e-01 -4.75905716e-01 1.48000404e-01 9.89576280e-01 1.36058497e+00 3.90182704e-01 1.42282620e-01 6.21799231e-01 -5.41434348e-01 9.33229625e-01 -4.27805156e-01 -6.54950023e-01 1.55256033e-01 -8.21085632e-01 4.13525343e-01 6.68679118e-01 -8.16145599e-01 -1.06985927e+00 1.15474440e-01 2.52402365e-01 -7.13135421e-01 1.73589401e-02 4.43376541e-01 4.97730702e-01 -5.96930422e-02 6.72538221e-01 4.21819419e-01 3.26570749e-01 -1.23732299e-01 2.36889988e-01 2.70490572e-02 5.83902597e-01 -8.67924929e-01 5.35165131e-01 5.16151115e-02 6.46248385e-02 -3.19190741e-01 -8.45557511e-01 -3.30335468e-01 -4.04033422e-01 -8.56024474e-02 5.21363676e-01 -1.00781929e+00 -1.25370860e+00 3.94632995e-01 -9.06183362e-01 -1.34942710e+00 -5.57116747e-01 7.31692255e-01 -1.34943807e+00 -1.23890840e-01 -6.04847550e-01 -7.10659266e-01 -1.86125204e-01 -1.43110847e+00 8.03846896e-01 2.94402540e-01 1.65266633e-01 -9.15453911e-01 1.46560803e-01 -9.72190872e-02 3.85672688e-01 -1.09015837e-01 9.34000671e-01 -5.49010158e-01 -5.96756041e-01 1.81699112e-01 8.46839044e-04 3.69163215e-01 9.14043165e-04 -7.49310493e-01 -5.23596227e-01 -8.45577478e-01 -8.56690481e-02 -1.25785577e+00 9.84529018e-01 2.11089030e-01 8.62055779e-01 -1.32302538e-01 -4.99947697e-01 2.85607219e-01 1.22798061e+00 5.20053685e-01 1.46690384e-01 4.58569705e-01 3.22354466e-01 4.39298838e-01 1.10037684e+00 5.34931421e-01 4.93952811e-01 6.06889427e-01 7.39102602e-01 5.12591660e-01 -4.52474579e-02 -4.91488904e-01 6.44577980e-01 4.52349186e-01 1.74522370e-01 -1.30876258e-01 -8.16300094e-01 6.32750273e-01 -2.06868505e+00 -9.21046495e-01 6.36772394e-01 1.94266558e+00 7.94697523e-01 1.81116581e-01 -1.22287191e-01 -4.80745375e-01 3.28527093e-01 -2.14160740e-01 -1.42325330e+00 -1.26052247e-02 3.10928226e-01 2.71960832e-02 4.60222274e-01 4.13025528e-01 -1.09883857e+00 1.08891165e+00 6.71304703e+00 6.67732000e-01 -8.85753751e-01 5.62058166e-02 2.07303882e-01 -8.56359378e-02 3.61633360e-01 -3.15860838e-01 -6.37766898e-01 9.25154686e-02 8.11462939e-01 -3.65390837e-01 8.99637699e-01 1.19066334e+00 -8.31329376e-02 -4.57424790e-01 -1.53749716e+00 9.64290082e-01 -5.98357469e-02 -7.57256687e-01 -1.33541822e-01 -1.73604563e-01 8.93549025e-01 6.35285079e-01 4.86484855e-01 1.08876288e+00 1.19967008e+00 -1.13833332e+00 6.90107167e-01 5.59637904e-01 4.15514261e-01 -5.47953963e-01 5.87142825e-01 7.59682596e-01 -7.45047987e-01 -6.66279078e-01 -4.93911624e-01 -1.59691676e-01 -3.95551547e-02 -3.65601271e-01 -1.34137309e+00 2.85918921e-01 7.39210129e-01 1.05904377e+00 -1.96689695e-01 1.06252408e+00 -3.65017176e-01 2.22015783e-01 -2.08020985e-01 -1.74891263e-01 5.21714270e-01 -1.76951252e-02 5.53544700e-01 6.36136413e-01 3.54293495e-01 1.44317567e-01 6.69705629e-01 7.12448955e-01 1.62098438e-01 -6.38983250e-01 -9.18347776e-01 -1.16485126e-01 4.62757200e-01 8.19863677e-01 -2.14077592e-01 -6.63065165e-02 -1.16096577e-02 9.50627863e-01 8.34132850e-01 5.65408468e-01 -5.02958536e-01 -1.45161942e-01 8.69458675e-01 -5.65126956e-01 3.82375002e-01 -5.89802146e-01 1.61261126e-01 -6.23784781e-01 -2.74844587e-01 -9.27492619e-01 3.06305051e-01 -9.04482603e-01 -1.38143086e+00 5.47882140e-01 3.74804949e-03 -1.11371875e+00 -6.34166360e-01 -8.99664283e-01 -3.39715242e-01 8.41544151e-01 -1.69467533e+00 -7.56733239e-01 -3.11471760e-01 6.18230760e-01 9.52517033e-01 -6.15114808e-01 9.68121529e-01 -3.69948179e-01 -2.76056111e-01 1.92847878e-01 4.35198635e-01 -2.56847143e-01 7.62502313e-01 -1.37928033e+00 2.41365358e-01 2.37608477e-01 -1.42373875e-01 3.59485835e-01 9.39111412e-01 -5.30088246e-01 -1.70609975e+00 -1.08721447e+00 -1.55969366e-01 -6.01714015e-01 6.76734567e-01 -3.69103923e-02 -7.53095210e-01 1.08068335e+00 1.56931713e-01 -2.41901442e-01 4.04479243e-02 -1.31731480e-03 -1.49602368e-01 -5.18382788e-02 -9.82824802e-01 6.66267931e-01 1.18160152e+00 -3.28484148e-01 -9.36916709e-01 5.77419162e-01 1.00143170e+00 -4.56455499e-01 -8.47578704e-01 3.76161516e-01 4.35591698e-01 -6.18070543e-01 9.58893418e-01 -8.34270418e-01 3.64986330e-01 1.39774103e-02 4.46074791e-02 -2.20049500e+00 -3.81581903e-01 -6.33493900e-01 3.36535275e-02 3.11231971e-01 3.16728204e-01 -6.71249032e-01 7.29889452e-01 3.70800585e-01 -5.15374959e-01 -6.79892182e-01 -7.58718252e-01 -1.06629956e+00 4.12669256e-02 -1.39998764e-01 4.82239686e-02 4.73486453e-01 1.97412729e-01 5.36467075e-01 -4.53895271e-01 1.19721770e-01 5.84579110e-01 8.43768641e-02 9.23965633e-01 -1.24895370e+00 -4.38154936e-01 -3.59786600e-01 -8.20702836e-02 -1.39629042e+00 5.02483428e-01 -7.81275511e-01 8.45328450e-01 -1.60087371e+00 2.30356932e-01 -8.81124020e-01 -2.52406925e-01 5.69292843e-01 -2.26651013e-01 -4.88194197e-01 4.48341399e-01 2.95538902e-01 -1.00630057e+00 1.18265712e+00 1.76122153e+00 -3.48133922e-01 -1.86231405e-01 1.58807069e-01 -3.20279807e-01 5.89691937e-01 7.53033817e-01 -3.83734196e-01 -1.01236165e+00 -8.00458848e-01 -1.48118392e-01 3.25826734e-01 3.26177984e-01 -1.34251928e+00 4.14229959e-01 -3.70513260e-01 2.75979519e-01 -2.05516830e-01 9.67239320e-01 -9.36627328e-01 -2.79619843e-01 8.77464712e-01 -5.87980866e-01 3.23970675e-01 4.84942615e-01 1.09188771e+00 7.97775835e-02 -3.84459466e-01 6.48119688e-01 -4.53579009e-01 -1.19340587e+00 4.64453608e-01 -3.46463501e-01 2.59009302e-01 1.23849809e+00 6.98312521e-02 -4.44325626e-01 -5.52410007e-01 -8.91710222e-01 1.03056538e+00 3.28059405e-01 5.46869934e-01 9.00314689e-01 -9.29518282e-01 -5.89592755e-01 8.85320380e-02 1.03613652e-01 2.41445396e-02 1.11687183e-01 4.96237189e-01 -2.31275812e-01 4.87798452e-01 -7.17282355e-01 -5.28091848e-01 -1.00217116e+00 7.78874993e-01 3.82450134e-01 -4.47812021e-01 -7.79381037e-01 9.00869668e-01 4.21773016e-01 -6.85230196e-01 6.81152582e-01 -1.65019333e-01 -2.59547949e-01 -5.30036628e-01 2.12536991e-01 1.82863474e-01 -4.57362980e-01 -2.26648390e-01 1.02570243e-01 2.06837773e-01 -2.91805655e-01 -2.59022117e-01 1.49477959e+00 -2.87374072e-02 4.51777995e-01 4.90817070e-01 6.14228249e-01 -1.12550437e+00 -2.23999238e+00 -4.54448491e-01 3.91172357e-02 -2.09302053e-01 -1.54789597e-01 -1.11433840e+00 -7.63816953e-01 8.90276909e-01 6.20026052e-01 -2.43725121e-01 7.32374310e-01 9.42808855e-03 4.78868306e-01 1.27158189e+00 1.15352404e+00 -1.11844623e+00 8.52212131e-01 1.16049242e+00 1.21912873e+00 -1.49600971e+00 -1.51065379e-01 1.31103620e-01 -1.07913709e+00 7.90940344e-01 1.13354838e+00 -2.51169950e-01 4.26046848e-01 -3.89242209e-02 -2.34242722e-01 -7.38217384e-02 -1.35521460e+00 -4.20663744e-01 -7.42166042e-02 1.05974412e+00 -3.15157413e-01 -3.84731404e-02 5.16649604e-01 3.24079305e-01 -2.37146333e-01 -4.10591960e-02 3.66783112e-01 1.20223641e+00 -6.59146786e-01 -6.51043355e-01 9.66858864e-02 6.01845622e-01 6.04613461e-02 1.88582659e-01 1.66709479e-02 7.93882191e-01 -2.37432376e-01 8.06672394e-01 8.18948299e-02 -4.42104578e-01 5.30827940e-01 -2.02711299e-03 9.52692509e-01 -8.13433886e-01 -3.76636952e-01 -5.73783457e-01 -3.51350452e-03 -6.83585405e-01 -2.15948120e-01 -7.26979792e-01 -1.30848122e+00 2.35870983e-02 9.39737856e-02 2.82271177e-01 5.54470718e-01 9.16161716e-01 3.98435593e-01 9.42063391e-01 4.53079164e-01 -1.26336408e+00 -1.15163243e+00 -1.23573124e+00 -5.10085464e-01 5.13041019e-01 5.12463868e-01 -1.35677397e+00 -1.18700378e-01 -1.60006844e-02]
[4.393594741821289, 1.066947340965271]
c4b9ae20-ab82-4b66-a704-01e29c17076d
towards-generalized-open-information
2211.15987
null
https://arxiv.org/abs/2211.15987v1
https://arxiv.org/pdf/2211.15987v1.pdf
Towards Generalized Open Information Extraction
Open Information Extraction (OpenIE) facilitates the open-domain discovery of textual facts. However, the prevailing solutions evaluate OpenIE models on in-domain test sets aside from the training corpus, which certainly violates the initial task principle of domain-independence. In this paper, we propose to advance OpenIE towards a more realistic scenario: generalizing over unseen target domains with different data distributions from the source training domains, termed Generalized OpenIE. For this purpose, we first introduce GLOBE, a large-scale human-annotated multi-domain OpenIE benchmark, to examine the robustness of recent OpenIE models to domain shifts, and the relative performance degradation of up to 70% implies the challenges of generalized OpenIE. Then, we propose DragonIE, which explores a minimalist graph expression of textual fact: directed acyclic graph, to improve the OpenIE generalization. Extensive experiments demonstrate that DragonIE beats the previous methods in both in-domain and out-of-domain settings by as much as 6.0% in F1 score absolutely, but there is still ample room for improvement.
['Bin Wang', 'Yongbin Li', 'Jian Sun', 'Tingwen Liu', 'Haiyang Yu', 'Jingyang Li', 'Zhenyu Zhang', 'Bowen Yu']
2022-11-29
null
null
null
null
['open-information-extraction']
['natural-language-processing']
[-6.24814443e-02 3.90861481e-01 -4.77915019e-01 -2.00188577e-01 -7.34458208e-01 -1.12993395e+00 5.92932820e-01 1.22868806e-01 -2.11445153e-01 1.12404180e+00 3.27603258e-02 -3.28443676e-01 -3.85637403e-01 -8.03697288e-01 -7.78655887e-01 -3.00727040e-01 -8.75725821e-02 7.69458771e-01 3.87011856e-01 -3.93377036e-01 -1.51324183e-01 -1.17679290e-01 -1.19323266e+00 4.21746790e-01 1.34517801e+00 7.48097003e-01 -1.72811419e-01 2.00528190e-01 -3.30043465e-01 6.42771840e-01 -7.40377843e-01 -8.39241505e-01 3.00311953e-01 -1.97605789e-02 -1.22818863e+00 -2.24096812e-02 4.62056249e-01 -7.64238164e-02 -3.53667170e-01 1.10728896e+00 3.61110985e-01 -8.97818655e-02 8.16718102e-01 -1.34098351e+00 -1.27577090e+00 8.50452125e-01 -3.81890893e-01 4.11410868e-01 5.17930210e-01 1.17721623e-02 1.26749027e+00 -7.29329944e-01 1.16258395e+00 1.07270944e+00 6.77934468e-01 4.50599432e-01 -1.13625574e+00 -7.27330387e-01 3.74111414e-01 2.51064867e-01 -1.42052925e+00 -6.66660431e-04 6.15943670e-01 -3.21639389e-01 9.18712854e-01 4.86760065e-02 -1.04135692e-01 1.76584351e+00 5.93004972e-02 9.61092174e-01 1.20556676e+00 -4.54610795e-01 2.27500573e-01 3.50690633e-01 4.75240022e-01 1.70116603e-01 7.26810634e-01 -5.54549433e-02 -3.56803060e-01 -3.20642471e-01 4.67270195e-01 -4.57935065e-01 -4.83815253e-01 -5.50992370e-01 -1.11974990e+00 8.58585358e-01 2.63706475e-01 4.13358718e-01 -8.74699205e-02 -7.73051560e-01 6.39635444e-01 6.53205335e-01 8.85632932e-01 8.45904112e-01 -1.00141060e+00 6.54014274e-02 -5.77809215e-01 5.73532104e-01 1.25738442e+00 1.19964159e+00 6.32513463e-01 -4.36297387e-01 4.59725074e-02 1.06951427e+00 -1.24650031e-01 4.68311995e-01 5.95283031e-01 -4.69265074e-01 1.02937794e+00 7.90300488e-01 7.28713945e-02 -9.19542611e-01 -2.00476810e-01 -6.68072581e-01 -4.70867693e-01 -5.03361762e-01 6.23169780e-01 -4.64178294e-01 -7.46721208e-01 1.70674014e+00 4.59942371e-01 -1.04178265e-02 4.50944513e-01 6.89435065e-01 1.10379469e+00 5.14774382e-01 9.40974131e-02 1.96877178e-02 1.34664023e+00 -7.02929854e-01 -7.40906298e-01 -6.15823746e-01 7.62957633e-01 -4.42705363e-01 1.19975698e+00 5.72180331e-01 -3.87651324e-01 -2.78079957e-01 -1.12441552e+00 -1.06908426e-01 -7.86291897e-01 -1.51239112e-01 3.75332892e-01 5.80083370e-01 -4.30885702e-01 4.94804680e-01 -1.29823059e-01 -3.66501302e-01 4.31419879e-01 1.12768732e-01 -5.68649352e-01 -6.00128710e-01 -1.91468966e+00 7.21152604e-01 9.10552979e-01 -6.14534140e-01 -6.70929790e-01 -9.97582734e-01 -8.98505151e-01 6.78968653e-02 1.05036330e+00 -6.44634187e-01 1.11523509e+00 -5.54022014e-01 -8.95669520e-01 1.01208770e+00 1.91504538e-01 -5.69980204e-01 5.00407934e-01 -5.52850783e-01 -8.46581757e-01 -5.85199669e-02 4.13482934e-01 3.18693042e-01 5.59807479e-01 -1.31235468e+00 -5.20077705e-01 -4.91095394e-01 3.59008372e-01 -1.67829305e-01 -6.18926167e-01 -2.26388261e-01 -2.01690599e-01 -7.31604815e-01 -6.59631863e-02 -9.21184003e-01 6.77995384e-02 -6.04069352e-01 -4.86324459e-01 -7.42217958e-01 8.09871912e-01 -4.81389314e-01 1.47337794e+00 -2.12710595e+00 -1.54140562e-01 -3.17845680e-02 6.68539703e-01 4.34292316e-01 -7.92345405e-02 5.47649026e-01 -3.11158776e-01 2.67961502e-01 -2.48049065e-01 1.18459396e-01 1.27593207e-03 5.34143865e-01 -5.36296964e-01 1.16819806e-01 3.40076298e-01 7.51781762e-01 -1.08217108e+00 -5.65241694e-01 -3.34170520e-01 -1.57332689e-01 -6.38706326e-01 -1.48343131e-01 -7.79238343e-01 2.06043676e-01 -6.93416953e-01 5.07283032e-01 9.55256104e-01 -5.68486929e-01 4.47767586e-01 2.60745823e-01 1.72631547e-01 5.52383959e-01 -1.09521067e+00 1.82455111e+00 -2.30971411e-01 6.27848744e-01 -3.03884745e-01 -9.97352123e-01 1.05941439e+00 3.75691831e-01 4.59539592e-01 -5.66200256e-01 2.23791137e-01 3.02579463e-01 -7.75293857e-02 -5.99961579e-01 4.86014664e-01 -1.84189156e-01 -4.69967246e-01 3.19609702e-01 6.43666804e-01 1.76886216e-01 5.50687790e-01 5.01129687e-01 1.20297480e+00 5.06589413e-02 5.22106469e-01 -4.70555782e-01 3.73642892e-01 2.75482744e-01 7.08519518e-01 6.63886845e-01 -2.47504532e-01 4.54655975e-01 7.90387571e-01 -2.74574459e-01 -8.10003519e-01 -1.14738083e+00 -5.51789880e-01 1.08139277e+00 3.01927119e-01 -6.28379703e-01 -7.26388812e-01 -1.50952232e+00 1.75727844e-01 8.92789960e-01 -5.59355736e-01 -5.37799392e-03 -2.70917058e-01 -7.26591885e-01 7.28767753e-01 2.58611232e-01 6.80593491e-01 -5.90368629e-01 3.79674584e-02 2.08250910e-01 -8.03991258e-01 -1.66111434e+00 -1.90692227e-02 3.29949081e-01 -6.06821895e-01 -1.14209759e+00 -6.31309152e-01 -6.12902284e-01 2.30055004e-01 1.13102481e-01 1.55080938e+00 -5.57341933e-01 -1.42334076e-02 1.55486062e-01 -8.72410417e-01 -5.29859304e-01 -4.38638389e-01 4.29300845e-01 2.83184182e-02 -2.50635892e-01 1.01214850e+00 -6.33886993e-01 -2.85008311e-01 4.99718755e-01 -1.00791383e+00 -4.24999267e-01 4.94606167e-01 9.42406774e-01 2.19101131e-01 3.15717399e-01 9.46853638e-01 -1.34656811e+00 8.03487539e-01 -1.03937483e+00 -2.74540335e-01 5.23589134e-01 -6.34268641e-01 2.63927609e-01 7.45770752e-01 -6.28492117e-01 -1.32391548e+00 -5.18770933e-01 -7.37553164e-02 -2.75916010e-01 -3.15573722e-01 6.38713717e-01 -4.17795062e-01 2.47565120e-01 1.30779183e+00 -4.47436003e-03 -5.11381447e-01 -6.55209839e-01 4.71663833e-01 8.64442348e-01 5.92520237e-01 -9.22689319e-01 7.50682473e-01 2.79482126e-01 -5.05892217e-01 -6.37003183e-01 -1.39516675e+00 -6.56239212e-01 -7.30640650e-01 3.36696267e-01 4.54285026e-01 -1.07382190e+00 -4.50277440e-02 1.53281838e-02 -1.11001301e+00 -1.46687984e-01 -3.20955038e-01 2.81165808e-01 -2.85514385e-01 5.59262514e-01 -4.16837931e-01 -1.62212417e-01 -2.94420067e-02 -6.02268398e-01 8.90486717e-01 -1.61206797e-01 -3.55794400e-01 -1.32175970e+00 3.75133485e-01 3.45216841e-01 -2.49700695e-01 3.79332095e-01 9.56343412e-01 -1.48615491e+00 -2.34335586e-01 -1.62844673e-01 -2.67771870e-01 5.25473058e-01 -3.10139023e-02 -4.07108366e-01 -1.05411494e+00 -1.99482396e-01 1.13087125e-01 -5.98519325e-01 7.64605999e-01 -1.01242155e-01 8.27036142e-01 -2.67089933e-01 -5.26411831e-01 4.21797007e-01 1.53635323e+00 -4.51249592e-02 6.58698678e-01 5.51547289e-01 4.34516996e-01 6.15510941e-01 9.25734878e-01 6.28877938e-01 3.85500133e-01 2.64596730e-01 1.73646063e-01 1.42166466e-01 -1.80054948e-01 -6.18626654e-01 2.37921447e-01 5.88819802e-01 3.04001808e-01 -6.71090126e-01 -1.09379113e+00 9.39946115e-01 -1.54024947e+00 -8.12766850e-01 -2.38408670e-01 1.89823759e+00 1.05252087e+00 3.42408478e-01 -7.67181488e-03 1.63629740e-01 6.42683029e-01 8.85084867e-02 -4.31853086e-01 -2.35408798e-01 -2.77585536e-01 2.01035753e-01 4.77333575e-01 -1.78979561e-02 -1.22976792e+00 1.02371562e+00 6.09929514e+00 1.22553718e+00 -7.72595942e-01 2.79138803e-01 2.38810122e-01 2.20551699e-01 -4.45147276e-01 1.01449437e-01 -1.15196788e+00 4.07090336e-01 8.89652908e-01 -3.83109868e-01 -5.43766022e-02 1.21735477e+00 -6.65563524e-01 1.07182488e-01 -1.36204588e+00 7.59887159e-01 1.00276962e-01 -1.01586139e+00 1.71916652e-02 2.35283762e-01 1.04522943e+00 2.31331661e-01 -1.22957133e-01 9.18203354e-01 7.13569283e-01 -4.95141625e-01 3.51203889e-01 -3.65297556e-01 9.20878947e-01 -4.99109268e-01 6.68052733e-01 8.03151667e-01 -7.61947930e-01 -1.46376252e-01 -5.04459023e-01 -1.25088710e-02 -2.51505166e-01 8.21351469e-01 -1.15730298e+00 9.47102666e-01 6.15330219e-01 7.70360172e-01 -7.07761645e-01 5.89356899e-01 -4.04618204e-01 7.39335597e-01 -3.45708847e-01 1.08385339e-01 2.59864330e-01 3.05845231e-01 6.50586188e-01 1.19743896e+00 3.76097560e-02 1.71318904e-01 3.04432184e-01 8.91814828e-01 -2.46070519e-01 7.51747042e-02 -1.02052784e+00 -8.26039240e-02 4.33042049e-01 7.37545252e-01 -2.88877904e-01 -3.31622273e-01 -7.83173561e-01 9.20429349e-01 5.22397935e-01 4.00326520e-01 -9.07850862e-01 -4.94407803e-01 5.89135885e-01 2.12994456e-01 3.28479946e-01 -3.78496796e-02 -2.53617138e-01 -1.62322533e+00 1.96166337e-01 -1.02156007e+00 9.23032165e-01 -3.87381613e-01 -1.70312750e+00 6.73508883e-01 3.19013089e-01 -1.36459088e+00 -4.02015448e-01 -8.00549626e-01 -1.20723762e-01 5.89013159e-01 -1.77420318e+00 -8.40354383e-01 -1.54319361e-01 8.38202417e-01 6.94067657e-01 -4.40509319e-02 7.12293446e-01 4.67608511e-01 -4.60834891e-01 8.20696533e-01 5.60826421e-01 4.04738128e-01 1.11439347e+00 -1.25922263e+00 7.57858694e-01 8.27915728e-01 2.94244558e-01 6.65072739e-01 8.37128222e-01 -9.30377364e-01 -9.03118849e-01 -1.03083158e+00 1.11808586e+00 -8.87475431e-01 9.77165818e-01 -5.62022090e-01 -1.14377069e+00 1.00186718e+00 1.82562679e-01 1.72217950e-01 7.84598947e-01 6.81720912e-01 -7.68067658e-01 2.75972664e-01 -1.18792534e+00 4.99064326e-01 1.31648219e+00 -2.50701070e-01 -1.28480661e+00 4.60693985e-01 9.30926859e-01 -2.87010312e-01 -1.10311019e+00 5.20039320e-01 9.84657109e-02 -8.10533822e-01 9.23744380e-01 -9.50846255e-01 7.38729894e-01 1.22933455e-01 -1.48393765e-01 -1.65625715e+00 -2.91291952e-01 -4.39667881e-01 6.42430484e-02 1.53619885e+00 6.30625665e-01 -8.15223634e-01 6.02838993e-01 3.56618673e-01 9.93349031e-02 -4.40253943e-01 -8.90285432e-01 -1.43174779e+00 5.23211122e-01 -4.77866530e-01 4.84463960e-01 1.26516318e+00 5.11206865e-01 7.46330738e-01 -2.06958339e-01 2.13918149e-01 5.03234267e-01 9.68904942e-02 8.42459857e-01 -1.58827901e+00 -2.77542353e-01 2.35925376e-01 -2.06163362e-01 -1.40448713e+00 4.33278114e-01 -9.69368935e-01 -2.53989518e-01 -1.23269725e+00 1.32104501e-01 -5.84480464e-01 -1.58762768e-01 2.19491497e-01 -3.31005126e-01 -2.69575089e-01 -4.05107513e-02 1.75699145e-01 -9.29150105e-01 5.54227650e-01 1.40215290e+00 -2.25892931e-01 -4.73175235e-02 -1.82581976e-01 -1.00248420e+00 6.41029298e-01 6.48252308e-01 -6.27070785e-01 -7.74863422e-01 -4.36055303e-01 1.69284865e-01 -8.55673477e-02 1.90396115e-01 -8.87449801e-01 -4.19715270e-02 -1.44210681e-02 9.49754491e-02 -3.38601112e-01 -1.57388613e-01 -8.53450298e-01 -3.64683121e-01 -1.04104891e-01 -2.57885098e-01 -4.47506934e-01 3.52921039e-01 8.04482102e-01 -4.58514541e-01 -3.54000241e-01 3.57036740e-01 -8.17500874e-02 -9.92914975e-01 2.18247905e-01 1.30198628e-01 1.11744273e+00 9.50125992e-01 -1.36667505e-01 -7.87426829e-01 -1.39621627e-02 -7.12490439e-01 3.07969421e-01 7.72360489e-02 6.44171417e-01 2.24574089e-01 -1.09297621e+00 -8.72440934e-01 1.34961948e-01 7.49818146e-01 1.33727789e-01 3.67375433e-01 4.61760342e-01 -2.31510196e-02 7.00486243e-01 -4.09349762e-02 -6.04336262e-01 -1.03842068e+00 9.15377080e-01 -5.84092103e-02 -9.46198285e-01 -6.98147714e-01 8.74003112e-01 6.07426643e-01 -6.99591935e-01 -3.52565311e-02 -3.17222953e-01 -1.56947784e-02 -9.80274007e-02 3.62893552e-01 2.20646366e-01 2.98691005e-01 -1.53223529e-01 -2.41362035e-01 1.44311100e-01 -4.39069688e-01 1.74621448e-01 1.12855971e+00 -1.61172196e-01 2.90804446e-01 1.75918132e-01 1.18939459e+00 6.68350384e-02 -1.10901570e+00 -7.92200089e-01 3.17158878e-01 -5.98330796e-01 -2.40917534e-01 -1.10409033e+00 -4.24229473e-01 6.56386793e-01 1.31634651e-02 3.78634393e-01 9.14828122e-01 2.95016736e-01 9.18565691e-01 4.28866506e-01 6.22093379e-01 -1.08383143e+00 -2.07153838e-02 8.30244958e-01 8.11874211e-01 -1.44940341e+00 2.56064236e-02 -9.07593608e-01 -8.78213763e-01 8.66970122e-01 7.47232854e-01 -9.84235108e-02 6.38775170e-01 2.36887723e-01 -1.55563504e-01 -3.27316076e-01 -9.06860948e-01 -2.70124406e-01 2.17743546e-01 8.48332763e-01 2.83663183e-01 1.23809557e-02 -3.90003830e-01 1.13147247e+00 -2.03137636e-01 1.64421111e-01 3.59671533e-01 6.83981240e-01 -3.42982054e-01 -1.08804667e+00 -4.10965145e-01 2.50170439e-01 -5.67215085e-01 -1.35258287e-01 -6.61200881e-01 1.31457102e+00 3.41200948e-01 9.22800899e-01 -2.51019746e-01 -3.50440919e-01 4.95777667e-01 3.62755060e-01 3.80001426e-01 -7.55322754e-01 -3.22053969e-01 -3.37750494e-01 5.02799511e-01 -8.55118856e-02 -1.41542079e-02 -5.97526729e-01 -9.24335182e-01 -2.95233339e-01 -5.81466496e-01 2.64604121e-01 2.07697108e-01 1.02225292e+00 4.85722631e-01 4.26374465e-01 4.09147382e-01 6.19580373e-02 -9.09599185e-01 -9.89879489e-01 -6.74235284e-01 7.32242763e-01 9.54579115e-02 -1.03478038e+00 -5.00082493e-01 -1.02612473e-01]
[9.977495193481445, 8.697132110595703]
a6d8a6e9-d2a2-4a53-989a-c431076226a3
incorporating-deep-q-network-with-multiclass
2307.03908
null
https://arxiv.org/abs/2307.03908v1
https://arxiv.org/pdf/2307.03908v1.pdf
Incorporating Deep Q -- Network with Multiclass Classification Algorithms
In this study, we explore how Deep Q-Network (DQN) might improve the functionality of multiclass classification algorithms. We will use a benchmark dataset from Kaggle to create a framework incorporating DQN with existing supervised multiclass classification algorithms. The findings of this study will bring insight into how deep reinforcement learning strategies may be used to increase multiclass classification accuracy. They have been used in a number of fields, including image recognition, natural language processing, and bioinformatics. This study is focused on the prediction of financial distress in companies in addition to the wider application of Deep Q-Network in multiclass classification. Identifying businesses that are likely to experience financial distress is a crucial task in the fields of finance and risk management. Whenever a business experiences serious challenges keeping its operations going and meeting its financial responsibilities, it is said to be in financial distress. It commonly happens when a company has a sharp and sustained recession in profitability, cash flow issues, or an unsustainable level of debt.
['Ravindranath Sawane', 'Noopur Zambare']
2023-07-08
null
null
null
null
['classification-1', 'management']
['methodology', 'miscellaneous']
[-8.73963833e-02 -4.08675261e-02 -4.14894938e-01 -5.88382483e-01 -3.20823073e-01 -5.67619324e-01 1.15560569e-01 3.24307293e-01 -2.92107224e-01 7.20382214e-01 1.60135254e-01 -7.11715698e-01 -4.27872986e-01 -1.09732640e+00 -5.86736798e-02 -4.87916172e-01 2.40504503e-01 4.02299583e-01 -2.58554697e-01 -4.17763054e-01 8.48292887e-01 9.47259665e-01 -9.74326849e-01 6.70417547e-01 3.17782491e-01 1.13782382e+00 -1.89467758e-01 5.05202413e-01 -8.11277516e-03 1.33417189e+00 -8.64880860e-01 -6.92461967e-01 1.34282619e-01 -2.64623433e-01 -1.07656729e+00 -9.28237289e-03 2.70703714e-02 -4.30670261e-01 -2.80329143e-03 8.78315330e-01 5.55757523e-01 2.46207774e-01 5.55675745e-01 -1.10267448e+00 -5.28806984e-01 2.63788253e-01 -4.23001885e-01 6.77355349e-01 1.22083113e-01 1.48857728e-01 1.00862026e+00 -5.47711730e-01 2.50371248e-01 1.17604613e+00 7.70044684e-01 2.66495496e-01 -9.02664661e-01 -7.93673277e-01 -1.53611869e-01 3.05172354e-01 -5.16064703e-01 -1.29223421e-01 5.04035294e-01 -5.75899303e-01 1.35889959e+00 -4.62945136e-05 4.39433187e-01 7.32186019e-01 7.23567843e-01 4.22416151e-01 8.86070251e-01 -4.21699077e-01 1.80771783e-01 2.85682768e-01 9.49660540e-02 1.46660894e-01 3.03053975e-01 1.74393877e-01 -3.39219391e-01 -3.47944945e-02 5.33731759e-01 1.97885156e-01 1.62136152e-01 1.19846106e-01 -9.03118193e-01 1.35589087e+00 3.96191955e-01 6.27228022e-01 -3.14651400e-01 2.26842597e-01 7.47515738e-01 6.51739657e-01 3.46676588e-01 9.90993857e-01 -5.23287058e-01 -4.61226106e-01 -6.76129937e-01 1.07094862e-01 7.80469537e-01 2.25479424e-01 5.24946451e-01 1.92548200e-01 1.20216981e-01 8.42999160e-01 -1.53581187e-01 -1.55271024e-01 8.11505914e-01 -1.18706477e+00 4.26624268e-01 7.88061917e-01 -8.46862793e-02 -1.17002833e+00 -6.70466483e-01 -2.87529558e-01 -6.92073047e-01 2.82501519e-01 1.87691838e-01 -1.18303448e-01 -2.69390285e-01 9.84709799e-01 -1.61542982e-01 -1.28693044e-01 3.60243171e-01 7.15796351e-01 5.66863596e-01 5.56564450e-01 1.32316470e-01 -2.15718746e-01 1.19351709e+00 -6.97602391e-01 -6.26590669e-01 -2.80530691e-01 9.18411791e-01 -5.26619971e-01 7.55105734e-01 9.09075916e-01 -8.10819566e-01 -5.44487357e-01 -8.96487713e-01 2.56578267e-01 -5.33080161e-01 -1.33715183e-01 7.97928631e-01 7.18406320e-01 -6.78924739e-01 8.98732126e-01 -5.32100201e-01 -4.17026252e-01 4.61544216e-01 3.39402765e-01 -2.65565842e-01 -8.15981701e-02 -1.32614088e+00 1.12796712e+00 3.44575912e-01 7.62811229e-02 -6.34842992e-01 -4.09131944e-01 -6.52487218e-01 1.11033320e-01 4.56440412e-02 -4.17230159e-01 1.22680032e+00 -1.26080287e+00 -7.92267680e-01 7.41548657e-01 3.63653153e-01 -8.45423758e-01 3.61917168e-01 -1.23313136e-01 -5.01628637e-01 1.36268035e-01 1.78372115e-01 1.27377346e-01 4.04451877e-01 -5.82756758e-01 -7.56692708e-01 -7.58998156e-01 9.91511047e-02 5.88102825e-02 -6.06387794e-01 5.03316343e-01 6.52850628e-01 -8.43204021e-01 -3.17737088e-02 -6.75632894e-01 -3.16485524e-01 -4.86253381e-01 -5.31320199e-02 -2.82565475e-01 8.80703151e-01 -6.65603161e-01 1.10619569e+00 -1.99183071e+00 -3.99878442e-01 2.06900816e-02 -1.24263227e-01 3.87842834e-01 2.12006599e-01 6.34976983e-01 -3.20660681e-01 4.93490309e-01 -8.20629671e-02 3.44668597e-01 -2.02193156e-01 1.36929095e-01 -2.86430329e-01 2.82098591e-01 7.75078058e-01 7.90442050e-01 -6.92010164e-01 -8.38719457e-02 -1.45831093e-01 -2.49264445e-02 -2.60051548e-01 2.60682553e-01 2.61868656e-01 2.29094446e-01 -4.39568967e-01 1.09313154e+00 5.64754248e-01 3.54127539e-03 1.47462949e-01 2.18018517e-01 2.26536486e-03 2.15983480e-01 -9.05832410e-01 7.37447858e-01 -4.76473242e-01 6.66323721e-01 -3.88788134e-01 -1.79921639e+00 1.27852368e+00 4.14493889e-01 3.14346224e-01 -9.63792026e-01 -7.79601410e-02 1.64410546e-01 1.14984341e-01 -7.53043056e-01 6.38866544e-01 -8.21322680e-01 2.30315067e-02 6.05395675e-01 -2.44653612e-01 -2.50786483e-01 1.77926317e-01 -2.47496024e-01 1.25395405e+00 -1.49687096e-01 2.88821608e-01 5.32936212e-03 5.45931458e-01 -3.94639559e-02 9.13633049e-01 4.34587032e-01 -6.25633895e-01 3.47503245e-01 1.13988256e+00 -7.41271794e-01 -8.06270242e-01 -5.98662853e-01 -3.07387561e-01 9.46048319e-01 -3.48221391e-01 2.34608799e-01 -1.89557120e-01 -6.25372946e-01 3.41751218e-01 7.16797650e-01 -4.96164083e-01 -6.31480813e-01 -4.07103509e-01 -9.50741410e-01 6.35126233e-01 9.18628216e-01 4.76899326e-01 -1.62681174e+00 -8.24273884e-01 5.41195452e-01 1.71285495e-01 -9.26083684e-01 2.63847530e-01 6.80805147e-01 -1.16619205e+00 -1.11078680e+00 -6.98121727e-01 -9.29504395e-01 2.39945635e-01 2.88798790e-02 1.04928660e+00 2.82795588e-03 -3.13313305e-01 1.06521532e-01 -5.18945277e-01 -5.77397466e-01 -5.96287131e-01 1.48215875e-01 -6.84076250e-02 -1.52792349e-01 6.71865165e-01 -1.23893842e-01 -6.40227973e-01 5.25843799e-01 -9.44279194e-01 -6.48978293e-01 2.42690831e-01 1.01746488e+00 2.96403497e-01 6.52103841e-01 1.60691905e+00 -1.04986668e+00 9.87675726e-01 -7.23251820e-01 -1.70186862e-01 2.12167799e-01 -8.93143594e-01 -5.47400773e-01 4.70307231e-01 -1.94997430e-01 -1.00663686e+00 -2.36025050e-01 -1.21216856e-01 5.55727892e-02 -1.28761947e-01 9.34392631e-01 2.99679369e-01 4.96339276e-02 6.42063200e-01 -1.22980602e-01 9.36276540e-02 -1.14652880e-01 -3.78811896e-01 9.77212667e-01 1.00152060e-01 -2.07718551e-01 1.09690107e-01 2.28562623e-01 -2.58565862e-02 -3.99565428e-01 -8.59645128e-01 -7.25672781e-01 -6.20356262e-01 -2.55924314e-01 7.69900382e-01 -5.74587941e-01 -6.32853091e-01 5.52427173e-01 -7.10887551e-01 -3.21272284e-01 -1.86687913e-02 3.82975698e-01 -5.98354936e-01 2.77056009e-01 -7.53048301e-01 -7.34896243e-01 -2.51790196e-01 -9.10956025e-01 4.97353196e-01 3.89054477e-01 -2.99477607e-01 -1.05696154e+00 4.85156029e-02 8.70125473e-01 3.16595107e-01 3.07270944e-01 1.16089022e+00 -7.80919433e-01 -3.14304372e-03 -4.62636977e-01 -2.81791866e-01 9.54444110e-01 2.88365394e-01 1.66426562e-02 -9.08619821e-01 -3.75776798e-01 1.63685128e-01 -7.57250786e-01 7.61017025e-01 3.02186638e-01 1.28484499e+00 -5.87709621e-02 1.28212810e-01 2.66707212e-01 1.45619869e+00 8.19769084e-01 8.32468331e-01 1.13316298e+00 1.63472369e-01 1.07309604e+00 1.26996183e+00 7.93772459e-01 9.55815464e-02 1.66570380e-01 7.21824706e-01 -5.13931960e-02 5.57535708e-01 4.56982762e-01 3.93249124e-01 3.23633552e-01 1.24513790e-01 -1.85499191e-01 -1.31233895e+00 6.33411646e-01 -1.57011425e+00 -1.12430847e+00 2.30649225e-02 1.96830535e+00 6.76056087e-01 4.05775279e-01 2.03584909e-01 4.10557508e-01 5.49837947e-01 -2.93315917e-01 -7.13222027e-01 -1.38784456e+00 -1.23758979e-01 2.40189776e-01 2.98826545e-01 -1.03611067e-01 -1.10196102e+00 7.22855508e-01 5.98804188e+00 2.31559753e-01 -1.21793187e+00 -4.54710037e-01 1.42469525e+00 1.12444736e-01 -1.17017746e-01 1.07710194e-02 -7.53372669e-01 1.34169310e-01 1.35791349e+00 -6.52564988e-02 1.32772684e-01 8.16622674e-01 4.66292202e-01 -3.29040080e-01 -9.29130614e-01 4.09098119e-01 -5.33245280e-02 -1.20346606e+00 -2.13464946e-01 7.29332268e-02 5.69048464e-01 -3.66492242e-01 9.84753370e-02 4.46794987e-01 1.35158271e-01 -1.37908185e+00 3.49211514e-01 4.49460030e-01 4.54871446e-01 -1.33895874e+00 1.29182136e+00 3.23862821e-01 -6.10359192e-01 -9.42706585e-01 -5.31729221e-01 -3.75404626e-01 -4.40931022e-02 4.75453228e-01 -1.04250741e+00 2.69270480e-01 1.09528399e+00 9.69640136e-01 -5.07293642e-01 8.20916593e-01 3.86593230e-02 7.89491832e-01 5.94748676e-01 -5.90990065e-04 3.03388566e-01 -2.97023773e-01 -2.06889704e-01 1.19168544e+00 2.76394188e-01 -1.87178180e-02 -2.23599479e-01 4.54409093e-01 1.08204253e-01 2.32378066e-01 -8.04332137e-01 -4.25981402e-01 -1.07168771e-01 1.13418484e+00 -8.42127144e-01 -2.13216871e-01 -6.25670195e-01 5.99519908e-01 7.03800768e-02 6.28730580e-02 -4.75523472e-01 -7.01965332e-01 4.84199673e-01 7.28657767e-02 3.08179945e-01 1.30968228e-01 -6.53366983e-01 -6.70167148e-01 -2.02957675e-01 -1.13849008e+00 6.20902359e-01 -8.30122769e-01 -1.41113830e+00 4.21838433e-01 -5.27598262e-01 -1.22057974e+00 -3.25754702e-01 -7.84994960e-01 -6.96417034e-01 7.12523520e-01 -1.72236001e+00 -8.05093884e-01 -1.30312383e-01 1.72165886e-01 3.29127878e-01 -6.52778327e-01 7.49648154e-01 2.25801632e-01 -5.61220586e-01 3.80862236e-01 2.50607640e-01 2.48026744e-01 7.79268682e-01 -1.18291831e+00 2.09815010e-01 5.37594616e-01 -2.60525852e-01 2.94027239e-01 2.62294441e-01 -5.21330357e-01 -7.06527889e-01 -1.13261092e+00 1.13550580e+00 -3.52393612e-02 8.41318846e-01 7.10162893e-02 -9.61997390e-01 6.19829357e-01 -7.42475688e-02 -1.38561979e-01 9.20104444e-01 -1.15835689e-01 8.46126825e-02 -5.06748557e-01 -1.28189647e+00 1.12041794e-01 1.27732307e-01 -5.33285797e-01 -5.29846728e-01 4.32475358e-01 4.50833321e-01 1.37908280e-01 -1.18847466e+00 3.62988174e-01 4.42275107e-01 -1.21087301e+00 7.86924481e-01 -9.75942314e-01 1.01017404e+00 4.37778592e-01 -7.78432712e-02 -1.18514800e+00 -3.92270029e-01 -1.33510962e-01 3.09130400e-01 1.24311149e+00 2.54849225e-01 -7.59221017e-01 9.24316883e-01 5.96847773e-01 -8.70946348e-02 -1.04824209e+00 -8.96084785e-01 -6.54556096e-01 6.19819701e-01 -3.35316420e-01 5.16061902e-01 1.26555979e+00 3.51786502e-02 -4.97468337e-02 -2.93718204e-02 -3.00658226e-01 -7.10859522e-02 1.89205542e-01 5.49463272e-01 -1.19571030e+00 -5.31505942e-02 -3.97031873e-01 -6.68720484e-01 -1.61639974e-01 2.55629867e-01 -6.84886992e-01 -3.96990567e-01 -1.45859385e+00 1.80785760e-01 -5.18140554e-01 -8.27126741e-01 5.66362500e-01 3.02053899e-01 9.59432721e-02 1.72862172e-01 -3.12392116e-02 -1.70896292e-01 2.01637894e-01 1.20355523e+00 -2.37341642e-01 -7.94192553e-02 6.26326442e-01 -1.11454785e+00 5.95198631e-01 1.16311228e+00 -5.19292474e-01 -2.77594477e-01 7.76227266e-02 6.86740279e-01 6.84409440e-01 2.61203527e-01 -6.06235087e-01 -1.57162920e-01 -4.74438399e-01 3.69738519e-01 -3.97911459e-01 3.51869278e-02 -7.09683418e-01 -1.73519120e-01 9.10392523e-01 -4.09177482e-01 4.43226755e-01 3.49704027e-01 2.99979419e-01 -6.80148661e-01 -8.68159294e-01 9.39025104e-01 -4.33298200e-01 -6.11264348e-01 -1.84484079e-01 -8.62614870e-01 -6.61112145e-02 1.20729589e+00 -5.57562292e-01 -3.65804136e-01 -4.07291174e-01 -8.08856487e-01 1.53694585e-01 2.33940125e-01 4.83957112e-01 8.00504267e-01 -1.13801503e+00 -6.09910011e-01 -3.76752950e-02 1.02014355e-01 -3.10001612e-01 5.77886961e-02 7.16485679e-01 -8.98111999e-01 6.02332473e-01 -5.93942642e-01 -7.73152709e-02 -1.05960381e+00 4.10771728e-01 6.81400239e-01 -6.12069249e-01 -3.19370449e-01 7.53865063e-01 8.98126960e-02 -3.06173116e-01 2.29621962e-01 -3.84985268e-01 -6.78896487e-01 5.93034923e-01 2.56669670e-01 5.49424589e-01 5.45538545e-01 -2.91601151e-01 -3.78523648e-01 2.67999321e-01 -2.97697306e-01 2.94407874e-01 1.70771849e+00 2.76238173e-01 -2.98021823e-01 5.93545079e-01 1.07172370e+00 -6.07390165e-01 -8.60087752e-01 1.69286564e-01 4.85048950e-01 -3.99804145e-01 2.08592072e-01 -9.01926696e-01 -1.30473483e+00 1.16789973e+00 6.35317385e-01 5.13479173e-01 1.26846087e+00 -2.25453407e-01 7.56685019e-01 4.73556608e-01 2.87166778e-02 -1.45927262e+00 6.70878291e-01 4.70379829e-01 9.36656237e-01 -1.57396197e+00 2.55866259e-01 2.84470081e-01 -1.03828025e+00 1.56628692e+00 6.78911865e-01 -3.23350020e-02 7.21956611e-01 -3.06599122e-02 1.77124351e-01 -2.28507727e-01 -6.72132969e-01 4.49787885e-01 -1.66223586e-01 5.77845454e-01 6.91150427e-01 -1.14760935e-01 -3.77762437e-01 7.28830755e-01 -9.30883586e-02 9.52359289e-02 8.75393391e-01 1.11209464e+00 -4.32531446e-01 -1.11191833e+00 -3.73829901e-01 8.33637357e-01 -1.20221043e+00 -9.37763005e-02 -5.30992329e-01 7.09383070e-01 6.03659190e-02 8.62790704e-01 2.08801225e-01 -1.27857298e-01 3.18968207e-01 -1.48587838e-01 -2.76443928e-01 -6.59119427e-01 -1.11732936e+00 -2.18309969e-01 1.93612128e-01 -2.88275778e-01 -4.77554023e-01 -1.06120646e+00 -1.35719168e+00 -1.69716939e-01 -3.48084360e-01 -2.37676099e-01 5.47597170e-01 7.01829612e-01 7.87274465e-02 5.10215700e-01 9.04742122e-01 -8.26704204e-02 -7.80636072e-01 -8.33928764e-01 -1.02239764e+00 2.45920092e-01 3.19037110e-01 -5.55468440e-01 -4.74614054e-01 -9.27905664e-02]
[4.653134822845459, 4.227510929107666]
4db98423-c9bb-4461-961e-2e9ad41e6c09
self-explaining-structures-improve-nlp-models
2012.01786
null
https://arxiv.org/abs/2012.01786v2
https://arxiv.org/pdf/2012.01786v2.pdf
Self-Explaining Structures Improve NLP Models
Existing approaches to explaining deep learning models in NLP usually suffer from two major drawbacks: (1) the main model and the explaining model are decoupled: an additional probing or surrogate model is used to interpret an existing model, and thus existing explaining tools are not self-explainable; (2) the probing model is only able to explain a model's predictions by operating on low-level features by computing saliency scores for individual words but are clumsy at high-level text units such as phrases, sentences, or paragraphs. To deal with these two issues, in this paper, we propose a simple yet general and effective self-explaining framework for deep learning models in NLP. The key point of the proposed framework is to put an additional layer, as is called by the interpretation layer, on top of any existing NLP model. This layer aggregates the information for each text span, which is then associated with a specific weight, and their weighted combination is fed to the softmax function for the final prediction. The proposed model comes with the following merits: (1) span weights make the model self-explainable and do not require an additional probing model for interpretation; (2) the proposed model is general and can be adapted to any existing deep learning structures in NLP; (3) the weight associated with each text span provides direct importance scores for higher-level text units such as phrases and sentences. We for the first time show that interpretability does not come at the cost of performance: a neural model of self-explaining features obtains better performances than its counterpart without the self-explaining nature, achieving a new SOTA performance of 59.1 on SST-5 and a new SOTA performance of 92.3 on SNLI.
['Jiwei Li', 'Fei Wu', 'Yuxian Meng', 'Xiaofei Sun', 'Qinghong Han', 'Chun Fan', 'Zijun Sun']
2020-12-03
null
null
null
null
['paraphrase-identification']
['natural-language-processing']
[ 1.71321541e-01 8.25853884e-01 -5.42320788e-01 -5.76683939e-01 -3.48013401e-01 -2.35398486e-01 4.44951415e-01 2.16450736e-01 4.38276026e-03 7.28106141e-01 3.66113305e-01 -3.65338683e-01 -1.04290776e-01 -5.75753629e-01 -8.49250376e-01 -3.67628187e-01 2.40376472e-01 6.15870178e-01 1.67405188e-01 -1.02584176e-01 1.42649904e-01 2.97034770e-01 -1.54370320e+00 5.32236814e-01 1.32854187e+00 1.03878391e+00 4.59181905e-01 2.93043762e-01 -7.53956735e-01 7.38944471e-01 -4.97310549e-01 -3.69787306e-01 -1.85182482e-01 -5.82826674e-01 -8.55668306e-01 -1.33968681e-01 2.00748190e-01 -1.76183924e-01 -4.09275629e-02 7.35292494e-01 1.23586588e-01 -1.65391475e-01 6.82205200e-01 -1.30118251e+00 -1.08106542e+00 1.20418715e+00 -4.58787471e-01 7.18805864e-02 5.48905991e-02 -1.84931710e-01 1.43253374e+00 -9.06035364e-01 1.71296254e-01 1.29630339e+00 5.26865423e-01 7.51850307e-01 -1.05175006e+00 -6.91216409e-01 5.50807595e-01 2.50709802e-01 -7.57026494e-01 -1.32226095e-01 7.57702112e-01 -4.99500036e-02 1.20269394e+00 2.62338430e-01 6.76413894e-01 1.15458143e+00 2.62630373e-01 1.11556959e+00 7.34851658e-01 -3.64818215e-01 1.94227174e-01 3.36192161e-01 5.44495940e-01 6.79777145e-01 2.32245490e-01 -9.95908007e-02 -5.18743038e-01 1.42387956e-01 5.90992033e-01 1.37289673e-01 -1.37415752e-01 -2.61580586e-01 -1.08647048e+00 9.97686982e-01 8.49975228e-01 6.08026683e-01 -3.74789596e-01 3.16232771e-01 1.35910377e-01 -1.56914759e-02 5.46464205e-01 5.78433454e-01 -8.22969198e-01 1.66456804e-01 -9.42904413e-01 9.50348154e-02 7.14064956e-01 8.81502092e-01 9.04975116e-01 2.34926894e-01 -1.90581843e-01 7.21199274e-01 4.21105862e-01 1.96314663e-01 8.40080619e-01 -4.93081361e-01 6.79756463e-01 8.96043420e-01 -9.68146548e-02 -9.13988829e-01 -8.42578709e-01 -9.54437852e-01 -9.04952645e-01 -1.29502490e-01 1.99859422e-02 -1.24063686e-01 -9.51575518e-01 1.94476199e+00 -8.94952118e-02 4.63300012e-02 3.13609332e-01 8.28281581e-01 1.05786622e+00 8.68360519e-01 3.43983382e-01 -1.51630178e-01 1.45511854e+00 -1.44301701e+00 -9.23787832e-01 -8.34890962e-01 4.33330268e-01 -4.80364889e-01 1.44337511e+00 1.18135810e-01 -1.14316857e+00 -7.44858325e-01 -1.04877353e+00 -4.83929902e-01 -4.29046601e-01 3.71091247e-01 8.83391917e-01 2.50578493e-01 -9.69331324e-01 4.82769370e-01 -6.10147417e-01 -2.50149548e-01 2.64758497e-01 4.61396992e-01 -1.40649691e-01 4.83312905e-01 -1.41500998e+00 1.10879183e+00 7.31219351e-01 -2.06674188e-02 -4.06065851e-01 -6.82109118e-01 -8.65885139e-01 8.75236511e-01 1.72679529e-01 -1.01468897e+00 1.06132686e+00 -1.31603670e+00 -1.34999466e+00 4.38958198e-01 -7.25491464e-01 -5.94527960e-01 2.28567272e-01 -4.53639865e-01 -3.55199784e-01 -1.06759399e-01 1.68513209e-01 9.26851690e-01 7.80235827e-01 -1.41308093e+00 -5.94852865e-01 -3.03990930e-01 1.12999432e-01 2.49921933e-01 -6.13412797e-01 -4.45601761e-01 -3.89724553e-01 -7.53354967e-01 3.78800750e-01 -7.64314592e-01 -6.72303140e-03 -3.87665391e-01 -6.78503573e-01 -3.31389993e-01 7.50386655e-01 -5.40257335e-01 1.42098403e+00 -1.88877106e+00 2.83096343e-01 -2.99936682e-01 4.44532901e-01 3.26262295e-01 -1.78655714e-01 2.62367070e-01 -5.02017081e-01 4.05494988e-01 -1.80742562e-01 -5.16356885e-01 1.63269624e-01 2.54568100e-01 -5.41304827e-01 -2.29958385e-01 4.74610925e-01 1.15452373e+00 -7.34176993e-01 -2.97136873e-01 2.30397046e-01 4.47893411e-01 -5.65306723e-01 -3.67460847e-02 -3.90940636e-01 1.66037142e-01 -5.42963028e-01 3.11223537e-01 4.98347014e-01 -5.36215842e-01 -8.77543166e-02 -8.84285793e-02 -7.99488574e-02 6.93613410e-01 -9.65881169e-01 1.43271291e+00 -4.84889209e-01 6.77610576e-01 -4.27859992e-01 -9.82019663e-01 1.11267662e+00 4.36217129e-01 2.78232582e-02 -4.81900215e-01 -5.72241209e-02 2.79439092e-01 2.56601926e-02 -4.15516943e-01 5.38997829e-01 -2.54641503e-01 -6.59041703e-02 5.17079890e-01 2.37354860e-01 2.14341983e-01 -1.22570731e-01 1.29487157e-01 8.40363860e-01 -8.19768161e-02 4.75851744e-01 -1.97121769e-01 5.99256754e-01 -1.32067800e-01 5.28884172e-01 5.75787306e-01 6.12132102e-02 5.57020247e-01 7.42139816e-01 -8.26697409e-01 -1.01680672e+00 -8.67414474e-01 -1.03033274e-01 9.51603174e-01 1.33620277e-01 -2.72767395e-01 -7.46192932e-01 -9.70239818e-01 -9.77867283e-03 1.16686809e+00 -6.98094547e-01 -3.17717284e-01 -3.60804975e-01 -4.34619904e-01 1.45362988e-01 8.12968612e-01 5.24407089e-01 -1.37672174e+00 -3.26667219e-01 2.63014019e-01 -4.40262139e-01 -1.06693590e+00 -1.85448155e-01 6.46626055e-01 -1.21357071e+00 -5.94584107e-01 -4.37510312e-01 -7.81097472e-01 6.90227151e-01 2.61453152e-01 1.17722368e+00 2.41501763e-01 4.33076084e-01 -2.73665518e-01 -2.78355420e-01 -7.56003976e-01 -2.11716995e-01 6.04545772e-01 -5.26227392e-02 -2.83144116e-02 5.34176648e-01 -4.98556495e-01 -3.53185773e-01 1.54658377e-01 -9.02633846e-01 5.00796854e-01 7.93610990e-01 1.01053989e+00 4.38818187e-01 3.15502770e-02 8.49775314e-01 -9.39885557e-01 5.65727055e-01 -4.61099982e-01 -2.24034935e-01 1.76445708e-01 -7.61421442e-01 4.98633116e-01 8.84064615e-01 -3.47133130e-01 -9.23939645e-01 6.05976470e-02 -2.20681801e-01 -1.37938917e-01 -1.32036820e-01 5.59054017e-01 -3.47015798e-01 3.39727700e-01 4.33347493e-01 1.97857201e-01 -2.59047180e-01 -5.94656885e-01 3.43595862e-01 5.39031029e-01 3.46374661e-01 -9.77543592e-02 9.08553720e-01 1.08508043e-01 -2.44812742e-01 -4.38533425e-01 -1.30526769e+00 -1.07779153e-01 -5.97937763e-01 1.64805576e-01 6.66692615e-01 -8.32056582e-01 -7.20217168e-01 -8.34801123e-02 -1.52938306e+00 3.93100940e-02 -3.26465845e-01 5.91982484e-01 -5.14612138e-01 2.09646180e-01 -5.71618199e-01 -6.05304480e-01 -4.55554366e-01 -1.01468742e+00 9.59460199e-01 4.74139243e-01 -4.72541153e-01 -1.14693499e+00 -3.49657208e-01 5.66816390e-01 5.05066097e-01 -8.95613953e-02 1.36351287e+00 -9.98941660e-01 -4.49377656e-01 -1.14006065e-01 -4.54041839e-01 2.88064510e-01 5.44640981e-03 -2.46267453e-01 -1.14987457e+00 1.60880029e-01 1.44063577e-01 -1.20242052e-01 1.08755100e+00 5.48748970e-01 1.30621970e+00 -5.31555235e-01 -3.07289302e-01 4.68857586e-01 1.21618569e+00 3.14318500e-02 5.10629475e-01 4.55714673e-01 6.83028162e-01 6.56184435e-01 4.33895230e-01 1.54885620e-01 5.02052128e-01 5.80764830e-01 8.20798159e-01 -5.36006212e-01 -8.26536193e-02 -5.32665968e-01 3.00427943e-01 7.97216237e-01 1.07347079e-01 -4.77162540e-01 -7.27263868e-01 5.56928217e-01 -2.17843485e+00 -7.98108697e-01 -3.44387740e-01 1.79684246e+00 5.98305166e-01 2.69445062e-01 -3.39029193e-01 1.29664272e-01 5.78583181e-01 2.89287597e-01 -6.98934138e-01 -7.60242701e-01 -1.28883719e-01 -1.83098838e-02 1.86703905e-01 6.81116343e-01 -8.69555056e-01 1.22866333e+00 6.16227865e+00 6.23096526e-01 -1.01747334e+00 8.01487640e-02 5.31408727e-01 -3.74060534e-02 -6.49872243e-01 6.02265149e-02 -9.79690015e-01 4.20903653e-01 8.68616998e-01 -1.63692370e-01 2.13604316e-01 1.05806839e+00 3.32638562e-01 2.79113084e-01 -1.30220914e+00 5.92652202e-01 -1.93808489e-02 -1.31234169e+00 5.55067718e-01 -3.17533351e-02 5.69501698e-01 -2.62490302e-01 1.09372944e-01 4.94213015e-01 -1.03837043e-01 -1.15872145e+00 7.22239077e-01 3.13771278e-01 2.97838241e-01 -8.78334761e-01 9.76028204e-01 6.28230333e-01 -9.86863077e-01 -1.51994422e-01 -6.68016016e-01 -2.64798343e-01 1.72159538e-01 6.45735145e-01 -8.09340715e-01 4.50568944e-01 3.94093156e-01 6.91587031e-01 -3.97011846e-01 6.90499306e-01 -7.98401117e-01 6.64071262e-01 -1.60978902e-02 -1.63784742e-01 5.88662088e-01 2.86483079e-01 4.74017411e-01 1.09957707e+00 5.02047300e-01 -1.18912123e-01 -2.40513101e-01 1.28505898e+00 -1.19422056e-01 1.63250789e-01 -3.22543889e-01 6.18131496e-02 3.13828975e-01 1.20711172e+00 -5.77835441e-01 -4.99393374e-01 -2.98314363e-01 1.04211569e+00 4.94701564e-01 4.00064498e-01 -1.11994600e+00 -1.44477531e-01 5.24360538e-01 -9.15250182e-03 4.31919187e-01 1.52737752e-01 -9.14823115e-01 -1.33364797e+00 1.12895079e-01 -5.84599257e-01 2.24958166e-01 -1.03489912e+00 -1.11503613e+00 8.91583979e-01 -2.75649101e-01 -9.97670949e-01 -2.97259122e-01 -5.58928132e-01 -5.57258844e-01 1.06634820e+00 -1.87832141e+00 -1.17669868e+00 -3.31484824e-01 3.83984834e-01 8.19218695e-01 -2.11627483e-01 1.02349019e+00 -2.87315194e-02 -6.21324360e-01 4.60738093e-01 -1.85339317e-01 -2.00370312e-01 3.37212831e-01 -1.58261526e+00 6.48793995e-01 6.75110936e-01 3.86650056e-01 7.24383950e-01 9.01736081e-01 -4.93783951e-01 -8.50434124e-01 -8.68547022e-01 1.73681068e+00 -4.03394878e-01 4.93883163e-01 -2.67599881e-01 -1.23789072e+00 8.74185562e-01 1.50275216e-01 -2.18362510e-01 6.46734178e-01 5.23316026e-01 -2.25598201e-01 -1.15089327e-01 -9.08332109e-01 4.99304324e-01 7.29299188e-01 -2.03833625e-01 -1.01194108e+00 3.88334543e-01 1.15876150e+00 -3.72321129e-01 -3.81625652e-01 3.27126831e-01 4.45851892e-01 -1.00880408e+00 8.21679533e-01 -6.60232961e-01 9.48151946e-01 -2.16450348e-01 2.51695246e-01 -1.29440165e+00 -6.07689679e-01 -3.34675491e-01 -3.36883992e-01 1.23355663e+00 9.78056848e-01 -6.81822360e-01 9.15250778e-01 6.62835896e-01 -3.34223449e-01 -9.31949079e-01 -8.00784647e-01 -5.18222570e-01 -1.53395399e-01 -4.34186757e-01 8.66411924e-01 8.38094175e-01 1.08156510e-01 9.75545645e-01 -5.07709742e-01 3.51886868e-01 3.84553403e-01 1.48700893e-01 4.34463859e-01 -1.38458550e+00 -3.55405211e-01 -3.74866217e-01 -3.53360102e-02 -1.23266125e+00 3.35639864e-01 -9.43933308e-01 -7.98220560e-02 -1.89164841e+00 3.64705443e-01 -1.95265055e-01 -4.39655870e-01 8.96698177e-01 -4.25146788e-01 -1.36833444e-01 3.22752684e-01 3.52464765e-01 -3.08975428e-01 7.17618525e-01 1.17174828e+00 -5.75917624e-02 -2.26855099e-01 1.44604146e-01 -1.13827085e+00 9.98459637e-01 7.77224362e-01 -4.45710152e-01 -5.06696165e-01 -7.89792418e-01 1.62547112e-01 4.36797775e-02 3.69585514e-01 -8.68978322e-01 1.50398105e-01 7.44227096e-02 5.32964647e-01 -7.47325003e-01 1.71908036e-01 -1.01650417e+00 -7.08425045e-03 5.59175670e-01 -5.94776392e-01 -1.73427872e-02 2.63995767e-01 4.75219727e-01 -3.86771291e-01 -4.25603181e-01 5.98917544e-01 1.20326634e-02 -5.23393214e-01 3.48656625e-02 -1.12609148e-01 -3.28572243e-01 7.89371252e-01 -2.96543121e-01 -2.97619015e-01 -4.59827065e-01 -7.55725861e-01 2.84344226e-01 1.77956924e-01 6.67530239e-01 4.52212632e-01 -1.29500580e+00 -4.37961072e-01 1.80739880e-01 -1.33254407e-02 -6.56555733e-03 2.18321219e-01 4.97215778e-01 -1.68298066e-01 9.20532763e-01 -8.18221048e-02 -4.67023373e-01 -9.03867006e-01 5.73711693e-01 1.51454598e-01 -6.43733203e-01 -6.89303100e-01 7.19465077e-01 6.62016451e-01 -4.32670772e-01 2.41468966e-01 -6.92649066e-01 -4.60411072e-01 8.43836740e-02 5.04306376e-01 -1.82655621e-02 -9.88880843e-02 -5.32896221e-01 -2.82161504e-01 5.62034190e-01 -4.70453836e-02 1.90649509e-01 1.53382528e+00 -2.08473772e-01 5.34668118e-02 5.76062739e-01 8.76311839e-01 -2.14753300e-01 -1.04213190e+00 -2.86720306e-01 4.35290895e-02 -1.22648859e-02 5.30707352e-02 -1.09814119e+00 -1.09893382e+00 1.12571156e+00 1.85426027e-02 3.51060092e-01 1.03411090e+00 1.57726943e-01 1.06616592e+00 1.27460867e-01 -9.82154310e-02 -8.04746866e-01 4.14616242e-02 6.27648592e-01 1.03321338e+00 -1.12861621e+00 -2.06262067e-01 -3.94554973e-01 -7.64437437e-01 1.30043685e+00 5.93748033e-01 1.14246987e-01 6.31015524e-02 1.31897271e-01 -5.55468686e-02 -2.67795563e-01 -9.50355470e-01 -5.15686497e-02 5.43817222e-01 2.55490869e-01 7.16257691e-01 3.26178931e-02 -5.28046429e-01 1.25491440e+00 -4.42578405e-01 -1.03541285e-01 2.15013087e-01 1.82712495e-01 -8.06798041e-01 -9.47533846e-01 -1.19576067e-01 2.06156433e-01 -3.99161518e-01 -3.12978476e-01 -5.29452085e-01 8.75489652e-01 9.57283974e-02 8.25981200e-01 -4.43726592e-03 -2.51940250e-01 2.61071056e-01 2.84564406e-01 -2.94010550e-01 -7.21464872e-01 -6.42859280e-01 -6.19490594e-02 -1.11196756e-01 -4.96916682e-01 -2.34997883e-01 -3.07040244e-01 -1.62535548e+00 -1.91097990e-01 -4.30575639e-01 2.61046559e-01 7.73306310e-01 1.20963562e+00 5.10492861e-01 8.38371277e-01 4.12084401e-01 -7.73983538e-01 -3.79936308e-01 -1.05896008e+00 -4.16528612e-01 1.36933938e-01 3.95761460e-01 -5.12561023e-01 -5.66236138e-01 -9.50770676e-02]
[9.231389045715332, 6.094292640686035]
3426de2e-34fb-40da-84fb-37aec80dd68d
zero-shot-transfer-for-implicit-discourse
1907.12885
null
https://arxiv.org/abs/1907.12885v1
https://arxiv.org/pdf/1907.12885v1.pdf
Zero-shot transfer for implicit discourse relation classification
Automatically classifying the relation between sentences in a discourse is a challenging task, in particular when there is no overt expression of the relation. It becomes even more challenging by the fact that annotated training data exists only for a small number of languages, such as English and Chinese. We present a new system using zero-shot transfer learning for implicit discourse relation classification, where the only resource used for the target language is unannotated parallel text. This system is evaluated on the discourse-annotated TED-MDB parallel corpus, where it obtains good results for all seven languages using only English training data.
['Robert Östling', 'Murathan Kurfali']
2019-07-30
null
null
null
null
['implicit-discourse-relation-classification']
['natural-language-processing']
[ 3.55700813e-02 6.81402504e-01 -5.43074965e-01 -2.44222328e-01 -8.39799643e-01 -4.73185152e-01 9.48163211e-01 3.21123183e-01 -5.13021231e-01 1.24072862e+00 5.35913646e-01 -5.03379941e-01 3.07458609e-01 -7.83625424e-01 -3.61864567e-01 -2.98781127e-01 -6.10588454e-02 8.07726085e-01 5.18191040e-01 -7.95052826e-01 -1.09414998e-02 -2.31259912e-01 -1.11089242e+00 8.46945226e-01 7.52207935e-01 8.16022038e-01 2.63689369e-01 5.24914145e-01 -4.55223411e-01 1.81778955e+00 -8.93585324e-01 -4.33807105e-01 -3.93602192e-01 -6.53149605e-01 -1.73502803e+00 -7.02989288e-03 2.41313260e-02 -2.11536512e-01 -2.75242925e-01 7.73840725e-01 3.40938091e-01 2.43168786e-01 6.95076644e-01 -7.29893506e-01 -6.31977141e-01 7.74449646e-01 -7.19587132e-02 5.34555733e-01 5.35251975e-01 -3.55563045e-01 1.24065220e+00 -6.10625207e-01 1.09851146e+00 1.31904829e+00 3.52248669e-01 4.82173979e-01 -9.35887694e-01 -2.28654012e-01 -7.01768920e-02 8.33098531e-01 -1.06848633e+00 -4.52094555e-01 6.86379015e-01 -6.96526706e-01 1.59118211e+00 2.71467477e-01 1.86975494e-01 1.21574378e+00 -1.22490890e-01 8.22607815e-01 8.05588126e-01 -9.29735839e-01 1.47420987e-01 1.20283797e-01 5.44626176e-01 2.78085083e-01 -4.54192549e-01 -4.00781274e-01 -4.24175322e-01 1.15922898e-01 1.53843954e-01 -7.13209450e-01 -3.18051398e-01 1.25757605e-01 -1.00924897e+00 1.03358829e+00 2.30563074e-01 7.79517591e-01 4.85567078e-02 -6.10097110e-01 1.05443132e+00 6.76720738e-01 1.01331890e+00 5.92415452e-01 -4.56469238e-01 -4.99604732e-01 -3.43366563e-01 2.43670031e-01 1.12241828e+00 1.02246654e+00 4.73869503e-01 -3.15295637e-01 -1.64659813e-01 1.11199677e+00 -2.87619382e-01 3.73598561e-02 5.71617007e-01 -7.56393194e-01 1.00003004e+00 8.75068128e-01 1.46989390e-01 -8.04270089e-01 -3.16015542e-01 2.66725928e-01 -6.29988134e-01 -1.24183975e-01 5.98695993e-01 -4.07954395e-01 -9.37180370e-02 1.61963499e+00 3.81441236e-01 -2.76920140e-01 6.99494302e-01 7.92558670e-01 1.39045548e+00 9.27729905e-01 8.71822834e-02 -5.83334863e-01 1.46738446e+00 -1.17851341e+00 -1.38843083e+00 -1.90702483e-01 1.09532487e+00 -6.67786002e-01 9.82911289e-01 -6.53427616e-02 -8.07901680e-01 -3.55522364e-01 -1.10051811e+00 -5.24467409e-01 -4.38381225e-01 1.29499221e-02 4.54769492e-01 1.83043569e-01 -4.60992813e-01 3.08606774e-01 -5.66227138e-01 -4.21594650e-01 2.85760313e-01 -1.37887588e-02 -3.82419288e-01 1.13311581e-01 -1.68502784e+00 1.50275290e+00 7.68482089e-01 -1.57390878e-01 -3.57470006e-01 -2.79444098e-01 -1.18666911e+00 -7.69157708e-02 5.98231137e-01 2.50524282e-02 1.41300654e+00 -1.15752149e+00 -1.63303339e+00 1.22545791e+00 -6.12331107e-02 -6.85274899e-01 4.13870305e-01 -2.98079610e-01 -4.40712959e-01 -3.62543911e-02 6.62543848e-02 -1.32527379e-02 3.90737981e-01 -7.87367046e-01 -5.15042067e-01 -1.32263720e-01 4.78640616e-01 4.32671338e-01 -4.75730360e-01 4.88874614e-01 1.56450123e-02 -3.15068185e-01 -4.78409201e-01 -6.52760863e-01 1.41487256e-01 -5.22442520e-01 -4.06959444e-01 -1.09134698e+00 1.25084889e+00 -9.46627200e-01 1.37050200e+00 -2.10863638e+00 3.78122300e-01 -6.92695737e-01 -2.59730034e-02 5.48967481e-01 1.63327605e-01 4.71000552e-01 -9.79389027e-02 -8.76308307e-02 -1.19874708e-01 2.44218903e-03 -3.35415065e-01 5.18388748e-01 -3.54403317e-01 3.43392789e-01 5.07896602e-01 6.51326358e-01 -1.16114533e+00 -8.59347343e-01 3.19103077e-02 -1.11391105e-01 -6.58833459e-02 7.11539447e-01 -4.62437093e-01 5.29579580e-01 -5.12294710e-01 1.63097888e-01 4.05501612e-02 -3.53345096e-01 6.18312061e-01 -5.11813499e-02 -2.63182759e-01 1.04409015e+00 -6.37135446e-01 1.39054489e+00 -4.82748389e-01 1.21838772e+00 -1.32374421e-01 -1.35124588e+00 8.27590823e-01 9.06333148e-01 1.57280102e-01 -6.24956667e-01 4.40607071e-01 9.23364162e-02 4.10930276e-01 -1.04091907e+00 5.01547456e-01 -3.38137597e-01 -3.79421353e-01 5.16154051e-01 2.14739516e-01 -7.96025619e-02 5.20277321e-01 1.67244915e-02 1.07954335e+00 -2.67540812e-02 1.00018597e+00 -3.11515719e-01 7.68057525e-01 4.64143842e-01 5.22081792e-01 -3.26718241e-02 -1.30006343e-01 2.93413281e-01 9.29082870e-01 -3.99218380e-01 -1.01634789e+00 -5.04671335e-01 -2.64971942e-01 1.15073562e+00 1.15895480e-01 -5.42238235e-01 -6.22768164e-01 -7.89678693e-01 -4.52903241e-01 9.17326510e-01 -7.00291991e-01 2.44873807e-01 -9.81531918e-01 -4.62149054e-01 4.36559647e-01 3.73801291e-01 5.96189797e-01 -1.36733758e+00 -6.32301688e-01 1.25731125e-01 -7.04859614e-01 -1.37797618e+00 -1.47475287e-01 2.10466415e-01 -3.67617786e-01 -1.35965204e+00 -3.46449077e-01 -1.12500226e+00 2.66764522e-01 -2.43716002e-01 1.29190075e+00 7.20358565e-02 1.70023933e-01 -1.69185579e-01 -6.95112467e-01 -3.54542345e-01 -8.59485388e-01 2.82273471e-01 -3.86631638e-01 -3.36330712e-01 6.05470002e-01 1.10353470e-01 2.47270733e-01 8.90371501e-02 -2.56673634e-01 1.98991433e-01 -1.74027726e-01 1.22675359e+00 1.04878031e-01 -1.34127125e-01 7.58538663e-01 -1.28873396e+00 8.83488536e-01 -6.39188886e-01 -1.50357515e-01 3.12227130e-01 1.26308158e-01 -1.60642609e-01 5.14550745e-01 -6.74320221e-01 -1.49630785e+00 -5.16971886e-01 -1.70836896e-01 1.62833393e-01 3.70129831e-02 7.31435239e-01 -4.27896172e-01 7.18568921e-01 9.52922225e-01 -4.26603734e-01 2.54237484e-02 -4.83274311e-01 3.49010795e-01 1.16002548e+00 3.29946995e-01 -4.87442195e-01 4.05485295e-02 -7.07321567e-03 -3.84576350e-01 -1.44346619e+00 -1.23759305e+00 -4.90702361e-01 -8.97662461e-01 -2.69143552e-01 1.26166558e+00 -7.65720427e-01 -6.20627820e-01 2.01856837e-01 -1.67407537e+00 -7.47797251e-01 -3.56587976e-01 4.90698427e-01 -5.73594689e-01 2.23551780e-01 -1.06492531e+00 -6.90804780e-01 -2.81860739e-01 -8.88241053e-01 3.57609749e-01 -1.45805299e-01 -8.14546585e-01 -1.11100852e+00 3.69733304e-01 4.06988591e-01 4.41280007e-03 2.21008107e-01 1.29023278e+00 -1.07627225e+00 -9.20734257e-02 9.67269838e-02 -1.62159532e-01 4.66492534e-01 3.98694158e-01 -2.13711724e-01 -8.26605439e-01 -3.87715660e-02 2.09117249e-01 -1.14215577e+00 2.90232331e-01 -2.57186498e-02 5.88915586e-01 -4.14377242e-01 -2.31365114e-01 -2.48522863e-01 8.96157980e-01 3.12335968e-01 5.75770795e-01 4.05613244e-01 5.50185978e-01 1.06575024e+00 9.04766798e-01 5.51239401e-02 4.11157161e-01 8.15725207e-01 -1.29906803e-01 4.94034169e-03 -3.36637944e-01 2.24378169e-01 1.22557186e-01 1.39279866e+00 -2.19888151e-01 -3.65288556e-01 -1.13071132e+00 6.64300382e-01 -2.01829123e+00 -1.05085802e+00 -3.25848520e-01 1.71678138e+00 1.55741286e+00 1.90419003e-01 -9.02404562e-02 1.02234505e-01 6.79863214e-01 3.37735564e-01 -2.01179102e-01 -4.38939542e-01 -3.05646002e-01 6.53091893e-02 -2.44709864e-01 7.68186033e-01 -1.43815720e+00 1.05650008e+00 6.04068089e+00 8.31837237e-01 -9.99357760e-01 7.42206693e-01 5.55737376e-01 2.66811818e-01 2.33438879e-01 -9.93535295e-02 -5.91411352e-01 2.21419081e-01 1.00754666e+00 -4.12964433e-01 -1.87218681e-01 9.45979774e-01 -2.35690206e-01 -2.70605475e-01 -1.37522805e+00 6.33077383e-01 1.81492612e-01 -1.52423859e+00 -3.78420651e-01 -4.40590322e-01 4.98242885e-01 -1.92466695e-02 -5.86295187e-01 8.47653329e-01 1.78720266e-01 -1.00790334e+00 4.83503699e-01 -2.18574658e-01 9.24434423e-01 -4.50535268e-01 1.01296151e+00 6.45695448e-01 -7.95154035e-01 3.28617007e-01 -2.88479805e-01 -6.06518328e-01 2.41513729e-01 -1.96369682e-02 -1.08535063e+00 3.78829658e-01 5.13001382e-01 7.57689118e-01 -1.34460628e-01 2.49409690e-01 -5.95255792e-01 7.12486029e-01 -7.76906237e-02 -4.18219358e-01 1.01993211e-01 2.50385236e-02 5.56394696e-01 1.30643213e+00 -1.21208042e-01 5.41823328e-01 3.00871015e-01 4.18762594e-01 -2.08138257e-01 4.60424840e-01 -7.27569044e-01 6.49993420e-02 3.29559147e-01 9.91316915e-01 -4.76081342e-01 -5.26376367e-01 -6.85238957e-01 6.51386499e-01 1.01069248e+00 2.31611058e-01 -4.79227751e-01 -1.64031103e-01 1.50001556e-01 4.87995654e-04 1.22806720e-01 -1.76756561e-01 -3.37379128e-02 -1.08034503e+00 -1.63905695e-01 -8.48141015e-01 5.92038453e-01 -5.38647354e-01 -1.55346191e+00 8.93967390e-01 2.68087775e-01 -1.14449060e+00 -7.13875890e-01 -7.44663477e-01 -5.91203034e-01 9.69505966e-01 -1.18746078e+00 -1.12626946e+00 -1.41773224e-02 3.93775284e-01 1.01416218e+00 -3.89518529e-01 1.26372027e+00 3.45851570e-01 -4.66354281e-01 2.37396896e-01 -1.51357397e-01 6.45774901e-01 9.16837394e-01 -1.30963504e+00 -2.30093952e-02 4.07009572e-01 -7.88734406e-02 3.23434770e-01 8.47885251e-01 -5.49251795e-01 -8.84884775e-01 -7.99138725e-01 1.35696208e+00 -4.41970646e-01 1.10502219e+00 -4.89838600e-01 -1.27860153e+00 9.07898903e-01 1.05966163e+00 -1.11269832e-01 1.01020670e+00 6.96272612e-01 -1.43054143e-01 2.84744173e-01 -7.24474669e-01 5.62970698e-01 6.07354522e-01 -6.94787085e-01 -1.43507719e+00 7.53472209e-01 9.50109243e-01 -8.82603228e-01 -1.02176058e+00 3.55068803e-01 -1.31190091e-01 -3.11596990e-01 5.89954138e-01 -9.40506876e-01 9.98112500e-01 -1.11053877e-01 -2.36235797e-01 -1.14068830e+00 1.93490371e-01 -3.40041667e-01 -3.23748022e-01 1.48391449e+00 7.30818927e-01 -2.87656456e-01 3.53759944e-01 5.16998231e-01 -1.03108615e-01 -7.29288697e-01 -1.41351628e+00 -7.81736016e-01 3.47241133e-01 -1.24175355e-01 1.09553084e-01 1.43046331e+00 9.33735013e-01 1.39723051e+00 -3.34884197e-01 -4.83175367e-01 -1.21814542e-01 3.35928887e-01 6.64326668e-01 -8.90425384e-01 -2.53129035e-01 -1.24833599e-01 -2.57613719e-01 -8.99905384e-01 8.06752026e-01 -1.00958276e+00 1.87746435e-01 -1.38829780e+00 1.23216502e-01 -4.42168295e-01 3.16542685e-01 4.07373965e-01 -1.54941320e-01 -1.73708275e-01 -1.21513516e-01 2.21431389e-01 -7.32284188e-01 9.20776308e-01 1.20439506e+00 -3.96840245e-01 -1.06850430e-01 -1.62760600e-01 2.95724161e-02 8.46070707e-01 5.24994135e-01 -2.35222369e-01 -4.76087004e-01 -3.64127278e-01 -3.15178111e-02 4.02148068e-01 -1.64173827e-01 -4.08142984e-01 2.05365699e-02 -1.59039512e-01 -1.98896661e-01 -6.40897512e-01 3.66387874e-01 -5.22012174e-01 -3.36996108e-01 1.94518238e-01 -6.42923176e-01 -1.83671162e-01 2.02540308e-01 2.07248822e-01 -6.50402427e-01 -5.48234701e-01 6.13007963e-01 -7.36986324e-02 -9.10295367e-01 -4.27204102e-01 -5.61298192e-01 6.46170199e-01 1.15892935e+00 3.98812801e-01 -8.14508200e-01 -3.21982384e-01 -6.59095824e-01 1.99441209e-01 7.36230835e-02 5.50311744e-01 2.36087173e-01 -1.20695341e+00 -9.32505667e-01 -3.48138630e-01 2.44091228e-01 1.98938027e-01 1.07973762e-01 6.63639843e-01 -3.81777823e-01 3.70039761e-01 -1.14240900e-01 -4.65847343e-01 -1.69817710e+00 7.70506799e-01 1.66700751e-01 -5.99246621e-01 -9.04297590e-01 6.15440130e-01 -3.21133528e-03 -3.35665196e-01 4.51165959e-02 -1.57730639e-01 -8.21392953e-01 4.71564680e-01 6.77630365e-01 6.34928420e-02 -5.61088882e-02 -1.06757796e+00 -2.92736799e-01 -1.95711225e-01 -1.85029954e-01 2.34385058e-02 1.31435668e+00 -1.56489283e-01 -4.02611732e-01 1.07163608e+00 1.23902881e+00 -1.26226738e-01 -7.64980733e-01 -4.32323456e-01 5.66408277e-01 -9.52987596e-02 -2.51765758e-01 -3.12509745e-01 -3.53520602e-01 9.69769359e-01 4.93358225e-02 5.63190043e-01 6.23723209e-01 4.04129416e-01 6.56091154e-01 6.36602044e-01 2.57602781e-01 -1.32231164e+00 1.24943815e-01 1.18183410e+00 1.20584631e+00 -1.46651685e+00 1.43177047e-01 -6.61378324e-01 -7.18625724e-01 9.81463075e-01 7.75271833e-01 -7.71807209e-02 6.28564000e-01 3.01753074e-01 1.46238834e-01 -2.19307244e-01 -8.42782259e-01 -1.51315555e-01 1.04048684e-01 5.70568323e-01 1.19630945e+00 7.35364780e-02 -6.99980021e-01 7.61058152e-01 -2.56052434e-01 -3.19118232e-01 5.46426892e-01 9.60852563e-01 -1.30670637e-01 -1.26275623e+00 -7.74227008e-02 2.94243187e-01 -5.02315760e-01 -7.31351078e-02 -4.06707883e-01 1.12959218e+00 -7.13918284e-02 1.10301900e+00 4.07661498e-01 6.19169371e-03 3.10948163e-01 8.82736668e-02 3.97834301e-01 -1.04225719e+00 -5.42719662e-01 1.65046356e-03 1.20500839e+00 -1.98752463e-01 -1.01255786e+00 -6.09169245e-01 -1.36659741e+00 -1.78922877e-01 -3.43362749e-01 4.07420576e-01 -6.26592934e-02 1.39437532e+00 -1.81973785e-01 7.71304429e-01 2.54631519e-01 -4.77502555e-01 -2.23995999e-01 -1.53693640e+00 -2.96304017e-01 6.57039225e-01 2.65964329e-01 -8.15329373e-01 1.11884959e-02 3.96337688e-01]
[10.813067436218262, 9.320127487182617]
d126cba8-6818-4675-89e4-d0f62b2c947d
applying-the-decisiveness-and-robustness
2006.00058
null
https://arxiv.org/abs/2006.00058v1
https://arxiv.org/pdf/2006.00058v1.pdf
Applying the Decisiveness and Robustness Metrics to Convolutional Neural Networks
We review three recently-proposed classifier quality metrics and consider their suitability for large-scale classification challenges such as applying convolutional neural networks to the 1000-class ImageNet dataset. These metrics, referred to as the "geometric accuracy," "decisiveness," and "robustness," are based on the generalized mean ($\rho$ equals 0, 1, and -2/3, respectively) of the classifier's self-reported and measured probabilities of correct classification. We also propose some minor clarifications to standardize the metric definitions. With these updates, we show some examples of calculating the metrics using deep convolutional neural networks (AlexNet and DenseNet) acting on large datasets (the German Traffic Sign Recognition Benchmark and ImageNet).
['Eduardo A. Barrera', 'Kenric P. Nelson', 'Christopher A. George']
2020-05-29
null
null
null
null
['traffic-sign-recognition']
['computer-vision']
[ 2.70724714e-01 -7.56773576e-02 -2.75902063e-01 -7.86152542e-01 -4.15668368e-01 -3.62499595e-01 5.98271847e-01 -1.39666066e-01 -8.52386177e-01 8.21871221e-01 -1.50954530e-01 -4.39620912e-01 -6.18823469e-01 -6.93259358e-01 -4.60159391e-01 -7.20741034e-01 -1.77302912e-01 1.05819285e-01 1.15539446e-01 -1.61746174e-01 4.15332764e-01 6.06458127e-01 -2.03249502e+00 3.33431721e-01 8.01158488e-01 1.85303020e+00 -6.45180464e-01 7.73921430e-01 2.25749642e-01 9.57192242e-01 -8.83634269e-01 -9.39334989e-01 2.53635675e-01 -1.57746933e-02 -7.38375068e-01 -3.39006126e-01 1.25222170e+00 -2.38061860e-01 -4.94124144e-01 1.01794899e+00 3.25115085e-01 -5.42812161e-02 8.71122837e-01 -1.61802661e+00 -6.44040048e-01 3.61417562e-01 1.08355008e-01 6.01452470e-01 -7.87741691e-02 4.53484982e-01 1.14854324e+00 -4.67746496e-01 5.48609972e-01 1.06725240e+00 1.12390947e+00 6.18403375e-01 -8.97963762e-01 -7.55554438e-01 3.39434966e-02 5.41409314e-01 -1.20649254e+00 -4.16586876e-01 3.15319210e-01 -5.75799823e-01 8.90383899e-01 3.98698270e-01 4.13349509e-01 1.21256649e+00 1.71463057e-01 7.84097552e-01 1.22486782e+00 -3.14390898e-01 1.56730302e-02 1.53260455e-01 8.39832127e-01 8.78037810e-01 6.29474878e-01 5.75617671e-01 -1.11189634e-01 -4.77844477e-02 4.25146341e-01 -4.29920614e-01 2.08314508e-02 -2.06409171e-01 -1.21525621e+00 6.69587076e-01 6.04757905e-01 2.28024974e-01 -2.03163520e-01 3.14822257e-01 6.90267980e-01 6.64905965e-01 1.59929857e-01 4.15542513e-01 -6.20425999e-01 -2.21630216e-01 -5.91048062e-01 3.09136838e-01 8.52185309e-01 8.02563608e-01 5.08010745e-01 2.59094983e-01 -3.91109318e-01 8.99220228e-01 1.30355224e-01 5.76609790e-01 5.21513879e-01 -7.90339768e-01 4.20390695e-01 4.80172068e-01 -7.58778304e-02 -9.88132715e-01 -5.15774965e-01 -7.51134992e-01 -9.16851103e-01 6.05477035e-01 8.99500549e-01 1.20815173e-01 -1.25175035e+00 1.34913492e+00 -3.86223733e-01 1.35152280e-01 -1.49171576e-02 7.00482726e-01 1.37270486e+00 -1.82947949e-01 4.14845049e-01 3.25685710e-01 1.30541599e+00 -7.65133560e-01 -2.79538006e-01 -3.01037785e-02 5.29854476e-01 -4.24256176e-01 9.17848527e-01 4.54370767e-01 -7.42285252e-01 -9.08757687e-01 -1.30337560e+00 2.87306547e-01 -8.97982538e-01 4.47736830e-01 5.66195071e-01 1.16802156e+00 -8.87311041e-01 9.48572040e-01 -4.55929488e-01 -2.89419472e-01 1.04322684e+00 3.30396503e-01 -5.15415192e-01 -8.36616731e-04 -1.07849264e+00 1.21495152e+00 2.28104338e-01 2.05033302e-01 -8.65478635e-01 -6.48343027e-01 -5.57038665e-01 1.32631313e-03 -3.48204941e-01 -3.97778630e-01 1.15954888e+00 -9.41736877e-01 -1.28917360e+00 1.30596375e+00 3.11281741e-01 -4.98035401e-01 8.60340059e-01 -7.89111480e-02 -8.44849229e-01 -1.75830618e-01 -1.82435319e-01 7.51599669e-01 6.62845194e-01 -9.51430798e-01 -9.91393924e-01 -1.44824803e-01 9.43801403e-02 -4.89871085e-01 -1.72968552e-01 1.35310128e-01 1.93809777e-01 -5.25445879e-01 6.47458658e-02 -7.53564060e-01 -1.37850657e-01 3.58165890e-01 -3.03708822e-01 -3.46056253e-01 8.58537257e-01 -5.02151966e-01 9.85742807e-01 -2.23648500e+00 -5.70764184e-01 6.01942778e-01 2.97721148e-01 8.06374073e-01 -4.71212536e-01 -4.55523819e-01 -3.70709747e-01 2.32020572e-01 -1.83453143e-01 1.43516362e-01 1.23641677e-01 2.30548397e-01 -6.59601316e-02 5.20934105e-01 4.88784879e-01 1.12762737e+00 -8.48868549e-01 -1.70390069e-01 3.68818432e-01 3.72501135e-01 -2.03643173e-01 -4.67084914e-01 4.41405684e-01 -5.28863072e-03 -1.47659063e-01 8.37277651e-01 7.45788157e-01 -1.90450415e-01 -2.11510569e-01 -5.83339274e-01 2.75808275e-01 1.47322431e-01 -1.12046826e+00 8.04893315e-01 -2.51355350e-01 1.23487842e+00 -3.90849620e-01 -1.04484665e+00 1.12395036e+00 2.02026777e-02 2.15630785e-01 -7.41866648e-01 4.59867418e-01 5.39881051e-01 2.92337865e-01 -3.48750710e-01 2.54678220e-01 7.96422586e-02 7.60393366e-02 5.12624420e-02 3.61476094e-01 2.32258394e-01 4.03723180e-01 -2.52459913e-01 1.11272323e+00 -2.96557724e-01 2.85809308e-01 -4.06821012e-01 6.98640406e-01 -2.38987908e-01 3.03046882e-01 1.08767211e+00 -1.00449038e+00 5.12601733e-01 7.65514970e-01 -7.64822185e-01 -9.47835326e-01 -1.30007172e+00 -5.57498813e-01 8.52871537e-01 -1.16994917e-01 -1.87724069e-01 -6.00545943e-01 -1.17633379e+00 4.57801580e-01 4.81456131e-01 -1.08123541e+00 -3.18213701e-01 -5.89265823e-01 -8.84253979e-01 1.24341238e+00 1.00059128e+00 9.01946425e-01 -1.04036343e+00 -6.33042693e-01 -8.68078396e-02 2.88527161e-02 -1.12582469e+00 2.11605236e-01 1.42328024e-01 -7.60002673e-01 -1.56516397e+00 -4.78044510e-01 -7.14424789e-01 4.84154910e-01 -2.25808963e-01 1.23993087e+00 3.23644876e-01 -3.31412017e-01 1.27129421e-01 -3.82159472e-01 -3.54546100e-01 -5.15729785e-01 8.63592923e-02 9.64264423e-02 7.64748896e-04 4.88901049e-01 -2.55845189e-01 -5.36343634e-01 7.59132743e-01 -7.10852623e-01 -5.42826474e-01 5.35324931e-01 1.06575453e+00 -3.47361527e-02 -4.04723227e-01 5.24789333e-01 -5.25867224e-01 7.27054834e-01 -1.09285833e-02 -5.70636094e-01 3.44698220e-01 -7.93454289e-01 -1.34428829e-01 4.22765255e-01 -4.57167774e-01 -6.14331663e-01 -3.48456472e-01 -3.74194682e-01 -3.13580096e-01 -5.29900551e-01 1.13554895e-01 1.72023982e-01 -5.51743031e-01 1.04040396e+00 -1.38857812e-01 1.38865958e-03 -1.85503975e-01 4.28990960e-01 8.12656045e-01 6.64062619e-01 -4.15223807e-01 5.88004589e-01 3.93140376e-01 1.09085992e-01 -5.62184691e-01 -5.87287247e-01 -1.20896801e-01 -7.58543015e-01 -3.53212833e-01 5.43506026e-01 -4.66232717e-01 -1.13757026e+00 9.17875350e-01 -8.92768204e-01 -1.08724356e-01 -5.43138683e-01 5.01497567e-01 -5.38362265e-01 1.94736138e-01 -3.65305066e-01 -5.07303417e-01 -3.26439083e-01 -1.25697243e+00 7.12803543e-01 3.61139476e-01 -2.01693073e-01 -9.02066886e-01 -3.19982350e-01 2.94684500e-01 8.87630463e-01 5.26054740e-01 7.66134679e-01 -7.97193885e-01 -1.50800422e-01 -7.48325348e-01 -8.81915689e-01 1.12195730e+00 7.15571642e-02 2.35116899e-01 -1.43416989e+00 9.94237838e-04 -6.97153866e-01 -4.53522235e-01 1.24123430e+00 4.30421740e-01 1.28466511e+00 -2.24780098e-01 -3.28208417e-01 6.16154373e-01 1.17459607e+00 1.15588792e-01 9.89132524e-01 4.93791282e-01 3.82459462e-01 3.39108080e-01 2.32807040e-01 2.64588565e-01 9.96584147e-02 5.00672519e-01 3.50658029e-01 1.11355774e-01 -4.13461745e-01 -4.18159962e-02 -5.75628481e-04 2.75653452e-01 -5.16402602e-01 -1.31725818e-01 -8.46989989e-01 2.53697217e-01 -1.45435023e+00 -1.08620322e+00 -1.59345970e-01 1.93803191e+00 1.85951546e-01 3.98768246e-01 1.79033399e-01 6.83432043e-01 6.43422604e-01 1.96447358e-01 -4.03918743e-01 -4.79274005e-01 -5.47711790e-01 6.57916546e-01 8.72943521e-01 2.39755422e-01 -1.45875132e+00 9.27627683e-01 7.95123911e+00 9.32766557e-01 -1.15825713e+00 8.94264728e-02 9.35164690e-01 2.17334971e-01 4.07576203e-01 -3.74797702e-01 -7.81585455e-01 4.18552816e-01 1.08399856e+00 2.81999558e-01 -1.47202648e-02 1.06904948e+00 -2.49357998e-01 1.16386235e-01 -1.04344463e+00 1.05662572e+00 -7.52817094e-02 -1.47782207e+00 1.43824890e-01 -1.39471546e-01 6.47344351e-01 3.64809871e-01 3.57044965e-01 4.28992867e-01 3.71521056e-01 -1.17750049e+00 8.79806221e-01 5.83197474e-01 1.21081674e+00 -3.84096533e-01 1.00466239e+00 -2.41442010e-01 -1.02557313e+00 -4.84833866e-01 -1.94263130e-01 9.62791368e-02 -9.29915756e-02 5.60846567e-01 -3.45418423e-01 3.59773636e-01 9.06196237e-01 7.94580400e-01 -8.90460074e-01 1.43315351e+00 -2.96191633e-01 6.19619370e-01 -2.34362975e-01 -2.80910581e-01 3.27459574e-01 1.58119798e-01 3.70188594e-01 1.35165989e+00 1.92893017e-02 -2.55149305e-01 -4.84165937e-01 8.05686891e-01 -4.99768555e-02 -1.76378533e-01 -2.73454636e-01 1.33947432e-01 6.13231584e-02 1.22278309e+00 -7.05356538e-01 -5.82542181e-01 -2.23372683e-01 5.62678933e-01 -2.35652775e-02 2.50724643e-01 -8.93301785e-01 -8.89970601e-01 1.16418779e+00 -2.52547890e-01 1.23455480e-01 3.41950916e-02 -7.57504582e-01 -1.02965844e+00 1.12672135e-01 -9.15770650e-01 5.32238901e-01 -5.96850753e-01 -1.58230889e+00 6.78623021e-01 -9.23529491e-02 -1.53959751e+00 1.82127520e-01 -1.70184445e+00 -7.05834210e-01 5.24477780e-01 -1.51075697e+00 -8.99198234e-01 -6.99943364e-01 3.74662966e-01 2.18709540e-02 -7.15595901e-01 6.44275010e-01 6.92084312e-01 -4.69350964e-01 1.27193010e+00 2.03546491e-02 7.42429733e-01 3.54519337e-01 -8.73749375e-01 6.87726080e-01 3.79619628e-01 -1.47977844e-01 2.04220101e-01 3.93963039e-01 -5.11499308e-02 -4.71080542e-01 -9.66583252e-01 9.14605737e-01 -7.58136094e-01 5.79490840e-01 7.98715726e-02 -4.76020873e-01 5.45404017e-01 -4.17714745e-01 3.95345032e-01 5.45615613e-01 9.48388071e-04 -8.87858391e-01 -5.70540071e-01 -1.58320165e+00 3.09018165e-01 1.44820201e+00 -2.30606586e-01 -4.37401623e-01 2.72010863e-01 -2.68303696e-02 -2.02292323e-01 -9.19820905e-01 8.49587977e-01 1.17251146e+00 -1.09385121e+00 1.07988250e+00 -1.36790276e+00 2.02321157e-01 -7.13174231e-03 -4.20428425e-01 -1.18166912e+00 -4.24719900e-01 -5.13822399e-02 1.42577097e-01 6.17212951e-01 6.37871027e-01 -1.05649567e+00 9.11817968e-01 4.10969049e-01 -8.92957821e-02 -6.86829031e-01 -1.38302064e+00 -1.07259250e+00 3.32743555e-01 -7.79108465e-01 6.76111221e-01 9.17916477e-01 -2.78930813e-01 -3.72954011e-01 -1.62322372e-02 -1.90795437e-01 5.79105914e-01 -6.56028807e-01 6.88912809e-01 -1.71421182e+00 3.19346905e-01 -1.30652392e+00 -1.45772338e+00 -6.43984616e-01 8.29379335e-02 -9.08198655e-01 -2.22849891e-01 -1.04695320e+00 -6.82922527e-02 -4.93115664e-01 -9.44458306e-01 4.77760375e-01 2.72544771e-01 5.73148966e-01 2.07525641e-01 -5.32739908e-02 -4.97389287e-01 2.30521202e-01 9.48143959e-01 -4.78897810e-01 2.99338728e-01 2.28917301e-01 -5.69452345e-01 6.89630151e-01 6.34152174e-01 -2.04311028e-01 1.50606826e-01 -2.36918524e-01 -3.09278723e-02 -6.99324906e-01 8.37333202e-01 -1.71248722e+00 -6.83268458e-02 1.68294720e-02 5.26213884e-01 -1.79167315e-01 1.38331085e-01 -6.15754902e-01 -2.92441249e-01 7.73264587e-01 -6.67792618e-01 -3.45836021e-02 1.57175735e-01 2.82939672e-01 -3.30121875e-01 -1.71856061e-01 1.29317474e+00 1.45118713e-01 -8.73551130e-01 3.08592290e-01 -3.22345167e-01 5.58811054e-03 8.75758708e-01 -5.69799304e-01 -7.02017844e-01 -1.46100983e-01 -8.20176780e-01 1.17183030e-02 -1.78710029e-01 8.62286985e-01 6.73143387e-01 -1.61499321e+00 -7.86589146e-01 3.86669040e-01 5.92995107e-01 -8.67694676e-01 7.71730149e-04 8.23411465e-01 -8.21646512e-01 6.20732009e-01 -6.85467601e-01 -7.36622095e-01 -1.27907646e+00 -2.01018415e-02 8.84943485e-01 -1.38494238e-01 -1.14547424e-01 8.92566383e-01 -2.86248475e-01 -5.36959529e-01 4.54857022e-01 -8.93628240e-01 -2.20275223e-01 9.25751925e-02 4.14180785e-01 8.61280203e-01 4.17529315e-01 -6.96764827e-01 -6.56613529e-01 6.63452625e-01 -1.04885986e-02 1.55847356e-01 1.20821416e+00 5.19151151e-01 2.08999738e-01 -3.07796244e-02 1.25370336e+00 -7.70320833e-01 -9.12183821e-01 -2.37740219e-01 2.46706977e-01 -5.36430180e-01 -1.41015232e-01 -1.06853461e+00 -1.38074756e+00 9.05913115e-01 1.30605078e+00 1.84585527e-01 7.97079742e-01 -2.84531564e-01 4.51780975e-01 8.71429861e-01 1.61327839e-01 -1.24287939e+00 -9.70251411e-02 8.17238808e-01 7.93286443e-01 -1.34854424e+00 -2.55056053e-01 -3.45469788e-02 -5.10326803e-01 1.27341664e+00 8.23407590e-01 -1.87469155e-01 1.10658693e+00 -7.97872469e-02 3.32372844e-01 5.49202785e-02 -3.29813272e-01 -3.93278271e-01 5.21716893e-01 8.27668846e-01 3.23398739e-01 1.92445636e-01 -3.04297984e-01 5.61958253e-01 -3.54985029e-01 3.34216326e-01 2.31514379e-01 6.62239552e-01 -4.47604448e-01 -6.24260783e-01 -1.34802461e-01 9.95620012e-01 -1.26535490e-01 1.76358029e-01 -3.31995636e-01 8.97831976e-01 4.83441830e-01 8.89040232e-01 2.93957174e-01 -9.11391318e-01 8.07359636e-01 1.30680636e-01 4.23931658e-01 8.15222114e-02 -5.04615843e-01 -9.94776249e-01 3.81444514e-01 -6.19666874e-01 -3.92607659e-01 -4.96765137e-01 -6.68042183e-01 -5.47700107e-01 -3.98858935e-01 -3.36025059e-01 7.38884449e-01 9.94573355e-01 2.45084897e-01 2.58817017e-01 2.66369849e-01 -5.94516456e-01 -7.80854583e-01 -1.22820282e+00 -4.35278744e-01 7.16103733e-01 4.06555265e-01 -9.41419303e-01 -5.27576447e-01 -2.65312612e-01]
[8.014339447021484, -0.7650129795074463]
41e86e31-1851-49fc-9bda-e76264af1730
machine-learning-and-chord-based-feature
1902.03283
null
http://arxiv.org/abs/1902.03283v1
http://arxiv.org/pdf/1902.03283v1.pdf
Machine learning and chord based feature engineering for genre prediction in popular Brazilian music
Music genre can be hard to describe: many factors are involved, such as style, music technique, and historical context. Some genres even have overlapping characteristics. Looking for a better understanding of how music genres are related to musical harmonic structures, we gathered data about the music chords for thousands of popular Brazilian songs. Here, 'popular' does not only refer to the genre named MPB (Brazilian Popular Music) but to nine different genres that were considered particular to the Brazilian case. The main goals of the present work are to extract and engineer harmonically related features from chords data and to use it to classify popular Brazilian music genres towards establishing a connection between harmonic relationships and Brazilian genres. We also emphasize the generalization of the method for obtaining the data, allowing for the replication and direct extension of this work. Our final model is a combination of multiple classification trees, also known as the random forest model. We found that features extracted from harmonic elements can satisfactorily predict music genre for the Brazilian case, as well as features obtained from the Spotify API. The variables considered in this work also give an intuition about how they relate to the genres.
['Walmes M. Zeviani', 'Bruna D. Wundervald']
2019-02-08
null
null
null
null
['music-genre-recognition']
['music']
[-8.60552117e-02 -3.64635706e-01 -1.79152712e-01 -1.64360017e-01 -5.98083258e-01 -1.06695914e+00 6.51559472e-01 2.45258734e-01 -2.02471763e-01 6.94066048e-01 3.77798378e-01 -1.04521073e-01 -7.58615613e-01 -9.02604282e-01 -2.78175205e-01 -5.87409139e-01 -1.52461812e-01 5.28323054e-01 4.66801286e-01 -5.58828890e-01 7.70695686e-01 5.53766429e-01 -1.96505332e+00 3.60943198e-01 3.96299034e-01 8.28808188e-01 2.28267953e-01 7.38162637e-01 -1.54556006e-01 6.89983547e-01 -9.27965701e-01 -3.34385335e-01 8.21238849e-03 -5.22246242e-01 -1.07424927e+00 -3.15542668e-01 2.18646973e-01 2.24026248e-01 2.27321431e-01 5.39295018e-01 2.91419923e-01 8.45543519e-02 7.65958369e-01 -9.96983528e-01 -6.45987168e-02 1.07560241e+00 -2.47604638e-01 1.98353842e-01 5.88365912e-01 -3.42688739e-01 1.40598750e+00 -4.46109951e-01 7.53635824e-01 9.17510211e-01 8.86835933e-01 9.52071324e-02 -1.17661214e+00 -7.51040399e-01 -3.86808932e-01 6.80212438e-01 -1.51752651e+00 -3.62162560e-01 9.28217411e-01 -7.70764589e-01 5.74648559e-01 5.78768313e-01 1.04989004e+00 8.92221153e-01 4.58742976e-02 4.25405771e-01 1.26866984e+00 -6.90554321e-01 3.87179628e-02 1.60054296e-01 2.03831434e-01 1.26951501e-01 -1.35156494e-02 5.35855256e-02 -7.41319776e-01 -2.86344647e-01 6.33194864e-01 -5.07768214e-01 -2.54515558e-01 -9.15913805e-02 -1.21269166e+00 8.77130330e-01 -1.42075559e-02 9.58726406e-01 -1.04636885e-01 -1.91755235e-01 5.30526161e-01 3.78932416e-01 -7.52925053e-02 1.05124998e+00 -6.60412312e-01 -6.92636967e-01 -1.14001536e+00 5.46691954e-01 1.13347244e+00 4.73736167e-01 6.82250321e-01 -8.35894644e-02 2.84336120e-01 9.56258595e-01 1.44168094e-01 1.16757788e-01 6.81498528e-01 -1.05721498e+00 1.38292104e-01 3.97374719e-01 -1.00551367e-01 -1.12495589e+00 -5.41708291e-01 -3.90554041e-01 -2.60347277e-01 1.63193747e-01 1.00692284e+00 3.17089170e-01 -7.57880360e-02 1.70225132e+00 1.74370892e-02 -1.77737579e-01 -2.18900889e-01 6.08159959e-01 8.51773500e-01 3.73259723e-01 -3.21819007e-01 -4.47063416e-01 1.43597567e+00 -4.46083128e-01 -4.81507748e-01 4.38808799e-01 2.72953868e-01 -1.14037013e+00 1.25655138e+00 9.66230810e-01 -1.00824583e+00 -7.69152582e-01 -9.94880199e-01 8.42446685e-02 -3.42913359e-01 -4.44624238e-02 6.65406287e-01 6.97999835e-01 -4.80737507e-01 1.12951279e+00 -6.71977878e-01 -2.38362446e-01 -1.05171427e-01 2.25758016e-01 -8.83047283e-02 7.12125123e-01 -1.06341588e+00 7.98758745e-01 3.63190204e-01 -2.98692435e-01 -4.32507366e-01 -4.54279304e-01 -1.59455165e-01 -1.07631184e-01 3.96830529e-01 -2.88653344e-01 1.31422031e+00 -1.09767556e+00 -1.35221171e+00 9.17236269e-01 1.52672857e-01 -1.83780417e-01 1.54726148e-01 1.01777889e-01 -5.70014000e-01 1.60509199e-02 5.47503158e-02 -3.36120091e-02 5.28656662e-01 -8.99187803e-01 -7.56772220e-01 -4.28530067e-01 1.46844819e-01 -1.22964829e-01 -3.08153093e-01 3.45969498e-01 -2.42281765e-01 -9.49721277e-01 1.80061489e-01 -1.10353899e+00 1.71922192e-01 -5.89723706e-01 -2.72832602e-01 -5.48082650e-01 3.19207013e-01 -5.25862873e-01 1.82042921e+00 -2.03931046e+00 4.39851850e-01 3.78972620e-01 -2.35882159e-02 -2.68215597e-01 2.48293981e-01 7.10677981e-01 -1.97769254e-01 1.19785629e-01 1.43482655e-01 3.37266684e-01 9.27675739e-02 3.51430446e-01 -4.39932883e-01 1.87443331e-01 -2.95909345e-01 4.41331238e-01 -6.55849159e-01 -5.22805572e-01 -1.01466805e-01 2.81551331e-01 -4.48878855e-01 -1.17724761e-01 -5.91425598e-02 3.07332963e-01 -3.35711569e-01 5.43065906e-01 2.16221553e-03 2.15610445e-01 1.48043409e-01 -4.12227720e-01 -6.12110913e-01 8.57818305e-01 -1.41394675e+00 1.73744595e+00 -2.40837678e-01 6.71582997e-01 -4.08486307e-01 -7.99469531e-01 1.07748055e+00 3.60030204e-01 4.43775535e-01 -2.90303171e-01 1.09841563e-01 5.50688267e-01 5.25454938e-01 -3.92758727e-01 8.06493282e-01 -3.83852214e-01 -2.35167697e-01 3.23396385e-01 7.25060329e-02 -4.11720276e-01 5.76273143e-01 -3.43024790e-01 8.10692608e-01 4.60142046e-01 7.44500101e-01 -4.31146681e-01 5.55116355e-01 1.44750521e-01 4.18557823e-01 5.32127678e-01 2.00993404e-01 5.75486124e-01 5.93950331e-01 -3.97881597e-01 -8.16071451e-01 -1.02623773e+00 -4.81056958e-01 1.20052528e+00 -2.75127292e-01 -9.75188673e-01 -3.95058960e-01 -2.26053029e-01 -1.97560281e-01 4.84656870e-01 -5.74903488e-01 1.58417806e-01 -5.67442417e-01 -5.46841085e-01 7.97409236e-01 2.49217644e-01 -9.39336717e-02 -1.19549322e+00 -7.67703831e-01 3.06551844e-01 -3.81783366e-01 -4.83869433e-01 -6.22755587e-02 3.24770689e-01 -8.36088061e-01 -1.40849447e+00 -3.02917987e-01 -5.14908910e-01 -4.62812483e-01 -1.97973669e-01 1.35061824e+00 -1.02804184e-01 -2.84675419e-01 2.45860696e-01 -6.55315876e-01 -6.35106444e-01 -5.78753114e-01 6.07879996e-01 8.93072411e-02 -2.58184314e-01 2.64406472e-01 -9.51962650e-01 -2.44646087e-01 4.82579023e-01 -7.44661510e-01 -3.12386066e-01 3.01699907e-01 3.84932518e-01 5.64686298e-01 2.50369251e-01 4.17545289e-01 -7.30434239e-01 5.33933878e-01 -2.39580810e-01 -2.39073709e-01 -1.75177976e-02 -5.41965544e-01 -1.59283876e-01 6.18317723e-01 -6.24078214e-01 -3.36597294e-01 -1.31065130e-01 -3.27418745e-01 -4.73579280e-02 -2.79810607e-01 6.30363762e-01 -1.13655858e-01 5.11201061e-02 8.29278648e-01 9.06395242e-02 -3.64185929e-01 -8.10147107e-01 5.26992157e-02 8.42049241e-01 6.41974509e-01 -8.84227216e-01 8.14036667e-01 1.89042076e-01 5.84811065e-03 -1.05514669e+00 -6.89243376e-01 -4.58846211e-01 -7.17493892e-01 -4.50330406e-01 6.51093245e-01 -3.41436148e-01 -8.50226283e-01 1.87378064e-01 -8.57900262e-01 8.50965977e-02 -6.13960922e-01 6.54909134e-01 -8.27908695e-01 2.43896067e-01 -6.01225913e-01 -8.29049349e-01 -2.33999174e-02 -8.77589881e-01 8.41676533e-01 5.31708784e-02 -9.77075636e-01 -7.84277678e-01 5.45193136e-01 3.03445876e-01 1.10533945e-01 3.11163604e-01 1.17052948e+00 -6.75084174e-01 -1.85905874e-01 6.96785524e-02 3.33350420e-01 3.81787792e-02 2.68775016e-01 3.69746119e-01 -1.04317272e+00 -4.45339009e-02 1.36695355e-01 -2.17327625e-01 5.49026489e-01 5.09206131e-02 9.46508348e-01 -1.19816884e-01 1.10419556e-01 4.47923154e-01 1.34479988e+00 3.61133307e-01 6.27353549e-01 7.78616428e-01 4.53330755e-01 8.69716227e-01 6.46229327e-01 3.92855853e-01 1.04312234e-01 1.34650326e+00 5.07935993e-02 4.36476439e-01 -6.16546273e-02 -2.43185163e-01 3.62151712e-01 1.27193606e+00 -7.94379711e-01 3.21130425e-01 -8.96370292e-01 4.44075078e-01 -1.51072395e+00 -1.21558285e+00 -5.55763543e-01 2.28745532e+00 9.80008125e-01 1.03545338e-01 8.14842522e-01 1.05514324e+00 2.73281425e-01 1.51167242e-02 6.94596767e-02 -5.22267461e-01 -1.09015398e-01 5.72008431e-01 5.85279316e-02 2.75071591e-01 -1.00264835e+00 6.01263285e-01 6.74603558e+00 1.02116442e+00 -1.02609670e+00 -1.08321168e-01 -1.29677519e-01 -7.22685009e-02 -1.60760164e-01 2.04676643e-01 -6.93039834e-01 3.43148112e-01 9.65367675e-01 -2.20000640e-01 6.88029826e-01 6.31540120e-01 1.48777023e-01 1.72368571e-01 -9.92385864e-01 7.73052096e-01 4.19230759e-02 -1.10969424e+00 -1.97435333e-03 2.53771544e-01 3.67489576e-01 -1.32272899e-01 -2.80420899e-01 1.51281103e-01 -1.59126207e-01 -9.37749803e-01 1.20193934e+00 6.18758082e-01 5.84823072e-01 -8.50155890e-01 5.36721706e-01 1.51069775e-01 -1.33424151e+00 -4.54607643e-02 -6.79684356e-02 -4.55758989e-01 -5.82145229e-02 1.94426075e-01 -5.90168715e-01 6.67337477e-01 8.57739627e-01 7.75219262e-01 -7.14792728e-01 1.05521703e+00 -6.70297723e-03 8.64904046e-01 -3.08469623e-01 -7.59930238e-02 -7.53997639e-02 -2.53782570e-01 6.97893798e-01 1.10830963e+00 4.87435848e-01 -1.83149248e-01 -4.20670211e-02 7.91646540e-01 7.41795719e-01 7.42038846e-01 -3.70631307e-01 -3.13150793e-01 3.54064286e-01 1.17917693e+00 -9.81214702e-01 1.37788616e-02 -3.29217732e-01 3.67087543e-01 -9.98580530e-02 -1.26115978e-01 -6.85964882e-01 -5.36521554e-01 5.98715007e-01 4.42612529e-01 4.05227512e-01 -2.47693881e-01 -2.36659229e-01 -9.92965698e-01 -2.71744877e-01 -1.26317918e+00 4.19000387e-01 -7.21110761e-01 -1.30235529e+00 6.50731146e-01 6.87258393e-02 -1.46741438e+00 -5.51573277e-01 -6.14628792e-01 -3.77493531e-01 8.27767611e-01 -9.44113910e-01 -9.49583530e-01 -5.92336021e-02 6.69155836e-01 3.77953738e-01 -3.48183513e-01 1.36991155e+00 1.94532961e-01 -1.59491543e-02 2.75970906e-01 2.66919751e-02 1.00179121e-01 7.75556684e-01 -1.42276382e+00 -8.48751217e-02 1.59902528e-01 9.77699101e-01 6.59413934e-01 9.36331987e-01 -3.11219603e-01 -9.35345411e-01 -3.18170309e-01 1.13468575e+00 -6.05465531e-01 1.07217753e+00 -3.90887633e-02 -7.61562705e-01 2.95320719e-01 -1.86014660e-02 -8.43030274e-01 1.25799966e+00 7.16704547e-01 -4.84187663e-01 -2.02898696e-01 -5.47461987e-01 4.34299141e-01 8.64743471e-01 -6.87292516e-01 -9.93532181e-01 -7.47459009e-02 1.40882418e-01 -7.17264861e-02 -1.22896206e+00 1.43301159e-01 1.14714861e+00 -1.26830590e+00 8.94407690e-01 -4.71852899e-01 5.68896592e-01 -5.21481752e-01 -3.96211892e-01 -1.01100886e+00 -3.07141244e-01 -5.57478845e-01 2.30544060e-01 1.40057397e+00 4.25158858e-01 -2.90238351e-01 5.87274432e-01 -2.64286995e-01 2.69314684e-02 -4.23465371e-01 -9.20141757e-01 -8.24114740e-01 3.01013231e-01 -8.88342500e-01 7.62890100e-01 1.20993996e+00 1.83832750e-01 5.49377680e-01 -1.62920654e-01 -3.65307570e-01 2.23136321e-01 6.91315234e-01 8.88178825e-01 -1.83004189e+00 -1.03489721e+00 -8.41144323e-01 -7.12921619e-01 -4.97208416e-01 -1.50524840e-01 -1.02805507e+00 -3.57948422e-01 -8.68395269e-01 2.88203992e-02 -5.68488598e-01 -4.63973910e-01 3.18887264e-01 2.81095415e-01 7.64953971e-01 5.10056198e-01 5.56961775e-01 -1.17513031e-01 -1.35159194e-01 1.05936015e+00 3.68068703e-02 -4.51224267e-01 4.87337142e-01 -7.21490741e-01 9.96627092e-01 7.32934892e-01 -5.98727643e-01 -1.84982300e-01 1.03096619e-01 6.33858681e-01 1.47837892e-01 5.99338226e-02 -1.15795565e+00 -2.34278850e-02 -1.65565103e-01 8.28050300e-02 -4.33062255e-01 3.47378910e-01 -6.96873784e-01 5.49560905e-01 3.32712829e-01 -3.33905071e-01 7.06112012e-02 6.04627281e-02 1.47077799e-01 -4.10595804e-01 -6.86244309e-01 3.64164263e-01 -4.48582061e-02 -3.25523227e-01 -3.98129165e-01 -4.44164366e-01 -3.30615416e-02 6.27129555e-01 -3.62788051e-01 1.85709342e-01 -3.22434485e-01 -1.00411665e+00 -5.57023942e-01 5.35849154e-01 5.76690078e-01 -1.72583356e-01 -1.20573354e+00 -6.49111211e-01 -2.80865617e-02 1.99713856e-01 -7.43619919e-01 -1.55138448e-01 7.50014365e-01 -4.65012938e-01 3.32346737e-01 -3.49892825e-01 -4.48309690e-01 -1.66151130e+00 4.56608236e-01 8.52721781e-02 -2.74065256e-01 -4.49975014e-01 4.40625161e-01 -3.17204565e-01 -2.51284003e-01 2.53189355e-01 -3.11165631e-01 -7.25611627e-01 7.12329924e-01 3.98441970e-01 5.49152553e-01 -4.61098999e-02 -8.04933429e-01 -3.44855696e-01 8.63698483e-01 3.14050853e-01 -3.24559063e-01 1.35025513e+00 1.23061016e-01 -4.07844365e-01 1.46628785e+00 6.87382698e-01 7.49062657e-01 -4.06231165e-01 1.73591748e-01 2.98851073e-01 -3.63623738e-01 -3.64002645e-01 -9.23132122e-01 -7.24291384e-01 6.96302891e-01 3.52498025e-01 8.58131349e-01 1.22814715e+00 1.51525840e-01 4.39519674e-01 1.70003384e-01 7.34896600e-01 -9.70208049e-01 -3.99322689e-01 4.92533267e-01 9.36188638e-01 -4.71191138e-01 6.82679340e-02 -2.84704775e-01 -4.01822537e-01 1.52220786e+00 2.56228051e-03 -1.31178409e-01 6.39577270e-01 2.39525124e-01 1.55960275e-02 -6.25793189e-02 -5.21715939e-01 -2.90264279e-01 5.24768651e-01 3.63839597e-01 8.74116421e-01 1.47544399e-01 -1.04425871e+00 1.19989598e+00 -1.29038429e+00 5.74773662e-02 4.78167534e-01 4.30885047e-01 -3.57579082e-01 -1.68910694e+00 -6.19137049e-01 3.24231505e-01 -7.14170516e-01 1.20654423e-03 -8.29284906e-01 1.00723827e+00 6.86586797e-01 9.43450391e-01 2.80508641e-02 -7.51110315e-01 4.77019280e-01 1.95875987e-01 8.39196980e-01 -4.29851890e-01 -1.15247953e+00 3.81109059e-01 2.95257479e-01 -3.51671517e-01 -6.52506769e-01 -1.00139034e+00 -8.63482654e-01 -3.78416657e-01 -3.41838032e-01 7.30048060e-01 7.20647633e-01 8.85023713e-01 -3.80194396e-01 5.61825931e-01 5.43354809e-01 -9.60256398e-01 -2.69581616e-01 -1.14126873e+00 -1.09435081e+00 3.78023833e-01 -2.32557416e-01 -5.50855041e-01 -5.40045321e-01 3.77068631e-02]
[15.937097549438477, 5.239450454711914]
4d7cc9be-fbbd-44ca-a863-e8bc36eb5e6d
superpixel-based-graph-laplacian
2007.14033
null
https://arxiv.org/abs/2007.14033v2
https://arxiv.org/pdf/2007.14033v2.pdf
Superpixel Based Graph Laplacian Regularization for Sparse Hyperspectral Unmixing
An efficient spatial regularization method using superpixel segmentation and graph Laplacian regularization is proposed for sparse hyperspectral unmixing method. Since it is likely to find spectrally similar pixels in a homogeneous region, we use a superpixel segmentation algorithm to extract the homogeneous regions by considering the image boundaries. We first extract the homogeneous regions, which are called superpixels, then a weighted graph in each superpixel is constructed by selecting $K$-nearest pixels in each superpixel. Each node in the graph represents the spectrum of a pixel and edges connect the similar pixels inside the superpixel. The spatial similarity is investigated using graph Laplacian regularization. Sparsity regularization for abundance matrix is provided using a weighted sparsity promoting norm. Experimental results on simulated and real data sets show the superiority of the proposed algorithm over the well-known algorithms in the literature.
['Taner Ince']
2020-07-28
null
null
null
null
['hyperspectral-unmixing']
['computer-vision']
[ 6.54104173e-01 6.46101311e-02 -2.54838467e-01 5.04138926e-03 -2.95203209e-01 -3.08658063e-01 -1.24209009e-01 -5.94697371e-02 -2.03878894e-01 6.35765910e-01 2.90645957e-02 1.79168805e-01 -2.02422440e-01 -8.89321864e-01 -4.35492098e-01 -1.09245801e+00 -4.93063219e-02 -2.22889669e-02 2.84285277e-01 2.90330052e-01 1.56407982e-01 2.99184710e-01 -1.11858559e+00 -1.26465350e-01 1.46470976e+00 8.67684901e-01 5.22745192e-01 1.39482751e-01 -1.54106513e-01 4.50122595e-01 -8.27862974e-03 3.15827549e-01 7.48739898e-01 -7.61669636e-01 -6.64372861e-01 1.00185752e+00 6.15201354e-01 1.37570634e-01 -6.97516352e-02 1.87192965e+00 -1.71328452e-03 2.74921358e-01 5.82843125e-01 -7.37920046e-01 -3.43932092e-01 5.81911922e-01 -1.48719454e+00 1.90802455e-01 -6.63562790e-02 -1.61722094e-01 9.63289618e-01 -7.33041286e-01 6.48657858e-01 8.30829024e-01 6.03055537e-01 -2.43607089e-01 -1.23379743e+00 -6.20785654e-01 1.16296411e-01 -2.27750689e-02 -1.72791755e+00 -1.02399506e-01 1.23160744e+00 -6.16359174e-01 3.25178623e-01 3.64398479e-01 6.99861526e-01 -1.77850217e-01 -2.44869754e-01 6.66068435e-01 1.35877264e+00 -4.70741630e-01 1.87597707e-01 -1.41874840e-02 4.77448702e-01 9.32917774e-01 4.98817086e-01 -3.58306915e-01 -2.25185513e-01 -2.78612077e-01 8.22006524e-01 1.43988848e-01 -6.07552350e-01 -4.39125210e-01 -1.08115518e+00 6.72146797e-01 8.17254007e-01 6.62151158e-01 -8.25920522e-01 -2.40137115e-01 -2.48221885e-02 -1.29560322e-01 6.93816245e-01 -2.27656420e-02 2.31362253e-01 9.54198062e-01 -1.58211970e+00 -2.86558837e-01 3.22249889e-01 8.07306528e-01 1.36376202e+00 2.51277447e-01 8.94948468e-02 9.76735532e-01 5.81724882e-01 4.46237892e-01 1.85181558e-01 -1.14435434e+00 3.29325914e-01 8.83489549e-01 1.38968766e-01 -1.27029824e+00 -1.43380299e-01 -4.63538438e-01 -1.08348346e+00 1.06729895e-01 2.92580187e-01 -1.60864666e-01 -1.13135469e+00 1.22839427e+00 5.33096433e-01 7.02649593e-01 -3.49582843e-02 1.18789065e+00 7.15953767e-01 8.86847138e-01 1.62380695e-01 -1.00434828e+00 1.14461696e+00 -9.01210606e-01 -6.59831226e-01 -1.52787670e-01 1.21211745e-01 -7.75492132e-01 6.14689529e-01 1.27845794e-01 -9.32006657e-01 -5.02524257e-01 -9.70665395e-01 4.50297922e-01 -5.17175347e-02 2.67329454e-01 5.63058019e-01 4.77046937e-01 -7.47957945e-01 4.62237060e-01 -7.88616180e-01 -5.01913607e-01 3.66306216e-01 1.91935882e-01 -2.05340818e-01 -1.11485615e-01 -8.73319626e-01 1.22338235e-01 8.15750599e-01 1.25469938e-01 -4.58508164e-01 -4.44090337e-01 -8.74019802e-01 -1.63869798e-01 2.96245307e-01 -3.82322907e-01 2.07108483e-01 -1.58524561e+00 -9.24386680e-01 1.05284739e+00 -3.56603324e-01 -1.43705562e-01 3.46052609e-02 4.26847488e-01 -4.89767462e-01 5.85323155e-01 1.67466775e-01 4.16449726e-01 1.08840954e+00 -1.46563852e+00 -6.69931948e-01 -4.89197969e-01 -4.31211919e-01 4.77171302e-01 -2.42580160e-01 -1.24450475e-01 -5.89584827e-01 -7.40308821e-01 1.13614392e+00 -8.04416656e-01 -4.98031765e-01 -2.90570706e-01 -8.31777811e-01 6.53982759e-02 1.09706569e+00 -9.69244123e-01 1.37277496e+00 -2.26737857e+00 2.39435568e-01 8.26644599e-01 2.80434102e-01 -9.51744150e-03 6.48109093e-02 -2.36394983e-02 -3.38819742e-01 -8.93222690e-02 -9.29217219e-01 3.87270272e-01 -4.67952251e-01 -1.31036758e-01 2.92521745e-01 8.94257665e-01 -4.86720204e-01 2.39507303e-01 -8.55919778e-01 -7.62844086e-01 2.10280463e-01 1.77361444e-01 -2.19663560e-01 5.44423610e-02 -2.75839239e-01 4.68639761e-01 -6.18410587e-01 6.71126485e-01 1.15795898e+00 -3.91580552e-01 1.72762379e-01 -6.61089301e-01 -4.90780860e-01 -6.15064502e-01 -1.77082229e+00 1.57558823e+00 2.75076091e-01 2.48658866e-01 7.05153346e-01 -1.35752857e+00 8.15102398e-01 1.33057922e-01 9.33490336e-01 4.64377217e-02 5.35273403e-02 2.47960895e-01 -2.89093107e-01 -6.52745664e-01 3.80119503e-01 -3.25946867e-01 5.13514519e-01 3.71346563e-01 -2.01899916e-01 -2.61774242e-01 4.94758099e-01 9.18887928e-02 3.60939205e-01 1.08293984e-02 1.71255082e-01 -6.30715728e-01 8.38555992e-01 2.14767799e-01 7.58198559e-01 2.88999349e-01 -1.95045784e-01 5.63446462e-01 6.44367337e-02 -1.09563790e-01 -8.23257923e-01 -9.52745974e-01 -2.37632751e-01 6.99764609e-01 5.38061202e-01 -5.08315815e-03 -9.65187490e-01 -3.35467994e-01 -1.15732402e-01 4.03192312e-01 -2.45690539e-01 2.42703781e-01 -2.76389629e-01 -1.33864617e+00 -2.23419398e-01 -1.53466925e-01 1.18353355e+00 -6.50621593e-01 -1.63700074e-01 6.74759150e-02 -4.14348304e-01 -8.62950325e-01 -4.96353060e-01 -7.17413351e-02 -9.32281673e-01 -1.16070342e+00 -9.10986185e-01 -1.27357185e+00 1.13231766e+00 8.56318414e-01 6.52672887e-01 -1.51857018e-01 -2.48230979e-01 2.23009899e-01 -2.99147636e-01 1.61363930e-01 2.21215608e-03 -2.01456100e-01 -2.04551011e-01 6.59909964e-01 2.31663048e-01 -6.17746055e-01 -7.30134249e-01 1.09620422e-01 -9.86705363e-01 2.61860974e-02 5.74434042e-01 5.24343312e-01 1.26387131e+00 7.77475238e-01 -1.69357136e-02 -1.24205315e+00 2.52225041e-01 -5.97374618e-01 -9.18523848e-01 3.18139732e-01 -4.95440632e-01 -1.70235962e-01 2.80617058e-01 -6.47537112e-02 -1.26407456e+00 6.26602292e-01 5.51289916e-01 -3.36648256e-01 -2.15260372e-01 7.97658980e-01 -2.68411517e-01 -3.38295817e-01 5.85772693e-01 4.77938086e-01 -8.45805556e-02 -3.83204162e-01 5.52836180e-01 7.18992889e-01 4.59926575e-01 -1.72483057e-01 8.98701906e-01 8.61526966e-01 7.82275721e-02 -1.43128610e+00 -7.69435048e-01 -8.81866097e-01 -7.12175190e-01 -3.36972088e-01 1.29939830e+00 -1.01112056e+00 -1.41422391e-01 3.73607516e-01 -6.53453887e-01 -3.88301648e-02 -2.29235351e-01 6.87619150e-01 -1.94079235e-01 1.03018737e+00 -4.97718990e-01 -7.72854626e-01 -2.04227239e-01 -9.47206080e-01 6.58871233e-01 5.84776640e-01 1.74042836e-01 -1.00752783e+00 -5.88056259e-02 5.98981500e-01 -3.95465381e-02 4.07152832e-01 8.71113837e-01 -1.83419675e-01 -7.95140326e-01 -4.48854342e-02 -4.72840458e-01 5.55915415e-01 3.06108117e-01 -3.09190173e-02 -7.62066960e-01 -2.34334841e-01 2.72407919e-01 6.65854439e-02 1.17035496e+00 1.01759672e+00 1.04229188e+00 -2.26459980e-01 -5.20209312e-01 8.89569700e-01 1.86018896e+00 4.23257910e-02 5.29893637e-01 8.19792151e-02 9.41887379e-01 6.07772648e-01 4.77558792e-01 2.13800266e-01 -1.87547028e-01 3.03539680e-04 4.45202500e-01 -5.31495154e-01 -1.63230114e-02 1.47663817e-01 1.12357885e-01 8.63956213e-01 -3.43514115e-01 -5.58173247e-02 -6.80582166e-01 5.64399064e-01 -1.76721120e+00 -1.03111064e+00 -8.41762900e-01 2.23334575e+00 6.38390779e-01 -3.25397909e-01 -3.19015421e-02 7.50476271e-02 1.33725834e+00 6.88721418e-01 -5.19382894e-01 3.66895914e-01 -5.13170004e-01 1.19693346e-01 9.57613170e-01 8.55381191e-01 -1.35397685e+00 1.11307383e+00 5.97291517e+00 1.06134045e+00 -8.64231110e-01 1.36060491e-01 9.14840400e-01 2.02076763e-01 -3.08865666e-01 1.51362881e-01 -4.09537226e-01 3.97415847e-01 4.91033532e-02 1.54132307e-01 5.25038660e-01 4.05514091e-01 6.52759552e-01 -7.05363333e-01 -1.22044772e-01 1.11702085e+00 1.29340515e-01 -9.71000910e-01 3.61206532e-02 -1.85072422e-01 1.37221599e+00 1.29878493e-02 -7.10545927e-02 -7.00063944e-01 3.06299537e-01 -7.93242514e-01 1.92130044e-01 6.33523345e-01 5.67797959e-01 -7.50736475e-01 2.98414052e-01 3.39616030e-01 -1.51950431e+00 2.69698560e-01 -6.12665415e-01 3.16577971e-01 5.97927254e-03 1.08522236e+00 -2.41902471e-01 6.74265027e-01 5.33502340e-01 9.36180592e-01 -3.51195872e-01 1.04155970e+00 -1.98390618e-01 9.14756715e-01 -4.53440696e-01 4.95795876e-01 4.63606238e-01 -1.61568224e+00 7.84181535e-01 1.04104459e+00 5.01504123e-01 5.22460938e-01 4.39243376e-01 1.09103060e+00 2.04233956e-02 5.50345719e-01 -5.78654826e-01 -1.22096762e-01 2.69001752e-01 1.48687565e+00 -1.08744133e+00 -3.30024779e-01 -5.72524488e-01 1.15834618e+00 -1.04911961e-01 8.13450277e-01 -4.84417766e-01 -2.09074646e-01 1.52780131e-01 2.23731682e-01 1.37173757e-01 -2.15753436e-01 -3.37733239e-01 -9.29506660e-01 -1.29733235e-01 -5.25433362e-01 7.03566968e-01 -8.40458512e-01 -9.91142869e-01 2.70200670e-01 -1.73895419e-01 -1.33677197e+00 6.64939523e-01 -2.61030465e-01 -6.27470851e-01 1.10222995e+00 -1.45487463e+00 -1.07241607e+00 -6.36142790e-01 6.90669239e-01 3.29717547e-01 5.97125068e-02 3.59973073e-01 -2.54919548e-02 -8.31117749e-01 -3.17572802e-01 4.57901746e-01 2.65115257e-02 2.16324583e-01 -1.03745914e+00 -7.75181651e-01 1.25062966e+00 -2.36958768e-02 5.22118568e-01 5.19667387e-01 -9.89151537e-01 -9.03887630e-01 -1.28628993e+00 5.48194885e-01 5.97221136e-01 6.69475257e-01 2.55647510e-01 -8.81683350e-01 5.04977763e-01 5.81989229e-01 2.65610684e-02 7.59011090e-01 -4.59850460e-01 8.65256712e-02 -1.98852077e-01 -1.26657319e+00 4.09729838e-01 7.48802662e-01 -3.34249079e-01 -2.69107938e-01 1.04429638e+00 3.03965181e-01 -8.62689912e-02 -8.59476447e-01 4.03469056e-01 -7.75277168e-02 -8.94026875e-01 1.05457914e+00 -3.98778394e-02 2.80802865e-02 -8.86053205e-01 -8.58996138e-02 -1.30648530e+00 -6.19397819e-01 -4.47101891e-01 5.77751279e-01 1.15355623e+00 5.78133047e-01 -4.09746468e-01 9.97581422e-01 2.02443123e-01 2.07359176e-02 -5.75686768e-02 -6.04174435e-01 -6.61999404e-01 -3.35373878e-01 -3.51539366e-02 1.53537229e-01 1.45378244e+00 8.20879415e-02 2.90351421e-01 -1.84617847e-01 7.75827348e-01 1.25867677e+00 6.16707742e-01 -2.09937189e-02 -1.27251077e+00 -7.03631416e-02 -4.76238966e-01 -9.28130671e-02 -8.92650306e-01 3.32535565e-01 -1.19702590e+00 2.84632500e-02 -1.75935423e+00 5.03963709e-01 -3.37465227e-01 -3.39422435e-01 8.42429027e-02 -4.30203646e-01 5.33878148e-01 -2.51443326e-01 4.99442875e-01 -4.82955754e-01 1.23154089e-01 1.31692159e+00 -4.66185391e-01 -5.70435762e-01 -1.97747096e-01 -4.89358038e-01 9.87338126e-01 8.01680326e-01 -2.01967567e-01 -4.15665954e-01 -1.86765969e-01 -3.30930725e-02 3.26985531e-02 5.21135218e-02 -1.20930517e+00 1.55913413e-01 -3.82043481e-01 2.67865568e-01 -6.18655741e-01 -7.72253796e-02 -1.12509203e+00 5.32605767e-01 3.75265270e-01 -1.09915726e-01 -8.14716160e-01 -1.61259323e-01 7.69206345e-01 -3.28499407e-01 -5.59077561e-01 1.29275763e+00 -3.93324703e-01 -1.00992811e+00 4.48566556e-01 -3.17996264e-01 -2.19934151e-01 1.16674924e+00 -4.06117231e-01 1.44377619e-01 -2.12517470e-01 -1.04620361e+00 1.55389234e-01 4.62210178e-01 -3.77646536e-01 5.19297600e-01 -1.17654729e+00 -6.21655345e-01 1.27624974e-01 1.23529792e-01 8.43757465e-02 3.29197943e-01 1.04688215e+00 -9.04641330e-01 -6.58655092e-02 -2.78640166e-02 -6.66104913e-01 -1.49640417e+00 4.52474624e-01 5.87565362e-01 1.94330573e-01 -4.23194557e-01 9.07832921e-01 2.89092869e-01 6.80215433e-02 -5.60111217e-02 -2.47298971e-01 -5.09505510e-01 9.22648981e-02 1.09187484e-01 5.74113429e-01 -5.35646200e-01 -1.24807799e+00 -5.73466644e-02 1.02397883e+00 5.65085471e-01 5.32524697e-02 1.18351102e+00 -3.33643794e-01 -9.70670581e-01 3.23245198e-01 1.03237975e+00 2.37618595e-01 -1.23010993e+00 -3.85411739e-01 -4.60014120e-02 -4.35708046e-01 4.55151677e-01 -2.49971956e-01 -1.64235139e+00 5.35800755e-01 7.19631910e-01 4.43147540e-01 1.33686233e+00 -8.28627869e-02 6.63614810e-01 -1.11111566e-01 4.69274223e-02 -1.37802565e+00 -3.36814612e-01 1.60078123e-01 4.70281392e-01 -1.16984069e+00 3.96126211e-01 -1.18180251e+00 -5.56670606e-01 9.64641154e-01 4.21274781e-01 -3.98505151e-01 8.51876915e-01 -2.37572402e-01 1.96886603e-02 -4.20700908e-01 4.92231458e-01 -5.22058368e-01 3.23089272e-01 3.64092946e-01 4.71976399e-01 1.78376168e-01 -6.63622916e-01 1.38515100e-01 2.44945660e-01 -3.27674091e-01 1.93404302e-01 5.03497541e-01 -1.02400434e+00 -6.49177134e-01 -8.06047797e-01 5.49705446e-01 -2.91676283e-01 -1.91359356e-01 -3.13600987e-01 4.07250762e-01 5.20094454e-01 1.12674987e+00 9.55926329e-02 -3.56124230e-02 -4.72962037e-02 2.43861258e-01 3.06229144e-01 -6.39143169e-01 -1.49268299e-01 8.53230536e-01 -2.30746239e-01 -2.21799284e-01 -1.05885947e+00 -6.70091391e-01 -1.56430733e+00 1.10720277e-01 -3.77365619e-01 4.26719904e-01 2.34318346e-01 6.74184382e-01 -1.65944934e-01 1.51247576e-01 8.13414633e-01 -6.42812610e-01 2.59216726e-01 -9.47857082e-01 -1.51512301e+00 4.53960955e-01 3.59718204e-02 -2.65299767e-01 -6.88122749e-01 5.68526864e-01]
[10.093822479248047, -2.00382924079895]
21255155-d06e-4f8d-8311-d44ce76de011
learn-to-race-challenge-2022-benchmarking
2205.02953
null
https://arxiv.org/abs/2205.02953v2
https://arxiv.org/pdf/2205.02953v2.pdf
Learn-to-Race Challenge 2022: Benchmarking Safe Learning and Cross-domain Generalisation in Autonomous Racing
We present the results of our autonomous racing virtual challenge, based on the newly-released Learn-to-Race (L2R) simulation framework, which seeks to encourage interdisciplinary research in autonomous driving and to help advance the state of the art on a realistic benchmark. Analogous to racing being used to test cutting-edge vehicles, we envision autonomous racing to serve as a particularly challenging proving ground for autonomous agents as: (i) they need to make sub-second, safety-critical decisions in a complex, fast-changing environment; and (ii) both perception and control must be robust to distribution shifts, novel road features, and unseen obstacles. Thus, the main goal of the challenge is to evaluate the joint safety, performance, and generalisation capabilities of reinforcement learning agents on multi-modal perception, through a two-stage process. In the first stage of the challenge, we evaluate an autonomous agent's ability to drive as fast as possible, while adhering to safety constraints. In the second stage, we additionally require the agent to adapt to an unseen racetrack through safe exploration. In this paper, we describe the new L2R Task 2.0 benchmark, with refined metrics and baseline approaches. We also provide an overview of deployment, evaluation, and rankings for the inaugural instance of the L2R Autonomous Racing Virtual Challenge (supported by Carnegie Mellon University, Arrival Ltd., AICrowd, Amazon Web Services, and Honda Research), which officially used the new L2R Task 2.0 benchmark and received over 20,100 views, 437 active participants, 46 teams, and 733 model submissions -- from 88+ unique institutions, in 58+ different countries. Finally, we release leaderboard results from the challenge and provide description of the two top-ranking approaches in cross-domain model transfer, across multiple sensor configurations and simulated races.
['Ivan Zhukov', 'Vrushank Vyas', 'Eric Nyberg', 'Jean Oh', 'Anirudh Koul', 'Max Kumskoy', 'Sahika Genc', 'Ayush Shivani', 'Jyotish Poonganam', 'Sidharth Kathpal', 'Siddha Ganju', 'Bingqing Chen', 'Jonathan Francis']
2022-05-05
null
null
null
null
['safe-exploration']
['robots']
[-1.59849271e-01 -1.70249324e-02 -1.40105963e-01 -2.59915829e-01 -9.77122605e-01 -8.04240048e-01 7.42551863e-01 -1.18407853e-01 -7.21504211e-01 8.27721953e-01 -1.48103803e-01 -5.66795230e-01 -1.98366478e-01 -5.29396355e-01 -1.03617990e+00 -2.63567477e-01 -5.95593452e-01 8.47835004e-01 3.69047433e-01 -9.58574116e-01 8.41619521e-02 5.92848599e-01 -2.01377940e+00 -2.30579395e-02 6.80882752e-01 8.75762701e-01 1.30020589e-01 7.91167200e-01 7.36250401e-01 5.98585308e-01 -3.14174980e-01 1.92757770e-01 4.52245057e-01 2.01367721e-01 -8.16863596e-01 -3.86821151e-01 3.20158064e-01 1.03919514e-01 -3.58256698e-01 6.20520651e-01 5.00062048e-01 6.15706325e-01 3.17123115e-01 -1.96170115e+00 3.48747708e-02 3.89905393e-01 -3.17953080e-01 1.79410309e-01 3.20263833e-01 7.99465179e-01 7.84440994e-01 -4.66902763e-01 6.05036497e-01 1.25883985e+00 5.39332151e-01 6.94413960e-01 -1.22298419e+00 -7.97545135e-01 5.21228969e-01 3.59532535e-01 -1.20705736e+00 -7.66039431e-01 2.55972594e-01 -5.36526263e-01 1.18095517e+00 1.26530677e-01 4.27922308e-01 1.51063061e+00 4.10502434e-01 5.54416180e-01 1.21253872e+00 2.32579932e-01 5.15151024e-01 1.28947377e-01 -2.26216856e-02 5.69487274e-01 -3.79236899e-02 9.72236812e-01 -6.18197381e-01 7.49762133e-02 -1.70804545e-01 -8.01101744e-01 9.00820196e-02 -6.84269905e-01 -1.33672333e+00 5.96328855e-01 3.52753848e-01 -3.07379931e-01 -2.82119960e-01 5.44433355e-01 5.41152596e-01 5.33854306e-01 -1.10742420e-01 7.26794004e-01 -5.95330060e-01 -6.08268619e-01 -2.00163767e-01 9.16668594e-01 7.18290150e-01 7.45347142e-01 5.95300794e-01 9.56044495e-02 3.66587788e-02 5.65874636e-01 2.81702548e-01 6.78568780e-01 8.14052075e-02 -1.37695515e+00 5.02428710e-01 1.31303191e-01 5.08725345e-01 -6.86874986e-01 -7.85479307e-01 -1.90530971e-01 -2.96328753e-01 8.69151473e-01 1.93736807e-01 -4.96041805e-01 -7.92636395e-01 1.98262310e+00 3.49087298e-01 3.27880569e-02 5.64435422e-01 9.98177290e-01 4.46973711e-01 4.64214116e-01 1.22064143e-01 3.84859800e-01 1.20334172e+00 -1.10509694e+00 -2.79529572e-01 -5.57687759e-01 6.50469780e-01 -4.44598407e-01 1.01100338e+00 6.17136776e-01 -9.63012099e-01 -8.12634766e-01 -1.28219402e+00 3.95391554e-01 -6.84559882e-01 -3.76235843e-01 4.17952269e-01 4.79864031e-01 -1.02078784e+00 2.71163970e-01 -8.59085977e-01 -4.42164063e-01 3.56296487e-02 2.90816188e-01 -5.51194310e-01 -1.64272219e-01 -1.51740706e+00 1.48020554e+00 2.47388437e-01 -6.64759129e-02 -1.73454452e+00 -8.13225448e-01 -9.39032733e-01 -4.92688686e-01 5.18508077e-01 -2.25011066e-01 1.43271947e+00 -3.57585639e-01 -1.50771356e+00 6.10862494e-01 2.36800194e-01 -6.50540411e-01 8.00972760e-01 -2.80704618e-01 -7.25723207e-01 -3.99113536e-01 2.27753937e-01 1.06604838e+00 2.84228742e-01 -1.48831308e+00 -1.09689927e+00 -2.04262421e-01 3.41230303e-01 5.52400351e-01 5.00252426e-01 -2.98304439e-01 -1.17037684e-01 4.43219527e-04 -5.26991427e-01 -1.39011955e+00 -4.60542828e-01 -5.71671128e-01 -8.46272558e-02 -3.14845681e-01 9.78725791e-01 -2.27766693e-01 4.74303782e-01 -2.28408504e+00 2.64294207e-01 2.23962501e-01 -6.47419319e-03 -1.00650914e-01 -4.93819982e-01 5.09505689e-01 1.35793880e-01 -4.80463654e-02 -2.08715662e-01 -3.09151471e-01 3.11791688e-01 3.44023556e-01 -5.63912451e-01 5.13381839e-01 1.36565194e-01 6.16540849e-01 -9.74453092e-01 -1.29908085e-01 2.39959046e-01 2.14560237e-02 -3.59096974e-01 9.47969556e-02 -2.97999501e-01 5.72714388e-01 -2.25063369e-01 3.93492311e-01 4.47490513e-01 4.75553513e-01 -1.69209316e-01 1.08867243e-01 -5.16374409e-01 7.90369436e-02 -1.28799903e+00 1.79413533e+00 -6.20000958e-01 7.02500701e-01 3.38330507e-01 -7.81370878e-01 9.11208987e-01 4.87861922e-03 5.41864574e-01 -1.15322649e+00 -1.60336103e-02 3.24180156e-01 6.51024356e-02 -3.76550555e-01 8.57718945e-01 2.29874775e-01 -6.81898415e-01 2.26633370e-01 -3.36617172e-01 -5.58993459e-01 3.26505691e-01 1.74732178e-01 1.22524250e+00 3.88166815e-01 -4.40586805e-01 -4.57932651e-01 2.89472193e-01 5.68615556e-01 5.83620012e-01 8.72813821e-01 -7.64460981e-01 8.58537629e-02 3.78705829e-01 -6.52995944e-01 -7.30702877e-01 -1.25097060e+00 -1.27556145e-01 1.27772295e+00 7.43719459e-01 -7.32256696e-02 -4.55831707e-01 -7.59830117e-01 2.73836106e-01 1.19354141e+00 -7.11500645e-01 -4.51528251e-01 -6.63031280e-01 -5.75313568e-01 9.36982393e-01 3.82255852e-01 5.40757537e-01 -9.93639052e-01 -1.24876952e+00 1.74335182e-01 -2.66353875e-01 -1.30957890e+00 -2.31906220e-01 4.19576198e-01 -6.71671927e-02 -1.29930878e+00 7.46927857e-02 -5.23785532e-01 -6.00181893e-02 2.34582424e-01 1.04859900e+00 -2.61815697e-01 -2.19115824e-01 5.60688794e-01 -2.79657543e-01 -4.64755237e-01 -5.59009731e-01 3.71368915e-01 5.71910858e-01 -3.20687711e-01 9.70912278e-02 -1.00280754e-01 -2.73038864e-01 9.73020613e-01 -5.40475070e-01 -1.43558234e-01 3.20740908e-01 5.56937873e-01 3.54537189e-01 8.68080407e-02 7.30302036e-01 -1.92033648e-01 7.01241970e-01 -7.00403214e-01 -1.04006195e+00 2.17946749e-02 -7.72452414e-01 6.69041574e-02 5.65868080e-01 -2.30417386e-01 -8.07727575e-01 1.25181958e-01 3.16582024e-02 -1.87618122e-01 -2.72854745e-01 2.14490920e-01 2.41310080e-03 -8.28040391e-03 1.00005937e+00 1.79885719e-02 2.47049987e-01 9.61563140e-02 5.35421789e-01 5.09082079e-01 7.75922775e-01 -1.10663033e+00 1.05116248e+00 4.61278111e-01 1.71971664e-01 -6.09831810e-01 -2.87706971e-01 -1.51677430e-01 -2.22003281e-01 -5.25321186e-01 9.73094404e-01 -1.14700854e+00 -1.13033223e+00 4.82138634e-01 -7.37619221e-01 -1.15567493e+00 -3.43835503e-01 4.21403497e-01 -7.99321115e-01 -2.07255065e-01 2.52288524e-02 -6.07950032e-01 2.14517653e-01 -1.69410753e+00 8.59598219e-01 2.77137667e-01 -5.95365129e-02 -6.21096611e-01 6.73555732e-01 5.18804073e-01 6.61436915e-01 3.47456455e-01 5.79198003e-01 -5.09847820e-01 -4.14691746e-01 -4.05303724e-02 -2.86108516e-02 1.42394662e-01 -3.19821417e-01 -2.50578821e-02 -8.73942435e-01 -6.66008830e-01 -7.65983164e-01 -9.63252783e-01 6.56531215e-01 -2.34097484e-02 8.49259377e-01 2.27014795e-01 -4.85737115e-01 2.95909822e-01 9.81679201e-01 3.64472151e-01 4.28829879e-01 7.08545566e-01 3.18761528e-01 9.24646854e-01 1.07726705e+00 1.79894015e-01 1.03931093e+00 9.17663276e-01 9.28581238e-01 3.38013507e-02 2.40046293e-01 -2.47466043e-01 6.96981668e-01 2.88439572e-01 6.63253590e-02 -1.25567034e-01 -1.15246379e+00 5.35504401e-01 -1.98036802e+00 -7.50906229e-01 -9.33215842e-02 2.31934214e+00 2.95242965e-01 4.46658522e-01 2.92195141e-01 -2.37074494e-01 2.77498215e-01 1.44459456e-01 -8.82429123e-01 -6.89846873e-01 -9.09520313e-03 -1.14095107e-01 7.96081543e-01 7.45038033e-01 -9.74569857e-01 1.18738639e+00 6.40718985e+00 5.89729011e-01 -9.15700793e-01 8.78098309e-02 4.71028358e-01 -2.05935821e-01 -9.10857022e-02 8.19718763e-02 -8.28180730e-01 2.63170868e-01 1.27409840e+00 -8.45358148e-02 8.38984549e-01 8.78041089e-01 2.98329353e-01 -3.45365703e-01 -1.24383104e+00 5.53183794e-01 -2.46242553e-01 -1.12049067e+00 -6.69934809e-01 7.08736405e-02 6.77657247e-01 8.57991338e-01 4.31511909e-01 1.07196581e+00 1.03062749e+00 -1.48726213e+00 1.04527628e+00 3.03154558e-01 7.19370484e-01 -9.30458486e-01 6.03733361e-01 2.20868438e-01 -1.31454146e+00 -3.86156172e-01 1.09382495e-01 -6.00358993e-02 2.20245257e-01 -1.93259314e-01 -5.27745962e-01 5.67688048e-01 1.04312599e+00 3.77665848e-01 -3.88726115e-01 6.81761444e-01 -1.67151242e-02 2.32691944e-01 -3.76945794e-01 8.44504610e-02 6.26060128e-01 2.14509234e-01 7.84759283e-01 9.56100881e-01 -2.14781746e-01 -1.24104559e-01 6.01531804e-01 5.62094271e-01 8.44340995e-02 -4.11938906e-01 -7.81319857e-01 4.73261714e-01 7.24892616e-01 1.22620189e+00 -1.97813541e-01 -6.56241775e-02 -3.85057889e-02 3.94241303e-01 2.74035275e-01 3.87517899e-01 -1.25271416e+00 -3.96535188e-01 1.31186223e+00 -2.16987003e-02 -1.68371648e-01 -4.82686460e-01 -4.62355232e-03 -5.23268223e-01 -2.50085473e-01 -1.28709888e+00 2.80793339e-01 -6.65365338e-01 -8.72282743e-01 6.89727068e-01 2.49100804e-01 -9.11650002e-01 -2.73451537e-01 -4.47856694e-01 -3.66561294e-01 6.55021012e-01 -1.79669094e+00 -8.61733735e-01 -5.19804835e-01 4.46391225e-01 6.17180586e-01 -4.79746878e-01 6.48783565e-01 1.94827303e-01 -7.68100917e-01 3.85070145e-01 -1.03867523e-01 -4.29560661e-01 8.04169714e-01 -1.20796239e+00 7.74665833e-01 6.41173899e-01 -2.70385951e-01 -7.84121752e-02 9.44155931e-01 -3.96631151e-01 -1.66809046e+00 -1.35635138e+00 2.34070763e-01 -9.10981536e-01 7.21565485e-01 -4.24752384e-01 -4.46808666e-01 6.83173835e-01 1.80773139e-01 1.94515532e-03 5.03869317e-02 6.62918612e-02 -3.63867551e-01 -4.01824325e-01 -1.05364180e+00 7.11203635e-01 1.01904547e+00 -1.82346746e-01 -3.33359927e-01 2.32643649e-01 7.90079176e-01 -6.75914168e-01 -6.77781701e-01 5.92701077e-01 6.52858198e-01 -7.08138764e-01 8.14518154e-01 -7.43721664e-01 7.49635696e-02 -5.77404857e-01 -6.81748033e-01 -1.62880015e+00 -9.11498070e-02 -8.49558949e-01 2.24645868e-01 6.58655405e-01 7.21844912e-01 -7.63377964e-01 5.73836148e-01 4.82925504e-01 -6.00272179e-01 -5.50071955e-01 -1.29398882e+00 -1.03138733e+00 2.82756835e-01 -8.15978408e-01 6.04731858e-01 5.95841527e-01 -1.76928788e-01 2.31550291e-01 -3.63675714e-01 4.33862776e-01 7.15413034e-01 -4.38189656e-01 1.27782190e+00 -1.09886467e+00 -1.20083109e-01 -4.15226787e-01 -2.01869115e-01 -8.22076559e-01 4.15684491e-01 -7.25723147e-01 5.92302024e-01 -1.35108078e+00 -3.93961042e-01 -1.05874431e+00 -2.93758333e-01 5.94937980e-01 3.46067935e-01 -6.83888793e-02 1.85096413e-01 1.77941658e-02 -8.67376566e-01 6.62352920e-01 8.66599500e-01 -2.86829680e-01 -3.47366840e-01 -4.26631831e-02 -5.59280097e-01 3.11035037e-01 9.09099758e-01 -2.46481076e-01 -5.24968266e-01 -4.74920064e-01 2.48314828e-01 2.52127685e-02 5.28353631e-01 -1.39141834e+00 2.74166286e-01 -5.63284874e-01 -2.59549499e-01 -4.06815797e-01 5.57216406e-01 -8.54917645e-01 1.07933491e-01 6.90808296e-01 -3.17157418e-01 6.17970169e-01 6.94975197e-01 5.31941295e-01 1.63417578e-01 4.37376499e-01 9.19810176e-01 1.81898117e-01 -1.19188333e+00 1.27055332e-01 -7.72414744e-01 4.74727869e-01 1.57704699e+00 -1.31947675e-03 -7.13933825e-01 -1.52645558e-01 -4.30728853e-01 1.21253109e+00 4.23244387e-01 1.22120297e+00 3.87388647e-01 -1.06803143e+00 -9.31713223e-01 1.52030706e-01 5.30233145e-01 6.46863878e-03 4.06388134e-01 7.20284045e-01 -3.38244140e-01 2.76779145e-01 -4.63450670e-01 -6.79945052e-01 -7.97139227e-01 3.74263376e-01 7.05959082e-01 -1.27393290e-01 -4.82989371e-01 5.73814511e-01 -8.58430490e-02 -1.05952561e+00 2.87543565e-01 3.45836729e-02 -6.21186756e-02 -1.25228941e-01 2.48018116e-01 5.77040195e-01 2.28003651e-01 -6.98963046e-01 -6.67413175e-01 3.74446064e-01 1.25139412e-02 -5.60170889e-01 8.96412432e-01 -6.13567308e-02 5.08268356e-01 3.81975919e-01 7.39659309e-01 -3.13587546e-01 -1.50973034e+00 2.47188866e-01 -5.52504091e-03 -2.37423349e-02 9.38844010e-02 -1.09332466e+00 -7.76776612e-01 5.12037814e-01 8.82645965e-01 3.07202693e-02 5.05911410e-01 -1.12010829e-01 5.70157945e-01 5.66640198e-01 6.30662739e-01 -1.55653024e+00 1.55670866e-01 6.89718723e-01 9.56625521e-01 -1.35860538e+00 -4.28667545e-01 3.67193341e-01 -9.64729786e-01 5.93400657e-01 1.17305267e+00 9.22307838e-03 3.13937396e-01 2.71701902e-01 3.22292387e-01 -1.64194673e-01 -1.27470016e+00 -1.27040461e-01 -1.37548357e-01 1.03122830e+00 -4.05968875e-01 3.78676027e-01 4.21589583e-01 1.37167394e-01 -3.85048121e-01 -3.34656626e-01 6.61268294e-01 8.98601115e-01 -4.41253364e-01 -8.24916840e-01 -3.40904593e-01 6.17080294e-02 2.48115286e-01 4.72145796e-01 -4.78735045e-02 1.13336384e+00 1.52543500e-01 1.34170949e+00 7.66267255e-02 -8.42938125e-01 9.56538141e-01 -3.19035411e-01 1.04786888e-01 -2.38957509e-01 -4.86637920e-01 -8.88581097e-01 5.61888695e-01 -1.08058453e+00 5.87529615e-02 -8.20057213e-01 -1.25886869e+00 -4.59400177e-01 1.57191396e-01 3.21901441e-01 1.11068201e+00 6.54088080e-01 6.22498930e-01 7.79982984e-01 8.58519912e-01 -9.98529851e-01 -7.74866700e-01 -5.48514187e-01 -9.98080149e-02 3.02568674e-02 4.91665453e-01 -9.80943739e-01 -3.71431082e-01 -4.77442741e-01]
[5.061178684234619, 1.2172703742980957]
41898670-cc22-4850-815e-f22de3506ad6
doubly-stochastic-subspace-clustering
2011.14859
null
https://arxiv.org/abs/2011.14859v2
https://arxiv.org/pdf/2011.14859v2.pdf
Doubly Stochastic Subspace Clustering
Many state-of-the-art subspace clustering methods follow a two-step process by first constructing an affinity matrix between data points and then applying spectral clustering to this affinity. Most of the research into these methods focuses on the first step of generating the affinity, which often exploits the self-expressive property of linear subspaces, with little consideration typically given to the spectral clustering step that produces the final clustering. Moreover, existing methods often obtain the final affinity that is used in the spectral clustering step by applying ad-hoc or arbitrarily chosen postprocessing steps to the affinity generated by a self-expressive clustering formulation, which can have a significant impact on the overall clustering performance. In this work, we unify these two steps by learning both a self-expressive representation of the data and an affinity matrix that is well-normalized for spectral clustering. In our proposed models, we constrain the affinity matrix to be doubly stochastic, which results in a principled method for affinity matrix normalization while also exploiting known benefits of doubly stochastic normalization in spectral clustering. We develop a general framework and derive two models: one that jointly learns the self-expressive representation along with the doubly stochastic affinity, and one that sequentially solves for one then the other. Furthermore, we leverage sparsity in the problem to develop a fast active-set method for the sequential solver that enables efficient computation on large datasets. Experiments show that our method achieves state-of-the-art subspace clustering performance on many common datasets in computer vision.
['Benjamin D. Haeffele', 'René Vidal', 'Derek Lim']
2020-11-30
null
null
null
null
['image-clustering']
['computer-vision']
[ 1.48958623e-01 -2.32055351e-01 -1.15866311e-01 -3.24027479e-01 -8.94177616e-01 -7.61706054e-01 4.09223765e-01 -1.75350189e-01 -3.54363114e-01 1.86954737e-01 3.71825755e-01 8.38926435e-02 -4.88300949e-01 -4.27863508e-01 -5.92997789e-01 -1.24209571e+00 1.05449267e-01 7.10183740e-01 2.41399258e-02 1.70192614e-01 -1.28391962e-02 4.24412817e-01 -1.44990277e+00 5.91681898e-02 1.16899693e+00 7.61850953e-01 4.93684784e-02 4.04061615e-01 -1.29581884e-01 3.58826339e-01 -5.29452525e-02 -8.80028680e-03 6.26608014e-01 -4.73753631e-01 -6.63385749e-01 5.25259793e-01 1.97700977e-01 1.65976837e-01 -1.32787287e-01 1.08765662e+00 2.71418482e-01 4.29297447e-01 9.10847127e-01 -1.31771111e+00 -2.36423597e-01 6.54325545e-01 -7.86052883e-01 -2.26727560e-01 6.34299368e-02 -9.96945202e-02 1.04612780e+00 -1.05979502e+00 5.68358958e-01 9.03516829e-01 6.08538449e-01 2.34207824e-01 -1.82145500e+00 -7.34090388e-01 1.28558978e-01 -6.91535249e-02 -1.80844676e+00 -4.85313475e-01 1.13336515e+00 -8.19046557e-01 3.73689801e-01 3.48844886e-01 7.03193367e-01 8.37010384e-01 -5.21678269e-01 8.40141535e-01 1.09966218e+00 -3.78080428e-01 5.40255189e-01 5.28931245e-02 3.12567800e-01 3.77615184e-01 8.62192214e-02 -1.71441644e-01 -4.28043455e-01 -5.64136267e-01 6.97608829e-01 1.09755903e-01 -2.14270502e-01 -1.23557329e+00 -1.08865511e+00 9.71595168e-01 2.26569355e-01 2.50797182e-01 -4.47002053e-01 -1.39488176e-01 1.50206208e-01 -6.21704012e-02 2.49398470e-01 3.38664204e-01 5.95100634e-02 1.94355726e-01 -1.43319106e+00 1.44424185e-01 7.15056181e-01 8.58991563e-01 1.12831843e+00 -2.27826893e-01 7.37860426e-02 9.72503364e-01 4.36105162e-01 4.00822818e-01 4.64279443e-01 -1.02982831e+00 1.75442308e-01 6.97039366e-01 3.33184861e-02 -9.49637949e-01 -2.68298835e-01 -5.54584682e-01 -9.25759196e-01 -1.76618602e-02 4.07489747e-01 -1.45768153e-03 -8.54322255e-01 1.96958721e+00 6.31224632e-01 4.17801470e-01 1.19937472e-01 1.00754333e+00 2.81455308e-01 3.53446543e-01 -1.83648974e-01 -5.53698301e-01 8.60502839e-01 -9.48835194e-01 -5.61620355e-01 6.74556941e-02 5.11038423e-01 -7.49231517e-01 1.10222256e+00 2.61148036e-01 -8.73405039e-01 -2.62265444e-01 -1.09074116e+00 9.32909641e-03 -1.01596788e-01 2.08521798e-01 6.34666979e-01 4.89828587e-01 -9.24541771e-01 4.95734334e-01 -1.02259517e+00 -4.83068526e-01 1.78306505e-01 4.86094922e-01 -3.01497817e-01 7.97141790e-02 -7.93680131e-01 2.55306900e-01 3.69123250e-01 -1.94542795e-01 -5.24858534e-01 -7.66409278e-01 -5.71120322e-01 -3.19240540e-02 6.25340521e-01 -8.42996657e-01 6.74458981e-01 -1.16302156e+00 -1.41813517e+00 7.05208123e-01 -2.67182052e-01 -2.26465225e-01 5.62765896e-01 -1.09247737e-01 -1.10445835e-01 2.36071169e-01 2.81120181e-01 4.18607265e-01 1.07564902e+00 -1.60250318e+00 -2.85212219e-01 -5.40741563e-01 -2.87277877e-01 5.02675533e-01 -6.77609324e-01 -1.44287154e-01 -1.08110833e+00 -7.18565166e-01 6.20252907e-01 -1.26033700e+00 -4.51632023e-01 -2.67944962e-01 -5.13695538e-01 2.29875632e-02 7.55336523e-01 -3.43157262e-01 1.42050087e+00 -2.42392182e+00 7.75822997e-01 8.32786143e-01 2.68011808e-01 -1.66188315e-01 3.24318856e-02 4.29693371e-01 -2.57256746e-01 -2.04267085e-01 -6.42854214e-01 -6.32705390e-01 1.07141204e-01 2.00313881e-01 -5.10092556e-01 6.71645105e-01 -1.71545282e-01 5.08931994e-01 -9.61551130e-01 -5.69390476e-01 9.56822485e-02 4.95852292e-01 -6.59536123e-01 3.57592881e-01 1.94826663e-01 4.16866392e-01 -2.78570622e-01 2.95288652e-01 6.81208611e-01 -4.17039007e-01 3.79143715e-01 -5.63324094e-01 -1.59636140e-01 -2.37579092e-01 -1.85793078e+00 1.82977796e+00 -4.80037518e-02 2.53920518e-02 4.44114059e-01 -1.09407306e+00 5.84695220e-01 2.55582631e-01 1.15945446e+00 1.83109388e-01 2.30425652e-02 3.07875663e-01 -1.97387561e-01 6.41241251e-03 2.94307858e-01 -1.04016237e-01 1.45928711e-01 6.50534034e-01 -2.36801095e-02 4.82626408e-02 4.34173375e-01 5.97197235e-01 7.48242259e-01 2.00473279e-01 1.51115090e-01 -5.96293449e-01 5.81107974e-01 1.28449783e-01 7.09002316e-01 5.41157663e-01 7.62485266e-02 8.86198103e-01 2.58344352e-01 1.58152245e-02 -1.05238593e+00 -1.21912670e+00 -1.70308128e-01 9.63527501e-01 1.67668760e-01 -5.55234969e-01 -1.09730983e+00 -5.85067570e-01 -6.20578416e-02 3.64538223e-01 -5.86050212e-01 -1.32722974e-01 -3.35494220e-01 -9.90147173e-01 3.04027647e-01 5.44193387e-01 2.58002549e-01 -3.64565670e-01 -1.70430318e-01 7.81246275e-02 -2.14389563e-01 -8.96190345e-01 -8.26154649e-01 3.45766157e-01 -9.32712376e-01 -1.05189884e+00 -5.50855458e-01 -5.74508190e-01 1.00288546e+00 5.56654513e-01 7.22283065e-01 -6.09884709e-02 -4.70646508e-02 5.59158266e-01 -2.85197943e-01 -1.14901997e-02 -5.47623523e-02 1.45113096e-01 4.76982534e-01 6.32907271e-01 2.92757928e-01 -8.69985223e-01 -3.90372932e-01 3.91742527e-01 -1.14709532e+00 1.57047257e-01 5.86576343e-01 8.38880122e-01 9.28480208e-01 1.60270467e-01 2.13925168e-01 -9.97741103e-01 4.31366533e-01 -5.31405687e-01 -5.49003720e-01 1.21941492e-01 -6.47041917e-01 2.43111253e-01 8.18478584e-01 -6.33150637e-01 -8.51848841e-01 9.12142277e-01 2.62723505e-01 -1.07691419e+00 8.71429313e-03 4.69955146e-01 -3.50494117e-01 3.48486789e-02 6.67608321e-01 2.85899132e-01 1.25066787e-01 -5.06570458e-01 6.53881550e-01 5.34597576e-01 6.78181529e-01 -7.96029210e-01 1.35827780e+00 8.56714427e-01 -4.99708578e-02 -8.83641243e-01 -8.79788458e-01 -1.18796539e+00 -1.07864666e+00 3.26504372e-02 7.97396302e-01 -8.76519978e-01 -3.41602683e-01 2.84909576e-01 -5.10655165e-01 -1.30657718e-01 -4.23677236e-01 5.44504523e-01 -6.39280260e-01 5.84378898e-01 -2.87904024e-01 -7.36752272e-01 -9.43793431e-02 -1.12242162e+00 1.02191126e+00 -1.46461636e-01 -2.80667990e-01 -9.74927008e-01 2.80841500e-01 4.46637601e-01 2.28893813e-02 1.65548816e-01 7.12856889e-01 -5.57415426e-01 -5.13352990e-01 5.02478033e-02 4.10912223e-02 1.95725515e-01 7.37692490e-02 -8.63526911e-02 -8.74801040e-01 -5.12707770e-01 1.70743410e-02 -1.31989226e-01 1.01630056e+00 4.37690735e-01 8.82627964e-01 -1.54709384e-01 -4.28234726e-01 9.96076405e-01 1.40968871e+00 -8.85418504e-02 2.62915909e-01 1.96783897e-02 1.10083354e+00 6.74142897e-01 4.43038017e-01 3.57324094e-01 2.56125927e-01 8.02390039e-01 5.04785627e-02 -3.09779704e-01 1.66353151e-01 -1.07353419e-01 2.94888586e-01 1.11177564e+00 -1.99540611e-02 2.98916578e-01 -1.01493943e+00 6.78586364e-01 -2.38209558e+00 -9.59208488e-01 -6.11362457e-02 2.41322255e+00 9.23434675e-01 -3.44426334e-01 5.43873072e-01 3.49356562e-01 6.94861531e-01 -4.41483893e-02 -6.60625756e-01 3.61038446e-01 -2.31765181e-01 -1.49524048e-01 4.52692688e-01 5.62399089e-01 -1.36513567e+00 8.10473263e-01 6.04941845e+00 8.01001191e-01 -1.05727315e+00 -3.42800491e-03 3.01634908e-01 -2.42233813e-01 -2.69770801e-01 2.56915271e-01 -4.87750500e-01 3.02277684e-01 5.63050568e-01 -2.20945984e-01 7.43997633e-01 9.46818769e-01 2.52809346e-01 1.04971007e-01 -1.26129270e+00 1.13627005e+00 -5.09552844e-02 -1.09320772e+00 8.90747309e-02 3.44176203e-01 1.00077081e+00 -2.32675135e-01 -1.25183363e-03 -1.85250640e-01 5.11775970e-01 -7.22507060e-01 7.53240824e-01 4.09655035e-01 5.80202818e-01 -8.23922694e-01 4.40382063e-01 3.58081043e-01 -1.29320240e+00 -1.06448486e-01 -2.38136828e-01 2.25653604e-01 1.14993248e-02 8.64763319e-01 -5.68332791e-01 6.09549046e-01 7.07269907e-01 7.31437266e-01 -4.37293857e-01 9.06516492e-01 1.41378865e-01 7.13182330e-01 -6.41887724e-01 5.32988250e-01 1.70871630e-01 -8.58853638e-01 7.27149546e-01 1.19891560e+00 2.19878331e-01 1.26441926e-01 5.83999693e-01 1.05055737e+00 6.61735311e-02 3.77336770e-01 -4.74241734e-01 -1.39024585e-01 6.13739729e-01 1.49263370e+00 -1.01505566e+00 -3.40131462e-01 -3.42797697e-01 1.04388928e+00 3.92608553e-01 6.70823097e-01 -6.89341545e-01 -1.08931519e-01 6.51139617e-01 1.26088113e-01 2.44839340e-01 -4.36916143e-01 -5.06345570e-01 -1.23900855e+00 -3.29888798e-02 -8.57617974e-01 5.32610297e-01 -3.55151772e-01 -1.38110220e+00 2.75096387e-01 1.27662376e-01 -1.19572783e+00 -1.76724657e-01 -2.48375759e-01 -4.08941418e-01 7.84593940e-01 -9.07373250e-01 -1.21167636e+00 -3.37717026e-01 8.99991930e-01 2.16030598e-01 -3.76051702e-02 6.32627666e-01 4.14511591e-01 -9.38935161e-01 4.95465457e-01 5.08982778e-01 1.26789525e-01 8.56190920e-01 -1.46546137e+00 -2.27504626e-01 1.10693586e+00 3.08699518e-01 1.20967293e+00 6.60347998e-01 -5.35499871e-01 -1.59013879e+00 -1.06010127e+00 2.51033872e-01 -4.15776938e-01 7.09790111e-01 -5.04697323e-01 -8.77066970e-01 6.16878927e-01 -1.56256676e-01 -1.03879668e-01 9.71024156e-01 3.34166288e-01 -4.50638026e-01 -1.53307214e-01 -8.81753862e-01 6.05026841e-01 1.07822478e+00 -6.39418066e-01 -3.98570716e-01 2.38023758e-01 3.28027248e-01 -1.68335915e-01 -8.15334499e-01 3.03349912e-01 5.67273915e-01 -8.95862043e-01 1.07369328e+00 -3.76746804e-01 3.77894826e-02 -7.14624107e-01 -3.27382267e-01 -1.25873828e+00 -6.58672988e-01 -6.72164738e-01 -2.11150497e-01 1.36407161e+00 2.82212257e-01 -4.07136470e-01 9.01536882e-01 6.55960679e-01 -1.58721045e-01 -6.23045623e-01 -6.39025688e-01 -8.35993946e-01 -7.50758350e-02 -2.84999192e-01 4.56995279e-01 1.29330873e+00 -7.22399801e-02 4.33663040e-01 -2.52186894e-01 3.76139343e-01 1.15002418e+00 3.98659408e-01 9.50225651e-01 -1.41620803e+00 -3.80749851e-01 -3.78902227e-01 -1.79525688e-01 -9.68454182e-01 2.05852330e-01 -9.84432757e-01 6.82149082e-02 -1.16131449e+00 6.78743839e-01 -4.47165668e-01 -2.66651571e-01 3.45273912e-01 -2.16648459e-01 3.04248899e-01 1.32730037e-01 7.39873290e-01 -5.89443445e-01 5.08112550e-01 7.68443108e-01 -9.80584323e-02 -6.70535922e-01 -2.82387376e-01 -8.32219005e-01 7.82231688e-01 3.10019314e-01 -3.28445852e-01 -7.82913029e-01 -6.66528568e-02 1.49743119e-02 -3.38401973e-01 9.18138325e-02 -9.46265221e-01 4.99122828e-01 -2.92519957e-01 1.89203545e-01 -4.62859273e-01 3.31680775e-01 -1.05537236e+00 4.98756707e-01 8.70802533e-03 -1.99996114e-01 -3.36737543e-01 -1.74025044e-01 6.62883341e-01 -2.24254966e-01 -4.73237634e-02 9.68644202e-01 1.14193648e-01 -3.29663038e-01 3.47937703e-01 -1.24043673e-01 5.97824976e-02 1.00848949e+00 -4.76079077e-01 3.68185818e-01 -3.76839966e-01 -9.61832225e-01 3.16486597e-01 9.22385573e-01 1.83614492e-01 1.85990795e-01 -1.55910492e+00 -5.75374782e-01 3.44333798e-01 3.40959132e-02 1.48865938e-01 8.81162062e-02 1.30427647e+00 -9.77659002e-02 1.54470980e-01 1.69297695e-01 -9.67865586e-01 -1.27565444e+00 7.33699262e-01 1.20470315e-01 -1.46548524e-01 -5.76936901e-01 5.72818100e-01 5.27231336e-01 -4.08517361e-01 2.30676189e-01 1.86441094e-01 -5.61541468e-02 3.67180228e-01 1.51815236e-01 4.73103225e-01 -7.24891424e-02 -9.81566370e-01 -3.42828810e-01 8.47644806e-01 8.44671652e-02 -3.23220551e-01 1.31144977e+00 -2.21742168e-01 -2.48201892e-01 6.00821257e-01 1.26594770e+00 2.67266184e-01 -1.36974061e+00 -3.49066824e-01 4.99063507e-02 -3.89006644e-01 1.69937804e-01 -2.61097521e-01 -1.04902136e+00 4.77748156e-01 3.93216789e-01 7.10365996e-02 1.26507878e+00 3.76136079e-02 5.61736047e-01 3.58360022e-01 1.91671848e-01 -1.43683994e+00 8.28864127e-02 2.37279475e-01 6.67861223e-01 -1.00330043e+00 1.87695920e-01 -8.59105229e-01 -5.67157149e-01 7.83820093e-01 2.96209246e-01 -5.51541448e-02 6.40240848e-01 2.36462772e-01 1.16841421e-01 -2.26589456e-01 -1.93286195e-01 -2.08099067e-01 6.01510108e-01 3.53158861e-01 3.47091109e-01 7.55038857e-02 -3.31801742e-01 6.44939125e-01 -2.00846523e-01 -3.69275540e-01 2.09204867e-01 6.98171973e-01 -2.57406890e-01 -1.29438555e+00 -6.74870491e-01 3.04737002e-01 -8.56716409e-02 4.36470620e-02 -8.39403212e-01 4.64181185e-01 -6.30604029e-02 9.03772950e-01 -3.68123390e-02 -3.66658956e-01 1.06945969e-01 2.75420994e-01 3.10505122e-01 -6.36693299e-01 -2.65479684e-01 6.34194136e-01 -1.95080474e-01 -6.99509978e-01 -6.37719512e-01 -1.04328299e+00 -1.45983779e+00 1.07650533e-02 -4.34718788e-01 5.00080645e-01 4.85436618e-01 7.57258356e-01 4.64218408e-01 1.73732162e-01 7.62563050e-01 -9.54620957e-01 -5.26191235e-01 -6.00727439e-01 -8.02971125e-01 9.26927745e-01 1.96741032e-03 -8.54697287e-01 -6.44375026e-01 2.57392645e-01]
[7.845475196838379, 4.361932277679443]
eec90b40-8b9b-46ed-9844-9315b0507f1c
cost-effective-photonic-super-resolution
2210.04280
null
https://arxiv.org/abs/2210.04280v1
https://arxiv.org/pdf/2210.04280v1.pdf
Cost-effective photonic super-resolution millimeter-wave joint radar-communication system using self-coherent detection
A cost-effective millimeter-wave (MMW) joint radar-communication (JRC) system with super resolution is proposed and experimentally demonstrated, using optical heterodyne up-conversion and self-coherent detection down-conversion techniques. The point lies in the designed coherent dual-band constant envelope linear frequency modulation-orthogonal frequency division multiplexing (LFM-OFDM) signal with opposite phase modulation indexes for the JRC system. Then the self-coherent detection, as a simple and low-cost means, is accordingly facilitated for both de-chirping of MMW radar and frequency down-conversion reception of MMW communication, which circumvents the costly high-speed mixers along with MMW local oscillators and more significantly achieves the real-time decomposition of radar and communication information. Furthermore, a super resolution radar range profile is realized through the coherent fusion processing of dual-band JRC signal. In experiments, a dual-band LFM-OFDM JRC signal centered at 54-GHz and 61-GHz is generated. The dual bands are featured with an identical instantaneous bandwidth of 2 GHz and carry an OFDM signal of 1 GBaud, which help to achieve a 6-Gbit/s data rate for communication and a 1.76-cm range resolution for radar.
['Bin Luo', 'Lianshan Yan', 'Wei Pan', 'Ningyuan Zhong', 'Xihua Zou', 'Peixuan Li', 'Wenlin Bai']
2022-10-09
null
null
null
null
['joint-radar-communication']
['robots']
[ 3.11824024e-01 -2.35587768e-02 -1.36847631e-03 -1.59934074e-01 -6.08500957e-01 -2.38956034e-01 5.83657205e-01 -7.34164178e-01 -4.63343114e-01 1.11272955e+00 2.01577842e-01 -2.58806586e-01 -7.05355942e-01 -1.00996017e+00 2.67460257e-01 -1.08520532e+00 -4.68375117e-01 9.67841372e-02 -2.65683651e-01 -1.22001901e-01 -4.11533862e-02 4.46342081e-01 -1.45865262e+00 -2.89940625e-01 1.03043497e+00 1.00662792e+00 2.42535517e-01 7.34881818e-01 2.42246658e-01 7.51949623e-02 -1.21826649e+00 -1.32014588e-01 2.37908930e-01 -6.32414222e-01 4.70319211e-01 -4.84844863e-01 4.75847095e-01 -4.67365444e-01 -7.59621859e-01 8.50798428e-01 3.19963276e-01 -4.00536537e-01 8.76548707e-01 -8.43126833e-01 -5.29515624e-01 4.44963664e-01 -6.40744805e-01 -1.25203148e-01 -1.35814652e-01 3.16454321e-01 6.26507461e-01 -5.84739223e-02 5.67328691e-01 8.99961948e-01 7.61019364e-02 2.51591384e-01 -1.19370651e+00 -1.41556132e+00 -9.96156812e-01 -1.34417899e-02 -1.29905522e+00 -5.07636845e-01 7.08298504e-01 -2.25463003e-01 6.37576282e-01 2.93724149e-01 6.92878127e-01 6.86475694e-01 8.56433094e-01 -7.69754350e-01 1.28778267e+00 -7.66384125e-01 -2.71589383e-02 -1.35320961e-01 8.29475373e-02 5.95564187e-01 9.70923722e-01 1.18652284e+00 -5.29518783e-01 9.85865109e-03 9.48227942e-01 -2.13828415e-01 -7.69318104e-01 2.40042165e-01 -1.19589400e+00 7.18847334e-01 2.15037599e-01 8.67132902e-01 -2.56593019e-01 5.85816085e-01 -7.79011071e-01 5.93597412e-01 6.78419545e-02 8.41648102e-01 4.23074007e-01 3.35929722e-01 -1.02746081e+00 3.91185246e-02 6.45063818e-01 9.77340877e-01 6.93100154e-01 4.82453436e-01 -2.92671859e-01 3.34770888e-01 7.26673484e-01 1.39134765e+00 2.70490766e-01 -7.51765788e-01 3.68642479e-01 -3.09239119e-01 2.25741908e-01 -7.82744527e-01 -4.56451714e-01 -1.51034546e+00 -1.06611907e+00 2.44537681e-01 1.59830943e-01 -3.04892808e-01 -1.00420475e+00 1.69999158e+00 -8.45586061e-02 6.48235381e-02 7.99977601e-01 7.87823796e-01 6.02881491e-01 7.92000115e-01 -2.97781914e-01 -7.06765413e-01 2.03658485e+00 -2.60981739e-01 -1.27860582e+00 -5.40560961e-01 3.07307057e-02 -1.08626997e+00 -1.75609425e-01 2.13312939e-01 -6.83414340e-01 -5.62309086e-01 -1.70600283e+00 3.00971061e-01 4.25762028e-01 1.17707811e-02 3.93140614e-01 1.05141461e+00 -1.35911077e-01 1.30355686e-01 -1.60695583e-01 3.76515090e-01 8.60508904e-02 -5.49996085e-02 1.41424552e-01 -1.70921117e-01 -1.13595879e+00 6.76571131e-01 -1.25121295e-01 -4.62060384e-02 -2.53385697e-02 -1.24045503e+00 -2.02306166e-01 -1.69245824e-01 -3.14660668e-01 -1.03099322e+00 6.45887256e-01 2.24152967e-01 -1.67885697e+00 4.79023993e-01 -1.07552325e-02 -7.46602714e-01 1.17491238e-01 3.99032310e-02 -1.32979000e+00 5.80236316e-01 2.04203606e-01 -1.17037751e-01 1.19070733e+00 -1.05021441e+00 -7.35614181e-01 -5.03953278e-01 -7.48894513e-01 -3.98182362e-01 3.33929747e-01 -5.07073700e-01 7.23715246e-01 -5.74530602e-01 8.11945796e-01 -4.44851935e-01 -9.60053056e-02 -5.63083589e-01 -1.44421950e-01 4.62811053e-01 1.03825617e+00 4.17601876e-02 1.00474954e+00 -2.42916703e+00 -4.29367244e-01 -5.56233153e-02 1.99856386e-01 -2.64803809e-03 -1.11704975e-01 7.28745162e-01 1.55034110e-01 -3.35886002e-01 -1.20225556e-01 1.46707952e-01 -1.94544360e-01 -3.37583214e-01 -5.68024993e-01 6.20624185e-01 1.05670966e-01 5.80465853e-01 -5.01543581e-01 -1.51474953e-01 -1.47190943e-01 3.88875932e-01 -2.96404302e-01 -8.07125568e-02 1.36882335e-01 6.56606615e-01 -4.52127725e-01 8.15223038e-01 1.41684926e+00 4.58936185e-01 8.51222873e-02 -5.17248273e-01 -7.29365170e-01 -4.85751405e-02 -9.64851916e-01 1.18852603e+00 -5.35193920e-01 6.54097557e-01 4.43127155e-01 -7.78544128e-01 1.69966185e+00 1.42292574e-01 3.25071812e-01 -1.27110839e+00 -2.16915965e-01 6.34945512e-01 2.21829727e-01 -4.84586030e-01 3.99432182e-01 -8.15202236e-01 -4.42462385e-01 4.54138190e-01 4.87497419e-01 -4.70433265e-01 3.52389887e-02 -2.86462516e-01 1.00910807e+00 -1.58437237e-01 3.45642656e-01 -3.43697727e-01 5.01292586e-01 -6.72801137e-02 4.85338986e-01 6.87667549e-01 2.49249235e-01 -1.46511987e-01 -3.22758555e-01 1.14613257e-01 -5.56471884e-01 -1.28502166e+00 -4.98375326e-01 -1.83793306e-01 6.76717460e-01 -3.35633382e-02 1.62632391e-01 6.77137673e-01 4.99528944e-01 7.56465316e-01 5.48912406e-01 -1.78029746e-01 -5.81971109e-01 -6.85410202e-01 5.35089731e-01 -4.43406910e-01 1.06933451e+00 -2.10129227e-02 -7.53857851e-01 6.30852580e-01 1.37240022e-01 -1.59972346e+00 2.47373283e-01 1.50528610e-01 -8.77708554e-01 -7.68305063e-01 -2.80668736e-01 -6.10951424e-01 2.56416142e-01 5.12069881e-01 5.02925992e-01 -7.73591101e-01 -9.96520042e-01 2.09186926e-01 5.44194393e-02 -5.52124456e-02 -6.23227730e-02 -5.03877997e-01 2.69046605e-01 -6.28240630e-02 3.66968364e-01 -1.17242694e+00 -7.57016838e-01 1.73992679e-01 -1.43516421e-01 1.33708129e-02 1.26240253e+00 5.17693877e-01 -2.45411769e-01 2.90467620e-01 9.02353585e-01 -8.34697336e-02 3.93629253e-01 -3.28016877e-02 -1.23041081e+00 -1.50283098e-01 -4.12200570e-01 1.43545523e-01 4.85904276e-01 -1.85063764e-01 -9.60199773e-01 -4.56170589e-01 1.30234480e-01 5.20415902e-01 -3.94940525e-02 2.68992811e-01 5.40141240e-02 -3.42592984e-01 5.73193014e-01 4.65958178e-01 1.36015430e-01 5.18052056e-02 6.45238757e-01 9.56708431e-01 9.97041345e-01 -2.33145684e-01 2.02427268e+00 5.36310136e-01 5.18448710e-01 -1.56743813e+00 -3.74402881e-01 -2.37614706e-01 1.43469095e-01 -1.08439848e-01 8.86111379e-01 -1.37945700e+00 -8.49589646e-01 3.15200269e-01 -1.17583656e+00 4.49133992e-01 -1.22207083e-01 1.59898317e+00 -3.13896835e-01 2.86104977e-01 -2.98616529e-01 -1.34897995e+00 -3.51100594e-01 -2.94862360e-01 5.87692678e-01 6.06657624e-01 -1.30092755e-01 -1.10064216e-01 2.06916369e-02 4.29148316e-01 1.08932436e+00 5.36974728e-01 9.64870214e-01 4.99301910e-01 -1.69443464e+00 -3.39485019e-01 -1.92409560e-01 -3.46401155e-01 2.32459858e-01 -5.59700787e-01 -7.60811985e-01 -2.48005092e-01 5.17513871e-01 3.00555348e-01 9.60610688e-01 6.13525331e-01 -3.48618686e-01 5.16185053e-02 -6.58463061e-01 8.52747798e-01 2.01595759e+00 5.71417153e-01 7.09655285e-01 -1.29463717e-01 -3.94524187e-01 2.70409495e-01 6.58062756e-01 2.73210466e-01 -9.58936736e-02 5.25227010e-01 1.37477979e-01 3.84202391e-01 -2.23359883e-01 2.12786600e-01 1.51942298e-01 4.38348323e-01 -3.31759334e-01 -1.15589850e-01 1.00554466e-01 5.27280681e-02 -1.23330915e+00 -1.26193178e+00 -4.42132860e-01 2.24692845e+00 5.05167484e-01 4.75831106e-02 -5.75982809e-01 -2.40177318e-01 5.86558878e-01 4.66250896e-01 1.03908092e-01 3.31669778e-01 -1.77689344e-01 1.04235697e+00 8.75660896e-01 9.92479146e-01 -7.43644178e-01 2.93282777e-01 4.83288813e+00 7.47675776e-01 -8.98105025e-01 1.10377274e-01 -4.18596178e-01 -1.48842394e-01 -4.94260013e-01 7.88656101e-02 -1.04821384e+00 4.87920016e-01 8.82422924e-01 -5.42413652e-01 -1.02866432e-02 1.50347784e-01 -5.15049584e-02 -6.45920217e-01 -5.67636132e-01 1.21361971e+00 -3.41871738e-01 -1.35112035e+00 -3.72833222e-01 6.64030254e-01 3.01892847e-01 -6.01501346e-01 1.17803104e-01 4.41867262e-01 -1.60765767e-01 -5.96233904e-01 2.07715690e-01 8.91100943e-01 1.08506429e+00 -7.97377229e-01 4.15718526e-01 6.50411665e-01 -1.40424562e+00 -3.63835424e-01 -3.71357501e-01 -1.39934808e-01 3.83165836e-01 1.59721100e+00 -3.71340603e-01 9.22611654e-01 -1.60195809e-02 -2.04761997e-01 4.60557580e-01 4.52464938e-01 -3.20890665e-01 4.20696855e-01 -4.82156366e-01 -1.44934252e-01 -2.06323087e-01 -7.23127425e-01 9.32131350e-01 9.01665330e-01 1.18743360e+00 5.66395164e-01 -1.69362828e-01 1.19929564e+00 5.20608500e-02 -6.63584471e-01 -5.71529925e-01 -2.67118603e-01 9.12283182e-01 1.40605032e+00 -1.07770287e-01 1.10804856e-01 -3.12411766e-02 2.87648231e-01 -7.76630223e-01 5.89320183e-01 -8.42607558e-01 -1.22437346e+00 1.22723317e+00 1.00512087e-01 3.70328546e-01 -1.19382203e+00 3.69059965e-02 -7.78322220e-01 -1.09861284e-01 1.18383326e-01 -2.83362389e-01 -1.53852537e-01 -9.10558224e-01 3.92098159e-01 -3.83303672e-01 -1.80447972e+00 -1.96311742e-01 -6.16962872e-02 -2.62524813e-01 1.20416582e+00 -2.36454368e+00 -8.02276850e-01 -5.07587373e-01 1.42119169e-01 -4.57096249e-01 -3.59857887e-01 1.13828671e+00 2.12041348e-01 3.06691974e-01 -3.68675636e-03 2.88953900e-01 -6.08892553e-02 9.41569328e-01 -6.05105221e-01 -1.83687642e-01 7.80785918e-01 -2.50860155e-01 7.94789374e-01 6.41127169e-01 -7.56125152e-01 -1.71150291e+00 -1.06818855e+00 7.96995580e-01 3.74090701e-01 5.04405677e-01 -4.49348241e-01 9.08258930e-02 -9.85325202e-02 5.39902389e-01 -5.02208993e-02 9.87976193e-01 -5.06139338e-01 -3.83822024e-01 -6.39652491e-01 -1.44016385e+00 2.70804167e-01 1.02338850e+00 -2.12669432e-01 -7.00198948e-01 2.79813290e-01 7.46999204e-01 -3.03852081e-01 -8.66858184e-01 6.07529938e-01 9.03318524e-01 -8.89142811e-01 1.08836091e+00 1.64201558e-01 3.91956419e-01 -8.19149613e-01 -5.75824559e-01 -1.16220725e+00 -7.85183907e-01 -1.03911757e+00 7.48533234e-02 1.11950374e+00 5.95729686e-02 -1.58111024e+00 5.26603520e-01 -7.83554614e-01 1.23007998e-01 -2.91975252e-02 -1.55491734e+00 -1.38578820e+00 -4.66036201e-01 -1.00487903e-01 2.71687984e-01 6.42869234e-01 4.79589514e-02 7.37656295e-01 -2.93281138e-01 9.17261899e-01 1.63465428e+00 8.24789941e-01 7.78567433e-01 -1.66599047e+00 -5.27281761e-01 -3.60842705e-01 -6.83854043e-01 -1.20173705e+00 -3.81432921e-01 -9.34616506e-01 -3.05941880e-01 -1.44896102e+00 -7.94134438e-01 -2.82894790e-01 3.71540010e-01 -5.84428966e-01 7.35565364e-01 5.70746899e-01 1.95882663e-01 1.30983844e-01 5.56331694e-01 7.63161182e-01 1.19071138e+00 -8.28956589e-02 -1.17022365e-01 4.33629125e-01 -6.06686413e-01 1.63577460e-02 4.45447266e-01 -3.24821562e-01 -1.44757047e-01 7.70834684e-02 -2.83549249e-01 5.95940709e-01 3.65135312e-01 -1.64347649e+00 5.99246621e-01 -1.63855255e-01 4.37844634e-01 -6.21815562e-01 7.41400719e-01 -8.08039010e-01 5.85483193e-01 1.30719829e+00 5.82418621e-01 -1.12023497e+00 -3.32061678e-01 8.28357577e-01 -3.48086059e-01 -1.33097157e-01 1.05795979e+00 -7.25425780e-02 -2.27033794e-01 -2.35615909e-01 -6.10602140e-01 -2.07807019e-01 1.30173659e+00 -2.95505643e-01 -1.40215242e+00 -3.73498768e-01 -2.49673977e-01 1.05768315e-01 -3.23052973e-01 -8.19354951e-02 5.17096341e-01 -1.48380697e+00 -9.20083165e-01 5.91975808e-01 -3.80819365e-02 -5.54219067e-01 4.76928651e-01 6.41891658e-01 -4.24220532e-01 9.46027219e-01 -3.02438289e-01 -7.03161180e-01 -1.03642046e+00 -3.48392934e-01 3.27963293e-01 -2.21637145e-01 -6.98233545e-01 3.30540091e-01 -4.18260634e-01 2.82455683e-01 -4.52966332e-01 -2.55850196e-01 -4.12225164e-02 2.25478590e-01 6.51802003e-01 -1.28784671e-01 -2.29639128e-01 5.01727080e-03 -4.71710920e-01 1.30918372e+00 3.99565488e-01 -3.87782961e-01 1.30855560e+00 1.32807627e-01 -2.59025514e-01 -1.47102356e-01 8.61434281e-01 8.05278480e-01 -4.22248602e-01 -3.42166185e-01 -2.45100364e-01 -6.47910297e-01 3.58749703e-02 -4.47898567e-01 -5.41039586e-01 3.90071601e-01 7.54757047e-01 4.96760875e-01 8.98129225e-01 -6.56179711e-02 7.37056494e-01 6.71415865e-01 9.91084456e-01 -1.00012851e+00 -6.71472773e-02 1.12035133e-01 5.37560284e-01 -4.33973610e-01 7.09557950e-01 -5.85838318e-01 2.95788109e-01 1.51746631e+00 1.24607161e-01 -3.14399511e-01 6.61998928e-01 6.39350891e-01 -2.09655818e-02 -2.91266620e-01 -4.79970157e-01 -3.68353933e-01 -3.17487210e-01 9.16047215e-01 3.55382502e-01 1.55624688e-01 -7.38821685e-01 6.37661219e-01 -8.48521233e-01 -2.35484138e-01 8.55983853e-01 6.38101161e-01 -1.14682186e+00 -9.60019588e-01 -7.61321187e-01 2.29076892e-01 1.54784828e-01 1.80822939e-01 4.14117038e-01 9.10721242e-01 1.43564016e-01 1.13939595e+00 2.28142291e-01 -2.05092102e-01 6.26388967e-01 2.21617863e-01 8.78546655e-01 -3.71531725e-01 3.58807325e-01 4.55688894e-01 5.03897071e-02 -2.25587189e-01 -4.54184860e-01 -1.45587400e-01 -1.06158710e+00 5.30835614e-02 -3.46397370e-01 4.30622607e-01 7.59721875e-01 6.36103034e-01 3.56689274e-01 5.54713488e-01 1.18637598e+00 -3.40490341e-01 -6.06479704e-01 -7.69125938e-01 -1.16400743e+00 -7.40708292e-01 7.99030721e-01 -5.24133921e-01 -7.88898408e-01 -5.55314124e-01]
[6.4202399253845215, 1.2502952814102173]
61ea3486-93b0-4a69-9987-bc98045f34c4
uniform-convergence-with-square-root
2306.13188
null
https://arxiv.org/abs/2306.13188v1
https://arxiv.org/pdf/2306.13188v1.pdf
Uniform Convergence with Square-Root Lipschitz Loss
We establish generic uniform convergence guarantees for Gaussian data in terms of the Rademacher complexity of the hypothesis class and the Lipschitz constant of the square root of the scalar loss function. We show how these guarantees substantially generalize previous results based on smoothness (Lipschitz constant of the derivative), and allow us to handle the broader class of square-root-Lipschitz losses, which includes also non-smooth loss functions appropriate for studying phase retrieval and ReLU regression, as well as rederive and better understand "optimistic rate" and interpolation learning guarantees.
['Nathan Srebro', 'Frederic Koehler', 'Zhen Dai', 'Lijia Zhou']
2023-06-22
null
null
null
null
['retrieval']
['methodology']
[-2.90014502e-02 3.19619834e-01 -2.49904290e-01 -3.63376230e-01 -1.34189510e+00 -4.94380295e-01 2.27272734e-01 4.69872952e-01 -5.17713606e-01 8.06169152e-01 -3.97459827e-02 -2.06720948e-01 -5.38833201e-01 -6.45412266e-01 -8.92202139e-01 -1.10346508e+00 -5.83284736e-01 3.44811410e-01 2.98147142e-01 -8.87620673e-02 9.16736871e-02 6.32021546e-01 -9.60801840e-01 -3.17653477e-01 6.81346655e-01 1.38114750e+00 -3.57667625e-01 3.80614191e-01 6.64992869e-01 3.46495599e-01 -2.01963887e-01 -2.05083102e-01 1.68065205e-01 6.11216426e-02 -6.90364301e-01 -2.41818801e-01 6.36100888e-01 -2.59269495e-03 -3.83876830e-01 1.15802360e+00 5.36775291e-01 4.92660165e-01 9.49451745e-01 -9.60995615e-01 -4.85227853e-01 2.09662169e-01 -5.19777894e-01 1.17222413e-01 -2.57318132e-02 -3.00771952e-01 1.15775728e+00 -6.77196562e-01 5.77131093e-01 1.16197896e+00 1.22042739e+00 3.72858793e-01 -1.32905197e+00 -5.10273516e-01 1.28018439e-01 -1.03392102e-01 -1.28911543e+00 -4.60864604e-01 3.86565238e-01 -5.66497147e-01 2.78748035e-01 1.29706964e-01 -1.75480977e-01 8.05197537e-01 4.24406826e-01 6.85520470e-01 7.43796766e-01 -3.65384609e-01 2.21040949e-01 3.26858342e-01 2.80939311e-01 9.29484785e-01 2.71516591e-01 2.14571923e-01 -2.65902072e-01 -2.60241300e-01 8.95587981e-01 -3.50774914e-01 -6.75318539e-01 -6.53848827e-01 -8.72437060e-01 1.15998149e+00 6.29362166e-01 1.09902643e-01 -8.64016190e-02 4.34392810e-01 3.00285071e-01 4.57890958e-01 7.58088350e-01 3.87686133e-01 -3.58170331e-01 1.32571891e-01 -6.89211965e-01 3.02730322e-01 8.33699942e-01 1.08320105e+00 7.22255468e-01 1.57192811e-01 -2.95941174e-01 5.00939548e-01 3.03945810e-01 7.33909845e-01 -1.03543118e-01 -1.18394220e+00 2.99005836e-01 -3.23210627e-01 3.95741254e-01 -5.73486984e-01 -4.31473136e-01 -5.14983058e-01 -7.01194286e-01 3.84374291e-01 5.98381102e-01 -1.84191078e-01 -4.35324639e-01 1.85147119e+00 2.30441555e-01 5.23692407e-02 -9.69264433e-02 8.56631637e-01 2.07707897e-01 2.77377933e-01 -2.17173263e-01 -3.01973969e-01 8.75209272e-01 -5.10202587e-01 -4.31985766e-01 1.72985241e-01 6.46647513e-01 -5.70398271e-01 1.21283019e+00 4.16858703e-01 -1.31698310e+00 -2.93409020e-01 -1.06240630e+00 -2.82847404e-01 6.28234297e-02 -1.10182613e-01 6.33452415e-01 3.04707140e-01 -9.52963114e-01 9.51331139e-01 -9.64231551e-01 2.63342768e-01 3.51041675e-01 1.55367911e-01 -3.26031893e-01 -7.45642511e-03 -9.48204815e-01 6.54917002e-01 9.64525491e-02 -9.32996124e-02 -7.16512561e-01 -1.29078400e+00 -8.65632057e-01 7.00449347e-02 -9.27496422e-03 -4.88985747e-01 1.04804134e+00 -7.29337513e-01 -1.29682541e+00 7.57210135e-01 9.64603275e-02 -6.19689167e-01 9.25115108e-01 -3.60771418e-01 9.40875560e-02 1.39988333e-01 -2.20602632e-01 9.62961093e-02 9.42248404e-01 -6.98459327e-01 -2.97667384e-01 -6.05488777e-01 4.25763540e-02 -8.18756148e-02 -1.82314068e-02 -2.89135009e-01 1.06189169e-01 -7.11289108e-01 4.90465760e-02 -5.07111430e-01 -3.11680913e-01 6.92922354e-01 1.92054123e-01 -1.71409801e-01 5.78368783e-01 -6.89231396e-01 8.81793320e-01 -2.57794881e+00 5.63960113e-02 3.07688624e-01 -6.51585730e-03 -3.43334258e-01 5.46141006e-02 1.10058829e-01 -4.10804944e-03 -6.63939938e-02 -4.71749574e-01 -4.61330801e-01 1.83490053e-01 1.92981333e-01 -4.43509489e-01 1.08619344e+00 1.22440301e-01 4.53377575e-01 -9.18643057e-01 -9.36497301e-02 -2.44506314e-01 6.18886292e-01 -6.79904103e-01 -1.12642840e-01 -1.39109850e-01 3.18962038e-01 -6.40442610e-01 1.77395195e-01 7.03044653e-01 -2.78205186e-01 -5.08475244e-01 -1.70987710e-01 -4.46277484e-02 3.36683929e-01 -1.35452461e+00 1.41480803e+00 -6.63697362e-01 8.54723334e-01 7.72837877e-01 -1.06399870e+00 6.23164475e-01 2.66538799e-01 6.12442255e-01 -1.18834890e-01 -6.36457354e-02 3.40388864e-01 -6.86487734e-01 -2.20836714e-01 9.30451900e-02 -7.17921376e-01 1.91303626e-01 -3.14787142e-02 -1.55616134e-01 -1.66455582e-01 -1.32139787e-01 -1.66384608e-01 7.31949806e-01 2.64713299e-02 -4.73213196e-02 -9.28347111e-01 4.19078439e-01 -3.71270686e-01 4.99349892e-01 6.01574659e-01 -1.20413199e-01 6.33498669e-01 5.07202327e-01 -3.10490523e-02 -1.07643104e+00 -1.27208614e+00 -9.20293033e-01 1.02771556e+00 2.20754638e-01 1.11485474e-01 -2.16134161e-01 -6.82086647e-01 4.93006021e-01 7.75625229e-01 -7.15237796e-01 -4.61174510e-02 -3.02130789e-01 -8.02677989e-01 5.34556627e-01 7.32437134e-01 1.02709651e-01 -2.40454197e-01 8.80928934e-02 1.37495773e-03 3.34283710e-01 -9.54271674e-01 -7.21932650e-01 4.56530154e-01 -7.26498306e-01 -1.11910069e+00 -6.19689226e-01 -8.22439373e-01 4.10848141e-01 -3.13291639e-01 8.57511759e-01 -4.33362812e-01 -1.65920720e-01 6.88393593e-01 7.51322284e-02 -2.81376123e-01 -8.11428800e-02 -3.54838461e-01 5.62827736e-02 -1.88082065e-02 -2.23874271e-01 -5.34462094e-01 -4.65372205e-01 8.60536546e-02 -7.23666131e-01 -7.38546968e-01 3.52017619e-02 7.90031314e-01 8.88550520e-01 1.61656857e-01 4.53329146e-01 -7.60683060e-01 4.64881510e-01 -4.84325767e-01 -1.30262160e+00 2.12759212e-01 -9.65996861e-01 4.91652995e-01 5.32583058e-01 -4.97023195e-01 -8.14203799e-01 -5.68866909e-01 -2.76235759e-01 -3.46587181e-01 4.26227450e-01 1.13992922e-01 2.11542517e-01 -7.19818532e-01 8.33484054e-01 -1.34684145e-01 -9.05296579e-02 -5.73337436e-01 3.71201396e-01 4.29468900e-01 8.00051808e-01 -8.78663480e-01 8.10722232e-01 7.14300454e-01 6.32498085e-01 -8.79129529e-01 -9.25397694e-01 -5.64115345e-01 -1.56944439e-01 2.62197673e-01 6.61374152e-01 -7.71795034e-01 -8.95881534e-01 1.43154651e-01 -7.62640834e-01 -4.03086871e-01 -7.99308121e-01 7.49530911e-01 -1.05740702e+00 4.93042916e-01 -9.59671319e-01 -9.79593754e-01 -3.14989716e-01 -9.13943529e-01 1.19640410e+00 -2.38702893e-02 3.91997486e-01 -1.54949534e+00 -1.49243148e-02 -2.74885386e-01 5.37868321e-01 4.63091075e-01 1.00052381e+00 -4.14958507e-01 -4.17150229e-01 -2.92806983e-01 -3.03777635e-01 7.21482456e-01 -4.26520884e-01 -8.33712220e-02 -9.11297858e-01 -4.75965977e-01 4.95201588e-01 -3.55597109e-01 1.00242853e+00 6.84291720e-01 1.07518482e+00 -4.66815203e-01 -2.05060840e-01 1.12295020e+00 1.55191767e+00 -5.99653482e-01 3.93860698e-01 1.98073849e-01 4.29630667e-01 4.72545177e-01 5.01722574e-01 5.61810076e-01 6.25764504e-02 5.02369165e-01 2.45565474e-01 8.23189318e-03 6.25383109e-02 -4.73229811e-02 3.78343523e-01 4.68841791e-01 1.76908061e-01 1.65990666e-01 -5.09806454e-01 3.71729851e-01 -1.74827003e+00 -6.46332622e-01 -6.18345290e-02 2.70913100e+00 9.41658258e-01 1.97602898e-01 4.98989858e-02 -9.08497255e-03 4.35056865e-01 1.84940472e-01 -5.00698447e-01 -1.82188943e-01 -3.22433889e-01 4.48573530e-01 1.21391571e+00 1.30860686e+00 -1.25657523e+00 3.71821135e-01 7.60660601e+00 1.07747245e+00 -9.83409166e-01 2.17987761e-01 2.90724605e-01 1.45527348e-01 -5.24178028e-01 -1.00328326e-01 -7.79438376e-01 3.13743591e-01 6.75185919e-01 -5.65177500e-02 5.25188804e-01 9.64131236e-01 -2.58058798e-03 3.73264164e-01 -1.34078598e+00 6.56640172e-01 -4.43234414e-01 -1.13274133e+00 -4.76420909e-01 -2.53571700e-02 6.37556612e-01 2.50502408e-01 3.04470181e-01 1.76962510e-01 1.06747106e-01 -8.65859687e-01 4.90688652e-01 6.42354965e-01 8.15307736e-01 -6.83773279e-01 6.19645774e-01 2.58647978e-01 -9.99922931e-01 6.62659034e-02 -4.65454727e-01 2.00486869e-01 9.61619779e-04 7.20246494e-01 -4.20275450e-01 5.64794838e-01 2.99261540e-01 6.33255899e-01 -6.35002628e-02 1.32884228e+00 -1.25898555e-01 3.54624689e-01 -9.38408256e-01 1.61936343e-01 2.36945972e-01 -4.10288960e-01 7.15002358e-01 1.20181870e+00 1.73963636e-01 -1.47794917e-01 3.31740141e-01 6.70277596e-01 -1.60203546e-01 8.04117918e-02 -1.53964221e-01 2.02018455e-01 2.42693871e-01 7.54682720e-01 -3.62651125e-02 1.96721449e-01 -4.33615267e-01 6.20715976e-01 2.52297819e-01 7.25328922e-01 -6.72526240e-01 -6.77291512e-01 8.63032758e-01 5.00262916e-01 4.53895688e-01 -4.84845489e-01 -4.17583883e-01 -1.12658525e+00 2.89384574e-01 -2.68644065e-01 5.90534568e-01 -2.99463809e-01 -1.60735190e+00 2.31356055e-01 7.76143223e-02 -7.17637837e-01 -1.92448825e-01 -9.78916407e-01 -4.99083996e-01 9.83709753e-01 -1.80930352e+00 -9.13490355e-01 1.41641945e-01 4.56295818e-01 3.95883471e-02 1.42077401e-01 5.48598170e-01 4.19361860e-01 -1.94476277e-01 9.25151587e-01 6.92761838e-01 -7.70262256e-02 7.07078397e-01 -1.41933417e+00 -1.17776364e-01 5.74678600e-01 -2.91841894e-01 4.43380594e-01 1.07638371e+00 -8.17600936e-02 -1.25383151e+00 -8.42426121e-01 5.14844418e-01 -3.37021947e-01 1.09716725e+00 -1.66610524e-01 -1.02693450e+00 9.08561647e-01 -4.54163671e-01 2.72321999e-01 2.28460848e-01 3.06276381e-01 -5.77040076e-01 -3.76755923e-01 -1.48892605e+00 8.39468911e-02 5.95498204e-01 -5.87889373e-01 -6.85780570e-02 6.99573696e-01 7.77108729e-01 -4.12268132e-01 -1.35541487e+00 7.75783300e-01 4.04523969e-01 -4.36933398e-01 1.02320123e+00 -9.01471615e-01 -2.38860607e-01 -9.55929011e-02 -5.46228766e-01 -6.46254659e-01 -4.25812244e-01 -9.46701407e-01 -2.79148251e-01 7.49716222e-01 3.96607995e-01 -7.17931211e-01 6.15516663e-01 5.00543177e-01 -2.69043922e-01 -1.10776246e+00 -1.13515568e+00 -1.11632097e+00 5.69560707e-01 -3.06237608e-01 -1.28376007e-01 7.70802498e-01 -9.40475464e-02 3.74175310e-02 -1.46184206e-01 3.14379573e-01 1.03236687e+00 -1.07349299e-01 2.17851833e-01 -1.32783282e+00 -6.96905851e-01 -7.50541329e-01 -5.40492415e-01 -1.44306958e+00 1.14223145e-01 -8.20104539e-01 3.26189160e-01 -9.20669138e-01 -3.42363775e-01 -8.90334785e-01 -4.08696562e-01 4.07661647e-02 -2.97931843e-02 6.58801058e-03 -3.11081380e-01 1.40367612e-01 -5.45160115e-01 7.23103225e-01 9.76140082e-01 -1.17020093e-01 -9.59495604e-02 5.30508935e-01 -3.39532971e-01 7.69250214e-01 1.99681208e-01 -3.92673045e-01 -3.66096973e-01 -2.22340107e-01 4.34223443e-01 -8.54462013e-02 2.86960453e-01 -7.96719491e-01 3.13054919e-02 -2.04916209e-01 6.50667250e-02 -1.19902320e-01 2.72267848e-01 -5.56160271e-01 -2.67429024e-01 2.44865790e-01 -6.01151466e-01 -2.81094939e-01 -6.23612814e-02 7.59129643e-01 -4.60921191e-02 -4.82343435e-01 1.49465942e+00 3.23661983e-01 -2.04652488e-01 5.42339981e-01 2.01068670e-01 5.97279906e-01 7.53379405e-01 3.13651651e-01 -2.40962699e-01 -4.89064485e-01 -8.87430131e-01 3.71909231e-01 4.97461051e-01 -1.41425878e-01 3.21103990e-01 -1.31109345e+00 -8.09551537e-01 2.21681491e-01 -7.85396025e-02 5.84673136e-02 4.44557704e-03 1.14764237e+00 -5.06118178e-01 2.16434956e-01 3.93545508e-01 -2.98939496e-01 -5.98448038e-01 6.30637467e-01 6.77153051e-01 -1.69073984e-01 -6.83582067e-01 1.12479389e+00 2.78686911e-01 4.22744561e-05 8.27352285e-01 -4.25252676e-01 5.11134982e-01 -2.42049605e-01 5.25462031e-01 5.56526184e-01 1.23072706e-01 -5.08457236e-02 -4.10347909e-01 5.28441608e-01 2.12239236e-01 -1.00270823e-01 1.28702211e+00 -6.80842251e-02 1.13244645e-01 6.10146344e-01 1.58904219e+00 1.68040320e-01 -1.69758224e+00 -4.51832354e-01 4.19398770e-02 -1.95165619e-01 7.77316689e-02 -3.57519388e-01 -6.13556862e-01 9.89437163e-01 6.27012014e-01 5.32728769e-02 9.16720986e-01 1.88399956e-01 5.63162506e-01 4.64836121e-01 2.67367840e-01 -9.82995808e-01 -1.90121084e-01 4.36730087e-01 9.05446351e-01 -1.10978758e+00 1.60608560e-01 -4.24375892e-01 -1.17053866e-01 1.11323822e+00 5.36549203e-02 -5.54373205e-01 1.28277636e+00 2.97253251e-01 -2.90199965e-01 1.81049839e-01 -4.21751112e-01 -1.00219846e-01 6.39999926e-01 6.87385082e-01 4.75953579e-01 -3.85787785e-02 -3.20096314e-01 5.25545895e-01 1.86936092e-02 -3.82072240e-01 1.05951011e-01 5.62029123e-01 -7.14566410e-01 -8.96175325e-01 -5.43933660e-02 1.51753709e-01 -6.36192381e-01 -1.08173355e-01 2.55474031e-01 8.00742745e-01 -2.19322801e-01 4.30179924e-01 -6.70524463e-02 3.35902989e-01 2.70761013e-01 -1.52256778e-02 8.39922428e-01 -2.90956795e-01 3.47502120e-02 1.18023388e-01 4.36439738e-02 -4.91527110e-01 -8.15840140e-02 -5.90618849e-01 -1.15846896e+00 -5.28364837e-01 -4.15208131e-01 3.58205318e-01 7.01449335e-01 7.92437196e-01 -6.83422312e-02 3.89294401e-02 6.90552235e-01 -3.91578943e-01 -1.38000274e+00 -8.93671870e-01 -1.01199329e+00 3.82941842e-01 9.60655689e-01 -5.38542807e-01 -7.79277146e-01 -2.32740223e-01]
[7.277900695800781, 4.11313009262085]
f63c3a3c-9091-4b02-aad3-c14838678367
deep-convolutional-neural-network-based-1
2005.11780
null
https://arxiv.org/abs/2005.11780v1
https://arxiv.org/pdf/2005.11780v1.pdf
Deep Convolutional Neural Network-based Bernoulli Heatmap for Head Pose Estimation
Head pose estimation is a crucial problem for many tasks, such as driver attention, fatigue detection, and human behaviour analysis. It is well known that neural networks are better at handling classification problems than regression problems. It is an extremely nonlinear process to let the network output the angle value directly for optimization learning, and the weight constraint of the loss function will be relatively weak. This paper proposes a novel Bernoulli heatmap for head pose estimation from a single RGB image. Our method can achieve the positioning of the head area while estimating the angles of the head. The Bernoulli heatmap makes it possible to construct fully convolutional neural networks without fully connected layers and provides a new idea for the output form of head pose estimation. A deep convolutional neural network (CNN) structure with multiscale representations is adopted to maintain high-resolution information and low-resolution information in parallel. This kind of structure can maintain rich, high-resolution representations. In addition, channelwise fusion is adopted to make the fusion weights learnable instead of simple addition with equal weights. As a result, the estimation is spatially more precise and potentially more accurate. The effectiveness of the proposed method is empirically demonstrated by comparing it with other state-of-the-art methods on public datasets.
['Yang Xing', 'Jie Liu', 'Chen Lv', 'Zhongxu Hu', 'Peng Hang']
2020-05-24
null
null
null
null
['head-pose-estimation']
['computer-vision']
[-2.62606084e-01 -1.45935128e-03 -4.81723845e-02 -8.02668810e-01 -5.38737416e-01 3.02777201e-01 1.69226646e-01 -5.00090532e-02 -7.89666593e-01 5.43371677e-01 2.53329992e-01 2.99460310e-02 2.20837463e-02 -7.05866992e-01 -6.77308619e-01 -9.37214077e-01 1.29070699e-01 5.74190635e-03 2.39211664e-01 -1.74944371e-01 9.71143395e-02 6.85142040e-01 -1.86958337e+00 -1.58719599e-01 8.18918467e-01 1.34184170e+00 2.73318738e-01 1.88962609e-01 -3.53625603e-02 5.11996567e-01 -6.60951495e-01 -1.59099653e-01 -5.85150346e-02 5.52793592e-02 -2.28376538e-01 -1.31816387e-01 4.48437423e-01 -2.44743526e-01 -4.28122461e-01 9.71870780e-01 8.65817904e-01 2.78726548e-01 4.26323861e-01 -1.33083332e+00 -2.40863502e-01 1.38788611e-01 -7.15714216e-01 5.69237545e-02 1.76515773e-01 2.99373269e-01 5.06990373e-01 -6.54274583e-01 -4.52668220e-02 1.34939063e+00 6.92129731e-01 2.86225408e-01 -4.84216481e-01 -1.06748772e+00 2.50731826e-01 7.79357553e-01 -1.67047727e+00 -4.58239377e-01 8.83807838e-01 -1.31286114e-01 5.17313361e-01 3.04072320e-01 6.80791497e-01 6.81148767e-01 3.72428656e-01 8.85351121e-01 9.51104581e-01 -1.01304851e-01 -4.15666625e-02 -1.68151816e-03 6.62151426e-02 7.32463062e-01 2.98272610e-01 -8.01363438e-02 -5.03431380e-01 2.90374815e-01 5.43465316e-01 3.80780756e-01 -5.85159004e-01 -1.55121639e-01 -9.71862316e-01 7.52844870e-01 1.22120190e+00 1.21916614e-01 -4.07475412e-01 7.63901845e-02 1.26592845e-01 -3.39381367e-01 2.79335350e-01 1.74085975e-01 -2.07927495e-01 -1.80938952e-02 -8.55545819e-01 1.18626438e-01 4.10379589e-01 8.02036941e-01 7.61717618e-01 4.01666239e-02 -2.76511192e-01 8.59958529e-01 5.08387089e-01 5.41937172e-01 6.60675287e-01 -5.55951834e-01 3.88748109e-01 6.07975245e-01 -7.00184554e-02 -1.14591050e+00 -1.07230163e+00 -6.11425400e-01 -1.02042568e+00 3.56998175e-01 3.56285900e-01 -2.97439098e-01 -9.03273761e-01 1.62202072e+00 4.61056232e-01 2.83431530e-01 -3.29084635e-01 1.27370238e+00 1.00770473e+00 6.46603286e-01 6.84254628e-04 -1.83322072e-01 1.53819501e+00 -9.44716632e-01 -1.02492189e+00 -5.03060937e-01 2.06392959e-01 -4.77844268e-01 7.64040172e-01 2.79373467e-01 -8.46383810e-01 -7.96641290e-01 -1.25717032e+00 -2.37947762e-01 -3.71135592e-01 2.82440364e-01 5.30648291e-01 3.84621918e-01 -8.26462567e-01 1.43159494e-01 -7.21182942e-01 -2.07243003e-02 4.63355094e-01 5.56798577e-01 -4.76213753e-01 -1.01727098e-01 -1.20478177e+00 1.18413174e+00 3.74383718e-01 7.72360861e-01 -1.93461686e-01 -3.17849517e-01 -1.16278887e+00 -1.72877014e-02 1.96691468e-01 -4.61473525e-01 1.07216704e+00 -4.84433353e-01 -1.56477010e+00 4.13379341e-01 -3.11870664e-01 -2.63255835e-01 4.60136324e-01 -3.86964411e-01 -4.20939416e-01 -1.90970048e-01 -8.96370336e-02 8.58752012e-01 1.01627171e+00 -7.95286119e-01 -7.48554945e-01 -7.06567943e-01 -2.89660484e-01 3.14821213e-01 -4.18722361e-01 -1.07159346e-01 -6.89153850e-01 -3.76626223e-01 3.93100441e-01 -7.77468324e-01 -1.57115340e-01 1.30946025e-01 -2.77757764e-01 -4.84400958e-01 7.55391717e-01 -6.99460924e-01 1.31942415e+00 -2.15731144e+00 5.55788130e-02 5.78328520e-02 4.04093266e-01 2.07361147e-01 2.21304938e-01 -3.57134223e-01 -3.36765572e-02 -4.73734349e-01 -1.29544392e-01 -4.30385649e-01 -1.25401348e-01 -7.07481289e-03 2.00062022e-01 8.02175641e-01 2.01717675e-01 9.13259983e-01 -5.56920528e-01 -5.68934858e-01 4.67924803e-01 8.21955681e-01 -3.50638390e-01 3.17725360e-01 2.77476370e-01 3.56457442e-01 -3.58517498e-01 5.78593433e-01 9.43295300e-01 1.46673009e-01 -2.60808200e-01 -5.74845552e-01 -2.39808321e-01 3.41614112e-02 -1.22859418e+00 1.49724925e+00 -6.20775104e-01 7.79004216e-01 7.33685121e-02 -7.62789071e-01 1.20005167e+00 8.26204047e-02 1.72810346e-01 -8.70738268e-01 5.47360480e-01 -1.71744600e-01 -1.26336515e-02 -6.48068488e-01 4.45779413e-01 5.41025512e-02 -1.86047256e-01 4.38035578e-02 -2.78139025e-01 -4.01253104e-02 -3.90623271e-01 -5.42629719e-01 5.50071478e-01 -2.52696685e-03 2.35125065e-01 3.63043286e-02 6.81263566e-01 -5.20918667e-01 7.70689428e-01 1.89092666e-01 -4.06258285e-01 7.14293420e-01 1.58208489e-01 -5.33340156e-01 -5.89111328e-01 -6.52987778e-01 -3.75416517e-01 1.02218914e+00 3.21020603e-01 -1.01661861e-01 -9.88785088e-01 -2.86428899e-01 1.47897467e-01 4.90357459e-01 -7.33230054e-01 -4.95742798e-01 -7.96622276e-01 -6.84977472e-01 3.05732340e-01 9.00646925e-01 9.39934313e-01 -9.72647429e-01 -6.88218713e-01 7.07100471e-03 -2.59802312e-01 -1.06271017e+00 -3.85501653e-01 2.89750136e-02 -6.67459726e-01 -8.53160441e-01 -7.61174679e-01 -8.46015990e-01 6.57364249e-01 2.98152626e-01 6.46622419e-01 1.14498995e-01 -2.50551462e-01 -3.14251870e-01 -4.92473766e-02 -8.12476575e-01 4.31220293e-01 1.72589526e-01 1.16239600e-01 1.92927375e-01 6.74204111e-01 -2.35569075e-01 -9.92112517e-01 3.15918624e-01 -5.35386980e-01 8.19740910e-03 8.51507723e-01 7.17902422e-01 3.92292857e-01 4.89215851e-02 3.99503469e-01 -2.17851505e-01 6.69702470e-01 -3.14407468e-01 -5.99467218e-01 1.27204945e-02 -4.42835450e-01 1.12841755e-01 4.58453506e-01 -3.18307281e-01 -1.00759804e+00 3.12131226e-01 -3.38960022e-01 -4.16453779e-01 -3.11643511e-01 1.92178845e-01 -4.22977060e-01 -1.06367730e-01 4.52572227e-01 6.25404567e-02 4.64207500e-01 -4.98254687e-01 3.25595289e-01 9.39688563e-01 7.04694629e-01 -2.51348820e-02 4.34124887e-01 2.98321664e-01 1.40404299e-01 -8.15176606e-01 -6.40730083e-01 -5.15794456e-01 -7.48519838e-01 -4.68401611e-01 1.02643239e+00 -9.34787452e-01 -1.06803823e+00 7.32091308e-01 -1.07540739e+00 1.52078912e-01 3.74527097e-01 6.50310576e-01 -1.35336071e-01 6.61811307e-02 -2.49363378e-01 -8.32987964e-01 -3.88283849e-01 -1.32830858e+00 1.17093921e+00 7.08773792e-01 8.19714963e-02 -5.91499925e-01 -4.26851362e-01 4.26004469e-01 5.07178664e-01 2.91619241e-01 3.08860004e-01 -7.38024265e-02 -4.32134539e-01 -6.48988008e-01 -2.88519949e-01 1.70666084e-01 1.26554782e-03 -1.30607367e-01 -1.28800762e+00 -3.31853598e-01 9.49536338e-02 -5.22108562e-02 9.47514892e-01 6.58281803e-01 1.50119305e+00 -2.75904775e-01 -3.84527177e-01 8.22820425e-01 1.02793479e+00 8.53929967e-02 7.58222938e-01 5.13932168e-01 8.85533392e-01 7.11851299e-01 6.91729903e-01 3.45228940e-01 7.84909070e-01 8.84260297e-01 5.83823800e-01 -3.85847241e-01 -6.69585168e-02 5.51136471e-02 1.13366850e-01 6.79283917e-01 -1.87111255e-02 8.00198242e-02 -6.73105240e-01 2.60316461e-01 -1.80016220e+00 -7.27177262e-01 3.53623554e-02 2.18044305e+00 7.04914212e-01 2.06579119e-01 1.82647675e-01 3.85890603e-01 8.43202651e-01 1.80603251e-01 -5.44955850e-01 -2.42865533e-01 3.13141108e-01 1.60837010e-01 6.21517420e-01 4.31381047e-01 -1.24008346e+00 5.03373027e-01 5.55929565e+00 6.38687551e-01 -1.50679815e+00 7.20370710e-02 5.72102368e-01 -2.93750942e-01 2.88680732e-01 -6.07812166e-01 -1.09202647e+00 6.86868608e-01 7.16450095e-01 2.58143812e-01 -2.45860573e-02 9.08093452e-01 2.47498691e-01 -2.62945652e-01 -6.55285180e-01 1.32863533e+00 2.31998965e-01 -9.02960360e-01 -4.46695805e-01 -1.03232086e-01 2.24692822e-01 -1.64685715e-02 1.71718150e-01 2.55065203e-01 -1.31616503e-01 -1.29065990e+00 7.23886609e-01 8.95447373e-01 7.36820757e-01 -1.09946644e+00 1.14985073e+00 4.75683659e-01 -1.44977510e+00 -4.06477422e-01 -5.49423158e-01 -1.77820578e-01 1.71179846e-01 5.17870903e-01 -6.99227929e-01 3.25109512e-01 8.55446517e-01 5.42930126e-01 -7.70904541e-01 1.41788256e+00 -3.95663470e-01 1.83488458e-01 -4.58460420e-01 -4.07075137e-01 1.12486027e-01 -3.25204292e-03 2.04641268e-01 1.16972303e+00 1.78433448e-01 7.96093941e-02 6.56625032e-02 5.06580532e-01 4.03346010e-02 1.55225277e-01 -4.80872422e-01 7.54482090e-01 4.59195644e-01 1.42601478e+00 -3.89795393e-01 -6.84558647e-03 -3.61605108e-01 5.51894724e-01 4.25004214e-01 1.52039915e-01 -9.58881259e-01 -7.71081269e-01 7.65119493e-01 4.94437478e-02 2.18771443e-01 -2.23071828e-01 -4.34894949e-01 -7.32904434e-01 9.79772061e-02 -4.18573946e-01 2.26540357e-01 -8.12735379e-01 -7.70779610e-01 7.91883111e-01 -4.87455577e-02 -1.29433298e+00 -1.53446555e-01 -6.55171812e-01 -8.02348077e-01 1.17157161e+00 -1.67617106e+00 -9.59132969e-01 -1.11218131e+00 5.70445061e-01 3.70151579e-01 3.28727849e-02 5.12843966e-01 3.46018106e-01 -1.05737877e+00 9.97547448e-01 -2.62531996e-01 2.37683550e-01 7.30613470e-01 -1.03512931e+00 2.07349136e-01 7.20255911e-01 -4.38418865e-01 5.50834060e-01 5.96193016e-01 -2.35119760e-01 -1.27548492e+00 -1.14192855e+00 5.55993855e-01 -2.57090509e-01 -3.80264217e-04 -2.45952249e-01 -9.83368874e-01 2.04712346e-01 3.59215215e-02 2.25590125e-01 4.10470724e-01 -1.92243103e-02 2.68228017e-02 -6.71996951e-01 -1.12166679e+00 3.28336149e-01 6.25231326e-01 -2.69998044e-01 -6.01207674e-01 1.75050661e-01 5.76271594e-01 -7.43643403e-01 -6.82013214e-01 5.16722739e-01 7.06356525e-01 -8.94671977e-01 9.66052711e-01 -1.81153417e-01 1.18849762e-01 -5.20363450e-01 2.12124825e-01 -1.18391848e+00 -4.49627936e-01 -3.37980151e-01 -1.14953004e-01 8.20661485e-01 2.27792606e-01 -7.47809947e-01 7.37831056e-01 7.55442381e-01 -3.14457744e-01 -1.10596204e+00 -1.08570337e+00 -3.21598411e-01 -2.08162725e-01 -3.74371320e-01 9.61376011e-01 3.82653534e-01 -2.05341965e-01 3.23139876e-01 -2.51603752e-01 2.42094114e-01 5.98803401e-01 -1.57478184e-01 8.69357288e-01 -1.29511142e+00 2.85175562e-01 -6.00522220e-01 -8.71784151e-01 -1.13167298e+00 1.83375418e-01 -4.61738974e-01 3.94359022e-01 -1.52953267e+00 -1.86088875e-01 -4.27543551e-01 -4.70636278e-01 5.73698044e-01 -4.15233880e-01 3.29535604e-01 -3.43251997e-03 5.74807301e-02 -2.63025165e-01 6.65750921e-01 1.44464624e+00 -9.64034274e-02 -2.22755954e-01 3.43333423e-01 -6.75610423e-01 7.86307871e-01 8.18651438e-01 -1.44423485e-01 -2.23035827e-01 -4.21902239e-01 -2.28553191e-01 -1.11921057e-01 3.38352740e-01 -1.24514031e+00 7.23769963e-01 5.87008148e-02 8.74165237e-01 -7.44995058e-01 5.80501854e-01 -7.71111131e-01 -1.64751559e-01 4.02073413e-01 3.72389108e-02 1.56668290e-01 3.63978297e-01 3.24227154e-01 -3.42605829e-01 9.52074677e-03 9.13063645e-01 1.38838917e-01 -9.01245892e-01 4.62362826e-01 1.18191458e-01 -3.82864684e-01 1.07010615e+00 -3.14021975e-01 -1.60743520e-01 -4.92857873e-01 -3.37922126e-01 3.75486821e-01 1.59989163e-01 6.14081860e-01 8.51222157e-01 -1.47365570e+00 -5.64212680e-01 5.67531168e-01 6.28415272e-02 3.57548118e-01 2.02778488e-01 1.02774990e+00 -5.01331627e-01 4.78993684e-01 -4.26719844e-01 -8.57886553e-01 -1.21870244e+00 3.29206854e-01 4.66514796e-01 3.17783207e-01 -6.66944861e-01 9.83212590e-01 1.46476567e-01 -1.91777214e-01 5.46812296e-01 -5.02547741e-01 -7.87122548e-01 2.18917266e-01 8.74952197e-01 2.42173895e-01 2.53476679e-01 -9.84335303e-01 -4.98251498e-01 7.74865329e-01 3.18396240e-02 2.33396634e-01 1.29454148e+00 -1.17355503e-01 -1.32035777e-01 1.50320828e-01 1.36700487e+00 -2.91273952e-01 -1.38780844e+00 -2.34014317e-01 -4.10126418e-01 -4.28225398e-01 5.35559237e-01 -3.91852587e-01 -1.38957536e+00 1.23069084e+00 9.33505416e-01 -1.43782601e-01 1.30776465e+00 -2.36937270e-01 8.93824220e-01 2.68766254e-01 1.43332422e-01 -1.09977984e+00 -1.66177869e-01 2.46254623e-01 9.83263612e-01 -1.44387138e+00 8.08603987e-02 -1.66109383e-01 -6.34124041e-01 1.18494236e+00 9.37740326e-01 1.52825862e-01 6.94211602e-01 2.66455770e-01 1.72093511e-01 -3.63763086e-02 -2.77303427e-01 -4.07816678e-01 6.47427619e-01 5.54024220e-01 3.64473343e-01 2.17329979e-01 -8.35509077e-02 6.73494875e-01 -5.15415311e-01 -1.69825330e-01 -4.28481912e-03 8.25564861e-01 -7.59768426e-01 -5.11583984e-01 -6.50018692e-01 4.94939685e-01 -2.24726424e-01 1.88474238e-01 8.40083212e-02 6.06615067e-01 3.59732985e-01 9.44908321e-01 3.25497895e-01 -6.77157342e-01 5.60971856e-01 -3.80300954e-02 2.26451144e-01 -2.53919721e-01 -1.80800900e-01 -1.25521943e-01 -3.18595737e-01 -6.88268781e-01 -1.98665813e-01 -4.09434795e-01 -1.23986256e+00 -2.92394817e-01 -5.76298475e-01 5.81316873e-02 9.21594083e-01 9.59795713e-01 2.32080027e-01 8.24768305e-01 8.11045706e-01 -1.20213223e+00 -3.52168798e-01 -1.21127999e+00 -2.91313052e-01 1.80297822e-01 5.08960068e-01 -1.16741824e+00 -1.64254680e-01 -2.89272696e-01]
[13.750085830688477, 0.2735692858695984]
9ece6ff1-1d96-4e3f-b473-6b835c4618a5
systematicity-emerges-in-transformers-when
null
null
https://aclanthology.org/2022.naacl-srw.1
https://aclanthology.org/2022.naacl-srw.1.pdf
Systematicity Emerges in Transformers when Abstract Grammatical Roles Guide Attention
Systematicity is thought to be a key inductive bias possessed by humans that is lacking in standard natural language processing systems such as those utilizing transformers. In this work, we investigate the extent to which the failure of transformers on systematic generalization tests can be attributed to a lack of linguistic abstraction in its attention mechanism. We develop a novel modification to the transformer by implementing two separate input streams: a role stream controls the attention distributions (i.e., queries and keys) at each layer, and a filler stream determines the values. Our results show that when abstract role labels are assigned to input sequences and provided to the role stream, systematic generalization is improved.
['Randall O’Reilly', 'Jacob Labe Russin', 'Ayush K Chakravarthy']
null
null
null
null
naacl-acl-2022-7
['systematic-generalization']
['reasoning']
[ 4.33443218e-01 9.82527342e-03 -2.12954402e-01 -5.47035038e-01 4.68685851e-03 -7.71351099e-01 6.07618272e-01 5.26394188e-01 -7.05706000e-01 5.08137465e-01 4.16526496e-01 -6.76811755e-01 -1.44884679e-02 -1.05067205e+00 -6.46012604e-01 -2.55805969e-01 1.85062476e-02 4.47714210e-01 7.44115591e-01 -4.75896984e-01 4.52925593e-01 5.11111617e-01 -1.78805530e+00 4.90896106e-01 8.17990839e-01 7.66312540e-01 1.88200310e-01 6.15344465e-01 -3.70166421e-01 9.19338048e-01 -7.79729307e-01 -1.95329681e-01 -2.07523834e-02 -3.47770929e-01 -1.00690818e+00 -3.56065840e-01 2.36087173e-01 -4.48518574e-01 9.04311147e-03 1.07965648e+00 3.81770283e-01 2.45344028e-01 6.61147118e-01 -1.17979801e+00 -1.02753413e+00 1.01050687e+00 -5.40779978e-02 7.20921874e-01 6.85839593e-01 4.35584337e-01 1.33748019e+00 -7.54496694e-01 3.05465311e-01 1.45550394e+00 5.60450554e-01 5.84883034e-01 -1.48915744e+00 -5.68392873e-01 4.29204255e-01 5.80860265e-02 -1.20570993e+00 -5.94679594e-01 4.46677417e-01 -3.22790176e-01 1.16380918e+00 1.29838303e-01 4.95299131e-01 6.61820292e-01 2.49041870e-01 7.48255014e-01 1.05449808e+00 -5.97664356e-01 3.37980658e-01 3.80491942e-01 4.39784706e-01 5.21103144e-01 3.57596248e-01 1.76414147e-01 -8.41387272e-01 -2.10056096e-01 8.82574439e-01 -2.17759430e-01 -2.44886845e-01 -2.21171290e-01 -9.03908432e-01 7.83815563e-01 5.20843744e-01 3.62658590e-01 -4.64099109e-01 -1.72010902e-02 4.48946834e-01 6.20973408e-01 -2.24757940e-02 1.03701925e+00 -5.30804455e-01 -1.79400310e-01 -2.11726323e-01 1.98997691e-01 6.82475805e-01 9.95289326e-01 6.67742908e-01 1.01986386e-01 -4.66142625e-01 5.32923877e-01 2.45544091e-01 4.29921091e-01 7.09616959e-01 -7.26733983e-01 2.16470778e-01 8.65492821e-01 9.05124843e-02 -5.96984148e-01 -2.46578515e-01 -3.07803482e-01 -8.12180042e-02 -1.90918267e-01 4.17825788e-01 8.27372000e-02 -6.60087943e-01 2.32692671e+00 1.52843475e-01 -5.38368165e-01 -3.54799032e-02 6.69655383e-01 4.40590531e-01 3.79201025e-01 6.90840185e-01 -6.47490844e-02 1.52828443e+00 -3.08395535e-01 -6.18202031e-01 -3.77380311e-01 8.60595942e-01 -3.97967845e-01 1.87190843e+00 1.76644281e-01 -1.25662446e+00 -6.36465073e-01 -1.14187133e+00 -2.47979462e-01 -5.97425640e-01 -6.04197919e-01 8.74658108e-01 6.42562330e-01 -1.10321879e+00 3.66366386e-01 -4.44941461e-01 -4.29688305e-01 2.54424185e-01 3.79643112e-01 -1.48941070e-01 7.05071986e-02 -1.67213750e+00 9.76225972e-01 3.46347839e-01 -1.28772259e-01 -8.32566559e-01 -5.89887917e-01 -9.88249481e-01 4.03443068e-01 2.42488801e-01 -6.48532927e-01 1.63545990e+00 -9.53654170e-01 -1.19848680e+00 7.42987335e-01 -3.97299886e-01 -4.73216623e-01 -1.66411161e-01 -2.43185595e-01 -1.77047208e-01 2.10407779e-01 2.22587422e-01 6.29784822e-01 8.39866102e-01 -1.18100905e+00 -6.46600366e-01 -4.67565417e-01 4.86120045e-01 1.66186064e-01 -5.16089261e-01 1.72478423e-01 1.37157589e-01 -5.72720230e-01 -8.07416886e-02 -5.60015321e-01 1.51650354e-01 -7.04964027e-02 -1.06567286e-01 -6.60172760e-01 2.96234012e-01 -2.29003713e-01 1.62949884e+00 -2.36674261e+00 3.44333053e-02 2.41337761e-01 2.37231836e-01 2.04031736e-01 -4.06041667e-02 4.61711317e-01 -1.49363250e-01 5.51861107e-01 -2.01430723e-01 6.34714961e-02 2.44027346e-01 3.08237314e-01 -5.50173938e-01 2.55289644e-01 2.60679245e-01 9.01705682e-01 -9.53394651e-01 -2.43172824e-01 -1.98572993e-01 3.91282365e-02 -8.74081910e-01 4.17716056e-01 -4.30659562e-01 -2.32331887e-01 -3.35151523e-01 5.62060595e-01 2.87530750e-01 -2.43241072e-01 6.79461509e-02 1.97960660e-01 -1.25978976e-01 1.09226358e+00 -9.02910709e-01 1.18068182e+00 -3.31653595e-01 4.36588377e-01 2.23353822e-02 -6.73267365e-01 6.36042833e-01 3.88516009e-01 -3.75096053e-01 -8.91589582e-01 2.26838868e-02 4.65466715e-02 5.88891089e-01 -3.24705273e-01 5.04956841e-01 -5.23655713e-01 -2.78217524e-01 8.84709418e-01 -1.22364154e-02 -1.41259164e-01 3.36800665e-01 5.56547225e-01 8.51054668e-01 -1.43891707e-01 4.11443919e-01 -5.86139619e-01 7.22883523e-01 -5.30279316e-02 4.61054087e-01 1.16296709e+00 -1.44291013e-01 -1.17219891e-02 8.17058444e-01 -3.33236754e-01 -9.85218167e-01 -1.05757332e+00 -8.75943154e-02 1.91684008e+00 -7.93653056e-02 -4.43673313e-01 -6.43798470e-01 -5.58972955e-01 2.18677580e-01 1.10472262e+00 -6.19083643e-01 -7.70351946e-01 -4.59520310e-01 -2.09905326e-01 6.53967500e-01 1.04147124e+00 2.88639843e-01 -1.42811453e+00 -9.44538236e-01 1.85584947e-01 3.42589676e-01 -7.94110537e-01 -6.75824344e-01 5.71376681e-01 -9.04878378e-01 -8.12503159e-01 1.11482561e-01 -6.09135151e-01 6.32821679e-01 -4.15015556e-02 1.28415334e+00 3.46255124e-01 2.11513281e-01 5.15169740e-01 -1.65668756e-01 -6.48040354e-01 -2.56215930e-01 2.64749914e-01 8.07690397e-02 -1.93775490e-01 6.90350354e-01 -5.30366600e-01 -3.48860979e-01 3.16423267e-01 -1.15088034e+00 -3.30411822e-01 5.34635723e-01 7.63989866e-01 2.68193521e-02 -5.60172424e-02 6.18221045e-01 -1.20184028e+00 1.21420014e+00 -5.22441506e-01 -4.24804389e-01 1.79455709e-02 -7.55681634e-01 4.25344795e-01 7.20419109e-01 -5.41065574e-01 -1.00802708e+00 -2.83899665e-01 9.65395421e-02 4.99226749e-02 -3.08865994e-01 6.15851521e-01 -3.96825731e-01 1.33961275e-01 7.44569242e-01 4.50366914e-01 1.40338242e-01 -3.41638952e-01 2.05389649e-01 5.24438739e-01 1.99405789e-01 -1.06308126e+00 6.42620742e-01 1.89520437e-02 -4.53938574e-01 -7.75416493e-01 -7.79733002e-01 -3.65140200e-01 -2.97615409e-01 1.03644013e-01 6.45593643e-01 -6.13802493e-01 -7.85720587e-01 2.62569904e-01 -1.17826843e+00 -5.78544438e-01 -4.80934888e-01 1.89687654e-01 -4.32697445e-01 -1.71342432e-01 -7.12511599e-01 -7.07689464e-01 -1.53520063e-01 -9.56825256e-01 7.47161746e-01 8.31652805e-02 -6.27705276e-01 -8.83422196e-01 -3.32997113e-01 -3.05000931e-01 8.08948517e-01 -6.29330516e-01 1.60700762e+00 -1.12359202e+00 -4.73107904e-01 -1.21568982e-02 -8.44745990e-03 3.05115521e-01 1.03217088e-01 -2.32242346e-01 -9.79335546e-01 -5.41921407e-02 1.48812041e-01 -4.08326924e-01 6.48661375e-01 -1.62526723e-02 1.09148288e+00 -4.06540543e-01 -2.39588723e-01 4.10089940e-01 1.15088487e+00 5.40923476e-01 5.64953327e-01 9.01516452e-02 3.52742881e-01 5.56203246e-01 5.34573048e-02 1.31499931e-01 3.88118982e-01 3.38389784e-01 -2.58321837e-02 1.50014192e-01 4.76682670e-02 -5.64449847e-01 2.61566967e-01 4.24659967e-01 2.31306270e-01 -2.23140717e-01 -1.00766015e+00 6.30792439e-01 -1.36216223e+00 -8.76007676e-01 4.04866606e-01 2.30084419e+00 1.30908275e+00 5.69711983e-01 -1.35776728e-01 3.50080997e-01 6.11690521e-01 -4.84738909e-02 -5.43712735e-01 -7.14491248e-01 -4.70751111e-04 3.17779303e-01 1.01088464e-01 8.16940904e-01 -4.96644229e-01 9.90692258e-01 7.57592773e+00 1.77971423e-01 -1.00036705e+00 -1.30588710e-01 1.95276886e-01 1.76285937e-01 -7.80902684e-01 -1.30844668e-01 -7.88644254e-01 3.46582234e-01 9.55775380e-01 -2.73068428e-01 3.23254764e-01 6.19494379e-01 -1.81341618e-01 -1.25474215e-01 -1.69980812e+00 2.88171321e-01 -1.08209290e-01 -8.08798313e-01 6.09636962e-01 -8.77922177e-02 7.82413129e-03 -5.39042592e-01 1.44177139e-01 4.04921144e-01 3.73413950e-01 -9.25554037e-01 9.43872809e-01 2.77977735e-01 5.52448213e-01 -6.08708143e-01 4.60779905e-01 2.92952806e-01 -9.91163433e-01 -2.99747616e-01 -1.87919155e-01 -5.13527095e-01 -1.71205461e-01 8.51400718e-02 -9.21317160e-01 -2.88033843e-01 4.45493251e-01 2.39423111e-01 -5.94698489e-01 7.32189536e-01 -7.68837690e-01 8.91313970e-01 -2.13195503e-01 -3.38062555e-01 4.19461019e-02 4.25253600e-01 5.33703208e-01 1.15743458e+00 -2.43227243e-01 1.61979780e-01 4.27801199e-02 1.02554524e+00 -7.79820085e-02 -1.17311209e-01 -6.56417072e-01 -1.76769406e-01 9.09801781e-01 6.50381565e-01 -5.78662336e-01 -4.80229855e-01 -3.94388795e-01 5.77033281e-01 4.23722297e-01 6.02550507e-01 -3.86454403e-01 -6.27433896e-01 6.44671261e-01 4.80110854e-01 2.89993495e-01 -8.92299861e-02 -5.38650393e-01 -8.72572362e-01 -1.46597371e-01 -8.94944668e-01 5.43408573e-01 -8.74140263e-01 -1.29019976e+00 3.50971311e-01 2.67028004e-01 -3.79385978e-01 -3.35401446e-01 -7.11213887e-01 -4.19541270e-01 1.26597619e+00 -1.31487978e+00 -6.06957495e-01 1.33054197e-01 7.79637456e-01 3.50953549e-01 2.49273982e-02 8.93621743e-01 -4.48401459e-02 -3.43680173e-01 6.69755280e-01 -9.21155274e-01 2.59052664e-01 5.13186038e-01 -1.46924448e+00 4.56307888e-01 9.09629226e-01 -1.49245262e-01 1.59622586e+00 7.65475690e-01 -3.17454666e-01 -1.42303741e+00 -4.71539974e-01 1.20436370e+00 -5.94721973e-01 6.33746684e-01 -6.79980099e-01 -1.40512848e+00 9.40438569e-01 3.18649620e-01 -7.64877871e-02 8.02792370e-01 2.71447331e-01 -7.09089875e-01 -2.29487345e-01 -1.12522531e+00 5.32141864e-01 1.04524159e+00 -9.55411375e-01 -1.29405522e+00 -2.66792774e-01 1.22235417e+00 7.50619769e-02 -5.07799387e-01 6.97213784e-02 4.58216906e-01 -7.60931790e-01 6.79505348e-01 -1.01078022e+00 2.45519653e-01 -3.79263222e-01 -1.77123681e-01 -1.37825024e+00 -6.55386031e-01 -3.94742697e-01 3.03153154e-02 1.14227855e+00 5.72928429e-01 -1.12324297e+00 2.99605876e-01 9.02370989e-01 -1.73983604e-01 -4.27535027e-01 -6.67927980e-01 -4.89068002e-01 1.31575927e-01 -4.37472045e-01 9.22292054e-01 6.71425223e-01 2.57916838e-01 7.75580466e-01 4.90234375e-01 9.80013162e-02 2.05428064e-01 -1.33441150e-01 2.03377455e-01 -1.40056336e+00 -2.04812810e-01 -5.18157542e-01 -1.15922883e-01 -1.10699022e+00 1.77273616e-01 -7.58594990e-01 2.29132995e-01 -1.17784858e+00 -1.25517204e-01 -5.77321053e-01 -5.17471194e-01 6.69004023e-01 -3.16062391e-01 -3.78103882e-01 1.23230577e-01 -2.16545444e-03 -3.76452595e-01 3.86674672e-01 9.76902485e-01 9.29284096e-02 -1.89868584e-01 -7.85787478e-02 -1.32744896e+00 7.73425996e-01 8.89923334e-01 -4.27519977e-01 -7.59606540e-01 -7.22049117e-01 5.16019046e-01 -1.84631184e-01 2.24881649e-01 -8.27369452e-01 4.89233464e-01 -9.60066393e-02 3.78066480e-01 -3.64466291e-03 -1.14196874e-01 -8.28109205e-01 -5.20913899e-01 4.69912589e-01 -9.04118180e-01 7.64981627e-01 6.16015851e-01 2.99001694e-01 -1.01419233e-01 -2.33599827e-01 5.10611236e-01 -4.39108968e-01 -6.81227922e-01 -6.74948692e-02 -7.35390484e-01 3.27305287e-01 5.13221562e-01 -1.67693153e-01 -3.70613158e-01 -8.53417516e-02 -3.79972845e-01 2.29915559e-01 3.78404558e-01 3.52625400e-01 6.49696827e-01 -1.18111670e+00 -3.75097543e-01 5.91742754e-01 2.55283564e-01 7.09454902e-03 -2.81220555e-01 4.48548645e-01 -7.53408223e-02 6.23020291e-01 -1.97085097e-01 -3.07716012e-01 -9.16415751e-01 7.75925934e-01 4.07158136e-01 1.79385990e-01 -1.41384169e-01 1.08065021e+00 3.94159436e-01 -2.23576546e-01 3.50614965e-01 -6.86275244e-01 -1.79536343e-01 1.74690604e-01 8.26878905e-01 -1.07977919e-01 -8.16073418e-02 -3.31212789e-01 -5.34762144e-01 8.66099522e-02 -2.58715034e-01 -3.82420659e-01 9.35622215e-01 1.33768413e-02 -3.59225661e-01 9.02694583e-01 5.73758900e-01 1.73292994e-01 -7.17385352e-01 -3.30315322e-01 1.26101583e-01 -3.76924068e-01 -2.03740418e-01 -9.31815684e-01 -4.78506029e-01 8.73631597e-01 9.83657688e-02 5.39588988e-01 1.02728581e+00 1.82284266e-01 2.94983566e-01 4.90956068e-01 4.00832534e-01 -8.30241799e-01 -6.38638884e-02 6.87497556e-01 8.91588867e-01 -6.78079069e-01 -2.90187746e-01 -3.10741782e-01 -4.54466432e-01 7.36753166e-01 9.88968253e-01 1.33321285e-01 4.02922601e-01 4.65070367e-01 -2.06846908e-01 -1.69520661e-01 -1.28472960e+00 -2.48886138e-01 -4.47770171e-02 6.55441403e-01 7.45904148e-01 -1.46970794e-01 -3.05227309e-01 6.88948035e-01 -3.56199443e-01 -5.19948900e-02 5.08987188e-01 1.27291107e+00 -7.12949336e-01 -7.86591828e-01 -3.09351265e-01 3.61883253e-01 -3.83241087e-01 -6.32258296e-01 -4.17648584e-01 7.14592993e-01 1.48764014e-01 7.11266816e-01 4.22979236e-01 -2.28880316e-01 7.03099191e-01 5.68837643e-01 3.74885619e-01 -1.15859580e+00 -9.26252663e-01 -5.29530287e-01 -1.22519195e-01 -5.29620886e-01 -4.64471616e-02 -6.06962919e-01 -1.42535460e+00 6.15743047e-04 -1.41927510e-01 4.72647756e-01 1.20411158e-01 7.92419970e-01 2.19142035e-01 5.17101169e-01 2.05981731e-01 -1.56462297e-01 -8.98058355e-01 -1.03296292e+00 -5.07844508e-01 5.19578993e-01 6.84510827e-01 -6.98245704e-01 -5.90730608e-01 -8.06118548e-02]
[10.456609725952148, 8.422676086425781]
33dc3a66-4d55-42f2-af07-c88778a5e489
bottlesum-unsupervised-and-self-supervised
1909.07405
null
https://arxiv.org/abs/1909.07405v2
https://arxiv.org/pdf/1909.07405v2.pdf
BottleSum: Unsupervised and Self-supervised Sentence Summarization using the Information Bottleneck Principle
The principle of the Information Bottleneck (Tishby et al. 1999) is to produce a summary of information X optimized to predict some other relevant information Y. In this paper, we propose a novel approach to unsupervised sentence summarization by mapping the Information Bottleneck principle to a conditional language modelling objective: given a sentence, our approach seeks a compressed sentence that can best predict the next sentence. Our iterative algorithm under the Information Bottleneck objective searches gradually shorter subsequences of the given sentence while maximizing the probability of the next sentence conditioned on the summary. Using only pretrained language models with no direct supervision, our approach can efficiently perform extractive sentence summarization over a large corpus. Building on our unsupervised extractive summarization (BottleSumEx), we then present a new approach to self-supervised abstractive summarization (BottleSumSelf), where a transformer-based language model is trained on the output summaries of our unsupervised method. Empirical results demonstrate that our extractive method outperforms other unsupervised models on multiple automatic metrics. In addition, we find that our self-supervised abstractive model outperforms unsupervised baselines (including our own) by human evaluation along multiple attributes.
['Yejin Choi', 'Jan Buys', 'Peter West', 'Ari Holtzman']
2019-09-16
bottlesum-unsupervised-and-self-supervised-1
https://aclanthology.org/D19-1389
https://aclanthology.org/D19-1389.pdf
ijcnlp-2019-11
['unsupervised-extractive-summarization', 'abstractive-sentence-summarization', 'unsupervised-sentence-summarization']
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
[ 8.12328756e-01 8.26894462e-01 -5.12607038e-01 -4.26625401e-01 -1.39189065e+00 -4.82817084e-01 6.15593195e-01 8.25569928e-01 -4.14130688e-01 9.33017850e-01 1.18556547e+00 -4.38607670e-02 8.64770543e-03 -4.62515563e-01 -7.42819190e-01 -2.57126957e-01 1.54940709e-01 6.28336668e-01 -2.96373814e-02 -9.71239284e-02 9.70759392e-01 -3.10028959e-02 -1.21971095e+00 5.61242223e-01 1.34231675e+00 3.38813215e-01 5.99195123e-01 1.26555431e+00 -3.23496878e-01 1.08457696e+00 -8.14241290e-01 -3.50316226e-01 -2.53903329e-01 -9.89181519e-01 -1.39521062e+00 2.97424585e-01 5.19955039e-01 -2.15550572e-01 -3.37258540e-02 9.09543276e-01 3.82252097e-01 3.37663770e-01 9.96986747e-01 -6.09405577e-01 -2.60392487e-01 1.28082907e+00 -4.54799831e-01 2.14014664e-01 6.77669525e-01 -1.62207827e-01 1.47589493e+00 -7.22244382e-01 7.85430133e-01 1.19789422e+00 3.57808769e-01 5.36574006e-01 -1.39302707e+00 2.68286355e-02 2.34636500e-01 -9.61818472e-02 -8.53676200e-01 -8.48595202e-01 6.71216309e-01 -3.35664824e-02 1.51230764e+00 5.29656053e-01 4.50212002e-01 6.66875482e-01 6.10765398e-01 1.36571550e+00 4.56483603e-01 -8.06291580e-01 2.70836651e-01 5.54133356e-02 5.97770870e-01 6.73254073e-01 2.20769688e-01 -7.20050454e-01 -8.97057772e-01 -2.32902586e-01 -2.69761503e-01 -2.37873301e-01 -3.46550085e-02 2.28944480e-01 -1.15485430e+00 8.41384172e-01 -9.62210596e-02 1.75550237e-01 -5.07150710e-01 7.26037398e-02 6.55845463e-01 3.79140735e-01 9.31318820e-01 8.88789475e-01 -3.90848428e-01 -1.77080452e-01 -1.55968595e+00 3.00340414e-01 1.15462554e+00 1.07488334e+00 7.86009848e-01 -2.68703490e-01 -5.04539669e-01 7.17311561e-01 9.69605520e-02 4.05667126e-01 6.96470916e-01 -1.18816781e+00 9.30393100e-01 4.15492326e-01 -1.44465566e-01 -6.92767859e-01 -2.65601158e-01 -2.83698171e-01 -8.63543868e-01 -6.50317669e-01 -3.53324175e-01 -2.64023453e-01 -4.54130560e-01 1.45660114e+00 -2.88402349e-01 -3.14176530e-01 5.80553293e-01 1.70992553e-01 7.82449722e-01 1.09060168e+00 -3.92936580e-02 -1.01153219e+00 9.33466494e-01 -1.28564990e+00 -8.17304254e-01 -3.78721744e-01 9.08603132e-01 -6.67947769e-01 6.76603019e-01 3.51734221e-01 -1.55935895e+00 -3.55914652e-01 -1.13897419e+00 -3.01590681e-01 2.68776059e-01 1.55896232e-01 2.69293785e-01 1.06394298e-01 -1.39167678e+00 8.62330496e-01 -8.75085533e-01 -6.92775130e-01 1.93824828e-01 1.81024939e-01 -1.53643563e-01 2.84419298e-01 -7.80632079e-01 9.10148025e-01 8.64143848e-01 -4.38524276e-01 -5.46746373e-01 -4.04168397e-01 -1.01930726e+00 3.28452468e-01 3.63346696e-01 -1.06934249e+00 1.76752114e+00 -9.37672496e-01 -1.56384373e+00 5.64842284e-01 -7.43532896e-01 -9.70183671e-01 1.03250101e-01 -5.96257925e-01 2.50422150e-01 6.55864120e-01 5.22865117e-01 6.90247178e-01 6.30468726e-01 -1.11785257e+00 -7.04079211e-01 -7.06666782e-02 -4.29826707e-01 5.66983342e-01 -4.91996855e-01 2.83214867e-01 -2.19049320e-01 -6.19360149e-01 -7.61379376e-02 -4.44951862e-01 -4.13065702e-01 -9.53621030e-01 -1.04412210e+00 -4.69663024e-01 3.78527790e-01 -9.14854705e-01 1.77052474e+00 -1.69047356e+00 5.59663415e-01 -2.24627748e-01 2.33726695e-01 -5.02709337e-02 -1.63115144e-01 1.04485440e+00 9.71565098e-02 4.18713540e-01 -7.98208296e-01 -7.88551569e-01 -8.53734165e-02 -7.02467859e-02 -6.38075471e-01 4.77257185e-02 2.67370522e-01 8.75944197e-01 -1.18528104e+00 -1.14294302e+00 -2.06693590e-01 -3.38756740e-01 -8.07433307e-01 3.06209445e-01 -5.05401015e-01 7.23685846e-02 -4.69942093e-01 7.80422390e-02 1.38299689e-01 -1.81918085e-01 2.98710436e-01 2.24106163e-01 -2.63005227e-01 7.70107925e-01 -5.83728909e-01 1.97746575e+00 -2.38684222e-01 8.08099687e-01 -2.46672913e-01 -1.14919901e+00 8.15094709e-01 3.15369755e-01 3.32687020e-01 -1.61119416e-01 -1.84181571e-01 1.21537812e-01 -4.41214114e-01 -5.47760904e-01 1.26693892e+00 -1.52655229e-01 -4.24840391e-01 1.07495975e+00 3.96355897e-01 -5.46896696e-01 7.30940521e-01 1.20103431e+00 1.22456121e+00 -3.95605080e-02 7.61718750e-01 -3.67973089e-01 3.83622319e-01 2.16047034e-01 3.17427397e-01 1.31188035e+00 1.83183402e-01 7.16256618e-01 6.90529883e-01 7.25432783e-02 -1.12785792e+00 -9.63212430e-01 3.00634682e-01 1.08838570e+00 -2.25162715e-01 -1.14546943e+00 -9.88063753e-01 -7.71727920e-01 -3.54081988e-01 1.37664163e+00 -1.87852055e-01 -1.19516678e-01 -6.80153847e-01 -5.20064831e-01 4.93073106e-01 2.99845040e-01 2.90236443e-01 -1.27464592e+00 -5.76154530e-01 3.22769761e-01 -6.86610460e-01 -7.41749048e-01 -7.95370579e-01 1.93160772e-01 -1.27458894e+00 -5.54201066e-01 -3.19581985e-01 -9.07767534e-01 6.64397120e-01 8.49754661e-02 1.22781479e+00 -6.08927794e-02 1.55191079e-01 5.33813894e-01 -4.46999639e-01 -5.66879809e-01 -1.06672907e+00 6.69524014e-01 5.06302416e-02 -1.99651241e-01 9.59456414e-02 -4.38604385e-01 -2.27067709e-01 -7.08924770e-01 -1.02878296e+00 4.63418007e-01 7.39298701e-01 6.14818335e-01 3.49440753e-01 -8.40747878e-02 8.33588183e-01 -1.09764051e+00 1.16009235e+00 -4.36498702e-01 1.54759452e-01 4.87960935e-01 -5.24706244e-01 6.01587892e-01 7.64405906e-01 1.36463925e-01 -1.35565996e+00 -3.30834128e-02 -1.07838651e-02 2.69278198e-01 1.44577958e-02 8.55540395e-01 9.09115598e-02 1.03020263e+00 6.17876887e-01 6.87454164e-01 4.81914580e-02 -5.16151547e-01 4.28698957e-01 9.29799378e-01 6.86501324e-01 -4.42808181e-01 4.01453197e-01 2.22789407e-01 -3.20133507e-01 -1.26014388e+00 -1.38266110e+00 -6.35641873e-01 -9.14597273e-01 -1.88089818e-01 8.35383713e-01 -8.03845584e-01 -1.45006418e-01 -9.31959320e-03 -1.56570148e+00 -3.42102200e-02 -6.77471638e-01 2.66564816e-01 -9.95574117e-01 9.38919783e-01 -6.31501734e-01 -9.36288416e-01 -9.67936158e-01 -5.27948856e-01 1.27943540e+00 1.81430086e-01 -9.23086822e-01 -1.02894509e+00 3.87803704e-01 2.45769486e-01 8.74794722e-02 -4.00705226e-02 8.81353796e-01 -1.07978845e+00 -3.19419712e-01 -1.21054016e-01 2.26657495e-01 5.13086557e-01 1.43728182e-01 1.13984220e-01 -4.59935725e-01 -2.18716443e-01 2.71113038e-01 -4.60175455e-01 1.53174472e+00 5.57549357e-01 8.54057908e-01 -9.21622694e-01 -2.59676933e-01 1.42326742e-01 1.18121839e+00 -2.89478481e-01 3.65365475e-01 1.27038851e-01 4.20178890e-01 7.90943861e-01 5.97758889e-01 4.00959194e-01 4.12083864e-01 9.82815027e-02 -1.89024672e-01 3.21046203e-01 8.08957890e-02 -5.24258375e-01 8.13243032e-01 1.68241286e+00 2.09546775e-01 -5.89791834e-01 -5.42725444e-01 6.43156648e-01 -2.07008314e+00 -1.36598122e+00 8.98389891e-02 1.95229614e+00 1.16136730e+00 2.60279298e-01 2.28730440e-02 -2.30393317e-02 6.84293032e-01 4.21896338e-01 -3.64538521e-01 -9.76303577e-01 -1.63154930e-01 1.14196621e-01 3.06449920e-01 8.84853363e-01 -1.06177557e+00 1.03263032e+00 6.70710230e+00 6.43277884e-01 -4.85148191e-01 -2.25148544e-01 5.39310873e-01 -2.65430689e-01 -5.50610602e-01 2.75942296e-01 -8.37917030e-01 2.51786739e-01 1.35401940e+00 -9.03411925e-01 6.75269440e-02 4.27661985e-01 6.01023853e-01 -4.43011194e-01 -1.22755158e+00 3.81332427e-01 7.22919106e-01 -1.51546192e+00 4.86761421e-01 -2.04099625e-01 8.91924679e-01 -2.77145728e-02 -4.85981822e-01 2.66920924e-01 3.28910738e-01 -6.17227733e-01 6.47399485e-01 8.60695004e-01 4.91229087e-01 -7.81295300e-01 4.80170786e-01 9.87394869e-01 -8.09827566e-01 9.34249833e-02 -3.70753139e-01 -1.85024932e-01 4.06826437e-01 4.38998818e-01 -9.62850690e-01 8.47788513e-01 6.76796883e-02 1.14683354e+00 -6.87164426e-01 7.13384449e-01 -4.17737365e-01 8.43103886e-01 -2.92660911e-02 -3.04487824e-01 1.55476451e-01 -2.15803578e-01 1.12496793e+00 1.75378180e+00 8.01471397e-02 1.37562141e-01 3.71016532e-01 4.54671234e-01 -3.44931215e-01 4.15827483e-01 -6.55135632e-01 -2.32087225e-01 2.77047217e-01 9.31716084e-01 -7.04189241e-01 -9.93364215e-01 -2.14244574e-02 1.19730413e+00 4.68832791e-01 2.51484066e-01 -1.58775464e-01 -6.83287799e-01 -1.99071988e-01 -2.14002147e-01 2.70971417e-01 -8.15808699e-02 -3.96842927e-01 -1.49856353e+00 3.77354100e-02 -8.11227679e-01 4.91139680e-01 -6.25593364e-01 -1.01364613e+00 3.18442702e-01 3.06117266e-01 -9.52036262e-01 -6.72340810e-01 2.13850230e-01 -9.99255657e-01 5.80596864e-01 -1.18582952e+00 -6.17599189e-01 4.16360021e-01 -1.26220807e-01 1.28616893e+00 -1.70397744e-01 8.98115695e-01 -5.72774589e-01 -5.10505378e-01 1.37879506e-01 3.21855873e-01 -1.02934159e-01 7.25111306e-01 -1.56411386e+00 6.00980580e-01 1.15409017e+00 1.21535167e-01 7.71282136e-01 1.10486364e+00 -9.18120861e-01 -1.18760169e+00 -1.15741515e+00 1.74167776e+00 -5.04319727e-01 5.02399564e-01 -8.37354511e-02 -6.62045598e-01 9.10858393e-01 9.40189242e-01 -1.20594811e+00 8.45578432e-01 7.92412162e-02 3.44832987e-02 1.41436204e-01 -6.47053063e-01 6.06169522e-01 7.99133897e-01 -3.37494642e-01 -1.39636719e+00 6.90705955e-01 1.20097494e+00 2.73188530e-03 -5.55460274e-01 3.48475128e-02 2.02694565e-01 -7.05557108e-01 4.46333915e-01 -8.80147099e-01 1.14071012e+00 1.54090986e-01 8.38838443e-02 -1.49992394e+00 -1.07656829e-01 -1.08035970e+00 -3.76919389e-01 1.28764582e+00 6.29847169e-01 -3.28916460e-01 5.39984524e-01 3.48269224e-01 -5.22277892e-01 -5.34273624e-01 -5.29078782e-01 -6.57509863e-01 1.21017739e-01 -2.17694759e-01 5.52553087e-02 2.84705132e-01 7.27181196e-01 1.06449902e+00 -3.49339962e-01 -2.97415972e-01 7.71126509e-01 2.98764408e-01 8.36474240e-01 -1.03988659e+00 -1.40700638e-01 -5.01923144e-01 4.46226373e-02 -1.37919962e+00 4.83618349e-01 -1.25658178e+00 4.50413525e-01 -2.16670513e+00 9.57380295e-01 6.01850390e-01 1.49732474e-02 1.52821913e-01 -4.16636199e-01 -3.59938562e-01 4.30892915e-01 3.85725349e-01 -1.34512770e+00 7.49333799e-01 8.88535440e-01 -3.82871181e-01 -5.54141641e-01 8.02201256e-02 -1.22125280e+00 7.54456043e-01 9.04144526e-01 -7.27206171e-01 -2.95358747e-01 -3.76810223e-01 1.43641397e-01 3.62153530e-01 -2.52936929e-01 -7.44055271e-01 7.27874219e-01 -1.27473203e-02 -5.65946065e-02 -1.12045300e+00 -9.93485376e-02 1.35138363e-01 -5.39085925e-01 5.17295063e-01 -1.10840786e+00 1.57760337e-01 -1.33330628e-01 6.66937292e-01 -3.78572464e-01 -7.26877630e-01 3.89041662e-01 -3.04510981e-01 -8.36345032e-02 -1.51900649e-01 -8.68306458e-01 4.36417997e-01 5.36536992e-01 5.18620536e-02 -1.36887714e-01 -7.43037820e-01 -5.26058555e-01 5.07209957e-01 3.66784722e-01 1.37981296e-01 7.09562421e-01 -6.87724352e-01 -1.31714714e+00 -2.23808974e-01 -1.31026417e-01 1.38851315e-01 -1.69316635e-01 6.01542413e-01 -4.50334549e-01 8.12198281e-01 3.00441295e-01 -5.05578876e-01 -1.27328527e+00 2.80885905e-01 -3.19290370e-01 -7.36765742e-01 -7.89953947e-01 4.70933855e-01 -1.11984141e-01 -2.45156601e-01 1.20257221e-01 -2.78804153e-01 -3.39208186e-01 2.75602818e-01 6.68608963e-01 4.16726708e-01 -1.32122383e-01 -4.77225780e-01 -3.10186623e-03 3.17208432e-02 -5.40567577e-01 -5.47158957e-01 1.53769612e+00 -4.03368324e-01 -5.64567089e-01 7.27016866e-01 1.29669976e+00 2.05137163e-01 -7.27726221e-01 -3.24443042e-01 5.70614994e-01 1.57473311e-01 -2.54825145e-01 -4.45527971e-01 -1.21308133e-01 5.20514667e-01 -6.36440933e-01 4.65789199e-01 1.15466416e+00 2.49474555e-01 8.44621062e-01 1.02468109e+00 -1.63138464e-01 -1.37375438e+00 1.91735849e-01 8.36085498e-01 9.37874973e-01 -9.98411298e-01 5.04758894e-01 -7.27604404e-02 -9.11372364e-01 1.19300258e+00 1.18759237e-01 -1.47680223e-01 2.15617836e-01 -1.01249514e-03 -4.07651633e-01 -2.40398616e-01 -1.37768912e+00 2.42634714e-02 1.46659195e-01 8.50624293e-02 5.40398121e-01 -1.43067883e-02 -6.96660876e-01 5.49648225e-01 -6.26332104e-01 -2.74654806e-01 9.64818358e-01 1.11009598e+00 -1.07127464e+00 -9.00137067e-01 -1.66261390e-01 8.59871686e-01 -5.02691090e-01 -3.69982094e-01 -8.76392543e-01 2.81196505e-01 -8.28888118e-01 1.29700732e+00 1.32742062e-01 -5.38695194e-02 1.55657753e-01 3.87999386e-01 3.27300876e-01 -1.29915297e+00 -7.18567312e-01 1.58955812e-01 3.86110008e-01 -1.71536788e-01 -6.06193244e-01 -1.09937692e+00 -1.29504967e+00 -2.16532126e-02 -2.73805469e-01 7.04074323e-01 5.20036340e-01 1.19331670e+00 3.66124243e-01 4.95565355e-01 8.67097139e-01 -8.28030705e-01 -8.18329275e-01 -1.29669166e+00 -2.60694653e-01 1.12125397e-01 5.10573447e-01 4.55736428e-01 -4.12026852e-01 5.35339534e-01]
[12.500020980834961, 9.499228477478027]
31754f43-49da-4987-98dc-121379f175ce
a-study-on-agreement-in-pico-span-annotations
1904.09557
null
http://arxiv.org/abs/1904.09557v1
http://arxiv.org/pdf/1904.09557v1.pdf
A Study on Agreement in PICO Span Annotations
In evidence-based medicine, relevance of medical literature is determined by predefined relevance conditions. The conditions are defined based on PICO elements, namely, Patient, Intervention, Comparator, and Outcome. Hence, PICO annotations in medical literature are essential for automatic relevant document filtering. However, defining boundaries of text spans for PICO elements is not straightforward. In this paper, we study the agreement of PICO annotations made by multiple human annotators, including both experts and non-experts. Agreements are estimated by a standard span agreement (i.e., matching both labels and boundaries of text spans), and two types of relaxed span agreement (i.e., matching labels without guaranteeing matching boundaries of spans). Based on the analysis, we report two observations: (i) Boundaries of PICO span annotations by individual human annotators are very diverse. (ii) Despite the disagreement in span boundaries, general areas of the span annotations are broadly agreed by annotators. Our results suggest that applying a standard agreement alone may undermine the agreement of PICO spans, and adopting both a standard and a relaxed agreements is more suitable for PICO span evaluation.
['Aixin Sun', 'Grace E. Lee']
2019-04-21
null
null
null
null
['pico']
['natural-language-processing']
[ 3.41664702e-01 4.17804182e-01 -4.97450858e-01 -2.66845912e-01 -9.07958031e-01 -1.04684103e+00 3.33557785e-01 1.05452907e+00 -4.53388989e-01 8.41541767e-01 5.14601469e-01 -6.04131758e-01 -6.97372139e-01 -4.21409398e-01 -3.33444774e-01 -3.47925454e-01 2.92778492e-01 4.24616843e-01 3.15299958e-01 2.24815235e-01 3.83776873e-01 2.79749423e-01 -1.17095363e+00 4.07996058e-01 1.15697563e+00 9.13069129e-01 -4.62936759e-02 2.86432415e-01 -9.32015553e-02 6.13819897e-01 -6.74535751e-01 -4.96879309e-01 2.19733089e-01 -5.88240266e-01 -1.11168802e+00 -2.66950637e-01 1.97441101e-01 1.37092456e-01 9.99459997e-02 1.08958077e+00 5.50449312e-01 -1.48007706e-01 8.70615363e-01 -9.45311308e-01 -4.11055058e-01 9.21782076e-01 -7.74797201e-02 3.48105319e-02 8.03269744e-01 2.94746570e-02 1.02451777e+00 -4.86369789e-01 9.58435118e-01 7.72580266e-01 9.02089953e-01 3.42898279e-01 -1.03046191e+00 -4.62287486e-01 3.56695727e-02 -8.79352391e-02 -1.51379108e+00 -2.41888285e-01 1.87680632e-01 -9.71495390e-01 5.98557293e-01 5.99154234e-01 5.98025858e-01 6.22479975e-01 7.40920603e-01 8.25828165e-02 1.13452756e+00 -5.49150705e-01 5.09701431e-01 2.74009049e-01 1.64375231e-01 3.21025074e-01 7.49082267e-01 -2.09441066e-01 -3.20034832e-01 -8.00860107e-01 3.29218686e-01 -2.08951443e-01 -5.11143804e-01 1.29989862e-01 -1.23278832e+00 5.63945413e-01 -5.53331859e-02 4.87459898e-01 -3.57427269e-01 -4.96608973e-01 8.81823421e-01 1.40751511e-01 2.08020076e-01 9.30637836e-01 -4.03939664e-01 -5.14711328e-02 -1.13822424e+00 1.78291097e-01 9.69340146e-01 9.55389977e-01 2.46543527e-01 -7.04517186e-01 -5.48147976e-01 8.30658138e-01 1.52353182e-01 1.53373003e-01 5.05235076e-01 -1.10474336e+00 3.50999862e-01 7.37289190e-01 4.39703137e-01 -9.02083039e-01 -5.69556057e-01 -9.93925110e-02 -7.76941240e-01 -3.66004743e-02 4.45990026e-01 -1.74847201e-01 -7.58537591e-01 1.68520868e+00 1.54187322e-01 -4.67201948e-01 -5.74509464e-02 8.15394700e-01 1.08101559e+00 1.29405782e-01 4.59484190e-01 -7.48487413e-01 1.94366217e+00 -5.24260342e-01 -1.27572417e+00 8.04317892e-02 1.00004137e+00 -1.05321145e+00 8.05503905e-01 3.38626504e-01 -1.22110212e+00 -2.62949109e-01 -9.92262125e-01 1.19100057e-01 -2.47461990e-01 -2.03004554e-01 1.33957326e-01 4.94334251e-01 -7.43161201e-01 5.90852499e-01 -4.07319278e-01 -2.31798545e-01 1.30008727e-01 1.54546589e-01 -3.92709255e-01 1.57578230e-01 -1.60045409e+00 1.15698087e+00 6.17161334e-01 -8.14302415e-02 -4.90158290e-01 -7.39039958e-01 -7.20755458e-01 5.18594943e-02 6.51580036e-01 -9.14690197e-01 1.12082064e+00 -3.60780388e-01 -7.81213105e-01 1.12196362e+00 1.79890804e-02 -2.97084063e-01 7.83968568e-01 -1.36112481e-01 -7.10879862e-01 3.49286854e-01 4.35888827e-01 1.52391613e-01 1.15149818e-01 -9.74033296e-01 -7.28456140e-01 -3.07013631e-01 3.02751474e-02 3.14798266e-01 -7.38538653e-02 3.31417561e-01 -3.58719260e-01 -5.78504264e-01 9.37378705e-02 -8.13893914e-01 -5.61905980e-01 1.11064017e-01 -3.05457801e-01 -6.05847061e-01 1.92391694e-01 -6.86527729e-01 2.05676436e+00 -1.90450609e+00 -5.83928823e-01 3.37817401e-01 5.16245186e-01 2.35949069e-01 3.37783962e-01 6.16908610e-01 -1.56828120e-01 6.39027596e-01 -3.60925466e-01 4.04553026e-01 7.94738808e-05 1.32085428e-01 -4.02193628e-02 4.45584804e-01 -1.96114898e-01 6.56873822e-01 -1.10288262e+00 -9.99180436e-01 5.37718050e-02 -6.02714578e-03 -2.16933787e-01 1.30760849e-01 9.14191902e-02 3.45347345e-01 -4.31235313e-01 5.22438169e-01 3.81349117e-01 -4.83574033e-01 5.12018681e-01 -3.30665231e-01 -2.25790232e-01 6.24729216e-01 -1.21429801e+00 1.28965497e+00 -2.12584659e-01 2.25423992e-01 -2.13939343e-02 -6.01363599e-01 9.22277093e-01 9.24241662e-01 6.45408452e-01 -2.68792629e-01 3.06053907e-02 6.55106366e-01 2.23992214e-01 -7.55258441e-01 3.26523155e-01 -2.43202627e-01 -2.84359217e-01 5.67390740e-01 -1.13784999e-01 -3.42344433e-01 3.96069199e-01 8.03689882e-02 1.19112110e+00 -2.59510159e-01 1.22281253e+00 -6.52592361e-01 4.48658288e-01 -5.34725329e-03 9.04062212e-01 9.84396815e-01 -4.50778484e-01 7.16687560e-01 9.08821940e-01 -3.73767108e-01 -9.86836612e-01 -8.33432734e-01 -6.36895597e-01 4.39577281e-01 5.83663024e-02 -9.01775956e-01 -7.05383539e-01 -1.02979326e+00 -3.13194662e-01 4.85394716e-01 -8.15051258e-01 5.81786260e-02 -4.28517461e-01 -6.07652068e-01 6.69330955e-01 5.95603347e-01 2.52001788e-02 -7.76299119e-01 -1.15054190e+00 3.64729732e-01 -3.64001960e-01 -1.21983802e+00 -7.04577327e-01 1.51524199e-02 -5.80920100e-01 -1.42395306e+00 -8.11224639e-01 -5.32207668e-01 6.05890274e-01 -1.91918224e-01 1.22460234e+00 2.60744691e-01 -1.23148309e-02 4.00345437e-02 -4.85713363e-01 -6.23670459e-01 -6.00922823e-01 -5.17286733e-02 -1.50398046e-01 -7.08535373e-01 2.52018839e-01 -1.16618797e-01 -8.45552683e-01 6.51821911e-01 -1.14536881e+00 -9.27945971e-02 5.82092285e-01 6.92694068e-01 5.91192007e-01 -4.28931803e-01 6.38553143e-01 -1.20565796e+00 1.03079236e+00 -3.70129377e-01 -3.63118201e-01 5.79293132e-01 -1.12325299e+00 -1.69308335e-01 3.90715361e-01 -4.53423679e-01 -6.84418917e-01 -2.92088687e-01 -1.88973457e-01 1.23051159e-01 -2.52230287e-01 8.43929052e-01 6.20161779e-02 3.44460845e-01 9.39085305e-01 -5.06586254e-01 -2.99972862e-01 -1.01863995e-01 7.21588358e-03 9.15663421e-01 2.76265800e-01 -6.77622974e-01 1.54009923e-01 1.52249768e-01 -1.85110748e-01 -1.46527931e-01 -9.75613177e-01 -6.88568711e-01 -2.88331926e-01 -1.33079737e-01 9.04045701e-01 -5.60859859e-01 -5.78029990e-01 -3.26161057e-01 -1.10481548e+00 -9.00869071e-02 -3.68113756e-01 8.56618524e-01 -2.81886816e-01 7.23639607e-01 -4.53880906e-01 -5.10466576e-01 -6.67987347e-01 -1.27196467e+00 8.49695444e-01 1.26458421e-01 -1.28241360e+00 -1.04950976e+00 2.79258490e-01 1.97692603e-01 -9.86677036e-02 4.94427413e-01 9.84911740e-01 -1.04578304e+00 5.07417202e-01 -4.16437387e-01 -2.14540467e-01 -7.21938983e-02 4.98905659e-01 1.87706742e-02 -4.93599892e-01 1.54959068e-01 2.00893015e-01 4.74957414e-02 2.77045429e-01 4.41549540e-01 1.14640963e+00 -6.84800506e-01 -6.27209187e-01 -4.88440618e-02 1.35517633e+00 5.32438636e-01 4.63582069e-01 2.56391406e-01 1.92193985e-01 8.92612398e-01 7.46819913e-01 3.93245339e-01 5.42783588e-02 5.63334405e-01 1.42513830e-02 4.52680374e-03 1.62186906e-01 1.14154130e-01 -2.02056527e-01 6.47024930e-01 -3.31600100e-01 -2.72952795e-01 -1.40753567e+00 7.52923310e-01 -1.88145220e+00 -5.21861732e-01 -2.99962044e-01 2.34352970e+00 1.34974027e+00 5.13115823e-01 -1.33846834e-01 8.47525150e-02 9.34184074e-01 -8.46978575e-02 -1.78746611e-01 -6.93893969e-01 1.51998639e-01 -8.37691501e-03 4.24505889e-01 4.27246779e-01 -5.90573907e-01 2.50798047e-01 7.42222452e+00 8.28870833e-01 -7.48395741e-01 9.85104144e-02 4.19694334e-01 3.16966593e-01 -3.30168992e-01 1.08712599e-01 -6.35137081e-01 6.91860497e-01 7.63565958e-01 -4.80421990e-01 -3.94328028e-01 4.99195665e-01 3.45895261e-01 -1.82084620e-01 -1.21164930e+00 5.62342584e-01 -8.95450637e-02 -1.37279201e+00 -3.05348281e-02 8.02790523e-02 9.21291471e-01 -5.77075899e-01 -4.52290863e-01 -2.06613198e-01 2.73054510e-01 -9.77472425e-01 6.07162356e-01 4.71935362e-01 9.73050833e-01 -1.30476683e-01 1.35213757e+00 9.07449797e-02 -1.01860809e+00 2.67165750e-01 -9.61335450e-02 4.45179194e-02 3.90998513e-01 1.05541956e+00 -1.01192760e+00 1.05331004e+00 6.39033139e-01 2.46210098e-01 -2.67455399e-01 1.21977282e+00 -3.95831317e-01 5.54047167e-01 -1.22516766e-01 1.43337876e-01 1.13968320e-01 -8.45917836e-02 4.95066851e-01 1.42155766e+00 2.62725890e-01 2.35351950e-01 1.50026321e-01 7.14027107e-01 2.68959347e-02 3.87637496e-01 -3.49607170e-01 -4.70267534e-02 8.05019557e-01 9.58410859e-01 -8.64854455e-01 -6.89749181e-01 -4.17832017e-01 1.32042721e-01 -1.93029955e-01 1.76272601e-01 -6.76167548e-01 -5.10170519e-01 1.69632748e-01 2.12486863e-01 -1.90161094e-01 3.48149717e-01 -8.73904765e-01 -5.73031902e-01 1.25882745e-01 -7.87574947e-01 9.70450878e-01 -5.23897469e-01 -1.19556606e+00 5.57716072e-01 3.44415069e-01 -1.70929790e+00 -5.72138876e-02 -4.47920263e-01 -3.73264551e-01 9.09768939e-01 -7.51827538e-01 -7.00109959e-01 -2.28489473e-01 7.51127228e-02 1.89036965e-01 4.75136697e-01 9.41236496e-01 7.52844214e-02 -1.55561790e-01 4.55196261e-01 -1.99768752e-01 2.23832920e-01 1.17567742e+00 -1.16444790e+00 -2.43833750e-01 4.03755784e-01 -5.08378625e-01 1.06694150e+00 1.01439226e+00 -9.23113227e-01 -3.01799655e-01 -7.31801867e-01 1.30955124e+00 -3.43537867e-01 5.08505881e-01 2.56640673e-01 -1.08395636e+00 5.25204659e-01 2.13071644e-01 -2.18239293e-01 1.28418303e+00 1.33030519e-01 -2.30166644e-01 1.34610474e-01 -1.04082382e+00 7.01007307e-01 8.99432182e-01 -3.04526836e-01 -1.14228284e+00 4.22313780e-01 4.94888246e-01 -5.35594344e-01 -1.61410165e+00 8.75799894e-01 8.33736539e-01 -4.23819959e-01 6.10901713e-01 -5.95488727e-01 4.63061482e-01 -4.39087749e-01 1.45236403e-01 -9.71704364e-01 -1.17242023e-01 -5.56595266e-01 2.89284199e-01 1.07389462e+00 7.79920399e-01 -5.70639849e-01 1.01289116e-01 1.02719402e+00 -3.76856804e-01 -1.10055661e+00 -9.44116831e-01 -6.40979171e-01 5.84465675e-02 -2.75718957e-01 6.04478359e-01 1.29814947e+00 7.14891374e-01 2.76437551e-01 -1.71245579e-02 -5.97491711e-02 4.24549393e-02 2.15975977e-02 2.38100246e-01 -1.47961473e+00 1.61860272e-01 -6.10486686e-01 -1.64693162e-01 -4.12155539e-01 -3.95346999e-01 -5.61222315e-01 1.58337012e-01 -2.05533266e+00 3.62643421e-01 -5.72384000e-01 -4.06995714e-01 5.27507901e-01 -5.37410080e-01 9.14004222e-02 -1.22191578e-01 4.92168814e-01 -6.50741160e-01 -2.15187296e-01 1.13729751e+00 -9.44193974e-02 -2.44185090e-01 -1.05586730e-01 -8.37700188e-01 1.02926719e+00 7.40142643e-01 -8.13934386e-01 -1.60676613e-01 -1.13201611e-01 8.20320189e-01 1.68601826e-01 -9.21192020e-02 -7.37998366e-01 3.40420008e-01 -4.73740786e-01 6.96263611e-02 -3.26448619e-01 -3.56009722e-01 -6.85210705e-01 4.75952297e-01 7.99497306e-01 -7.10966051e-01 3.62873562e-02 5.52871563e-02 8.63369703e-02 -1.73935235e-01 -8.68330717e-01 4.27108169e-01 -3.97409260e-01 -1.81539714e-01 -5.84754534e-02 -4.41207588e-01 2.98629612e-01 9.18055236e-01 -4.25447524e-01 -3.69071543e-01 -1.58630595e-01 -1.03306174e+00 3.67774308e-01 5.25553823e-01 1.18945062e-01 2.15968415e-01 -1.13553524e+00 -9.35177028e-01 -6.60929561e-01 4.56036150e-01 8.19965824e-02 2.46946454e-01 1.27137232e+00 -4.82828468e-01 5.97853899e-01 9.68525410e-02 -4.57667381e-01 -1.46585834e+00 6.49169385e-01 7.54189417e-02 -7.55585909e-01 -4.52478707e-01 1.42996073e-01 3.67048800e-01 -2.33472630e-01 1.93822518e-01 -5.93457997e-01 -5.00801623e-01 3.08396548e-01 6.72035456e-01 3.90810549e-01 3.18298221e-01 -3.86123538e-01 -5.60575843e-01 7.54052997e-01 -1.38691310e-02 -1.66920438e-01 7.30879426e-01 -1.28577381e-01 -1.85960293e-01 5.72065830e-01 5.65178812e-01 3.73659372e-01 -3.06660384e-01 -3.02311461e-02 1.68559477e-01 -1.64344132e-01 -4.56407011e-01 -1.04823828e+00 -1.78692415e-01 4.24133569e-01 1.81775346e-01 3.82178187e-01 9.82214510e-01 1.66144967e-01 2.87246585e-01 -6.86659664e-02 1.75128251e-01 -1.37541294e+00 -3.65454674e-01 1.44396827e-01 1.02738273e+00 -8.71747136e-01 4.32929337e-01 -7.77516007e-01 -6.45370185e-01 1.03339934e+00 3.41930598e-01 4.96594101e-01 5.23644030e-01 2.03947097e-01 1.09684721e-01 -2.60295242e-01 -6.03008449e-01 2.75732905e-01 6.62352920e-01 2.06254035e-01 7.71401763e-01 1.75465897e-01 -1.65049386e+00 9.35896695e-01 -1.67012870e-01 2.87548542e-01 3.84613425e-01 8.17911267e-01 -2.47205213e-01 -8.09508204e-01 -6.95363939e-01 7.39459038e-01 -9.09179211e-01 -1.32170409e-01 -3.28717470e-01 6.85782611e-01 4.24856603e-01 1.07991803e+00 -2.64791727e-01 -4.17929851e-02 4.75574464e-01 4.12831604e-02 1.85221225e-01 -7.48416245e-01 -8.59432399e-01 1.08623557e-01 4.58972245e-01 -2.58053094e-01 -6.60322905e-01 -3.88243735e-01 -1.17339718e+00 2.12811604e-01 -6.48631454e-01 8.51576865e-01 2.31171682e-01 1.30569673e+00 2.09030449e-01 4.62334335e-01 1.53694123e-01 1.96689188e-01 -5.36620021e-01 -1.11789083e+00 -2.77521312e-01 4.85976011e-01 8.14059936e-03 -4.76549178e-01 -3.66266638e-01 2.29102641e-01]
[8.453250885009766, 8.683704376220703]
083f4742-fd3d-4660-b573-4890bc566867
direction-of-arrival-estimation-for-non
2011.02083
null
https://arxiv.org/abs/2011.02083v1
https://arxiv.org/pdf/2011.02083v1.pdf
Direction of Arrival Estimation for Non-Coherent Sub-Arrays via Joint Sparse and Low-Rank Signal Recovery
Estimating the directions of arrival (DOAs) of multiple sources from a single snapshot obtained by a coherent antenna array is a well-known problem, which can be addressed by sparse signal reconstruction methods, where the DOAs are estimated from the peaks of the recovered high-dimensional signal. In this paper, we consider a more challenging DOA estimation task where the array is composed of non-coherent sub-arrays (i.e., sub-arrays that observe different unknown phase shifts due to using low-cost unsynchronized local oscillators). We formulate this problem as the reconstruction of a joint sparse and low-rank matrix and solve its convex relaxation. While the DOAs can be estimated from the solution of the convex problem, we further show how an improvement is obtained if instead one estimates from this solution the phase shifts, creates "phase-corrected" observations and applies another final (plain, coherent) sparsity-based DOA estimation. Numerical experiments show that the proposed approach outperforms strategies that are based on non-coherent processing of the sub-arrays as well as other sparsity-based methods.
['Oded Bialer', 'Tom Tirer']
2020-11-04
null
null
null
null
['direction-of-arrival-estimation']
['audio']
[ 2.66941398e-01 -1.64624900e-01 4.62265849e-01 7.59986788e-02 -9.10042048e-01 -8.70023370e-01 2.51923770e-01 -1.37586057e-01 -1.58008978e-01 7.34244108e-01 6.06822014e-01 2.80716449e-01 -5.86522102e-01 -3.61029238e-01 -8.38132858e-01 -1.36017835e+00 -3.26106340e-01 3.15979272e-01 -1.87252238e-01 -2.33061947e-02 5.81871681e-02 3.84516090e-01 -1.25118339e+00 -1.97534591e-01 5.97317398e-01 6.29150689e-01 1.29766524e-01 6.61101162e-01 4.25633699e-01 5.03245413e-01 -7.09871709e-01 3.01441878e-01 4.02933389e-01 -6.32867336e-01 3.29303443e-01 1.10488804e-02 4.27249610e-01 8.02315474e-02 -3.82691443e-01 1.07141376e+00 5.63768804e-01 -1.05928957e-01 2.98845977e-01 -9.86845970e-01 1.51279703e-01 5.25083244e-01 -6.96605623e-01 9.83932614e-02 5.02558470e-01 -3.62586528e-01 5.77717543e-01 -9.71939564e-01 3.79139930e-01 8.39496315e-01 8.14579844e-01 -3.65620285e-01 -1.36607373e+00 -4.75911111e-01 -3.62982035e-01 -9.03236419e-02 -1.67262697e+00 -9.49462593e-01 9.94434059e-01 -2.65633941e-01 2.82176584e-01 3.59701961e-01 5.28479159e-01 7.06385672e-01 1.83168128e-01 2.21163288e-01 1.08736432e+00 -4.89864081e-01 4.74115580e-01 -3.78603905e-01 9.33375303e-03 3.63151103e-01 9.70892251e-01 5.97337522e-02 -8.98413002e-01 -4.49263841e-01 5.55570662e-01 -1.88134208e-01 -9.36979175e-01 -6.27509475e-01 -1.68585122e+00 4.92970407e-01 3.02408874e-01 8.15224171e-01 -8.48485351e-01 3.63516062e-01 -4.59783912e-01 3.61647099e-01 2.19299003e-01 6.75918162e-01 -1.26267612e-01 9.62053165e-02 -1.41662407e+00 2.44730771e-01 1.07453704e+00 6.78216398e-01 8.95950794e-01 5.70900202e-01 5.83083555e-02 4.74864632e-01 6.09776556e-01 1.27417600e+00 2.31901318e-01 -8.23876083e-01 3.71209741e-01 -2.73156196e-01 6.44213259e-01 -1.46450567e+00 -5.93279123e-01 -1.31527936e+00 -1.03596938e+00 -3.18892747e-01 8.05268824e-01 -8.50631297e-01 -6.55145824e-01 1.92926955e+00 2.52551675e-01 9.02422547e-01 2.73770094e-01 1.02273095e+00 3.23353410e-01 9.66301501e-01 -7.02105999e-01 -8.86986494e-01 1.05201936e+00 -5.31624198e-01 -1.11942673e+00 -7.57515788e-01 2.35288531e-01 -1.03360724e+00 -2.11043477e-01 7.36769617e-01 -1.10672832e+00 -2.62145936e-01 -1.30088317e+00 5.67007422e-01 3.66459787e-01 2.44038612e-01 9.75625589e-02 6.06440783e-01 -1.01165915e+00 5.40850200e-02 -8.70303273e-01 5.27448617e-02 -3.56932551e-01 1.88201070e-01 -2.83122838e-01 -2.43131071e-01 -6.60111904e-01 5.03266335e-01 -8.84078220e-02 5.24968863e-01 -6.33308768e-01 -7.25226521e-01 -6.71535730e-01 1.02512583e-01 5.23109555e-01 -6.29158258e-01 6.66484296e-01 -9.50978220e-01 -1.26663280e+00 1.23619623e-01 -7.82856822e-01 -3.46663624e-01 -1.47216871e-01 -2.09812596e-01 -3.99400800e-01 2.90772527e-01 2.84666806e-01 -3.35351169e-01 1.15551078e+00 -1.50087070e+00 -2.01643854e-01 -2.79011130e-01 -6.25521719e-01 2.82491706e-02 1.72154799e-01 -4.85220373e-01 -2.33090386e-01 -6.91036284e-01 1.07831395e+00 -1.27272332e+00 -4.95453626e-01 -4.20656264e-01 -4.17370975e-01 7.80831873e-01 2.61995345e-01 -6.97721899e-01 1.16295373e+00 -2.36471629e+00 4.95896846e-01 7.10902691e-01 2.63819098e-01 -2.04919785e-01 -2.27151126e-01 7.65420377e-01 -2.37963095e-01 -3.56501907e-01 -2.29684055e-01 -3.71483028e-01 -3.95009905e-01 1.62145551e-02 -2.53067434e-01 8.91135395e-01 -2.72494227e-01 2.94951171e-01 -8.53533864e-01 1.74305975e-01 -1.27706453e-01 4.51655865e-01 -5.57591856e-01 2.03308478e-01 2.65380919e-01 8.82477641e-01 -5.15494645e-01 2.58991003e-01 1.03190506e+00 -1.96923628e-01 5.11199474e-01 -5.98336875e-01 -4.72184777e-01 -1.42954707e-01 -1.95133507e+00 1.86390209e+00 -4.73115802e-01 7.53891766e-01 6.55805707e-01 -1.26499867e+00 8.13658178e-01 5.56839287e-01 7.39597142e-01 -4.63578433e-01 -9.74493474e-02 6.50903344e-01 1.45863086e-01 -4.07808185e-01 2.64665693e-01 -1.51839167e-01 -1.72526374e-01 2.60096908e-01 -2.60799285e-02 -9.41528976e-02 5.34009896e-02 4.49743755e-02 1.39035487e+00 -1.55462563e-01 5.68559408e-01 -3.35900068e-01 4.73666817e-01 -2.07073823e-01 8.52561653e-01 9.30669308e-01 3.97790343e-01 7.85840690e-01 2.50695050e-01 -4.12747525e-02 -7.36309111e-01 -8.76076579e-01 1.63318533e-02 1.63971826e-01 1.58008799e-01 -3.75918627e-01 -3.98566186e-01 2.25964352e-01 -7.03248531e-02 4.36779171e-01 -3.01332980e-01 2.53264427e-01 -9.57963049e-01 -1.04494882e+00 -2.05531362e-02 -2.42756590e-01 4.10400540e-01 1.85464233e-01 -4.69666690e-01 5.55906057e-01 -5.69875360e-01 -1.24818850e+00 -1.63995802e-01 3.57375890e-01 -7.80400753e-01 -5.91870248e-01 -9.08523202e-01 -5.82772315e-01 7.18908846e-01 8.88564467e-01 7.73751795e-01 -2.90044874e-01 2.63061374e-01 5.73758602e-01 -2.73329020e-01 -3.02044481e-01 1.49926886e-01 -3.91155183e-01 3.24976981e-01 7.97072887e-01 -4.23976779e-01 -1.10802710e+00 -5.17192662e-01 4.87831414e-01 -5.81325471e-01 -1.33698925e-01 7.98333883e-01 5.92291355e-01 6.69556022e-01 1.48071319e-01 6.21523321e-01 -4.62363273e-01 2.56549239e-01 -8.11865628e-01 -7.38946080e-01 -5.47287613e-02 -1.47142783e-01 3.31189275e-01 5.57035685e-01 -2.92196035e-01 -1.00986969e+00 4.89062339e-01 2.59303510e-01 -2.69726425e-01 1.62082106e-01 8.72089028e-01 -1.37544215e-01 -4.43404555e-01 5.81009924e-01 4.50300843e-01 -3.02794516e-01 -4.45623040e-01 2.12839544e-01 3.13166887e-01 4.98451680e-01 -1.67314410e-01 1.38995278e+00 6.11362875e-01 5.97217143e-01 -1.40178025e+00 -8.49749029e-01 -6.32690489e-01 -2.68909693e-01 -1.72626466e-01 4.16619182e-01 -1.24522853e+00 -3.02913576e-01 3.73705983e-01 -1.14184916e+00 1.67459577e-01 -1.26816303e-01 1.08212101e+00 -1.68507740e-01 4.50923234e-01 9.37834531e-02 -8.90544355e-01 1.48310661e-01 -6.14544749e-01 9.36526477e-01 2.66946584e-01 -3.41673307e-02 -8.21799159e-01 7.26638734e-01 2.03984771e-02 5.55076420e-01 5.03567576e-01 3.09210420e-01 -3.02987218e-01 -9.13425207e-01 -3.85565430e-01 3.85505080e-01 -1.99510813e-01 1.21156529e-01 -6.98057353e-01 -8.04689288e-01 -5.25369585e-01 6.32162094e-01 3.55494022e-01 4.38005626e-01 9.59000409e-01 1.57147020e-01 -4.67803955e-01 -5.89080930e-01 8.15642834e-01 1.89506650e+00 -5.75157721e-03 4.04784590e-01 -1.31529674e-01 5.07483542e-01 2.72323340e-01 3.96997660e-01 6.20443225e-01 1.61461055e-01 8.29390109e-01 3.90853435e-01 7.95262456e-02 -8.56698826e-02 2.02046871e-01 2.03675658e-01 1.15234828e+00 -1.40301660e-01 -6.25892341e-01 -6.26459122e-01 6.16442859e-01 -2.01643944e+00 -1.03372228e+00 -7.40295708e-01 2.48953366e+00 3.08835536e-01 -2.90486962e-01 -2.00845450e-01 2.03561842e-01 4.89752799e-01 4.62818712e-01 -2.31150016e-01 2.82805979e-01 -4.27027971e-01 2.78520614e-01 7.37336993e-01 6.77784324e-01 -6.99297369e-01 6.64117858e-02 6.35221672e+00 3.86754274e-01 -1.12334037e+00 1.19235493e-01 -1.55940264e-01 -1.49258852e-01 -4.99060035e-01 2.16999933e-01 -5.63256860e-01 5.39968491e-01 1.05728745e+00 -2.95544446e-01 4.26057696e-01 1.10463589e-01 4.60508525e-01 -5.64787149e-01 -1.00400150e+00 1.08776069e+00 2.96158880e-01 -1.12949491e+00 -4.57359701e-01 2.21974924e-01 1.06732798e+00 -1.65132284e-01 -1.62289456e-01 -2.52221555e-01 -1.63229853e-01 -4.82286364e-01 5.79823494e-01 9.99472439e-01 1.50874421e-01 -3.43861401e-01 8.02679360e-01 5.56609154e-01 -1.24794304e+00 -2.67297357e-01 -7.54336268e-02 -2.12671414e-01 5.90928733e-01 1.50648475e+00 -5.09215653e-01 9.83381093e-01 2.18216836e-01 8.19591701e-01 -5.66440634e-02 1.57841969e+00 -2.18193799e-01 1.04248762e+00 -9.00378168e-01 1.02848165e-01 5.63861281e-02 -5.67474484e-01 1.11384833e+00 9.17022586e-01 1.19687998e+00 3.97002280e-01 1.41684592e-01 5.02534211e-01 1.07933305e-01 -2.37873986e-01 -6.28071487e-01 2.46747121e-01 5.83792865e-01 1.21902609e+00 -4.69514906e-01 4.91313227e-02 -4.05641317e-01 5.30347824e-01 -4.46969151e-01 7.66274571e-01 -4.38212156e-01 -1.62901923e-01 3.84169519e-01 1.51101127e-01 5.46564221e-01 -8.10074747e-01 -1.54583737e-01 -1.34967482e+00 1.10227734e-01 -8.37684333e-01 -1.30253121e-01 -8.53044629e-01 -7.22285569e-01 3.12981993e-01 -1.26234397e-01 -1.78152621e+00 -5.51114142e-01 5.68524152e-02 -5.57602823e-01 7.89949596e-01 -1.49744844e+00 -7.17608631e-01 -2.82905340e-01 3.98317784e-01 -6.67964444e-02 1.38248324e-01 6.97174668e-01 4.54474926e-01 -2.76863217e-01 -4.37902212e-02 7.08559036e-01 -2.43614614e-01 7.05403864e-01 -9.64554369e-01 -4.58940536e-01 1.23860180e+00 3.68804365e-01 5.27691126e-01 1.34038532e+00 -4.95156884e-01 -1.94544423e+00 -5.63729107e-01 1.00106156e+00 -3.50165702e-02 5.75106859e-01 -2.79062986e-01 -5.17699599e-01 6.88576579e-01 3.09998930e-01 2.71082014e-01 8.05939853e-01 -2.10951790e-01 2.35356584e-01 -4.11164433e-01 -7.35349357e-01 1.19314298e-01 7.33087897e-01 -3.72351781e-02 -5.54493427e-01 4.71919835e-01 9.80450884e-02 -4.12581503e-01 -6.64176583e-01 1.55022740e-01 4.45444643e-01 -9.18715000e-01 1.21366763e+00 1.34336010e-01 -1.09262012e-01 -8.89000416e-01 -3.90749961e-01 -1.66667736e+00 -5.53372800e-01 -9.40528154e-01 -4.08407599e-02 8.49536240e-01 1.59194306e-01 -7.24956512e-01 5.70427120e-01 -3.30192924e-01 -5.91401346e-02 -7.45768100e-02 -1.25057030e+00 -7.46935248e-01 -6.06249690e-01 -2.51292557e-01 1.76139832e-01 9.29059863e-01 -1.83827400e-01 5.02351701e-01 -9.01180506e-01 1.18478823e+00 1.23646450e+00 3.86071414e-01 8.68959188e-01 -1.14400744e+00 -7.82629728e-01 3.00756186e-01 -2.67422110e-01 -1.49968314e+00 -3.46608669e-01 -5.50116241e-01 4.32680309e-01 -1.32446313e+00 -2.88680971e-01 -3.93779844e-01 7.30186626e-02 -4.26166737e-03 2.00848728e-02 2.19404474e-01 1.33235663e-01 3.68835032e-01 -3.59486699e-01 3.54871988e-01 7.92892277e-01 1.10386364e-01 -2.84064144e-01 3.56385231e-01 -4.57984537e-01 6.95053399e-01 2.26660833e-01 -9.20064569e-01 -2.16898918e-01 -6.82616234e-01 7.52918303e-01 7.89447784e-01 1.97688073e-01 -1.44347179e+00 6.98914349e-01 2.09460214e-01 3.14117491e-01 -5.09490013e-01 5.77775002e-01 -9.26437855e-01 8.83488178e-01 5.06053329e-01 1.26726866e-01 -2.21525162e-01 -7.41750672e-02 8.93855155e-01 -4.14940238e-01 -3.35975885e-01 4.65019524e-01 5.05871437e-02 -1.28620863e-01 -1.90607056e-01 -7.27683485e-01 -1.30012691e-01 7.94807613e-01 -1.28866181e-01 -1.24692410e-01 -9.75064218e-01 -8.24306428e-01 7.13950619e-02 -5.21031246e-02 -3.44500214e-01 2.87074089e-01 -1.29001355e+00 -9.29491162e-01 1.47994325e-01 -3.01260769e-01 -6.05215132e-02 4.01941359e-01 1.12766790e+00 -2.19520524e-01 5.03587723e-01 -6.32747561e-02 -7.38049269e-01 -1.02146888e+00 9.76784825e-02 3.58371228e-01 -2.90084690e-01 -8.43531266e-02 5.28527319e-01 2.76868232e-02 -3.04739028e-02 -2.25667268e-01 -2.56681591e-01 -1.34009898e-01 2.17703775e-01 3.62782359e-01 4.33713406e-01 2.85521541e-02 -7.89073110e-01 -5.02109766e-01 1.27532709e+00 7.76538551e-01 -4.00654197e-01 1.54999721e+00 -4.21178401e-01 -2.69132316e-01 4.50467944e-01 1.17559898e+00 1.16630149e+00 -8.41886818e-01 -3.42896551e-01 -1.93769053e-01 -4.53949034e-01 2.33305231e-01 -5.48771441e-01 -1.07980466e+00 4.11910743e-01 6.13200307e-01 3.79975975e-01 1.26401639e+00 -1.57941848e-01 3.62554878e-01 7.21190035e-01 5.35809934e-01 -4.68948066e-01 5.92967942e-02 3.24144870e-01 9.32071388e-01 -6.34352744e-01 3.39760065e-01 -5.30390859e-01 2.54929215e-01 1.21953309e+00 -2.92521149e-01 -5.24023473e-01 8.42375576e-01 3.85252863e-01 -3.21312062e-02 -8.74645412e-02 -3.06841373e-01 -8.68015513e-02 2.10429341e-01 3.49108279e-01 1.07853919e-01 -2.98516676e-02 -3.69973898e-01 4.72001702e-01 -1.84392348e-01 -9.69135240e-02 9.23711896e-01 8.97919536e-01 -5.08472621e-01 -1.05935085e+00 -1.09439707e+00 2.10923359e-01 -2.09938392e-01 1.11543678e-01 1.46323159e-01 3.79253626e-01 -2.71446351e-03 1.04039681e+00 -1.21101849e-01 -1.08191609e-01 4.50470358e-01 -2.45356143e-01 3.32785904e-01 -4.83341098e-01 -2.52261370e-01 6.82941794e-01 3.77775639e-01 -5.55294693e-01 -6.48476124e-01 -9.58712876e-01 -8.53588164e-01 2.91682184e-01 -3.73467326e-01 4.91395354e-01 8.22466075e-01 8.97614777e-01 4.48376894e-01 5.51132321e-01 1.09484446e+00 -9.92972970e-01 -3.06204140e-01 -5.01931429e-01 -8.52490544e-01 -8.83051604e-02 8.44552636e-01 -3.44662011e-01 -9.78869081e-01 -6.18631542e-02]
[6.494715690612793, 1.3287110328674316]
3a02c5f8-3485-4a11-a159-d1fe9cf5ad23
global-context-aware-attention-lstm-networks
null
null
http://openaccess.thecvf.com/content_cvpr_2017/html/Liu_Global_Context-Aware_Attention_CVPR_2017_paper.html
http://openaccess.thecvf.com/content_cvpr_2017/papers/Liu_Global_Context-Aware_Attention_CVPR_2017_paper.pdf
Global Context-Aware Attention LSTM Networks for 3D Action Recognition
Long Short-Term Memory (LSTM) networks have shown superior performance in 3D human action recognition due to their power in modeling the dynamics and dependencies in sequential data. Since not all joints are informative for action analysis and the irrelevant joints often bring a lot of noise, we need to pay more attention to the informative ones. However, original LSTM does not have strong attention capability. Hence we propose a new class of LSTM network, Global Context-Aware Attention LSTM (GCA-LSTM), for 3D action recognition, which is able to selectively focus on the informative joints in the action sequence with the assistance of global contextual information. In order to achieve a reliable attention representation for the action sequence, we further propose a recurrent attention mechanism for our GCA-LSTM network, in which the attention performance is improved iteratively. Experiments show that our end-to-end network can reliably focus on the most informative joints in each frame of the skeleton sequence. Moreover, our network yields state-of-the-art performance on three challenging datasets for 3D action recognition.
['Ling-Yu Duan', 'Ping Hu', 'Gang Wang', 'Alex C. Kot', 'Jun Liu']
2017-07-01
null
null
null
cvpr-2017-7
['action-analysis', 'one-shot-3d-action-recognition', '3d-human-action-recognition']
['computer-vision', 'computer-vision', 'computer-vision']
[ 2.48778746e-01 -5.55964783e-02 -3.00777227e-01 -2.01117560e-01 -6.44120812e-01 2.40924001e-01 4.79724497e-01 -4.31996316e-01 -3.65484685e-01 3.94538939e-01 6.24895155e-01 7.33181983e-02 6.55329376e-02 -3.90172899e-01 -7.12753832e-01 -8.54950905e-01 -8.48540291e-02 2.85242468e-01 5.23637652e-01 -4.51931693e-02 1.30016968e-01 5.48913479e-01 -1.05413997e+00 4.35154647e-01 4.82788801e-01 9.47015107e-01 3.85337800e-01 5.22816658e-01 -1.80957437e-01 1.37728608e+00 -5.36793470e-01 7.12731928e-02 -1.59867629e-01 -7.48914242e-01 -9.51956391e-01 1.89941823e-01 1.27674401e-01 -6.23368025e-01 -5.87921321e-01 8.51880133e-01 6.64514899e-01 3.94143850e-01 5.26342571e-01 -8.88337553e-01 -5.24434090e-01 3.94342601e-01 -5.83215654e-01 4.27923828e-01 3.58757496e-01 1.17540710e-01 1.04568028e+00 -8.58831167e-01 5.06697774e-01 1.46785200e+00 2.61142045e-01 6.39322460e-01 -6.12015963e-01 -2.51100600e-01 8.28605652e-01 7.48493373e-01 -9.01516795e-01 -2.90653914e-01 8.97283018e-01 -1.96379840e-01 1.34115660e+00 -9.64014903e-02 7.16762364e-01 1.49073768e+00 5.70480406e-01 1.44265139e+00 3.63329500e-01 -3.77077192e-01 1.96981262e-02 -9.54841316e-01 -1.74041018e-01 6.67601824e-01 -4.19914335e-01 -2.30458692e-01 -6.30902350e-01 2.18961626e-01 1.07554448e+00 3.18931222e-01 -7.03463797e-04 -3.63610148e-01 -1.45709133e+00 4.88454431e-01 5.52357256e-01 6.55603051e-01 -8.24720562e-01 4.56970125e-01 6.27351820e-01 1.70274004e-01 5.58123887e-01 4.19837087e-02 -5.02278924e-01 -4.36122984e-01 -3.08313757e-01 -7.97862485e-02 4.15103734e-02 5.95717669e-01 2.83607274e-01 1.56762853e-01 -6.67719007e-01 9.28573489e-01 2.88815737e-01 2.99248159e-01 7.17830658e-01 -9.22158003e-01 5.82646847e-01 7.04732656e-01 2.08146479e-02 -9.26776588e-01 -4.02518332e-01 -2.92663932e-01 -8.52271497e-01 8.04845914e-02 1.60579398e-01 5.23683690e-02 -1.12383473e+00 1.77903223e+00 2.50932783e-01 2.58942097e-01 -4.93064485e-02 1.07302558e+00 4.06962365e-01 5.35399258e-01 2.96391428e-01 -9.92267579e-02 1.07717884e+00 -1.22862387e+00 -8.13479900e-01 -4.54947233e-01 6.96256995e-01 -3.91191572e-01 8.76589358e-01 8.47897157e-02 -8.91595602e-01 -7.19824731e-01 -7.66743660e-01 -1.88090473e-01 3.78079340e-02 1.73554093e-01 4.36454207e-01 -2.86358684e-01 -5.27090430e-01 7.20069349e-01 -1.44157612e+00 -4.30054337e-01 4.97879505e-01 2.32426032e-01 -4.29707050e-01 1.82240494e-02 -1.19552886e+00 9.95776236e-01 2.64205456e-01 6.61726177e-01 -1.12345159e+00 1.22425996e-01 -9.14040864e-01 -9.28908363e-02 6.23268723e-01 -5.58461964e-01 1.33735001e+00 -1.09387398e+00 -1.74724460e+00 4.67685699e-01 -2.23902240e-01 -4.78828847e-01 3.93653512e-01 -6.09137952e-01 -1.12595141e-01 3.55010629e-01 5.27759036e-03 5.94233274e-01 1.09918630e+00 -6.26469195e-01 -6.19006753e-01 -5.32906592e-01 2.23307461e-02 2.38304868e-01 -2.69516736e-01 1.73053235e-01 -5.44755638e-01 -9.17558014e-01 3.84634912e-01 -1.01982164e+00 -3.75031352e-01 -1.57643467e-01 -2.14405462e-01 -5.31686008e-01 8.45151365e-01 -6.97974861e-01 1.04419911e+00 -2.04070258e+00 6.44770145e-01 -2.41444170e-01 -1.44342855e-01 4.90793139e-01 -3.77177924e-01 3.10366809e-01 -4.46656123e-02 -2.39931509e-01 -6.69451952e-02 -3.07342470e-01 -3.70499003e-03 5.96866786e-01 -1.18236594e-01 3.68657619e-01 3.68736267e-01 1.10822546e+00 -8.77578318e-01 -5.11047304e-01 4.31313545e-01 5.34044862e-01 -2.38487661e-01 3.39622468e-01 -4.19792742e-01 7.27512538e-01 -1.04204869e+00 6.42241716e-01 -6.25117496e-03 -2.30001017e-01 3.47914882e-02 -8.31295252e-02 8.15805942e-02 3.01103741e-01 -8.12514842e-01 2.24861717e+00 -2.99676597e-01 2.72462666e-01 -1.85734257e-01 -1.13271821e+00 8.42383444e-01 4.62618232e-01 6.70759797e-01 -9.00465131e-01 2.90200740e-01 -1.62207370e-03 1.20428331e-01 -8.35961938e-01 -4.01754938e-02 -3.80260125e-02 -9.39695165e-03 5.28627336e-01 1.35134771e-01 4.70349640e-01 6.31888658e-02 -1.05281360e-01 1.12194765e+00 5.50925195e-01 8.25324506e-02 2.38240227e-01 8.25596631e-01 -5.19650280e-01 7.64890492e-01 4.49791968e-01 -4.83254492e-01 5.85375309e-01 5.90615332e-01 -6.52783215e-01 -7.82925904e-01 -4.35468197e-01 5.90242922e-01 1.36300838e+00 -6.98758885e-02 -1.86113149e-01 -6.75251007e-01 -9.04691637e-01 -2.97667891e-01 4.31221515e-01 -8.77657115e-01 -5.29256403e-01 -1.02645624e+00 -2.56648213e-01 3.82947952e-01 1.02443278e+00 6.76024497e-01 -1.63702369e+00 -8.99764895e-01 5.17459989e-01 -2.83632517e-01 -1.04460347e+00 -6.14530921e-01 3.01149208e-02 -1.02978408e+00 -1.15084958e+00 -1.08595252e+00 -7.27201164e-01 4.93279070e-01 1.44713596e-01 6.13942862e-01 -1.10720232e-01 1.08308807e-01 3.19936246e-01 -6.02374256e-01 1.65019613e-02 -4.40162309e-02 1.34050339e-01 -1.23882033e-01 3.60742092e-01 4.04549509e-01 -5.74319720e-01 -5.33938110e-01 4.18582439e-01 -5.78296006e-01 -5.70944771e-02 8.47964048e-01 9.65479553e-01 6.48414731e-01 -3.33758801e-01 5.24047732e-01 -3.28650802e-01 2.55795121e-01 -4.86695543e-02 1.63949486e-02 2.83651352e-01 1.90602422e-01 3.61721128e-01 6.12825036e-01 -4.05167013e-01 -1.11005116e+00 2.00555399e-01 -4.18753386e-01 -9.17687178e-01 -1.86884031e-01 5.38864255e-01 -3.06372762e-01 1.84713989e-01 1.78631738e-01 3.77820462e-01 6.84223026e-02 -8.48051250e-01 2.28550527e-02 3.30095679e-01 3.59858632e-01 -4.01195735e-01 5.30580953e-02 3.24157357e-01 4.82012220e-02 -7.98734009e-01 -1.00275469e+00 -1.66241139e-01 -1.09477878e+00 -3.76105458e-01 1.12362194e+00 -7.01385379e-01 -6.59972787e-01 9.54198122e-01 -1.09285069e+00 -5.38763404e-01 -1.36638895e-01 7.70544767e-01 -7.49318361e-01 5.32271564e-01 -8.50858927e-01 -8.26645613e-01 -2.99379408e-01 -1.26131642e+00 1.21149683e+00 -4.61378992e-02 -1.81187093e-01 -6.76627696e-01 6.87508378e-03 1.34181097e-01 6.84621409e-02 3.68745565e-01 8.36202323e-01 -4.59265530e-01 -4.77889746e-01 -3.08507234e-01 -6.44311449e-03 4.66701657e-01 2.86863089e-01 -7.94446766e-02 -5.32663226e-01 -1.45931140e-01 -3.39707993e-02 -4.21814680e-01 1.17167735e+00 6.02685392e-01 1.19151199e+00 -9.76525918e-02 -2.20645875e-01 2.16128156e-01 7.51053572e-01 3.43438178e-01 8.33195031e-01 2.73427367e-01 1.00234735e+00 4.07445818e-01 8.34029317e-01 4.07816976e-01 8.67592543e-02 7.56933987e-01 5.72425306e-01 6.10116571e-02 -5.43250851e-02 -4.78532076e-01 6.10904872e-01 7.61919796e-01 -5.35596728e-01 -1.37463599e-01 -5.90813577e-01 4.53736037e-01 -2.33169174e+00 -9.61437166e-01 9.52086598e-02 2.07805777e+00 4.92464751e-01 4.58467185e-01 2.43011296e-01 5.64745031e-02 6.18082285e-01 6.73486531e-01 -7.01799035e-01 -1.89860374e-01 1.39109090e-01 8.94495174e-02 1.22240186e-01 3.29554021e-01 -1.17299461e+00 1.03316605e+00 5.87722826e+00 8.68900895e-01 -1.06557798e+00 2.16524471e-02 3.86252970e-01 -1.82147831e-01 1.81852981e-01 -2.33290136e-01 -5.48396289e-01 4.25480962e-01 7.12320924e-01 2.58266181e-01 -2.27327690e-01 8.35389137e-01 4.96764511e-01 -8.85331556e-02 -9.51525211e-01 7.96421528e-01 7.54220635e-02 -8.27319741e-01 2.85845041e-01 -4.21995781e-02 4.67400074e-01 -8.60246345e-02 -2.10918918e-01 3.51491153e-01 -4.19753976e-02 -8.71882200e-01 6.37566984e-01 8.87842059e-01 3.65713984e-01 -9.31704879e-01 7.21060872e-01 4.90135968e-01 -1.21444929e+00 -2.37784728e-01 -5.04233599e-01 -2.04531580e-01 3.82596880e-01 2.63218135e-01 -2.61819631e-01 4.06929880e-01 5.76716483e-01 1.27822566e+00 -4.67600256e-01 7.47838378e-01 -6.82011008e-01 3.21136743e-01 -6.25420064e-02 -9.54676718e-02 7.24105299e-01 -5.28172031e-03 5.43595612e-01 8.22351336e-01 2.44143263e-01 3.14590931e-01 3.07830960e-01 3.65811586e-01 8.44620466e-02 -1.94684431e-01 -4.45519924e-01 -2.69854248e-01 -1.95491463e-01 6.85460448e-01 -5.63931346e-01 -2.70362765e-01 -4.36325699e-01 1.30697238e+00 4.11040634e-01 4.21613932e-01 -7.60126233e-01 -1.27747402e-01 6.72427833e-01 -2.97580361e-01 6.33921981e-01 -4.89326477e-01 2.06890494e-01 -1.05763102e+00 1.25767678e-01 -8.21533740e-01 6.44437730e-01 -7.28222430e-01 -1.06496251e+00 4.56794441e-01 -2.35535607e-01 -1.27831960e+00 -4.17407721e-01 -4.45961148e-01 -5.47422230e-01 5.23084223e-01 -1.17132914e+00 -1.27158761e+00 -9.28722844e-02 7.24451840e-01 9.60706532e-01 3.77110252e-03 6.56856954e-01 2.45700836e-01 -7.32173443e-01 2.75899172e-01 -2.40620658e-01 2.20356330e-01 5.67682266e-01 -1.02138543e+00 5.27718544e-01 7.76924968e-01 1.81098282e-01 3.95366877e-01 3.87588710e-01 -7.42793500e-01 -1.30849612e+00 -9.94065166e-01 9.79628980e-01 -2.09463149e-01 5.20211816e-01 -8.29968229e-02 -9.59832549e-01 9.77064610e-01 -8.73243660e-02 2.27765124e-02 2.33409062e-01 4.38509583e-02 1.86104625e-02 -3.74592990e-02 -4.33333874e-01 5.33975482e-01 1.31714988e+00 -5.70365012e-01 -8.62702966e-01 3.05374920e-01 6.57899618e-01 -4.03477341e-01 -6.34096742e-01 5.08411765e-01 6.51879966e-01 -9.27789330e-01 9.27747667e-01 -7.91606903e-01 5.15382290e-01 -2.51420319e-01 5.06987348e-02 -1.02062333e+00 -4.99992162e-01 -3.35858226e-01 -3.94231349e-01 8.13093245e-01 1.21044971e-01 -2.77707458e-01 7.51258910e-01 3.99138987e-01 -4.48759437e-01 -8.35272908e-01 -1.12931979e+00 -7.31391072e-01 -1.33410960e-01 -4.43120450e-01 3.42063010e-01 5.02688289e-01 -1.17296278e-01 3.51528019e-01 -8.78827155e-01 -1.72538787e-01 3.66504073e-01 2.47189924e-01 7.08306193e-01 -7.91520953e-01 -2.29755089e-01 -3.67276341e-01 -5.48873544e-01 -1.57585204e+00 3.50387573e-01 -3.78637224e-01 2.46198311e-01 -1.75441158e+00 1.30918965e-01 2.53721625e-01 -6.94260478e-01 8.16341221e-01 -2.60640949e-01 -2.90913861e-02 1.42857686e-01 1.65011540e-01 -9.78095293e-01 1.08814847e+00 1.73407423e+00 -2.07443669e-01 -2.08615154e-01 1.61410019e-01 -5.35671692e-03 8.18854749e-01 4.77833778e-01 -2.34676868e-01 -1.99534088e-01 -6.91645265e-01 -2.76669532e-01 2.32035384e-01 3.86830628e-01 -1.11211216e+00 2.27031112e-01 -2.13528335e-01 5.79324782e-01 -9.04489696e-01 4.28618968e-01 -6.92164421e-01 -1.36186525e-01 6.02235973e-01 -5.82888424e-01 -2.05654442e-01 -1.32690534e-01 6.23219311e-01 -4.36913013e-01 4.38788421e-02 5.00704944e-01 -4.22498673e-01 -1.09113503e+00 6.35256529e-01 -5.85998535e-01 -2.26876870e-01 8.35882962e-01 -6.57305866e-02 7.72588700e-02 -3.09099764e-01 -1.09621751e+00 2.94729143e-01 1.15024477e-01 6.87593758e-01 5.68650961e-01 -1.64747369e+00 -4.71401840e-01 1.90979496e-01 2.97930557e-03 -1.11865200e-01 6.25119269e-01 9.05526876e-01 -1.15995407e-01 6.60249472e-01 -4.89024878e-01 -5.40212750e-01 -1.12625706e+00 5.78489304e-01 4.39034581e-01 -3.03380102e-01 -9.31381106e-01 1.00064945e+00 1.18942663e-01 6.39837459e-02 4.52553362e-01 -6.33682132e-01 -3.68890524e-01 8.31236765e-02 5.92511177e-01 2.93788046e-01 -1.24465048e-01 -7.98359692e-01 -5.91249704e-01 6.79937124e-01 3.14352848e-02 1.87024996e-01 1.33762646e+00 -1.00870766e-01 4.46144380e-02 6.56079054e-01 1.14053571e+00 -6.77130282e-01 -1.76026249e+00 -2.78204888e-01 8.67135450e-02 -5.14191031e-01 1.45094851e-02 -4.04282302e-01 -1.25129366e+00 1.18392420e+00 2.79642314e-01 -9.26984549e-02 1.08410358e+00 -1.68760400e-02 1.12345374e+00 5.19061625e-01 2.50920326e-01 -1.27675056e+00 6.05587184e-01 8.55194986e-01 1.07674348e+00 -9.78416502e-01 -1.33109316e-01 1.04482390e-01 -7.89429843e-01 1.21026504e+00 7.06847012e-01 -3.32909733e-01 3.37037772e-01 -2.86944449e-01 -1.01681173e-01 -3.22328418e-01 -8.04713428e-01 -4.84838456e-01 3.59004408e-01 3.73258293e-01 2.85887450e-01 -2.91242540e-01 -2.25549877e-01 4.43781555e-01 5.37542462e-01 7.78755248e-02 1.54835256e-02 1.07399786e+00 -6.03621602e-01 -1.09732664e+00 -3.57114337e-02 2.03442246e-01 -4.05537516e-01 3.37100267e-01 -5.60370684e-01 6.23754263e-01 -1.42085701e-01 5.79953790e-01 -7.54689798e-02 -3.91618490e-01 6.38987064e-01 3.69916111e-01 6.10638857e-01 -4.93752509e-01 -2.29038790e-01 4.40658569e-01 2.26990767e-02 -1.02970791e+00 -7.42282093e-01 -7.71833241e-01 -1.36364257e+00 1.59943491e-01 -1.05921954e-01 -1.53213471e-01 1.22934006e-01 1.36993945e+00 4.47741270e-01 9.11998630e-01 5.04057705e-01 -1.09301341e+00 -4.26750243e-01 -1.31689417e+00 -5.30817330e-01 3.54045033e-01 3.61862391e-01 -1.00434959e+00 -9.21909735e-02 -1.49312839e-01]
[7.934614658355713, 0.4385104477405548]
7a495859-1f42-48d9-a1fa-252677d79999
d-lema-deep-learning-ensembles-from-multiple
2012.07206
null
https://arxiv.org/abs/2012.07206v2
https://arxiv.org/pdf/2012.07206v2.pdf
D-LEMA: Deep Learning Ensembles from Multiple Annotations -- Application to Skin Lesion Segmentation
Medical image segmentation annotations suffer from inter- and intra-observer variations even among experts due to intrinsic differences in human annotators and ambiguous boundaries. Leveraging a collection of annotators' opinions for an image is an interesting way of estimating a gold standard. Although training deep models in a supervised setting with a single annotation per image has been extensively studied, generalizing their training to work with datasets containing multiple annotations per image remains a fairly unexplored problem. In this paper, we propose an approach to handle annotators' disagreements when training a deep model. To this end, we propose an ensemble of Bayesian fully convolutional networks (FCNs) for the segmentation task by considering two major factors in the aggregation of multiple ground truth annotations: (1) handling contradictory annotations in the training data originating from inter-annotator disagreements and (2) improving confidence calibration through the fusion of base models' predictions. We demonstrate the superior performance of our approach on the ISIC Archive and explore the generalization performance of our proposed method by cross-dataset evaluation on the PH2 and DermoFit datasets.
['Ghassan Hamarneh', 'Saeed Izadi', 'Kumar Abhishek', 'Zahra Mirikharaji']
2020-12-14
null
null
null
null
['skin-lesion-segmentation']
['medical']
[ 4.12755698e-01 6.60603940e-01 -8.17477554e-02 -9.39547181e-01 -1.26244533e+00 -5.71678638e-01 2.65248418e-01 3.75411958e-01 -8.04334402e-01 8.45779896e-01 -1.33835584e-01 -1.41047817e-02 -6.50519282e-02 -2.33446330e-01 -8.10205400e-01 -7.98563957e-01 4.57326263e-01 9.10852015e-01 5.52758515e-01 2.91837186e-01 1.52718474e-03 8.60843062e-02 -1.24673235e+00 6.97411478e-01 1.12010896e+00 1.30010962e+00 2.70227175e-02 5.18350005e-01 -7.94827342e-02 6.33478642e-01 -7.74048984e-01 -8.41774642e-01 1.14187852e-01 -1.69462413e-01 -1.04609632e+00 3.04208934e-01 5.63547134e-01 -1.23520911e-01 4.32862252e-01 1.19521058e+00 4.38364387e-01 -4.55946326e-01 6.47668540e-01 -1.09115112e+00 -5.52273691e-01 8.63171756e-01 -6.05423033e-01 -2.84893010e-02 -1.39674827e-01 1.10126637e-01 9.10130084e-01 -4.33756977e-01 4.91228372e-01 9.48995948e-01 1.00133014e+00 5.70677042e-01 -1.30402899e+00 -3.96521747e-01 -5.31065697e-03 2.42265970e-01 -1.41310620e+00 -2.30137467e-01 4.35138375e-01 -7.07472384e-01 3.48028302e-01 1.24951757e-01 2.51430392e-01 1.19615841e+00 -8.78618509e-02 9.41138983e-01 1.32795119e+00 -3.81187588e-01 3.02786857e-01 4.78094578e-01 4.22854781e-01 4.92782384e-01 3.43291253e-01 -3.86936337e-01 -3.58389467e-01 -2.72552937e-01 2.47003675e-01 -4.50760245e-01 -3.30093235e-01 -3.22126150e-01 -9.14091766e-01 7.51705766e-01 5.29863179e-01 3.13719511e-01 -3.78211230e-01 9.64091793e-02 4.61002111e-01 -1.22843713e-01 6.03150904e-01 3.62173587e-01 -6.96653366e-01 2.36878052e-01 -1.25650775e+00 4.28791270e-02 8.74703646e-01 6.97037160e-01 5.85097790e-01 -5.99345565e-01 -2.93581456e-01 9.45775330e-01 3.37862462e-01 2.88777091e-02 4.34180379e-01 -1.04191673e+00 2.36818194e-01 6.21491313e-01 2.93289751e-01 -7.18936741e-01 -6.28839493e-01 -6.77266061e-01 -9.63651955e-01 8.41741040e-02 8.17924500e-01 -2.65181869e-01 -1.10264206e+00 1.75055575e+00 3.05078149e-01 7.62588456e-02 -2.55987048e-02 9.95481312e-01 8.83913577e-01 -9.11115780e-02 3.59500229e-01 -1.28725961e-01 1.40598595e+00 -9.79531705e-01 -7.42766619e-01 -4.65519726e-02 6.07925534e-01 -5.36019742e-01 5.23563683e-01 6.87321007e-01 -9.15959895e-01 -6.25570059e-01 -8.95121515e-01 -3.66414003e-02 -2.39737764e-01 6.11796021e-01 2.11912841e-01 6.45731091e-01 -1.11253822e+00 5.25841475e-01 -9.32524681e-01 -1.52389988e-01 7.30488181e-01 5.72444975e-01 -4.08715636e-01 1.57171428e-01 -1.19037116e+00 9.45966065e-01 7.87510216e-01 6.77090049e-01 -5.37716508e-01 -6.05116189e-01 -4.50769991e-01 2.22565793e-02 5.93742788e-01 -6.85864091e-01 1.29887235e+00 -1.44298983e+00 -1.13932097e+00 1.22644544e+00 1.28973112e-01 -7.01363683e-01 1.05841017e+00 -2.60425717e-01 -1.07850157e-01 7.52078444e-02 1.61513999e-01 9.29275751e-01 5.22312343e-01 -1.43629408e+00 -7.40707517e-01 -6.53096497e-01 -1.19426042e-01 -1.70044944e-01 -6.13680705e-02 -3.07883024e-01 -5.10313213e-01 -3.84871751e-01 2.18759939e-01 -1.24692142e+00 -3.71823281e-01 9.95949209e-02 -6.32354438e-01 -4.59373921e-01 4.67005134e-01 -6.34476006e-01 8.84983122e-01 -1.93828797e+00 2.03906208e-01 2.58570701e-01 3.34779501e-01 4.14871216e-01 1.54530302e-01 -1.52947605e-01 2.19759196e-01 2.02507511e-01 -5.35824418e-01 -6.55921876e-01 -6.94310963e-02 4.63866502e-01 4.22957353e-02 4.49422389e-01 3.32326412e-01 6.34680867e-01 -6.65406466e-01 -8.92022848e-01 -5.56425983e-03 5.44189274e-01 -4.32708114e-01 1.20681763e-01 -2.33291000e-01 8.97072613e-01 -4.27653939e-01 4.85706210e-01 6.85401618e-01 -5.85404694e-01 4.31597412e-01 -5.24820149e-01 2.48673216e-01 -2.08622158e-01 -1.19277227e+00 1.83959401e+00 -2.08591893e-01 4.47945684e-01 2.15855941e-01 -1.10739934e+00 8.58122706e-01 4.62171286e-01 4.39263642e-01 -2.75593817e-01 2.89958268e-01 5.84350944e-01 1.36968046e-01 -6.14150226e-01 2.18418255e-01 -1.62747726e-01 5.38197421e-02 2.22564027e-01 5.07254660e-01 1.65736321e-02 2.72660315e-01 5.07789999e-02 9.12334204e-01 8.38415846e-02 2.38798484e-01 -4.53853428e-01 6.20173872e-01 -6.07765689e-02 7.50992656e-01 9.86369848e-01 -6.26458883e-01 8.15738082e-01 7.30916321e-01 -7.43909359e-01 -8.85440707e-01 -7.57201195e-01 -6.41171515e-01 7.22632945e-01 9.06865820e-02 -1.73237190e-01 -9.31699812e-01 -1.18259561e+00 -1.18017107e-01 6.30187035e-01 -1.13090014e+00 2.00179175e-01 -2.47003138e-01 -1.01898098e+00 7.19370425e-01 7.81063974e-01 6.24361932e-01 -6.94573760e-01 -7.71461964e-01 1.51652396e-01 -4.79989290e-01 -1.33731234e+00 -3.95972095e-02 4.62666512e-01 -7.08295226e-01 -1.32246983e+00 -9.15134788e-01 -3.99224102e-01 5.96684337e-01 -4.39766973e-01 1.42351663e+00 1.08416550e-01 -1.45445317e-01 2.22479999e-01 -2.75487363e-01 -7.20885158e-01 -6.38768494e-01 3.92593861e-01 -2.74982035e-01 1.76252455e-01 4.58743781e-01 -1.12261288e-01 -7.45622098e-01 5.92911661e-01 -1.11474395e+00 9.92610902e-02 7.07517803e-01 8.78129780e-01 7.33508468e-01 -3.55851829e-01 5.75564086e-01 -1.25491560e+00 2.77033329e-01 -3.12823713e-01 -6.61741853e-01 6.70937002e-01 -5.45255482e-01 1.72807664e-01 1.56886593e-01 -8.88551474e-02 -1.23741889e+00 3.84485006e-01 -2.99193919e-01 -2.29553044e-01 -4.91292059e-01 4.35952634e-01 1.28962234e-01 1.22651458e-01 7.20533192e-01 -4.01592791e-01 -1.18086906e-02 -3.80273461e-01 1.53910503e-01 8.11871946e-01 7.13205934e-01 -7.25548148e-01 1.22911103e-01 5.41125774e-01 -6.38070777e-02 -1.98970258e-01 -1.57537973e+00 -5.64450383e-01 -1.10497630e+00 -2.57667214e-01 1.32850945e+00 -8.67765069e-01 -6.00840390e-01 4.84911859e-01 -1.38792169e+00 -1.41250581e-01 -6.78547099e-02 3.29531610e-01 -3.78518969e-01 3.84063929e-01 -4.89553511e-01 -7.21875489e-01 -3.57201666e-01 -1.69866049e+00 1.34042811e+00 3.63245070e-01 -4.20925587e-01 -1.01291215e+00 2.10848793e-01 8.54574442e-01 2.24191859e-01 3.06789935e-01 4.06456560e-01 -1.21977258e+00 -2.10308909e-01 -2.42922828e-01 -3.47684383e-01 6.14171445e-01 -1.60213187e-01 5.44226952e-02 -1.30961466e+00 -2.12688483e-02 -6.48001060e-02 -7.17759907e-01 1.07467806e+00 5.43245077e-01 1.42026818e+00 1.01627767e-01 -3.87162507e-01 2.72152901e-01 1.19955218e+00 -3.21730584e-01 4.99395251e-01 3.84890527e-01 5.11636734e-01 7.80035615e-01 4.37095821e-01 2.45722786e-01 2.54434645e-01 5.83448648e-01 5.80861211e-01 -1.64554387e-01 1.60261452e-01 3.74218762e-01 -3.98334533e-01 3.35727721e-01 -3.84855926e-01 -4.28623825e-01 -1.12050116e+00 5.63873410e-01 -2.08759212e+00 -4.58941042e-01 -4.60774034e-01 1.84571660e+00 1.01803899e+00 2.18869761e-01 -1.53808832e-01 -6.97456524e-02 9.10286188e-01 -2.69138753e-01 -5.13213158e-01 3.01568327e-03 -1.72402337e-01 7.73543268e-02 6.38492107e-01 2.07894400e-01 -1.26228595e+00 4.72102582e-01 5.50370741e+00 7.37900078e-01 -8.89813900e-01 3.20139080e-01 1.22628915e+00 2.87037462e-01 1.05906948e-01 -1.91019595e-01 -8.06399167e-01 4.80793327e-01 8.75375450e-01 7.49679565e-01 -4.41385895e-01 8.63127172e-01 -1.92484334e-01 -3.88194442e-01 -1.09879875e+00 7.40842640e-01 1.69405699e-01 -1.27335584e+00 -2.76634753e-01 2.35141418e-03 9.87176180e-01 9.76850912e-02 -7.56841749e-02 7.70112947e-02 2.73781419e-01 -9.47931945e-01 7.47473359e-01 4.70489770e-01 5.14781356e-01 -4.62489903e-01 1.40463078e+00 3.03401917e-01 -4.89448190e-01 1.06369786e-01 -9.67870429e-02 5.80250025e-01 2.27154717e-01 8.02510321e-01 -9.32997882e-01 6.39733493e-01 9.37252283e-01 3.64593655e-01 -7.77298808e-01 1.00164366e+00 -2.46634156e-01 6.50336027e-01 -3.45421046e-01 4.26834345e-01 3.94247502e-01 1.00229174e-01 9.89928395e-02 1.17125833e+00 6.93475977e-02 -6.89373314e-02 7.09553584e-02 1.02827942e+00 -2.19604284e-01 -1.22457139e-01 1.30636185e-01 3.00724030e-01 -4.17395309e-02 1.50542414e+00 -1.05829358e+00 -4.62343127e-01 -2.42731288e-01 8.62156034e-01 4.89491463e-01 1.41824946e-01 -1.02834141e+00 1.69371247e-01 -7.39396438e-02 -1.28912091e-01 3.42854112e-01 1.73809156e-01 -5.96581399e-01 -9.69762146e-01 5.45542128e-02 -7.32298255e-01 6.76378727e-01 -6.65045202e-01 -1.57765949e+00 8.66875947e-01 -3.91249396e-02 -9.60623443e-01 1.18472002e-01 -8.56503963e-01 -1.75224990e-01 7.77646422e-01 -1.41239607e+00 -1.19442916e+00 -6.22633636e-01 2.91513145e-01 3.75058770e-01 1.75467178e-01 6.88046098e-01 3.29628587e-01 -4.81018335e-01 5.33695698e-01 -1.78091213e-01 2.01399550e-01 1.02805936e+00 -1.49528098e+00 -1.80564597e-01 5.10182381e-01 9.51882750e-02 2.92244524e-01 8.80779803e-01 -5.90430439e-01 -4.72069740e-01 -8.76090765e-01 6.09920025e-01 -6.86748624e-01 5.24120212e-01 1.00180320e-01 -1.26955616e+00 6.70199037e-01 2.21468374e-01 2.45875329e-01 1.04186392e+00 2.94075608e-01 -2.81600803e-01 -1.38325766e-02 -1.26397538e+00 2.44664419e-02 6.13097012e-01 -1.73491865e-01 -4.77151453e-01 4.02756691e-01 3.01755130e-01 -6.61810935e-01 -9.83222008e-01 8.06955755e-01 5.84367871e-01 -1.22796786e+00 5.55335402e-01 -4.87155378e-01 3.65515321e-01 -3.20775956e-01 -4.04583327e-02 -1.09702122e+00 2.15949506e-01 -5.59653193e-02 4.07354712e-01 1.14972353e+00 5.39857090e-01 -2.44211033e-01 7.99802661e-01 9.84491885e-01 -3.02482873e-01 -6.37996018e-01 -1.04632974e+00 -2.22002521e-01 1.49330661e-01 -4.96996462e-01 1.26341507e-01 8.46921086e-01 -4.74584162e-01 1.78165153e-01 -2.80785263e-01 3.25562865e-01 6.43218458e-01 -2.16406703e-01 5.99187016e-01 -1.56346083e+00 -5.42792499e-01 -2.47627258e-01 -4.04717416e-01 -4.50674623e-01 2.83531785e-01 -7.32884407e-01 5.08811533e-01 -1.32921243e+00 5.24612367e-01 -4.98481691e-01 -3.00520867e-01 3.88166130e-01 -3.22696209e-01 6.10772729e-01 -4.38668840e-02 3.79860580e-01 -1.05590498e+00 1.32660002e-01 9.72731292e-01 -1.25045747e-01 1.80149049e-01 1.31573156e-01 -4.15120244e-01 1.07625222e+00 5.31612337e-01 -5.78343332e-01 -1.48163632e-01 -6.20148659e-01 3.56248081e-01 -3.48107889e-02 5.34709513e-01 -9.42225873e-01 2.38937870e-01 2.69477844e-01 4.66868192e-01 -5.25140226e-01 7.97054768e-02 -8.93015027e-01 2.45960146e-01 3.30306232e-01 -7.41892636e-01 -3.14615220e-01 8.64408761e-02 9.04206514e-01 -3.50313097e-01 -4.85173911e-01 9.47529435e-01 -3.57051522e-01 -3.88833284e-01 -1.53831383e-02 3.79289910e-02 6.93646520e-02 8.15440774e-01 2.52944455e-02 -3.26962113e-01 -1.91429947e-02 -1.29555655e+00 3.22890908e-01 2.35882431e-01 1.56744242e-01 -1.10063264e-02 -8.53886604e-01 -7.68787324e-01 -2.38475785e-01 2.37989306e-01 4.16779190e-01 5.03467381e-01 1.26142097e+00 -4.49996322e-01 2.81355232e-01 -1.00136727e-01 -1.30521536e+00 -1.18470204e+00 1.57089144e-01 6.07338488e-01 -6.91057622e-01 4.89863381e-02 1.10241926e+00 2.69846529e-01 -4.93797719e-01 3.80554020e-01 -3.51226658e-01 -1.87452629e-01 2.28881747e-01 2.74846852e-01 9.51282308e-02 4.41972584e-01 -6.33808494e-01 -3.89641613e-01 3.63215595e-01 -2.93291569e-01 1.24134563e-01 1.17133164e+00 -1.04203828e-01 -8.12806115e-02 5.18385530e-01 8.33740234e-01 -4.94973838e-01 -1.20942783e+00 -2.20798090e-01 3.57376158e-01 -9.91434902e-02 3.02592479e-02 -1.10187769e+00 -1.16780758e+00 8.91618192e-01 8.15476537e-01 1.80153415e-01 8.63097489e-01 2.87491113e-01 3.67897093e-01 2.90156573e-01 1.42129317e-01 -1.33966219e+00 -5.57167493e-02 1.85174588e-02 7.08934605e-01 -1.91957664e+00 -6.03701510e-02 -3.27525139e-01 -9.06261981e-01 1.25828636e+00 6.35357797e-01 1.36596903e-01 5.01084626e-01 7.95793980e-02 3.46056908e-01 -2.50814497e-01 -5.51926136e-01 -1.77407980e-01 6.39856756e-01 3.66235852e-01 7.41214156e-01 -9.78635275e-04 -5.43500960e-01 8.08535993e-01 2.80287862e-01 3.32475185e-01 3.58980745e-01 7.01051116e-01 -1.40587792e-01 -1.09508622e+00 -2.75767624e-01 3.25823992e-01 -9.21833575e-01 2.02376202e-01 -4.55070078e-01 7.00050116e-01 5.81809759e-01 8.26467276e-01 6.93993568e-02 1.38356432e-01 6.33802935e-02 1.84242621e-01 2.77681559e-01 -5.43614566e-01 -7.42540121e-01 1.23807400e-01 1.78809881e-01 -4.18028921e-01 -1.14515936e+00 -5.91070890e-01 -9.82261956e-01 4.35617983e-01 -4.92944241e-01 2.07563519e-01 7.79453516e-01 1.33455241e+00 1.38216078e-01 6.64204717e-01 -1.68434769e-01 -7.16415405e-01 -5.83725333e-01 -1.09786499e+00 -3.43579531e-01 6.87940001e-01 1.32481232e-01 -6.99167788e-01 -2.77878284e-01 3.43733698e-01]
[14.65034294128418, -2.1045825481414795]
36f44186-ffd0-4f38-a2de-78c88b1228b9
iiitk-dravidianlangtech-eacl2021-offensive
null
null
https://aclanthology.org/2021.dravidianlangtech-1.30
https://aclanthology.org/2021.dravidianlangtech-1.30.pdf
IIITK@DravidianLangTech-EACL2021: Offensive Language Identification and Meme Classification in Tamil, Malayalam and Kannada
This paper describes the IIITK team’s submissions to the offensive language identification, and troll memes classification shared tasks for Dravidian languages at DravidianLangTech 2021 workshop@EACL 2021. Our best configuration for Tamil troll meme classification achieved 0.55 weighted average F1 score, and for offensive language identification, our system achieved weighted F1 scores of 0.75 for Tamil, 0.95 for Malayalam, and 0.71 for Kannada. Our rank on Tamil troll meme classification is 2, and offensive language identification in Tamil, Malayalam and Kannada are 3, 3 and 4 respectively.
['Bharathi Raja Chakravarthi', 'Ruba Priyadharshini', 'Sajeetha Thavareesan', 'Parameswari Krishnamurthy', 'Nikhil Ghanghor']
2021-04-17
null
null
null
null
['meme-classification']
['natural-language-processing']
[-5.81272900e-01 -4.75913942e-01 -4.74065393e-01 2.65805542e-01 -9.87007201e-01 -1.36167979e+00 1.27708042e+00 -3.61147337e-02 -8.65652323e-01 8.29699337e-01 4.53859657e-01 -7.32986569e-01 8.18918943e-02 -3.13447118e-01 1.35737821e-01 -4.91174430e-01 -2.81649902e-02 7.74116337e-01 -2.76852697e-01 -7.34668970e-01 7.84704447e-01 3.36139202e-01 -5.11232793e-01 4.37792391e-01 6.01848602e-01 7.67064929e-01 -6.57173395e-01 1.31382644e+00 -1.95554435e-01 1.41253710e+00 -8.56140494e-01 -3.58617455e-01 3.78034681e-01 -2.81834424e-01 -1.05585515e+00 -6.30570233e-01 9.12079096e-01 -3.40756439e-02 -5.28704822e-01 9.20794487e-01 5.96374810e-01 -5.83359562e-02 1.17736256e+00 -9.65621412e-01 -7.64697015e-01 1.20828998e+00 -8.01196218e-01 5.55731475e-01 2.47282982e-01 -1.49327546e-01 9.83470559e-01 -1.07536674e+00 4.00001347e-01 1.45943451e+00 8.28113973e-01 6.02131307e-01 -6.90609396e-01 -1.13377357e+00 -2.53108680e-01 -3.39398891e-01 -1.64292514e+00 -6.43965840e-01 2.86725491e-01 -7.93303430e-01 8.17616820e-01 5.33004701e-01 2.81594217e-01 1.20356047e+00 4.51769650e-01 1.17097163e+00 1.48479879e+00 -2.12668702e-01 -2.28984267e-01 4.89133835e-01 4.06471491e-01 7.20607758e-01 3.62953357e-02 -3.66140530e-02 -8.76732767e-01 -4.17935342e-01 9.20688361e-02 -6.44729614e-01 3.56299907e-01 7.89557159e-01 -1.30907011e+00 1.16179633e+00 -1.05553813e-01 7.11250782e-01 3.41911435e-01 9.13884193e-02 9.94521320e-01 8.28281641e-01 9.48217869e-01 6.29770637e-01 -3.54652017e-01 -2.19277397e-01 -1.31683314e+00 8.70041773e-02 8.48857939e-01 7.51564860e-01 2.61160582e-01 6.06066644e-01 2.03281701e-01 1.24698031e+00 1.83288530e-01 9.81118143e-01 8.05321097e-01 -1.88815430e-01 6.40434980e-01 1.67127118e-01 3.55047882e-02 -8.88200462e-01 -4.93144751e-01 -3.23832512e-01 -7.91203439e-01 -2.11890265e-01 2.88087845e-01 -4.35610086e-01 -1.10814357e+00 1.18604779e+00 -4.78223026e-01 -3.66556287e-01 6.26396835e-01 5.17314076e-01 1.10879362e+00 9.69177783e-01 2.14364663e-01 1.42824695e-01 1.15797651e+00 -8.55370045e-01 -3.45121741e-01 -2.91302264e-01 1.04125643e+00 -1.36637235e+00 9.16387618e-01 4.52822506e-01 -7.61459053e-01 -2.86839426e-01 -8.52880716e-01 1.59630269e-01 -7.56153226e-01 4.53584522e-01 6.85856342e-01 1.25689566e+00 -8.59332502e-01 1.71477377e-01 -1.79895490e-01 -5.55698872e-01 1.19232975e-01 4.09817934e-01 -4.36028928e-01 5.50060153e-01 -1.51542807e+00 1.05723011e+00 6.58429623e-01 -2.44578227e-01 -9.84401226e-01 -6.92586541e-01 -5.55669665e-01 -6.09006464e-01 -2.23183990e-01 4.23589677e-01 9.21342313e-01 -7.23488569e-01 -1.47150826e+00 1.51530635e+00 4.81976599e-01 -3.95370841e-01 6.69980884e-01 -2.97639281e-01 -9.55991089e-01 -2.41701350e-01 3.21402758e-01 4.51854736e-01 4.64801431e-01 -7.64987946e-01 -8.44984531e-01 -2.07996815e-01 -3.90216380e-01 1.32579491e-01 -6.13965571e-01 7.37474263e-01 1.35573208e-01 -7.67608643e-01 -6.90686852e-02 -1.08026910e+00 1.67536885e-01 -9.90382910e-01 -6.71385109e-01 -3.15774977e-01 8.82461846e-01 -1.10648358e+00 1.39639652e+00 -2.03961682e+00 -1.07727617e-01 4.08127874e-01 4.07598227e-01 3.36245716e-01 -1.71121642e-01 7.42547333e-01 2.00381234e-01 6.01318657e-01 2.41421089e-01 -7.84369484e-02 -1.87994521e-02 -2.56560236e-01 -6.30195796e-01 6.90594256e-01 -4.62088257e-01 7.95080900e-01 -8.43455255e-01 -3.90928298e-01 1.93251312e-01 4.83043268e-02 -6.22155108e-02 -2.77100891e-01 4.41558093e-01 2.39973694e-01 -4.52879786e-01 9.30714130e-01 6.47827446e-01 3.63768101e-01 1.27030402e-01 1.09263755e-01 -6.76838577e-01 2.57380038e-01 -5.49019098e-01 8.96159470e-01 -6.02348685e-01 1.06446636e+00 -4.99791317e-02 -2.55330801e-01 1.34086478e+00 2.57422715e-01 3.93143922e-01 -6.17205501e-01 5.82845509e-01 8.83409977e-01 2.31549770e-01 3.57272150e-03 9.69715238e-01 -2.42852330e-01 -8.81830692e-01 5.02002299e-01 -1.10162944e-01 -2.31975049e-01 1.34301424e-01 6.46669090e-01 5.69965720e-01 -4.66667861e-01 5.27402103e-01 -1.08564472e+00 9.05073881e-01 7.86939934e-02 1.34430379e-01 9.40898955e-01 -4.00998890e-01 3.45094532e-01 1.21571571e-01 -7.17329979e-01 -1.07957017e+00 -1.09524179e+00 -1.27745375e-01 1.65022600e+00 -2.20038429e-01 -6.93354309e-01 -4.44080442e-01 -8.08387339e-01 -1.41966492e-01 6.15961373e-01 -5.96280515e-01 -4.35693637e-02 -7.92686403e-01 -8.37495267e-01 1.76438260e+00 1.42121002e-01 1.01771450e+00 -7.49084592e-01 -5.64577915e-02 1.61659606e-02 -1.96869165e-01 -8.27199042e-01 -9.12178516e-01 8.62355307e-02 -4.34898466e-01 -6.62086487e-01 -1.05472612e+00 -8.08627367e-01 -1.21215448e-01 -3.29067051e-01 9.74939942e-01 -1.14137717e-01 9.90976393e-02 6.53793216e-02 -2.55170554e-01 -3.39291662e-01 -7.37349689e-01 5.96726358e-01 5.05287528e-01 -4.32320498e-02 4.79674816e-01 7.79680610e-02 7.34731182e-02 2.06978783e-01 -2.95721799e-01 -5.74934110e-02 -9.51159559e-03 8.60949457e-01 -3.24353784e-01 -4.24170613e-01 3.89963418e-01 -1.16525948e+00 7.72803009e-01 -5.70729315e-01 -4.43387359e-01 2.89687514e-01 -3.31986636e-01 -2.14790359e-01 8.39435577e-01 -6.94678009e-01 -7.48673737e-01 -1.51849806e-01 -1.96485519e-01 1.27824977e-01 2.90878624e-01 5.46954215e-01 2.99504578e-01 -3.53613377e-01 9.22330499e-01 4.97530103e-01 -5.53896844e-01 -5.88596940e-01 1.62336916e-01 1.48836589e+00 3.91579688e-01 -5.60413063e-01 9.76660907e-01 1.60469890e-01 -5.19772708e-01 -1.30769467e+00 -7.19458342e-01 -6.20349169e-01 -6.91069126e-01 -3.21804434e-01 8.90786469e-01 -9.59258199e-01 -7.86019564e-01 9.84039068e-01 -9.06259298e-01 -7.72032589e-02 2.95914829e-01 4.14823294e-01 -4.29419987e-02 2.39467919e-01 -9.59258616e-01 -1.00203300e+00 -9.09124970e-01 -7.99210131e-01 8.80286396e-01 -1.24692395e-01 -7.66567230e-01 -1.36733878e+00 4.71979648e-01 3.70945215e-01 5.87067306e-01 -5.12570925e-02 8.04962397e-01 -1.17152953e+00 3.89637083e-01 -2.26811305e-01 -2.32532740e-01 4.62292373e-01 1.70807824e-01 -5.29714525e-02 -8.21612895e-01 -4.91472483e-01 -5.41501760e-01 -7.25267529e-01 9.43053365e-01 -1.56048760e-01 2.53516525e-01 -3.74269009e-01 -2.89421454e-02 4.97478038e-01 9.79914308e-01 3.52248847e-01 3.83811414e-01 4.10923332e-01 9.54645991e-01 3.27303439e-01 3.18128735e-01 4.78998452e-01 2.74897635e-01 4.71712559e-01 -3.29996973e-01 1.71506196e-01 -2.77284384e-01 -4.51707959e-01 8.85701776e-01 1.25750923e+00 3.33664417e-02 -3.63402724e-01 -1.53590798e+00 6.47379100e-01 -1.31348753e+00 -7.50196576e-01 -4.04526621e-01 1.81124747e+00 7.60252655e-01 -1.35443181e-01 3.92303467e-01 1.04999796e-01 4.85392392e-01 2.65345454e-01 -2.60654092e-02 -8.95506561e-01 -5.38730681e-01 1.07844070e-01 1.12017572e+00 1.09823596e+00 -1.53421605e+00 2.09616494e+00 6.65633345e+00 1.28819239e+00 -1.45613360e+00 1.99871436e-01 5.57952881e-01 -1.18008181e-01 5.49884029e-02 -1.92120969e-01 -1.20802188e+00 5.70588112e-01 1.24092638e+00 -6.26765907e-01 3.59259665e-01 5.39224803e-01 -1.59739211e-01 1.64890051e-01 -5.90457916e-01 1.19316149e+00 4.68399674e-01 -1.37235177e+00 1.67437941e-01 2.17444003e-01 9.15614307e-01 7.33665705e-01 4.96957630e-01 5.96180201e-01 7.53789306e-01 -1.35537493e+00 9.69506741e-01 -1.95499137e-02 8.02766502e-01 -1.14072418e+00 5.97393632e-01 4.09055591e-01 -1.00969827e+00 4.42772964e-03 -3.62576216e-01 5.42666838e-02 -4.13694590e-01 1.55322887e-02 -8.43367696e-01 1.64591581e-01 5.17501116e-01 9.92578030e-01 -7.95425534e-01 2.99202025e-01 -5.26143610e-03 1.11302388e+00 -6.45801008e-01 -5.66819847e-01 9.16000605e-01 1.23917356e-01 7.26167679e-01 1.81316841e+00 9.63010341e-02 -1.40689835e-01 5.54925501e-01 -9.66796800e-02 -1.38666466e-01 6.04845464e-01 -6.75353825e-01 -3.76611203e-01 3.60817075e-01 1.07693291e+00 -6.74558997e-01 -6.12667739e-01 3.31581533e-01 1.01673317e+00 -1.97870374e-01 6.98324218e-02 -5.92981696e-01 -5.70048690e-01 5.00697017e-01 -4.54101771e-01 -5.90592504e-01 -5.02655745e-01 -5.53154886e-01 -1.43175066e+00 -7.83226788e-01 -1.38892257e+00 8.96636724e-01 -7.42344111e-02 -1.53117979e+00 1.06273293e+00 -1.36473998e-01 -1.24104607e+00 -2.62053639e-01 -1.06136608e+00 -4.54447985e-01 1.04217017e+00 -8.68313611e-01 -1.73534501e+00 5.00140011e-01 5.41445017e-01 6.69699609e-01 -1.33032930e+00 7.79216647e-01 3.54555339e-01 -3.81889015e-01 1.16302669e+00 2.62584448e-01 5.80926061e-01 6.53071582e-01 -9.71054494e-01 4.82350171e-01 4.94365036e-01 3.53089213e-01 7.50937462e-01 5.52402318e-01 -7.43926167e-01 -1.48211908e+00 -8.60930741e-01 1.37318826e+00 -8.66470516e-01 1.35486352e+00 -5.28200209e-01 -3.42366159e-01 7.78348804e-01 3.65759045e-01 -6.59057379e-01 8.05240214e-01 3.39994669e-01 -8.88329983e-01 3.67038637e-01 -1.13557911e+00 8.65785837e-01 6.71062469e-01 -9.35811698e-01 -2.12807700e-01 8.48602295e-01 -1.65767357e-01 -1.41051516e-01 -9.28353488e-01 -2.42905885e-01 7.92549074e-01 -2.52773046e-01 7.29161859e-01 -9.02997077e-01 2.15326071e-01 -1.30530477e-01 -5.01754820e-01 -1.09435308e+00 -8.44416916e-02 -8.35264921e-01 4.32728797e-01 1.08738911e+00 7.54426003e-01 -7.38201141e-01 8.23146999e-01 -3.19096744e-01 2.11338148e-01 1.70893192e-01 -1.11367738e+00 -1.02587545e+00 1.10524619e+00 -4.35071856e-01 -2.05053255e-01 1.43493664e+00 1.35104597e-01 8.35280538e-01 -1.27845180e+00 -2.41139308e-01 7.01320171e-01 -3.28368008e-01 6.75854027e-01 -9.66584504e-01 -1.38943475e-02 -6.62815273e-01 -4.37809765e-01 -7.69604445e-01 4.65254873e-01 -1.56389523e+00 -5.35318911e-01 -8.47244740e-01 1.58301726e-01 -4.54553872e-01 -3.30475926e-01 7.01610088e-01 4.25265491e-01 1.29542589e+00 7.04915524e-01 4.85345066e-01 -3.11753184e-01 -4.64262180e-02 5.56460440e-01 -7.52819479e-01 -2.39650860e-01 -1.79037064e-01 -7.52391756e-01 7.71895111e-01 1.19182563e+00 -4.09870714e-01 -1.70643982e-02 -5.05192697e-01 2.29561493e-01 -2.55812287e-01 -3.26287448e-01 -8.41073453e-01 5.40420227e-02 -2.58726060e-01 3.68449599e-01 -7.36199439e-01 4.03766125e-01 -1.20930731e-01 -3.27159166e-01 8.01504016e-01 -5.42945445e-01 1.37132183e-01 4.01436597e-01 -4.93566781e-01 -1.43109009e-01 2.88947448e-02 1.24697924e+00 -1.45624191e-01 -8.25990796e-01 2.46112525e-01 -1.34554648e+00 3.96779150e-01 6.57271028e-01 -3.32146615e-01 -3.27000797e-01 -4.75599051e-01 -5.70685029e-01 3.10033143e-01 3.03928442e-02 8.67277145e-01 5.20817161e-01 -1.12625253e+00 -1.53526962e+00 4.24148366e-02 2.02129483e-01 -1.70811307e+00 -2.26894513e-01 7.01959014e-01 -8.36984694e-01 5.70984840e-01 -2.48882860e-01 -1.89841554e-01 -1.68977726e+00 -2.73250610e-01 5.47253966e-01 -3.86304885e-01 -2.35482067e-01 9.55869257e-01 -2.64112558e-02 -1.05203760e+00 -1.67273059e-01 5.09570658e-01 -3.41535389e-01 3.57758790e-01 7.11843133e-01 5.85760772e-01 -1.27183601e-01 -1.20866716e+00 -7.85183966e-01 8.78528893e-01 -6.18701756e-01 -5.29780030e-01 7.21207440e-01 2.48696119e-01 -4.41930711e-01 7.15426803e-01 1.46433055e+00 8.04105639e-01 1.64604962e-01 1.88686401e-01 5.40935457e-01 -1.62737012e-01 3.06041658e-01 -1.26415586e+00 -6.82655632e-01 9.13540721e-01 5.01369894e-01 -9.96555854e-03 4.96489257e-01 -1.77568793e-01 9.06829417e-01 6.40630901e-01 3.87142897e-01 -1.23666835e+00 -2.40745068e-01 1.65431571e+00 8.43948185e-01 -9.90489602e-01 -1.31086204e-02 9.53620952e-03 -8.38756025e-01 1.21782911e+00 2.40235493e-01 -1.36508137e-01 5.64939618e-01 4.21198606e-01 8.00311804e-01 1.05575480e-01 -5.22359252e-01 2.76716053e-01 4.32282567e-01 2.88864613e-01 9.93796706e-01 3.31867158e-01 -6.34764433e-01 3.85924131e-02 -8.73864532e-01 -5.64350784e-01 5.75741351e-01 5.68160534e-01 -4.18111920e-01 -9.70357537e-01 -4.30153936e-01 3.75599056e-01 -9.08285737e-01 -6.37821436e-01 -1.32842636e+00 9.53252614e-01 -1.27927557e-01 9.40318704e-01 2.68674325e-02 -1.03147984e+00 -2.20348060e-01 -4.55763787e-02 1.73204079e-01 -4.89735097e-01 -1.47932065e+00 1.31487653e-01 7.33478010e-01 1.42894536e-01 2.74160296e-01 -2.14646205e-01 -1.20436680e+00 -1.00314033e+00 1.86148286e-02 7.97186136e-01 6.04789376e-01 7.15187311e-01 -2.67002672e-01 -3.77673030e-01 6.96499169e-01 -4.62018192e-01 -4.19084400e-01 -1.10917795e+00 -9.06667769e-01 -6.87570870e-02 2.39774644e-01 -2.33093146e-02 -4.61505920e-01 -3.78099889e-01]
[9.674905776977539, 10.700288772583008]
e9f4e8ef-cc02-4155-aad1-bceddd1b4094
learning-unnormalized-statistical-models-via
2306.07485
null
https://arxiv.org/abs/2306.07485v1
https://arxiv.org/pdf/2306.07485v1.pdf
Learning Unnormalized Statistical Models via Compositional Optimization
Learning unnormalized statistical models (e.g., energy-based models) is computationally challenging due to the complexity of handling the partition function. To eschew this complexity, noise-contrastive estimation~(NCE) has been proposed by formulating the objective as the logistic loss of the real data and the artificial noise. However, as found in previous works, NCE may perform poorly in many tasks due to its flat loss landscape and slow convergence. In this paper, we study it a direct approach for optimizing the negative log-likelihood of unnormalized models from the perspective of compositional optimization. To tackle the partition function, a noise distribution is introduced such that the log partition function can be written as a compositional function whose inner function can be estimated with stochastic samples. Hence, the objective can be optimized by stochastic compositional optimization algorithms. Despite being a simple method, we demonstrate that it is more favorable than NCE by (1) establishing a fast convergence rate and quantifying its dependence on the noise distribution through the variance of stochastic estimators; (2) developing better results for one-dimensional Gaussian mean estimation by showing our objective has a much favorable loss landscape and hence our method enjoys faster convergence; (3) demonstrating better performance on multiple applications, including density estimation, out-of-distribution detection, and real image generation.
['Lijun Zhang', 'Tianbao Yang', 'Changyou Chen', 'Lingyu Wu', 'Jiayu Qin', 'Wei Jiang']
2023-06-13
null
null
null
null
['density-estimation']
['methodology']
[ 3.96144181e-01 -7.82416761e-02 -4.17250022e-02 -1.07854858e-01 -1.02380443e+00 -4.14516628e-01 5.13110578e-01 -6.32310733e-02 -6.54443562e-01 8.81969392e-01 -7.73321241e-02 -1.66216597e-01 -1.77073181e-01 -6.76098466e-01 -9.19735014e-01 -1.10183823e+00 1.32088542e-01 4.95257825e-01 1.54349491e-01 1.44138128e-01 1.48650587e-01 4.78852808e-01 -1.61553156e+00 -5.23808002e-01 9.46700811e-01 1.04419065e+00 2.18552262e-01 5.43367207e-01 -1.25579476e-01 4.42247123e-01 -4.51039612e-01 -3.38346839e-01 8.79901089e-03 -8.29554081e-01 -6.40366852e-01 -1.49972379e-01 4.16995674e-01 1.78751558e-01 6.92991987e-02 1.50578952e+00 6.43431067e-01 3.81243408e-01 1.11927474e+00 -1.18053544e+00 -4.40211862e-01 4.32979822e-01 -5.30687690e-01 3.17503102e-02 5.30562252e-02 -1.21300951e-01 1.19176900e+00 -8.80389094e-01 5.03463507e-01 1.17698050e+00 6.83909714e-01 4.52470452e-01 -1.91923273e+00 -4.95785058e-01 -1.90479919e-01 1.51598439e-01 -1.71001947e+00 -3.65123421e-01 7.42634356e-01 -4.61375862e-01 5.37374020e-01 3.16319644e-01 3.61174077e-01 1.01463568e+00 2.94024348e-01 8.01564217e-01 1.15243316e+00 -6.92509055e-01 4.91054207e-01 2.42657766e-01 -3.40781063e-02 6.27453685e-01 2.56193995e-01 1.01399772e-01 -4.12172735e-01 -1.67583764e-01 4.94667292e-01 -5.84728658e-01 -4.64069754e-01 -4.53896284e-01 -9.91899431e-01 8.13198745e-01 2.42856205e-01 3.16916168e-01 -1.43496960e-01 5.42141140e-01 2.72877425e-01 2.29284652e-02 5.13021469e-01 3.81220907e-01 -1.12629257e-01 -2.62715012e-01 -1.03021061e+00 4.16702598e-01 8.71120751e-01 5.83528101e-01 7.44603992e-01 1.39840856e-01 1.07694350e-01 9.02820170e-01 3.06886613e-01 8.60450029e-01 2.44373351e-01 -9.05118883e-01 1.04290150e-01 5.82728274e-02 1.36908352e-01 -7.50703037e-01 -2.69426882e-01 -6.00329399e-01 -1.08996069e+00 3.05251539e-01 7.65645862e-01 6.59043640e-02 -6.22583091e-01 2.22482586e+00 2.43731707e-01 -2.08871327e-02 -1.95656225e-01 6.88033998e-01 1.81681573e-01 5.90064526e-01 8.45073760e-02 -5.33948541e-01 1.05857360e+00 -4.73835558e-01 -8.81145775e-01 5.29795885e-02 3.26737404e-01 -7.34718204e-01 1.11363208e+00 4.19615835e-01 -1.08768380e+00 -1.93109795e-01 -1.09530890e+00 -2.96731330e-02 -1.45704731e-01 7.32492507e-02 1.64566264e-01 7.45116413e-01 -1.08520639e+00 8.04100871e-01 -9.46247041e-01 -3.51269901e-01 3.39625031e-01 2.98568964e-01 -9.56271961e-02 1.96278512e-01 -7.87012339e-01 9.66506720e-01 2.66529769e-01 -1.42237008e-01 -5.50478101e-01 -7.77273178e-01 -9.31026220e-01 1.17534392e-01 3.04072082e-01 -7.85658538e-01 9.81666565e-01 -8.87050927e-01 -1.74104750e+00 4.53378528e-01 -2.74128109e-01 -3.00918400e-01 6.68746710e-01 -5.33800982e-02 -2.50627995e-02 -3.97236980e-02 -2.38433033e-02 3.54524791e-01 1.01694632e+00 -1.19054699e+00 -8.76417309e-02 -1.57555714e-01 -3.94621849e-01 -1.17558911e-01 -2.27016583e-01 -7.86609426e-02 -1.67422667e-02 -8.52409959e-01 1.63984343e-01 -8.33059669e-01 -1.77689269e-01 1.45857707e-01 -4.74170178e-01 1.09812267e-01 5.53260207e-01 -5.81246555e-01 1.43064380e+00 -2.17008400e+00 2.74132401e-01 4.35342997e-01 1.82583705e-01 -2.46025361e-02 -1.13943815e-02 1.91529661e-01 -4.43420373e-02 2.47304067e-01 -8.51971269e-01 -3.11915606e-01 3.60506415e-01 1.23930022e-01 -4.52956930e-02 6.53084695e-01 4.19846535e-01 8.46983612e-01 -9.33206677e-01 -5.56581497e-01 1.12207994e-01 5.24268150e-01 -5.85494816e-01 2.33635958e-02 -1.17703877e-01 1.69982851e-01 -8.23099911e-02 3.53605539e-01 6.86861694e-01 -1.70867726e-01 6.16464615e-02 -3.74042064e-01 -8.15495849e-03 -1.02792822e-01 -1.49785280e+00 1.19369459e+00 -4.95655119e-01 7.69457459e-01 3.16207886e-01 -1.13591111e+00 5.89558959e-01 2.41922792e-02 4.64864016e-01 -4.91913080e-01 1.67924717e-01 4.15940732e-01 6.43276703e-03 -1.74070925e-01 2.82608986e-01 -7.00240970e-01 -2.63485536e-02 4.67400789e-01 1.78076074e-01 -3.60244453e-01 2.17496783e-01 -6.96796253e-02 9.55959499e-01 1.86840191e-01 4.33984697e-01 -7.05125213e-01 4.89295036e-01 -4.14574146e-01 2.75174767e-01 7.33188808e-01 -8.61488208e-02 6.75188065e-01 5.23106098e-01 2.08497539e-01 -1.17274106e+00 -1.43341398e+00 -5.17314792e-01 7.68771827e-01 4.20880616e-02 -2.54863054e-01 -9.81441677e-01 -2.76254952e-01 -1.23775735e-01 9.15652812e-01 -4.68460292e-01 -2.18748808e-01 -4.47013110e-01 -1.35789037e+00 5.13967156e-01 4.11075503e-01 3.77977520e-01 -5.18422961e-01 -3.43692690e-01 2.20338270e-01 -3.32142174e-01 -9.50661778e-01 -7.81226754e-01 4.27073479e-01 -6.88969612e-01 -9.37795401e-01 -6.18146539e-01 -4.63516414e-01 5.77093542e-01 -1.82851389e-01 1.14493203e+00 -2.61369973e-01 -2.80053675e-01 3.67013007e-01 7.85989910e-02 -2.90597022e-01 -6.70937717e-01 -6.70583220e-03 2.20504567e-01 1.41810209e-01 -6.35389462e-02 -9.10722554e-01 -2.98327953e-01 3.38968426e-01 -1.07680726e+00 -2.14080647e-01 4.32410449e-01 9.63024974e-01 8.92010331e-01 1.55457363e-01 5.62634945e-01 -5.30057192e-01 5.44611454e-01 -2.88824588e-01 -7.82591581e-01 2.61847436e-01 -7.18036652e-01 5.85840225e-01 5.90506852e-01 -5.35316944e-01 -9.57111418e-01 -4.00407426e-02 -1.70324072e-01 -2.15415791e-01 2.02427387e-01 2.98702419e-01 -2.97869563e-01 -1.38847336e-01 7.25694597e-01 3.07150334e-01 2.43564487e-01 -4.21283871e-01 4.16177303e-01 4.29041266e-01 5.40150940e-01 -7.98094273e-01 9.51559007e-01 3.39021385e-01 4.83076990e-01 -1.03848612e+00 -6.90856457e-01 -2.19788596e-01 -3.32203299e-01 -1.70278713e-01 8.57502699e-01 -3.33180428e-01 -7.94462740e-01 3.94583076e-01 -9.23368752e-01 -1.05311632e-01 -5.71114063e-01 6.12923384e-01 -7.46564567e-01 4.87358332e-01 -3.47166479e-01 -1.33158243e+00 -1.94614485e-01 -1.09122086e+00 9.96766627e-01 -7.88173079e-02 -2.27228835e-01 -1.21150982e+00 1.14999972e-01 -2.34495297e-01 5.53102255e-01 3.17407370e-01 1.35189450e+00 -3.52376997e-01 -4.36506540e-01 -1.59391209e-01 -2.85037369e-01 7.22650468e-01 -7.48552978e-02 2.52322227e-01 -9.62682366e-01 -8.23882744e-02 2.32679591e-01 -7.59309083e-02 8.35551023e-01 6.92957222e-01 1.17025983e+00 -9.74362716e-02 -1.36509284e-01 6.15474045e-01 1.67715347e+00 -1.75306961e-01 5.41105092e-01 3.77071723e-02 4.27808940e-01 4.47820336e-01 1.68354198e-01 2.62626588e-01 -6.64647967e-02 7.18289375e-01 2.42488235e-01 1.72754079e-01 -2.36733049e-01 -1.16652563e-01 5.25035560e-01 9.27302778e-01 6.74595013e-02 -2.48236477e-01 -6.84961140e-01 2.24169448e-01 -1.79953206e+00 -1.04084957e+00 -2.26547584e-01 2.44422650e+00 8.93753409e-01 1.07674368e-01 1.02805637e-01 1.54106364e-01 6.50913298e-01 1.14613064e-01 -5.16673505e-01 -2.07672521e-01 -3.39597464e-01 4.02215272e-01 4.51590002e-01 7.62619793e-01 -1.02669334e+00 3.55170697e-01 6.84647274e+00 1.42085993e+00 -9.26446140e-01 1.92252204e-01 5.92052042e-01 -1.00652337e-01 -4.96606797e-01 -2.18468800e-01 -6.61350965e-01 6.80893838e-01 8.65848839e-01 -3.67774278e-01 5.33471525e-01 7.00200200e-01 1.58208072e-01 -4.09560084e-01 -1.04797447e+00 1.07572305e+00 -7.85672888e-02 -9.41687822e-01 5.56370281e-02 2.74925649e-01 5.33909500e-01 -8.66760314e-02 -2.68041110e-03 1.32837266e-01 -3.23009454e-02 -1.04060495e+00 9.12398517e-01 6.96976483e-01 8.96529377e-01 -7.04212487e-01 7.38979995e-01 5.03003240e-01 -1.01555252e+00 2.53538758e-01 -2.99284101e-01 9.08222347e-02 2.00179175e-01 1.03980589e+00 -3.78322750e-01 3.27919126e-01 2.82298476e-01 3.08392614e-01 -1.17672563e-01 1.08675075e+00 5.79630025e-02 7.02708781e-01 -8.10997903e-01 -1.93622738e-01 -9.53748450e-02 -7.41582334e-01 7.92103350e-01 1.28649008e+00 4.90629286e-01 -3.24529797e-01 3.40429880e-02 1.27153003e+00 -2.25161582e-01 3.11983764e-01 -3.71961057e-01 -2.57614981e-02 1.43877253e-01 9.96481001e-01 -7.63490319e-01 4.62990813e-02 -7.68093988e-02 7.08733022e-01 1.61069885e-01 5.09841025e-01 -9.68827188e-01 -4.76310939e-01 6.16947889e-01 1.21207282e-01 2.80423284e-01 -1.81697667e-01 -3.44748318e-01 -1.09598279e+00 2.28565291e-01 -5.13875723e-01 -2.15763543e-02 -4.81356561e-01 -1.43329358e+00 5.29126227e-01 2.42802039e-01 -1.02591622e+00 -2.40077004e-01 -7.04943657e-01 -3.95699799e-01 8.83161604e-01 -1.31599474e+00 -6.89931929e-01 9.21911299e-02 1.37088105e-01 1.80827186e-01 1.87592700e-01 7.13436306e-01 4.00315195e-01 -6.16684198e-01 7.26268113e-01 4.71929163e-01 -3.56567442e-01 4.81407106e-01 -1.56331813e+00 1.13958586e-02 8.41277242e-01 2.46753544e-02 3.77723277e-01 9.62764919e-01 -4.00947899e-01 -1.13583362e+00 -7.57271945e-01 7.75563240e-01 -4.54625636e-01 8.56349409e-01 -5.38856983e-01 -9.47625339e-01 1.48754001e-01 -2.42670521e-01 -3.01190820e-02 5.70102036e-01 -1.93951935e-01 -1.98358446e-01 -8.67849439e-02 -1.21890986e+00 4.58357960e-01 9.09183562e-01 -5.64839900e-01 -2.30690405e-01 2.22155556e-01 4.48604792e-01 -1.08176775e-01 -8.28365505e-01 4.12478209e-01 5.40557265e-01 -8.12852323e-01 7.68579781e-01 -2.66634971e-01 3.00871462e-01 -3.18324417e-01 -4.53260839e-01 -1.24005961e+00 -7.16487691e-02 -6.52822673e-01 -2.52411067e-01 1.18528593e+00 4.00207400e-01 -7.32266784e-01 5.88535726e-01 3.98388326e-01 1.55982673e-01 -9.65254128e-01 -1.28085279e+00 -1.00833976e+00 1.47088483e-01 -7.72382140e-01 1.53180823e-01 5.02009630e-01 -3.83856475e-01 3.44936162e-01 -1.94086641e-01 4.09003580e-03 9.38541353e-01 -1.34825006e-01 4.63752478e-01 -1.18037641e+00 -4.12194312e-01 -8.81652594e-01 -2.89403051e-01 -1.13092220e+00 2.76953101e-01 -9.55465317e-01 3.83039325e-01 -9.71360147e-01 3.23157936e-01 -2.68304169e-01 -1.92410648e-01 -1.66390449e-01 -2.18831539e-01 1.94153398e-01 -1.89902578e-02 3.73303443e-02 -4.87235576e-01 8.29482436e-01 1.07115078e+00 -2.50075255e-02 8.12960640e-02 5.82282878e-02 -3.19461435e-01 8.49836349e-01 4.61950570e-01 -6.65837765e-01 -2.85260171e-01 1.42425567e-01 3.00354034e-01 -2.15758309e-01 4.21749473e-01 -1.01351035e+00 5.95113523e-02 2.00845525e-02 1.53596222e-01 -3.70607436e-01 3.86288613e-01 -6.83424473e-01 2.76469260e-01 3.08689505e-01 -1.86497107e-01 -1.96730986e-01 1.13917932e-01 7.14870691e-01 -2.41360217e-01 -6.84853435e-01 1.07684696e+00 1.56285897e-01 -1.33282870e-01 -7.01945089e-03 -3.65164578e-01 4.09761429e-01 5.26134491e-01 -2.37644419e-01 1.33382559e-01 -4.48630542e-01 -7.27490664e-01 -1.37590930e-01 5.68730235e-01 -2.94315219e-01 4.36573297e-01 -1.28868127e+00 -6.28401518e-01 1.36817217e-01 -1.69585064e-01 -1.36399537e-01 1.49489092e-02 1.15716875e+00 -5.03260732e-01 -1.32038696e-02 3.75714481e-01 -7.44415283e-01 -9.86896098e-01 2.45121747e-01 6.13247871e-01 -4.27737087e-01 -1.37068272e-01 7.66264141e-01 3.28518093e-01 -4.22577202e-01 2.70798057e-01 -3.53157163e-01 3.14064920e-01 1.48991689e-01 4.86689448e-01 6.19663477e-01 3.17507647e-02 -5.35981297e-01 -3.87844086e-01 7.37086058e-01 3.66789520e-01 -3.12754810e-01 1.18553722e+00 -1.06398351e-01 -4.11590874e-01 5.74519575e-01 1.50366664e+00 1.74232215e-01 -1.22942412e+00 -1.73271969e-01 9.47503895e-02 -3.10934544e-01 2.18015179e-01 -5.02776325e-01 -7.23190188e-01 7.59786367e-01 6.36188090e-01 2.70972699e-01 1.13027179e+00 -7.88878873e-02 6.75210536e-01 2.77052522e-01 2.92584032e-01 -1.13407779e+00 2.58062407e-03 4.03935492e-01 7.45478272e-01 -1.12339652e+00 9.71137807e-02 -4.80372518e-01 -5.49322665e-02 1.11812580e+00 1.21892035e-01 4.11213748e-02 9.12041306e-01 4.10678118e-01 -2.58862108e-01 5.28317429e-02 -3.51564676e-01 -3.15180510e-01 4.97280419e-01 4.40770417e-01 2.57521152e-01 6.84922412e-02 -3.35001379e-01 4.81147647e-01 -2.30422914e-01 -1.79392740e-01 4.29791868e-01 5.76371312e-01 -3.68323416e-01 -1.10242426e+00 -2.98418313e-01 4.69559371e-01 -4.75900114e-01 -2.59780347e-01 -9.41220000e-02 7.98828125e-01 -7.95803815e-02 7.25712538e-01 -7.10364208e-02 3.87279764e-02 3.02867025e-01 3.49797755e-01 5.95370710e-01 -3.84221673e-01 -8.53801519e-02 1.57133281e-01 -8.15765783e-02 -3.87765706e-01 -5.21841288e-01 -6.33729875e-01 -8.77142966e-01 -2.32031718e-01 -6.44665778e-01 1.78686678e-01 9.02260303e-01 9.67534781e-01 -3.69896889e-02 4.74457979e-01 4.11211163e-01 -8.93175721e-01 -9.74784255e-01 -7.87263215e-01 -8.11788559e-01 4.13940996e-01 2.73773432e-01 -7.49270082e-01 -7.39792824e-01 -1.25624180e-01]
[7.145923137664795, 3.9552276134490967]
7910dee7-bb08-4675-ae89-6eedef44ded7
learn-more-for-food-recognition-via
2303.05073
null
https://arxiv.org/abs/2303.05073v1
https://arxiv.org/pdf/2303.05073v1.pdf
Learn More for Food Recognition via Progressive Self-Distillation
Food recognition has a wide range of applications, such as health-aware recommendation and self-service restaurants. Most previous methods of food recognition firstly locate informative regions in some weakly-supervised manners and then aggregate their features. However, location errors of informative regions limit the effectiveness of these methods to some extent. Instead of locating multiple regions, we propose a Progressive Self-Distillation (PSD) method, which progressively enhances the ability of network to mine more details for food recognition. The training of PSD simultaneously contains multiple self-distillations, in which a teacher network and a student network share the same embedding network. Since the student network receives a modified image from its teacher network by masking some informative regions, the teacher network outputs stronger semantic representations than the student network. Guided by such teacher network with stronger semantics, the student network is encouraged to mine more useful regions from the modified image by enhancing its own ability. The ability of the teacher network is also enhanced with the shared embedding network. By using progressive training, the teacher network incrementally improves its ability to mine more discriminative regions. In inference phase, only the teacher network is used without the help of the student network. Extensive experiments on three datasets demonstrate the effectiveness of our proposed method and state-of-the-art performance.
['Jiang Tian', 'Linhu Liu', 'Yaohui Zhu']
2023-03-09
null
null
null
null
['food-recognition']
['computer-vision']
[ 3.92519951e-01 3.57501626e-01 -4.46297646e-01 -5.43328106e-01 -3.51976514e-01 -4.88935381e-01 1.64404169e-01 5.15871763e-01 -4.10240293e-01 4.41465527e-01 2.07493320e-01 1.35240108e-01 1.74021721e-02 -1.15748131e+00 -6.85314178e-01 -9.73902881e-01 -3.76343466e-02 1.72766268e-01 5.55896223e-01 1.07591180e-02 -4.53729741e-02 3.84507068e-02 -1.32434154e+00 5.02080262e-01 1.04766917e+00 1.01853049e+00 6.22514367e-01 2.86149740e-01 -2.98570007e-01 8.15260589e-01 -2.04492256e-01 1.05872422e-01 2.21129283e-01 -4.20369267e-01 -5.72001040e-01 3.03242266e-01 5.79164289e-02 -3.75847816e-01 -1.49324566e-01 1.44252396e+00 2.77068257e-01 4.59931046e-01 3.36735815e-01 -9.92382586e-01 -9.00881708e-01 1.11042929e+00 -5.04986107e-01 3.13222617e-01 2.37907365e-01 1.18008815e-02 8.33759129e-01 -8.62224042e-01 2.10319668e-01 9.85247552e-01 3.62105012e-01 2.43908033e-01 -1.01692832e+00 -7.60086179e-01 7.83213139e-01 4.06528637e-02 -1.26417696e+00 -1.74549595e-01 1.08620250e+00 -6.37668446e-02 7.14800179e-01 9.28350613e-02 7.70628095e-01 5.05567253e-01 -5.66839576e-01 1.39576626e+00 8.44246209e-01 -3.43152136e-01 2.08327204e-01 5.19233942e-01 3.60907465e-01 8.59472275e-01 8.85994062e-02 3.31091136e-02 -3.93468976e-01 1.63971633e-01 7.79039562e-01 5.69396377e-01 -3.05449367e-01 -6.65991783e-01 -1.27805102e+00 1.02936542e+00 1.04082513e+00 4.43410784e-01 -6.10758245e-01 -2.68723994e-01 2.83178300e-01 3.25039208e-01 5.74346602e-01 3.00908148e-01 -5.69183826e-01 4.79809344e-01 -9.59136009e-01 -2.35148758e-01 5.71426988e-01 7.79877543e-01 1.08402598e+00 -1.71998367e-01 1.18921176e-01 9.49880421e-01 5.84710479e-01 2.72653967e-01 6.43043995e-01 -4.62579757e-01 4.32444870e-01 1.22548521e+00 -1.30004004e-01 -1.01651525e+00 -2.85655618e-01 -7.04632461e-01 -9.90981579e-01 -3.38462042e-03 3.54791462e-01 -4.42169374e-03 -9.38042164e-01 1.60589051e+00 7.99445629e-01 3.33717585e-01 2.66341895e-01 1.01759100e+00 9.97918844e-01 8.10489535e-01 2.08549440e-01 -2.45760418e-02 1.33477271e+00 -1.53486848e+00 -2.97373354e-01 -3.72205824e-01 6.59656107e-01 -6.72613144e-01 5.98757863e-01 3.06361198e-01 -8.58585000e-01 -8.74242723e-01 -1.00245345e+00 1.33916467e-01 -4.22788620e-01 1.74852729e-01 4.59998429e-01 1.65219337e-01 -6.68502450e-01 5.62456250e-01 -7.05590367e-01 -4.64606106e-01 5.68447232e-01 3.09013963e-01 -3.23799193e-01 -3.86758715e-01 -1.15559185e+00 4.01664108e-01 8.44136655e-01 7.68674314e-02 -7.21030056e-01 -6.60959423e-01 -1.23349774e+00 1.24296337e-01 5.36116838e-01 -4.87229824e-01 9.95952189e-01 -1.41055691e+00 -1.21334887e+00 5.52682161e-01 4.73386236e-02 -2.41179079e-01 2.71377742e-01 -2.81915396e-01 -5.46450138e-01 3.86646867e-01 3.73483717e-01 1.15167284e+00 7.25433111e-01 -1.09860396e+00 -1.11458242e+00 -4.83835191e-01 1.30127072e-01 3.28411698e-01 -3.74993503e-01 -3.69011909e-01 -3.42458397e-01 -6.69622540e-01 5.67598701e-01 -6.71676338e-01 -3.07109475e-01 2.87980795e-01 -4.63966548e-01 -4.68398482e-01 9.06979978e-01 -4.76423621e-01 1.00109482e+00 -2.54025555e+00 -1.61471710e-01 4.29645419e-01 1.26100808e-01 2.79002994e-01 -5.09941757e-01 2.50864234e-02 -6.07759468e-02 -2.22728655e-01 -2.13551700e-01 2.21717790e-01 -1.80332392e-01 4.38839763e-01 -2.44095037e-03 3.77095908e-01 2.97860175e-01 8.27667356e-01 -1.38199854e+00 -5.65642655e-01 4.30717736e-01 3.14006507e-01 -5.24296463e-01 1.72147304e-01 -2.27239490e-01 2.10926756e-01 -9.24414337e-01 5.84283352e-01 6.22738063e-01 -4.70992625e-01 1.70665503e-01 -5.68570554e-01 -1.19344793e-01 3.24781090e-01 -1.48248672e+00 1.65574408e+00 -9.71695259e-02 -2.00019434e-01 1.52842831e-02 -1.40099609e+00 1.02816796e+00 1.34825990e-01 4.76971745e-01 -6.97040558e-01 -1.97025776e-01 4.23216373e-02 -2.74981896e-04 -5.64103007e-01 2.04546433e-02 1.22767173e-01 1.13373533e-01 5.32672584e-01 2.85485014e-02 5.67091882e-01 2.00790524e-01 1.47824615e-01 6.84852123e-01 4.09782916e-01 3.82772535e-01 -3.35152566e-01 5.90071321e-01 1.66037560e-01 6.74301207e-01 6.37425721e-01 -4.66883779e-01 6.82870001e-02 -1.67829812e-01 -5.58207273e-01 -8.09349775e-01 -1.22290206e+00 1.98916838e-01 1.57309389e+00 4.38462347e-01 -8.93876255e-02 -4.12793875e-01 -1.37504768e+00 1.67241395e-01 3.68087411e-01 -7.66672671e-01 -3.34284842e-01 -6.18385553e-01 -4.93863493e-01 1.60729781e-01 7.72493064e-01 1.04750299e+00 -1.27423823e+00 -6.54397726e-01 5.66494405e-01 -1.93663090e-01 -4.88427430e-01 -7.45178819e-01 4.04068172e-01 -1.01323247e+00 -1.24098647e+00 -7.89584458e-01 -1.48450387e+00 1.14304352e+00 7.44802535e-01 9.07827735e-01 2.62094826e-01 2.42027873e-03 -1.52486190e-01 -4.54132080e-01 -6.29568100e-02 -3.87109816e-01 1.35603055e-01 -2.28648052e-01 -2.06697434e-01 7.29104400e-01 -3.26293766e-01 -9.11234379e-01 3.16394895e-01 -1.02293479e+00 1.30199865e-01 8.16002488e-01 8.00543308e-01 8.25901449e-01 4.03757662e-01 7.82970667e-01 -8.75728667e-01 1.24469906e-01 -6.88082993e-01 -2.59916037e-01 1.30094364e-01 -4.77270633e-01 4.07360792e-01 6.27527237e-01 -8.41101527e-01 -9.47561562e-01 4.34923172e-01 -6.85798302e-02 -9.41831917e-02 -4.39195484e-01 5.50776720e-01 -1.99438259e-01 4.11250234e-01 4.19368178e-01 4.60430205e-01 3.56965810e-02 -5.75416327e-01 4.16656047e-01 4.03511137e-01 5.61604202e-01 -2.29296789e-01 6.45231783e-01 1.06311858e-01 -6.45454764e-01 -4.87578809e-01 -1.09317112e+00 -8.14548552e-01 -5.92473984e-01 -5.35368547e-02 7.91296065e-01 -1.04722548e+00 -3.78302425e-01 2.38444924e-01 -4.93615180e-01 -2.21796736e-01 -4.56468254e-01 6.64088070e-01 1.10948697e-01 2.42396668e-01 -4.69219267e-01 -5.51359594e-01 -2.07904905e-01 -1.06408107e+00 7.96295404e-01 6.82181895e-01 -9.28267837e-02 -1.05049765e+00 -4.62457351e-02 1.77298233e-01 1.52531311e-01 5.91598451e-02 8.31369042e-01 -1.01024091e+00 -3.14314455e-01 1.72653273e-01 -3.48121196e-01 2.39100233e-01 6.43710494e-01 -4.81268048e-01 -8.54011416e-01 -2.71676600e-01 -1.71939775e-01 -2.47063160e-01 1.21303558e+00 2.50178933e-01 9.84327435e-01 -5.85587859e-01 -5.53397059e-01 3.75180006e-01 1.47036135e+00 2.76811272e-01 1.57719746e-01 4.92028385e-01 6.71138585e-01 6.73459709e-01 4.83507872e-01 6.57482892e-02 4.81566101e-01 2.54269540e-01 5.13737440e-01 -6.31646097e-01 1.51505172e-02 -5.22884548e-01 5.98802686e-01 6.69554472e-01 4.79917705e-01 2.52243668e-01 -1.43467113e-01 7.81403363e-01 -1.87351584e+00 -9.11959291e-01 2.04327717e-01 1.98125231e+00 9.89423752e-01 -4.21510860e-02 3.30219805e-01 1.64515138e-01 7.11700797e-01 2.42219314e-01 -1.05285239e+00 -3.83982249e-02 1.25264198e-01 -1.01800688e-01 1.63912296e-01 2.85920650e-01 -1.18811524e+00 8.34182739e-01 4.97917271e+00 6.46373510e-01 -9.26239669e-01 -9.15220529e-02 6.85893238e-01 1.14644594e-01 -2.23540470e-01 -3.29229206e-01 -7.98818588e-01 5.48072040e-01 3.71269435e-01 3.84626180e-01 3.11036974e-01 1.23109174e+00 1.47252176e-02 -3.38748455e-01 -1.08720016e+00 4.13294643e-01 9.88287553e-02 -9.93364334e-01 -4.19233330e-02 -1.54506996e-01 7.28794277e-01 3.12100407e-02 -8.95969942e-02 5.05833268e-01 6.77489281e-01 -6.80216193e-01 4.85668570e-01 2.21285746e-01 1.83448777e-01 -8.67446184e-01 7.59668529e-01 6.72952354e-01 -1.66236222e+00 -8.39408264e-02 -5.19400239e-01 6.82698339e-02 -2.43057385e-01 9.51344371e-01 -7.61864662e-01 4.80374992e-01 7.51662910e-01 7.28004098e-01 -4.47180659e-01 7.92595148e-01 -4.71122026e-01 5.52319467e-01 -6.00321114e-01 -1.07371680e-01 5.73149502e-01 -2.93875068e-01 1.79350570e-01 1.16102028e+00 6.26456812e-02 2.99268514e-01 1.05280399e+00 7.15428054e-01 -3.89991701e-02 3.62445831e-01 -4.48897243e-01 -4.90661769e-04 4.44631666e-01 1.31166661e+00 -1.02346492e+00 -7.80508876e-01 -3.46530199e-01 1.17227697e+00 4.35398012e-01 2.59731561e-01 -5.54143727e-01 -2.96204954e-01 4.65077847e-01 -1.19401008e-01 6.97043836e-01 2.25192994e-01 1.65678978e-01 -9.03323591e-01 -1.35322303e-01 -5.77749789e-01 8.33596706e-01 -6.07235849e-01 -1.34384489e+00 3.63264412e-01 -1.57094136e-01 -8.42109978e-01 -1.52886823e-01 -1.77758694e-01 -6.38663948e-01 7.11138427e-01 -1.78019345e+00 -1.03438723e+00 -2.39781529e-01 7.59312570e-01 6.65376663e-01 1.05691426e-01 7.63255358e-01 1.42672077e-01 -5.41436315e-01 4.36678797e-01 -2.40435966e-05 4.94019449e-01 5.09744048e-01 -1.27515590e+00 -1.54648989e-01 7.43956864e-01 3.23933333e-01 6.29162431e-01 4.70086187e-01 -7.50797212e-01 -9.88576353e-01 -1.57445800e+00 7.49854207e-01 2.12277696e-01 3.33535075e-01 -3.47223654e-02 -1.05202258e+00 4.39236909e-01 -4.59981523e-03 8.73510838e-02 7.79285491e-01 2.61909012e-02 -4.29187953e-01 -3.38087350e-01 -1.45794630e+00 2.96986848e-01 7.22273648e-01 -3.62839580e-01 -8.61276269e-01 3.70907217e-01 7.46847987e-01 3.05844713e-02 -8.11569154e-01 3.86019766e-01 4.02294010e-01 -7.55502343e-01 1.01915693e+00 -2.80813485e-01 1.98689193e-01 -6.52055740e-01 -1.67623106e-02 -1.55391610e+00 -6.55139387e-01 1.29166409e-01 -4.39333655e-02 1.15326476e+00 4.86037403e-01 -7.10526288e-01 8.12361896e-01 1.74258560e-01 -1.72445238e-01 -8.48653793e-01 -3.56280237e-01 -5.21845520e-01 -2.35086471e-01 4.70698103e-02 7.94016898e-01 1.15612876e+00 7.18788505e-02 2.15625152e-01 3.39775160e-02 4.92816091e-01 5.08557737e-01 5.78635871e-01 3.59751672e-01 -9.76739168e-01 -2.76790321e-01 -1.09222688e-01 -3.94821987e-02 -1.56161189e+00 -2.24053830e-01 -1.08417308e+00 2.99998999e-01 -1.59449375e+00 4.10905272e-01 -5.46756268e-01 -8.25249910e-01 1.05264592e+00 -5.33201039e-01 3.33482206e-01 5.44727743e-02 2.15326678e-02 -7.50341356e-01 2.42662922e-01 1.33123827e+00 -3.69475514e-01 -5.02202153e-01 -5.41713275e-02 -9.45635021e-01 8.40034962e-01 9.43921745e-01 -5.61494708e-01 -7.57732749e-01 -2.71489829e-01 -1.38361499e-01 -3.97740871e-01 3.20407748e-01 -8.12367857e-01 1.88942030e-01 -1.79003533e-02 1.08440220e+00 -5.80476522e-01 -2.37338871e-01 -8.79979074e-01 -8.93188044e-02 8.03492725e-01 -5.18720686e-01 -6.92212731e-02 -7.63256773e-02 3.94173712e-01 -2.03736022e-01 -3.04545581e-01 7.15511799e-01 -5.05782068e-01 -8.52686644e-01 4.17945683e-01 -1.59798577e-01 -2.22348288e-01 9.12943423e-01 -3.83818030e-01 -1.23877220e-01 1.04786225e-01 -9.19755220e-01 5.73499203e-01 3.12216192e-01 4.13816750e-01 7.57539630e-01 -1.37381780e+00 -5.38259208e-01 5.58632612e-01 8.61169845e-02 3.76181960e-01 2.03773871e-01 6.75665259e-01 1.26047894e-01 1.23924740e-01 1.08078822e-01 -7.38616109e-01 -1.13160026e+00 8.93702626e-01 7.51019493e-02 -3.92827481e-01 -8.89588475e-01 7.31549442e-01 6.72487795e-01 -4.62503880e-01 1.87124297e-01 -7.58979201e-01 -4.61927593e-01 9.92738008e-02 9.06236172e-01 5.49077280e-02 -2.79375345e-01 -4.42706406e-01 -2.14952260e-01 3.15264344e-01 -3.72790873e-01 3.19683492e-01 1.44268286e+00 -2.15905502e-01 1.71212763e-01 2.13266596e-01 1.16697371e+00 -3.62935066e-01 -1.40934002e+00 -5.87817848e-01 -2.47349486e-01 -3.30830365e-01 1.70727387e-01 -1.01995099e+00 -1.45382881e+00 6.04638577e-01 6.81630433e-01 1.97912708e-01 1.35835695e+00 1.59246594e-01 9.14019942e-01 2.75015831e-01 1.99579924e-01 -1.06103289e+00 2.51309127e-01 1.77181780e-01 3.84787917e-01 -1.31592965e+00 -2.18520939e-01 -4.43787098e-01 -5.35031378e-01 9.40477192e-01 7.62684047e-01 -2.59035140e-01 7.73928285e-01 2.20320106e-01 1.41284540e-01 -1.53142869e-01 -4.14683700e-01 -2.91343391e-01 3.12420845e-01 6.34119034e-01 2.32121721e-01 8.67755860e-02 2.64987886e-01 6.14288688e-01 7.28830472e-02 -2.49246676e-02 -1.80251747e-01 9.56729293e-01 -1.10826981e+00 -8.79413426e-01 -3.28056037e-01 4.47229773e-01 -3.52933146e-02 -2.17677876e-01 2.66267243e-03 6.24691546e-01 4.58080590e-01 1.03084886e+00 2.06437126e-01 -1.74333811e-01 1.35574311e-01 1.66527301e-01 2.95767635e-01 -6.88743770e-01 -7.54121304e-01 4.43072915e-01 -9.88077000e-02 -4.99230891e-01 -7.51268327e-01 -5.35431325e-01 -1.55771327e+00 4.57283594e-02 -5.42607963e-01 4.36728269e-01 2.42274851e-01 1.02423966e+00 1.03907347e-01 5.24060786e-01 8.15787911e-01 -3.35473746e-01 -4.46735203e-01 -9.54021692e-01 -6.04172826e-01 5.82138062e-01 3.28038603e-01 -5.06023884e-01 -1.16887242e-01 1.79923162e-01]
[11.391328811645508, 4.163058280944824]
c1f21150-30d7-4bea-968c-dc434a0cffc5
motr-end-to-end-multiple-object-tracking-with
2105.03247
null
https://arxiv.org/abs/2105.03247v4
https://arxiv.org/pdf/2105.03247v4.pdf
MOTR: End-to-End Multiple-Object Tracking with Transformer
Temporal modeling of objects is a key challenge in multiple object tracking (MOT). Existing methods track by associating detections through motion-based and appearance-based similarity heuristics. The post-processing nature of association prevents end-to-end exploitation of temporal variations in video sequence. In this paper, we propose MOTR, which extends DETR and introduces track query to model the tracked instances in the entire video. Track query is transferred and updated frame-by-frame to perform iterative prediction over time. We propose tracklet-aware label assignment to train track queries and newborn object queries. We further propose temporal aggregation network and collective average loss to enhance temporal relation modeling. Experimental results on DanceTrack show that MOTR significantly outperforms state-of-the-art method, ByteTrack by 6.5% on HOTA metric. On MOT17, MOTR outperforms our concurrent works, TrackFormer and TransTrack, on association performance. MOTR can serve as a stronger baseline for future research on temporal modeling and Transformer-based trackers. Code is available at https://github.com/megvii-research/MOTR.
['Xiangyu Zhang', 'Tiancai Wang', 'Yuang Zhang', 'Yichen Wei', 'Bin Dong', 'Fangao Zeng']
2021-05-07
null
null
null
null
['multiple-object-tracking-with-transformer']
['computer-vision']
[-1.48110792e-01 -4.81398493e-01 -7.22358108e-01 -3.64785314e-01 -9.70213234e-01 -5.29385626e-01 4.25077885e-01 5.52266315e-02 -4.49604124e-01 5.71151972e-01 3.85258384e-02 1.79855540e-01 -1.73324928e-01 -2.39670932e-01 -8.09683800e-01 -5.33021867e-01 -4.20453072e-01 6.79529607e-01 9.56319869e-01 2.32860178e-01 -5.15299439e-02 3.11817855e-01 -1.61401689e+00 3.80865663e-01 2.80112386e-01 1.09807992e+00 4.76753592e-01 9.02116537e-01 1.97560668e-01 1.29076529e+00 -3.20644200e-01 -1.82380825e-01 3.11542839e-01 -2.30276749e-01 -4.22316313e-01 -1.55413866e-01 1.08372188e+00 -1.36474907e-01 -7.52494752e-01 6.66071177e-01 6.40021205e-01 3.81884396e-01 1.28512591e-01 -1.56256926e+00 -4.16226119e-01 8.69886696e-01 -9.56505299e-01 9.01684701e-01 1.70002982e-01 2.12709963e-01 8.22856426e-01 -9.89308894e-01 6.91747725e-01 1.30763793e+00 1.02677417e+00 7.20257521e-01 -1.10270357e+00 -1.02268839e+00 4.84617084e-01 6.29710436e-01 -1.62192464e+00 -6.36406541e-01 1.49877563e-01 -3.09747964e-01 7.63563275e-01 5.92894435e-01 7.45152056e-01 8.18015337e-01 2.14061245e-01 1.24196672e+00 7.45925486e-01 -1.34761229e-01 -2.38844216e-01 -2.83189982e-01 2.33119145e-01 7.13196158e-01 2.96041705e-02 2.91256845e-01 -1.19370043e+00 -1.74867079e-01 5.70768416e-01 7.31101632e-02 -9.55634937e-02 -2.75548100e-01 -1.43794155e+00 4.11830872e-01 3.17471176e-01 2.08183095e-01 -5.54945767e-01 7.95605838e-01 3.73509377e-01 1.56598061e-01 4.30445403e-01 -6.36553243e-02 -3.54164153e-01 -4.91507202e-01 -1.29692459e+00 4.11713988e-01 3.02204162e-01 1.33808482e+00 4.65730488e-01 -1.29613280e-01 -9.41149354e-01 5.51493526e-01 5.15363574e-01 6.54852808e-01 3.57889116e-01 -1.32681298e+00 1.18443854e-01 2.01962262e-01 9.53953862e-02 -6.13844514e-01 -2.49163032e-01 -5.01487792e-01 -1.85084239e-01 -8.88887867e-02 3.20482731e-01 5.53435870e-02 -8.13603818e-01 1.62634552e+00 8.22374284e-01 1.05253899e+00 -3.97933960e-01 9.60294664e-01 7.46489346e-01 3.20430070e-01 2.77320147e-01 -3.55790049e-01 1.48193073e+00 -1.32145762e+00 -9.23026323e-01 1.28927017e-02 6.82774663e-01 -9.10368919e-01 2.65729755e-01 1.64251015e-01 -1.42060804e+00 -7.20940590e-01 -6.07003629e-01 2.54444808e-01 1.29529759e-01 1.73966914e-01 5.41867375e-01 3.73772949e-01 -1.15589178e+00 4.80811417e-01 -1.44709289e+00 -7.66093433e-01 5.43521643e-01 5.97598195e-01 5.56719899e-02 -4.01550792e-02 -6.72294974e-01 7.22629488e-01 4.01785672e-01 1.30220996e-02 -1.16407895e+00 -1.05887675e+00 -5.11721611e-01 -4.09579754e-01 6.04401171e-01 -7.71086574e-01 1.66840577e+00 -3.66875589e-01 -8.89892697e-01 6.54600322e-01 -6.57554090e-01 -9.23610032e-01 6.39589489e-01 -6.04386687e-01 -5.83912969e-01 -6.73306212e-02 2.46127516e-01 8.46086979e-01 5.39708495e-01 -1.01184893e+00 -1.32305253e+00 -6.63420632e-02 -2.14490876e-01 1.94225207e-01 -3.15012187e-01 3.88525754e-01 -1.17522132e+00 -6.48932278e-01 -4.87917624e-02 -1.28648937e+00 -8.73234347e-02 2.65370905e-01 -6.83553144e-02 -4.70313638e-01 1.16721976e+00 -2.58330047e-01 1.69553971e+00 -1.98163819e+00 -1.62842318e-01 -1.23629197e-01 3.61612201e-01 1.43662199e-01 -1.14400566e-01 1.53614163e-01 3.30591649e-01 -4.82645184e-01 4.86908555e-01 -7.49269843e-01 -1.50934113e-02 2.13730395e-01 -7.33241737e-02 7.90110111e-01 -1.67336568e-01 1.02042198e+00 -1.13825047e+00 -1.06275010e+00 2.86424845e-01 4.74742532e-01 -4.30106521e-01 1.34966061e-01 -3.96341473e-01 5.55750549e-01 -5.36834896e-01 8.72318804e-01 5.70823491e-01 -5.06557643e-01 1.50282318e-02 -4.98772323e-01 -4.10238475e-01 1.92181095e-01 -1.12531066e+00 2.00137615e+00 -2.47728229e-02 8.88412356e-01 -1.25643611e-01 -4.29307759e-01 5.19021451e-01 3.41243446e-01 1.05200827e+00 -6.92089915e-01 6.47142250e-03 -1.06495321e-01 -1.65912381e-04 -2.58865684e-01 1.03150237e+00 5.35604239e-01 2.97000110e-01 3.37238044e-01 -7.44306818e-02 1.02178109e+00 5.15692055e-01 4.65010971e-01 1.47119045e+00 5.70905209e-01 -1.67665467e-01 -7.33255669e-02 2.46735290e-01 2.21195206e-01 9.31916356e-01 1.12774742e+00 -5.04527569e-01 4.36489522e-01 -5.93185961e-01 -5.56805313e-01 -5.46854198e-01 -1.15483725e+00 -1.20456204e-01 1.67097390e+00 5.55163085e-01 -6.44244909e-01 -1.84824020e-01 -5.73703110e-01 1.08386561e-01 4.52540129e-01 -8.11998725e-01 1.35154918e-01 -1.12643647e+00 -7.72164226e-01 5.99585354e-01 8.65931988e-01 1.65953979e-01 -9.03853714e-01 -8.26634288e-01 5.44871151e-01 -5.01582503e-01 -1.17977166e+00 -1.02672625e+00 3.27575058e-02 -7.41267145e-01 -9.70289826e-01 -6.22896016e-01 -4.28441316e-01 2.30083331e-01 6.47122800e-01 1.26115668e+00 3.14595968e-01 -2.89216369e-01 7.17802227e-01 -4.30429012e-01 -4.91224676e-01 -1.34113252e-01 2.24130210e-02 3.11403304e-01 3.86134312e-02 5.83595216e-01 -3.78836125e-01 -8.44503760e-01 8.43145072e-01 -3.54611933e-01 -6.29148781e-02 3.65138173e-01 3.85557294e-01 9.57337499e-01 -3.93023312e-01 2.28548154e-01 -2.78787494e-01 -2.38935128e-01 -5.04234970e-01 -6.82569683e-01 5.83012521e-01 -3.82464528e-01 -2.16844648e-01 -3.88868123e-01 -9.39611793e-01 -8.28640282e-01 1.84923813e-01 4.51383501e-01 -1.04376352e+00 1.65555432e-01 -1.06008865e-01 4.30526406e-01 -1.42028913e-01 4.77446258e-01 2.60961801e-01 -3.88713837e-01 -4.84651923e-01 2.66703337e-01 1.16341248e-01 8.44721735e-01 -6.16549492e-01 8.52334440e-01 7.20007837e-01 -1.07395008e-01 -3.95111620e-01 -1.01244617e+00 -1.12321973e+00 -3.72889310e-01 -8.29649091e-01 1.00780523e+00 -1.35198665e+00 -9.76164103e-01 7.45632052e-02 -1.06408465e+00 -4.77701008e-01 -2.87009597e-01 8.07212591e-01 -4.81394202e-01 5.45035256e-03 -5.58215737e-01 -9.74708378e-01 -3.29910934e-01 -8.32694411e-01 1.23660111e+00 3.40215534e-01 -3.69396925e-01 -8.17496657e-01 2.52015322e-01 3.50705475e-01 2.45435953e-01 1.45230189e-01 -4.51897264e-01 -5.13591349e-01 -1.33710349e+00 2.17876099e-02 -6.14531152e-02 -5.97826004e-01 4.47197855e-02 -4.96818982e-02 -8.84951115e-01 -6.31101787e-01 -5.02192080e-01 -2.35960990e-01 1.05592096e+00 8.35953414e-01 9.32301223e-01 -7.57715553e-02 -1.14609981e+00 5.66316307e-01 1.21230686e+00 2.21932381e-01 2.69008666e-01 5.41466713e-01 8.63839447e-01 3.05991545e-02 1.37572646e+00 7.28969812e-01 7.14053750e-01 1.34488201e+00 3.27378899e-01 1.90478489e-01 -5.69079280e-01 9.53257363e-03 5.56925356e-01 6.46567285e-01 -1.87229261e-01 -5.50574720e-01 -7.70650685e-01 8.62728059e-01 -2.45230246e+00 -1.38510180e+00 -4.70396549e-01 2.09587765e+00 6.20177805e-01 9.24010724e-02 7.23241329e-01 -4.20544714e-01 9.20554161e-01 -9.05545428e-02 -4.61332381e-01 5.03251851e-01 -9.81774926e-02 -3.34886581e-01 7.80878067e-01 2.71330923e-01 -1.35964096e+00 9.60250139e-01 5.35960531e+00 1.05021381e+00 -7.98782945e-01 6.18741989e-01 -5.24077192e-02 -7.26332247e-01 4.02085274e-01 -6.78901523e-02 -1.55796349e+00 3.99319381e-01 9.96324420e-01 -1.76094785e-01 1.69524014e-01 4.34639305e-01 2.91596860e-01 -1.12967618e-01 -1.33484626e+00 1.02489424e+00 -4.83887531e-02 -1.51178527e+00 -4.42942470e-01 5.98974153e-02 6.29781187e-01 6.11860454e-01 1.26259789e-01 3.64954352e-01 5.95173240e-01 -3.81622642e-01 1.19051218e+00 6.86891556e-01 3.32579970e-01 -2.96497971e-01 3.27590823e-01 6.38703536e-03 -2.12129164e+00 3.52354459e-02 -2.09568292e-01 3.05424720e-01 5.59111178e-01 -1.45232240e-02 -7.78520644e-01 7.10290015e-01 1.25093305e+00 1.15532172e+00 -7.73078859e-01 1.71358991e+00 4.76064920e-01 7.95612097e-01 -7.51480222e-01 2.20604584e-01 1.13921084e-01 4.80795652e-01 8.04006517e-01 1.41262841e+00 3.99935722e-01 8.12180862e-02 7.26980746e-01 3.59808087e-01 1.27510533e-01 -2.80026853e-01 -1.53827772e-01 2.60058731e-01 9.91510868e-01 1.30624604e+00 -9.64257121e-01 -5.51355600e-01 -4.80026960e-01 5.95391870e-01 1.08856879e-01 3.84723581e-02 -1.47971272e+00 5.01556754e-01 6.06626987e-01 1.67246267e-01 8.76026213e-01 6.21076636e-02 2.55193889e-01 -8.77504826e-01 7.32961223e-02 -4.91680503e-01 8.75636220e-01 -6.08148873e-01 -1.14015472e+00 4.78412211e-01 2.19859719e-01 -1.75392699e+00 6.61745295e-02 4.49802913e-02 -4.99108762e-01 2.70467401e-01 -1.25264800e+00 -1.53111839e+00 -4.11719799e-01 6.11919761e-01 8.18655133e-01 -4.92594251e-03 2.81296372e-01 8.45570326e-01 -4.77110237e-01 9.19907033e-01 -1.43946316e-02 4.11070101e-02 9.74892855e-01 -1.08832324e+00 3.89224917e-01 7.71664798e-01 4.55895036e-01 3.53601396e-01 8.85675609e-01 -9.03913498e-01 -1.48120201e+00 -1.64062548e+00 6.61515594e-01 -9.66495037e-01 7.91802227e-01 -1.38903350e-01 -9.15319443e-01 1.02220869e+00 3.62247050e-01 5.46132505e-01 6.13228500e-01 4.42846306e-02 -8.38848129e-02 -1.44024000e-01 -7.44857907e-01 3.34777862e-01 1.39780223e+00 -1.50075490e-02 -2.28019416e-01 4.62116152e-01 8.38091075e-01 -7.59569168e-01 -1.18862593e+00 4.70591575e-01 9.36486125e-01 -6.83043182e-01 1.03657103e+00 -2.10859776e-01 -4.78196114e-01 -8.91440094e-01 -2.49705732e-01 -4.88972604e-01 -5.90415597e-01 -9.36623096e-01 -8.63191426e-01 1.47698140e+00 2.45514750e-01 2.28976160e-02 1.07359755e+00 2.54104644e-01 -2.54142374e-01 -6.33309543e-01 -9.34452236e-01 -1.30357337e+00 -6.54838502e-01 -4.82604444e-01 4.15276945e-01 8.61277819e-01 -5.34642518e-01 -5.71154468e-02 -7.27920890e-01 4.42011833e-01 1.13274109e+00 1.42024890e-01 8.29479873e-01 -9.71849918e-01 -4.14599955e-01 -2.02880695e-01 -3.86373252e-01 -1.21277916e+00 -1.90669924e-01 -7.44363308e-01 2.22438544e-01 -1.27135992e+00 3.71976167e-01 -7.08723128e-01 -6.65811300e-01 6.19961143e-01 -1.36915892e-01 6.38698041e-01 5.80513239e-01 4.53612059e-01 -1.84729874e+00 4.61999327e-01 9.80696559e-01 -2.27936387e-01 -1.41565517e-01 6.15751930e-02 3.91752161e-02 5.06742418e-01 6.28504097e-01 -1.08071470e+00 -9.26666856e-02 -5.73671818e-01 -2.09360331e-01 1.97585955e-01 5.32002509e-01 -1.46885514e+00 8.44133854e-01 -6.31281585e-02 3.14306051e-01 -1.46685183e+00 7.06217647e-01 -7.69167483e-01 6.25114501e-01 5.91438472e-01 -4.04970288e-01 5.82371712e-01 4.65049267e-01 7.72750020e-01 1.24447994e-01 2.82817781e-01 5.50207794e-01 2.35489801e-01 -1.05256724e+00 7.57793784e-01 -2.11940005e-01 8.33653733e-02 1.16035581e+00 -4.50286061e-01 -2.48704076e-01 -8.05246532e-02 -7.44724035e-01 9.04813290e-01 3.42190474e-01 7.07869470e-01 4.04203862e-01 -1.65244508e+00 -8.94366086e-01 -6.26955092e-01 2.81022489e-01 -3.66851926e-01 3.91404390e-01 1.48067737e+00 -1.80680510e-02 2.91565716e-01 1.22987613e-01 -1.29030943e+00 -2.00742936e+00 5.42446077e-01 1.63523868e-01 -2.47056708e-01 -6.81803823e-01 1.06528544e+00 3.06033045e-01 2.55750984e-01 6.15567982e-01 -1.16522960e-01 -1.20823821e-02 -1.64237887e-01 8.38770688e-01 5.90094388e-01 -3.82297099e-01 -7.99434900e-01 -7.09870458e-01 5.86844683e-01 -3.48567009e-01 -4.03261818e-02 1.06994545e+00 -4.63295937e-01 2.59842247e-01 4.79384899e-01 6.77997649e-01 -2.27568850e-01 -1.46481061e+00 -5.49993396e-01 1.80723742e-01 -5.71247935e-01 3.34946999e-05 -6.79427564e-01 -1.12246954e+00 2.03117251e-01 1.18532693e+00 -1.46214798e-01 9.67020273e-01 3.92242342e-01 8.04895580e-01 2.99789190e-01 5.01502633e-01 -7.23269045e-01 8.41968283e-02 2.20402315e-01 4.87140745e-01 -1.19765282e+00 2.10307583e-01 -9.24072117e-02 -2.96921402e-01 6.04928792e-01 1.05143940e+00 2.24159911e-01 5.05847991e-01 5.03064156e-01 8.72721970e-02 -2.96703309e-01 -1.26281786e+00 -3.67377967e-01 4.89674687e-01 5.16491354e-01 5.74954152e-01 -1.96491271e-01 -3.85794267e-02 -1.39419511e-01 1.43733621e-01 1.66143835e-01 1.24464417e-02 1.17527294e+00 -4.14941072e-01 -1.14916408e+00 -7.57037818e-01 3.79940748e-01 -6.60772443e-01 9.19658169e-02 1.70995995e-01 7.42165327e-01 3.39584976e-01 9.75493848e-01 2.24262357e-01 -5.12260497e-01 1.05038948e-01 -3.91357005e-01 7.99927711e-01 -3.41306329e-01 -8.69123936e-01 3.97262037e-01 2.01549515e-01 -7.44523168e-01 -1.05700707e+00 -1.22566390e+00 -1.31072688e+00 -3.04804802e-01 -6.93133235e-01 2.00374767e-01 1.76702783e-01 5.99780142e-01 4.27356184e-01 8.13913167e-01 1.37097195e-01 -9.36524749e-01 3.87722030e-02 -9.76929903e-01 -8.86833221e-02 2.35149205e-01 4.87482041e-01 -1.00183904e+00 3.74567568e-01 3.80851150e-01]
[6.2995219230651855, -2.0284571647644043]
e651ed6e-a046-44d2-943b-97a20f7ba5dc
automatic-identification-and-classification
2203.05840
null
https://arxiv.org/abs/2203.05840v1
https://arxiv.org/pdf/2203.05840v1.pdf
Automatic Identification and Classification of Bragging in Social Media
Bragging is a speech act employed with the goal of constructing a favorable self-image through positive statements about oneself. It is widespread in daily communication and especially popular in social media, where users aim to build a positive image of their persona directly or indirectly. In this paper, we present the first large scale study of bragging in computational linguistics, building on previous research in linguistics and pragmatics. To facilitate this, we introduce a new publicly available data set of tweets annotated for bragging and their types. We empirically evaluate different transformer-based models injected with linguistic information in (a) binary bragging classification, i.e., if tweets contain bragging statements or not; and (b) multi-class bragging type prediction including not bragging. Our results show that our models can predict bragging with macro F1 up to 72.42 and 35.95 in the binary and multi-class classification tasks respectively. Finally, we present an extensive linguistic and error analysis of bragging prediction to guide future research on this topic.
['Nikolaos Aletras', 'A. Seza Doğruöz', 'Daniel Preoţiuc-Pietro', 'Mali Jin']
2022-03-11
null
https://aclanthology.org/2022.acl-long.273
https://aclanthology.org/2022.acl-long.273.pdf
acl-2022-5
['type-prediction']
['computer-code']
[ 2.40185544e-01 6.53189898e-01 -6.35788083e-01 -7.70396948e-01 -6.74144804e-01 -1.09921440e-01 1.07089531e+00 5.59943557e-01 -5.35763383e-01 6.15305126e-01 9.69379902e-01 -2.03091785e-01 8.66971612e-02 -7.24841595e-01 -3.36013108e-01 -4.33639884e-01 2.53323615e-01 6.05524004e-01 -1.06456362e-01 -5.66640913e-01 3.52562368e-01 1.87837856e-03 -1.45537353e+00 7.25378633e-01 5.57490170e-01 8.31915855e-01 1.90728992e-01 3.73216659e-01 -1.16239890e-01 1.29710996e+00 -5.06104589e-01 -8.27147782e-01 -2.88807422e-01 -9.35004056e-02 -1.07549608e+00 2.25466207e-01 2.94815212e-01 -1.23629048e-02 2.98248708e-01 9.47562099e-01 3.80879283e-01 -6.33250847e-02 5.85651815e-01 -1.22770262e+00 -6.60291076e-01 1.18201947e+00 -4.47841436e-01 1.45490333e-01 5.77800989e-01 -5.72669059e-02 1.56419361e+00 -8.03520441e-01 6.54378831e-01 1.63795817e+00 8.04165184e-01 5.76325655e-01 -1.11684465e+00 -7.05401123e-01 -4.12516668e-02 4.23519574e-02 -1.12389815e+00 -7.94124722e-01 9.02397156e-01 -4.73925889e-01 1.05929756e+00 5.62224686e-01 9.30751026e-01 1.07386422e+00 1.03700250e-01 1.02722669e+00 1.52265084e+00 -6.27396882e-01 -1.48050338e-01 6.37920141e-01 3.32167357e-01 6.53458238e-01 -2.99923688e-01 -6.24952801e-02 -8.31289530e-01 -5.48469089e-02 7.34562725e-02 -4.03865427e-01 2.02470198e-01 4.73991215e-01 -1.33416522e+00 1.06111169e+00 2.61220127e-01 7.27826118e-01 -6.21621311e-01 -1.12953126e-01 2.45711774e-01 1.73355535e-01 1.14901686e+00 3.39938223e-01 -3.12282890e-01 -3.74524891e-01 -8.80355835e-01 4.07740265e-01 7.58254409e-01 5.26198685e-01 6.73388779e-01 -3.70379061e-01 -1.48382977e-01 1.25719845e+00 4.78520542e-01 6.12964272e-01 6.63886011e-01 -6.25482321e-01 2.31889293e-01 6.12845063e-01 1.40378609e-01 -1.16207910e+00 -7.70336807e-01 -3.87303561e-01 -6.25968695e-01 -2.32313439e-01 3.28924507e-01 1.57900706e-01 -2.06352338e-01 1.67176819e+00 2.17886433e-01 -2.79702306e-01 2.57772982e-01 6.14602447e-01 1.18437481e+00 5.56128383e-01 4.00197297e-01 -4.56750929e-01 1.61472201e+00 -9.62573409e-01 -1.16259193e+00 -6.01150215e-01 1.02294135e+00 -1.10231078e+00 1.52174020e+00 3.23244572e-01 -1.06160021e+00 -2.63379723e-01 -7.12175965e-01 -2.36331835e-01 -3.24200690e-01 1.97009757e-01 6.34869039e-01 8.65782082e-01 -1.11920512e+00 3.15690517e-01 -3.63533765e-01 -5.86878121e-01 4.09791201e-01 1.78461969e-01 -2.68715829e-01 4.61076945e-01 -1.29876637e+00 9.97516751e-01 5.59201166e-02 -4.31347966e-01 -1.02768846e-01 -6.82242632e-01 -8.79994094e-01 -2.92443335e-01 9.75320563e-02 -3.39148909e-01 1.41665709e+00 -1.15015435e+00 -1.20363522e+00 1.76456940e+00 -5.20506978e-01 -1.03173113e+00 2.26523802e-01 -2.35348210e-01 -5.16829312e-01 -2.66934484e-01 4.20309931e-01 8.92185688e-01 4.94743168e-01 -1.09149659e+00 -7.10043728e-01 -3.81039232e-01 2.87694514e-01 3.48591775e-01 -7.03002214e-01 4.50987607e-01 2.58714318e-01 -6.19474471e-01 8.59126300e-02 -7.37351596e-01 3.65155265e-02 -3.50004733e-01 -7.76849449e-01 -7.03977883e-01 6.64254844e-01 -6.33815169e-01 1.29668355e+00 -2.11815834e+00 -2.97829658e-01 1.85890198e-02 4.65108722e-01 8.18173364e-02 3.45944107e-01 4.78821307e-01 1.01192128e-02 4.46355164e-01 1.09127529e-01 -7.88660288e-01 2.73158401e-01 1.37973994e-01 -1.55932933e-01 5.13907671e-01 -5.50178885e-02 8.32897425e-01 -8.94898772e-01 -7.61080086e-01 2.82311469e-01 3.30567330e-01 -6.93663538e-01 8.28230157e-02 -5.30778877e-02 1.07547835e-01 -1.61394656e-01 4.89020586e-01 2.37008616e-01 -3.91350210e-01 2.05167770e-01 -2.92816043e-01 -4.11326826e-01 8.21100175e-01 -5.85999489e-01 9.56220746e-01 -6.69737160e-01 6.85925663e-01 9.09032077e-02 -9.86988306e-01 1.07042038e+00 1.85141250e-01 3.83520156e-01 -8.74635637e-01 1.79857895e-01 2.20131457e-01 -1.08804539e-01 -4.16471243e-01 8.86088848e-01 -9.73836482e-01 -2.47151867e-01 6.08508229e-01 -4.05688852e-01 -3.42355490e-01 1.31942466e-01 1.99485078e-01 6.50966287e-01 -3.29273522e-01 8.00521791e-01 -6.76988780e-01 6.73161983e-01 -1.53121695e-01 2.41782099e-01 5.02003312e-01 -3.65999937e-01 9.18106735e-02 4.73904997e-01 -4.93086278e-01 -7.72544801e-01 -5.37088156e-01 -3.65323305e-01 1.43767798e+00 -4.29114662e-02 -6.02981329e-01 -7.89265752e-01 -3.90723526e-01 -1.94806010e-01 1.21581316e+00 -7.74591267e-01 7.89724812e-02 -3.47702533e-01 -8.90661895e-01 4.86947805e-01 8.87613818e-02 6.26658440e-01 -1.24773049e+00 -3.95534098e-01 3.45204882e-02 -7.33611941e-01 -1.25752616e+00 -3.07604402e-01 -2.32840165e-01 -3.83884072e-01 -7.36020625e-01 -2.72641510e-01 -1.05780244e+00 3.52331042e-01 2.27709394e-02 1.26009583e+00 3.47719908e-01 3.90375704e-01 2.64669836e-01 -3.22082579e-01 -7.15733349e-01 -8.26411545e-01 -6.48363978e-02 1.55600816e-01 1.49892479e-01 6.81622267e-01 -2.80607790e-01 -1.40293792e-01 2.82254755e-01 -5.56186616e-01 7.07648098e-01 1.41206250e-01 7.15748668e-01 2.51950562e-01 -6.24689013e-02 6.70769095e-01 -1.13075137e+00 9.39006329e-01 -4.79678750e-01 2.00658768e-01 -1.40184671e-01 -6.66470528e-01 -4.15177494e-01 5.94936013e-02 -2.10023955e-01 -1.02132726e+00 -2.86895424e-01 -8.11766207e-01 4.98904765e-01 1.09556168e-02 5.51737726e-01 3.22191119e-01 2.78737217e-01 6.86609745e-01 -2.52275653e-02 1.34178013e-01 -2.44127378e-01 5.90411834e-02 1.17313755e+00 1.84146702e-01 -2.83800274e-01 1.70700058e-01 6.86637104e-01 -4.77176130e-01 -1.28797090e+00 -1.53414226e+00 -6.61224663e-01 -2.19373122e-01 -7.16399789e-01 8.48514318e-01 -7.97239125e-01 -9.13668513e-01 5.07236362e-01 -9.65239048e-01 -5.25593877e-01 -2.85827339e-01 3.02753836e-01 -5.99981606e-01 1.65627658e-01 -6.49668157e-01 -1.26585758e+00 -4.77940679e-01 -9.79556382e-01 1.22735596e+00 -1.07170492e-01 -9.88870323e-01 -1.33315325e+00 -1.83327660e-01 9.46537673e-01 2.16928765e-01 -1.30855264e-02 5.80814064e-01 -9.27493036e-01 3.84826601e-01 1.44415004e-02 -5.03260195e-02 2.18081445e-01 7.04196617e-02 -2.57642835e-01 -1.00952768e+00 5.37190735e-02 3.01261067e-01 -5.55820167e-01 5.00750482e-01 2.98545033e-01 7.19942629e-01 -7.24876404e-01 -2.41056949e-01 -2.51762658e-01 8.98359179e-01 -1.22435600e-01 6.23058021e-01 6.02496326e-01 1.59372434e-01 1.16372037e+00 9.83556151e-01 7.76774347e-01 8.84914041e-01 7.80938566e-01 3.01812023e-01 -2.92725414e-01 -1.41741499e-01 -4.42017645e-01 4.38544720e-01 9.00710225e-01 1.49277607e-02 -2.67940491e-01 -1.08287036e+00 6.95585430e-01 -1.46225250e+00 -1.19612050e+00 -3.53451580e-01 1.83541298e+00 1.09648252e+00 3.04784954e-01 3.14757735e-01 6.38853967e-01 8.12890232e-01 4.47361797e-01 2.86349952e-01 -4.32475924e-01 -4.27881539e-01 -5.02177738e-02 1.84410527e-01 1.02210641e+00 -1.19695449e+00 1.29912472e+00 5.80223799e+00 8.78283203e-01 -9.08866525e-01 3.64816815e-01 1.07937539e+00 1.21416450e-01 -5.02275586e-01 -2.28439733e-01 -9.40334678e-01 2.34490767e-01 9.92962122e-01 -1.97806761e-01 1.44178018e-01 6.12784624e-01 6.40035331e-01 -2.18459576e-01 -7.54073679e-01 9.68093455e-01 1.57245815e-01 -1.40473306e+00 -3.26064348e-01 1.31670162e-01 5.14665663e-01 -3.34393680e-02 6.00696765e-02 2.74702609e-01 2.24458903e-01 -9.65287685e-01 1.31194925e+00 2.49373630e-01 4.00440097e-01 -3.93569142e-01 8.32716584e-01 6.77216232e-01 -6.60881341e-01 1.77782010e-02 6.47951886e-02 -6.67272806e-01 4.10896659e-01 7.91147888e-01 -1.17276371e+00 -2.37360179e-01 6.89919114e-01 1.19321597e+00 -3.89679611e-01 1.40677765e-01 -2.16531426e-01 9.04801130e-01 -3.32140505e-01 -6.94658220e-01 4.36794788e-01 2.89686793e-03 6.17288709e-01 1.46689582e+00 4.61474545e-02 1.34707913e-01 1.33315846e-01 6.63411975e-01 -1.72281433e-02 6.52906597e-01 -3.55994105e-01 -2.15245754e-01 3.31748247e-01 1.28889179e+00 -6.38301075e-01 -4.46857095e-01 -2.61505932e-01 4.88308877e-01 4.29250032e-01 -2.97329336e-01 -6.23586833e-01 3.93285751e-01 3.50218654e-01 7.13446438e-01 -3.68814796e-01 1.36285931e-01 -5.77317357e-01 -7.72135437e-01 -2.48089895e-01 -6.34989202e-01 1.78620711e-01 -8.93787146e-01 -1.33344805e+00 4.10806030e-01 -1.20566286e-01 -6.76218569e-01 -2.24342689e-01 -4.38314795e-01 -3.40559334e-01 5.31666577e-01 -1.46811604e+00 -1.40643024e+00 -2.73660515e-02 3.14041346e-01 5.36242723e-01 -3.71250100e-02 8.37219656e-01 3.37900996e-01 -1.91479281e-01 2.86599219e-01 -5.68260849e-01 -7.96558559e-02 6.83883667e-01 -1.23744226e+00 3.09303850e-02 2.73158789e-01 2.91279763e-01 5.24634182e-01 1.18195784e+00 -5.59993684e-01 -7.64186203e-01 -7.74246871e-01 2.14425230e+00 -1.98732972e-01 9.81675327e-01 -2.68460333e-01 -4.50907111e-01 7.94942200e-01 3.42767477e-01 -4.90607411e-01 1.01044071e+00 4.76833463e-01 -1.68308541e-02 9.12333801e-02 -1.56716490e+00 7.13420391e-01 9.28102374e-01 -4.02519017e-01 -7.16151714e-01 9.32459831e-01 6.23741746e-01 -2.49583676e-01 -9.30543780e-01 5.99994585e-02 4.84211951e-01 -1.28856647e+00 9.42479253e-01 -3.21804196e-01 7.46574461e-01 2.73335606e-01 -3.40209156e-01 -1.06154060e+00 -2.31463481e-02 -5.47204912e-01 5.94386697e-01 1.41096437e+00 5.42777300e-01 -7.91522563e-01 6.50879502e-01 4.51082855e-01 -2.46639431e-01 -8.56742859e-01 -1.04811394e+00 -1.44718811e-01 8.60136598e-02 -8.64593148e-01 5.87038517e-01 1.02584004e+00 4.79013562e-01 7.08226025e-01 -5.61241210e-01 -3.05202544e-01 3.19023252e-01 -1.05394378e-01 6.67321265e-01 -1.18939328e+00 6.53811693e-02 -5.29346764e-01 -2.68974125e-01 -1.01604950e+00 4.67601210e-01 -1.02222395e+00 6.13255054e-02 -1.46899068e+00 4.24066067e-01 -6.66300893e-01 2.74088413e-01 6.36203110e-01 1.84011057e-01 6.12848520e-01 1.21789044e-02 3.18045765e-01 -4.65024441e-01 4.70106870e-01 1.18409181e+00 -1.19077384e-01 -1.87030621e-02 -5.40244542e-02 -9.28426623e-01 1.20227802e+00 1.07415307e+00 -2.46634752e-01 -2.12473124e-01 7.69074187e-02 6.02962077e-01 -1.19629607e-01 4.52435613e-01 -6.02956116e-01 -2.79925406e-01 -2.51183212e-01 -3.47666174e-01 -5.71497262e-01 7.48147905e-01 -4.12720203e-01 -2.18272924e-01 4.14570361e-01 -8.34587872e-01 -1.21762067e-01 -1.51809976e-01 1.72364399e-01 -1.62172750e-01 -2.79645801e-01 9.72966909e-01 -4.88523282e-02 -4.91777539e-01 -3.07611436e-01 -6.51759744e-01 2.54646152e-01 8.48207355e-01 6.82135448e-02 -3.11523825e-01 -8.25815082e-01 -8.52241039e-01 -1.60656407e-01 1.68813244e-01 4.90817994e-01 5.61436713e-01 -1.37289643e+00 -8.95058274e-01 -1.76812634e-02 2.13725582e-01 -7.36579180e-01 6.88997889e-03 1.15245938e+00 -1.13022365e-01 4.82593477e-01 2.32563376e-01 -5.25465667e-01 -1.70373797e+00 9.20857787e-02 9.37882289e-02 -4.84529734e-01 -5.00160396e-01 9.18274045e-01 9.51929614e-02 -6.98335052e-01 -1.02368295e-01 -1.83222622e-01 -7.74505675e-01 2.69031852e-01 7.00661123e-01 1.47860378e-01 -1.26602709e-01 -1.57022071e+00 -3.33398342e-01 9.06641334e-02 6.89579919e-02 -3.48070174e-01 1.28857422e+00 -4.02291030e-01 -5.85939884e-01 8.61200452e-01 1.13377571e+00 3.46698254e-01 -2.54715204e-01 -3.01218748e-01 1.26584113e-01 -3.36715549e-01 2.95108467e-01 -7.78543174e-01 -6.49082839e-01 5.44614017e-01 1.87275827e-01 7.35951126e-01 6.87328458e-01 5.06536603e-01 7.65842140e-01 7.14020878e-02 2.52821892e-01 -1.28662479e+00 6.12382069e-02 3.89249265e-01 1.30622792e+00 -1.37152183e+00 -1.69288237e-02 -7.62521803e-01 -9.54915822e-01 6.78441882e-01 2.21422270e-01 2.41841495e-01 8.43488872e-01 1.30089447e-01 1.64035201e-01 -4.95599508e-01 -8.10552120e-01 -1.27498925e-01 1.29376993e-01 2.97181547e-01 1.10224867e+00 5.28645754e-01 -8.69348466e-01 8.68401408e-01 -1.00268769e+00 -1.92352727e-01 4.81361538e-01 5.10238290e-01 -6.08265221e-01 -1.01947689e+00 -2.62590051e-01 6.68884277e-01 -6.50168538e-01 -3.97903025e-01 -5.46160161e-01 6.73004270e-01 9.61004049e-02 1.45312238e+00 1.90163955e-01 -5.25647461e-01 1.20062992e-01 -1.80143435e-02 3.34596448e-02 -6.51648760e-01 -1.00570524e+00 -2.94971429e-02 1.08405638e+00 -2.60515392e-01 -1.19334185e+00 -1.04631114e+00 -1.24509680e+00 -5.37054896e-01 -2.16013029e-01 1.89617917e-01 6.46732390e-01 1.36353028e+00 -1.45170242e-01 2.09379047e-01 3.50469202e-01 -5.08761704e-01 -3.31976324e-01 -1.28345490e+00 -5.92369080e-01 6.36463225e-01 8.82192627e-02 -5.91220558e-01 -5.05154967e-01 6.58951998e-02]
[9.075824737548828, 10.605148315429688]
5a361fb5-0f3b-4e32-aef4-d8bc5e4e2a8f
dirichlet-survival-process-scalable-inference
2212.05996
null
https://arxiv.org/abs/2212.05996v1
https://arxiv.org/pdf/2212.05996v1.pdf
Dirichlet-Survival Process: Scalable Inference of Topic-Dependent Diffusion Networks
Information spread on networks can be efficiently modeled by considering three features: documents' content, time of publication relative to other publications, and position of the spreader in the network. Most previous works model up to two of those jointly, or rely on heavily parametric approaches. Building on recent Dirichlet-Point processes literature, we introduce the Houston (Hidden Online User-Topic Network) model, that jointly considers all those features in a non-parametric unsupervised framework. It infers dynamic topic-dependent underlying diffusion networks in a continuous-time setting along with said topics. It is unsupervised; it considers an unlabeled stream of triplets shaped as \textit{(time of publication, information's content, spreading entity)} as input data. Online inference is conducted using a sequential Monte-Carlo algorithm that scales linearly with the size of the dataset. Our approach yields consequent improvements over existing baselines on both cluster recovery and subnetworks inference tasks.
['Sabine Loudcher', 'Julien Velcin', 'Gaël Poux-Médard']
2022-12-12
null
null
null
null
['point-processes']
['methodology']
[ 4.04034136e-03 1.38034612e-01 -5.60544312e-01 -1.30393282e-01 -7.78671026e-01 -7.36444533e-01 1.45481455e+00 3.47185016e-01 -3.18089545e-01 5.79899311e-01 4.38725471e-01 -2.98898667e-01 -3.93750966e-01 -8.85618269e-01 -6.27845943e-01 -8.15697551e-01 -3.02117229e-01 1.41037130e+00 4.36505347e-01 4.70649451e-01 4.35289174e-01 5.90452887e-02 -7.28556216e-01 -2.37309471e-01 4.20896620e-01 4.82154489e-01 -2.66526919e-02 7.95478523e-01 -2.49804020e-01 6.93586051e-01 -6.20464087e-01 -4.15824533e-01 -3.33360471e-02 -8.28194395e-02 -7.52224088e-01 1.35582179e-01 -3.41550887e-01 -3.16975385e-01 -5.07766664e-01 6.84624314e-01 1.19037982e-02 3.35576348e-02 1.38365912e+00 -1.40910828e+00 -2.29654506e-01 9.83375132e-01 -9.52127457e-01 5.21393478e-01 4.62724492e-02 -9.55558121e-02 1.16250956e+00 -4.57887203e-01 9.11752105e-01 1.31898904e+00 6.25864923e-01 -2.96522439e-01 -1.52492523e+00 -4.87920612e-01 2.67104119e-01 -2.15280503e-01 -1.28943062e+00 -1.61248013e-01 5.36187053e-01 -8.84635150e-01 6.41589999e-01 -1.68695688e-01 5.41182697e-01 1.49242568e+00 1.15407042e-01 6.55203223e-01 7.54896402e-01 -1.84791401e-01 6.59873247e-01 5.66618778e-02 1.59079611e-01 2.54551768e-01 2.90853709e-01 -3.37420642e-01 -7.11174846e-01 -8.69076431e-01 8.03979278e-01 2.81024843e-01 8.07789415e-02 -1.39390334e-01 -1.34594369e+00 1.02268815e+00 -1.38273641e-01 1.87565267e-01 -6.53836131e-01 3.16448301e-01 1.18000664e-01 1.07561648e-01 1.15685391e+00 -1.47933796e-01 -5.43730140e-01 -4.21638370e-01 -1.52374661e+00 4.10577893e-01 1.52581191e+00 1.02999520e+00 8.84611607e-01 -6.39111400e-01 -2.91388869e-01 4.47914243e-01 7.08910406e-01 4.10988212e-01 -4.74877283e-02 -8.93397570e-01 5.19293845e-01 2.66384542e-01 3.35161686e-01 -8.89442980e-01 -1.26028791e-01 -1.90061212e-01 -5.96196592e-01 -5.15522122e-01 3.52982968e-01 -7.40677953e-01 -7.31561542e-01 1.48900402e+00 4.79181558e-01 6.29577696e-01 -2.97280461e-01 3.03968936e-01 2.16518506e-01 9.23908114e-01 2.20633820e-01 -1.88757986e-01 1.19394350e+00 -7.56922185e-01 -5.92051506e-01 2.52317667e-01 1.92079470e-01 -6.02100968e-01 9.49583948e-02 2.92584509e-01 -9.69534516e-01 2.68840164e-01 -3.11311483e-01 1.03065662e-01 -5.67740262e-01 -1.20650567e-01 6.02350891e-01 6.22568011e-01 -1.35266054e+00 6.58588767e-01 -1.12930501e+00 -5.65761328e-01 5.10260284e-01 2.00901493e-01 2.45747641e-01 -1.11214876e-01 -1.03396642e+00 2.68234521e-01 1.54028073e-01 -2.38191396e-01 -1.26274419e+00 -1.00750315e+00 -2.70710707e-01 1.53469503e-01 4.83326107e-01 -9.30032730e-01 1.23482919e+00 -2.96478868e-01 -1.51911139e+00 5.00458956e-01 -5.57812631e-01 -5.12362957e-01 6.39177024e-01 5.64152841e-03 -3.38839442e-01 3.49262387e-01 3.87243807e-01 5.07012904e-01 1.07998943e+00 -1.18196094e+00 -6.85020387e-01 -3.89157295e-01 -2.67725676e-01 -4.37141769e-03 -2.86506653e-01 2.33495966e-01 -9.61888373e-01 -6.37851834e-01 -1.50312871e-01 -1.06030440e+00 -3.18082303e-01 -1.08009845e-01 -1.09210062e+00 -6.58947468e-01 8.52674425e-01 -6.59803867e-01 1.15295935e+00 -1.74723244e+00 2.28606850e-01 6.27656460e-01 5.17174780e-01 -4.78244364e-01 1.93691671e-01 9.71968651e-01 3.96273047e-01 4.82658029e-01 -2.88900793e-01 -8.32461953e-01 2.66624659e-01 2.56629325e-02 -1.52056545e-01 7.30316997e-01 -8.87547433e-02 6.78233624e-01 -7.59342790e-01 -6.27872527e-01 -3.38977158e-01 6.44553840e-01 -4.82892543e-01 -9.48696062e-02 -5.91462970e-01 4.71102923e-01 -8.12948883e-01 3.63215119e-01 4.85460162e-01 -7.74724305e-01 2.93185353e-01 3.67909700e-01 -2.35973313e-01 6.35014772e-02 -1.12889576e+00 1.45401239e+00 -1.45114258e-01 7.01165438e-01 3.00303340e-01 -6.20318651e-01 3.93415630e-01 6.80815279e-01 8.50381136e-01 4.56586897e-01 8.41816291e-02 -3.31898838e-01 -4.23634559e-01 -1.54287919e-01 3.23242605e-01 -1.94910802e-02 -4.07670178e-02 1.19972944e+00 1.31167561e-01 2.39119396e-01 3.82967919e-01 9.17609990e-01 1.50523543e+00 -2.90200919e-01 -2.18452103e-02 -2.49030426e-01 -1.89049289e-01 -2.77689174e-02 1.71189114e-01 1.27557230e+00 1.21630572e-01 2.40213260e-01 1.07989061e+00 3.61668795e-01 -1.26899445e+00 -1.27706647e+00 -1.90085799e-01 1.26912153e+00 -2.77824670e-01 -4.60103601e-01 -6.76482856e-01 -4.29579467e-01 1.37682602e-01 6.15465581e-01 -7.92429984e-01 5.01816511e-01 -1.27933294e-01 -8.86637688e-01 4.95150208e-01 1.61552817e-01 8.86090621e-02 -7.49714375e-01 2.14825720e-01 3.06039035e-01 -1.89456359e-01 -1.02761233e+00 -4.73148167e-01 -1.33013546e-01 -1.00061393e+00 -9.28641438e-01 -9.59532857e-01 -3.84685904e-01 6.04221940e-01 -3.74011397e-02 1.00979054e+00 -4.20648634e-01 -3.42264801e-01 6.96388364e-01 -9.11346450e-03 -4.06731427e-01 -2.44150087e-01 3.92869830e-01 -1.84555739e-01 2.15222389e-01 4.51103181e-01 -8.32200170e-01 -6.65287077e-01 -2.95466017e-02 -8.41203809e-01 -3.10241282e-01 4.85204548e-01 2.56304592e-01 2.15233743e-01 6.14285886e-01 3.94486457e-01 -1.44110370e+00 7.91565418e-01 -1.46182883e+00 -3.92223865e-01 4.10459340e-02 -5.12367785e-01 -2.01866571e-02 3.92987244e-02 -3.16802204e-01 -1.26744390e+00 -3.87970626e-01 5.76252878e-01 -2.77571499e-01 -5.39042830e-01 7.22174764e-01 1.03063010e-01 5.23461401e-01 9.20457989e-02 2.49672793e-02 -1.11820810e-01 -5.99766254e-01 5.61138332e-01 6.84169054e-01 1.40938208e-01 -6.31760478e-01 7.97049463e-01 1.05975616e+00 -1.34061232e-01 -1.03556263e+00 -7.45067477e-01 -1.05968153e+00 -1.01946354e+00 -2.09528673e-02 7.22099900e-01 -1.08670807e+00 -6.73649132e-01 5.13707578e-01 -1.14915037e+00 -4.62729245e-01 5.74457273e-02 5.79596877e-01 -3.38660747e-01 3.10319155e-01 -1.02597678e+00 -9.67498958e-01 -1.27281517e-01 -4.86965090e-01 1.19043708e+00 -2.69170702e-01 -2.70437062e-01 -1.84361923e+00 3.92601460e-01 2.16622651e-01 4.92210120e-01 1.28869951e-01 8.96738052e-01 -9.90555465e-01 -9.08433199e-01 -2.94172913e-01 -3.30724478e-01 -2.55446553e-01 -1.28684580e-01 3.80564660e-01 -6.06699049e-01 -8.22430998e-02 -1.44340590e-01 3.82924885e-01 1.07665229e+00 8.53963315e-01 8.86151195e-01 -6.88451350e-01 -1.09312737e+00 1.92763209e-01 1.24845183e+00 -2.17316911e-01 2.32603550e-01 5.34319840e-02 5.13981760e-01 6.24657452e-01 2.30421335e-03 1.03757799e+00 7.63106883e-01 2.07515255e-01 7.76599869e-02 -4.28562276e-02 4.45583135e-01 -2.33818278e-01 1.82836831e-01 6.84906304e-01 -1.26331314e-01 -6.95279896e-01 -1.06352770e+00 9.10819888e-01 -1.70362294e+00 -1.17718101e+00 -3.71992797e-01 1.88500488e+00 9.82420564e-01 3.13704498e-02 5.12818515e-01 -3.46178174e-01 9.94505405e-01 3.71366471e-01 -5.06144285e-01 3.84943038e-01 1.12681292e-01 -1.21975109e-01 7.12784350e-01 6.14072859e-01 -1.15875471e+00 7.63937175e-01 6.48376465e+00 6.89992726e-01 -4.78643805e-01 4.37835276e-01 5.72788417e-01 -2.24665269e-01 -3.39147002e-01 2.63898343e-01 -1.09847736e+00 7.59806454e-01 1.29887319e+00 -3.29741776e-01 3.53699565e-01 3.72049332e-01 4.69589204e-01 -3.64181936e-01 -1.11273825e+00 4.21390593e-01 2.65246313e-02 -1.18819380e+00 -8.01093131e-02 7.59716213e-01 1.13978338e+00 3.81798953e-01 2.08699629e-01 -3.56724113e-02 1.16270781e+00 -5.76130271e-01 4.43577856e-01 8.23219657e-01 3.98940235e-01 -5.89779377e-01 3.79731834e-01 7.35565364e-01 -7.85753310e-01 7.53462985e-02 3.93502004e-02 3.01348567e-01 6.63928390e-01 1.18090403e+00 -1.10199487e+00 3.52654427e-01 5.57179928e-01 1.08871794e+00 -2.62282282e-01 1.11941338e+00 -8.53085592e-02 1.40462172e+00 -7.95329988e-01 -3.87850851e-02 3.77687901e-01 -3.92145753e-01 8.77779424e-01 1.30850780e+00 3.31889570e-01 4.56909463e-03 2.71702409e-01 1.10024941e+00 -2.71546960e-01 -2.72855252e-01 -4.75496978e-01 -2.38017857e-01 8.31445038e-01 1.21839559e+00 -1.08834934e+00 -5.54956257e-01 -1.50464639e-01 7.51095176e-01 5.80276400e-02 9.32646871e-01 -5.92418313e-01 -1.00627899e-01 2.49555200e-01 5.03270745e-01 8.44305217e-01 -4.04104203e-01 -4.67824265e-02 -1.04024410e+00 -4.43959624e-01 -3.42605785e-02 2.93446362e-01 -3.53493840e-01 -1.72551346e+00 -8.59657954e-03 2.55640090e-01 -5.36387920e-01 -2.11421609e-01 -6.74868375e-03 -1.02243483e+00 8.76945674e-01 -1.25887465e+00 -1.14641416e+00 2.92015612e-01 5.83955884e-01 3.85601133e-01 -1.10619418e-01 4.08392727e-01 -7.19743688e-03 -7.12637186e-01 -9.83394831e-02 6.76620543e-01 6.69519976e-02 3.76778752e-01 -1.36144960e+00 5.86775839e-01 3.74521285e-01 1.05700277e-01 8.51412714e-01 6.17247224e-01 -1.04010761e+00 -1.00857902e+00 -1.23397911e+00 9.49363351e-01 -7.99566507e-01 1.31164289e+00 -4.95949388e-01 -5.49388468e-01 1.10808039e+00 3.79500628e-01 -5.72639704e-01 9.61841702e-01 6.45764530e-01 -1.13225944e-01 3.90099913e-01 -7.94073045e-01 4.59724545e-01 6.41404748e-01 -5.18155158e-01 -1.54896066e-01 8.21463108e-01 5.95304072e-01 2.93853343e-01 -1.22614658e+00 -3.96409214e-01 -1.08146039e-03 -4.19275075e-01 8.53760660e-01 -5.39036393e-01 4.09814984e-01 8.63025784e-02 4.27531809e-01 -1.19744575e+00 -4.61088359e-01 -1.00230658e+00 -5.15617430e-01 1.47911465e+00 6.21309042e-01 -5.31108737e-01 8.70215654e-01 4.95313138e-01 5.34524918e-01 -3.55146378e-01 -8.64986300e-01 -3.63567859e-01 7.89237842e-02 -1.39535859e-01 3.70453209e-01 9.31626320e-01 9.77709517e-02 5.49532354e-01 -1.55647069e-01 1.90846562e-01 1.15380526e+00 -1.19078904e-01 6.77676141e-01 -1.97419083e+00 -3.15476298e-01 -3.03283453e-01 1.20926991e-01 -1.23689008e+00 2.57629812e-01 -6.40542388e-01 3.42150070e-02 -1.62604082e+00 6.39996231e-01 -6.45965517e-01 1.10570878e-01 8.04411098e-02 9.94711965e-02 -3.42499107e-01 -4.53307986e-01 8.21749330e-01 -1.03051078e+00 4.23465699e-01 7.64061749e-01 1.11588515e-01 -1.05044872e-01 4.63521719e-01 -4.50278968e-01 7.66957521e-01 6.93781316e-01 -7.38917172e-01 -3.00989449e-01 -2.78073043e-01 1.71627834e-01 5.21933496e-01 4.17113841e-01 -5.09893894e-01 9.17408884e-01 1.16530895e-01 1.86585218e-01 -1.04337656e+00 5.34821033e-01 -5.78748524e-01 1.16858527e-01 -8.36469159e-02 -6.26367748e-01 -2.00274289e-01 -4.98242289e-01 1.72547448e+00 3.53499837e-02 -1.88758239e-01 1.55646548e-01 -2.29219958e-01 9.13187936e-02 7.36255288e-01 -9.28479850e-01 2.00505510e-01 1.05563533e+00 1.79104432e-01 -3.00182968e-01 -7.52471268e-01 -1.11241412e+00 2.91269094e-01 3.18864882e-01 -1.76513400e-02 -5.70309870e-02 -6.18126094e-01 -7.59892046e-01 -3.41855824e-01 -4.83113676e-01 1.20681569e-01 2.35579714e-01 1.15345848e+00 -9.19659734e-02 6.64987028e-01 7.07646132e-01 -6.31287932e-01 -7.83298135e-01 4.76952046e-01 -2.30626583e-01 -6.53384507e-01 -5.26849389e-01 7.08265841e-01 -1.76040381e-01 -3.05256873e-01 3.32912594e-01 1.05829448e-01 -2.07114682e-01 4.10000950e-01 2.47782350e-01 7.82920539e-01 -3.29324961e-01 -2.86946416e-01 6.99159056e-02 3.24714966e-02 -4.14726526e-01 -8.32798362e-01 1.62894595e+00 -3.47353429e-01 -4.65503335e-01 1.01216018e+00 1.19023299e+00 -1.01730384e-01 -1.40806735e+00 -6.63349807e-01 3.50750178e-01 -1.00455150e-01 1.54232726e-01 -5.85743070e-01 -8.13855231e-01 6.98560953e-01 -2.56023377e-01 6.75734222e-01 2.40271911e-01 5.61637998e-01 5.40077090e-01 8.29553157e-02 1.71166882e-01 -1.03157938e+00 6.18215231e-03 4.17121798e-01 2.32915610e-01 -9.12902832e-01 1.55377463e-01 -5.60420871e-01 -4.12754685e-01 7.70452917e-01 -6.33162260e-02 -1.30176663e-01 1.64341843e+00 2.62432396e-01 -4.88805205e-01 -6.93588018e-01 -1.15158176e+00 1.65687442e-01 -4.02933620e-02 4.20553744e-01 2.12713942e-01 4.15256284e-02 1.89204261e-01 3.54648888e-01 -7.33818412e-02 -9.76629332e-02 5.54564416e-01 8.77378345e-01 -4.90407348e-01 -8.20118845e-01 -2.27348626e-01 9.21743214e-01 -6.71352386e-01 -3.37861389e-01 -2.92874217e-01 4.16343421e-01 -3.35877746e-01 9.21367049e-01 5.55885613e-01 4.18832690e-01 -3.73298973e-01 -4.23259325e-02 -1.11157753e-01 -8.35723579e-01 -3.51141840e-01 4.41688210e-01 -1.80589594e-02 -1.76009294e-02 -4.01098192e-01 -1.25324821e+00 -7.89749861e-01 -6.00109458e-01 -3.48957270e-01 3.79932791e-01 7.77245283e-01 9.96288478e-01 5.39824367e-01 3.10812503e-01 6.58065617e-01 -9.95757163e-01 -3.98622513e-01 -1.26268160e+00 -1.10227203e+00 1.49032623e-01 1.16890945e-01 -5.65556347e-01 -6.49549127e-01 4.18019980e-01]
[7.1276140213012695, 5.219654560089111]
9ca694c9-9ac5-41db-96e3-2826acbbb791
direct-multi-view-multi-person-3d-pose
2111.04076
null
https://arxiv.org/abs/2111.04076v2
https://arxiv.org/pdf/2111.04076v2.pdf
Direct Multi-view Multi-person 3D Pose Estimation
We present Multi-view Pose transformer (MvP) for estimating multi-person 3D poses from multi-view images. Instead of estimating 3D joint locations from costly volumetric representation or reconstructing the per-person 3D pose from multiple detected 2D poses as in previous methods, MvP directly regresses the multi-person 3D poses in a clean and efficient way, without relying on intermediate tasks. Specifically, MvP represents skeleton joints as learnable query embeddings and let them progressively attend to and reason over the multi-view information from the input images to directly regress the actual 3D joint locations. To improve the accuracy of such a simple pipeline, MvP presents a hierarchical scheme to concisely represent query embeddings of multi-person skeleton joints and introduces an input-dependent query adaptation approach. Further, MvP designs a novel geometrically guided attention mechanism, called projective attention, to more precisely fuse the cross-view information for each joint. MvP also introduces a RayConv operation to integrate the view-dependent camera geometry into the feature representations for augmenting the projective attention. We show experimentally that our MvP model outperforms the state-of-the-art methods on several benchmarks while being much more efficient. Notably, it achieves 92.3% AP25 on the challenging Panoptic dataset, improving upon the previous best approach [36] by 9.8%. MvP is general and also extendable to recovering human mesh represented by the SMPL model, thus useful for modeling multi-person body shapes. Code and models are available at https://github.com/sail-sg/mvp.
['Jiashi Feng', 'Shuicheng Yan', 'Yujun Cai', 'Jianfeng Zhang', 'Tao Wang']
2021-11-07
direct-multi-view-multi-person-3d-pose-1
http://proceedings.neurips.cc/paper/2021/hash/6da9003b743b65f4c0ccd295cc484e57-Abstract.html
http://proceedings.neurips.cc/paper/2021/file/6da9003b743b65f4c0ccd295cc484e57-Paper.pdf
neurips-2021-12
['3d-pose-estimation', '3d-multi-person-pose-estimation']
['computer-vision', 'computer-vision']
[-2.85034567e-01 5.45189083e-02 3.28758918e-02 -2.48670965e-01 -1.21859825e+00 -5.90163887e-01 5.15531957e-01 -1.08205549e-01 -4.09805626e-01 2.75035977e-01 3.93115729e-01 4.00394499e-01 2.57281214e-01 -6.50622666e-01 -9.96445239e-01 -3.72159481e-01 2.28423148e-01 9.22047675e-01 1.61307618e-01 -1.12812862e-01 -6.31988347e-02 5.66225886e-01 -1.25765383e+00 4.90476377e-02 4.38384175e-01 7.03320444e-01 -4.96400520e-02 8.57571423e-01 3.75329167e-01 1.88473091e-01 -2.97926664e-01 -7.39766955e-01 2.77733207e-01 -1.24621086e-01 -6.99851692e-01 3.20604503e-01 9.64751840e-01 -5.58279157e-01 -4.25355256e-01 7.09476531e-01 6.11477137e-01 9.95914638e-02 6.80939615e-01 -9.85093772e-01 -5.77087760e-01 -1.14826597e-01 -8.78253698e-01 -6.42190650e-02 8.93634140e-01 1.49944022e-01 1.12328589e+00 -1.39103603e+00 6.01027668e-01 1.53619790e+00 7.33431280e-01 4.87041652e-01 -1.13489401e+00 -3.35399985e-01 2.94122368e-01 4.36338857e-02 -1.51910150e+00 -2.48622164e-01 8.44348192e-01 -5.48665166e-01 8.65379930e-01 3.24363053e-01 1.01733661e+00 1.17040920e+00 1.84368774e-01 8.17602694e-01 6.46905243e-01 -1.51999488e-01 -2.95259565e-01 -1.71983466e-01 -1.15470342e-01 1.13069916e+00 1.53366372e-01 -1.49953485e-01 -4.76575136e-01 -2.18427494e-01 1.19635522e+00 3.39797467e-01 -2.83262789e-01 -8.07427347e-01 -1.38456595e+00 7.46432126e-01 6.73620582e-01 -2.78367162e-01 -3.97357851e-01 4.19578582e-01 1.78825274e-01 -2.60457516e-01 6.01844370e-01 1.65805563e-01 -5.91128886e-01 5.66534884e-02 -6.39872193e-01 5.99650979e-01 4.76460010e-01 9.79318500e-01 6.45381868e-01 -2.77467310e-01 -1.42461866e-01 7.67022014e-01 5.32110035e-01 6.85252130e-01 -5.77969328e-02 -1.29402888e+00 7.30966806e-01 6.70956254e-01 1.69843376e-01 -9.48530197e-01 -4.31496441e-01 -4.68470305e-01 -6.03396118e-01 -8.51110667e-02 4.97521996e-01 7.77364895e-02 -8.88986111e-01 1.51368570e+00 6.82880461e-01 -1.28467664e-01 -4.45500821e-01 1.21841145e+00 8.26878488e-01 4.77248490e-01 -1.88089058e-01 3.18551689e-01 1.66528988e+00 -1.28826511e+00 -2.24330470e-01 -2.99174190e-01 1.25986144e-01 -5.30708134e-01 1.01760685e+00 4.01818216e-01 -1.44644237e+00 -7.17227519e-01 -7.86720872e-01 -6.36701047e-01 -1.20121799e-01 3.87628019e-01 4.63239461e-01 4.04060930e-01 -9.07042563e-01 3.21468234e-01 -1.07598770e+00 -4.46339905e-01 3.80740404e-01 3.68769437e-01 -6.22059226e-01 -2.65556544e-01 -7.76017368e-01 7.05066502e-01 -1.11496791e-01 1.65806115e-01 -9.14869070e-01 -8.66947234e-01 -1.20462751e+00 -1.02524057e-01 5.61601579e-01 -1.48350394e+00 1.05455542e+00 -4.47792262e-01 -1.41017032e+00 9.61229563e-01 -3.07126403e-01 2.35842615e-02 8.17595065e-01 -8.37789178e-01 1.39513850e-01 4.93245929e-01 3.06322545e-01 8.54555964e-01 8.50492656e-01 -1.32102191e+00 -9.52157453e-02 -7.79445410e-01 2.56726503e-01 5.22011399e-01 5.74894398e-02 -1.84521809e-01 -1.31569433e+00 -6.34668052e-01 3.88276249e-01 -1.17343283e+00 -2.93397427e-01 6.27861381e-01 -6.09624505e-01 -2.06887007e-01 3.37388933e-01 -8.76735330e-01 6.35389984e-01 -1.86167884e+00 7.69685030e-01 -5.74632734e-02 4.06775862e-01 -1.83270931e-01 -4.45847958e-02 3.76067281e-01 5.08925691e-02 -1.03279702e-01 -2.89529204e-01 -9.88275349e-01 1.50517309e-02 1.68223903e-01 8.17701370e-02 8.21423113e-01 2.60556310e-01 1.15110660e+00 -7.73015618e-01 -5.09065807e-01 5.45979917e-01 9.86738861e-01 -9.45286036e-01 3.00110549e-01 -1.00551844e-01 7.19247580e-01 -4.21073765e-01 8.03715348e-01 5.81415653e-01 -4.94954467e-01 -5.64273261e-02 -4.52292502e-01 1.59952670e-01 -1.25831272e-02 -1.06852376e+00 2.29229951e+00 -4.32818830e-01 5.69939055e-02 1.01352207e-01 -7.58226693e-01 5.19910753e-01 2.05432996e-01 7.45989978e-01 -2.38377362e-01 5.27116582e-02 -1.88398257e-01 -6.71000242e-01 -4.58383232e-01 3.95116806e-01 -2.27293726e-02 -3.20335209e-01 1.14664301e-01 2.59759128e-01 -1.09484799e-01 -1.94375440e-01 2.22459570e-01 6.31425500e-01 6.89072609e-01 2.53194541e-01 1.45792216e-01 7.08048284e-01 -3.20344627e-01 4.42309141e-01 3.02187502e-01 4.23998712e-03 1.21660292e+00 2.61673421e-01 -4.75201190e-01 -9.86130178e-01 -1.66233993e+00 1.23645261e-01 9.78626251e-01 6.63517267e-02 -6.97957039e-01 -5.97116590e-01 -7.51913130e-01 2.47854918e-01 2.77116120e-01 -7.12290704e-01 1.77064598e-01 -7.46434689e-01 -4.54431117e-01 2.58523345e-01 7.44390190e-01 3.51752847e-01 -4.42359775e-01 -5.16412258e-01 -1.50684658e-02 -5.48101902e-01 -1.22652888e+00 -8.91743362e-01 -4.72339869e-01 -8.04419637e-01 -1.23521757e+00 -1.13630331e+00 -3.50295931e-01 7.63560653e-01 2.78448671e-01 1.21183383e+00 6.32776022e-02 -3.91456455e-01 9.47235107e-01 -1.38188094e-01 -9.82364416e-02 1.35693282e-01 8.96908715e-02 8.34020898e-02 6.36968613e-02 2.26910785e-02 -5.90625525e-01 -9.44859505e-01 2.28428125e-01 -3.75443518e-01 1.08560085e-01 5.22718549e-01 6.68593824e-01 9.48565066e-01 -6.28732383e-01 -5.35324169e-03 -5.89797854e-01 -9.24765170e-02 -3.31923664e-01 -2.83237100e-01 8.71041119e-02 -2.86106206e-02 -1.57788489e-02 3.57892394e-01 -2.05607533e-01 -7.88754284e-01 3.68825644e-01 -4.37092662e-01 -7.11371958e-01 -1.89871445e-01 -1.01878112e-02 -3.84169668e-01 1.54446438e-01 3.28001559e-01 2.80687451e-01 2.32470091e-02 -8.53403330e-01 5.22032499e-01 1.40899867e-01 6.26601875e-01 -7.69304335e-01 9.09018874e-01 8.71880591e-01 -7.93467648e-03 -7.31474459e-01 -9.78125155e-01 -5.66848516e-01 -1.07149756e+00 -2.61364698e-01 1.22432017e+00 -1.33429301e+00 -7.97902465e-01 2.94618696e-01 -1.27874207e+00 1.68015510e-02 -1.09611869e-01 5.12417972e-01 -7.61990905e-01 5.92865825e-01 -7.63297260e-01 -6.26253843e-01 -3.38695705e-01 -1.23983967e+00 1.91610968e+00 -1.60076529e-01 -2.62342781e-01 -8.47410977e-01 9.77931619e-02 8.90191853e-01 -2.84505755e-01 5.02717972e-01 5.12450695e-01 -1.79299951e-01 -7.48990774e-01 -2.95870543e-01 -1.29709572e-01 7.91329592e-02 -1.37771383e-01 -2.61432350e-01 -9.31587696e-01 -5.55963814e-01 -2.62234420e-01 -1.96818158e-01 7.95036137e-01 4.49880630e-01 1.06910181e+00 -1.85969129e-01 -3.66115600e-01 9.44975257e-01 1.37415016e+00 -5.61762094e-01 3.27943414e-01 1.21424675e-01 1.22002542e+00 5.69771051e-01 3.91975015e-01 4.93393302e-01 1.00353980e+00 1.06371057e+00 5.62282026e-01 -3.72665860e-02 -1.90626398e-01 -5.32912314e-01 4.05491859e-01 7.90314615e-01 -5.47150970e-01 8.27360898e-02 -6.77455723e-01 3.36250842e-01 -1.65845609e+00 -7.06398547e-01 -1.88134834e-01 2.18828058e+00 3.60317230e-01 -3.48509252e-02 4.54431713e-01 -1.95671588e-01 3.38218838e-01 3.26309860e-01 -5.81872523e-01 6.20367639e-02 2.55967498e-01 1.32688761e-01 3.46965671e-01 6.86954439e-01 -1.10182154e+00 7.70785987e-01 5.43039799e+00 2.74085641e-01 -5.59759855e-01 3.48455250e-01 2.60139853e-01 -4.66534376e-01 -2.75362551e-01 -2.52764940e-01 -8.60630989e-01 1.73586547e-01 5.15823305e-01 5.04362822e-01 2.40301818e-01 7.31919408e-01 4.63667102e-02 4.50301245e-02 -1.21456385e+00 1.22177672e+00 3.96535993e-01 -1.04920375e+00 2.28957757e-01 2.72575796e-01 4.47172821e-01 4.24983762e-02 -4.06208970e-02 1.91639394e-01 -1.92462951e-02 -7.77994275e-01 9.54501808e-01 7.33993232e-01 8.24836493e-01 -7.57988930e-01 3.91414940e-01 3.57347965e-01 -1.45108855e+00 6.78666532e-02 -2.51868635e-01 -1.53043736e-02 3.98455769e-01 3.51069510e-01 -4.11346257e-01 9.87359822e-01 8.39793622e-01 8.46240997e-01 -7.18558073e-01 6.89555228e-01 -2.86225438e-01 9.28473994e-02 -3.78636301e-01 4.53534633e-01 -2.91273650e-02 -2.42553025e-01 6.31505966e-01 9.80719924e-01 2.37694800e-01 -5.49319871e-02 3.32746118e-01 8.83088648e-01 4.83291857e-02 3.00450865e-02 -3.79097074e-01 4.56905901e-01 1.21255212e-01 1.20070684e+00 -4.42569703e-01 -2.99027622e-01 -5.69952726e-01 1.54124069e+00 5.25745332e-01 3.45894188e-01 -1.09547555e+00 1.69626117e-01 7.98120975e-01 3.89671028e-01 5.09344161e-01 -5.37245572e-01 -1.29204914e-01 -1.51700592e+00 4.40110028e-01 -6.18252873e-01 4.21115279e-01 -9.75985229e-01 -1.22851872e+00 4.69577610e-01 1.84821084e-01 -1.16093040e+00 -3.08600843e-01 -6.16931617e-01 -2.51560777e-01 9.55642939e-01 -1.22273481e+00 -1.68520534e+00 -4.15971726e-01 6.77832961e-01 8.42665732e-01 2.75362492e-01 8.27332258e-01 1.45978898e-01 -5.47069252e-01 5.27472556e-01 -5.71208000e-01 3.17645967e-01 7.78607786e-01 -1.34067512e+00 7.08439648e-01 6.51272714e-01 2.50457704e-01 6.42412782e-01 4.10193115e-01 -5.45408547e-01 -1.70549297e+00 -9.93505657e-01 6.62597537e-01 -1.15712476e+00 2.65986353e-01 -6.01065159e-01 -5.60904622e-01 9.29939926e-01 -1.39366388e-01 2.70828485e-01 5.36248505e-01 6.41154498e-02 -5.92243433e-01 -1.85021516e-02 -9.36036050e-01 5.66238105e-01 1.36763918e+00 -5.59115112e-01 -6.54626608e-01 3.65278900e-01 7.54850984e-01 -7.71656334e-01 -1.10668194e+00 2.20903754e-01 7.62820959e-01 -1.04456925e+00 1.64319551e+00 -4.73722160e-01 3.91848236e-01 -3.13065946e-01 -3.04656714e-01 -1.02580774e+00 -3.37044626e-01 -2.43678808e-01 -4.04302597e-01 8.10582757e-01 8.27063471e-02 -4.31332916e-01 8.68207037e-01 5.38390577e-01 -1.48112416e-01 -1.04877830e+00 -9.13095593e-01 -3.60839069e-01 3.29403765e-02 -4.69356954e-01 3.77420485e-01 6.02609873e-01 -5.53297102e-01 3.76851857e-01 -4.68557686e-01 5.57212651e-01 9.84999180e-01 1.23226292e-01 1.07234919e+00 -1.19005799e+00 -6.33599401e-01 -7.67668784e-02 -3.89256239e-01 -1.52325439e+00 4.77905087e-02 -8.43802929e-01 -2.57447213e-01 -1.67942643e+00 3.76604587e-01 1.01116471e-01 8.55452195e-02 2.24735588e-01 -3.08183461e-01 5.39701164e-01 5.01485765e-01 1.89282060e-01 -5.30785620e-01 6.53390586e-01 1.54109097e+00 -6.94918185e-02 6.55936971e-02 1.09325372e-01 -5.57719827e-01 9.38685179e-01 3.52978170e-01 -2.69885778e-01 -2.78080285e-01 -6.00356936e-01 1.23087004e-01 2.46255055e-01 1.03736711e+00 -1.00012589e+00 6.38804808e-02 1.10230759e-01 8.87721479e-01 -9.52941597e-01 1.04122865e+00 -7.17610598e-01 1.66573405e-01 3.91661853e-01 8.39995500e-03 3.16775352e-01 -1.07532121e-01 8.40503514e-01 1.53583080e-01 2.87565619e-01 5.65684557e-01 -5.39278030e-01 -3.41869086e-01 7.45710373e-01 1.81012243e-01 9.64683890e-02 8.36400807e-01 -1.95245698e-01 1.53929397e-01 -4.07305628e-01 -1.11885369e+00 1.80298790e-01 7.45159686e-01 5.51442742e-01 8.36882412e-01 -1.37526977e+00 -7.30874002e-01 3.25074196e-01 1.25574514e-01 3.93353313e-01 4.62337703e-01 9.74165142e-01 -6.22426271e-01 3.23153049e-01 -3.54623720e-02 -9.51757491e-01 -1.20842993e+00 5.37122488e-01 4.66108710e-01 -2.62619764e-01 -9.79423046e-01 8.64677012e-01 4.28878546e-01 -6.88475370e-01 8.10689181e-02 -2.83187926e-01 8.93295035e-02 -1.22713715e-01 4.61058825e-01 3.71617466e-01 -7.67478496e-02 -9.41281855e-01 -5.24877667e-01 1.29114473e+00 3.01486645e-02 -2.48717159e-01 1.40442860e+00 -2.52900034e-01 3.57946828e-02 4.74196315e-01 1.43699002e+00 1.51012659e-01 -1.63032937e+00 -1.19916402e-01 -5.29835284e-01 -5.14546454e-01 -2.94640332e-01 -4.99023497e-01 -1.19078374e+00 9.90282118e-01 3.44648689e-01 -4.33706969e-01 8.01726103e-01 3.62396538e-01 7.95386672e-01 1.19479775e-01 4.67706680e-01 -7.23444283e-01 4.05365080e-01 2.13942483e-01 1.25016928e+00 -1.15695012e+00 3.98623377e-01 -7.23336697e-01 -5.31878710e-01 9.82988596e-01 7.04937637e-01 -3.88561785e-01 6.28903329e-01 -1.47363901e-01 -1.96382686e-01 -4.03645277e-01 -3.70648026e-01 -3.32538299e-02 6.74408197e-01 5.58973432e-01 4.11126733e-01 4.95567545e-02 7.11075962e-02 5.89062095e-01 -1.96052253e-01 -2.88519889e-01 1.52340576e-01 5.37912846e-01 -4.49396446e-02 -1.04335523e+00 -6.11736476e-01 1.24676526e-01 -3.65922987e-01 1.74188957e-01 -2.81702071e-01 9.16120708e-01 1.26084059e-01 4.89501506e-01 1.30535915e-01 -3.14259708e-01 6.74358189e-01 4.71721105e-02 7.57392704e-01 -7.88004279e-01 -3.71753067e-01 3.27308923e-01 2.82250866e-02 -9.97344613e-01 -4.63243186e-01 -8.48854244e-01 -1.01956034e+00 -1.32848799e-01 8.69242847e-02 -2.71986455e-01 3.85178268e-01 8.07222366e-01 3.51559907e-01 3.90897065e-01 3.09209406e-01 -1.39902735e+00 -4.37923372e-01 -7.97518373e-01 -2.14930236e-01 5.35912037e-01 4.28286403e-01 -9.68323469e-01 -4.12905067e-02 -4.67230864e-02]
[7.0486602783203125, -0.9362524151802063]
74a4e163-66f7-4d83-a373-453b65da98d1
skghoi-spatial-semantic-knowledge-graph-for
2303.04253
null
https://arxiv.org/abs/2303.04253v3
https://arxiv.org/pdf/2303.04253v3.pdf
TMHOI: Translational Model for Human-Object Interaction Detection
Detecting human-object interactions (HOIs) is an intricate challenge in the field of computer vision. Existing methods for HOI detection heavily rely on appearance-based features, but these may not fully capture all the essential characteristics necessary for accurate detection. To overcome these challenges, we propose an innovative graph-based approach called TMGHOI (Translational Model for Human-Object Interaction Detection). Our method effectively captures the sentiment representation of HOIs by integrating both spatial and semantic knowledge. By representing HOIs as a graph, where the interaction components serve as nodes and their spatial relationships as edges. To extract crucial spatial and semantic information, TMGHOI employs separate spatial and semantic encoders. Subsequently, these encodings are combined to construct a knowledge graph that effectively captures the sentiment representation of HOIs. Additionally, the ability to incorporate prior knowledge enhances the understanding of interactions, further boosting detection accuracy. We conducted extensive evaluations on the widely-used HICO-DET datasets to demonstrate the effectiveness of TMGHOI. Our approach outperformed existing state-of-the-art graph-based methods by a significant margin, showcasing its potential as a superior solution for HOI detection. We are confident that TMGHOI has the potential to significantly improve the accuracy and efficiency of HOI detection. Its integration of spatial and semantic knowledge, along with its computational efficiency and practicality, makes it a valuable tool for researchers and practitioners in the computer vision community. As with any research, we acknowledge the importance of further exploration and evaluation on various datasets to establish the generalizability and robustness of our proposed method.
['Shuteng Niu', 'Qing Tian', 'Acharya Kamal', 'Houbing Song', 'Alvaro Velasquez', 'Qizhen Lan', 'Lijing Zhu']
2023-03-07
null
null
null
null
['human-object-interaction-detection']
['computer-vision']
[ 2.33085513e-01 -1.61903828e-01 -2.20053792e-01 9.71969664e-02 -2.96391279e-01 -4.02866155e-01 3.14965218e-01 2.24693269e-01 -6.17158972e-02 7.60612488e-02 2.61401445e-01 -6.60073161e-02 -1.28923990e-02 -8.50815058e-01 -5.74035645e-01 -6.26590014e-01 -2.51257598e-01 -7.67390281e-02 5.60454667e-01 -1.41883090e-01 1.17760763e-01 6.06788635e-01 -1.72936940e+00 9.16913822e-02 8.07353199e-01 1.21756804e+00 -8.73581618e-02 5.53289235e-01 4.64295715e-01 8.56405556e-01 -4.69529659e-01 -2.04262465e-01 2.71944612e-01 -3.55228424e-01 -3.51315349e-01 1.89877465e-01 5.16009867e-01 -3.30690622e-01 -5.26672006e-01 6.79999113e-01 4.28368300e-01 -4.24890667e-02 6.62909746e-01 -1.43353140e+00 -5.31690717e-01 -1.30377144e-01 -8.16518784e-01 1.50067002e-01 4.77451324e-01 3.09355110e-01 1.23367524e+00 -9.13318098e-01 7.19474435e-01 9.15883124e-01 7.45228291e-01 2.69741099e-02 -8.70951116e-01 -6.77749217e-01 9.32528565e-05 4.02109891e-01 -1.57182848e+00 -1.83994263e-01 8.75639081e-01 -5.01825750e-01 1.00194585e+00 3.13191444e-01 1.03900599e+00 7.29315817e-01 9.78219416e-03 1.24086499e+00 7.38354981e-01 -6.13246918e-01 -2.66401079e-02 2.11870283e-01 1.44044966e-01 1.07982886e+00 5.94276428e-01 2.40272686e-01 -7.02626526e-01 -2.29725726e-02 7.67210543e-01 9.83122140e-02 -2.05321342e-01 -6.04229271e-01 -1.13352156e+00 8.87904108e-01 8.32744360e-01 6.26021549e-02 -2.92168677e-01 1.14717595e-01 2.82859325e-01 -3.65782768e-01 1.82306647e-01 1.72713473e-01 4.37190413e-01 2.67264694e-01 -7.55899131e-01 2.40624636e-01 4.72057700e-01 1.07346821e+00 5.95854402e-01 -2.26949051e-01 -2.77138382e-01 6.26260161e-01 3.02886546e-01 5.79885662e-01 -4.88420278e-02 -8.07859600e-01 2.75911778e-01 1.07098734e+00 8.65939260e-02 -1.61184525e+00 -5.03707767e-01 -6.63416445e-01 -5.88306248e-01 1.59717336e-01 1.57662764e-01 6.09784247e-03 -7.56207705e-01 1.40870893e+00 6.24265850e-01 1.86077669e-01 -3.20732176e-01 1.01950145e+00 8.47639143e-01 3.10028464e-01 -9.25438926e-02 2.30351105e-01 1.53033710e+00 -9.72249806e-01 -4.68930244e-01 -3.77347857e-01 7.09347546e-01 -7.17393696e-01 8.19176495e-01 1.35570630e-01 -7.31682062e-01 -2.98100621e-01 -1.22663379e+00 -1.24539584e-01 -3.21462333e-01 3.67187023e-01 9.08267677e-01 5.76296985e-01 -8.54688585e-01 -1.29865870e-01 -5.99602222e-01 -6.32551730e-01 5.70829272e-01 2.29519278e-01 -3.43005031e-01 -2.75084466e-01 -9.81517673e-01 6.40273333e-01 3.61537039e-01 2.81581849e-01 -7.63431549e-01 -3.40728760e-01 -1.08566618e+00 1.82288943e-03 5.36614418e-01 -8.67922008e-01 7.67447412e-01 -4.34273660e-01 -7.58715391e-01 7.01128423e-01 -4.09710974e-01 -3.94423157e-01 3.93799722e-01 -1.21621892e-01 -2.13428512e-01 5.02061963e-01 1.96660787e-01 6.50100112e-01 8.60876918e-01 -1.27551794e+00 -1.03726125e+00 -4.55033481e-01 2.33083278e-01 4.52198833e-01 -5.40942430e-01 3.03419586e-02 -8.53524923e-01 -4.48282123e-01 1.43956199e-01 -1.18646908e+00 1.50787517e-01 2.28774026e-01 -5.92351854e-01 -2.24027634e-01 9.79831398e-01 -6.06053770e-01 1.43664122e+00 -2.31052971e+00 5.62802590e-02 3.45743924e-01 6.74343228e-01 2.84288853e-01 -1.40267788e-02 5.39113939e-01 3.87834281e-01 -1.00275151e-01 -3.99034321e-02 -1.23371847e-01 -2.19977602e-01 -1.02761574e-01 1.64649650e-01 7.60886908e-01 2.38695040e-01 1.25385940e+00 -1.04290986e+00 -5.31194329e-01 5.64807594e-01 7.38760889e-01 -3.81475657e-01 1.28278241e-01 6.10281676e-02 1.49700224e-01 -5.17923236e-01 7.99742997e-01 3.28053594e-01 -5.01608253e-01 3.79723273e-02 -4.34280962e-01 3.39498557e-02 -6.81040064e-02 -1.10046554e+00 1.05069327e+00 -1.78705499e-01 7.80768692e-01 -6.32164255e-02 -9.88873065e-01 6.59048378e-01 1.30827099e-01 6.78400815e-01 -8.17253113e-01 1.92991599e-01 -1.06872022e-02 -8.64838660e-02 -4.52048331e-01 2.61723638e-01 2.17560843e-01 -2.76746005e-02 6.26685917e-02 -2.69386649e-01 6.98461607e-02 1.86214700e-01 5.20281613e-01 1.17539692e+00 -7.83603713e-02 4.17845100e-01 1.33449003e-01 2.67559201e-01 2.34571561e-01 3.28670800e-01 9.09115970e-01 -4.07742590e-01 6.03003502e-01 2.05088943e-01 -1.64022103e-01 -7.48893440e-01 -9.40003753e-01 4.42283005e-02 8.97447228e-01 4.32216078e-01 -4.99308735e-01 -6.70565486e-01 -7.35013783e-01 2.00281203e-01 2.84358263e-01 -7.44109273e-01 -2.43311137e-01 -3.71822327e-01 -5.17911017e-01 2.56641626e-01 8.28228831e-01 5.05608261e-01 -9.05112088e-01 -8.89193118e-01 7.70439673e-03 -3.40563446e-01 -1.35327470e+00 -5.18054664e-01 -1.70765683e-01 -4.13274854e-01 -1.28586698e+00 -6.09004915e-01 -7.68984616e-01 6.78157508e-01 1.09274852e+00 7.00661719e-01 3.79459471e-01 -8.46485674e-01 6.17852926e-01 -5.93016922e-01 -6.64905965e-01 1.86133847e-01 -5.66547178e-02 -2.51724064e-01 2.02809989e-01 4.16897416e-01 -2.43345261e-01 -1.05614877e+00 3.99117053e-01 -6.07354403e-01 2.36355320e-01 6.80835068e-01 7.00686455e-01 3.64718378e-01 3.79755646e-01 9.98282731e-02 -8.58244598e-01 3.04006368e-01 -4.35676694e-01 -4.97846276e-01 1.35511562e-01 -4.47094887e-01 -3.81901562e-01 2.76313096e-01 -1.45859793e-01 -9.84042704e-01 1.96067780e-01 3.74862373e-01 -2.73835003e-01 -9.03838202e-02 5.86390316e-01 4.63596778e-03 -4.64839578e-01 6.54416025e-01 4.29305062e-02 7.73736928e-03 -1.06752530e-01 4.45761800e-01 7.33954608e-01 6.32184148e-01 -1.10362232e-01 8.54393303e-01 9.18547034e-01 1.73449978e-01 -1.12432551e+00 -1.03109813e+00 -1.20732045e+00 -4.68481809e-01 -3.25999498e-01 9.71309304e-01 -1.03067207e+00 -7.65310049e-01 5.12512863e-01 -9.09134626e-01 1.20063638e-02 2.09994063e-01 3.55058372e-01 -2.37735406e-01 5.36510229e-01 -5.51403642e-01 -1.01648426e+00 -3.06188256e-01 -9.28368509e-01 1.41683650e+00 1.93504885e-01 -1.75840408e-01 -9.79319036e-01 -3.28213036e-01 8.70969772e-01 7.11348355e-02 4.78822887e-01 7.25666523e-01 -1.41075388e-01 -8.05348158e-01 -7.20844090e-01 -6.65808082e-01 7.91721269e-02 -2.10805610e-02 -3.14542465e-02 -8.07708621e-01 -3.07053030e-01 -3.60806376e-01 -1.36949435e-01 8.03448260e-01 4.05360222e-01 7.88249731e-01 7.93895870e-02 -5.88454485e-01 3.51555139e-01 1.37732017e+00 7.01683983e-02 5.41473269e-01 4.26397413e-01 1.22703457e+00 6.28849804e-01 8.73110116e-01 5.25405943e-01 7.32977092e-01 9.97704625e-01 5.36794364e-01 -5.60407817e-01 -5.21711111e-01 -2.40538701e-01 1.52466819e-01 3.95849943e-01 -9.84908268e-02 -2.37447381e-01 -1.07315183e+00 6.86166048e-01 -2.01236081e+00 -8.25674176e-01 -3.53260875e-01 2.22507477e+00 1.78019494e-01 5.13511747e-02 3.84858727e-01 3.13890070e-01 5.17436683e-01 4.25584428e-02 -2.75616556e-01 9.97405499e-02 5.83710782e-02 -1.78837150e-01 5.88159740e-01 1.65290117e-01 -1.32840681e+00 9.08264458e-01 6.03850174e+00 5.57892203e-01 -1.09768081e+00 -6.41773716e-02 1.47475705e-01 -9.67012569e-02 1.79886017e-02 -1.26944140e-01 -9.86538231e-01 1.73896909e-01 3.57152939e-01 6.57543913e-02 5.10543466e-01 7.67374635e-01 2.56818116e-01 -4.20455068e-01 -9.62009788e-01 9.92627859e-01 4.96583015e-01 -1.18602228e+00 -2.44551733e-01 2.80647695e-01 7.25879908e-01 3.59502137e-02 8.80843848e-02 3.85346562e-02 6.58418313e-02 -1.00242293e+00 7.00304329e-01 1.53676450e-01 6.88098192e-01 -5.89591920e-01 9.74815488e-01 1.91613719e-01 -1.60194504e+00 -2.44887277e-01 -9.64933187e-02 -1.35397553e-01 1.06211282e-01 3.67726654e-01 -1.06355155e+00 6.81746840e-01 6.07765198e-01 8.33874881e-01 -7.80083179e-01 1.15989220e+00 -3.20402354e-01 5.82159996e-01 -5.64852118e-01 -5.36920652e-02 1.39385760e-01 3.76654640e-02 4.18408632e-01 1.17748666e+00 9.30082276e-02 4.81468774e-02 5.69392145e-01 6.21297657e-01 1.31772518e-01 1.90265298e-01 -8.27111363e-01 -2.51414448e-01 5.28816402e-01 1.15779948e+00 -8.30017209e-01 -2.54524350e-01 -8.49813223e-01 9.56388950e-01 4.46537852e-01 4.48571891e-01 -9.16822016e-01 -4.81057316e-01 3.18393260e-01 3.46138179e-01 4.31213796e-01 -3.17619413e-01 -4.16820735e-01 -9.64519441e-01 2.79682517e-01 -7.41151810e-01 5.59920728e-01 -5.33536136e-01 -9.63814497e-01 2.21340671e-01 7.43714301e-03 -1.24800634e+00 1.36920316e-02 -6.43961966e-01 -4.40262645e-01 5.40650606e-01 -1.53199017e+00 -1.74479449e+00 -8.82618964e-01 4.93171155e-01 2.92042494e-01 2.29585409e-01 5.76396525e-01 1.78406507e-01 -7.82792330e-01 6.57674789e-01 -2.02639386e-01 3.52181464e-01 3.42033327e-01 -9.55290675e-01 1.76904365e-01 9.62799668e-01 4.31417078e-01 7.19579220e-01 5.76619923e-01 -6.77222490e-01 -1.66230941e+00 -1.16408610e+00 6.32065177e-01 -6.85820460e-01 5.17231762e-01 -4.42411631e-01 -6.52095914e-01 6.98998749e-01 -3.07059139e-01 2.30563655e-01 4.82751101e-01 1.57041356e-01 -3.67493659e-01 -1.88355576e-02 -7.46091545e-01 6.11303687e-01 1.29685962e+00 -6.11453235e-01 -3.35721314e-01 3.11625987e-01 2.89242744e-01 -2.99460381e-01 -6.65359437e-01 6.25358522e-01 8.01014066e-01 -1.14475119e+00 1.18080783e+00 7.04834685e-02 1.58049449e-01 -4.23149228e-01 9.99133363e-02 -8.94389033e-01 -4.86997217e-01 -8.11593160e-02 -3.36715251e-01 8.85728121e-01 5.93174957e-02 -5.30572653e-01 8.06707561e-01 3.94282311e-01 1.33809686e-01 -9.16802406e-01 -4.72576141e-01 -7.08846569e-01 -5.56573153e-01 -6.13679945e-01 8.49118680e-02 6.68821037e-01 1.53823107e-01 3.95213783e-01 -4.70125854e-01 4.84130204e-01 8.79581988e-01 1.02825470e-01 1.04164553e+00 -1.23027086e+00 -2.42800340e-01 -4.09266859e-01 -8.51051033e-01 -8.57210994e-01 -8.87824148e-02 -7.87554264e-01 -4.90872860e-02 -1.81887352e+00 5.62090039e-01 -3.61348689e-01 -2.72042960e-01 5.28415978e-01 -3.73928905e-01 8.31591666e-01 2.28943482e-01 2.65574664e-01 -7.86563039e-01 3.28070492e-01 1.06071758e+00 -8.76629278e-02 -2.45252669e-01 -3.69100273e-01 -5.95039845e-01 6.37282908e-01 5.52741766e-01 -1.46093577e-01 -6.34874582e-01 -1.76638573e-01 -1.15456827e-01 -2.89770097e-01 6.97222531e-01 -1.12390530e+00 2.94378608e-01 -5.49918264e-02 4.33362633e-01 -4.82277662e-01 3.67358118e-01 -7.80575752e-01 5.86952120e-02 3.54113132e-01 1.30295277e-01 -4.19846237e-01 1.27267405e-01 6.71640515e-01 -2.03905866e-01 9.39043537e-02 4.89963561e-01 5.96016161e-02 -1.02110052e+00 2.99090862e-01 -1.84259459e-01 -1.91900373e-01 1.41997647e+00 -4.77905512e-01 -2.86063612e-01 -5.62405169e-01 -2.93136716e-01 2.95056581e-01 6.49854362e-01 4.88122672e-01 7.86799729e-01 -1.05001235e+00 -4.80379015e-01 4.19304997e-01 6.58445477e-01 -1.18596397e-01 1.87391445e-01 1.12265599e+00 -5.15191317e-01 5.29049993e-01 1.09588429e-01 -7.95812607e-01 -1.63661170e+00 6.11821234e-01 8.47785324e-02 1.34057298e-01 -9.88213062e-01 6.92157209e-01 5.95896304e-01 1.15273975e-01 4.33083117e-01 -1.84950326e-02 -1.78312570e-01 -7.57178515e-02 5.94226658e-01 4.93438661e-01 -1.23008922e-01 -8.56145263e-01 -4.93505746e-01 6.00062907e-01 7.16822743e-02 1.40455067e-01 8.88682783e-01 -2.16266140e-01 9.50404257e-02 1.32027194e-01 1.04768205e+00 1.48597911e-01 -1.04439712e+00 -1.67779744e-01 -8.17178413e-02 -6.24021649e-01 2.64458507e-01 -7.32850730e-01 -8.89742076e-01 9.92701054e-01 5.30090988e-01 7.41493702e-02 1.04501200e+00 1.89924315e-01 8.72491658e-01 -1.08643007e-02 6.48429155e-01 -7.36289799e-01 3.00506413e-01 1.26327053e-01 7.36916602e-01 -1.32852948e+00 3.35917063e-02 -1.13364625e+00 -6.87790275e-01 7.16287374e-01 5.39381862e-01 4.87113073e-02 2.36786813e-01 1.19399309e-01 2.79015936e-02 -5.41504741e-01 -2.55622417e-01 -6.56375349e-01 7.25753248e-01 7.42336869e-01 4.21604037e-01 1.24145389e-01 -2.25525349e-01 1.10117570e-01 5.17143160e-02 -5.71989976e-02 5.67477457e-02 9.64840055e-01 -4.59230214e-01 -6.68720484e-01 -3.66947353e-01 4.82261598e-01 -1.28899822e-02 1.07962020e-01 -5.73652864e-01 8.34238946e-01 1.97437018e-01 1.13797557e+00 -2.94162840e-01 -5.18972635e-01 3.95860761e-01 -3.08662474e-01 4.64936107e-01 -7.96208382e-01 -2.22857326e-01 1.38339967e-01 1.97309077e-01 -8.19975376e-01 -1.32260397e-01 -4.29095060e-01 -1.34354258e+00 -1.25983819e-01 -5.24378002e-01 -2.31692359e-01 5.29135704e-01 8.85620415e-01 6.01153791e-01 4.93852556e-01 4.33763772e-01 -8.54394674e-01 -2.17863232e-01 -6.60657704e-01 -5.04630864e-01 9.44809198e-01 2.74140954e-01 -1.15105939e+00 -3.28774691e-01 -6.75130934e-02]
[9.538262367248535, 1.3809270858764648]
ac92284b-a69b-41e8-8504-62a00cdc9706
free-form-video-inpainting-with-3d-gated
1904.10247
null
https://arxiv.org/abs/1904.10247v3
https://arxiv.org/pdf/1904.10247v3.pdf
Free-form Video Inpainting with 3D Gated Convolution and Temporal PatchGAN
Free-form video inpainting is a very challenging task that could be widely used for video editing such as text removal. Existing patch-based methods could not handle non-repetitive structures such as faces, while directly applying image-based inpainting models to videos will result in temporal inconsistency (see http://bit.ly/2Fu1n6b ). In this paper, we introduce a deep learn-ing based free-form video inpainting model, with proposed 3D gated convolutions to tackle the uncertainty of free-form masks and a novel Temporal PatchGAN loss to enhance temporal consistency. In addition, we collect videos and design a free-form mask generation algorithm to build the free-form video inpainting (FVI) dataset for training and evaluation of video inpainting models. We demonstrate the benefits of these components and experiments on both the FaceForensics and our FVI dataset suggest that our method is superior to existing ones. Related source code, full-resolution result videos and the FVI dataset could be found on Github https://github.com/amjltc295/Free-Form-Video-Inpainting .
['Winston Hsu', 'Kuan-Ying Lee', 'Ya-Liang Chang', 'Zhe Yu Liu']
2019-04-23
free-form-video-inpainting-with-3d-gated-1
http://openaccess.thecvf.com/content_ICCV_2019/html/Chang_Free-Form_Video_Inpainting_With_3D_Gated_Convolution_and_Temporal_PatchGAN_ICCV_2019_paper.html
http://openaccess.thecvf.com/content_ICCV_2019/papers/Chang_Free-Form_Video_Inpainting_With_3D_Gated_Convolution_and_Temporal_PatchGAN_ICCV_2019_paper.pdf
iccv-2019-10
['video-inpainting']
['computer-vision']
[ 2.50062287e-01 -1.54766023e-01 -8.15965161e-02 -3.88519049e-01 -5.52536368e-01 -2.92486757e-01 3.16128492e-01 -7.31654704e-01 2.20978539e-03 7.34240830e-01 1.17234983e-01 -3.70654613e-02 -7.27226911e-03 -4.95077670e-01 -1.21241355e+00 -4.15860623e-01 3.43118720e-02 4.24891785e-02 8.44916031e-02 -1.05511732e-01 2.36399204e-01 3.87504339e-01 -1.46283746e+00 6.29013598e-01 9.04080510e-01 9.38148618e-01 3.64250690e-01 5.32659292e-01 -8.27699080e-02 6.85216784e-01 -4.48625594e-01 -4.88933653e-01 5.41525245e-01 -6.41764224e-01 -4.94154662e-01 2.18253806e-01 8.18880022e-01 -8.81686568e-01 -7.70703197e-01 9.51840222e-01 4.44352061e-01 1.44316107e-01 4.07168269e-01 -1.18941200e+00 -7.21813142e-01 1.24083824e-01 -8.96548390e-01 8.71488824e-02 5.06085634e-01 1.83148861e-01 4.22539681e-01 -1.15336382e+00 8.18185568e-01 1.49304211e+00 6.28739297e-01 7.97372103e-01 -9.01969552e-01 -9.36281264e-01 3.55308726e-02 4.57256228e-01 -1.36630046e+00 -7.57353604e-01 8.45699430e-01 -3.15246820e-01 6.56257570e-01 3.26191157e-01 8.11089516e-01 1.13702357e+00 3.55794072e-01 9.20419335e-01 8.17378819e-01 -3.24521273e-01 -1.05044678e-01 -4.64367807e-01 -4.86108124e-01 8.54409933e-01 -3.43580544e-02 3.43103856e-01 -8.02648842e-01 6.04426377e-02 1.20156837e+00 3.20231378e-01 -4.15557921e-01 2.23654345e-01 -8.79981756e-01 6.55454755e-01 1.47314802e-01 8.34967718e-02 -1.89230129e-01 3.41612786e-01 2.33000249e-01 5.42302072e-01 7.81103849e-01 -3.59448306e-02 -2.38295794e-01 -5.97873479e-02 -1.32992578e+00 5.77810884e-01 3.56231570e-01 1.05207169e+00 8.45423579e-01 1.03944510e-01 -3.69625539e-01 1.04562235e+00 3.06588292e-01 4.11259025e-01 4.31840271e-01 -1.38117981e+00 3.17899942e-01 9.12675187e-02 -9.62593313e-03 -9.54497099e-01 4.86725271e-02 2.56580830e-01 -7.11568952e-01 2.15393782e-01 1.75236821e-01 -1.72401980e-01 -1.03031480e+00 1.58772969e+00 4.33106393e-01 5.99625349e-01 -6.11920059e-01 9.29964781e-01 1.01583767e+00 9.21872973e-01 -4.23585534e-01 -3.96190375e-01 1.12576389e+00 -1.25945604e+00 -1.20760369e+00 -1.14830203e-01 1.48433834e-01 -1.18949521e+00 8.21211696e-01 4.57209587e-01 -1.33841515e+00 -6.78045273e-01 -8.23943079e-01 -4.20068890e-01 1.91210270e-01 1.05151542e-01 3.99901003e-01 1.25988945e-01 -1.13283992e+00 7.97174215e-01 -8.71431112e-01 -2.50045717e-01 7.24849522e-01 1.24549061e-01 -6.53132200e-01 -3.08445513e-01 -1.14021909e+00 6.41123295e-01 1.30835474e-01 3.33289474e-01 -1.19012654e+00 -8.12216640e-01 -7.96113789e-01 -2.01463327e-01 5.84729850e-01 -4.92852539e-01 1.24461174e+00 -1.29346514e+00 -1.52526796e+00 7.40477502e-01 -4.87784892e-01 -2.00253367e-01 6.97137654e-01 -4.78386045e-01 -2.81327397e-01 4.17687714e-01 6.08535707e-02 9.75349367e-01 1.65804756e+00 -1.00056100e+00 -2.45319590e-01 -8.34665671e-02 -7.64183104e-02 8.76686126e-02 -6.46743998e-02 3.94956432e-02 -8.32590818e-01 -1.23289275e+00 -1.77068844e-01 -7.37436533e-01 2.84674376e-01 6.69690609e-01 -2.09706530e-01 1.11787066e-01 1.17960286e+00 -1.30396152e+00 1.28611290e+00 -2.17355466e+00 3.53099853e-02 -3.07928681e-01 1.17397401e-03 5.15365541e-01 -4.28171813e-01 5.29243171e-01 -1.92224324e-01 6.76595941e-02 -4.55476016e-01 -5.65492332e-01 -3.75990987e-01 1.72140390e-01 -2.63438821e-01 4.65122968e-01 5.03537178e-01 8.58809829e-01 -6.07299924e-01 -4.50046241e-01 2.71223247e-01 7.56090999e-01 -7.39763260e-01 5.08105040e-01 -3.83174986e-01 5.83100498e-01 -1.42642990e-01 8.23559105e-01 1.00654066e+00 1.55893594e-01 -1.18835606e-01 -3.49271387e-01 -3.94092128e-02 -7.55632147e-02 -9.69728529e-01 2.00983071e+00 -3.15224797e-01 7.41928160e-01 4.37029094e-01 -7.15310693e-01 5.66362858e-01 3.38033348e-01 5.44926226e-01 -8.02827418e-01 1.02182157e-01 -3.21007036e-02 -1.54532343e-01 -7.16670454e-01 3.21998924e-01 2.02782881e-02 6.54482603e-01 3.68465990e-01 2.27990851e-01 -2.02503085e-01 3.53571087e-01 2.11887538e-01 9.50209498e-01 6.71826422e-01 -4.01776023e-02 -2.51204520e-02 3.16908568e-01 -5.20855427e-01 6.53466165e-01 2.65512377e-01 -1.75321534e-01 1.14205933e+00 4.32054281e-01 -5.24383187e-01 -9.49856639e-01 -6.66670024e-01 -9.65620577e-02 7.43378401e-01 6.94837719e-02 -5.82017124e-01 -8.30107391e-01 -5.15156269e-01 2.14142855e-02 4.46257442e-01 -7.98850179e-01 -1.70357395e-02 -6.64557755e-01 -2.59573668e-01 3.65640044e-01 1.88667223e-01 5.85071623e-01 -1.22301769e+00 -2.24112257e-01 1.18168421e-01 -4.30001289e-01 -8.90335441e-01 -1.12966025e+00 -2.92122692e-01 -8.33255947e-01 -1.16787124e+00 -8.79117727e-01 -7.77428210e-01 6.59716487e-01 5.00220716e-01 8.72243106e-01 4.21287060e-01 -4.75962877e-01 2.68149555e-01 -4.49720740e-01 -2.69748658e-01 -3.73661876e-01 -4.93776411e-01 -9.23044607e-02 2.15619877e-01 -8.65563098e-03 -7.50094593e-01 -8.56597424e-01 3.07940155e-01 -1.31692791e+00 2.90432513e-01 3.11955184e-01 9.61872756e-01 6.30867660e-01 -1.32825091e-01 1.75741896e-01 -7.66878903e-01 3.96369755e-01 -3.17123234e-01 -6.13768637e-01 7.67512545e-02 -2.70450801e-01 -1.83671445e-01 5.08304238e-01 -6.09218836e-01 -1.25035179e+00 -1.17331728e-01 -4.52223629e-01 -1.05138350e+00 2.88444683e-02 1.72203526e-01 -1.92396060e-01 -1.33626595e-01 2.90683657e-01 1.95216596e-01 3.77110302e-01 -6.26554787e-01 2.67134070e-01 3.38279516e-01 2.70035595e-01 -5.06869674e-01 8.28802824e-01 6.11452043e-01 -1.90128267e-01 -8.61209810e-01 -5.12978673e-01 1.28750175e-01 -3.31945717e-01 -3.87519568e-01 5.64347565e-01 -9.72336650e-01 -4.04771417e-01 7.16758251e-01 -1.29101539e+00 -7.18038082e-01 -1.24057770e-01 2.18008429e-01 -6.28145039e-01 7.32554913e-01 -8.83801281e-01 -4.67353702e-01 -3.77602071e-01 -1.02642846e+00 1.16036785e+00 2.89619505e-01 -3.32064158e-03 -7.76175141e-01 5.56930602e-02 3.71966034e-01 3.56940895e-01 2.12584957e-01 5.25859296e-01 7.51473010e-02 -5.78559458e-01 2.19896525e-01 -2.83007920e-02 6.27539515e-01 2.16383472e-01 3.70505184e-01 -8.92250776e-01 -4.05527264e-01 1.53333530e-01 -3.40045929e-01 1.02114022e+00 7.19667315e-01 1.45635450e+00 -5.23255408e-01 -1.08919315e-01 1.00588655e+00 1.14869678e+00 3.71336520e-01 1.10310400e+00 -4.73765880e-02 8.23089719e-01 5.04566431e-01 8.36259544e-01 6.75905228e-01 -2.68799961e-02 8.45821083e-01 3.55854452e-01 -6.63396418e-02 -5.42616963e-01 -3.83816689e-01 5.88209271e-01 7.12017655e-01 -6.40295148e-02 -2.47527480e-01 -3.48612666e-01 4.16269541e-01 -1.83391500e+00 -1.15401721e+00 6.71609864e-02 1.94233346e+00 1.15317392e+00 -2.74020821e-01 -1.09798443e-02 -4.39783037e-02 7.76958168e-01 2.63498485e-01 -5.35328209e-01 -3.30333680e-01 1.62684582e-02 3.94851178e-01 2.22410411e-01 8.95224094e-01 -9.87452328e-01 1.05257404e+00 5.07185173e+00 1.16514456e+00 -1.31609392e+00 3.30332965e-01 8.10184538e-01 -3.80023926e-01 -2.77598053e-01 2.25068107e-02 -6.02751434e-01 9.08426940e-01 6.14712954e-01 1.36744261e-01 8.53779793e-01 3.98076981e-01 6.22796476e-01 -2.16364488e-01 -9.99505758e-01 1.33498919e+00 2.44940192e-01 -1.44587469e+00 1.91532761e-01 -1.85655832e-01 7.50516355e-01 -3.34665239e-01 3.07678990e-03 -2.83844266e-02 -3.36010277e-01 -9.99190569e-01 9.50947285e-01 6.49545074e-01 1.23996663e+00 -5.80951214e-01 2.59723216e-01 -4.64887246e-02 -1.04230654e+00 1.55466348e-01 -3.80518466e-01 -8.51870254e-02 2.20093310e-01 6.85390651e-01 -3.50278586e-01 5.85936010e-01 9.10142839e-01 1.01304543e+00 -4.33676213e-01 1.01707363e+00 -2.69788176e-01 6.26245379e-01 -2.38160700e-01 6.08241796e-01 -1.93224996e-01 -4.28513974e-01 4.89755213e-01 9.55834508e-01 6.63121283e-01 2.64099568e-01 -2.00164407e-01 9.33241785e-01 -2.57100195e-01 -6.63892478e-02 -6.62506521e-01 -2.19026104e-01 4.53219920e-01 1.22028136e+00 -4.68886077e-01 -1.61891788e-01 -5.82248986e-01 1.42154038e+00 5.60454652e-02 4.23055947e-01 -1.10774958e+00 -3.68731678e-01 7.55878270e-01 4.86004204e-01 5.73174357e-01 -1.54789686e-01 2.14799777e-01 -1.46867585e+00 2.39584297e-01 -1.05828333e+00 2.64844075e-02 -8.81726027e-01 -1.13799357e+00 5.10452032e-01 1.01137348e-01 -1.42303050e+00 -1.88128397e-01 -5.04023433e-01 -6.90294743e-01 6.36251330e-01 -1.36751497e+00 -1.22173274e+00 -4.49255288e-01 9.55179095e-01 8.77575636e-01 -4.28973064e-02 3.56887609e-01 6.52505517e-01 -5.04555702e-01 6.17001653e-01 8.27241912e-02 -9.75575969e-02 1.13008285e+00 -4.28635448e-01 3.59283388e-01 1.12824655e+00 -1.79897435e-02 5.39025068e-01 6.36631906e-01 -8.81549478e-01 -1.66078055e+00 -1.22643757e+00 4.67422426e-01 -2.32863873e-02 2.22353011e-01 -3.32227856e-01 -9.25777674e-01 7.57415295e-01 4.42351431e-01 1.58184037e-01 2.79473215e-01 -6.21026456e-01 -2.17469260e-01 -2.43791744e-01 -1.33361411e+00 5.45369446e-01 1.32715237e+00 -3.55627596e-01 -1.75513506e-01 4.76321220e-01 6.94343567e-01 -5.80199659e-01 -5.97430229e-01 2.82786101e-01 6.74689293e-01 -1.04680395e+00 7.77201355e-01 -1.70004338e-01 7.87574470e-01 -4.65199530e-01 -1.75990406e-02 -9.69814718e-01 -2.98563629e-01 -1.11771894e+00 -4.89112318e-01 1.24275780e+00 -1.58476576e-01 -4.54118073e-01 5.96934021e-01 2.43772507e-01 -2.31577650e-01 -8.40408385e-01 -1.00030529e+00 -6.35562658e-01 -9.58041698e-02 -3.60462278e-01 3.32797140e-01 7.86709428e-01 -4.31184888e-01 -1.38262525e-01 -9.36232924e-01 -1.70097977e-01 5.20978689e-01 -6.68260530e-02 6.79033458e-01 -6.62634790e-01 -4.25124675e-01 -1.06697984e-01 -1.26039609e-02 -1.12367535e+00 1.63438901e-01 -5.81172168e-01 -1.40244678e-01 -1.36262250e+00 6.63356483e-02 -5.11031672e-02 2.22391769e-01 6.50033474e-01 -1.14079513e-01 5.23459494e-01 3.38142455e-01 1.54766828e-01 -2.41271257e-01 8.65020156e-01 1.59893596e+00 -1.96878985e-01 -1.88614300e-03 -2.95384943e-01 -4.18191493e-01 4.86935556e-01 5.76624632e-01 -3.66044968e-01 -5.45683026e-01 -6.22827411e-01 -5.82482703e-02 3.17511946e-01 3.86793494e-01 -8.08778286e-01 -1.38719767e-01 -2.51165122e-01 6.37187421e-01 -4.81168300e-01 5.50492704e-01 -7.13806391e-01 5.09787858e-01 3.94987375e-01 -1.79980823e-03 9.52920243e-02 5.06618023e-01 3.76776785e-01 -4.06943023e-01 -2.95126677e-01 9.23188090e-01 -2.43923485e-01 -6.37355447e-01 6.20238721e-01 -1.90385312e-01 -1.78098481e-03 9.02229965e-01 -2.77231157e-01 -8.04414675e-02 -7.31997788e-01 -5.53045571e-01 8.92789569e-03 6.69958353e-01 5.04648983e-01 1.06245756e+00 -1.42429066e+00 -8.26791286e-01 5.31754017e-01 -4.41449672e-01 1.90333631e-02 7.58047044e-01 8.76498401e-01 -9.23323512e-01 3.99290957e-02 -5.73895693e-01 -4.15324897e-01 -1.38191903e+00 5.89261293e-01 2.09754303e-01 9.97457504e-02 -6.04229629e-01 1.15094864e+00 4.86437440e-01 -4.07142341e-02 2.43661776e-01 -2.96615273e-01 2.45723352e-01 -6.65958524e-02 7.24313080e-01 2.52882123e-01 1.10660456e-02 -5.25494099e-01 -3.32399696e-01 5.51124036e-01 -3.29187423e-01 -6.99287057e-02 1.35659766e+00 -1.58377498e-01 -3.20421100e-01 -5.51702864e-02 1.22120035e+00 -2.44959369e-02 -1.69659352e+00 9.66592506e-02 -6.42374933e-01 -9.73696470e-01 -2.55974382e-01 -5.83051860e-01 -1.44340134e+00 8.26522350e-01 7.71014214e-01 -3.97260457e-01 1.45679998e+00 -3.47480953e-01 1.04368293e+00 -8.09404626e-02 2.52124876e-01 -8.92943919e-01 3.26385349e-01 3.43147248e-01 1.52312648e+00 -1.10690105e+00 2.09363922e-01 -4.73920614e-01 -4.08133775e-01 1.27162683e+00 6.98952675e-01 -2.57610172e-01 7.63885677e-01 3.75972658e-01 -9.40847304e-03 4.25045975e-02 -8.22606325e-01 2.51189351e-01 1.45586595e-01 5.80849826e-01 5.33910930e-01 -3.18406641e-01 -6.15389764e-01 3.02359849e-01 1.95551172e-01 2.87748814e-01 5.25282979e-01 1.01176286e+00 7.71247745e-02 -1.43043303e+00 -4.71407682e-01 4.68465358e-01 -4.12382543e-01 -2.17510372e-01 -9.42851454e-02 5.55684149e-01 3.75777662e-01 7.98669815e-01 -3.82967331e-02 -3.82277071e-01 -1.76243130e-02 -8.14699605e-02 8.38129163e-01 -3.59492540e-01 -2.78113335e-01 4.01801825e-01 -1.25906467e-01 -9.78717089e-01 -5.11586130e-01 -4.97002453e-01 -8.90930057e-01 -5.49483657e-01 -5.29241934e-02 -2.56299496e-01 2.99954265e-01 5.85738242e-01 6.57372415e-01 3.66072655e-01 5.01119673e-01 -1.32403648e+00 -6.12986349e-02 -1.12430179e+00 -4.94383574e-01 5.24285555e-01 4.13693964e-01 -7.38034248e-01 -2.45615542e-01 4.98380154e-01]
[10.940652847290039, -1.2598581314086914]
e7fdff61-5f24-4a99-b3fb-5d7a7ef7277d
deep-factor-model-a-novel-approach-for-motion
2304.00102
null
https://arxiv.org/abs/2304.00102v1
https://arxiv.org/pdf/2304.00102v1.pdf
Deep Factor Model: A Novel Approach for Motion Compensated Multi-Dimensional MRI
Recent quantitative parameter mapping methods including MR fingerprinting (MRF) collect a time series of images that capture the evolution of magnetization. The focus of this work is to introduce a novel approach termed as Deep Factor Model(DFM), which offers an efficient representation of the multi-contrast image time series. The higher efficiency of the representation enables the acquisition of the images in a highly undersampled fashion, which translates to reduced scan time in 3D high-resolution multi-contrast applications. The approach integrates motion estimation and compensation, making the approach robust to subject motion during the scan.
['Mathews Jacob', 'Vincent Magnotta', 'Curtis Corum', 'James H. Holmes', 'Yan Chen']
2023-03-31
null
null
null
null
['motion-estimation']
['computer-vision']
[ 3.40555638e-01 -4.42668140e-01 -1.61194101e-01 -2.73464382e-01 -4.99255270e-01 -1.64733082e-01 6.02388084e-01 -6.94693103e-02 -7.64033973e-01 6.73441350e-01 7.74354935e-02 5.73643036e-02 -5.52619994e-01 -4.01220053e-01 -3.08387488e-01 -8.72485995e-01 -4.21190351e-01 3.71661007e-01 5.62700450e-01 -1.92610435e-02 5.23557544e-01 8.37334394e-01 -1.25539625e+00 7.30219483e-02 6.25042439e-01 6.91024303e-01 7.59348333e-01 4.78789955e-01 3.24475378e-01 7.63574421e-01 -2.20809117e-01 2.53844440e-01 5.00782300e-03 -4.48837161e-01 -9.75113153e-01 -6.81521669e-02 2.70056635e-01 -6.12388909e-01 -4.09947634e-01 9.51036453e-01 6.49016738e-01 4.17285830e-01 6.03577435e-01 -6.57839477e-01 -9.69398916e-02 4.29624528e-01 -7.28759468e-01 1.06398141e+00 2.49151036e-01 -2.43261456e-01 1.48704335e-01 -7.94389606e-01 9.54286456e-01 7.96253264e-01 3.69232148e-01 2.97891945e-01 -1.32471192e+00 -2.29063667e-02 -5.02526045e-01 5.11735559e-01 -1.16793084e+00 -2.95389116e-01 9.48331475e-01 -4.29814368e-01 6.75677359e-01 2.77526349e-01 6.36425436e-01 7.60366976e-01 9.89894629e-01 4.17531699e-01 1.95382297e+00 -4.45771247e-01 1.12179294e-02 -3.84767920e-01 2.12081850e-01 6.45168245e-01 1.93665951e-01 4.14790809e-01 -5.70844889e-01 -3.10570091e-01 1.07767379e+00 -1.29514653e-02 -4.93351221e-01 -4.89211857e-01 -1.58187044e+00 5.22941530e-01 1.50640234e-01 7.58083701e-01 -7.04645216e-01 -1.27358183e-01 4.44596350e-01 2.69895166e-01 1.32660404e-01 4.83136922e-01 2.22672150e-01 -5.34566827e-02 -1.41846287e+00 2.71959037e-01 1.32347912e-01 2.58313924e-01 3.97491992e-01 1.62363052e-01 -5.35869189e-02 5.67691982e-01 2.05076650e-01 4.73667055e-01 9.66456056e-01 -1.19912279e+00 -1.05664790e-01 -7.80534819e-02 -2.74629518e-02 -8.57991338e-01 -8.52075636e-01 -4.31388378e-01 -9.41135705e-01 1.92697763e-01 3.13605487e-01 4.35412526e-01 -6.93478703e-01 1.51270914e+00 4.56177950e-01 1.09400004e-01 -3.77419770e-01 1.13241851e+00 3.60259742e-01 3.28352869e-01 -1.66376412e-01 -7.07043827e-01 1.30083847e+00 -6.60608172e-01 -1.05798614e+00 1.89223260e-01 1.13888495e-01 -6.53801918e-01 7.05088496e-01 3.77546817e-01 -1.17805338e+00 -5.28960943e-01 -1.21370387e+00 1.83122754e-01 -1.81887031e-03 -5.00867903e-01 4.92326051e-01 2.32235804e-01 -1.05258727e+00 9.48492110e-01 -1.17147648e+00 2.17192426e-01 -9.57233540e-04 3.95204812e-01 -5.96890271e-01 -1.15108781e-01 -1.27783310e+00 1.23855698e+00 3.96027505e-01 1.14513487e-01 -8.90507281e-01 -9.67162967e-01 -5.62057555e-01 -5.29768109e-01 5.06493077e-02 -3.48912865e-01 9.07779753e-01 -2.89584875e-01 -1.40888357e+00 9.33559358e-01 -2.78713107e-01 -4.62519467e-01 6.76774919e-01 1.09242797e-01 -6.24146819e-01 9.74047899e-01 6.34716079e-02 2.94308931e-01 1.24104583e+00 -8.33639979e-01 1.08010747e-01 -6.00902438e-01 -3.57231200e-01 1.69698410e-02 3.94936830e-01 1.92300349e-01 1.97226256e-01 -5.53870320e-01 3.31088006e-01 -9.04559433e-01 -3.69881243e-01 -3.43418181e-01 6.00516871e-02 5.67062616e-01 7.42144763e-01 -9.14180815e-01 1.11499703e+00 -2.00024700e+00 2.94069141e-01 1.97156146e-01 7.54669189e-01 4.24148180e-02 8.48693177e-02 2.97962427e-01 -2.76861757e-01 -4.58893389e-01 -4.07154799e-01 1.68436348e-01 -5.24122357e-01 1.55712070e-03 -4.89371456e-02 1.00478041e+00 -1.29039928e-01 1.01901913e+00 -8.94326389e-01 -6.62331760e-01 3.79304737e-01 5.11034548e-01 -2.22977653e-01 2.60208815e-01 4.63989317e-01 1.23595619e+00 -4.86925155e-01 2.59703338e-01 9.19035554e-01 -2.75382668e-01 3.61486882e-01 -4.83838439e-01 -6.39602900e-01 -7.39556411e-03 -9.09981966e-01 1.78979754e+00 -3.56658511e-02 4.41979289e-01 6.98233396e-02 -9.89225388e-01 7.47035384e-01 5.14303505e-01 1.00033033e+00 -1.29178894e+00 3.11367977e-02 3.85055244e-01 8.51944387e-02 -4.80356991e-01 6.52097344e-01 -4.46318477e-01 2.22214475e-01 6.92041576e-01 -7.65916985e-03 -6.07198030e-02 1.79323375e-01 -1.36385262e-01 8.23362589e-01 6.80481791e-02 4.60892916e-01 -8.54208708e-01 6.73669994e-01 -3.34411621e-01 2.97305137e-01 7.64364541e-01 -5.18620610e-01 4.79294926e-01 6.84171245e-02 -5.69780767e-01 -1.55498695e+00 -1.03756261e+00 -7.68740475e-01 2.94233382e-01 1.70207337e-01 -9.35441535e-03 -7.67536223e-01 -8.91450420e-02 -2.44449779e-01 -1.17235795e-01 -6.71990097e-01 -9.02042240e-02 -1.27214265e+00 -1.12586033e+00 1.81140989e-01 1.62121028e-01 5.09258807e-01 -9.49521840e-01 -1.30341649e+00 5.75880051e-01 -4.51723844e-01 -1.08653462e+00 -4.78417009e-01 1.02706686e-01 -1.33051741e+00 -9.34809327e-01 -1.09741187e+00 -4.10678983e-01 3.97115439e-01 2.26675019e-01 8.26057553e-01 -1.78431526e-01 -4.61226046e-01 2.60416627e-01 -9.32308957e-02 3.90607148e-01 -4.60194975e-01 -1.12406202e-01 1.83544889e-01 1.51026070e-01 -7.56501034e-02 -8.57876182e-01 -8.88056695e-01 4.06804562e-01 -1.10602534e+00 8.01788643e-02 5.66501141e-01 9.63773847e-01 1.11978376e+00 -1.54118031e-01 4.97295409e-01 -6.99160516e-01 4.83911157e-01 -3.02070051e-01 -6.34799361e-01 1.89232200e-01 -7.56090760e-01 4.06419843e-01 4.44479883e-01 -6.46928787e-01 -9.81699705e-01 -4.43912864e-01 4.36043963e-02 -3.20168734e-01 9.42282304e-02 3.85142356e-01 3.13410282e-01 -7.51659572e-01 4.58374113e-01 7.17082739e-01 3.57136577e-01 -5.42896450e-01 -8.69075581e-02 1.54313996e-01 8.86026680e-01 -6.65660262e-01 3.27839822e-01 6.34178460e-01 6.81601107e-01 -1.03915524e+00 -3.37392867e-01 -3.89351606e-01 -9.99352336e-01 -6.76061153e-01 7.70375252e-01 -3.84235799e-01 -6.83929563e-01 6.88900471e-01 -7.64275551e-01 -2.78736025e-01 -2.44352110e-02 1.13593709e+00 -8.96076679e-01 8.06770146e-01 -9.17215049e-01 -3.51601809e-01 -3.52797180e-01 -1.57406616e+00 7.11690068e-01 1.20008796e-01 -2.06735477e-01 -9.70699787e-01 2.70017087e-01 2.09203124e-01 7.71235764e-01 6.13690019e-01 1.05046570e+00 9.17708129e-02 -5.88368416e-01 2.31300488e-01 1.41055286e-01 9.68535766e-02 -1.53150514e-01 -6.14969611e-01 -5.82581699e-01 -3.79801244e-01 7.31850564e-01 1.10199889e-02 4.87944722e-01 9.39693928e-01 1.07709932e+00 3.75946820e-01 -1.17659159e-02 7.21827924e-01 1.42274857e+00 3.16380322e-01 8.37707162e-01 4.98019457e-01 4.24801588e-01 4.82053071e-01 6.91620052e-01 2.16341466e-01 1.25578344e-01 9.48620558e-01 -3.02549768e-02 -1.15743883e-01 -1.59281909e-01 -1.95615366e-02 -7.00961947e-02 1.38691902e+00 -4.62644339e-01 4.79308635e-01 -7.82976270e-01 3.38392049e-01 -1.45782137e+00 -1.11912775e+00 -3.06652606e-01 2.17756915e+00 7.43246496e-01 -2.35247597e-01 6.28888011e-02 2.67634302e-01 8.49361122e-01 4.42099243e-01 -5.44226944e-01 -1.08666591e-01 -1.32163271e-01 3.41193974e-01 6.12825394e-01 6.38856113e-01 -7.03553557e-01 1.40972078e-01 8.01707268e+00 4.69324261e-01 -1.48866749e+00 4.22345519e-01 2.77966231e-01 8.80441070e-02 -1.80353135e-01 -1.79313868e-01 -3.67052943e-01 6.17083251e-01 1.47850585e+00 -2.66532540e-01 5.22585571e-01 1.44586191e-01 5.01845300e-01 -3.70964050e-01 -6.07098103e-01 9.09357905e-01 -1.00786418e-01 -1.25080478e+00 -1.62966490e-01 3.13704431e-01 3.09873432e-01 1.65979400e-01 1.10587500e-01 -4.04237181e-01 -7.99988091e-01 -7.03371942e-01 8.19692254e-01 9.50001061e-01 1.11195529e+00 -7.43267298e-01 3.99205625e-01 2.14321211e-01 -9.15384114e-01 2.99678296e-01 -2.39085793e-01 2.92900085e-01 6.97257161e-01 7.34863937e-01 -3.73997778e-01 6.75069749e-01 4.73667055e-01 3.52108151e-01 -3.47958297e-01 9.83065546e-01 3.04703087e-01 1.85473695e-01 1.05373533e-02 5.22138774e-01 1.22765392e-01 -4.65773404e-01 8.52680981e-01 7.65196741e-01 2.63355672e-01 1.09209590e-01 4.14959267e-02 6.88428462e-01 4.09780949e-01 -2.00116292e-01 -4.09611017e-01 -3.17629911e-02 2.28806347e-01 9.83446419e-01 -8.19761276e-01 -8.12508091e-02 -3.37118477e-01 8.86420906e-01 1.06803879e-01 1.80200413e-01 -5.86031854e-01 -7.43684992e-02 -9.49049741e-02 5.55858970e-01 1.44096509e-01 -5.87856352e-01 3.49508017e-01 -1.15874100e+00 7.00277463e-02 -8.31749737e-01 1.31229669e-01 -6.52835727e-01 -9.79370356e-01 7.74006128e-01 5.99349618e-01 -9.67158735e-01 -3.61115903e-01 -3.46415877e-01 -1.27321914e-01 9.49483037e-01 -1.67695737e+00 -7.17747450e-01 -8.76932144e-02 6.68795049e-01 -4.04117145e-02 8.14523697e-02 6.11050308e-01 6.11313105e-01 9.84458253e-03 1.37871772e-01 1.78226799e-01 -3.94872278e-01 6.68437183e-01 -1.20738256e+00 5.08463979e-02 7.62699008e-01 -5.25023818e-01 8.25595796e-01 6.80474520e-01 -6.63508475e-01 -1.42874289e+00 -4.57670331e-01 6.51296198e-01 -1.01666249e-01 4.90198404e-01 -7.56252510e-03 -1.30224478e+00 3.11619222e-01 -4.15295735e-02 3.34975094e-01 6.13763690e-01 -6.47689164e-01 6.21440709e-02 -1.50094898e-02 -1.48295271e+00 6.93647861e-02 5.41987300e-01 -7.68864095e-01 -6.72958195e-01 -1.49930462e-01 3.21509421e-01 -5.89555264e-01 -1.57659137e+00 3.02273333e-01 7.67263830e-01 -1.00994682e+00 9.99674559e-01 -7.72461742e-02 2.06514180e-01 -3.46062839e-01 3.54791097e-02 -1.08507824e+00 -5.73185205e-01 -6.76857531e-01 -5.48179805e-01 4.77226079e-01 -1.87096760e-01 -8.68864834e-01 3.66277575e-01 2.09204152e-01 -1.43583044e-01 -5.61588943e-01 -1.25595331e+00 -6.15569413e-01 -1.21761389e-01 1.73952505e-01 6.16673291e-01 1.04158902e+00 -2.52627939e-01 -2.16148898e-01 -5.24380386e-01 -4.33862992e-02 1.23361015e+00 2.74574786e-01 -3.13972950e-01 -1.10270929e+00 -3.77110362e-01 -1.57126307e-01 -5.21937788e-01 -6.31955564e-01 2.56029703e-02 -8.56033862e-01 -2.29762942e-01 -7.49438524e-01 4.24035132e-01 -3.02038938e-01 -5.43601811e-01 -3.48499328e-01 5.29026426e-02 3.16001356e-01 -1.53805986e-01 6.78034544e-01 -2.10108653e-01 3.61474246e-01 1.90085506e+00 1.59144938e-01 1.76235110e-01 -2.67509580e-01 -5.23669757e-02 6.17424287e-02 5.12033403e-01 -5.41512489e-01 -4.20733869e-01 -1.83239430e-01 -1.33147806e-01 7.82173455e-01 3.80961806e-01 -8.64836335e-01 1.38483837e-01 -4.35378179e-02 4.85098124e-01 -6.96285844e-01 1.40751854e-01 -6.46143258e-01 6.67605221e-01 7.79059410e-01 -1.73676625e-01 6.37882352e-01 -4.15548831e-02 1.52715430e-01 -5.34343779e-01 -3.81635398e-01 1.39587450e+00 -3.76275569e-01 -6.41482413e-01 4.98997927e-01 -3.12797636e-01 -1.17257945e-01 6.67229295e-01 -1.36977166e-01 -1.38787627e-01 1.57177329e-01 -7.75921822e-01 -3.83550912e-01 5.82246065e-01 -1.46612804e-02 6.73939824e-01 -1.50780082e+00 -4.48240519e-01 2.53227234e-01 -3.28814417e-01 -4.26093280e-01 1.01114154e+00 1.57850480e+00 -7.34501302e-01 5.32501698e-01 -8.93498898e-01 -6.65799141e-01 -7.57186234e-01 6.40518904e-01 7.52812088e-01 -6.26150191e-01 -1.15930164e+00 -6.29513618e-03 -9.92605835e-02 -1.00141957e-01 -6.10230505e-01 1.50710225e-01 -4.17106867e-01 -1.10574804e-01 1.14500940e+00 6.17683113e-01 2.92135537e-01 -9.53039110e-01 -4.96150792e-01 7.00881004e-01 -1.91865966e-01 -3.66304487e-01 1.44391751e+00 -4.15454060e-01 -3.42283607e-01 6.22781456e-01 1.29913354e+00 -1.97234541e-01 -1.31866562e+00 -2.51297385e-01 3.47783081e-02 -5.01480043e-01 6.03354096e-01 -5.44729531e-01 -8.35212946e-01 8.03826690e-01 9.36514735e-01 -1.35123402e-01 1.02651763e+00 -3.45417172e-01 8.84719014e-01 -2.67912209e-01 8.04210782e-01 -8.68382692e-01 -7.86220953e-02 1.37388170e-01 8.18242908e-01 -8.09414268e-01 1.97256655e-01 -1.83152910e-02 -2.75061250e-01 1.25104392e+00 -2.58658472e-02 3.97116952e-02 4.38619733e-01 4.26634848e-01 -1.91923603e-01 -4.58396256e-01 -2.48633638e-01 3.26749146e-01 2.94057935e-01 6.69808030e-01 3.62540364e-01 7.90139586e-02 -7.98364758e-01 5.02734520e-02 1.56696573e-01 1.66627467e-01 6.72216535e-01 1.16069543e+00 -2.31303334e-01 -1.15170062e+00 -4.36897695e-01 2.27775902e-01 -7.51947105e-01 2.53873169e-01 3.93602699e-01 6.57591105e-01 -4.24817890e-01 3.23775798e-01 -1.17606439e-01 1.21087074e-01 2.86835611e-01 5.56284040e-02 1.24238348e+00 7.78760435e-03 -4.08120245e-01 1.37416855e-01 -2.96737731e-01 -8.84727299e-01 -8.56014490e-01 -8.06913316e-01 -1.32846189e+00 -3.59585702e-01 -8.64647701e-02 3.20785791e-01 7.59962320e-01 8.72595549e-01 8.57824460e-02 5.59013069e-01 8.30502689e-01 -1.04092109e+00 -5.09998918e-01 -8.27583909e-01 -1.06100833e+00 3.89144719e-01 7.09793866e-01 -9.75054681e-01 -8.45111832e-02 -2.20981151e-01]
[13.51130485534668, -2.4118289947509766]
07fb11cc-b074-4daa-ba60-7cf9c41afdf7
snowflakenet-point-cloud-completion-by
2108.04444
null
https://arxiv.org/abs/2108.04444v2
https://arxiv.org/pdf/2108.04444v2.pdf
SnowflakeNet: Point Cloud Completion by Snowflake Point Deconvolution with Skip-Transformer
Point cloud completion aims to predict a complete shape in high accuracy from its partial observation. However, previous methods usually suffered from discrete nature of point cloud and unstructured prediction of points in local regions, which makes it hard to reveal fine local geometric details on the complete shape. To resolve this issue, we propose SnowflakeNet with Snowflake Point Deconvolution (SPD) to generate the complete point clouds. The SnowflakeNet models the generation of complete point clouds as the snowflake-like growth of points in 3D space, where the child points are progressively generated by splitting their parent points after each SPD. Our insight of revealing detailed geometry is to introduce skip-transformer in SPD to learn point splitting patterns which can fit local regions the best. Skip-transformer leverages attention mechanism to summarize the splitting patterns used in the previous SPD layer to produce the splitting in the current SPD layer. The locally compact and structured point cloud generated by SPD is able to precisely capture the structure characteristic of 3D shape in local patches, which enables the network to predict highly detailed geometries, such as smooth regions, sharp edges and corners. Our experimental results outperform the state-of-the-art point cloud completion methods under widely used benchmarks. Code will be available at https://github.com/AllenXiangX/SnowflakeNet.
['Zhizhong Han', 'Wen Zheng', 'Pengfei Wan', 'Yan-Pei Cao', 'Yu-Shen Liu', 'Xin Wen', 'Peng Xiang']
2021-08-10
null
http://openaccess.thecvf.com//content/ICCV2021/html/Xiang_SnowflakeNet_Point_Cloud_Completion_by_Snowflake_Point_Deconvolution_With_Skip-Transformer_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Xiang_SnowflakeNet_Point_Cloud_Completion_by_Snowflake_Point_Deconvolution_With_Skip-Transformer_ICCV_2021_paper.pdf
iccv-2021-1
['point-cloud-completion']
['computer-vision']
[-2.29567423e-01 1.48000523e-01 1.88255087e-01 -1.40827149e-01 -6.53330684e-01 -5.20840526e-01 5.30420661e-01 6.30113035e-02 4.73646849e-01 4.35626835e-01 1.31067351e-01 -4.84318137e-02 1.36652052e-01 -1.09521794e+00 -1.26246989e+00 -5.82445323e-01 2.19466053e-02 8.25358152e-01 1.89357147e-01 -4.42198038e-01 1.26036093e-01 9.78604674e-01 -1.70280683e+00 4.38060611e-01 1.10856402e+00 9.30936217e-01 5.76381683e-01 3.37835610e-01 -4.70677763e-01 1.03221826e-01 -4.66044806e-02 -2.38874644e-01 6.86674893e-01 3.60369116e-01 -4.68487531e-01 1.94636971e-01 6.19534373e-01 -5.12673557e-01 -2.59009153e-01 8.11851263e-01 -4.63340851e-03 6.11982979e-02 6.98435187e-01 -1.22695208e+00 -1.00126898e+00 2.31850162e-01 -8.77656996e-01 -2.49112800e-01 2.15822488e-01 2.69298375e-01 1.02166176e+00 -1.54115474e+00 5.51737309e-01 1.19408083e+00 1.14541078e+00 2.60177225e-01 -1.15419996e+00 -7.75100827e-01 2.47788191e-01 -3.53546053e-01 -1.59467053e+00 -1.16715766e-01 1.18010294e+00 -5.32755315e-01 1.00594592e+00 2.90654510e-01 9.17242646e-01 6.76001489e-01 1.27251580e-01 7.89916098e-01 6.94545627e-01 1.57707389e-02 3.50126289e-02 -3.23804498e-01 8.09179246e-02 5.21109462e-01 2.38595679e-01 2.64099002e-01 -3.14578593e-01 -2.76558578e-01 1.31659877e+00 7.30236828e-01 -2.16918916e-01 -3.93984318e-01 -1.05562186e+00 6.09462500e-01 1.09547901e+00 -1.14368014e-01 -8.16356361e-01 7.16140270e-02 -1.85492411e-01 -3.42744626e-02 8.20412815e-01 2.45728850e-01 -5.55442989e-01 2.76373804e-01 -1.05368805e+00 6.96514130e-01 3.84466171e-01 1.29789233e+00 1.27466547e+00 3.24999425e-03 8.42358395e-02 5.99163055e-01 3.71043921e-01 6.50138736e-01 -2.31113076e-01 -1.12113142e+00 7.84036934e-01 1.06546569e+00 3.28142673e-01 -9.57007527e-01 -1.92158848e-01 -6.39599383e-01 -1.10410583e+00 4.91540760e-01 -8.77967104e-02 7.02988878e-02 -1.07615232e+00 9.69754636e-01 5.06420016e-01 6.02627337e-01 -1.92922935e-01 1.02534509e+00 1.15812194e+00 1.10300124e+00 -4.74206120e-01 3.60855341e-01 1.04364848e+00 -1.01132619e+00 1.19149514e-01 -1.17079906e-01 2.97273755e-01 -5.31041563e-01 1.07716036e+00 1.58675656e-01 -1.18921030e+00 -6.54307365e-01 -6.68136954e-01 -3.99890989e-01 2.26480630e-03 1.61724631e-02 5.20966053e-01 -2.41688907e-01 -1.08850861e+00 8.71720910e-01 -8.88003707e-01 8.66471678e-02 9.46274817e-01 2.02735052e-01 -3.64343762e-01 -2.35323593e-01 -5.37085116e-01 2.11074919e-01 -5.56389801e-02 2.45792478e-01 -7.97121644e-01 -1.61379254e+00 -8.08060288e-01 2.91821688e-01 -6.28368556e-02 -1.07660341e+00 1.07504046e+00 -6.58278108e-01 -8.75613749e-01 8.82140458e-01 -3.80049288e-01 -3.38893384e-01 3.73048037e-01 -2.26645514e-01 1.96000621e-01 -1.48290157e-01 3.76550913e-01 9.92961287e-01 7.97596753e-01 -1.75053632e+00 -5.40026665e-01 -5.44991016e-01 -2.01399609e-01 4.16447788e-01 4.33917940e-01 -5.50965488e-01 -4.79615957e-01 -6.48650050e-01 5.40488243e-01 -7.37909019e-01 -3.35886240e-01 2.66079843e-01 -4.91588414e-01 -2.16707602e-01 7.23204851e-01 -5.58919549e-01 6.08337045e-01 -2.22852612e+00 6.01787642e-02 9.84029993e-02 5.72601736e-01 -6.64127097e-02 -2.68288583e-01 6.07047021e-01 -1.36401102e-01 1.74639478e-01 -3.59174490e-01 -9.01393116e-01 -1.05451673e-01 3.07585895e-01 -8.31792772e-01 2.92806119e-01 4.35042500e-01 1.11966610e+00 -7.31459558e-01 -7.95751512e-02 2.97621906e-01 6.94408298e-01 -6.34726107e-01 8.38592574e-02 -4.92531002e-01 5.19337952e-01 -7.49112546e-01 1.08512294e+00 1.31964386e+00 -3.87773693e-01 -7.97202945e-01 -1.42967880e-01 -4.10175502e-01 4.19167839e-02 -8.36377382e-01 1.73411465e+00 -3.21382910e-01 2.94698209e-01 2.17268139e-01 -4.28149521e-01 1.31560588e+00 -1.70988496e-02 5.79207301e-01 -3.58848631e-01 -1.23083346e-01 1.99559391e-01 -5.79312861e-01 -6.85757846e-02 6.61790907e-01 -2.57216066e-01 2.73668081e-01 -8.24974850e-03 -4.52125192e-01 -6.13655746e-01 -5.05152285e-01 5.30667752e-02 7.18852103e-01 3.60170394e-01 -3.04707378e-01 -2.14775484e-02 1.39866337e-01 4.38027471e-01 6.53543472e-01 4.92084742e-01 3.13942552e-01 1.36814904e+00 9.45041180e-02 -8.62410367e-01 -1.41175461e+00 -1.23569715e+00 -1.70514762e-01 4.25075173e-01 4.19268936e-01 -2.24428222e-01 -4.07131284e-01 -3.11140418e-01 3.42354715e-01 6.71023011e-01 -6.35495782e-01 1.59626886e-01 -4.46294248e-01 -3.04652810e-01 1.30958006e-01 6.06849194e-01 3.59002322e-01 -1.27841473e+00 -9.70358625e-02 1.19798072e-01 -7.33380094e-02 -8.83184075e-01 -4.99832690e-01 -1.03963375e-01 -1.42970634e+00 -8.34878743e-01 -8.39976907e-01 -7.60676503e-01 9.89304900e-01 7.36637831e-01 1.36343884e+00 3.47385705e-01 1.45173043e-01 -4.55550514e-02 -4.08817470e-01 -4.98714596e-01 -9.82331559e-02 1.24129346e-02 -2.06681252e-01 -4.13998179e-02 3.48702699e-01 -1.07724631e+00 -7.97864377e-01 2.75933623e-01 -7.90858507e-01 4.30228144e-01 6.73939466e-01 6.14631474e-01 1.20848680e+00 6.55645505e-04 4.14426380e-04 -5.76629043e-01 2.40360364e-01 -7.94525683e-01 -6.21905386e-01 -2.13009626e-01 -7.72789642e-02 -1.38555899e-01 6.68860793e-01 -3.96973267e-02 -7.39422381e-01 2.60233968e-01 -2.46892810e-01 -1.30579317e+00 -4.01205838e-01 3.45086008e-01 1.09726675e-01 -2.02928975e-01 5.41069746e-01 6.52784824e-01 3.86382192e-02 -8.76166821e-01 6.36014789e-02 1.82459831e-01 5.26789069e-01 -6.36102498e-01 1.08669031e+00 7.82195032e-01 1.08913504e-01 -8.98546040e-01 -7.53362000e-01 -4.86277401e-01 -8.93487573e-01 -5.73833920e-02 6.16838574e-01 -1.23872602e+00 -4.32766527e-01 4.83974218e-01 -1.42356026e+00 -4.22939330e-01 -5.38348079e-01 -7.51466081e-02 -4.25058722e-01 3.49267870e-01 -3.77169132e-01 -5.83465755e-01 -6.48903072e-01 -8.50174248e-01 1.79832244e+00 1.77031770e-01 1.39841378e-01 -7.09816396e-01 1.52239516e-01 3.66244525e-01 -2.16142014e-02 4.41564709e-01 6.69649720e-01 -1.35936305e-01 -1.25560701e+00 -2.39133686e-01 -2.17792243e-01 2.08744988e-01 -3.93972136e-02 1.02231070e-01 -7.12254345e-01 -2.04859033e-01 -4.58052605e-02 1.10857666e-01 8.90499651e-01 6.30924165e-01 1.29082263e+00 -2.98179895e-01 -5.31922519e-01 1.21032798e+00 1.55872905e+00 -3.59141886e-01 6.79845989e-01 7.92115703e-02 1.04532659e+00 2.96284288e-01 4.51860279e-01 4.91212100e-01 6.36001170e-01 4.56812799e-01 1.02304733e+00 -1.95562214e-01 -1.56309515e-01 -8.07817996e-01 1.10814221e-01 6.53592110e-01 -4.42119896e-01 2.75897328e-02 -1.03456128e+00 7.08739698e-01 -1.66423619e+00 -8.45524311e-01 -6.12581611e-01 1.98268795e+00 2.43733034e-01 -3.23918574e-02 -1.30461633e-01 -1.49965465e-01 6.05661094e-01 1.17594026e-01 -6.74283087e-01 3.49354185e-02 -2.35314369e-01 1.09049611e-01 4.17313665e-01 5.96454680e-01 -6.66274786e-01 1.03303552e+00 5.10955667e+00 7.40855992e-01 -9.92393017e-01 -6.81044385e-02 4.68989968e-01 -1.54311107e-02 -7.84546137e-01 1.65785134e-01 -9.33078885e-01 4.99361813e-01 8.13923404e-02 6.24780916e-02 3.70418459e-01 8.18779230e-01 3.84238809e-01 3.65875602e-01 -8.01260650e-01 1.14595330e+00 -2.90372223e-01 -1.94118190e+00 4.15878803e-01 2.63826013e-01 1.09330976e+00 6.86078131e-01 -5.91214746e-03 1.43660590e-01 3.61582100e-01 -9.82744515e-01 8.25889766e-01 8.68811965e-01 7.02340245e-01 -5.75462162e-01 4.12051678e-01 8.19509983e-01 -1.33647501e+00 1.47413947e-02 -8.08956325e-01 -2.37135246e-01 1.36255875e-01 7.42451549e-01 -8.99478495e-01 7.48259246e-01 9.17805970e-01 1.11963952e+00 -2.92508721e-01 1.30218840e+00 6.33978918e-02 4.96329218e-01 -7.46672630e-01 3.02447200e-01 4.39053714e-01 -7.26972222e-01 1.00340068e+00 6.45136774e-01 7.32200384e-01 3.17693025e-01 1.91777512e-01 1.46084726e+00 -1.58317178e-01 -8.32882598e-02 -7.64631510e-01 3.73210400e-01 6.35498405e-01 1.10001934e+00 -5.39902508e-01 -6.89702183e-02 -3.05541992e-01 5.78895152e-01 4.76786822e-01 2.78063208e-01 -5.05133331e-01 1.66564837e-01 9.09139514e-01 8.40847075e-01 6.92723036e-01 -4.09272015e-01 -7.68515289e-01 -1.15509439e+00 3.42487514e-01 -2.85762370e-01 -8.64841416e-02 -1.48335910e+00 -1.52081859e+00 7.19171345e-01 -1.60968721e-01 -1.67648375e+00 4.18261856e-01 -3.61455977e-01 -1.16046381e+00 1.13889718e+00 -1.50820708e+00 -1.61383760e+00 -7.30440795e-01 5.16966343e-01 8.21511984e-01 8.29017311e-02 4.71915275e-01 -8.82775262e-02 -2.03049973e-01 1.12052716e-01 1.72218621e-01 -1.47388339e-01 4.92647924e-02 -1.05390072e+00 1.05317163e+00 5.96043289e-01 1.00626238e-01 4.14443433e-01 4.76022452e-01 -1.07882524e+00 -1.34158444e+00 -1.43413186e+00 6.57056510e-01 -6.33349001e-01 2.21333861e-01 -3.89365554e-01 -1.33523464e+00 5.83382607e-01 -2.60587543e-01 2.36990899e-01 -6.65291399e-02 -1.57329857e-01 -1.92946225e-01 -1.31151304e-01 -1.01761758e+00 4.86637682e-01 1.20009100e+00 -1.38856724e-01 -7.36586630e-01 5.35006225e-01 9.79128718e-01 -5.92052341e-01 -6.56816661e-01 4.47607189e-01 5.61793745e-02 -1.02999604e+00 1.19151723e+00 -4.31994170e-01 9.77631032e-01 -5.72631538e-01 -1.21737100e-01 -1.34485948e+00 -6.58921301e-01 -4.84878927e-01 -3.68442833e-02 1.01437426e+00 4.02590185e-01 -5.80659628e-01 1.25743616e+00 4.15133625e-01 -8.76548052e-01 -1.16147614e+00 -7.17496634e-01 -4.39976335e-01 4.05127645e-01 -3.94612968e-01 1.33451581e+00 7.62688100e-01 -7.15323508e-01 -3.35544050e-02 -1.34221971e-01 7.28519559e-01 8.40892434e-01 7.28555560e-01 1.03492153e+00 -1.60789287e+00 1.20254189e-01 -3.07346523e-01 -2.28156909e-01 -1.45816672e+00 -1.74131617e-02 -1.12924957e+00 -1.76949188e-01 -1.80544245e+00 -8.41843560e-02 -9.13234651e-01 2.28541225e-01 6.03168607e-01 -1.00447729e-01 3.28635454e-01 2.71142155e-01 6.40357494e-01 -6.18689023e-02 1.01243269e+00 1.63259470e+00 -1.42816558e-01 -4.56137538e-01 3.93489659e-01 -7.58265436e-01 7.10650921e-01 6.69609010e-01 -4.67992991e-01 -1.17150918e-01 -8.37091029e-01 4.67686206e-01 1.58697203e-01 8.04255664e-01 -9.83654022e-01 2.59007007e-01 -1.51205406e-01 5.95324934e-01 -1.44556236e+00 6.93441272e-01 -9.93024230e-01 4.82822925e-01 -1.56161319e-02 2.32936755e-01 5.17166173e-03 2.11361542e-01 5.98471761e-01 -1.26661763e-01 1.96027700e-02 3.86530817e-01 -3.18315476e-01 -4.82298553e-01 1.07330287e+00 2.80889541e-01 -2.56869972e-01 8.39019179e-01 -6.96001828e-01 -5.13738394e-02 -2.24013805e-01 -7.19911993e-01 5.83579063e-01 1.05436385e+00 4.82452124e-01 1.02701271e+00 -1.40834010e+00 -1.02700150e+00 6.64496303e-01 4.18121926e-02 1.27514648e+00 6.39411330e-01 5.11617661e-01 -8.77717376e-01 9.61674079e-02 -1.91337496e-01 -9.74258065e-01 -9.33846474e-01 4.15858656e-01 4.27431405e-01 1.46012917e-01 -1.31504893e+00 8.61383438e-01 6.23064995e-01 -7.30875373e-01 -2.11734965e-01 -7.69082606e-01 1.92498192e-01 -4.17384267e-01 1.01306424e-01 2.07833216e-01 1.01771675e-01 -7.20553279e-01 8.64197761e-02 1.03956461e+00 9.46448594e-02 4.31842566e-01 1.89575016e+00 -1.08509362e-02 -1.61293715e-01 1.73442543e-01 8.52061450e-01 6.85499832e-02 -1.83668351e+00 -2.87012398e-01 -7.34294236e-01 -6.62545979e-01 1.13627523e-01 -4.63982999e-01 -1.16123235e+00 8.77147853e-01 -7.36567599e-05 5.16679399e-02 8.54965329e-01 2.22948477e-01 1.05124092e+00 9.11661908e-02 6.25567079e-01 -3.33727241e-01 -4.00685877e-01 5.70810020e-01 1.48316681e+00 -9.94645238e-01 2.00379007e-02 -8.11849058e-01 -5.52188694e-01 1.01012850e+00 5.91669321e-01 -7.20758438e-01 8.58704925e-01 1.39576405e-01 -4.16304022e-01 -6.53282404e-01 -7.26879299e-01 1.89456448e-01 4.33633834e-01 6.36685252e-01 -2.69279361e-01 2.54575729e-01 5.76085865e-01 8.13003004e-01 -5.58818102e-01 -2.39208024e-02 2.99134523e-01 5.98005295e-01 -4.35528636e-01 -7.48634219e-01 -8.00612509e-01 7.76549160e-01 2.22879007e-01 -1.25755861e-01 -4.22085285e-01 5.08815885e-01 3.05363119e-01 3.05165738e-01 5.49172938e-01 -2.79068440e-01 5.44354796e-01 -1.84676662e-01 1.39638305e-01 -7.29042292e-01 -4.41695005e-01 -4.36082669e-03 -4.53782052e-01 -5.10746241e-01 6.37580305e-02 -9.05282855e-01 -1.35265934e+00 -4.17336881e-01 -1.02773746e-02 -2.47399658e-02 3.95983458e-01 5.96576571e-01 8.99182737e-01 1.48823902e-01 7.30340183e-01 -1.61307287e+00 -2.29240328e-01 -8.49889517e-01 -6.95922792e-01 3.59829187e-01 5.49122572e-01 -6.53603196e-01 -3.49735886e-01 -8.60653445e-02]
[8.380166053771973, -3.611647129058838]
28ec7d3d-3a30-4a18-91a0-214771c6cb6f
penalizing-proposals-using-classifiers-for
2205.13219
null
https://arxiv.org/abs/2205.13219v2
https://arxiv.org/pdf/2205.13219v2.pdf
Penalizing Proposals using Classifiers for Semi-Supervised Object Detection
Obtaining gold standard annotated data for object detection is often costly, involving human-level effort. Semi-supervised object detection algorithms solve the problem with a small amount of gold-standard labels and a large unlabelled dataset used to generate silver-standard labels. But training on the silver standard labels does not produce good results, because they are machine-generated annotations. In this work, we design a modified loss function to train on large silver standard annotated sets generated by a weak annotator. We include a confidence metric associated with the annotation as an additional term in the loss function, signifying the quality of the annotation. We test the effectiveness of our approach on various test sets and use numerous variations to compare the results with some of the current approaches to object detection. In comparison with the baseline where no confidence metric is used, we achieved a 4% gain in mAP with 25% labeled data and 10% gain in mAP with 50% labeled data by using the proposed confidence metric.
['Pallab Dasgupta', 'Somnath Hazra']
2022-05-26
null
null
null
null
['semi-supervised-object-detection']
['computer-vision']
[ 4.59415287e-01 5.15520334e-01 1.20077170e-01 -6.72013760e-01 -1.09584951e+00 -6.59210682e-01 5.87904871e-01 3.13494802e-01 -8.83887708e-01 8.26590896e-01 -2.25625917e-01 7.96139464e-02 3.76974881e-01 -4.95506823e-01 -7.16926098e-01 -6.40330434e-01 2.52363503e-01 6.73334956e-01 8.54545534e-01 3.07877243e-01 2.15851143e-01 8.16059113e-02 -1.64210546e+00 2.79902935e-01 6.50801837e-01 1.11461210e+00 3.56842428e-01 6.52877331e-01 -2.43551955e-02 5.43374658e-01 -9.40058172e-01 -5.30097783e-01 5.72317481e-01 -4.97415036e-01 -1.04807568e+00 2.57059723e-01 6.21096611e-01 8.70195106e-02 4.20655012e-01 1.03010416e+00 5.38734138e-01 -2.02387512e-01 7.44084597e-01 -1.19703662e+00 -2.37044737e-01 5.86353660e-01 -7.58205712e-01 1.09484028e-02 1.68093741e-01 5.20372987e-02 1.08830917e+00 -1.01731706e+00 6.47292316e-01 1.05871093e+00 1.02235961e+00 4.52161282e-01 -1.45044851e+00 -5.89724183e-01 -1.36959955e-01 -2.72052974e-01 -1.57608044e+00 -3.37198913e-01 1.80561960e-01 -5.10247886e-01 4.35404241e-01 2.62167186e-01 2.14085057e-01 6.70352399e-01 -4.21747208e-01 6.77184463e-01 1.08785880e+00 -9.00958061e-01 2.70853013e-01 6.34192288e-01 1.55072540e-01 7.13971376e-01 5.50407588e-01 -2.06516962e-02 -1.12506770e-01 -1.40102878e-01 2.78009206e-01 -4.62813228e-01 -8.30700062e-03 -4.58584309e-01 -1.05259299e+00 8.75351429e-01 5.60565352e-01 4.67241257e-02 -1.72641188e-01 1.00584909e-01 4.01569605e-01 1.14497378e-01 7.04396009e-01 7.06605852e-01 -3.82184178e-01 1.68024510e-01 -9.64868486e-01 -1.49950637e-02 8.12577248e-01 9.77511406e-01 7.80084670e-01 -2.14494124e-01 -4.77358460e-01 8.15354168e-01 4.42673355e-01 3.38045269e-01 2.84641474e-01 -9.73353088e-01 2.48310864e-01 7.50993788e-01 5.55841923e-01 -5.05358338e-01 -2.73563623e-01 -3.81555587e-01 -1.95903271e-01 6.29203498e-01 9.33281720e-01 -3.61943305e-01 -1.16415393e+00 1.66420281e+00 4.28042769e-01 -3.26634273e-02 -4.53784466e-02 7.90783763e-01 5.63954890e-01 3.84996831e-01 2.29496822e-01 -6.90984866e-03 1.21966815e+00 -1.14730787e+00 -3.84730041e-01 -4.06578690e-01 8.82623553e-01 -7.83105195e-01 1.19659293e+00 2.16386527e-01 -6.74807250e-01 -5.88555455e-01 -1.12995708e+00 2.35811412e-01 -3.62876624e-01 4.28611875e-01 3.01014185e-01 7.74224281e-01 -9.56949949e-01 3.90890479e-01 -5.63148201e-01 -5.97184420e-01 5.14505446e-01 3.10623765e-01 -4.16279078e-01 7.68036917e-02 -8.04531336e-01 1.18652248e+00 7.23364353e-01 -1.99962661e-01 -9.93747950e-01 -3.29333365e-01 -6.73068702e-01 -2.13148743e-01 5.88523388e-01 -2.06836313e-01 1.49890637e+00 -9.92663145e-01 -9.18383837e-01 1.38922739e+00 1.84006885e-01 -4.61793005e-01 8.55638564e-01 -2.82988250e-01 -2.87810527e-03 -6.07771762e-02 5.04077673e-01 9.52253461e-01 5.76495588e-01 -1.46883786e+00 -1.12279236e+00 -9.20626372e-02 -1.47462208e-02 5.68016507e-02 -1.35991931e-01 3.09541702e-01 -4.00026083e-01 -4.07014489e-01 -2.45238319e-01 -1.00102961e+00 -3.76243204e-01 3.50610703e-01 -3.08944046e-01 -3.57635975e-01 7.59022593e-01 -2.64246792e-01 9.04730976e-01 -2.19859505e+00 -4.37959701e-01 1.53342173e-01 6.89407066e-02 5.91625154e-01 -1.53464749e-01 -3.35192978e-02 1.79872885e-01 1.73637345e-01 -3.48604083e-01 -4.90946561e-01 -7.72181451e-02 1.79339454e-01 3.86644714e-02 2.78745055e-01 5.10333419e-01 6.50827825e-01 -1.19241655e+00 -6.86849654e-01 -3.70069668e-02 3.28421146e-01 -2.46511728e-01 2.40023285e-01 -2.77782589e-01 4.90481704e-02 -1.87257960e-01 4.66338307e-01 3.77509266e-01 -4.94591653e-01 -3.75937894e-02 -1.39100015e-01 3.85348313e-02 -2.69045327e-02 -1.44877183e+00 1.38024604e+00 -1.82237878e-01 6.56577110e-01 3.12522352e-02 -7.24038124e-01 1.09767222e+00 1.76491976e-01 2.47097667e-02 -2.90186018e-01 1.19256236e-01 4.07498300e-01 -3.16863507e-02 -1.75748631e-01 4.35451657e-01 -1.27931191e-02 -2.05120631e-03 6.13007545e-01 2.22846985e-01 -2.44488642e-01 4.23112392e-01 2.15485841e-01 1.31195843e+00 1.47704512e-01 3.01816016e-01 -2.79367417e-01 3.43604177e-01 2.73177356e-01 4.17570233e-01 9.79288757e-01 -2.69215882e-01 7.82594740e-01 3.50926846e-01 -4.52691823e-01 -1.23708606e+00 -6.48401260e-01 -2.50061303e-01 1.33668613e+00 -4.16741893e-03 -3.15870136e-01 -9.65295315e-01 -1.21267343e+00 -1.23328052e-01 6.01927519e-01 -7.56369829e-01 -3.72280777e-02 -2.94529766e-01 -1.11245692e+00 7.58960903e-01 6.51545525e-01 4.78562891e-01 -1.16424525e+00 -8.43045354e-01 3.51460949e-02 -4.02047820e-02 -1.08425808e+00 -3.48795146e-01 6.37400091e-01 -4.70661014e-01 -1.11973715e+00 -6.48910701e-01 -8.35498571e-01 9.89587903e-01 1.06402576e-01 1.31213236e+00 1.40074283e-01 -5.64529955e-01 -6.16250522e-02 -5.25842965e-01 -8.78165245e-01 -4.84281123e-01 3.75267230e-02 -7.77758211e-02 -9.74799097e-02 4.69183862e-01 2.64535695e-01 -3.39262366e-01 7.45414972e-01 -8.25562119e-01 -3.32020283e-01 5.58391213e-01 8.77531707e-01 6.49903774e-01 -1.12655170e-01 6.52822971e-01 -1.16567361e+00 2.53071427e-01 -2.27436453e-01 -9.05330062e-01 2.49084517e-01 -8.93729687e-01 2.49885246e-01 1.29155755e-01 -5.62741339e-01 -1.05842388e+00 5.54387867e-01 5.41571938e-02 4.43423130e-02 -1.21724226e-01 9.43462104e-02 -6.54709116e-02 -2.64174372e-01 1.26088274e+00 -3.48754108e-01 -1.30451292e-01 -4.66491580e-01 3.43440890e-01 7.04565525e-01 5.27761161e-01 -4.09604818e-01 7.98133671e-01 3.83657455e-01 -2.88172126e-01 -1.43175364e-01 -1.47407818e+00 -6.00500107e-01 -6.56707168e-01 -6.67822734e-02 7.51072407e-01 -7.00982094e-01 -1.05233872e-02 3.09307247e-01 -1.14852691e+00 -4.80373025e-01 -6.52295291e-01 4.15139884e-01 -3.99728656e-01 1.85002655e-01 -2.49769360e-01 -9.55755949e-01 -2.81572580e-01 -9.47573900e-01 1.27967095e+00 9.78252664e-02 -3.36474597e-01 -5.86179733e-01 1.11881539e-01 1.90914541e-01 2.02912226e-01 3.72837842e-01 1.78914621e-01 -9.78167534e-01 -1.24942958e-01 -7.17187226e-01 -6.40745521e-01 6.95204973e-01 3.84861305e-02 -7.28037953e-02 -1.22885752e+00 -1.65914595e-01 -3.66152406e-01 -7.67911255e-01 8.62831116e-01 6.69004768e-02 8.26158643e-01 5.26141783e-04 -5.19626856e-01 1.18576422e-01 1.32277632e+00 -7.85251800e-03 5.32310486e-01 4.84284103e-01 5.49826145e-01 7.24548578e-01 1.03264201e+00 2.82055944e-01 4.02098075e-02 6.14536762e-01 4.12987530e-01 -3.61609280e-01 -1.43206075e-01 -1.70705840e-01 7.95802027e-02 -1.28247544e-01 -1.22651257e-01 -4.68561828e-01 -1.11311495e+00 5.41117132e-01 -1.85655630e+00 -7.24265993e-01 -3.95196259e-01 2.37809682e+00 1.14273691e+00 5.47056556e-01 3.21777761e-01 3.19238394e-01 9.58646715e-01 -4.03212398e-01 -2.82760471e-01 1.00347072e-01 5.14124744e-02 3.12744975e-02 6.49437189e-01 4.54876065e-01 -1.37823093e+00 8.99571180e-01 7.37148046e+00 7.59532213e-01 -9.15464342e-01 2.75185823e-01 4.91487145e-01 4.97448780e-02 3.37819964e-01 -4.90130410e-02 -1.04848874e+00 6.23267710e-01 9.32622313e-01 1.23389520e-01 -2.68389225e-01 1.11895812e+00 3.72289796e-03 -4.07070547e-01 -1.18889439e+00 7.52890468e-01 1.85928941e-01 -8.80551934e-01 -3.67155194e-01 -7.84689263e-02 8.28629494e-01 1.05961986e-01 -4.62165534e-01 1.51746482e-01 7.81249940e-01 -7.77872801e-01 1.04110479e+00 1.47554442e-01 8.64884913e-01 -4.97525394e-01 1.22339082e+00 2.97281146e-01 -9.76239085e-01 2.37990811e-01 -4.25290465e-01 4.93857749e-02 1.46607429e-01 5.23389339e-01 -1.27851748e+00 1.19992822e-01 7.65614629e-01 6.01210333e-02 -1.10550320e+00 1.43776941e+00 -4.86599892e-01 7.10472882e-01 -5.18625021e-01 -2.89055128e-02 3.21898550e-01 3.50133419e-01 2.71640092e-01 1.62102962e+00 9.35861543e-02 -3.57108682e-01 4.16822404e-01 6.87166631e-01 -1.27670839e-01 5.40944785e-02 -4.82957125e-01 1.77692875e-01 4.40806866e-01 1.42787540e+00 -1.11551249e+00 -5.51264584e-01 -2.76015967e-01 9.16787505e-01 3.55731934e-01 -2.00695433e-02 -7.25720942e-01 -6.82467222e-01 -1.79560930e-01 2.51359403e-01 2.42711008e-01 3.79423529e-01 -3.30867767e-01 -6.40616775e-01 1.44787788e-01 -5.16600311e-01 5.79621494e-01 -7.68252254e-01 -1.30891681e+00 8.19159210e-01 1.25963569e-01 -1.18214118e+00 -3.16655368e-01 -6.47640705e-01 -3.90637577e-01 8.68430793e-01 -1.13001168e+00 -1.12097740e+00 -5.59379280e-01 3.25965956e-02 3.42261076e-01 8.92517343e-02 8.57701540e-01 4.26837951e-01 -2.99187064e-01 6.97885334e-01 -1.76184863e-01 3.55537802e-01 9.96137798e-01 -1.63843346e+00 2.60525137e-01 9.10126507e-01 2.03746557e-01 1.74480036e-01 9.17518914e-01 -4.44255620e-01 -3.59701008e-01 -1.31544495e+00 8.35475862e-01 -8.74662578e-01 5.38687229e-01 -4.06468302e-01 -9.60683107e-01 6.33958936e-01 -3.43006141e-02 4.37195212e-01 6.28845513e-01 -6.52661100e-02 -5.33867836e-01 8.31688643e-02 -1.52637017e+00 2.08135433e-02 9.44049001e-01 -1.07271343e-01 -5.15920937e-01 4.60740447e-01 6.42700136e-01 -1.46200910e-01 -4.83049661e-01 5.56922615e-01 4.91070718e-01 -4.75985676e-01 6.50440097e-01 -4.01339203e-01 5.47994971e-02 -8.50902319e-01 9.51212272e-03 -1.20000446e+00 -1.85402766e-01 -1.20390035e-01 2.96675593e-01 1.40337121e+00 9.39597487e-01 -3.64442170e-01 8.58406842e-01 6.71561420e-01 -5.59540093e-02 -3.84013861e-01 -5.12405694e-01 -9.66943085e-01 -3.41841161e-01 -3.05808216e-01 1.43191487e-01 7.75357008e-01 -2.52003640e-01 6.12503052e-01 -2.16730237e-01 -2.94651301e-03 7.07085431e-01 -2.49825358e-01 9.22919750e-01 -1.57319379e+00 -2.18938321e-01 -2.37445116e-01 -5.70511639e-01 -4.79499578e-01 -8.57372880e-02 -7.96523094e-01 6.59694612e-01 -1.56837153e+00 4.31105852e-01 -6.58239305e-01 -1.75072610e-01 8.68789375e-01 -4.00543302e-01 9.62880790e-01 5.18486984e-02 9.67404097e-02 -1.09956348e+00 8.76131654e-02 7.21940160e-01 -1.21526137e-01 -9.35667977e-02 -9.53908488e-02 -6.93655968e-01 9.67153251e-01 6.60378039e-01 -8.62002492e-01 -1.74307421e-01 -4.10013705e-01 1.14835702e-01 -7.62122810e-01 3.33231688e-01 -1.08532238e+00 4.32252511e-02 9.33284536e-02 4.14298922e-01 -2.77473301e-01 4.79610972e-02 -7.71724463e-01 7.93650225e-02 4.26264018e-01 -4.45211142e-01 -2.36927286e-01 1.82358697e-02 5.40509999e-01 -1.42855003e-01 -6.54752851e-01 1.10432935e+00 -1.59041137e-01 -7.65003145e-01 -1.67786583e-01 5.88013325e-03 1.52739927e-01 1.25488222e+00 -3.25118363e-01 -1.50323033e-01 5.22835292e-02 -7.25332737e-01 3.12343359e-01 5.10202825e-01 2.43319467e-01 5.91105521e-02 -1.18006301e+00 -9.52181756e-01 -1.49962634e-01 5.64200878e-01 2.50757009e-01 -7.05584943e-01 4.38429594e-01 -5.65352559e-01 -1.44742383e-02 -9.72533226e-03 -7.77485669e-01 -1.32760024e+00 3.57096702e-01 1.65262237e-01 -3.19668621e-01 -3.26405048e-01 1.09876549e+00 8.20078049e-03 -5.09372711e-01 5.55285454e-01 -4.97137569e-02 -8.02200884e-02 2.51194835e-01 7.22011268e-01 3.97037208e-01 7.81722963e-02 -4.22990888e-01 -4.47940558e-01 3.16680044e-01 -1.63614452e-02 -2.52141833e-01 1.08184862e+00 9.41721275e-02 1.78493947e-01 4.73872721e-01 7.84561157e-01 -1.43622145e-01 -1.27712131e+00 -3.75176400e-01 5.78959823e-01 -4.84152496e-01 -1.17892735e-01 -9.90116954e-01 -7.96661973e-01 4.85718489e-01 9.13691461e-01 2.86841869e-01 6.94666088e-01 3.68200064e-01 5.29208705e-02 5.20917475e-01 5.92516243e-01 -1.13533747e+00 2.78949261e-01 3.05263042e-01 6.63696229e-01 -1.73298275e+00 -5.92597276e-02 -4.16692376e-01 -7.56380558e-01 6.35013759e-01 9.57653463e-01 -1.06850743e-01 2.78132796e-01 4.20506179e-01 3.36577654e-01 -2.05877379e-01 -3.92434418e-01 -4.85093802e-01 2.89592326e-01 5.86301982e-01 5.05796969e-01 -3.35397050e-02 -5.23214638e-01 3.62859517e-01 6.12528622e-02 4.41364236e-02 2.93275595e-01 1.10367644e+00 -8.40044200e-01 -1.05871522e+00 -5.17755568e-01 6.47547603e-01 -6.37271404e-01 5.11957221e-02 -6.40779138e-01 6.23309493e-01 4.37022299e-01 1.02601683e+00 1.40608475e-02 -2.58953720e-01 4.05937195e-01 2.28970483e-01 2.03262448e-01 -1.16147852e+00 -4.03588355e-01 -1.01174489e-01 3.00590396e-01 -2.69681692e-01 -6.61479235e-01 -4.78042483e-01 -1.22122943e+00 3.22790056e-01 -9.87846851e-01 4.04716402e-01 5.95628083e-01 7.32828557e-01 7.16581270e-02 2.07756370e-01 3.84096652e-01 -8.94211411e-01 -6.29207134e-01 -1.33271837e+00 -3.25461417e-01 6.26477897e-01 1.11820973e-01 -7.72580206e-01 -5.24879515e-01 3.65295440e-01]
[9.261161804199219, 1.2303614616394043]
3099bbc3-feb7-4165-b676-df50f2006a9e
fedcp-separating-feature-information-for
2307.01217
null
https://arxiv.org/abs/2307.01217v1
https://arxiv.org/pdf/2307.01217v1.pdf
FedCP: Separating Feature Information for Personalized Federated Learning via Conditional Policy
Recently, personalized federated learning (pFL) has attracted increasing attention in privacy protection, collaborative learning, and tackling statistical heterogeneity among clients, e.g., hospitals, mobile smartphones, etc. Most existing pFL methods focus on exploiting the global information and personalized information in the client-level model parameters while neglecting that data is the source of these two kinds of information. To address this, we propose the Federated Conditional Policy (FedCP) method, which generates a conditional policy for each sample to separate the global information and personalized information in its features and then processes them by a global head and a personalized head, respectively. FedCP is more fine-grained to consider personalization in a sample-specific manner than existing pFL methods. Extensive experiments in computer vision and natural language processing domains show that FedCP outperforms eleven state-of-the-art methods by up to 6.69%. Furthermore, FedCP maintains its superiority when some clients accidentally drop out, which frequently happens in mobile settings. Our code is public at https://github.com/TsingZ0/FedCP.
['Haibing Guan', 'Ruhui Ma', 'Zhengui Xue', 'Tao Song', 'Hao Wang', 'Yang Hua', 'Jianqing Zhang']
2023-07-01
null
null
null
null
['federated-learning', 'personalized-federated-learning']
['methodology', 'methodology']
[-2.21771255e-01 -1.00155197e-01 -5.66448867e-01 -7.43986547e-01 -8.89256239e-01 -3.28393102e-01 5.06052673e-01 4.28476110e-02 -2.16705739e-01 7.27006137e-01 3.94114345e-01 -2.35938802e-01 -1.13898307e-01 -6.71924531e-01 -5.96586406e-01 -8.73853743e-01 7.15486109e-02 3.48194242e-01 1.08486257e-01 3.90737355e-01 -2.45151483e-02 2.05277562e-01 -1.36873055e+00 5.76215148e-01 9.04381216e-01 1.06662321e+00 3.75166386e-02 2.76791841e-01 -3.29095125e-01 6.83330894e-01 -1.81410551e-01 -8.39094579e-01 1.80994764e-01 -1.75757054e-02 -7.04199135e-01 -6.77224919e-02 2.30560422e-01 -4.83185828e-01 -3.20269525e-01 1.14664388e+00 5.26176929e-01 2.09657513e-02 2.48340726e-01 -1.37307727e+00 -5.45399785e-01 8.61582041e-01 -4.72371697e-01 -1.66612193e-02 2.38021255e-01 2.20937490e-01 9.24600899e-01 -5.86327314e-01 3.55464876e-01 1.27411520e+00 5.15964746e-01 6.77853584e-01 -1.06223941e+00 -8.78714800e-01 5.85044622e-01 3.32219154e-01 -1.15698087e+00 -5.68968773e-01 7.55086780e-01 -2.69168764e-01 2.61276484e-01 5.59624434e-01 1.29504368e-01 1.38206220e+00 -5.71933249e-03 1.34458148e+00 9.82880294e-01 4.48609404e-02 3.09139788e-01 4.73352462e-01 3.91888738e-01 4.00152117e-01 1.36482581e-01 -3.31782512e-02 -4.47345853e-01 -8.52704227e-01 2.11084664e-01 6.81223512e-01 -3.70808691e-01 -3.30088228e-01 -8.45411181e-01 6.34114027e-01 -8.87854323e-02 9.09365416e-02 -5.05902350e-01 -3.16326737e-01 4.26585346e-01 9.35512483e-02 4.14852470e-01 -3.94546598e-01 -9.37905550e-01 -3.63970809e-02 -5.82039356e-01 3.39719325e-01 1.04594135e+00 1.17589414e+00 1.06250727e+00 -6.14589691e-01 -6.44182265e-01 8.44827235e-01 3.56632024e-01 2.87826359e-01 8.51182103e-01 -1.07769573e+00 5.02639890e-01 6.18031263e-01 1.69229150e-01 -9.01293457e-01 2.34895907e-02 -4.88032907e-01 -7.59420693e-01 -4.91431415e-01 1.98686063e-01 -4.23712134e-01 -2.92964399e-01 1.79755092e+00 7.79730678e-01 5.70243061e-01 -1.32907212e-01 6.56878710e-01 6.32090807e-01 3.20538908e-01 3.07282656e-01 -4.83839601e-01 1.36406624e+00 -1.01423776e+00 -5.51705182e-01 9.18371975e-02 4.97014105e-01 -5.38797736e-01 9.45538282e-01 4.80302036e-01 -6.25715852e-01 -2.45214198e-02 -3.43805552e-01 2.24958986e-01 -2.10588634e-01 -5.32617308e-02 6.31191969e-01 8.86074007e-01 -6.04844928e-01 6.78244293e-01 -7.74708927e-01 -3.61337692e-01 8.41275692e-01 3.11265230e-01 -3.48453790e-01 -2.71028101e-01 -9.99713421e-01 -1.67676751e-02 -1.36228194e-02 -4.73580182e-01 -7.02606022e-01 -9.15341854e-01 -3.79035205e-01 2.73900479e-01 6.58969939e-01 -7.30385900e-01 1.48358822e+00 -8.19040418e-01 -1.55289364e+00 5.46771705e-01 -4.04428989e-01 -2.75030971e-01 9.17511106e-01 -2.84626275e-01 -5.16361356e-01 -2.50796139e-01 6.50631189e-02 -2.46590033e-01 8.90685976e-01 -1.23190916e+00 -1.08480966e+00 -7.02416122e-01 -2.13127971e-01 -3.54735628e-02 -7.25992501e-01 9.02040862e-03 -9.84648049e-01 -4.25006598e-01 -2.00347900e-01 -7.68451333e-01 -2.60768324e-01 -1.24798164e-01 -5.96498728e-01 -5.11917710e-01 1.08179045e+00 -5.17383516e-01 1.36672056e+00 -2.34622598e+00 -3.93333912e-01 2.24189028e-01 3.01444203e-01 4.08049941e-01 2.53044404e-02 5.38026214e-01 2.63147533e-01 9.42713171e-02 -1.20154023e-01 -7.70033598e-01 2.30969682e-01 3.03905606e-01 -2.58837521e-01 5.42019546e-01 -5.44220746e-01 5.12982368e-01 -8.83875728e-01 -6.72120988e-01 -7.84010813e-02 5.53550303e-01 -7.78802693e-01 4.83940065e-01 -2.42673457e-01 6.12631023e-01 -1.05937779e+00 7.72204459e-01 1.11716163e+00 -2.68426985e-01 5.62779665e-01 -3.08707803e-02 2.57151172e-04 2.58176714e-01 -1.11473644e+00 1.42999005e+00 -2.59101421e-01 -2.11638689e-01 5.12274802e-01 -7.74296224e-01 6.41970992e-01 4.22344416e-01 7.59293020e-01 -4.17956561e-01 2.82807708e-01 4.38624993e-02 -7.23340750e-01 -6.50575757e-01 2.55258411e-01 1.84186369e-01 6.50568977e-02 6.40355825e-01 -1.12598844e-01 1.00168169e+00 -2.39012972e-01 2.47431040e-01 1.13121676e+00 -1.92895532e-01 1.36780500e-01 -1.68397799e-01 8.38489115e-01 -8.02306950e-01 1.25791466e+00 9.09654200e-01 -6.83633685e-01 5.92638791e-01 4.23394650e-01 -2.09619761e-01 -1.94572777e-01 -8.81562769e-01 4.81387526e-02 1.46652830e+00 1.12280793e-01 -6.16984785e-01 -6.77515805e-01 -1.30371499e+00 4.64491457e-01 7.06197798e-01 -3.96728873e-01 -1.61045939e-01 -2.48483688e-01 -6.44348800e-01 2.57242829e-01 1.66541170e-02 5.67275941e-01 -8.05899560e-01 -3.70465592e-02 2.87407130e-01 -3.64569783e-01 -9.15112674e-01 -8.20946515e-01 -2.82727361e-01 -6.91264927e-01 -1.12894070e+00 -4.72325236e-01 -3.97438109e-01 4.74903315e-01 3.14557374e-01 7.51833856e-01 -1.26970321e-01 2.11881846e-02 4.96415377e-01 -4.29008871e-01 -2.57909834e-01 -1.74137708e-02 3.61485660e-01 6.72200918e-02 9.40043688e-01 6.11528635e-01 -7.89452910e-01 -7.57047474e-01 4.43662882e-01 -7.94886589e-01 -4.19366807e-01 4.23311114e-01 7.69509017e-01 4.92452711e-01 6.74884841e-02 4.92789149e-01 -1.69034827e+00 6.72742486e-01 -1.03017521e+00 -3.02654177e-01 3.69485021e-01 -9.06336963e-01 -1.60084084e-01 6.94358587e-01 -5.49062669e-01 -1.45317400e+00 9.23090726e-02 -2.08125964e-01 -4.39366788e-01 -3.98708940e-01 9.52240378e-02 -8.08727920e-01 1.82668492e-01 2.55011439e-01 2.77119696e-01 -1.37925968e-02 -1.08138227e+00 2.21269637e-01 1.09861112e+00 3.61599892e-01 -8.54297161e-01 5.60572386e-01 4.08054382e-01 -5.11991560e-01 -3.04504007e-01 -6.75010562e-01 -5.24576485e-01 2.27325223e-03 1.59091190e-01 2.38952518e-01 -7.41853952e-01 -1.12310815e+00 5.95528305e-01 -8.18388164e-01 -6.47330731e-02 -1.90679014e-01 4.07237887e-01 -3.59401524e-01 5.11688232e-01 -6.28339112e-01 -9.44349468e-01 -5.57505012e-01 -1.02443373e+00 7.76486099e-01 4.42896694e-01 2.15807736e-01 -8.02241385e-01 9.09209251e-03 4.21788812e-01 5.10896444e-01 -7.93646872e-02 6.90059066e-01 -1.02510142e+00 -5.88645160e-01 -3.16439956e-01 -1.29527465e-01 3.03940356e-01 4.00830328e-01 -1.16782069e-01 -1.13341451e+00 -4.29783225e-01 2.12484628e-01 1.04391068e-01 6.04618073e-01 1.37992397e-01 1.68171895e+00 -1.05150187e+00 -6.07001305e-01 7.56673694e-01 1.19644058e+00 8.04126486e-02 2.48696685e-01 6.87006041e-02 6.14141941e-01 7.53209174e-01 4.52984869e-01 9.89982069e-01 7.47108519e-01 4.77403820e-01 5.20903230e-01 3.99383843e-01 3.19483638e-01 -4.84808713e-01 3.14956158e-01 4.79663968e-01 3.19918931e-01 -1.67664215e-01 -4.63861018e-01 4.61519331e-01 -2.25121641e+00 -1.00820220e+00 1.28981858e-01 2.38817763e+00 8.45055580e-01 -2.97329605e-01 1.92203149e-01 -2.18996286e-01 9.87376511e-01 1.73304364e-01 -8.70039821e-01 -1.74111903e-01 9.21994448e-02 -3.51689786e-01 7.04852045e-01 2.45421931e-01 -1.16105628e+00 7.58743823e-01 4.90811300e+00 1.04624569e+00 -1.00281060e+00 4.68586892e-01 9.04376626e-01 -8.55710730e-02 -4.86404747e-01 -3.02292295e-02 -9.06037867e-01 9.79227245e-01 7.72764862e-01 -4.46804315e-01 5.10889769e-01 1.16148376e+00 8.69266093e-02 2.98697412e-01 -9.75290239e-01 1.13264704e+00 -2.28512064e-01 -1.07566595e+00 8.45976267e-03 1.97732762e-01 5.32573760e-01 1.27947211e-01 3.27086687e-01 3.93403441e-01 5.13174832e-01 -5.15264452e-01 3.19975823e-01 8.15389931e-01 3.14939678e-01 -8.42234015e-01 4.29876238e-01 6.76326334e-01 -9.30684030e-01 -3.94428730e-01 -2.51118451e-01 3.37560892e-01 -1.55411005e-01 7.87354529e-01 -5.07832527e-01 7.51571894e-01 1.11302817e+00 7.38887191e-01 -4.47949260e-01 1.03163457e+00 1.33902833e-01 8.93384159e-01 -2.56310642e-01 2.41380066e-01 -1.90040588e-01 -1.97384879e-01 5.41833103e-01 1.15700674e+00 1.75623581e-01 5.91417588e-02 2.41870180e-01 3.59711587e-01 -3.72631878e-01 4.15219814e-01 -7.60088041e-02 1.60761535e-01 6.57260418e-01 1.35898042e+00 5.07933535e-02 -3.97436351e-01 -6.00043297e-01 9.82096016e-01 4.26926047e-01 5.08143246e-01 -6.82957053e-01 -2.64471531e-01 1.22537410e+00 1.42311469e-01 4.07369524e-01 3.37457627e-01 6.46992922e-02 -1.44397831e+00 1.76128626e-01 -9.30218637e-01 8.79329026e-01 -9.63710845e-02 -1.76293337e+00 5.09754539e-01 -3.62271339e-01 -1.08036649e+00 -6.77036494e-02 -1.45903572e-01 -7.27782786e-01 7.53907442e-01 -1.42507720e+00 -1.04953206e+00 -2.39438981e-01 1.23139083e+00 1.55152068e-01 -2.11284474e-01 8.23038220e-01 6.72822654e-01 -1.03156292e+00 1.25285470e+00 5.55255473e-01 1.82414684e-03 9.45380509e-01 -7.78060973e-01 -2.77811792e-02 6.90834761e-01 -3.05995405e-01 8.43305826e-01 5.23776591e-01 -4.85585123e-01 -1.60798621e+00 -1.32663047e+00 9.47876513e-01 -3.18791181e-01 3.28446716e-01 -2.41028264e-01 -1.12568581e+00 8.30260694e-01 -1.37387533e-02 3.47392708e-01 9.79211509e-01 1.26373842e-01 -6.52767956e-01 -6.69618189e-01 -1.79713225e+00 4.44553316e-01 9.95109797e-01 -3.65720034e-01 -7.67698605e-03 4.22674268e-01 8.66384327e-01 -2.70067811e-01 -7.91140079e-01 1.70959443e-01 6.32086635e-01 -1.33509398e+00 5.82451522e-01 -6.69700384e-01 -1.96655542e-01 -5.76744974e-02 -2.32820183e-01 -8.49452794e-01 -4.55061495e-01 -1.07313490e+00 -4.89170849e-01 1.83884501e+00 8.03080797e-02 -1.23103881e+00 1.17863894e+00 1.08906949e+00 2.67035544e-01 -8.03969383e-01 -9.16316390e-01 -6.61114335e-01 -1.89017713e-01 -4.72417325e-01 1.38834047e+00 1.00405586e+00 -6.61526918e-02 -1.84821501e-01 -6.08866572e-01 2.72776365e-01 9.79671538e-01 2.11166561e-01 8.42283845e-01 -1.16068792e+00 -4.78729039e-01 -3.16752583e-01 1.12039931e-02 -9.85806525e-01 2.91860640e-01 -8.11500371e-01 -4.35365856e-01 -1.12339497e+00 3.85661840e-01 -7.18209088e-01 -7.61750877e-01 6.33550406e-01 -2.91214496e-01 -3.25654477e-01 1.22218870e-01 3.73218805e-01 -8.33159208e-01 6.81466997e-01 8.36935282e-01 1.21358614e-02 -3.15740198e-01 6.47224545e-01 -1.24171746e+00 4.88754392e-01 1.01138604e+00 -6.08841121e-01 -3.24880928e-01 -2.33016461e-01 -4.76344943e-01 -3.71086672e-02 7.34387115e-02 -6.35002553e-01 4.14706588e-01 -5.15465438e-01 4.96106595e-02 -3.44450623e-01 -6.02514297e-03 -1.09085834e+00 2.60014147e-01 3.37258011e-01 -2.85284102e-01 -4.68179315e-01 -2.99622774e-01 8.09529483e-01 2.04683058e-02 1.86260983e-01 5.91387153e-01 -6.04504608e-02 -2.92057574e-01 1.07482755e+00 -2.05916956e-01 -4.17186320e-02 1.01944482e+00 2.77017355e-01 -4.00396794e-01 -3.81561011e-01 -6.31511092e-01 6.23372257e-01 3.18134338e-01 4.76597458e-01 2.92108715e-01 -1.22985470e+00 -5.74884593e-01 3.84454101e-01 9.59038958e-02 -5.88267632e-02 6.98921084e-01 8.84839594e-01 1.96468249e-01 2.67313212e-01 2.70048261e-01 -3.58593196e-01 -1.37685061e+00 7.17952669e-01 1.94265366e-01 -2.17085391e-01 -5.32566130e-01 7.65958667e-01 2.27830261e-01 -6.87207818e-01 6.13602638e-01 1.28328606e-01 -9.38453376e-02 -1.11818865e-01 7.58728802e-01 5.88347793e-01 -8.37184489e-03 -6.59527600e-01 -5.44785142e-01 1.38611689e-01 -4.84687686e-01 2.67935246e-01 1.10647678e+00 -2.84181923e-01 -1.33032233e-01 1.20069377e-01 1.51854098e+00 2.92019695e-01 -1.34038794e+00 -8.56688738e-01 -4.64555956e-02 -8.65877390e-01 -2.24622697e-01 -6.58108473e-01 -1.52030087e+00 3.32632661e-01 5.40534377e-01 1.34595767e-01 1.21362925e+00 2.04438660e-02 9.95904148e-01 3.04348264e-02 6.35086417e-01 -7.57236123e-01 -5.41509271e-01 2.50507444e-01 3.79871488e-01 -1.20378029e+00 -6.30827472e-02 -4.06364799e-01 -6.91799104e-01 6.01486981e-01 6.04343235e-01 1.89686090e-01 1.16532123e+00 -1.19140245e-01 3.22025940e-02 4.17653143e-01 -9.20010090e-01 1.62409127e-01 -4.52160090e-02 5.36145210e-01 1.68444902e-01 3.74386638e-01 -5.20126760e-01 1.38800824e+00 2.40624994e-02 7.12713748e-02 8.96246135e-02 1.07059872e+00 -8.77971798e-02 -1.61380565e+00 -2.64605314e-01 6.27917171e-01 -8.89958560e-01 2.10976720e-01 -1.05016522e-01 4.12152074e-02 2.03391641e-01 1.05358422e+00 -2.18900785e-01 -6.22935891e-01 2.08519816e-01 5.99043667e-02 -1.67170346e-01 -4.24861312e-01 -7.93061018e-01 -3.56748514e-02 -1.27761468e-01 -1.00690329e+00 -7.02958554e-02 -8.93527746e-01 -1.03800297e+00 -5.55334985e-01 1.24510133e-03 4.71852422e-01 5.76082587e-01 7.23871589e-01 9.91872668e-01 9.44431350e-02 1.13315952e+00 -4.92403567e-01 -7.24984109e-01 -6.22153580e-01 -7.95221627e-01 3.98186713e-01 5.05156279e-01 -3.13409835e-01 -4.43218708e-01 -2.14114070e-01]
[5.849491596221924, 6.342510223388672]
42746194-d990-42ff-b0d3-df1e33905b39
explanationlp-abductive-reasoning-for
2010.13128
null
https://arxiv.org/abs/2010.13128v1
https://arxiv.org/pdf/2010.13128v1.pdf
ExplanationLP: Abductive Reasoning for Explainable Science Question Answering
We propose a novel approach for answering and explaining multiple-choice science questions by reasoning on grounding and abstract inference chains. This paper frames question answering as an abductive reasoning problem, constructing plausible explanations for each choice and then selecting the candidate with the best explanation as the final answer. Our system, ExplanationLP, elicits explanations by constructing a weighted graph of relevant facts for each candidate answer and extracting the facts that satisfy certain structural and semantic constraints. To extract the explanations, we employ a linear programming formalism designed to select the optimal subgraph. The graphs' weighting function is composed of a set of parameters, which we fine-tune to optimize answer selection performance. We carry out our experiments on the WorldTree and ARC-Challenge corpus to empirically demonstrate the following conclusions: (1) Grounding-Abstract inference chains provides the semantic control to perform explainable abductive reasoning (2) Efficiency and robustness in learning with a fewer number of parameters by outperforming contemporary explainable and transformer-based approaches in a similar setting (3) Generalisability by outperforming SOTA explainable approaches on general science question sets.
['André Freitas', 'Marco Valentino', 'Mokanarangan Thayaparan']
2020-10-25
null
null
null
null
['science-question-answering', 'answer-selection']
['miscellaneous', 'natural-language-processing']
[ 2.65543312e-01 1.13538861e+00 -5.45153618e-01 -6.95133150e-01 -9.98406887e-01 -5.72134614e-01 5.32881498e-01 3.41528535e-01 9.41716060e-02 7.42464721e-01 5.07391334e-01 -7.68897772e-01 -8.03717434e-01 -8.83782744e-01 -8.75215590e-01 2.02797055e-02 4.96437699e-02 1.13276350e+00 3.14960301e-01 -2.39148572e-01 4.87630546e-01 2.39487574e-01 -1.63982689e+00 5.53654432e-01 1.30573297e+00 8.43647122e-01 -3.83815020e-02 5.18832505e-01 -5.89881182e-01 1.12452221e+00 -2.26012364e-01 -9.14579809e-01 2.57938020e-02 -6.38991833e-01 -1.64852834e+00 3.39498855e-02 4.85849291e-01 4.91135903e-02 -9.35822353e-02 6.91791713e-01 -2.37017512e-01 1.31561786e-01 5.96314073e-01 -1.51098752e+00 -6.16661191e-01 1.22832465e+00 6.70205848e-03 2.58866578e-01 1.01512289e+00 5.68760000e-02 1.83780682e+00 -6.71904147e-01 6.74850583e-01 1.67002499e+00 1.93342924e-01 7.10469007e-01 -1.26690197e+00 -2.74242073e-01 5.24059355e-01 6.74082279e-01 -8.97403836e-01 -1.08346254e-01 6.29274607e-01 -7.96033591e-02 1.21573877e+00 8.12646508e-01 5.37374318e-01 8.72610569e-01 1.46289229e-01 6.70745492e-01 8.57379138e-01 -6.51470125e-01 4.50967014e-01 -1.02046579e-02 7.53390074e-01 1.13913441e+00 5.50058126e-01 -2.27408096e-01 -5.73486149e-01 -4.99969751e-01 2.94017643e-01 -2.26578131e-01 -3.94171953e-01 -2.58613169e-01 -1.19834661e+00 1.17915201e+00 4.49435025e-01 6.86537400e-02 -4.39745218e-01 4.69788134e-01 -2.23539665e-01 5.51295519e-01 1.31736286e-02 1.18770170e+00 -7.37489581e-01 4.94568050e-01 -4.20338959e-01 6.78968370e-01 1.32190835e+00 1.05443525e+00 9.12698388e-01 -3.99090439e-01 -2.91496277e-01 2.31581315e-01 3.27906281e-01 2.32878372e-01 8.80496949e-02 -1.38430762e+00 7.20910966e-01 1.14768302e+00 1.29389778e-01 -9.61425304e-01 -5.27529478e-01 -2.81777829e-01 -1.01971157e-01 -2.51935244e-01 5.21176815e-01 -1.58497952e-02 -6.85954571e-01 1.67384171e+00 4.91110235e-01 2.65897792e-02 1.31339103e-01 9.38164711e-01 1.09005129e+00 5.20645678e-01 2.89459378e-01 -3.77655565e-03 1.73524368e+00 -9.92323875e-01 -6.66369617e-01 -5.14336884e-01 5.49300075e-01 -2.36923426e-01 1.39171743e+00 3.08019131e-01 -1.13430583e+00 -1.94878802e-01 -9.70271945e-01 -5.08542061e-01 -1.96289301e-01 -3.70820463e-01 1.04962909e+00 2.91634172e-01 -6.81090891e-01 4.16593403e-01 -3.33855838e-01 -1.29479483e-01 1.90750808e-01 4.81298149e-01 -1.63190305e-01 -4.41650838e-01 -1.23827600e+00 1.12519968e+00 6.24117911e-01 -4.33391243e-01 -5.11449933e-01 -7.54220843e-01 -1.02923870e+00 5.33828855e-01 1.04459786e+00 -1.36583877e+00 1.06736410e+00 -5.19566596e-01 -1.06045604e+00 7.29347885e-01 -5.56624711e-01 -6.33318841e-01 1.42713338e-02 -4.62910533e-02 -5.23614407e-01 4.28543240e-01 3.13275635e-01 6.41727686e-01 3.80864859e-01 -1.30459011e+00 -6.14459336e-01 -3.55063558e-01 6.94049835e-01 8.11891481e-02 2.22501978e-01 -2.72545427e-01 -3.47310454e-01 -8.30158219e-02 7.26826310e-01 -8.57853889e-01 -3.95905018e-01 -3.19509029e-01 -6.56048715e-01 -7.03710914e-01 3.00460666e-01 -5.07550478e-01 1.07111990e+00 -1.49575496e+00 5.51072180e-01 5.27706265e-01 6.41245246e-01 -5.25469184e-01 -2.91091383e-01 4.06591684e-01 -1.60231383e-03 2.64402896e-01 -1.23316363e-01 9.20054168e-02 4.34634328e-01 5.77863455e-01 -5.71966767e-01 -7.41344467e-02 1.49321318e-01 1.20185745e+00 -9.02241945e-01 -6.91992640e-01 -3.62218656e-02 -1.94515109e-01 -1.00035524e+00 3.53103429e-01 -1.04148924e+00 6.86398670e-02 -9.00345802e-01 5.67645609e-01 1.45873860e-01 -7.61328042e-01 3.03176492e-01 -1.51460171e-01 3.93383801e-01 7.85788596e-01 -1.24790990e+00 1.50884700e+00 -3.88426691e-01 2.22952709e-01 -3.95804971e-01 -8.23822379e-01 9.36142266e-01 1.77271217e-01 -1.51380002e-01 -4.73826468e-01 -7.93818012e-02 4.02980745e-01 -1.80617198e-02 -1.02996838e+00 2.49009088e-01 -4.25225317e-01 -2.47317985e-01 4.79250908e-01 1.27741992e-01 -5.64640045e-01 2.56671816e-01 6.39855206e-01 1.20078826e+00 4.39044759e-02 4.07307357e-01 -4.47159737e-01 7.65357137e-01 4.37867701e-01 2.82796264e-01 8.39379549e-01 4.73398298e-01 2.47178003e-01 8.69089603e-01 -8.85788858e-01 -6.13960624e-01 -9.34109390e-01 2.98547894e-01 1.05120885e+00 4.16246176e-01 -4.87471938e-01 -5.76923788e-01 -9.86235619e-01 9.48911533e-02 1.66193497e+00 -5.93125582e-01 -4.56234440e-02 -7.39295721e-01 -2.81421952e-02 3.05318683e-01 2.57503331e-01 9.10045654e-02 -1.24087286e+00 -7.06567645e-01 -8.01290646e-02 -8.24230611e-01 -1.06093192e+00 -2.19017863e-01 2.72636980e-01 -8.68247747e-01 -1.57054174e+00 2.85607636e-01 -5.90468705e-01 7.17303216e-01 1.28335357e-01 1.60291374e+00 7.89674163e-01 2.11119518e-01 5.27569711e-01 -1.75045103e-01 -4.12600845e-01 -2.25180596e-01 2.85136789e-01 -4.48950440e-01 -3.70690078e-01 5.72714567e-01 -3.47429663e-01 -3.16948920e-01 1.24377921e-01 -8.36289465e-01 7.56180435e-02 2.85577178e-01 5.30856192e-01 6.52478814e-01 -9.70324203e-02 5.77190697e-01 -1.28323364e+00 8.27400804e-01 -7.39765644e-01 -5.93545198e-01 7.10965872e-01 -8.36132526e-01 8.94921064e-01 6.83394790e-01 4.47529741e-02 -1.25822675e+00 -2.23308235e-01 -2.70178635e-03 2.23014176e-01 -1.30770490e-01 6.75018311e-01 -3.40739876e-01 2.46874452e-01 9.02368724e-01 -2.59752035e-01 -4.90225047e-01 -2.05825195e-01 7.42755651e-01 8.53675976e-02 6.06214046e-01 -1.10979676e+00 9.73124385e-01 2.28440344e-01 3.16492528e-01 -2.07252145e-01 -1.55352104e+00 -2.40347221e-01 -2.19581261e-01 1.94308266e-01 8.41391623e-01 -4.61177349e-01 -1.04578400e+00 -9.52140749e-01 -1.23083985e+00 1.72057047e-01 -5.00348687e-01 4.63986278e-01 -7.12572038e-01 2.51713604e-01 -3.40332508e-01 -6.88195229e-01 -2.93812394e-01 -9.72521365e-01 8.55870008e-01 1.32514745e-01 -9.22572136e-01 -9.67151880e-01 6.18958063e-02 9.17929232e-01 7.18164369e-02 3.64115477e-01 1.79152775e+00 -1.10466671e+00 -9.95570898e-01 7.12824538e-02 -1.10074960e-01 -4.64408159e-01 -4.04433668e-01 -4.45790976e-01 -6.22204065e-01 2.03540921e-01 -7.04080164e-02 -3.45921665e-01 6.72218144e-01 2.44134307e-01 1.17733467e+00 -7.61895359e-01 -4.31024045e-01 4.09336954e-01 1.39889097e+00 -1.80657148e-01 5.56203723e-01 4.70446914e-01 4.12620366e-01 9.81829405e-01 4.19014394e-01 -3.33010368e-02 8.07973087e-01 3.08589518e-01 6.87882066e-01 2.06045970e-01 -3.11533600e-04 -6.63994431e-01 -3.42900991e-01 2.03574598e-01 -1.35206059e-01 -2.34456807e-01 -8.85571957e-01 6.81305230e-01 -2.02593613e+00 -1.12522745e+00 -6.19000554e-01 1.78415227e+00 7.54973412e-01 1.71187744e-01 -3.70784141e-02 1.63222522e-01 3.67131382e-01 -1.45403609e-01 -5.82109392e-01 -4.96798098e-01 -1.48222819e-01 4.27150279e-01 2.77031250e-02 1.10353959e+00 -3.29288810e-01 9.71862733e-01 6.27446747e+00 3.02599758e-01 -1.69907957e-01 -2.44467840e-01 3.48518342e-01 7.10350275e-02 -1.46768749e+00 5.84254444e-01 -4.96213824e-01 -1.50064051e-01 8.47386718e-01 -1.19447872e-01 7.76195228e-01 7.21435010e-01 -2.91272312e-01 1.45909533e-01 -1.49858010e+00 3.80073726e-01 -6.03680201e-02 -1.72737384e+00 4.60038781e-01 -2.41529062e-01 5.97261012e-01 -5.10938108e-01 -4.01059359e-01 3.02829862e-01 4.78666812e-01 -1.26917350e+00 7.95026183e-01 5.44724286e-01 2.75374651e-01 -7.62138009e-01 5.80929875e-01 4.92051721e-01 -8.98283005e-01 -4.04288203e-01 -3.75231385e-01 -1.60041064e-01 2.42714100e-02 3.99450213e-01 -7.51622736e-01 7.91240394e-01 3.79259497e-01 6.27471805e-02 -5.14348745e-01 6.75260365e-01 -9.29219067e-01 6.58060551e-01 -1.14705786e-01 -2.21236393e-01 3.40647697e-01 -6.85542524e-02 4.78193820e-01 1.05611157e+00 2.88965940e-01 7.35873222e-01 -1.49890512e-01 1.28865588e+00 -1.59820765e-01 -1.75722107e-01 -2.48782843e-01 1.90109849e-01 6.85636282e-01 1.03819835e+00 -4.98456329e-01 -4.92753983e-01 -2.88653970e-01 4.98016506e-01 5.41670561e-01 3.62523675e-01 -7.76563942e-01 -2.67755091e-01 2.81624794e-01 2.21807808e-01 1.84758127e-01 2.53217578e-01 -5.74306786e-01 -1.29544973e+00 -3.52039710e-02 -1.16066289e+00 1.14879990e+00 -1.03014898e+00 -1.06991088e+00 7.96770215e-01 4.27908063e-01 -4.59319264e-01 -4.68808264e-01 -5.23399413e-01 -6.67319477e-01 8.47055793e-01 -1.55944347e+00 -1.10181022e+00 -4.19269532e-01 5.41840553e-01 5.51138401e-01 1.24141216e-01 8.14611733e-01 -5.12058377e-01 -2.42961615e-01 8.62069279e-02 -1.02870750e+00 -6.27533734e-01 -2.64815651e-02 -1.55496776e+00 3.76400769e-01 6.84221864e-01 4.19434249e-01 8.32802534e-01 1.28936255e+00 -4.48089451e-01 -1.64095879e+00 -5.81990242e-01 1.62697756e+00 -7.37566710e-01 5.03237665e-01 9.78893191e-02 -1.10015929e+00 1.19197214e+00 4.11625683e-01 -4.12696064e-01 5.35423398e-01 4.73200828e-01 -7.15972483e-01 2.67892241e-01 -1.28534567e+00 7.72568762e-01 1.25494754e+00 -1.42868072e-01 -1.46852815e+00 5.40569484e-01 1.11150599e+00 -3.59944761e-01 -7.02104032e-01 1.80978969e-01 3.17650020e-01 -9.17256653e-01 9.26082969e-01 -1.15933681e+00 7.59132206e-01 -4.27072316e-01 -2.24235177e-01 -1.08180976e+00 -5.32031536e-01 -7.12612689e-01 -4.18123245e-01 7.16175973e-01 9.93035913e-01 -6.37048244e-01 7.60638714e-01 1.13127255e+00 -5.83892725e-02 -9.90164399e-01 -7.74503171e-01 -3.10054988e-01 -2.11879149e-01 -3.41668665e-01 1.25407946e+00 7.40152717e-01 5.10005176e-01 8.14139664e-01 2.24585515e-02 4.64913934e-01 7.74655640e-01 8.05470586e-01 6.62450433e-01 -1.41280794e+00 -4.77151543e-01 -2.03643382e-01 1.95586011e-01 -9.75830376e-01 4.94041681e-01 -1.05884588e+00 -1.57701164e-01 -2.03511953e+00 2.11887643e-01 -1.98400974e-01 8.81355479e-02 6.32825255e-01 -4.66046274e-01 -4.77184236e-01 7.94497505e-02 6.21299585e-03 -8.08138132e-01 1.64036557e-01 1.36206412e+00 -6.39783591e-02 1.16194412e-01 -7.69659355e-02 -1.27887774e+00 8.30147326e-01 6.06537163e-01 -5.47066271e-01 -8.02957475e-01 -4.82613295e-01 6.85841501e-01 4.88083214e-01 3.95605326e-01 -5.28462946e-01 3.22444111e-01 -7.49238908e-01 7.63437971e-02 -2.66548872e-01 1.93067253e-01 -8.90235722e-01 2.14711204e-01 5.30048132e-01 -9.61813152e-01 1.84440270e-01 -3.39562446e-02 5.67285240e-01 -4.00861651e-02 -4.91174936e-01 2.50163645e-01 -1.78627774e-01 -5.45877934e-01 7.67207518e-02 -1.28449756e-03 4.08010125e-01 6.81704640e-01 -6.51229993e-02 -4.92929697e-01 -5.87132156e-01 -5.97286940e-01 5.63392818e-01 8.32909346e-02 2.48527572e-01 8.10853958e-01 -1.14139259e+00 -7.48528063e-01 -1.61118001e-01 3.17693263e-01 -1.84665024e-01 -1.45648150e-02 3.90785635e-01 -3.19276810e-01 7.48277426e-01 1.85896635e-01 -1.32216915e-01 -9.20638144e-01 6.95618629e-01 3.39038581e-01 -4.99772638e-01 -6.82280958e-01 8.19472790e-01 -4.38384339e-02 -5.89618623e-01 1.25053197e-01 -6.29662275e-01 -3.67321372e-01 -4.95061964e-01 2.81115294e-01 5.41940928e-01 -1.37197226e-01 -5.62543161e-02 -4.65703934e-01 3.57136905e-01 2.80467749e-01 1.62957385e-02 1.32347298e+00 -9.51709971e-02 -1.64329186e-01 -1.18874021e-01 6.04555130e-01 1.19608209e-01 -6.74397469e-01 -3.45845610e-01 5.13483524e-01 -4.01164323e-01 -2.96985596e-01 -1.01212049e+00 -5.66468060e-01 3.98443252e-01 -4.81762499e-01 6.69535637e-01 7.43612409e-01 6.57855630e-01 7.44893372e-01 8.14030111e-01 3.30065310e-01 -5.62962055e-01 2.64868606e-02 2.58115739e-01 1.25361621e+00 -9.14603651e-01 2.64087841e-02 -7.62044668e-01 -6.56056523e-01 1.18917024e+00 6.90512836e-01 -3.71518061e-02 8.31709951e-02 -2.01201618e-01 -2.19219550e-01 -1.03467894e+00 -1.16568327e+00 -8.67788717e-02 7.95847833e-01 1.69174537e-01 3.98843646e-01 1.08662784e-01 -3.82046878e-01 8.62369001e-01 -6.46440387e-01 -3.27971667e-01 3.71848047e-01 4.37396586e-01 -8.73000979e-01 -9.65532899e-01 -4.12719458e-01 4.66656297e-01 -2.21324563e-01 1.94534995e-02 -7.52883315e-01 9.38594162e-01 -7.46916980e-02 1.38305569e+00 -3.99094909e-01 -6.75272197e-02 6.12320721e-01 2.32955471e-01 6.84982598e-01 -6.50899529e-01 -5.62680066e-01 -6.34761333e-01 6.24349773e-01 -7.35410035e-01 -2.39004657e-01 -2.02577055e-01 -1.71373212e+00 -2.93963909e-01 -4.03477371e-01 6.78569078e-01 2.03190729e-01 1.58400118e+00 3.79775614e-01 5.17477751e-01 1.65201083e-01 3.78290936e-02 -8.34325194e-01 -4.71258491e-01 -2.16560200e-01 6.09629691e-01 1.33714780e-01 -4.94835049e-01 -5.74450076e-01 -1.58980101e-01]
[10.833137512207031, 7.782648086547852]
6582102a-e716-4ee2-a5f7-d8c845406190
weakly-supervised-video-anomaly-detection
2101.10030
null
https://arxiv.org/abs/2101.10030v3
https://arxiv.org/pdf/2101.10030v3.pdf
Weakly-supervised Video Anomaly Detection with Robust Temporal Feature Magnitude Learning
Anomaly detection with weakly supervised video-level labels is typically formulated as a multiple instance learning (MIL) problem, in which we aim to identify snippets containing abnormal events, with each video represented as a bag of video snippets. Although current methods show effective detection performance, their recognition of the positive instances, i.e., rare abnormal snippets in the abnormal videos, is largely biased by the dominant negative instances, especially when the abnormal events are subtle anomalies that exhibit only small differences compared with normal events. This issue is exacerbated in many methods that ignore important video temporal dependencies. To address this issue, we introduce a novel and theoretically sound method, named Robust Temporal Feature Magnitude learning (RTFM), which trains a feature magnitude learning function to effectively recognise the positive instances, substantially improving the robustness of the MIL approach to the negative instances from abnormal videos. RTFM also adapts dilated convolutions and self-attention mechanisms to capture long- and short-range temporal dependencies to learn the feature magnitude more faithfully. Extensive experiments show that the RTFM-enabled MIL model (i) outperforms several state-of-the-art methods by a large margin on four benchmark data sets (ShanghaiTech, UCF-Crime, XD-Violence and UCSD-Peds) and (ii) achieves significantly improved subtle anomaly discriminability and sample efficiency. Code is available at https://github.com/tianyu0207/RTFM.
['Gustavo Carneiro', 'Johan W. Verjans', 'Rajvinder Singh', 'Yuanhong Chen', 'Guansong Pang', 'Yu Tian']
2021-01-25
null
http://openaccess.thecvf.com//content/ICCV2021/html/Tian_Weakly-Supervised_Video_Anomaly_Detection_With_Robust_Temporal_Feature_Magnitude_Learning_ICCV_2021_paper.html
http://openaccess.thecvf.com//content/ICCV2021/papers/Tian_Weakly-Supervised_Video_Anomaly_Detection_With_Robust_Temporal_Feature_Magnitude_Learning_ICCV_2021_paper.pdf
iccv-2021-1
['anomaly-detection-in-surveillance-videos', 'anomaly-detection-in-surveillance-videos']
['computer-vision', 'methodology']
[ 2.49558941e-01 -3.86649132e-01 -3.16536099e-01 -3.52478325e-01 -9.06177461e-01 -3.76484275e-01 6.59818709e-01 4.97712269e-02 -2.72183597e-01 4.31790739e-01 8.57208073e-02 -1.28089618e-02 -7.26380944e-02 -4.62608576e-01 -9.56507206e-01 -7.56381154e-01 -6.92085862e-01 2.82065552e-02 1.29473552e-01 7.58536011e-02 2.09288105e-01 3.08650315e-01 -1.58080494e+00 6.20252609e-01 8.34322274e-01 1.43084145e+00 -4.54507440e-01 7.51005590e-01 7.74346069e-02 1.26934600e+00 -5.70505083e-01 -2.07662180e-01 3.04035574e-01 -3.32145095e-01 -6.52251661e-01 1.66423187e-01 7.61407077e-01 -6.33863330e-01 -7.06788242e-01 8.98613513e-01 2.14114979e-01 3.23311239e-01 6.65087759e-01 -1.74115419e+00 -6.32045150e-01 4.90552559e-02 -9.08183575e-01 1.06567764e+00 4.88836616e-01 3.78379017e-01 9.70338345e-01 -1.03319407e+00 2.80617625e-01 1.05641985e+00 6.98581994e-01 4.87061620e-01 -7.94297636e-01 -7.99135208e-01 5.60480475e-01 6.69390500e-01 -1.24181569e+00 -3.70951533e-01 6.19449437e-01 -5.31630278e-01 1.00024974e+00 2.94838607e-01 5.07188201e-01 1.49126339e+00 8.17804784e-02 1.14505410e+00 6.07325435e-01 7.28311539e-02 8.63374677e-04 -3.10133010e-01 1.93979755e-01 6.80066824e-01 3.18540670e-02 7.77270868e-02 -5.26885688e-01 -2.39063740e-01 5.18032372e-01 4.55349237e-01 -2.74996966e-01 6.41679764e-03 -1.18669546e+00 8.05024922e-01 2.14800656e-01 2.16786295e-01 -5.02657831e-01 1.86550424e-01 8.85915339e-01 4.84778702e-01 7.68444717e-01 4.11871970e-02 -4.19655949e-01 -3.60722899e-01 -7.31958687e-01 2.62475580e-01 4.51748252e-01 7.00680017e-01 4.30397451e-01 2.86105335e-01 -4.24034804e-01 6.86797142e-01 -4.45500240e-02 3.61778408e-01 5.15959561e-01 -7.51638412e-01 6.36016130e-01 5.40812135e-01 -5.59307188e-02 -1.20788479e+00 -2.55926073e-01 -3.15169066e-01 -7.14822531e-01 -4.28156219e-02 4.80049938e-01 -1.67506427e-01 -1.01940525e+00 1.59248269e+00 1.27846196e-01 1.01398015e+00 -5.08315079e-02 8.74963701e-01 7.65932441e-01 8.60241473e-01 2.55502701e-01 -2.31320232e-01 9.08262730e-01 -8.36447179e-01 -6.22971117e-01 -3.64213526e-01 7.65895963e-01 -3.21907461e-01 9.15958703e-01 3.24691892e-01 -8.41426730e-01 -3.42668623e-01 -7.56382823e-01 4.28332716e-01 -3.15355420e-01 -1.53835535e-01 5.37461936e-01 1.70496076e-01 -6.58339500e-01 5.88070333e-01 -8.24751854e-01 -1.72649294e-01 8.19306850e-01 2.60772914e-01 -5.83602548e-01 -2.45297804e-01 -1.25074995e+00 4.50358033e-01 3.37095737e-01 1.93623558e-01 -1.08599317e+00 -8.64382327e-01 -1.20070899e+00 -7.90172741e-02 5.91846526e-01 -1.14500567e-01 1.05170238e+00 -1.57364666e+00 -6.82075381e-01 7.94555902e-01 -2.37161964e-01 -6.48717284e-01 5.86799622e-01 -4.40279871e-01 -7.97407687e-01 4.08956200e-01 2.57954150e-01 3.49018902e-01 1.14556134e+00 -8.75550985e-01 -8.27222705e-01 -2.79635817e-01 -6.96291551e-02 6.90475330e-02 -4.19351935e-01 2.45172828e-01 -4.01411265e-01 -9.65080261e-01 -1.38811484e-01 -7.25700796e-01 -2.50754990e-02 -1.45016506e-01 -3.65912288e-01 -2.98977464e-01 1.19514596e+00 -7.71507263e-01 1.43958735e+00 -2.32182217e+00 -6.80648983e-02 2.18288407e-01 2.14635089e-01 3.72568816e-01 -3.19635928e-01 1.32041305e-01 -4.18167323e-01 -2.02813558e-02 -2.23661274e-01 -4.98964749e-02 -2.09970966e-01 2.90417641e-01 -3.64425451e-01 7.33788192e-01 6.82377577e-01 7.87047684e-01 -1.12548077e+00 -3.37883443e-01 2.26206198e-01 2.74320215e-01 -5.21496534e-01 2.89320201e-01 6.00345880e-02 5.23085833e-01 -4.25271839e-01 1.05000877e+00 4.49787498e-01 -2.42737949e-01 -3.89577001e-01 -1.18370824e-01 1.40781611e-01 -9.76668596e-02 -9.68417108e-01 1.21045887e+00 -6.64459588e-03 7.49954879e-01 -3.18749398e-01 -1.35288703e+00 3.99396598e-01 4.88189608e-01 7.95147955e-01 -8.93508732e-01 -1.21264488e-01 2.12292790e-01 -1.46801054e-01 -9.25281465e-01 1.08382717e-01 1.14798546e-01 -1.48663791e-02 9.74556282e-02 1.12315156e-01 7.46738851e-01 3.98890138e-01 3.53755921e-01 1.47241306e+00 2.44698692e-02 1.86037317e-01 1.94903374e-01 5.59751272e-01 -2.04269275e-01 8.61698925e-01 8.35470974e-01 -5.56648135e-01 6.59620821e-01 7.32425153e-01 -6.75898612e-01 -8.09382200e-01 -1.21751177e+00 -1.53175458e-01 1.19344509e+00 1.06786795e-01 -3.16435605e-01 -4.64964390e-01 -1.19229066e+00 9.70288962e-02 4.64878380e-01 -7.64065444e-01 -4.46662158e-01 -6.54622614e-01 -9.59995031e-01 4.67368126e-01 7.37130344e-01 3.83420199e-01 -1.16298783e+00 -2.86227971e-01 3.81895751e-02 -3.75589281e-01 -1.34024775e+00 -6.47846699e-01 -4.33215015e-02 -5.75887859e-01 -1.26565337e+00 -5.55475116e-01 -5.07126212e-01 6.25581086e-01 5.93341850e-02 1.07593966e+00 1.62971184e-01 -4.14476633e-01 6.29665911e-01 -5.98065257e-01 -3.63849878e-01 -4.22023190e-03 -4.71326530e-01 3.66113424e-01 5.01887560e-01 6.48176789e-01 -4.61140424e-01 -5.67609668e-01 1.87481061e-01 -9.68568742e-01 -3.78700048e-01 4.85317051e-01 9.49152470e-01 6.79872870e-01 5.08821383e-02 8.74661326e-01 -6.99736059e-01 2.47649163e-01 -1.06849647e+00 -2.09301263e-01 -1.49126828e-01 -1.46503463e-01 -3.72507572e-01 7.33698249e-01 -6.43144727e-01 -7.38837779e-01 -1.38352215e-01 -9.87024307e-02 -9.22821999e-01 -4.64248717e-01 3.29690963e-01 1.14825137e-01 -1.13700004e-02 4.35259163e-01 3.55776429e-01 -1.42454922e-01 -2.33214408e-01 -2.57546276e-01 3.89581233e-01 6.49180651e-01 -3.74775946e-01 6.64056957e-01 5.40440321e-01 -1.70994982e-01 -8.49867523e-01 -1.10933757e+00 -8.15322757e-01 -5.06439745e-01 -5.50471485e-01 7.78517544e-01 -1.08696830e+00 -3.79826337e-01 6.48629963e-01 -7.73731530e-01 -3.06740075e-01 -2.97987103e-01 4.76832956e-01 -5.21728814e-01 4.39956188e-01 -7.31412649e-01 -8.19038689e-01 -1.14639759e-01 -8.90154898e-01 1.12148273e+00 2.74932999e-02 -1.62722424e-01 -9.24748063e-01 -2.76758410e-02 2.25223318e-01 1.15461998e-01 7.83345044e-01 7.32547998e-01 -1.03189600e+00 -3.36784124e-01 -5.65197408e-01 -2.29723126e-01 5.63814402e-01 4.15109433e-02 1.51314110e-01 -1.01514900e+00 -3.31446648e-01 -2.33609647e-01 -3.68405074e-01 1.02542102e+00 5.45150697e-01 1.67185676e+00 -5.50807595e-01 -1.83778107e-01 6.88650608e-01 1.06039083e+00 2.80835032e-01 6.89322710e-01 4.13802713e-01 8.66733849e-01 3.45640689e-01 7.64028907e-01 6.06908500e-01 2.13678777e-01 5.11954129e-01 6.78100049e-01 -1.07321076e-01 1.65825561e-01 7.69166201e-02 7.71897316e-01 3.52895260e-01 -2.04268068e-01 -2.67775416e-01 -8.72465193e-01 7.08569109e-01 -2.11476898e+00 -1.49657297e+00 -2.14547172e-01 2.13266182e+00 3.58453810e-01 2.52328694e-01 4.42513883e-01 3.03810298e-01 6.92823112e-01 3.59220505e-01 -6.70823872e-01 -2.49971926e-01 -1.16775066e-01 -1.62817165e-01 2.17451274e-01 -5.30899577e-02 -1.69474089e+00 4.53556001e-01 5.69038391e+00 7.92213380e-01 -1.01982307e+00 5.34989536e-02 9.38828886e-01 -5.98911643e-01 3.11433852e-01 -5.95390141e-01 -3.78563762e-01 8.50330949e-01 1.12088275e+00 4.76014614e-02 2.27467492e-01 8.35418165e-01 2.61656314e-01 1.07801318e-01 -1.13740349e+00 1.02513099e+00 2.21075058e-01 -1.16786838e+00 2.96682268e-02 -2.02865064e-01 7.14052141e-01 2.31826216e-01 2.45253026e-01 6.67507172e-01 -3.76173556e-01 -1.05849409e+00 6.27565503e-01 4.71176982e-01 7.36390591e-01 -9.33065593e-01 8.87650609e-01 1.04515642e-01 -1.26965892e+00 -5.65706789e-01 -1.33328050e-01 -5.05105555e-02 -8.11168179e-03 6.44604087e-01 -3.69967848e-01 3.68662417e-01 1.12568915e+00 1.30286753e+00 -6.16117954e-01 1.05788171e+00 7.63165802e-02 8.73722196e-01 -1.13993749e-01 4.84756738e-01 5.80195367e-01 -6.22731037e-02 8.63804698e-01 1.43314993e+00 2.86932439e-01 4.79208566e-02 3.04242343e-01 4.50828671e-01 -6.86087981e-02 -2.53291819e-02 -7.66288042e-01 -1.01364456e-01 5.70296347e-02 1.25188839e+00 -4.78746265e-01 -3.79750729e-01 -8.85109723e-01 1.05834258e+00 2.28750214e-01 5.12525916e-01 -1.21353781e+00 -2.56929547e-01 9.08633947e-01 4.52434123e-02 2.92230636e-01 1.98823169e-01 2.05223009e-01 -1.41071486e+00 3.35432112e-01 -9.93474245e-01 8.60612571e-01 -3.62371296e-01 -1.61120498e+00 5.74655116e-01 -2.59243418e-02 -1.62467217e+00 -4.02965575e-01 -6.79212868e-01 -9.92118537e-01 3.52694541e-01 -1.39796674e+00 -9.34355736e-01 -4.22651201e-01 9.46505249e-01 7.10021198e-01 -2.49190673e-01 4.38631833e-01 5.94204664e-01 -9.56053019e-01 7.76918590e-01 1.03155179e-02 4.98190731e-01 7.49430478e-01 -1.21523273e+00 1.70615226e-01 1.23633480e+00 -2.28836276e-02 4.03857343e-02 5.59334695e-01 -6.45781219e-01 -1.24531996e+00 -1.53899050e+00 4.21331704e-01 -5.71355224e-01 8.76089573e-01 -1.61076009e-01 -1.24877965e+00 9.97257352e-01 -2.17174515e-01 8.56142581e-01 6.51559949e-01 -1.70555502e-01 -4.46580559e-01 1.04210840e-03 -1.19742060e+00 3.93561870e-01 1.19218218e+00 -4.84511495e-01 -4.39369082e-01 5.69518447e-01 3.90126854e-01 -4.08929288e-01 -7.01617420e-01 6.82481468e-01 3.98970127e-01 -1.03573918e+00 1.06411338e+00 -9.68913198e-01 4.81345952e-01 -8.84809420e-02 -1.70207024e-01 -1.23495638e+00 -2.57582009e-01 -3.61573040e-01 -9.65843141e-01 1.10585690e+00 1.79760531e-01 -6.41662776e-01 5.04798472e-01 4.21243131e-01 -2.91716874e-01 -1.02497327e+00 -1.00319672e+00 -1.02844477e+00 -3.36771578e-01 -7.75703669e-01 3.83675367e-01 1.07394195e+00 2.74890522e-03 -1.88113928e-01 -6.75691187e-01 4.66819882e-01 5.78531682e-01 -1.94111168e-01 2.18394727e-01 -9.26901460e-01 -9.51395705e-02 -3.80448401e-01 -8.86223257e-01 -5.02586961e-01 3.59433621e-01 -7.47854829e-01 -5.50108552e-02 -9.84926760e-01 2.86380917e-01 -1.10834308e-01 -6.88009083e-01 5.53262174e-01 -4.10744667e-01 5.45053482e-01 -1.33032829e-01 1.00812301e-01 -1.07552135e+00 4.02225345e-01 8.11218381e-01 -8.45956132e-02 -3.45443264e-02 1.01092190e-01 -3.37791413e-01 1.00587237e+00 7.04263985e-01 -4.03448552e-01 -1.50943100e-01 -2.08143577e-01 -2.03308798e-02 -7.86369815e-02 7.20359147e-01 -1.11472929e+00 -1.02584757e-01 -1.88927934e-01 7.30108917e-01 -4.42950100e-01 1.56728461e-01 -6.99823856e-01 -2.98917919e-01 3.75139445e-01 -3.85909140e-01 2.02379897e-01 2.11324573e-01 9.33771014e-01 -4.04701918e-01 8.81729797e-02 5.93404531e-01 5.92188984e-02 -1.17066705e+00 9.12073851e-01 -4.35182780e-01 4.07480240e-01 1.34801280e+00 -2.25002870e-01 -3.20312709e-01 -4.37668830e-01 -6.78602099e-01 4.50955778e-01 3.15754004e-02 6.77748680e-01 7.92900264e-01 -1.56549788e+00 -8.42103720e-01 3.36850256e-01 4.62779760e-01 -1.85251415e-01 6.82172239e-01 1.27080917e+00 -1.88554883e-01 2.25600034e-01 -7.87972510e-02 -7.15609372e-01 -1.21606529e+00 5.71793497e-01 4.21871156e-01 -1.63181812e-01 -7.98288643e-01 8.42035532e-01 4.16568637e-01 -2.61090249e-02 3.91926199e-01 -1.45575777e-01 -2.36358851e-01 6.47762641e-02 9.51834738e-01 5.46649635e-01 4.96821404e-02 -8.95159483e-01 -5.51666558e-01 2.66065329e-01 -2.11071149e-01 6.39733851e-01 1.36617327e+00 9.30420831e-02 1.75293759e-01 4.63897973e-01 1.41858709e+00 -2.52882838e-01 -1.60014331e+00 -2.48043388e-01 8.26728418e-02 -7.04855144e-01 -1.61399208e-02 -5.23038328e-01 -1.33722723e+00 6.29178166e-01 7.20337093e-01 2.96456903e-01 1.36257005e+00 1.25014052e-01 9.15529788e-01 1.72957093e-01 -3.44403386e-02 -9.96876717e-01 5.26884973e-01 5.19571066e-01 8.46566558e-01 -1.75699258e+00 -4.05979931e-01 -7.20793456e-02 -9.49667573e-01 1.09313023e+00 9.93608356e-01 -3.35922271e-01 4.40409034e-01 1.72486141e-01 -2.34053954e-01 -3.47846031e-01 -7.68238902e-01 -1.17208645e-01 7.15238154e-01 5.19408584e-01 2.84426004e-01 -1.41943723e-01 1.93375960e-01 6.18537128e-01 5.06058455e-01 -2.14394554e-01 3.21983010e-01 8.12619209e-01 -2.25474805e-01 -3.44556689e-01 -3.14127445e-01 1.14015830e+00 -1.01800632e+00 -5.16720638e-02 -5.22563532e-02 7.76401818e-01 1.98025629e-01 7.76816189e-01 5.30419469e-01 -4.86344367e-01 3.59411508e-01 1.99220240e-01 1.51711643e-01 -3.47017497e-01 -4.13393497e-01 -8.75664949e-02 8.15144554e-02 -1.11714232e+00 -2.73000687e-01 -9.55675125e-01 -1.29378128e+00 -2.25145623e-01 -2.15804931e-02 -1.22876473e-01 -4.35466133e-02 1.11865044e+00 3.04691344e-01 6.54638410e-01 7.78670192e-01 -9.12816107e-01 -2.90541351e-01 -7.50030696e-01 -5.34230351e-01 9.02163744e-01 9.12892997e-01 -7.40835667e-01 -7.10220814e-01 -1.55693628e-02]
[7.851905822753906, 1.590477705001831]
5128f321-a66f-4e01-806c-5f32e9eb77a1
detecting-out-of-distribution-inputs-in-deep
1910.10307
null
https://arxiv.org/abs/1910.10307v1
https://arxiv.org/pdf/1910.10307v1.pdf
Detecting Out-of-Distribution Inputs in Deep Neural Networks Using an Early-Layer Output
Deep neural networks achieve superior performance in challenging tasks such as image classification. However, deep classifiers tend to incorrectly classify out-of-distribution (OOD) inputs, which are inputs that do not belong to the classifier training distribution. Several approaches have been proposed to detect OOD inputs, but the detection task is still an ongoing challenge. In this paper, we propose a new OOD detection approach that can be easily applied to an existing classifier and does not need to have access to OOD samples. The detector is a one-class classifier trained on the output of an early layer of the original classifier fed with its original training set. We apply our approach to several low- and high-dimensional datasets and compare it to the state-of-the-art detection approaches. Our approach achieves substantially better results over multiple metrics.
['Taylor Denounden', 'Krzysztof Czarnecki', 'Rick Salay', 'Sachin Vernekar', 'Vahdat Abdelzad', 'Buu Phan']
2019-10-23
null
null
null
null
['one-class-classifier']
['methodology']
[ 2.28591278e-01 9.74532142e-02 -2.35000044e-01 -3.81245852e-01 -4.83761579e-01 -5.34354031e-01 6.22466207e-01 3.73484135e-01 -4.06903982e-01 3.32862318e-01 -1.98592722e-01 -2.83678532e-01 3.11750442e-01 -9.04519975e-01 -5.98015726e-01 -5.58464468e-01 7.25743547e-02 4.94927198e-01 6.09423459e-01 3.08865547e-01 1.40728816e-01 6.21908307e-01 -1.85110927e+00 7.68354774e-01 5.37211895e-01 1.30191338e+00 -1.70949697e-01 7.44384468e-01 -2.20069796e-01 4.35930580e-01 -1.14218903e+00 -1.39666930e-01 3.60501826e-01 -3.65321666e-01 -4.78699833e-01 -5.53655289e-02 8.07477415e-01 -4.47697699e-01 -2.19620019e-01 1.20248747e+00 5.55632889e-01 -4.02916372e-01 1.16975868e+00 -1.32211232e+00 -6.50592029e-01 5.22792041e-02 -5.11141539e-01 5.49858570e-01 6.26526847e-02 -1.29932314e-01 7.75177002e-01 -1.19578457e+00 6.83570564e-01 1.32963169e+00 5.59523463e-01 5.60052216e-01 -1.36049712e+00 -7.11556375e-01 -3.87289701e-03 -1.07350767e-01 -1.12958193e+00 -4.14774716e-02 4.30317461e-01 -6.97789729e-01 9.52809572e-01 -4.65752110e-02 3.41358364e-01 1.22598445e+00 1.70652419e-01 1.14006436e+00 9.45581555e-01 -6.38706625e-01 5.10757864e-01 3.74011844e-01 5.19841075e-01 5.61664462e-01 4.63836521e-01 2.69399911e-01 -3.31136137e-01 -1.57217190e-01 2.55263358e-01 6.99096248e-02 -2.09686179e-02 -4.44252014e-01 -7.23108232e-01 1.00625849e+00 4.55469370e-01 3.94191444e-01 -3.27131391e-01 -1.28988877e-01 4.04300034e-01 4.62120920e-01 6.92663014e-01 3.74402791e-01 -3.71454239e-01 1.20481454e-01 -7.85224974e-01 2.49735922e-01 9.51774299e-01 7.78218269e-01 5.08230925e-01 -2.95970619e-01 -3.80910009e-01 9.17082489e-01 9.00104120e-02 1.84259146e-01 5.34205258e-01 -4.26673293e-01 1.69840217e-01 8.56100023e-01 -1.34760693e-01 -7.93212175e-01 -3.32635313e-01 -6.48278236e-01 -8.24914515e-01 5.19743979e-01 5.63575625e-01 -6.21030070e-02 -1.31162643e+00 1.19077671e+00 3.92501891e-01 -2.95402259e-01 -1.39570519e-04 6.54823005e-01 9.08748984e-01 5.39383411e-01 -1.09463319e-01 2.88310081e-01 1.17893624e+00 -7.04895377e-01 -5.43951213e-01 -3.47525507e-01 5.98795533e-01 -7.42154241e-01 7.81734169e-01 6.78544402e-01 -4.02160406e-01 -6.03092194e-01 -1.29500890e+00 2.36794859e-01 -9.08841252e-01 3.90647739e-01 2.83299208e-01 7.37320602e-01 -7.44238377e-01 4.70675886e-01 -4.72984374e-01 -3.85105878e-01 9.74665046e-01 3.97070408e-01 -1.76513359e-01 -3.02403212e-01 -8.32623303e-01 7.90247858e-01 4.63161349e-01 -3.86642009e-01 -1.13232994e+00 -3.65395457e-01 -6.43502176e-01 -8.91969875e-02 2.29186699e-01 -8.28164592e-02 1.41076505e+00 -1.10612500e+00 -9.21317458e-01 1.18770051e+00 -1.69545695e-01 -4.11473811e-01 7.52894938e-01 -3.54646921e-01 -2.83268780e-01 -1.41341895e-01 1.64019004e-01 6.50037408e-01 1.21388578e+00 -1.18924665e+00 -8.78934741e-01 -3.35736722e-01 -1.07731707e-01 -3.03659052e-01 -6.05969369e-01 -7.04975240e-03 -2.80975133e-01 -7.23864973e-01 1.37622297e-01 -7.77468026e-01 8.39964971e-02 3.15653503e-01 -6.73559785e-01 -5.92965245e-01 1.13457322e+00 -2.19607502e-02 1.10406971e+00 -2.33669233e+00 -2.81006247e-01 9.25215110e-02 3.85203153e-01 6.67989492e-01 -1.15243867e-01 1.35175928e-01 -1.61307469e-01 9.64805931e-02 -1.13368168e-01 -2.89946318e-01 1.26316816e-01 9.11229178e-02 -4.69732493e-01 5.14701009e-01 4.92329866e-01 4.03473169e-01 -8.24886143e-01 -2.84702510e-01 1.89449005e-02 2.23098189e-01 -2.59767771e-01 4.01587009e-01 -1.17504992e-01 -1.91914633e-01 -1.41856506e-01 9.26597357e-01 8.30524921e-01 -9.53938514e-02 -1.80468172e-01 4.50288579e-02 -1.14353588e-02 2.91587830e-01 -1.13722098e+00 9.48909760e-01 -1.09370604e-01 1.03360248e+00 -1.93745956e-01 -1.16402483e+00 1.08691847e+00 1.06669396e-01 2.34354381e-02 -6.69684887e-01 2.93950617e-01 5.01670361e-01 3.05443287e-01 -4.30817008e-01 1.52345881e-01 7.73342233e-03 4.05642912e-02 3.82060021e-01 3.77522022e-01 1.93457350e-01 2.29429901e-01 -1.19924545e-02 1.25554562e+00 -2.62101322e-01 5.13756394e-01 -3.22325140e-01 2.67994374e-01 1.04308296e-02 3.99251163e-01 1.17422032e+00 -3.52034897e-01 6.98421717e-01 8.71585429e-01 -7.24704504e-01 -1.03984487e+00 -1.00158596e+00 -5.66588521e-01 1.00062430e+00 -6.55679926e-02 -2.09277973e-01 -9.04521286e-01 -1.21943402e+00 4.29399282e-01 4.38384175e-01 -8.91483307e-01 -1.51838869e-01 -2.36463174e-01 -1.00463963e+00 5.82235396e-01 7.48387396e-01 2.15197861e-01 -1.02641523e+00 -6.73515379e-01 2.61975318e-01 6.00257814e-01 -9.18203652e-01 -8.95391926e-02 8.08882415e-01 -7.70685613e-01 -1.14493322e+00 -8.68820667e-01 -1.02164888e+00 8.87635946e-01 7.33245239e-02 1.10222483e+00 -1.52154022e-03 -7.90116847e-01 -1.30024180e-01 -2.15004578e-01 -1.13813424e+00 -6.60136759e-01 3.29978347e-01 1.46950871e-01 7.54297078e-02 1.09191513e+00 -6.46944419e-02 -3.74527574e-01 4.01390016e-01 -8.87025476e-01 -5.72583318e-01 7.35786378e-01 1.00777698e+00 4.70313519e-01 2.11282894e-02 7.74360538e-01 -1.11780906e+00 7.91300535e-01 -4.92416441e-01 -6.33277953e-01 -2.41364554e-01 -4.97884095e-01 1.26985297e-01 7.79588997e-01 -7.27034628e-01 -5.81590414e-01 3.16073239e-01 -3.04190099e-01 -5.40847421e-01 -6.96146548e-01 5.01909442e-02 2.73103435e-02 1.10671885e-01 9.30233538e-01 -8.78712162e-02 -1.67205840e-01 -8.16147327e-01 -1.47629641e-02 1.22616506e+00 2.77831733e-01 -1.51788339e-01 6.51701987e-01 4.25241292e-01 -1.53499857e-01 -9.38228428e-01 -1.14884770e+00 -5.97899139e-01 -8.22568595e-01 -2.52230354e-02 5.70988297e-01 -7.02601016e-01 -2.58856654e-01 6.98197067e-01 -1.15719223e+00 -1.74105540e-01 -1.93146169e-01 2.64679313e-01 5.80692627e-02 5.21173291e-02 -4.90255952e-01 -9.24547374e-01 -2.00853452e-01 -9.87869382e-01 1.12169874e+00 2.44408563e-01 -3.93490762e-01 -6.43463075e-01 3.21730152e-02 -3.94138902e-01 3.57070595e-01 2.89787650e-01 9.08687174e-01 -1.12311125e+00 -1.65128335e-01 -8.13423932e-01 -2.95636892e-01 6.90829873e-01 2.11233646e-01 5.61092906e-02 -1.42347205e+00 -3.36590677e-01 -2.61710644e-01 -5.26979208e-01 1.29277980e+00 2.81064749e-01 1.44111741e+00 -5.37721142e-02 -7.23851800e-01 3.54393899e-01 1.17912173e+00 3.01889032e-01 4.04934555e-01 4.19828951e-01 2.84521520e-01 7.09179819e-01 6.20541036e-01 2.15918988e-01 -2.15032503e-01 4.81964856e-01 4.40490872e-01 -3.66755128e-01 -1.47743270e-01 -5.34301735e-02 2.38835916e-01 1.39337569e-01 5.65354407e-01 -4.68107939e-01 -9.68206465e-01 6.93027854e-01 -1.58336306e+00 -7.90739715e-01 -2.41754606e-01 2.02801180e+00 6.44150138e-01 7.42100775e-01 2.42818505e-01 5.00721157e-01 7.59409010e-01 -6.62558675e-02 -8.72726440e-01 -6.33905172e-01 1.27769306e-01 3.81539315e-01 3.45624119e-01 2.38851532e-02 -1.68174374e+00 5.46157479e-01 7.22412682e+00 6.56922758e-01 -1.15552211e+00 -9.06722806e-03 7.29013979e-01 2.10673977e-02 2.65372902e-01 -5.84510863e-01 -1.13032317e+00 5.85899711e-01 5.48129141e-01 3.42729956e-01 -1.63383633e-01 1.30856979e+00 -3.47898692e-01 -3.28395367e-01 -1.51319313e+00 9.12367046e-01 1.97158888e-01 -1.09589899e+00 -1.54297695e-01 2.10855544e-01 7.13793635e-01 1.53519779e-01 9.23645124e-02 5.09189188e-01 1.10250279e-01 -1.05748594e+00 5.42148769e-01 1.17881112e-01 8.20885181e-01 -7.04873145e-01 1.07977223e+00 5.90617955e-01 -8.60508144e-01 -2.00040624e-01 -5.76411545e-01 -3.11552491e-02 -5.45236170e-01 9.64778304e-01 -1.08776438e+00 -6.84962571e-02 9.75464165e-01 6.88640475e-01 -7.01996565e-01 1.20365584e+00 -1.49345696e-01 6.22014642e-01 -4.48019475e-01 -3.49227220e-01 2.59031892e-01 3.34404588e-01 3.72598380e-01 1.38707542e+00 1.69656083e-01 -2.88961560e-01 1.94999382e-01 9.94802356e-01 -3.61494422e-01 -1.31855398e-01 -9.61292207e-01 -9.38637555e-02 4.54231322e-01 1.04503119e+00 -7.76700974e-01 -6.79133892e-01 -4.48242545e-01 8.70949447e-01 3.31090152e-01 3.27756628e-02 -4.74901587e-01 -8.84119093e-01 8.73519182e-01 5.84066026e-02 5.45821607e-01 2.87267894e-01 -2.58575916e-01 -9.94545221e-01 1.65635213e-01 -8.98335159e-01 5.58644414e-01 -1.59480020e-01 -1.66032636e+00 6.31096363e-01 -2.91610897e-01 -1.40234673e+00 -1.35161802e-01 -1.25151372e+00 -5.88550329e-01 7.10815430e-01 -1.48879707e+00 -3.61990362e-01 -3.83343190e-01 1.53131932e-01 5.27258456e-01 -3.43776554e-01 8.25471163e-01 4.84586418e-01 -5.43849766e-01 7.48468161e-01 3.39869469e-01 5.53903341e-01 7.11062968e-01 -1.48558652e+00 4.45567876e-01 7.24840283e-01 1.12192437e-01 2.20905766e-01 5.65959811e-01 -3.27724308e-01 -9.41119373e-01 -1.19016790e+00 8.01794529e-01 -5.01078904e-01 1.97657421e-01 -6.13511384e-01 -9.51810718e-01 4.39525068e-01 -2.15462148e-02 3.64618003e-01 6.06388509e-01 4.74057794e-02 -3.16988140e-01 -1.36584610e-01 -1.18067372e+00 1.01753294e-01 8.33997190e-01 -4.66774583e-01 -8.61348510e-01 3.04778159e-01 1.81743503e-01 -4.28584248e-01 -4.90491480e-01 3.01587403e-01 6.00489974e-01 -9.14376199e-01 8.27742696e-01 -6.12959146e-01 3.34292352e-01 -2.87258297e-01 -1.02980748e-01 -1.24488056e+00 -1.00129865e-01 -1.51687399e-01 -3.71608675e-01 1.09520721e+00 3.36287498e-01 -6.76744938e-01 8.52185369e-01 8.46311301e-02 2.77069286e-02 -9.18140173e-01 -7.87379086e-01 -1.06080329e+00 5.29300794e-02 -2.71574378e-01 3.74086678e-01 8.24039876e-01 -2.96280593e-01 3.67682576e-01 -6.64087832e-02 1.81556523e-01 7.50180483e-01 1.05021767e-01 7.55053401e-01 -1.69031274e+00 -1.17312789e-01 -4.74993140e-01 -6.63953424e-01 -9.79188859e-01 -8.02823454e-02 -9.68452513e-01 2.46029377e-01 -1.36519980e+00 1.32371873e-01 -3.83740187e-01 -3.40379804e-01 6.11408353e-01 -1.99902407e-03 6.03296340e-01 -8.97306427e-02 2.68430263e-01 -3.92356813e-01 3.23060036e-01 7.21242130e-01 -3.95685524e-01 -8.08610693e-02 7.13164061e-02 -3.99869621e-01 9.66820419e-01 7.54986525e-01 -1.05368602e+00 5.79150543e-02 -1.71347141e-01 -2.07881466e-01 -6.65946901e-01 2.55922168e-01 -1.23358810e+00 -1.03304563e-02 2.93387264e-01 9.69755828e-01 -7.77110875e-01 5.15260827e-03 -6.05978429e-01 -3.92548740e-01 8.94625247e-01 -4.97840524e-01 -1.71394199e-01 1.81923598e-01 5.96412182e-01 -3.57582569e-01 -5.01895368e-01 1.09563422e+00 9.40222386e-03 -5.99037528e-01 1.32621735e-01 -7.30523050e-01 1.18507601e-01 1.24399352e+00 -1.27038598e-01 -4.28664207e-01 7.57719949e-02 -5.26312888e-01 -9.48821381e-02 2.70711392e-01 6.88181818e-01 5.90711474e-01 -1.22610021e+00 -4.40825045e-01 5.06036580e-01 4.48152959e-01 1.50674973e-02 -6.30227089e-01 2.73036391e-01 -4.40595478e-01 4.68436271e-01 -1.17160134e-01 -9.24719810e-01 -1.28904200e+00 6.11691117e-01 5.97907960e-01 1.66437104e-02 -4.83038545e-01 8.53926241e-01 2.98222810e-01 -5.32698572e-01 6.65624738e-01 -5.01062393e-01 -2.24581227e-01 4.81136471e-01 7.98651576e-01 1.48379818e-01 2.98150063e-01 -2.11323053e-01 -5.67867875e-01 3.59190583e-01 -2.43143752e-01 3.65149766e-01 1.23316598e+00 6.05747700e-01 1.59588501e-01 5.74790299e-01 1.57481718e+00 -3.57691437e-01 -1.07952666e+00 -2.41098464e-01 3.10549468e-01 -5.94840348e-01 1.37850419e-01 -8.22183847e-01 -9.41634476e-01 1.23920512e+00 1.07626116e+00 6.66014731e-01 8.18457663e-01 2.55046904e-01 5.23010135e-01 5.13037920e-01 9.39047411e-02 -1.33604026e+00 3.31095427e-01 5.98506451e-01 5.22037089e-01 -1.45646417e+00 -1.88917190e-01 -3.04838866e-01 -1.63396940e-01 1.34935808e+00 8.69965315e-01 -3.57927918e-01 7.91604757e-01 4.49106991e-01 1.31274030e-01 -1.95996821e-01 -7.91295230e-01 -3.79141986e-01 4.26857948e-01 5.78482509e-01 4.38121408e-01 -5.19500934e-02 -1.55936792e-01 2.76348054e-01 3.25773180e-01 6.77118450e-03 3.35865706e-01 1.06179190e+00 -5.78073919e-01 -9.99093235e-01 -3.47309440e-01 1.01219356e+00 -5.26529670e-01 9.29639190e-02 -8.23130906e-01 6.91139638e-01 3.84870440e-01 5.84967494e-01 3.78591388e-01 -4.15546477e-01 5.34432948e-01 2.16663972e-01 8.90755653e-02 -8.68472338e-01 -4.16439503e-01 -1.31431103e-01 -7.88140222e-02 -4.82749611e-01 -1.46659061e-01 -6.87151313e-01 -9.05243754e-01 1.88083217e-01 -4.76197302e-01 -2.05792576e-01 5.94940186e-01 7.95679867e-01 4.01197642e-01 5.87987304e-01 5.22376537e-01 -7.00593591e-01 -7.66292393e-01 -9.53150392e-01 -5.33846021e-01 3.89602423e-01 7.53265202e-01 -8.14863861e-01 -5.60012698e-01 -1.91349968e-01]
[9.342658042907715, 2.9456374645233154]
2f364245-258e-48f8-b017-506f6dda1096
morpheus-a-neural-network-for-jointly
null
null
https://aclanthology.org/W19-4205
https://aclanthology.org/W19-4205.pdf
Morpheus: A Neural Network for Jointly Learning Contextual Lemmatization and Morphological Tagging
In this study, we present Morpheus, a joint contextual lemmatizer and morphological tagger. Morpheus is based on a neural sequential architecture where inputs are the characters of the surface words in a sentence and the outputs are the minimum edit operations between surface words and their lemmata as well as the morphological tags assigned to the words. The experiments on the datasets in nearly 100 languages provided by SigMorphon 2019 Shared Task 2 organizers show that the performance of Morpheus is comparable to the state-of-the-art system in terms of lemmatization. In morphological tagging, on the other hand, Morpheus significantly outperforms the SigMorphon baseline. In our experiments, we also show that the neural encoder-decoder architecture trained to predict the minimum edit operations can produce considerably better results than the architecture trained to predict the characters in lemmata directly as in previous studies. According to the SigMorphon 2019 Shared Task 2 results, Morpheus has placed 3rd in lemmatization and reached the 9th place in morphological tagging among all participant teams.
['A. C{\\"u}neyd Tantu{\\u{g}}', 'Eray Yildiz']
2019-08-01
null
null
null
ws-2019-8
['morphological-tagging']
['natural-language-processing']
[ 1.80171475e-01 3.06887627e-01 1.11049667e-01 -4.43038195e-01 -9.53530133e-01 -9.26082492e-01 3.33771735e-01 4.79685664e-01 -8.35499167e-01 5.21600664e-01 4.04426754e-01 -5.98466694e-01 3.48306924e-01 -7.08868861e-01 -8.79923761e-01 -3.11040580e-01 2.93755412e-01 6.77586436e-01 1.39734700e-01 -8.42706189e-02 9.81109366e-02 1.80976138e-01 -7.96921313e-01 7.29587495e-01 8.18825126e-01 6.01368248e-01 2.95135289e-01 4.65800792e-01 -3.13081652e-01 2.54373461e-01 -6.61005378e-01 -1.08615565e+00 4.11882669e-01 -4.08051670e-01 -1.05180407e+00 -4.11605209e-01 8.82213295e-01 2.83973515e-01 -1.01889729e-01 1.10401046e+00 4.15190041e-01 5.75348511e-02 6.81813121e-01 -3.10504347e-01 -1.22259748e+00 1.47235000e+00 -1.37698933e-01 3.95996302e-01 7.42484480e-02 -1.31999940e-01 1.42824090e+00 -1.05202818e+00 9.24834549e-01 1.23828399e+00 9.28799391e-01 4.76064324e-01 -1.21596646e+00 -2.86473900e-01 1.83406800e-01 -1.25523627e-01 -1.24191141e+00 -3.33881408e-01 2.17098027e-01 -4.68054473e-01 1.58523011e+00 -4.36309502e-02 4.10168290e-01 6.86942279e-01 5.67696452e-01 6.97518051e-01 9.63778615e-01 -7.92162299e-01 -7.16009736e-02 8.09003133e-03 1.77555189e-01 9.34122205e-01 4.54211950e-01 -2.54233450e-01 -3.56399566e-01 1.46877527e-01 5.07164836e-01 -4.51582253e-01 1.53349295e-01 3.24957997e-01 -1.46167409e+00 6.69676065e-01 2.37391785e-01 6.61490679e-01 -3.23274523e-01 8.71932805e-02 6.62653804e-01 4.72327530e-01 6.11888289e-01 9.40236926e-01 -9.55631495e-01 -7.35141933e-02 -9.89772975e-01 -1.87517196e-01 8.08919251e-01 8.92250717e-01 5.51775098e-01 9.21777338e-02 -6.24118298e-02 7.37759233e-01 -8.72195587e-02 4.22689348e-01 4.76884276e-01 -6.84070289e-01 8.91692996e-01 6.77384555e-01 -2.83427596e-01 -4.91328001e-01 -3.85895759e-01 -1.27006680e-01 -4.12069172e-01 -1.37168337e-02 4.87727880e-01 -5.97378254e-01 -1.12420201e+00 1.75378716e+00 -1.33013815e-01 -5.45352399e-01 1.38465315e-01 2.62250245e-01 6.65462375e-01 7.58962214e-01 4.47474808e-01 -7.67856464e-02 1.55715466e+00 -8.60692620e-01 -8.01058710e-01 -5.57174206e-01 9.42138255e-01 -1.18227875e+00 1.13403749e+00 2.19917640e-01 -1.47162926e+00 -7.10097313e-01 -1.11403465e+00 -3.20967436e-01 -8.32516432e-01 4.19521570e-01 7.53550410e-01 4.69798863e-01 -1.22333169e+00 8.49001169e-01 -9.92292345e-01 -6.36506796e-01 4.61779395e-03 4.48809415e-01 -5.24922073e-01 4.66769159e-01 -1.27244079e+00 1.31724393e+00 7.31929958e-01 3.74140590e-02 -4.06595409e-01 -6.21946216e-01 -9.51364636e-01 -1.12789594e-01 4.08629961e-02 -4.30441767e-01 1.40748811e+00 -1.01563954e+00 -1.26333320e+00 1.31155121e+00 -4.67075333e-02 -6.42751992e-01 2.01178305e-02 -3.56244773e-01 -4.68756646e-01 -1.47869840e-01 3.00539285e-01 7.96069026e-01 1.32113859e-01 -7.87885904e-01 -9.11958754e-01 5.57826385e-02 -1.72281101e-01 2.15131700e-01 -1.81317747e-01 5.24622500e-01 -3.22295874e-01 -7.82226682e-01 9.87614617e-02 -8.27437520e-01 -2.37870976e-01 -7.46095777e-01 -4.63534713e-01 -3.84853035e-01 7.10010603e-02 -1.11843717e+00 1.30557835e+00 -2.17302918e+00 2.24305034e-01 -2.43913248e-01 -3.10460746e-01 4.95606244e-01 -1.86670437e-01 7.38557577e-01 -3.48611176e-01 4.35327023e-01 -3.86811078e-01 -7.06268728e-01 2.68029690e-01 2.18963861e-01 -2.91552663e-01 3.85155350e-01 3.63033324e-01 9.92290676e-01 -8.43372703e-01 -5.03214717e-01 -1.07601099e-01 9.66919735e-02 -2.47279927e-01 1.67888075e-01 -1.41418085e-01 -1.59022897e-01 1.29255593e-01 5.34918189e-01 4.25811976e-01 2.97160447e-01 4.48062867e-01 -1.48730814e-01 -3.55102569e-01 1.08034563e+00 -6.71760976e-01 1.91293705e+00 -7.06835806e-01 6.34329855e-01 -3.23297195e-02 -5.44328511e-01 7.15213299e-01 6.32489145e-01 4.90135662e-02 -4.88922596e-01 1.20675988e-01 5.68952739e-01 2.19096646e-01 -3.15347254e-01 7.58913934e-01 -3.57503146e-01 -6.07160807e-01 4.84210044e-01 4.23407465e-01 -2.92925835e-01 5.93921959e-01 3.09145540e-01 1.18752658e+00 1.43916056e-01 5.17094195e-01 -4.61356878e-01 2.76269346e-01 7.31216371e-02 6.36642814e-01 5.34472585e-01 2.13059187e-01 4.97698367e-01 5.96432626e-01 -4.21445787e-01 -1.19954610e+00 -1.32817793e+00 -9.01670381e-02 1.22394025e+00 -5.13179064e-01 -5.45022368e-01 -1.17728722e+00 -1.04157817e+00 -1.34802073e-01 1.08690512e+00 -8.66656363e-01 8.97056609e-02 -1.05576777e+00 -5.26996911e-01 9.59048748e-01 7.88733661e-01 1.77697167e-01 -1.74215555e+00 -1.85341761e-01 4.14581329e-01 -1.74662471e-01 -1.14833081e+00 -7.67254770e-01 7.27439880e-01 -6.38287723e-01 -7.69817412e-01 -2.01710612e-01 -1.58413696e+00 6.31782532e-01 -2.92061448e-01 1.20461726e+00 -1.24029256e-01 -2.44998951e-02 -2.86967814e-01 -4.87774432e-01 -3.98455590e-01 -7.58652627e-01 6.56250834e-01 -1.17679879e-01 -2.90722400e-01 5.15831292e-01 -5.96747220e-01 -3.62036750e-02 -3.27715665e-01 -8.20447564e-01 1.08645052e-01 8.69267881e-01 4.44299608e-01 5.69885194e-01 -2.12467864e-01 3.40630710e-01 -1.46500337e+00 3.98265988e-01 -1.24082476e-01 -5.97380340e-01 3.04343492e-01 -4.90359426e-01 1.55655742e-01 8.63343000e-01 -2.69120276e-01 -9.26422238e-01 1.24319248e-01 -4.29463625e-01 3.23093176e-01 -1.98491156e-01 6.52321100e-01 -4.47478592e-01 4.99187022e-01 5.72440028e-01 -1.27933010e-01 -5.66924214e-01 -9.64083493e-01 4.67974246e-01 5.14708638e-01 7.57868648e-01 -4.27071363e-01 7.97452927e-01 1.47853717e-01 -2.22678065e-01 -4.33470011e-01 -1.17052090e+00 -3.64591032e-01 -1.20621264e+00 4.36504126e-01 1.21900296e+00 -8.23040307e-01 -1.00844502e-01 4.82376158e-01 -1.65686703e+00 -6.15888059e-01 -6.54489636e-01 2.85379559e-01 -4.33533192e-01 2.74911344e-01 -1.16786849e+00 -2.35409111e-01 -5.49655974e-01 -9.41752672e-01 9.91862774e-01 -8.36521536e-02 -4.64915693e-01 -1.28402853e+00 2.01823890e-01 1.09653205e-01 4.63188291e-02 2.30217218e-01 1.54372156e+00 -1.10738325e+00 1.09844524e-02 -1.24241590e-01 -1.01818286e-01 5.73333800e-01 2.09808692e-01 -1.92758848e-03 -4.95301038e-01 -1.40372112e-01 -2.18417123e-01 -1.43483847e-01 1.00785446e+00 2.03877226e-01 4.52535361e-01 -2.06167415e-01 -4.69592176e-02 6.16653979e-01 1.37546325e+00 2.82051533e-01 5.13965130e-01 4.98614579e-01 7.34136760e-01 5.18759608e-01 5.38884997e-01 -2.18672797e-01 2.41684094e-01 4.59444791e-01 9.56860259e-02 -1.82622209e-01 -5.46325624e-01 -4.51802701e-01 9.79603410e-01 1.55152440e+00 3.13324124e-01 -5.01415312e-01 -9.54314232e-01 8.36583972e-01 -1.58895969e+00 -6.52589738e-01 -3.77647609e-01 2.03406191e+00 1.26291084e+00 3.03479970e-01 -3.58116031e-01 -1.98025629e-01 9.12548721e-01 2.59663731e-01 7.78588653e-02 -1.16363537e+00 -3.31725270e-01 8.10485482e-01 6.93956435e-01 8.03592622e-01 -1.32196379e+00 1.85344803e+00 6.61680174e+00 8.46591771e-01 -6.94755077e-01 4.65338320e-01 1.20686524e-01 -1.55740902e-02 -2.05337241e-01 2.69541472e-01 -1.04768431e+00 2.75491148e-01 1.32567072e+00 1.15338922e-01 3.72836888e-01 4.28833783e-01 -4.33063805e-02 -3.93346138e-02 -1.27655756e+00 3.03544015e-01 1.19753867e-01 -1.17138910e+00 2.69678324e-01 3.51272523e-02 9.62946475e-01 2.62483150e-01 -5.72416671e-02 3.03276092e-01 6.94835126e-01 -9.29727376e-01 1.07473195e+00 3.14637452e-01 1.11576068e+00 -7.85536349e-01 1.01209927e+00 -6.01907708e-02 -1.06910276e+00 4.16191250e-01 -5.73136270e-01 -2.04149056e-02 4.45950598e-01 4.26529080e-01 -1.10279536e+00 4.03366923e-01 1.66471153e-01 4.32307959e-01 -7.83254087e-01 8.23787332e-01 -9.55135047e-01 1.04365540e+00 -2.63536051e-02 2.90453583e-02 5.61884940e-01 -2.48978123e-01 3.66378099e-01 2.02613068e+00 1.56172544e-01 -1.08230628e-01 1.24722436e-01 3.29183549e-01 -4.64020342e-01 5.36217213e-01 -2.91700333e-01 -3.53728980e-01 4.30261672e-01 1.32329023e+00 -9.20679152e-01 -5.68228364e-01 -3.21232766e-01 1.12663901e+00 7.71317363e-01 7.11781010e-02 -7.55985200e-01 -8.58546615e-01 5.94223499e-01 1.14095703e-01 5.43082774e-01 -4.02118087e-01 -4.89132285e-01 -8.54912162e-01 -4.83093783e-02 -8.07689786e-01 4.30932343e-01 -5.50976455e-01 -1.14232814e+00 8.90822411e-01 -2.88540155e-01 -5.14259458e-01 3.37441303e-02 -1.04158330e+00 -5.65319180e-01 1.02569652e+00 -1.00023782e+00 -1.47895277e+00 5.17474413e-01 2.02350631e-01 5.90680897e-01 -1.98224008e-01 1.00024521e+00 2.32633263e-01 -5.01514733e-01 6.51734650e-01 2.34402925e-01 8.28233004e-01 9.11188662e-01 -1.74243581e+00 1.36531293e+00 1.34750533e+00 7.39148915e-01 7.13677585e-01 5.07298768e-01 -1.08136463e+00 -7.78459013e-01 -1.27108264e+00 1.90066290e+00 -5.61232507e-01 1.01542652e+00 -7.43183613e-01 -7.86761940e-01 1.14857101e+00 8.60437095e-01 -4.75668728e-01 6.13425970e-01 2.66255617e-01 -4.73690242e-01 1.25072449e-01 -5.77377856e-01 5.74183166e-01 1.17649412e+00 -6.32318735e-01 -1.12163925e+00 6.42103016e-01 9.77874041e-01 -3.85137409e-01 -7.34356582e-01 4.06885473e-03 1.61813155e-01 -6.04296982e-01 2.48945072e-01 -8.29619706e-01 4.28705543e-01 -6.08683787e-02 -2.11649686e-01 -1.41022038e+00 -4.67012078e-01 -6.89610541e-01 4.07196224e-01 1.60933971e+00 1.06495500e+00 -3.35390419e-01 5.66570342e-01 2.19396427e-02 -7.03297019e-01 -4.69065368e-01 -9.75160539e-01 -7.98691690e-01 6.03363931e-01 -3.70459944e-01 3.11624408e-01 7.91680992e-01 7.60693103e-02 5.11897147e-01 1.15398824e-01 5.77504709e-02 5.03172651e-02 2.48555746e-03 1.55210719e-01 -1.01527715e+00 -3.04813623e-01 -4.67783034e-01 -3.26191068e-01 -8.70261908e-01 5.75364649e-01 -1.46928656e+00 3.62386525e-01 -1.85409033e+00 1.61095843e-01 -2.51252919e-01 -2.86709547e-01 9.89893556e-01 -1.80172041e-01 3.16625535e-01 2.43013203e-01 -1.05531238e-01 -5.52227855e-01 -1.91902276e-02 8.07992160e-01 -6.54306635e-02 2.08030753e-02 -1.93538040e-01 -7.25349367e-01 8.76652718e-01 7.59499073e-01 -8.73359919e-01 2.29520813e-01 -9.63870287e-01 5.30959010e-01 -3.90962929e-01 -2.54238665e-01 -9.06113803e-01 1.52138159e-01 3.53275128e-02 1.16239429e-01 -5.31691194e-01 2.06487641e-01 -3.75825375e-01 -7.30957910e-02 4.84256566e-01 -2.88661182e-01 7.01650620e-01 2.80484468e-01 -1.01734884e-01 -7.22486749e-02 -5.70641994e-01 7.74970472e-01 -3.01044732e-01 -4.61003631e-01 5.23252413e-02 -8.71400177e-01 4.90682185e-01 6.00566208e-01 -5.88632897e-02 -2.00267792e-01 1.54323056e-01 -8.48955631e-01 -2.64602184e-01 5.04408240e-01 2.35034212e-01 1.46743849e-01 -1.12418020e+00 -8.16309512e-01 1.88856963e-02 -3.09104860e-01 -2.29176074e-01 -4.04735684e-01 5.42849898e-01 -7.72732496e-01 5.44219434e-01 -8.31656530e-02 2.23357826e-01 -1.15427232e+00 5.79399705e-01 2.04831347e-01 -7.34416425e-01 -3.56799275e-01 1.15381634e+00 1.39563568e-02 -4.93090421e-01 -5.76656722e-02 -5.72979867e-01 1.73175216e-01 3.36696953e-01 1.93443969e-01 7.96593800e-02 5.59797227e-01 -6.72273576e-01 -3.95784020e-01 2.41251796e-01 -3.88650954e-01 -4.73455518e-01 1.43775845e+00 1.78074449e-01 -4.85761076e-01 7.30363607e-01 1.15389705e+00 7.42947161e-01 -8.53922009e-01 -1.98334962e-01 5.35473347e-01 2.91144162e-01 -2.56296396e-01 -1.17403436e+00 -9.50585186e-01 8.82035673e-01 -1.12921698e-02 1.28539607e-01 7.17633724e-01 1.00178599e-01 1.12516034e+00 4.31038946e-01 2.50856519e-01 -1.38967478e+00 -3.57755691e-01 1.08915973e+00 6.23811662e-01 -6.63344085e-01 -1.82614386e-01 -4.69970971e-01 -6.97136283e-01 1.05948412e+00 3.80493999e-01 -3.49599570e-01 3.43210071e-01 4.77056175e-01 4.31471586e-01 -1.07144162e-01 -6.41690671e-01 -3.28545541e-01 2.14764804e-01 4.25295204e-01 9.29391265e-01 2.95378655e-01 -6.28831804e-01 6.89178050e-01 -6.79572940e-01 -8.22800219e-01 4.17589664e-01 7.06995487e-01 -4.21832651e-01 -1.59733987e+00 -1.55421989e-02 2.72853851e-01 -9.46897745e-01 -8.98831844e-01 -8.49105418e-01 1.01634169e+00 6.17018402e-01 7.56085753e-01 2.50325561e-01 -3.01038772e-01 5.06939054e-01 4.93751884e-01 5.19502342e-01 -1.30592442e+00 -1.49878180e+00 -2.51882285e-01 5.15235960e-01 -1.83849648e-01 1.12797029e-01 -1.17544186e+00 -1.50800610e+00 -1.49248600e-01 -2.22328201e-01 5.03606737e-01 6.59850657e-01 1.03680849e+00 1.26958087e-01 3.75635296e-01 1.31906345e-01 -5.99868238e-01 -5.68999112e-01 -1.23806763e+00 -5.91967046e-01 4.69458252e-01 -3.46328527e-01 -5.56046106e-02 -1.82403952e-01 2.56641835e-01]
[10.434189796447754, 10.056742668151855]
4a832185-98d6-4950-9ab3-e6fffdd9d29c
explaining-image-classifiers-using
null
null
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/6192_ECCV_2020_paper.php
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123730392.pdf
Explaining Image Classifiers using Statistical Fault Localization
The black-box nature of deep neural networks (DNNs) makes it impossible to understand why a particular output is produced, creating demand for “Explainable AI”. In this paper, we show that statistical fault localization (SFL) techniques from software engineering deliver high quality explanations of the outputs of DNNs, where we define an explanation as a minimal subset of features sufficient for making the same decision as for the original input. We present an algorithm and a tool called DeepCover, which synthesizes a ranking of the features of the inputs using SFL and constructs explanations for the decisions of the DNN based on this ranking. We compare explanations produced by DeepCover with those of the state-of-the-art tools GradCAM, LIME, SHAP, RISE and Extremal and show that explanations generated by DeepCover are consistently better across a broad set of experiments. On a benchmark set with known ground truth, DeepCover achieves 76.7% accuracy, which is 6% better than the second best Extremal.
['Daniel\xa0Kroening', 'Xiaowei\xa0Huang', 'Hana\xa0Chockler', 'Youcheng\xa0Sun']
null
null
null
null
eccv-2020-8
['fault-localization']
['computer-code']
[ 2.07237244e-01 7.42385626e-01 -1.00609578e-01 -6.34987712e-01 -3.04793209e-01 -4.69130009e-01 5.17577589e-01 -1.49871454e-01 4.19449449e-01 7.45429516e-01 1.13661319e-01 -5.64095676e-01 -5.07368088e-01 -6.27982378e-01 -1.04041588e+00 -3.65473360e-01 1.03098825e-02 3.68253440e-01 2.57843196e-01 6.19659498e-02 4.52469677e-01 5.10377467e-01 -1.94089866e+00 8.58024418e-01 7.97368169e-01 1.01724005e+00 -3.82802002e-02 8.08167458e-01 -1.80837989e-01 1.08827305e+00 -9.72709298e-01 -2.24273309e-01 -5.41653074e-02 -6.20135844e-01 -9.22560096e-01 -8.89393166e-02 6.58682585e-01 -3.56399566e-01 -2.83640593e-01 9.76512671e-01 -5.82527034e-02 -5.08729219e-01 6.46208465e-01 -2.01191807e+00 -9.29261148e-01 1.06960356e+00 1.91236749e-01 6.60685897e-02 2.79981680e-02 6.58046380e-02 1.14828408e+00 -9.14242804e-01 4.97718662e-01 1.20561457e+00 7.49992192e-01 9.42159593e-01 -1.17323077e+00 -5.20041585e-01 9.11552459e-03 3.34657013e-01 -8.03452373e-01 -3.08732718e-01 3.72019589e-01 -4.24419969e-01 1.53708255e+00 3.73742968e-01 4.50305909e-01 1.15543211e+00 7.68071711e-01 6.28253460e-01 6.85427248e-01 -5.52810609e-01 6.60851538e-01 -1.20696567e-01 5.31601310e-01 1.01064277e+00 8.58770490e-01 1.64400890e-01 -7.60543168e-01 -9.82451886e-02 7.25662351e-01 2.56799281e-01 -2.29435876e-01 -1.92156628e-01 -1.00474179e+00 5.85228801e-01 4.38280255e-01 3.68793815e-01 -5.18777311e-01 8.22375834e-01 1.31805435e-01 4.03946102e-01 1.65812463e-01 8.23315918e-01 -1.09288669e+00 4.27789465e-02 -6.41172826e-01 2.01398358e-01 9.54787254e-01 1.08794236e+00 8.68967533e-01 5.54865658e-01 -1.00435428e-01 8.51066485e-02 3.52454305e-01 2.62172073e-01 4.06924397e-01 -1.48018181e+00 5.81394099e-02 1.14620638e+00 -1.47244915e-01 -7.16304779e-01 -3.10929328e-01 -7.09808230e-01 -6.05880499e-01 6.45585477e-01 1.80328891e-01 -2.51148582e-01 -1.13208377e+00 1.54257226e+00 -4.57858920e-01 -6.76233962e-04 3.79655182e-01 6.26189888e-01 8.01206410e-01 3.92256021e-01 -4.02608573e-01 3.21271479e-01 7.37640738e-01 -9.48092520e-01 -3.87076259e-01 -6.83825791e-01 6.83168054e-01 7.33290762e-02 9.08622265e-01 9.22091842e-01 -6.69991016e-01 -5.47420979e-01 -1.51084125e+00 2.76476234e-01 -1.85167447e-01 3.08080912e-01 8.03377688e-01 2.34605491e-01 -1.38494337e+00 1.09503746e+00 -8.56352508e-01 -2.09180355e-01 4.84487295e-01 5.24987459e-01 -5.11697233e-01 -5.13707027e-02 -7.32823610e-01 9.05304372e-01 3.01844418e-01 -1.37593463e-01 -1.47782683e+00 -5.78255355e-01 -7.10027277e-01 4.61618721e-01 2.83007652e-01 -5.88497341e-01 1.53947580e+00 -1.20195305e+00 -7.32961178e-01 4.05731797e-01 2.39479337e-02 -8.01646590e-01 1.93655893e-01 -3.74514878e-01 -4.20204639e-01 -5.25420271e-02 1.51181817e-01 6.79761231e-01 4.91642445e-01 -1.45027673e+00 -7.63183713e-01 -2.10356042e-01 2.54138440e-01 -7.10517764e-01 -1.68940783e-01 -4.09816176e-01 6.68061376e-02 6.20231405e-02 3.61571789e-01 -6.95631683e-01 -2.57759213e-01 1.32091856e-02 -9.68851149e-01 -3.63611281e-01 7.65094578e-01 -4.22163099e-01 9.15214002e-01 -1.88883638e+00 -5.24686277e-02 1.17601246e-01 7.27045894e-01 -1.44542456e-01 -1.00815915e-01 2.20615909e-01 -6.35817230e-01 4.33550805e-01 -2.04289600e-01 -2.37239897e-01 3.98153245e-01 6.47407413e-01 -4.25791055e-01 1.81214556e-01 5.72154880e-01 6.15033805e-01 -7.73021698e-01 -1.06285609e-01 -5.61795719e-02 2.65750945e-01 -6.13051713e-01 1.63162053e-01 -5.43032467e-01 -2.94578582e-01 -2.53327012e-01 7.86057889e-01 1.41028836e-01 -3.73908341e-01 2.80678552e-02 2.17564166e-01 8.09991807e-02 4.50045586e-01 -8.75781655e-01 1.29402578e+00 -1.59779817e-01 1.27645886e+00 -6.53842151e-01 -7.28589475e-01 1.27622354e+00 5.14153183e-01 -1.46229804e-01 -3.44182253e-01 1.33909643e-01 4.53748852e-01 3.02255601e-01 -4.68162417e-01 1.01482347e-01 5.23772389e-02 -3.28188650e-02 6.12265527e-01 3.64768147e-01 1.63237184e-01 -4.76468587e-03 1.39804989e-01 1.98769450e+00 -2.08402593e-02 2.66545177e-01 -3.78524274e-01 8.48117396e-02 2.76165962e-01 7.56047666e-01 1.06349385e+00 1.02249257e-01 8.60625029e-01 1.26714635e+00 -9.97337759e-01 -1.06403828e+00 -9.36102450e-01 2.90517449e-01 6.29656494e-01 -2.44637564e-01 -3.56476426e-01 -1.17160857e+00 -1.22719419e+00 1.03137746e-01 1.44174325e+00 -1.00839365e+00 -4.51299399e-01 1.35694718e-04 -3.02481949e-02 6.18751466e-01 9.79598761e-01 1.63097471e-01 -1.42470360e+00 -1.02823174e+00 7.50333890e-02 3.03622544e-01 -8.67683053e-01 3.48050117e-01 1.01641583e+00 -1.04713619e+00 -1.24196935e+00 1.05439685e-02 -5.20929992e-01 1.09314930e+00 -6.60931394e-02 1.60725176e+00 7.80252516e-01 -1.43320650e-01 -8.62805471e-02 -2.85638332e-01 -5.50345659e-01 -8.45652103e-01 1.04399761e-02 2.15874493e-01 -6.38815165e-01 4.67142850e-01 -6.59535348e-01 -1.15977183e-01 2.00378343e-01 -9.62925494e-01 2.50687003e-01 8.73974025e-01 6.62784755e-01 3.55580568e-01 3.19170326e-01 6.13050759e-01 -1.02292991e+00 4.58432674e-01 -5.70345581e-01 -4.05025244e-01 3.31960112e-01 -9.53745365e-01 7.50827730e-01 8.83058429e-01 -1.17831282e-01 -5.64547420e-01 7.32468516e-02 1.48114100e-01 -4.40750003e-01 -5.42987883e-01 3.69134814e-01 -3.05805653e-01 3.14422309e-01 1.22360563e+00 -2.72959080e-02 -3.58835697e-01 -3.55835736e-01 4.57767993e-02 6.01417482e-01 8.67313325e-01 -3.58623624e-01 6.04418635e-01 5.11148870e-02 -2.80749872e-02 3.97963785e-02 -9.43837643e-01 4.81823027e-01 -5.12701571e-01 -1.99770942e-01 4.65763122e-01 -2.86227554e-01 -4.48335171e-01 4.66699228e-02 -1.62593675e+00 -3.41965526e-01 -4.28691477e-01 1.79329723e-01 -5.54863393e-01 -3.88386488e-01 -3.10296834e-01 -6.86178327e-01 -2.62942135e-01 -1.26987267e+00 7.41467774e-01 2.43960768e-01 -7.93169856e-01 -6.28396571e-01 -2.20036864e-01 -2.19983786e-01 2.58130163e-01 5.48631191e-01 1.31509531e+00 -1.35801339e+00 -6.40774131e-01 -3.26498389e-01 -1.25397757e-01 7.47174263e-01 5.73675074e-02 4.47370201e-01 -1.25119889e+00 3.65162015e-01 -7.69867152e-02 2.25163568e-02 7.25929022e-01 3.23606730e-01 1.27993584e+00 -4.31527019e-01 -3.60912979e-01 2.68026829e-01 1.43418431e+00 2.71821201e-01 6.20595455e-01 5.15578568e-01 5.39523244e-01 4.83411789e-01 2.66285032e-01 3.46891969e-01 1.48339169e-02 -1.15728192e-01 1.24444520e+00 -3.84910544e-03 -7.72148520e-02 -1.34976730e-01 4.86971438e-01 2.74046659e-01 6.72002509e-02 -4.15456295e-01 -1.24322999e+00 7.66983271e-01 -2.05875921e+00 -6.43744588e-01 -4.61182684e-01 1.66228127e+00 3.75375062e-01 6.55751228e-01 -5.36044121e-01 6.85030580e-01 4.37848955e-01 -3.94148588e-01 -8.32762182e-01 -7.68700838e-01 -9.55257714e-02 1.08295709e-01 3.42190742e-01 3.65665764e-01 -5.01890600e-01 6.35405004e-01 6.98080492e+00 1.84232742e-01 -6.45083249e-01 -1.51116326e-01 5.93259513e-01 2.97565579e-01 -6.82172537e-01 5.01282811e-02 -5.52607059e-01 5.88790327e-02 1.31861913e+00 -1.78805888e-01 3.96532804e-01 1.40227091e+00 -1.29575685e-01 9.89296213e-02 -1.95886791e+00 3.64171386e-01 -2.16616001e-02 -1.70946264e+00 1.41451862e-02 -4.05578464e-02 9.61837113e-01 1.78675856e-02 -8.89174119e-02 5.64927720e-02 6.83190525e-01 -1.41619110e+00 1.12226474e+00 6.40510321e-01 5.19151926e-01 -8.76073122e-01 1.19933271e+00 3.76151770e-01 -3.34444284e-01 -3.77229631e-01 -4.76801246e-01 -3.56717348e-01 -7.08577156e-01 7.15095937e-01 -1.30333829e+00 2.48875454e-01 7.42826581e-01 7.19175637e-01 -6.87834978e-01 7.97210395e-01 -9.55168188e-01 7.84628153e-01 1.08068265e-01 -1.23633087e-01 1.56259984e-01 7.78196394e-01 3.29223424e-01 1.07737970e+00 6.16968632e-01 -2.78527021e-01 -4.87902731e-01 1.50219035e+00 -1.32548571e-01 -8.21916640e-01 -7.97966421e-01 -2.75682628e-01 5.84185958e-01 1.09228873e+00 -7.30984151e-01 -4.85150099e-01 -8.37487653e-02 8.09619129e-01 7.60544017e-02 1.47402808e-01 -7.67759323e-01 -3.76834005e-01 7.79428840e-01 6.46576062e-02 1.37048587e-01 1.59335777e-01 -8.32137823e-01 -5.82588196e-01 2.09690426e-02 -1.14373791e+00 3.21421325e-02 -1.31178939e+00 -1.05620944e+00 1.17619741e+00 -1.77718818e-01 -1.11366785e+00 -7.27084994e-01 -9.42405164e-01 -8.00499320e-01 6.91951632e-01 -9.10212994e-01 -5.23181558e-01 -5.55252552e-01 -4.93553467e-02 5.46407878e-01 -4.63916123e-01 1.07591915e+00 -2.70515651e-01 -4.55614060e-01 2.73581684e-01 -2.10450947e-01 -3.00891064e-02 8.70630741e-02 -1.70179987e+00 1.01528621e+00 1.02236545e+00 3.56572986e-01 5.47983885e-01 1.30479085e+00 -5.42571902e-01 -1.29865706e+00 -1.05462384e+00 1.08357787e+00 -6.18955910e-01 5.29452264e-01 -1.22751735e-01 -9.12309408e-01 1.01154125e+00 2.78455883e-01 -1.38786612e-02 3.08372408e-01 7.32851215e-04 -4.33944225e-01 -6.54770136e-02 -1.15974343e+00 4.57450539e-01 1.09658575e+00 -4.13251728e-01 -7.02743769e-01 1.94964677e-01 1.05273581e+00 -2.71083176e-01 -3.62847269e-01 2.60438859e-01 4.90604639e-01 -1.54962909e+00 4.98430848e-01 -8.74467790e-01 1.22028685e+00 -5.15739024e-01 -3.97990167e-01 -1.61555815e+00 -2.93599397e-01 -3.06211680e-01 -2.51471132e-01 9.73898411e-01 8.98253083e-01 -6.21830881e-01 9.28711057e-01 7.95663893e-01 -8.26660752e-01 -9.39367056e-01 -6.55101955e-01 -7.35341966e-01 -3.29990327e-01 -8.03847909e-01 9.67643201e-01 5.99766970e-01 1.34433964e-02 7.65252113e-02 1.03451788e-01 4.68836397e-01 6.24264896e-01 -7.06623355e-03 5.65138757e-01 -1.65421116e+00 -2.84338087e-01 -2.62993902e-01 -7.33005702e-01 -4.18880910e-01 4.38031405e-01 -7.57067978e-01 4.97813761e-01 -2.05268240e+00 2.26468787e-01 6.74633235e-02 -4.63795245e-01 1.23684001e+00 2.78684467e-01 -1.56290710e-01 -1.31196290e-01 -5.93517385e-02 -5.26537716e-01 1.41435012e-01 7.41706312e-01 -9.46638882e-02 1.72422186e-01 -2.78830647e-01 -1.21766543e+00 1.07593548e+00 7.14605629e-01 -1.07824957e+00 -2.69637793e-01 -8.33703399e-01 3.76936466e-01 -4.46955189e-02 6.84058726e-01 -1.44729722e+00 1.50677711e-01 -7.65187889e-02 5.90611398e-01 -4.66266274e-01 -1.73071191e-01 -8.40750873e-01 4.09260809e-01 7.91247725e-01 -6.80833697e-01 3.67166072e-01 2.43958622e-01 2.25692332e-01 -1.12668253e-01 -5.42644203e-01 2.26777226e-01 1.33541316e-01 -8.55692685e-01 -1.79056287e-01 -4.37743396e-01 -3.31236839e-01 6.23773336e-01 -2.89116949e-01 -8.13913524e-01 -3.74749303e-01 -5.16675889e-01 -4.10095453e-02 4.67174977e-01 4.59994465e-01 8.87900889e-01 -1.24444437e+00 -4.76129502e-01 1.99889258e-01 1.18930042e-01 3.96858677e-02 -2.39136070e-01 2.81059146e-01 -5.20541906e-01 5.21188498e-01 -3.49161983e-01 -5.02437949e-01 -1.09234035e+00 1.13409460e-01 6.67247236e-01 1.48194388e-01 -6.81742966e-01 9.42097068e-01 1.12914637e-01 -4.26988423e-01 2.65669733e-01 -9.22219098e-01 1.59943551e-01 -6.40321791e-01 5.11796057e-01 2.97620803e-01 2.35081658e-01 4.25478630e-02 -5.03700197e-01 -1.56456456e-01 2.70032436e-01 6.52422905e-02 1.85185909e+00 6.13872945e-01 -8.30690563e-02 6.63850188e-01 6.64512992e-01 -5.66695869e-01 -1.50799322e+00 1.58565730e-01 4.36886519e-01 -1.33400410e-01 7.89001212e-02 -1.17265391e+00 -1.25266612e+00 1.11756599e+00 3.46895695e-01 7.00173795e-01 1.05404937e+00 2.03660920e-01 3.26727897e-01 6.39253020e-01 4.22035992e-01 -5.40824771e-01 1.11916326e-01 5.42270243e-01 1.04495573e+00 -7.17111111e-01 -3.80041182e-01 -4.06343527e-02 -4.46383864e-01 1.59670901e+00 8.69055867e-01 -4.15809065e-01 7.60320202e-02 5.89658141e-01 -6.33635819e-02 -5.78120708e-01 -1.42809176e+00 3.09163094e-01 2.20084965e-01 5.47895372e-01 2.71222144e-01 7.11116269e-02 4.46212083e-01 1.26010835e+00 -3.77438039e-01 2.07405463e-01 8.30207050e-01 7.79554307e-01 -8.69075477e-01 -6.89716339e-01 -3.25291067e-01 5.95691085e-01 -3.36891294e-01 -1.00759417e-01 -9.44704533e-01 9.69880641e-01 4.34945673e-01 1.01368916e+00 3.71212997e-02 -1.02454185e+00 4.66922581e-01 1.80695370e-01 1.64886892e-01 -8.95856798e-01 -4.97122496e-01 -6.42149270e-01 3.21145445e-01 -7.08575547e-01 1.15399271e-01 -4.21280652e-01 -1.98774755e+00 -2.73904562e-01 -2.59314448e-01 1.34505838e-01 8.20614994e-01 1.40294158e+00 4.01031762e-01 1.07055640e+00 3.32694709e-01 -6.00014806e-01 -5.18432617e-01 -8.11739564e-01 -4.90237474e-01 -8.88023004e-02 5.14619172e-01 -5.30669510e-01 -8.47822428e-01 9.86574218e-02]
[8.8857421875, 5.713723659515381]
4fd29c63-a2f0-44c0-8183-433a748c9959
query-utterance-attention-with-joint-modeling
2303.04487
null
https://arxiv.org/abs/2303.04487v3
https://arxiv.org/pdf/2303.04487v3.pdf
Query-Utterance Attention with Joint modeling for Query-Focused Meeting Summarization
Query-focused meeting summarization (QFMS) aims to generate summaries from meeting transcripts in response to a given query. Previous works typically concatenate the query with meeting transcripts and implicitly model the query relevance only at the token level with attention mechanism. However, due to the dilution of key query-relevant information caused by long meeting transcripts, the original transformer-based model is insufficient to highlight the key parts related to the query. In this paper, we propose a query-aware framework with joint modeling token and utterance based on Query-Utterance Attention. It calculates the utterance-level relevance to the query with a dense retrieval module. Then both token-level query relevance and utterance-level query relevance are combined and incorporated into the generation process with attention mechanism explicitly. We show that the query relevance of different granularities contributes to generating a summary more related to the query. Experimental results on the QMSum dataset show that the proposed model achieves new state-of-the-art performance.
['Yajing Xu', 'Bo Xiao', 'Bin Duan', 'Xingxian Liu']
2023-03-08
null
null
null
null
['meeting-summarization']
['natural-language-processing']
[ 4.11661327e-01 3.58364910e-01 -1.80776611e-01 -2.97787666e-01 -1.98091483e+00 -2.73970097e-01 7.26585150e-01 5.90522647e-01 -1.71210602e-01 7.26103723e-01 1.16812813e+00 3.00516665e-01 -4.73714992e-02 -5.65406799e-01 -5.20127416e-01 -3.42747688e-01 1.59058318e-01 6.69173360e-01 2.79108763e-01 -5.19901335e-01 6.20562494e-01 -2.19747365e-01 -1.50153911e+00 8.26777935e-01 1.15446222e+00 5.57237208e-01 5.22098362e-01 9.32813227e-01 -5.14364660e-01 9.39545453e-01 -1.18840659e+00 1.99497044e-02 -3.33443165e-01 -7.52764165e-01 -1.15323019e+00 2.67821867e-02 4.62633073e-01 -3.50207537e-01 -1.76939949e-01 6.71932280e-01 8.63005638e-01 2.51118541e-01 4.31055486e-01 -8.72624874e-01 -6.72986031e-01 9.97427046e-01 -4.47343320e-01 3.84452820e-01 8.42270613e-01 -3.78074020e-01 1.32912648e+00 -1.02233219e+00 4.92109805e-01 1.52888322e+00 -3.74005139e-02 3.82898659e-01 -7.68018425e-01 -3.90493482e-01 5.49287200e-01 3.83680195e-01 -1.32333469e+00 -5.99340439e-01 8.66868556e-01 2.00838409e-02 1.24902272e+00 6.23285174e-01 5.62777638e-01 6.19208992e-01 2.07697287e-01 1.24776256e+00 1.59157455e-01 -5.07005811e-01 -1.20021431e-02 -1.61215082e-01 3.52490515e-01 7.54928291e-02 -2.03198984e-01 -6.75237536e-01 -8.78919065e-01 -3.39291155e-01 1.81568488e-01 -1.29194126e-01 -3.49808514e-01 4.22969699e-01 -1.20496547e+00 7.35583365e-01 2.38362148e-01 4.31022346e-01 -9.24247146e-01 2.64813215e-01 5.48407555e-01 2.09242806e-01 7.52668262e-01 5.16927660e-01 -1.79710582e-01 -8.25883001e-02 -1.26125944e+00 6.91290677e-01 6.94492459e-01 1.24969041e+00 6.96459830e-01 -1.10367440e-01 -1.28717864e+00 8.29568744e-01 1.78938389e-01 3.68476570e-01 5.36179960e-01 -8.95951986e-01 6.22644722e-01 7.07016230e-01 3.93252492e-01 -7.95449317e-01 -9.28234160e-02 -7.39158273e-01 -6.98870242e-01 -9.77797389e-01 -3.74686658e-01 -1.06849790e-01 -7.22621322e-01 1.62759662e+00 2.12878600e-01 9.29261092e-03 3.46889377e-01 7.15266526e-01 1.50660908e+00 1.14907050e+00 3.50163728e-02 -7.86999047e-01 1.61714876e+00 -1.18274546e+00 -1.37607431e+00 -2.37470478e-01 6.33623064e-01 -1.01696217e+00 9.42447722e-01 -8.49313289e-02 -1.51192307e+00 -7.16885328e-01 -8.26681793e-01 -4.27154630e-01 2.68606573e-01 1.08233713e-01 4.82666976e-04 2.63351426e-02 -1.31788337e+00 1.07309846e-02 -4.14039105e-01 -5.16223013e-01 -2.48061001e-01 1.97044566e-01 2.92901844e-01 -1.59828395e-01 -1.59432471e+00 5.17043591e-01 2.54121274e-01 4.94894832e-02 -8.25900078e-01 -7.12983847e-01 -8.13701749e-01 4.70287591e-01 5.54745317e-01 -1.04762781e+00 1.93972874e+00 -5.10035217e-01 -1.24657142e+00 3.39103550e-01 -1.10580266e+00 -3.53943557e-01 7.02928752e-02 -4.80327576e-01 -9.42185819e-02 4.28210169e-01 4.97459799e-01 8.27358723e-01 5.92849195e-01 -1.11589015e+00 -6.94341540e-01 -1.83437556e-01 2.11924195e-01 8.73685837e-01 -6.62532970e-02 1.74393415e-01 -9.01225388e-01 -4.62519109e-01 2.09872425e-01 -3.14975858e-01 -8.77613425e-02 -1.05571783e+00 -5.59434831e-01 -8.52131248e-01 6.86805904e-01 -7.02784359e-01 1.85591280e+00 -1.87545049e+00 2.94930651e-03 -3.96628171e-01 -1.50011377e-02 -7.15403557e-02 -4.30958360e-01 1.36517441e+00 2.96284914e-01 2.42313236e-01 -1.98830962e-02 -6.37611091e-01 1.28937913e-02 -1.33184880e-01 -7.54912734e-01 -2.20852405e-01 4.23972309e-01 1.18036509e+00 -1.19976997e+00 -9.60075617e-01 -2.29993865e-01 3.25484216e-01 -4.70651478e-01 5.15466511e-01 -3.71724367e-01 2.81053185e-01 -8.64031255e-01 5.14847755e-01 3.31230164e-01 -5.30407131e-02 -1.90236136e-01 -2.60410577e-01 7.17535662e-03 7.83936262e-01 -7.14841545e-01 1.91913056e+00 -4.15786743e-01 3.04396778e-01 5.99001572e-02 -6.60174727e-01 9.71013367e-01 7.50412345e-01 4.06318486e-01 -8.26796889e-01 -2.05271021e-01 1.11824600e-02 -2.29155838e-01 -5.38222551e-01 1.38974929e+00 -9.00964290e-02 -4.13613588e-01 4.71107781e-01 -3.16018127e-02 -4.15916741e-01 3.99197310e-01 8.68137419e-01 9.73425686e-01 -6.87911510e-02 2.21214250e-01 -1.30651161e-01 6.30510032e-01 8.60718116e-02 4.89600390e-01 1.02743864e+00 -3.95408049e-02 7.42362022e-01 4.21928257e-01 1.80418849e-01 -6.32969379e-01 -5.44563472e-01 4.01961982e-01 1.53254342e+00 3.34255487e-01 -7.50105202e-01 -7.52245128e-01 -2.70519108e-01 -4.11829621e-01 1.00461006e+00 -4.97939914e-01 -2.79790431e-01 -5.90061545e-01 -3.58577698e-01 3.54912043e-01 3.01545441e-01 3.58680815e-01 -1.25524545e+00 -4.92289066e-01 6.82034254e-01 -1.07173526e+00 -8.90130162e-01 -1.07351255e+00 -3.01398724e-01 -7.89898753e-01 -5.19963384e-01 -1.08165205e+00 -8.03474903e-01 3.70112330e-01 5.95696747e-01 1.24470627e+00 1.58439711e-01 2.44228944e-01 5.95638156e-01 -6.47024453e-01 -6.00893557e-01 -3.91936153e-01 4.26264226e-01 -3.76753688e-01 1.36868842e-02 2.84342378e-01 -2.32999995e-01 -6.33618832e-01 -1.49481192e-01 -1.05891156e+00 1.57047257e-01 6.66379690e-01 7.54302502e-01 5.19441724e-01 -3.52033824e-01 1.31499827e+00 -6.05669379e-01 1.30405974e+00 -3.79454404e-01 8.78805667e-02 5.02113461e-01 -1.30975574e-01 1.35015935e-01 2.44923249e-01 -1.87604710e-01 -1.31570160e+00 -4.39312160e-01 -4.94586490e-02 3.57976258e-02 1.92188546e-01 8.01881135e-01 -2.79275149e-01 9.03725028e-01 2.99230099e-01 6.03725255e-01 -5.08463860e-01 -2.68692374e-01 2.88239509e-01 7.21931100e-01 3.87010813e-01 -4.90266174e-01 2.61442810e-01 3.20127010e-02 -4.60141182e-01 -9.29470897e-01 -9.25792634e-01 -9.39761996e-01 -1.74735844e-01 -3.28632265e-01 7.35771894e-01 -1.14562023e+00 -4.79682893e-01 -3.07190940e-02 -1.63556528e+00 1.76580712e-01 -4.37069416e-01 2.84938782e-01 -4.57667798e-01 5.42874277e-01 -5.74449182e-01 -1.36071920e+00 -1.01151705e+00 -1.02774405e+00 1.71551359e+00 3.67336124e-01 -4.15532440e-01 -4.72002447e-01 7.30896071e-02 3.32934141e-01 3.93466830e-01 -1.13895521e-01 7.18183517e-01 -8.70620549e-01 -7.53473043e-01 -3.57791632e-01 -1.29501373e-01 -3.06676894e-01 2.36151457e-01 -1.29725024e-01 -1.06768668e+00 -1.63432464e-01 8.42114687e-02 -2.85020679e-01 1.09818494e+00 5.51730096e-01 6.81080520e-01 -6.58569276e-01 -2.47540623e-01 -2.23124459e-01 1.09655702e+00 1.93569303e-01 5.33696234e-01 -1.24187432e-01 3.90806377e-01 8.76012504e-01 9.90107894e-01 5.42214811e-01 7.15998888e-01 6.00634038e-01 2.65728503e-01 2.93747503e-02 8.08984973e-04 -4.22807664e-01 4.20657694e-01 1.26465774e+00 4.49704438e-01 -6.67470455e-01 -4.76305813e-01 1.01340294e+00 -2.05138493e+00 -1.11903298e+00 -1.45890549e-01 2.08647180e+00 8.75165641e-01 -6.31679744e-02 -6.91759065e-02 -9.92031321e-02 7.63014972e-01 3.60997081e-01 -4.41842854e-01 -5.79046905e-01 -2.04884484e-02 -1.86668351e-01 -1.49756446e-01 7.50139952e-01 -6.28565073e-01 1.03550208e+00 6.01344728e+00 9.36895192e-01 -7.08350897e-01 -6.62299469e-02 4.47702765e-01 -2.04112008e-01 -7.76516438e-01 3.24061187e-03 -9.45972919e-01 -8.58302694e-04 1.01471591e+00 -9.44509387e-01 -1.64402291e-01 3.48585635e-01 6.02389395e-01 -3.33573073e-01 -1.17177069e+00 6.03083849e-01 2.70693362e-01 -1.20755494e+00 5.82576036e-01 -1.13649093e-01 6.56163633e-01 -3.42109978e-01 -2.31723085e-01 7.60493219e-01 -6.93898201e-02 -7.01852024e-01 6.85961664e-01 6.30584419e-01 5.01652122e-01 -8.69907260e-01 1.05336237e+00 6.40967727e-01 -1.45570850e+00 1.93261236e-01 -3.75623643e-01 -7.46017471e-02 5.64445078e-01 4.41085666e-01 -1.29078424e+00 1.14320290e+00 3.53758276e-01 4.50499922e-01 -2.48984843e-01 8.85873795e-01 -9.89340828e-04 6.53441906e-01 -1.10004194e-01 -2.08041802e-01 4.80008602e-01 1.92781195e-01 6.90948188e-01 1.37533510e+00 4.09330070e-01 3.84154856e-01 4.14464325e-01 6.83544099e-01 -1.39985874e-01 5.13478279e-01 -3.62966627e-01 -1.07160047e-01 8.00882578e-01 1.04614294e+00 -2.58916289e-01 -8.55551839e-01 -6.55382872e-02 9.74129081e-01 -1.83463115e-02 5.94848573e-01 -4.09656674e-01 -5.49765527e-01 1.81418538e-01 -9.77344513e-02 1.84249923e-01 2.03183025e-01 1.27097920e-01 -8.54938686e-01 3.13970894e-01 -7.24641562e-01 3.63817394e-01 -8.72764826e-01 -7.86315143e-01 5.38108289e-01 3.15565675e-01 -1.14680338e+00 -7.76554227e-01 8.12963486e-01 -9.03196990e-01 1.28922844e+00 -1.70788741e+00 -1.06647432e+00 -2.13268042e-01 1.22587793e-01 1.26082516e+00 3.33733439e-01 8.50975752e-01 -1.31610725e-02 -2.66135812e-01 4.66459602e-01 -2.78673351e-01 -2.57502943e-01 8.78035784e-01 -1.14328301e+00 4.26930755e-01 5.90007603e-01 -1.20014690e-01 6.79191947e-01 9.10776019e-01 -8.55386734e-01 -1.28993154e+00 -1.08781183e+00 1.83783734e+00 -2.88113117e-01 1.05202243e-01 -8.34510624e-02 -1.15789282e+00 3.41837376e-01 8.49113464e-01 -7.46387064e-01 6.77649736e-01 -4.93688099e-02 2.22182676e-01 -1.87305450e-01 -8.09110701e-01 5.19990921e-01 6.69164240e-01 -6.04131758e-01 -1.15804589e+00 3.92420679e-01 1.56360698e+00 -2.86224216e-01 -5.97067237e-01 3.82800549e-01 2.93472528e-01 -5.98266721e-01 6.83796763e-01 -3.73527825e-01 3.72138590e-01 -2.86481351e-01 -1.11739621e-01 -1.17882133e+00 -2.99126506e-01 -7.52969444e-01 -6.57341257e-02 1.59254444e+00 3.25507969e-01 -4.81591076e-02 4.54995960e-01 4.70118582e-01 -6.58260226e-01 -6.65990651e-01 -9.23403084e-01 -3.28561097e-01 -2.27305442e-01 -2.30596945e-01 8.38823020e-01 3.74807537e-01 3.37376207e-01 9.96369660e-01 -3.47396761e-01 1.28652036e-01 3.06135595e-01 2.77026623e-01 7.51076818e-01 -9.52540100e-01 3.98142673e-02 -4.23645943e-01 1.99113101e-01 -1.49059010e+00 -2.58432725e-03 -5.81756234e-01 4.74106997e-01 -2.42088604e+00 4.52522665e-01 3.24184150e-01 -1.86762065e-01 2.40398124e-02 -7.51998961e-01 -2.85920084e-01 2.30905384e-01 1.63772643e-01 -1.05320644e+00 9.96330976e-01 1.42356515e+00 -4.29977268e-01 -5.28519690e-01 -3.84435616e-02 -1.01274276e+00 1.19360298e-01 5.47818601e-01 -3.76887858e-01 -6.63660228e-01 -5.46404183e-01 8.05437714e-02 6.37590408e-01 -9.23808888e-02 -6.90687716e-01 5.72341979e-01 -1.13690615e-01 -5.30468347e-03 -1.45088732e+00 4.64611709e-01 -2.37852305e-01 -2.22202212e-01 3.06442618e-01 -9.69627142e-01 1.86350122e-01 2.64432102e-01 6.89965844e-01 -6.51664495e-01 -1.41036063e-01 -6.42585829e-02 -1.99613944e-01 -3.97673011e-01 1.27668634e-01 -5.33007085e-01 2.68526316e-01 4.30790454e-01 -5.97910658e-02 -1.17407672e-01 -1.12323570e+00 -4.04432058e-01 9.42578197e-01 -5.65364435e-02 5.18563807e-01 7.36153781e-01 -1.29359376e+00 -1.44980788e+00 -1.91166461e-01 4.82434601e-01 3.68413031e-01 6.14567935e-01 5.78197420e-01 2.32303068e-01 1.01323295e+00 5.64235568e-01 -6.80870056e-01 -1.25837862e+00 3.67916912e-01 -4.33243364e-02 -7.63656676e-01 -2.50913382e-01 9.18971002e-01 6.63164556e-01 -1.72743928e-02 3.62364024e-01 -6.77402794e-01 -6.18344724e-01 3.96356165e-01 9.53615129e-01 1.57893062e-01 1.50347665e-01 -7.33196199e-01 -3.00511092e-01 4.62792605e-01 -2.52118945e-01 -4.77151424e-01 9.00537491e-01 -5.36996365e-01 -8.49007815e-02 6.29384041e-01 1.03497386e+00 -4.61871549e-02 -6.73506558e-01 -5.74986041e-01 1.56174242e-01 -7.20774904e-02 -3.75300571e-02 -7.15332031e-01 -5.02443790e-01 7.71813571e-01 -6.09372742e-02 3.09696436e-01 1.29404688e+00 1.54107407e-01 9.52993989e-01 4.73993391e-01 6.09931834e-02 -1.05542374e+00 3.02946836e-01 6.85459197e-01 1.43953741e+00 -9.46220875e-01 -3.77870686e-02 -2.96552986e-01 -6.23503685e-01 7.43306458e-01 5.94828546e-01 1.33513853e-01 -1.07554188e-02 -2.56401360e-01 6.88536316e-02 -1.10734709e-01 -1.33955061e+00 -3.60566109e-01 5.72241545e-01 3.57955098e-01 8.08415592e-01 -1.41557351e-01 -5.81531227e-01 7.28434145e-01 -1.79189965e-01 -5.44295311e-02 5.98657727e-01 9.17400956e-01 -9.40439403e-01 -9.87367570e-01 -5.01103938e-01 4.26603794e-01 -5.87125003e-01 -3.96798402e-01 -6.06380880e-01 2.22027704e-01 -5.28058767e-01 1.56145811e+00 1.83754101e-01 -8.44734758e-02 5.47601104e-01 4.02713209e-01 1.06899679e-01 -1.06889713e+00 -1.06124151e+00 6.81037307e-01 2.59907693e-01 -3.58659953e-01 -3.89366657e-01 -4.30485338e-01 -1.33774018e+00 2.31112435e-01 -3.90812010e-01 8.85694921e-01 3.28134447e-01 8.66015196e-01 7.59511173e-01 1.00403261e+00 5.75739861e-01 -7.77711987e-01 -5.71711540e-01 -1.58500469e+00 -3.02341580e-01 1.82484478e-01 6.40090883e-01 3.59120145e-02 -5.62807173e-02 -7.64319077e-02]
[12.66324234008789, 9.423822402954102]
410fbb80-644a-42b0-a4b0-f3ec7090d4f9
ml-doctor-holistic-risk-assessment-of
2102.02551
null
https://arxiv.org/abs/2102.02551v2
https://arxiv.org/pdf/2102.02551v2.pdf
ML-Doctor: Holistic Risk Assessment of Inference Attacks Against Machine Learning Models
Inference attacks against Machine Learning (ML) models allow adversaries to learn sensitive information about training data, model parameters, etc. While researchers have studied, in depth, several kinds of attacks, they have done so in isolation. As a result, we lack a comprehensive picture of the risks caused by the attacks, e.g., the different scenarios they can be applied to, the common factors that influence their performance, the relationship among them, or the effectiveness of possible defenses. In this paper, we fill this gap by presenting a first-of-its-kind holistic risk assessment of different inference attacks against machine learning models. We concentrate on four attacks -- namely, membership inference, model inversion, attribute inference, and model stealing -- and establish a threat model taxonomy. Our extensive experimental evaluation, run on five model architectures and four image datasets, shows that the complexity of the training dataset plays an important role with respect to the attack's performance, while the effectiveness of model stealing and membership inference attacks are negatively correlated. We also show that defenses like DP-SGD and Knowledge Distillation can only mitigate some of the inference attacks. Our analysis relies on a modular re-usable software, ML-Doctor, which enables ML model owners to assess the risks of deploying their models, and equally serves as a benchmark tool for researchers and practitioners.
['Yang Zhang', 'Mario Fritz', 'Emiliano De Cristofaro', 'Michael Backes', 'Zhikun Zhang', 'Ahmed Salem', 'Xinlei He', 'Rui Wen', 'Yugeng Liu']
2021-02-04
null
null
null
null
['membership-inference-attack']
['computer-vision']
[ 1.84731588e-01 9.69770700e-02 -2.33353093e-01 -1.40373155e-01 -6.03407502e-01 -1.06889760e+00 7.20770299e-01 1.24436744e-01 -2.19966903e-01 3.33587527e-01 -3.08043003e-01 -9.90405858e-01 -1.19947366e-01 -8.73080254e-01 -8.72958601e-01 -5.64292073e-01 -5.27957343e-02 3.60454947e-01 3.80764812e-01 -2.11614862e-01 2.87340134e-01 5.56236029e-01 -1.28875947e+00 7.04580024e-02 3.90561700e-01 7.17450202e-01 -5.80843151e-01 6.03976011e-01 3.04655999e-01 9.01021063e-01 -8.54255497e-01 -9.39577281e-01 1.82281747e-01 3.75769590e-03 -7.90430546e-01 -5.32468796e-01 4.41869557e-01 -4.69230145e-01 -3.09468597e-01 1.13616574e+00 3.57881933e-01 -5.82634509e-01 5.02593040e-01 -1.73033249e+00 6.10582419e-02 8.92022967e-01 -1.47863731e-01 9.68466029e-02 9.97110605e-02 4.62224394e-01 7.84683883e-01 -1.48353308e-01 3.82369369e-01 1.27154660e+00 5.26747763e-01 7.19667673e-01 -1.33642232e+00 -1.03142166e+00 -1.45165354e-01 1.78115651e-01 -1.31137180e+00 -5.40602565e-01 5.98519325e-01 -4.36421007e-01 7.67920315e-01 6.54024899e-01 8.63923356e-02 1.43852246e+00 2.84132719e-01 3.72670621e-01 1.17207456e+00 -4.48129207e-01 2.10420400e-01 6.00148916e-01 3.08708251e-01 5.30894578e-01 5.51615000e-01 4.50798273e-01 -1.71714053e-01 -7.93192267e-01 3.77413452e-01 -4.10146207e-01 -3.14720243e-01 -3.55120331e-01 -5.64915955e-01 8.45551550e-01 9.07813162e-02 1.10367961e-01 1.97348937e-01 3.83499414e-01 5.57991505e-01 4.77465063e-01 -7.38970637e-02 5.37805259e-01 -7.19048738e-01 6.67321607e-02 -5.79122126e-01 2.75753438e-01 1.16949797e+00 4.84297544e-01 6.01702571e-01 -2.29754243e-02 3.37733328e-01 4.50774282e-02 3.56737763e-01 4.32156920e-01 2.14657038e-02 -6.39488697e-01 3.62187862e-01 3.48117650e-01 -1.07864685e-01 -1.20167291e+00 -3.89890485e-02 -3.51642430e-01 -5.15728533e-01 3.85202259e-01 5.94770014e-01 -2.99965948e-01 -4.04785842e-01 1.88708711e+00 2.18134463e-01 4.96973723e-01 1.33696049e-01 3.43073070e-01 3.25680643e-01 2.32591480e-01 2.76674539e-01 6.49831519e-02 1.37592924e+00 -2.50439048e-01 -2.39626274e-01 -2.74163514e-01 7.54220784e-01 -5.88786066e-01 8.19774389e-01 3.71396452e-01 -1.00122547e+00 -2.19336510e-01 -1.21790385e+00 4.14042592e-01 -6.26087248e-01 -4.79371369e-01 5.10893345e-01 1.34379125e+00 -6.06917977e-01 6.40684187e-01 -9.21674788e-01 7.37009868e-02 3.48621339e-01 4.71601486e-01 -2.48574734e-01 1.69503719e-01 -1.53466225e+00 1.10308921e+00 4.96512115e-01 -2.07158700e-01 -1.15204287e+00 -9.58106637e-01 -6.47433877e-01 1.16007276e-01 4.73436534e-01 -6.91549540e-01 1.09686744e+00 -4.14444238e-01 -9.84426618e-01 8.06144416e-01 2.63698012e-01 -8.55441332e-01 6.31714284e-01 -1.11310735e-01 -5.17074943e-01 1.53317517e-02 -5.37777007e-01 1.80982172e-01 9.65113878e-01 -1.30508649e+00 -3.67107868e-01 -4.03282315e-01 6.64451003e-01 -2.84806818e-01 -4.21722233e-01 4.90909487e-01 -1.10884383e-01 -2.72190750e-01 -4.09919530e-01 -1.05627513e+00 -1.27675191e-01 -2.18284220e-01 -7.23795474e-01 2.87976921e-01 1.00995600e+00 -4.05630112e-01 1.45653808e+00 -2.25668049e+00 5.11786761e-03 4.96724695e-01 1.32242650e-01 6.45068944e-01 2.82656699e-01 4.51019257e-01 -6.14627376e-02 7.08404839e-01 -1.69567108e-01 -2.99918741e-01 2.45777592e-01 3.10812086e-01 -8.59159470e-01 3.43808055e-01 1.51402980e-01 7.48139679e-01 -4.10384983e-01 -3.27330500e-01 2.64420331e-01 6.55683339e-01 -5.09982586e-01 6.66634142e-02 -3.59802574e-01 2.26293087e-01 -4.79025930e-01 4.47972566e-01 5.81247330e-01 5.48772849e-02 3.81913632e-01 -2.77700067e-01 3.28896791e-01 3.66266817e-01 -1.12552559e+00 7.16923296e-01 -4.64750290e-01 3.84909600e-01 1.81791067e-01 -5.94533086e-01 6.22476637e-01 2.64554799e-01 -2.47284561e-01 -4.99265641e-02 3.00778300e-01 1.98542818e-01 2.96851873e-01 -2.36780778e-01 6.33955076e-02 8.79448578e-02 -3.58356386e-01 6.51866376e-01 -3.12170506e-01 -7.15855360e-02 -1.50412276e-01 2.36846268e-01 1.09624755e+00 -2.65081048e-01 2.90161192e-01 -1.73577398e-01 5.93510151e-01 -2.20836803e-01 3.61891448e-01 1.02858615e+00 2.42996905e-02 4.30655591e-02 1.00068259e+00 -4.31755662e-01 -6.94015980e-01 -1.00316024e+00 -2.33092085e-01 7.47160912e-01 3.37921083e-02 -7.21117914e-01 -1.04171443e+00 -1.06077874e+00 1.49449378e-01 1.13249981e+00 -5.64601302e-01 -6.51030421e-01 -4.78004724e-01 -6.55978084e-01 1.29143977e+00 3.99852067e-01 5.09642899e-01 -7.54758000e-01 -7.45837390e-01 -2.35668600e-01 -9.25851334e-03 -1.19444788e+00 -9.90445465e-02 7.74350315e-02 -6.53676212e-01 -1.43907022e+00 6.43161237e-01 -2.14093938e-01 5.01645088e-01 -2.15751022e-01 1.17213762e+00 4.25779432e-01 -2.46415123e-01 3.39778483e-01 1.57630183e-02 -8.22861552e-01 -1.10313058e+00 -3.17511521e-02 5.74100316e-02 -1.32372826e-01 4.79729623e-01 -6.56359315e-01 -1.63713604e-01 4.79431599e-01 -1.08081770e+00 -3.79106194e-01 3.99598032e-01 4.57566470e-01 4.68033217e-02 3.82434577e-01 1.34500355e-01 -1.33925819e+00 5.33767939e-01 -3.69608372e-01 -7.94992149e-01 4.83407289e-01 -6.39063001e-01 4.18779776e-02 6.37579620e-01 -4.82410252e-01 -6.80067539e-01 -2.88669616e-01 -3.04347873e-01 -4.94403481e-01 -4.13700938e-01 4.36157018e-01 -6.54359996e-01 -3.03355992e-01 7.53468812e-01 1.49713516e-01 1.80329457e-02 -3.36149871e-01 1.14913054e-01 4.03608531e-01 2.94520140e-01 -7.67450809e-01 1.30431747e+00 3.45851988e-01 3.27018619e-01 -6.42601609e-01 -5.92250824e-01 2.50336528e-01 -3.45108509e-01 1.64701954e-01 5.33129275e-01 -5.74309230e-01 -9.28068995e-01 8.83873582e-01 -9.65349019e-01 -3.62255096e-01 1.98375866e-01 3.90521577e-03 -1.11122131e-01 4.78737414e-01 -7.06638932e-01 -7.28467524e-01 -2.49049649e-01 -1.45314324e+00 4.44297284e-01 -5.50049506e-02 -2.29690030e-01 -1.14993215e+00 -2.84934729e-01 5.66398799e-01 6.35762155e-01 3.83489519e-01 1.28522933e+00 -1.11750209e+00 -5.81423223e-01 -4.83087212e-01 1.56553090e-01 5.43176353e-01 -3.27311099e-01 3.56371880e-01 -1.22067964e+00 -3.76750320e-01 2.18420416e-01 -3.40095639e-01 6.05230510e-01 -4.94661257e-02 1.35484111e+00 -7.16786027e-01 -3.48986775e-01 6.58482194e-01 1.38070571e+00 -9.16602835e-02 9.03888702e-01 3.09928298e-01 5.53042650e-01 3.90302062e-01 3.68989736e-01 9.32172164e-02 6.78574815e-02 8.14349473e-01 8.98611188e-01 1.00469492e-01 4.65530604e-01 -2.86292970e-01 4.74631757e-01 -9.29552987e-02 2.03281045e-01 7.53317960e-03 -8.98605108e-01 -1.92932248e-01 -1.36086595e+00 -9.18061197e-01 -3.06791682e-02 2.39298511e+00 8.57553005e-01 7.23543525e-01 1.50827114e-02 3.96969527e-01 4.45542276e-01 7.90278241e-02 -4.10629928e-01 -6.02008641e-01 8.29760879e-02 2.73574054e-01 5.18225014e-01 6.73887849e-01 -1.12855279e+00 8.28762054e-01 6.20458460e+00 8.96106005e-01 -1.15291703e+00 -1.38316318e-01 5.33887804e-01 6.92087552e-03 -2.37846151e-01 3.95458996e-01 -9.72001791e-01 5.88714540e-01 1.12425601e+00 -1.46806926e-01 4.56444710e-01 9.58956599e-01 -2.37060964e-01 1.83912367e-01 -1.22761130e+00 3.99511337e-01 2.51198336e-02 -1.21472204e+00 5.23455851e-02 3.75275522e-01 -5.35961553e-05 -1.05625331e-01 2.80602336e-01 3.77611786e-01 5.35904109e-01 -1.20686412e+00 6.53049946e-01 1.95321918e-01 5.44579029e-01 -1.04369748e+00 7.30728805e-01 5.92397690e-01 -5.96022546e-01 -2.54776984e-01 -7.52684325e-02 -9.82289091e-02 -1.77846104e-01 2.61701584e-01 -8.02660108e-01 4.72555697e-01 4.70196337e-01 -9.48883444e-02 -8.74248922e-01 3.36586028e-01 -6.89137876e-01 1.09875619e+00 -4.44156736e-01 1.90140679e-01 -7.62083977e-02 2.56855756e-01 4.89855587e-01 1.08197689e+00 -1.76151037e-01 -1.07967444e-01 1.36124091e-02 7.54178524e-01 1.32921889e-01 -4.73336965e-01 -4.99728978e-01 -1.45555707e-02 8.86267364e-01 1.18603265e+00 -2.34366849e-01 -3.51279415e-02 -1.84923515e-01 3.11832935e-01 8.37485120e-02 8.34898837e-03 -1.10937572e+00 -2.12010890e-01 9.98073459e-01 1.91702411e-01 3.73616489e-03 -1.16897695e-01 -1.30631417e-01 -9.92524922e-01 -5.66122979e-02 -1.47484076e+00 5.54062784e-01 -2.66466171e-01 -1.02998018e+00 3.27650547e-01 3.27821910e-01 -6.40794277e-01 -3.68577451e-01 -4.79113907e-01 -8.31018925e-01 8.71466458e-01 -1.30564487e+00 -1.34098566e+00 1.98653683e-01 6.21226549e-01 -2.02215582e-01 -2.22518101e-01 1.11523962e+00 9.90061164e-02 -9.13684845e-01 1.08840942e+00 -2.91768909e-01 4.27900165e-01 4.37395483e-01 -9.27967429e-01 6.14453852e-01 9.05522525e-01 3.01053107e-01 1.07050812e+00 1.01237607e+00 -3.96285832e-01 -1.37155557e+00 -9.58048522e-01 6.35409236e-01 -9.64232862e-01 8.46044004e-01 -4.58276004e-01 -9.51625228e-01 8.64097953e-01 -1.47994414e-01 -1.27747208e-01 8.13459992e-01 6.88761398e-02 -9.76329327e-01 -3.41085307e-02 -1.40069425e+00 8.25946927e-01 5.83117783e-01 -7.38002241e-01 -2.87651569e-01 2.68354584e-02 6.88329995e-01 -2.36966684e-01 -9.04020786e-01 4.64313358e-01 6.34904146e-01 -1.25197279e+00 1.20622921e+00 -8.56348932e-01 2.41684258e-01 -7.16816857e-02 -2.47001916e-01 -8.85045409e-01 3.46556045e-02 -6.69550776e-01 -4.63207990e-01 1.40378225e+00 5.15012980e-01 -1.02024543e+00 6.71959877e-01 9.04245198e-01 3.39975834e-01 -7.34385490e-01 -7.43189275e-01 -5.93193889e-01 4.06257510e-01 -6.90970540e-01 9.40120995e-01 1.05804896e+00 -2.97626793e-01 1.41100168e-01 -2.52808511e-01 6.89742625e-01 9.39406812e-01 -2.30976328e-01 1.12619090e+00 -1.12631679e+00 -7.25845277e-01 -4.28315848e-01 -4.36919987e-01 -2.16976956e-01 2.35022202e-01 -6.80542111e-01 -6.74100578e-01 -6.44125044e-01 1.05063953e-01 -4.74917948e-01 -2.98432946e-01 7.62054205e-01 -1.13571174e-01 1.46497801e-01 4.29322362e-01 2.09030062e-01 -1.43492743e-01 -3.00781131e-01 4.64138538e-01 -6.31996896e-03 1.06749475e-01 3.03171158e-01 -7.35653341e-01 9.54781175e-01 1.15121305e+00 -6.06639922e-01 -5.04595339e-01 -1.64142519e-01 3.30618829e-01 -1.84204146e-01 8.50765526e-01 -1.09411097e+00 1.68927908e-01 -2.03098938e-01 -1.87595859e-02 -4.62635383e-02 3.01450819e-01 -8.88478637e-01 3.93338859e-01 8.62323284e-01 -3.62832695e-01 -1.74589753e-01 4.58140045e-01 3.42310637e-01 2.30272204e-01 -5.06424725e-01 7.40082443e-01 -7.02152848e-02 -3.16969991e-01 6.24607541e-02 -1.42182261e-01 1.38066247e-01 1.08995771e+00 6.52580187e-02 -5.96183598e-01 -2.72527337e-01 -5.96528590e-01 7.80053288e-02 6.78189874e-01 3.74537081e-01 2.34176993e-01 -7.06663072e-01 -7.48323977e-01 4.98791218e-01 -4.88423603e-03 -3.50151598e-01 1.23023674e-01 5.82215726e-01 -3.08478206e-01 1.18231148e-01 -2.02611517e-02 -3.07739079e-01 -1.80453360e+00 6.69708490e-01 4.42711562e-01 -5.44236302e-01 -1.39555126e-01 6.43528104e-01 9.10305604e-02 -4.00555193e-01 4.75534409e-01 1.86222538e-01 1.16527513e-01 -1.52094454e-01 5.04528761e-01 2.94867009e-01 4.52211574e-02 -3.35515410e-01 -4.66847271e-01 2.37343252e-01 -3.77048999e-01 1.35729298e-01 9.16523159e-01 2.46696532e-01 -1.94435805e-01 8.33624154e-02 9.32288766e-01 8.56188163e-02 -9.14458811e-01 -1.56734079e-01 -1.56676263e-01 -4.81556088e-01 -7.70875737e-02 -9.70220089e-01 -8.85412395e-01 1.02080119e+00 1.93674475e-01 5.54204583e-01 9.84592676e-01 -1.18585475e-01 5.96106648e-01 3.68975937e-01 5.08739352e-01 -6.03556931e-01 -2.04701841e-01 2.98650771e-01 5.68890691e-01 -8.54690194e-01 1.22660875e-01 -5.40552676e-01 -3.99249971e-01 8.48773897e-01 4.48411405e-01 1.20062709e-01 7.39948153e-01 6.64040327e-01 6.64879605e-02 9.67070386e-02 -8.62637758e-01 4.30790931e-01 -1.05833232e-01 5.80993772e-01 -1.28467634e-01 3.48705165e-02 2.03945458e-01 4.92298931e-01 -3.76464456e-01 -2.41790026e-01 5.17441154e-01 8.50145280e-01 -1.11888930e-01 -1.45629358e+00 -6.63108468e-01 2.46354744e-01 -8.84455919e-01 -6.38283230e-03 -6.30325675e-01 9.95658576e-01 1.61357298e-01 1.00569773e+00 -3.35202068e-01 -6.77338421e-01 2.09146455e-01 1.91584498e-01 4.50097471e-01 -3.55967820e-01 -1.00264454e+00 -4.95968282e-01 3.29853654e-01 -4.27306026e-01 3.27931702e-01 -5.94806135e-01 -7.37148285e-01 -7.74149716e-01 -3.90040189e-01 1.02233782e-01 7.17522800e-01 7.58871198e-01 3.21288466e-01 2.50038028e-01 6.94501221e-01 -4.57854837e-01 -1.48521590e+00 -5.70583522e-01 -4.04691815e-01 4.35053051e-01 1.41279221e-01 -4.56891716e-01 -9.80909169e-01 -3.09812635e-01]
[5.884347438812256, 7.436987400054932]
40f38670-11a9-47c4-b55c-c96024c0466a
making-parameter-efficient-tuning-more
null
null
https://aclanthology.org/2022.coling-1.615
https://aclanthology.org/2022.coling-1.615.pdf
Making Parameter-efficient Tuning More Efficient: A Unified Framework for Classification Tasks
Large pre-trained language models (PLMs) have demonstrated superior performance in industrial applications. Recent studies have explored parameter-efficient PLM tuning, which only updates a small amount of task-specific parameters while achieving both high efficiency and comparable performance against standard fine-tuning. However, all these methods ignore the inefficiency problem caused by the task-specific output layers, which is inflexible for us to re-use PLMs and introduces non-negligible parameters. In this work, we focus on the text classification task and propose plugin-tuning, a framework that further improves the efficiency of existing parameter-efficient methods with a unified classifier. Specifically, we re-formulate both token and sentence classification tasks into a unified language modeling task, and map label spaces of different tasks into the same vocabulary space. In this way, we can directly re-use the language modeling heads of PLMs, avoiding introducing extra parameters for different tasks. We conduct experiments on six classification benchmarks. The experimental results show that plugin-tuning can achieve comparable performance against fine-tuned PLMs, while further saving around 50% parameters on top of other parameter-efficient methods.
['Wei Wu', 'Rui Xie', 'Xuanjing Huang', 'Qi Zhang', 'Tao Gui', 'Xuanting Chen', 'Yicheng Zou', 'Ruotian Ma', 'Xin Zhou']
null
null
null
null
coling-2022-10
['sentence-classification']
['natural-language-processing']
[-2.13412300e-01 -3.84254724e-01 -4.76414204e-01 -5.28874278e-01 -8.90421987e-01 -5.29736340e-01 3.08928728e-01 2.29267851e-02 -6.44112527e-01 6.69724464e-01 -1.65010080e-01 -5.29926360e-01 2.74409145e-01 -5.00378907e-01 -4.05081838e-01 -5.77058971e-01 5.30057251e-01 4.74883616e-01 3.31612855e-01 1.28871307e-01 1.78018019e-01 2.37641875e-02 -1.02702069e+00 2.04609692e-01 8.91454875e-01 7.76122391e-01 6.31317735e-01 2.95669347e-01 -5.45885980e-01 4.08437729e-01 -7.70580649e-01 -3.03449810e-01 -1.34199515e-01 5.54298516e-03 -7.70824254e-01 8.48487392e-02 3.03498834e-01 -1.55378073e-01 4.72134352e-02 8.66702974e-01 3.99365783e-01 1.74275503e-01 5.18196940e-01 -1.19558835e+00 -4.40529943e-01 7.87968636e-01 -6.54994726e-01 -1.40258759e-01 -3.11520576e-01 -7.99812973e-02 8.71774554e-01 -7.47911036e-01 3.07654552e-02 1.58514714e+00 6.59432232e-01 5.43509007e-01 -1.33593583e+00 -1.11652446e+00 6.01300061e-01 -2.53542781e-01 -1.32405293e+00 -3.37892920e-01 4.23441559e-01 -5.17019033e-01 1.42177427e+00 -9.68124270e-02 -5.32167815e-02 9.90243614e-01 4.64357644e-01 8.39010954e-01 8.04525673e-01 -4.23570991e-01 1.43020302e-01 5.12649775e-01 4.33035791e-01 6.12115443e-01 2.34149337e-01 -4.77699310e-01 -2.29261339e-01 -3.39087218e-01 7.49816656e-01 2.04606563e-01 1.01690955e-01 -3.91253263e-01 -1.40807080e+00 9.51615632e-01 6.64358661e-02 5.57833731e-01 7.49361441e-02 3.21854174e-01 9.10328448e-01 2.65192389e-01 8.04694891e-01 4.87457871e-01 -1.02248108e+00 -2.35615209e-01 -9.20799792e-01 -9.84571204e-02 7.12707341e-01 1.10623646e+00 8.99686158e-01 1.35022581e-01 -4.07634944e-01 1.36084247e+00 2.50825286e-01 3.69460255e-01 9.12027121e-01 -6.45421624e-01 7.59662032e-01 6.59601450e-01 -6.19392321e-02 -4.76441205e-01 -5.48225939e-01 -7.28628635e-01 -9.03377533e-01 -3.66023749e-01 8.01142976e-02 -2.86421627e-01 -7.62209237e-01 1.90922368e+00 4.87755463e-02 2.29542583e-01 -1.03979282e-01 4.26404476e-01 4.12878722e-01 8.33112061e-01 4.81441975e-01 -1.10018820e-01 1.59880042e+00 -1.72259676e+00 -7.66949058e-01 -5.05647898e-01 1.23646498e+00 -9.07203496e-01 1.70583558e+00 3.57537717e-01 -8.46254826e-01 -7.43784189e-01 -1.14599597e+00 -1.67667985e-01 -3.35995615e-01 6.94251597e-01 7.58232236e-01 6.10868394e-01 -5.91171741e-01 4.76986885e-01 -8.90657246e-01 -1.99015051e-01 2.57396489e-01 5.26955366e-01 -1.01748511e-01 2.05436394e-01 -1.15960515e+00 7.94589579e-01 5.92216134e-01 -1.44522801e-01 -7.85204291e-01 -9.29232121e-01 -8.45997214e-01 2.19375804e-01 5.97797990e-01 -8.16411555e-01 1.58527792e+00 -4.70170707e-01 -1.73694146e+00 6.78758204e-01 -4.49511230e-01 -1.94030151e-01 2.13155061e-01 -4.18410987e-01 -3.53108346e-01 -5.79461753e-01 -7.46653927e-03 5.33096611e-01 7.22737849e-01 -9.95302379e-01 -7.84488022e-01 -5.48795564e-03 1.15469985e-01 1.68567285e-01 -1.02954590e+00 1.57438338e-01 -9.11328256e-01 -7.42750943e-01 -2.54561096e-01 -9.13537562e-01 -4.18691874e-01 -4.63014901e-01 -3.52237314e-01 -5.62016904e-01 6.79118812e-01 -3.29028279e-01 1.68748200e+00 -2.17437458e+00 -1.78898409e-01 -1.78300723e-01 2.23261848e-01 6.36193514e-01 -3.05659026e-01 2.33797595e-01 1.50681585e-01 2.83413470e-01 -2.33081579e-01 -9.74122882e-01 2.95445383e-01 3.51610333e-01 -3.52786958e-01 9.05747041e-02 -1.72366753e-01 8.98383915e-01 -7.38808036e-01 -5.70382118e-01 3.19711238e-01 1.69094786e-01 -5.55247009e-01 5.16919196e-02 -4.44989383e-01 1.38702437e-01 -6.60082161e-01 6.52340874e-02 6.50206447e-01 -4.96089876e-01 2.08513349e-01 -1.29357502e-01 5.94733506e-02 7.52666593e-01 -1.20293558e+00 1.89123881e+00 -1.26438868e+00 2.88655102e-01 1.15730114e-01 -9.72274542e-01 9.48321939e-01 2.73914933e-01 1.97029352e-01 -4.45678771e-01 1.39583319e-01 2.00962692e-01 -3.93031627e-01 -1.76408634e-01 4.87301320e-01 -8.18699077e-02 -4.18965697e-01 6.50401294e-01 2.11133972e-01 -1.88303757e-02 1.86601534e-01 -4.05283719e-02 7.08404064e-01 5.57632893e-02 4.11968082e-01 -4.05849755e-01 7.60185003e-01 -4.40264583e-01 6.85080886e-01 7.04286516e-01 1.03783280e-01 1.02861188e-01 3.35756302e-01 -2.26820976e-01 -8.54161203e-01 -8.43644142e-01 -1.53897792e-01 1.74276114e+00 1.13296434e-02 -9.04162228e-01 -8.81116033e-01 -7.75440335e-01 4.03795280e-02 8.27260673e-01 -2.58249193e-01 -1.47005826e-01 -6.68707192e-01 -1.00448573e+00 6.02194369e-01 5.65226257e-01 5.24170876e-01 -8.96335363e-01 -2.11391859e-02 4.04881895e-01 -2.88832307e-01 -1.34036112e+00 -8.07630301e-01 2.54034072e-01 -1.08666480e+00 -3.53913486e-01 -6.44134641e-01 -1.04987633e+00 5.12781560e-01 3.33205909e-01 1.18261278e+00 -9.24335644e-02 7.03646382e-03 -3.86318229e-02 5.75646805e-03 -4.48694527e-01 -5.45072079e-01 8.01229954e-01 6.61249533e-02 -1.89389139e-01 5.08414030e-01 -4.68985707e-01 -2.41299927e-01 5.37529290e-01 -7.32276618e-01 1.35089278e-01 5.40811062e-01 7.33995676e-01 4.92889851e-01 4.83279377e-02 8.05439651e-01 -1.10016572e+00 7.78344095e-01 -2.29885876e-01 -5.48726916e-01 4.88177866e-01 -8.24788094e-01 2.16039956e-01 8.00877631e-01 -7.62868047e-01 -1.11543322e+00 -1.03410557e-01 -1.28070444e-01 -2.64745742e-01 -4.42230236e-03 3.38291436e-01 -3.09877813e-01 2.66769201e-01 3.07732552e-01 1.40466034e-01 -2.70968616e-01 -8.88213933e-01 2.31030464e-01 1.06559384e+00 1.71268478e-01 -9.80947077e-01 6.50458455e-01 1.67667195e-01 -4.44570780e-01 -5.40073931e-01 -1.05449104e+00 -7.93489277e-01 -5.48960268e-01 4.96147573e-01 5.46000183e-01 -1.18322742e+00 -6.25275433e-01 4.90550220e-01 -1.11399543e+00 -6.16936028e-01 8.76793116e-02 5.99818230e-01 -4.74640995e-01 5.64393818e-01 -8.11137080e-01 -4.87091452e-01 -5.36672175e-01 -1.39566743e+00 1.28908300e+00 -6.93988428e-02 -2.75786310e-01 -1.29685140e+00 -1.76129878e-01 3.61450583e-01 5.40165901e-01 -5.75779498e-01 1.07224619e+00 -8.06738198e-01 -1.75457940e-01 -4.17671315e-02 -1.92664832e-01 6.26890957e-01 2.83613592e-01 -3.48646700e-01 -1.08987522e+00 -6.08823955e-01 6.47693574e-02 -2.76445568e-01 1.00977790e+00 3.01557392e-01 1.46369791e+00 -3.12865786e-02 -6.75719261e-01 7.13716447e-01 1.14422119e+00 -1.67646986e-02 1.30796030e-01 3.58027309e-01 7.97926068e-01 3.27090859e-01 6.75758719e-01 3.11115772e-01 4.48431790e-01 8.85814309e-01 -1.33580312e-01 -3.07527930e-01 1.58102736e-02 -2.92785555e-01 6.21581852e-01 1.20641327e+00 4.71338630e-01 -3.34520549e-01 -7.22535670e-01 5.05216300e-01 -2.03817916e+00 -1.92353219e-01 1.50226519e-01 2.26181507e+00 1.21462071e+00 4.44608510e-01 -3.54206525e-02 -1.63664937e-01 7.69275069e-01 1.06121033e-01 -5.46432495e-01 -5.41725457e-01 3.79072189e-01 1.32342190e-01 4.80734855e-01 6.14772439e-01 -1.26940656e+00 1.31194043e+00 6.17438745e+00 1.57241511e+00 -1.20959055e+00 5.00128448e-01 3.62116486e-01 -2.35839620e-01 -1.36109650e-01 -2.93726120e-02 -1.48856521e+00 4.44213599e-01 1.20166028e+00 -3.54235470e-01 2.06377774e-01 1.08656621e+00 2.98577994e-01 1.94795623e-01 -1.26898348e+00 1.04142869e+00 -7.22819120e-02 -9.83210146e-01 1.52027398e-01 -2.87057422e-02 6.11915529e-01 1.48399323e-01 -5.95507305e-03 8.67471218e-01 4.27085102e-01 -8.23349595e-01 5.40617168e-01 4.22058217e-02 7.66720593e-01 -6.04076445e-01 8.13422441e-01 6.54792607e-01 -1.35900903e+00 -9.25158039e-02 -7.59038031e-01 -7.84248412e-02 2.22981796e-01 8.74832332e-01 -8.35416317e-01 4.11966622e-01 5.45684695e-01 5.61410129e-01 -5.21276534e-01 7.39407718e-01 -1.34720504e-01 8.07088017e-01 -2.50971586e-01 1.38630276e-03 3.05496812e-01 -6.73777238e-02 1.65945187e-01 1.67228496e+00 1.76836953e-01 -6.28204226e-01 5.99933803e-01 7.81855285e-01 -2.20681578e-01 3.21088612e-01 -1.54145792e-01 4.38704751e-02 5.95354199e-01 1.43137813e+00 -6.50823832e-01 -5.97055137e-01 -5.23101151e-01 8.66780043e-01 3.24912995e-01 2.87931472e-01 -1.00690114e+00 -3.76520157e-01 9.42386627e-01 -1.35162979e-01 2.06183195e-01 -4.88302261e-01 -3.45705420e-01 -1.30048001e+00 1.07109889e-01 -7.63559520e-01 2.55850822e-01 -2.97293425e-01 -1.27647114e+00 5.77648401e-01 7.80970231e-02 -1.10704494e+00 -2.45629445e-01 -6.21436357e-01 -3.21322918e-01 1.00738358e+00 -1.67787576e+00 -1.28804171e+00 -1.15947677e-02 3.97014707e-01 9.33396637e-01 -1.51421383e-01 9.97702777e-01 5.24601460e-01 -9.54441071e-01 1.11792505e+00 4.25960481e-01 3.95406038e-02 1.30100310e+00 -1.19954383e+00 5.85304201e-01 4.03918177e-01 -8.44897106e-02 1.05719364e+00 2.81018376e-01 -4.03010756e-01 -8.63660693e-01 -1.43439424e+00 1.20170319e+00 -4.29933071e-01 6.70667708e-01 -9.63685930e-01 -9.03911412e-01 7.60250688e-01 2.79487129e-02 -1.80634588e-01 7.87661374e-01 5.18479586e-01 -3.67234319e-01 -3.11107188e-01 -6.58398151e-01 6.24036670e-01 8.86696160e-01 -5.34619689e-01 -4.30373430e-01 4.39978153e-01 1.07962537e+00 -2.47479379e-01 -9.47031677e-01 3.51704419e-01 3.24631602e-01 -3.91985714e-01 9.09665227e-01 -5.81195891e-01 4.71604615e-02 -1.58386037e-01 4.90780398e-02 -1.39820242e+00 -3.90809685e-01 -5.39006948e-01 2.05816422e-02 1.63569784e+00 5.17850935e-01 -1.00070441e+00 4.29960847e-01 3.81300867e-01 -2.01718763e-01 -7.01016426e-01 -5.93847573e-01 -1.11788321e+00 5.04038334e-01 -5.48589110e-01 7.02764869e-01 9.15787399e-01 -1.08875915e-01 7.79030383e-01 -3.43434483e-01 -1.00574277e-01 3.25940043e-01 2.97493935e-01 8.14429641e-01 -1.12991738e+00 -4.98483926e-01 -6.01220250e-01 2.30291396e-01 -1.51463747e+00 5.18170536e-01 -1.04267943e+00 1.11482903e-01 -1.40217340e+00 4.73386407e-01 -7.19808042e-01 -5.11764050e-01 9.27121937e-01 -3.52313906e-01 8.53320286e-02 2.34683543e-01 2.62330621e-01 -8.64985704e-01 5.37668347e-01 1.14275622e+00 -2.41964802e-01 -2.82961041e-01 3.02725900e-02 -7.88097084e-01 9.61369336e-01 9.49278951e-01 -6.37397707e-01 -6.27722025e-01 -7.31219113e-01 1.84125155e-02 -4.22275662e-01 -1.91111952e-01 -8.26464891e-01 1.36208773e-01 -1.34826526e-01 -1.08581111e-02 -2.87983179e-01 1.74685821e-01 -6.89892352e-01 -2.72092849e-01 2.26884723e-01 -5.03030241e-01 -1.31719291e-01 5.48363864e-01 3.17068338e-01 -2.21651092e-01 -5.42900920e-01 7.06091881e-01 4.79240268e-02 -6.08111441e-01 2.02921450e-01 -4.22524244e-01 -6.77241907e-02 8.96017969e-01 3.93905304e-02 -1.73982114e-01 6.96934164e-02 -4.84261304e-01 4.61805075e-01 3.15852284e-01 7.58943379e-01 -3.57185937e-02 -1.18290412e+00 -5.17552435e-01 1.22848280e-01 1.71192810e-01 1.12662174e-01 2.46279210e-01 6.39558733e-01 1.12751476e-03 9.61347580e-01 3.15151632e-01 -5.87696195e-01 -1.20721173e+00 7.39008844e-01 2.84686178e-01 -9.10967767e-01 -4.85434264e-01 6.25936329e-01 8.61028731e-01 -9.51131523e-01 4.14194256e-01 -7.05323160e-01 -1.31527722e-01 -1.78633317e-01 3.81428480e-01 2.08491072e-01 2.76853800e-01 -6.69875070e-02 -3.46393406e-01 7.06392884e-01 -3.57871294e-01 2.64230758e-01 9.78123784e-01 -2.69018173e-01 -7.76797310e-02 7.83548653e-01 1.28125393e+00 6.44451678e-02 -9.83708382e-01 -6.39908791e-01 8.33313819e-03 -3.40193436e-02 2.74361670e-01 -7.60756314e-01 -9.40808475e-01 1.11307871e+00 2.34695822e-01 -8.98700282e-02 1.01809597e+00 -8.66148546e-02 1.17614436e+00 6.31318569e-01 4.20226395e-01 -1.30109227e+00 -4.75204960e-02 7.46073186e-01 4.61690724e-01 -1.07913423e+00 -9.14525092e-02 -6.64478719e-01 -4.98601735e-01 1.03229463e+00 8.04652512e-01 1.05927773e-01 6.88657701e-01 6.09420955e-01 1.01792356e-02 3.08639050e-01 -1.04391909e+00 1.63263008e-01 1.36636525e-01 1.88137665e-01 8.03479552e-01 2.23789424e-01 -4.16196764e-01 1.06196475e+00 -3.38803530e-02 1.49591759e-01 3.33727896e-02 7.47748911e-01 -4.40317214e-01 -1.57607019e+00 -1.44780055e-01 4.44655448e-01 -5.65842450e-01 -3.00031245e-01 1.70323417e-01 6.71056628e-01 1.17444888e-01 9.89919305e-01 9.81154218e-02 -3.14445406e-01 2.62569070e-01 2.61359930e-01 2.38439143e-01 -8.36718440e-01 -7.02017069e-01 2.44718239e-01 1.14105232e-01 -3.93664181e-01 -1.28920719e-01 -1.47954226e-01 -1.29957235e+00 -2.26864636e-01 -5.89200914e-01 3.49108249e-01 7.04314232e-01 1.00166917e+00 5.44966578e-01 8.80137026e-01 4.58741963e-01 -6.88931584e-01 -1.03473246e+00 -1.27266479e+00 -5.00574648e-01 1.11514412e-01 1.77848656e-02 -8.20276558e-01 -2.43268982e-01 2.27049626e-02]
[10.885740280151367, 8.311603546142578]
c71b5a87-6ca3-4599-918d-fe659bff0358
data-augmentation-for-low-resource-keyphrase
2305.17968
null
https://arxiv.org/abs/2305.17968v1
https://arxiv.org/pdf/2305.17968v1.pdf
Data Augmentation for Low-Resource Keyphrase Generation
Keyphrase generation is the task of summarizing the contents of any given article into a few salient phrases (or keyphrases). Existing works for the task mostly rely on large-scale annotated datasets, which are not easy to acquire. Very few works address the problem of keyphrase generation in low-resource settings, but they still rely on a lot of additional unlabeled data for pretraining and on automatic methods for pseudo-annotations. In this paper, we present data augmentation strategies specifically to address keyphrase generation in purely resource-constrained domains. We design techniques that use the full text of the articles to improve both present and absent keyphrase generation. We test our approach comprehensively on three datasets and show that the data augmentation strategies consistently improve the state-of-the-art performance. We release our source code at https://github.com/kgarg8/kpgen-lowres-data-aug.
['Cornelia Caragea', 'Jishnu Ray Chowdhury', 'Krishna Garg']
2023-05-29
null
null
null
null
['keyphrase-generation']
['natural-language-processing']
[ 1.57724023e-01 1.75913990e-01 -6.16494894e-01 7.73656229e-03 -1.21547210e+00 -7.99871147e-01 8.53768468e-01 5.07440746e-01 -4.45282161e-01 9.93194878e-01 7.32426047e-01 -1.11381181e-01 3.06876361e-01 -5.18768668e-01 -8.95835519e-01 -2.43948266e-01 1.69284269e-01 3.61353487e-01 1.58390984e-01 -3.98034155e-01 5.02561688e-01 5.18220849e-02 -1.37733936e+00 5.02224386e-01 8.94360602e-01 5.60549319e-01 3.56411248e-01 8.22579801e-01 -4.23959225e-01 8.75641823e-01 -5.98035693e-01 -5.15463769e-01 3.17122787e-01 -4.68133211e-01 -1.02128828e+00 -5.00258466e-04 7.66168475e-01 -3.91095668e-01 -5.24523675e-01 1.05054569e+00 4.05146748e-01 2.52496880e-02 4.78501439e-01 -1.32756269e+00 -8.04349124e-01 1.33094001e+00 -6.21822834e-01 4.07720655e-01 3.62181604e-01 -3.27779531e-01 1.44632578e+00 -1.26068485e+00 7.51333773e-01 8.86073232e-01 5.26138425e-01 4.78682637e-01 -9.81913865e-01 -8.28194380e-01 2.21542895e-01 2.35503361e-01 -1.22255898e+00 -4.30815578e-01 8.36568356e-01 -2.53469586e-01 6.76175416e-01 2.89987653e-01 5.60854912e-01 1.26872981e+00 -1.44755110e-01 1.26281631e+00 1.00622964e+00 -6.03122652e-01 -2.25277975e-01 7.70551115e-02 3.78780216e-01 4.62195843e-01 4.66252446e-01 -4.53489900e-01 -7.56088555e-01 -1.50396109e-01 6.80228174e-01 -6.62175789e-02 -3.12572300e-01 1.33414464e-02 -1.84981716e+00 8.00133586e-01 1.47678062e-01 2.27009818e-01 -5.41201293e-01 -7.21763372e-02 3.70800108e-01 1.76272795e-01 8.09025764e-01 1.21451366e+00 -7.47897387e-01 -1.79972962e-01 -1.17354596e+00 7.90456891e-01 1.02342474e+00 1.21089971e+00 7.83929229e-01 -3.29885304e-01 -5.74600279e-01 8.48369122e-01 -2.81924695e-01 3.01389873e-01 4.37764823e-01 -6.81017816e-01 8.66718173e-01 6.33907378e-01 4.33686227e-01 -8.99829924e-01 -2.42265180e-01 -2.90358216e-01 -7.01698840e-01 -6.33792818e-01 3.15412223e-01 -5.93873441e-01 -1.12257349e+00 1.42377794e+00 1.81412399e-01 1.73396409e-01 4.85736579e-02 5.95919371e-01 1.36297202e+00 9.94210303e-01 5.75218499e-02 -2.22344056e-01 1.46571648e+00 -1.42756772e+00 -1.08125150e+00 -3.95340919e-01 5.44586539e-01 -1.06842554e+00 1.37625337e+00 2.01862976e-01 -1.12770045e+00 -3.74793917e-01 -8.53294015e-01 -4.29600716e-01 -6.94231033e-01 5.65931201e-01 5.58288217e-01 -1.02228589e-01 -6.99649096e-01 3.88427883e-01 -4.24332082e-01 -3.49584550e-01 2.34514982e-01 -1.60042346e-01 -3.52385849e-01 8.87647942e-02 -1.51506710e+00 7.35133946e-01 7.63269901e-01 -2.71602869e-01 -6.94923937e-01 -1.14535809e+00 -7.54730821e-01 -1.42444313e-01 1.04082584e+00 -6.49011075e-01 1.62833381e+00 -4.53446448e-01 -1.01786542e+00 9.47168648e-01 -1.45839959e-01 -4.87392068e-01 3.81111443e-01 -8.14655006e-01 -2.57831037e-01 2.65173703e-01 4.80258852e-01 9.64930356e-01 7.17395842e-01 -1.24003410e+00 -9.81177568e-01 8.14795420e-02 3.49663258e-01 4.83160913e-01 -6.67652786e-01 1.87439248e-01 -8.80159497e-01 -1.39038563e+00 -3.09242249e-01 -8.53909671e-01 -2.74866313e-01 -5.26644886e-01 -8.67713332e-01 -3.29152256e-01 6.62259340e-01 -7.85575926e-01 1.66627836e+00 -1.78518295e+00 3.64256091e-02 -4.55477506e-01 2.83591181e-01 1.42111138e-01 -1.79021388e-01 7.68955529e-01 -6.77782521e-02 4.20975357e-01 -1.48489118e-01 -2.89122999e-01 -2.24584937e-02 -1.88142620e-02 -8.50446463e-01 -2.36712098e-01 2.96762764e-01 1.08364880e+00 -1.03868675e+00 -6.12926841e-01 -2.47352213e-01 -1.05258346e-01 -3.50577921e-01 3.56398851e-01 -5.68600357e-01 1.13705069e-01 -5.68576634e-01 4.94255990e-01 3.56518567e-01 -4.62319642e-01 -1.86139986e-01 -3.71181458e-01 -1.88116357e-01 5.42422831e-01 -1.14519536e+00 1.71238542e+00 -2.62213647e-01 4.85360414e-01 -1.27222598e-01 -6.62401319e-01 6.16278708e-01 3.59418958e-01 6.26227736e-01 -3.11033934e-01 3.29240151e-02 2.86309361e-01 -3.79296660e-01 -1.41406104e-01 1.31734061e+00 2.10793465e-01 -4.28342640e-01 5.93020082e-01 2.39204034e-01 -3.08071256e-01 8.80112469e-01 7.90698588e-01 9.05978441e-01 -7.64492378e-02 4.42251354e-01 -2.66650617e-01 2.59545028e-01 4.10256118e-01 2.79288143e-01 1.03196931e+00 3.91124070e-01 7.56007433e-01 5.38892806e-01 -2.25995153e-01 -1.31640828e+00 -5.59868634e-01 1.54943690e-01 1.29165208e+00 1.31551594e-01 -1.03009868e+00 -7.13606477e-01 -8.88865948e-01 5.94600290e-02 8.67443383e-01 -7.32980132e-01 7.34512284e-02 -5.24355233e-01 -7.19782829e-01 4.50692534e-01 4.85911548e-01 2.91352302e-01 -1.18889177e+00 -2.47825161e-01 1.85231835e-01 -5.12057245e-01 -1.54367530e+00 -7.03116059e-01 1.15172684e-01 -5.62194884e-01 -9.87089753e-01 -1.16065073e+00 -6.87768340e-01 7.79606342e-01 4.14318591e-01 1.45115101e+00 5.76037588e-03 -3.47767472e-02 3.37496907e-01 -8.35461080e-01 -8.35034370e-01 -4.15791482e-01 6.22065365e-01 -5.57366610e-02 -2.99085200e-01 3.37858021e-01 -2.22823367e-01 -3.53405803e-01 -1.19327351e-01 -9.03706074e-01 6.29243076e-01 7.45075881e-01 7.76129067e-01 6.53936207e-01 -4.83974107e-02 5.60948431e-01 -1.23814571e+00 1.09869671e+00 -5.51881790e-01 -4.04692292e-01 2.54022449e-01 -5.93703151e-01 1.54715493e-01 7.00637519e-01 -5.11487842e-01 -8.47319782e-01 -2.02331692e-01 9.66209769e-02 -3.14420193e-01 -7.86598306e-04 7.25138366e-01 2.52342254e-01 4.95343834e-01 5.91459751e-01 3.21176380e-01 -4.73734409e-01 -7.16416001e-01 8.06371987e-01 5.32453060e-01 6.29376650e-01 -8.04394603e-01 1.05743611e+00 1.98973835e-01 -3.32274228e-01 -7.00951099e-01 -1.52751386e+00 -7.08140492e-01 -6.34943306e-01 2.63089743e-02 5.12249708e-01 -1.24916172e+00 6.72190562e-02 3.99726093e-01 -1.12159622e+00 -3.19995493e-01 -5.88675976e-01 2.43662581e-01 -2.40393549e-01 4.29389477e-01 -5.53505301e-01 -2.50760287e-01 -8.62948537e-01 -7.58218169e-01 1.14287806e+00 3.51198554e-01 -3.22273284e-01 -6.36759698e-01 6.55164719e-02 2.35584468e-01 1.67251915e-01 6.34698197e-02 6.41238689e-01 -1.09966481e+00 -3.59763026e-01 -3.17518055e-01 -2.88875222e-01 1.65728152e-01 3.77012849e-01 2.54599862e-02 -5.43649793e-01 -6.29315013e-03 -5.28976381e-01 -6.64588749e-01 1.16633666e+00 2.09398597e-01 1.36773562e+00 -7.72117257e-01 -3.18357885e-01 2.41289988e-01 1.01580608e+00 -1.84056357e-01 2.79783338e-01 4.19332325e-01 1.01422751e+00 5.11619806e-01 9.04359758e-01 4.91914272e-01 7.32036114e-01 3.79662722e-01 -9.22545511e-03 -1.98742494e-01 -2.54735321e-01 -5.70088923e-01 1.83740243e-01 8.81666601e-01 -3.01043876e-02 -4.34324980e-01 -9.66962039e-01 1.06245697e+00 -1.98548377e+00 -8.83595645e-01 -9.59837586e-02 1.79746318e+00 1.58204532e+00 2.23284364e-01 2.54387468e-01 -5.46883419e-02 6.85417771e-01 4.08101946e-01 -2.67538249e-01 7.42197260e-02 -1.96100086e-01 1.56762421e-01 5.42258859e-01 3.74371082e-01 -1.38035893e+00 1.38110304e+00 6.21630096e+00 7.73237646e-01 -8.02654922e-01 -1.00097038e-01 5.50826311e-01 -8.80271271e-02 -3.66236210e-01 -5.31268157e-02 -1.27297318e+00 4.38237667e-01 7.69729018e-01 -6.38439655e-01 9.29805562e-02 1.01366055e+00 4.26974259e-02 -1.25948086e-01 -1.03273690e+00 8.58404517e-01 7.56313652e-02 -1.66406786e+00 3.83929551e-01 -3.06065083e-01 1.06875491e+00 2.95944717e-02 -1.09265810e-02 3.40735227e-01 5.09079814e-01 -5.83684504e-01 7.29743958e-01 2.73672014e-01 7.75764406e-01 -6.14047706e-01 6.26001358e-01 2.36332476e-01 -1.04205811e+00 2.11908326e-01 -1.86448276e-01 1.52035775e-02 1.58798203e-01 6.98450089e-01 -8.39633763e-01 5.68311334e-01 7.13691473e-01 9.06530082e-01 -8.86862695e-01 8.40237558e-01 -6.26453221e-01 7.00040102e-01 -2.59597480e-01 -2.20287591e-01 3.50866139e-01 1.69747010e-01 6.55212164e-01 1.52473664e+00 1.68074489e-01 1.53983206e-01 4.70393449e-01 6.22093737e-01 -6.78937733e-01 3.52524638e-01 -4.12749559e-01 -6.42070413e-01 6.41699314e-01 1.64828265e+00 -8.09669733e-01 -8.59720409e-01 -4.37897891e-01 8.05030107e-01 1.92238450e-01 1.78849727e-01 -6.30218983e-01 -5.49012184e-01 3.92271638e-01 1.71921581e-01 3.19705546e-01 -1.62423179e-01 -1.92883119e-01 -1.54589939e+00 1.36520371e-01 -1.23347402e+00 5.18561482e-01 -8.53376687e-01 -1.38974154e+00 4.01064932e-01 4.35422033e-01 -1.12410235e+00 -3.66539836e-01 -3.06748420e-01 -4.56914485e-01 6.58358634e-01 -1.59689629e+00 -1.25805581e+00 -3.38710099e-01 3.40720117e-01 9.44175124e-01 -1.27363160e-01 7.07920134e-01 1.26447588e-01 -4.67097759e-01 4.66354519e-01 -1.09491557e-01 3.71034294e-01 1.07868731e+00 -1.53833687e+00 9.31265473e-01 1.24132288e+00 3.96744162e-01 6.83129191e-01 8.82184088e-01 -8.81488502e-01 -1.26964498e+00 -9.29207027e-01 1.01511192e+00 -6.17042542e-01 1.13783360e+00 -5.42831123e-01 -9.00556266e-01 7.46603429e-01 6.09704971e-01 -7.76117370e-02 5.35099030e-01 2.26114675e-01 -3.09540451e-01 2.24093616e-01 -5.56833804e-01 1.00918114e+00 7.89690554e-01 -3.60193700e-01 -9.64189768e-01 5.34344733e-01 1.04612374e+00 -7.18612850e-01 -7.10461855e-01 4.58410859e-01 3.06795299e-01 -2.08077088e-01 8.25621188e-01 -7.70505130e-01 9.94939089e-01 -3.48076493e-01 1.10301897e-01 -1.31429958e+00 -2.93307781e-01 -8.02487135e-01 -6.28467202e-01 1.34442759e+00 7.29321480e-01 -1.39321774e-01 7.16948330e-01 3.53119045e-01 -5.54701760e-02 -7.11586535e-01 -1.35198101e-01 -3.90103191e-01 -8.73586312e-02 -1.59208015e-01 4.07108605e-01 1.18220556e+00 2.43961494e-02 8.18113089e-01 -6.09265625e-01 -3.01656052e-02 4.03209865e-01 2.58687913e-01 1.01441634e+00 -1.07128203e+00 8.62665921e-02 -3.54552060e-01 3.76932621e-01 -1.09477770e+00 4.24375050e-02 -7.77128339e-01 5.13104200e-02 -1.95419645e+00 4.54064190e-01 -1.45984143e-01 -2.59608209e-01 8.73777986e-01 -8.78794968e-01 1.91643700e-01 2.69040912e-01 3.73376369e-01 -7.52999425e-01 4.05155212e-01 1.35323691e+00 -1.35167524e-01 -2.60839850e-01 -1.89832360e-01 -1.23816645e+00 6.47806346e-01 8.67974579e-01 -5.14055908e-01 -4.82531399e-01 -2.67019749e-01 4.01797056e-01 -3.59614968e-01 2.12700889e-01 -7.57726014e-01 2.38485202e-01 -3.09975177e-01 3.17562997e-01 -9.02729452e-01 6.74137846e-02 -4.85218436e-01 -2.97324061e-01 2.44212300e-02 -7.28086829e-01 2.99495876e-01 4.58283305e-01 3.68834615e-01 -2.54091144e-01 -3.59567702e-01 3.65733951e-01 -4.23910111e-01 -9.11138654e-01 5.22068739e-01 -2.21349955e-01 5.54332137e-01 7.34345078e-01 3.08053225e-01 -4.67413396e-01 -4.76873428e-01 -4.10331368e-01 3.24869007e-01 3.86539578e-01 7.83656240e-01 4.89228755e-01 -1.26818204e+00 -1.10224032e+00 -2.76764333e-01 4.29885954e-01 2.43544936e-01 -5.77792637e-02 4.71141249e-01 -4.03679132e-01 4.88782734e-01 -1.39836483e-02 6.63350523e-02 -1.12843025e+00 5.66916466e-01 -2.27260739e-01 -6.97576046e-01 -7.16307223e-01 8.39682043e-01 5.06723784e-02 -2.96107352e-01 1.06670901e-01 -4.47375625e-01 -4.46791142e-01 5.88781714e-01 8.24581563e-01 2.56703496e-02 -2.56146509e-02 -4.32162851e-01 -7.08220750e-02 3.30362678e-01 -7.53770053e-01 -1.31688327e-01 1.37336314e+00 4.10714969e-02 -1.40705854e-01 3.98035467e-01 9.11807537e-01 2.97923863e-01 -9.35558379e-01 -6.23532534e-01 1.63585469e-01 -2.10306615e-01 1.14888586e-01 -8.60995233e-01 -7.10487604e-01 4.21395004e-01 -2.02571139e-01 2.92112768e-01 1.03151405e+00 1.99792862e-01 9.72120583e-01 6.43739283e-01 -8.12097415e-02 -1.32187295e+00 2.92995811e-01 7.12925255e-01 1.17975271e+00 -1.40824461e+00 4.99969214e-01 -4.15327758e-01 -9.20709908e-01 9.84728813e-01 6.57615721e-01 1.67625491e-02 5.51142693e-01 2.51348525e-01 4.03620563e-02 -1.76782519e-01 -8.63779068e-01 -3.21121365e-01 6.86905861e-01 1.50648907e-01 5.56318939e-01 -1.47259966e-01 -4.04707551e-01 7.53022492e-01 -5.56158781e-01 -1.02980345e-01 8.66993129e-01 1.03908348e+00 -3.54063094e-01 -1.11105514e+00 -1.67606935e-01 6.88134074e-01 -1.00180829e+00 -5.71218193e-01 -7.77719676e-01 7.71071136e-01 -2.52939194e-01 6.54841542e-01 -1.49621278e-01 -2.23037705e-01 3.39130580e-01 1.15500942e-01 2.24510074e-01 -1.05577838e+00 -4.13124770e-01 1.91821814e-01 3.58612210e-01 -1.83045730e-01 -5.48013568e-01 -4.46965188e-01 -1.05668950e+00 -2.79878974e-01 -1.45459607e-01 4.40047741e-01 3.43975544e-01 7.21682906e-01 5.50159097e-01 5.80173135e-01 4.61820453e-01 -8.12263727e-01 -2.22432062e-01 -1.27755904e+00 -2.97417164e-01 5.15353739e-01 3.77040744e-01 -3.40367943e-01 -5.23535982e-02 4.34678733e-01]
[12.298221588134766, 8.896663665771484]
4e52da4d-7e1f-4d95-82e9-da1250566f99
modeling-and-utilizing-user-s-internal-state
2012.03118
null
https://arxiv.org/abs/2012.03118v1
https://arxiv.org/pdf/2012.03118v1.pdf
Modeling and Utilizing User's Internal State in Movie Recommendation Dialogue
Intelligent dialogue systems are expected as a new interface between humans and machines. Such an intelligent dialogue system should estimate the user's internal state (UIS) in dialogues and change its response appropriately according to the estimation result. In this paper, we model the UIS in dialogues, taking movie recommendation dialogues as examples, and construct a dialogue system that changes its response based on the UIS. Based on the dialogue data analysis, we model the UIS as three elements: knowledge, interest, and engagement. We train the UIS estimators on a dialogue corpus with the modeled UIS's annotations. The estimators achieved high estimation accuracy. We also design response change rules that change the system's responses according to each UIS. We confirmed that response changes using the result of the UIS estimators improved the system utterances' naturalness in both dialogue-wise evaluation and utterance-wise evaluation.
['Sadao Kurohashi', 'Ribeka Tanaka', 'Takashi Kodama']
2020-12-05
null
null
null
null
['movie-recommendation']
['miscellaneous']
[-1.02137730e-01 8.50623310e-01 -3.07002086e-02 -9.98218000e-01 -2.03336060e-01 -7.16417074e-01 9.07443345e-01 -1.21817462e-01 -4.31453615e-01 7.93869257e-01 7.37361729e-01 -4.62570302e-02 2.80566067e-01 -7.24448442e-01 8.13881904e-02 1.47082031e-01 4.38176751e-01 6.54140353e-01 4.74109501e-01 -6.75838649e-01 6.80869401e-01 -8.81201923e-02 -1.28929377e+00 7.16734171e-01 1.00457585e+00 8.74826312e-01 1.53993860e-01 1.21444213e+00 -5.76512873e-01 1.24229860e+00 -9.82571483e-01 -2.25283593e-01 -2.40189865e-01 -7.78514922e-01 -1.50062919e+00 1.89529985e-01 -6.01189852e-01 -7.02766478e-01 -9.17906836e-02 8.01024079e-01 3.77643019e-01 5.95467389e-01 7.64682472e-01 -1.15672624e+00 -4.13783759e-01 7.34016240e-01 2.45497420e-01 -3.14209491e-01 1.14672542e+00 1.82236150e-01 6.41382992e-01 -5.39524198e-01 6.03678107e-01 1.61107528e+00 4.47291404e-01 1.09268904e+00 -7.79340088e-01 -7.70288706e-02 2.79291332e-01 1.42857328e-01 -7.38268673e-01 -6.40045762e-01 6.01962328e-01 -6.18793845e-01 9.30416226e-01 7.73362935e-01 4.40751880e-01 7.26393580e-01 -1.89499278e-02 9.36817765e-01 7.73502052e-01 -6.66373014e-01 2.70399958e-01 9.25961673e-01 6.27341032e-01 5.75189054e-01 -8.02706122e-01 -5.32252967e-01 -3.25064600e-01 -2.38134816e-01 6.46833718e-01 -5.02322972e-01 -2.39236221e-01 3.59796256e-01 -1.07167423e+00 8.75998437e-01 -2.28321865e-01 3.98704499e-01 -3.31469357e-01 -6.29892170e-01 6.53643191e-01 6.19427323e-01 5.68847179e-01 9.12880838e-01 -9.03618753e-01 -7.04927206e-01 1.20336711e-01 1.87572949e-02 1.72762382e+00 1.00226998e+00 4.84014422e-01 -2.43718639e-01 -8.42528105e-01 1.38271093e+00 4.68959987e-01 2.19966620e-01 6.58284903e-01 -1.18185163e+00 1.17731400e-01 1.05292642e+00 7.50191212e-01 -9.80485499e-01 -5.38359404e-01 5.64152956e-01 -6.01055980e-01 -3.30420405e-01 4.09039050e-01 -7.27559268e-01 -1.97680086e-01 1.63299978e+00 6.25201344e-01 -5.93353331e-01 2.57863522e-01 7.17761457e-01 1.15561616e+00 7.99683630e-01 -4.79320250e-02 -7.28084862e-01 1.47024786e+00 -9.47616756e-01 -1.46518362e+00 2.27002010e-01 9.56110775e-01 -6.11895680e-01 1.34920835e+00 3.82085592e-01 -9.50657845e-01 -8.02738130e-01 -6.54843330e-01 3.24681252e-01 -2.36714426e-02 3.44187468e-01 3.11389655e-01 5.56227744e-01 -9.33965564e-01 3.73432308e-01 -2.89641589e-01 -3.97745311e-01 -7.86655962e-01 2.74247020e-01 9.11089778e-02 8.68185937e-01 -1.96928561e+00 1.09946692e+00 1.41418964e-01 1.23070866e-01 -2.95744091e-01 7.11018071e-02 -8.89637589e-01 -7.02357143e-02 2.61880308e-01 -4.43863004e-01 2.13878727e+00 -1.01063168e+00 -2.61573243e+00 3.25932860e-01 -1.37330517e-02 -8.84303004e-02 5.81531644e-01 5.59970140e-02 -4.49671805e-01 -2.75630653e-01 -1.92069590e-01 4.09484446e-01 3.69556934e-01 -9.23456490e-01 -9.34616685e-01 -8.84763300e-02 4.04578626e-01 7.57075548e-01 -3.69923502e-01 1.11645326e-01 -2.89934784e-01 3.26693505e-01 -1.29665673e-01 -6.93878591e-01 -1.94961280e-01 -5.26282251e-01 -4.60207164e-01 -7.78978884e-01 4.08380687e-01 -4.95344728e-01 1.62343359e+00 -1.61386061e+00 -6.26706481e-02 -7.20297471e-02 1.56767532e-01 3.27781379e-01 9.64700580e-02 5.12093425e-01 6.01479411e-01 1.36900619e-01 2.55490035e-01 -1.13761321e-01 3.79797935e-01 -3.67732532e-02 1.39380366e-01 -2.56948441e-01 -1.90521076e-01 5.06265938e-01 -9.56490099e-01 -4.78855938e-01 2.85080016e-01 -1.99797943e-01 -6.63936555e-01 1.44664812e+00 -5.12371957e-01 7.53920913e-01 -6.61330998e-01 -1.10106923e-01 1.16522811e-01 -7.16182590e-02 3.64714503e-01 -3.08893412e-01 -3.02498192e-01 5.68005562e-01 -9.26952362e-01 1.14694643e+00 -6.59613550e-01 3.67781103e-01 5.13644936e-03 -4.30377305e-01 1.32558155e+00 6.73247695e-01 6.43110797e-02 -4.82269168e-01 1.25757754e-01 -1.56297609e-01 2.95979887e-01 -1.15659118e+00 6.91263914e-01 2.35567942e-01 -5.01342475e-01 8.01883221e-01 6.31216615e-02 -3.96207452e-01 5.42968418e-03 3.69858742e-01 8.24655652e-01 -1.20771274e-01 7.71296620e-01 -4.58019748e-02 1.05516827e+00 -1.38606742e-01 2.19174951e-01 1.03265834e+00 -3.03252310e-01 -2.66534872e-02 7.72375286e-01 -4.34861928e-01 -5.37225187e-01 -2.93762445e-01 4.35123816e-02 1.51975191e+00 3.21979038e-02 -3.75577718e-01 -1.24584770e+00 -1.00400770e+00 -4.82988626e-01 1.09895229e+00 -4.55846220e-01 -3.58978301e-01 -2.82376021e-01 -2.09936291e-01 4.49751556e-01 3.28618884e-02 8.49882841e-01 -1.36621261e+00 -3.74612689e-01 4.70080107e-01 -5.68359733e-01 -7.04041004e-01 -7.61639535e-01 -2.50775695e-01 -2.34783754e-01 -8.82978916e-01 -2.31392831e-01 -5.21511197e-01 6.02993429e-01 -2.15344086e-01 9.58630860e-01 2.91766733e-01 5.66988945e-01 6.89401031e-01 -7.25216568e-01 -1.41651884e-01 -1.32527864e+00 1.01840042e-01 2.35315785e-01 -2.43576597e-02 3.07319820e-01 1.78180844e-01 -4.90650207e-01 8.16625953e-01 -6.48050010e-01 5.31896651e-01 -1.48095831e-01 9.03717518e-01 -4.55544144e-01 -3.07508260e-01 9.88692880e-01 -1.40138173e+00 1.71823657e+00 -3.92988205e-01 -1.64086014e-01 6.45885289e-01 -8.17797542e-01 1.99567288e-01 7.36992002e-01 -4.58780676e-01 -1.91228175e+00 -1.71285406e-01 -4.16409522e-01 7.11087167e-01 -5.17450035e-01 5.12534678e-01 -2.68626392e-01 3.39678794e-01 7.67470717e-01 -1.30659183e-02 1.43001154e-01 -2.39093766e-01 4.28302526e-01 1.71190751e+00 1.86776519e-01 -6.92348063e-01 -5.51501587e-02 -5.20439208e-01 -1.06658304e+00 -7.13110447e-01 -7.61507273e-01 -4.19402033e-01 -7.16208398e-01 -9.94286895e-01 6.50837004e-01 -2.83960104e-01 -1.52409625e+00 6.59327805e-01 -1.60233867e+00 -4.93700415e-01 1.19814053e-01 3.31004411e-01 -6.22959733e-01 4.18842256e-01 -8.89342785e-01 -1.56220233e+00 -5.23446500e-01 -1.06880510e+00 6.02519691e-01 5.33977747e-01 -1.02264249e+00 -1.21571147e+00 2.52903491e-01 3.21035028e-01 4.59110767e-01 -3.55263501e-01 7.69840360e-01 -1.25256395e+00 2.60807842e-01 -2.36313194e-01 6.00911640e-02 2.49638334e-01 4.05677468e-01 1.15459278e-01 -9.20893192e-01 3.02880615e-01 4.41450953e-01 -4.26323116e-01 -1.07833005e-01 -1.43626640e-02 8.10371578e-01 -1.07214868e+00 8.15192387e-02 -3.23656052e-01 3.71117294e-01 8.95441949e-01 5.51995516e-01 2.44037434e-02 1.59107536e-01 9.66378987e-01 1.20237124e+00 8.31634760e-01 6.71960354e-01 8.66687119e-01 -3.54329757e-02 1.23042852e-01 3.88931781e-01 -3.44500750e-01 5.83804965e-01 1.23786557e+00 7.98856989e-02 -5.01259625e-01 -6.27547085e-01 -2.52656881e-02 -2.26473689e+00 -7.56327331e-01 -2.45429620e-01 1.99206471e+00 1.50543177e+00 1.89050272e-01 6.82147220e-02 -3.94908756e-01 8.85750353e-01 5.82791381e-02 -5.08974016e-01 -1.06230223e+00 7.21319079e-01 -5.43823898e-01 -2.44672954e-01 1.11704195e+00 -5.42569160e-01 1.12369406e+00 5.89530516e+00 1.77714527e-01 -6.65005028e-01 -4.50234041e-02 6.57014489e-01 5.46873689e-01 -4.14532810e-01 -1.06074765e-01 -7.60174036e-01 5.56581914e-01 1.00587487e+00 -4.46823448e-01 4.29763556e-01 6.06683969e-01 5.28991878e-01 -3.66436601e-01 -1.23988211e+00 4.53987718e-01 -5.93478307e-02 -9.60278034e-01 -2.35372745e-02 -4.67255890e-01 1.95063606e-01 -8.71204138e-01 -6.19858146e-01 8.69430184e-01 6.09386265e-01 -5.65679610e-01 2.14767575e-01 1.12674379e+00 6.58031285e-01 -2.94900477e-01 9.86931324e-01 1.07248294e+00 -6.99872434e-01 2.55113900e-01 -1.63403913e-01 -4.87528682e-01 1.02821931e-01 -1.40696168e-01 -1.79219067e+00 6.74662832e-03 -3.26665118e-02 1.63981989e-01 -2.06767008e-01 2.52833396e-01 -2.50859141e-01 6.42728806e-01 4.33660485e-03 -8.19968641e-01 -2.29976639e-01 -4.36228365e-01 3.58910322e-01 1.31712031e+00 -2.22777233e-01 7.37781644e-01 3.54864240e-01 6.87727451e-01 4.33137864e-02 4.28346992e-01 -3.15314114e-01 1.35167286e-01 9.62474406e-01 1.28296244e+00 -1.57136992e-01 -7.41951704e-01 -2.85932869e-02 9.53387916e-01 1.91894531e-01 1.70743376e-01 -6.92801118e-01 -4.81008112e-01 3.35033923e-01 -1.82954803e-01 -8.24009299e-01 3.95747274e-01 -1.03907198e-01 -9.61846948e-01 -4.88622487e-01 -9.30540979e-01 1.59725308e-01 -8.34425271e-01 -9.62140501e-01 8.93075049e-01 -8.67752433e-02 -1.02777195e+00 -8.60765278e-01 -1.98241100e-01 -5.84720552e-01 7.61541665e-01 -5.02505183e-01 -4.48456407e-01 -3.48416895e-01 1.70683295e-01 1.04198337e+00 -1.25268728e-01 1.41101027e+00 -2.13350862e-01 -5.93413234e-01 5.90201437e-01 -4.62350160e-01 2.04545408e-01 9.75778639e-01 -1.35165083e+00 2.58790135e-01 1.56390835e-02 -5.02143323e-01 6.17875636e-01 9.34296012e-01 -6.31484926e-01 -9.88781214e-01 -4.49562699e-01 9.71811116e-01 -4.30920541e-01 5.50389051e-01 -1.32420674e-01 -1.11578751e+00 4.97372329e-01 5.49378514e-01 -8.45432162e-01 7.85371125e-01 3.55806202e-01 2.19305396e-01 3.05107683e-01 -1.34576082e+00 6.10296190e-01 7.21183300e-01 -4.90012676e-01 -9.61783826e-01 4.69282299e-01 8.54690313e-01 -8.16487074e-01 -1.18715882e+00 3.81710112e-01 6.90570831e-01 -8.41239750e-01 4.24562305e-01 -9.62348402e-01 1.62292317e-01 -5.13877459e-02 1.68832377e-01 -1.56775868e+00 -2.94680506e-01 -6.58998191e-01 -3.66344079e-02 1.36593306e+00 5.94295979e-01 -6.36008203e-01 3.07279527e-01 1.78778660e+00 -5.13937557e-03 -4.55336869e-01 -3.53874534e-01 3.09514031e-02 -6.85530424e-01 -9.11357030e-02 7.40187943e-01 9.72522855e-01 1.06663334e+00 1.02124393e+00 -6.81637943e-01 -1.41162261e-01 -2.69718766e-01 -2.33499572e-01 1.25037110e+00 -1.20097923e+00 -3.03330421e-01 -1.26684606e-01 2.49902442e-01 -1.58383632e+00 2.36343384e-01 -2.27776870e-01 5.24603069e-01 -1.53351700e+00 2.91847587e-02 -1.25569969e-01 3.17219287e-01 3.19493860e-01 -4.59038317e-01 -7.93545604e-01 -5.87164201e-02 2.24292144e-01 -1.00431204e+00 6.57339633e-01 1.32118845e+00 2.46194024e-02 -8.25214148e-01 6.52352512e-01 -4.67126995e-01 8.93779993e-01 7.93878436e-01 -6.13372624e-02 -4.84604836e-01 7.35279173e-02 9.70388129e-02 8.73734951e-01 -5.49370170e-01 -3.54654521e-01 5.45274317e-01 -5.41639209e-01 1.40806526e-01 -2.23102883e-01 2.93085217e-01 -5.67789018e-01 -4.50751297e-02 3.45576406e-01 -1.36097145e+00 -3.32095236e-01 -1.61398113e-01 2.37449706e-01 -1.75503388e-01 -5.93667507e-01 5.14232457e-01 -2.32235745e-01 -6.19676292e-01 -1.07687749e-01 -1.01818192e+00 -7.01424554e-02 7.40576804e-01 -2.37443939e-01 -3.10232490e-01 -1.17325985e+00 -8.75550807e-01 5.72393119e-01 8.11490938e-02 4.77375388e-01 5.13593137e-01 -9.74571168e-01 -5.23880243e-01 2.00017676e-01 1.09731212e-01 -2.89645791e-01 1.49061918e-01 4.13335502e-01 -1.49658516e-01 2.37912416e-01 1.08045943e-01 -4.61314350e-01 -1.45305097e+00 -2.27686897e-01 6.11136675e-01 -2.99771905e-01 1.40634492e-01 7.36827254e-01 1.63972735e-01 -1.32042527e+00 4.60440546e-01 -2.57306490e-02 -1.09716630e+00 1.64756820e-01 6.56173468e-01 2.28033245e-01 -2.26273000e-01 -3.34864438e-01 1.21500544e-01 -3.13960105e-01 -2.40694478e-01 -4.48532701e-01 8.95531058e-01 -6.92263901e-01 -1.27024353e-01 9.42680538e-01 7.82246351e-01 -1.47404537e-01 -1.05805516e+00 -3.59711319e-01 2.98952788e-01 -5.77706337e-01 -4.30569917e-01 -1.24252832e+00 -2.39282072e-01 2.93795943e-01 2.36972168e-01 1.04291153e+00 6.39213204e-01 -2.05677181e-01 6.56378031e-01 7.65519798e-01 5.47009289e-01 -1.69033325e+00 2.09532186e-01 1.07729781e+00 1.09691012e+00 -1.25172877e+00 -3.12640369e-01 -1.75933376e-01 -1.33829772e+00 1.34712040e+00 1.36796772e+00 5.52033246e-01 5.83777249e-01 1.94714908e-02 4.50064391e-01 -7.97029436e-02 -1.38362610e+00 1.42145202e-01 2.71639287e-01 3.41449797e-01 9.30678070e-01 2.98078537e-01 -6.63781285e-01 9.91262317e-01 -2.16305822e-01 -3.06677576e-02 7.68051386e-01 3.84664148e-01 -9.04582143e-01 -9.87615228e-01 -6.73746616e-02 5.80640733e-01 2.54023403e-01 2.52693266e-01 -1.04529786e+00 3.61981332e-01 -4.17197347e-01 1.58404410e+00 -1.92005008e-01 -8.39530230e-01 8.67533267e-01 2.71047205e-01 -4.16101329e-02 -1.02141690e+00 -1.13882875e+00 -3.36361617e-01 8.32055926e-01 -2.68826544e-01 -2.43839145e-01 -3.09458613e-01 -1.50604212e+00 -2.46988699e-01 -8.62483263e-01 7.98563540e-01 6.36140764e-01 1.11654365e+00 1.64766610e-01 4.07616079e-01 1.51228142e+00 -2.86526114e-01 -1.15272164e+00 -1.68759239e+00 -3.36733460e-01 4.85611469e-01 -1.82691112e-01 -3.73347253e-01 -3.84116501e-01 -1.60456821e-01]
[12.916879653930664, 8.004467010498047]
05aa360e-6503-40bf-8034-ed3129aff948
is-swarm-intelligence-able-to-create-mazes
1601.06580
null
http://arxiv.org/abs/1601.06580v1
http://arxiv.org/pdf/1601.06580v1.pdf
Is swarm intelligence able to create mazes?
In this paper, the idea of applying Computational Intelligence in the process of creation board games, in particular mazes, is presented. For two different algorithms the proposed idea has been examined. The results of the experiments are shown and discussed to present advantages and disadvantages.
['Dawid Polap', 'Christian Napoli', 'Marcin Wozniak', 'Emiliano Tramontana']
2016-01-25
null
null
null
null
['board-games']
['playing-games']
[-4.45582235e-04 4.21128422e-01 4.61226493e-01 -1.07697472e-01 5.27578652e-01 -4.73711610e-01 4.44786429e-01 -1.55943647e-01 -6.66458666e-01 1.35576570e+00 -1.71910256e-01 -5.78094006e-01 -9.32464838e-01 -1.22309721e+00 1.11684024e-01 -3.89788508e-01 -4.27101254e-01 6.99185431e-01 3.81640315e-01 -9.77423429e-01 9.08734441e-01 4.70264316e-01 -1.66064882e+00 8.53618532e-02 8.33831251e-01 4.21564639e-01 5.18186629e-01 8.97804141e-01 -3.36830825e-01 1.03107786e+00 -9.50301409e-01 -3.04454863e-01 4.62789387e-01 -5.08347929e-01 -9.27534342e-01 -5.31341247e-02 -8.51235151e-01 -1.22687809e-01 -1.08377881e-01 1.05252016e+00 2.76765674e-01 2.94085413e-01 9.36728418e-01 -1.14599538e+00 1.15661249e-01 5.24589062e-01 -1.88373789e-01 2.95514524e-01 8.77693176e-01 -1.95555226e-03 5.48931241e-01 -1.13810837e-01 5.73439240e-01 9.49351192e-01 4.72882241e-01 3.71005088e-01 -8.75674725e-01 -4.83832747e-01 -2.56457746e-01 4.24003780e-01 -1.41222239e+00 3.13584894e-01 8.39152455e-01 -2.47190401e-01 1.04472363e+00 4.54948813e-01 1.18857908e+00 9.23215449e-02 5.70705712e-01 5.24114609e-01 1.71194828e+00 -7.51654327e-01 6.04464710e-01 2.13054493e-01 4.73224789e-01 5.54851115e-01 9.19954538e-01 2.71230072e-01 1.23771615e-01 -1.12820424e-01 7.25822747e-01 -5.15386105e-01 3.07377547e-01 -3.14049959e-01 -3.62234861e-01 1.02670991e+00 2.51951605e-01 7.96599090e-01 -3.71859491e-01 -2.00149447e-01 2.33278409e-01 2.88031220e-01 -4.35768850e-02 1.02047253e+00 -4.12018836e-01 -5.09246945e-01 -2.14823514e-01 7.07053304e-01 1.14166689e+00 5.47458947e-01 2.42306411e-01 4.94729616e-02 4.47582066e-01 1.99388787e-01 4.59392548e-01 -1.29214421e-01 4.67565030e-01 -6.43944442e-01 3.37086797e-01 8.47887516e-01 4.53643531e-01 -1.31094742e+00 -1.04664290e+00 1.00033633e-01 -3.35033596e-01 1.15758312e+00 3.87037039e-01 -4.96382982e-01 -7.69211650e-01 8.65959883e-01 1.16368599e-01 -1.85453758e-01 6.96205914e-01 3.08694214e-01 1.20360386e+00 6.83439493e-01 1.31843230e-02 4.73588407e-02 1.36653197e+00 -1.08321702e+00 -9.78842437e-01 1.83064818e-01 5.00083983e-01 -5.82032800e-01 4.57032382e-01 9.43453074e-01 -1.28593004e+00 -4.88049120e-01 -1.19096839e+00 5.69818735e-01 -8.95012915e-01 -2.50041515e-01 1.30225778e+00 1.42602849e+00 -9.47497427e-01 7.49539495e-01 -4.12245423e-01 -4.63032395e-01 -2.28805050e-01 5.90757072e-01 -1.40513688e-01 4.33977753e-01 -1.13499618e+00 1.16719794e+00 1.12350476e+00 2.20619068e-01 -2.68055946e-01 2.68122226e-01 -5.21584570e-01 1.61655694e-01 2.26965651e-01 -3.93796176e-01 9.75491583e-01 -6.43791556e-01 -1.81077206e+00 5.81536353e-01 3.22754353e-01 -2.91122198e-01 7.52247870e-01 3.39482337e-01 -3.88291597e-01 -2.22003814e-02 -8.58356655e-02 9.84591767e-02 5.92643023e-03 -1.15968728e+00 -9.28963840e-01 -2.63836056e-01 6.99556112e-01 2.72874594e-01 3.73314500e-01 1.11983508e-01 -4.32296023e-02 -3.93191457e-01 3.95212293e-01 -5.24621964e-01 -9.89695966e-01 -9.50666487e-01 2.59528697e-01 -1.25894010e-01 2.03157231e-01 -6.34888053e-01 1.24665248e+00 -1.65177691e+00 -9.07693990e-03 8.36503088e-01 -1.51339816e-02 1.56453356e-01 3.96687180e-01 7.92855859e-01 -3.11618950e-02 5.69502115e-02 1.33982107e-01 5.15047610e-01 7.33027875e-04 4.69912767e-01 3.90112698e-01 -2.11339921e-01 -1.81112796e-01 3.31236660e-01 -7.74790585e-01 -1.31336913e-01 3.82357150e-01 -7.20398501e-02 -3.61070037e-01 -1.20220222e-02 1.18387967e-01 1.92120492e-01 -7.20184088e-01 5.55174947e-01 7.56407320e-01 4.26098228e-01 4.03154522e-01 5.49823046e-01 -5.65376520e-01 2.69267231e-01 -1.65506136e+00 1.27931654e+00 -1.02213986e-01 4.46159571e-01 -3.23810995e-01 -9.97130454e-01 1.14206839e+00 4.79904741e-01 2.41360545e-01 -8.84338498e-01 5.93251824e-01 4.02072340e-01 6.66345358e-01 -6.95355475e-01 7.21493959e-01 3.25628789e-03 -2.59753633e-02 4.03781325e-01 -9.33315456e-02 -4.21932399e-01 8.91397893e-01 -3.08895439e-01 1.18423927e+00 3.42966586e-01 9.59723711e-01 -3.94805670e-01 9.20807302e-01 7.53403962e-01 1.22214936e-01 8.20679128e-01 -2.92468041e-01 6.62657246e-02 7.44179547e-01 -7.66530752e-01 -1.04041016e+00 -7.69560575e-01 3.98773737e-02 5.26380777e-01 3.97943914e-01 -4.63996530e-01 -1.11813653e+00 -3.27959418e-01 -4.16167289e-01 7.50612736e-01 -5.58207333e-01 4.16362077e-01 -4.07394409e-01 -1.05177426e+00 3.82163316e-01 6.38713017e-02 7.59310782e-01 -1.38759470e+00 -1.27073729e+00 6.48275018e-01 1.78011477e-01 -4.38448370e-01 1.18329322e+00 3.15425277e-01 -1.10710537e+00 -1.09429133e+00 -4.53686386e-01 -7.91073322e-01 6.13386273e-01 3.84033978e-01 8.25862110e-01 4.23910141e-01 -4.81449723e-01 5.35748191e-02 -7.47389734e-01 -1.02488494e+00 -1.28791869e-01 -4.41959091e-02 -4.66583014e-01 -8.52838993e-01 4.50238556e-01 -7.97861457e-01 -2.41797417e-01 1.62371561e-01 -6.43180847e-01 1.35951519e-01 4.54983354e-01 6.91333890e-01 -2.13068396e-01 1.07504880e+00 2.40816817e-01 -1.08403087e+00 1.05832458e+00 -2.29930699e-01 -8.51565599e-01 -7.94108026e-03 -4.82138813e-01 -1.26776576e-01 3.07992518e-01 1.28812522e-01 -1.04965580e+00 -3.89866441e-01 -4.19180423e-01 1.05799592e+00 -5.47514021e-01 6.00392580e-01 -2.92888343e-01 -5.72314143e-01 7.04625547e-01 -7.06216693e-02 -3.01165320e-02 -3.55936676e-01 -2.08443955e-01 3.98318052e-01 -9.44489390e-02 -3.94786477e-01 4.00648504e-01 9.74519104e-02 1.79666981e-01 -6.15147352e-01 3.37347150e-01 -1.98514104e-01 -2.99904257e-01 -6.71652973e-01 7.08136976e-01 -1.56723022e-01 -7.09295571e-01 6.48180485e-01 -1.01483154e+00 2.69738771e-02 -3.27688009e-01 3.78787667e-01 -9.72753286e-01 -7.57495537e-02 -3.43248367e-01 -1.26123762e+00 1.71932252e-03 -1.06592774e+00 -2.62130722e-02 8.68011355e-01 -1.51594654e-01 -1.01022696e+00 2.66773194e-01 3.04411530e-01 1.42796546e-01 3.41835409e-01 6.69711411e-01 -6.73755348e-01 -2.94087470e-01 -3.48463863e-01 7.18974695e-02 -3.01546484e-01 6.29890412e-02 -6.46369830e-02 -6.07077181e-01 1.85730249e-01 1.63566098e-01 1.50802329e-01 6.11471906e-02 2.94428140e-01 4.98608172e-01 1.25652567e-01 -3.97955924e-01 2.66004711e-01 1.84812987e+00 1.64475536e+00 1.39591122e+00 1.38765705e+00 -2.55128145e-01 6.60411596e-01 1.21969676e+00 5.74481905e-01 1.51476795e-02 3.73381674e-01 3.74071449e-01 6.94017112e-02 3.72626275e-01 3.30604166e-01 -1.87703207e-01 4.80746716e-01 -9.47354078e-01 -3.12889040e-01 -1.12991428e+00 3.78157385e-02 -1.77838409e+00 -1.01044416e+00 -2.91993022e-01 1.47297442e+00 -1.38353510e-02 6.69248223e-01 3.69299501e-01 8.55133891e-01 4.12395895e-01 -1.57960728e-01 3.76129985e-01 -1.13236821e+00 1.37742415e-01 7.36591756e-01 4.16696966e-01 7.58321702e-01 -9.99885023e-01 8.10906649e-01 7.36224794e+00 4.93307650e-01 -6.86121464e-01 6.82022572e-02 1.81317031e-01 5.44595361e-01 2.43793830e-01 1.67006075e-01 -6.99668750e-02 3.75937968e-01 4.30344701e-01 -3.30402792e-01 6.97386682e-01 6.76834226e-01 1.01396684e-02 -8.88010085e-01 8.74081552e-02 6.84487164e-01 -1.42641485e-01 -9.93290961e-01 -2.34772880e-02 2.37114169e-02 5.81774354e-01 -6.92963779e-01 -1.77338675e-01 1.88206404e-01 5.56806862e-01 -1.03279901e+00 7.06546843e-01 4.52280045e-01 -3.48494291e-01 -1.46910965e+00 1.22425008e+00 2.39313319e-01 -1.06925285e+00 -2.68512338e-01 -6.12498999e-01 -1.48918915e+00 -1.40014544e-01 -4.03212726e-01 -8.96836340e-01 1.29024422e+00 6.94872022e-01 -1.42474547e-01 -4.84661847e-01 1.75096285e+00 -4.49382573e-01 2.64834493e-01 -2.79768705e-01 -9.42775726e-01 6.69773459e-01 -7.87683129e-01 3.74197751e-01 7.57254779e-01 3.37447137e-01 8.09018433e-01 -2.21489727e-01 6.45955801e-01 1.18892825e+00 5.68275750e-01 -9.86590922e-01 -1.56482439e-02 1.28156871e-01 8.37733209e-01 -1.50392699e+00 -3.35662991e-01 -2.26360768e-01 3.96658570e-01 -1.80491909e-01 9.25316960e-02 -8.92814636e-01 -1.10498559e+00 3.54881287e-01 -2.61856228e-01 1.93970427e-02 -2.44610712e-01 -8.75279486e-01 -5.26514053e-01 -3.53163958e-01 -6.64201438e-01 3.27705503e-01 -8.94446552e-01 -4.99241948e-01 7.00642347e-01 4.23761010e-01 -9.53114927e-01 -1.58821672e-01 -1.28601551e+00 -9.12705481e-01 1.07081759e+00 -7.08008647e-01 -5.84688544e-01 -3.03395897e-01 1.96411267e-01 4.08145130e-01 -7.52054274e-01 7.87047684e-01 -5.36346287e-02 -2.29298100e-01 5.25682531e-02 4.17384982e-01 -9.62610096e-02 -3.31237257e-01 -1.22347212e+00 4.65171814e-01 5.17830968e-01 -1.92041025e-01 3.01480293e-01 1.17093003e+00 -7.71225810e-01 -6.82727814e-01 1.93693079e-02 6.27708554e-01 5.00765331e-02 5.16695321e-01 1.40017882e-01 -2.20040917e-01 3.65773171e-01 3.89230996e-01 -1.13527608e+00 6.20415747e-01 4.28618528e-02 5.55627942e-01 3.47547770e-01 -1.49834013e+00 1.01711500e+00 6.22463644e-01 4.16939914e-01 -9.92053092e-01 -2.82048911e-01 8.58345553e-02 -5.16986549e-01 -2.00243294e-01 2.55093217e-01 5.71469367e-01 -1.40842354e+00 9.78505313e-01 -7.72983670e-01 -5.11526279e-02 -3.04820597e-01 1.87770814e-01 -1.24367702e+00 -5.37897766e-01 -4.66654599e-01 6.37897432e-01 5.70500910e-01 4.75999415e-01 -9.60088193e-01 1.28273857e+00 4.83582258e-01 2.39502892e-01 -2.61110038e-01 -8.84704590e-01 -6.02172315e-01 -1.89304650e-01 -5.36428452e-01 8.11635315e-01 6.12031996e-01 6.70377195e-01 1.63078472e-01 -9.99135822e-02 1.07621819e-01 3.77692968e-01 -4.76404428e-01 8.41880918e-01 -1.49631655e+00 -1.83067337e-01 -6.37275636e-01 -1.27444732e+00 -1.41537517e-01 -5.05115926e-01 -1.81225404e-01 -2.35176280e-01 -1.82773352e+00 -2.97043800e-01 -4.83615369e-01 -3.48612726e-01 -1.14405170e-01 -8.27069655e-02 1.13017093e-02 5.43935895e-01 -3.01803768e-01 -1.66289449e-01 -9.53908339e-02 1.11496198e+00 -5.21288700e-02 -4.73036945e-01 1.31189793e-01 -5.88249862e-01 1.13224900e+00 1.50234747e+00 -4.18520540e-01 -6.76691115e-01 1.91849023e-01 4.62873012e-01 1.17401011e-01 -2.61744201e-01 -1.68329453e+00 -4.57736813e-02 -2.87130892e-01 3.30450565e-01 -5.43822825e-01 4.07848775e-01 -1.22903228e+00 6.36945009e-01 1.27059734e+00 1.30401641e-01 6.59086227e-01 3.08414817e-01 2.27303818e-01 -2.13134378e-01 -1.23982024e+00 3.53090852e-01 -5.49974859e-01 -1.15612888e+00 -6.98858202e-01 -1.11008239e+00 -4.74959761e-01 1.62089145e+00 -9.11825955e-01 1.95492461e-01 -4.32699531e-01 -1.07286894e+00 3.64482515e-02 3.68321300e-01 -2.61216789e-01 6.44123852e-01 -9.94891286e-01 -1.60869017e-01 5.20261563e-03 -1.52614906e-01 -6.28747165e-01 4.03393358e-01 3.22826654e-01 -1.76083267e+00 4.81748641e-01 -1.37816846e+00 2.80603081e-01 -1.51976752e+00 6.86308205e-01 2.43025422e-01 -6.15616679e-01 -4.85666782e-01 5.13198078e-01 -9.72639471e-02 -3.75037879e-01 7.75144920e-02 -1.86714768e-01 -1.38212109e+00 -3.56093258e-01 7.05676138e-01 6.16935849e-01 -1.05809480e-01 -1.79498181e-01 1.75786037e-02 4.06442612e-01 2.81280696e-01 -6.38979673e-01 1.19653463e+00 3.28596048e-02 -4.64610845e-01 2.54316866e-01 -1.22119330e-01 -1.88099489e-01 -5.10283351e-01 5.33265769e-01 4.57547307e-01 -7.40121603e-01 -3.65510017e-01 -1.11428463e+00 -4.76138383e-01 5.10901749e-01 6.54521883e-01 7.03979194e-01 1.31731582e+00 -7.61145115e-01 2.18088850e-01 6.22069776e-01 1.12013566e+00 -1.27892160e+00 -4.53111589e-01 9.74090993e-01 4.59540188e-01 -6.22173011e-01 3.19451746e-03 -7.31358111e-01 -5.53008139e-01 1.56151867e+00 6.86466098e-01 -5.44709265e-01 5.48935294e-01 4.79612589e-01 6.16246462e-03 -4.62174743e-01 -4.87204432e-01 -3.18737656e-01 -2.99014211e-01 1.26315629e+00 2.59106785e-01 2.97286302e-01 -1.44864523e+00 7.57958710e-01 -5.11566877e-01 1.80975810e-01 1.04570043e+00 1.90245068e+00 -7.01317549e-01 -1.32717335e+00 -1.08457077e+00 8.55722502e-02 -3.98963988e-01 2.63327956e-01 -8.09367657e-01 1.45865548e+00 4.21454608e-01 1.04568207e+00 -1.65961742e-01 -5.52228272e-01 4.55729872e-01 -1.73110008e-01 7.33533621e-01 -2.04355434e-01 -7.52895474e-01 -4.48083609e-01 5.49595714e-01 5.43937320e-03 -3.92063111e-01 -4.11589444e-01 -1.18060315e+00 -5.17054617e-01 -1.94882706e-01 9.45518196e-01 7.45241582e-01 6.32208347e-01 -3.55647206e-01 7.55390584e-01 5.32438494e-02 -8.97604465e-01 1.25340670e-01 -8.44101429e-01 -1.00649071e+00 -3.91242057e-02 -5.04688740e-01 -1.02526999e+00 1.16669610e-01 -7.59426892e-01]
[3.4573404788970947, 1.5077226161956787]
35423515-86d1-4fd8-ba28-39bac1dc04fa
conformal-uncertainty-sets-for-robust
2105.14957
null
https://arxiv.org/abs/2105.14957v2
https://arxiv.org/pdf/2105.14957v2.pdf
Conformal Uncertainty Sets for Robust Optimization
Decision-making under uncertainty is hugely important for any decisions sensitive to perturbations in observed data. One method of incorporating uncertainty into making optimal decisions is through robust optimization, which minimizes the worst-case scenario over some uncertainty set. We connect conformal prediction regions to robust optimization, providing finite sample valid and conservative ellipsoidal uncertainty sets, aptly named conformal uncertainty sets. In pursuit of this connection we explicitly define Mahalanobis distance as a potential conformity score in full conformal prediction. We also compare the coverage and optimization performance of conformal uncertainty sets, specifically generated with Mahalanobis distance, to traditional ellipsoidal uncertainty sets on a collection of simulated robust optimization examples.
['Bruce Cox', 'Chancellor Johnstone']
2021-05-31
null
null
null
null
['decision-making-under-uncertainty', 'multi-target-regression', 'decision-making-under-uncertainty']
['medical', 'miscellaneous', 'reasoning']
[ 3.03428769e-01 7.28732944e-01 4.46261428e-02 -8.64429355e-01 -1.48190379e+00 -9.77846146e-01 5.89247167e-01 3.66284758e-01 -1.18133454e-02 1.16089797e+00 5.69416285e-01 -3.66007388e-01 -1.17847848e+00 -9.08499062e-01 -7.35658228e-01 -4.92637366e-01 -6.87962323e-02 9.89610553e-01 -1.53044313e-01 1.46394193e-01 5.15787482e-01 6.48886621e-01 -1.02698243e+00 -2.45332792e-01 1.31580520e+00 1.22615969e+00 -5.44746518e-01 2.77319163e-01 3.20168555e-01 1.11123741e-01 -4.46454972e-01 -6.31607831e-01 3.90083760e-01 3.01726937e-01 -1.38317749e-01 -1.71358064e-01 2.33739451e-01 7.45918080e-02 2.30598748e-01 1.15129483e+00 3.68036449e-01 3.49409759e-01 1.47877824e+00 -1.19633830e+00 -7.52065122e-01 9.59651828e-01 -2.64845014e-01 5.06290421e-03 4.52981442e-01 -3.45274746e-01 8.65346909e-01 -8.97137880e-01 3.21173131e-01 1.39100373e+00 8.04422200e-01 1.91395819e-01 -1.38190186e+00 -3.02802205e-01 -3.98793429e-01 -4.65005219e-01 -1.69936073e+00 -6.12881243e-01 4.31869864e-01 -6.24473393e-01 4.40810889e-01 7.46836662e-01 -1.65347070e-01 8.32403839e-01 1.33872783e+00 8.40960890e-02 9.79060650e-01 -2.45899230e-01 7.72891223e-01 2.50174671e-01 -1.55999303e-01 1.42049938e-01 6.64434493e-01 8.42643738e-01 1.34314224e-01 -6.04328156e-01 3.94331247e-01 -5.72626442e-02 -4.28075165e-01 -6.49949729e-01 -9.81281102e-01 1.00926936e+00 1.94899857e-01 -4.23494965e-01 1.02151282e-01 9.47906449e-02 -1.92095548e-01 1.67272896e-01 6.22180760e-01 9.15672243e-01 -2.49502048e-01 2.70626873e-01 -5.01801431e-01 3.28910857e-01 8.46339822e-01 1.12540674e+00 1.18846722e-01 1.05557226e-01 -5.86487472e-01 3.61378312e-01 7.94999659e-01 9.96176600e-01 -3.06282878e-01 -1.23481309e+00 5.77809513e-01 1.30264193e-01 6.62950635e-01 -8.92044961e-01 -6.05505943e-01 -3.74453127e-01 -5.06612897e-01 5.23170710e-01 3.75806361e-01 -5.44772089e-01 -6.26186490e-01 1.47483754e+00 3.66844714e-01 -3.60107511e-01 5.64455509e-01 6.44194424e-01 2.62477219e-01 6.29004121e-01 -3.84318382e-01 -5.63476801e-01 7.72549868e-01 -1.24227539e-01 -5.47196984e-01 4.23261747e-02 1.55707866e-01 -6.57309175e-01 5.20749032e-01 1.27539605e-01 -9.79355633e-01 1.38192773e-01 -1.34881783e+00 5.15556872e-01 1.56697500e-02 -7.95575380e-01 1.99334770e-01 9.34477389e-01 -6.59614682e-01 8.85055006e-01 -3.71832401e-01 2.59310514e-01 1.81473628e-01 2.55448580e-01 -3.45170081e-01 2.19958395e-01 -1.11266220e+00 1.32082534e+00 4.40695196e-01 -2.00035069e-02 -9.44155574e-01 -8.92031252e-01 -1.23796725e+00 2.73976452e-03 5.57325780e-01 -3.71584177e-01 9.74550605e-01 -1.37293795e-02 -1.14653277e+00 2.23607585e-01 2.70034254e-01 -4.94604886e-01 6.90188468e-01 8.16946849e-02 -7.21919775e-01 -3.75403553e-01 3.66007499e-02 2.21678272e-01 1.00575364e+00 -1.36391962e+00 -2.23302603e-01 -4.94978726e-01 -2.54107773e-01 4.28095341e-01 3.92500758e-01 1.45585746e-01 2.35674426e-01 -8.14197838e-01 3.72044742e-01 -1.04872036e+00 -6.88145876e-01 -3.25088561e-01 -7.26192236e-01 -3.02203707e-02 2.53869057e-01 -3.40294391e-01 1.15821099e+00 -1.59810150e+00 3.39462101e-01 1.07620311e+00 -3.23028825e-02 -7.78121054e-01 3.26205522e-01 2.73920059e-01 3.78222205e-02 3.77052426e-01 -8.29984963e-01 -8.32948182e-03 5.50744295e-01 1.66062102e-01 -6.46835446e-01 7.99574554e-01 9.26821828e-02 8.93868089e-01 -5.74889898e-01 -3.18170696e-01 1.72069430e-01 1.15171529e-01 -2.20215976e-01 -2.23127902e-01 -2.10133106e-01 4.16966736e-01 -6.34487450e-01 1.06802690e+00 6.42858982e-01 1.41375482e-01 -2.90160686e-01 2.23351046e-01 9.29235220e-02 -4.85331565e-01 -1.31452680e+00 1.17399633e+00 5.42586623e-03 2.71643579e-01 -1.86558530e-01 -4.54534024e-01 1.07364249e+00 1.79972738e-01 4.84958798e-01 -1.39908224e-01 2.71271676e-01 9.17221233e-02 -2.97791570e-01 1.86409965e-01 4.88881707e-01 -4.40594316e-01 -7.84577906e-01 6.76377892e-01 -2.00497106e-01 -9.81461942e-01 -3.49836886e-01 -9.90683958e-02 7.45321631e-01 -1.16124034e-01 3.59440386e-01 -1.05054641e+00 -1.50320372e-02 -1.62870497e-01 7.29735613e-01 6.80812597e-01 -1.82649761e-01 1.01535225e+00 5.26503265e-01 -2.66190529e-01 -1.35969281e+00 -1.70609200e+00 -1.39663804e+00 2.14424476e-01 8.46446007e-02 -9.17722061e-02 -6.71176016e-01 -7.42385030e-01 6.83097303e-01 1.38590181e+00 -1.01193893e+00 -3.60899895e-01 1.65214702e-01 -8.99664223e-01 2.95414478e-01 4.80016410e-01 -4.88080233e-01 -5.29648900e-01 -1.94002941e-01 1.48003310e-01 3.94032806e-01 -5.72677612e-01 -7.79620111e-01 1.76330194e-01 -7.12950051e-01 -9.03469384e-01 -5.71538329e-01 2.12867364e-01 6.07704699e-01 -4.91764426e-01 1.06404269e+00 -8.71115446e-01 -1.77454710e-01 7.60840654e-01 -9.79450643e-02 -1.07773280e+00 -4.28871393e-01 -8.65245342e-01 7.91223347e-01 -1.88506350e-01 -1.32618444e-02 -2.41222292e-01 -3.12316895e-01 7.95047879e-01 -6.82280719e-01 -8.93715441e-01 1.55112058e-01 5.07542074e-01 1.14576292e+00 2.79564559e-01 5.43634713e-01 -5.91946185e-01 8.80677044e-01 -6.58062279e-01 -1.13314033e+00 6.91589117e-01 -1.00301123e+00 5.91254950e-01 -1.39489368e-01 3.06125470e-02 -1.11745369e+00 -3.48914504e-01 4.59642738e-01 -3.99976939e-01 2.33182043e-01 5.62143743e-01 -1.78676948e-01 -3.46861213e-01 1.18212163e+00 -5.19319654e-01 -2.13994995e-01 2.32898127e-02 2.92138606e-01 6.09493375e-01 4.54255164e-01 -1.05380559e+00 7.61785150e-01 3.94284457e-01 3.27848434e-01 -2.29293242e-01 -1.03084540e+00 2.28102848e-01 -5.30378819e-01 -1.74187839e-01 8.44868302e-01 -5.02620816e-01 -3.38994801e-01 -3.23331475e-01 -6.10054672e-01 1.45755246e-01 -9.33997691e-01 6.72210932e-01 -9.14423645e-01 9.83814374e-02 2.58969128e-01 -1.12158072e+00 -2.76835054e-01 -1.40233076e+00 1.16585004e+00 2.16879219e-01 -4.52525318e-01 -1.20471799e+00 4.03818488e-01 2.19996631e-01 2.42800057e-01 8.93940449e-01 7.74138153e-01 -8.92354488e-01 -2.06419021e-01 -5.55536985e-01 -4.89986092e-02 3.14870686e-03 -2.14574739e-01 1.74970761e-01 -8.41753125e-01 -3.07558477e-01 3.58649522e-01 -1.50147051e-01 5.02942204e-01 9.37453151e-01 1.16854227e+00 -2.29887664e-01 -4.60739881e-01 6.45722091e-01 1.30712378e+00 4.52185571e-01 5.17988145e-01 -3.44324112e-02 1.63297087e-01 7.71708310e-01 8.20895851e-01 7.03211784e-01 2.49474868e-01 4.98213142e-01 4.59222496e-01 6.47713244e-01 6.92474842e-01 -2.27416411e-01 1.16521187e-01 9.58227962e-02 1.81979220e-02 -3.58120263e-01 -1.11038232e+00 1.10472195e-01 -1.74362969e+00 -9.30985868e-01 1.76175669e-01 2.61451101e+00 5.45860827e-01 2.86322176e-01 -4.82668847e-01 -3.11580926e-01 7.35850930e-01 -1.75142080e-01 -8.22441876e-01 -6.48119986e-01 -3.47435236e-01 -2.48976931e-01 8.82195830e-01 1.10255921e+00 -1.22381532e+00 1.82278052e-01 7.78341103e+00 9.46138442e-01 -1.06692813e-01 -3.66777368e-02 8.33683491e-01 -3.25210899e-01 -1.19297481e+00 -7.06988722e-02 -9.93316174e-01 5.11948764e-01 1.07704449e+00 -7.19379246e-01 1.30430371e-01 7.16539025e-01 8.16534758e-02 -1.21832192e-01 -1.40201294e+00 4.05115038e-01 -2.24052705e-02 -1.40071082e+00 -1.47073656e-01 1.99428216e-01 1.39951134e+00 -1.34233296e-01 4.52138603e-01 -2.30189059e-02 9.62577403e-01 -1.61372340e+00 7.26872325e-01 1.28051555e+00 1.17469954e+00 -1.35223055e+00 8.19905400e-01 1.40490547e-01 -6.66456699e-01 -1.76013678e-01 -7.82815218e-01 4.68278557e-01 4.43297386e-01 1.06107044e+00 -6.46678984e-01 4.67405230e-01 6.01695657e-01 1.95262685e-01 -1.05146609e-01 1.11965501e+00 2.96431500e-02 1.16308350e-02 -7.95801342e-01 1.89923599e-01 4.96732444e-02 -6.42265856e-01 1.14248991e+00 5.87920249e-01 8.38649988e-01 4.03387606e-01 -1.41123936e-01 1.13368845e+00 2.00877964e-01 -8.88056755e-02 -1.17272568e+00 2.26099163e-01 8.19961071e-01 6.15275145e-01 -4.32535648e-01 3.22040528e-01 3.13744321e-02 2.80473024e-01 -6.54920191e-02 3.44162434e-01 -7.92895615e-01 -4.26934540e-01 5.31606495e-01 -2.47824118e-01 -2.86561310e-01 2.82966755e-02 -8.78883541e-01 -9.66847539e-01 -5.48661985e-02 -4.47225779e-01 6.70271814e-01 -8.66200387e-01 -1.57315159e+00 4.07390058e-01 4.32990581e-01 -1.35147202e+00 -2.83962160e-01 -6.26118898e-01 -8.35836589e-01 1.06874037e+00 -6.21270597e-01 -7.11659789e-01 2.62439758e-01 3.68153483e-01 1.60415769e-01 -4.42306548e-01 8.60002875e-01 -6.60611272e-01 -3.82296234e-01 7.21703529e-01 9.02365565e-01 -7.03871608e-01 9.22298193e-01 -1.60940278e+00 1.29570246e-01 6.80716872e-01 -1.31187961e-01 3.73116583e-01 1.14222634e+00 -1.02670133e+00 -1.19615340e+00 -1.20565641e+00 2.31461331e-01 -1.41901588e+00 7.93242037e-01 -1.59769263e-02 -5.57961226e-01 9.22910571e-01 -2.13500470e-01 -9.82613415e-02 7.75049925e-01 1.24136254e-01 -8.03753436e-02 4.46105041e-02 -1.83841825e+00 5.95544815e-01 7.67152667e-01 -2.18738303e-01 -8.83119702e-01 6.14172578e-01 1.12573564e+00 -5.22102773e-01 -1.71737528e+00 1.22643471e+00 5.63053071e-01 -6.96124256e-01 1.04321337e+00 -5.90169847e-01 1.55182883e-01 -3.25032651e-01 -9.67041910e-01 -1.73224962e+00 -3.13522607e-01 -6.98555529e-01 1.07839122e-01 9.14896309e-01 1.09098196e+00 -8.41174901e-01 4.93958980e-01 1.39192724e+00 -3.81845355e-01 -8.03125978e-01 -1.30525625e+00 -6.72018528e-01 6.58529401e-01 -7.82299340e-01 8.36944103e-01 7.82909572e-01 2.64923364e-01 -3.89501125e-01 -1.94484085e-01 5.27788460e-01 1.27289760e+00 1.71355322e-01 1.11332551e-01 -1.56825042e+00 -7.08035603e-02 -3.03704113e-01 -4.01759833e-01 9.67769772e-02 1.40799567e-01 -8.51862252e-01 1.56415790e-01 -1.19411302e+00 6.99634552e-02 -9.32590961e-01 -1.46232694e-01 -9.59226564e-02 9.48651321e-03 1.27802603e-03 -1.38729945e-01 1.15433075e-01 -3.55902940e-01 7.85497427e-01 1.14032972e+00 -1.94403544e-01 -2.08032176e-01 4.00492668e-01 -1.02897751e+00 1.02056134e+00 5.69574058e-01 -4.04155701e-01 -6.36820436e-01 1.56548172e-02 5.88726223e-01 6.42999351e-01 -2.70442754e-01 -8.47361863e-01 7.36084655e-02 -8.01793873e-01 7.06943810e-01 -5.43460131e-01 1.73115283e-01 -9.15188789e-01 6.82637274e-01 -1.19043931e-01 -4.99904424e-01 -2.99085514e-03 2.59988576e-01 9.12613153e-01 7.40316585e-02 -5.86034298e-01 8.07465971e-01 3.70278001e-01 -3.38480383e-01 4.67360437e-01 -7.94545934e-02 2.25813910e-01 1.79103267e+00 -9.46311653e-02 -3.64992887e-01 -5.66015422e-01 -1.10696614e+00 6.30818427e-01 7.29448140e-01 2.27150455e-01 7.85256922e-01 -1.57353675e+00 -1.08338308e+00 1.50968760e-01 3.32714885e-01 1.94227397e-01 6.22729138e-02 5.17288446e-01 -2.17049047e-01 5.53784490e-01 5.38554229e-02 -6.07797503e-01 -5.52879333e-01 6.81304455e-01 6.66891277e-01 8.93693566e-02 -2.61482477e-01 9.73003566e-01 2.37974778e-01 -8.62572908e-01 1.10364944e-01 -2.57102132e-01 -9.32806507e-02 -1.00444444e-01 4.72857565e-01 6.64014339e-01 -3.52813229e-02 -4.66636240e-01 -2.45950684e-01 8.70395482e-01 4.25635606e-01 -5.83561122e-01 8.62433255e-01 -2.77621597e-01 8.96103494e-03 5.02829552e-01 7.32030213e-01 3.39570314e-01 -1.56279635e+00 4.40193005e-02 7.07922056e-02 -5.36995769e-01 4.01991699e-03 -9.82314289e-01 -3.73775959e-01 3.54015619e-01 5.33858776e-01 1.89290375e-01 4.13780004e-01 1.05456926e-01 -1.91473156e-01 4.19416428e-01 5.67108393e-01 -9.92147923e-01 -5.80125988e-01 2.99869716e-01 1.75308049e+00 -1.51304877e+00 3.65545869e-01 -2.10006341e-01 -1.01738751e+00 9.15993929e-01 2.85346210e-01 -5.34384772e-02 1.13076448e+00 7.56568372e-01 -1.13248877e-01 -4.10843492e-02 -5.64820945e-01 4.93989736e-01 1.05538738e+00 6.94069445e-01 -1.53974503e-01 6.32263362e-01 3.53982002e-02 9.38688159e-01 -4.55882758e-01 -6.96087956e-01 4.88414288e-01 6.47390664e-01 -7.94233859e-01 -5.35328031e-01 -9.84564185e-01 8.39492142e-01 -3.37676495e-01 1.23585328e-01 -1.13757886e-01 3.70851874e-01 -3.76849920e-01 1.08273923e+00 2.71986037e-01 -2.13822573e-01 4.09637481e-01 -1.89232498e-01 2.65804261e-01 -3.68707597e-01 3.43502164e-01 -2.08492249e-01 3.03928196e-01 -6.56147838e-01 2.83581376e-01 -9.57650840e-01 -1.11854315e+00 -2.96914220e-01 -5.02548873e-01 4.18932498e-01 5.96570492e-01 9.68559325e-01 1.77981660e-01 1.75528795e-01 7.28918493e-01 -8.83093178e-01 -1.51438677e+00 -7.27068663e-01 -9.53450322e-01 -1.83464110e-01 1.99779779e-01 -9.10501122e-01 -5.79917371e-01 -7.09334791e-01]
[5.287027359008789, 3.897360324859619]
62e67138-6166-4897-b92f-c6dc2f06b2ad
transfer-learning-for-relation-extraction-via
1908.08507
null
https://arxiv.org/abs/1908.08507v1
https://arxiv.org/pdf/1908.08507v1.pdf
Transfer Learning for Relation Extraction via Relation-Gated Adversarial Learning
Relation extraction aims to extract relational facts from sentences. Previous models mainly rely on manually labeled datasets, seed instances or human-crafted patterns, and distant supervision. However, the human annotation is expensive, while human-crafted patterns suffer from semantic drift and distant supervision samples are usually noisy. Domain adaptation methods enable leveraging labeled data from a different but related domain. However, different domains usually have various textual relation descriptions and different label space (the source label space is usually a superset of the target label space). To solve these problems, we propose a novel model of relation-gated adversarial learning for relation extraction, which extends the adversarial based domain adaptation. Experimental results have shown that the proposed approach outperforms previous domain adaptation methods regarding partial domain adaptation and can improve the accuracy of distance supervised relation extraction through fine-tuning.
['Wei zhang', 'Ningyu Zhang', 'Jiaoyan Chen', 'Huajun Chen', 'Shumin Deng', 'Zhanlin Sun']
2019-08-22
null
null
null
null
['partial-domain-adaptation']
['methodology']
[ 3.20754230e-01 3.29838037e-01 -7.81113923e-01 -7.01690495e-01 -6.03408694e-01 -4.95780796e-01 5.80584824e-01 2.45666161e-01 -3.18150073e-01 1.17572236e+00 1.00075297e-01 -5.92740178e-02 -2.56361842e-01 -1.05265141e+00 -6.44452691e-01 -5.09366989e-01 2.27271140e-01 8.96221578e-01 3.73986959e-01 -3.79905015e-01 -2.41849005e-01 1.48531258e-01 -8.10786188e-01 1.27429709e-01 9.54770803e-01 9.53518867e-01 -1.42107978e-01 4.06981371e-02 -2.75814205e-01 1.01886904e+00 -8.68936718e-01 -7.03844845e-01 1.45027861e-01 -5.35343289e-01 -1.18229818e+00 -3.56288301e-03 -6.77252561e-02 1.71529315e-02 -4.77774948e-01 1.11836338e+00 2.41938785e-01 1.72022656e-01 6.33459151e-01 -1.29218769e+00 -1.09225118e+00 1.05886567e+00 -3.73720616e-01 2.41694868e-01 2.87253171e-01 -2.57317096e-01 1.11602545e+00 -6.19647920e-01 9.40988541e-01 9.43199039e-01 6.42059803e-01 6.57834411e-01 -1.32966709e+00 -9.10809577e-01 4.04451340e-01 3.44246596e-01 -1.57483625e+00 -1.73778787e-01 1.16869259e+00 -2.53744096e-01 8.31753433e-01 -6.74128383e-02 2.78593183e-01 1.51413131e+00 -4.33716446e-01 6.55941308e-01 9.31852102e-01 -4.21261877e-01 2.75063992e-01 4.70810294e-01 2.03085124e-01 2.93069243e-01 3.22470158e-01 -7.01359706e-03 -5.05271673e-01 -2.71062613e-01 5.07145226e-01 -5.11188619e-02 -4.17325646e-01 -5.38593054e-01 -1.11002362e+00 6.75117850e-01 4.56756413e-01 5.35654068e-01 -1.05851457e-01 -5.78085244e-01 6.45825744e-01 5.03609657e-01 5.54081798e-01 5.29607058e-01 -9.31663394e-01 1.22122869e-01 -5.41251063e-01 2.09836259e-01 8.07819128e-01 1.46780121e+00 9.21716332e-01 -3.18473965e-01 -2.42705941e-01 8.99578691e-01 -3.56211029e-02 5.19720018e-01 6.98665559e-01 -4.25304502e-01 1.04051697e+00 9.51308131e-01 9.48223546e-02 -9.59402442e-01 -3.00364941e-01 -2.87856430e-01 -8.80214453e-01 -3.78648460e-01 5.57471514e-01 -3.32655400e-01 -7.08690226e-01 1.59757006e+00 4.90786076e-01 7.91406929e-02 4.97419506e-01 7.85180688e-01 9.54775453e-01 2.83056647e-01 3.01214278e-01 -2.91254610e-01 1.13019013e+00 -8.99524391e-01 -9.16244507e-01 -4.53697830e-01 6.92944229e-01 -2.97948837e-01 1.03021932e+00 2.20354646e-01 -4.50154752e-01 -5.15036941e-01 -1.17487514e+00 1.34701310e-02 -3.50640029e-01 5.80512658e-02 6.95194125e-01 4.70620722e-01 -1.47088200e-01 6.12159848e-01 -4.94585872e-01 -2.24299625e-01 7.02388227e-01 3.07734579e-01 -5.79582393e-01 -1.32327512e-01 -1.65915132e+00 7.83615947e-01 9.76178050e-01 -1.87738970e-01 -5.31138599e-01 -3.68994236e-01 -9.50309277e-01 -9.44510549e-02 8.16154480e-01 -5.26295841e-01 1.07572126e+00 -8.91242623e-01 -1.41403031e+00 9.96419013e-01 -2.31545530e-02 -7.52637327e-01 3.44614953e-01 -2.74425358e-01 -8.14484358e-01 -6.73034340e-02 1.92863330e-01 8.69130120e-02 7.37633407e-01 -1.08990622e+00 -5.09850740e-01 -3.62665355e-01 -7.29823392e-03 1.08710095e-01 -5.56515038e-01 -3.81061621e-03 -1.39202818e-01 -8.08838785e-01 -4.15320881e-02 -6.45414889e-01 -1.47443429e-01 -5.49610615e-01 -4.96068895e-01 -4.20273095e-01 8.97607088e-01 -4.72261459e-01 1.13582909e+00 -2.13822556e+00 5.93320057e-02 1.43157631e-01 1.87012553e-02 4.83320177e-01 1.43049181e-01 1.05423689e-01 -1.59385771e-01 1.73288900e-02 -2.88607210e-01 6.16388097e-02 -2.19472706e-01 5.18138647e-01 -2.74598479e-01 2.55429894e-02 3.84142697e-01 7.64848351e-01 -1.32361162e+00 -7.09771037e-01 -1.51578739e-01 5.62852807e-02 2.20999066e-02 2.58105040e-01 -6.05220199e-01 6.90850317e-01 -7.80891299e-01 5.29522240e-01 6.60693467e-01 -3.52822483e-01 5.47000825e-01 -1.17301188e-01 8.19017828e-01 4.88332421e-01 -1.18467855e+00 1.82101011e+00 -3.82502526e-01 2.52044320e-01 -4.39730018e-01 -1.35354340e+00 1.54726315e+00 5.32863796e-01 2.32516944e-01 -3.49282920e-01 2.42658183e-01 1.75839841e-01 -3.16957869e-02 -7.08367765e-01 1.55298844e-01 -5.29556692e-01 -2.31461421e-01 2.41656989e-01 3.02822143e-01 2.10247878e-02 1.55314550e-01 3.97155546e-02 1.25207877e+00 2.57719696e-01 5.85746408e-01 3.04711014e-01 7.01244235e-01 1.21073194e-01 1.16299510e+00 2.92897582e-01 -2.63346940e-01 3.76065791e-01 5.47680914e-01 -4.16502923e-01 -7.39302039e-01 -8.16002905e-01 -3.60742897e-01 8.16505432e-01 3.58348995e-01 -4.58262533e-01 -5.09019494e-01 -1.63763893e+00 1.64601095e-02 7.16659367e-01 -5.88675439e-01 -4.87498611e-01 -5.96465230e-01 -7.84464478e-01 5.32165229e-01 7.80557811e-01 6.02077901e-01 -1.31774831e+00 2.08488911e-01 3.45268190e-01 -3.11139286e-01 -1.59585786e+00 -1.20548688e-01 3.98002088e-01 -9.36075926e-01 -1.17832172e+00 -2.11046010e-01 -1.00584280e+00 7.20413268e-01 -1.46925732e-01 1.27002800e+00 -2.45784581e-01 3.13584328e-01 -3.91964585e-01 -6.02002382e-01 -3.23187709e-01 -5.38264096e-01 4.47618157e-01 1.20549239e-01 -2.65895892e-02 1.19882083e+00 -7.14468598e-01 -1.09550655e-01 3.26612979e-01 -6.91147566e-01 -4.52393115e-01 5.36108255e-01 1.17037404e+00 5.86968064e-01 5.47383010e-01 8.43950510e-01 -1.68116581e+00 6.65950119e-01 -5.24347007e-01 -3.23742986e-01 5.80845892e-01 -8.13642800e-01 3.10871989e-01 9.65252519e-01 -7.23768651e-01 -1.41504395e+00 1.56881288e-01 3.17243427e-01 -2.92853057e-01 -5.69792509e-01 3.34789723e-01 -8.18563461e-01 3.66109341e-01 1.11710477e+00 -4.98881154e-02 -3.35342288e-01 -4.03606832e-01 4.15277064e-01 7.01191902e-01 5.79202116e-01 -8.35333049e-01 1.11528480e+00 2.23307267e-01 -2.06099391e-01 -1.68601200e-01 -1.25776899e+00 -1.31808311e-01 -1.06426454e+00 4.75380123e-01 6.34549916e-01 -8.83405983e-01 -1.06987886e-01 5.61400466e-02 -1.07207549e+00 -1.84830055e-01 -6.12226665e-01 5.11769652e-01 -2.42548510e-01 1.55918643e-01 -5.45120001e-01 -4.55632925e-01 -3.61712545e-01 -5.67921460e-01 7.61150599e-01 3.47118706e-01 -4.15509433e-01 -1.15437376e+00 7.95903876e-02 4.65073586e-01 -2.21151918e-01 3.89434904e-01 8.93423617e-01 -1.33132696e+00 -2.88992316e-01 -5.27271688e-01 -1.35440230e-01 4.58976895e-01 6.00499511e-01 -3.64212930e-01 -7.85793900e-01 9.27188918e-02 -1.09665981e-02 -5.53831279e-01 4.69680876e-01 -2.37132043e-01 7.72292912e-01 -3.27817053e-01 -7.88158476e-01 2.83568114e-01 1.13530219e+00 1.90373272e-01 1.53460592e-01 5.03467262e-01 8.61124575e-01 6.76784575e-01 1.18365157e+00 1.80045053e-01 2.89061278e-01 6.43080771e-01 -1.48264587e-01 1.14506505e-01 1.69923138e-02 -5.90460658e-01 2.79536303e-02 4.71293360e-01 -6.61201105e-02 -2.46153027e-02 -9.44486201e-01 7.26938426e-01 -1.94702435e+00 -8.64403129e-01 9.85841304e-02 1.95359838e+00 1.63298643e+00 6.28154337e-01 1.43308297e-01 4.07230318e-01 8.35686088e-01 -9.85401347e-02 -6.48245156e-01 -3.05538662e-02 -1.64194420e-01 3.51461649e-01 4.54958826e-01 3.34722400e-01 -1.24322939e+00 1.21723831e+00 5.24896479e+00 8.74320149e-01 -8.02761555e-01 2.05641806e-01 3.22119057e-01 1.59862876e-01 -3.01575392e-01 1.52129278e-01 -9.84087348e-01 3.94931942e-01 7.65002608e-01 -2.62775004e-01 6.39609247e-02 1.11345434e+00 -3.47858846e-01 2.65324771e-01 -1.39280128e+00 7.33871281e-01 2.20504683e-02 -9.60171342e-01 -1.45252958e-01 -2.26016298e-01 8.39591503e-01 -3.27565670e-01 -2.35544279e-01 4.89131063e-01 7.46291459e-01 -8.44686449e-01 1.51516885e-01 2.29164198e-01 7.70658314e-01 -7.59899974e-01 1.06804812e+00 4.36906844e-01 -1.03130448e+00 1.19146615e-01 -4.02772486e-01 -8.98412913e-02 -5.18363938e-02 8.37697566e-01 -1.17635942e+00 8.62636685e-01 6.44027650e-01 1.08086371e+00 -5.68770111e-01 3.32637966e-01 -9.03337359e-01 5.29651225e-01 -1.72856167e-01 -5.95996752e-02 -2.63644934e-01 -3.42163444e-02 2.60234237e-01 9.59989488e-01 -1.85019374e-01 1.29391462e-01 2.70563543e-01 6.95701301e-01 -3.23183089e-01 1.00209564e-01 -8.36072326e-01 1.05983838e-01 8.53778243e-01 1.05237293e+00 -5.13275027e-01 -4.57226574e-01 -6.77264750e-01 9.33372855e-01 5.79378128e-01 4.47335809e-01 -7.59504318e-01 -6.40671968e-01 3.82770956e-01 -1.21981487e-01 4.78037983e-01 2.07948238e-01 -5.59372187e-01 -1.29452825e+00 2.99963146e-01 -9.70039368e-01 5.75873613e-01 -2.74025768e-01 -1.63552010e+00 8.29456985e-01 -4.56169285e-02 -1.38575840e+00 -2.93171495e-01 -3.42938811e-01 -1.91762000e-01 6.95217252e-01 -1.37830234e+00 -1.05324948e+00 -1.88799128e-01 7.73252428e-01 2.59302646e-01 -4.95593399e-01 1.07996142e+00 3.50495070e-01 -7.14202404e-01 9.49147463e-01 -1.21971555e-01 6.99395299e-01 1.05765069e+00 -1.31202340e+00 3.19063842e-01 7.21685946e-01 4.16699320e-01 5.67979515e-01 5.44591010e-01 -8.34697247e-01 -6.29283071e-01 -1.31139183e+00 1.11324787e+00 -6.77998602e-01 7.89818823e-01 -2.60457963e-01 -1.21004784e+00 1.06291592e+00 -5.13247699e-02 4.42464054e-01 9.62320924e-01 5.54215670e-01 -6.38600647e-01 -5.20827055e-01 -1.44449174e+00 3.51525486e-01 1.28804874e+00 -6.06489360e-01 -9.93822455e-01 3.90961915e-01 7.88726270e-01 -4.75337178e-01 -1.18716586e+00 6.14462495e-01 -1.24479048e-01 -4.51655835e-01 8.68447185e-01 -8.19809139e-01 3.46247643e-01 -4.48260754e-01 7.33250380e-02 -1.22918284e+00 -1.16609305e-01 -2.87748098e-01 -4.64121103e-01 1.75189984e+00 6.44964218e-01 -4.66703355e-01 9.04611945e-01 7.64216423e-01 3.73843670e-01 -5.22722065e-01 -5.96100509e-01 -1.16973317e+00 8.26204270e-02 -1.37449965e-01 9.60071325e-01 1.41141486e+00 2.69130379e-01 1.19068134e+00 -2.66800106e-01 3.71295691e-01 4.93365198e-01 4.37584966e-01 7.51126528e-01 -1.45456743e+00 -4.24240977e-01 -4.37822975e-02 -4.11263913e-01 -7.70483732e-01 4.29379284e-01 -8.99742424e-01 -1.55517170e-02 -1.16689301e+00 -5.71282068e-03 -8.17953110e-01 -3.18796486e-01 8.18471909e-01 -5.94029725e-01 -1.31903291e-01 -3.68719757e-01 1.81585535e-01 -6.82299197e-01 5.61953664e-01 1.23138714e+00 -3.40487897e-01 -2.80289829e-01 2.37575144e-01 -7.68056154e-01 6.65902019e-01 8.94984543e-01 -8.30493093e-01 -7.76040494e-01 -2.62931705e-01 1.84712231e-01 -2.95515731e-02 -1.15384189e-02 -8.57920229e-01 7.03195259e-02 -3.37757885e-01 4.25150245e-01 -1.68343633e-01 -2.05364153e-02 -1.08341384e+00 -9.58845615e-02 3.18952464e-02 -4.26092982e-01 -5.88491976e-01 -1.36447161e-01 8.67818236e-01 -5.22172928e-01 -2.37678692e-01 5.38567424e-01 -1.39163271e-01 -6.04675829e-01 2.60627329e-01 3.10189724e-01 5.73087335e-01 1.21438026e+00 1.42694265e-01 -1.65062726e-01 1.99979860e-02 -9.25217211e-01 1.61683753e-01 2.39630356e-01 4.75642174e-01 3.71286064e-01 -1.58849955e+00 -6.86385214e-01 9.53618661e-02 4.60778326e-01 6.67014122e-01 -2.41956875e-01 2.61444181e-01 6.97254762e-02 1.59037679e-01 -3.34907137e-03 -3.25616866e-01 -1.09824455e+00 1.08119297e+00 1.28368035e-01 -6.96257055e-01 -6.23495996e-01 1.00318730e+00 -2.23026201e-01 -5.86876750e-01 2.15190262e-01 -3.00327986e-01 -3.10126632e-01 -8.92321914e-02 2.75992185e-01 1.78118851e-02 1.45686895e-01 -3.43131781e-01 -4.54475462e-01 4.20696020e-01 -3.92528802e-01 2.68983603e-01 1.30366421e+00 7.21167475e-02 5.90333939e-02 5.18312931e-01 9.39346850e-01 2.36013927e-03 -1.05825877e+00 -1.13623857e+00 5.14016747e-01 -3.93182099e-01 -4.88620400e-01 -8.12256694e-01 -9.96284962e-01 4.79632288e-01 3.47861387e-02 2.04991102e-01 1.16326165e+00 2.38657296e-01 1.05544198e+00 4.44400191e-01 4.92557436e-01 -1.14488053e+00 8.98945704e-03 3.85799080e-01 5.99906802e-01 -1.43453169e+00 5.81256300e-02 -8.12899947e-01 -8.82913649e-01 7.37108767e-01 1.03089011e+00 -3.86171378e-02 6.80672526e-01 3.21266651e-01 1.91770673e-01 -4.05622926e-03 -5.62270701e-01 -1.79712027e-01 1.87042758e-01 1.10950017e+00 4.49174404e-01 1.14447154e-01 -2.97202498e-01 1.17332041e+00 -2.63374209e-01 7.94777498e-02 -1.05485223e-01 8.06785405e-01 7.15775415e-02 -1.81164527e+00 -1.89555764e-01 2.82884210e-01 -4.10252869e-01 -7.19034225e-02 -6.02129400e-01 6.95482850e-01 3.81257981e-01 9.86477673e-01 -2.51854479e-01 -5.70828855e-01 5.22698998e-01 3.66211534e-01 4.72115189e-01 -8.89326632e-01 -4.65280116e-01 -4.57943022e-01 3.88361573e-01 2.04687403e-03 -5.86339951e-01 -6.03939354e-01 -1.41642153e+00 -2.65828352e-02 -5.45180023e-01 4.97705966e-01 -1.14823714e-01 1.24120760e+00 2.69981205e-01 5.19545138e-01 6.49479449e-01 1.65632322e-01 -3.91663939e-01 -1.33370078e+00 -5.69358528e-01 9.71925259e-01 2.31917262e-01 -7.36676037e-01 8.37337226e-03 3.23860228e-01]
[9.201451301574707, 8.493167877197266]
7b485709-1ce2-4c19-93b2-b364a271f419
online-3d-bin-packing-with-constrained-deep
2006.14978
null
https://arxiv.org/abs/2006.14978v5
https://arxiv.org/pdf/2006.14978v5.pdf
Online 3D Bin Packing with Constrained Deep Reinforcement Learning
We solve a challenging yet practically useful variant of 3D Bin Packing Problem (3D-BPP). In our problem, the agent has limited information about the items to be packed into the bin, and an item must be packed immediately after its arrival without buffering or readjusting. The item's placement also subjects to the constraints of collision avoidance and physical stability. We formulate this online 3D-BPP as a constrained Markov decision process. To solve the problem, we propose an effective and easy-to-implement constrained deep reinforcement learning (DRL) method under the actor-critic framework. In particular, we introduce a feasibility predictor to predict the feasibility mask for the placement actions and use it to modulate the action probabilities output by the actor during training. Such supervisions and transformations to DRL facilitate the agent to learn feasible policies efficiently. Our method can also be generalized e.g., with the ability to handle lookahead or items with different orientations. We have conducted extensive evaluation showing that the learned policy significantly outperforms the state-of-the-art methods. A user study suggests that our method attains a human-level performance.
['Qijin She', 'Yin Yang', 'Hang Zhao', 'Kai Xu', 'Chenyang Zhu']
2020-06-26
null
null
null
null
['3d-bin-packing']
['miscellaneous']
[-2.31668368e-01 1.55134246e-01 -5.30982137e-01 -1.32719144e-01 -4.50672567e-01 -6.89239740e-01 4.34635282e-02 3.34928006e-01 -6.36108637e-01 8.66813540e-01 -8.82302001e-02 -6.76632404e-01 -3.69148748e-03 -7.04422712e-01 -1.13946450e+00 -6.74967587e-01 -5.65729976e-01 9.59897578e-01 1.46847695e-01 -2.94448018e-01 2.13391542e-01 6.49729133e-01 -1.08472800e+00 -7.23961666e-02 5.66956878e-01 1.17579234e+00 5.99792063e-01 7.61840820e-01 7.79570863e-02 6.73841596e-01 -6.11677706e-01 5.92804924e-02 7.13074625e-01 -1.57186553e-01 -6.93253219e-01 4.51952487e-01 -3.07556540e-01 -7.40649879e-01 -2.18685597e-01 5.37232339e-01 4.30152118e-01 4.58248436e-01 4.42139208e-01 -1.31748950e+00 -4.33527887e-01 5.22707582e-01 -6.46793544e-01 4.55391139e-01 2.97198892e-01 5.78298330e-01 8.81896734e-01 -2.60381162e-01 3.61673176e-01 1.14570999e+00 9.62850358e-03 4.65561837e-01 -1.28283954e+00 -1.52551308e-01 8.17037106e-01 1.03602789e-01 -7.91016757e-01 -1.71913654e-01 3.24480891e-01 -2.79900134e-01 1.35331309e+00 -7.80134052e-02 1.00515187e+00 7.73146152e-01 4.68034595e-01 9.48712051e-01 8.09525728e-01 -4.32592988e-01 8.68353367e-01 -2.36497417e-01 -8.64844695e-02 6.26494586e-01 4.13688391e-01 3.56396765e-01 -2.06126690e-01 1.41508177e-01 9.00275886e-01 1.03317685e-01 8.30770470e-03 -8.78186941e-01 -8.91429663e-01 8.27596664e-01 3.88967991e-01 -3.84298474e-01 -6.23126864e-01 3.68567050e-01 2.51166403e-01 2.57715851e-01 6.94682598e-02 6.08422577e-01 -4.73874748e-01 -2.39911571e-01 -5.14295399e-01 6.87830567e-01 1.07076812e+00 1.34214842e+00 4.11821187e-01 1.12795215e-02 -3.66976172e-01 3.93899769e-01 2.80019104e-01 5.06582260e-01 -4.64247502e-02 -9.44324315e-01 9.81842935e-01 1.02491647e-01 8.87052774e-01 -5.24392605e-01 -5.75302958e-01 -1.88366652e-01 -4.55776155e-01 3.47308159e-01 2.99886972e-01 -4.62266713e-01 -1.16554117e+00 1.53032422e+00 3.40405464e-01 -3.81059200e-02 -1.14909904e-02 1.28216934e+00 -8.99861977e-02 9.67557132e-01 -4.99190688e-02 -3.63345742e-01 9.86567914e-01 -1.24823213e+00 -5.00446916e-01 -4.96180922e-01 2.15863124e-01 -4.59059179e-01 6.34149194e-01 4.92218733e-01 -1.54152060e+00 -3.09410423e-01 -1.24590111e+00 3.41009438e-01 -1.68018699e-01 -2.75540799e-01 6.75023019e-01 3.02648813e-01 -8.08176160e-01 6.32467568e-01 -1.29017663e+00 -1.76958311e-02 2.59470701e-01 7.53297389e-01 6.98388517e-02 -8.27739760e-02 -8.10428083e-01 1.14252579e+00 4.78847742e-01 1.38010174e-01 -1.42127061e+00 -1.51527926e-01 -9.31407750e-01 3.37945104e-01 9.87360120e-01 -5.98676383e-01 1.85834873e+00 -5.22895753e-01 -1.84350801e+00 3.18493545e-01 6.05578944e-02 -7.67201364e-01 2.60558337e-01 -2.35614449e-01 1.87107801e-01 7.83395246e-02 2.25238167e-02 6.50866628e-01 7.95363843e-01 -1.13334084e+00 -8.38633895e-01 -3.97544913e-02 7.00116277e-01 6.04015291e-01 1.69524163e-01 -3.30969512e-01 -6.33646429e-01 -3.60812664e-01 -1.63893595e-01 -1.14523637e+00 -6.96286082e-01 -1.40013024e-01 -3.22894335e-01 -1.11855179e-01 1.86852038e-01 -3.11649233e-01 9.65129375e-01 -1.91661680e+00 4.05208290e-01 1.79541394e-01 -2.06277966e-01 1.42268255e-01 -1.39599308e-01 6.22854888e-01 2.66429126e-01 -5.00882193e-02 1.34747043e-01 -5.06321609e-01 4.12527323e-01 6.38523579e-01 -3.13849479e-01 4.81608152e-01 1.59821242e-01 7.31015563e-01 -9.01917040e-01 -1.67307287e-01 2.58731723e-01 -2.15996027e-01 -1.06427920e+00 6.58512354e-01 -7.45027065e-01 2.59399116e-01 -6.17426157e-01 4.75683808e-01 6.35577321e-01 -2.14762047e-01 4.48689640e-01 4.15490896e-01 -1.95945367e-01 3.70914161e-01 -1.29051650e+00 1.46160984e+00 -5.33964396e-01 -4.67704497e-02 3.80732298e-01 -1.16484892e+00 5.93641758e-01 -9.13752392e-02 2.73814261e-01 -8.64558160e-01 1.30330905e-01 -1.35872260e-01 -1.46995019e-02 -4.30033684e-01 7.13356853e-01 -1.79645661e-02 -2.13943943e-01 3.58828247e-01 -1.54551417e-01 -2.98607200e-01 4.35012400e-01 -1.62431356e-02 9.25268948e-01 2.22346634e-01 4.32805300e-01 -1.13957413e-01 -6.26614876e-03 1.39189810e-01 7.17426121e-01 1.05438352e+00 -1.68415204e-01 -1.20422341e-01 7.54740655e-01 -4.70567673e-01 -1.09929132e+00 -9.81153548e-01 2.84336895e-01 1.15586209e+00 6.28389120e-01 3.80530313e-04 -5.59042454e-01 -5.33633113e-01 4.44312483e-01 7.30019450e-01 -3.90557826e-01 5.64677976e-02 -7.20358193e-01 -4.28181827e-01 -4.25561786e-01 7.97482312e-01 2.29586616e-01 -1.00146961e+00 -1.05930591e+00 5.90921760e-01 2.63713926e-01 -1.10593343e+00 -7.85715997e-01 8.57588410e-01 -7.08450139e-01 -8.71766746e-01 -5.22732735e-01 -9.45139885e-01 8.79667759e-01 4.63774383e-01 9.44641650e-01 -1.27863333e-01 -5.84861822e-02 1.87278062e-01 -4.60950643e-01 -3.54030252e-01 -1.24448620e-01 4.96281870e-02 1.46603823e-01 -6.16540611e-01 -1.02510110e-01 -2.11398587e-01 -7.46161759e-01 4.85303015e-01 -6.07767761e-01 2.70903736e-01 5.73327243e-01 9.26365912e-01 8.64854872e-01 1.32151753e-01 2.86814302e-01 -4.54637468e-01 6.53139055e-01 -4.55433637e-01 -1.20206368e+00 1.03036195e-01 -3.19050312e-01 3.59370053e-01 8.91052186e-01 -4.41252559e-01 -7.42476761e-01 5.86277731e-02 1.85945909e-02 -4.51221704e-01 2.10143197e-02 4.12917495e-01 -1.43202171e-01 2.00140655e-01 -4.84640971e-02 -1.18911564e-01 -1.14941180e-01 -2.93791741e-01 2.67730802e-01 2.95121670e-01 2.01657921e-01 -9.16312099e-01 4.23943549e-01 2.92127933e-02 4.84420024e-02 -1.85663983e-01 -8.37027788e-01 -2.76136816e-01 -3.05951357e-01 -6.85581341e-02 6.91050589e-01 -8.48895490e-01 -1.34294248e+00 2.00790882e-01 -9.75003183e-01 -1.16349506e+00 -2.99797744e-01 2.67519534e-01 -9.97833669e-01 6.40491992e-02 -8.19448531e-01 -9.70250309e-01 -6.08221777e-02 -1.31812096e+00 8.17656875e-01 4.31783646e-01 2.87419677e-01 -6.76801980e-01 1.49421534e-03 6.22539520e-02 3.72288913e-01 1.44391894e-01 8.97238314e-01 -5.09002388e-01 -8.87389004e-01 1.10871784e-01 1.03024147e-01 -1.58018507e-02 -7.36907423e-02 -2.86320627e-01 -1.30273700e-01 -7.38274097e-01 -2.20164150e-01 -4.04564887e-01 5.26808798e-01 5.69359541e-01 1.36739874e+00 -6.09385431e-01 -4.22037512e-01 4.63469356e-01 1.34071898e+00 6.81884944e-01 1.61532208e-01 5.84483147e-01 1.43153548e-01 1.51452392e-01 1.04592860e+00 8.86645317e-01 5.48583567e-01 9.50362086e-01 1.00023997e+00 2.03678280e-01 5.03118634e-01 -3.96823287e-01 3.77730012e-01 3.47702414e-01 8.93788263e-02 -8.15967441e-01 -6.98673069e-01 2.84301013e-01 -2.16092420e+00 -8.47330928e-01 6.80407643e-01 2.26972437e+00 6.72750831e-01 7.50026584e-01 2.10111231e-01 -2.65593559e-01 6.46152914e-01 1.15052266e-02 -1.05532670e+00 -9.42653358e-01 5.21084011e-01 1.08048707e-01 1.01351357e+00 7.26839900e-01 -1.13589239e+00 1.03592837e+00 6.55236053e+00 4.23445404e-01 -8.87667358e-01 -2.87439436e-01 6.61938787e-01 -4.13117498e-01 1.04781471e-01 -1.33687630e-01 -1.05723846e+00 6.31124079e-01 8.08090329e-01 -3.00434930e-03 1.09994900e+00 9.93505895e-01 5.08209109e-01 -5.98920882e-01 -1.39238620e+00 6.69612646e-01 -2.58933663e-01 -1.20513165e+00 -3.22259724e-01 2.55156696e-01 6.87040985e-01 -2.71568507e-01 5.42331673e-02 6.13751769e-01 7.27731645e-01 -8.07869256e-01 9.64834034e-01 2.70753771e-01 4.62184876e-01 -1.06435716e+00 6.42989576e-01 6.00336969e-01 -1.02919710e+00 -5.67971110e-01 -6.02782905e-01 -3.97464126e-01 6.36055052e-01 2.15282366e-01 -1.11266625e+00 2.71129578e-01 3.95676643e-01 3.14466029e-01 1.49600238e-01 1.15595329e+00 -4.46803421e-01 2.59215921e-01 -4.75202113e-01 -4.72761810e-01 6.22962654e-01 -2.14273423e-01 3.32468003e-01 8.96818101e-01 1.25520751e-01 4.67261851e-01 8.76668811e-01 6.94573998e-01 1.16424285e-01 -3.70148867e-01 -2.21950814e-01 -8.02281499e-02 5.01634300e-01 8.64732444e-01 -8.95241916e-01 -2.05024451e-01 -2.26607978e-01 9.41556633e-01 6.80135548e-01 2.95407623e-01 -1.19361484e+00 -1.85092241e-02 8.10360610e-01 -6.67603016e-02 9.41098094e-01 -6.39387071e-01 2.34805532e-02 -8.25235426e-01 -7.28437454e-02 -6.43106103e-01 2.17675224e-01 -5.98112702e-01 -1.14908338e+00 2.97157317e-01 1.01459615e-01 -8.84879768e-01 -2.46972114e-01 -8.13393474e-01 -5.20238757e-01 6.80519521e-01 -1.77948976e+00 -3.53395909e-01 3.98859158e-02 4.54190373e-01 7.01910675e-01 -2.48538833e-02 5.56767464e-01 -6.05404042e-02 -8.84947300e-01 3.56150687e-01 1.35930225e-01 -1.43505976e-01 3.05574596e-01 -1.45670152e+00 6.24450147e-01 7.51154602e-01 -4.11181927e-01 2.08306924e-01 8.40338111e-01 -5.94286680e-01 -1.99570417e+00 -9.43863034e-01 1.33721650e-01 4.42139730e-02 4.48735744e-01 -3.58058035e-01 -3.32589835e-01 8.53396297e-01 1.30647317e-01 -6.33809939e-02 3.02387148e-01 -1.01119496e-01 2.75213569e-01 -7.22342208e-02 -9.67764318e-01 6.05786204e-01 1.01143301e+00 2.94443429e-01 -6.32356942e-01 4.52337861e-01 9.44358885e-01 -1.12571418e+00 -4.52970922e-01 5.78025095e-02 1.89215750e-01 -7.39283085e-01 6.75575435e-01 -8.82085502e-01 3.08806628e-01 -1.46505564e-01 -1.73064157e-01 -1.58676362e+00 -6.23196304e-01 -8.14438522e-01 -5.07647812e-01 5.58848739e-01 3.85185897e-01 -4.73469764e-01 9.23490644e-01 8.92145753e-01 -4.29422826e-01 -1.14047635e+00 -9.23053622e-01 -8.54980052e-01 -4.08840328e-02 -5.83572760e-02 6.35754287e-01 2.41107300e-01 3.92013490e-01 -9.56924725e-03 -5.24197042e-01 5.84398568e-01 3.12407911e-01 5.68901241e-01 7.46491671e-01 -3.41570020e-01 -8.66587579e-01 -2.71502435e-01 1.05575800e-01 -1.96351802e+00 4.36125807e-02 -4.98530865e-01 4.93137002e-01 -1.59919071e+00 1.49885580e-01 -7.83238411e-01 -2.25324258e-01 6.32763863e-01 -3.40057700e-03 -5.29777229e-01 7.08177507e-01 -1.20638512e-01 -1.11489034e+00 5.59235275e-01 1.58692074e+00 -3.84866185e-02 -5.24203122e-01 2.74096161e-01 -5.25608599e-01 2.76543349e-01 1.00798678e+00 -2.75083870e-01 -2.96369374e-01 -5.80492616e-01 2.01868653e-01 7.04017103e-01 -4.73013259e-02 -7.23067641e-01 3.23651046e-01 -7.51451433e-01 3.85111958e-01 -5.25968313e-01 4.32643473e-01 -9.61836934e-01 -2.18163237e-01 6.63299799e-01 -2.76091039e-01 4.25621033e-01 3.55049402e-01 6.52591109e-01 3.32971305e-01 -2.71244138e-01 6.18005455e-01 -2.13354617e-01 -5.98080158e-01 5.65747142e-01 -7.52512336e-01 1.45155594e-01 1.37492144e+00 -7.23145381e-02 -2.06713364e-01 -3.52864325e-01 -8.01155686e-01 1.08370221e+00 4.60218459e-01 1.89426973e-01 5.40574074e-01 -9.48318422e-01 -1.51603192e-01 2.78750360e-01 -4.05580342e-01 4.01418447e-01 7.68191665e-02 3.90345395e-01 -6.47234559e-01 6.54769659e-01 -4.60733145e-01 -3.85975510e-01 -6.43871844e-01 1.17669272e+00 3.42962742e-01 -7.71832705e-01 -5.42417884e-01 6.34035528e-01 -8.92115459e-02 6.95640072e-02 6.82139993e-01 -6.29815519e-01 1.00781821e-01 -2.12065488e-01 3.10824692e-01 2.15947449e-01 -1.29205778e-01 2.26222247e-01 -2.64809132e-01 8.52735490e-02 -4.44320530e-01 -8.62406343e-02 1.51233542e+00 -5.55340461e-02 4.69444275e-01 1.05948307e-01 5.11778712e-01 -3.97376478e-01 -1.93462801e+00 9.48305354e-02 -3.17391366e-01 -4.88617271e-01 -1.17352106e-01 -8.71048391e-01 -9.07112837e-01 6.17378414e-01 1.60927415e-01 3.95610660e-01 8.50393713e-01 -2.64224321e-01 6.92860842e-01 5.77717483e-01 9.48067784e-01 -1.24672878e+00 2.79526681e-01 7.66215742e-01 6.99326932e-01 -1.09176195e+00 8.41223300e-02 -9.11493674e-02 -9.04237330e-01 1.03473830e+00 8.88366461e-01 -4.00531024e-01 4.01607871e-01 5.29208243e-01 -3.81525338e-01 2.13729113e-01 -1.11744523e+00 -2.24048913e-01 -3.45629126e-01 4.11717266e-01 -2.20703542e-01 2.32741654e-01 -8.96578431e-02 4.02356684e-01 -2.67457757e-02 -3.37520629e-01 6.13286853e-01 1.26713991e+00 -8.06066036e-01 -1.03129423e+00 -3.64462167e-01 2.19774410e-01 -1.41914859e-01 3.08777571e-01 2.13721126e-01 8.28387380e-01 -5.95134497e-02 8.64153624e-01 3.84060860e-01 1.46018296e-01 4.64458764e-01 -1.95466846e-01 6.46324933e-01 -8.66272092e-01 -3.27098221e-01 2.02215239e-01 3.46981846e-02 -7.66367972e-01 -1.85373664e-01 -4.87981588e-01 -1.30302572e+00 -4.24282700e-01 -2.85186976e-01 2.29884729e-01 5.12451589e-01 8.49795818e-01 3.35869640e-01 7.30749667e-01 8.47026885e-01 -1.44039094e+00 -9.42418396e-01 -4.33961689e-01 -5.79139829e-01 3.07541080e-02 5.54321527e-01 -9.12456632e-01 2.47537275e-03 -3.31341416e-01]
[4.926309108734131, 2.6476681232452393]
fb9535eb-c91d-4a9c-88e5-324aad4a9701
i-msv-2022-indic-multilingual-and-multi
2302.13209
null
https://arxiv.org/abs/2302.13209v1
https://arxiv.org/pdf/2302.13209v1.pdf
I-MSV 2022: Indic-Multilingual and Multi-sensor Speaker Verification Challenge
Speaker Verification (SV) is a task to verify the claimed identity of the claimant using his/her voice sample. Though there exists an ample amount of research in SV technologies, the development concerning a multilingual conversation is limited. In a country like India, almost all the speakers are polyglot in nature. Consequently, the development of a Multilingual SV (MSV) system on the data collected in the Indian scenario is more challenging. With this motivation, the Indic- Multilingual Speaker Verification (I-MSV) Challenge 2022 has been designed for understanding and comparing the state-of-the-art SV techniques. For the challenge, approximately $100$ hours of data spoken by $100$ speakers has been collected using $5$ different sensors in $13$ Indian languages. The data is divided into development, training, and testing sets and has been made publicly available for further research. The goal of this challenge is to make the SV system robust to language and sensor variations between enrollment and testing. In the challenge, participants were asked to develop the SV system in two scenarios, viz. constrained and unconstrained. The best system in the constrained and unconstrained scenario achieved a performance of $2.12\%$ and $0.26\%$ in terms of Equal Error Rate (EER), respectively.
['S. R. Mahadeva Prasanna', 'Mrinmoy Bhattacharjee', 'Jagabandhu Mishra']
2023-02-26
null
null
null
null
['speaker-verification']
['speech']
[-1.73006877e-01 -2.39047334e-02 5.02665080e-02 -7.16909051e-01 -1.35576069e+00 -7.66790390e-01 4.15829211e-01 -8.20766836e-02 -3.17753881e-01 6.75698876e-01 4.50002104e-02 -5.25049567e-01 6.07480049e-01 -1.33173645e-01 -5.17522693e-01 -3.48001808e-01 7.76576772e-02 1.52489990e-01 -1.72690749e-01 -3.07542920e-01 -2.00184602e-02 3.45033199e-01 -1.44605124e+00 -1.10182390e-02 7.19770610e-01 1.13833892e+00 4.05215612e-03 6.90788269e-01 -4.15980630e-02 4.08495903e-01 -9.29597080e-01 -6.47902787e-01 4.61760283e-01 -3.76917422e-01 -6.72770917e-01 -1.01518929e-01 6.88157916e-01 -5.42938970e-02 -9.57808271e-02 1.23144174e+00 1.10486925e+00 -6.09339215e-03 2.15377808e-02 -1.36011815e+00 -6.63249791e-01 8.52571666e-01 -2.37363264e-01 1.12451427e-01 8.61632884e-01 -1.78689640e-02 8.45297754e-01 -9.55848098e-01 5.61962485e-01 1.17590261e+00 7.97610939e-01 7.99899459e-01 -1.17338407e+00 -1.23175550e+00 -5.77810705e-02 1.80650249e-01 -1.84643269e+00 -1.11953032e+00 8.08595657e-01 -4.58243996e-01 7.55586028e-01 5.01267910e-01 2.64227837e-01 1.09075725e+00 -2.73281097e-01 6.73565209e-01 1.52703834e+00 -4.70309824e-01 6.64922073e-02 1.08560252e+00 3.63494784e-01 4.00928319e-01 -2.23919287e-01 1.88634962e-01 -8.91821384e-01 -1.37541160e-01 -1.46596774e-03 -7.06763327e-01 -2.89714307e-01 1.98830187e-01 -8.86982679e-01 9.28098917e-01 5.97938150e-02 2.97433108e-01 -1.09845087e-01 -6.13784134e-01 4.86956060e-01 5.59125960e-01 3.52293581e-01 5.48930690e-02 -2.93638825e-01 -4.39851671e-01 -1.06392753e+00 2.80365974e-01 9.38359082e-01 9.91232336e-01 4.19081450e-01 4.96115834e-01 6.04837388e-02 1.13902438e+00 7.49927580e-01 9.82425749e-01 5.76138854e-01 -5.65047264e-01 7.87815154e-01 3.47328216e-01 1.29123300e-01 -9.01929021e-01 8.11878592e-03 -8.96965042e-02 -5.48810184e-01 1.20428726e-01 3.32993180e-01 -4.72460687e-01 -6.87754929e-01 1.85014498e+00 3.16552460e-01 -1.95927285e-02 4.26592708e-01 7.77842283e-01 1.07402873e+00 5.46350002e-01 -2.42267355e-01 -2.90013313e-01 1.39460778e+00 -5.67689478e-01 -9.70065117e-01 -3.32541496e-01 1.39689669e-01 -1.17004824e+00 1.24064338e+00 1.19184040e-01 -9.41215634e-01 -5.74927986e-01 -1.17064035e+00 3.12628686e-01 -3.56064051e-01 -7.71117881e-02 1.15541257e-01 1.58773184e+00 -1.18797100e+00 -1.50858834e-01 -5.05842149e-01 -3.47666740e-01 1.01118416e-01 4.59532142e-01 -3.95887226e-01 -1.09093348e-02 -1.57095897e+00 8.35506141e-01 -3.41992855e-01 1.96670443e-01 -9.27723587e-01 -4.02982503e-01 -1.08151972e+00 -5.86219907e-01 -2.19241306e-01 3.54914546e-01 1.43292034e+00 -7.94174016e-01 -1.72138369e+00 1.06402743e+00 -4.82667327e-01 -4.96474981e-01 6.25744462e-01 1.11377485e-01 -1.26472080e+00 -3.80216211e-01 3.01345021e-01 1.73002616e-01 6.39035106e-01 -1.08525431e+00 -5.72057962e-01 -5.98535776e-01 -3.66031259e-01 3.39022949e-02 -4.34987098e-02 7.43961692e-01 -2.90734023e-01 -4.76588517e-01 8.92190933e-02 -1.20154154e+00 3.85365039e-01 -5.37087560e-01 -5.87486148e-01 -1.40026659e-01 9.99499440e-01 -1.27786088e+00 1.10699582e+00 -2.37050247e+00 -5.20108640e-01 2.42903918e-01 -3.34242612e-01 3.74461323e-01 1.38269007e-01 2.31155902e-01 9.87983122e-02 9.79799554e-02 -2.37699360e-01 -6.76612556e-01 2.69338727e-01 -1.91872105e-01 -1.78490281e-01 6.24110341e-01 -3.82809460e-01 4.86894697e-01 -2.01209739e-01 -4.27882642e-01 -4.70687933e-02 5.21609187e-01 -8.81008580e-02 3.44754487e-01 3.91646534e-01 3.98110390e-01 1.39379399e-02 1.15654945e+00 9.42128122e-01 4.76432234e-01 5.17976731e-02 2.04752937e-01 -3.08768660e-01 3.60502779e-01 -1.52432454e+00 1.38369548e+00 -3.44043374e-01 9.62548316e-01 8.37211967e-01 -6.71918511e-01 1.09035599e+00 8.43639195e-01 1.62734061e-01 -7.85444677e-01 1.48434862e-01 5.31213820e-01 -7.57989958e-02 -4.84649092e-01 5.80423892e-01 -2.47115895e-01 -4.02083308e-01 4.40549433e-01 -1.50087014e-01 -1.00272655e-01 -8.15119445e-02 -3.83974351e-02 4.83448774e-01 -4.70593274e-01 -1.16688758e-01 -3.31047535e-01 7.52677262e-01 -1.91182300e-01 8.06776345e-01 4.17862624e-01 -9.74445045e-01 3.64467621e-01 -2.00534672e-01 2.56170869e-01 -6.09690666e-01 -1.06467474e+00 -3.98951232e-01 8.52918208e-01 2.42011622e-02 1.15296310e-02 -9.40523326e-01 -3.74250799e-01 -3.32664885e-02 6.71501398e-01 -1.36808559e-01 3.26255977e-01 -4.09900159e-01 -3.51544857e-01 1.15940928e+00 2.06689179e-01 9.25679624e-01 -1.00379407e+00 -1.35035768e-01 1.58288792e-01 -5.33091545e-01 -1.34098113e+00 -9.09187794e-01 -3.00131500e-01 -1.19955033e-01 -7.78380990e-01 -6.51089370e-01 -1.03996015e+00 1.99820042e-01 6.39748797e-02 6.89638138e-01 -5.82567275e-01 1.32995592e-02 2.05102205e-01 -1.49991408e-01 -8.42015624e-01 -5.58788240e-01 -1.37352526e-01 6.79659247e-01 2.88725883e-01 7.83852935e-01 -1.50355414e-01 -2.00274333e-01 6.02190554e-01 -3.44483525e-01 -6.31087005e-01 1.64731920e-01 5.09533882e-01 2.10164174e-01 -2.44743973e-01 8.86258066e-01 -7.06078172e-01 9.47274864e-01 -2.06424251e-01 -7.84713984e-01 1.76698461e-01 -7.72399902e-01 -3.11874002e-01 1.77957490e-01 -3.79509270e-01 -8.91366005e-01 8.44924003e-02 -5.01864254e-01 -1.18579820e-01 -8.18228051e-02 5.49254596e-01 -5.70463479e-01 -2.03719601e-01 5.47168076e-01 4.27307487e-01 8.12648609e-02 -4.16370898e-01 3.47725227e-02 1.68303418e+00 6.90029979e-01 -2.70555705e-01 7.57216334e-01 -9.31853130e-02 -9.11684871e-01 -1.08397853e+00 -1.32332355e-01 -5.93116641e-01 -1.92033917e-01 -3.37814450e-01 7.99182117e-01 -1.16800964e+00 -8.43278885e-01 9.56265330e-01 -8.86385322e-01 -1.92325354e-01 2.43733842e-02 6.72325194e-01 -2.41316855e-02 2.37066522e-01 -5.20523965e-01 -1.23906767e+00 -5.58956921e-01 -1.62659216e+00 7.50189781e-01 1.45880342e-01 -4.46714044e-01 -7.43501484e-01 1.77670747e-01 1.13826191e+00 6.89328074e-01 4.01219539e-02 1.31144077e-01 -7.91232765e-01 1.30112525e-02 -6.00613952e-01 8.12310819e-03 6.82495475e-01 2.34405369e-01 -2.39182293e-01 -1.54327703e+00 -4.96474415e-01 1.97224602e-01 -3.80518019e-01 8.02754685e-02 1.32987827e-01 4.86634105e-01 -4.49634403e-01 1.52061999e-01 1.58115998e-01 1.04162943e+00 3.79660070e-01 3.80256891e-01 7.77853355e-02 4.75180894e-01 6.52069330e-01 5.28309107e-01 2.79849946e-01 7.19701231e-01 9.61210310e-01 -1.18203750e-02 2.69548774e-01 2.56999675e-02 -2.67175198e-01 9.80151474e-01 9.93857324e-01 2.61387169e-01 3.38745043e-02 -1.00600398e+00 6.01507545e-01 -1.15329599e+00 -1.01896620e+00 -4.37478796e-02 2.36446643e+00 1.06701493e+00 -1.31986469e-01 4.39311862e-01 4.22560573e-01 1.02549553e+00 1.79416567e-01 -4.04380739e-01 -6.94231510e-01 -2.43150458e-01 6.56308532e-02 4.30720478e-01 8.88391435e-01 -1.00151312e+00 7.72185922e-01 6.47173595e+00 4.80289161e-01 -1.59611714e+00 2.00101420e-01 5.62381208e-01 1.21064544e-01 1.76214930e-02 -3.73817772e-01 -1.21440697e+00 5.70545495e-01 1.56312454e+00 -2.04035148e-01 4.61255074e-01 8.76560152e-01 4.19311047e-01 1.16058916e-01 -8.63666892e-01 1.40400136e+00 4.74209160e-01 -1.01499593e+00 -5.87668896e-01 1.22269869e-01 5.75540423e-01 3.36778134e-01 2.19541296e-01 5.25603235e-01 3.49444717e-01 -1.04361022e+00 1.18283415e+00 -3.46543565e-02 1.11154461e+00 -6.89273417e-01 9.54629421e-01 4.11835521e-01 -1.37146735e+00 3.14761847e-02 2.06316337e-01 1.86447769e-01 3.52012128e-01 3.10650617e-01 -1.03663409e+00 4.28869158e-01 8.76073599e-01 9.11581814e-02 -3.43906075e-01 4.40781891e-01 -9.83288698e-03 8.02540481e-01 -2.89821774e-01 5.77843785e-02 -1.49881527e-01 -1.20438904e-01 5.49586236e-01 1.16312528e+00 4.19384301e-01 -5.60473762e-02 6.72542974e-02 5.52601933e-01 -3.72622997e-01 1.84949279e-01 -5.24056017e-01 8.00656062e-03 8.35932732e-01 8.78719509e-01 1.26334354e-01 -1.31678462e-01 -3.85991812e-01 8.64062428e-01 -1.57180294e-01 2.67068833e-01 -6.84166431e-01 -2.22732559e-01 7.58240521e-01 6.82445765e-02 -4.63646092e-02 -1.69915020e-01 -3.38804483e-01 -1.01734757e+00 3.46433222e-01 -1.43359649e+00 2.53883421e-01 -2.58760482e-01 -1.12208831e+00 9.75043356e-01 -3.59107047e-01 -1.07090342e+00 -4.85321611e-01 -2.01749548e-01 -4.76468831e-01 1.52665770e+00 -1.43139362e+00 -1.20813787e+00 -1.77021399e-01 8.47342670e-01 6.21255457e-01 -6.96613193e-01 1.13762355e+00 7.64301479e-01 -8.07727396e-01 1.34728622e+00 6.30097166e-02 3.80395412e-01 9.23054099e-01 -9.56782341e-01 3.00809264e-01 1.08107650e+00 7.53311962e-02 5.58321416e-01 7.84051895e-01 -5.04456460e-01 -1.61470807e+00 -1.00308049e+00 1.45539737e+00 -5.61925948e-01 4.16270435e-01 -7.68424273e-01 -5.97709656e-01 7.07309186e-01 3.55831027e-01 -7.82018434e-03 1.02211750e+00 3.91266048e-02 -4.49704260e-01 -3.00920904e-01 -1.71621382e+00 1.52291924e-01 4.02833045e-01 -1.21201038e+00 -3.53694588e-01 -1.21774331e-01 3.20146412e-01 -4.02665675e-01 -9.89238441e-01 1.26805753e-01 6.93452418e-01 -5.30818999e-01 7.18843341e-01 -6.09399527e-02 -3.98937106e-01 -3.91033798e-01 -8.17004621e-01 -1.17500436e+00 3.48882139e-01 -7.86391854e-01 2.16991112e-01 1.93627560e+00 9.32511628e-01 -9.93689001e-01 7.32709944e-01 1.09174979e+00 -6.22088499e-02 -1.64649650e-01 -1.27573359e+00 -9.42279458e-01 -4.33443636e-02 -7.69209504e-01 7.36271560e-01 1.16266179e+00 1.65683180e-01 2.69758135e-01 -4.02289867e-01 5.00693977e-01 5.93278348e-01 -5.17943092e-02 8.66826236e-01 -9.02852178e-01 -9.02811438e-02 -1.82654411e-01 -3.75176430e-01 -7.17445910e-01 1.64912775e-01 -7.95086026e-01 1.94889396e-01 -9.09327388e-01 -1.08471856e-01 -5.07215917e-01 -9.60762054e-02 2.57661432e-01 -4.95597646e-02 3.19773138e-01 3.51987749e-01 2.44251370e-01 4.49161604e-02 3.20356220e-01 3.79941404e-01 -4.67439234e-01 -3.15004766e-01 4.12534684e-01 -6.90299928e-01 2.45053247e-01 9.40497100e-01 -2.61026472e-01 -2.28768706e-01 -2.11845785e-01 -4.81222689e-01 2.62910575e-01 -2.71148346e-02 -1.03749776e+00 2.64932871e-01 3.95094939e-02 -2.02598184e-01 -5.94984710e-01 5.58843493e-01 -6.83242619e-01 2.45728076e-01 2.79164851e-01 -2.37526104e-01 2.41889954e-01 3.10963869e-01 1.20397739e-01 -6.27807856e-01 1.08446963e-01 8.55930567e-01 1.40362665e-01 -6.04711115e-01 1.51076227e-01 -2.61952013e-01 -4.42906376e-03 9.52655554e-01 -2.45737478e-01 -1.95615426e-01 -5.22934079e-01 -6.57089710e-01 2.26632968e-01 2.96417356e-01 5.68573356e-01 5.37064493e-01 -1.35268939e+00 -1.04998600e+00 4.67641920e-01 2.67109543e-01 -4.66564059e-01 2.11530864e-01 6.91282272e-01 -4.09838140e-01 6.25399351e-01 -1.47295650e-02 -4.66058493e-01 -1.88496995e+00 1.37359453e-02 4.04347837e-01 4.47946377e-02 2.95258639e-03 1.03891897e+00 -5.81932425e-01 -9.97310102e-01 4.06240106e-01 1.55452102e-01 -1.45030305e-01 1.76627830e-01 7.66755342e-01 3.54759306e-01 3.00833851e-01 -1.43026233e+00 -6.85083628e-01 2.19753861e-01 1.50452107e-01 -8.34902823e-01 9.15191591e-01 -3.19807738e-01 9.67868567e-02 6.99025810e-01 1.22219324e+00 5.70128500e-01 -7.72314966e-01 -3.10238987e-01 -1.41593471e-01 -4.11476254e-01 -1.11670338e-01 -7.94275403e-01 -9.98178959e-01 8.01581144e-01 1.17412257e+00 3.67021888e-01 5.93424559e-01 -1.49129614e-01 8.52352321e-01 -9.78224277e-02 4.02096868e-01 -1.28078365e+00 -5.97834408e-01 4.36461031e-01 9.83624339e-01 -1.64324772e+00 -4.56748575e-01 -2.77426690e-01 -9.03021395e-01 4.02783245e-01 4.26615119e-01 5.96012414e-01 8.16810906e-01 2.02649161e-01 9.01656926e-01 -1.58117842e-02 -9.42191184e-02 3.02067429e-01 6.92235976e-02 7.57505953e-01 6.79381967e-01 5.60995877e-01 -8.56064260e-02 7.47543812e-01 -8.13489974e-01 -3.00409019e-01 2.62880236e-01 9.03046906e-01 -1.51935592e-01 -1.23508036e+00 -8.06797922e-01 1.88865200e-01 -7.45159090e-01 -3.00485338e-03 -6.16247118e-01 5.57587028e-01 3.56145054e-02 1.70164788e+00 -3.97035629e-01 -6.18110061e-01 5.62905192e-01 3.26153547e-01 -1.82285219e-01 -4.35446620e-01 -8.54852378e-01 -2.58664668e-01 5.31135321e-01 -2.93011427e-01 -4.30869758e-01 -1.19065833e+00 -1.03545117e+00 -6.87794447e-01 -3.10862243e-01 4.87217695e-01 1.22857487e+00 7.80024409e-01 2.46994734e-01 6.05036616e-02 1.00818443e+00 -5.31113327e-01 -7.67935395e-01 -1.28767025e+00 -7.12348938e-01 3.58009458e-01 3.57268214e-01 -2.60396928e-01 -4.82020706e-01 3.60618643e-02]
[14.263619422912598, 6.204905986785889]
4cb0ffca-8aa2-4bc7-b9ec-6e0428969c7d
exploring-the-representation-power-of-splade
2306.16680
null
https://arxiv.org/abs/2306.16680v1
https://arxiv.org/pdf/2306.16680v1.pdf
Exploring the Representation Power of SPLADE Models
The SPLADE (SParse Lexical AnD Expansion) model is a highly effective approach to learned sparse retrieval, where documents are represented by term impact scores derived from large language models. During training, SPLADE applies regularization to ensure postings lists are kept sparse -- with the aim of mimicking the properties of natural term distributions -- allowing efficient and effective lexical matching and ranking. However, we hypothesize that SPLADE may encode additional signals into common postings lists to further improve effectiveness. To explore this idea, we perform a number of empirical analyses where we re-train SPLADE with different, controlled vocabularies and measure how effective it is at ranking passages. Our findings suggest that SPLADE can effectively encode useful ranking signals in documents even when the vocabulary is constrained to terms that are not traditionally useful for ranking, such as stopwords or even random words.
['Guido Zuccon', 'Shengyao Zhuang', 'Joel Mackenzie']
2023-06-29
null
null
null
null
['retrieval']
['methodology']
[ 8.07416886e-02 -1.89679682e-01 -7.59213746e-01 3.64724211e-02 -1.18971229e+00 -6.38700068e-01 8.02454710e-01 6.13135993e-01 -5.03574848e-01 6.17120326e-01 1.09596527e+00 -5.39818257e-02 -4.88655657e-01 -7.66083300e-01 -7.65896201e-01 -2.64145464e-01 -2.59075552e-01 5.44375777e-01 -5.08746952e-02 -3.07507426e-01 4.02781963e-01 2.18098328e-01 -1.59594440e+00 5.81810415e-01 5.11105299e-01 4.34961498e-01 3.77497703e-01 2.79710650e-01 -3.47572893e-01 9.02217507e-01 -5.87024510e-01 -2.26144135e-01 2.91083664e-01 -3.21627110e-01 -4.60129946e-01 -5.47394216e-01 6.58283949e-01 -4.39192563e-01 -7.16077268e-01 9.23973203e-01 3.29629600e-01 5.09200513e-01 9.51264679e-01 -3.63634199e-01 -7.78618157e-01 1.19125700e+00 -1.98080361e-01 4.22639370e-01 5.92235208e-01 -3.92367765e-02 1.81937551e+00 -8.30277026e-01 8.71345580e-01 1.31027877e+00 6.74253464e-01 1.60427555e-01 -1.36508119e+00 -4.86580789e-01 1.71101615e-01 -2.45462880e-01 -1.50764298e+00 -5.80243587e-01 5.41606426e-01 -3.46658319e-01 1.18912041e+00 4.40247595e-01 4.96084839e-01 1.13280272e+00 2.56709486e-01 6.50681794e-01 6.90169036e-01 -6.31894112e-01 2.88035274e-02 2.74462432e-01 4.86888647e-01 4.21398401e-01 6.06681883e-01 1.92579880e-01 -8.67940605e-01 -6.03708565e-01 4.22023743e-01 6.26783893e-02 -2.55355567e-01 -1.45408615e-01 -9.80595648e-01 1.17281783e+00 3.76933455e-01 7.24023998e-01 -6.11200809e-01 1.73459560e-01 4.01433229e-01 4.73478973e-01 3.07580799e-01 1.11418307e+00 -4.52452004e-01 1.15641877e-01 -1.32027102e+00 4.73655492e-01 7.77090967e-01 7.75161505e-01 7.52858102e-01 -6.51038066e-02 -7.29198396e-01 1.00864232e+00 3.28366458e-01 6.36455178e-01 8.15790176e-01 -7.33784735e-01 3.54841590e-01 2.37247482e-01 -5.81602678e-02 -1.22168899e+00 -8.74685273e-02 -6.40250742e-01 -5.21477163e-01 -7.08680689e-01 -7.81583786e-03 2.70314276e-01 -7.37308264e-01 1.73932469e+00 -5.66027641e-01 -1.63275585e-01 -2.07355455e-01 6.77931964e-01 5.87279499e-01 8.59252751e-01 3.27706397e-01 -4.48762089e-01 1.23411322e+00 -5.77149451e-01 -7.26836562e-01 -5.11776149e-01 8.94512534e-01 -6.48063123e-01 1.54905796e+00 2.46454254e-01 -1.10767829e+00 -2.12897986e-01 -7.95769930e-01 -2.19441891e-01 -3.82316232e-01 -3.08748931e-02 8.55137408e-01 2.03228503e-01 -9.96596038e-01 4.41911221e-01 -3.10422987e-01 -8.96626860e-02 2.58947670e-01 -3.76856737e-02 -1.47348374e-01 -3.73331577e-01 -1.62229407e+00 9.00711477e-01 1.34694725e-01 -5.39688706e-01 -1.20239723e+00 -8.32618773e-01 -8.79096150e-01 4.68448579e-01 2.08326370e-01 -7.52662778e-01 9.96780753e-01 -6.17579162e-01 -7.83859074e-01 6.05474234e-01 -3.62185508e-01 -4.98916179e-01 -4.76098180e-01 -2.93815225e-01 -1.49601355e-01 2.80662090e-01 2.02732906e-01 3.45449626e-01 7.85497844e-01 -9.99237776e-01 1.42521011e-02 -2.51840413e-01 -5.09278364e-02 2.32701257e-01 -8.15149248e-01 1.52258292e-01 -4.77930754e-01 -1.03310144e+00 -1.63204879e-01 -7.05556571e-01 -2.26249516e-01 -5.11296213e-01 4.00652699e-02 -3.05419981e-01 -4.26282883e-02 -6.42160058e-01 1.97543848e+00 -2.04317069e+00 1.77557915e-01 8.15042257e-01 3.02730680e-01 4.47461307e-02 -6.24652684e-01 6.97161019e-01 3.31732959e-01 3.28532696e-01 2.02337682e-01 -7.69326165e-02 7.67448843e-02 -8.13979132e-04 -7.16500998e-01 1.99871659e-01 -2.41761506e-01 1.09251595e+00 -9.50013161e-01 -4.93371159e-01 -2.47695208e-01 4.48811263e-01 -7.95282543e-01 -3.77940349e-02 -5.07626355e-01 -3.46679986e-01 -6.23387218e-01 5.07978439e-01 2.86217593e-02 -4.05523926e-01 1.40009150e-01 2.51177717e-02 2.23320812e-01 9.72925186e-01 -7.21433043e-01 1.60246611e+00 -6.46628082e-01 8.71621311e-01 -1.31084681e-01 -7.24837184e-01 5.59579849e-01 1.29044086e-01 3.17663908e-01 -8.23816836e-01 -7.77256265e-02 1.67892307e-01 -2.17015579e-01 -3.10712367e-01 8.28467488e-01 -3.05370808e-01 -1.43449888e-01 5.14436960e-01 7.13908747e-02 -2.71068096e-01 4.92023408e-01 7.54722536e-01 1.36268640e+00 -6.42754078e-01 8.96834135e-02 -3.03410321e-01 1.52866423e-01 6.05012886e-02 -5.01122326e-02 1.24335420e+00 4.67627227e-01 4.28551555e-01 3.81081909e-01 1.97824001e-01 -9.98473227e-01 -9.58153188e-01 -1.59925073e-01 1.51249397e+00 -2.44273990e-01 -1.10876203e+00 -1.28166437e-01 -3.72975945e-01 3.53770554e-01 9.01288867e-01 -4.77669865e-01 -5.15801132e-01 -3.77414197e-01 -2.74791062e-01 5.86236894e-01 4.22617227e-01 -3.93514037e-01 -1.12579012e+00 5.60027659e-02 1.98629759e-02 -1.39093488e-01 -5.38929701e-01 -7.77369380e-01 5.38640141e-01 -8.89509976e-01 -5.61222792e-01 -7.27246940e-01 -6.01046860e-01 6.61089957e-01 3.91057521e-01 1.65533745e+00 3.23964626e-01 -9.72098559e-02 6.99984968e-01 -4.83202934e-01 -1.14998937e-01 -1.69294700e-01 3.50789040e-01 -2.75340751e-02 -6.90550685e-01 7.18451262e-01 -4.28455025e-01 -4.59605962e-01 -1.55114204e-01 -1.18672442e+00 -7.03871012e-01 8.30204725e-01 9.87003803e-01 5.79477072e-01 -9.49214026e-02 5.62305987e-01 -9.49528217e-01 1.38168001e+00 -7.62107074e-01 -5.03070235e-01 2.55818993e-01 -7.63635635e-01 3.66189599e-01 3.82464617e-01 -6.47629201e-01 -4.93276864e-01 -6.16765857e-01 -8.59830063e-03 -4.23046440e-01 3.34704250e-01 1.29185748e+00 4.78986919e-01 1.65204927e-02 1.00384474e+00 3.58896762e-01 -1.13216370e-01 -4.25047010e-01 4.53521430e-01 6.00708604e-01 -2.05250651e-01 -7.41317689e-01 7.35912859e-01 5.04306667e-02 -2.86399782e-01 -9.68577147e-01 -1.10568833e+00 -9.13677394e-01 1.25591189e-01 1.11880936e-01 3.60891193e-01 -1.24763811e+00 -1.10950537e-01 -4.51454133e-01 -7.20083058e-01 -2.19625577e-01 -5.97696185e-01 6.76412940e-01 -2.02398419e-01 2.52284169e-01 -9.20759261e-01 -5.66566169e-01 -2.84337580e-01 -7.39640594e-01 1.14006281e+00 -2.50128001e-01 -6.98701143e-01 -9.65694666e-01 3.82366657e-01 1.05668232e-02 7.97960877e-01 -5.64006031e-01 1.25592005e+00 -9.66141284e-01 -5.50830305e-01 -5.20845294e-01 2.57285666e-02 2.12339118e-01 -5.87152056e-02 -4.40901369e-01 -6.95122778e-01 -4.65231031e-01 -1.29651949e-01 -6.09040797e-01 1.40029871e+00 5.72558999e-01 9.51917589e-01 -5.54613769e-01 -4.46319491e-01 2.51955986e-01 1.33967960e+00 -2.60293275e-01 5.31082034e-01 2.17299193e-01 2.05194533e-01 5.46872616e-01 3.58726680e-01 4.77715999e-01 -1.04568794e-01 5.84199309e-01 -1.72542617e-01 3.61269265e-01 -4.70783375e-02 -7.44120538e-01 4.62230265e-01 1.11015093e+00 4.62702125e-01 -2.94521451e-01 -6.85871065e-01 7.45406806e-01 -1.19874132e+00 -1.08576095e+00 3.35771829e-01 2.32494307e+00 1.24677444e+00 2.85823345e-01 -1.01508886e-01 -2.23428473e-01 1.79600134e-01 5.20640433e-01 1.99967921e-02 -9.72508118e-02 -3.64997149e-01 4.10881221e-01 4.31050807e-01 7.66624033e-01 -5.48670650e-01 9.22092974e-01 7.56639481e+00 1.04452467e+00 -7.29396105e-01 -4.68586721e-02 2.43273601e-01 -5.31594813e-01 -1.02172422e+00 -1.12327524e-02 -9.62070942e-01 3.03746164e-01 1.07712495e+00 -3.62298667e-01 6.51848078e-01 7.61117160e-01 6.07358851e-02 7.94920027e-02 -1.08071077e+00 5.67199290e-01 3.06605816e-01 -1.41589701e+00 5.43199658e-01 4.71906178e-02 8.46919537e-01 1.36347547e-01 3.47109437e-01 8.03361475e-01 3.52358907e-01 -1.16655910e+00 4.09134269e-01 5.81377625e-01 8.22086096e-01 -4.03550386e-01 5.88665605e-01 4.39849138e-01 -8.74786556e-01 -2.05823570e-01 -8.30242395e-01 4.76269983e-02 -3.88062969e-02 6.73268259e-01 -5.89581072e-01 2.15189792e-02 3.58275026e-01 6.38340294e-01 -5.62358141e-01 9.83550489e-01 -6.45308271e-02 9.11938071e-01 -2.97876477e-01 -3.08165342e-01 3.66112560e-01 6.18003421e-02 7.18548417e-01 1.31952596e+00 3.27630967e-01 -4.86569591e-02 -2.55213231e-02 6.32506728e-01 -3.79722118e-01 3.74040604e-01 -1.08146727e+00 -5.63639760e-01 7.18052268e-01 7.00302243e-01 -2.15195999e-01 -3.59670281e-01 -4.52008069e-01 4.74901259e-01 4.06475931e-01 6.72697246e-01 -3.37805331e-01 -2.67597169e-01 3.82618695e-01 4.65300173e-01 2.37826139e-01 -1.81223005e-01 -6.35698363e-02 -1.33832443e+00 -1.68395981e-01 -1.20644796e+00 3.58064115e-01 -8.08032095e-01 -1.47743309e+00 4.05877143e-01 2.40737915e-01 -8.67710769e-01 -3.35844517e-01 -3.45463991e-01 -5.65539338e-02 9.00101602e-01 -1.44363248e+00 -7.57200897e-01 2.10817665e-01 5.08076370e-01 6.55125380e-01 -3.84963721e-01 8.14241827e-01 3.46996963e-01 1.25924582e-02 7.23104894e-01 4.28010851e-01 -2.16895640e-01 7.53057718e-01 -9.18607771e-01 -2.41884351e-01 3.50166023e-01 7.79439211e-01 1.63975227e+00 6.88682079e-01 -8.63693178e-01 -1.39908659e+00 -8.41193199e-01 1.18682432e+00 -6.02020860e-01 7.38771498e-01 -1.67576969e-01 -9.08068657e-01 7.25415528e-01 1.55769885e-01 -5.46117842e-01 8.42039526e-01 6.41129553e-01 -6.51215494e-01 -6.83878213e-02 -6.62798882e-01 6.64038420e-01 8.57183337e-01 -9.32375193e-01 -1.10655391e+00 5.49759686e-01 9.58074450e-01 9.87718403e-02 -5.67283213e-01 2.47850493e-01 4.16469693e-01 -2.89189756e-01 1.22655702e+00 -7.80874133e-01 5.56173563e-01 9.19107050e-02 -5.11406422e-01 -1.35940146e+00 -6.21590793e-01 -4.84587014e-01 -3.78130913e-01 9.21080589e-01 5.61443329e-01 -2.99395472e-01 5.52264810e-01 2.29403853e-01 6.01575933e-02 -4.97541726e-01 -4.84709799e-01 -7.04687178e-01 2.05118924e-01 -4.32162315e-01 3.13615292e-01 8.14424813e-01 3.13250542e-01 6.21174216e-01 -2.78578252e-01 -2.51163840e-01 3.02715600e-01 -1.94877431e-01 2.85913706e-01 -1.31650043e+00 -3.72578532e-01 -6.38454318e-01 6.48724660e-02 -1.45284390e+00 5.12359679e-01 -1.29235256e+00 9.59725156e-02 -1.33221209e+00 4.89058554e-01 -4.50609803e-01 -6.18937254e-01 2.28490934e-01 -1.62882984e-01 1.46159455e-02 -3.99699174e-02 4.33589607e-01 -7.48824298e-01 5.89498699e-01 1.00470543e+00 -3.74678493e-01 -2.23692596e-01 -1.74471989e-01 -1.08634484e+00 3.58821899e-01 3.09871912e-01 -5.95107377e-01 -6.62438452e-01 -5.33283651e-01 9.42395866e-01 5.46504892e-02 2.16508098e-02 -4.34323758e-01 3.10440868e-01 -5.56952730e-02 1.06807351e-01 -4.66161311e-01 1.44729719e-01 -8.48906219e-01 -1.21182539e-01 1.94459006e-01 -1.26673448e+00 6.23661131e-02 2.92848289e-01 6.95887268e-01 -4.26995933e-01 -6.09047949e-01 2.49877423e-01 -1.85701922e-01 -3.05442303e-01 9.38622206e-02 -6.04071617e-01 2.73364604e-01 3.46055895e-01 3.00217360e-01 -1.99003011e-01 -6.86957836e-01 -3.78283232e-01 6.78449720e-02 3.79018694e-01 2.73856491e-01 7.74823368e-01 -1.41173255e+00 -7.39809453e-01 2.46617183e-01 3.78436387e-01 -4.65980053e-01 -2.35596791e-01 4.23468351e-01 -1.83004320e-01 1.11237800e+00 4.82746869e-01 -9.88931283e-02 -8.80323887e-01 5.95614254e-01 -2.26055775e-02 -7.11833775e-01 -5.20891666e-01 9.52207685e-01 1.98668256e-01 2.52159350e-02 5.32975018e-01 -3.45159978e-01 -3.03433567e-01 2.56879061e-01 5.86866260e-01 -3.20198312e-02 8.56404528e-02 -2.81125754e-01 -1.74967363e-01 4.59618628e-01 -3.75236481e-01 -2.24073380e-01 1.34917521e+00 -3.76385599e-02 -3.32137644e-01 4.49075758e-01 1.27886581e+00 7.79953778e-01 -3.73504609e-01 -4.75429595e-01 2.57601500e-01 -5.59147477e-01 4.42984521e-01 -7.35905170e-01 -4.88344193e-01 4.67043370e-01 4.42258939e-02 2.08585218e-01 7.91507185e-01 2.09052220e-01 6.90361202e-01 7.49251425e-01 3.97402167e-01 -8.19004953e-01 3.74943167e-01 8.02830100e-01 9.28542674e-01 -8.66910696e-01 1.90297157e-01 1.78523600e-01 -4.35565978e-01 5.30652404e-01 2.27395311e-01 -1.83918953e-01 6.80599988e-01 2.96673905e-02 -3.34773183e-01 -4.72459584e-01 -1.13592982e+00 -1.83068126e-01 7.48784840e-01 2.58485228e-01 7.85704851e-01 -2.67771840e-01 -6.36780679e-01 6.75617576e-01 -1.99213088e-01 -2.55592942e-01 1.51978895e-01 6.45180106e-01 -6.18129730e-01 -8.10494304e-01 -2.04924956e-01 1.21015406e+00 -6.93363547e-01 -8.83835912e-01 -3.79462749e-01 3.79097462e-01 -4.82212543e-01 7.49130845e-01 -8.71431753e-02 -2.73885489e-01 8.52728933e-02 2.63883591e-01 1.93634838e-01 -1.15455115e+00 -7.15130150e-01 2.31671944e-01 1.15531556e-01 -5.42656422e-01 -3.24082039e-02 -4.98653173e-01 -6.54316485e-01 -6.81971535e-02 -6.65331542e-01 6.89198852e-01 3.16062093e-01 6.12366796e-01 2.75981247e-01 5.24173558e-01 4.04525876e-01 -3.29390109e-01 -1.00193954e+00 -1.17936254e+00 -7.29234993e-01 4.79092419e-01 4.12037462e-01 -4.52960223e-01 -7.72745430e-01 -3.10957849e-01]
[11.471431732177734, 7.630578517913818]
b260227f-8af6-4eb0-a108-3e9fa3173b19
leveraging-virtual-and-real-person-for
1811.02074
null
http://arxiv.org/abs/1811.02074v1
http://arxiv.org/pdf/1811.02074v1.pdf
Leveraging Virtual and Real Person for Unsupervised Person Re-identification
Person re-identification (re-ID) is a challenging problem especially when no labels are available for training. Although recent deep re-ID methods have achieved great improvement, it is still difficult to optimize deep re-ID model without annotations in training data. To address this problem, this study introduces a novel approach for unsupervised person re-ID by leveraging virtual and real data. Our approach includes two components: virtual person generation and training of deep re-ID model. For virtual person generation, we learn a person generation model and a camera style transfer model using unlabeled real data to generate virtual persons with different poses and camera styles. The virtual data is formed as labeled training data, enabling subsequently training deep re-ID model in supervision. For training of deep re-ID model, we divide it into three steps: 1) pre-training a coarse re-ID model by using virtual data; 2) collaborative filtering based positive pair mining from the real data; and 3) fine-tuning of the coarse re-ID model by leveraging the mined positive pairs and virtual data. The final re-ID model is achieved by iterating between step 2 and step 3 until convergence. Experimental results on two large-scale datasets, Market-1501 and DukeMTMC-reID, demonstrate the effectiveness of our approach and shows that the state of the art is achieved in unsupervised person re-ID.
['Fengxiang Yang', 'Zhun Zhong', 'Shaozi Li', 'Zhiming Luo', 'Sheng Lian']
2018-11-05
null
null
null
null
['unsupervised-person-re-identification']
['computer-vision']
[-9.79482159e-02 4.17850502e-02 2.36214444e-01 -6.11637235e-01 -4.71619189e-01 -4.29463118e-01 8.38313520e-01 -8.68856907e-02 -7.42863894e-01 7.19344258e-01 4.08211827e-01 3.30606014e-01 2.30613306e-01 -7.95445561e-01 -6.75075650e-01 -3.26772660e-01 1.92887083e-01 1.21107602e+00 -9.25028101e-02 -1.30916104e-01 -9.58848521e-02 2.97166228e-01 -1.66229880e+00 2.36980170e-02 1.06583631e+00 4.13216770e-01 4.53831330e-02 6.86731637e-01 6.96661994e-02 1.39065027e-01 -7.39731789e-01 -8.92398715e-01 7.43353784e-01 -5.40154397e-01 -8.60033274e-01 1.44681528e-01 3.74969393e-01 -5.36411464e-01 -8.95088911e-02 9.49237168e-01 9.78877723e-01 5.25762916e-01 7.42236197e-01 -1.25271130e+00 -9.62532640e-01 5.40360391e-01 -6.58548951e-01 -7.30322301e-02 5.28334081e-01 4.62735593e-02 5.73781312e-01 -1.10429502e+00 6.05934203e-01 1.36595559e+00 8.27670932e-01 1.26637518e+00 -1.23457718e+00 -9.66324806e-01 1.93377972e-01 1.10639408e-01 -1.78173423e+00 -2.92271644e-01 4.92008299e-01 -6.77422106e-01 7.67384410e-01 -1.62356004e-01 8.24584067e-01 1.26941001e+00 -5.06722867e-01 7.81850517e-01 7.52596021e-01 -5.07297695e-01 -2.69902050e-02 4.84111935e-01 2.61294544e-01 4.34474289e-01 2.53071457e-01 2.67280191e-01 -4.28106010e-01 -2.13358045e-01 6.70390010e-01 1.07597008e-01 -6.97311312e-02 -8.95980895e-02 -1.10715437e+00 7.28640378e-01 3.01116049e-01 -2.76283603e-02 -2.47860685e-01 -3.52165431e-01 3.21968198e-01 1.96028605e-01 3.95804971e-01 2.56015569e-01 -1.57019734e-01 2.49486476e-01 -8.02635968e-01 6.34250820e-01 6.95987403e-01 9.88419235e-01 6.88551545e-01 -3.73604357e-01 -2.65779972e-01 1.21256316e+00 3.30214322e-01 6.08767033e-01 9.16715145e-01 -3.44094664e-01 6.03585660e-01 6.84037447e-01 4.71644253e-01 -7.12848663e-01 -3.31839114e-01 -2.77619213e-01 -9.95084286e-01 3.43426168e-02 4.97087508e-01 -4.23468530e-01 -1.09946132e+00 1.83786035e+00 6.40411496e-01 5.17241478e-01 1.32848680e-01 9.69218791e-01 1.04136145e+00 2.84143537e-01 1.65197462e-01 1.88681632e-01 1.25164425e+00 -1.10730374e+00 -3.55516821e-01 -1.95535317e-01 5.91571569e-01 -3.53642464e-01 6.66190386e-01 1.69214472e-01 -8.85563791e-01 -1.16572666e+00 -9.91326749e-01 6.77131787e-02 -4.51352417e-01 4.85130250e-01 2.12926552e-01 8.65915000e-01 -1.09655547e+00 4.88500923e-01 -3.48884732e-01 -5.42927563e-01 4.52028722e-01 7.87618995e-01 -7.13671625e-01 -5.26305698e-02 -1.28768444e+00 5.42438567e-01 6.15138173e-01 2.25260004e-01 -8.86047781e-01 -7.13464916e-01 -7.73201585e-01 -1.83707163e-01 1.10003792e-01 -1.13312721e+00 1.06118882e+00 -1.15392566e+00 -1.46587157e+00 1.21007991e+00 -1.96801692e-01 -4.04209286e-01 1.02476275e+00 -3.05472344e-01 -5.24616241e-01 -3.88055742e-01 4.30571079e-01 7.67386138e-01 8.23249400e-01 -1.61747992e+00 -8.57300937e-01 -6.41042590e-01 -1.44483656e-01 4.24436539e-01 -4.35789883e-01 8.15380551e-03 -7.77722716e-01 -6.24903262e-01 -3.04406613e-01 -1.18169963e+00 -2.22536802e-01 -5.48555732e-01 -5.92158675e-01 -5.56424379e-01 2.67169207e-01 -8.39187384e-01 8.66457760e-01 -1.72125828e+00 6.10411912e-02 4.38665479e-01 3.58221561e-01 6.49993896e-01 -3.05566162e-01 1.20848469e-01 -3.29754919e-01 8.80882218e-02 1.41757995e-01 -1.11112869e+00 -2.74416637e-02 -2.11494401e-01 3.41848210e-02 3.35643798e-01 -1.18867807e-01 9.06507015e-01 -1.01617455e+00 -5.05677581e-01 3.15079719e-01 4.30962920e-01 -4.84876812e-01 5.19674182e-01 3.16176742e-01 6.31590068e-01 -1.17899656e-01 4.54579562e-01 7.89675653e-01 -1.90254316e-01 1.57082006e-01 -1.90300643e-01 1.21408068e-01 -3.46035182e-01 -1.57912493e+00 1.37198174e+00 -1.08934201e-01 1.21730983e-01 -3.18240345e-01 -7.67129958e-01 1.05465543e+00 1.71638012e-01 4.98336673e-01 -5.87652564e-01 2.22610459e-01 -4.52945381e-02 -3.34100395e-01 -2.89196461e-01 8.16567183e-01 -1.88843198e-02 -7.09651336e-02 6.89527988e-01 3.61984491e-01 6.88906968e-01 1.29379302e-01 2.56546229e-01 4.31849360e-01 1.75386503e-01 8.58726129e-02 -1.19928196e-02 7.69486070e-01 -1.60026953e-01 6.82906806e-01 9.52093482e-01 -3.31639528e-01 7.56529570e-01 -3.14115435e-01 -4.78075624e-01 -1.16193736e+00 -1.05081320e+00 1.56267777e-01 1.07392156e+00 4.47618157e-01 -4.32256430e-01 -1.01021159e+00 -1.00266421e+00 1.29237965e-01 3.13870043e-01 -1.00363338e+00 -5.14979241e-03 -5.91738999e-01 -1.12637365e+00 5.30021548e-01 6.73740387e-01 8.71867120e-01 -8.27116787e-01 2.96656400e-01 7.26270452e-02 -4.56002086e-01 -9.43410873e-01 -6.87997758e-01 -5.06153286e-01 -3.91740710e-01 -1.10316169e+00 -1.16679239e+00 -8.62248659e-01 9.97200072e-01 4.75464672e-01 9.98655200e-01 3.42926919e-01 -3.78220007e-02 3.74568909e-01 -3.91672432e-01 -5.46979845e-01 -3.65308464e-01 6.45077676e-02 6.53762281e-01 3.98467302e-01 7.51967430e-01 -2.32448503e-01 -7.05505371e-01 5.87618947e-01 -3.62825215e-01 1.46385998e-01 3.52834433e-01 9.67712700e-01 6.24056697e-01 4.16959338e-02 7.34642386e-01 -9.73231971e-01 6.25611722e-01 -4.07466024e-01 -3.18840772e-01 3.51126909e-01 -6.54060185e-01 -1.30079508e-01 5.39817989e-01 -6.44558549e-01 -1.41330707e+00 3.19274455e-01 -2.38434598e-01 -3.44685376e-01 -3.34614575e-01 6.03042915e-03 -3.79887104e-01 1.49965972e-01 8.50423157e-01 7.89430663e-02 -1.15197502e-01 -5.15566170e-01 2.15621367e-01 8.71174216e-01 9.18041885e-01 -4.60741639e-01 1.10080528e+00 5.89520097e-01 -5.81873059e-01 -4.48038042e-01 -6.56955183e-01 -6.47973657e-01 -1.09745240e+00 -3.63800585e-01 1.10821438e+00 -1.32185018e+00 -7.88045168e-01 8.27673793e-01 -9.35127616e-01 -3.38877320e-01 -2.48891652e-01 4.39186931e-01 -4.94653806e-02 5.48386931e-01 -4.46414381e-01 -7.48192966e-01 -5.88313699e-01 -7.92952299e-01 9.86313283e-01 5.89003086e-01 -2.87699819e-01 -9.75051343e-01 3.85049403e-01 6.10718071e-01 -1.44877344e-01 1.00471579e-01 1.22854300e-01 -1.04507852e+00 -1.65556699e-01 -5.39304614e-01 -2.28787303e-01 1.90778077e-01 1.55673325e-01 -5.03277600e-01 -1.05157924e+00 -4.34493721e-01 -6.72636449e-01 -2.90062636e-01 7.07376778e-01 5.89588396e-02 9.15154994e-01 -2.51682778e-03 -7.43921399e-01 7.84100056e-01 1.22236252e+00 -1.03772417e-01 4.67517614e-01 3.98438424e-01 1.15420270e+00 7.58434892e-01 6.59397185e-01 4.92532730e-01 8.91657114e-01 8.51720452e-01 1.50502762e-02 -2.80770481e-01 -3.90455686e-02 -7.34837115e-01 9.01657045e-02 2.67046332e-01 -6.77537620e-01 -2.94170022e-01 -8.47842515e-01 6.98233187e-01 -2.08776402e+00 -1.19825077e+00 -6.31367341e-02 2.46022964e+00 5.39527655e-01 -2.74686456e-01 7.28570521e-01 -9.28324014e-02 1.16460228e+00 -5.14303625e-01 -6.44972324e-01 2.10736156e-01 4.74783257e-02 -1.52892649e-01 4.22377318e-01 4.94123518e-01 -1.16258812e+00 1.14192617e+00 5.50567150e+00 4.71158981e-01 -6.95864439e-01 1.27332374e-01 4.98941720e-01 5.24804592e-02 5.44718979e-03 -3.79430801e-01 -1.50564313e+00 6.59457505e-01 5.91856718e-01 -1.12763315e-01 3.67679358e-01 7.58637786e-01 1.55613646e-01 1.23050056e-01 -1.23179138e+00 1.62488580e+00 4.29183900e-01 -1.05016720e+00 2.47420177e-01 2.12771192e-01 9.39295113e-01 -2.41089061e-01 -3.21229786e-01 4.89004135e-01 9.48799312e-01 -9.47592199e-01 7.30875611e-01 6.27487540e-01 6.37014270e-01 -9.83171403e-01 9.95716929e-01 4.43154901e-01 -1.31520951e+00 -1.92367107e-01 -3.52105975e-01 3.31145763e-01 3.50506842e-01 2.70579785e-01 -1.00036979e+00 7.96948493e-01 1.06358826e+00 8.01970661e-01 -7.70536721e-01 8.08514416e-01 -8.54080170e-02 8.42234716e-02 -9.27482918e-02 3.13298881e-01 -5.25085330e-01 -2.54477739e-01 1.90199450e-01 1.21190190e+00 1.80236742e-01 1.06425643e-01 3.21491003e-01 7.32608795e-01 -1.38149902e-01 -7.43200257e-02 -2.89792806e-01 3.56538266e-01 5.35620928e-01 1.21642923e+00 -4.91360456e-01 -6.22629285e-01 -3.09560627e-01 1.55786848e+00 5.41366339e-01 3.61323863e-01 -7.87097573e-01 -3.27177905e-02 6.91720009e-01 1.19972415e-01 -2.64710803e-02 -3.09708100e-02 -1.49349859e-02 -1.31032574e+00 -2.49453992e-01 -6.72367275e-01 7.01223135e-01 -5.39222360e-01 -1.73782444e+00 7.34611809e-01 1.57646373e-01 -1.29199028e+00 -6.61904931e-01 -3.62698555e-01 -4.14011031e-01 1.33568048e+00 -1.10844481e+00 -1.55595624e+00 -8.01550448e-01 1.00251150e+00 3.09238374e-01 -5.56091011e-01 7.30059743e-01 4.24854457e-01 -8.32908809e-01 1.25696611e+00 -3.15260850e-02 6.45170987e-01 9.98506606e-01 -1.27385354e+00 8.18867981e-01 1.05567098e+00 -1.00450493e-01 7.69072711e-01 3.52384090e-01 -1.09368360e+00 -8.71765435e-01 -1.32538879e+00 9.20519471e-01 -7.03180194e-01 2.15981267e-02 -6.63065493e-01 -6.53708577e-01 9.07770216e-01 -2.89009631e-01 -2.17198625e-01 1.10252130e+00 2.36144528e-01 -1.29533783e-01 -3.95123847e-04 -1.36769116e+00 5.89643598e-01 1.52892888e+00 -3.71772051e-01 -4.11228746e-01 3.19842875e-01 2.68395811e-01 -3.67360711e-01 -8.29089642e-01 1.81182846e-01 6.89954281e-01 -7.55464554e-01 1.40292847e+00 -4.99797910e-01 2.07091384e-02 -4.39162970e-01 2.39276722e-01 -1.53940856e+00 -5.44636726e-01 -4.05755997e-01 1.27822816e-01 1.70532382e+00 2.11180285e-01 -6.98865414e-01 9.82466578e-01 1.00255549e+00 1.68544799e-01 -1.64242819e-01 -3.36642176e-01 -6.37488127e-01 -1.79083824e-01 -1.39378846e-01 1.00137365e+00 1.08455539e+00 -3.62789065e-01 3.68809521e-01 -6.84315026e-01 3.35067302e-01 9.67656553e-01 -1.19530715e-01 1.55374014e+00 -1.50017095e+00 -4.13197637e-01 1.68247509e-03 -2.68557340e-01 -1.08939993e+00 3.35687459e-01 -1.02503955e+00 -2.63495813e-03 -1.55035460e+00 6.06871545e-01 -7.96218276e-01 5.58208376e-02 4.56418961e-01 -6.52137458e-01 4.22347248e-01 2.30304152e-01 5.43253124e-01 -4.55477625e-01 5.48316181e-01 9.85494733e-01 -1.19133078e-01 -7.00852394e-01 1.35655478e-01 -8.36842299e-01 5.64702332e-01 6.90977812e-01 -2.13250905e-01 -5.08378327e-01 -4.06861246e-01 -3.59883569e-02 -3.33069891e-01 6.51004791e-01 -1.08228922e+00 3.14271808e-01 1.17981054e-01 1.10408807e+00 -6.19768918e-01 3.06336641e-01 -4.00517136e-01 2.96427816e-01 1.87139824e-01 -1.17310487e-01 9.08316150e-02 -8.82194117e-02 7.05340445e-01 1.81286752e-01 -2.34127834e-01 7.15905964e-01 -2.72730678e-01 -9.53308344e-01 6.68843269e-01 2.89794840e-02 2.53243279e-02 1.00587988e+00 -5.56182206e-01 2.19823606e-02 -2.56467879e-01 -1.09788072e+00 5.29959023e-01 6.35235906e-01 7.20489502e-01 5.14714122e-01 -1.43421793e+00 -9.62674618e-01 4.70726490e-01 2.42719874e-01 6.78216144e-02 5.30965567e-01 2.31183201e-01 -1.24880485e-01 -7.74164051e-02 -2.55882382e-01 -6.02676272e-01 -1.62530112e+00 6.25097394e-01 6.38761401e-01 -3.18501800e-01 -6.86538279e-01 9.67295527e-01 3.71003658e-01 -9.08791423e-01 1.80643842e-01 4.84309852e-01 -6.17802501e-01 2.61955578e-02 1.00055146e+00 4.13565129e-01 -1.05528191e-01 -1.16529942e+00 -3.40467274e-01 5.29287100e-01 -3.71286690e-01 -3.76255751e-01 1.11279666e+00 -3.85497779e-01 2.04404041e-01 -3.18084694e-02 8.49804401e-01 -1.51892066e-01 -1.19115067e+00 -3.07964325e-01 -4.09452915e-01 -5.32189608e-01 -5.91355681e-01 -6.97418511e-01 -8.61852705e-01 4.00924414e-01 8.95824015e-01 -3.43292892e-01 8.29297662e-01 -6.53994689e-03 8.72520089e-01 1.89705551e-01 4.99752402e-01 -1.54924762e+00 1.94586113e-01 3.68713081e-01 6.99649453e-01 -1.52163517e+00 -1.64048374e-01 -3.34127486e-01 -7.43717134e-01 5.89796007e-01 1.02411652e+00 -9.16063860e-02 7.09439874e-01 -2.08406731e-01 1.27687737e-01 1.90774992e-01 4.22998816e-02 -3.62918019e-01 3.49802792e-01 1.14083672e+00 8.18038508e-02 3.39420497e-01 -1.41155645e-02 1.04363585e+00 -6.74237907e-01 1.36404037e-01 2.29193479e-01 5.14202058e-01 8.00546855e-02 -1.34598994e+00 -5.12209833e-01 5.19344389e-01 -1.76212981e-01 1.73877642e-01 -6.89966917e-01 6.61680520e-01 6.00451946e-01 1.05168903e+00 8.36212784e-02 -8.97058547e-01 4.55245763e-01 -1.89009652e-01 4.98553544e-01 -6.92939937e-01 -8.55537057e-01 -5.59818029e-01 1.35394260e-01 -2.15454370e-01 -3.92958313e-01 -7.52830207e-01 -1.04821610e+00 -5.81177175e-01 -3.18486273e-01 1.66913375e-01 4.16699708e-01 1.05193210e+00 3.34631264e-01 1.51707992e-01 5.58557868e-01 -1.23884892e+00 -3.11278582e-01 -1.01657164e+00 -4.80631739e-01 9.13904727e-01 8.42324272e-02 -7.42620468e-01 -9.11810920e-02 3.10623258e-01]
[14.8267822265625, 1.0881916284561157]
cc0facb2-c209-4a2e-8faf-1665b0bfca45
implicit-feature-refinement-for-instance
2112.04709
null
https://arxiv.org/abs/2112.04709v1
https://arxiv.org/pdf/2112.04709v1.pdf
Implicit Feature Refinement for Instance Segmentation
We propose a novel implicit feature refinement module for high-quality instance segmentation. Existing image/video instance segmentation methods rely on explicitly stacked convolutions to refine instance features before the final prediction. In this paper, we first give an empirical comparison of different refinement strategies,which reveals that the widely-used four consecutive convolutions are not necessary. As an alternative, weight-sharing convolution blocks provides competitive performance. When such block is iterated for infinite times, the block output will eventually convergeto an equilibrium state. Based on this observation, the implicit feature refinement (IFR) is developed by constructing an implicit function. The equilibrium state of instance features can be obtained by fixed-point iteration via a simulated infinite-depth network. Our IFR enjoys several advantages: 1) simulates an infinite-depth refinement network while only requiring parameters of single residual block; 2) produces high-level equilibrium instance features of global receptive field; 3) serves as a plug-and-play general module easily extended to most object recognition frameworks. Experiments on the COCO and YouTube-VIS benchmarks show that our IFR achieves improved performance on state-of-the-art image/video instance segmentation frameworks, while reducing the parameter burden (e.g.1% AP improvement on Mask R-CNN with only 30.0% parameters in mask head). Code is made available at https://github.com/lufanma/IFR.git
['Xiangyu Zhang', 'Xiu Li', 'Jiangpeng Yan', 'Bin Dong', 'Tiancai Wang', 'Lufan Ma']
2021-12-09
null
null
null
null
['video-instance-segmentation']
['computer-vision']
[ 1.51031837e-01 9.47597176e-02 -2.04162151e-01 -3.02855134e-01 -7.50363410e-01 -2.92669743e-01 3.47182751e-01 -3.35407168e-01 -6.12506390e-01 4.30574208e-01 -4.43621635e-01 -4.17070925e-01 1.02200486e-01 -7.07787633e-01 -9.43498254e-01 -7.28827536e-01 -1.30680548e-02 1.97002918e-01 6.56428874e-01 -2.45190598e-02 1.92743093e-01 5.18095136e-01 -1.51421213e+00 3.33772063e-01 9.17151392e-01 1.30579829e+00 4.97664005e-01 7.47526348e-01 -1.76689997e-01 5.55637419e-01 -3.33979338e-01 -4.24214065e-01 6.29069090e-01 -5.75207174e-02 -1.08971548e+00 2.91908503e-01 4.37220991e-01 -5.46500862e-01 -3.25621307e-01 9.65245247e-01 2.19554439e-01 1.46190852e-01 2.77992249e-01 -9.60596621e-01 -4.21402693e-01 6.08943582e-01 -8.09082329e-01 3.46741587e-01 -1.81904599e-01 3.65439653e-01 9.59331512e-01 -1.11673713e+00 4.29479599e-01 8.49235415e-01 5.71537554e-01 6.59342289e-01 -1.27847600e+00 -8.29934955e-01 4.68873322e-01 6.09848239e-02 -1.40937555e+00 -2.98829168e-01 4.85682279e-01 -2.15147123e-01 1.02324033e+00 3.24424267e-01 8.59915912e-01 5.40741563e-01 6.15182929e-02 1.00124526e+00 9.83668089e-01 3.58699933e-02 -6.66538775e-02 -1.14140593e-01 3.66155773e-01 9.92059588e-01 6.25836402e-02 -1.38070419e-01 -2.12852061e-01 1.56629309e-01 1.19570351e+00 2.78042316e-01 -4.74815845e-01 -9.19504929e-03 -1.00367606e+00 5.45874119e-01 7.29185581e-01 2.02717975e-01 -2.57597983e-01 4.35676098e-01 1.17173925e-01 2.79981822e-01 5.06695092e-01 1.66840553e-01 -5.76759636e-01 -7.15640187e-02 -1.31243336e+00 1.31834820e-01 5.26768386e-01 9.57904935e-01 1.20855093e+00 -1.63841676e-02 -8.62456933e-02 9.23765004e-01 1.57931715e-01 2.69755602e-01 3.89487237e-01 -1.19744706e+00 2.73975968e-01 6.89723134e-01 -4.73295636e-02 -4.94616598e-01 -2.76957601e-01 -6.93661749e-01 -9.15644646e-01 2.99786568e-01 4.03798670e-01 -5.07076271e-02 -1.42401791e+00 1.23706949e+00 4.27311748e-01 7.15515733e-01 -3.42934906e-01 9.45432186e-01 1.04774427e+00 6.96947038e-01 -1.56620756e-01 1.11312829e-01 1.37348664e+00 -1.28194022e+00 -2.68769860e-01 -1.13124222e-01 4.78501469e-01 -6.44110978e-01 9.54758704e-01 5.02940714e-01 -1.40373468e+00 -7.97858894e-01 -9.33119476e-01 -1.62012070e-01 -2.27611363e-01 8.01272541e-02 7.73751795e-01 4.80992019e-01 -1.37912393e+00 8.00222993e-01 -9.71383035e-01 1.48128718e-01 8.12535465e-01 8.52870166e-01 -3.71381998e-01 1.61137715e-01 -8.00173521e-01 3.84842426e-01 2.89065093e-01 3.42992246e-01 -8.66666496e-01 -1.11247575e+00 -7.65114009e-01 5.75723834e-02 5.07618427e-01 -6.80202901e-01 1.39608884e+00 -1.21706128e+00 -1.58877540e+00 8.19074810e-01 -2.58336216e-01 -5.04308164e-01 6.17442012e-01 -3.44590276e-01 5.64373508e-02 3.04042995e-01 -7.15146884e-02 1.06701863e+00 1.00463617e+00 -1.13995278e+00 -8.33302259e-01 -9.78344604e-02 3.81652564e-01 2.51796514e-01 -1.08684517e-01 -1.17289379e-01 -1.07028973e+00 -5.92882633e-01 1.51354983e-01 -8.14554334e-01 -6.48762822e-01 7.27270171e-03 -2.75363237e-01 -1.15022674e-01 6.23138130e-01 -4.44512308e-01 1.27228224e+00 -2.08326101e+00 -4.79585417e-02 2.86716700e-01 4.72973526e-01 5.41823328e-01 -1.76946253e-01 -2.88506180e-01 -1.58418626e-01 3.01684678e-01 -4.45725203e-01 -5.36396682e-01 -3.44226927e-01 5.06974645e-02 -1.23576194e-01 4.60601509e-01 4.30085748e-01 1.19640887e+00 -6.13892674e-01 -4.00026500e-01 3.84768337e-01 5.49600065e-01 -8.83796334e-01 6.06892519e-02 -1.18653253e-02 3.36178929e-01 -3.92468989e-01 5.99190652e-01 7.95265377e-01 -6.03651583e-01 -1.50579810e-01 -1.77205324e-01 -3.26527178e-01 2.56869882e-01 -1.19706738e+00 1.72993934e+00 -3.48005086e-01 3.93688858e-01 1.20350115e-01 -9.30450261e-01 6.95601702e-01 1.98946163e-01 5.11175871e-01 -5.99582672e-01 2.52238244e-01 2.98482388e-01 6.23738766e-02 -5.88165708e-02 4.39610124e-01 3.01445514e-01 2.26749808e-01 1.11851804e-01 1.90742344e-01 -8.65656976e-03 2.48373330e-01 1.47979245e-01 1.02284837e+00 2.16330454e-01 -4.11548726e-02 -4.17475134e-01 7.07107902e-01 -1.27556071e-01 6.10029936e-01 7.83585548e-01 -3.09419334e-02 8.90180528e-01 4.33358222e-01 -5.77866971e-01 -9.30326700e-01 -8.84905577e-01 -3.87257427e-01 8.44152153e-01 3.40228945e-01 -3.88778031e-01 -1.07485294e+00 -6.47683501e-01 -2.32227191e-01 1.57965928e-01 -7.36087203e-01 1.61161676e-01 -7.28188097e-01 -6.55309439e-01 4.26789880e-01 7.04522610e-01 8.29538345e-01 -1.14496291e+00 -7.25898683e-01 2.54585236e-01 7.67454356e-02 -1.12374389e+00 -6.11087918e-01 2.16753095e-01 -1.15901232e+00 -1.03415835e+00 -9.51073289e-01 -8.33347380e-01 8.41153324e-01 4.71673399e-01 1.02988911e+00 6.21158183e-01 -3.28679860e-01 1.80072203e-01 -2.09835827e-01 1.19516760e-01 1.09503828e-01 2.91143507e-01 -3.31002533e-01 2.94960700e-02 9.79699939e-02 -5.31825542e-01 -1.07199776e+00 4.64146316e-01 -9.95184779e-01 4.94580954e-01 5.05016029e-01 9.14683104e-01 9.83970463e-01 -2.39139587e-01 3.69022936e-01 -1.04997969e+00 1.12432390e-01 -1.73496604e-01 -7.13770747e-01 5.80866858e-02 -4.51825410e-01 -1.33927703e-01 5.43638885e-01 -4.33232039e-01 -1.04602575e+00 1.51512355e-01 -4.72572267e-01 -6.18020296e-01 -1.53632671e-01 1.51221275e-01 3.19271721e-02 -1.66489631e-01 2.43473575e-01 2.34232768e-01 -2.32238472e-02 -4.24992561e-01 4.25496161e-01 5.18975079e-01 4.09817219e-01 -6.18007481e-01 7.94153929e-01 6.08020246e-01 -3.38192284e-01 -6.72702968e-01 -7.39190459e-01 -5.94568431e-01 -5.87296605e-01 -2.19657436e-01 8.23567927e-01 -9.63084221e-01 -9.04027104e-01 7.58360028e-01 -9.95727122e-01 -9.15188551e-01 -3.23079586e-01 1.79535255e-01 -4.98476207e-01 2.35401392e-01 -8.07298958e-01 -3.85800987e-01 -5.55782914e-01 -1.52537119e+00 1.02397418e+00 4.75381702e-01 3.44480909e-02 -7.08779097e-01 -4.83631223e-01 3.36482435e-01 3.19936067e-01 -1.34688526e-01 4.18194264e-01 -3.20916831e-01 -9.22675908e-01 1.17233977e-01 -4.26019400e-01 4.56050336e-01 3.39616761e-02 1.34245217e-01 -1.00775003e+00 -2.60372490e-01 -1.22551560e-01 -6.69285357e-02 1.13066506e+00 6.65519238e-01 1.77167666e+00 -1.33159176e-01 -2.07660839e-01 1.18079770e+00 1.41992652e+00 1.43887818e-01 8.41870844e-01 1.91579401e-01 8.27869654e-01 1.96026161e-01 5.57583570e-01 2.87080616e-01 2.49191284e-01 4.95521694e-01 4.18479234e-01 -5.50831199e-01 -2.58096337e-01 1.40216008e-01 2.37071961e-01 7.06450641e-01 -4.57730711e-01 8.68433192e-02 -7.69262731e-01 6.34171069e-01 -1.83156669e+00 -7.04229116e-01 -2.70575136e-01 2.11339402e+00 8.76560867e-01 3.08104813e-01 1.66244775e-01 4.02542986e-02 6.52320623e-01 1.91866070e-01 -5.27137399e-01 -3.21787506e-01 3.28370988e-01 7.67607391e-01 7.07966447e-01 6.19626224e-01 -1.12264407e+00 1.11602330e+00 4.99414492e+00 1.14994967e+00 -1.18259811e+00 2.43880495e-01 1.12634158e+00 -2.19547451e-01 -1.85494095e-01 -1.47237971e-01 -9.67950106e-01 3.65701050e-01 6.08085692e-01 1.65437356e-01 4.77201909e-01 7.52034605e-01 7.40415901e-02 -6.55727647e-03 -8.46594274e-01 9.69934583e-01 -4.20155197e-01 -1.62620008e+00 6.80475235e-02 1.31098583e-01 6.80437148e-01 2.95558840e-01 1.88611865e-01 1.97837278e-01 1.12661742e-01 -1.02839482e+00 7.88920701e-01 1.88076094e-01 1.07718277e+00 -8.76154184e-01 5.89099824e-01 1.57421038e-01 -1.42036951e+00 3.23106758e-02 -3.83780718e-01 5.41207977e-02 1.03736177e-01 5.31317055e-01 -4.50806707e-01 4.49012190e-01 1.03167200e+00 7.13993311e-01 -4.36790526e-01 1.12921107e+00 -7.29059801e-02 6.17911220e-01 -3.82616043e-01 3.06359351e-01 6.09344363e-01 -2.04310715e-01 2.77600944e-01 1.39020956e+00 1.62244439e-01 3.35413456e-01 9.24384445e-02 8.00593555e-01 -2.54972637e-01 1.22353313e-02 -1.53634056e-01 3.79311532e-01 3.26495059e-02 1.52718365e+00 -1.14502621e+00 -4.19275880e-01 -6.10209048e-01 1.05208719e+00 3.82451445e-01 4.10829455e-01 -1.11142659e+00 -2.96989560e-01 7.92658567e-01 2.81615525e-01 7.70173252e-01 -5.96295558e-02 -4.11664963e-01 -1.13477325e+00 -2.40523648e-03 -7.75023043e-01 2.69643981e-02 -3.47389877e-01 -8.21149826e-01 8.55578363e-01 -1.74516499e-01 -1.03183162e+00 1.59248024e-01 -6.24406338e-01 -5.69977045e-01 8.56219471e-01 -1.75337768e+00 -9.35168803e-01 -4.10243243e-01 7.74115503e-01 9.39866781e-01 3.01527292e-01 5.79855680e-01 4.66248482e-01 -7.17739165e-01 7.42275655e-01 -2.96533853e-01 3.23435307e-01 1.79819793e-01 -1.24793661e+00 6.85107470e-01 8.19155514e-01 1.25116631e-01 6.82372153e-01 2.39616007e-01 -4.79305774e-01 -1.07297361e+00 -1.14671767e+00 3.52309883e-01 -9.99150276e-02 6.08459055e-01 -3.30927610e-01 -1.03482199e+00 6.42639637e-01 2.21447721e-01 4.97758150e-01 4.16706890e-01 -1.17301449e-01 -1.64110631e-01 -2.30685353e-01 -1.00743353e+00 7.29838431e-01 1.24830186e+00 -1.48633748e-01 -1.46232709e-01 5.63411601e-02 7.28240907e-01 -7.43420601e-01 -9.46121216e-01 6.07970119e-01 5.75365365e-01 -1.10799778e+00 9.03615654e-01 -3.12518477e-01 5.58299124e-01 -4.37454283e-01 1.12963855e-01 -7.54272044e-01 -4.74395454e-01 -9.16091204e-01 -2.75981516e-01 9.83928442e-01 6.60445929e-01 -7.51410663e-01 8.65159512e-01 6.76047385e-01 -3.34795713e-01 -1.57793283e+00 -8.27524602e-01 -4.43109006e-01 -5.33934981e-02 -6.67260885e-01 6.11552358e-01 4.70036060e-01 -5.19702435e-01 -7.08864406e-02 -2.30536517e-02 2.27171808e-01 4.36304212e-01 9.37373862e-02 6.92138970e-01 -9.65791762e-01 -5.11830747e-01 -6.05091691e-01 -3.28521818e-01 -1.76656008e+00 -1.05379485e-02 -8.17265391e-01 -3.85562107e-02 -1.35385835e+00 2.34572843e-01 -7.47601032e-01 -5.32313824e-01 7.02817440e-01 -2.85722941e-01 7.38672376e-01 3.17656130e-01 8.30481648e-02 -6.07923627e-01 2.30749950e-01 1.50138330e+00 -2.33097948e-04 -2.08240584e-01 1.67992964e-01 -6.25910938e-01 8.81754577e-01 9.60145593e-01 -3.89181346e-01 -3.43556553e-01 -4.60881293e-01 -4.96867076e-02 -1.64651111e-01 4.59115207e-01 -8.59664321e-01 1.04677834e-01 4.36848216e-02 3.72764647e-01 -5.64224660e-01 3.96370798e-01 -8.10916603e-01 1.43041030e-01 5.22666752e-01 9.88934655e-03 -1.60385445e-02 3.71522933e-01 2.13352308e-01 -1.13502942e-01 -4.14576948e-01 9.88597870e-01 -3.00758868e-01 -7.71993101e-01 7.39993870e-01 -1.26316115e-01 -2.74932757e-02 9.56907332e-01 -7.18790948e-01 -1.32941768e-01 4.64325817e-03 -7.88130045e-01 3.02241117e-01 5.16465902e-01 2.30047673e-01 7.12867260e-01 -8.95026386e-01 -5.80683589e-01 2.84019470e-01 -4.01016295e-01 5.09609818e-01 3.09589475e-01 1.03195071e+00 -7.61849165e-01 1.67139322e-01 1.14024833e-01 -8.42182219e-01 -1.25980842e+00 1.16864339e-01 5.06548882e-01 -4.05707717e-01 -8.97121847e-01 1.22132802e+00 6.76301062e-01 -2.38936856e-01 1.76470131e-01 -5.17865121e-01 1.11653870e-02 -8.21404457e-02 5.56105077e-01 2.58517921e-01 8.49665180e-02 -5.43663323e-01 -2.45612785e-01 6.58429325e-01 -4.65770394e-01 4.97980528e-02 1.51994848e+00 1.51225366e-02 -1.51242921e-02 9.17014033e-02 1.18185163e+00 -2.48869807e-01 -1.79489660e+00 -1.24533683e-01 -3.54799688e-01 -4.45220679e-01 1.76166639e-01 -4.87207204e-01 -1.77781713e+00 7.90331602e-01 5.00555754e-01 1.30866513e-01 1.30877221e+00 -7.36390874e-02 1.02269030e+00 8.17303807e-02 3.67613614e-01 -8.23072314e-01 -6.94710910e-02 4.63734478e-01 6.68641686e-01 -1.02012992e+00 -6.95570335e-02 -6.35689020e-01 -5.02283573e-01 1.03671908e+00 8.08101654e-01 -5.33557713e-01 8.24773788e-01 4.18380111e-01 -5.71078993e-02 -1.29557103e-01 -5.91555655e-01 -3.10117811e-01 4.60293561e-01 1.39737263e-01 4.40022677e-01 -2.35737562e-02 -2.29087308e-01 5.67790687e-01 -1.20734014e-01 1.26744742e-02 2.82821596e-01 5.81995845e-01 -3.71152669e-01 -9.92961884e-01 1.37954233e-02 6.78590894e-01 -7.97150850e-01 -4.00827020e-01 1.81830168e-01 8.25480998e-01 2.09897175e-01 5.78307092e-01 2.77859360e-01 -1.49227738e-01 1.64483264e-01 -2.50432581e-01 4.81102914e-01 -6.44901097e-01 -1.06062150e+00 2.35227138e-01 -3.18594366e-01 -1.01593471e+00 -3.86092842e-01 -4.49269980e-01 -1.63741171e+00 -2.71303743e-01 -5.60720325e-01 -4.58508842e-02 3.73164743e-01 8.01773489e-01 4.26452219e-01 7.73414195e-01 4.37613934e-01 -1.17397368e+00 -1.29719734e-01 -7.21137762e-01 -3.96593153e-01 1.47016972e-01 2.56728560e-01 -4.60058719e-01 -2.29168117e-01 9.29304883e-02]
[9.516234397888184, 0.059627994894981384]
b2aadaf1-be43-46b2-bc21-73d518c8eba1
explore-more-guidance-a-task-aware
2204.05953
null
https://arxiv.org/abs/2204.05953v3
https://arxiv.org/pdf/2204.05953v3.pdf
Explore More Guidance: A Task-aware Instruction Network for Sign Language Translation Enhanced with Data Augmentation
Sign language recognition and translation first uses a recognition module to generate glosses from sign language videos and then employs a translation module to translate glosses into spoken sentences. Most existing works focus on the recognition step, while paying less attention to sign language translation. In this work, we propose a task-aware instruction network, namely TIN-SLT, for sign language translation, by introducing the instruction module and the learning-based feature fuse strategy into a Transformer network. In this way, the pre-trained model's language ability can be well explored and utilized to further boost the translation performance. Moreover, by exploring the representation space of sign language glosses and target spoken language, we propose a multi-level data augmentation scheme to adjust the data distribution of the training set. We conduct extensive experiments on two challenging benchmark datasets, PHOENIX-2014-T and ASLG-PC12, on which our method outperforms former best solutions by 1.65 and 1.42 in terms of BLEU-4. Our code is published at https://github.com/yongcaoplus/TIN-SLT.
['Hwang Kai', 'Zhengdao Li', 'Long Hu', 'Guangyong Chen', 'Min Chen', 'Xianzhi Li', 'Wei Li', 'Yong Cao']
2022-04-12
null
https://aclanthology.org/2022.findings-naacl.205
https://aclanthology.org/2022.findings-naacl.205.pdf
findings-naacl-2022-7
['sign-language-recognition', 'sign-language-translation']
['computer-vision', 'computer-vision']
[ 2.71584332e-01 -2.51212984e-01 -4.33937967e-01 -4.88122612e-01 -8.55937302e-01 -3.54482323e-01 6.47955477e-01 -9.21750903e-01 -4.79895473e-01 4.86063272e-01 6.02467477e-01 -2.16902152e-01 3.53106976e-01 -5.37212074e-01 -6.22413933e-01 -7.08495557e-01 6.61943436e-01 3.88119549e-01 3.43141444e-02 -2.27717951e-01 -3.20288055e-02 -2.32240222e-02 -1.26101327e+00 2.73709118e-01 1.13222480e+00 9.03541327e-01 1.26661614e-01 3.63937497e-01 -2.65383244e-01 6.45094931e-01 -4.44138825e-01 -3.66763324e-01 2.70749599e-01 -8.36025000e-01 -4.42059368e-01 6.11479469e-02 4.24102962e-01 -5.55266798e-01 -5.83476365e-01 8.46061468e-01 8.10066462e-01 -6.90650195e-02 5.37965238e-01 -1.13315356e+00 -6.03066862e-01 6.17192268e-01 -2.04686791e-01 -3.23177367e-01 2.56722093e-01 5.42507052e-01 7.73679137e-01 -1.01781821e+00 6.31476283e-01 1.16485548e+00 2.70294517e-01 9.74830210e-01 -6.49562597e-01 -9.09368455e-01 1.99630931e-01 3.43715817e-01 -1.13690233e+00 -5.43156266e-01 7.09495187e-01 -8.87291729e-02 6.93964958e-01 1.86709270e-01 8.14637721e-01 1.28403080e+00 -7.29641914e-02 1.35640693e+00 1.32722449e+00 -4.41609025e-01 -1.68921605e-01 -2.19254375e-01 8.09224509e-03 8.15089881e-01 3.94580048e-03 1.58566639e-01 -6.16090775e-01 2.97101527e-01 6.67221129e-01 6.78522140e-02 -4.96995747e-01 -1.59614623e-01 -1.40173876e+00 4.95652884e-01 3.55194569e-01 2.82178998e-01 -3.13077062e-01 1.13460824e-01 4.80999053e-01 4.62296665e-01 1.50375396e-01 -1.01783872e-01 -3.38601589e-01 -2.22168297e-01 -7.52824545e-01 -7.52758607e-02 5.97747564e-01 1.09606922e+00 3.94610405e-01 2.31527448e-01 -5.81189275e-01 9.57826614e-01 6.26442552e-01 9.78369951e-01 9.61552382e-01 -4.61933285e-01 8.72234344e-01 7.59921551e-01 -3.26533854e-01 -2.68192440e-01 -2.15156265e-02 -2.84099340e-01 -6.48151994e-01 7.56397191e-03 2.43641123e-01 -2.63380915e-01 -1.50234652e+00 1.66041028e+00 1.23508036e-01 1.64693534e-01 2.44453296e-01 1.20199335e+00 1.09381104e+00 5.51972449e-01 3.24349664e-02 -1.62416808e-02 1.36812675e+00 -1.37609446e+00 -7.50660181e-01 -2.32653618e-01 5.66193163e-01 -8.92242789e-01 1.51033199e+00 2.52229959e-01 -8.86149764e-01 -3.55211347e-01 -7.68540442e-01 -2.07817927e-01 -1.83040962e-01 7.51426816e-01 3.69615555e-01 2.72124946e-01 -5.92280805e-01 1.18367359e-01 -9.01827097e-01 -4.57651258e-01 5.06925583e-01 2.77976513e-01 -1.44652590e-01 -3.80327523e-01 -1.04781771e+00 8.59477997e-01 2.42714331e-01 4.62040037e-01 -7.24543273e-01 -2.27924064e-01 -6.79139256e-01 -2.31760427e-01 3.13728541e-01 -7.99538612e-01 1.24392045e+00 -8.60054851e-01 -2.16483760e+00 6.43148899e-01 -3.36130321e-01 6.15298282e-03 8.31243277e-01 -2.53034770e-01 -4.79726613e-01 -1.14809386e-01 -2.40735799e-01 5.75226068e-01 8.89682472e-01 -9.47832286e-01 -6.02716386e-01 -2.78024375e-01 -1.86739296e-01 5.03366828e-01 -2.65142947e-01 1.15020156e-01 -8.13588440e-01 -8.30079854e-01 1.43410623e-01 -9.38182592e-01 1.92196865e-03 6.84835715e-03 -3.10755700e-01 -3.14822674e-01 6.26455009e-01 -9.64576900e-01 1.30333352e+00 -2.00576019e+00 3.98169577e-01 2.78249562e-01 -4.98202220e-02 6.23030245e-01 -4.82356936e-01 3.25210482e-01 3.55062574e-01 -1.83118597e-01 -3.11376244e-01 -4.91190225e-01 2.10781321e-01 4.72314507e-01 -3.15442204e-01 1.46159381e-01 1.34220600e-01 1.27297640e+00 -6.25909507e-01 -4.29222286e-01 2.45343715e-01 5.20573318e-01 -4.70559269e-01 2.73989201e-01 -1.78684220e-01 5.22748351e-01 -4.90445465e-01 8.46893907e-01 4.95618641e-01 1.39708184e-02 1.03425249e-01 -2.68770397e-01 -1.22081466e-01 5.63267469e-01 -6.87027216e-01 1.89315939e+00 -6.64804399e-01 4.37243998e-01 -1.74141735e-01 -6.25887156e-01 1.09431922e+00 3.66663456e-01 2.07604885e-01 -8.82707596e-01 4.75704551e-01 6.65113032e-01 4.26868151e-04 -7.15936482e-01 5.54930307e-02 -2.74302009e-02 1.13809388e-02 4.78475004e-01 -4.24780771e-02 -3.45000662e-02 2.29729265e-01 -2.50340730e-01 9.24955130e-01 4.24450725e-01 -1.21277779e-01 1.86789095e-01 6.12088978e-01 -1.11621141e-01 5.37398517e-01 1.71079040e-01 -2.10033104e-01 5.09841323e-01 8.46395791e-02 -2.13620141e-01 -7.45799839e-01 -9.36097682e-01 2.00687587e-01 1.01612091e+00 3.31764743e-02 -2.09635690e-01 -7.67032623e-01 -8.70872796e-01 -2.19776139e-01 6.27753258e-01 -3.07083458e-01 -3.50268722e-01 -9.11857009e-01 -4.87045228e-01 6.47645772e-01 5.86088717e-01 8.61396730e-01 -1.33729124e+00 -3.30916733e-01 3.08617689e-02 -5.62032521e-01 -1.04910827e+00 -1.06878889e+00 -2.66153753e-01 -8.21892738e-01 -8.37057650e-01 -1.04784048e+00 -1.11388075e+00 7.96231449e-01 -1.09416246e-01 5.34422815e-01 -1.05074709e-02 1.57737862e-02 1.40512630e-01 -6.07980311e-01 -2.80950189e-01 -3.63631278e-01 2.08783180e-01 -4.77157943e-02 1.38855547e-01 6.78702831e-01 -4.54093426e-01 -5.87023318e-01 4.81063366e-01 -8.06486487e-01 2.93538928e-01 1.13981533e+00 1.04098547e+00 5.06371915e-01 -7.23969162e-01 2.85066634e-01 -3.31749797e-01 6.58402085e-01 -7.50689507e-02 -4.36486244e-01 5.10834098e-01 -6.30552173e-01 2.35096097e-01 5.75677037e-01 -7.79392362e-01 -9.58217978e-01 1.01011671e-01 -2.94191331e-01 -4.35557812e-01 1.05445057e-01 4.87919629e-01 -3.94906670e-01 -1.18627362e-02 2.89026439e-01 8.43135118e-01 2.77456820e-01 -5.45802951e-01 3.31826359e-01 1.04488885e+00 4.80108410e-01 -3.07896703e-01 8.76258433e-01 1.23088121e-01 -4.19900686e-01 -4.43154275e-01 -5.58928847e-01 -1.24212645e-01 -4.39060271e-01 -2.70515203e-01 5.97687721e-01 -9.18286324e-01 -5.04148722e-01 7.71596730e-01 -9.27880883e-01 -5.46512365e-01 -2.98729718e-01 7.23886311e-01 -5.96347570e-01 3.10205132e-01 -4.94429648e-01 -3.44535798e-01 -7.30656385e-01 -1.29224181e+00 1.10297871e+00 2.25497857e-01 -9.72285029e-03 -5.00218868e-01 7.60710016e-02 7.34436572e-01 6.00063145e-01 -1.28604144e-01 4.93577838e-01 -3.43980938e-01 -6.97637320e-01 -1.16649427e-01 -3.15485179e-01 6.80286467e-01 3.46527725e-01 -2.53317714e-01 -6.64112210e-01 -3.15373659e-01 -3.31641763e-01 -3.88841778e-01 1.10307562e+00 1.35915399e-01 7.55140960e-01 -5.10583937e-01 -4.64833379e-02 8.33268881e-01 1.08990180e+00 1.59397215e-01 7.76879311e-01 5.46161644e-02 7.49888599e-01 2.93512404e-01 5.85106730e-01 1.78805992e-01 7.19568908e-01 7.30423689e-01 -1.45445485e-02 3.55697190e-03 -6.95281923e-01 -5.46432078e-01 8.43695641e-01 1.42832530e+00 -3.26153815e-01 -1.95720494e-01 -7.02671111e-01 3.99333984e-01 -1.88252985e+00 -6.38071835e-01 1.78801358e-01 1.95905852e+00 1.06163692e+00 -1.32225022e-01 5.31333946e-02 3.78045775e-02 4.31992114e-01 8.36966187e-02 -6.53106689e-01 -1.77236065e-01 -1.13707326e-01 2.77774602e-01 4.12461251e-01 5.55038214e-01 -8.30786288e-01 1.34717607e+00 5.15197563e+00 8.37335050e-01 -1.53419876e+00 5.33354282e-02 1.41320564e-02 -2.46262476e-01 -3.27422559e-01 -1.65154517e-01 -7.89690137e-01 7.25175858e-01 5.60969055e-01 -1.05902471e-01 7.41170645e-01 5.29336989e-01 4.16592121e-01 3.58604789e-01 -7.96626806e-01 9.88348603e-01 3.17939848e-01 -9.11323905e-01 3.65465999e-01 6.22615330e-02 5.27838767e-01 3.96251291e-01 4.83480021e-02 5.32261729e-01 1.42714903e-01 -7.10448027e-01 6.63414419e-01 7.98874080e-01 1.10676372e+00 -3.74807954e-01 8.11985075e-01 8.60147923e-02 -1.22327685e+00 8.52164701e-02 -7.84645136e-03 2.24770412e-01 2.17110485e-01 1.19650662e-01 -5.74843645e-01 4.96202648e-01 3.70767742e-01 7.91389942e-01 -3.88384342e-01 1.01320577e+00 -7.97692060e-01 9.04440880e-01 -5.54224253e-01 -2.72516906e-01 1.84989274e-01 -2.92762280e-01 3.64098549e-01 1.18973613e+00 5.05575716e-01 -2.94582336e-03 1.77290350e-01 5.80105901e-01 -1.75027266e-01 3.29326928e-01 -3.08503956e-01 -1.05139688e-01 2.91852802e-01 8.69422555e-01 -7.67524391e-02 -4.70451444e-01 -4.55046147e-01 1.28144860e+00 -1.76529679e-02 5.69823682e-01 -8.45961034e-01 -4.49271530e-01 6.07028127e-01 -1.39921159e-01 3.23078603e-01 -2.21906692e-01 -3.69504355e-02 -1.64083743e+00 5.04631519e-01 -1.21100724e+00 1.12914406e-01 -6.29447460e-01 -1.08554864e+00 5.85956693e-01 -2.60969996e-01 -1.56858194e+00 -2.12074116e-01 -6.89076126e-01 -5.05770862e-01 8.68107796e-01 -1.70072389e+00 -1.64655757e+00 -5.16991854e-01 7.11431921e-01 6.36712193e-01 -3.83194715e-01 5.73690772e-01 6.07390702e-01 -6.12038016e-01 1.03322816e+00 4.33394909e-02 3.72689337e-01 8.79779398e-01 -6.74335480e-01 2.99162954e-01 8.16506624e-01 1.21724710e-01 4.46630716e-01 1.99641675e-01 -5.23699105e-01 -1.46112025e+00 -1.13532853e+00 1.11751509e+00 -2.28595719e-01 5.27508199e-01 -2.24283904e-01 -6.09922230e-01 6.91005766e-01 1.80261984e-01 -1.30112082e-01 4.07575250e-01 -3.95170718e-01 -3.63990813e-01 -2.41340771e-01 -9.03296590e-01 9.09304917e-01 1.41855335e+00 -4.70943570e-01 -5.02128124e-01 2.55190164e-01 5.12911141e-01 -6.05822504e-01 -7.78624833e-01 4.14663732e-01 9.68905270e-01 -2.99933612e-01 5.61715126e-01 -5.40137589e-01 4.58827257e-01 -4.00935411e-01 -2.29439050e-01 -1.30347502e+00 1.09552957e-01 -5.28615654e-01 -1.34900495e-01 1.12132239e+00 3.65274042e-01 -8.11138451e-01 7.17452288e-01 2.31126666e-01 -3.26737791e-01 -8.73772681e-01 -1.00226450e+00 -7.98761368e-01 -6.57497440e-03 -2.38247305e-01 7.36345828e-01 6.41725540e-01 -1.81817770e-01 3.99693310e-01 -5.75272977e-01 -2.39728838e-01 4.74221349e-01 1.98448718e-01 9.56952214e-01 -6.71287954e-01 -1.53782189e-01 -6.98990345e-01 -1.75023243e-01 -1.48152101e+00 -2.12562010e-02 -1.08458674e+00 3.03636044e-01 -1.72401690e+00 3.72520089e-02 -1.30472347e-01 -2.61012226e-01 9.43634033e-01 -1.84410572e-01 2.26735070e-01 3.27183634e-01 3.11143577e-01 -3.61580968e-01 1.03553188e+00 1.70551860e+00 -2.41241693e-01 -2.16869473e-01 8.34771693e-02 -4.34108347e-01 4.93264049e-01 1.06237888e+00 -2.07759976e-01 -2.40946770e-01 -8.70787263e-01 -3.33212167e-01 -3.29676300e-01 2.14286402e-01 -8.69749308e-01 1.67066321e-01 -1.83842987e-01 -1.99811548e-01 -4.63409811e-01 2.45165467e-01 -7.76232541e-01 -2.18943954e-01 5.69857240e-01 -3.45547408e-01 -2.59452879e-01 6.51040003e-02 1.95112973e-01 -3.21996152e-01 4.40865830e-02 6.46917820e-01 9.08362195e-02 -7.36724317e-01 4.65426862e-01 -3.67580354e-01 -3.57863158e-02 6.78044319e-01 -8.18918720e-02 -4.02211457e-01 -3.54784369e-01 -4.81407344e-01 4.99616385e-01 2.28402868e-01 6.42516971e-01 8.87700438e-01 -1.69079089e+00 -8.77254963e-01 4.71160322e-01 3.94621283e-01 -1.56883791e-01 3.30686234e-02 9.60233867e-01 -5.58125317e-01 2.19266698e-01 -3.12097400e-01 -4.33189362e-01 -1.31395471e+00 -1.62934333e-01 3.21790278e-01 -2.63808101e-01 -6.29271805e-01 6.90208554e-01 -1.99920237e-01 -8.48007143e-01 3.44575077e-01 -7.38471687e-01 4.49028835e-02 -2.71432012e-01 3.22740167e-01 1.16694920e-01 -2.68873632e-01 -7.25137174e-01 -3.41792256e-01 8.52464497e-01 -7.13559762e-02 -2.07492083e-01 1.11979580e+00 -5.59636354e-02 -2.48089619e-02 1.43538415e-01 1.12001669e+00 -1.31472811e-01 -1.04661143e+00 -5.66084445e-01 -1.39346182e-01 -2.86569327e-01 -1.37610510e-01 -1.23190689e+00 -1.19650543e+00 8.62744212e-01 7.03176856e-01 -8.40313971e-01 1.44226205e+00 -4.75931875e-02 1.26259112e+00 5.21887243e-01 2.83188611e-01 -9.46003854e-01 1.15505993e-01 7.03335226e-01 1.15619218e+00 -1.27545118e+00 -3.51171821e-01 -1.93316132e-01 -6.92140818e-01 9.49922681e-01 6.39661610e-01 -9.09418315e-02 3.74504328e-01 1.32346794e-01 5.74328840e-01 1.48963302e-01 -5.67089617e-01 -3.76877457e-01 5.21043360e-01 3.28162134e-01 3.33147287e-01 1.65282875e-01 -7.52915084e-01 5.65266013e-01 -3.41156602e-01 4.88092691e-01 5.29605672e-02 8.74842644e-01 -3.19037497e-01 -1.44934404e+00 -6.57577291e-02 4.35629219e-01 1.37507021e-01 -2.48102352e-01 -4.79688048e-01 6.17358983e-01 1.73525944e-01 6.92554712e-01 -2.40085408e-01 -6.79958045e-01 6.41012847e-01 3.03540319e-01 6.37250662e-01 -4.91991729e-01 -4.38085824e-01 1.29917651e-01 1.56232744e-01 -5.38669586e-01 -4.58289027e-01 -7.14721024e-01 -1.33073592e+00 -6.56515360e-02 -2.11661160e-01 -1.34910464e-01 6.47601902e-01 1.05329096e+00 2.85941899e-01 4.99049544e-01 4.73745883e-01 -3.79753202e-01 -6.79208040e-01 -1.27281034e+00 -1.31741941e-01 3.19438428e-01 2.54044235e-01 -4.58036095e-01 -2.28873596e-01 1.59548968e-02]
[9.208115577697754, -6.520098686218262]
46d160fb-d22d-471a-ad29-6a0a82e92f65
kimera-from-slam-to-spatial-perception-with
2101.06894
null
https://arxiv.org/abs/2101.06894v3
https://arxiv.org/pdf/2101.06894v3.pdf
Kimera: from SLAM to Spatial Perception with 3D Dynamic Scene Graphs
Humans are able to form a complex mental model of the environment they move in. This mental model captures geometric and semantic aspects of the scene, describes the environment at multiple levels of abstractions (e.g., objects, rooms, buildings), includes static and dynamic entities and their relations (e.g., a person is in a room at a given time). In contrast, current robots' internal representations still provide a partial and fragmented understanding of the environment, either in the form of a sparse or dense set of geometric primitives (e.g., points, lines, planes, voxels) or as a collection of objects. This paper attempts to reduce the gap between robot and human perception by introducing a novel representation, a 3D Dynamic Scene Graph(DSG), that seamlessly captures metric and semantic aspects of a dynamic environment. A DSG is a layered graph where nodes represent spatial concepts at different levels of abstraction, and edges represent spatio-temporal relations among nodes. Our second contribution is Kimera, the first fully automatic method to build a DSG from visual-inertial data. Kimera includes state-of-the-art techniques for visual-inertial SLAM, metric-semantic 3D reconstruction, object localization, human pose and shape estimation, and scene parsing. Our third contribution is a comprehensive evaluation of Kimera in real-life datasets and photo-realistic simulations, including a newly released dataset, uHumans2, which simulates a collection of crowded indoor and outdoor scenes. Our evaluation shows that Kimera achieves state-of-the-art performance in visual-inertial SLAM, estimates an accurate 3D metric-semantic mesh model in real-time, and builds a DSG of a complex indoor environment with tens of objects and humans in minutes. Our final contribution shows how to use a DSG for real-time hierarchical semantic path-planning. The core modules in Kimera are open-source.
['Luca Carlone', 'Arjun Gupta', 'Jingnan Shi', 'Yun Chang', 'Nathan Hughes', 'Marcus Abate', 'Andrew Violette', 'Antoni Rosinol']
2021-01-18
null
null
null
null
['scene-parsing']
['computer-vision']
[-3.07723641e-01 3.25887464e-02 3.50070477e-01 -4.72240806e-01 -2.15209916e-01 -4.49028015e-01 6.48822844e-01 4.32681769e-01 -2.06470490e-01 4.87750530e-01 1.83383122e-01 2.09314078e-02 -2.62960121e-02 -1.10409939e+00 -8.74418855e-01 -1.61298797e-01 -3.88638705e-01 1.26822424e+00 5.97887695e-01 -5.46172619e-01 2.63659637e-02 7.73251116e-01 -1.76156676e+00 -9.61814970e-02 4.05812800e-01 8.34186852e-01 7.58921862e-01 7.42442071e-01 7.13079143e-03 5.36715865e-01 -4.01267707e-01 2.71875411e-01 1.79654479e-01 1.25047499e-02 -7.56678939e-01 2.40438029e-01 4.44575280e-01 -9.08056833e-03 -3.81135195e-01 7.52534211e-01 3.00518274e-01 3.27278137e-01 3.40962380e-01 -1.54333329e+00 -1.10528633e-01 -3.16606387e-02 -5.15730772e-03 -3.05857748e-01 1.22373652e+00 1.68381244e-01 4.78456646e-01 -8.34653139e-01 1.00467896e+00 1.64545906e+00 9.11508083e-01 9.48872566e-02 -1.12066805e+00 -2.30232880e-01 2.98323214e-01 2.59258151e-01 -1.73287868e+00 -1.89829573e-01 5.67150593e-01 -4.32669163e-01 1.44393516e+00 3.01013678e-01 1.15926969e+00 1.02855766e+00 3.75745624e-01 3.91098499e-01 8.93973231e-01 -2.84312159e-01 6.55200958e-01 -3.87100652e-02 4.10574861e-02 1.11515045e+00 3.44358474e-01 -1.52911335e-01 -6.87823951e-01 -2.29152322e-01 1.03616345e+00 1.77123845e-01 -8.43030661e-02 -1.25952101e+00 -1.51932383e+00 3.10102999e-01 9.72546160e-01 9.16832089e-02 -3.77767295e-01 2.64570862e-01 5.32972515e-02 -2.44887426e-01 6.86867684e-02 2.22090647e-01 -4.04645205e-01 -1.27030924e-01 -3.74604583e-01 5.41381598e-01 9.48416054e-01 1.47322249e+00 1.10194445e+00 -3.85824174e-01 5.12292147e-01 3.81301850e-01 5.00054598e-01 9.78756368e-01 2.14153603e-01 -1.21279502e+00 5.03343523e-01 7.65244305e-01 5.01292348e-01 -1.32834566e+00 -9.79705334e-01 -6.81582019e-02 -6.38171315e-01 2.24512994e-01 1.85507033e-02 3.53908837e-01 -1.07619178e+00 1.59368992e+00 6.59363985e-01 2.14897513e-01 -4.64020148e-02 9.49412763e-01 9.01994646e-01 3.93716156e-01 -1.32512795e-02 3.89288068e-01 1.56504476e+00 -9.15790141e-01 -4.01395828e-01 -8.08132887e-01 5.04982591e-01 -2.00426057e-01 9.27209198e-01 1.08523242e-01 -8.27047706e-01 -6.79450154e-01 -1.01915634e+00 -4.17758465e-01 -7.40161359e-01 -3.67359519e-01 7.96919346e-01 2.83946961e-01 -1.28657961e+00 1.65681645e-01 -1.20261502e+00 -1.08252037e+00 3.03519573e-02 1.78723127e-01 -7.70018935e-01 -3.32323283e-01 -7.51783729e-01 1.05347323e+00 4.85854864e-01 -1.51750118e-01 -1.06205177e+00 -3.92843455e-01 -1.51028967e+00 -1.24238193e-01 4.43977892e-01 -1.39248908e+00 1.19408631e+00 -3.32615338e-02 -1.03197324e+00 9.22678888e-01 -4.40449864e-01 -3.32715183e-01 3.83322895e-01 -1.99247167e-01 -2.16068313e-01 9.59609449e-03 4.58581299e-01 7.45181501e-01 8.00602809e-02 -1.91760707e+00 -6.48463607e-01 -8.45978856e-01 3.50552261e-01 6.44922674e-01 5.75546682e-01 -5.29180825e-01 -6.62440300e-01 -2.50426959e-03 1.01995385e+00 -1.24455965e+00 -6.72788501e-01 -4.26330566e-02 -4.33056891e-01 2.24750340e-01 6.08234227e-01 -3.67923737e-01 7.54915178e-01 -2.00327754e+00 3.12049448e-01 2.96196252e-01 1.65616587e-01 -4.39696401e-01 2.28756055e-01 5.90281606e-01 4.83740985e-01 -9.94882584e-02 -1.97463214e-01 -1.05276239e+00 3.54041457e-01 6.79496050e-01 4.62435782e-02 6.29964650e-01 -7.13826418e-01 8.18959236e-01 -1.28564727e+00 -5.69383442e-01 7.66901612e-01 7.00607777e-01 -4.43451136e-01 2.03968231e-02 -1.82811424e-01 4.97554541e-01 -5.76874375e-01 7.97873378e-01 6.03544176e-01 -1.80375099e-01 2.13763490e-01 6.24283366e-02 -2.21978754e-01 4.33946311e-01 -1.53646755e+00 2.59098339e+00 -5.62208951e-01 3.31671566e-01 1.89507306e-01 -3.16031337e-01 7.14844525e-01 -8.28767288e-03 4.34821218e-01 -6.07240736e-01 6.57327101e-02 4.43575531e-02 -8.91826689e-01 -3.05222839e-01 8.67410779e-01 1.80289596e-01 -6.86447740e-01 -1.11094788e-01 -2.21362665e-01 -8.12875271e-01 -8.22870880e-02 3.13370109e-01 1.22341096e+00 4.62132990e-01 7.64390290e-01 -1.90861687e-01 2.29360059e-01 5.12906075e-01 3.75895977e-01 9.02457595e-01 -8.20718110e-02 7.49917626e-01 -2.20667288e-01 -7.00484931e-01 -9.56596434e-01 -1.52884924e+00 4.39906418e-02 7.18191504e-01 9.38967645e-01 -7.57274628e-01 -5.93563676e-01 -1.28246322e-01 1.33062318e-01 6.68799877e-01 -5.56889057e-01 1.26420259e-01 -6.22042120e-01 -2.47288331e-01 8.92316699e-02 5.35756111e-01 6.97507143e-01 -9.41541970e-01 -1.46279454e+00 2.00107649e-01 -4.61376816e-01 -1.49425030e+00 1.15570776e-01 5.31318560e-02 -6.80070817e-01 -1.14710808e+00 1.07622147e-01 -7.40796149e-01 7.46782422e-01 6.54647231e-01 1.47587228e+00 -2.04827055e-01 -5.18970549e-01 9.48678911e-01 -3.74537796e-01 -4.09917951e-01 -1.44896535e-02 -3.41310233e-01 3.63830268e-01 -5.55650830e-01 8.93842056e-02 -6.92442000e-01 -5.27155757e-01 4.22799796e-01 -2.49595746e-01 3.95662814e-01 4.35185768e-02 1.02835849e-01 1.01507425e+00 6.47656294e-03 -6.00543141e-01 -3.69189918e-01 1.01426944e-01 -4.71079975e-01 -4.99773830e-01 1.14733718e-01 5.15723228e-02 -1.77360773e-01 2.58100390e-01 3.46004730e-03 -9.65427995e-01 3.46689880e-01 1.10995673e-01 -1.68475017e-01 -6.28500104e-01 4.35072452e-01 -3.66302699e-01 3.37843299e-02 7.63004363e-01 1.38403224e-02 -4.44537371e-01 -4.11008626e-01 5.26098669e-01 2.58446902e-01 8.29972386e-01 -6.31468177e-01 6.05449677e-01 1.03199673e+00 3.85513008e-01 -8.09451520e-01 -5.65860510e-01 -8.28016758e-01 -1.23874581e+00 -1.75948277e-01 8.43882024e-01 -1.02010417e+00 -9.11753535e-01 4.28993225e-01 -1.23573554e+00 -6.33019209e-01 -3.35811257e-01 5.24506271e-01 -9.54899788e-01 1.06120110e-01 -3.52435261e-01 -6.93504632e-01 2.85701990e-01 -1.03858864e+00 1.58612251e+00 -1.11476354e-01 -4.33518708e-01 -9.89147484e-01 2.36792415e-01 2.39524066e-01 -5.60466498e-02 9.05776978e-01 4.08793241e-01 4.23825122e-02 -9.28851545e-01 -1.18334867e-01 8.37269798e-02 -5.05844951e-01 8.94714296e-02 -5.80703259e-01 -6.72083974e-01 -2.67832309e-01 -6.97487146e-02 3.50411758e-02 3.25212002e-01 3.33622903e-01 6.71207130e-01 4.99644093e-02 -1.02398324e+00 7.49026418e-01 1.63303792e+00 2.91864812e-01 6.25720382e-01 7.64763474e-01 9.02852416e-01 4.14710760e-01 8.33859026e-01 6.96660101e-01 1.29199708e+00 9.98206913e-01 9.17063177e-01 -6.08375110e-02 -7.06195906e-02 -5.96231520e-01 2.88300514e-02 5.15125871e-01 -2.21241236e-01 -1.70245320e-01 -1.35065258e+00 4.14225817e-01 -2.01426673e+00 -7.90162921e-01 -3.66773993e-01 2.24048662e+00 1.53601095e-02 4.75732163e-02 -1.52726918e-01 -3.64541747e-02 3.43308032e-01 1.15662985e-01 -3.98702025e-01 -7.06097037e-02 -1.44451195e-02 -2.61392981e-01 6.20238245e-01 9.04172599e-01 -1.07218432e+00 1.22188711e+00 5.80352545e+00 1.01774722e-01 -5.83767116e-01 1.39555439e-01 -1.63103625e-01 1.52847096e-01 8.61243680e-02 1.61769941e-01 -7.75449395e-01 -3.12826149e-02 6.81571484e-01 1.37022823e-01 5.82948387e-01 1.23827982e+00 1.43168673e-01 -5.55502892e-01 -1.15674293e+00 1.26855981e+00 2.44311005e-01 -1.12695587e+00 -2.53580481e-01 1.94461465e-01 5.06852925e-01 3.13548982e-01 -3.19107413e-01 2.67596960e-01 6.33370876e-01 -1.01463366e+00 1.53399670e+00 6.42115057e-01 5.47063887e-01 -4.15994048e-01 5.39390981e-01 7.21471846e-01 -1.78000998e+00 3.45179252e-02 -3.37574750e-01 -3.86959463e-01 5.42654872e-01 1.92281768e-01 -1.23291016e+00 8.49534810e-01 1.07315254e+00 6.05173528e-01 -6.28635108e-01 1.05011666e+00 -1.83335125e-01 -4.12360519e-01 -5.89884043e-01 -4.73493151e-02 1.00642756e-01 -1.53165132e-01 6.34567797e-01 1.14831758e+00 4.74521726e-01 2.48152107e-01 8.12127650e-01 7.18993187e-01 3.90697628e-01 -3.41391414e-01 -9.44423437e-01 7.08639622e-01 7.38876343e-01 9.33759987e-01 -9.76926088e-01 -4.87075061e-01 1.20525993e-01 1.02844942e+00 3.04095805e-01 3.56311440e-01 -8.59948277e-01 -1.07225096e-02 9.01871681e-01 4.43532348e-01 -1.43533155e-01 -1.05886984e+00 -3.57975245e-01 -9.57036078e-01 8.59108865e-02 -3.97029072e-01 1.06301747e-01 -1.36302900e+00 -5.20320356e-01 6.18389606e-01 2.97192037e-01 -1.17575371e+00 -2.43365020e-01 -5.56906104e-01 1.20541148e-01 6.42886281e-01 -1.05352116e+00 -1.33116961e+00 -1.21841729e+00 7.75468886e-01 5.50384104e-01 4.51659381e-01 1.06106293e+00 -1.17753223e-01 5.26333041e-02 -3.71116757e-01 -2.14775547e-01 -1.36360496e-01 1.00755826e-01 -1.35382700e+00 7.46190190e-01 5.60812593e-01 2.17936143e-01 6.87882900e-01 9.87492800e-01 -8.74116063e-01 -1.49207103e+00 -9.06119704e-01 8.56107652e-01 -1.04955506e+00 1.64287925e-01 -1.00289905e+00 -4.92992491e-01 1.17178702e+00 -4.77039635e-01 1.06704623e-01 1.13912992e-01 7.85972327e-02 -1.97159782e-01 1.14779398e-01 -1.42675567e+00 6.32284641e-01 2.00365257e+00 -4.61166799e-01 -6.65339112e-01 4.54635382e-01 9.11035180e-01 -1.10974407e+00 -5.64456344e-01 5.25185585e-01 4.72521633e-01 -1.43266380e+00 1.49698997e+00 -4.56955982e-04 -3.69196683e-01 -6.25734985e-01 -5.74386895e-01 -1.47354627e+00 -4.03939188e-01 -2.01598793e-01 -1.03327204e-02 5.44861555e-01 -2.18443573e-01 -5.78112960e-01 7.02064216e-01 4.68746096e-01 -5.59656024e-01 -3.44882429e-01 -1.09447896e+00 -8.13500404e-01 -7.84007967e-01 -8.71093452e-01 9.98338759e-01 7.19323158e-01 -7.40595758e-02 3.41189094e-02 1.58653751e-01 7.09408820e-01 6.88711822e-01 -2.46501360e-02 1.24987400e+00 -1.42083037e+00 1.57428995e-01 -2.03729309e-02 -1.09233510e+00 -1.21109891e+00 1.05890699e-01 -5.30733705e-01 4.03415978e-01 -2.37746191e+00 -2.02244043e-01 -8.01989555e-01 3.41712534e-01 3.23285103e-01 5.19571722e-01 -1.15954742e-01 1.36040181e-01 1.45260453e-01 -1.04619586e+00 5.80122590e-01 1.01250458e+00 -8.70751143e-02 -3.78191888e-01 -4.15126175e-01 1.98008642e-02 1.12242973e+00 5.31060219e-01 -2.45017424e-01 -4.29807395e-01 -5.59986472e-01 2.21517995e-01 1.15869485e-01 8.60491872e-01 -1.63323009e+00 5.07615209e-01 -2.75947571e-01 3.56219530e-01 -8.45654786e-01 1.17919755e+00 -1.02258587e+00 7.63219237e-01 5.71424365e-01 4.61106777e-01 3.76959354e-01 1.52398437e-01 6.45054281e-01 5.44582494e-02 2.14478284e-01 2.45088890e-01 -8.13440621e-01 -1.27110279e+00 2.62389302e-01 -4.34390493e-02 -2.90965438e-01 1.06333017e+00 -6.47730708e-01 -2.84760386e-01 -2.31106654e-01 -8.92974615e-01 3.22080851e-01 1.19785607e+00 6.47971153e-01 9.10229266e-01 -1.26806939e+00 -2.34560594e-01 2.02777982e-01 3.42641443e-01 6.75866127e-01 2.83203512e-01 3.83263499e-01 -1.05780387e+00 6.39976680e-01 -1.92406550e-01 -9.06129122e-01 -1.26472938e+00 5.10577679e-01 3.40696752e-01 -2.72722002e-02 -9.08183813e-01 7.42890358e-01 4.03174579e-01 -8.53938580e-01 2.23852977e-01 -7.73701489e-01 2.21053101e-02 -4.86357212e-01 2.10165024e-01 5.60812593e-01 -4.22441512e-02 -1.31685126e+00 -7.62283385e-01 9.09667671e-01 9.36393738e-01 -2.69906074e-01 1.11911178e+00 -6.16713405e-01 -2.85483479e-01 9.06203389e-01 7.96615839e-01 -1.26500214e-02 -1.12345207e+00 -1.55247509e-01 -1.15252517e-01 -6.30328834e-01 -4.25245315e-01 -6.71349049e-01 -1.02661632e-01 5.35593688e-01 5.29518008e-01 1.12560198e-01 6.95193410e-01 4.79975164e-01 5.39270520e-01 5.66773713e-01 1.80073297e+00 -8.39947701e-01 1.67405725e-01 8.21453512e-01 1.13779759e+00 -9.76163924e-01 6.92631751e-02 -8.14519763e-01 -4.44083035e-01 6.73591852e-01 5.00464916e-01 1.06391206e-01 4.35025394e-01 2.19937727e-01 -1.35629356e-01 -5.47270119e-01 -1.96267903e-01 -3.35027456e-01 4.43006977e-02 1.00031757e+00 -2.45401844e-01 5.66433907e-01 4.80340153e-01 2.09044382e-01 -9.59074855e-01 -3.88712913e-01 3.47237349e-01 1.36415315e+00 -7.42934406e-01 -5.64005256e-01 -8.70604455e-01 -2.64098108e-01 5.28722644e-01 3.05615753e-01 -2.37022296e-01 9.81395006e-01 6.08821571e-01 9.71745372e-01 2.44516060e-01 -4.57036614e-01 7.55320549e-01 -1.60905436e-01 8.16271305e-01 -9.54109013e-01 -1.46059766e-01 -3.74015898e-01 1.43477365e-01 -1.15164089e+00 -2.98865855e-01 -7.64089286e-01 -1.92558074e+00 -1.92722976e-01 2.92512685e-01 -1.53253153e-01 1.17481673e+00 7.39496946e-01 4.42631245e-01 4.89067793e-01 4.07031104e-02 -1.59515035e+00 2.69191742e-01 -6.64588332e-01 -6.94076478e-01 6.58396423e-01 3.40608239e-01 -9.94876862e-01 -1.50958627e-01 -5.33475280e-02]
[4.850970268249512, 0.3603166937828064]
6443eb50-79f3-4480-ba97-002581a6efb4
clicking-matters-towards-interactive-human
2111.06162
null
https://arxiv.org/abs/2111.06162v2
https://arxiv.org/pdf/2111.06162v2.pdf
Clicking Matters:Towards Interactive Human Parsing
In this work, we focus on Interactive Human Parsing (IHP), which aims to segment a human image into multiple human body parts with guidance from users' interactions. This new task inherits the class-aware property of human parsing, which cannot be well solved by traditional interactive image segmentation approaches that are generally class-agnostic. To tackle this new task, we first exploit user clicks to identify different human parts in the given image. These clicks are subsequently transformed into semantic-aware localization maps, which are concatenated with the RGB image to form the input of the segmentation network and generate the initial parsing result. To enable the network to better perceive user's purpose during the correction process, we investigate several principal ways for the refinement, and reveal that random-sampling-based click augmentation is the best way for promoting the correction effectiveness. Furthermore, we also propose a semantic-perceiving loss (SP-loss) to augment the training, which can effectively exploit the semantic relationships of clicks for better optimization. To the best knowledge, this work is the first attempt to tackle the human parsing task under the interactive setting. Our IHP solution achieves 85\% mIoU on the benchmark LIP, 80\% mIoU on PASCAL-Person-Part and CIHP, 75\% mIoU on Helen with only 1.95, 3.02, 2.84 and 1.09 clicks per class respectively. These results demonstrate that we can simply acquire high-quality human parsing masks with only a few human effort. We hope this work can motivate more researchers to develop data-efficient solutions to IHP in the future.
['Yunchao Wei', 'Yidong Li', 'Songhe Feng', 'Congyan Lang', 'Liqian Liang', 'Yutong Gao']
2021-11-11
null
null
null
null
['human-parsing']
['computer-vision']
[ 6.38349414e-01 4.10896122e-01 -4.07029130e-02 -5.28662145e-01 -8.34675133e-01 -4.31119889e-01 4.23385650e-02 -4.43535522e-02 -6.40237093e-01 4.60244507e-01 -1.46701396e-01 -1.32950202e-01 2.78389394e-01 -6.69851661e-01 -9.03065979e-01 -4.35462952e-01 4.17938054e-01 3.20627183e-01 6.68344080e-01 -2.31464636e-02 1.29538417e-01 -2.01742239e-02 -1.56455374e+00 1.72492400e-01 1.07750213e+00 1.05828989e+00 4.88170475e-01 7.27301240e-01 -1.18420944e-01 2.37232640e-01 -6.08115494e-01 -7.15891480e-01 2.21323490e-01 -4.35141325e-01 -7.90548325e-01 3.88031453e-02 5.80626369e-01 -3.22318941e-01 9.84753445e-02 1.19873226e+00 6.68618798e-01 6.22290485e-02 3.84431660e-01 -1.20664060e+00 -4.11935568e-01 5.36242783e-01 -8.70061457e-01 -8.62937272e-02 4.79714006e-01 2.83712029e-01 1.06399429e+00 -5.31716704e-01 5.52476883e-01 1.22297156e+00 5.84370613e-01 9.40119386e-01 -8.26346874e-01 -8.57331514e-01 2.39189908e-01 6.10181838e-02 -1.05565333e+00 -9.19948444e-02 7.90859640e-01 -1.97259769e-01 3.75874519e-01 5.15359521e-01 7.05535173e-01 8.87772322e-01 -3.30348551e-01 1.23854673e+00 1.18398023e+00 -4.00740027e-01 -6.82110935e-02 1.84131101e-01 3.07883620e-01 8.90408635e-01 9.00148973e-02 -1.89514413e-01 -5.56793034e-01 2.56889284e-01 6.99228525e-01 -1.51946962e-01 -2.74761200e-01 -2.04688698e-01 -9.00979817e-01 6.91076040e-01 5.69935858e-01 -3.59430164e-02 -2.60359138e-01 1.70270801e-01 1.87636182e-01 -3.20628881e-01 2.83226073e-01 3.81662935e-01 -5.81735551e-01 -2.02143103e-01 -9.43545520e-01 2.50787526e-01 5.74149191e-01 9.48732138e-01 6.84113622e-01 -3.18975180e-01 -3.47331256e-01 8.82938802e-01 3.20884466e-01 5.36102533e-01 1.70971647e-01 -1.13565290e+00 6.85498476e-01 6.95954204e-01 1.64108947e-01 -8.14112484e-01 -4.23010021e-01 -3.12342435e-01 -5.27691901e-01 6.17964864e-02 7.07567036e-01 -1.62086114e-01 -1.17076528e+00 1.80243671e+00 5.68586349e-01 5.58330715e-02 -2.30870679e-01 1.05061269e+00 8.82197559e-01 3.95930290e-01 4.83335555e-01 4.77845855e-02 1.87093449e+00 -1.15530658e+00 -7.27122962e-01 -4.26602453e-01 4.28319186e-01 -8.22635353e-01 1.68327093e+00 4.60823238e-01 -1.11646688e+00 -5.75310946e-01 -1.02788401e+00 -1.04386330e-01 -2.37653956e-01 3.98711801e-01 7.26644754e-01 8.42353106e-01 -6.32187545e-01 5.76844633e-01 -8.59209657e-01 -3.22377294e-01 6.40841484e-01 4.13684428e-01 -6.29146025e-02 -7.25493059e-02 -1.09730387e+00 3.92431647e-01 3.15548211e-01 3.95780690e-02 -4.73831922e-01 -6.54317439e-01 -7.32571363e-01 -1.55142307e-01 8.39364767e-01 -6.91267490e-01 1.26988959e+00 -1.01862359e+00 -1.42914522e+00 9.27740693e-01 -1.78956702e-01 -3.10660452e-01 7.51782596e-01 -5.51226676e-01 -2.49596853e-02 2.19139218e-01 3.40208799e-01 1.10584688e+00 6.67669535e-01 -1.42793536e+00 -8.85377824e-01 -4.87911493e-01 2.71469921e-01 3.59867334e-01 -2.32131407e-01 -2.66630109e-02 -1.16921175e+00 -7.26577878e-01 2.25017458e-01 -1.13082922e+00 -1.58685803e-01 -2.01274361e-02 -9.13333595e-01 -3.31629425e-01 6.55513883e-01 -9.08808708e-01 1.15245152e+00 -2.09432936e+00 1.05606064e-01 5.19252270e-02 1.24754161e-01 4.39166605e-01 -2.88029462e-02 -1.70052707e-01 2.40743563e-01 3.17369789e-01 -4.46732223e-01 -7.00631559e-01 3.10673565e-02 1.15978658e-01 1.08151294e-01 4.74180765e-02 3.45472097e-02 9.24637854e-01 -7.70269334e-01 -7.88731515e-01 9.81498063e-02 4.04168457e-01 -7.34320283e-01 3.24687570e-01 -3.10819387e-01 5.60102224e-01 -4.98562813e-01 6.28182471e-01 8.31748962e-01 -1.19106501e-01 -1.31328240e-01 -4.76800561e-01 1.72733739e-01 -2.70541832e-02 -1.15922451e+00 1.80108976e+00 -3.20278108e-01 2.91202515e-01 9.80091766e-02 -7.83476770e-01 5.06417334e-01 3.52799296e-02 2.86610097e-01 -8.04310143e-01 1.70127690e-01 5.19769313e-03 -1.44887730e-01 -5.34544766e-01 4.47422355e-01 2.33844250e-01 -1.73299760e-01 2.64280945e-01 -1.67452991e-01 -1.76950963e-03 2.15618938e-01 1.82127178e-01 8.23451400e-01 2.51949668e-01 -5.13071604e-02 1.30685344e-01 3.38358223e-01 -1.05332695e-01 6.64711714e-01 9.15800869e-01 -4.29263294e-01 7.43996143e-01 4.38873380e-01 -4.26424630e-02 -8.09185266e-01 -9.19101417e-01 9.59035382e-02 1.31735873e+00 4.72227454e-01 -4.07062292e-01 -1.43305206e+00 -9.91539896e-01 -3.53567839e-01 5.67796528e-01 -4.83855337e-01 1.12961322e-01 -6.59332335e-01 -7.87391305e-01 5.96837819e-01 7.39195287e-01 9.76714790e-01 -1.23356903e+00 -7.75993466e-01 -2.94191632e-02 -4.56915617e-01 -1.25206125e+00 -6.08275235e-01 -9.87374857e-02 -6.91580355e-01 -1.23317480e+00 -8.48966300e-01 -7.69360542e-01 7.37867177e-01 -2.03732215e-02 9.28543270e-01 2.68935174e-01 -4.89248157e-01 3.33664507e-01 -4.74103779e-01 -4.16229099e-01 -8.15852433e-02 1.10418707e-01 -3.30526352e-01 -1.36235312e-01 2.43826941e-01 -2.76787549e-01 -1.01881814e+00 4.26591635e-01 -6.38671875e-01 4.05310929e-01 6.61043048e-01 6.56383038e-01 7.80368686e-01 5.40361479e-02 1.44522294e-01 -1.27144945e+00 4.18774843e-01 -7.48131424e-03 -4.16600704e-01 2.38257349e-01 -4.44474220e-01 -7.82424510e-02 3.48169982e-01 -4.89005208e-01 -1.17884731e+00 3.95407766e-01 -3.06796670e-01 -1.19845130e-01 -3.29463899e-01 -8.17979649e-02 -5.16600609e-01 -3.49626541e-02 4.24564511e-01 7.44670406e-02 -4.20914710e-01 -5.63668966e-01 5.56235254e-01 5.47044575e-01 6.95389748e-01 -5.81045747e-01 6.95727408e-01 3.33177000e-01 -2.90075719e-01 -5.68228304e-01 -1.03968668e+00 -5.93164146e-01 -4.79643524e-01 -1.27580717e-01 1.40089035e+00 -7.03971624e-01 -9.06643629e-01 4.39903080e-01 -1.12581813e+00 -4.06986058e-01 -3.75833437e-02 6.07323423e-02 -4.32239175e-01 5.75553536e-01 -6.40578926e-01 -8.95597339e-01 -4.52333093e-01 -1.27841079e+00 1.36463702e+00 6.09068274e-01 -1.45039335e-01 -4.24098372e-01 -4.48363692e-01 1.07941091e+00 3.43849286e-02 1.59438327e-01 7.21572995e-01 -5.29753804e-01 -7.03204870e-01 -1.95277035e-01 -5.68999171e-01 3.47524971e-01 -1.20230414e-01 -2.50554591e-01 -1.02921915e+00 1.11416290e-02 -3.18090856e-01 -2.14024261e-01 8.22993517e-01 3.53507847e-01 1.56821263e+00 -2.58889288e-01 -3.51056606e-01 6.26759052e-01 1.15435624e+00 2.45187044e-01 6.58759594e-01 1.61408767e-01 1.04444730e+00 6.81198299e-01 1.06656814e+00 3.70449394e-01 4.95584607e-01 8.14644694e-01 4.85188961e-01 -5.14660656e-01 -3.70897055e-01 -4.40869421e-01 -6.31771907e-02 3.90103549e-01 -3.14206243e-01 -2.35688195e-01 -6.49777591e-01 3.06333393e-01 -1.73824894e+00 -5.56508064e-01 -2.25238383e-01 2.09414554e+00 9.73344207e-01 3.55295986e-01 2.09215239e-01 1.66822940e-01 8.00979733e-01 -5.24203468e-04 -4.50882912e-01 -1.58496220e-02 2.49691337e-01 4.65051293e-01 6.62088215e-01 3.80850405e-01 -1.41489053e+00 1.25051808e+00 5.00895548e+00 9.88856494e-01 -8.29929233e-01 2.42085740e-01 6.85427547e-01 2.04829752e-01 -3.92455561e-03 -8.14274326e-02 -1.00691283e+00 6.22526109e-01 5.74841797e-01 4.19838458e-01 3.62715930e-01 8.79710019e-01 1.79115295e-01 -3.88939053e-01 -8.72972786e-01 1.00855696e+00 -1.13501862e-01 -7.18113720e-01 -1.49521604e-01 -1.85854852e-01 4.13264632e-01 -4.79142338e-01 1.81294352e-01 3.10385942e-01 1.08227238e-01 -8.53159368e-01 6.11493409e-01 3.54257882e-01 6.32363021e-01 -6.88606083e-01 7.51180589e-01 3.52911949e-01 -1.19803548e+00 1.32034138e-01 -1.40256956e-01 2.58699536e-01 2.59799212e-01 3.66042495e-01 -8.27287972e-01 3.78673553e-01 9.77413237e-01 2.30146021e-01 -7.59969234e-01 1.01891792e+00 -5.30876458e-01 6.92746758e-01 -4.01390642e-01 -1.05235711e-01 3.98899801e-02 9.50445011e-02 4.11744207e-01 1.14226401e+00 1.40591323e-01 1.63123384e-01 3.07317317e-01 8.12351644e-01 -1.66199252e-01 3.02745968e-01 2.95232888e-02 2.10289583e-02 3.43332440e-01 1.27679384e+00 -9.89762008e-01 -3.40063661e-01 -1.11162528e-01 1.52551472e+00 1.43588349e-01 3.87086630e-01 -1.21191251e+00 -4.87353265e-01 2.65757769e-01 2.26410076e-01 2.58074760e-01 7.78756514e-02 -4.63606089e-01 -9.06975448e-01 2.10402042e-01 -8.73694360e-01 3.40444207e-01 -7.83568203e-01 -8.34637582e-01 4.63793188e-01 -5.57878390e-02 -7.70816505e-01 7.57314041e-02 -5.41662931e-01 -3.97675484e-01 6.37339294e-01 -1.35225284e+00 -1.27091777e+00 -6.85514450e-01 4.50469404e-01 7.49216318e-01 3.19425821e-01 5.63465178e-01 4.68153447e-01 -7.16915965e-01 9.42213058e-01 -6.42256200e-01 2.69952893e-01 7.02665925e-01 -1.51961255e+00 3.61000687e-01 7.83061922e-01 1.40392125e-01 4.28642094e-01 8.52969408e-01 -7.39074588e-01 -9.17456925e-01 -9.82970476e-01 6.30183756e-01 -3.78337830e-01 9.92851928e-02 -3.27262819e-01 -7.48922884e-01 4.16758478e-01 1.03425235e-01 -1.60175577e-01 5.22435844e-01 6.79751858e-02 -2.27123484e-01 -9.70637053e-03 -1.22652459e+00 7.76830256e-01 1.31221974e+00 -2.78860062e-01 -3.44883233e-01 3.18535328e-01 9.89398241e-01 -6.06522977e-01 -6.79282844e-01 5.66640079e-01 4.23265010e-01 -8.93596590e-01 1.05464256e+00 -3.09839696e-01 3.36960763e-01 -2.58196175e-01 1.20506145e-01 -7.94143677e-01 1.53704017e-01 -5.58375418e-01 2.67932326e-01 1.38651371e+00 4.77612495e-01 -3.17237020e-01 1.18099403e+00 7.47334063e-01 -4.76095937e-02 -8.40924561e-01 -6.74665213e-01 -2.93077409e-01 -2.05115259e-01 -5.70263684e-01 4.54337329e-01 5.12932718e-01 -4.20393527e-01 1.22936815e-01 -4.80081111e-01 3.10607255e-01 8.19896817e-01 -1.09835960e-01 9.02956605e-01 -9.47959900e-01 -5.41288555e-01 -2.89264828e-01 -1.03555150e-01 -1.20355666e+00 -1.13414928e-01 -4.37330902e-01 3.53397667e-01 -1.38312399e+00 3.59833777e-01 -6.99455261e-01 -7.96416700e-02 6.69237614e-01 -6.37833953e-01 5.66716731e-01 3.96041930e-01 5.97642153e-04 -7.86637664e-01 1.63816661e-01 1.48307419e+00 9.03831050e-03 -3.40980381e-01 3.07801276e-01 -7.22105801e-01 9.72441316e-01 7.40623057e-01 -4.30275053e-01 -4.12646294e-01 -2.65693843e-01 -3.47312018e-02 -3.29268537e-02 5.72446406e-01 -1.02950144e+00 1.63039997e-01 6.54769083e-03 2.63185799e-01 -6.20876431e-01 4.34742302e-01 -6.07684791e-01 -1.12117194e-01 3.51310492e-01 -2.88670599e-01 -3.42455328e-01 4.02333774e-02 6.50726974e-01 -3.28962924e-03 -2.46423200e-01 8.42006266e-01 -3.45349669e-01 -8.62454534e-01 2.80713648e-01 2.26329684e-01 2.14372888e-01 9.13182080e-01 -2.65304804e-01 -1.05274059e-01 -2.40579218e-01 -9.79666829e-01 3.60904157e-01 2.30747402e-01 2.99484253e-01 4.68271911e-01 -9.43103373e-01 -3.00192654e-01 -3.87575030e-02 -6.47498518e-02 2.17411175e-01 3.80168706e-01 8.20565999e-01 -5.53305089e-01 1.10827364e-01 1.65351969e-03 -5.89916408e-01 -1.67261863e+00 3.10963511e-01 2.23194644e-01 -3.05867106e-01 -5.87655663e-01 1.15602815e+00 3.70185047e-01 -3.69896948e-01 6.05720520e-01 -2.58345187e-01 -1.94834426e-01 -7.17874197e-03 2.86193520e-01 2.95517474e-01 -2.43230700e-01 -4.55130875e-01 -2.93528050e-01 7.84928024e-01 -2.46456452e-02 -2.06699505e-01 9.55490291e-01 -1.48233578e-01 2.57743895e-01 4.96688820e-02 9.73136365e-01 8.66469964e-02 -1.37205100e+00 9.19883996e-02 -7.32379407e-02 -4.37758565e-01 -2.69211054e-01 -1.18841279e+00 -1.30663836e+00 1.06045866e+00 8.50261569e-01 -7.34474789e-03 1.29115021e+00 1.42844051e-01 1.20597172e+00 -1.42491236e-02 2.98193753e-01 -1.31614137e+00 2.15250954e-01 -9.74649389e-04 6.41868234e-01 -1.31280458e+00 -2.01092437e-01 -1.08133411e+00 -8.12958479e-01 7.41758227e-01 9.38553393e-01 1.41192719e-01 1.74150974e-01 8.37595314e-02 7.28474483e-02 -5.37097007e-02 -8.55747238e-02 -6.00610733e-01 3.71964961e-01 6.04920149e-01 3.81144643e-01 2.03822449e-01 -4.77157861e-01 1.00886190e+00 -3.46140027e-01 -8.79905596e-02 4.84364405e-02 7.30304062e-01 -5.18285275e-01 -1.25749350e+00 -2.59946018e-01 2.45582208e-01 -6.44143462e-01 -2.99067814e-02 -3.71671826e-01 7.94906259e-01 5.40012240e-01 8.77886474e-01 -2.82053590e-01 -2.74007350e-01 5.87243438e-01 1.16216257e-01 4.16961670e-01 -5.97745299e-01 -5.76349378e-01 2.00587079e-01 1.63557976e-01 -8.13950241e-01 -3.89750153e-01 -4.33527350e-01 -1.51593637e+00 2.17539947e-02 -4.06597644e-01 -6.25347197e-02 6.84050679e-01 7.58905888e-01 1.54726550e-01 6.18527889e-01 8.93232822e-02 -7.48357475e-01 -2.68981129e-01 -7.13180006e-01 -3.24136376e-01 7.93413997e-01 -1.35214344e-01 -6.44210935e-01 -3.30477297e-01 1.25912189e-01]
[9.071419715881348, 0.154776468873024]
b154a2e6-dc25-4ae1-a169-25790e9e395b
skeleton-based-action-analysis-for-adhd
2304.09751
null
https://arxiv.org/abs/2304.09751v1
https://arxiv.org/pdf/2304.09751v1.pdf
Skeleton-based action analysis for ADHD diagnosis
Attention Deficit Hyperactivity Disorder (ADHD) is a common neurobehavioral disorder worldwide. While extensive research has focused on machine learning methods for ADHD diagnosis, most research relies on high-cost equipment, e.g., MRI machine and EEG patch. Therefore, low-cost diagnostic methods based on the action characteristics of ADHD are desired. Skeleton-based action recognition has gained attention due to the action-focused nature and robustness. In this work, we propose a novel ADHD diagnosis system with a skeleton-based action recognition framework, utilizing a real multi-modal ADHD dataset and state-of-the-art detection algorithms. Compared to conventional methods, the proposed method shows cost-efficiency and significant performance improvement, making it more accessible for a broad range of initial ADHD diagnoses. Through the experiment results, the proposed method outperforms the conventional methods in accuracy and AUC. Meanwhile, our method is widely applicable for mass screening.
['Syed Mohsen Naqvi', 'Rajesh Nair', 'Yi Li', 'YiChun Li']
2023-04-14
null
null
null
null
['skeleton-based-action-recognition', 'action-analysis', 'action-recognition-in-videos']
['computer-vision', 'computer-vision', 'computer-vision']
[ 3.88603836e-01 -2.52889782e-01 -5.64484000e-01 -2.44180173e-01 -6.19276345e-01 1.35521933e-01 7.02138841e-02 2.36462682e-01 -1.29811361e-01 5.67063093e-01 -8.48499611e-02 -2.74454318e-02 -5.22527874e-01 -6.28364503e-01 -2.28476226e-01 -7.13601887e-01 1.31630272e-01 5.84302485e-01 3.26578617e-01 3.81381601e-01 4.73538011e-01 5.14599502e-01 -1.78326488e+00 6.85407519e-02 1.64257264e+00 9.66113746e-01 5.62486172e-01 -3.45141254e-02 2.49705128e-02 4.51469868e-01 -4.84781146e-01 -1.31923378e-01 -1.83914065e-01 -5.50936997e-01 -4.53240395e-01 2.66289920e-01 3.46437275e-01 -5.45046031e-01 -3.72515082e-01 1.23659492e+00 7.42246866e-01 1.58804163e-01 5.97503603e-01 -1.09303284e+00 -5.74814916e-01 2.82959282e-01 -9.09595847e-01 5.05135655e-01 4.59053338e-01 1.29184946e-01 5.87382555e-01 -3.26413929e-01 4.40963730e-02 9.78312671e-01 2.60023803e-01 8.02298129e-01 -8.40408325e-01 -1.10053229e+00 3.74794036e-01 1.00318122e+00 -1.10521746e+00 -1.90838218e-01 5.89148700e-01 -4.10570920e-01 8.47987771e-01 1.28130883e-01 1.09771991e+00 1.01176488e+00 1.38988078e-01 9.48553145e-01 1.17417884e+00 1.15728145e-03 4.01763678e-01 -4.96953815e-01 7.35496208e-02 4.98798549e-01 7.70700097e-01 -3.16698760e-01 -3.16575676e-01 7.88993910e-02 7.29705870e-01 2.63330579e-01 -2.32170194e-01 -3.04571331e-01 -8.39231372e-01 4.38502997e-01 1.97827697e-01 4.05836254e-01 -5.73510945e-01 -9.74034369e-02 4.91048127e-01 -5.84602654e-01 2.50938863e-01 1.25687540e-01 -1.88667513e-02 -5.13330340e-01 -1.02654350e+00 2.71503925e-01 -6.85085580e-02 6.53486311e-01 -1.95271105e-01 2.63697505e-01 -1.87698483e-01 1.13263524e+00 3.73914421e-01 5.58122039e-01 8.41129839e-01 -7.22458661e-01 4.93332326e-01 9.30050433e-01 -4.85899955e-01 -8.72585714e-01 -6.14209890e-01 -4.96743321e-01 -7.98610389e-01 -5.19887544e-02 1.46060821e-03 2.14232937e-01 -1.00680006e+00 1.43967772e+00 4.86963034e-01 5.70669115e-01 -3.53394687e-01 9.91197109e-01 7.42310166e-01 1.48896828e-01 2.50788420e-01 -4.27876025e-01 1.34165525e+00 -8.42420578e-01 -7.82816708e-01 -1.12242609e-01 5.37960291e-01 -2.24133223e-01 6.15123987e-01 8.68874311e-01 -1.04619014e+00 -1.20697103e-01 -9.45035160e-01 2.86966741e-01 6.20835461e-02 4.84480768e-01 8.72945070e-01 7.31405437e-01 -4.84940320e-01 3.17616791e-01 -1.21338606e+00 -5.68342030e-01 7.57080674e-01 6.59111917e-01 -3.46153647e-01 -2.03099817e-01 -6.47139907e-01 8.85953307e-01 2.59892941e-01 -1.05598763e-01 -8.22615147e-01 -4.73973960e-01 -6.41182780e-01 1.58713520e-01 3.39984953e-01 -4.34793234e-01 1.36499429e+00 -3.79006147e-01 -1.65328300e+00 4.73434359e-01 -5.35887107e-02 -3.21667433e-01 2.93017238e-01 -1.57221511e-01 -6.77290559e-01 7.19524443e-01 1.20066926e-01 3.79760951e-01 7.97929704e-01 -4.61693019e-01 -8.68903220e-01 -1.05253744e+00 -9.85947922e-02 2.62157530e-01 -5.06182730e-01 1.22258939e-01 -9.45234373e-02 -4.99395490e-01 5.18757880e-01 -9.02812421e-01 5.96918724e-02 -1.87658740e-03 -2.31063887e-01 -3.74636620e-01 8.21748853e-01 -6.49693668e-01 1.34381783e+00 -1.88015842e+00 -1.41230095e-02 -2.93510258e-02 3.59471172e-01 8.31139386e-01 1.91407084e-01 5.19846156e-02 -3.36558223e-01 -1.13010742e-01 -1.63596436e-01 3.53018016e-01 -2.48172864e-01 -1.36388168e-01 3.00256461e-01 8.19097459e-01 8.07490125e-02 4.48491752e-01 -8.43597889e-01 -4.44612682e-01 5.08152366e-01 2.90421128e-01 -5.31742930e-01 9.54732448e-02 1.12618744e-01 5.54121435e-01 -8.87775123e-01 9.74795222e-01 7.16275156e-01 -4.60322499e-02 -2.04969365e-02 1.49466144e-02 -8.57250094e-02 3.27188045e-01 -1.15212560e+00 1.52581584e+00 -1.00074396e-01 9.82360244e-02 2.19905772e-03 -1.28605890e+00 7.49983847e-01 5.37077367e-01 6.42436802e-01 -1.02534163e+00 2.35489935e-01 5.57082653e-01 4.96741086e-01 -1.02547348e+00 -2.95595109e-01 1.22065499e-01 5.87306976e-01 2.43228093e-01 -1.46016076e-01 9.53644961e-02 3.74342442e-01 -1.00837439e-01 1.45798278e+00 1.03637621e-01 5.34420848e-01 1.59647450e-01 6.10019147e-01 -2.78927237e-01 6.74168348e-01 2.37992913e-01 -3.54267895e-01 4.22888845e-01 1.76378280e-01 3.83344181e-02 -5.35443366e-01 -6.56651080e-01 -5.99510968e-01 5.52374661e-01 2.08979741e-01 -1.52223945e-01 -9.05377507e-01 -5.12996674e-01 -5.57309017e-02 6.56118512e-01 -2.93444246e-01 -4.03192818e-01 -3.27227056e-01 -6.28076434e-01 5.18466294e-01 7.65754223e-01 8.56691539e-01 -9.98540163e-01 -5.91318250e-01 2.68752337e-01 1.47731677e-01 -8.25734973e-01 -6.99313432e-02 -3.22405279e-01 -1.25195634e+00 -1.24946964e+00 -1.07701933e+00 -7.27331758e-01 6.15458906e-01 6.28283322e-01 2.01274440e-01 5.52209280e-02 -5.08343160e-01 3.70464563e-01 -3.41842681e-01 -4.59730297e-01 2.86391914e-01 1.32203987e-02 3.01945210e-01 6.23716414e-02 4.82080847e-01 -8.06305408e-01 -8.76607656e-01 1.35050878e-01 -6.03532732e-01 -1.11316919e-01 1.09717166e+00 4.84120429e-01 6.66817009e-01 1.34749055e-01 1.01933897e+00 -5.20976007e-01 7.40718126e-01 -4.98884201e-01 -5.81234396e-01 1.85595438e-01 -8.65274787e-01 -3.43439937e-01 3.47708195e-01 -6.07874691e-01 -1.14242721e+00 -8.74874964e-02 -2.28248850e-01 -4.12276149e-01 -5.12354374e-01 4.03445959e-01 -4.81709719e-01 -8.03737789e-02 2.30883926e-01 2.83617914e-01 -1.99410170e-01 -7.10565448e-01 -2.33196974e-01 9.07833338e-01 5.73932052e-01 -4.54651266e-01 -5.49910292e-02 1.49780899e-01 1.17939889e-01 -8.94091845e-01 -5.82250059e-01 -6.22674763e-01 -2.26593837e-01 -3.23451519e-01 9.76911366e-01 -7.33759284e-01 -8.77700150e-01 6.95748389e-01 -1.02398741e+00 3.54910940e-01 3.42569739e-01 1.26397443e+00 -4.00291234e-01 3.92709166e-01 -4.01087850e-01 -9.15915668e-01 -5.60020089e-01 -1.29505014e+00 1.01778078e+00 4.50274438e-01 -1.25416577e-01 -5.06180584e-01 2.33134702e-01 8.00042510e-01 5.94308451e-02 -1.54873535e-01 1.03786647e+00 -7.36717761e-01 -3.61014396e-01 -3.32490712e-01 -4.51016337e-01 1.72018170e-01 2.17496261e-01 -3.50370407e-02 -7.57120192e-01 -1.38788419e-02 1.93675697e-01 -2.90725499e-01 4.98937130e-01 5.89584410e-01 1.71350527e+00 8.71507544e-03 -5.72895050e-01 3.63993317e-01 9.74573255e-01 9.71003294e-01 7.53237188e-01 4.56763178e-01 7.88421273e-01 3.71933728e-01 7.23717451e-01 4.73556399e-01 2.59408712e-01 6.20321035e-01 6.22290969e-01 1.95467293e-01 -1.90313831e-02 9.26947370e-02 -3.28044519e-02 9.69915926e-01 -4.13939685e-01 -9.30414274e-02 -9.74702239e-01 5.38013399e-01 -1.90271926e+00 -1.14587057e+00 -4.18189168e-01 2.14483047e+00 5.69357634e-01 6.43982142e-02 1.46948710e-01 3.80547285e-01 9.91960287e-01 -2.29558200e-01 -1.04389131e+00 -3.14422578e-01 3.69270951e-01 3.64160001e-01 5.04168533e-02 -5.62898993e-01 -8.07580471e-01 2.89420217e-01 5.99416208e+00 1.05950308e+00 -1.25410426e+00 2.81268358e-01 8.65281820e-02 -4.23909992e-01 1.60008505e-01 -4.95968312e-01 -6.33519471e-01 7.85202622e-01 6.52369916e-01 -6.18323535e-02 3.74607950e-01 8.40344191e-01 4.48726386e-01 -3.34663957e-01 -7.94854522e-01 1.29817307e+00 2.18580276e-01 -9.11212802e-01 -1.32804766e-01 3.00409764e-01 4.73213106e-01 -2.53816634e-01 3.57387848e-02 1.86638296e-01 -4.60202485e-01 -9.05635953e-01 3.29088092e-01 2.69360691e-01 8.17450166e-01 -7.21221149e-01 6.10635817e-01 3.23624074e-01 -1.14930022e+00 -2.74360418e-01 -1.48259819e-01 -5.70463054e-02 -1.15874827e-01 5.46092093e-01 -4.44305420e-01 4.18404490e-01 5.78138649e-01 8.76813233e-01 -4.19833571e-01 1.78970087e+00 -4.60421026e-01 8.05831969e-01 -8.04622546e-02 6.56715184e-02 1.26982436e-01 -3.51550162e-01 4.05511945e-01 7.16703832e-01 6.76631093e-01 5.38670182e-01 6.95535764e-02 5.78114092e-01 8.28001648e-02 2.66449064e-01 -5.84776819e-01 -3.15591246e-01 3.37861240e-01 1.30212557e+00 -7.39723086e-01 -2.72275925e-01 -6.49253368e-01 4.20145333e-01 2.25567192e-01 -1.22247286e-01 -1.02799821e+00 -3.35923493e-01 3.40009183e-01 1.93041608e-01 2.31605902e-01 2.18318105e-01 -3.16318035e-01 -1.08899057e+00 2.73297336e-02 -1.23375893e+00 2.27314889e-01 -7.96872139e-01 -9.30414736e-01 1.48290873e-01 1.02540679e-01 -1.48795986e+00 5.64357359e-03 -4.61957008e-01 -9.54877079e-01 4.23458397e-01 -9.96461451e-01 -8.76113832e-01 -3.00074458e-01 3.78112137e-01 7.32379436e-01 -1.30844340e-01 5.59685886e-01 6.85376942e-01 -1.38480186e+00 3.41749310e-01 8.14849287e-02 -3.35552782e-01 4.63374823e-01 -7.79363513e-01 -4.28157091e-01 6.31544709e-01 -2.88790524e-01 6.09584510e-01 2.67048895e-01 -9.24424648e-01 -1.33065701e+00 -8.85788202e-01 3.59360665e-01 4.20006663e-01 6.05753601e-01 2.53854185e-01 -1.03132033e+00 3.29081982e-01 9.97128189e-02 -3.19296002e-01 5.74648678e-01 -1.07315302e-01 9.60957035e-02 -2.35064879e-01 -1.32430792e+00 4.46742505e-01 1.30637848e+00 -1.55848354e-01 -8.62836063e-01 3.85891259e-01 2.48363204e-02 -5.00966728e-01 -7.21296549e-01 7.17600405e-01 7.99728334e-01 -7.37839341e-01 5.66606581e-01 -4.06306446e-01 3.72530758e-01 -1.24078907e-01 2.89525717e-01 -1.22743511e+00 -3.54832292e-01 8.86308253e-02 -3.07860315e-01 1.29117393e+00 -9.18368995e-02 -7.16742337e-01 6.92802548e-01 6.56103015e-01 -3.64021271e-01 -1.14764667e+00 -9.96075392e-01 -9.11823153e-01 -5.10353267e-01 -4.64313328e-01 7.26593614e-01 7.27634430e-01 2.64501423e-01 1.51812345e-01 -2.81936233e-03 2.09317103e-01 5.10749757e-01 1.47033438e-01 1.25702322e-01 -1.30337560e+00 -1.56853601e-01 -5.41218042e-01 -9.38294113e-01 -5.11243999e-01 1.40221462e-01 -8.35339665e-01 -3.19494635e-01 -1.91640687e+00 4.99282926e-01 -1.25971213e-01 -4.00454819e-01 6.84373915e-01 -1.98778920e-02 6.66228831e-02 -2.31959239e-01 4.39078435e-02 -7.10560203e-01 6.64714873e-01 1.47461271e+00 -2.88185298e-01 -2.51700312e-01 1.28846288e-01 -5.55393994e-01 1.01730263e+00 1.00668657e+00 -5.93437254e-01 -7.38338709e-01 -3.58586043e-01 -3.29813749e-01 3.63285840e-02 2.04044059e-02 -1.32534969e+00 2.04596549e-01 -3.48806173e-01 4.24098790e-01 -7.40189672e-01 4.30644900e-01 -5.52694201e-01 2.90993992e-02 5.84190667e-01 -4.16075066e-02 -2.01726988e-01 9.40451249e-02 4.37197357e-01 -8.78230259e-02 -5.53768992e-01 7.36754537e-01 6.55784607e-02 -5.24490297e-01 5.57873905e-01 -5.45562744e-01 -1.32717080e-02 1.30096292e+00 -4.26026046e-01 -4.27738845e-01 -3.64299156e-02 -5.12112796e-01 1.11300342e-01 1.28704056e-01 2.55151123e-01 7.62043953e-01 -1.28916335e+00 -2.68129557e-01 1.14047006e-01 1.36856541e-01 -6.49993047e-02 5.48379302e-01 1.52024066e+00 -2.80272156e-01 7.21338570e-01 -5.08126318e-01 -6.97950542e-01 -1.43218124e+00 4.15417075e-01 -4.40285839e-02 -1.37324883e-02 -8.37143719e-01 3.34385574e-01 2.33609393e-01 1.70307010e-01 4.14167464e-01 -3.68859291e-01 -6.02022171e-01 -1.30092995e-02 7.91703045e-01 1.02810121e+00 3.40333492e-01 -4.69591558e-01 -5.84076107e-01 5.34199178e-01 -1.60493255e-01 2.04739556e-01 1.34503639e+00 1.91354468e-01 -3.11608287e-03 1.56636313e-01 5.34844220e-01 -1.83489501e-01 -7.73117065e-01 4.10607129e-01 -1.80324137e-01 -5.32822013e-01 3.99318337e-01 -6.60514355e-01 -1.34392071e+00 1.13712144e+00 9.67443943e-01 1.16255939e-01 1.20380759e+00 -1.45007402e-01 8.79106045e-01 2.39081144e-01 6.64478421e-01 -1.07352304e+00 -6.13065064e-02 3.80687183e-03 6.41594887e-01 -9.14645910e-01 3.47199202e-01 -3.78249288e-01 -4.97314125e-01 8.99380565e-01 1.12640417e+00 -6.26041293e-02 1.50426224e-01 -2.46015579e-01 -3.52245569e-01 -3.20142567e-01 -5.15589833e-01 -2.85780072e-01 3.34198207e-01 8.50735784e-01 3.19409996e-01 1.20543338e-01 -9.36189592e-01 9.41933811e-01 2.69722015e-01 3.05889487e-01 2.39724770e-01 8.66897166e-01 -5.44552624e-01 -1.05842674e+00 -4.55237508e-01 1.25752783e+00 -5.49866080e-01 1.68071687e-01 -2.32495904e-01 6.16928101e-01 3.57006997e-01 9.11046028e-01 -1.80741146e-01 -2.56226867e-01 4.58044499e-01 -1.02636702e-01 8.24740708e-01 -9.25714850e-01 -1.96795210e-01 2.05440879e-01 -8.12525302e-03 -6.30626202e-01 -5.13229609e-01 -9.79749501e-01 -1.66492200e+00 -2.19714101e-02 -6.41406357e-01 -2.76934594e-01 7.97565579e-01 1.16224635e+00 5.34680426e-01 6.82323396e-01 2.69110680e-01 -6.49772167e-01 -3.05206180e-01 -9.24937665e-01 -1.02359784e+00 -9.51901835e-04 -2.34948397e-01 -1.37873054e+00 -3.49215940e-02 -5.05093992e-01]
[12.76740550994873, 2.988612413406372]
5b99765e-f97b-4508-9ce9-35ddda2b0703
frustratingly-easy-model-ensemble-for
null
null
https://aclanthology.org/D18-1449
https://aclanthology.org/D18-1449.pdf
Frustratingly Easy Model Ensemble for Abstractive Summarization
Ensemble methods, which combine multiple models at decoding time, are now widely known to be effective for text-generation tasks. However, they generally increase computational costs, and thus, there have been many studies on compressing or distilling ensemble models. In this paper, we propose an alternative, simple but effective unsupervised ensemble method, \textit{post-ensemble}, that combines multiple models by selecting a majority-like output in post-processing. We theoretically prove that our method is closely related to kernel density estimation based on the von Mises-Fisher kernel. Experimental results on a news-headline-generation task show that the proposed method performs better than the current ensemble methods.
['Hayato Kobayashi']
2018-10-01
null
null
null
emnlp-2018-10
['headline-generation']
['natural-language-processing']
[ 2.59339243e-01 -1.09831885e-01 1.94909304e-01 -4.33167994e-01 -9.54592168e-01 -3.81706327e-01 8.13936532e-01 2.58548051e-01 -4.34027255e-01 1.12323833e+00 3.43256801e-01 -5.11181653e-01 -1.56560596e-02 -6.46051288e-01 -4.51948196e-01 -8.45740020e-01 1.40669599e-01 5.40471733e-01 -4.39270847e-02 -1.75568551e-01 5.55842638e-01 -7.39691481e-02 -1.49566448e+00 2.31140554e-01 1.65674841e+00 7.13422596e-01 1.96088612e-01 8.97251844e-01 -1.19279668e-01 5.18185437e-01 -7.88671136e-01 -9.15680170e-01 -1.39076144e-01 -8.21160734e-01 -5.28712034e-01 -2.59312481e-01 2.01179013e-01 -2.84090519e-01 -2.28997052e-01 1.00419056e+00 8.33729029e-01 5.02729475e-01 1.19479144e+00 -8.95701885e-01 -5.17514944e-01 1.00611794e+00 -3.76543105e-01 1.97150379e-01 1.66622102e-01 -5.25019109e-01 8.56570542e-01 -1.05839789e+00 2.09031850e-01 1.07538211e+00 5.85230589e-01 4.86614645e-01 -1.25996804e+00 -6.55205369e-01 1.64039489e-02 3.51061761e-01 -1.34678268e+00 -6.39090359e-01 4.87456411e-01 -2.72071332e-01 9.09329951e-01 3.23484927e-01 1.75570309e-01 1.22435153e+00 5.41935563e-01 1.03497159e+00 1.25584972e+00 -6.74935400e-01 3.11128318e-01 2.21947819e-01 2.85390407e-01 5.37496567e-01 2.10001007e-01 6.63338602e-02 -6.07152879e-01 -4.80649561e-01 3.40313733e-01 -2.57208616e-01 -4.56551939e-01 3.49657178e-01 -1.10072029e+00 9.36656594e-01 -3.79130512e-01 3.18740577e-01 -3.30073178e-01 -9.45148021e-02 2.98024148e-01 2.58103669e-01 1.29268026e+00 1.08319268e-01 -3.30000609e-01 -5.45869946e-01 -1.34523773e+00 4.25825745e-01 1.18825483e+00 8.45251858e-01 2.88600355e-01 -2.42003445e-02 -4.82329786e-01 8.71974468e-01 3.13840061e-01 5.72519422e-01 5.84070921e-01 -4.55209613e-01 6.51717126e-01 1.18395761e-01 9.74935368e-02 -9.18119192e-01 -4.89588864e-02 -5.28053343e-01 -1.36608469e+00 -3.16705704e-01 1.07612111e-01 -5.49143076e-01 -7.58193970e-01 1.50720632e+00 3.89553048e-02 6.55703485e-01 1.39902756e-01 3.83134186e-01 4.65944201e-01 9.18787777e-01 -6.99752867e-02 -5.20309865e-01 8.74413967e-01 -9.94042397e-01 -1.07718110e+00 1.81491435e-01 7.25560367e-01 -9.14846718e-01 3.93577427e-01 7.48209417e-01 -1.01906979e+00 -4.58904982e-01 -9.84185755e-01 2.04566851e-01 -3.12971979e-01 4.20495033e-01 4.15992409e-01 9.33583438e-01 -9.53930974e-01 9.99084473e-01 -5.33006728e-01 -1.68286264e-02 1.11279137e-01 1.73409164e-01 -2.45485976e-01 -3.11491881e-02 -1.22696173e+00 1.03576708e+00 7.30992973e-01 1.34969592e-01 -5.22386730e-01 -2.11401582e-01 -7.68106699e-01 8.79032463e-02 1.99225083e-01 -7.60000110e-01 1.25677514e+00 -7.21288025e-01 -1.80695081e+00 7.74783641e-02 -7.73536265e-01 -6.37699723e-01 3.88212740e-01 -5.63566506e-01 -5.68015337e-01 -2.19486058e-01 -4.22030240e-01 1.40820146e-01 1.07520556e+00 -1.12416720e+00 -6.21286333e-01 -2.97956347e-01 -5.10425270e-01 2.64133036e-01 -4.88280982e-01 -9.90212485e-02 -2.07228705e-01 -1.07266867e+00 -1.06233224e-01 -8.48060966e-01 -3.29770058e-01 -8.66094589e-01 -6.38876915e-01 -6.13581240e-01 4.94443595e-01 -1.12083864e+00 2.02717328e+00 -1.81850886e+00 3.57517272e-01 3.76921088e-01 2.62555629e-02 3.91714156e-01 -8.50396901e-02 6.28423572e-01 2.12383702e-01 2.76502222e-01 -5.48938155e-01 -7.43009210e-01 1.22981109e-02 8.18179995e-02 -5.22398651e-01 1.34089470e-01 -1.40251489e-02 5.78625441e-01 -8.27130079e-01 -6.87957883e-01 -3.59763801e-02 5.16941607e-01 -3.54585230e-01 1.08845167e-01 -2.21483603e-01 3.36853296e-01 -1.60195276e-01 2.24193111e-01 7.03878641e-01 1.21780951e-03 9.75682661e-02 9.82725993e-02 -1.09477285e-02 2.29055941e-01 -1.20398414e+00 1.50581706e+00 -4.16284293e-01 6.76177979e-01 -3.95744771e-01 -1.05185843e+00 8.43984902e-01 3.52041036e-01 1.09237768e-02 4.07934152e-02 1.89360097e-01 2.49180585e-01 -3.41612957e-02 -1.93187907e-01 9.97473955e-01 -7.07348213e-02 -4.68425974e-02 7.31820285e-01 3.33278149e-01 -1.11210309e-01 4.86510366e-01 4.06384170e-01 8.20645928e-01 2.54927158e-01 4.41261232e-01 -9.31781232e-02 4.67050731e-01 -4.68825638e-01 2.74099559e-01 1.12537575e+00 2.37578928e-01 5.05035937e-01 2.23675936e-01 1.71052396e-01 -9.95855212e-01 -9.88164008e-01 -1.44909352e-01 9.65060234e-01 -8.30581710e-02 -9.50541079e-01 -1.04371583e+00 -8.06404352e-01 -3.15504074e-01 1.25962675e+00 -2.84224153e-01 -1.69134125e-01 -4.62430507e-01 -9.45752501e-01 6.47733688e-01 5.16162932e-01 5.45914590e-01 -5.85599840e-01 2.16314029e-02 6.32520139e-01 -6.61981046e-01 -8.31899405e-01 -5.77015400e-01 -3.22646499e-02 -1.11086249e+00 -5.21811604e-01 -1.02239811e+00 -3.64763558e-01 6.24546409e-01 9.48768258e-02 1.11057973e+00 -3.17062438e-02 3.04312974e-01 9.69492793e-02 -7.64299095e-01 -7.52551198e-01 -5.27684510e-01 3.58013093e-01 1.70556292e-01 1.95977330e-01 3.96257907e-01 -4.13805336e-01 -1.37526646e-01 1.31875917e-01 -9.64751184e-01 2.40209728e-01 5.78454554e-01 1.09593189e+00 4.28298593e-01 3.79870683e-01 7.24905789e-01 -9.67594743e-01 1.09460282e+00 -4.76213247e-01 -2.30867833e-01 5.81129670e-01 -7.13861704e-01 3.84638399e-01 5.30446112e-01 -2.99652219e-01 -1.35626328e+00 -3.53909224e-01 -1.72080055e-01 6.01469427e-02 -1.29536882e-01 8.62574756e-01 2.57261600e-02 2.87249982e-01 4.22083676e-01 6.72441304e-01 -7.16281310e-02 -5.58974206e-01 3.17994446e-01 1.07358623e+00 2.17805281e-01 -4.75282729e-01 6.68743968e-01 1.06694445e-01 -2.50189483e-01 -9.17441010e-01 -9.43547964e-01 -2.91716754e-01 -4.56747532e-01 -2.37888962e-01 6.33399665e-01 -7.48739660e-01 -1.85909271e-01 7.81886101e-01 -1.32964504e+00 3.44129875e-02 3.84961635e-01 8.20605755e-01 -5.55420637e-01 6.03498042e-01 -4.85705703e-01 -1.30431497e+00 -6.40294611e-01 -7.30709314e-01 9.75304902e-01 2.89962143e-01 -2.72206873e-01 -1.13833129e+00 3.38568181e-01 8.70111063e-02 5.14656067e-01 -7.89853632e-02 8.64335835e-01 -9.52177882e-01 -9.07860249e-02 -1.61273897e-01 8.43820199e-02 6.58445120e-01 -1.10567346e-01 1.82546273e-01 -8.45711052e-01 -2.10496232e-01 -9.83030051e-02 1.24200612e-01 1.33253133e+00 3.56004626e-01 1.44434643e+00 -4.07999605e-01 -5.09596407e-01 2.54657388e-01 1.14294875e+00 3.93590517e-02 8.59342158e-01 -1.69692367e-01 4.00017619e-01 3.65565747e-01 4.93590087e-01 6.69371903e-01 3.49317372e-01 4.51386303e-01 -1.85771167e-01 3.32809269e-01 4.95055616e-01 -2.61386335e-01 8.22349250e-01 1.55086887e+00 -5.30207217e-01 -8.82402182e-01 -4.64848369e-01 2.46671230e-01 -2.11971283e+00 -1.40482473e+00 -2.36053571e-01 2.27835608e+00 1.02769566e+00 9.06723365e-03 5.02093025e-02 2.62683034e-01 9.26263809e-01 -1.10201761e-01 -1.48222819e-01 -3.80319297e-01 -3.30632448e-01 5.85467696e-01 3.05198610e-01 4.38900143e-01 -1.20217371e+00 7.10859060e-01 6.73434496e+00 1.33350515e+00 -5.45065284e-01 1.77514881e-01 6.43951058e-01 1.59385335e-02 -3.83059293e-01 -4.38383929e-02 -1.04610837e+00 8.83263230e-01 1.34224784e+00 -5.86698830e-01 2.57095963e-01 5.71888089e-01 -1.61716193e-02 -4.21412349e-01 -9.12249446e-01 9.77911651e-01 5.92982173e-01 -1.11119533e+00 5.87397665e-02 6.63225204e-02 1.01390326e+00 -3.42615366e-01 -5.90664148e-02 5.22128165e-01 4.78312939e-01 -9.74089503e-01 4.27354604e-01 9.81688380e-01 3.53114396e-01 -1.10465467e+00 1.11029112e+00 8.97949100e-01 -8.72691214e-01 1.95814297e-01 -3.73644054e-01 1.60884470e-01 5.75129151e-01 1.17330337e+00 -6.22385859e-01 9.28852856e-01 3.15553546e-01 6.13239229e-01 -5.47033489e-01 1.21893072e+00 -2.55588800e-01 1.05061281e+00 -3.08728606e-01 -3.83278340e-01 -3.02568264e-02 -3.89566332e-01 6.93782449e-01 1.31857598e+00 9.07836258e-01 1.27786294e-01 -1.24788724e-01 4.70052809e-01 8.50439945e-04 2.21704051e-01 -5.32192647e-01 -1.72428936e-01 2.71112293e-01 1.05605590e+00 -5.02345681e-01 -7.52417505e-01 1.02328826e-02 1.21831381e+00 1.61353528e-01 2.82575816e-01 -9.03641939e-01 -6.46221042e-01 9.58935022e-02 -3.94875318e-01 3.77178699e-01 -4.95504379e-01 -1.97859704e-01 -1.43782842e+00 -7.75765181e-02 -7.98997819e-01 1.53426483e-01 -6.08602285e-01 -1.44303691e+00 6.09730005e-01 2.44598657e-01 -1.16092360e+00 -6.17264986e-01 -5.49822211e-01 -6.31051898e-01 9.65322495e-01 -1.14667547e+00 -5.64118981e-01 -4.74147673e-04 3.66111577e-01 6.10322237e-01 -3.04089308e-01 1.12277150e+00 1.55391276e-01 -6.52375877e-01 7.09728003e-01 7.78100252e-01 -4.94264327e-02 8.26075912e-01 -1.33430934e+00 1.68036237e-01 7.78009415e-01 3.96795541e-01 6.30797148e-01 7.42928267e-01 -7.67577052e-01 -9.36816990e-01 -9.56377327e-01 1.25381303e+00 -3.95359159e-01 2.23996416e-01 -2.54057288e-01 -8.76172841e-01 3.55692029e-01 6.23967171e-01 -8.69117379e-01 1.08778536e+00 2.23677039e-01 -7.21579641e-02 1.71986923e-01 -1.02291012e+00 6.10501051e-01 7.43744969e-01 -2.88907349e-01 -7.66206801e-01 2.52474427e-01 4.47597384e-01 -2.52131134e-01 -7.56987751e-01 4.27449822e-01 4.12627637e-01 -9.38811600e-01 4.44054037e-01 -4.12071317e-01 4.97106373e-01 -1.75173908e-01 -9.03390124e-02 -2.04530430e+00 -1.75006539e-01 -8.13816965e-01 -7.05575049e-01 1.37210345e+00 5.99987805e-01 -8.03375006e-01 2.69354075e-01 2.94426978e-01 -4.99784239e-02 -5.90779901e-01 -9.25932527e-01 -8.73381913e-01 1.28142819e-01 -4.74448442e-01 5.06427765e-01 6.95531785e-01 1.10301793e-01 4.85350311e-01 -5.78503251e-01 -1.86646923e-01 7.56631851e-01 -1.73018619e-01 7.45743990e-01 -1.24941552e+00 -4.34761643e-01 -5.17461002e-01 -1.34018436e-01 -1.33282423e+00 4.05403525e-01 -8.18281949e-01 1.91162333e-01 -1.46734583e+00 1.61509961e-01 -2.43097395e-01 -3.22033495e-01 -2.02429611e-02 -6.50409639e-01 -1.55438304e-01 -1.04797622e-02 4.75628264e-02 -5.77506661e-01 8.27061117e-01 8.83319020e-01 6.01079389e-02 -4.66391109e-02 2.78260410e-01 -6.47838831e-01 6.00673556e-01 1.00830197e+00 -4.53523636e-01 -2.63144463e-01 -2.40368083e-01 2.58724391e-01 -4.30805609e-02 -1.75656322e-02 -9.07051027e-01 2.26400778e-01 2.41283968e-01 3.20300162e-01 -1.07478607e+00 2.28784442e-01 -4.26332235e-01 4.90760021e-02 1.05552986e-01 -4.24531519e-01 2.82389641e-01 -4.38351892e-02 8.53436530e-01 -3.32871348e-01 -7.06904590e-01 2.83276469e-01 7.61483014e-02 -3.79081547e-01 1.24062411e-01 -6.58897519e-01 -1.16659313e-01 8.80370736e-01 -4.73067164e-03 -8.80800113e-02 -6.44765854e-01 -6.54364884e-01 -1.32511212e-02 -1.43029373e-02 3.27892125e-01 5.92642784e-01 -1.24809325e+00 -1.31914449e+00 5.28976470e-02 -1.51006162e-01 -2.01458365e-01 3.22887659e-01 9.85934019e-01 -1.90713838e-01 3.53540778e-01 5.04047751e-01 -4.81270343e-01 -1.23636961e+00 1.05966292e-01 -9.88147631e-02 -6.82172954e-01 -1.80605486e-01 8.97195399e-01 -3.03349614e-01 -2.43675455e-01 7.41550103e-02 3.72389108e-02 -3.10193658e-01 1.55540556e-01 8.77512455e-01 7.02051759e-01 9.04460475e-02 -4.59330738e-01 9.02049094e-02 2.19640508e-01 -3.90622318e-01 -5.89922190e-01 1.06919658e+00 -2.05365732e-01 -3.62390399e-01 6.35934293e-01 8.65213275e-01 -1.94976032e-02 -6.48118317e-01 -2.84074455e-01 8.83703679e-02 -2.51515657e-01 8.55304226e-02 -7.21170664e-01 -4.75468814e-01 9.48155224e-01 3.65195394e-01 3.73134971e-01 1.23788941e+00 -2.90454060e-01 7.60001242e-01 4.04877156e-01 4.33831096e-01 -1.35728991e+00 -2.63452828e-01 9.32198584e-01 7.10556388e-01 -1.05614841e+00 6.53420463e-02 -1.71388492e-01 -6.48198783e-01 1.19840515e+00 5.01420438e-01 1.36731386e-01 8.87895107e-01 2.80148387e-01 -3.78399998e-01 5.48786521e-01 -1.05452538e+00 -5.26936539e-02 4.50611234e-01 4.19609338e-01 6.30178630e-01 2.43400410e-01 -6.55358255e-01 7.50593364e-01 -4.24679607e-01 -8.20511393e-03 3.47576559e-01 7.34843314e-01 -5.15186310e-01 -1.27349436e+00 -4.31608498e-01 9.35516596e-01 -4.79367495e-01 -5.65782845e-01 -3.22803617e-01 1.75782785e-01 -6.99159922e-04 1.26207685e+00 1.46126375e-01 -5.37068486e-01 -1.05924733e-01 5.25384724e-01 7.22554028e-01 -4.65407670e-01 -4.38928962e-01 1.87576205e-01 3.18410516e-01 1.10561959e-01 -5.31218410e-01 -5.34876883e-01 -7.75481939e-01 -3.96964371e-01 -9.98663247e-01 5.50876081e-01 7.84295619e-01 1.28884530e+00 5.19135177e-01 2.96837568e-01 6.44636869e-01 -8.86273265e-01 -8.73160064e-01 -1.49095714e+00 -6.04519904e-01 -5.53250825e-03 -9.01102126e-02 -5.69452763e-01 -5.52759945e-01 1.91700250e-01]
[11.94503116607666, 9.262096405029297]
ffe5e07e-c2a9-4909-b884-01ebbae6412d
deepaste-inpainting-for-pasting
2112.10600
null
https://arxiv.org/abs/2112.10600v2
https://arxiv.org/pdf/2112.10600v2.pdf
DeePaste -- Inpainting for Pasting
One of the challenges of supervised learning training is the need to procure an substantial amount of tagged data. A well-known method of solving this problem is to use synthetic data in a copy-paste fashion, so that we cut objects and paste them onto relevant backgrounds. Pasting the objects naively results in artifacts that cause models to give poor results on real data. We present a new method for cleanly pasting objects on different backgrounds so that the dataset created gives competitive performance on real data. The main emphasis is on the treatment of the border of the pasted object using inpainting. We show state-of-the-art results both on instance detection and foreground segmentation
['Levi Kassel Michael Werman']
2021-12-20
null
null
null
null
['foreground-segmentation']
['computer-vision']
[ 6.97481036e-01 1.45157784e-01 4.78717200e-02 -3.79141927e-01 -8.93224955e-01 -5.86442351e-01 4.56981361e-01 1.90130845e-01 -4.87907588e-01 5.96114516e-01 -3.05020660e-01 -5.00246622e-02 3.04574311e-01 -5.62711775e-01 -9.65032578e-01 -6.85576618e-01 1.37329891e-01 5.39509952e-01 7.28739142e-01 1.23457983e-01 2.03082979e-01 5.27559817e-01 -1.42415142e+00 6.42618477e-01 7.62683868e-01 6.89768970e-01 2.29242474e-01 8.30103219e-01 -5.68617225e-01 8.51393104e-01 -9.69490767e-01 -5.95260620e-01 4.47141647e-01 -7.04293311e-01 -9.18337882e-01 4.77362990e-01 8.48769248e-01 -7.87920952e-02 3.68974358e-03 9.16549027e-01 6.50132969e-02 -3.82334590e-02 6.19041622e-01 -1.26425743e+00 -1.40592515e-01 3.26590657e-01 -9.32773709e-01 2.36207262e-01 7.80505221e-03 -7.10701868e-02 4.84245688e-01 -7.28044033e-01 8.08216333e-01 1.09132028e+00 7.92356610e-01 5.50305367e-01 -1.42581046e+00 -3.73797089e-01 7.21661747e-02 -2.11323887e-01 -1.02686906e+00 -4.47562516e-01 6.92824900e-01 -4.56514806e-01 4.69430566e-01 3.84244680e-01 5.71525931e-01 6.87212408e-01 2.22772643e-01 9.48178709e-01 1.11418355e+00 -8.10813069e-01 3.25436383e-01 2.73707062e-02 1.11303881e-01 6.91826344e-01 2.96990663e-01 -2.09905669e-01 -3.16982359e-01 1.06412023e-02 8.55992138e-01 1.30339235e-01 -1.49721980e-01 -6.01705074e-01 -1.18517983e+00 5.59074163e-01 2.71178216e-01 2.62536556e-01 -2.09416255e-01 2.57876068e-01 3.73954594e-01 1.21509843e-01 7.27341950e-01 5.47484040e-01 -6.91493094e-01 -2.67957225e-02 -1.46651959e+00 2.55334526e-01 6.93222046e-01 7.81939805e-01 6.85067356e-01 -2.04450279e-01 3.56357619e-02 7.35748231e-01 -4.46019731e-02 2.18676761e-01 1.09442063e-01 -1.34612286e+00 4.63444561e-01 3.92505139e-01 1.69648901e-01 -6.84521616e-01 9.34506431e-02 -3.09517086e-02 -3.59244317e-01 7.83834636e-01 9.85952914e-01 -3.20565492e-01 -1.44397175e+00 1.32770443e+00 4.36518669e-01 2.74719775e-01 -3.00909523e-02 4.66142386e-01 5.41702271e-01 6.69175923e-01 5.32026179e-02 -1.65559724e-01 9.56772625e-01 -1.18002522e+00 -7.38545835e-01 -5.23420811e-01 5.96280515e-01 -1.05005252e+00 9.77903962e-01 6.03159845e-01 -1.37579966e+00 -5.21308303e-01 -8.76885176e-01 -5.41887172e-02 -3.47392142e-01 4.89468202e-02 5.79091430e-01 4.83200729e-01 -6.32911086e-01 9.46089506e-01 -1.08418512e+00 -3.93012881e-01 7.09743023e-01 2.34746650e-01 -3.63370597e-01 -2.06665233e-01 -4.04132992e-01 9.19229388e-01 4.13373947e-01 -7.31753977e-03 -6.88493490e-01 -8.52163136e-01 -7.74365306e-01 -1.51615068e-01 5.34579098e-01 -1.79761514e-01 1.38058031e+00 -1.36217511e+00 -1.07089913e+00 1.18141079e+00 -1.08368210e-01 -3.75567377e-01 9.75838959e-01 -5.60403585e-01 -1.03382111e-01 1.15348049e-01 1.05257906e-01 7.99521565e-01 1.13718915e+00 -1.69691098e+00 -8.19891930e-01 -1.07428230e-01 -4.22856897e-01 -1.23956636e-01 8.98618326e-02 1.68174118e-01 -7.44430661e-01 -1.06380618e+00 2.53925174e-01 -8.08425128e-01 -3.61475378e-01 3.49298149e-01 -2.33436584e-01 1.42094404e-01 1.25706410e+00 -1.00994051e+00 7.17243910e-01 -2.26395869e+00 -3.98226604e-02 -9.10949036e-02 1.06234603e-01 5.79126298e-01 6.47381991e-02 1.27026096e-01 -4.13753763e-02 -1.32427085e-02 -6.23391569e-01 -5.94170928e-01 -3.68604362e-01 5.24820507e-01 -4.43286031e-01 3.34224641e-01 3.94345403e-01 6.79743290e-01 -9.77246046e-01 -6.90893054e-01 1.34841070e-01 3.89610857e-01 -4.75757957e-01 4.46344465e-01 -5.97320914e-01 2.73128271e-01 -3.69338170e-02 5.08054912e-01 8.51304889e-01 -8.14292803e-02 -7.16634467e-02 2.45103184e-02 9.38005671e-02 1.40842304e-01 -1.41887176e+00 1.73259985e+00 -2.39039987e-01 8.90929580e-01 2.65948266e-01 -5.56071341e-01 6.02685153e-01 -7.10365502e-03 2.57984638e-01 -2.07268342e-01 -7.93012138e-03 1.48354143e-01 -1.72816649e-01 -4.85856265e-01 3.64262789e-01 -2.00139984e-01 2.74941415e-01 5.30689836e-01 -6.58442453e-03 -4.53281581e-01 3.32449108e-01 2.69870460e-01 1.03485870e+00 4.80874687e-01 -5.60740158e-02 -1.30054653e-01 2.10187510e-02 3.28020006e-01 5.97416282e-01 6.17549419e-01 5.47200441e-02 1.07642281e+00 6.28358841e-01 -4.56662536e-01 -1.00738549e+00 -9.04031634e-01 1.55146316e-01 1.16414785e+00 -5.59366159e-02 -1.01654530e-01 -1.32552850e+00 -9.04382706e-01 1.71131715e-02 8.33472848e-01 -7.94173956e-01 9.79972780e-02 -7.83996165e-01 -6.49902701e-01 3.62422466e-01 7.99678504e-01 4.96644169e-01 -1.10797966e+00 -9.82149720e-01 1.31328031e-01 -1.11552037e-01 -1.05247176e+00 -3.42316896e-01 7.22685337e-01 -1.14230108e+00 -1.06721461e+00 -9.53621387e-01 -1.00663102e+00 1.08624029e+00 3.90620351e-01 1.27407908e+00 3.49582046e-01 -3.92975450e-01 -7.12901577e-02 -2.25824654e-01 -3.66815865e-01 -7.31782734e-01 -2.37919480e-01 -4.04875338e-01 5.33587486e-03 7.79711083e-02 -8.50034654e-02 -1.09929383e-01 2.91071981e-01 -1.07278800e+00 2.05129609e-01 4.21370655e-01 5.97478211e-01 6.63809001e-01 1.94890797e-01 -2.23509334e-02 -1.30440831e+00 1.43774599e-01 1.15160510e-01 -6.62724853e-01 4.01884079e-01 -2.06531748e-01 9.20961648e-02 4.43830252e-01 -6.09718382e-01 -1.28266788e+00 4.10854429e-01 1.40100792e-01 -6.15732968e-01 -3.97485197e-01 -1.55237794e-01 -2.44038776e-01 -9.20102820e-02 6.90048039e-01 -1.41486853e-01 -1.27886981e-01 -8.52658391e-01 4.62422311e-01 3.79461110e-01 7.49498963e-01 -4.71694767e-01 7.54870176e-01 5.45906901e-01 -1.54985622e-01 -7.64559984e-01 -9.27132607e-01 -3.00421804e-01 -9.95034397e-01 -5.07082511e-03 9.30759430e-01 -5.38487613e-01 2.14498147e-01 7.20610976e-01 -1.17624104e+00 -8.46828520e-01 -6.37804568e-01 -1.25406440e-02 -4.17659283e-01 3.16047549e-01 -7.11677194e-01 -6.47037625e-01 1.27398461e-01 -8.39906454e-01 9.85910773e-01 3.07835191e-01 -2.64590293e-01 -8.26872230e-01 -1.35931233e-02 3.69672537e-01 1.21157296e-01 5.44682682e-01 8.90495360e-01 -4.30632740e-01 -5.31326413e-01 -2.36385092e-01 -2.23545343e-01 5.19350588e-01 3.44245970e-01 3.70231807e-01 -1.12071383e+00 -1.61062256e-01 1.24301855e-02 -2.77243793e-01 1.11320949e+00 3.56827646e-01 1.08787966e+00 1.77219734e-02 -6.83438838e-01 4.64337915e-01 1.29461050e+00 3.11888278e-01 6.53270721e-01 1.14527300e-01 8.64596367e-01 6.93443000e-01 6.91372216e-01 -7.54017308e-02 -3.27190578e-01 4.38408315e-01 2.02802971e-01 -5.51278472e-01 -3.42980415e-01 -2.01421767e-01 -4.48012948e-02 2.08300561e-01 2.00938433e-01 -2.52169877e-01 -9.68179584e-01 6.37967527e-01 -1.73291636e+00 -8.84979427e-01 -4.43189532e-01 2.24772215e+00 9.95660603e-01 5.58824599e-01 1.36643410e-01 2.73093790e-01 7.65179098e-01 -3.21055278e-02 -2.49854431e-01 -3.10531020e-01 2.56847680e-01 3.09256136e-01 6.25974655e-01 7.05614269e-01 -1.22855520e+00 1.04949641e+00 7.20380354e+00 5.74679494e-01 -1.14410758e+00 8.82276595e-02 7.95919955e-01 -1.59746125e-01 1.69596344e-01 2.73729004e-02 -6.93321049e-01 5.83509982e-01 6.64680660e-01 3.27908635e-01 1.25291690e-01 8.55788589e-01 1.39367193e-01 -7.31475711e-01 -1.22455645e+00 7.48582125e-01 6.55869916e-02 -1.37044513e+00 -2.33587667e-01 -1.81740016e-01 7.85126686e-01 -1.72655299e-01 -2.99489707e-01 -5.02961129e-02 3.21866244e-01 -7.96347082e-01 7.98317969e-01 2.78690696e-01 6.67810917e-01 -5.39387703e-01 5.15479982e-01 2.87257403e-01 -8.07016432e-01 2.54707247e-01 -4.25438762e-01 -1.31712571e-01 1.23068966e-01 8.11557770e-01 -8.53875160e-01 4.90084961e-02 7.20036864e-01 4.15531129e-01 -8.26965094e-01 1.56145763e+00 -1.68527514e-01 8.28642666e-01 -5.39888561e-01 4.07658041e-01 8.00117031e-02 -8.29640552e-02 2.80262884e-02 1.59733284e+00 -2.73470953e-02 -7.36141652e-02 8.20975527e-02 7.37524927e-01 -1.19789340e-01 -2.07629800e-01 -5.95573902e-01 9.38116685e-02 1.01882383e-01 1.15561950e+00 -1.53199613e+00 -6.82943106e-01 -4.11521077e-01 1.23608196e+00 1.48150876e-01 3.23748410e-01 -7.31496811e-01 -4.81887996e-01 3.16298425e-01 4.21351105e-01 6.63408101e-01 -1.00549236e-01 -5.17391384e-01 -9.07580495e-01 1.22817799e-01 -7.78391361e-01 3.84935439e-01 -9.16335285e-01 -1.10934150e+00 3.05782169e-01 7.42405504e-02 -9.79793847e-01 1.36706799e-01 -5.82501829e-01 -7.97967494e-01 6.89556122e-01 -1.14259875e+00 -9.42037463e-01 -4.68811959e-01 1.97337821e-01 8.88426065e-01 2.82654792e-01 4.97444987e-01 2.59399086e-01 -3.82648975e-01 1.73310548e-01 2.45532930e-01 2.70200998e-01 7.46087551e-01 -1.50084424e+00 7.67014980e-01 1.08544123e+00 5.47935307e-01 2.61202931e-01 9.50486660e-01 -7.47503936e-01 -8.37383091e-01 -1.05897045e+00 7.81483471e-01 -6.44868910e-01 2.15809822e-01 -5.25941074e-01 -1.35001576e+00 7.64095306e-01 3.63626659e-01 3.67974907e-01 3.34814101e-01 -3.52945000e-01 -1.83438510e-01 2.81027053e-02 -1.34551501e+00 4.28968728e-01 6.97605133e-01 -7.78419077e-02 -8.49929690e-01 5.39924145e-01 5.53194344e-01 -7.01779604e-01 -1.97952628e-01 1.14412360e-01 2.03425601e-01 -1.01491964e+00 7.99299955e-01 -7.33414412e-01 4.81814593e-01 -4.48044688e-01 2.56308347e-01 -1.27782500e+00 1.01421930e-01 -8.72640789e-01 -3.05415019e-02 1.32641208e+00 3.14378589e-01 -1.36601403e-01 1.20221138e+00 7.53170192e-01 1.74919233e-01 -5.27881742e-01 -5.22621214e-01 -7.94673264e-01 2.95120794e-02 -1.73158646e-01 6.68391883e-02 1.01748967e+00 -3.97249669e-01 1.86109133e-02 -1.24850072e-01 -3.81540731e-02 6.89140022e-01 4.47363220e-02 9.03332293e-01 -1.22844005e+00 -3.25915039e-01 -1.47403985e-01 -1.09555095e-01 -9.34690297e-01 -2.15711787e-01 -4.70685840e-01 3.57341468e-01 -1.57303798e+00 6.03599399e-02 -4.97806787e-01 6.79772496e-02 5.00042558e-01 -3.75227213e-01 4.78196830e-01 3.04591268e-01 1.93271890e-01 -4.95831490e-01 -1.14327490e-01 1.16388071e+00 4.66526262e-02 -3.13809663e-01 1.09544881e-01 -3.64074975e-01 1.04713643e+00 7.09824741e-01 -9.55716372e-01 -2.40819156e-01 -5.71095943e-01 -2.16670960e-01 -7.19652250e-02 4.27461147e-01 -1.17412841e+00 -3.17910016e-02 -3.15578341e-01 7.59169281e-01 -6.95329189e-01 3.40216100e-01 -9.41144884e-01 -2.51933793e-03 4.50963825e-01 -3.86387646e-01 1.70650020e-01 4.09987181e-01 4.34991896e-01 -2.69753635e-02 -6.34004474e-01 1.22172868e+00 -3.97257060e-01 -6.02155328e-01 -2.56635934e-01 -3.25383127e-01 4.48243886e-01 1.21893752e+00 -2.40331084e-01 -1.01318538e-01 -1.02596179e-01 -7.45913982e-01 -8.87404755e-02 8.21143866e-01 2.07016706e-01 2.28014067e-01 -8.29972088e-01 -3.86855066e-01 3.53587598e-01 -3.76727194e-01 3.91198039e-01 -4.00297791e-01 4.36493635e-01 -1.03347719e+00 -1.30662158e-01 -2.52310336e-01 -6.57806814e-01 -1.59008133e+00 6.67087376e-01 4.59734857e-01 -1.21182159e-01 -9.41892326e-01 1.09008968e+00 9.60564911e-02 -1.05826613e-02 5.35750270e-01 -5.42575121e-01 3.61896485e-01 1.86756372e-01 5.69030404e-01 4.28469837e-01 3.18878174e-01 -1.64538711e-01 -2.16027737e-01 3.53451133e-01 -3.56790185e-01 -2.44291395e-01 1.46241283e+00 3.55317980e-01 -4.63229902e-02 3.73134285e-01 1.02997971e+00 1.24806605e-01 -1.84533513e+00 -1.77719682e-01 2.35518590e-01 -8.29841197e-01 -2.11468395e-02 -8.98014069e-01 -1.23565173e+00 1.01303792e+00 5.30027032e-01 4.03887093e-01 8.98717225e-01 -9.81571712e-03 8.59004557e-01 3.11749279e-01 1.35552973e-01 -1.20515931e+00 1.49285719e-01 2.55150706e-01 5.11123896e-01 -1.18912506e+00 2.42919952e-01 -6.73229218e-01 -6.10790014e-01 1.07355416e+00 6.02624774e-01 -2.07566902e-01 3.81620616e-01 8.26109350e-01 5.76794326e-01 -1.15516827e-01 -3.38616282e-01 8.98757763e-03 -2.43293177e-02 6.70777261e-01 4.40141082e-01 -2.95757174e-01 8.77595171e-02 9.48362797e-02 2.26403162e-01 7.32252151e-02 6.30759060e-01 1.63545835e+00 -5.31017423e-01 -1.19875610e+00 -7.04350710e-01 4.67479467e-01 -6.14009798e-01 8.92311856e-02 -7.12722480e-01 8.34737241e-01 3.30175549e-01 6.94455326e-01 2.11936384e-01 2.68573016e-01 2.87014037e-01 4.43640590e-01 6.55750215e-01 -8.53540242e-01 -6.53325021e-01 3.93345118e-01 -1.14196755e-01 -3.83922160e-01 -3.56390536e-01 -7.72123516e-01 -1.50577760e+00 5.70078380e-03 -3.53627861e-01 4.33593243e-03 5.65364897e-01 1.00854206e+00 7.08225071e-02 5.21267951e-01 1.57920644e-01 -1.26899362e+00 -2.43728369e-01 -6.15411222e-01 -4.17098194e-01 7.30042815e-01 4.88346666e-01 -5.40689886e-01 -3.26700628e-01 7.41714835e-01]
[9.673212051391602, 0.2980881929397583]
a2fefdd1-5dff-4156-9e66-948bc319dd18
recursive-context-aware-lexical
null
null
https://aclanthology.org/D19-1491
https://aclanthology.org/D19-1491.pdf
Recursive Context-Aware Lexical Simplification
This paper presents a novel architecture for recursive context-aware lexical simplification, REC-LS, that is capable of (1) making use of the wider context when detecting the words in need of simplification and suggesting alternatives, and (2) taking previous simplification steps into account. We show that our system outputs lexical simplifications that are grammatically correct and semantically appropriate, and outperforms the current state-of-the-art systems in lexical simplification.
['Sian Gooding', 'Ekaterina Kochmar']
2019-11-01
null
null
null
ijcnlp-2019-11
['lexical-simplification']
['natural-language-processing']
[ 2.04425544e-01 2.70006597e-01 -8.59932303e-02 -3.84625673e-01 -6.51708841e-01 -3.96734446e-01 3.49549621e-01 8.07535827e-01 -7.75753379e-01 7.13728487e-01 6.75304711e-01 -6.34057343e-01 1.80630296e-01 -8.69520843e-01 -3.81584287e-01 1.42859176e-01 4.37063426e-01 8.57589722e-01 2.25261450e-01 -8.30572784e-01 6.00538492e-01 5.35859704e-01 -1.76883650e+00 2.92591065e-01 1.20432091e+00 2.29604587e-01 3.42196822e-01 7.31123507e-01 -4.70785618e-01 5.10320604e-01 -9.42819893e-01 -5.81734002e-01 -5.11309691e-02 -5.43382406e-01 -1.04103601e+00 -4.09132838e-01 5.73652446e-01 -7.29419738e-02 6.14851192e-02 1.09185386e+00 5.79237878e-01 6.96021080e-01 6.01383686e-01 -3.85461092e-01 -8.03328395e-01 1.13855207e+00 1.10270709e-01 4.78209943e-01 8.20136011e-01 -2.10378990e-01 9.79521871e-01 -1.13855815e+00 9.71463323e-01 1.88140595e+00 7.25068212e-01 7.44959712e-01 -1.10196030e+00 -6.65725887e-01 6.42351449e-01 2.64484733e-01 -1.75238991e+00 -6.23766601e-01 4.32814628e-01 4.48059253e-02 2.06343174e+00 6.24168038e-01 6.89456165e-01 4.47101682e-01 2.38083571e-01 4.40837085e-01 6.29727244e-01 -1.15110934e+00 -7.62639418e-02 -3.24583262e-01 4.60975558e-01 4.05818462e-01 5.26957333e-01 -2.18781516e-01 -2.82199085e-01 -1.79326802e-01 2.26481453e-01 -4.59221214e-01 1.33313891e-02 5.70917368e-01 -7.68259943e-01 8.44560981e-01 3.49968858e-02 4.49440539e-01 -5.11630356e-01 4.69065830e-02 5.70574701e-01 4.67501163e-01 4.06412512e-01 9.81253922e-01 -3.76581401e-01 9.06404704e-02 -1.00830650e+00 6.32972658e-01 8.67779315e-01 1.35925925e+00 6.68026209e-01 2.63555318e-01 -1.21029556e-01 1.01549315e+00 3.92063856e-02 5.85063398e-01 6.09533668e-01 -7.89054513e-01 2.73070455e-01 6.46689534e-01 -1.14981800e-01 -4.82760012e-01 -5.49662769e-01 3.64991017e-02 -3.09709519e-01 -1.37838989e-01 -2.88684368e-01 1.29721850e-01 -9.51135755e-01 1.80867684e+00 2.82809198e-01 -3.26869577e-01 6.14851862e-02 2.63575405e-01 1.15657139e+00 5.31328559e-01 9.22060549e-01 -5.94681323e-01 1.37460208e+00 -6.88017249e-01 -1.13973701e+00 -2.53964871e-01 1.05864513e+00 -1.30240321e+00 1.22212672e+00 3.15679371e-01 -1.45225632e+00 -6.19927049e-01 -1.08829892e+00 -7.84003973e-01 -7.07780480e-01 -2.61120051e-01 6.75663829e-01 6.46350801e-01 -1.09521616e+00 8.51345003e-01 -2.89818317e-01 -8.14517558e-01 -1.59640074e-01 4.28582937e-01 -3.05155873e-01 -1.61714002e-01 -1.44801521e+00 1.47211325e+00 6.49291575e-01 -1.58261314e-01 -3.23327184e-01 -5.34603536e-01 -1.09169054e+00 -1.82999447e-02 4.48391974e-01 -1.10807300e+00 1.54885304e+00 -8.33476007e-01 -1.36127114e+00 7.91081727e-01 -6.42306626e-01 -3.56489986e-01 -3.04915961e-02 -6.90766394e-01 -6.37366712e-01 -3.40116322e-01 2.50381559e-01 7.95099497e-01 5.41031957e-01 -7.07215607e-01 -9.36583698e-01 -1.06226578e-01 2.50925034e-01 7.37977624e-01 8.12246576e-02 4.98673677e-01 -4.80843574e-01 -1.20001352e+00 -8.47037882e-02 -9.26995873e-01 -2.88924575e-01 -7.03400731e-01 -4.42897111e-01 -7.86145926e-01 4.05228943e-01 -1.10395598e+00 2.15559077e+00 -1.51890874e+00 5.49100816e-01 1.14261016e-01 -3.75597253e-02 7.29965866e-01 -2.04103976e-01 6.83630228e-01 -2.38855526e-01 7.01923788e-01 -8.62530898e-03 -6.08603001e-01 -5.68452477e-02 3.01267773e-01 -2.47556180e-01 -1.77189842e-01 1.77268349e-02 1.03030968e+00 -8.61636519e-01 -6.48614228e-01 5.95404863e-01 1.84338331e-01 -7.95702934e-01 -7.07911253e-02 -2.16937780e-01 -1.23537458e-01 -1.42596379e-01 7.10704565e-01 3.56350422e-01 7.75517941e-01 3.79599422e-01 -1.36335611e-01 -2.16046736e-01 1.11047924e+00 -1.23425674e+00 1.52506340e+00 -6.38871908e-01 4.58379656e-01 -4.37551111e-01 -3.11541200e-01 5.63075066e-01 8.70175064e-02 -5.00290453e-01 -8.87117028e-01 2.84530818e-01 3.16190600e-01 5.07420264e-02 -1.98900387e-01 1.04492402e+00 -1.91448897e-01 -3.50904644e-01 5.66510499e-01 -4.27101329e-02 -6.94762766e-01 6.46834791e-01 2.91888386e-01 9.02841330e-01 1.71970248e-01 1.34462905e+00 -4.62300539e-01 8.46695125e-01 -1.87304199e-01 5.45626044e-01 8.44999254e-01 8.52428675e-02 2.53352195e-01 -8.75643492e-02 -4.73139524e-01 -1.09774315e+00 -7.84723580e-01 2.18057036e-01 1.60993242e+00 -4.14483808e-02 -1.06332529e+00 -9.94759560e-01 -4.81831878e-01 1.27921015e-01 1.73000908e+00 -4.92752343e-01 -1.74749702e-01 -1.25525475e+00 -7.15714842e-02 8.11653972e-01 3.77799064e-01 -2.07850605e-01 -1.49265945e+00 -2.96798617e-01 5.44514656e-01 -4.37929273e-01 -6.90235019e-01 -5.32575011e-01 1.36777744e-01 -9.34517741e-01 -1.01156652e+00 9.23767313e-02 -8.68239701e-01 4.53849465e-01 2.84181058e-01 1.51011825e+00 8.24274540e-01 -1.17080264e-01 -1.42105013e-01 -3.73880357e-01 -7.07436621e-01 -8.22314799e-01 2.59917885e-01 4.26164046e-02 -1.02443910e+00 6.42604589e-01 -4.98180181e-01 -5.51173575e-02 -1.79052562e-01 -8.75083983e-01 -8.14911351e-02 2.45825499e-01 4.61586326e-01 9.60733533e-01 -1.11690350e-01 6.32379353e-01 -1.54784715e+00 1.24399734e+00 -6.64772689e-02 -8.89082253e-02 4.69890177e-01 -7.87546456e-01 1.28252447e-01 8.98570418e-01 -1.89255267e-01 -1.08473516e+00 -2.75516450e-01 -7.41267383e-01 1.97651356e-01 -1.76250488e-01 4.86034811e-01 -3.04523796e-01 -1.59576610e-01 8.49445581e-01 -2.46851176e-01 -6.24367774e-01 -6.76735282e-01 8.72332156e-01 5.73863447e-01 4.64511693e-01 -6.29599512e-01 4.31747347e-01 -7.58522302e-02 -1.87657773e-01 -9.92957294e-01 -7.75095820e-01 -4.38334197e-01 -9.02504623e-01 5.32732047e-02 3.15832108e-01 -6.28132522e-01 4.56681475e-02 2.26993144e-01 -1.46687245e+00 6.82565878e-05 -7.69941688e-01 9.60661322e-02 -2.72726357e-01 7.19996333e-01 -6.08797312e-01 -4.51201707e-01 -7.91199744e-01 -9.40259576e-01 9.18637276e-01 1.69133991e-01 -1.23745155e+00 -9.65603113e-01 1.27954066e-01 -1.53547272e-01 3.92289191e-01 -1.68516159e-01 1.51808500e+00 -9.15152550e-01 -7.59267388e-03 -1.53352723e-01 2.32206956e-01 2.23264292e-01 1.44557372e-01 2.74041295e-01 -4.97177422e-01 1.36256963e-01 -2.51985848e-01 9.78047475e-02 7.32870340e-01 4.92773831e-01 7.78570533e-01 -3.05997729e-01 -4.42525983e-01 6.05093062e-01 9.37830091e-01 3.16995263e-01 8.43209863e-01 3.85767817e-01 7.21947908e-01 4.81202364e-01 8.51433992e-01 3.14936787e-02 7.60207593e-01 7.49879956e-01 5.00594191e-02 6.63603016e-04 -7.37652600e-01 -3.61222804e-01 1.07827038e-01 1.22470689e+00 7.36245587e-02 -4.12541032e-01 -6.67860627e-01 7.46100366e-01 -1.85203314e+00 -6.78541362e-01 -1.75014988e-01 1.91473591e+00 1.07809079e+00 1.99143752e-01 -1.67057425e-01 1.35441557e-01 8.55355322e-01 -6.89871609e-02 -3.60997528e-01 -1.58590817e+00 -4.96195138e-01 7.68862665e-01 1.60029337e-01 1.11374533e+00 -1.05441415e+00 1.96249223e+00 7.59130573e+00 1.02027786e+00 -6.72856450e-01 5.28414212e-02 8.90886709e-02 5.81650212e-02 -6.72610164e-01 1.84229791e-01 -9.50552464e-01 -8.60699788e-02 8.90466332e-01 -6.09062493e-01 7.63850749e-01 6.41825080e-01 -3.69764008e-02 -2.68001676e-01 -1.16587639e+00 6.64776802e-01 3.99940073e-01 -1.05578208e+00 1.03443182e+00 -6.12752974e-01 5.11007965e-01 -3.68091494e-01 -3.58691543e-01 5.10733962e-01 2.96797752e-01 -1.06954074e+00 1.02986395e+00 5.36688030e-01 6.97863638e-01 -1.10594749e+00 7.12543130e-01 1.95570096e-01 -1.30799556e+00 2.97682285e-01 -4.39627916e-01 -5.01045644e-01 3.38608593e-01 3.68221015e-01 -6.37600124e-01 4.41377670e-01 3.83618504e-01 4.71045762e-01 -8.68372858e-01 7.75537789e-01 -9.83237028e-01 4.00883824e-01 -4.33469057e-01 -1.78531483e-01 -1.80542156e-01 1.24952696e-01 8.50228190e-01 1.94279408e+00 1.08889602e-01 5.80392480e-01 7.75534958e-02 3.99734169e-01 -1.17770851e-01 7.85136938e-01 -5.19258440e-01 1.19408600e-01 1.14225566e+00 7.77094007e-01 -7.77908862e-01 -7.96969414e-01 -1.01634394e-02 9.09840226e-01 5.86431861e-01 5.73097132e-02 -4.32881325e-01 -8.53409648e-01 8.42516243e-01 -2.90516317e-01 3.04125249e-01 -9.94291306e-02 -6.86350107e-01 -1.01350427e+00 -1.87811151e-01 -9.62037742e-01 6.62895143e-01 -7.56685317e-01 -9.70294535e-01 7.62354136e-01 1.42945826e-01 -4.96876270e-01 -3.17510784e-01 -3.21247309e-01 -4.87372637e-01 1.29046845e+00 -1.50591266e+00 -1.04960442e+00 2.89061517e-01 2.92850077e-01 9.16399121e-01 2.20085651e-01 1.24401772e+00 2.02194348e-01 -4.72050846e-01 8.51236105e-01 -4.77136821e-01 -5.39417624e-01 7.58991182e-01 -1.09321880e+00 1.10194099e+00 1.24917674e+00 1.81441978e-02 1.26943862e+00 9.49869990e-01 -1.12156427e+00 -8.56005371e-01 -1.12033784e+00 2.09935856e+00 -2.97282547e-01 3.61454785e-01 -1.09058924e-01 -1.03401673e+00 6.79162920e-01 1.82506397e-01 -9.22935128e-01 6.70234442e-01 3.27528894e-01 -3.58294517e-01 2.40101799e-01 -1.32119656e+00 1.24551916e+00 1.51192522e+00 -6.99488640e-01 -1.45344341e+00 3.37541819e-01 1.06880963e+00 -5.53654730e-01 -4.52866793e-01 3.63991797e-01 4.14611131e-01 -4.74200308e-01 8.77955794e-01 -8.93740177e-01 -6.29040673e-02 -1.33090943e-01 -1.15843855e-01 -1.50967002e+00 -6.33531272e-01 -9.11114514e-01 -3.54357734e-02 1.02181327e+00 4.90595996e-01 -4.10900414e-01 -2.95973811e-02 4.85399127e-01 -6.05034828e-01 -5.29971540e-01 -1.03139389e+00 -4.45417225e-01 1.91063136e-01 -7.31774926e-01 8.94060612e-01 7.78547525e-01 1.57107368e-01 7.95242846e-01 4.94507477e-02 -1.34780675e-01 2.32769735e-02 -1.36899784e-01 5.28752863e-01 -1.12056589e+00 3.30463827e-01 -8.65142465e-01 -1.60803691e-01 -9.52530503e-01 5.44753790e-01 -9.08387065e-01 -1.01326585e-01 -1.75047100e+00 -2.29564667e-01 -2.93956548e-01 -4.61352943e-03 5.66009879e-01 -6.09254181e-01 2.13825360e-01 2.81361639e-01 9.82323363e-02 -6.35729015e-01 3.52457345e-01 7.34216034e-01 6.33556023e-02 -2.97421336e-01 -4.06748444e-01 -1.14795530e+00 1.08974218e+00 5.84234178e-01 -7.59849608e-01 -9.14311185e-02 -5.14269710e-01 3.57680172e-01 -5.24873972e-01 -4.31018740e-01 -6.14693999e-01 2.11682558e-01 -1.71689630e-01 4.39989427e-03 -6.76497877e-01 4.87021171e-02 -4.38729107e-01 2.25025415e-01 4.77781385e-01 -3.35312665e-01 6.65784299e-01 6.00418150e-01 4.38206978e-02 -1.70085463e-03 -6.64185584e-01 6.44109190e-01 -2.33452499e-01 -1.01604426e+00 -3.25775862e-01 -8.27632487e-01 3.05004448e-01 7.35476673e-01 -2.88672417e-01 -2.77953267e-01 -1.54399782e-01 -6.89502299e-01 5.03954366e-02 6.88666344e-01 5.28150201e-01 6.05167806e-01 -1.12726963e+00 -5.95935464e-01 1.09704562e-01 1.49247438e-01 -3.13461900e-01 -1.06659740e-01 4.54799496e-02 -1.02401900e+00 6.97297096e-01 1.44992899e-02 1.15598999e-01 -1.51529980e+00 7.14244902e-01 2.00506657e-01 -3.75403255e-01 -6.33909941e-01 1.07195497e+00 -1.33353218e-01 -6.09967232e-01 1.92964420e-01 -6.51267231e-01 -5.47658205e-01 5.37080131e-02 6.07524335e-01 4.87337291e-01 6.77308738e-01 -9.01280284e-01 -5.50477326e-01 6.81348026e-01 -3.90397906e-01 -6.23854734e-02 8.86096179e-01 -5.28732657e-01 -4.74075705e-01 2.41485193e-01 3.96133125e-01 3.64852995e-01 -6.59979973e-03 -3.10575515e-01 3.59709859e-01 -2.20015079e-01 -6.58144727e-02 -1.09892035e+00 -3.22426200e-01 6.03794634e-01 -1.10085294e-01 3.87687348e-02 1.29072046e+00 -3.56545776e-01 1.05723846e+00 9.28243995e-01 3.77103001e-01 -1.43911135e+00 -7.76813507e-01 1.19674587e+00 1.03185666e+00 -4.72547829e-01 4.99112248e-01 -8.60116899e-01 -4.75796461e-01 1.22987485e+00 4.88093853e-01 -3.99803579e-01 4.86429840e-01 1.01024859e-01 9.10720080e-02 -4.36037593e-02 -9.11526144e-01 -3.80942285e-01 3.83861125e-01 7.78151631e-01 7.28515983e-01 3.94561648e-01 -1.21932900e+00 6.58743262e-01 -5.71289837e-01 -3.52773786e-01 4.24560457e-01 1.03985286e+00 -6.37214005e-01 -1.53173280e+00 -2.18900591e-01 3.59007359e-01 -3.58126402e-01 -9.31173384e-01 -9.53359723e-01 1.04468489e+00 4.38651055e-01 1.01708114e+00 -2.34920666e-01 -1.98872000e-01 1.13785243e+00 3.76805127e-01 6.82312906e-01 -1.23305726e+00 -1.31652939e+00 2.92792898e-02 7.06131220e-01 -6.88608527e-01 -1.11275250e-02 -8.18582296e-01 -1.32306778e+00 -6.32429779e-01 -5.05844533e-01 1.12194061e-01 4.86422092e-01 1.07066453e+00 4.36074793e-01 6.49053335e-01 1.26488954e-02 -1.09105837e+00 -1.53154805e-01 -1.09192657e+00 -1.61700666e-01 5.13402104e-01 -5.98267354e-02 -6.56775475e-01 -1.06095292e-01 -1.32996157e-01]
[10.909907341003418, 10.387351036071777]
ab76b450-1198-463e-bd27-27f213584cf5
fb-mstcn-a-full-band-single-channel-speech
2203.07684
null
https://arxiv.org/abs/2203.07684v1
https://arxiv.org/pdf/2203.07684v1.pdf
FB-MSTCN: A Full-Band Single-Channel Speech Enhancement Method Based on Multi-Scale Temporal Convolutional Network
In recent years, deep learning-based approaches have significantly improved the performance of single-channel speech enhancement. However, due to the limitation of training data and computational complexity, real-time enhancement of full-band (48 kHz) speech signals is still very challenging. Because of the low energy of spectral information in the high-frequency part, it is more difficult to directly model and enhance the full-band spectrum using neural networks. To solve this problem, this paper proposes a two-stage real-time speech enhancement model with extraction-interpolation mechanism for a full-band signal. The 48 kHz full-band time-domain signal is divided into three sub-channels by extracting, and a two-stage processing scheme of `masking + compensation' is proposed to enhance the signal in the complex domain. After the two-stage enhancement, the enhanced full-band speech signal is restored by interval interpolation. In the subjective listening and word accuracy test, our proposed model achieves superior performance and outperforms the baseline model overall by 0.59 MOS and 4.0% WAcc for the non-personalized speech denoising task.
['Mingjiang Wang', 'Heng Li', 'Yukun Qian', 'Xuyi Zhuang', 'Lu Zhang', 'Zehua Zhang']
2022-03-15
null
null
null
null
['speech-denoising']
['speech']
[ 3.22654575e-01 -2.93121815e-01 1.67210072e-01 -3.75420034e-01 -1.16734433e+00 -1.22270718e-01 -1.20872989e-01 3.29525806e-02 -6.56177640e-01 5.31947792e-01 5.00060976e-01 -2.93868512e-01 -9.62935165e-02 -5.60692728e-01 -3.36711466e-01 -9.99076724e-01 -6.16567321e-02 -8.02749932e-01 8.83397013e-02 -3.72407854e-01 -1.87761754e-01 -1.16625046e-02 -1.56934273e+00 4.04361486e-01 1.17910111e+00 1.17810464e+00 7.17245102e-01 8.59824359e-01 1.83218643e-01 7.59199336e-02 -8.83724034e-01 -2.64785498e-01 5.92765436e-02 -4.67399746e-01 -1.59866482e-01 -1.06714284e-02 -5.12073562e-02 -6.56148255e-01 -6.09528720e-01 1.52572274e+00 1.28809714e+00 3.51234496e-01 9.89435241e-02 -7.34670520e-01 -4.42818165e-01 6.19288146e-01 -5.25373280e-01 3.01750600e-01 4.98599280e-03 -7.19593689e-02 6.40789688e-01 -9.96340454e-01 -6.92603067e-02 1.10360301e+00 9.76718068e-01 3.29513460e-01 -9.42631721e-01 -8.85432363e-01 -2.51042068e-01 6.38458371e-01 -1.16293454e+00 -8.04956317e-01 1.15026474e+00 8.45355839e-02 9.94815707e-01 2.18228847e-01 4.91742194e-01 7.92772174e-01 -4.34229933e-02 5.49233615e-01 1.20464647e+00 -5.57728708e-01 4.77038287e-02 -1.26635715e-01 -8.60442296e-02 7.22441822e-02 -2.38571599e-01 5.45441866e-01 -6.12165809e-01 2.05363616e-01 5.16357005e-01 -3.70341063e-01 -6.70142174e-01 7.36618638e-01 -8.10716510e-01 4.34094012e-01 3.40222627e-01 5.60231149e-01 -5.92917264e-01 -1.12874277e-01 5.09554744e-01 4.26191360e-01 8.57631266e-01 -4.84946091e-03 -6.10910118e-01 -3.14117700e-01 -1.23936546e+00 7.80759528e-02 3.57601404e-01 4.77444261e-01 3.82844061e-01 5.98214567e-01 -1.05074398e-01 1.38521087e+00 3.69554609e-01 3.67418349e-01 5.74630260e-01 -7.84523129e-01 6.16640389e-01 -4.67133343e-01 7.58813694e-02 -7.50199497e-01 -2.65050799e-01 -9.44308877e-01 -1.27500451e+00 1.36017695e-01 1.51779920e-01 -5.21687567e-01 -6.60643995e-01 1.73444557e+00 3.94082874e-01 7.13696629e-02 2.60469705e-01 1.13711214e+00 7.53853261e-01 1.28194273e+00 1.58519581e-01 -8.54511619e-01 1.30156446e+00 -9.73360419e-01 -1.55754066e+00 -1.07080303e-01 -4.05594632e-02 -1.14616978e+00 9.14902985e-01 7.03694403e-01 -1.35068297e+00 -1.07942319e+00 -1.22471786e+00 -2.66415685e-01 -1.57528482e-02 3.74017268e-01 1.72095284e-01 9.93685663e-01 -1.16114724e+00 6.30946815e-01 -4.02414620e-01 3.80187362e-01 1.91099331e-01 1.34778231e-01 -1.56225115e-01 1.62210509e-01 -1.77392769e+00 5.71902931e-01 3.85117263e-01 3.71729106e-01 -5.44897974e-01 -8.86068285e-01 -8.40160787e-01 4.00480747e-01 9.41079482e-02 -1.99600905e-01 1.51288354e+00 -8.59250605e-01 -1.71964133e+00 3.01720887e-01 -4.21067774e-01 -4.14014161e-01 2.04639554e-01 -2.64881432e-01 -1.19014359e+00 1.97766036e-01 -3.38550776e-01 1.51668981e-01 1.30706799e+00 -9.78746295e-01 -7.71946073e-01 -3.30633789e-01 -3.53815615e-01 3.28426629e-01 -6.57732010e-01 1.32529691e-01 -2.14162216e-01 -1.36269677e+00 4.05109018e-01 -1.48952276e-01 -4.93773259e-02 -2.03594133e-01 -5.54733463e-02 1.89223245e-01 9.34875011e-01 -1.72628236e+00 1.64778316e+00 -2.76384592e+00 -2.36209065e-01 -2.14000717e-01 -1.30447999e-01 7.62980878e-01 -1.53507441e-01 1.75360590e-01 -2.76986241e-01 -1.12762921e-01 -2.99254715e-01 -5.97846150e-01 7.99373612e-02 -3.01482856e-01 -1.84173688e-01 3.75931948e-01 7.64238983e-02 1.93837881e-01 -6.55344427e-01 -3.13823730e-01 2.30267778e-01 1.17995393e+00 -4.89693969e-01 3.08180362e-01 4.19085532e-01 3.77578676e-01 2.14713797e-01 4.81944025e-01 1.18218422e+00 5.53036153e-01 -1.59846276e-01 -6.95276380e-01 -2.80813485e-01 6.00238681e-01 -1.40260756e+00 1.66486359e+00 -7.11638987e-01 7.86837637e-01 8.08374703e-01 -8.00225854e-01 8.54910672e-01 8.17574799e-01 1.43281117e-01 -9.92281377e-01 8.90883952e-02 4.49350983e-01 2.80683309e-01 -6.46060050e-01 4.94912624e-01 -4.71519798e-01 3.78483236e-01 1.06635846e-01 5.72765507e-02 -1.95839614e-01 -1.52471617e-01 -3.04185182e-01 6.37041748e-01 -3.25360149e-01 1.86374038e-01 9.13591981e-02 7.88579345e-01 -7.25401580e-01 6.37045205e-01 2.65196860e-01 -6.67911828e-01 5.08143783e-01 -1.33863702e-01 3.17660481e-01 -1.13466954e+00 -1.04224145e+00 -2.78027028e-01 1.27953434e+00 -3.85890901e-02 -1.41328484e-01 -1.03038180e+00 1.24459453e-01 -3.54775935e-01 6.82354510e-01 -3.52844410e-02 -3.07560861e-01 -6.13788068e-01 -7.33729482e-01 6.79617941e-01 4.43637908e-01 9.88892555e-01 -8.58084679e-01 1.88780818e-02 7.23931611e-01 -6.07931554e-01 -9.98112023e-01 -8.67515445e-01 2.60938942e-01 -7.74499059e-01 -2.65562952e-01 -1.02917647e+00 -1.22666192e+00 2.59615511e-01 3.29712719e-01 4.34485346e-01 -8.49341676e-02 1.03464248e-02 -2.97967851e-01 -4.00561363e-01 -3.84413183e-01 -4.55313414e-01 -3.60129148e-01 8.58873129e-02 2.01814145e-01 1.11506358e-01 -6.70853734e-01 -8.02144349e-01 1.48798332e-01 -8.55719149e-01 -1.82687119e-01 7.63964295e-01 1.07374632e+00 4.93389279e-01 1.02288735e+00 1.20849597e+00 1.64046735e-01 1.22015250e+00 -2.44100112e-02 -4.13407266e-01 -2.13725433e-01 -2.75592953e-01 -5.52628815e-01 6.74002945e-01 -7.83007979e-01 -1.74686158e+00 -2.71640033e-01 -9.73553002e-01 4.15383577e-02 -6.98517114e-02 6.54968381e-01 -6.26473427e-01 1.58904418e-01 4.14358646e-01 5.78789830e-01 2.03758199e-02 -7.80559719e-01 5.59467562e-02 1.46251965e+00 8.49125445e-01 -9.75509360e-02 7.09071696e-01 7.22326040e-02 -4.43654388e-01 -1.14424145e+00 -3.82156879e-01 -4.65442628e-01 -6.57916516e-02 -1.63532227e-01 7.88905323e-01 -1.14422143e+00 -4.55635130e-01 1.01801872e+00 -1.23215222e+00 -3.24913979e-01 -1.59403786e-01 9.30060983e-01 -2.27539271e-01 7.10652649e-01 -1.14465237e+00 -1.14468324e+00 -8.16723049e-01 -1.13216746e+00 7.02234864e-01 2.94977158e-01 2.11466834e-01 -5.56135476e-01 -3.26801687e-01 2.72240937e-01 9.73973215e-01 -5.16951501e-01 7.28984058e-01 -7.37773106e-02 8.23097005e-02 -2.17884779e-01 -6.10793056e-03 9.68796909e-01 2.57606119e-01 -5.28554499e-01 -1.42967010e+00 -3.50097597e-01 7.93923676e-01 4.26440351e-02 5.84122837e-01 8.01799357e-01 1.40150452e+00 -2.41314858e-01 2.60457337e-01 6.98256195e-01 1.10920358e+00 5.76036870e-01 8.94680440e-01 2.03933697e-02 -6.42018914e-02 5.84302008e-01 7.54654527e-01 3.83750856e-01 2.24007107e-02 4.28001642e-01 1.29975215e-01 -3.89992028e-01 -5.69803655e-01 -1.39112085e-01 3.80142182e-01 1.38987374e+00 6.05207644e-02 -6.86385334e-02 -3.17525625e-01 6.08461916e-01 -1.07119930e+00 -1.14285123e+00 -1.52374983e-01 2.04155397e+00 1.42243898e+00 8.77829343e-02 -3.58678922e-02 9.43854749e-01 1.06522739e+00 2.48367324e-01 -4.59656447e-01 -3.56120437e-01 -4.10643011e-01 4.71420050e-01 2.56110877e-01 7.38671601e-01 -1.12407911e+00 3.92263204e-01 5.41966248e+00 1.39545453e+00 -1.27665055e+00 4.12021101e-01 7.62511671e-01 7.44956881e-02 1.14365630e-01 -6.26225531e-01 -5.40557504e-01 6.22577965e-01 1.06025112e+00 -2.73883075e-01 7.30937958e-01 4.64653641e-01 7.76196599e-01 1.51176959e-01 -3.39853346e-01 1.12239492e+00 -1.68497935e-01 -7.36528456e-01 -5.37784457e-01 -1.43489078e-01 4.35738832e-01 -3.92048419e-01 3.17081898e-01 4.99702126e-01 -6.33270264e-01 -8.16649020e-01 8.02585721e-01 8.77379999e-02 1.21549416e+00 -9.74693596e-01 7.09344447e-01 4.99261349e-01 -1.46040356e+00 -2.68459857e-01 -2.26219997e-01 -2.26956997e-02 3.52214843e-01 9.61099207e-01 -5.25468051e-01 5.02285063e-01 8.78358662e-01 1.18170269e-01 2.89656162e-01 1.23709702e+00 -3.64583045e-01 9.54494357e-01 -1.74747571e-01 3.40917647e-01 -1.47669807e-01 -1.49263749e-02 5.56446493e-01 1.45183623e+00 5.26984215e-01 5.04832923e-01 -5.02998292e-01 3.99571151e-01 -1.20922767e-01 -7.17307851e-02 1.75972581e-01 1.20519653e-01 5.74825287e-01 1.12764037e+00 2.73018181e-02 -2.71785170e-01 -3.63863409e-01 8.82906795e-01 -4.41577673e-01 6.76643074e-01 -9.35478508e-01 -1.17540717e+00 5.83295465e-01 -1.31193921e-01 2.93176800e-01 -2.39089936e-01 -3.06038439e-01 -6.64498627e-01 2.19706163e-01 -1.18124378e+00 -4.43194844e-02 -8.96362722e-01 -1.09893477e+00 7.04321563e-01 -5.49634755e-01 -1.20871162e+00 -3.78040336e-02 -3.60303104e-01 -5.08886874e-01 1.38668382e+00 -1.87875021e+00 -7.12067306e-01 -1.57693744e-01 7.16644228e-01 8.72706711e-01 1.33723646e-01 6.88091457e-01 1.06712735e+00 -2.74855673e-01 7.94833124e-01 2.76281029e-01 -1.30362272e-01 7.08767295e-01 -1.01470792e+00 3.04913521e-01 1.03147066e+00 -5.91648102e-01 4.57500488e-01 7.83047974e-01 -5.71647048e-01 -1.01108491e+00 -1.08212709e+00 1.07450581e+00 7.72951961e-01 3.64998549e-01 -2.72995412e-01 -1.23290718e+00 -9.21183005e-02 3.80742878e-01 -2.07630053e-01 6.32178307e-01 -4.36006308e-01 -9.07445420e-03 -4.61459607e-01 -1.33060336e+00 5.51411092e-01 6.42708361e-01 -8.73101532e-01 -6.31949723e-01 -9.15970095e-03 1.06117356e+00 -4.91493762e-01 -9.46382344e-01 4.55419093e-01 3.53403777e-01 -7.52951086e-01 1.10878944e+00 1.56623609e-02 1.80794850e-01 -4.52373236e-01 -3.39429140e-01 -1.67996347e+00 -1.40867427e-01 -1.04212487e+00 -7.62081742e-02 1.41405678e+00 3.48043591e-01 -7.04105377e-01 2.86253244e-01 -7.58737465e-03 -5.70972085e-01 -4.11706388e-01 -1.02445662e+00 -6.14202619e-01 -2.22692579e-01 -6.65965617e-01 5.54995179e-01 6.76583469e-01 1.79980889e-01 2.52554286e-02 -5.50014675e-01 5.68772435e-01 7.11757779e-01 -4.27978545e-01 1.76908284e-01 -6.11612439e-01 -4.94079888e-01 -3.39665562e-01 7.98215494e-02 -1.38487732e+00 -1.17338896e-01 -2.48241022e-01 4.25634712e-01 -1.42142403e+00 -2.51078784e-01 7.47310594e-02 -3.78263950e-01 -1.73486937e-02 -4.49402779e-01 1.23109572e-01 -3.88842821e-02 -2.36794159e-01 1.87573075e-01 8.77988338e-01 1.39342308e+00 -2.78076440e-01 -2.64772981e-01 1.84808776e-01 -4.74385172e-01 5.88660300e-01 8.79477084e-01 -1.72273561e-01 -3.31299305e-01 -4.97229308e-01 -3.94989550e-01 5.14682710e-01 1.24973103e-01 -1.04847515e+00 3.55361789e-01 2.96048433e-01 3.46941471e-01 -8.13786507e-01 7.84878314e-01 -8.10507596e-01 -9.11841821e-03 4.74264145e-01 -3.72255594e-01 -4.12573189e-01 4.85893041e-01 4.69421566e-01 -7.40980148e-01 -4.46606502e-02 1.04543412e+00 2.45524153e-01 -4.56946611e-01 7.32790455e-02 -4.40273613e-01 -3.24635059e-01 4.49973911e-01 -1.73251957e-01 -2.48975754e-01 -6.78097188e-01 -9.44497526e-01 -4.07075077e-01 -2.78895766e-01 5.92922643e-02 7.75978625e-01 -1.19403625e+00 -8.71766925e-01 3.94741744e-01 -6.72985077e-01 -2.39503786e-01 1.03763068e+00 8.41126800e-01 -1.99931309e-01 1.00265764e-01 3.55872931e-03 -2.48705149e-01 -1.31455398e+00 2.93803722e-01 6.34166896e-01 1.12957126e-02 -3.84544790e-01 9.54556406e-01 -5.15649058e-02 1.37731776e-01 5.31432688e-01 -1.79430678e-01 -1.87101096e-01 -1.27497330e-01 9.11769986e-01 6.06570542e-01 4.11887407e-01 -7.44485378e-01 1.80196855e-02 3.35119039e-01 1.84941456e-01 -5.27928889e-01 1.40055466e+00 -4.36913341e-01 -7.32594877e-02 -2.15178262e-02 1.33657432e+00 3.26861650e-01 -1.19726264e+00 -2.24885821e-01 -5.18114746e-01 -5.42090714e-01 7.26493180e-01 -1.06747854e+00 -1.10432422e+00 1.23935044e+00 9.92921233e-01 4.44914609e-01 1.98388875e+00 -6.96934223e-01 1.34763253e+00 -1.45167693e-01 -5.91214821e-02 -1.33373737e+00 1.43901929e-01 4.05985296e-01 1.13302243e+00 -9.85028327e-01 -3.49552423e-01 -5.38691401e-01 -1.04116186e-01 1.08734858e+00 3.46084803e-01 3.37278694e-01 6.88981712e-01 5.12394845e-01 2.81508833e-01 6.00375772e-01 -2.12572828e-01 -5.23959324e-02 8.25699717e-02 7.55607069e-01 4.92514402e-01 -6.95715323e-02 -4.85432148e-01 1.06236088e+00 -4.82106388e-01 -1.79354459e-01 2.06874058e-01 5.45943141e-01 -7.22375214e-01 -8.78632665e-01 -9.26230252e-01 1.30593464e-01 -8.91733587e-01 -3.84136349e-01 3.23715210e-01 2.15158597e-01 8.92429426e-02 1.73990512e+00 -1.57494783e-01 -4.09010082e-01 5.22035241e-01 -1.87579133e-02 1.14510536e-01 3.41346152e-02 -6.36329949e-01 9.72783148e-01 6.09619580e-02 -1.79843456e-01 -1.66641548e-01 -3.54351491e-01 -1.15530264e+00 -2.07913429e-01 -6.67931259e-01 1.80816576e-01 1.14087474e+00 5.30896246e-01 1.81797475e-01 9.29826081e-01 7.73765922e-01 -9.38823998e-01 -9.37696993e-01 -1.32928669e+00 -9.17202830e-01 3.43801707e-01 9.03262496e-01 -1.09717868e-01 -6.63439214e-01 1.98395342e-01]
[14.978921890258789, 5.910974979400635]
635d0d33-fb21-42ca-b4e5-e27b7626ab34
openvidial-2-0-a-larger-scale-open-domain
2109.12761
null
https://arxiv.org/abs/2109.12761v2
https://arxiv.org/pdf/2109.12761v2.pdf
OpenViDial 2.0: A Larger-Scale, Open-Domain Dialogue Generation Dataset with Visual Contexts
In order to better simulate the real human conversation process, models need to generate dialogue utterances based on not only preceding textual contexts but also visual contexts. However, with the development of multi-modal dialogue learning, the dataset scale gradually becomes a bottleneck. In this report, we release OpenViDial 2.0, a larger-scale open-domain multi-modal dialogue dataset compared to the previous version OpenViDial 1.0. OpenViDial 2.0 contains a total number of 5.6 million dialogue turns extracted from either movies or TV series from different resources, and each dialogue turn is paired with its corresponding visual context. We hope this large-scale dataset can help facilitate future researches on open-domain multi-modal dialog generation, e.g., multi-modal pretraining for dialogue generation.
['Jiwei Li', 'Rongbin Ouyang', 'Xiaofei Sun', 'Xiaoya Li', 'Yuxian Meng', 'Shuhe Wang']
2021-09-27
null
null
null
null
['multi-modal-dialogue-generation']
['natural-language-processing']
[-2.23303393e-01 2.10355178e-01 1.45785362e-01 -3.76883239e-01 -9.43198204e-01 -9.39878047e-01 9.02828336e-01 -3.96794677e-01 4.49858652e-03 9.59248126e-01 8.42901111e-01 -5.63414618e-02 8.25540483e-01 -8.63390565e-01 -1.13701962e-01 -4.44872230e-01 5.15130758e-01 8.36869001e-01 1.57921582e-01 -6.91151679e-01 1.85778979e-02 -1.28813922e-01 -1.03266943e+00 9.23789620e-01 6.29491210e-01 6.26621783e-01 5.63907623e-01 8.05893660e-01 -4.71338958e-01 8.13158274e-01 -1.04915094e+00 -5.31925321e-01 -2.88005024e-01 -9.85426009e-01 -1.15040183e+00 2.54921496e-01 1.08786374e-01 -6.76568270e-01 -3.64607126e-01 6.02028251e-01 9.32148814e-01 3.76621991e-01 7.49686182e-01 -1.13711345e+00 -8.56133997e-01 6.90043271e-01 -3.26482087e-01 -1.02256097e-01 8.71557295e-01 5.90789258e-01 1.04683077e+00 -8.01193237e-01 1.10421181e+00 1.67876291e+00 1.53924689e-01 1.05963516e+00 -1.00649142e+00 -3.85246634e-01 5.14040999e-02 3.02046109e-02 -8.92726421e-01 -5.02130747e-01 9.93229568e-01 -4.02126789e-01 6.73548281e-01 1.88337028e-01 5.47010720e-01 1.81167340e+00 -2.11020216e-01 1.08975422e+00 1.05307662e+00 -2.86814332e-01 -2.87067890e-01 2.89173275e-01 -3.16346973e-01 5.39989769e-01 -7.93684542e-01 -3.62546027e-01 -2.82162011e-01 -1.43157244e-01 1.05340600e+00 -5.65914512e-01 -1.80852681e-01 -1.32840693e-01 -1.53733599e+00 1.27090812e+00 1.75554603e-01 3.30805540e-01 -8.67171958e-02 -5.12156427e-01 8.77707124e-01 4.12875295e-01 3.61500263e-01 4.86537367e-01 -1.19647190e-01 -4.92167860e-01 -3.21810782e-01 3.74033153e-01 8.69909346e-01 9.92577910e-01 6.95028603e-01 1.16286233e-01 -5.69181740e-01 1.32169485e+00 3.42935294e-01 4.45107937e-01 4.94585097e-01 -9.95887935e-01 8.78514707e-01 6.47551060e-01 5.09507768e-02 -8.73417258e-01 -4.96324688e-01 6.22106493e-01 -1.06318426e+00 -2.24388584e-01 7.49279737e-01 -9.16632295e-01 -2.54572183e-01 1.65465021e+00 5.86965501e-01 -2.77082592e-01 5.92845738e-01 9.47088301e-01 1.75732219e+00 1.24731529e+00 -1.00886822e-01 -3.33295882e-01 1.56886268e+00 -1.04066086e+00 -9.66897011e-01 -6.78311288e-02 5.05142748e-01 -1.06872034e+00 1.58707595e+00 2.04158694e-01 -1.00094581e+00 -7.26686299e-01 -7.34865546e-01 -2.99982816e-01 -1.39926195e-01 1.70892894e-01 4.13647532e-01 2.10465401e-01 -6.51727319e-01 -2.56864548e-01 -5.22063375e-02 -1.17251746e-01 -9.52188820e-02 -4.22430277e-01 -2.98523545e-01 1.74347579e-01 -1.69236755e+00 7.78976262e-01 3.25852692e-01 -7.38618569e-03 -9.08031344e-01 -3.33786421e-02 -1.07480395e+00 -4.80013132e-01 3.83328736e-01 -4.83589411e-01 1.66418266e+00 -7.92708397e-01 -1.78790486e+00 1.03069997e+00 -9.24667940e-02 -1.15353860e-01 7.50008583e-01 -9.18938145e-02 -3.46246302e-01 3.62868786e-01 8.66240114e-02 7.09722996e-01 6.69277847e-01 -1.48183763e+00 -5.57101846e-01 -9.03399140e-02 5.84613502e-01 6.81886494e-01 -3.16210911e-02 -4.57323110e-03 -5.53655922e-01 -5.55340767e-01 -5.49298346e-01 -9.86371100e-01 -1.90627500e-01 -4.42384660e-01 -5.58170557e-01 -5.44035912e-01 9.23758149e-01 -6.50154948e-01 9.80543256e-01 -1.99173713e+00 4.59441125e-01 -5.53661764e-01 3.34067136e-01 1.07049003e-01 -1.58247128e-01 7.26959288e-01 3.94914299e-01 -1.62555620e-01 8.70780945e-02 -3.82360607e-01 7.11928084e-02 -1.05851777e-02 -2.83881813e-01 -4.93012704e-02 -1.40929043e-01 9.63134885e-01 -9.33627605e-01 -9.23076153e-01 3.52343231e-01 9.53156576e-02 -1.43324867e-01 7.49712646e-01 -6.81801379e-01 8.80324006e-01 -4.80283976e-01 1.98049709e-01 3.90495032e-01 -3.90837967e-01 2.81835914e-01 -2.16114536e-01 -2.22798973e-01 1.96290046e-01 -7.79031157e-01 1.96300149e+00 -7.85698235e-01 9.82340157e-01 1.80714399e-01 -5.42964518e-01 1.14307046e+00 7.95479000e-01 1.36476561e-01 -8.12500715e-01 2.12992221e-01 -4.08365339e-01 -1.16570950e-01 -7.80227661e-01 9.20186102e-01 -2.00181410e-01 -7.39938319e-01 7.38592386e-01 1.40573680e-01 -3.90511870e-01 3.78496319e-01 6.09781384e-01 2.79115528e-01 -7.18021318e-02 3.53763908e-01 1.92773819e-01 5.40740013e-01 2.55141616e-01 3.85436684e-01 3.33336055e-01 -2.95225114e-01 7.19957530e-01 9.35038567e-01 -1.33103400e-01 -1.10079765e+00 -1.05427527e+00 5.48003130e-02 1.24057448e+00 2.09015921e-01 -3.08743030e-01 -7.80520201e-01 -7.03774393e-01 -4.79095221e-01 5.42556703e-01 -4.81692940e-01 2.09500566e-01 -6.32785618e-01 -5.25328755e-01 7.72475183e-01 2.42636338e-01 7.49674737e-01 -1.63972437e+00 -7.40862265e-02 2.88145512e-01 -9.30303514e-01 -1.21046722e+00 -6.89595461e-01 -6.70522392e-01 -2.67856032e-01 -1.04435694e+00 -1.13478613e+00 -1.16642857e+00 1.37925148e-01 1.92860156e-01 1.46402633e+00 -3.54871392e-01 8.30544978e-02 2.77573884e-01 -6.90778315e-01 -1.21011592e-01 -9.51498091e-01 1.41058087e-01 -3.62423420e-01 -1.50320530e-01 1.86044127e-01 -1.78687871e-01 -3.37559432e-01 5.42126417e-01 -5.46484113e-01 8.20192635e-01 2.03956351e-01 1.06642318e+00 1.90445989e-01 -4.20242250e-01 1.02505684e+00 -8.64188492e-01 1.39517450e+00 -6.26485109e-01 -2.20202684e-01 4.16785121e-01 2.84023672e-01 -3.19779366e-01 7.68479407e-01 -5.19653618e-01 -1.73184097e+00 -1.50940359e-01 -2.93238789e-01 -1.08902231e-01 -5.47827542e-01 4.62050706e-01 -3.83351982e-01 7.42056608e-01 6.02568209e-01 3.10393482e-01 1.93367582e-02 -3.15969884e-01 7.05350578e-01 1.01364779e+00 5.44240296e-01 -7.30610549e-01 3.33254546e-01 -4.33881059e-02 -5.63196301e-01 -1.25141943e+00 -5.25448859e-01 -1.54398307e-01 -6.42543852e-01 -5.90093613e-01 1.23400247e+00 -1.04505670e+00 -8.86758029e-01 6.49300754e-01 -1.53477800e+00 -8.05376351e-01 2.59070128e-01 6.83846474e-02 -4.59731221e-01 5.49383044e-01 -1.06151903e+00 -7.27789879e-01 -3.47148627e-01 -1.05369532e+00 8.42133522e-01 4.65624511e-01 -3.76153499e-01 -1.22058570e+00 4.91877437e-01 7.14841902e-01 -5.25649227e-02 2.07712874e-01 6.37160897e-01 -5.52681804e-01 -1.36042088e-01 2.82216668e-01 -2.55183667e-01 2.87577137e-02 2.05685824e-01 1.80317596e-01 -8.19178164e-01 8.05493444e-02 -1.77192241e-01 -1.07371223e+00 2.18310058e-01 5.01118340e-02 6.05774879e-01 -4.86585021e-01 1.10710867e-01 -9.54446420e-02 7.07570374e-01 4.64781910e-01 5.63398302e-01 -7.94260204e-02 8.65373909e-01 9.08294857e-01 9.72176254e-01 7.46173680e-01 9.43580151e-01 7.07773685e-01 1.86332062e-01 -1.51172340e-01 1.11911051e-01 -3.64417881e-01 2.15150207e-01 1.07385719e+00 9.29182246e-02 -4.98693824e-01 -8.27099144e-01 5.58067799e-01 -1.79063272e+00 -1.35992682e+00 -2.33777076e-01 1.54999113e+00 1.34191263e+00 -1.61874201e-02 6.49763823e-01 -4.82958436e-01 8.79342079e-01 6.71027660e-01 -5.42177439e-01 -3.17176938e-01 -3.96205962e-01 -5.23340940e-01 -5.92445195e-01 6.12889349e-01 -1.04409540e+00 1.12875867e+00 5.89383841e+00 7.67650723e-01 -1.03368974e+00 -9.34877805e-03 7.47958779e-01 -3.34510095e-02 -3.30289006e-01 -2.67234057e-01 -8.39861512e-01 5.17683089e-01 8.17831099e-01 -3.53086114e-01 2.44643420e-01 8.42126608e-01 2.56396025e-01 -2.58101225e-02 -8.11874747e-01 1.08085370e+00 9.00257304e-02 -1.46762085e+00 1.12867579e-01 -1.40791729e-01 6.68725252e-01 -3.28287870e-01 -8.96656513e-02 6.79947674e-01 5.51585853e-01 -8.62626493e-01 2.43101373e-01 3.50392252e-01 1.05332816e+00 -7.89884269e-01 5.37705481e-01 4.66918021e-01 -1.46893799e+00 5.32312691e-01 -1.92877769e-01 -8.03726614e-02 7.47913301e-01 8.94518867e-02 -1.10152161e+00 3.83167952e-01 1.60749599e-01 6.22963250e-01 -3.36207509e-01 3.24986339e-01 -1.96522102e-01 2.81514376e-01 1.32125378e-01 -4.52738881e-01 3.65664095e-01 -2.86821842e-01 4.19019431e-01 1.28826499e+00 -9.66231972e-02 3.61327440e-01 4.60403949e-01 5.64904928e-01 -8.25696811e-02 2.68059880e-01 -7.75139809e-01 -2.66436428e-01 6.81371570e-01 1.51059949e+00 -3.34990770e-01 -7.51843452e-01 -6.65810585e-01 1.17562211e+00 2.59175807e-01 2.50245571e-01 -8.80309343e-01 -3.93117100e-01 5.77181816e-01 -3.71012837e-01 -1.99934795e-01 -2.26191312e-01 -4.70001698e-02 -1.30872965e+00 -3.77254814e-01 -1.32243967e+00 4.63340491e-01 -1.01635933e+00 -1.42941523e+00 1.03651154e+00 -4.40457799e-02 -1.34551775e+00 -9.04072106e-01 -3.65167886e-01 -7.45421946e-01 8.93746436e-01 -7.36952543e-01 -1.16163194e+00 -2.66117454e-01 8.76016736e-01 1.29348421e+00 -3.33849609e-01 1.12792444e+00 6.03313521e-02 -6.75460160e-01 4.99354064e-01 -1.04249001e-01 5.91317177e-01 1.21077073e+00 -1.30410016e+00 5.90378106e-01 1.72431156e-01 -7.03236163e-02 2.64394432e-01 7.73231030e-01 -5.06181538e-01 -9.14304972e-01 -5.79682112e-01 5.11024058e-01 -4.45260644e-01 7.32222497e-01 -5.34229398e-01 -9.46100175e-01 6.39240563e-01 1.05315971e+00 -7.20246673e-01 8.15455377e-01 1.33452475e-01 8.75282586e-02 3.97534072e-01 -9.90610242e-01 1.00982809e+00 8.52885842e-01 -7.41893888e-01 -9.94529009e-01 2.42575988e-01 8.72794569e-01 -9.41053927e-01 -1.03171074e+00 -3.23510878e-02 4.72356319e-01 -1.02404594e+00 7.59872556e-01 -5.17657518e-01 9.18011427e-01 1.76204965e-01 -1.30383400e-02 -1.63845682e+00 7.32850209e-02 -8.21009457e-01 5.15748747e-02 1.67414582e+00 4.70707953e-01 -2.13426083e-01 6.07500196e-01 4.20447290e-01 -1.67427018e-01 -4.04329151e-01 -6.12852693e-01 3.57911997e-02 1.36030599e-01 6.20579720e-02 4.75212336e-01 1.13461876e+00 4.45468217e-01 1.39471197e+00 -9.56822634e-01 -3.27914834e-01 1.50464416e-01 6.35616839e-01 1.47549415e+00 -1.17315412e+00 -3.90437901e-01 -2.59471059e-01 3.23996007e-01 -1.73423147e+00 1.77592143e-01 -3.02614033e-01 7.66318142e-02 -1.68936348e+00 9.61566251e-03 -2.79441595e-01 6.15312397e-01 2.02783078e-01 -2.77889967e-01 6.13343194e-02 2.43317634e-01 4.27289866e-02 -8.32395792e-01 9.81174111e-01 1.95775867e+00 -1.29609898e-01 -4.74791676e-01 1.86791644e-02 -5.04873395e-01 7.07402945e-01 9.71407533e-01 2.73919731e-01 -6.30170286e-01 -2.10525692e-01 -8.93153250e-02 1.05602968e+00 5.66509515e-02 -4.61643726e-01 -1.30130142e-01 -5.71519852e-01 2.91536391e-01 -8.20741713e-01 7.99031615e-01 -1.86240807e-01 -2.57185578e-01 -3.00467983e-02 -6.13341093e-01 1.43766329e-01 2.58021832e-01 2.84926057e-01 -4.05934781e-01 -5.44090793e-02 6.41936660e-01 -4.93820190e-01 -1.04696786e+00 9.59063843e-02 -7.71211624e-01 6.05017304e-01 9.95771527e-01 1.93887800e-01 -7.55940795e-01 -1.05364418e+00 -9.49729621e-01 5.89840949e-01 2.97347754e-01 7.25322843e-01 6.66630089e-01 -1.58243203e+00 -9.11321700e-01 -2.61071831e-01 1.97009385e-01 1.87241718e-01 8.77488971e-01 1.98854536e-01 -2.88618982e-01 3.70276451e-01 -5.06011367e-01 -4.37575132e-01 -1.61981535e+00 2.94879556e-01 4.87231873e-02 -3.85984540e-01 -5.63607574e-01 7.74466157e-01 5.38561285e-01 -7.09978878e-01 -2.55072638e-02 3.46320003e-01 -5.97967148e-01 3.90507638e-01 7.94548988e-01 1.19887911e-01 -7.10168898e-01 -1.03650331e+00 1.90489233e-01 3.81910205e-02 -1.08124986e-01 -6.33752823e-01 7.43975580e-01 -5.84999800e-01 1.54860094e-01 1.01597822e+00 9.93118942e-01 1.67982534e-01 -1.40451229e+00 -3.04398775e-01 -5.49088240e-01 -2.21417919e-01 -6.18725359e-01 -7.02273369e-01 -8.28231454e-01 9.42819178e-01 6.94579184e-02 6.42236114e-01 6.99152708e-01 2.00898990e-01 9.73397315e-01 5.19702077e-01 2.68911064e-01 -1.15920877e+00 6.46074414e-01 8.49558830e-01 1.32815385e+00 -1.54671741e+00 -3.08191478e-01 -1.22664496e-01 -1.71687210e+00 9.34063792e-01 1.25896931e+00 1.73387021e-01 1.22999653e-01 -7.42892623e-02 7.04884946e-01 -1.33238927e-01 -9.06131864e-01 -2.52317097e-02 1.08736726e-02 7.97363579e-01 7.43762553e-01 1.84037119e-01 -1.32560402e-01 7.51502454e-01 -5.03206015e-01 -5.69960237e-01 6.42240942e-01 4.17086631e-01 -4.21610206e-01 -1.12399697e+00 -5.04094362e-01 2.37711132e-01 -7.16971755e-02 -6.80596083e-02 -8.34406793e-01 8.39624941e-01 -3.87635767e-01 1.27266526e+00 1.85284123e-01 -4.34982151e-01 3.41157466e-01 -1.28304614e-02 2.68656582e-01 -6.17479861e-01 -4.02196467e-01 1.33885682e-01 7.37973273e-01 -1.33717373e-01 -3.23522061e-01 -5.56227088e-01 -1.33071876e+00 -5.80009103e-01 1.50590828e-02 2.17268616e-01 1.05074756e-01 7.88480997e-01 2.25969315e-01 4.75102216e-01 7.94705808e-01 -9.34603631e-01 -1.80730954e-01 -1.39752209e+00 -3.73923838e-01 4.48411614e-01 8.32018107e-02 -5.14267445e-01 -4.26302617e-03 3.64895687e-02]
[12.827534675598145, 7.838142395019531]
c8e52a9f-bff6-4983-8611-80f18c60c846
robust-optimization-structure-control-co
2306.08472
null
https://arxiv.org/abs/2306.08472v1
https://arxiv.org/pdf/2306.08472v1.pdf
Robust Optimization, Structure/Control co-design, Distributed Optimization, Monolithic Optimization, Robust Control, Parametric Uncertainty
This paper presents an end-to-end framework for robust structure/control optimization of an industrial benchmark. When dealing with space structures, a reduction of the spacecraft mass is paramount to minimize the mission cost and maximize the propellant availability. However, a lighter design comes with a bigger structural flexibility and the resulting impact on control performance. Two optimization architectures (distributed and monolithic) are proposed in order to face this issue. In particular the Linear Fractional Transformation (LFT) framework is exploited to formally set the two optimization problems by including parametric uncertainties. Large sets of uncertainties have to be indeed taken into account in spacecraft control design due to the impossibility to completely validate structural models in micro-gravity conditions with on-ground experiments and to the evolution of spacecraft dynamics during the mission (structure degradation and fuel consumption). In particular the Two-Input Two-Output Port (TITOP) multi-body approach is used to build the flexible dynamics in a minimal LFT form. The two proposed optimization algorithms are detailed and their performance are compared on an ESA future exploration mission, the ENVISION benchmark. With both approaches, an important reduction of the mass is obtained by coping with the mission's control performance/stability requirements and a large set of uncertainties.
['Finn Ankersen', 'Pedro Simplicio', 'Mark Watt', 'Andy Kiley', 'Daniel Alazard', 'Francesco Sanfedino']
2023-06-14
null
null
null
null
['distributed-optimization']
['methodology']
[-1.37388840e-01 4.29405957e-01 3.10984999e-01 1.31671965e-01 7.96022117e-02 -8.31474781e-01 7.22546399e-01 2.89435148e-01 -3.21539193e-01 1.15821326e+00 -2.56317496e-01 -1.17913522e-01 -1.09100306e+00 -5.31601846e-01 -4.96874690e-01 -9.03776109e-01 -1.66808784e-01 8.14032555e-01 -2.64413804e-01 -5.32745063e-01 5.19668050e-02 9.23245132e-01 -1.46199620e+00 -7.07328320e-01 9.98794913e-01 1.09985662e+00 3.04533780e-01 1.63309053e-01 1.93782076e-01 1.87816709e-01 -4.72570390e-01 1.94644436e-01 5.44136286e-01 4.60064039e-02 -5.08262992e-01 3.67242247e-01 -3.02856028e-01 2.17545897e-01 2.98824281e-01 1.03878629e+00 4.26089644e-01 6.36182845e-01 7.11266637e-01 -1.11442149e+00 6.39630377e-01 2.64308155e-01 -1.77215468e-02 -2.09186539e-01 5.02541736e-02 1.89820796e-01 5.22704840e-01 -5.76054752e-01 5.91910660e-01 1.20552886e+00 1.96201041e-01 -1.44816756e-01 -1.45510757e+00 1.77114859e-01 2.42496226e-02 -2.34742790e-01 -1.40362179e+00 -2.65126169e-01 5.34003854e-01 -8.72161031e-01 8.03658247e-01 7.63144791e-01 8.18213105e-01 4.86347944e-01 4.31583524e-01 -2.23180592e-01 8.45397294e-01 -2.35070407e-01 5.22901356e-01 1.94827884e-01 -2.08971202e-01 1.15088798e-01 7.23518252e-01 3.80657315e-01 1.92050666e-01 4.46792729e-02 4.23393339e-01 -4.19280559e-01 -3.39752078e-01 -6.50623143e-01 -9.95885670e-01 5.54960668e-01 3.71428251e-01 6.38567865e-01 -3.61008316e-01 -8.84001106e-02 3.80605400e-01 3.70024264e-01 4.28543985e-02 7.54722357e-01 -1.91170007e-01 9.59997475e-02 -8.27750802e-01 5.23322642e-01 9.84935522e-01 7.54921198e-01 3.24946940e-01 6.82361603e-01 6.06116578e-02 1.66469932e-01 4.86290723e-01 3.76849145e-01 9.32552069e-02 -6.78501189e-01 4.06747967e-01 6.15618348e-01 5.59267104e-01 -8.45947206e-01 -7.65876770e-01 -1.20036674e+00 -5.95360041e-01 7.97876835e-01 1.82697147e-01 -3.42815876e-01 -3.52811068e-01 1.49732637e+00 7.16832936e-01 -5.88184297e-01 2.12466136e-01 1.21333051e+00 -2.05982655e-01 6.10911489e-01 -1.80407956e-01 -4.99358207e-01 1.27570033e+00 -4.24435139e-01 -7.73512483e-01 -2.65379041e-01 2.53916860e-01 -7.50472069e-01 4.21195388e-01 5.02853513e-01 -1.11386847e+00 -3.97716045e-01 -1.48166764e+00 5.78898907e-01 -2.60089129e-01 2.76104897e-01 -1.75619975e-01 7.09473133e-01 -6.53205872e-01 8.62457514e-01 -6.65955603e-01 -2.68978834e-01 -6.50437117e-01 4.95541602e-01 -1.29135743e-01 6.06569469e-01 -1.15180695e+00 1.39761698e+00 7.90034831e-01 3.38026583e-01 -7.15159118e-01 -6.73813820e-01 -5.63612819e-01 2.28697747e-01 5.68559766e-01 -8.58321786e-01 8.20663810e-01 -7.74122059e-01 -1.80804121e+00 1.59900263e-01 5.34389257e-01 -5.46402454e-01 1.03670228e+00 4.52392213e-02 -2.17135459e-01 -1.29579529e-01 -4.21669126e-01 -8.25456227e-04 9.90175009e-01 -1.08240831e+00 -2.66333640e-01 -1.86970457e-01 -1.05941519e-02 3.97471398e-01 -1.24455348e-01 -3.08587700e-01 1.81235954e-01 -5.67703366e-01 -7.11246207e-02 -1.35009718e+00 -3.16564649e-01 -4.10402507e-01 -3.34801018e-01 4.45270628e-01 6.55947387e-01 -7.33978808e-01 9.46403801e-01 -1.74301100e+00 1.22596431e+00 5.30947268e-01 -3.78773093e-01 -5.27663976e-02 5.50558329e-01 7.28064597e-01 -2.01799467e-01 -2.91886359e-01 -3.54569077e-01 -2.75752395e-01 6.06974587e-02 1.21493004e-01 -2.19439700e-01 6.44302487e-01 -7.14326948e-02 6.53522238e-02 -3.44567835e-01 -9.37031582e-02 4.43345338e-01 2.48638794e-01 -4.50312793e-01 4.67593633e-02 -3.61034304e-01 8.43449175e-01 -6.34731412e-01 4.05874342e-01 4.37228620e-01 4.82765734e-01 3.42017591e-01 -3.20679843e-01 -8.09065461e-01 -4.08244252e-01 -1.56464028e+00 1.47529769e+00 -5.91425717e-01 -2.67569814e-02 9.79926825e-01 -1.07320762e+00 1.18674743e+00 3.58410567e-01 7.84435153e-01 -4.28541869e-01 5.91327071e-01 6.18320405e-01 1.63015753e-01 -1.39538020e-01 6.62621975e-01 -6.21569633e-01 6.45503253e-02 -3.84513199e-01 -1.31031767e-01 -6.91434085e-01 3.18155825e-01 -3.00782293e-01 4.80781376e-01 1.66424036e-01 2.04810411e-01 -1.18849587e+00 1.16091466e+00 6.25693724e-02 7.85387099e-01 -3.86217803e-01 3.66539121e-01 1.21884793e-02 5.86500108e-01 -8.31885543e-03 -1.18704450e+00 -4.00832653e-01 -1.49718881e-01 1.18212759e-01 1.51287287e-01 1.40049323e-01 -4.61718291e-01 1.31635770e-01 4.08017695e-01 1.16284978e+00 -1.57587484e-01 -2.32173815e-01 -4.44613963e-01 -5.52584469e-01 -6.45666271e-02 -1.37996912e-01 -1.48115709e-01 -1.02805912e-01 -1.08511090e+00 4.62481320e-01 1.47860587e-01 -7.25533664e-01 -9.99195874e-03 2.02073634e-01 -1.09057856e+00 -9.73754287e-01 -5.02396822e-01 9.27894097e-03 5.20005047e-01 -4.10350621e-01 6.10988438e-01 -1.92101091e-01 -4.03036684e-01 3.80873919e-01 -2.36806311e-02 -3.40613782e-01 -4.51329440e-01 -1.86998039e-01 5.97049534e-01 5.62441256e-03 -1.05685115e+00 -5.25783598e-01 -2.94464707e-01 8.10086548e-01 -1.09654176e+00 -2.24064320e-01 2.60586888e-01 6.03590667e-01 5.58006525e-01 4.65005487e-01 3.95727843e-01 -1.02659706e-02 4.80472088e-01 -4.10553247e-01 -1.41802692e+00 6.27856106e-02 -6.74005151e-01 3.49040598e-01 7.05190301e-01 -2.35210031e-01 -1.22355866e+00 6.80902600e-02 1.02681369e-01 -5.36099792e-01 3.97420794e-01 7.27583706e-01 -3.51654977e-01 -4.70577002e-01 2.18708426e-01 -1.89217374e-01 6.76012576e-01 -5.77519417e-01 -6.10403344e-02 6.11918271e-02 1.77424878e-01 -9.84107792e-01 1.07382786e+00 2.55446136e-01 9.15648401e-01 -8.79925132e-01 3.30041386e-02 -3.34617496e-02 -3.63811553e-01 -5.71680605e-01 4.58805293e-01 -7.42298067e-01 -8.11433852e-01 -8.51116329e-02 -7.45714545e-01 -3.55070047e-02 -5.75542808e-01 6.44601345e-01 -8.48821759e-01 4.05247271e-01 1.52188435e-01 -1.26269841e+00 -4.04935181e-01 -1.59924841e+00 5.23956358e-01 1.58287093e-01 -1.29126504e-01 -9.41579700e-01 3.39451730e-02 -1.27275409e-02 7.05271661e-01 1.01871502e+00 5.90207577e-01 -6.81721345e-02 -5.31634390e-01 -1.35600865e-01 6.49427116e-01 2.75064856e-01 -1.28722787e-01 3.16736512e-02 -2.53181159e-01 -9.89190340e-01 5.70348084e-01 8.36252943e-02 1.27227724e-01 2.21967086e-01 2.86025584e-01 -2.43872747e-01 -1.70596227e-01 3.48524600e-01 1.92784274e+00 3.67205828e-01 2.38327697e-01 6.48801923e-01 -3.01047927e-03 9.77755964e-01 1.24031639e+00 8.65809500e-01 -3.67649376e-01 1.22968304e+00 1.09198511e+00 3.79948974e-01 3.02470148e-01 4.30846602e-01 5.03679812e-01 3.08044851e-01 -3.45047325e-01 -9.37901288e-02 -7.96531200e-01 3.54567140e-01 -1.67037499e+00 -5.39871514e-01 -2.69828558e-01 2.75046301e+00 1.10634774e-01 2.44235426e-01 4.48789746e-02 2.31930658e-01 7.42967486e-01 -1.32720858e-01 -3.08789402e-01 -4.36561406e-01 -1.97849423e-01 -4.92213964e-01 9.58038986e-01 6.26132131e-01 -6.64912760e-01 -1.25244558e-01 4.35600567e+00 5.77618897e-01 -1.28931797e+00 -1.75640613e-01 9.32780504e-02 -4.16019976e-01 -2.99266577e-01 4.31269825e-01 -6.57559216e-01 5.80389321e-01 1.09446323e+00 -6.10154688e-01 4.86168087e-01 8.68875802e-01 5.63886344e-01 -1.92783147e-01 -6.24475777e-01 6.12328231e-01 -3.21098268e-01 -9.38921392e-01 -4.20099497e-01 4.76363093e-01 3.95999908e-01 -2.33730063e-01 -1.27690300e-01 -7.05023436e-03 -5.66501975e-01 -4.47129935e-01 1.48836911e+00 9.21259880e-01 5.89599967e-01 -9.65092838e-01 6.69450045e-01 4.57067102e-01 -1.18875611e+00 -5.97930968e-01 -2.02515367e-02 7.42217898e-02 9.09736991e-01 8.23830605e-01 -6.99678183e-01 1.38217592e+00 2.03531384e-01 -1.66210402e-02 -8.63053128e-02 1.21468544e+00 3.08321148e-01 -8.23797360e-02 -7.92283416e-01 -1.54664174e-01 2.34860152e-01 -8.99619937e-01 1.65937603e+00 4.78142887e-01 8.94623041e-01 -2.57821113e-01 8.96811113e-02 1.07853770e+00 7.06024528e-01 1.78938404e-01 -4.18677151e-01 -2.29335114e-01 1.13327771e-01 1.41135037e+00 -5.47541976e-01 1.09781824e-01 1.68374300e-01 2.13477179e-01 -4.68074381e-01 5.08725345e-02 -9.55663145e-01 -3.35451871e-01 8.58006060e-01 5.67681551e-01 5.53642474e-02 -6.91267252e-01 1.90308560e-02 -6.84919953e-01 -6.89782649e-02 -4.98340786e-01 1.81798741e-01 -4.14531887e-01 -6.41886294e-01 4.69835669e-01 3.80715996e-01 -1.72230279e+00 -5.91562510e-01 -6.39006317e-01 -4.92549628e-01 9.85244274e-01 -1.25055289e+00 -7.32529104e-01 -5.76766394e-02 9.88228694e-02 2.78466344e-01 -2.76904345e-01 3.42195064e-01 4.15904880e-01 -5.82628727e-01 -2.55339652e-01 5.39396107e-01 -1.09807944e+00 3.47185820e-01 -9.89245713e-01 -3.60479623e-01 8.44349563e-01 -9.88232613e-01 5.76505005e-01 1.54104853e+00 -8.62222970e-01 -2.02616596e+00 -7.67013371e-01 2.53828257e-01 2.91015506e-01 8.27027678e-01 1.08422190e-01 -5.27135134e-01 3.36179644e-01 3.08934093e-01 -4.02831137e-01 -6.35161474e-02 -2.08480224e-01 6.75061882e-01 -2.21075892e-01 -1.34916496e+00 3.90270561e-01 3.39705974e-01 1.30649686e-01 -6.60583496e-01 2.30538473e-01 5.43038785e-01 -4.20921177e-01 -1.33850038e+00 7.46001899e-01 2.14852244e-01 -6.19953513e-01 9.58060384e-01 -1.75611675e-01 -3.49379390e-01 -8.10852051e-01 1.74676944e-02 -1.43345058e+00 -9.07263979e-02 -9.44877625e-01 -2.59135105e-02 1.23025024e+00 2.41696775e-01 -7.44016111e-01 1.79281697e-01 7.36084759e-01 -5.17023861e-01 -3.95270497e-01 -1.52545738e+00 -1.29504538e+00 -1.84357867e-01 1.86885566e-01 3.54974002e-01 6.62981868e-01 3.39131802e-02 7.67318755e-02 -2.61657119e-01 4.25164253e-01 6.52405858e-01 1.15378231e-01 6.83562636e-01 -1.41899931e+00 -5.94521403e-01 -3.07826817e-01 -3.00495625e-01 9.97740030e-02 -1.76445708e-01 -5.94673097e-01 -2.48956293e-01 -1.10441613e+00 -7.25283444e-01 -3.80451560e-01 4.51894142e-02 -3.25982481e-01 4.36510056e-01 -6.42878473e-01 4.50062335e-01 1.04321025e-01 2.94060349e-01 1.04331791e+00 9.54084873e-01 -2.75320411e-01 -2.46778399e-01 2.67021835e-01 2.02118501e-01 2.64622927e-01 5.11295617e-01 -1.71601072e-01 -5.34535348e-01 -1.20503321e-01 4.61216331e-01 6.09979451e-01 1.99962243e-01 -1.58416700e+00 4.80636768e-03 -2.24065527e-01 -1.77387848e-01 -4.69704151e-01 5.90902567e-01 -1.41223896e+00 1.32375348e+00 9.51323211e-01 3.53309631e-01 -3.33942063e-02 4.67745751e-01 5.17556489e-01 -5.06499887e-01 -6.36056185e-01 1.05769956e+00 1.11106664e-01 -2.17141092e-01 -2.65943438e-01 -2.75008798e-01 -7.77577043e-01 1.31952798e+00 -1.36772782e-01 9.18049887e-02 -2.01451313e-02 -1.05345452e+00 5.35423994e-01 6.13186061e-01 3.22104394e-01 -1.50800824e-01 -1.01961970e+00 -4.71171439e-01 -1.12806611e-01 -2.44943529e-01 -9.55000967e-02 6.89298332e-01 1.21754408e+00 -6.72466934e-01 8.03464472e-01 -5.81990778e-01 -4.24357504e-01 -1.09516454e+00 6.99210346e-01 6.61274970e-01 -3.46907288e-01 -1.67021006e-01 3.29501241e-01 -2.02310115e-01 -1.99807107e-01 -1.43915787e-01 -3.84607881e-01 -1.89392582e-01 5.39846003e-01 3.04712616e-02 7.28905678e-01 5.56431115e-01 -5.93889654e-01 -2.55689114e-01 5.78024685e-01 7.47274220e-01 -3.80542934e-01 1.32227361e+00 -1.98688492e-01 -1.18918791e-01 1.57904491e-01 5.54956794e-01 1.94383070e-01 -1.12156975e+00 4.10780102e-01 1.80238001e-02 -3.03280711e-01 3.00584942e-01 -5.26086450e-01 -7.83837438e-01 2.28963777e-01 5.29161513e-01 3.39131445e-01 9.70124245e-01 -7.42447793e-01 4.17952947e-02 3.36327881e-01 7.36018956e-01 -1.44882929e+00 -3.47489059e-01 4.55267817e-01 1.67154169e+00 -3.73210132e-01 4.21607047e-01 -5.47894359e-01 -5.42284548e-01 1.36033237e+00 2.73415267e-01 -3.31720449e-02 3.72773290e-01 3.25112373e-01 -7.18971908e-01 1.21150821e-01 -2.08353207e-01 -1.43293709e-01 2.47094288e-01 -3.14126849e-01 8.53289962e-02 1.18692897e-01 -1.08213317e+00 5.81254005e-01 1.17392264e-01 -3.09750319e-01 2.09156364e-01 1.20821631e+00 -6.09896183e-01 -1.27638662e+00 -1.20406222e+00 -1.04878433e-01 -1.44585641e-03 6.74981534e-01 -8.66347104e-02 1.11935258e+00 9.66590866e-02 6.24852359e-01 -2.05822259e-01 -1.71794251e-01 8.67175758e-01 1.88581735e-01 2.84210086e-01 -1.76480502e-01 -7.36932158e-01 2.69551724e-01 4.88674819e-01 -4.16356713e-01 -1.44679755e-01 -8.13661695e-01 -1.20272827e+00 2.98250094e-02 -4.18051273e-01 5.87777734e-01 1.41744947e+00 7.05852330e-01 3.64283919e-01 1.04534996e+00 5.95912099e-01 -1.03822064e+00 -1.12689793e+00 -8.93270552e-01 -8.35518479e-01 -8.97717252e-02 1.30819660e-02 -1.32288516e+00 -5.35275519e-01 -4.23080206e-01]
[5.381500244140625, 2.370124340057373]
126a262f-850c-4393-abc1-0e1a9829fbaf
task-wise-sampling-convolutions-for-arbitrary
2209.02200
null
https://arxiv.org/abs/2209.02200v1
https://arxiv.org/pdf/2209.02200v1.pdf
Task-wise Sampling Convolutions for Arbitrary-Oriented Object Detection in Aerial Images
Arbitrary-oriented object detection (AOOD) has been widely applied to locate and classify objects with diverse orientations in remote sensing images. However, the inconsistent features for the localization and classification tasks in AOOD models may lead to ambiguity and low-quality object predictions, which constrains the detection performance. In this paper, an AOOD method called task-wise sampling convolutions (TS-Conv) is proposed. TS-Conv adaptively samples task-wise features from respective sensitive regions and maps these features together in alignment to guide a dynamic label assignment for better predictions. Specifically, sampling positions of the localization convolution in TS-Conv is supervised by the oriented bounding box (OBB) prediction associated with spatial coordinates. While sampling positions and convolutional kernel of the classification convolution are designed to be adaptively adjusted according to different orientations for improving the orientation robustness of features. Furthermore, a dynamic task-aware label assignment (DTLA) strategy is developed to select optimal candidate positions and assign labels dynamicly according to ranked task-aware scores obtained from TS-Conv. Extensive experiments on several public datasets covering multiple scenes, multimodal images, and multiple categories of objects demonstrate the effectiveness, scalability and superior performance of the proposed TS-Conv.
['Ran Tao', 'Hao Wang', 'Xiang-Gen Xia', 'Wei Li', 'Zhanchao Huang']
2022-09-06
null
null
null
null
['object-detection-in-aerial-images']
['computer-vision']
[ 3.32462400e-01 -5.47006130e-01 3.14075984e-02 -8.20530653e-01 -4.55751777e-01 -5.09154558e-01 3.72753412e-01 -5.11355996e-02 -4.10963953e-01 2.44830027e-01 2.01008245e-02 1.00287274e-01 -5.77414215e-01 -6.67984068e-01 -3.98866177e-01 -1.09913039e+00 -1.10307969e-01 2.85737723e-01 5.13761640e-01 1.26266122e-01 5.62280118e-01 9.75593090e-01 -1.51712847e+00 4.49924678e-01 1.10675430e+00 1.39365768e+00 8.64081502e-01 1.62203684e-01 -6.45259991e-02 4.11915332e-01 -2.66449541e-01 2.60654300e-01 3.80148083e-01 6.62963912e-02 -4.16617125e-01 5.01544952e-01 6.30804479e-01 -3.17442328e-01 1.50108948e-01 1.04342270e+00 5.22168458e-01 2.60711581e-01 7.92427599e-01 -1.04052806e+00 -6.88843429e-01 1.82258576e-01 -6.41851187e-01 4.34026450e-01 -3.92334849e-01 1.37542859e-01 1.11034071e+00 -1.35284960e+00 2.44762033e-01 1.19846511e+00 5.21411479e-01 1.20090835e-01 -1.20762932e+00 -4.42630470e-01 5.82009077e-01 2.95798838e-01 -1.67030168e+00 -9.70887393e-02 7.60221481e-01 -7.24136174e-01 6.49597287e-01 3.41764212e-01 5.68090141e-01 4.50584620e-01 1.56348884e-01 6.30643368e-01 9.73583698e-01 -2.45985061e-01 1.34846851e-01 3.40522170e-01 1.79784104e-01 6.14665270e-01 1.97530076e-01 -5.07179610e-02 -5.59760094e-01 2.37720599e-03 5.95973492e-01 2.34601751e-01 -2.49153510e-01 -5.69946945e-01 -1.42007613e+00 6.41620755e-01 9.79808211e-01 1.83809716e-02 -7.82744527e-01 -1.57608226e-01 2.95203894e-01 -2.39746660e-01 5.60211837e-01 4.06991392e-01 -5.80153465e-01 8.61544788e-01 -4.64045137e-01 3.23778421e-01 -1.52982637e-01 8.57999027e-01 1.01445258e+00 -1.22337252e-01 -7.40753710e-01 1.24608922e+00 4.04858142e-01 6.06981695e-01 2.68601269e-01 -5.65807283e-01 5.41672051e-01 9.59193051e-01 2.98166662e-01 -1.33747447e+00 -7.15202987e-01 -6.14044487e-01 -6.24315917e-01 -5.11125810e-02 2.00094029e-01 -2.28920411e-02 -9.32365656e-01 1.31711125e+00 6.64110243e-01 -1.00091532e-01 -2.30825260e-01 1.50939643e+00 5.93684077e-01 8.55605721e-01 2.98575640e-01 2.25854412e-01 1.49741495e+00 -7.55844533e-01 -1.86872631e-01 -5.91551006e-01 5.61589658e-01 -6.21535420e-01 1.06867564e+00 1.84949994e-01 -2.47901812e-01 -8.36468756e-01 -8.40719223e-01 2.44394034e-01 -2.48384267e-01 1.03779113e+00 5.09950697e-01 2.31990233e-01 -5.53048909e-01 1.49842978e-01 -4.17684346e-01 -1.97892308e-01 5.91119230e-01 2.46631265e-01 -2.39837319e-02 -8.85578841e-02 -8.80535364e-01 5.21772921e-01 6.21162713e-01 7.33423471e-01 -1.00338781e+00 -4.48099732e-01 -5.78358412e-01 5.02513275e-02 3.19032460e-01 -1.40429273e-01 8.03242266e-01 -1.15499938e+00 -9.28534508e-01 5.04682720e-01 -1.65359333e-01 -1.16594303e-02 2.23574385e-01 -6.97480664e-02 -3.35842758e-01 8.87658298e-02 4.93270367e-01 1.01623249e+00 9.26382422e-01 -1.30661964e+00 -1.32499397e+00 -6.30637586e-01 -7.09423050e-02 6.19200528e-01 -5.56974828e-01 2.08332799e-02 -1.21717550e-01 -6.30259633e-01 8.31645548e-01 -8.40963304e-01 -3.36795688e-01 3.08976352e-01 -3.60161394e-01 -4.56445128e-01 8.83777559e-01 -4.72149611e-01 9.31917548e-01 -2.15842915e+00 4.93152775e-02 2.70282120e-01 -7.20740557e-02 6.01012707e-02 -2.63381988e-01 -9.24637392e-02 1.05441630e-01 -1.68261707e-01 -4.23332192e-02 2.07910269e-01 -2.23540723e-01 1.11101689e-02 -3.37631643e-01 4.44489002e-01 5.14443994e-01 4.20895875e-01 -7.11226463e-01 -5.94223976e-01 4.41565275e-01 4.37736720e-01 -3.30559492e-01 2.04561010e-01 -3.60943109e-01 5.61417699e-01 -1.02564704e+00 1.02568567e+00 8.83456588e-01 -2.92986959e-01 -2.86813062e-02 -6.52146935e-01 -4.77966636e-01 1.83717325e-01 -1.29545975e+00 8.92380297e-01 -3.64592761e-01 4.12720144e-01 -3.45722705e-01 -9.83914852e-01 1.40612245e+00 -1.43934518e-01 3.11094552e-01 -6.82000577e-01 -1.64984301e-01 2.52936363e-01 -7.35537037e-02 -7.70754337e-01 5.96744120e-01 2.45744154e-01 8.65896791e-02 -1.15037575e-01 -4.00581807e-01 2.87795693e-01 -1.35969400e-01 -3.03183347e-01 3.13925028e-01 2.25413993e-01 2.52623260e-01 -4.69395816e-01 7.44766176e-01 3.60687077e-01 6.81956887e-01 7.68506348e-01 -2.42253929e-01 5.42679667e-01 4.08463888e-02 -9.48470652e-01 -8.14759254e-01 -6.24333739e-01 -6.77084327e-01 1.46821976e+00 5.52586377e-01 3.29102427e-01 -3.98163587e-01 -8.64215255e-01 9.89475753e-03 5.85271657e-01 -6.97671592e-01 7.92162679e-03 -3.99792016e-01 -1.05136514e+00 3.00725102e-02 7.04818368e-01 6.69314027e-01 -1.25143421e+00 -6.60978854e-01 2.71748096e-01 -3.33210677e-02 -7.47157037e-01 -5.11791825e-01 3.22360367e-01 -7.73492754e-01 -8.36481392e-01 -5.42686582e-01 -9.66343105e-01 1.06720841e+00 8.29909265e-01 4.13891196e-01 -1.41566619e-01 -1.07297711e-01 -2.58161664e-01 -4.70160037e-01 -3.14106137e-01 3.53131145e-01 1.99853748e-01 -1.38895944e-01 6.39999390e-01 4.51202840e-01 -3.86280380e-02 -8.49201858e-01 7.55073130e-01 -7.55151093e-01 1.04490547e-02 7.00736701e-01 1.04427207e+00 8.10874581e-01 -6.66492507e-02 5.68881512e-01 -5.94923973e-01 1.06501408e-01 -4.59809005e-01 -9.62031722e-01 5.09216428e-01 -5.45425594e-01 -1.16891362e-01 5.98812044e-01 -5.35453320e-01 -1.03853571e+00 3.71383339e-01 4.15174782e-01 -2.77055085e-01 -2.71599084e-01 5.41974366e-01 -2.55839288e-01 -1.26295283e-01 7.66222954e-01 4.08735603e-01 -4.60955828e-01 -4.32152063e-01 -9.91492271e-02 9.69217181e-01 3.15157697e-02 -3.97077113e-01 5.85611165e-01 4.47428942e-01 -2.32799619e-01 -5.89197338e-01 -1.23329973e+00 -4.75949615e-01 -7.63801754e-01 -3.36277097e-01 7.82409370e-01 -1.12053311e+00 -4.19405103e-01 4.75019276e-01 -1.01876771e+00 -7.53516331e-02 1.14826173e-01 5.92860818e-01 3.79783586e-02 -1.64624408e-01 -3.69380265e-02 -9.36221719e-01 -3.73842716e-01 -1.35423696e+00 1.44533706e+00 6.15071595e-01 3.09478700e-01 -3.85344028e-01 -4.97188389e-01 3.33094209e-01 3.43838602e-01 -2.28696868e-01 6.88125849e-01 -5.03879547e-01 -9.16409314e-01 -1.52304322e-01 -6.82365537e-01 2.43004829e-01 2.25709647e-01 1.18701328e-02 -9.52214181e-01 -1.90824777e-01 -4.85846698e-01 -2.60064989e-01 8.50244224e-01 6.32591367e-01 1.63630533e+00 -1.87259033e-01 -5.12066185e-01 7.00354934e-01 1.37392652e+00 3.53721827e-01 1.29513770e-01 5.66182554e-01 8.64246368e-01 8.33083808e-01 1.34237981e+00 6.11233950e-01 2.84079164e-01 8.62703681e-01 5.96450567e-01 5.14347702e-02 2.79252619e-01 -9.52494889e-03 7.51900151e-02 -7.01856166e-02 -4.51019183e-02 -1.85612932e-01 -6.63922429e-01 6.12786829e-01 -1.68260574e+00 -7.46946394e-01 -1.81824282e-01 1.95401013e+00 4.24889088e-01 -1.00289047e-01 -3.38844657e-02 -2.48701185e-01 1.06424916e+00 3.08969021e-01 -7.50932932e-01 2.24703282e-01 -1.15499431e-02 -5.46969295e-01 6.59164667e-01 2.31595784e-01 -1.46993649e+00 7.93231845e-01 4.97376537e+00 1.02405703e+00 -1.55005205e+00 -6.46051466e-02 8.19266498e-01 1.56090438e-01 -1.00626782e-01 -3.80336940e-02 -1.38498032e+00 4.33070213e-01 -9.33453143e-02 4.75920677e-01 3.67804855e-01 1.21891773e+00 4.88159120e-01 -9.15130153e-02 -4.00443316e-01 5.81003308e-01 -1.21247828e-01 -1.12786174e+00 2.39440277e-01 -2.57027186e-02 7.47440696e-01 1.45971820e-01 1.25667483e-01 2.20548678e-02 -1.16626255e-01 -5.96195400e-01 1.13966346e+00 4.12585825e-01 6.17517412e-01 -5.55839360e-01 5.72554708e-01 3.53848934e-01 -1.45693278e+00 -6.61372781e-01 -9.09744442e-01 2.30659410e-01 -9.04618502e-02 5.42871475e-01 -1.15776157e+00 3.44162494e-01 1.10089040e+00 7.98148930e-01 -7.89252162e-01 1.05656886e+00 -1.29888445e-01 5.24579167e-01 -2.54620045e-01 -2.72099674e-01 4.97030258e-01 -2.41191596e-01 5.07588029e-01 9.64381933e-01 4.49796677e-01 -4.18421179e-02 5.02785265e-01 9.01185274e-01 3.70683283e-01 1.41593382e-01 -2.24118531e-01 2.64944971e-01 8.63902032e-01 1.62185812e+00 -8.27654064e-01 -8.49976242e-02 -1.19503222e-01 4.33089256e-01 4.35308248e-01 3.74961168e-01 -8.94928873e-01 -2.47872531e-01 4.27555233e-01 2.24208117e-01 6.35734379e-01 -1.69620827e-01 -3.24728251e-01 -7.56446123e-01 4.03779596e-02 -5.09769797e-01 4.95622158e-01 -8.82630885e-01 -1.38786888e+00 6.70051873e-01 3.08236722e-02 -1.62852097e+00 5.85534096e-01 -6.81314468e-01 -4.43263501e-01 9.92625952e-01 -1.62486517e+00 -1.25660789e+00 -8.27817440e-01 1.94249198e-01 7.01778173e-01 -7.44697452e-02 3.18931222e-01 1.83713943e-01 -8.08339179e-01 1.67930916e-01 1.26072213e-01 9.12057087e-02 5.68212152e-01 -9.50285196e-01 -1.47860512e-01 7.44697034e-01 -9.29579735e-02 3.91238332e-01 2.78510958e-01 -5.92414320e-01 -9.89962816e-01 -1.58524644e+00 6.86577082e-01 3.52362245e-02 2.76342452e-01 -2.00273558e-01 -9.16724503e-01 5.07487237e-01 -5.65465569e-01 4.60303336e-01 3.24755996e-01 -2.24135190e-01 -1.02711476e-01 -8.35683644e-01 -1.00933039e+00 4.61804003e-01 9.19582844e-01 -1.78108528e-01 -1.89046845e-01 6.26007199e-01 5.12480557e-01 -2.41255477e-01 -5.14101744e-01 7.39627063e-01 4.68900889e-01 -6.76058054e-01 9.89585996e-01 -3.05359066e-01 1.29496962e-01 -9.91091371e-01 -4.20350850e-01 -9.64280605e-01 -6.98714972e-01 3.37668329e-01 6.72088504e-01 1.11369693e+00 2.88206309e-01 -6.68845475e-01 4.18509841e-01 2.05774531e-01 -3.62917393e-01 -7.99018383e-01 -8.42118502e-01 -4.07742232e-01 -7.42165923e-01 -9.75119770e-02 7.20100939e-01 8.40855002e-01 -7.73885965e-01 1.76847935e-01 -2.45487034e-01 1.02152002e+00 4.64436293e-01 5.70604980e-01 5.32110691e-01 -1.19291329e+00 6.22726567e-02 -3.26449782e-01 -2.73467004e-01 -1.32698190e+00 -1.96309969e-01 -8.23942125e-01 4.44678724e-01 -1.34425640e+00 1.25953093e-01 -1.10395265e+00 -4.38810229e-01 7.90502727e-01 -3.26355755e-01 3.65711778e-01 -8.78276750e-02 5.66052020e-01 -6.17674708e-01 6.24154568e-01 1.21034849e+00 -2.82162398e-01 -2.85192549e-01 1.82214379e-01 -4.63287562e-01 4.05321091e-01 7.17843115e-01 -5.44945359e-01 -2.59233683e-01 -5.25300980e-01 -4.52930294e-02 -2.65524656e-01 6.00976646e-01 -8.56610715e-01 8.33047032e-02 -5.96312106e-01 7.83346593e-01 -9.75504935e-01 4.75186892e-02 -9.07569051e-01 3.98463458e-02 4.73831236e-01 -4.67609793e-01 -3.74462992e-01 -7.50282630e-02 6.52949512e-01 -4.66770455e-02 -3.94297868e-01 8.77179027e-01 7.74434432e-02 -1.17793000e+00 4.07591134e-01 -1.95438847e-01 -5.51418543e-01 1.02874124e+00 -4.24334675e-01 -3.08155954e-01 2.70059347e-01 -5.35744667e-01 3.91951829e-01 9.78597477e-02 4.13825333e-01 6.54939950e-01 -1.17040396e+00 -7.03845143e-01 3.86186779e-01 5.57976902e-01 4.52748984e-01 5.86332619e-01 9.06282187e-01 -5.78360200e-01 3.53506148e-01 -8.17623660e-02 -1.12653244e+00 -9.97746527e-01 2.50903875e-01 7.45843530e-01 1.50689483e-01 -3.68959486e-01 8.83859515e-01 5.32112956e-01 -6.05254769e-01 8.79378617e-02 -1.99953109e-01 -6.00035787e-01 1.74423799e-01 4.60779607e-01 2.32823819e-01 3.49205062e-02 -6.28627300e-01 -6.15529120e-01 7.14641690e-01 -1.58993915e-01 4.09508675e-01 1.39703143e+00 -3.34877431e-01 -4.79018092e-02 2.01767847e-01 5.78743517e-01 -2.73285151e-01 -1.78347468e+00 -4.44021910e-01 2.16433723e-02 -7.45704293e-01 2.67267585e-01 -8.43681037e-01 -1.18887067e+00 7.47030318e-01 9.29049015e-01 5.44448756e-02 1.17182624e+00 1.47106750e-02 2.18721241e-01 4.58100855e-01 2.62168795e-01 -9.90028143e-01 1.93581253e-01 3.72117281e-01 8.76342952e-01 -1.19904268e+00 -9.02155042e-02 -4.60024804e-01 -8.86232674e-01 1.12169015e+00 1.08446348e+00 2.50797737e-02 4.51703221e-01 -1.74151108e-01 2.11811841e-01 -2.19437212e-01 -2.90162444e-01 -1.21403240e-01 7.21799910e-01 5.15863121e-01 1.53730735e-01 1.31153807e-01 -2.51026422e-01 4.79326487e-01 4.79948193e-01 -4.12400693e-01 -1.94401935e-01 7.75347471e-01 -9.76760924e-01 -5.42334735e-01 -7.69038439e-01 5.23877144e-01 1.29765734e-01 2.51605129e-03 -8.41636509e-02 3.09585690e-01 5.16368032e-01 7.71280646e-01 3.07977319e-01 -4.68187511e-01 2.13870868e-01 -2.37063974e-01 -3.82117629e-02 -5.26085019e-01 -4.33795810e-01 2.55772173e-01 -1.99906558e-01 -1.76766545e-01 -5.02425194e-01 -7.49064744e-01 -1.27738976e+00 3.64731461e-01 -7.97809362e-01 -5.66859543e-02 4.99855876e-01 8.99654150e-01 4.78970438e-01 4.09667999e-01 1.15943742e+00 -1.12525725e+00 -6.65874600e-01 -1.14076293e+00 -7.83304155e-01 -3.38626583e-03 1.28790915e-01 -9.63037193e-01 -2.14648247e-01 -1.29885584e-01]
[8.84967041015625, -0.7477682828903198]
8332d878-1a75-4028-9fc4-afefc5ee904c
class-adaptive-self-training-for-relation
2306.09697
null
https://arxiv.org/abs/2306.09697v1
https://arxiv.org/pdf/2306.09697v1.pdf
Class-Adaptive Self-Training for Relation Extraction with Incompletely Annotated Training Data
Relation extraction (RE) aims to extract relations from sentences and documents. Existing relation extraction models typically rely on supervised machine learning. However, recent studies showed that many RE datasets are incompletely annotated. This is known as the false negative problem in which valid relations are falsely annotated as 'no_relation'. Models trained with such data inevitably make similar mistakes during the inference stage. Self-training has been proven effective in alleviating the false negative problem. However, traditional self-training is vulnerable to confirmation bias and exhibits poor performance in minority classes. To overcome this limitation, we proposed a novel class-adaptive re-sampling self-training framework. Specifically, we re-sampled the pseudo-labels for each class by precision and recall scores. Our re-sampling strategy favored the pseudo-labels of classes with high precision and low recall, which improved the overall recall without significantly compromising precision. We conducted experiments on document-level and biomedical relation extraction datasets, and the results showed that our proposed self-training framework consistently outperforms existing competitive methods on the Re-DocRED and ChemDisgene datasets when the training data are incompletely annotated. Our code is released at https://github.com/DAMO-NLP-SG/CAST.
['Hwee Tou Ng', 'Lidong Bing', 'Lu Xu', 'Qingyu Tan']
2023-06-16
null
null
null
null
['relation-extraction']
['natural-language-processing']
[ 2.22204402e-01 4.43009019e-01 -5.98345876e-01 -6.61897838e-01 -9.43167984e-01 -4.41546082e-01 4.56454068e-01 5.42296171e-01 -3.67729813e-01 1.40499830e+00 -1.33811040e-02 -2.74137914e-01 -7.14040920e-02 -9.97867763e-01 -6.01833701e-01 -6.50571823e-01 2.81474233e-01 6.58360779e-01 8.95717815e-02 6.18813597e-02 -5.46303950e-02 8.53304416e-02 -1.25116396e+00 6.09646320e-01 1.20883107e+00 5.73718429e-01 -8.62684324e-02 3.24491471e-01 -2.18949929e-01 7.91324794e-01 -7.61356950e-01 -7.56851196e-01 -8.65507945e-02 -6.16078973e-01 -1.22402227e+00 -2.06320390e-01 -6.48645684e-02 -4.58650254e-02 -7.67711997e-02 1.22223544e+00 3.93948764e-01 -2.91285664e-01 6.34695113e-01 -1.25673139e+00 -6.06612861e-01 1.02611208e+00 -5.85764468e-01 1.72667742e-01 3.48869771e-01 -3.07717323e-01 1.19096041e+00 -9.52084661e-01 7.19781816e-01 9.39243913e-01 6.91205263e-01 6.79112971e-01 -1.20188022e+00 -1.15559864e+00 -1.26753464e-01 1.51042253e-01 -1.70515239e+00 -5.25273323e-01 5.21105289e-01 -2.26139158e-01 8.90801251e-01 5.16716719e-01 3.83881062e-01 9.70451772e-01 1.17722871e-02 7.78490841e-01 1.16085243e+00 -3.92626554e-01 5.77961840e-02 3.26321781e-01 3.13573688e-01 5.61932445e-01 7.03588128e-01 -1.76597834e-01 -4.69437361e-01 -4.35841054e-01 2.40636855e-01 -1.79772988e-01 -2.80732214e-01 4.04984020e-02 -1.13067722e+00 6.21201217e-01 2.15787187e-01 4.39257741e-01 -1.05010867e-01 -5.82493484e-01 5.16894698e-01 1.69353992e-01 7.62644231e-01 5.49116910e-01 -9.86775815e-01 1.86488852e-01 -8.15236628e-01 -9.91034321e-03 7.88939118e-01 1.10320830e+00 6.25302434e-01 -5.72333276e-01 -3.50649714e-01 1.15240574e+00 -5.57551384e-02 1.92695767e-01 6.36214137e-01 -5.16329587e-01 5.27658522e-01 9.19463158e-01 -6.77219173e-03 -9.80033576e-01 -5.84387481e-01 -7.61146367e-01 -1.23086929e+00 -5.54280102e-01 5.14110386e-01 -2.62560993e-01 -8.58698905e-01 1.69573462e+00 5.97565413e-01 2.10737325e-02 3.38357329e-01 6.72756612e-01 1.27984631e+00 2.82395065e-01 4.03914839e-01 -6.00880086e-01 1.43510139e+00 -6.08406007e-01 -1.14611220e+00 -1.21449009e-01 1.07929885e+00 -7.06845224e-01 6.82413876e-01 3.39517057e-01 -6.05148673e-01 -2.64094591e-01 -1.03721511e+00 -1.60503797e-02 -2.86574036e-01 5.72339356e-01 7.94238329e-01 5.02625942e-01 -3.33715409e-01 6.15762413e-01 -7.00479686e-01 -1.85844183e-01 8.23134840e-01 3.85380507e-01 -7.59961903e-01 -1.52327009e-02 -1.37962520e+00 7.94253111e-01 8.76902282e-01 2.48347640e-01 -3.66729498e-01 -6.81460261e-01 -6.01061046e-01 -8.04763511e-02 6.65109754e-01 -5.43137074e-01 1.15617275e+00 -5.45049846e-01 -9.61161077e-01 1.15759814e+00 -3.81317616e-01 -3.97986650e-01 3.84747297e-01 -3.39916259e-01 -5.86140275e-01 -6.22382015e-02 2.92472214e-01 2.60453045e-01 2.90642858e-01 -1.07619441e+00 -6.75508499e-01 -4.29646701e-01 -1.33906201e-01 -1.86450984e-02 -1.89853281e-01 -9.99259353e-02 -6.48851022e-02 -5.02747178e-01 5.05659878e-01 -7.27313519e-01 -1.29909784e-01 -4.80923444e-01 -8.60064387e-01 -5.80748498e-01 4.60705519e-01 -2.95614541e-01 1.30466545e+00 -1.90714288e+00 -3.93602401e-01 4.59794104e-02 6.05479419e-01 6.72799945e-01 3.12175006e-01 3.50799441e-01 -3.58978868e-01 3.36556494e-01 -1.69301525e-01 9.69773382e-02 -5.61992943e-01 2.41898447e-01 -1.50054753e-01 2.92472631e-01 3.87836009e-01 7.10542679e-01 -1.31258559e+00 -9.88489509e-01 -1.84111401e-01 2.40896210e-01 -2.74600565e-01 2.08519995e-01 -9.25829262e-02 2.97611922e-01 -4.88195449e-01 7.01441705e-01 7.89126337e-01 -5.94303191e-01 6.52619123e-01 -5.36043763e-01 3.30072969e-01 6.66137457e-01 -1.05628955e+00 1.22968447e+00 -3.37781757e-02 2.65416622e-01 -4.48385954e-01 -1.13616824e+00 1.08939528e+00 4.21261728e-01 3.90262127e-01 -3.16002727e-01 1.98556185e-01 4.35867012e-01 1.14586040e-01 -7.05259204e-01 2.76724279e-01 -3.14651310e-01 2.30198681e-01 1.15579516e-01 8.71313810e-02 2.81710535e-01 2.71679193e-01 3.18971664e-01 1.07143009e+00 1.31527334e-01 8.46433222e-01 -1.13778755e-01 5.41764557e-01 3.19096595e-01 1.12279427e+00 5.62868118e-01 -5.77708445e-02 3.82123917e-01 7.82123923e-01 -7.66615346e-02 -6.42518282e-01 -6.78266048e-01 -5.17728686e-01 6.17894530e-01 -1.07389584e-01 -8.94277573e-01 -5.06101191e-01 -1.15173578e+00 -1.48981616e-01 5.86660802e-01 -5.61568797e-01 -1.83261335e-01 -2.33060211e-01 -1.38628328e+00 7.81578362e-01 3.13519210e-01 5.84036767e-01 -7.04643667e-01 4.92392853e-02 2.76657522e-01 -5.11589885e-01 -1.39795995e+00 -1.05409734e-02 4.57044721e-01 -8.64595830e-01 -1.66511786e+00 -3.93675387e-01 -7.42915154e-01 1.02137911e+00 4.35363837e-02 1.02491176e+00 1.71315834e-01 -8.59762803e-02 -5.77557981e-01 -5.25717795e-01 -4.95079666e-01 -5.51363409e-01 2.50566363e-01 6.03145063e-02 -1.97421148e-01 9.53465641e-01 -3.87897640e-01 -5.85380495e-01 2.81783938e-01 -6.36960089e-01 2.34835580e-01 6.47196770e-01 1.18099666e+00 8.16467583e-01 1.91270947e-01 1.10004187e+00 -1.75560939e+00 5.61602592e-01 -5.35729706e-01 -1.53134629e-01 4.44230497e-01 -9.37720597e-01 3.53904776e-02 6.13836527e-01 -3.97449136e-01 -9.58439410e-01 2.20001638e-01 -3.32517028e-01 2.20760942e-01 -1.29935637e-01 5.99174321e-01 -3.81909460e-01 4.56793398e-01 7.23403871e-01 -2.36869324e-04 -1.40135661e-01 -3.30151290e-01 5.82131892e-02 1.13191187e+00 3.88921618e-01 -5.12452662e-01 4.46217865e-01 1.74556613e-01 1.22279096e-02 -4.35346723e-01 -1.51535308e+00 -4.61663246e-01 -5.85257232e-01 2.89976776e-01 4.38767314e-01 -9.72510338e-01 -6.51399672e-01 2.33146906e-01 -8.82501245e-01 8.05779248e-02 -1.51963562e-01 6.13236308e-01 1.25272989e-01 3.44515771e-01 -5.35803378e-01 -8.25302720e-01 -6.43461108e-01 -6.98162198e-01 8.46196711e-01 2.63119727e-01 -4.63073969e-01 -6.22464359e-01 -2.79239786e-04 3.26025218e-01 -1.91145867e-01 1.86979339e-01 7.78879821e-01 -1.21515763e+00 1.07097570e-02 -4.22490686e-01 -3.79804403e-01 2.09679067e-01 6.19897187e-01 -4.18306626e-02 -1.10627913e+00 9.32119489e-02 -3.22545022e-01 -4.18266952e-01 7.51413047e-01 -9.17278975e-02 1.18466008e+00 -4.46436465e-01 -8.00916672e-01 2.83011079e-01 1.14487314e+00 7.48005360e-02 3.83304864e-01 1.44019604e-01 6.18464530e-01 5.13975441e-01 1.06181943e+00 2.47168377e-01 4.04237151e-01 4.56704497e-01 -8.28158855e-02 -1.54489756e-01 3.98356169e-02 -3.87393087e-01 -2.59292990e-01 5.79943061e-01 3.54707129e-02 -2.41016284e-01 -9.05212283e-01 6.54402018e-01 -1.79204011e+00 -7.58885443e-01 -4.64646101e-01 1.98303246e+00 1.77879298e+00 3.74320239e-01 3.67331356e-02 6.39426947e-01 7.02489436e-01 -3.96632403e-01 -2.79890746e-01 2.96667051e-02 -2.89206624e-01 2.51772314e-01 2.59104997e-01 3.33616048e-01 -1.14492214e+00 9.63893473e-01 5.44181681e+00 9.67250705e-01 -7.07068145e-01 2.42182329e-01 7.01638103e-01 9.37256739e-02 -2.55657047e-01 9.82257649e-02 -1.22726667e+00 3.81297857e-01 8.48622739e-01 -2.80853271e-01 -4.09677595e-01 6.79472506e-01 -6.74999133e-02 -1.10007174e-01 -1.12739813e+00 7.68460631e-01 -1.84240296e-01 -1.32650900e+00 -1.20833255e-01 -1.06444255e-01 5.09494841e-01 -1.93303734e-01 -5.61892509e-01 3.72352540e-01 3.44901115e-01 -8.89584184e-01 1.23217531e-01 3.28374714e-01 1.02736354e+00 -5.93837678e-01 1.16672409e+00 4.75465655e-01 -7.47057974e-01 2.44858995e-01 -3.89640152e-01 6.66057542e-02 -1.92250073e-01 1.37424600e+00 -1.15236092e+00 7.53201842e-01 5.89978039e-01 6.59218192e-01 -5.56637168e-01 8.05341005e-01 -4.55445766e-01 7.11526275e-01 -2.77552992e-01 -1.27191305e-01 -3.92161936e-01 1.98563561e-01 1.73261568e-01 1.26017165e+00 -1.12739339e-01 4.94101197e-01 -1.97892990e-02 6.31026208e-01 -1.90534756e-01 2.49628425e-01 -3.62433314e-01 -9.84953567e-02 6.23974681e-01 1.28462017e+00 -8.34733248e-01 -5.16194522e-01 -1.84663191e-01 5.51501393e-01 6.29047990e-01 1.03824988e-01 -6.07859790e-01 -5.18778324e-01 2.30295315e-01 2.69641411e-02 -1.01718850e-01 2.40900472e-01 -3.49832118e-01 -1.24218833e+00 -7.82031640e-02 -7.72114635e-01 6.55542672e-01 -3.90190125e-01 -1.37163639e+00 7.66233146e-01 -2.17719227e-02 -1.24841082e+00 -7.00720698e-02 -3.21259707e-01 9.59880129e-02 5.92721581e-01 -1.20816493e+00 -9.87626076e-01 -2.00021401e-01 4.40882780e-02 2.48859033e-01 -2.42441334e-02 1.13486075e+00 5.89133441e-01 -9.37094033e-01 1.05360973e+00 -1.46875724e-01 4.09792453e-01 9.97826993e-01 -1.19193709e+00 -2.37451810e-02 5.09212732e-01 -2.97962241e-02 9.48450923e-01 7.32749522e-01 -9.49153841e-01 -8.18986535e-01 -1.12245047e+00 1.44029438e+00 -3.82050037e-01 4.30292010e-01 -3.32933545e-01 -1.06949830e+00 6.40875995e-01 -3.20435971e-01 1.85842142e-01 1.18570948e+00 4.22342241e-01 -3.43125850e-01 -2.04214424e-01 -1.28153825e+00 2.93021888e-01 1.11645460e+00 -3.94994706e-01 -6.14647031e-01 7.18840122e-01 4.85012323e-01 -6.23233855e-01 -1.01979077e+00 7.17707396e-01 5.23660421e-01 -5.88116288e-01 7.39377499e-01 -8.30265403e-01 5.05990326e-01 -3.65907758e-01 1.17413789e-01 -1.04612780e+00 -2.54218698e-01 -1.99268222e-01 -3.81687343e-01 1.69199133e+00 8.66005361e-01 -6.30877852e-01 7.85163164e-01 5.56500137e-01 1.15626752e-01 -1.03660405e+00 -6.70577288e-01 -6.11438155e-01 -9.50604230e-02 -1.82014346e-01 7.44875073e-01 1.26641512e+00 3.56905550e-01 7.02140331e-01 -2.61279613e-01 1.56114668e-01 5.07853091e-01 1.48931742e-01 6.13785267e-01 -1.26869988e+00 -1.44092888e-01 4.56992388e-02 -1.76618382e-01 -6.74269497e-01 1.52366474e-01 -1.11069989e+00 5.65511733e-02 -1.44260263e+00 5.83343863e-01 -6.73477471e-01 -2.52872080e-01 8.46788526e-01 -7.16603279e-01 3.54220569e-01 -6.10832155e-01 3.61293167e-01 -4.91930336e-01 3.38987052e-01 1.27162433e+00 1.53929647e-02 -2.44471610e-01 1.82663202e-01 -1.01989269e+00 5.96417367e-01 1.14056337e+00 -9.01162446e-01 -3.93921912e-01 1.37707405e-02 6.09561205e-01 -1.44538864e-01 9.77761820e-02 -5.34176469e-01 1.11466929e-01 -1.67813405e-01 5.04294872e-01 -7.35173285e-01 -7.00108856e-02 -5.84426880e-01 1.12206355e-01 4.33443576e-01 -5.22986174e-01 -4.67303753e-01 5.35514504e-02 4.58186597e-01 -1.19571656e-01 -2.56409168e-01 5.51219285e-01 -5.86192161e-02 -1.81822419e-01 2.77494043e-02 2.17196923e-02 2.02153802e-01 8.97467613e-01 8.75097960e-02 -6.53774977e-01 1.02121972e-01 -7.50688195e-01 2.86362588e-01 -3.61195914e-02 1.70910254e-01 6.07006431e-01 -1.11935759e+00 -9.22609448e-01 1.45186141e-01 3.77931535e-01 3.94439697e-01 1.32204471e-02 9.29594755e-01 -2.40187317e-01 4.13911343e-01 2.75013238e-01 -5.12058616e-01 -1.52123129e+00 4.58815128e-01 2.70632565e-01 -3.65885437e-01 -5.68077147e-01 1.01827645e+00 -6.67408332e-02 -5.21429300e-01 2.88195293e-02 -1.79737940e-01 -3.77527416e-01 5.22995666e-02 5.37857056e-01 1.75592452e-01 3.20413619e-01 -3.73198837e-01 -7.14355230e-01 -4.67179623e-03 -5.85716307e-01 3.38847369e-01 1.17331970e+00 7.64669552e-02 -3.36521953e-01 3.72963965e-01 8.67257714e-01 1.23054631e-01 -4.42189664e-01 -4.55197871e-01 2.45063350e-01 -3.94757509e-01 -1.23766348e-01 -9.63608742e-01 -8.50677848e-01 3.32006186e-01 1.62624270e-01 1.34872362e-01 9.53830481e-01 1.50819868e-01 6.00332201e-01 3.34964633e-01 2.69777000e-01 -9.17447984e-01 -4.62793618e-01 2.96827286e-01 5.22972882e-01 -1.46931636e+00 5.26755691e-01 -1.07909238e+00 -4.81826752e-01 7.62013793e-01 8.60144079e-01 2.14323655e-01 6.94188178e-01 3.43649268e-01 -5.16444817e-02 -2.91281849e-01 -8.19242656e-01 -6.26454055e-02 4.10328805e-01 4.83734071e-01 1.04346251e+00 3.17814738e-01 -9.74423647e-01 1.08104241e+00 -3.88514966e-01 2.57973790e-01 3.13327730e-01 9.10189509e-01 -4.70204428e-02 -1.32693541e+00 -4.22237143e-02 9.76573169e-01 -8.57450008e-01 -1.84963569e-01 -5.80853343e-01 6.91136420e-01 3.51861179e-01 1.10991657e+00 -4.07166898e-01 -6.52041614e-01 1.70669049e-01 1.31785840e-01 4.18044478e-01 -8.49310577e-01 -4.84409600e-01 -4.39574122e-02 5.42044222e-01 -1.95073381e-01 -6.66879535e-01 -5.76023281e-01 -1.46915126e+00 -1.00643747e-01 -8.84923577e-01 6.96955860e-01 8.79528150e-02 1.08143950e+00 3.39551836e-01 6.63695097e-01 4.23207343e-01 2.68920243e-01 -4.21108842e-01 -1.25481808e+00 -4.36406642e-01 2.49433443e-01 2.55820811e-01 -6.81510806e-01 -2.54695237e-01 2.69436650e-03]
[9.093921661376953, 8.614165306091309]
e45a9bcf-2468-4d8d-b4c8-32c305fd4d00
a-mutual-learning-method-for-salient-object
null
null
http://openaccess.thecvf.com/content_CVPR_2019/html/Wu_A_Mutual_Learning_Method_for_Salient_Object_Detection_With_Intertwined_CVPR_2019_paper.html
http://openaccess.thecvf.com/content_CVPR_2019/papers/Wu_A_Mutual_Learning_Method_for_Salient_Object_Detection_With_Intertwined_CVPR_2019_paper.pdf
A Mutual Learning Method for Salient Object Detection With Intertwined Multi-Supervision
Though deep learning techniques have made great progress in salient object detection recently, the predicted saliency maps still suffer from incomplete predictions due to the internal complexity of objects and inaccurate boundaries caused by strides in convolution and pooling operations. To alleviate these issues, we propose to train saliency detection networks by exploiting the supervision from not only salient object detection, but also foreground contour detection and edge detection. First, we leverage salient object detection and foreground contour detection tasks in an intertwined manner to generate saliency maps with uniform highlight. Second, the foreground contour and edge detection tasks guide each other simultaneously, thereby leading to preciser foreground contour prediction and reducing the local noises for edge prediction. In addition, we develop a novel mutual learning module (MLM) which serves as the building block of our method. Each MLM consists of multiple network branches trained in a mutual learning manner, which improves the performance by a large margin. Extensive experiments on seven challenging datasets demonstrate that the proposed method has delivered state-of-the-art results in both salient object detection and edge detection.
[' Errui Ding', ' Huchuan Lu', ' Dong Wang', ' Wenlong Guan', ' Mengyang Feng', 'Runmin Wu']
2019-06-01
null
null
null
cvpr-2019-6
['contour-detection']
['computer-vision']
[ 4.63514507e-01 3.26225050e-02 -2.80421585e-01 -2.73099691e-01 -4.98061478e-01 -1.03290610e-01 3.94820958e-01 3.91494520e-02 -1.52480274e-01 4.77713376e-01 1.51160985e-01 -4.33635190e-02 3.32292765e-01 -6.36876941e-01 -8.01642060e-01 -7.28626192e-01 1.99562788e-01 -2.37212062e-01 1.18584943e+00 1.68564186e-01 3.58615458e-01 1.85570523e-01 -1.52538645e+00 3.56321871e-01 1.33668935e+00 1.13157034e+00 7.42943108e-01 4.45204526e-01 -1.51830986e-01 1.47512448e+00 -1.71877444e-01 -1.91998765e-01 2.78810486e-02 -3.32275182e-01 -5.40402591e-01 2.46840030e-01 1.62260115e-01 -4.14505154e-01 -2.81119734e-01 1.20401037e+00 3.34582984e-01 -1.40222735e-04 3.83693874e-01 -1.16389489e+00 -6.52653754e-01 6.50553584e-01 -1.21395373e+00 4.04214174e-01 -7.34461993e-02 -1.77811168e-03 1.09438932e+00 -1.19224465e+00 3.11400712e-01 9.79383588e-01 5.15300512e-01 2.99814224e-01 -1.10174584e+00 -7.02980876e-01 5.22011638e-01 3.03348929e-01 -1.23099613e+00 -2.81717539e-01 1.19564593e+00 -1.76412404e-01 3.67006868e-01 1.37293279e-01 5.94727755e-01 4.74988401e-01 3.16796787e-02 1.67522025e+00 6.88458502e-01 -1.58350751e-01 -3.17473151e-02 2.33330563e-01 -7.05347881e-02 7.78391182e-01 2.01722994e-01 -2.42652129e-02 -4.62530106e-01 2.28099033e-01 1.08635819e+00 1.95966825e-01 -4.00851816e-01 -6.52761519e-01 -1.24640107e+00 6.84494317e-01 9.43849623e-01 1.12063766e-01 -3.80440056e-01 9.85996798e-02 1.37014270e-01 -2.98693538e-01 5.05416512e-01 1.36255575e-02 -2.62564510e-01 4.09767956e-01 -1.15273952e+00 -2.49542040e-03 4.20641124e-01 9.71137166e-01 8.82287920e-01 9.29207206e-02 -2.93634593e-01 6.28744721e-01 3.83776218e-01 2.85186529e-01 3.57040048e-01 -5.24103999e-01 2.58520633e-01 8.11542571e-01 2.78524637e-01 -1.36663735e+00 -4.13727015e-01 -6.86909080e-01 -8.24170828e-01 2.53052652e-01 3.10377747e-01 -1.85291290e-01 -9.65889573e-01 1.54624271e+00 4.76423442e-01 7.15770781e-01 -1.85221180e-01 1.25617588e+00 8.56551409e-01 7.54170418e-01 3.78881425e-01 1.10037369e-03 1.07832491e+00 -1.39054525e+00 -6.52074873e-01 -6.36714697e-01 3.51816893e-01 -1.05105102e+00 7.12619424e-01 -1.50669172e-01 -1.27576864e+00 -6.56195939e-01 -1.08734763e+00 -2.55992949e-01 -2.74072643e-02 4.28892106e-01 7.81421900e-01 1.38791591e-01 -8.41690898e-01 3.43026012e-01 -9.82498467e-01 1.46089017e-01 8.52824092e-01 1.63420126e-01 1.49235398e-01 2.26752549e-01 -9.86703396e-01 8.26635420e-01 7.23951519e-01 2.90666908e-01 -6.64790630e-01 -6.28137767e-01 -1.01895356e+00 2.82704771e-01 5.83012462e-01 -5.36495268e-01 1.11279929e+00 -1.20239651e+00 -1.15805197e+00 6.58026636e-01 -3.24892372e-01 -5.50516903e-01 7.68322229e-01 -3.78538162e-01 -3.20090383e-01 9.67645124e-02 2.15850160e-01 1.06503463e+00 9.52631116e-01 -1.27844095e+00 -1.25798094e+00 3.22251320e-02 -3.81865650e-01 2.31079772e-01 -2.02148587e-01 3.80846523e-02 -5.34658492e-01 -9.07250583e-01 5.00460267e-01 -3.74450982e-01 -3.04053783e-01 2.00258225e-01 -4.34500277e-01 -1.39454901e-01 1.26710880e+00 -6.14294946e-01 1.09220695e+00 -2.20246792e+00 1.23009263e-02 -2.50324458e-01 4.15156245e-01 2.90720850e-01 -2.65876693e-03 -3.50343823e-01 -5.65924169e-03 -1.51689813e-01 -4.23197776e-01 -4.75742340e-01 -3.14088583e-01 -9.40907672e-02 -6.22276068e-01 2.50857472e-01 7.13677108e-01 1.15993309e+00 -1.11547613e+00 -9.66145039e-01 3.90971482e-01 3.31390858e-01 -3.24392349e-01 2.22842038e-01 -2.21662879e-01 2.10918888e-01 -5.14473021e-01 8.09943140e-01 8.64346862e-01 -6.08099282e-01 -1.59417167e-01 -2.93516994e-01 -3.94438833e-01 2.60539263e-01 -1.20594227e+00 1.51898909e+00 -4.70924228e-02 6.93874598e-01 -3.87164056e-02 -8.85092556e-01 9.62556541e-01 1.27213538e-01 2.63952047e-01 -6.61225080e-01 1.48671031e-01 2.33857468e-01 -4.34994046e-03 -2.48937428e-01 5.36433339e-01 8.08894485e-02 2.40037084e-01 4.89562005e-01 4.94889263e-03 2.15973973e-01 -8.37177485e-02 1.39667824e-01 5.36642015e-01 3.03623497e-01 2.06298620e-01 -3.85569960e-01 5.33179402e-01 -1.38801292e-01 1.05325091e+00 4.61529195e-01 -5.19658983e-01 8.33364427e-01 1.70738801e-01 -3.37438613e-01 -8.03195357e-01 -1.08883131e+00 8.55996683e-02 1.18405282e+00 9.29658711e-01 1.20229758e-01 -6.98516786e-01 -6.24780476e-01 -1.82830811e-01 6.23789072e-01 -7.31277883e-01 -3.43842179e-01 -6.99409783e-01 -8.25379491e-01 -1.96545683e-02 8.48536134e-01 8.12065482e-01 -1.42897463e+00 -9.25986230e-01 3.13322693e-01 -1.32921875e-01 -1.15615404e+00 -7.22898364e-01 1.65270135e-01 -8.56993377e-01 -9.44467783e-01 -8.57912540e-01 -1.32561743e+00 6.47811651e-01 9.75897431e-01 8.93652856e-01 2.33217075e-01 -3.01209867e-01 -3.57741326e-01 -1.54829606e-01 -5.40565252e-01 7.37422109e-02 1.31281152e-01 -3.82515758e-01 3.27094376e-01 2.59573579e-01 -4.22445059e-01 -8.87325764e-01 4.01385456e-01 -7.90307164e-01 6.23281002e-01 1.03522551e+00 7.59138227e-01 5.51478922e-01 -7.17393160e-02 7.23094165e-01 -7.13005662e-01 1.54076919e-01 -3.80190879e-01 -7.53526568e-01 3.35386753e-01 -2.41135910e-01 -1.54536575e-01 1.83974028e-01 -3.57554972e-01 -1.38681948e+00 3.34231704e-01 2.59869337e-01 -3.43614250e-01 -2.11903397e-02 1.87388107e-01 -2.57718325e-01 -1.70530424e-01 2.31671542e-01 6.49074256e-01 -1.70597613e-01 -4.09798831e-01 3.28125656e-01 3.74322057e-01 7.92748153e-01 -1.86690222e-02 8.91749859e-01 4.88842249e-01 -3.55889201e-01 -4.09622848e-01 -1.24260008e+00 -4.79286343e-01 -6.34955585e-01 -2.14912593e-01 9.06306624e-01 -1.13465142e+00 -4.41199869e-01 7.38731980e-01 -1.14870012e+00 -3.31264257e-01 -4.00161222e-02 2.38047361e-01 -2.70867765e-01 3.55942190e-01 -6.57023728e-01 -8.55079949e-01 -5.26188135e-01 -1.11480868e+00 1.08495498e+00 8.35433245e-01 1.96449175e-01 -8.17884684e-01 -4.77902204e-01 5.68634011e-02 3.77885520e-01 1.10930227e-01 5.30598938e-01 -4.20739830e-01 -1.09956896e+00 -2.31305007e-02 -9.77704763e-01 1.05472639e-01 4.85117674e-01 -3.03771198e-02 -8.86540413e-01 -5.31297624e-02 2.92037874e-02 -1.87124327e-01 1.21209288e+00 4.50710446e-01 1.03400517e+00 1.15936548e-02 -5.74088633e-01 4.89826053e-01 1.17381442e+00 -2.10295934e-02 4.06509817e-01 2.56246984e-01 9.66611683e-01 6.12812996e-01 8.59466136e-01 3.48535001e-01 3.68417501e-01 3.65130097e-01 5.36279142e-01 -7.28454709e-01 -3.22224468e-01 -4.37343955e-01 3.93248871e-02 5.58393598e-01 2.83945441e-01 2.39975482e-01 -7.28358865e-01 8.41168761e-01 -2.12630987e+00 -9.10750866e-01 -1.76396117e-01 1.95692694e+00 9.93975878e-01 5.49263299e-01 1.53637305e-01 -1.15791053e-01 1.06726396e+00 3.66129577e-01 -7.86918938e-01 3.04902703e-01 -1.95533469e-01 -2.26734564e-01 3.78133893e-01 3.32520574e-01 -1.50132847e+00 1.36024535e+00 5.80081940e+00 7.59933233e-01 -1.33600366e+00 -6.03202656e-02 9.30586398e-01 1.36850297e-01 -1.71245396e-01 4.28523645e-02 -6.83930159e-01 6.56842709e-01 -9.24653187e-03 -7.83783644e-02 3.45769897e-02 1.07907844e+00 1.01409607e-01 -2.16852337e-01 -6.00195348e-01 7.90460646e-01 -8.42248201e-02 -1.43697369e+00 -4.08120677e-02 -4.64946806e-01 1.04856825e+00 8.36309940e-02 7.47597292e-02 5.63297942e-02 2.05390498e-01 -6.75515890e-01 9.40006375e-01 4.29803103e-01 1.06347963e-01 -9.20434475e-01 6.98095381e-01 4.43963110e-01 -1.51514876e+00 -4.40723449e-02 -5.48305631e-01 -1.46963641e-01 2.92126596e-01 7.60208488e-01 -6.65421784e-01 2.96994001e-01 6.42149150e-01 9.77252185e-01 -5.31283855e-01 1.47296369e+00 -4.95413363e-01 5.49411297e-01 -1.83578387e-01 -2.78635751e-02 3.26709092e-01 -1.25207931e-01 4.60181206e-01 1.25866342e+00 -9.97711569e-02 3.61050665e-02 4.07632172e-01 1.31804645e+00 -5.57411835e-02 -1.92764867e-02 -9.16238129e-02 2.76700616e-01 4.71668571e-01 1.54643750e+00 -1.07850027e+00 -3.75027806e-01 -5.40977001e-01 9.45972204e-01 4.95836467e-01 3.55817050e-01 -1.05195594e+00 -5.19441009e-01 5.04804552e-01 -1.58128127e-01 6.07499778e-01 6.44805059e-02 -5.78437746e-01 -1.16691709e+00 1.03229076e-01 -2.41464108e-01 1.21855393e-01 -6.90710127e-01 -9.70934510e-01 2.93188751e-01 -5.67251623e-01 -1.30499732e+00 3.14679623e-01 -3.03147167e-01 -9.59762692e-01 7.91627824e-01 -1.99597919e+00 -1.23142838e+00 -3.27016711e-01 2.73901999e-01 6.48935795e-01 2.68029094e-01 1.86844230e-01 2.02450573e-01 -7.31675148e-01 4.10877228e-01 -2.65812218e-01 3.82093310e-01 6.10500991e-01 -1.25930488e+00 5.65823138e-01 1.28055382e+00 4.80313227e-02 2.43668407e-01 5.06146967e-01 -6.85387135e-01 -7.83031404e-01 -1.40382302e+00 7.89485931e-01 -2.38871966e-02 7.24734902e-01 -4.45059866e-01 -1.28636026e+00 6.09639049e-01 8.80730376e-02 2.95814037e-01 2.53733724e-01 -2.96178490e-01 -7.45453984e-02 1.16241528e-02 -6.56106889e-01 7.76064396e-01 8.85008335e-01 -3.08267772e-01 -6.69330120e-01 1.83462560e-01 9.76745546e-01 -4.65143710e-01 -1.41618013e-01 6.44527495e-01 3.34146708e-01 -1.06384814e+00 9.92761016e-01 -2.27409869e-01 4.67974573e-01 -6.50142014e-01 2.68166244e-01 -7.89366603e-01 -4.88409340e-01 -4.16221410e-01 -3.57245356e-01 1.27807200e+00 4.31550503e-01 -3.81923884e-01 8.69579673e-01 5.59804499e-01 -3.01245838e-01 -9.39445794e-01 -5.00049353e-01 -2.68739015e-01 -5.04745722e-01 -1.35700285e-01 3.45060498e-01 8.29781294e-01 -4.31366675e-02 3.28250259e-01 -5.29546559e-01 3.53793740e-01 7.38043606e-01 8.81969452e-01 5.02357602e-01 -1.25920844e+00 -5.74887469e-02 -7.83068955e-01 -3.08281958e-01 -1.49184287e+00 -1.46661773e-02 -5.92215061e-01 5.15034139e-01 -1.35679138e+00 5.90962708e-01 -3.16940188e-01 -7.17926621e-01 4.06486809e-01 -1.03594398e+00 3.07955593e-01 2.84720808e-01 2.39881083e-01 -1.02402103e+00 7.26973176e-01 1.54197323e+00 -1.57809883e-01 -3.09454948e-01 6.53470084e-02 -7.64547288e-01 1.04178870e+00 6.50016665e-01 -2.48720765e-01 -2.95721471e-01 -5.13148189e-01 -2.28979990e-01 -1.14238612e-01 6.45306885e-01 -1.04859674e+00 6.23883784e-01 -1.45233601e-01 7.84164906e-01 -9.46360946e-01 4.45409343e-02 -5.57118952e-01 -5.30668497e-01 4.39272404e-01 -1.81081191e-01 -4.42742109e-01 2.81509727e-01 7.29490340e-01 -3.41139853e-01 7.06843287e-02 9.31170285e-01 7.52940923e-02 -1.17231381e+00 2.73656487e-01 -2.47154683e-02 -1.79964617e-01 9.93240952e-01 -1.89490825e-01 -1.70795754e-01 -1.89225003e-01 -5.65150440e-01 4.98228073e-01 4.24715877e-01 5.71510553e-01 6.79557502e-01 -1.17362821e+00 -6.76762700e-01 1.86771423e-01 -7.98390061e-02 3.51535857e-01 2.39375994e-01 1.08213258e+00 -2.41315588e-01 1.46703854e-01 -2.56986380e-01 -6.78752840e-01 -9.42041814e-01 6.45671189e-01 2.08734408e-01 -1.50241433e-02 -6.60865903e-01 1.13725054e+00 7.92487681e-01 -2.50793546e-02 2.34882146e-01 -3.03088844e-01 -3.03778052e-01 -3.26545425e-02 7.56721795e-01 1.89939246e-01 -2.81041116e-01 -6.48727298e-01 -2.87251562e-01 4.97722417e-01 -4.64666337e-01 4.02203172e-01 1.07171285e+00 -3.64030659e-01 -3.61983962e-02 2.83770233e-01 8.55325520e-01 -2.18375012e-01 -1.86831045e+00 -5.21601140e-01 2.26029381e-01 -4.59237874e-01 3.85200351e-01 -5.44018507e-01 -1.29379082e+00 8.87018681e-01 5.31501114e-01 3.88959944e-02 1.23179853e+00 -1.55317336e-01 1.07056308e+00 -2.86727082e-02 2.72572130e-01 -1.00739801e+00 2.11889163e-01 1.98813930e-01 5.52691162e-01 -1.65585065e+00 -5.66730574e-02 -7.80713916e-01 -7.00304568e-01 8.24621797e-01 8.02314878e-01 -3.25160682e-01 4.79858428e-01 2.07517117e-01 1.14711963e-01 1.24198444e-01 -4.83083397e-01 -4.49949652e-01 4.32736278e-01 3.02870661e-01 3.83658975e-01 1.04722656e-01 -1.12668127e-01 6.78869128e-01 3.69722992e-01 -2.49874480e-02 2.47477263e-01 9.98113334e-01 -9.35513556e-01 -7.40288377e-01 -1.33950785e-01 3.16864669e-01 -3.38570803e-01 -2.94346720e-01 -3.24698389e-01 6.03402853e-01 -7.89784789e-02 8.52965891e-01 2.05547437e-01 -2.24632010e-01 -8.88544321e-02 -2.68391371e-01 -4.92566265e-03 -4.96241659e-01 -1.18823268e-01 3.23043615e-01 -5.03480077e-01 -4.81436282e-01 -3.47744912e-01 -7.05227196e-01 -1.61137438e+00 2.05022812e-01 -7.25076616e-01 -1.49202961e-02 1.09463334e-01 1.01530182e+00 4.32545245e-01 7.51325071e-01 5.43228984e-01 -1.10819519e+00 -2.18533397e-01 -9.75833833e-01 -4.68143702e-01 3.20087224e-01 4.21900213e-01 -7.91571558e-01 -9.00499001e-02 2.02305600e-01]
[9.789856910705566, -0.3667083978652954]
451bc5a5-c052-4f30-8b09-319bb4db97cb
unit3d-a-unified-transformer-for-3d-dense
2212.00836
null
https://arxiv.org/abs/2212.00836v1
https://arxiv.org/pdf/2212.00836v1.pdf
UniT3D: A Unified Transformer for 3D Dense Captioning and Visual Grounding
Performing 3D dense captioning and visual grounding requires a common and shared understanding of the underlying multimodal relationships. However, despite some previous attempts on connecting these two related tasks with highly task-specific neural modules, it remains understudied how to explicitly depict their shared nature to learn them simultaneously. In this work, we propose UniT3D, a simple yet effective fully unified transformer-based architecture for jointly solving 3D visual grounding and dense captioning. UniT3D enables learning a strong multimodal representation across the two tasks through a supervised joint pre-training scheme with bidirectional and seq-to-seq objectives. With a generic architecture design, UniT3D allows expanding the pre-training scope to more various training sources such as the synthesized data from 2D prior knowledge to benefit 3D vision-language tasks. Extensive experiments and analysis demonstrate that UniT3D obtains significant gains for 3D dense captioning and visual grounding.
['Angel X. Chang', 'Matthias Nießner', 'Xinlei Chen', 'Ronghang Hu', 'Dave Zhenyu Chen']
2022-12-01
null
null
null
null
['dense-captioning', '3d-dense-captioning']
['computer-vision', 'computer-vision']
[ 2.13657722e-01 3.84333819e-01 -2.17058837e-01 -5.88620305e-01 -9.30346727e-01 -6.65816128e-01 6.50512934e-01 -3.18067700e-01 1.19886115e-01 5.46574831e-01 6.04848385e-01 -3.84414643e-01 2.73597091e-01 -4.18116063e-01 -1.00003839e+00 -3.36943150e-01 8.59299302e-02 5.22904813e-01 -3.94058138e-01 -2.90527791e-01 -2.13449970e-01 2.61675358e-01 -1.28318667e+00 6.40132725e-01 7.43820131e-01 9.82151985e-01 4.30930167e-01 6.65185094e-01 -4.18502569e-01 7.32493222e-01 -3.05866539e-01 -4.79638815e-01 3.48966420e-01 -4.30233181e-01 -8.99633467e-01 3.69753063e-01 8.90631735e-01 -4.68005806e-01 -3.57059062e-01 7.70895898e-01 6.22397423e-01 -1.67894036e-01 5.23555279e-01 -1.43358123e+00 -1.06369269e+00 2.81004220e-01 -6.81273937e-01 -2.10808575e-01 5.84229469e-01 3.72412086e-01 1.24952149e+00 -1.06917834e+00 5.11910141e-01 1.45680153e+00 4.59333658e-01 8.44612420e-01 -1.44083643e+00 -5.59491634e-01 4.30894703e-01 -8.05491880e-02 -1.17597640e+00 -2.69677818e-01 6.79823339e-01 -5.82237899e-01 1.03691554e+00 -6.40117303e-02 5.73467135e-01 1.49174809e+00 -2.01126531e-01 1.02771568e+00 1.13598394e+00 -8.22159499e-02 -1.50579661e-01 -5.89232333e-02 6.24964200e-02 9.19832408e-01 1.00722320e-01 -2.75329612e-02 -9.19880331e-01 1.32442012e-01 8.34532559e-01 -2.59705186e-01 -2.01574370e-01 -6.83060169e-01 -1.64438272e+00 6.85360491e-01 7.74710596e-01 -7.95482397e-02 -2.01528683e-01 5.14225125e-01 2.07321763e-01 1.77369580e-01 3.55135530e-01 5.68607926e-01 -3.75867605e-01 1.61398977e-01 -7.15996325e-01 2.19538227e-01 4.84262586e-01 1.33876097e+00 1.07713687e+00 5.24553321e-02 -5.05558252e-01 5.91206670e-01 6.68873429e-01 9.00436640e-01 -2.27409929e-01 -8.68273854e-01 8.60146344e-01 7.81398296e-01 5.18769622e-02 -8.44126523e-01 -2.65321732e-01 -5.19855738e-01 -9.92051840e-01 1.15109466e-01 3.05395693e-01 -3.04620296e-01 -1.22751689e+00 2.20235014e+00 1.03653714e-01 1.39505252e-01 2.17391208e-01 1.26291680e+00 1.27985775e+00 7.27281988e-01 2.41314679e-01 4.28033173e-01 1.40922117e+00 -1.05684257e+00 -6.05549574e-01 -5.95871985e-01 5.10498226e-01 -7.78756440e-01 1.22368169e+00 -1.90337971e-01 -1.15758395e+00 -5.98554134e-01 -9.17604625e-01 -5.68311930e-01 -2.38457903e-01 7.83133134e-02 1.03881192e+00 1.90871507e-01 -1.34886801e+00 -1.39143124e-01 -6.22511685e-01 -4.47966337e-01 5.68302393e-01 3.10715914e-01 -6.08085990e-01 -2.74434209e-01 -1.11516249e+00 1.03856492e+00 2.31617734e-01 3.72781724e-01 -1.30993402e+00 -8.00127447e-01 -1.25401831e+00 -1.28058031e-01 1.31308138e-01 -1.58066225e+00 1.34485662e+00 -5.49936831e-01 -1.26772726e+00 1.49169159e+00 -3.38158250e-01 -2.41269201e-01 1.09503791e-01 -2.84542143e-01 2.11057231e-01 3.56865861e-02 1.61605701e-01 1.34542704e+00 7.27578104e-01 -1.58067644e+00 -1.19959436e-01 -2.74344593e-01 3.63590509e-01 5.70621669e-01 -9.97368768e-02 -3.42324615e-01 -6.68178082e-01 -4.40099120e-01 3.94115895e-02 -8.52268875e-01 -1.19567759e-01 2.57578254e-01 -5.45954943e-01 3.23527865e-02 5.20649195e-01 -4.55388665e-01 5.89888752e-01 -2.00279903e+00 8.83974075e-01 -1.02691121e-01 5.07315993e-01 1.56641677e-02 -5.77125013e-01 5.61161101e-01 -3.16898227e-02 -1.31473765e-01 -2.83436328e-01 -8.06922734e-01 4.54191923e-01 4.91000444e-01 -6.09293759e-01 1.16062164e-01 7.20999062e-01 1.46839964e+00 -9.98726070e-01 -5.46173751e-01 2.75456250e-01 4.88599837e-01 -6.22030497e-01 7.57939219e-01 -5.71109712e-01 5.48110902e-01 -4.10840273e-01 8.79172266e-01 6.19170725e-01 -6.46645486e-01 1.07099272e-01 -5.75580776e-01 1.72177359e-01 1.96268648e-01 -6.64060712e-01 2.52744222e+00 -4.74171430e-01 7.52690494e-01 2.91082174e-01 -1.01506829e+00 1.09661758e+00 2.64538139e-01 2.23909169e-01 -9.68550026e-01 4.17609625e-02 2.90129967e-02 -4.54048425e-01 -5.60705841e-01 5.04902005e-01 -2.78311163e-01 -4.07046020e-01 4.43507642e-01 5.42143524e-01 -3.99663866e-01 -2.04287738e-01 5.06399691e-01 7.00388730e-01 4.84036833e-01 -1.34302601e-01 -3.03789005e-02 1.60483912e-01 1.52727455e-01 2.05937754e-02 6.80502713e-01 3.82434539e-02 8.91300142e-01 5.45264065e-01 -2.30687693e-01 -1.19410574e+00 -1.32130682e+00 1.76934991e-02 9.60388005e-01 4.13797915e-01 -4.08834755e-01 -4.32612032e-01 -6.20748222e-01 9.59340036e-02 5.07595420e-01 -7.26468086e-01 -1.33749813e-01 -3.01256984e-01 -3.47422242e-01 6.56080544e-01 3.63600880e-01 6.26096129e-01 -7.22267032e-01 -1.61347657e-01 -1.87637091e-01 -4.49305475e-01 -1.42652547e+00 -3.84461254e-01 2.68407047e-01 -7.94705868e-01 -7.67398000e-01 -1.14599752e+00 -9.33760107e-01 6.16323352e-01 6.78904772e-01 1.60034847e+00 -1.56162396e-01 1.20391645e-01 7.85092652e-01 -1.67065427e-01 -2.59440958e-01 -2.67847151e-01 2.12660998e-01 -4.06809837e-01 -4.32071537e-02 2.01651767e-01 -6.16031528e-01 -3.92136127e-01 2.42522657e-01 -6.78449810e-01 7.59172797e-01 8.29435229e-01 7.23602295e-01 5.27253091e-01 -1.03962028e+00 3.72461855e-01 -3.53862584e-01 5.59779644e-01 -5.96636474e-01 -3.27338904e-01 2.20360458e-01 -1.63299188e-01 3.59580159e-01 -8.99839625e-02 -1.84129983e-01 -8.39038610e-01 2.52762642e-02 -1.31745666e-01 -7.31639743e-01 -2.59597272e-01 3.67270976e-01 -4.39878523e-01 1.15235627e-01 3.51306409e-01 2.42467597e-01 3.19229037e-01 -4.38088894e-01 9.47200835e-01 2.66282201e-01 6.84610248e-01 -8.51328433e-01 1.08970284e+00 2.91446358e-01 1.50167897e-01 -5.39566338e-01 -1.31799531e+00 -2.23960400e-01 -5.14264464e-01 -1.29210353e-01 1.41058683e+00 -1.43867373e+00 -7.27555215e-01 3.92362416e-01 -1.46798551e+00 -5.98315477e-01 -1.12228073e-01 2.53061742e-01 -6.66369915e-01 2.42953598e-01 -2.46246457e-01 -3.16155732e-01 -3.32102001e-01 -1.15295970e+00 1.79206979e+00 -2.52030361e-02 -1.67926773e-01 -1.06251550e+00 2.70979285e-01 6.46480560e-01 1.86204076e-01 2.99638420e-01 8.98386717e-01 -1.33880302e-01 -9.05193090e-01 1.59195960e-01 -7.54707992e-01 3.63236189e-01 1.43967807e-01 -4.64950800e-01 -1.01593423e+00 -1.66619465e-01 -6.05870962e-01 -8.27483952e-01 9.18995380e-01 6.09409660e-02 7.76446223e-01 -8.93221274e-02 -1.50727972e-01 8.70822132e-01 1.21209180e+00 -3.90380412e-01 4.91885692e-01 1.04017086e-01 1.08350980e+00 5.45182168e-01 3.03331167e-01 5.49234934e-02 8.23325098e-01 7.65841246e-01 9.81425822e-01 -6.72667921e-01 -4.99301851e-01 -5.71622908e-01 4.41307753e-01 6.89300954e-01 1.85081914e-01 -2.57706434e-01 -1.02568626e+00 4.57344890e-01 -1.85556269e+00 -9.07488942e-01 1.69686347e-01 1.74689412e+00 8.83971870e-01 -2.30781212e-01 -5.56666404e-02 -5.21102428e-01 6.03917897e-01 4.65608805e-01 -5.57579696e-01 -2.96911508e-01 -4.48428214e-01 -1.29597843e-01 5.89684360e-02 5.53708971e-01 -8.99532378e-01 1.02248549e+00 6.57695103e+00 3.18684936e-01 -1.01147962e+00 -6.95493221e-02 4.52138722e-01 -1.78309336e-01 -9.24970388e-01 -8.34915265e-02 -8.53864431e-01 3.90447080e-02 3.61270338e-01 7.75025338e-02 1.72847643e-01 4.45835769e-01 -8.51290207e-03 2.66056269e-01 -1.45566738e+00 1.31948090e+00 1.30113974e-01 -1.54715717e+00 6.13429427e-01 1.12112641e-01 9.06467319e-01 1.75677881e-01 2.84378916e-01 3.60844165e-01 3.69239330e-01 -1.33959758e+00 9.86150026e-01 5.14386356e-01 8.66978526e-01 -2.96468258e-01 5.53542912e-01 2.11752392e-02 -1.04112613e+00 3.12207520e-01 -1.10985763e-01 -1.95057206e-02 4.22095805e-01 3.82638752e-01 -7.67253458e-01 6.93049788e-01 4.40900236e-01 1.02720809e+00 -4.78593886e-01 6.25132620e-01 -3.59132349e-01 2.55597293e-01 -8.87447596e-02 9.78185013e-02 7.65694797e-01 -8.76834318e-02 6.42128408e-01 1.20084572e+00 2.62649924e-01 -1.30764544e-01 6.12126999e-02 1.19710326e+00 -1.79615200e-01 -3.27208459e-01 -9.88568544e-01 -2.09209949e-01 1.46995023e-01 1.16747165e+00 -1.04848340e-01 -1.90555811e-01 -6.56437159e-01 1.02892482e+00 5.14054179e-01 6.93262577e-01 -8.36983263e-01 1.39227167e-01 1.14980292e+00 2.19506770e-02 3.36342812e-01 -8.46438050e-01 -4.48682487e-01 -1.43923199e+00 -1.26590848e-01 -8.89630437e-01 1.85665786e-01 -1.35919893e+00 -1.38861561e+00 5.86023152e-01 1.24261506e-01 -1.12263548e+00 -2.23675549e-01 -7.51914799e-01 -4.15722340e-01 1.13277197e+00 -1.78406751e+00 -1.72226763e+00 -7.61200905e-01 9.29174304e-01 3.05420727e-01 -9.65680853e-02 8.22177649e-01 2.01432139e-01 -3.49700123e-01 5.76252401e-01 -5.53095698e-01 9.61708054e-02 7.99485683e-01 -1.34674811e+00 4.63030845e-01 6.38317943e-01 2.87522614e-01 4.41047400e-01 6.03797257e-01 -3.79929096e-01 -1.90747321e+00 -1.12807655e+00 7.08448827e-01 -6.97989106e-01 5.95467925e-01 -8.10090482e-01 -7.43891060e-01 6.70651197e-01 6.41781449e-01 -1.82238460e-01 6.89962685e-01 2.68955082e-01 -7.35302925e-01 -5.66848414e-03 -6.36952281e-01 6.82488263e-01 1.50147843e+00 -1.00393474e+00 -8.07796955e-01 3.99951905e-01 1.01999044e+00 -5.72387457e-01 -7.52409220e-01 3.39579731e-01 2.92468816e-01 -7.76138544e-01 1.21126723e+00 -7.96902418e-01 8.28806043e-01 -4.21965539e-01 -6.45017087e-01 -1.12811553e+00 -1.36984706e-01 -7.19799340e-01 -2.19258934e-01 1.21037316e+00 5.33140779e-01 -1.84265733e-01 7.34892845e-01 2.49738544e-01 -5.50674260e-01 -7.11159348e-01 -7.40066469e-01 -5.12838602e-01 1.42958343e-01 -4.41930205e-01 5.53599238e-01 9.20402169e-01 -2.22123474e-01 9.38876748e-01 -6.77421033e-01 4.13453996e-01 8.66813481e-01 5.22972524e-01 1.14437926e+00 -9.48442698e-01 -6.50835410e-02 -2.77945668e-01 -3.93049628e-01 -1.73761737e+00 4.21117246e-01 -1.36950231e+00 7.86502585e-02 -2.00088811e+00 2.84077853e-01 -2.45483473e-01 -3.20279151e-02 7.63872862e-01 -6.35943487e-02 4.09300894e-01 2.45894581e-01 5.66230677e-02 -8.01760912e-01 9.76255596e-01 1.89705253e+00 -4.50106144e-01 -2.00849436e-02 -4.76386994e-01 -1.00898969e+00 2.83053130e-01 4.93075401e-01 -8.29020888e-02 -7.47622490e-01 -1.26832545e+00 5.70771635e-01 2.67722290e-02 8.98216367e-01 -5.42989612e-01 5.75803444e-02 -1.59149215e-01 2.45269939e-01 -7.58606434e-01 6.43575966e-01 -7.16875434e-01 -8.93724486e-02 -4.12585177e-02 -4.20402646e-01 3.12072443e-05 5.87017238e-01 4.80015874e-01 -4.54653203e-01 3.26494098e-01 4.67423558e-01 -1.57850564e-01 -9.90906894e-01 5.60487568e-01 -5.98290935e-02 2.99779356e-01 8.77011120e-01 -2.26984531e-01 -4.07467335e-01 -5.94959617e-01 -8.33183408e-01 5.62064767e-01 3.90564531e-01 5.81359446e-01 8.18706870e-01 -1.68018186e+00 -8.51462066e-01 2.15527460e-01 4.43761468e-01 3.12104523e-01 2.65207618e-01 5.74887872e-01 -1.70330569e-01 6.93553746e-01 -4.85327959e-01 -1.05937016e+00 -9.74945307e-01 5.62492251e-01 4.19273019e-01 -4.18805145e-02 -4.79054481e-01 9.50546205e-01 6.29953742e-01 -5.86633086e-01 3.49726826e-01 -2.60070175e-01 2.19179586e-01 8.66361558e-02 2.16981262e-01 -2.73770183e-01 -2.46520475e-01 -6.04251266e-01 -3.95837843e-01 7.71817446e-01 1.64931417e-01 -1.33078128e-01 1.21994293e+00 -4.07764226e-01 -6.99268878e-02 3.93970281e-01 1.19315135e+00 -4.95533258e-01 -1.54620707e+00 -3.73604447e-01 -3.90914887e-01 -1.87874034e-01 3.23765054e-02 -7.77287960e-01 -8.95156920e-01 1.31878698e+00 2.05009699e-01 -1.74234763e-01 1.03597009e+00 6.14511907e-01 7.32683599e-01 5.63238144e-01 2.10469067e-01 -2.57627010e-01 5.99024475e-01 7.62666702e-01 1.27054012e+00 -1.32030582e+00 -2.23757714e-01 -3.56827021e-01 -8.01200211e-01 7.55480647e-01 8.62201333e-01 8.47243294e-02 1.79633170e-01 1.29099146e-01 1.12717122e-01 -3.30343872e-01 -9.14252698e-01 -5.54180264e-01 7.26475835e-01 8.80583465e-01 2.66315907e-01 -2.03593582e-01 6.19419575e-01 3.46201539e-01 1.35451062e-02 -2.70002425e-01 1.92228965e-02 6.12015903e-01 -1.47967607e-01 -1.02900612e+00 -1.46821171e-01 -1.15065962e-01 1.07608393e-01 -3.21951985e-01 -6.92510307e-01 8.09514046e-01 1.19964555e-02 5.70778787e-01 1.95898488e-01 -4.78118211e-01 3.46085459e-01 -1.76840685e-02 8.09756458e-01 -7.55998850e-01 -3.03624779e-01 -1.66653439e-01 2.73460418e-01 -7.04677582e-01 -6.36138022e-01 -1.67716473e-01 -1.05746043e+00 -3.08509856e-01 1.18773997e-01 -1.10984124e-01 6.31471753e-01 9.10965979e-01 7.47003853e-01 4.85188186e-01 2.99299628e-01 -1.18542230e+00 -3.43158662e-01 -6.87533259e-01 -1.57718971e-01 4.69504684e-01 5.87967217e-01 -7.82914400e-01 -8.16773623e-02 8.68780538e-02]
[8.181294441223145, -3.3220622539520264]
7928faa3-8f7a-44b4-91ed-def38557319b
non-local-attention-learning-on-large
null
null
https://ieeexplore.ieee.org/document/9006463
https://xiaoyuxin1002.github.io/docs/NLAH.pdf
Non-local Attention Learning on Large Heterogeneous Information Networks
Heterogeneous information network (HIN) summarizes rich structural information in real-world datasets and plays an important role in many big data applications. Recently, graph neural networks have been extended to the representation learning of HIN. One very recent advancement is the hierarchical attention mechanism which incorporates both nodewise and semantic-wise attention. However, since HIN is more likely to be densely connected given its diverse types of edges, repeatedly applying graph convolutional layers can make the node embeddings indistinguishable very quickly. In order to avoid oversmoothness, existing graph neural networks targeting HIN generally suffer from a shallow structure. Consequently, those approaches ignore information beyond the local neighborhood. This design flaw violates the concept of non-local learning, which emphasizes the importance of capturing long-range dependencies. To properly address this limitation, we propose a novel framework of non-local attention in heterogeneous information networks (NLAH). Our framework utilizes a non-local attention structure to complement the hierarchical attention mechanism. In this way, it leverages both local and non-local information simultaneously. Moreover, a weighted sampling schema is designed for NLAH to reduce the computation cost for largescale datasets. Extensive experiments on three different realworld heterogeneous information networks illustrate that our framework exhibits extraordinary scalability and outperforms state-of-the-art baselines with significant margins.
['ChengXiang Zhai', 'Zecheng Zhang', 'Yuxin Xiao', 'Carl Yang']
2019-12-12
null
null
null
2019-ieee-international-conference-on-big-2
['heterogeneous-node-classification']
['graphs']
[-2.37738296e-01 2.95498937e-01 -6.22361779e-01 -3.15166414e-01 -2.81378776e-01 -2.20317483e-01 4.67832148e-01 3.05265188e-01 -1.43661425e-01 4.88319039e-01 4.75265741e-01 -2.68738329e-01 -1.93181574e-01 -1.28444684e+00 -5.61039567e-01 -6.00254536e-01 6.10408485e-02 1.62527099e-01 4.85260427e-01 -2.86969662e-01 -4.89915609e-02 3.49022329e-01 -9.09095109e-01 1.58233956e-01 9.29353058e-01 9.78145599e-01 1.72827423e-01 2.10133761e-01 -4.68232930e-01 1.06744039e+00 -2.63436347e-01 -5.91441512e-01 1.53253868e-01 -2.99981445e-01 -7.77255297e-01 -1.46131873e-01 3.74596268e-01 -2.37754941e-01 -9.01054204e-01 1.23785198e+00 3.40502739e-01 2.21232716e-02 3.18345368e-01 -1.33425367e+00 -1.10249829e+00 8.87257099e-01 -7.27411807e-01 1.84801742e-01 6.60146251e-02 1.10856190e-01 1.62632501e+00 -6.22545362e-01 3.09550464e-01 1.16969585e+00 7.96660781e-01 1.22318573e-01 -9.64520216e-01 -5.95415711e-01 6.95316315e-01 1.51609346e-01 -1.38331568e+00 6.49280027e-02 1.04981005e+00 -1.23657092e-01 8.69073570e-01 8.99400488e-02 6.78397119e-01 1.02481949e+00 8.28279629e-02 9.33545470e-01 4.81798232e-01 1.17659084e-01 -8.33378881e-02 -1.43817455e-01 4.38211292e-01 8.94491076e-01 5.98204732e-01 -3.11741084e-01 -1.68367326e-01 -2.76470669e-02 9.03666377e-01 4.80375081e-01 -2.29723260e-01 -4.37051028e-01 -1.11559236e+00 9.23157752e-01 1.33673394e+00 4.79729116e-01 -4.07689899e-01 2.29531348e-01 5.34064114e-01 1.81115776e-01 5.50979376e-01 2.09155634e-01 -2.64662832e-01 3.26330870e-01 -4.97317195e-01 -1.96578577e-01 6.70151472e-01 1.02905285e+00 9.73846734e-01 6.13010041e-02 -3.28499138e-01 7.17348397e-01 3.57296973e-01 4.38285358e-02 3.61766666e-01 -4.70904857e-01 7.64678955e-01 1.16731787e+00 -5.31389832e-01 -1.56403053e+00 -5.60466766e-01 -7.89577007e-01 -1.37505984e+00 -3.63420129e-01 7.35546872e-02 3.69303375e-02 -8.45297575e-01 1.70898974e+00 2.73807973e-01 2.39800721e-01 -2.94352591e-01 8.46207023e-01 1.13797998e+00 5.72568655e-01 6.06842190e-02 1.93280146e-01 1.06447506e+00 -1.43363404e+00 -6.98680460e-01 -2.30624571e-01 4.87246513e-01 -1.23120464e-01 1.14314914e+00 -2.20125586e-01 -9.44713175e-01 -4.93598521e-01 -1.02692342e+00 -3.59348118e-01 -5.68188906e-01 -3.79866928e-01 9.62814212e-01 3.05612504e-01 -1.00940812e+00 4.13155645e-01 -6.92766964e-01 -3.83906126e-01 6.29009306e-01 3.04237366e-01 -3.04676175e-01 -3.81232291e-01 -1.36297953e+00 3.65220457e-01 4.13548172e-01 2.51556695e-01 -4.88477230e-01 -6.69726074e-01 -1.05081475e+00 5.79908490e-01 6.44166410e-01 -7.64597654e-01 7.20110536e-01 -8.09786260e-01 -1.23675191e+00 3.32036138e-01 -1.29155397e-01 -2.70652413e-01 3.32033753e-01 8.82991701e-02 -4.86831933e-01 1.72897071e-01 1.59343153e-01 3.54505688e-01 4.82249230e-01 -1.10461187e+00 -3.03114921e-01 -2.91305780e-01 4.04379338e-01 8.02304074e-02 -8.57406557e-01 -3.25592220e-01 -9.97443259e-01 -7.79853523e-01 2.60103583e-01 -6.60534203e-01 -3.30118090e-01 -2.12950751e-01 -7.07892299e-01 -3.90533149e-01 8.03974688e-01 -2.52911925e-01 1.60127485e+00 -1.98211110e+00 4.14742827e-02 3.63869816e-01 8.77312005e-01 2.85754830e-01 -4.50127721e-01 5.98116577e-01 2.52004806e-02 4.52575117e-01 -1.80718422e-01 -2.50365227e-01 4.75738235e-02 1.27428591e-01 1.03330366e-01 3.12908620e-01 3.27063203e-01 1.36407506e+00 -1.16528237e+00 -5.14896929e-01 2.78608617e-03 5.66562235e-01 -6.52820289e-01 2.28588954e-01 1.81420445e-02 1.32065266e-01 -7.29027212e-01 6.43975079e-01 6.03941500e-01 -7.95665741e-01 2.89124280e-01 -4.88176882e-01 2.01376945e-01 2.75460154e-01 -8.15210819e-01 1.66194439e+00 -3.73455077e-01 3.95213783e-01 2.01612979e-01 -1.20976210e+00 7.25342512e-01 7.74624348e-02 5.91052830e-01 -7.13597417e-01 7.50210136e-02 -9.39910412e-02 1.14423208e-01 -3.38380605e-01 4.17337537e-01 1.79355577e-01 5.73747121e-02 3.13039184e-01 -3.46037820e-02 4.77918118e-01 1.33674771e-01 6.95625365e-01 1.36543095e+00 -3.98856193e-01 2.50153512e-01 -3.84109318e-01 4.17071164e-01 -4.58083332e-01 8.90347660e-01 7.46935844e-01 -3.30373853e-01 7.14589715e-01 8.10121238e-01 -4.60510135e-01 -6.89659774e-01 -9.65166986e-01 1.43460304e-01 1.03971481e+00 4.06176001e-01 -7.88681269e-01 -4.91227120e-01 -1.08848536e+00 1.37311995e-01 3.90575193e-02 -7.43437767e-01 -4.39350575e-01 -5.91473758e-01 -7.78800070e-01 3.26605767e-01 7.47303367e-01 7.19164670e-01 -9.29671943e-01 1.62744522e-01 2.92485237e-01 -3.00587192e-02 -1.10432303e+00 -7.13292539e-01 -2.18040675e-01 -6.77976429e-01 -1.18895900e+00 -6.31243825e-01 -9.31401730e-01 8.46222818e-01 7.48133898e-01 1.30332887e+00 4.76117253e-01 6.36372566e-02 1.80394098e-01 -4.06586498e-01 6.93330765e-02 2.15686604e-01 6.79434061e-01 -3.28186035e-01 2.55562752e-01 4.32935834e-01 -6.59214199e-01 -7.51497805e-01 1.74477190e-01 -1.03965735e+00 1.34715978e-02 8.93162966e-01 1.00702620e+00 3.82126927e-01 2.91896462e-01 7.16155052e-01 -1.29201460e+00 8.15746903e-01 -9.61750507e-01 -2.46805206e-01 3.26620013e-01 -5.43744922e-01 1.15163885e-01 1.03150344e+00 -2.38762513e-01 -8.74489665e-01 -4.08423215e-01 -1.27251953e-01 -3.33793700e-01 1.86777934e-01 9.16735470e-01 -4.96442258e-01 -1.71646804e-01 9.06101149e-03 -9.20680016e-02 -2.51709610e-01 -4.49837655e-01 2.53559738e-01 3.83355349e-01 1.28950208e-01 -4.62089777e-01 7.99800396e-01 4.18581694e-01 6.51701316e-02 -7.00620890e-01 -9.96673346e-01 -3.91854972e-01 -4.35780972e-01 1.05306089e-01 6.40508294e-01 -7.83815444e-01 -7.15028822e-01 3.65307748e-01 -9.18222070e-01 -2.66282171e-01 -1.86844859e-02 3.10837001e-01 1.12926073e-01 4.75551277e-01 -9.64163423e-01 -4.09611434e-01 -3.51522058e-01 -1.13850474e+00 7.36380219e-01 1.72480583e-01 2.15599239e-01 -1.26292908e+00 -2.02164724e-01 1.90250456e-01 6.80306375e-01 1.53724700e-01 1.04847753e+00 -6.11963689e-01 -8.56381059e-01 -2.45288029e-01 -8.95578504e-01 1.04171701e-01 2.61940360e-01 -3.34876701e-02 -6.61373556e-01 -4.67575431e-01 -4.81355697e-01 -2.54086733e-01 1.24940968e+00 1.91398174e-01 1.46250820e+00 -3.68248522e-01 -3.61501575e-01 7.36570477e-01 1.48278737e+00 -2.08274782e-01 4.25226152e-01 1.44871518e-01 1.43299294e+00 5.52760839e-01 5.24193905e-02 2.05151036e-01 9.58477378e-01 3.85957032e-01 6.90416873e-01 -4.77174610e-01 -3.00156921e-01 -6.15968287e-01 1.11956254e-01 1.19072545e+00 1.66902244e-01 -4.29277420e-01 -8.88827145e-01 6.18899524e-01 -1.94579828e+00 -7.77140439e-01 -1.61609516e-01 1.91386330e+00 4.36097115e-01 2.74353057e-01 1.07567742e-01 -1.51291922e-01 8.76129329e-01 6.72780454e-01 -7.07905293e-01 -3.25297448e-03 -2.10155651e-01 -2.29410857e-01 4.98742133e-01 4.11559641e-01 -9.96192753e-01 8.32864165e-01 5.41553068e+00 7.07134902e-01 -9.14681494e-01 5.28562702e-02 6.77827239e-01 1.51531532e-01 -7.91752100e-01 -1.32649839e-01 -5.97950101e-01 6.93231940e-01 6.14592850e-01 -1.41234159e-01 2.44517431e-01 7.28725612e-01 -1.07189611e-01 4.33251202e-01 -7.22769022e-01 8.12111080e-01 -6.30517304e-02 -1.35991943e+00 2.73344725e-01 1.67121395e-01 7.75386870e-01 4.47602779e-01 -5.94928265e-02 4.91123050e-01 5.27308702e-01 -9.07677233e-01 2.77400404e-01 2.60199398e-01 5.38903296e-01 -8.33556116e-01 8.95840466e-01 1.77409619e-01 -1.89481807e+00 -1.51933566e-01 -4.56757814e-01 -3.29490192e-02 9.86210629e-02 7.96990991e-01 -2.87184507e-01 9.10460472e-01 6.30929589e-01 1.25570726e+00 -7.66153693e-01 9.39390361e-01 -2.72532076e-01 4.94436949e-01 -2.59227097e-01 -3.95633429e-02 6.57487988e-01 -3.53511930e-01 3.16764891e-01 1.10843599e+00 4.25015390e-02 -5.78731261e-02 4.30113703e-01 8.67532074e-01 -5.37929893e-01 5.19890189e-02 -7.50613332e-01 -2.01658815e-01 5.30624092e-01 1.33940887e+00 -6.67738140e-01 -1.96320400e-01 -9.48832154e-01 8.80988896e-01 9.48119640e-01 6.34653926e-01 -7.24242330e-01 -7.14481235e-01 5.78343630e-01 1.16754346e-01 2.84893930e-01 -1.97267249e-01 -1.81962579e-01 -1.40253246e+00 9.25680622e-02 -5.44683695e-01 6.76921248e-01 -2.84132451e-01 -1.78775060e+00 8.87626648e-01 -3.19055974e-01 -9.27147985e-01 3.56892943e-01 -2.90834606e-01 -9.08366919e-01 6.75341964e-01 -1.95344627e+00 -1.47836673e+00 -6.24764740e-01 6.68755174e-01 2.60953933e-01 6.96573481e-02 3.47580820e-01 4.83139426e-01 -1.04583704e+00 8.33358407e-01 -6.73414469e-02 5.65085053e-01 4.70021427e-01 -1.29385722e+00 6.36981785e-01 8.54649961e-01 5.39853573e-02 7.59351671e-01 -5.84584335e-03 -6.68053985e-01 -1.43927610e+00 -1.34654486e+00 7.49639988e-01 -4.51764315e-02 9.23552513e-01 -3.97018909e-01 -1.18206131e+00 7.95036793e-01 2.41017550e-01 4.92881358e-01 5.82681000e-01 3.79867852e-01 -5.97947896e-01 -4.14305180e-01 -8.74201536e-01 6.87734425e-01 1.34409404e+00 -6.87430024e-01 -1.61986679e-01 2.20391914e-01 1.12400186e+00 1.03022397e-01 -1.12883091e+00 5.67380667e-01 3.05788368e-01 -1.09877741e+00 8.91206026e-01 -6.45764410e-01 3.26339126e-01 -2.17902750e-01 1.02167524e-01 -1.20883048e+00 -7.60413229e-01 -5.99328399e-01 -3.94887775e-01 1.42807889e+00 3.21377695e-01 -9.75947440e-01 9.63771462e-01 5.63482940e-01 -2.38126472e-01 -1.03777122e+00 -5.77347159e-01 -6.98475003e-01 -3.81350629e-02 -1.09031260e-01 8.52962554e-01 1.09934735e+00 -9.75299906e-03 7.07191348e-01 -3.34172219e-01 2.25848809e-01 5.99371552e-01 2.58367985e-01 6.12697601e-01 -1.31588161e+00 -1.59729689e-01 -7.08942115e-01 -6.04052782e-01 -1.40563214e+00 3.98332715e-01 -1.01065969e+00 -3.56747448e-01 -1.77547407e+00 4.25663084e-01 -6.86447263e-01 -7.78878629e-01 5.37472785e-01 -5.62305748e-01 1.88157380e-01 1.76343352e-01 1.78488955e-01 -9.12359595e-01 7.96095610e-01 1.30181253e+00 -2.54267871e-01 2.76938826e-02 -2.46752501e-01 -9.33546007e-01 5.57915866e-01 8.77736330e-01 -2.15270326e-01 -6.96636438e-01 -8.12870622e-01 3.70164484e-01 -3.10233831e-01 8.60191733e-02 -7.86455989e-01 4.94456857e-01 -4.09349315e-02 6.01404645e-02 -4.30440664e-01 4.57768403e-02 -9.11155522e-01 -1.34690911e-01 2.46622041e-01 -2.82941043e-01 2.10741639e-01 -2.89750218e-01 9.23521340e-01 -4.91606504e-01 2.86899537e-01 6.16581559e-01 -1.84029475e-01 -7.35095322e-01 1.01901066e+00 1.58674747e-01 2.68010080e-01 9.30952907e-01 8.42581242e-02 -5.98659158e-01 -4.15524513e-01 -2.02048615e-01 5.56888938e-01 5.16887248e-01 3.93267840e-01 5.23703754e-01 -1.47373557e+00 -6.00730836e-01 3.63010734e-01 2.54226983e-01 3.56161982e-01 3.48992288e-01 1.00736034e+00 -3.84999424e-01 4.18163449e-01 1.60319030e-01 -2.89683342e-01 -6.95053101e-01 8.47064972e-01 1.69230804e-01 -5.34372211e-01 -8.03659797e-01 8.39350045e-01 5.52608132e-01 -3.80277276e-01 2.78264046e-01 -2.48169407e-01 -1.33526519e-01 -8.13063607e-02 3.93568784e-01 2.28423566e-01 -5.25068305e-02 -6.20363593e-01 -2.00086147e-01 4.96624649e-01 -2.95006335e-01 6.34687603e-01 1.24758005e+00 -3.03036630e-01 -2.48381287e-01 3.00825894e-01 1.58402562e+00 1.20940618e-01 -1.20622659e+00 -5.53984880e-01 -2.63287831e-04 -3.91719073e-01 2.93347389e-01 -3.01345557e-01 -1.73123229e+00 9.70726430e-01 -2.69596994e-01 6.85231388e-01 1.14861572e+00 2.74142306e-02 1.17544794e+00 3.72380883e-01 2.28884265e-01 -8.20408463e-01 4.09472175e-02 4.78533536e-01 5.36632538e-01 -1.53120112e+00 -4.29584645e-02 -6.18434131e-01 -5.79580307e-01 9.14801002e-01 9.37679172e-01 -2.00388566e-01 8.42239559e-01 6.47356063e-02 -2.62351930e-01 -4.53546494e-01 -6.20556772e-01 -3.64657372e-01 3.02345395e-01 3.49373907e-01 3.80906850e-01 -4.10764553e-02 -1.80399045e-01 6.15699649e-01 3.63078296e-01 -4.82356578e-01 2.49346614e-01 7.13759303e-01 -2.16639668e-01 -9.69053209e-01 3.64500165e-01 5.65647840e-01 -4.43492681e-01 -3.64597648e-01 -5.21274090e-01 8.56893599e-01 -1.78023577e-01 8.61078382e-01 1.54161239e-02 -4.63488221e-01 1.74150825e-01 -6.05348289e-01 -5.96307330e-02 -5.08521974e-01 -6.95398808e-01 -1.15430593e-01 -9.92368907e-02 -7.61025548e-01 -2.80265212e-01 -4.01706173e-04 -1.21447051e+00 -5.88131249e-01 -2.42818281e-01 1.82366759e-01 1.77665800e-01 6.39515400e-01 6.83223724e-01 8.27100635e-01 8.01196933e-01 -6.05863392e-01 -2.27234676e-01 -7.78269231e-01 -6.32769287e-01 6.23360276e-01 4.05832171e-01 -5.47417581e-01 -4.23666209e-01 -7.52216578e-01]
[7.270221710205078, 6.2669243812561035]
e5fc6dcb-f991-4c03-aa71-dd14451ef8e1
classification-and-online-clustering-of-zero
2305.00605
null
https://arxiv.org/abs/2305.00605v1
https://arxiv.org/pdf/2305.00605v1.pdf
Classification and Online Clustering of Zero-Day Malware
A large amount of new malware is constantly being generated, which must not only be distinguished from benign samples, but also classified into malware families. For this purpose, investigating how existing malware families are developed and examining emerging families need to be explored. This paper focuses on the online processing of incoming malicious samples to assign them to existing families or, in the case of samples from new families, to cluster them. We experimented with seven prevalent malware families from the EMBER dataset, with four in the training set and three additional new families in the test set. Based on the classification score of the multilayer perceptron, we determined which samples would be classified and which would be clustered into new malware families. We classified 97.21% of streaming data with a balanced accuracy of 95.33%. Then, we clustered the remaining data using a self-organizing map, achieving a purity from 47.61% for four clusters to 77.68% for ten clusters. These results indicate that our approach has the potential to be applied to the classification and clustering of zero-day malware into malware families.
['Róbert Lórencz', 'Martin Jureček', 'Olha Jurečková']
2023-05-01
null
null
null
null
['online-clustering']
['computer-vision']
[ 2.21598998e-01 -2.66422033e-01 -8.87275040e-02 -2.96978861e-01 1.68172553e-01 -6.81871116e-01 6.59086883e-01 6.35544717e-01 -2.21477985e-01 5.62613130e-01 -3.61003995e-01 -5.01977742e-01 8.61896351e-02 -7.93204367e-01 -4.08060431e-01 -1.02358615e+00 -4.73366112e-01 7.32515097e-01 4.53760356e-01 2.20941558e-01 4.42459881e-01 7.82520533e-01 -1.82576752e+00 6.68627858e-01 7.44836807e-01 8.43212664e-01 1.00351304e-01 7.97108889e-01 -2.89247066e-01 7.01508820e-01 -8.95131767e-01 -1.30758002e-01 1.12380005e-01 7.75042847e-02 -6.08918369e-01 2.31885344e-01 -2.48202175e-01 -3.83794069e-01 7.95763284e-02 1.06380641e+00 -2.79119879e-01 -2.39617079e-01 8.34771812e-01 -1.37189198e+00 -1.69546768e-01 7.15608776e-01 -3.87139767e-01 4.06215966e-01 1.69383004e-01 5.11146523e-02 4.91639495e-01 -7.75302112e-01 7.75376678e-01 1.05317819e+00 4.73884761e-01 4.71495479e-01 -1.08092737e+00 -8.12271059e-01 -1.85346574e-01 4.24849033e-01 -1.16827703e+00 -5.50717652e-01 6.12360120e-01 -1.14029360e+00 9.97842312e-01 2.04500541e-01 7.34133482e-01 1.01934266e+00 3.59408587e-01 2.13416189e-01 8.94075215e-01 -1.03152767e-02 3.78785640e-01 6.37684703e-01 6.89112604e-01 3.68437350e-01 5.50551593e-01 6.06183559e-02 1.56351939e-01 -4.47493941e-01 6.98367655e-02 3.04797590e-01 -1.33537948e-01 -2.11618826e-01 -9.44993019e-01 1.02517533e+00 2.11342096e-01 6.54580057e-01 -4.77898747e-01 -5.63571751e-01 6.31494761e-01 1.60608575e-01 3.99666995e-01 4.19705659e-01 -3.98470312e-01 2.19887774e-02 -5.85950196e-01 -1.66116938e-01 8.30592036e-01 4.25684959e-01 8.49596441e-01 2.06433579e-01 4.88393545e-01 5.10892868e-01 1.97975323e-01 3.40616405e-01 6.10230029e-01 -5.43823004e-01 1.20356180e-01 9.36624169e-01 -2.14431033e-01 -1.05515206e+00 -3.23387265e-01 -1.77498311e-01 -7.53436208e-01 -2.53253710e-02 1.67018518e-01 -9.31750461e-02 -7.92294085e-01 1.21872127e+00 2.98536807e-01 -1.48522601e-01 1.06841102e-01 1.94416389e-01 3.86701465e-01 1.08208001e+00 -2.72129029e-01 -5.44496775e-01 1.05681455e+00 -3.68040711e-01 -3.08571041e-01 2.11548135e-01 3.10990632e-01 -4.96558070e-01 4.53762978e-01 6.49942935e-01 -4.11047637e-01 -4.93025780e-01 -8.68169665e-01 1.20427775e+00 -4.21657085e-01 -1.82821929e-01 3.07152867e-01 7.96309173e-01 -1.17053699e+00 5.62168241e-01 -8.66937757e-01 -5.42333066e-01 3.32437664e-01 2.86559135e-01 -2.83180177e-01 1.25970170e-01 -8.73343229e-01 3.61061633e-01 9.93994355e-01 -3.79109144e-01 -9.77124095e-01 -3.84245515e-01 -5.57408988e-01 3.81975472e-02 -4.08888534e-02 -3.43701318e-02 7.45797515e-01 -9.52962518e-01 -1.01559794e+00 5.92031538e-01 -2.05148742e-01 -5.13872385e-01 -4.20399643e-02 6.22269750e-01 -6.67672276e-01 3.92792612e-01 3.89515460e-02 5.01167595e-01 1.16443646e+00 -1.54352701e+00 -1.07265604e+00 -5.17842650e-01 -3.19111556e-01 -6.05395734e-01 -8.26848984e-01 2.88336486e-01 1.54816061e-01 -4.05868322e-01 1.01075238e-02 -1.05426466e+00 1.23459779e-01 -6.95814669e-01 -3.21064174e-01 -2.55799353e-01 1.29455459e+00 -7.90645361e-01 1.28751922e+00 -2.27191305e+00 -8.11166987e-02 4.61318374e-01 6.13480508e-01 4.89945352e-01 1.35944173e-01 3.72191340e-01 -2.52561063e-01 3.42667311e-01 -2.63209313e-01 -6.75042253e-03 -3.65930229e-01 1.16188310e-01 -4.87567514e-01 4.37072814e-01 5.47890663e-01 1.24434844e-01 -9.77773845e-01 -1.52650326e-01 2.49469746e-02 4.41487581e-01 -4.30659533e-01 2.60002524e-01 8.69113728e-02 4.96096373e-01 -1.36692956e-01 9.80856299e-01 6.64471209e-01 -9.28841680e-02 -1.60460230e-02 1.54527515e-01 -6.28382936e-02 -1.35864373e-02 -6.95023835e-01 1.51601359e-01 1.17940284e-01 8.33741009e-01 1.93517819e-01 -1.08756220e+00 1.16548729e+00 2.65557796e-01 6.92295134e-01 -1.06749155e-01 3.64293814e-01 2.23294929e-01 5.75625956e-01 -5.29075205e-01 5.98611355e-01 -6.09579831e-02 1.23851188e-01 4.89090651e-01 2.04832122e-01 3.70032996e-01 6.10717833e-01 -5.20064421e-02 1.40832746e+00 -7.80191243e-01 2.33950123e-01 -3.60665947e-01 6.57253087e-01 6.34333566e-02 5.04064798e-01 4.03601348e-01 -6.30038202e-01 6.35366514e-02 5.98158002e-01 -4.78627145e-01 -9.60428119e-01 -1.29567647e+00 -4.30046737e-01 7.50721574e-01 -2.69228667e-01 -2.52695709e-01 -7.89169669e-01 -6.45386219e-01 -2.30885684e-01 7.19386399e-01 -4.04770702e-01 -3.64876747e-01 -4.08267289e-01 -8.14592063e-01 3.95995736e-01 -9.64012928e-03 2.98602004e-02 -1.34073794e+00 -5.46392441e-01 2.49178886e-01 3.67033780e-02 -8.60855997e-01 1.29098147e-01 5.95115781e-01 -7.62591779e-01 -1.30695617e+00 -2.24720433e-01 -7.67704368e-01 9.29979026e-01 3.17964435e-01 7.48870075e-01 -3.13388258e-02 -1.54824704e-01 2.01696873e-01 -7.50550568e-01 -4.14804906e-01 -1.24980915e+00 -9.29851085e-03 6.19854391e-01 1.77732840e-01 4.50866431e-01 -5.26134193e-01 2.54839770e-02 3.94829035e-01 -8.84052992e-01 -5.16819417e-01 2.20428765e-01 4.33314800e-01 1.57966003e-01 6.56736135e-01 6.66799605e-01 -6.59373581e-01 5.63729405e-01 -1.09075165e+00 -3.99106622e-01 -1.17809020e-01 -3.21591079e-01 -3.02439868e-01 1.34706247e+00 -8.48063588e-01 -5.09503603e-01 2.14492038e-01 6.14256226e-02 -6.12314463e-01 -5.27038515e-01 2.05064192e-01 -8.53992999e-02 1.71718568e-01 7.56297648e-01 2.47738123e-01 4.67903130e-02 -5.85221574e-02 -1.67922214e-01 1.10444331e+00 3.07444066e-01 -1.48545653e-01 8.55987430e-01 3.81521016e-01 -4.97308105e-01 -1.22109962e+00 2.38710001e-01 -7.07660973e-01 -7.40372241e-01 -3.83134604e-01 8.29646170e-01 -3.16320956e-01 -5.00570178e-01 5.42486906e-01 -1.20461488e+00 3.96213122e-02 -1.51765347e-01 3.73458087e-01 -5.51727694e-03 5.02968192e-01 -7.44735658e-01 -8.98773730e-01 -4.52068299e-01 -1.38311863e+00 6.53528750e-01 -8.16537663e-02 -5.34160614e-01 -9.20424998e-01 1.01962455e-01 1.88194104e-02 2.81363726e-01 2.84424603e-01 9.57665026e-01 -1.20773363e+00 -1.10235982e-01 -5.50477266e-01 3.34474095e-03 3.61429363e-01 6.15707695e-01 7.70987749e-01 -9.42836046e-01 -6.48302197e-01 4.38677818e-01 1.03551805e-01 8.10981154e-01 4.79139909e-02 7.71818995e-01 -4.63356704e-01 -4.55983967e-01 2.83403754e-01 8.54910672e-01 1.29076707e+00 1.19557075e-01 1.81629121e-01 6.58707023e-01 1.02171803e+00 3.76920521e-01 5.35412133e-01 -3.72792371e-02 8.15647021e-02 5.33210993e-01 5.51979303e-01 6.56921089e-01 3.01927388e-01 7.03852177e-01 1.22334313e+00 3.04805458e-01 -2.56722778e-01 -1.33733690e+00 4.17015553e-01 -1.30693114e+00 -1.15278733e+00 -2.01507270e-01 1.94829845e+00 4.60562766e-01 3.98508638e-01 4.39036608e-01 6.25178516e-01 1.21661389e+00 -1.93182856e-01 -4.10899878e-01 -8.65868270e-01 2.55246907e-01 -8.76925066e-02 1.33735433e-01 2.06643015e-01 -1.56737614e+00 3.23185414e-01 5.94571304e+00 7.18312323e-01 -1.46230757e+00 -3.04223955e-01 7.98999429e-01 2.60248721e-01 -4.09941049e-03 -8.27858523e-02 -7.61331379e-01 8.68089557e-01 1.38478982e+00 -2.18067750e-01 5.38700402e-01 8.87353957e-01 6.36851788e-02 9.14209187e-02 -8.56777847e-01 8.28858972e-01 1.62335649e-01 -8.90063763e-01 7.49879777e-02 3.14275801e-01 5.68592966e-01 4.93131056e-02 -1.37239486e-01 2.04764411e-01 2.15662345e-01 -7.22835183e-01 4.54244614e-01 3.55877191e-01 4.21303183e-01 -8.28546524e-01 7.76146591e-01 7.66160846e-01 -1.33194888e+00 -4.38886672e-01 -2.98990071e-01 -9.28350538e-02 -2.11131543e-01 7.75421023e-01 -1.53631461e+00 2.00030550e-01 7.73312032e-01 7.21813858e-01 -6.38541579e-01 8.68996978e-01 2.54902482e-01 8.84678721e-01 -2.67244905e-01 -2.41212994e-01 2.87221801e-02 -1.92241177e-01 5.09503603e-01 1.17892683e+00 4.58554059e-01 -4.41684633e-01 2.86887735e-01 6.34787202e-01 3.10939848e-01 -9.76662189e-02 -7.58882523e-01 -4.35291618e-01 7.05354393e-01 1.28439629e+00 -1.15661836e+00 -7.32241988e-01 1.28220454e-01 5.00274718e-01 4.19529006e-02 -4.00654785e-02 -8.37884128e-01 -5.66963494e-01 5.71457565e-01 4.46631372e-01 3.25942397e-01 -1.76723197e-01 3.14589925e-02 -9.96545911e-01 -1.31537125e-01 -9.15065110e-01 2.36648872e-01 -2.35420823e-01 -1.65693641e+00 1.15806508e+00 8.78467932e-02 -1.43973553e+00 -5.91437757e-01 -7.44452059e-01 -8.41241777e-01 2.48866633e-01 -6.41497076e-01 -5.39846122e-01 -3.02686483e-01 3.63105536e-01 2.99414724e-01 -5.39398909e-01 6.94646537e-01 4.09270823e-01 -7.35876083e-01 2.39990160e-01 6.28990829e-01 8.04222301e-02 3.85907799e-01 -8.91120791e-01 2.48202741e-01 9.26215172e-01 -2.27284431e-01 6.72163308e-01 5.17373681e-01 -8.28491688e-01 -1.17426658e+00 -1.51783121e+00 7.57370949e-01 -1.68100968e-01 1.11249197e+00 -4.69673514e-01 -1.19212723e+00 3.51965159e-01 -5.37825711e-02 -4.07966018e-01 8.18497956e-01 -5.78624129e-01 -2.03225538e-01 -1.79186407e-02 -1.51192844e+00 1.71266600e-01 5.11370778e-01 -3.11879486e-01 -3.61968189e-01 1.37985408e-01 6.12959504e-01 3.47469032e-01 -9.16291118e-01 4.66616273e-01 3.36455822e-01 -1.05428696e+00 6.90194070e-01 -3.14824134e-01 2.86377698e-01 -2.84672916e-01 -1.33590892e-01 -1.13956332e+00 -5.46216011e-01 -8.46973211e-02 -2.70374328e-01 1.42326379e+00 1.56590849e-01 -9.39570606e-01 6.34930491e-01 3.50029208e-02 -1.13998763e-01 -6.33039236e-01 -9.07087445e-01 -8.54720831e-01 6.96532279e-02 -2.81107366e-01 3.25434715e-01 1.12328541e+00 1.30115092e-01 1.93674028e-01 -3.54908057e-03 -3.91862132e-02 4.68012780e-01 -1.08049743e-01 6.50661409e-01 -1.51208138e+00 -2.90471524e-01 -5.95492899e-01 -6.92802250e-01 -3.55140865e-02 2.15091527e-01 -9.93161500e-01 -1.42538249e-02 -9.35724497e-01 3.10779512e-01 -1.73294932e-01 -6.06817529e-02 2.11430073e-01 -1.06390864e-01 1.48118883e-01 8.30100328e-02 5.95491886e-01 -2.63122946e-01 1.39674097e-01 2.87567675e-01 -2.64955133e-01 -3.55662167e-01 1.48421392e-01 -2.77146161e-01 9.57162023e-01 1.36082399e+00 -2.64455676e-01 -1.84891537e-01 1.89984322e-01 -3.27407211e-01 -2.12756395e-01 -1.10219479e-01 -1.21916330e+00 -1.06717512e-01 -7.78154880e-02 3.30271661e-01 -9.46578324e-01 1.86093748e-01 -9.66411769e-01 4.58971113e-01 1.09614539e+00 3.96624982e-01 1.49091676e-01 1.81213483e-01 2.44583026e-01 -1.44003361e-01 -4.44035709e-01 8.77673864e-01 1.24218963e-01 -5.97646296e-01 1.36679709e-01 -1.26817954e+00 -4.41881269e-01 1.51388037e+00 -4.89657462e-01 -3.14463377e-01 5.88563010e-02 -6.49796844e-01 -8.51869136e-02 7.20470965e-01 2.90735036e-01 6.91951931e-01 -1.04375172e+00 -5.68295062e-01 4.79711235e-01 1.48666903e-01 -4.37133908e-01 1.26334652e-02 5.62398493e-01 -7.67157912e-01 4.90703344e-01 -3.65211487e-01 -9.93517160e-01 -1.53975964e+00 1.05046558e+00 -1.20417327e-01 -2.56623089e-01 -1.50726527e-01 3.61172885e-01 -5.71289472e-03 -6.31962180e-01 2.66119808e-01 -1.63940534e-01 -6.64962292e-01 4.16184366e-01 6.41731560e-01 5.40845990e-01 1.06631853e-01 -8.98529887e-01 -5.11307120e-01 2.25543812e-01 -9.93949845e-02 4.96204257e-01 1.22418070e+00 4.92799103e-01 -6.64677382e-01 7.36040771e-01 1.58906174e+00 -1.73242539e-01 -7.12307990e-01 2.44186163e-01 3.00226778e-01 -4.22117919e-01 -7.43279278e-01 -1.48112938e-01 -8.78309429e-01 8.19215655e-01 6.26821518e-01 9.48481858e-01 1.22909701e+00 -3.72301042e-02 6.29259944e-01 5.10762930e-01 4.52022374e-01 -4.48804915e-01 5.32715991e-02 8.47604871e-01 4.23885822e-01 -1.10273027e+00 -3.85961443e-01 -2.43988633e-01 -1.49847984e-01 1.33824432e+00 3.99675339e-01 -3.83097649e-01 6.33422792e-01 4.97371942e-01 -3.05746704e-01 3.22233438e-02 -8.56755853e-01 3.21321696e-01 -7.92498514e-02 8.86685848e-01 1.51329637e-01 3.82188410e-01 -1.65963233e-01 3.39518815e-01 -3.73770177e-01 -5.04611194e-01 6.74805403e-01 8.93920660e-01 -9.20902967e-01 -8.18380117e-01 -8.39425027e-01 1.08738232e+00 -2.04375044e-01 1.88655794e-01 -9.97833133e-01 4.17722970e-01 4.16138709e-01 1.16920257e+00 5.66852510e-01 -1.17125177e+00 -1.29016668e-01 9.43105444e-02 -9.21451077e-02 -5.58881760e-01 -6.97500587e-01 -2.54361611e-02 -9.63694230e-02 1.37471288e-01 5.26594631e-02 -6.35192454e-01 -1.18169487e+00 -5.29671311e-01 -2.44643867e-01 2.79202223e-01 7.36708403e-01 4.58924353e-01 1.37226388e-01 5.65167427e-01 1.26406348e+00 -1.19504142e+00 -3.07833523e-01 -8.89540970e-01 -2.61302233e-01 2.52820015e-01 2.77116746e-01 -5.01533866e-01 -7.15613723e-01 2.32653052e-01]
[14.414718627929688, 9.639486312866211]
f700fd19-6b44-466d-8692-45940dd0c6b6
unsupervised-dependency-parsing-with-acoustic
null
null
https://aclanthology.org/Q13-1006
https://aclanthology.org/Q13-1006.pdf
Unsupervised Dependency Parsing with Acoustic Cues
Unsupervised parsing is a difficult task that infants readily perform. Progress has been made on this task using text-based models, but few computational approaches have considered how infants might benefit from acoustic cues. This paper explores the hypothesis that word duration can help with learning syntax. We describe how duration information can be incorporated into an unsupervised Bayesian dependency parser whose only other source of information is the words themselves (without punctuation or parts of speech). Our results, evaluated on both adult-directed and child-directed utterances, show that using word duration can improve parse quality relative to words-only baselines. These results support the idea that acoustic cues provide useful evidence about syntactic structure for language-learning infants, and motivate the use of word duration cues in NLP tasks with speech.
['Sharon Goldwater', 'John K Pate']
2013-01-01
null
null
null
tacl-2013-1
['unsupervised-dependency-parsing']
['natural-language-processing']
[ 2.31084719e-01 4.45302993e-01 -1.88024983e-01 -1.04029989e+00 -8.07920277e-01 -6.15459740e-01 2.05527961e-01 7.08395600e-01 -9.87469256e-01 2.58872300e-01 7.38811731e-01 -5.54387808e-01 3.91715169e-01 -5.13462603e-01 -6.83137238e-01 -3.59926432e-01 -2.16704145e-01 2.89258331e-01 3.82274806e-01 1.60869136e-01 6.73920512e-02 1.22104630e-01 -1.31785679e+00 2.63197005e-01 5.07880747e-01 9.58220214e-02 6.31721854e-01 8.93851399e-01 -4.00570720e-01 5.35372794e-01 -6.83720469e-01 -4.00746822e-01 -3.86124969e-01 -3.20159614e-01 -6.37981713e-01 -3.51130426e-01 3.40397507e-01 -7.11745799e-01 5.48455007e-02 5.62567115e-01 5.28849542e-01 2.83528324e-02 5.75040221e-01 -2.32630357e-01 -5.10396540e-01 1.34906507e+00 -9.17143077e-02 8.21465015e-01 4.08986837e-01 -2.47059748e-01 1.35762918e+00 -7.36191511e-01 4.62286144e-01 1.45165837e+00 4.86127287e-01 8.88846815e-01 -1.35338080e+00 -6.16641283e-01 5.24813354e-01 -2.64787257e-01 -5.85011065e-01 -9.70767200e-01 8.58064711e-01 -5.60032427e-01 1.64405954e+00 -1.15532465e-01 6.11371994e-01 1.16317654e+00 -2.22050305e-02 1.05128503e+00 1.02756286e+00 -9.14170682e-01 9.44368243e-02 -1.27782017e-01 6.28083825e-01 5.97398698e-01 2.89398581e-01 5.28200686e-01 -9.00427639e-01 3.73314917e-01 2.74372727e-01 -6.86614990e-01 1.03273109e-01 2.65825123e-01 -7.85787642e-01 9.46760833e-01 -3.41813743e-01 6.13048673e-01 3.10859866e-02 4.69872952e-01 5.54488540e-01 6.38688207e-02 5.35299957e-01 3.29026520e-01 -9.38163519e-01 -7.68559337e-01 -6.71105444e-01 -8.22122172e-02 7.12989092e-01 1.03079438e+00 4.84590977e-01 1.46344543e-01 1.48465395e-01 1.36398065e+00 8.78301382e-01 6.57680213e-01 4.61072654e-01 -7.52906084e-01 6.94071770e-01 -2.29294807e-01 -3.97849441e-01 -2.95820534e-01 -5.61229289e-01 2.29769722e-01 4.18020666e-01 -1.98427945e-01 6.67656302e-01 -5.02724946e-01 -8.88447404e-01 2.34874558e+00 2.84342229e-01 -3.66822869e-01 9.31468681e-02 1.43467665e-01 1.12452137e+00 5.51622272e-01 7.18156576e-01 -4.76395845e-01 1.40015471e+00 -5.15222669e-01 -9.14235771e-01 -6.97261691e-01 9.07535493e-01 -7.59076595e-01 8.84666800e-01 3.48165035e-01 -1.29021406e+00 -3.98998499e-01 -1.02228427e+00 -1.67865574e-01 -2.05190763e-01 -4.50899720e-01 1.07043087e+00 1.34357941e+00 -1.01392472e+00 6.29008412e-01 -1.40921021e+00 -3.95374566e-01 1.93134889e-01 2.92685568e-01 -3.09001952e-01 5.05239144e-02 -9.45196033e-01 9.78733957e-01 3.62944365e-01 -2.67396748e-01 -6.51926577e-01 -3.22653681e-01 -1.34246552e+00 -2.54889965e-01 -3.21060568e-01 7.23786354e-02 2.02447939e+00 -7.70585001e-01 -1.47163725e+00 9.43035007e-01 -3.94676149e-01 -8.81652310e-02 -6.41830802e-01 -5.82026124e-01 -1.52637035e-01 3.99249107e-01 1.13245443e-01 9.06906843e-01 4.59481359e-01 -7.92744219e-01 -6.09387398e-01 -4.06837225e-01 -2.98940510e-01 2.03209206e-01 -3.09381485e-01 9.02683198e-01 -1.66397631e-01 -8.72967899e-01 2.42339492e-01 -7.02417672e-01 1.47916138e-01 -3.56052190e-01 9.37347040e-02 -8.81668210e-01 -5.39624598e-03 -9.50287461e-01 1.15957141e+00 -2.00789022e+00 -3.80088568e-01 -3.01365763e-01 -4.27552402e-01 -9.07942504e-02 -1.55934617e-01 4.19143826e-01 -6.95987744e-03 4.44301486e-01 -2.04395369e-01 -5.64488947e-01 6.66320464e-03 7.44056582e-01 5.68605103e-02 2.19591081e-01 5.90649784e-01 9.01662111e-01 -9.63893175e-01 -5.75825572e-01 4.18294687e-03 3.77313197e-01 -6.07866168e-01 3.90095741e-01 -2.18162790e-01 3.06888998e-01 -2.48675928e-01 4.00130451e-01 1.37065068e-01 6.03838623e-01 4.19548243e-01 7.22949624e-01 -4.17646289e-01 1.53897309e+00 -4.65303987e-01 1.60468006e+00 -4.07374412e-01 8.09324622e-01 1.71085313e-01 -8.53261650e-01 5.22570908e-01 4.70368654e-01 -1.69905245e-01 -6.15850091e-01 2.26193041e-01 1.14918098e-01 7.86197484e-01 -7.51835287e-01 -1.15991086e-01 -4.97532994e-01 -3.87437642e-02 4.61215705e-01 2.89885312e-01 -5.03101468e-01 2.02932388e-01 -1.35261029e-01 1.18269432e+00 4.13869053e-01 6.51766509e-02 -4.29013073e-01 -2.65641332e-01 -2.89270729e-01 6.93485796e-01 7.50187993e-01 -1.15338430e-01 5.11970937e-01 1.98331237e-01 1.03904111e-02 -6.35544002e-01 -1.20531225e+00 -5.31060636e-01 1.97505784e+00 -6.33153141e-01 -3.27349156e-01 -8.64230573e-01 -5.91642439e-01 -2.77705878e-01 1.20431614e+00 -4.73666281e-01 3.15970421e-01 -1.12060094e+00 -5.52296877e-01 6.18081808e-01 1.05060375e+00 -5.69741905e-01 -1.51978052e+00 -9.37488794e-01 8.17407608e-01 1.21047482e-01 -1.11107028e+00 -4.87266541e-01 8.28628242e-01 -1.14973319e+00 -7.00953841e-01 -3.84138674e-01 -1.21768713e+00 3.86243254e-01 -1.06850425e-02 1.04109573e+00 1.79036751e-01 -4.02754173e-02 7.89751589e-01 -8.11901510e-01 -9.15119410e-01 -7.00575829e-01 -2.19632998e-01 4.50173467e-02 -9.27606225e-01 8.59900892e-01 -9.13393497e-01 -3.08960646e-01 -2.98361093e-01 -4.65865403e-01 -4.20454778e-02 4.82723206e-01 6.63640738e-01 1.89941388e-03 -5.14722943e-01 7.75108874e-01 -8.58972371e-01 2.37277687e-01 -2.71818370e-01 -5.16920149e-01 -5.78541048e-02 -3.42164665e-01 3.69920731e-01 2.63994068e-01 -6.22284651e-01 -1.20490384e+00 1.47763416e-01 -8.25375199e-01 4.98240471e-01 -7.56563365e-01 4.74195033e-01 -2.16857761e-01 3.75113487e-01 4.83863771e-01 -2.08739057e-01 -1.73142433e-01 -7.35743999e-01 3.29905242e-01 4.30463135e-01 4.29628581e-01 -1.20421541e+00 4.44095135e-01 -3.07487160e-01 -5.18488705e-01 -1.25064600e+00 -9.34012949e-01 -4.29509819e-01 -8.16810608e-01 1.26140401e-01 1.13454545e+00 -8.41893733e-01 -2.62139648e-01 2.58820504e-01 -1.56818628e+00 -7.45456457e-01 1.03244178e-01 9.79022026e-01 -4.15571600e-01 4.02130574e-01 -9.17138100e-01 -1.03007233e+00 -8.57582837e-02 -8.54829133e-01 6.47681773e-01 2.76759923e-01 -6.18476331e-01 -1.13864517e+00 4.61617559e-01 3.08283925e-01 5.56461513e-02 -1.41419381e-01 1.29220200e+00 -1.18330646e+00 -1.59868658e-01 2.21487224e-01 2.55023897e-01 5.03066361e-01 8.90868902e-02 1.90186247e-01 -1.21324587e+00 2.55779270e-02 3.94932479e-02 -5.06296396e-01 7.66902506e-01 6.75816715e-01 5.16884029e-01 -1.46492422e-01 -1.70004889e-01 2.32648671e-01 9.38745141e-01 5.52603900e-01 -1.51882544e-01 -2.22623914e-01 4.02890325e-01 1.29916692e+00 3.90945584e-01 1.21876532e-02 8.32459390e-01 1.34732768e-01 -1.29986480e-01 4.70892727e-01 -3.81811112e-01 -2.77761161e-01 6.24913692e-01 1.46670842e+00 2.02645898e-01 -2.38987342e-01 -1.17036271e+00 1.19083965e+00 -1.16868949e+00 -5.96635282e-01 -2.12915435e-01 1.95291352e+00 1.29176760e+00 5.20684123e-01 -1.08042837e-03 1.68300405e-01 5.10332167e-01 2.30925947e-01 3.20090987e-02 -1.19198751e+00 1.87388316e-01 8.65633249e-01 5.50560988e-02 7.74157643e-01 -8.99638116e-01 1.11443651e+00 7.27940798e+00 2.29432788e-02 -6.47606015e-01 9.36482772e-02 4.25277412e-01 2.41158098e-01 -5.46372712e-01 -1.24475144e-01 -1.14512873e+00 2.68350065e-01 1.44609261e+00 2.69992769e-01 2.53013194e-01 7.47261882e-01 1.48588687e-01 -6.35646820e-01 -1.85749352e+00 3.71500492e-01 1.14651144e-01 -4.13605094e-01 -7.63567388e-01 -3.94988805e-01 2.12616950e-01 3.55358094e-01 -2.33961269e-01 4.05209512e-01 4.25056010e-01 -1.03680444e+00 8.24443579e-01 -1.46498397e-01 5.97530603e-01 -4.77431744e-01 3.34737062e-01 4.78414625e-01 -1.15467179e+00 3.30035836e-01 -2.22985342e-01 -4.99653876e-01 1.77516058e-01 1.86852202e-01 -1.16952717e+00 -4.73609775e-01 5.78106940e-01 2.82447129e-01 -5.38249731e-01 7.99600482e-01 -9.46025431e-01 1.65704346e+00 -6.89836085e-01 -5.98854899e-01 -1.03830531e-01 3.60621333e-01 4.07266349e-01 1.93582964e+00 1.13845132e-02 6.70383096e-01 -1.93869337e-01 2.85688251e-01 2.94402689e-01 5.58917165e-01 -7.61443317e-01 -3.67297649e-01 7.17082262e-01 6.33387387e-01 -1.00488007e+00 6.76187426e-02 -9.96145904e-01 3.79256785e-01 6.84224606e-01 -5.20860739e-02 -2.31573418e-01 -2.61857510e-01 3.53624821e-01 -3.13668787e-01 5.60381413e-01 -7.80855536e-01 -4.66644526e-01 -6.48416340e-01 -1.34527450e-03 -4.85747218e-01 2.29117364e-01 -4.27475184e-01 -1.05719864e+00 1.65557340e-01 4.09780502e-01 2.62718480e-02 -5.93771875e-01 -9.65878189e-01 -7.98942983e-01 7.82164395e-01 -1.40918958e+00 -7.72869945e-01 4.89138752e-01 -1.78452283e-01 1.04122567e+00 2.50694633e-01 1.08127081e+00 -2.08011180e-01 -3.04726094e-01 7.85383344e-01 -3.53009582e-01 4.59814698e-01 5.05596519e-01 -1.62151349e+00 8.43799829e-01 9.79061306e-01 6.22501850e-01 8.83858085e-01 7.54256606e-01 -7.99656212e-01 -1.08669543e+00 -4.18156296e-01 1.24976015e+00 -9.23265815e-01 4.94750887e-01 -5.90566993e-01 -9.28505480e-01 6.30821884e-01 4.75757390e-01 -3.18465292e-01 1.23598325e+00 7.66903341e-01 -4.80313003e-01 2.38066301e-01 -8.14130545e-01 2.82419443e-01 1.31054282e+00 -6.12257540e-01 -1.55660844e+00 5.94041534e-02 1.19104111e+00 -1.72340497e-01 -6.55705512e-01 5.16015053e-01 6.49073064e-01 -7.29120255e-01 5.61552703e-01 -4.68942940e-01 3.23307335e-01 4.04822588e-01 -2.41788700e-01 -1.34963679e+00 -7.18722716e-02 -4.93476927e-01 5.24083860e-02 1.67801082e+00 8.11843514e-01 -2.36876383e-01 4.91775900e-01 9.54181492e-01 -5.98159730e-01 -5.61631501e-01 -9.03529227e-01 -6.02534950e-01 6.23200715e-01 -1.17367923e+00 2.40296740e-02 5.18927872e-01 2.44601518e-01 6.06007934e-01 4.57412809e-01 2.03140631e-01 3.27489793e-01 -4.44320381e-01 2.31689643e-02 -1.24419916e+00 -3.80049288e-01 -3.32552254e-01 -9.25734267e-02 -1.01072013e+00 4.38505888e-01 -6.58187926e-01 8.57395709e-01 -1.43729484e+00 -1.35482118e-01 -4.85000670e-01 -5.42655215e-02 6.91869020e-01 -4.03808057e-01 -3.23501468e-01 1.46156043e-01 -6.23205721e-01 -2.15261891e-01 1.60033271e-01 5.13362229e-01 3.99739176e-01 -2.43641168e-01 1.34164944e-01 -6.39499605e-01 1.14559400e+00 7.36641645e-01 -8.48012209e-01 -4.05039757e-01 -8.71978164e-01 1.42286345e-01 3.22638839e-01 -5.40057182e-01 -6.63784802e-01 9.52938199e-02 -2.80968606e-01 2.39010453e-01 -6.11311197e-01 1.62007034e-01 -4.18827087e-01 -7.64751911e-01 1.97099298e-01 -3.95153522e-01 2.07794607e-01 5.18429577e-01 1.49415687e-01 1.67674974e-01 -9.95061934e-01 6.64633393e-01 1.09746633e-02 -4.10619080e-01 -2.50662833e-01 -1.01619101e+00 5.73207319e-01 3.95587355e-01 8.03602561e-02 5.86946011e-02 -6.70047626e-02 -8.91618907e-01 1.09560221e-01 1.60862729e-01 3.91901225e-01 5.84466577e-01 -8.74054372e-01 -6.16088629e-01 2.14458004e-01 6.42974302e-02 1.31894112e-01 -4.41871285e-01 3.52042913e-01 -4.08608764e-01 4.83392447e-01 3.18037361e-01 -4.89146560e-01 -1.57697678e+00 5.11559486e-01 -2.05939427e-01 3.81437331e-01 -4.37064976e-01 1.69836879e+00 2.99469501e-01 -1.98423937e-01 8.19723010e-01 -7.58400500e-01 -3.78296524e-01 2.30343431e-01 7.39759505e-01 2.64223404e-02 -2.66192198e-01 -4.51367527e-01 -5.17067969e-01 4.51427966e-01 -2.68534154e-01 -6.12273693e-01 1.45653999e+00 -1.43457249e-01 8.70376676e-02 1.04419386e+00 1.08044052e+00 5.04982233e-01 -1.02700198e+00 -1.11650981e-01 7.06240535e-01 -7.22454786e-02 -1.62360653e-01 -7.29270816e-01 -3.55245978e-01 1.32916081e+00 1.79779157e-01 1.73733890e-01 7.27733850e-01 5.42710721e-01 5.98585665e-01 5.50905764e-01 1.45977244e-01 -1.18859327e+00 -9.45613906e-03 7.71600008e-01 6.29451334e-01 -1.31754506e+00 -2.57502466e-01 -4.23129171e-01 -3.34418327e-01 1.21691287e+00 7.08170593e-01 3.64039950e-02 6.82676733e-01 5.95061183e-01 1.46383613e-01 7.00745210e-02 -1.03099108e+00 -4.52465385e-01 2.61856586e-01 8.14386368e-01 1.26367068e+00 2.04062730e-01 -7.12485135e-01 9.51683402e-01 -5.41291177e-01 -8.32541585e-01 5.06474197e-01 1.15781033e+00 -9.38316762e-01 -1.35874295e+00 -1.75091967e-01 3.89162868e-01 -1.09110343e+00 -7.35050499e-01 -2.13037521e-01 7.66425848e-01 1.92565948e-01 1.39915001e+00 4.66775568e-03 1.17190972e-01 5.90169569e-03 6.36292696e-01 7.92652071e-01 -1.67473423e+00 -4.79228467e-01 3.99354666e-01 8.13646555e-01 -2.32792318e-01 -4.23150539e-01 -1.23763371e+00 -1.63262236e+00 5.53211689e-01 -5.20831585e-01 2.30011865e-01 9.26842391e-01 1.08441079e+00 -6.02612793e-01 2.66210884e-01 4.67548728e-01 -4.78335321e-01 -1.76350981e-01 -1.07148182e+00 -1.42251492e-01 -2.95320362e-01 5.86313426e-01 -4.35169399e-01 -5.37842214e-01 2.60875911e-01]
[10.568007469177246, 9.505419731140137]
e339de18-ed70-407c-81e6-88a9f81ee17e
argus-efficient-activity-detection-system-for
null
null
http://openaccess.thecvf.com/content_WACVW_2020/html/w5/Liu_Argus_Efficient_Activity_Detection_System_for_Extended_Video_Analysis_WACVW_2020_paper.html
http://openaccess.thecvf.com/content_WACVW_2020/papers/w5/Liu_Argus_Efficient_Activity_Detection_System_for_Extended_Video_Analysis_WACVW_2020_paper.pdf
Argus: Efficient Activity Detection System for Extended Video Analysis
We propose an Efficient Activity Detection System, Argus, for Extended Video Analysis in the surveillance scenario. For the spatial-temporal event detection in the surveillance video, we first generate video proposals by applying object detection and tracking algorithm which shared the detection features. After that, we extract several different features and apply sequential activity classification with them. Finally, we eliminate inaccurate events and fuse all the predictions from different features. The proposed system wins Trecvid Activities in Extended Video (ActEV) challenge 2019. It achieves the first place with 60.5 mean weighted Pmiss, out-performing the second place system by 14.5 and the baseline R-C3D by 29.0. In TRECVID 2019 Challenge, the proposed system wins the first place with pAUDC@ 0.2 tfa 0.48407
['Xiaojun Chang', 'Po-Yao Huang', 'Junwei Liang', 'Guoliang Kang', 'Wenhe Liu', 'Liangke Gui', 'Jing Wen', 'Yijun Qian', 'Peng Chen']
2020-03-02
null
null
null
proceedings-of-the-ieee-winter-conference-on
['video-object-tracking']
['computer-vision']
[ 3.00921679e-01 -1.13405399e-02 6.35621417e-03 -2.32474118e-01 -9.54426467e-01 -7.48586714e-01 8.64451885e-01 8.32948983e-02 -9.98151898e-01 7.23001599e-01 2.22629815e-01 8.76852348e-02 1.98938161e-01 -2.32396632e-01 -9.43272114e-01 -8.13785672e-01 -3.02994192e-01 -7.47823939e-02 9.33701992e-01 3.50110531e-01 6.38115257e-02 3.12630206e-01 -1.47132421e+00 5.78241348e-01 4.28907961e-01 1.36053038e+00 -1.62894860e-01 1.12275755e+00 6.48593783e-01 1.11456954e+00 -6.87132359e-01 -5.38553596e-01 3.22050780e-01 -1.78453788e-01 -4.49955314e-01 1.00094154e-02 6.60450339e-01 -6.27820909e-01 -4.70488876e-01 7.45986164e-01 3.96228522e-01 1.18174836e-01 8.09042275e-01 -1.32725132e+00 1.20782472e-01 -2.88050040e-03 -9.43631232e-01 7.44078577e-01 7.76973307e-01 2.16832101e-01 5.45079052e-01 -1.10116899e+00 6.20744169e-01 8.33449781e-01 5.75125694e-01 4.45084155e-01 -7.06391215e-01 -6.88714802e-01 6.30610287e-02 3.00618172e-01 -1.53105724e+00 -2.15997905e-01 1.33993737e-02 -6.09695315e-01 1.06761849e+00 3.50945711e-01 8.47876847e-01 1.50567603e+00 4.37575042e-01 1.12149262e+00 7.00192630e-01 4.57269624e-02 1.50697008e-01 -1.60805449e-01 3.08176249e-01 7.36528397e-01 4.07461643e-01 1.88074157e-01 -5.20512283e-01 -3.22471201e-01 1.77451134e-01 3.00351083e-01 -1.20555438e-01 2.44082779e-01 -1.37792552e+00 4.37342554e-01 -1.75650209e-01 -8.25783759e-02 -7.30368614e-01 -9.21996161e-02 5.27657926e-01 1.35216475e-01 3.03970098e-01 1.10409692e-01 -5.17422199e-01 -3.15367162e-01 -1.11826456e+00 5.10733664e-01 4.74761069e-01 1.06877589e+00 1.45177752e-01 -2.24157855e-01 -8.39720309e-01 1.88902289e-01 1.25302732e-01 8.69672894e-01 2.77998805e-01 -1.08628094e+00 3.21072876e-01 4.20252711e-01 5.20944357e-01 -7.72771657e-01 -3.94281775e-01 -1.79248676e-01 -7.36848354e-01 -9.48809739e-03 4.23369586e-01 -5.30675292e-01 -1.02372897e+00 1.43631053e+00 4.16184992e-01 6.16859138e-01 1.13120347e-01 6.60034120e-01 5.95387220e-01 9.81143534e-01 5.08379519e-01 -6.27551913e-01 1.75106490e+00 -1.03082108e+00 -6.74106777e-01 -9.98816118e-02 4.58720446e-01 -6.51372492e-01 2.48200759e-01 7.26699591e-01 -9.57827926e-01 -6.33251786e-01 -1.01061094e+00 5.09092808e-01 -3.42464522e-02 2.93659657e-01 4.52814847e-02 4.95010883e-01 -7.72466302e-01 2.16824651e-01 -9.84722614e-01 -4.51168656e-01 3.68425757e-01 1.69313848e-01 -5.18306315e-01 -4.56887819e-02 -1.04843962e+00 8.19212258e-01 6.65436089e-01 -3.43324542e-01 -1.41829193e+00 -5.43983281e-01 -6.69653177e-01 -8.99439305e-02 7.80451596e-01 -4.65608925e-01 1.20176911e+00 -6.35203719e-01 -1.04035878e+00 6.95875466e-01 -3.10832292e-01 -8.86681318e-01 8.14221740e-01 -4.84312475e-01 -6.41384959e-01 2.75504112e-01 1.94841967e-04 6.48716748e-01 7.96666205e-01 -5.30090809e-01 -1.27389979e+00 -2.04096451e-01 -1.13365889e-01 2.38482058e-01 8.95285755e-02 4.84738618e-01 -8.50000739e-01 -6.56723678e-01 -3.33670676e-01 -1.07591271e+00 -2.55753785e-01 -1.91397503e-01 3.00041656e-03 -3.59001935e-01 7.60389447e-01 -9.20828044e-01 1.25522542e+00 -2.29092193e+00 -3.10464323e-01 1.34407744e-01 2.90821433e-01 5.11052728e-01 -1.01030461e-01 7.26033822e-02 1.54734448e-01 -3.08934867e-01 2.23882794e-01 -2.04428956e-01 -4.09299463e-01 -1.96212992e-01 7.20281061e-03 5.09579301e-01 2.37000927e-01 6.47868991e-01 -9.85621035e-01 -7.47059047e-01 2.61842489e-01 3.71194899e-01 -4.72047955e-01 3.39673162e-01 1.89198612e-03 2.50016421e-01 -4.71113414e-01 5.05768836e-01 6.71381950e-01 -1.30805254e-01 1.82540894e-01 -1.53972894e-01 -2.45547146e-01 -3.08697999e-01 -1.16243207e+00 1.44847488e+00 3.41328740e-01 6.41997457e-01 -4.40480530e-01 -6.38316453e-01 5.02882421e-01 6.22140706e-01 9.70235109e-01 -4.14920300e-01 1.48505688e-01 -3.05193961e-01 -2.58367151e-01 -6.93195701e-01 5.42039216e-01 6.97084963e-01 -3.17179590e-01 1.58623576e-01 3.59388381e-01 7.10793555e-01 4.78903383e-01 4.87410337e-01 1.57282031e+00 2.70111591e-01 7.74265945e-01 -5.45531437e-02 5.22563517e-01 -3.46696675e-02 7.46558666e-01 1.12516475e+00 -5.91537952e-01 5.47469497e-01 3.81065637e-01 -8.10504317e-01 -9.05839443e-01 -1.17367029e+00 1.36188343e-01 1.21838987e+00 -2.03293208e-02 -7.82840490e-01 -9.70196009e-01 -1.06364036e+00 -4.87872452e-01 3.49483132e-01 -7.97657311e-01 -6.87037706e-02 -5.00172317e-01 -8.48347723e-01 8.63602161e-01 5.67388117e-01 6.05961323e-01 -9.26711738e-01 -1.03391063e+00 2.10643366e-01 -5.09051442e-01 -1.41409290e+00 -5.96234381e-01 -1.74227238e-01 -4.03494477e-01 -1.42979193e+00 -9.78409529e-01 -4.27632600e-01 3.84978265e-01 3.48459244e-01 8.39904249e-01 -3.74615669e-01 -3.03885967e-01 3.05029839e-01 -4.49882150e-01 -5.86177945e-01 -4.80302572e-02 -1.50197148e-01 2.29772687e-01 1.44870549e-01 8.84596884e-01 1.58264473e-01 -7.72900403e-01 4.99322206e-01 -6.49278820e-01 -9.21751335e-02 5.24869025e-01 5.98634362e-01 6.19211912e-01 -2.35794276e-01 1.04174510e-01 -2.62954205e-01 9.81563553e-02 -5.32571316e-01 -9.53717411e-01 5.56664526e-01 -2.73690313e-01 -2.43981838e-01 7.80870244e-02 -6.32568657e-01 -1.15293002e+00 5.98062217e-01 1.54308707e-01 -4.09947425e-01 -3.26358050e-01 2.56421179e-01 -4.06805193e-04 3.12292337e-01 8.65355551e-01 2.79778600e-01 -4.99609023e-01 -1.34058714e-01 -1.23976581e-01 4.69113261e-01 6.63925409e-01 7.40093216e-02 4.48286265e-01 3.25181842e-01 -2.33048514e-01 -9.04806495e-01 -9.38159883e-01 -6.16413534e-01 -2.73293495e-01 -4.37873214e-01 1.33229458e+00 -1.30370986e+00 -8.83310676e-01 6.49556696e-01 -1.28383327e+00 2.19927728e-01 -1.04712039e-01 8.87397110e-01 -3.65624160e-01 4.90596831e-01 -3.53564590e-01 -1.04307938e+00 -4.41382229e-01 -9.71125185e-01 1.11611962e+00 2.32084766e-01 -3.65032971e-01 -1.21095195e-01 4.02009904e-01 4.01512742e-01 7.68176792e-03 5.38915396e-01 -3.51605564e-01 -8.52359653e-01 -4.91197705e-01 -6.22462273e-01 -1.11618333e-01 2.77455688e-01 -7.15000629e-02 9.27102044e-02 -1.08145821e+00 -2.84550577e-01 -2.11015776e-01 -1.13065159e-02 1.05857420e+00 7.00991213e-01 1.02681208e+00 -2.85236955e-01 -4.86745059e-01 3.41979951e-01 9.30382431e-01 6.34532094e-01 7.01399267e-01 -1.22024929e-02 2.85578966e-01 3.24656427e-01 1.00954199e+00 8.52755308e-01 7.17951506e-02 7.44503319e-01 2.51123995e-01 1.03962637e-01 2.85820723e-01 -1.45365700e-01 7.03830421e-01 1.25814527e-01 -5.94213665e-01 -4.49991733e-01 -7.68630624e-01 3.56182545e-01 -2.20527911e+00 -1.55513024e+00 -7.80811682e-02 2.11253238e+00 3.26100439e-01 3.78301948e-01 4.44583386e-01 -8.83689448e-02 7.10486054e-01 2.08619401e-01 -1.60430625e-01 2.14870811e-01 -8.89355242e-02 -1.02247991e-01 8.54825795e-01 1.74927175e-01 -1.71435571e+00 8.76709461e-01 6.50388908e+00 8.53121221e-01 -5.29489994e-01 2.73511589e-01 3.52788270e-01 -4.54668462e-01 7.85150945e-01 -3.53577346e-01 -1.01950991e+00 7.34355509e-01 1.42132545e+00 -2.20310390e-01 6.26403242e-02 1.01181591e+00 3.08210939e-01 -4.35956627e-01 -8.70790541e-01 1.13059855e+00 2.65722781e-01 -1.20911205e+00 -4.05041128e-02 3.71614993e-02 7.73639977e-01 2.03163862e-01 -1.90122679e-01 4.15222883e-01 3.26719940e-01 -6.20414734e-01 6.92706645e-01 6.61123395e-01 6.75245762e-01 -5.83579838e-01 1.00016451e+00 4.37781692e-01 -1.37952042e+00 -8.36501718e-02 -1.35170311e-01 1.56451374e-01 3.49583507e-01 2.54466683e-01 -9.28407133e-01 4.20225441e-01 1.00955594e+00 5.20232797e-01 -5.20738661e-01 1.30889165e+00 8.19182023e-02 6.85592532e-01 -5.13572872e-01 -1.41393300e-03 2.23427534e-01 3.69868010e-01 4.68483239e-01 1.54362309e+00 4.18069363e-01 2.94745445e-01 4.04905438e-01 -2.13830590e-01 -7.07889199e-02 -2.63505608e-01 -5.26994169e-01 3.49202096e-01 2.68350184e-01 1.18937838e+00 -5.16502857e-01 -8.86628270e-01 -5.31180859e-01 1.18569982e+00 -2.38050163e-01 2.47705579e-01 -1.53956580e+00 -1.56784758e-01 3.73745322e-01 -3.80745716e-02 5.03308594e-01 1.36955604e-01 6.18707955e-01 -1.32480967e+00 1.51902050e-01 -7.50281215e-01 8.67230773e-01 -6.55288577e-01 -8.22083354e-01 7.49323308e-01 4.17056769e-01 -1.46067882e+00 -4.12953377e-01 -5.89845479e-01 -3.73638779e-01 3.22227329e-01 -7.67828286e-01 -9.98430669e-01 -6.15807950e-01 8.54424775e-01 8.11901033e-01 -2.41240293e-01 5.04743338e-01 4.26359653e-01 -5.56165874e-01 5.75573504e-01 -9.62792784e-02 2.83460826e-01 8.14101994e-01 -8.50172162e-01 5.50707221e-01 1.13045502e+00 8.75313506e-02 1.06185764e-01 6.51920259e-01 -8.79911900e-01 -9.19365764e-01 -1.40780747e+00 1.03125620e+00 -7.82653451e-01 4.37750190e-01 -1.80938825e-01 -5.84467649e-01 7.94236302e-01 1.90320894e-01 2.71613032e-01 6.78483963e-01 -3.94256443e-01 -3.19907874e-01 4.60679233e-02 -1.22953176e+00 3.40177059e-01 1.03855538e+00 -2.15285182e-01 -8.22143912e-01 2.98389494e-01 5.03068089e-01 -3.44634235e-01 -6.41146243e-01 5.47237217e-01 9.70813096e-01 -6.92350686e-01 8.27989459e-01 -9.24907267e-01 4.16012481e-02 -5.08225501e-01 -2.09818542e-01 -7.12654054e-01 -5.25259793e-01 -6.08809114e-01 -5.13416409e-01 6.74009681e-01 4.37227279e-01 -3.39003913e-02 7.58402050e-01 3.48045498e-01 1.30770296e-01 -2.17645347e-01 -1.03803241e+00 -8.12925458e-01 -7.19830036e-01 -5.59640646e-01 2.10419729e-01 5.35447598e-01 -1.56445339e-01 6.27515167e-02 -1.06700063e+00 4.60292399e-01 7.05733240e-01 -5.27013421e-01 7.12484300e-01 -1.20065141e+00 -2.53457397e-01 2.04393864e-01 -5.35210311e-01 -1.03582060e+00 -3.95449340e-01 -5.91581017e-02 -4.98113520e-02 -9.85534012e-01 5.97274601e-01 6.82948411e-01 -8.45885336e-01 3.27020198e-01 -3.42803091e-01 3.66520196e-01 1.89776793e-01 5.58205992e-02 -1.45532477e+00 9.28845704e-02 4.59965616e-01 -9.10720825e-02 -6.71572536e-02 1.57405362e-01 -1.51781783e-01 9.59251106e-01 8.12903345e-01 -7.27807581e-01 -1.67971089e-01 -1.43232703e-01 -2.32502967e-01 2.79084414e-01 4.44020540e-01 -1.45511353e+00 1.95451304e-01 -2.55930871e-01 8.06735933e-01 -1.01413381e+00 4.17039663e-01 -9.48218107e-01 2.83216596e-01 7.83314049e-01 -2.29788795e-01 2.64191031e-01 8.29932019e-02 9.67438936e-01 5.02144098e-02 6.59769624e-02 6.09103560e-01 -7.34418333e-02 -8.24175298e-01 5.15637040e-01 -1.00848031e+00 -5.42783216e-02 1.70023298e+00 -3.33969474e-01 -2.60698169e-01 -5.49627207e-02 -9.62824404e-01 2.98280686e-01 1.19266182e-01 5.52459598e-01 5.86278677e-01 -1.12278795e+00 -9.05608237e-01 3.19849811e-02 4.03494537e-01 -5.05961597e-01 5.04752815e-01 9.35734272e-01 -3.99815500e-01 3.30431461e-01 -2.70895779e-01 -6.24289930e-01 -1.79470646e+00 6.49967372e-01 6.85578436e-02 -4.52418596e-01 -4.14731503e-01 5.53683579e-01 2.03719690e-01 5.46336532e-01 3.83468717e-01 -2.34197229e-01 -4.10678536e-01 1.15368158e-01 1.22411966e+00 6.27367198e-01 -6.49340153e-02 -6.40445769e-01 -7.38642097e-01 2.71767437e-01 -2.95648307e-01 -3.07035655e-01 1.05553889e+00 1.70893505e-01 7.34181106e-01 -1.52408466e-01 7.97184646e-01 -1.32516205e-01 -1.66001987e+00 -1.42477453e-01 1.93903059e-01 -3.80878776e-01 -1.12923607e-01 -7.91133106e-01 -7.63044000e-01 3.67777109e-01 9.59838033e-01 -7.91860372e-03 1.15619636e+00 -5.19806072e-02 5.30918777e-01 5.73325872e-01 3.40020776e-01 -9.46460843e-01 8.15448612e-02 4.79678601e-01 6.43166840e-01 -1.34395850e+00 1.10905781e-01 -9.78045836e-02 -1.03730667e+00 6.59254611e-01 5.79927325e-01 -2.11934388e-01 4.61519659e-01 3.04080814e-01 -3.52247238e-01 -8.61452371e-02 -1.07016599e+00 4.39272746e-02 4.64007914e-01 6.12941444e-01 1.43346623e-01 -6.49179565e-03 -3.62978429e-01 5.21708488e-01 4.67084378e-01 3.32507879e-01 3.31033945e-01 1.02793860e+00 -5.71802497e-01 -3.55134636e-01 -3.32905859e-01 4.59112912e-01 -9.57605898e-01 1.30818322e-01 -2.83552766e-01 5.86655915e-01 2.82812893e-01 9.84798729e-01 2.05686361e-01 -6.32124305e-01 4.57358330e-01 -6.20515225e-03 2.87500620e-01 -8.10336918e-02 -6.07890427e-01 2.95051754e-01 3.33360404e-01 -9.63480175e-01 -6.33201301e-01 -8.15519810e-01 -7.30679274e-01 -2.08457410e-01 -1.64553985e-01 2.33105779e-01 1.98572129e-01 6.85839415e-01 4.82433259e-01 2.76423275e-01 4.35686082e-01 -5.40791094e-01 -2.83324420e-01 -9.79420006e-01 -2.58767426e-01 3.17379683e-01 3.89769346e-01 -6.29840136e-01 -1.65986612e-01 4.36795503e-01]
[8.3054838180542, 0.4329882264137268]
5e565518-ea66-4d23-b253-9b0055a2dceb
impact-of-acoustic-noise-on-alzheimer-s
2203.17110
null
https://arxiv.org/abs/2203.17110v2
https://arxiv.org/pdf/2203.17110v2.pdf
Impact of Environmental Noise on Alzheimer's Disease Detection from Speech: Should You Let a Baby Cry?
Research related to automatically detecting Alzheimer's disease (AD) is important, given the high prevalence of AD and the high cost of traditional methods. Since AD significantly affects the acoustics of spontaneous speech, speech processing and machine learning (ML) provide promising techniques for reliably detecting AD. However, speech audio may be affected by different types of background noise and it is important to understand how the noise influences the accuracy of ML models detecting AD from speech. In this paper, we study the effect of fifteen types of environmental noise from five different categories on the performance of four ML models trained with three types of acoustic representations. We perform a thorough analysis showing how ML models and acoustic features are affected by different types of acoustic noise. We show that acoustic noise is not necessarily harmful - certain types of noise are beneficial for AD detection models and help increasing accuracy by up to 4.8\%. We provide recommendations on how to utilize acoustic noise in order to achieve the best performance results with the ML models deployed in real world.
['Jekaterina Novikova']
2022-03-31
null
null
null
null
['alzheimer-s-disease-detection']
['medical']
[ 2.90675700e-01 -3.21864933e-01 2.41651878e-01 -4.44811642e-01 -1.16579723e+00 6.92794099e-02 2.54363716e-01 2.15895638e-01 -5.05717158e-01 4.07824129e-01 5.29091656e-01 -2.00191587e-01 1.02135979e-01 -6.26304150e-01 -3.55649322e-01 -4.29182023e-01 -4.20530707e-01 3.98380384e-02 3.71902376e-01 3.80743947e-03 -2.26759031e-01 4.68433470e-01 -1.80266345e+00 7.27146506e-01 6.67571545e-01 9.89133656e-01 6.18915379e-01 1.01585484e+00 -3.08105946e-01 5.87600589e-01 -1.30506194e+00 3.77815068e-02 -1.16923966e-01 -4.30278331e-01 -3.73722315e-01 -4.11144905e-02 3.48175585e-01 -4.84290957e-01 -2.29793996e-01 9.80358362e-01 8.88160110e-01 -1.45677954e-01 6.35666251e-01 -8.12728286e-01 -2.91156709e-01 4.05319124e-01 1.07278138e-01 8.29009831e-01 5.82479298e-01 1.59936950e-01 7.02135324e-01 -4.45369840e-01 1.69408426e-01 1.52790356e+00 7.71683514e-01 6.52459145e-01 -1.21176136e+00 -7.28894114e-01 1.74885631e-01 4.49338317e-01 -1.09450686e+00 -9.77424800e-01 7.44644165e-01 -4.04331982e-01 8.74744177e-01 3.26588720e-01 7.54430950e-01 1.41807199e+00 2.11879283e-01 3.92345339e-01 1.07285881e+00 -5.11887252e-01 3.55098426e-01 2.13908613e-01 3.86428863e-01 1.87946990e-01 1.81036234e-01 -2.89473608e-02 -4.22434151e-01 -7.18391657e-01 2.76810467e-01 -2.97868043e-01 -1.02279507e-01 3.29592139e-01 -8.57495904e-01 8.20755005e-01 1.55320987e-01 5.64796686e-01 -3.91029269e-01 2.38028958e-01 4.87207025e-01 1.80941880e-01 4.69966561e-01 3.08169991e-01 -2.59108067e-01 -2.02119872e-01 -6.45849645e-01 2.62586832e-01 6.07103765e-01 4.88448799e-01 -5.70307076e-02 3.30398828e-01 1.27407953e-01 1.68569756e+00 7.28581727e-01 8.85300815e-01 5.73391199e-01 -8.71395946e-01 2.40774795e-01 1.90855935e-01 -7.55863637e-02 -9.01266336e-01 -6.79388285e-01 -4.34307486e-01 -3.28575283e-01 1.23021737e-01 4.20475900e-01 -2.67317265e-01 -8.42085242e-01 1.45130217e+00 -1.45676747e-01 1.75497502e-01 -1.33926822e-02 5.37456274e-01 8.75695348e-01 6.36877894e-01 5.30248702e-01 -3.63039166e-01 1.59723365e+00 -1.61043599e-01 -1.14257348e+00 -7.98413098e-01 6.64534748e-01 -9.36964571e-01 1.08350909e+00 4.53667402e-01 -7.96695471e-01 -3.64125788e-01 -1.16709411e+00 3.70113343e-01 -1.05313003e-01 1.22042336e-01 2.15523735e-01 1.23042512e+00 -9.77477431e-01 1.26107350e-01 -1.01192105e+00 -5.95804632e-01 4.18781281e-01 1.64251789e-01 -1.13022774e-01 -1.97801858e-01 -1.26629460e+00 9.40128565e-01 -3.25318724e-01 1.62705049e-01 -6.48326218e-01 -5.11993170e-01 -6.62426591e-01 -2.55968481e-01 -1.99024037e-01 -3.24312359e-01 1.47890770e+00 -6.90125704e-01 -9.87218142e-01 4.42808717e-01 -4.22830492e-01 -5.84026694e-01 3.39785039e-01 -4.32811975e-01 -1.09588873e+00 5.63545287e-01 4.87135351e-02 1.99851871e-01 7.27083683e-01 -1.12663007e+00 -5.09441435e-01 -5.62700629e-01 -5.14482558e-01 -2.52830297e-01 -3.38471740e-01 6.06493115e-01 3.35023701e-01 -7.58476794e-01 2.03552768e-01 -8.40046108e-01 -1.44170985e-01 1.10284351e-01 -1.70469552e-01 -1.60629347e-01 7.52077758e-01 -9.11928952e-01 1.21017027e+00 -2.32716060e+00 -8.53050411e-01 8.25281218e-02 1.72104388e-01 2.79197961e-01 -3.54956910e-02 2.61464491e-02 9.32629928e-02 2.78833508e-01 -1.50084332e-01 -1.88337669e-01 -2.14212433e-01 2.87722677e-01 8.09272677e-02 3.76890957e-01 3.56687248e-01 1.18478626e-01 -6.56098604e-01 -2.60825723e-01 2.62204409e-01 8.70960057e-01 -4.97230083e-01 2.42710069e-01 7.20945746e-02 1.10843003e-01 -5.45563579e-01 5.18924892e-01 4.82491165e-01 3.31959188e-01 2.39835400e-02 -1.76632494e-01 1.05506316e-01 8.25415313e-01 -1.18806493e+00 7.35687435e-01 -3.07513505e-01 7.48437762e-01 2.56515086e-01 -9.08417225e-01 1.06789052e+00 4.85964566e-01 3.72813046e-01 -8.67028654e-01 -6.66474476e-02 3.50921333e-01 5.27768970e-01 -9.21230912e-01 -1.42643973e-01 -2.51314133e-01 2.75774866e-01 1.66204020e-01 -3.77611578e-01 3.11465174e-01 -1.27662942e-01 -3.31472754e-02 1.37379336e+00 -8.85441780e-01 9.95605811e-02 -2.23691210e-01 2.64771998e-01 -2.74380803e-01 6.04441285e-01 8.94034445e-01 -5.66032887e-01 5.76619744e-01 2.73589611e-01 -5.68782054e-02 -7.46836364e-01 -1.20505679e+00 -4.73764002e-01 9.55033541e-01 -4.53032911e-01 -4.57525998e-01 -6.04409099e-01 -2.08736002e-01 -1.45215228e-01 1.05620217e+00 -2.10391328e-01 -4.36376244e-01 -5.24534583e-01 -1.25732267e+00 9.66283679e-01 7.47798383e-01 2.78374165e-01 -9.86888349e-01 -5.41417718e-01 3.33103359e-01 -3.91151816e-01 -1.26341701e+00 -1.39927134e-01 3.92563552e-01 -6.96956694e-01 -9.39017355e-01 -6.20844364e-01 -8.63875806e-01 2.47477919e-01 3.08201104e-01 8.28790486e-01 -1.11597307e-01 -6.11596882e-01 7.15499520e-01 -5.63331664e-01 -8.28719795e-01 -1.05369818e+00 -3.72271001e-01 2.38142952e-01 -8.18742141e-02 6.03986859e-01 -6.32999778e-01 -4.55811977e-01 2.16780007e-01 -5.81252515e-01 -8.11783791e-01 5.35526872e-01 3.57053161e-01 3.42710376e-01 2.95234054e-01 9.41290617e-01 -4.52276796e-01 1.00783658e+00 -4.73051786e-01 1.27644256e-01 -2.44538948e-01 -2.53711998e-01 -2.29238778e-01 2.85892248e-01 -5.87003529e-01 -8.76372397e-01 -1.79203935e-02 -5.52385271e-01 1.08963460e-01 -6.91775739e-01 1.48213312e-01 -3.21138501e-01 2.53648221e-01 9.47417438e-01 6.47388250e-02 2.37843797e-01 -7.07985520e-01 -2.28934124e-01 1.12908697e+00 9.39224213e-02 -1.28427207e-01 2.55444139e-01 2.58496314e-01 -5.60391009e-01 -1.53098011e+00 -3.83169502e-01 -4.89573747e-01 -1.56989232e-01 -2.88951010e-01 1.08808899e+00 -8.05826545e-01 -3.36969011e-02 6.10551298e-01 -1.08431554e+00 -1.61874056e-01 2.67847329e-01 7.63800502e-01 -1.91278309e-01 3.02109092e-01 -6.60248280e-01 -1.35468221e+00 -2.78239012e-01 -1.39508593e+00 8.35257709e-01 -1.95940971e-01 -7.19172299e-01 -5.96597791e-01 -5.03764115e-02 4.98375535e-01 4.70171243e-01 -2.37109050e-01 1.15449154e+00 -9.33957279e-01 2.08154544e-01 -1.70301571e-01 -2.10482273e-02 6.39248252e-01 4.56606060e-01 -2.01480851e-01 -1.24963546e+00 1.22710489e-01 3.28938574e-01 1.72904483e-03 6.78624034e-01 7.33877599e-01 7.81843901e-01 -1.90713242e-01 -2.92940617e-01 -1.99778855e-01 8.89676869e-01 8.61825764e-01 7.23576546e-01 4.62479770e-01 2.28070408e-01 4.98327881e-01 1.51250765e-01 2.80578226e-01 -4.13886718e-02 6.43898368e-01 2.76709020e-01 1.97247341e-02 -3.62322867e-01 3.76915455e-01 8.11454475e-01 7.01438010e-01 3.23944539e-01 -2.15519052e-02 -9.17017758e-01 3.92135650e-01 -1.04453826e+00 -9.37839329e-01 -3.90187353e-01 2.25157309e+00 6.49927378e-01 2.58377939e-01 5.18326283e-01 5.89315236e-01 8.04428041e-01 1.28959671e-01 -1.43322811e-01 -3.95309687e-01 -8.59450325e-02 1.73354596e-01 1.59438059e-01 3.90244693e-01 -1.14161491e+00 7.41629526e-02 7.67999983e+00 1.72468618e-01 -1.00905204e+00 1.07708395e-01 4.16335672e-01 -3.19246262e-01 -5.86957447e-02 -8.66274536e-01 -6.73186958e-01 7.44006574e-01 1.51082921e+00 3.30588311e-01 -7.40238652e-02 7.88500905e-01 8.64443004e-01 -2.33242095e-01 -9.36650991e-01 8.63729954e-01 7.32658803e-02 -6.23050690e-01 -4.70330492e-02 7.07469583e-02 -1.42175913e-01 7.24311471e-02 -1.27222329e-01 5.96963204e-02 -7.71542639e-02 -8.22574556e-01 5.36685705e-01 2.90205896e-01 2.33421549e-01 -7.14622200e-01 8.03539217e-01 -1.19874440e-01 -1.05684447e+00 -2.44406343e-01 -3.60270917e-01 1.44836649e-01 2.71048814e-01 8.87384772e-01 -1.21344984e+00 -1.81259081e-01 9.51711595e-01 3.89478505e-01 -5.79709172e-01 1.22904932e+00 1.35010347e-01 1.23331904e+00 -5.32148898e-01 -3.05416107e-01 -1.98624447e-01 1.97845578e-01 7.21512198e-01 1.55194986e+00 4.89907295e-01 5.09981550e-02 7.56000131e-02 5.72439015e-01 3.97600174e-01 1.97221398e-01 -5.01611590e-01 -1.91637486e-01 6.55772626e-01 4.98026073e-01 -4.91818070e-01 -1.57702371e-01 -6.67196214e-01 4.71551180e-01 -3.04750174e-01 2.35136166e-01 -4.81788546e-01 -1.14692472e-01 9.27368641e-01 5.25881410e-01 5.90603845e-03 -2.21400917e-01 -2.58258164e-01 -3.38406593e-01 1.44935563e-01 -1.06183183e+00 3.72467339e-01 -7.60483623e-01 -1.44585252e+00 6.26114607e-01 -2.61272788e-01 -1.13465488e+00 -1.31914109e-01 -5.78906476e-01 -5.09691179e-01 6.61799550e-01 -1.07888341e+00 -4.98187751e-01 -2.26722270e-01 3.18146378e-01 8.00714254e-01 -2.05193833e-01 1.09424460e+00 6.98747158e-01 -4.88703698e-01 5.29742658e-01 3.86980101e-02 1.82360962e-01 7.41681814e-01 -1.08834052e+00 2.50230312e-01 5.15077412e-01 3.49493623e-02 4.91765648e-01 9.90553916e-01 -7.46010721e-01 -1.04463089e+00 -1.13839626e+00 7.62200236e-01 -2.12204546e-01 7.66220868e-01 -3.02032799e-01 -1.06263304e+00 3.07797700e-01 -3.43495309e-01 -3.02701384e-01 9.59180355e-01 1.71157449e-01 -2.71313459e-01 -3.58003438e-01 -1.23024762e+00 6.10192597e-01 8.38745713e-01 -6.98644817e-01 -7.98234522e-01 3.28775972e-01 6.93101943e-01 3.31396371e-01 -8.18509996e-01 2.97876149e-01 6.51759803e-01 -7.60329902e-01 1.21315825e+00 -3.67561311e-01 -1.00096777e-01 -3.41607668e-02 -5.41077971e-01 -1.41082406e+00 -2.12914169e-01 1.13036640e-01 3.02232027e-01 1.08626771e+00 5.74924529e-01 -8.90429795e-01 3.50281894e-01 5.02903044e-01 -2.33944729e-01 -4.00391668e-01 -9.86640751e-01 -9.18047667e-01 -9.62110981e-02 -1.18088126e+00 -8.15979466e-02 4.67547983e-01 -1.01577183e-02 2.09597677e-01 -1.86523706e-01 5.48331976e-01 4.46257859e-01 -9.86546397e-01 6.94941878e-02 -1.40019858e+00 -1.10090323e-01 -2.02579767e-01 -9.52212930e-01 -4.14996058e-01 -1.27194509e-01 -6.29002154e-01 1.20427467e-01 -1.56543100e+00 -1.11165382e-01 -3.50313306e-01 -2.88235635e-01 1.17461354e-01 -1.61102593e-01 3.59672271e-02 1.54965267e-01 2.62817321e-03 -2.54658721e-02 3.38742197e-01 5.64701438e-01 -4.47195888e-01 -3.68095398e-01 4.15429324e-01 -7.17364192e-01 7.62826979e-01 9.04887319e-01 -5.35649121e-01 -5.88882208e-01 -2.61642069e-01 -3.82293224e-01 -2.98855305e-01 5.03013074e-01 -1.27513957e+00 -1.28563076e-01 3.41647267e-01 5.22538185e-01 -4.26524073e-01 6.72708750e-01 -7.61225462e-01 -5.83824925e-02 6.66044354e-01 -5.32934427e-01 -7.91549906e-02 3.21951598e-01 7.17774630e-01 3.99346873e-02 -3.53587061e-01 1.00822890e+00 -1.82268947e-01 -4.26264793e-01 -4.57730830e-01 -1.34523952e+00 -3.62464860e-02 4.87184197e-01 -8.20947737e-02 -2.46341050e-01 -4.48363572e-01 -1.21734595e+00 -3.28921080e-01 -2.76228398e-01 4.77331311e-01 6.19211614e-01 -1.10823274e+00 -6.75489545e-01 1.89413205e-01 2.34149117e-02 -5.44310629e-01 1.24180421e-01 8.47715914e-01 -3.15906525e-01 2.67665356e-01 1.05862789e-01 -7.26938605e-01 -1.75712895e+00 4.51887436e-02 6.91858828e-01 4.20035332e-01 -8.84750068e-01 6.95684373e-01 -9.04762968e-02 1.59385368e-01 6.72870874e-01 -5.21888673e-01 -3.17962199e-01 1.31050527e-01 1.13636076e+00 7.82041013e-01 5.37281752e-01 -4.19073969e-01 -6.27588928e-01 5.63692376e-02 -3.50656688e-01 -1.91228256e-01 1.18018603e+00 4.74956352e-03 1.60714269e-01 8.19839001e-01 1.21593916e+00 -4.98429388e-02 -5.64586699e-01 6.98352084e-02 2.55172670e-01 -8.93421471e-02 2.75840133e-01 -7.65729725e-01 -8.22536409e-01 7.41686523e-01 1.52000606e+00 5.44054449e-01 1.01877880e+00 2.65727013e-01 9.22933340e-01 4.27735895e-01 2.42399663e-01 -1.05332661e+00 1.75695032e-01 2.86488086e-02 8.51168633e-01 -8.27133536e-01 -3.54467571e-01 -6.06503904e-01 -4.30402935e-01 1.01792407e+00 3.42345029e-01 -1.43625602e-01 8.58418405e-01 7.53715992e-01 5.89969277e-01 -4.95391823e-02 -5.37013769e-01 -2.49764368e-01 6.52790740e-02 1.15361118e+00 5.39722621e-01 2.08066151e-01 -2.19943583e-01 8.52680802e-01 -3.52185816e-01 -2.97638625e-01 5.53913355e-01 9.65057969e-01 -1.03587413e+00 -1.05410576e+00 -8.65994751e-01 1.12968314e+00 -6.67112708e-01 3.93243432e-02 -6.66691720e-01 4.06323224e-01 1.31968185e-01 1.58484423e+00 1.43817902e-01 -2.30112970e-01 6.68052733e-01 5.18775046e-01 7.90661350e-02 -7.13893533e-01 -1.76770583e-01 4.63146776e-01 8.55511844e-01 -4.14070010e-01 -4.57792997e-01 -1.05271876e+00 -1.26982927e+00 1.92196935e-01 -1.85602382e-01 -1.36614934e-01 6.34729445e-01 1.06436026e+00 2.41709843e-01 9.24185276e-01 2.89242864e-01 -4.05336082e-01 -4.85110074e-01 -1.23079264e+00 -8.30882668e-01 4.04685140e-02 6.08303607e-01 -7.97947705e-01 -7.99257696e-01 1.52453706e-01]
[13.947796821594238, 5.397353649139404]
f496e9a3-fc51-4477-99c2-16b1b4c2c6e8
mt-vae-learning-motion-transformations-to
1808.04545
null
http://arxiv.org/abs/1808.04545v1
http://arxiv.org/pdf/1808.04545v1.pdf
MT-VAE: Learning Motion Transformations to Generate Multimodal Human Dynamics
Long-term human motion can be represented as a series of motion modes---motion sequences that capture short-term temporal dynamics---with transitions between them. We leverage this structure and present a novel Motion Transformation Variational Auto-Encoders (MT-VAE) for learning motion sequence generation. Our model jointly learns a feature embedding for motion modes (that the motion sequence can be reconstructed from) and a feature transformation that represents the transition of one motion mode to the next motion mode. Our model is able to generate multiple diverse and plausible motion sequences in the future from the same input. We apply our approach to both facial and full body motion, and demonstrate applications like analogy-based motion transfer and video synthesis.
['Sunil Hadap', 'Ersin Yumer', 'Xinchen Yan', 'Kalyan Sunkavalli', 'Honglak Lee', 'Eli Shechtman', 'Akash Rastogi', 'Ruben Villegas']
2018-08-14
mt-vae-learning-motion-transformations-to-1
http://openaccess.thecvf.com/content_ECCV_2018/html/Xinchen_Yan_Generating_Multimodal_Human_ECCV_2018_paper.html
http://openaccess.thecvf.com/content_ECCV_2018/papers/Xinchen_Yan_Generating_Multimodal_Human_ECCV_2018_paper.pdf
eccv-2018-9
['human-pose-forecasting', 'human-dynamics']
['computer-vision', 'computer-vision']
[ 2.33149484e-01 2.88609087e-01 -3.47571343e-01 -1.67171165e-01 -5.21778882e-01 -4.51741576e-01 1.04922152e+00 -9.93339121e-01 1.64105058e-01 7.73833215e-01 8.76964211e-01 8.32668170e-02 2.63731450e-01 -7.64357924e-01 -9.25656676e-01 -6.47792220e-01 -3.30763310e-02 2.27775991e-01 4.49573770e-02 -3.70210350e-01 -6.90912604e-02 3.08236957e-01 -1.37543118e+00 6.07828438e-01 8.66099745e-02 1.10627726e-01 2.72877216e-01 1.26495087e+00 1.09927244e-01 1.08312988e+00 -2.06256837e-01 -2.22285315e-01 1.97604686e-01 -1.11246526e+00 -1.02639234e+00 3.05620253e-01 2.16575593e-01 -5.88578939e-01 -7.90742517e-01 3.24389935e-01 2.46254802e-01 4.75532740e-01 1.04918933e+00 -1.51410401e+00 -7.64386117e-01 1.87417641e-01 -1.95551500e-01 -2.74168044e-01 8.74042749e-01 7.38643587e-01 9.80510116e-01 -8.83322001e-01 1.48794341e+00 1.53309941e+00 5.50848424e-01 1.57295084e+00 -1.44032073e+00 -2.71872938e-01 -3.62823844e-01 2.36161172e-01 -9.17523205e-01 -5.99500179e-01 7.33204961e-01 -7.54517853e-01 9.62086082e-01 3.36275250e-02 1.05158830e+00 1.70482671e+00 6.06076717e-01 8.04430485e-01 1.95847362e-01 -2.40669668e-01 7.31437132e-02 -6.72132015e-01 -7.72044718e-01 8.12863469e-01 -4.64216352e-01 3.38157505e-01 -8.04019451e-01 -1.17620692e-01 1.36107552e+00 -3.55260730e-01 -2.65369952e-01 -3.31548959e-01 -1.58905327e+00 8.07161510e-01 7.36004189e-02 6.16429672e-02 -6.04681551e-01 8.74043286e-01 9.13766846e-02 2.48315215e-01 -9.29231197e-02 1.43611208e-01 3.47279534e-02 -2.74784654e-01 -1.03349137e+00 8.45864832e-01 6.31733954e-01 8.89212310e-01 7.35441983e-01 4.98441309e-01 -3.76571804e-01 2.85740227e-01 3.40079367e-01 4.82652545e-01 5.91756165e-01 -1.71402788e+00 2.51305938e-01 -8.90158415e-02 4.58646387e-01 -8.63624752e-01 6.22607134e-02 4.70372349e-01 -8.39311242e-01 2.20359072e-01 3.29063267e-01 -2.02136412e-01 -7.87268758e-01 2.16654515e+00 4.15227503e-01 7.62758493e-01 1.50400341e-01 7.66485095e-01 5.51115572e-01 1.17291188e+00 1.26008019e-01 -2.01744735e-01 8.60801756e-01 -9.26138341e-01 -7.48120844e-01 3.15010846e-02 4.68328804e-01 -6.34285510e-01 8.37557256e-01 -7.06160441e-02 -1.68776155e+00 -6.91229522e-01 -7.87278771e-01 -1.40780538e-01 4.51543570e-01 -3.25603604e-01 1.82567030e-01 2.03887671e-02 -1.38073623e+00 1.06437445e+00 -1.04493237e+00 -3.19777369e-01 8.27440619e-02 2.64336795e-01 -6.30316615e-01 2.17049733e-01 -1.17172706e+00 6.32688880e-01 1.77849814e-01 -9.52497572e-02 -1.21835077e+00 -8.31858695e-01 -1.10816479e+00 -1.64596617e-01 -3.77223015e-01 -1.83165503e+00 1.49564219e+00 -1.18213296e+00 -1.87182009e+00 7.31825709e-01 -7.09295094e-01 -2.92032540e-01 6.42595589e-01 -1.03622779e-01 -2.14551449e-01 2.86639124e-01 4.61434536e-02 1.36475611e+00 1.21189070e+00 -9.88106072e-01 -4.71327007e-01 3.21545422e-01 -3.04684371e-01 3.25810760e-01 1.16906077e-01 -1.18411899e-01 -3.03758740e-01 -9.70510602e-01 -2.83745795e-01 -1.18304884e+00 -3.24299902e-01 4.16149080e-01 -2.11613804e-01 1.52041027e-02 8.32037210e-01 -6.28210306e-01 1.12583923e+00 -1.73994529e+00 9.96873498e-01 -6.97483197e-02 -1.65666074e-01 -9.75237787e-02 -5.00624061e-01 6.56451523e-01 -2.16737971e-01 1.14345483e-01 -2.57107735e-01 -3.23161751e-01 -4.10297662e-02 4.85235661e-01 -6.02590382e-01 1.45938426e-01 4.85688925e-01 1.41231787e+00 -1.08100176e+00 -4.44511622e-01 2.57693470e-01 7.05522418e-01 -7.83299148e-01 5.87067068e-01 -5.04345000e-01 1.18652320e+00 -2.80366957e-01 1.77593395e-01 8.45163763e-02 -2.77468473e-01 1.10948116e-01 -3.25074866e-02 2.80137211e-01 -2.47139595e-02 -7.57293582e-01 2.03424454e+00 -4.91991550e-01 7.95133948e-01 -2.92649806e-01 -4.29454684e-01 5.36208391e-01 7.62448549e-01 7.67170548e-01 -1.67197183e-01 -1.83818921e-01 -2.56084967e-02 -1.87053233e-01 -7.65049458e-01 5.40614486e-01 -5.42454481e-01 -2.99876872e-02 6.16023958e-01 2.47980610e-01 -2.59968102e-01 -1.26406863e-01 -1.43964484e-03 9.35338676e-01 9.15488899e-01 2.87042648e-01 1.05673388e-01 5.38768172e-01 -3.65829021e-02 5.30946970e-01 2.45930165e-01 -1.23349782e-02 9.00315046e-01 3.20038825e-01 -5.96029699e-01 -1.61176538e+00 -1.51035178e+00 6.74300313e-01 7.57933855e-01 -1.47444278e-01 -3.95478249e-01 -7.55468547e-01 -1.59264833e-01 -1.78688720e-01 6.55958295e-01 -7.51042485e-01 -3.15897495e-01 -1.28430820e+00 -6.58515841e-02 5.42573929e-01 6.64251864e-01 -2.88645457e-03 -1.50289297e+00 -7.32669055e-01 4.47871119e-01 -6.03673041e-01 -9.91121709e-01 -1.08829153e+00 -8.96500349e-01 -9.89661872e-01 -7.00348198e-01 -1.07478249e+00 -1.01701558e+00 3.15187007e-01 -1.04893081e-01 1.24220395e+00 -1.00418665e-01 -4.59716320e-01 6.15347445e-01 5.32745980e-02 2.65571654e-01 -1.12077403e+00 -4.59740162e-01 1.23461567e-01 2.85172135e-01 -3.58355314e-01 -8.66737008e-01 -9.39742208e-01 1.54016882e-01 -9.78093922e-01 4.32858855e-01 2.20911670e-02 1.01423883e+00 5.69556534e-01 -7.52428412e-01 2.96874166e-01 -3.69961321e-01 2.76737303e-01 -7.08618283e-01 -1.26333505e-01 1.37811631e-01 2.04409495e-01 3.43599290e-01 5.89511752e-01 -7.12380648e-01 -1.18645859e+00 2.80031025e-01 -2.11493924e-01 -9.12620842e-01 -1.19351231e-01 -8.88911635e-03 -7.27829859e-02 3.40958804e-01 5.90252459e-01 5.18314719e-01 7.11875781e-02 8.32649413e-03 9.60465550e-01 -1.51510602e-02 1.07524669e+00 -5.99570274e-01 9.82033193e-01 8.19571733e-01 4.81675237e-01 -9.17180300e-01 -5.23440540e-03 1.18723571e-01 -8.08828831e-01 -5.30808985e-01 1.36353838e+00 -9.51470912e-01 -6.92696929e-01 4.16588157e-01 -1.60951877e+00 -7.19591737e-01 -4.89333153e-01 4.00004387e-01 -1.35733235e+00 4.12887245e-01 -7.65404999e-01 -5.71828306e-01 -1.95868224e-01 -9.61693764e-01 1.15583920e+00 1.01426795e-01 -9.63371992e-01 -1.30470657e+00 5.29659271e-01 -1.21931806e-01 4.29232344e-02 8.82498145e-01 9.21777368e-01 4.18540061e-01 -9.10531640e-01 3.52483571e-01 6.86541975e-01 -9.99899581e-02 2.60185719e-01 4.38251078e-01 -5.20869553e-01 -2.12878302e-01 -4.38787222e-01 -1.96837395e-01 6.35262787e-01 7.25790560e-01 5.56406021e-01 -6.00834250e-01 -4.33860242e-01 8.56650472e-01 1.06472301e+00 1.72871068e-01 9.11023259e-01 -1.38914457e-03 9.26974475e-01 8.85317564e-01 3.47878665e-01 4.16352153e-01 2.31764629e-01 9.60723341e-01 8.25475827e-02 3.70766312e-01 -4.72296357e-01 -8.50799799e-01 9.36762035e-01 8.24300230e-01 -3.52422178e-01 -1.50606573e-01 -6.06922507e-01 9.34905350e-01 -1.93810570e+00 -1.60464191e+00 3.71317714e-02 1.74620974e+00 7.03913569e-01 -3.71864825e-01 4.83471066e-01 -3.31530064e-01 6.42924488e-01 3.24567705e-01 -5.43546438e-01 -3.76669198e-01 7.48323947e-02 4.16051537e-01 -2.37171352e-01 7.57844627e-01 -8.52859914e-01 1.14347684e+00 7.47475004e+00 5.26573062e-01 -1.04497063e+00 -1.74003974e-01 3.11731219e-01 -2.46529907e-01 -8.92368674e-01 1.00262456e-01 -2.63398260e-01 3.20393920e-01 1.14769435e+00 -3.74164701e-01 2.87098348e-01 3.93395364e-01 5.59135497e-01 5.19995809e-01 -1.43659842e+00 9.24015224e-01 -2.78701723e-01 -1.94510281e+00 8.27315032e-01 -2.21500653e-04 9.67312098e-01 -6.01575732e-01 2.65226483e-01 -1.31749094e-01 4.51220006e-01 -1.35349822e+00 1.00725091e+00 8.85092318e-01 1.12201154e+00 -6.33204639e-01 -1.71514124e-01 4.32204306e-01 -1.47210109e+00 6.83315769e-02 -8.97209123e-02 1.75864995e-01 7.55964935e-01 -2.18991444e-01 -5.99431992e-01 5.47349513e-01 1.68583393e-01 9.69827831e-01 2.06831113e-01 3.38202178e-01 -2.06383586e-01 4.70526487e-01 2.70298690e-01 3.62485707e-01 9.08216462e-02 -3.31272870e-01 8.81205082e-01 1.00430942e+00 7.29053020e-01 2.30294943e-01 -1.93193451e-01 1.13576651e+00 1.37800500e-01 -3.74365181e-01 -9.94871795e-01 7.73057118e-02 3.33770186e-01 8.07620823e-01 -7.46319816e-02 -2.70023853e-01 -3.64773631e-01 1.43467534e+00 -4.38734815e-02 4.90103304e-01 -8.74707341e-01 6.31026551e-02 1.40001714e+00 1.27887830e-01 3.41873437e-01 -4.51672167e-01 1.14843458e-01 -1.28603995e+00 -2.54522502e-01 -5.96286178e-01 1.07048310e-01 -1.07304096e+00 -1.00856543e+00 5.01323998e-01 9.42355096e-02 -1.52254581e+00 -1.46890783e+00 -1.81425050e-01 -9.31232691e-01 8.19686651e-01 -8.61106753e-01 -1.40110075e+00 -8.12603068e-03 7.94974208e-01 8.49751174e-01 -3.21492404e-01 8.35903764e-01 -1.59601837e-01 4.81980853e-02 5.56528270e-01 -2.92488754e-01 1.07499734e-01 4.90619451e-01 -8.41333866e-01 1.08437300e+00 7.07649291e-01 2.60855049e-01 4.24850792e-01 8.51252079e-01 -6.49556279e-01 -1.31227839e+00 -1.29788911e+00 9.17060852e-01 -7.44344652e-01 4.76952136e-01 -2.25384440e-02 -8.53106558e-01 9.41676199e-01 3.61976355e-01 -1.45183608e-01 5.26164293e-01 -1.06371355e+00 -8.63042995e-02 5.06538510e-01 -8.04669499e-01 1.20849824e+00 1.51232851e+00 -6.81065083e-01 -6.12239063e-01 -1.94114268e-01 7.54848123e-01 -4.12739098e-01 -8.73419225e-01 1.72714517e-01 1.04072011e+00 -8.03457141e-01 1.33155012e+00 -1.17016029e+00 1.16982281e+00 -1.93833753e-01 -7.87742138e-02 -1.27739692e+00 -5.00113904e-01 -1.44443798e+00 -5.60357630e-01 7.13174164e-01 1.07583255e-01 3.51747088e-02 1.08113253e+00 4.74790275e-01 3.94600183e-02 -5.46315730e-01 -9.04198527e-01 -7.35750377e-01 4.53247070e-01 -1.30559653e-01 6.88717306e-01 8.62660587e-01 -3.43556285e-01 6.87768087e-02 -9.37155187e-01 -1.53900117e-01 4.60787535e-01 1.25740200e-01 1.07744002e+00 -6.20450139e-01 -7.12349355e-01 -2.89471865e-01 -4.35368180e-01 -1.19853878e+00 6.65920377e-01 -9.23723817e-01 1.15678303e-01 -1.36230409e+00 9.71962884e-02 5.45977116e-01 4.11975235e-01 1.86946943e-01 -1.98023662e-01 1.32523281e-02 4.49143827e-01 5.58709443e-01 6.19158037e-02 1.00162411e+00 1.74032974e+00 1.51112005e-01 -2.75069535e-01 -1.98155031e-01 7.65950680e-02 7.22632527e-01 3.87625664e-01 -3.94529253e-01 -7.22042143e-01 -3.45755160e-01 -4.28148136e-02 1.00689340e+00 5.33911407e-01 -8.83058548e-01 -3.28817628e-02 -6.97116613e-01 4.86726224e-01 -2.01451942e-01 5.22337615e-01 -2.95784205e-01 9.11110997e-01 6.74842298e-01 -5.91320395e-01 3.06845367e-01 -4.06571925e-02 7.62204707e-01 -9.59493518e-02 2.15259477e-01 6.39748335e-01 -2.60192871e-01 -8.29562545e-01 5.00795007e-01 -7.32746422e-01 5.21903895e-02 1.00139046e+00 -5.61692238e-01 1.71095341e-01 -1.03258967e+00 -1.24897182e+00 -5.63799515e-02 6.10871851e-01 6.59908891e-01 9.04906690e-01 -1.96986723e+00 -1.00333905e+00 1.82903245e-01 -1.96447968e-01 -2.70555377e-01 3.04642677e-01 2.59641141e-01 -8.07056248e-01 1.53650403e-01 -7.59340703e-01 -6.41438961e-01 -1.17225218e+00 2.51546562e-01 4.31730658e-01 -1.60567164e-01 -8.22467208e-01 5.61756074e-01 4.57235157e-01 -8.07787329e-02 -4.35409665e-01 -2.06781253e-02 4.90069166e-02 -4.13538367e-01 4.38359261e-01 3.80926102e-01 -8.47240865e-01 -1.15143204e+00 -1.26915991e-01 8.98301423e-01 6.91546202e-01 -9.50249672e-01 1.09063935e+00 -2.08266303e-01 6.70428649e-02 5.93438983e-01 1.38956177e+00 -3.92480224e-01 -1.84863532e+00 2.26537332e-01 -4.49968800e-02 -4.84935641e-01 -8.12555730e-01 7.94995576e-02 -8.37031305e-01 8.68753016e-01 9.34074074e-02 -5.63935995e-01 8.98928523e-01 -4.57859896e-02 1.31506288e+00 1.06180549e-01 3.56567025e-01 -6.73801720e-01 6.72573328e-01 4.06255454e-01 1.10073876e+00 -5.60241103e-01 -5.04929662e-01 -2.00499594e-01 -8.86181951e-01 1.30282152e+00 3.07311743e-01 -4.30332601e-01 5.59429824e-01 3.59594002e-02 -1.05390273e-01 1.49997219e-01 -1.24276185e+00 1.21892221e-01 5.80249548e-01 1.03434587e+00 4.65851665e-01 -6.44789636e-02 -7.04508275e-02 2.10849822e-01 -3.39549899e-01 2.91796058e-01 6.68491602e-01 6.90888107e-01 -3.54999304e-02 -1.35948777e+00 -3.38006556e-01 -1.71155274e-01 -6.07287511e-02 2.92037904e-01 -2.52040058e-01 7.11460769e-01 -6.05994370e-03 7.29884744e-01 2.24889413e-01 -5.86139321e-01 3.28880809e-02 4.00150716e-01 9.23532367e-01 -5.53534746e-01 -2.33907312e-01 1.45349234e-01 1.52684608e-03 -1.16626275e+00 -6.80729985e-01 -8.81696820e-01 -1.29801643e+00 -4.60380226e-01 6.92656636e-01 -2.91695178e-01 6.41658604e-02 5.71032047e-01 3.33083272e-01 4.46463913e-01 6.40634358e-01 -1.23991859e+00 -2.90161252e-01 -4.45677727e-01 -2.78308868e-01 9.81241703e-01 6.59129500e-01 -2.88344979e-01 7.32940994e-03 1.03735769e+00]
[7.398108959197998, -0.20243282616138458]
aa00f972-88a6-4433-9eea-097037575ea0
qrnet-optimal-regulator-design-with-lqr
2009.05686
null
https://arxiv.org/abs/2009.05686v2
https://arxiv.org/pdf/2009.05686v2.pdf
QRnet: optimal regulator design with LQR-augmented neural networks
In this paper we propose a new computational method for designing optimal regulators for high-dimensional nonlinear systems. The proposed approach leverages physics-informed machine learning to solve high-dimensional Hamilton-Jacobi-Bellman equations arising in optimal feedback control. Concretely, we augment linear quadratic regulators with neural networks to handle nonlinearities. We train the augmented models on data generated without discretizing the state space, enabling application to high-dimensional problems. We use the proposed method to design a candidate optimal regulator for an unstable Burgers' equation, and through this example, demonstrate improved robustness and accuracy compared to existing neural network formulations.
['Tenavi Nakamura-Zimmerer', 'Wei Kang', 'Qi Gong']
2020-09-11
null
null
null
null
['physics-informed-machine-learning']
['graphs']
[-1.66273177e-01 4.30856079e-01 -4.56039071e-01 5.27752519e-01 -7.27624297e-01 -4.79588896e-01 3.91237944e-01 -3.61689001e-01 -1.00938477e-01 1.09974229e+00 5.88541254e-02 -5.25139809e-01 -4.23119873e-01 -1.40816495e-01 -7.58934498e-01 -8.52887094e-01 -2.98791319e-01 4.98953253e-01 -3.03125143e-01 -5.57135940e-01 3.04073066e-01 8.34272802e-01 -7.44224787e-01 -5.26228905e-01 9.96028244e-01 1.04552233e+00 -3.53072017e-01 7.98683703e-01 7.30077386e-01 5.72407126e-01 -7.86976889e-02 5.44474900e-01 6.21641934e-01 -2.88257807e-01 -4.64496583e-01 -4.43689972e-02 3.19784850e-01 -4.02361691e-01 -8.54654849e-01 7.05771804e-01 5.05760312e-01 5.91877162e-01 1.11720538e+00 -9.14012372e-01 -1.09548998e+00 -4.77203503e-02 -1.04332633e-01 3.35475475e-01 -7.82408491e-02 6.63926363e-01 9.13808584e-01 -7.97479749e-01 4.61130083e-01 1.26092923e+00 9.04560745e-01 7.97760487e-01 -1.39673305e+00 -3.42385679e-01 1.16963536e-01 -9.65079367e-02 -1.26503217e+00 -4.28927600e-01 7.27799654e-01 -8.62140954e-01 8.62743914e-01 -1.79906547e-01 6.06750667e-01 7.40737319e-01 7.48129368e-01 3.70684028e-01 7.11724997e-01 -3.62510949e-01 4.00014669e-01 -1.59908697e-01 6.65415004e-02 9.11092460e-01 2.43396401e-01 7.92824268e-01 -2.96826720e-01 -3.63108814e-01 1.07608426e+00 -5.05392969e-01 -4.49201733e-01 -4.33737367e-01 -8.44475389e-01 1.16166735e+00 6.98130310e-01 -1.36371434e-01 -5.99028587e-01 3.35334897e-01 2.70404458e-01 1.27941400e-01 6.71075046e-01 1.25687408e+00 -4.16926950e-01 3.78924906e-01 -3.82124096e-01 5.81799567e-01 8.40107441e-01 8.52812111e-01 3.99757236e-01 6.84927225e-01 -1.23108394e-01 4.31824267e-01 4.92546856e-01 6.43490195e-01 1.80269942e-01 -1.42014682e+00 2.48712655e-02 2.53872007e-01 8.00330102e-01 -9.58921790e-01 -8.01636636e-01 -2.99577504e-01 -1.32486224e+00 3.23114485e-01 2.84483463e-01 -6.70772493e-01 -8.74162257e-01 1.54757547e+00 5.41660130e-01 4.21406120e-01 2.30616614e-01 1.40547717e+00 2.56249309e-01 1.01392794e+00 -3.21679682e-01 -5.26120782e-01 7.96676993e-01 -9.32193518e-01 -9.10113275e-01 -1.11860082e-01 4.24289972e-01 -1.50075495e-01 6.78303421e-01 3.15342486e-01 -1.20925128e+00 -2.06268370e-01 -1.09275806e+00 9.77006778e-02 -1.50806546e-01 2.07356200e-01 2.76056975e-01 -9.83099118e-02 -1.13220429e+00 1.05085266e+00 -7.91824520e-01 6.84671551e-02 -1.30336091e-01 6.00590885e-01 2.21261308e-02 7.35722423e-01 -1.53987241e+00 1.35001361e+00 2.68623710e-01 4.84027445e-01 -9.01520967e-01 -1.12590241e+00 -7.71684229e-01 -3.30689371e-01 4.11246777e-01 -9.43302572e-01 1.40463340e+00 -3.20521891e-01 -2.09256554e+00 2.16068506e-01 -1.61402479e-01 -5.49540043e-01 1.59658819e-01 -3.14095527e-01 1.05866387e-01 1.36213601e-01 -1.40849099e-01 1.78209156e-01 1.00476813e+00 -1.13678443e+00 1.88801158e-02 9.90426838e-02 -1.11271730e-02 -9.40293074e-02 -1.24935523e-01 -3.80296826e-01 1.06677033e-01 -5.72614908e-01 1.77500382e-01 -1.29306078e+00 -9.06248152e-01 1.43789366e-01 -4.48268354e-01 -7.65612498e-02 7.46542692e-01 -8.13400030e-01 1.14728653e+00 -1.46384537e+00 8.68564963e-01 3.25730771e-01 1.51935756e-01 5.44133186e-01 1.30153641e-01 5.08619428e-01 6.64660111e-02 9.00961757e-02 -3.30545127e-01 -1.19783677e-01 5.90803288e-02 2.46502191e-01 -6.62228644e-01 8.19102645e-01 6.82465792e-01 7.73897052e-01 -6.91988587e-01 -1.58966675e-01 2.36317858e-01 5.35679102e-01 -8.32568705e-01 3.05380106e-01 -3.62752825e-01 1.13620543e+00 -8.87369812e-01 3.45466018e-01 3.19798708e-01 -2.95535117e-01 -2.33929217e-01 -1.46300778e-01 -3.27087581e-01 -4.36991267e-02 -1.14231682e+00 1.19620514e+00 -6.28995240e-01 5.25039256e-01 9.32162702e-01 -1.63197470e+00 7.71995306e-01 4.35288846e-01 6.43475235e-01 -4.80593532e-01 3.21621627e-01 2.46703148e-01 -2.13693291e-01 -6.40303612e-01 4.02444750e-01 -5.65942347e-01 -1.05686173e-01 9.92464200e-02 -4.88265567e-02 -6.54261768e-01 -1.37796998e-01 3.21355984e-02 9.55765367e-01 1.94088239e-02 4.05282289e-01 -7.97939122e-01 9.40883577e-01 4.78339463e-01 3.96144927e-01 7.80737042e-01 -4.05801862e-01 1.59481212e-01 5.15985250e-01 -5.57593465e-01 -1.31904137e+00 -5.76507330e-01 -1.68773100e-01 7.43226528e-01 1.86616540e-01 2.68136114e-01 -4.78200495e-01 1.62993580e-01 5.64082682e-01 4.96793568e-01 -5.56145072e-01 -5.37246585e-01 -1.05669510e+00 -4.68794078e-01 1.83109164e-01 2.78021425e-01 -1.20426089e-01 -5.50850272e-01 -2.14661166e-01 4.41894472e-01 5.19688070e-01 -8.57597291e-01 -5.42826593e-01 1.29166424e-01 -7.17353463e-01 -1.09124637e+00 -5.32622576e-01 -6.13305509e-01 5.13420463e-01 -3.49419832e-01 6.87697411e-01 -6.84769750e-02 -4.23182279e-01 6.26539469e-01 3.40247333e-01 -1.36452183e-01 -8.03012609e-01 1.54875759e-02 8.76535654e-01 -2.58550376e-01 -6.16454601e-01 -3.40198517e-01 -5.30619740e-01 3.32345515e-01 -4.92111802e-01 -2.07052588e-01 4.05725092e-01 1.15018570e+00 5.93782723e-01 -4.27692771e-01 1.07849073e+00 -2.62763500e-01 1.05043614e+00 -6.41662419e-01 -1.37735927e+00 -1.53452188e-01 -7.18360066e-01 6.28841937e-01 1.37508464e+00 -8.14116240e-01 -9.03363168e-01 2.57247627e-01 5.19277267e-02 -8.06658208e-01 3.01557571e-01 5.35279036e-01 4.94642347e-01 -6.65590465e-01 9.42632794e-01 -4.83778305e-02 5.22388041e-01 -2.86003530e-01 6.56089127e-01 5.51658213e-01 7.26162374e-01 -8.33549500e-01 9.38486397e-01 3.17833543e-01 6.33522511e-01 -8.02930117e-01 -9.61482346e-01 -5.04457057e-01 -7.65539825e-01 -2.63581127e-01 7.07200468e-01 -7.24052072e-01 -1.20364130e+00 3.29964876e-01 -1.29232347e+00 -5.45062840e-01 -2.18168780e-01 5.16874552e-01 -1.06507683e+00 -1.83930293e-01 -1.03622532e+00 -1.24684787e+00 -4.59925026e-01 -1.08428752e+00 1.06195080e+00 1.59704477e-01 -3.31848767e-03 -1.52893567e+00 5.12646735e-01 -2.27726161e-01 7.51405239e-01 3.84159982e-01 6.02562070e-01 -1.85531497e-01 -5.04250884e-01 -3.26538920e-01 4.84785885e-02 2.74096251e-01 -2.65742749e-01 -1.44624785e-02 -6.42847478e-01 -4.73159283e-01 1.91217452e-01 -3.78848851e-01 6.75117850e-01 8.59106183e-01 5.35941184e-01 -7.76402652e-01 -5.27450442e-01 4.39316541e-01 1.36272502e+00 1.27539977e-01 -2.30300844e-01 1.94286168e-01 7.78856575e-01 4.07775700e-01 4.24644530e-01 4.74596858e-01 -3.01625486e-02 6.75957441e-01 3.40555698e-01 2.48733386e-02 5.87504208e-01 5.78174554e-02 3.35003346e-01 7.93492973e-01 1.56810910e-01 1.70373395e-01 -9.06165600e-01 4.79742318e-01 -1.91087127e+00 -5.60928881e-01 -1.41879946e-01 1.72465754e+00 7.73924172e-01 -3.48816574e-01 -1.04390524e-01 -5.70458531e-01 7.24345207e-01 -6.54599145e-02 -1.18671238e+00 -5.82608044e-01 1.05741844e-01 -1.15805805e-01 8.18705678e-01 9.57540214e-01 -1.28453958e+00 8.27751756e-01 8.00080299e+00 4.84872252e-01 -1.24361598e+00 -4.09638844e-02 4.59937930e-01 -3.85946408e-02 2.49673888e-01 -1.72456503e-01 -9.20949459e-01 5.35121076e-02 1.29957366e+00 -6.09685719e-01 7.33520150e-01 7.51369655e-01 1.06994760e+00 8.78911242e-02 -8.62329364e-01 5.49177110e-01 -2.38913268e-01 -1.77125466e+00 -3.70049179e-01 -1.03622591e-02 1.10869122e+00 -1.97555363e-01 5.89876734e-02 2.78437287e-01 5.19503057e-01 -1.07194507e+00 5.21928012e-01 8.34324658e-01 4.34236139e-01 -2.75204748e-01 3.03636640e-01 5.34552395e-01 -9.54847455e-01 -3.86411846e-01 -3.43120575e-01 -3.63347203e-01 6.24591231e-01 4.14126188e-01 -6.31885409e-01 1.29000708e-01 5.22958376e-02 8.83491814e-01 1.88514769e-01 9.12317693e-01 -2.50554420e-02 5.70824981e-01 -5.91195226e-01 -2.05835894e-01 4.72941458e-01 -6.64090931e-01 9.06409264e-01 7.70711780e-01 2.99241126e-01 5.75444818e-01 3.37155282e-01 1.20331693e+00 8.04341957e-02 -1.37398720e-01 -9.23500061e-01 1.07918233e-01 1.15904994e-01 1.07263672e+00 -9.49870646e-02 -3.34086210e-01 -2.04004683e-02 3.32521141e-01 4.64348197e-01 6.88126564e-01 -8.22514534e-01 -1.26500791e-02 9.81291890e-01 -5.45754135e-02 1.06078468e-01 -6.17643595e-01 -1.94454119e-01 -1.08459580e+00 -1.50638640e-01 -5.18288612e-01 1.14113674e-01 -6.10454023e-01 -1.41665494e+00 8.81211534e-02 2.29213640e-01 -1.12490439e+00 -6.10648572e-01 -1.02131677e+00 -7.21909225e-01 1.01827836e+00 -1.25452054e+00 -8.35959196e-01 3.68356973e-01 1.14102818e-01 1.24791764e-01 -8.48706812e-02 4.21053112e-01 -9.37377810e-02 -8.93020988e-01 -1.00161552e-01 7.97581017e-01 -2.77317226e-01 3.54300678e-01 -1.31736875e+00 2.10732043e-01 7.27557838e-01 -9.08349574e-01 6.34004235e-01 1.15792012e+00 -6.96573794e-01 -2.02617407e+00 -1.23673141e+00 4.92618866e-02 -1.13915652e-01 1.36147678e+00 -2.43687890e-02 -1.12718952e+00 6.50975645e-01 2.78666168e-01 9.07155424e-02 -1.53333843e-01 -4.57524449e-01 2.04591542e-01 2.10253969e-01 -1.13310194e+00 5.08778632e-01 6.93837225e-01 -3.93659681e-01 -4.31206644e-01 7.42480576e-01 8.58184218e-01 -7.41366446e-01 -1.11794567e+00 4.43893939e-01 3.07734579e-01 -4.49130163e-02 1.04720414e+00 -1.13635576e+00 3.23502481e-01 -2.38086313e-01 3.67902786e-01 -1.92571664e+00 -4.36757773e-01 -1.31022084e+00 -5.53216219e-01 3.17823261e-01 4.13048387e-01 -8.30587447e-01 5.00320792e-01 7.95774281e-01 -4.72907722e-01 -1.10551989e+00 -1.23449695e+00 -7.92043746e-01 8.68603349e-01 -1.52588561e-01 -1.52862206e-01 6.23286247e-01 3.01134676e-01 2.70027459e-01 -8.00942719e-01 6.83976889e-01 6.13950074e-01 -2.65926659e-01 4.61537451e-01 -7.58830667e-01 -2.22430393e-01 -5.90019107e-01 1.45275993e-02 -1.27029669e+00 8.57885718e-01 -6.11603320e-01 4.91777450e-01 -1.38720393e+00 -5.10520816e-01 -3.27295721e-01 3.59163433e-02 8.50184355e-03 -8.16021785e-02 -4.74298261e-02 -5.96007816e-02 2.04278886e-01 -3.19891363e-01 9.19652164e-01 1.36554587e+00 -1.60454482e-01 -4.88556772e-01 -2.55920589e-02 -3.79093111e-01 5.15687168e-01 8.95648003e-01 -1.06065132e-01 -2.28149533e-01 -2.65958812e-02 -1.14185929e-01 5.31901360e-01 3.97154301e-01 -8.81115556e-01 4.72497821e-01 -6.05681419e-01 -1.03670344e-01 -1.18540369e-01 4.21002716e-01 -4.03098285e-01 -2.02490136e-01 8.56947422e-01 -4.71425176e-01 8.09342507e-03 4.73674685e-01 7.78306127e-01 1.44700869e-03 -8.64727795e-02 1.27385426e+00 6.39656484e-02 -2.55205870e-01 2.89903373e-01 -9.84421849e-01 1.92923740e-01 9.47960854e-01 4.81436998e-01 -2.54751712e-01 -5.51797330e-01 -9.13207769e-01 5.83343148e-01 2.48049691e-01 4.76908404e-03 4.10414636e-01 -1.19536650e+00 -4.53349411e-01 2.30473116e-01 -3.86044234e-01 -2.59369556e-02 1.02279633e-01 9.62268114e-01 -5.30603826e-01 7.94304252e-01 1.79435611e-01 -6.11790180e-01 -4.37614560e-01 7.50198126e-01 1.10113657e+00 -3.02228063e-01 -4.30610478e-01 5.67766070e-01 -1.01396456e-01 -6.27970636e-01 -4.58141454e-02 -5.54332554e-01 -8.15993994e-02 -3.21862578e-01 1.71299413e-01 4.50076222e-01 -2.79824823e-01 -7.34973252e-01 -6.52673990e-02 7.34182060e-01 4.68285024e-01 -1.19844988e-01 1.18616807e+00 9.19349398e-03 -5.28029315e-02 2.82604158e-01 1.30475092e+00 -4.56661522e-01 -1.52895248e+00 -1.86163887e-01 2.97232550e-02 1.34100124e-01 4.01046395e-01 -1.83484003e-01 -8.43845308e-01 5.30738235e-01 3.07273805e-01 2.86253989e-01 5.65249741e-01 -1.67202741e-01 6.23176634e-01 8.65860224e-01 3.35866399e-03 -1.08883905e+00 1.88515820e-02 9.69787836e-01 1.20293760e+00 -1.27848792e+00 1.74053073e-01 -1.90753251e-01 -3.85813206e-01 1.30250919e+00 7.48585522e-01 -9.29553390e-01 9.84951198e-01 2.82976210e-01 -7.18639493e-02 2.94562839e-02 -1.06847513e+00 1.46169633e-01 6.10804737e-01 2.91673720e-01 1.04102314e-01 -2.61961311e-01 -3.18926662e-01 1.10268869e-01 2.58300185e-01 -2.17110924e-02 7.55431294e-01 8.35456729e-01 -6.34736419e-01 -5.57148159e-01 -6.34982526e-01 5.31859517e-01 -4.29617800e-02 1.78168699e-01 -2.19958454e-01 7.73437202e-01 -4.33767349e-01 7.40699232e-01 -7.21750921e-03 6.64325878e-02 5.39125860e-01 -1.99070033e-02 1.22665569e-01 -3.60933244e-01 -2.78547406e-01 1.13228269e-01 -1.62000924e-01 -4.92407024e-01 -1.76637650e-01 -3.93443048e-01 -1.28061724e+00 -1.96574956e-01 -4.41242665e-01 3.23807418e-01 4.10898626e-01 1.05386746e+00 6.67465568e-01 4.14765507e-01 5.31027079e-01 -1.70678198e+00 -1.35740614e+00 -9.31561410e-01 -4.03839231e-01 1.20158970e-01 1.09629023e+00 -1.26342070e+00 -8.55353951e-01 -9.61035937e-02]
[6.299037456512451, 3.3827760219573975]
6a51b8fb-dc1a-4d4d-96b2-36dd48e6140a
open-access-to-orbit-and-runaway-space-debris
2202.07442
null
https://arxiv.org/abs/2202.07442v1
https://arxiv.org/pdf/2202.07442v1.pdf
Open access to orbit and runaway space debris growth
As Earth's orbits fill with satellites and debris, debris-producing collisions between orbiting bodies become more likely. Runaway space debris growth, known as Kessler Syndrome, may render Earth's orbits unusable for centuries. We present a dynamic physico-economic model of Earth orbit use under rational expectations with endogenous collision risk and Kessler Syndrome. When satellites can be destroyed in collisions with debris and other satellites, the open-access equilibrium manifold allows for multiple steady states. When debris can collide to produce more debris, at least one steady state may be a tipping point and Kessler Syndrome can occur along equilibrium paths. We show open access is increasingly and inefficiently likely to cause Kessler Syndrome as satellites become more profitable. Calibrated simulations reveal Kessler Syndrome is expected to occur in low-Earth orbit around 2048 under recent historical sectoral growth trends, and may occur as early as 2035 if the space economy grows consistent with projections by major investment banks. These results highlight the urgent need for modeling and policy approaches which incorporate open access and positive feedbacks in debris growth.
['Giacomo Rondina', 'Akhil Rao']
2022-02-12
null
null
null
null
['2048']
['playing-games']
[-9.71183479e-01 5.92919350e-01 -4.17167068e-01 8.62265289e-01 3.02476466e-01 -6.76282763e-01 9.84901726e-01 -2.30655476e-01 -2.31176242e-01 1.16357291e+00 4.79174465e-01 -8.81393075e-01 -6.79014772e-02 -9.05282497e-01 -8.24448645e-01 -4.67163414e-01 -6.59575045e-01 6.40285373e-01 8.06904808e-02 -3.18092644e-01 -2.53555551e-02 6.18196130e-01 -1.47773659e+00 -7.47849047e-01 9.27401960e-01 5.46807230e-01 4.10650581e-01 3.59325290e-01 -7.84986690e-02 2.56755412e-01 -3.00857276e-01 -4.84358072e-02 9.40417528e-01 2.72393245e-02 -5.81850469e-01 -4.60358143e-01 -6.32778645e-01 -2.36340091e-01 -2.94797897e-01 7.77466416e-01 3.70917432e-02 3.84056754e-02 8.50952327e-01 -1.12200427e+00 -4.46434200e-01 5.49299717e-01 -6.66834772e-01 5.65532684e-01 -3.39400917e-02 7.33167470e-01 7.32482016e-01 -9.11152601e-01 7.42301762e-01 1.17185748e+00 6.73453212e-01 -1.44055516e-01 -8.32120836e-01 -4.03382242e-01 -2.20301032e-01 -4.25285280e-01 -1.21328568e+00 -3.71398985e-01 -1.66851297e-01 -7.51560748e-01 1.33214414e+00 6.79849029e-01 1.42383492e+00 2.10507005e-01 1.25515461e+00 1.49981245e-01 7.17721641e-01 -2.61847556e-01 2.24412069e-01 -9.58299413e-02 -3.34157944e-01 2.05848932e-01 1.75274658e+00 6.14540637e-01 -2.59348273e-01 -3.47027183e-01 7.64020681e-01 -4.52311128e-01 -2.76819825e-01 -6.84002042e-02 -1.39924550e+00 3.91780525e-01 4.64336336e-01 2.45786875e-01 -7.18219936e-01 2.08647400e-01 -6.81794211e-02 2.41069108e-01 3.76999348e-01 7.65334308e-01 -1.50264785e-01 -8.69237334e-02 -9.71732974e-01 5.15571654e-01 8.57417762e-01 7.47838557e-01 5.12897134e-01 6.85448125e-02 2.51898497e-01 2.92678084e-02 7.35075533e-01 1.13735485e+00 2.66739458e-01 -7.99686134e-01 3.87186438e-01 2.52654046e-01 9.30701613e-01 -6.39423072e-01 -6.72516704e-01 -7.13171184e-01 -5.48513949e-01 4.36761677e-01 3.04557085e-01 -3.37813824e-01 -5.83048344e-01 1.25402176e+00 3.83438319e-01 -1.33029252e-01 3.87182683e-01 9.10768926e-01 -2.70287693e-01 8.86618137e-01 6.02467842e-02 -6.76608086e-01 1.25930583e+00 -4.16524678e-01 -7.44262874e-01 -5.76100826e-01 8.62549663e-01 -6.71700239e-01 6.00928545e-01 -1.84256017e-01 -1.34127247e+00 5.60045123e-01 -9.74847436e-01 5.02099633e-01 -4.86026295e-02 -4.90134448e-01 7.97921956e-01 4.35946584e-01 -8.56176198e-01 5.38015127e-01 -1.14511383e+00 -5.12516260e-01 3.46045047e-01 -1.17514161e-02 1.70587629e-01 4.72823054e-01 -1.29264021e+00 1.34005892e+00 6.27886057e-02 -3.87006663e-02 -7.19333470e-01 -7.47085094e-01 -2.79855281e-01 1.70979485e-01 2.03563254e-02 -1.11615694e+00 1.05482471e+00 -4.07083273e-01 -6.86238885e-01 5.05467713e-01 4.42344636e-01 -1.10544598e+00 6.13652170e-01 -2.43118703e-01 -4.77045685e-01 -1.40395403e-01 4.16981608e-01 3.18621665e-01 1.89075932e-01 -1.10222840e+00 -4.01215494e-01 -9.39569399e-02 -2.06037000e-01 6.29100859e-01 -2.67838873e-02 3.34546298e-01 4.00756896e-01 -7.48937607e-01 2.42688209e-01 -1.49880016e+00 -1.99726298e-01 -3.15439403e-02 -2.43444622e-01 1.01331688e-01 4.08374399e-01 -5.25473714e-01 1.29333103e+00 -1.77812779e+00 -1.78627148e-01 -4.20606174e-02 5.54060750e-02 -5.66421151e-02 6.34459496e-01 5.49021423e-01 1.38769940e-01 6.07803285e-01 -1.70654804e-01 2.22334638e-01 2.37042569e-02 1.87677607e-01 -3.20451647e-01 6.77501678e-01 -1.15054794e-01 6.10000253e-01 -7.31000543e-01 2.82482244e-03 -2.40161538e-01 -5.01341701e-01 -4.21390176e-01 -3.59645456e-01 -2.84536690e-01 3.43563370e-02 -7.10384697e-02 8.62555861e-01 9.75789309e-01 -2.23156452e-01 -1.26330033e-01 6.14626050e-01 -1.16323590e+00 3.39297295e-01 -7.30587721e-01 7.06140161e-01 -4.36837703e-01 3.99140775e-01 3.43254387e-01 2.50352342e-02 2.78420657e-01 -4.34552096e-02 2.63209492e-01 -4.34880465e-01 1.91060215e-01 6.30385876e-01 4.83816117e-01 -3.16042751e-01 1.04238784e+00 -7.82464623e-01 -1.02313422e-01 8.65314066e-01 -5.31292737e-01 -2.52239376e-01 -1.62803814e-01 5.82580149e-01 9.95268941e-01 -5.07535040e-01 2.05058962e-01 -1.25713289e+00 -3.88528109e-01 4.32922363e-01 5.71238220e-01 2.38593787e-01 -9.99926552e-02 -2.93472290e-01 4.54056114e-01 -3.86319727e-01 -1.28903759e+00 -1.19603598e+00 -4.12352920e-01 3.90700549e-02 5.24886072e-01 -2.46402651e-01 -2.69023538e-01 1.66978359e-01 6.62568629e-01 9.65543330e-01 -2.62720555e-01 -2.73572415e-01 -2.18393713e-01 -1.25106299e+00 3.50339681e-01 6.89661726e-02 4.10706997e-01 -4.75100428e-01 -7.40250945e-01 -4.28031571e-03 -1.46792844e-01 -2.20919669e-01 -3.72977108e-02 9.55745485e-03 -1.03236163e+00 -7.12446451e-01 -9.34475183e-01 3.05837274e-01 7.11678743e-01 7.42675006e-01 6.87109709e-01 3.06854397e-01 -2.48346373e-01 1.19475819e-01 3.13204348e-01 -9.40907776e-01 -4.11936104e-01 -2.96835840e-01 9.73291397e-01 -6.01713061e-01 -2.87662655e-01 -3.47299784e-01 -9.94591475e-01 5.49469054e-01 -5.12146592e-01 -6.61259368e-02 5.66250563e-01 2.30061516e-01 2.65131980e-01 2.54999608e-01 9.19162512e-01 9.46179405e-02 1.01412974e-01 -1.40858829e+00 -7.57052720e-01 -1.09454729e-01 -9.73509729e-01 1.71217784e-01 -1.32334948e-01 8.13153572e-04 -1.28084278e+00 -5.52918494e-01 5.96215546e-01 -2.84908324e-01 5.48291922e-01 8.30276012e-01 1.16693690e-01 1.49424048e-02 8.08458805e-01 -2.59434342e-01 4.13101345e-01 -3.94418269e-01 3.21671277e-01 6.52309179e-01 3.79744589e-01 -4.76066321e-01 1.19367266e+00 9.03025985e-01 6.57093972e-02 -9.51401055e-01 -6.49521723e-02 -1.46940932e-01 -2.38283142e-01 -7.31332302e-01 6.32449567e-01 -1.58745384e+00 -4.11011547e-01 5.73858142e-01 -8.93349349e-01 -4.63050336e-01 -5.14520824e-01 8.19857836e-01 -4.67086941e-01 2.08816051e-01 -1.91864625e-01 -1.23061275e+00 -9.44959559e-03 -5.61284304e-01 2.37461165e-01 4.07845676e-01 -3.32634538e-01 -8.30169141e-01 1.62707105e-01 -1.05826348e-01 6.50021434e-01 8.75637382e-02 5.60505271e-01 4.87719430e-03 -1.38679254e+00 1.94167793e-01 3.90139446e-02 -4.03698832e-01 2.01855779e-01 -2.24500913e-02 -1.15710184e-01 -4.63757008e-01 1.98202562e-02 4.31742489e-01 7.70953596e-01 3.97119403e-01 -4.83425744e-02 -9.95582521e-01 -9.96908069e-01 3.98595750e-01 1.29533315e+00 9.00018141e-02 7.03321695e-01 1.02755225e+00 -5.86941419e-03 3.38404417e-01 1.11802757e+00 7.69026220e-01 3.92822087e-01 1.75237492e-01 6.38326705e-01 1.24783933e-01 -4.51464839e-02 -8.56694207e-02 3.57015043e-01 6.94076002e-01 -2.86851108e-01 -1.22477703e-01 -1.44838011e+00 8.95688176e-01 -1.20497489e+00 -1.02318072e+00 -6.48494124e-01 2.67204785e+00 4.99277741e-01 4.51206386e-01 1.86697826e-01 -5.54517686e-01 5.62130392e-01 -7.71337822e-02 -4.35520291e-01 -9.59305987e-02 -2.62261838e-01 -9.08142745e-01 1.57173264e+00 5.00065923e-01 -2.24831149e-01 5.76092422e-01 7.30166245e+00 1.21425994e-01 -9.69621122e-01 2.20609158e-01 3.41058701e-01 -4.94329453e-01 -1.07150149e+00 5.57101488e-01 -8.43142211e-01 7.33845830e-01 1.38130558e+00 -1.31519914e+00 1.76472127e-01 3.44714552e-01 7.03616738e-01 -8.61631632e-01 -1.30023211e-01 3.78683418e-01 -4.98462766e-01 -1.43777502e+00 -8.29745680e-02 6.85406208e-01 7.37106144e-01 6.07215166e-01 -1.05521232e-01 -7.40040094e-02 6.24233544e-01 -5.15621543e-01 1.08060718e+00 9.65538442e-01 8.65177214e-01 -6.81902289e-01 2.72658408e-01 6.08963490e-01 -8.60026360e-01 -4.09503162e-01 -3.81618798e-01 -5.02231061e-01 8.42100561e-01 1.01971030e+00 -7.64915347e-01 5.15765965e-01 5.62857389e-01 1.15814500e-01 -9.72009897e-02 1.27426016e+00 1.94039077e-01 3.85834068e-01 -9.68335211e-01 2.48806253e-01 1.39562830e-01 -4.39253122e-01 1.32839441e+00 1.85280174e-01 1.01318038e+00 3.00807565e-01 -5.49339473e-01 8.16970766e-01 5.48330434e-02 -2.44684890e-01 -1.13110185e+00 -3.82826954e-01 9.12876904e-01 7.24824369e-01 -1.01159132e+00 -2.07389906e-01 -3.65756780e-01 3.83938283e-01 -4.93265927e-01 -1.37192726e-01 -6.80376112e-01 -1.41772153e-02 1.16434860e+00 1.08505440e+00 -3.87144268e-01 -6.03878438e-01 -6.76252961e-01 -1.23223448e+00 -3.38553876e-01 -8.19812939e-02 -3.65631282e-02 -9.97842610e-01 -5.21728337e-01 1.26192868e-01 1.58110604e-01 -1.41640425e+00 4.42188606e-02 -1.19182989e-02 -9.02980387e-01 9.84201908e-01 -1.11414504e+00 -7.27911532e-01 3.61730129e-01 -2.71725744e-01 2.95055270e-01 -1.03279568e-01 1.30957678e-01 -1.06509179e-01 -4.17473227e-01 -6.77921548e-02 8.61425579e-01 -9.77483571e-01 4.58531708e-01 -7.52992630e-01 1.14097428e+00 9.25829649e-01 -4.54852164e-01 6.82445288e-01 1.11886680e+00 -1.84906125e+00 -1.55145991e+00 -1.05549037e+00 6.70503855e-01 -4.66078490e-01 1.19031370e+00 -6.93584755e-02 -7.45026290e-01 9.62353766e-01 8.31435844e-02 -3.49451452e-01 3.16923410e-01 4.79250923e-02 3.65148544e-01 3.65194440e-01 -8.40397000e-01 7.65031338e-01 1.04723585e+00 -5.97973987e-02 -6.10246599e-01 6.08097672e-01 6.35743082e-01 -8.16605464e-02 -7.20175087e-01 3.04967523e-01 6.83965027e-01 -4.21443075e-01 7.84620762e-01 -3.73592854e-01 -6.78300932e-02 -3.56063783e-01 2.05049127e-01 -1.59582436e+00 -5.01995802e-01 -1.20293009e+00 -4.89676138e-03 9.72395301e-01 6.22925758e-01 -8.75425458e-01 2.08680511e-01 7.40065336e-01 -5.13559759e-01 -3.48038733e-01 -1.29912436e+00 -1.60092771e+00 6.11140072e-01 1.50885761e-01 7.10949302e-01 7.48859167e-01 3.43479991e-01 -2.73039848e-01 -1.63118318e-01 6.16442680e-01 8.26151073e-01 -3.08235258e-01 4.95291352e-01 -1.50741100e+00 2.27239549e-01 -3.45480353e-01 -1.99297699e-03 -5.39819539e-01 -3.74767333e-01 -9.32932079e-01 -1.26255214e-01 -1.68403006e+00 -5.62649071e-02 -1.16803193e+00 2.10653856e-01 1.03466541e-01 4.90054548e-01 -2.52283186e-01 5.56833863e-01 1.12017059e+00 8.71845521e-03 6.91459000e-01 1.00125754e+00 2.06769645e-01 -2.63039798e-01 -3.20466273e-02 -8.15766454e-01 9.21577156e-01 6.94973648e-01 -3.65529120e-01 -7.41507439e-03 -7.65846074e-02 1.09872603e+00 5.56988716e-01 4.26638842e-01 -1.30186987e+00 3.11555117e-02 -5.10662079e-01 -6.08104803e-02 -8.30431461e-01 1.16490992e-02 -5.76909244e-01 1.19705999e+00 1.02110267e+00 6.70956910e-01 1.08149257e-02 5.66933036e-01 8.08215439e-01 6.73820257e-01 -1.37664765e-01 8.92207682e-01 -7.70332990e-03 1.99772105e-01 7.53034130e-02 -1.18708551e+00 -3.84524643e-01 1.53671956e+00 -5.58130816e-02 -7.04671264e-01 -6.98218420e-02 -9.74786639e-01 6.99421346e-01 1.36402810e+00 4.19530898e-01 -1.71479702e-01 -1.42473066e+00 -3.47735107e-01 -1.26057431e-01 -3.01347464e-01 -3.28593016e-01 3.68022382e-01 9.92973208e-01 -8.49177420e-01 5.44031441e-01 -4.19903934e-01 -4.51503471e-02 -5.56892514e-01 6.38260305e-01 5.87391853e-01 6.48853183e-02 -9.06665981e-01 7.34694123e-01 2.88069457e-01 3.51807475e-01 -7.30241418e-01 -4.45467800e-01 3.23749155e-01 4.70647663e-01 5.21261036e-01 5.49645543e-01 -9.88850296e-02 -7.45858908e-01 -3.01097274e-01 -8.46463367e-02 1.47059634e-01 -2.69842923e-01 1.06632805e+00 -6.69376194e-01 -1.91786632e-01 7.51849711e-01 1.39529720e-01 2.68422365e-01 -1.06883192e+00 2.87682027e-01 -1.28445536e-01 -5.57557106e-01 1.02526858e-01 -7.02258289e-01 -4.99259979e-01 1.02030195e-01 2.35871762e-01 4.49715704e-01 2.86827892e-01 2.79351532e-01 7.90535212e-01 2.23054811e-01 8.82397354e-01 -1.05725503e+00 -4.88237858e-01 2.98808515e-01 1.39855754e+00 -4.84425455e-01 5.04420221e-01 -2.78960258e-01 -3.16426903e-01 2.73790359e-01 3.32517892e-01 -7.76803419e-02 1.10503995e+00 4.03016388e-01 -6.63656235e-01 -3.72612834e-01 -1.02876210e+00 -1.76408187e-01 -6.57529712e-01 1.22297041e-01 -3.26865435e-01 6.56005323e-01 -1.04842293e+00 6.07842624e-01 -4.16312635e-01 -2.30976522e-01 1.35482407e+00 9.25604701e-01 -1.18405604e+00 -6.47138417e-01 -1.12127030e+00 1.09355175e+00 -4.15235281e-01 -2.00097799e-01 -6.61122426e-03 7.80415297e-01 -3.24162468e-02 2.36030370e-01 6.91962540e-01 2.80738592e-01 -2.00578555e-01 1.17433563e-01 1.34870455e-01 -4.36689377e-01 -8.13309029e-02 -5.44641465e-02 4.46358472e-01 -1.50820047e-01 6.14785179e-02 -1.48290908e+00 -1.61613989e+00 -8.13383937e-01 -8.49509120e-01 2.53850549e-01 8.03765893e-01 8.65078926e-01 5.80955565e-01 8.56242031e-02 5.26513278e-01 -9.00010467e-01 -5.40381789e-01 -9.30176079e-01 -1.14717758e+00 -5.54670870e-01 2.47815356e-01 -1.31166196e+00 -9.48930323e-01 -1.84583738e-01]
[6.100048542022705, 3.634866237640381]
b01e6b29-e39a-4318-b439-fcdfd7407e67
teacher-student-training-and-triplet-loss-to
2111.10561
null
https://arxiv.org/abs/2111.10561v1
https://arxiv.org/pdf/2111.10561v1.pdf
Teacher-Student Training and Triplet Loss to Reduce the Effect of Drastic Face Occlusion
We study a series of recognition tasks in two realistic scenarios requiring the analysis of faces under strong occlusion. On the one hand, we aim to recognize facial expressions of people wearing Virtual Reality (VR) headsets. On the other hand, we aim to estimate the age and identify the gender of people wearing surgical masks. For all these tasks, the common ground is that half of the face is occluded. In this challenging setting, we show that convolutional neural networks (CNNs) trained on fully-visible faces exhibit very low performance levels. While fine-tuning the deep learning models on occluded faces is extremely useful, we show that additional performance gains can be obtained by distilling knowledge from models trained on fully-visible faces. To this end, we study two knowledge distillation methods, one based on teacher-student training and one based on triplet loss. Our main contribution consists in a novel approach for knowledge distillation based on triplet loss, which generalizes across models and tasks. Furthermore, we consider combining distilled models learned through conventional teacher-student training or through our novel teacher-student training based on triplet loss. We provide empirical evidence showing that, in most cases, both individual and combined knowledge distillation methods bring statistically significant performance improvements. We conduct experiments with three different neural models (VGG-f, VGG-face, ResNet-50) on various tasks (facial expression recognition, gender recognition, age estimation), showing consistent improvements regardless of the model or task.
['Radu Tudor Ionescu', 'Georgian Duta', 'Mariana-Iuliana Georgescu']
2021-11-20
null
null
null
null
['age-estimation', 'age-estimation']
['computer-vision', 'miscellaneous']
[ 1.16435841e-01 5.79566002e-01 -1.13468945e-01 -5.55713952e-01 -5.66692472e-01 -3.22066694e-01 3.80436599e-01 -3.11475426e-01 -4.88360375e-01 7.57765353e-01 -1.05800167e-01 -1.06616959e-01 -1.14566483e-01 -3.62090409e-01 -9.47477520e-01 -6.56235576e-01 -5.84577508e-02 3.81787926e-01 -4.42674756e-01 -2.79344711e-02 -1.63532913e-01 6.91702604e-01 -1.70446193e+00 1.38829142e-01 6.41475141e-01 1.44754946e+00 -3.27739686e-01 5.64860106e-01 9.16492417e-02 7.55589724e-01 -4.99314517e-01 -8.34968269e-01 3.41388017e-01 5.85835800e-02 -7.37995565e-01 -4.95824851e-02 9.50049222e-01 -3.73703659e-01 -3.25683028e-01 8.24957967e-01 7.32720912e-01 -6.07595406e-02 6.38169885e-01 -1.38488829e+00 -4.09399152e-01 2.85118550e-01 -6.67501271e-01 -9.49586704e-02 3.67008269e-01 1.32989231e-02 4.07288522e-01 -7.95495629e-01 5.85710824e-01 1.36738622e+00 8.53133798e-01 1.14557815e+00 -1.12777603e+00 -9.78086650e-01 2.91141480e-01 1.05326824e-01 -1.56334567e+00 -7.01480150e-01 5.97364366e-01 -6.01512134e-01 6.41135037e-01 2.17889175e-01 3.94653589e-01 1.52008009e+00 -1.01464115e-01 6.71635449e-01 1.30407023e+00 -1.97957993e-01 1.99232865e-02 6.35314226e-01 -5.09995669e-02 9.89212990e-01 -2.78386623e-02 2.21418411e-01 -5.09160161e-01 -2.73708463e-01 7.71751940e-01 -1.94974944e-01 -4.43246692e-01 -4.44471687e-01 -5.70176661e-01 7.57056653e-01 2.47251064e-01 2.25355655e-01 -3.43495011e-01 2.60354370e-01 4.59153384e-01 2.88818061e-01 6.82415009e-01 2.61330694e-01 -6.55584574e-01 6.46312535e-02 -1.00398684e+00 1.34039611e-01 9.50863481e-01 7.01155841e-01 6.61503553e-01 1.57894343e-01 -4.33936387e-01 7.83605516e-01 4.84121591e-02 2.93230206e-01 3.49521488e-01 -9.28164661e-01 -1.08259752e-01 2.12014392e-01 -1.16616257e-01 -7.17143297e-01 -5.61017394e-01 -4.55874622e-01 -7.03846157e-01 4.31466937e-01 3.66739541e-01 -4.72578794e-01 -1.35890639e+00 2.28953695e+00 3.26244295e-01 4.81706589e-01 -1.72241762e-01 6.12616241e-01 1.22752380e+00 -4.66656946e-02 2.34820068e-01 -2.49157414e-01 1.51280582e+00 -6.76572502e-01 -6.35453939e-01 -3.48295248e-03 5.96938193e-01 -3.45932126e-01 7.04948068e-01 6.33716047e-01 -1.09900129e+00 -4.56503391e-01 -7.86777914e-01 -3.66312638e-03 -3.70536715e-01 5.27147889e-01 9.22188103e-01 1.11691058e+00 -1.41302752e+00 6.65251076e-01 -7.26643145e-01 -2.95556754e-01 8.35791528e-01 1.01318514e+00 -6.81919217e-01 1.34332985e-01 -8.59659910e-01 8.07482421e-01 -1.10169038e-01 2.24720448e-01 -1.06134582e+00 -1.16101170e+00 -9.34872448e-01 1.52890652e-01 3.26570064e-01 -1.05154896e+00 1.09546947e+00 -1.49600935e+00 -1.68342483e+00 1.45215857e+00 -1.95023417e-01 -4.13648337e-01 8.60287607e-01 -1.39445305e-01 -1.48463681e-01 3.10934447e-02 -4.80024219e-01 8.51889908e-01 1.10293698e+00 -1.29732776e+00 -2.05634266e-01 -7.96707928e-01 1.62832931e-01 -1.15614668e-01 -5.24984539e-01 1.22283682e-01 -2.40683541e-01 -3.77245873e-01 -3.97154301e-01 -1.00059438e+00 6.85051840e-04 4.14936483e-01 -3.78638327e-01 -1.84312135e-01 6.80260718e-01 -8.66426289e-01 7.56983340e-01 -2.23578382e+00 1.97397023e-01 3.41865122e-01 5.58612406e-01 3.70589137e-01 -1.81965217e-01 -2.49141648e-01 -4.00250226e-01 -8.83132889e-05 -7.57701620e-02 -7.71566212e-01 -1.18900299e-01 1.80329859e-01 -2.74206817e-01 4.58504498e-01 1.72484875e-01 8.45557511e-01 -4.39692557e-01 -3.49183053e-01 -7.12632760e-02 1.01891279e+00 -6.99949622e-01 1.57576323e-01 1.12242222e-01 7.78513670e-01 -1.56623513e-01 5.69615543e-01 8.95445466e-01 3.89776984e-03 1.79360002e-01 -2.40278840e-01 4.37747240e-01 -3.52497339e-01 -9.10310149e-01 1.60580492e+00 -8.83439660e-01 5.19510686e-01 4.94318306e-01 -1.08285284e+00 8.62123787e-01 5.98144650e-01 3.88708949e-01 -4.54704672e-01 4.34751540e-01 8.71384889e-02 -2.38943413e-01 -4.73086536e-01 1.90820798e-01 -4.69988436e-01 3.00983101e-01 -9.95824188e-02 4.00135130e-01 1.47346988e-01 -2.05713779e-01 -1.57911152e-01 9.04229760e-01 5.95784225e-02 2.40618060e-03 -8.89424831e-02 5.86675704e-01 -8.15075636e-01 5.94665289e-01 4.34606105e-01 -2.66029894e-01 5.59672117e-01 9.09824491e-01 -3.84341270e-01 -4.68229085e-01 -1.10363996e+00 -2.45904759e-01 1.12155628e+00 -3.30500364e-01 -1.34335384e-01 -8.47938538e-01 -9.42032039e-01 1.89856574e-01 3.72315586e-01 -1.30050933e+00 -3.78939718e-01 -4.13688570e-01 -6.56692743e-01 6.10165834e-01 7.96161413e-01 1.77270278e-01 -7.39039719e-01 -3.33593339e-01 -4.76027727e-01 1.25487477e-01 -1.45417166e+00 -1.70020685e-02 -4.48023602e-02 -6.91338956e-01 -9.92365420e-01 -1.06878233e+00 -6.18176460e-01 8.92363787e-01 -4.02136475e-01 1.08919787e+00 1.46605611e-01 -4.70598400e-01 8.26062262e-01 7.95358792e-02 -6.86401069e-01 9.73583572e-03 1.29152136e-02 3.27447474e-01 3.74890715e-01 2.82084584e-01 -8.61773133e-01 -5.49175620e-01 1.62631333e-01 -5.70873976e-01 -2.84407467e-01 5.84249556e-01 6.85546398e-01 3.66430432e-01 -5.40306628e-01 4.47895795e-01 -1.09321499e+00 3.12209308e-01 -3.27592731e-01 -5.74977338e-01 2.82136291e-01 -2.70958483e-01 -7.93407485e-02 3.74911398e-01 -6.39346600e-01 -1.10925138e+00 2.81364053e-01 -4.39150631e-01 -1.10533237e+00 -1.65636390e-01 1.27028018e-01 -1.06380142e-01 -5.36335051e-01 6.35957956e-01 -2.43483726e-02 1.83049336e-01 -4.53003019e-01 1.82334155e-01 4.31356996e-01 5.38819373e-01 -7.53388941e-01 6.32409573e-01 7.02832699e-01 6.48578554e-02 -8.90252888e-01 -7.50914872e-01 5.94877712e-02 -3.89487565e-01 -1.08667277e-01 8.76005173e-01 -1.02365530e+00 -1.29137099e+00 4.31579381e-01 -1.00135374e+00 -5.00928521e-01 -3.01702082e-01 4.90222514e-01 -7.56253183e-01 8.69341418e-02 -4.28225189e-01 -9.84417975e-01 -3.61348927e-01 -1.14640546e+00 1.29943848e+00 2.53226131e-01 -9.37179103e-02 -1.05046833e+00 -9.50431004e-02 4.04083908e-01 4.60001498e-01 5.51523805e-01 7.57882833e-01 -7.26786494e-01 -2.83073753e-01 -2.08073303e-01 -2.59247899e-01 4.63979989e-01 -3.48314047e-02 -5.71435951e-02 -1.62802362e+00 -4.48076516e-01 -7.28459805e-02 -6.93352461e-01 1.13728976e+00 6.97241485e-01 1.80741274e+00 -5.08577079e-02 -4.27251279e-01 9.87135887e-01 1.03362143e+00 -1.35009944e-01 7.80822575e-01 -4.43223000e-01 6.98417664e-01 8.84463072e-01 5.21993972e-02 3.38669926e-01 3.17998022e-01 7.76217818e-01 3.21927130e-01 -3.71863693e-01 -1.78761289e-01 2.72155609e-02 1.89277515e-01 1.21506371e-01 -4.75498348e-01 3.26138109e-01 -6.39961898e-01 4.01899219e-01 -1.36075675e+00 -5.58595717e-01 4.69646275e-01 2.26589274e+00 6.86683178e-01 -2.13018328e-01 2.61863977e-01 -2.04114500e-03 5.11000872e-01 -4.92159948e-02 -5.37622392e-01 -5.44381738e-01 1.60704032e-02 6.87932014e-01 2.54742146e-01 4.38493609e-01 -9.98884320e-01 6.85116172e-01 6.12121534e+00 8.28358769e-01 -1.52227056e+00 3.18201095e-01 1.00072658e+00 -4.14307296e-01 1.19415261e-02 -4.70724523e-01 -8.07924807e-01 1.04459472e-01 9.41439509e-01 6.83286935e-02 2.71192163e-01 9.48831439e-01 -2.45169878e-01 8.86298940e-02 -1.41788363e+00 1.30377460e+00 2.62248009e-01 -1.00309253e+00 -1.05894744e-01 1.90490648e-01 6.36202157e-01 -4.05258238e-01 5.59096932e-01 6.50372148e-01 1.43033545e-02 -1.53162920e+00 3.68501276e-01 6.62033319e-01 1.27086413e+00 -6.63663745e-01 7.05689788e-01 -1.64096281e-01 -8.53089690e-01 -8.93411785e-02 -1.08966827e-01 2.20462129e-01 -1.83001041e-01 5.49441218e-01 -7.41446376e-01 5.11836886e-01 5.98857164e-01 3.80382478e-01 -3.81991714e-01 7.25069702e-01 -1.64576501e-01 3.00412506e-01 -1.60915047e-01 4.76299912e-01 -3.23272854e-01 2.74603456e-01 2.57550359e-01 1.05552590e+00 2.05006570e-01 7.09091425e-02 -1.50951028e-01 8.87840688e-01 -3.83119076e-01 7.96600655e-02 -5.62174559e-01 1.81129709e-01 -4.60066646e-03 1.49663889e+00 -3.02002043e-01 -1.43476248e-01 -3.19215029e-01 9.64428186e-01 5.46088934e-01 4.48173285e-01 -7.39490271e-01 -2.55320132e-01 9.78402197e-01 2.28536353e-01 3.77291143e-01 1.68586403e-01 -9.62332934e-02 -1.02546358e+00 1.20703764e-01 -7.33027279e-01 2.75147289e-01 -5.41576862e-01 -1.21646142e+00 7.37110198e-01 -7.42283538e-02 -6.62909806e-01 -3.70542765e-01 -9.32575285e-01 -4.38829839e-01 7.90767908e-01 -1.67593658e+00 -1.42919385e+00 -5.19132793e-01 7.58078873e-01 -1.81916449e-02 -9.19535905e-02 9.30567324e-01 5.24213195e-01 -6.42739952e-01 1.33548450e+00 -3.46278846e-01 2.41219461e-01 6.78917348e-01 -1.02746546e+00 -1.56731024e-01 1.75774157e-01 -4.02256884e-02 5.70746720e-01 5.41024745e-01 -2.65555382e-01 -1.29294670e+00 -9.34664190e-01 5.26821077e-01 -6.30843461e-01 1.79412335e-01 -7.30685830e-01 -8.96211445e-01 9.34021115e-01 -1.34721324e-01 5.01756310e-01 8.18112493e-01 5.60975373e-01 -6.12935543e-01 -2.60695845e-01 -1.47345757e+00 1.99916422e-01 1.13719916e+00 -6.29873335e-01 -1.11558037e-02 3.18149507e-01 4.42697406e-01 -6.35611176e-01 -1.01064217e+00 8.07671785e-01 8.52128625e-01 -9.24552441e-01 9.57515895e-01 -8.62055361e-01 4.76370722e-01 3.62829685e-01 -2.79162340e-02 -1.24474955e+00 2.63411909e-01 -6.81768596e-01 -1.86395392e-01 1.01914883e+00 3.24312568e-01 -6.84113920e-01 1.22592676e+00 8.16993177e-01 2.26258889e-01 -1.13184547e+00 -1.10513139e+00 -7.62798965e-01 3.22390765e-01 -2.66557187e-01 3.28631401e-01 9.95153308e-01 -2.73823738e-01 7.53276050e-02 -4.92528170e-01 3.12250368e-02 3.51248056e-01 -2.06995249e-01 8.90313625e-01 -1.35015023e+00 -4.69179839e-01 -4.25885260e-01 -6.99869514e-01 -5.87194383e-01 7.19881177e-01 -7.94425607e-01 -3.58958095e-01 -1.12916565e+00 3.61917078e-01 -2.50941128e-01 -2.37421483e-01 6.33823097e-01 -4.24657986e-02 4.32538778e-01 3.15106139e-02 -4.80554223e-01 -2.48798534e-01 6.91305637e-01 9.44011807e-01 4.41641398e-02 -8.67456943e-02 1.77147493e-01 -7.25371957e-01 8.02409410e-01 3.81975949e-01 -2.77673811e-01 -3.38924825e-01 -2.70211875e-01 1.79597139e-01 6.13799877e-02 6.45757139e-01 -8.10750842e-01 6.63051903e-02 2.48148620e-01 5.54808319e-01 6.51443675e-02 8.59157205e-01 -5.69931865e-01 -7.78385922e-02 3.63712907e-01 -9.31530818e-02 -3.95904690e-01 6.77893758e-01 3.23823243e-01 -6.01874152e-03 3.00022960e-02 9.03090954e-01 3.16154137e-02 -3.93792510e-01 5.11025906e-01 -5.14396541e-02 5.63470460e-02 9.74948823e-01 -2.63900846e-01 -1.51148006e-01 -5.71863770e-01 -1.33476591e+00 1.53848212e-02 1.47130802e-01 3.92170727e-01 5.51115334e-01 -1.16209114e+00 -7.67229259e-01 3.56404930e-01 7.98472837e-02 -3.51394683e-01 5.89581490e-01 1.26277232e+00 -6.23969324e-02 3.30771863e-01 -3.40259612e-01 -4.52029198e-01 -1.65719974e+00 6.31846070e-01 6.98776424e-01 -2.16559470e-01 -6.22760095e-02 1.33039689e+00 9.01405811e-01 -6.54670358e-01 6.13818169e-01 -1.00820139e-01 -2.82407731e-01 7.32234940e-02 5.89747429e-01 2.12666959e-01 2.89608508e-01 -5.55934131e-01 -4.29049432e-01 7.66876519e-01 -2.24760309e-01 1.85177952e-01 1.21207678e+00 2.73664922e-01 9.31787938e-02 1.17109045e-02 1.34483266e+00 -1.92377456e-02 -1.06090069e+00 -1.63787365e-01 -4.36411202e-01 -5.18669844e-01 6.97764382e-02 -8.94362628e-01 -1.73284507e+00 1.04472613e+00 1.04324460e+00 -5.17862797e-01 1.41007864e+00 8.79739225e-02 3.73566300e-01 2.15464339e-01 3.05227250e-01 -7.49876916e-01 4.87179682e-02 7.90911317e-02 8.66828918e-01 -1.23079729e+00 -1.19645029e-01 -7.26475418e-01 -4.13607180e-01 8.16925943e-01 7.97192872e-01 8.48913193e-02 8.26087892e-01 2.67537087e-01 1.79995850e-01 -3.95050019e-01 -6.43050790e-01 -2.03321412e-01 4.23716307e-01 7.85396636e-01 4.26080793e-01 -1.55646838e-02 -1.44554228e-01 8.71966362e-01 -4.21659052e-01 2.97496438e-01 3.22163612e-01 6.21501029e-01 1.69590995e-01 -8.97172570e-01 -1.73911437e-01 5.30521035e-01 -8.47326100e-01 1.54542133e-01 -4.90434825e-01 7.12450147e-01 2.73754478e-01 4.97966260e-01 1.75644174e-01 -3.57181847e-01 3.79985183e-01 1.05932184e-01 1.07008016e+00 -4.42518175e-01 -6.56369328e-01 -4.33820963e-01 1.63249955e-01 -7.15937138e-01 -2.67952889e-01 -5.30271709e-01 -8.69795561e-01 -2.09109947e-01 -2.04027683e-01 -1.36288747e-01 8.29099178e-01 8.52935672e-01 3.61121088e-01 5.71158290e-01 4.98128653e-01 -9.45926189e-01 -5.36977291e-01 -7.83221066e-01 -7.55163491e-01 2.56732166e-01 4.71318692e-01 -1.07432127e+00 -3.75457168e-01 -1.95983410e-01]
[13.447114944458008, 1.1882466077804565]
00a90ccd-f88e-4a2b-93c9-afa1c65ebe8d
location-aware-feature-selection-for-scene
2004.10999
null
https://arxiv.org/abs/2004.10999v2
https://arxiv.org/pdf/2004.10999v2.pdf
Location-Aware Feature Selection Text Detection Network
Regression-based text detection methods have already achieved promising performances with simple network structure and high efficiency. However, they are behind in accuracy comparing with recent segmentation-based text detectors. In this work, we discover that one important reason to this case is that regression-based methods usually utilize a fixed feature selection way, i.e. selecting features in a single location or in neighbor regions, to predict components of the bounding box, such as the distances to the boundaries or the rotation angle. The features selected through this way sometimes are not the best choices for predicting every component of a text bounding box and thus degrade the accuracy performance. To address this issue, we propose a novel Location-Aware feature Selection text detection Network (LASNet). LASNet selects suitable features from different locations to separately predict the five components of a bounding box and gets the final bounding box through the combination of these components. Specifically, instead of using the classification score map to select one feature for predicting the whole bounding box as most of the existing methods did, the proposed LASNet first learn five new confidence score maps to indicate the prediction accuracy of the bounding box components, respectively. Then, a Location-Aware Feature Selection mechanism (LAFS) is designed to weightily fuse the top-$K$ prediction results for each component according to their confidence score, and to combine the all five fused components into a final bounding box. As a result, LASNet predicts the more accurate bounding boxes by using a learnable feature selection way. The experimental results demonstrate that our LASNet achieves state-of-the-art performance with single-model and single-scale testing, outperforming all existing regression-based detectors.
['Haojie Li', 'Wanli Ouyang', 'Zengyuan Guo', 'Wen Gao', 'Zilin Wang', 'Zhihui Wang']
2020-04-23
null
null
null
null
['scene-text-detection']
['computer-vision']
[-2.92019427e-01 -5.16431391e-01 -3.23674619e-01 -4.47354347e-01 -6.87369406e-01 -3.30262810e-01 3.00672352e-01 3.02846819e-01 -4.28949028e-01 4.61192280e-01 -2.53825784e-01 4.33448069e-02 -1.45193696e-01 -1.04761338e+00 -4.83299106e-01 -7.11870670e-01 3.27770263e-01 5.50303102e-01 1.02871013e+00 -1.53804913e-01 5.20028830e-01 4.84509438e-01 -1.53353190e+00 2.12422043e-01 1.17629743e+00 1.25416315e+00 1.66435346e-01 3.35586131e-01 -5.51606297e-01 2.55148977e-01 -6.37309372e-01 -6.13688640e-02 1.67336732e-01 -2.50486434e-01 -3.85123342e-01 -1.65590450e-01 2.52072990e-01 -4.35023457e-01 -2.78436691e-01 8.54115784e-01 1.88117161e-01 1.31558284e-01 6.89706981e-01 -9.47377563e-01 5.59768677e-02 7.16360748e-01 -8.43170166e-01 1.08531073e-01 3.19839448e-01 -1.72541067e-01 1.28238547e+00 -1.01289630e+00 4.01278973e-01 8.97676647e-01 7.92915821e-01 2.57485569e-01 -8.63938391e-01 -9.06037033e-01 6.28446877e-01 4.28168923e-02 -1.82831967e+00 -3.11978590e-02 8.48940849e-01 -2.33855411e-01 6.59680724e-01 3.42230678e-01 6.89556301e-01 4.67578769e-01 1.90903753e-01 1.12491632e+00 8.55301440e-01 -3.86525661e-01 9.97030511e-02 3.71631503e-01 4.36057091e-01 8.56511652e-01 3.55182469e-01 -3.85312676e-01 -6.65116131e-01 -1.58063456e-01 5.28958797e-01 2.94485420e-01 -1.63918495e-01 -2.06584811e-01 -1.12650681e+00 7.82926500e-01 8.37835371e-01 6.37057602e-01 -4.70034517e-02 -2.20248811e-02 1.45698339e-01 -2.36652195e-01 5.51111102e-01 2.52537787e-01 -5.11777163e-01 1.24629103e-01 -1.37651193e+00 3.12990665e-01 6.47960424e-01 7.17535555e-01 1.05198729e+00 -2.79388279e-01 -2.80481964e-01 9.08751428e-01 3.84977609e-01 2.75972784e-01 5.43193221e-01 7.23374188e-02 7.54726589e-01 1.34965432e+00 7.54755884e-02 -1.16313922e+00 -8.38285029e-01 -3.31753135e-01 -6.77643478e-01 -9.92796645e-02 4.59899217e-01 -2.05828965e-01 -1.08673978e+00 1.22509539e+00 5.25718451e-01 1.08420536e-01 -3.94265771e-01 9.21124518e-01 8.03915799e-01 8.08212161e-01 -8.04832131e-02 4.17987630e-02 1.30933928e+00 -8.17843437e-01 -3.02184075e-01 -3.42019469e-01 7.80535817e-01 -7.00716138e-01 7.43260741e-01 4.31367040e-01 -5.07224023e-01 -5.17221153e-01 -1.28361523e+00 3.10517520e-01 -6.99148417e-01 7.29383230e-01 4.67639953e-01 5.32668233e-01 -7.51465440e-01 5.43059111e-01 -7.08109736e-01 -3.29374254e-01 2.96194136e-01 3.89436990e-01 -1.63521674e-02 -1.65737681e-02 -1.06163359e+00 5.05682588e-01 5.18281341e-01 3.68422195e-02 -3.65828931e-01 -2.47445285e-01 -6.47550404e-01 1.65671200e-01 5.45682073e-01 -2.29645908e-01 7.77401686e-01 -6.54651105e-01 -1.03269351e+00 3.28599334e-01 -2.98561633e-01 -2.42347285e-01 5.98517179e-01 -1.65871665e-01 -3.44287813e-01 1.01094827e-01 2.11721182e-01 5.65289855e-01 1.04324687e+00 -1.06551170e+00 -1.21660948e+00 -3.62238973e-01 -3.34599793e-01 1.92784965e-01 -5.77993512e-01 -3.60861607e-02 -7.03177273e-01 -5.52403331e-01 7.57630587e-01 -6.49270892e-01 -1.30396619e-01 -1.43394306e-01 -7.78491855e-01 -6.30452275e-01 1.22612154e+00 -4.41488862e-01 1.68455040e+00 -2.08178639e+00 -1.41391635e-01 5.95008552e-01 3.96555692e-01 2.75172621e-01 3.32008451e-01 1.41615957e-01 3.43687743e-01 3.43499631e-01 1.34865707e-03 -1.96888462e-01 -1.56707928e-01 -3.48842829e-01 -1.38078570e-01 3.75197977e-01 2.30384603e-01 5.47076821e-01 -5.48874557e-01 -8.96145999e-01 4.94774997e-01 3.27191949e-01 -3.18866909e-01 -3.20622176e-02 -3.47958505e-01 4.19707485e-02 -1.05732858e+00 8.24364662e-01 7.31529057e-01 -1.80404827e-01 -2.44117770e-02 -6.65871426e-02 -2.82707065e-01 1.41616434e-01 -1.60037827e+00 1.10514867e+00 -1.06727146e-01 4.85950798e-01 -2.53571153e-01 -8.16373587e-01 1.57665658e+00 1.23492011e-03 5.41807353e-01 -2.98327893e-01 2.34445974e-01 4.67225283e-01 -1.65207207e-01 -1.08934842e-01 6.06104910e-01 2.76464760e-01 -2.74366051e-01 3.81935716e-01 -2.32847318e-01 -1.06348544e-01 1.95904970e-01 1.41128510e-01 9.94885147e-01 1.36744872e-01 3.04791868e-01 5.94315603e-02 7.37475514e-01 1.60407394e-01 6.38272345e-01 8.87068689e-01 -2.39022747e-01 6.80121005e-01 4.93559659e-01 -5.82349181e-01 -7.85712719e-01 -5.76567531e-01 -4.02181983e-01 1.14093423e+00 4.46308523e-01 -5.24144888e-01 -8.46636415e-01 -1.19217551e+00 6.61945641e-02 4.38829631e-01 -5.76326132e-01 -2.89403666e-02 -6.71108663e-01 -8.42896402e-01 3.82649720e-01 5.80259323e-01 9.34660494e-01 -8.59822750e-01 -3.85305673e-01 2.12474257e-01 -8.01877007e-02 -8.12671185e-01 -4.89845455e-01 3.89094561e-01 -7.89613962e-01 -1.00754571e+00 -6.20288253e-01 -7.56662965e-01 6.99073434e-01 5.12242436e-01 4.87125814e-01 5.06964982e-01 -2.36140996e-01 -3.45879287e-01 -7.35404968e-01 -1.09913640e-01 6.88032880e-02 5.03846109e-01 -1.09281890e-01 1.61050647e-01 5.64758360e-01 -7.61573715e-03 -6.75824046e-01 8.04640293e-01 -5.27594209e-01 -5.76312169e-02 6.42697871e-01 7.95167983e-01 8.10546219e-01 5.36532521e-01 5.52952051e-01 -7.20289409e-01 5.23266315e-01 -2.76016921e-01 -6.04084909e-01 2.46688962e-01 -6.05755985e-01 -1.19110718e-01 9.00617063e-01 -3.68032545e-01 -8.45552862e-01 3.51538777e-01 -5.19829802e-02 -1.16335973e-01 -2.20672131e-01 5.24877071e-01 -8.38233754e-02 -8.90340507e-02 5.98119438e-01 4.67599303e-01 -5.25592864e-01 -3.73754680e-01 -5.95539585e-02 8.25062037e-01 -1.36878997e-01 -3.42228144e-01 8.88605773e-01 4.35847729e-01 -1.79142579e-01 -5.86141646e-01 -8.36343169e-01 -8.60538661e-01 -9.54975724e-01 -2.23978281e-01 7.65465617e-01 -6.03299141e-01 -6.15911543e-01 5.94199896e-01 -9.83269155e-01 1.96171418e-01 1.46657109e-01 3.81563753e-01 -6.34927601e-02 2.17079207e-01 -3.72932941e-01 -9.10048246e-01 -3.92368823e-01 -1.25008380e+00 1.27994347e+00 6.83666945e-01 -8.38832185e-02 -4.65665638e-01 -1.44008800e-01 1.92095518e-01 1.80707127e-01 -1.03810079e-01 7.58695900e-01 -1.04636919e+00 -5.98674953e-01 -8.37473929e-01 -5.40535927e-01 -1.05452389e-01 -9.42427516e-02 3.48558486e-01 -8.10404778e-01 -1.55564025e-01 -5.02747834e-01 6.67710975e-02 1.21221101e+00 4.79054749e-01 1.15216088e+00 1.32784516e-01 -1.04043114e+00 5.20658433e-01 1.26207578e+00 3.13416421e-01 2.30249390e-01 4.66848433e-01 6.50007725e-01 3.09558809e-01 1.07383811e+00 5.96122444e-01 1.60543203e-01 7.01935172e-01 2.83638716e-01 -1.20399252e-01 1.80314466e-01 -3.08690667e-01 2.11219445e-01 3.97485137e-01 7.24148452e-02 -3.09410185e-01 -9.07292843e-01 3.06655079e-01 -2.00269794e+00 -5.98900855e-01 -1.47436261e-01 2.22232485e+00 4.60049957e-01 4.78031963e-01 3.08707595e-01 3.47733438e-01 1.01064467e+00 2.46556655e-01 -6.99345708e-01 -1.27385287e-02 -1.08262859e-01 5.37502170e-02 4.76896167e-01 6.35429770e-02 -1.39497042e+00 1.25656092e+00 5.56081915e+00 1.15715146e+00 -1.11360610e+00 -2.74413913e-01 5.84760308e-01 9.97388288e-02 9.55288038e-02 1.58803549e-03 -1.62320602e+00 3.03701639e-01 2.92243063e-01 2.01174334e-01 5.84709905e-02 1.06060314e+00 1.15680017e-01 -3.74412894e-01 -6.41130447e-01 6.22243822e-01 9.86056030e-02 -1.13845181e+00 3.34534496e-02 -1.82642147e-01 5.66314042e-01 -1.25619501e-01 -1.26913711e-01 3.46276194e-01 1.51032060e-01 -8.11599076e-01 6.15259588e-01 4.20069844e-01 4.36226487e-01 -8.25408399e-01 9.26409185e-01 5.82597554e-01 -1.71849322e+00 -2.50364363e-01 -4.61917073e-01 2.85680205e-01 -2.54382581e-01 6.50566518e-01 -1.11820698e+00 3.83514673e-01 9.94139194e-01 6.86015844e-01 -7.23729908e-01 1.34677339e+00 -2.94324696e-01 5.89686632e-01 -6.06288671e-01 -7.85301447e-01 3.74444962e-01 -5.29525355e-02 5.04154086e-01 1.10552180e+00 4.60049689e-01 -1.79756880e-01 3.71421874e-01 8.48376870e-01 -3.90889533e-02 6.18058741e-01 -2.93458164e-01 2.70191759e-01 7.32082427e-01 1.50016713e+00 -1.26102340e+00 -2.79860049e-01 -3.76245260e-01 6.16533637e-01 4.14397776e-01 8.99407268e-02 -8.69228005e-01 -8.66544485e-01 1.93650469e-01 3.42160910e-01 6.45873725e-01 -1.19209811e-01 -5.09881020e-01 -9.79700208e-01 1.51910886e-01 -3.18031371e-01 4.37990099e-01 -5.42917371e-01 -1.02557206e+00 6.30695283e-01 -1.89402863e-01 -1.44731939e+00 8.96785408e-02 -5.89820385e-01 -7.42446065e-01 7.57419229e-01 -1.34810984e+00 -1.19408298e+00 -4.92308855e-01 4.46152955e-01 6.50989532e-01 -1.11083128e-01 4.49134588e-01 -7.47207226e-03 -1.00360632e+00 7.66004920e-01 2.06962138e-01 4.16758150e-01 7.84588695e-01 -1.09319937e+00 1.99394882e-01 7.66199052e-01 1.89549893e-01 5.18650174e-01 2.32041851e-01 -8.72945905e-01 -9.89972293e-01 -1.01057398e+00 7.97690511e-01 -1.52707666e-01 4.95163649e-01 -4.08953428e-01 -1.00629604e+00 3.88399065e-01 -5.64508557e-01 8.38203281e-02 2.90392280e-01 1.94619313e-01 -2.83734620e-01 -3.56670797e-01 -1.18231368e+00 5.00326633e-01 6.51007593e-01 4.30817204e-03 -3.20273131e-01 2.17127189e-01 5.87898672e-01 -3.56167346e-01 -5.33745766e-01 4.15777534e-01 5.66290021e-01 -1.02611911e+00 6.75742447e-01 -7.46144801e-02 2.66136348e-01 -6.51429236e-01 7.19675347e-02 -1.10951555e+00 -2.42150575e-01 -1.33091420e-01 1.40370503e-01 1.28757036e+00 7.07020164e-01 -7.60075152e-01 1.11996365e+00 2.71324217e-01 3.74269905e-04 -9.87975895e-01 -1.13741708e+00 -2.72551984e-01 -8.66819397e-02 -4.75268126e-01 8.37919414e-01 6.95817173e-01 1.01568617e-01 1.47856668e-01 -2.63910890e-01 2.65966773e-01 1.68785140e-01 3.83278489e-01 7.84111440e-01 -1.55993152e+00 -2.49536671e-02 -6.57712340e-01 -3.97231787e-01 -1.33820522e+00 -5.28068878e-02 -6.70790672e-01 4.08602476e-01 -1.46833849e+00 3.14567566e-01 -1.00078952e+00 -3.25109303e-01 5.32132030e-01 -4.16135430e-01 1.78159788e-01 6.37520105e-02 4.34312552e-01 -7.59764016e-01 1.99098170e-01 1.11750329e+00 -1.19852103e-01 -6.42283022e-01 3.64285678e-01 -5.16076922e-01 8.59514832e-01 7.98529625e-01 -5.72008550e-01 1.26263231e-01 -2.13822676e-03 1.10148862e-01 -8.68228227e-02 8.04770514e-02 -1.16919553e+00 5.34859240e-01 -1.62627503e-01 8.61976206e-01 -1.11873329e+00 1.60832256e-01 -9.01038468e-01 -4.59432125e-01 3.34535986e-01 -2.58009247e-02 -5.81492364e-01 9.98270139e-02 5.87220967e-01 -1.30420789e-01 -4.50454712e-01 5.92968524e-01 1.63170040e-01 -7.07247555e-01 3.45187694e-01 -2.78152347e-01 -3.39941680e-01 1.16555393e+00 -6.05835617e-01 -4.17313695e-01 -5.17547550e-03 -4.55781937e-01 3.36093515e-01 2.69077510e-01 2.69497544e-01 7.13679373e-01 -9.33629155e-01 -5.22203207e-01 2.95890361e-01 1.91226482e-01 2.29753986e-01 1.39178857e-01 7.33368576e-01 -4.22237098e-01 5.03677249e-01 1.86687067e-01 -7.88330734e-01 -1.32240868e+00 1.86362043e-01 4.09822583e-01 -5.50116122e-01 -6.20983422e-01 8.92830133e-01 8.30753222e-02 -3.23635995e-01 1.38949439e-01 -3.98217022e-01 -5.24833560e-01 1.79025412e-01 4.62057441e-01 2.04759091e-01 1.40404701e-01 -6.61248446e-01 -5.14417529e-01 1.07498300e+00 -4.73458230e-01 2.77381092e-01 1.05141783e+00 9.23620388e-02 1.88800380e-01 2.99450487e-01 7.07018673e-01 1.59174964e-01 -1.10525906e+00 -3.23651791e-01 -1.57383636e-01 -4.34859514e-01 2.00879067e-01 -7.14872301e-01 -1.08519149e+00 7.17281163e-01 4.30717379e-01 3.71726751e-01 1.07695305e+00 7.78660774e-02 8.43922615e-01 3.47530425e-01 3.64246219e-01 -1.36912894e+00 7.58252339e-03 5.41962266e-01 4.28899378e-01 -1.27539778e+00 2.74639964e-01 -6.64673686e-01 -4.80556190e-01 1.34870338e+00 9.89491761e-01 -2.27293387e-01 8.56792808e-01 2.42113218e-01 -1.65088117e-01 -7.68804401e-02 -3.62297833e-01 -3.10403019e-01 4.71197456e-01 2.37378120e-01 4.59980398e-01 1.93355173e-01 -4.36127067e-01 8.29098463e-01 -1.17876068e-01 -3.63007337e-01 -6.20486140e-02 8.20318103e-01 -1.09418225e+00 -9.33514118e-01 -6.62633121e-01 9.58387554e-01 -2.04886392e-01 4.35998514e-02 -6.03970051e-01 9.69052672e-01 2.95891732e-01 9.29198503e-01 1.31919995e-01 -8.16389620e-01 3.48883510e-01 7.06608817e-02 -8.46501067e-02 -5.33032417e-01 -7.67100513e-01 2.48173371e-01 -2.07248256e-01 -3.13408881e-01 9.08490494e-02 -6.45005524e-01 -1.65244973e+00 -2.40376711e-01 -1.12063348e+00 2.01860830e-01 6.80202842e-01 1.02124286e+00 9.37363803e-02 4.42021161e-01 8.44528317e-01 -7.59371758e-01 -3.33383620e-01 -1.07136822e+00 -7.31739938e-01 -3.99413751e-03 3.26300524e-02 -7.99997032e-01 -5.47993481e-01 -5.53453982e-01]
[12.098637580871582, 2.3023080825805664]
8a18df9c-0d45-4836-a055-09a9af700a8f
bootstrapping-text-anonymization-models-with-1
2205.06895
null
https://arxiv.org/abs/2205.06895v1
https://arxiv.org/pdf/2205.06895v1.pdf
Bootstrapping Text Anonymization Models with Distant Supervision
We propose a novel method to bootstrap text anonymization models based on distant supervision. Instead of requiring manually labeled training data, the approach relies on a knowledge graph expressing the background information assumed to be publicly available about various individuals. This knowledge graph is employed to automatically annotate text documents including personal data about a subset of those individuals. More precisely, the method determines which text spans ought to be masked in order to guarantee $k$-anonymity, assuming an adversary with access to both the text documents and the background information expressed in the knowledge graph. The resulting collection of labeled documents is then used as training data to fine-tune a pre-trained language model for text anonymization. We illustrate this approach using a knowledge graph extracted from Wikidata and short biographical texts from Wikipedia. Evaluation results with a RoBERTa-based model and a manually annotated collection of 553 summaries showcase the potential of the approach, but also unveil a number of issues that may arise if the knowledge graph is noisy or incomplete. The results also illustrate that, contrary to most sequence labeling problems, the text anonymization task may admit several alternative solutions.
['Ildikó Pilán', 'Lilja Øvrelid', 'Pierre Lison', 'Anthi Papadopoulou']
2022-05-13
null
https://aclanthology.org/2022.lrec-1.476
https://aclanthology.org/2022.lrec-1.476.pdf
lrec-2022-6
['text-anonymization']
['natural-language-processing']
[ 2.92534292e-01 6.47940993e-01 -3.00369471e-01 -5.41661561e-01 -7.55185306e-01 -1.00533378e+00 6.66199148e-01 5.91487110e-01 -5.79005659e-01 1.09765697e+00 4.86693054e-01 -1.31085023e-01 -7.22453892e-02 -8.48360956e-01 -6.31251216e-01 -5.16605616e-01 1.76423013e-01 7.60906577e-01 -1.68519437e-01 -6.96406979e-03 6.47381023e-02 5.77670336e-01 -1.08159816e+00 3.30364347e-01 8.66500318e-01 4.51917619e-01 -4.28970307e-01 5.02795875e-01 -2.80975401e-01 7.24426091e-01 -8.60575974e-01 -1.24435532e+00 3.69516194e-01 -2.62207806e-01 -1.27338719e+00 2.58873522e-01 4.86853480e-01 -1.06800154e-01 -3.31151903e-01 1.18970978e+00 3.94151241e-01 -1.38833031e-01 7.18877375e-01 -1.36176944e+00 -6.54721975e-01 9.28309441e-01 -2.01993525e-01 3.58182788e-02 4.66209501e-01 -8.70959535e-02 1.02584291e+00 -1.44066766e-01 9.53853250e-01 9.40453112e-01 8.48389506e-01 5.34833610e-01 -1.50996614e+00 -3.84083360e-01 -9.02945697e-02 -1.67005092e-01 -1.51629138e+00 -4.94251490e-01 6.25033438e-01 -4.56394702e-01 4.77507085e-01 5.92089951e-01 3.10128957e-01 1.07309806e+00 -2.67037541e-01 6.06992364e-01 9.23692405e-01 -5.89793205e-01 2.55861908e-01 9.31187749e-01 5.15474021e-01 7.75059402e-01 7.51801729e-01 -3.33771735e-01 -5.34148395e-01 -8.62545371e-01 -6.78509241e-04 -5.56117117e-01 -4.90908891e-01 -8.08748305e-01 -8.93914223e-01 6.68915033e-01 -3.71635407e-02 3.25891435e-01 -6.26275130e-03 -3.03051293e-01 7.09727705e-01 3.63515437e-01 5.37310719e-01 5.44870913e-01 -3.90107155e-01 3.29150647e-01 -8.50936174e-01 3.08474988e-01 1.42439961e+00 9.98958349e-01 1.15824234e+00 -3.11158687e-01 -7.25576878e-02 5.26318908e-01 6.87637776e-02 3.16772789e-01 4.91991162e-01 -6.00351632e-01 8.72300804e-01 8.67512703e-01 5.34350097e-01 -1.25592399e+00 -1.37354448e-01 -1.12534547e-02 -5.47168553e-01 -6.41791597e-02 9.14739549e-01 -1.36129662e-01 -5.06902099e-01 1.57253182e+00 4.13120240e-01 -3.10172975e-01 3.12269539e-01 5.23904622e-01 5.23207068e-01 1.81810707e-01 8.79551247e-02 -6.77699819e-02 1.39054942e+00 -4.15948719e-01 -8.26537549e-01 -4.37747082e-03 1.10544372e+00 -3.83653253e-01 9.27279353e-01 2.21233726e-01 -6.12460613e-01 -3.46644297e-02 -9.95263696e-01 -8.94893035e-02 -9.13707316e-01 1.03445716e-01 5.76386929e-01 1.45550323e+00 -9.62343693e-01 5.46285689e-01 -6.37356699e-01 -7.23001957e-01 6.23721123e-01 4.41580176e-01 -8.80799532e-01 -2.95017082e-02 -1.28721011e+00 6.26567483e-01 8.17929924e-01 -1.34558484e-01 -4.74249452e-01 -3.20373446e-01 -8.98540735e-01 9.47997570e-02 3.48196298e-01 -6.90831661e-01 8.06549847e-01 -1.15981138e+00 -1.10805988e+00 1.36893606e+00 -1.16884358e-01 -6.85420930e-01 7.77198315e-01 -4.65299077e-02 -5.05893230e-01 1.73233181e-01 1.41491726e-01 2.41443180e-02 7.78749347e-01 -1.38554823e+00 -3.92953545e-01 -7.16103435e-01 9.13952589e-02 1.07329823e-01 -6.60495102e-01 1.64134298e-02 -3.22537541e-01 -7.39269078e-01 -3.53098601e-01 -9.71182525e-01 -2.46915534e-01 -2.29525283e-01 -1.08209753e+00 1.25264153e-01 8.27370644e-01 -1.01953268e+00 1.41516399e+00 -1.70787466e+00 1.20781558e-02 7.24558294e-01 1.27701283e-01 3.90835166e-01 1.16214760e-01 6.60370946e-01 -1.61074400e-01 6.71545327e-01 -5.06614804e-01 -4.39683646e-01 2.02732056e-01 1.82211265e-01 -5.16943514e-01 7.25342870e-01 -1.92500249e-01 6.98245108e-01 -8.71311903e-01 -4.89862055e-01 -6.32635206e-02 7.55729377e-02 -3.46081018e-01 -6.37938781e-03 -3.82750332e-01 3.33018631e-01 -6.40591681e-01 4.38194752e-01 5.29217720e-01 -3.52229103e-02 5.26399434e-01 8.56397375e-02 3.43948752e-01 1.13831781e-01 -1.33584726e+00 1.63432884e+00 1.07932212e-02 4.66247767e-01 1.24167651e-01 -1.03701532e+00 7.02354848e-01 4.28651571e-01 2.41045937e-01 -8.36930349e-02 1.03722364e-01 -6.19795658e-02 -6.25070333e-01 -6.13119781e-01 5.91368437e-01 2.02439353e-02 -3.60925108e-01 9.34781015e-01 5.42627089e-02 5.07676005e-02 1.57734796e-01 6.56323493e-01 1.06785750e+00 -2.91095860e-02 3.33122045e-01 -3.65849108e-01 8.86130750e-01 6.98679164e-02 4.73761231e-01 8.37415874e-01 -1.03311338e-01 2.35071659e-01 6.93108857e-01 -5.55462658e-01 -1.11106908e+00 -6.18704498e-01 1.29924426e-02 7.48180628e-01 -2.13630229e-01 -7.62253761e-01 -1.28512776e+00 -1.29515493e+00 1.73046812e-01 9.50846970e-01 -7.54054546e-01 -1.84300929e-01 -3.83214772e-01 -7.38351166e-01 1.05031896e+00 6.06455430e-02 2.74190515e-01 -7.53306091e-01 -1.39945552e-01 -1.20498754e-01 -4.13298577e-01 -1.01077223e+00 -5.50904751e-01 -1.84435710e-01 -5.66291213e-01 -1.31495249e+00 -1.08036205e-01 -5.67195714e-01 1.02502394e+00 -3.01870137e-01 9.47531819e-01 3.55494693e-02 -2.05299139e-01 6.75070584e-01 -1.72337711e-01 -3.49367827e-01 -8.62120211e-01 3.26586187e-01 4.48972695e-02 3.96971852e-01 7.20782876e-01 -4.71635312e-01 2.77200323e-02 1.64215654e-01 -1.19592452e+00 -4.43820268e-01 1.17143793e-02 6.88872755e-01 2.48030350e-01 2.52987713e-01 5.17819345e-01 -1.78893852e+00 6.33186758e-01 -4.01035666e-01 -7.19699740e-01 6.61334515e-01 -4.67937797e-01 3.11035812e-01 8.88469517e-01 -2.30922669e-01 -1.20439053e+00 3.45209211e-01 2.73675352e-01 -7.26238266e-02 -2.28101403e-01 4.10400867e-01 -8.33176672e-01 1.81051027e-02 8.89875889e-01 1.00765057e-01 4.28918330e-03 -5.21717966e-01 7.18627989e-01 9.54438448e-01 5.73192716e-01 -6.97226882e-01 1.07621980e+00 6.62886918e-01 -2.58846402e-01 -8.20384502e-01 -7.57395327e-01 -3.25503200e-01 -9.92849410e-01 2.30040014e-01 7.83033907e-01 -7.82275021e-01 -6.15796685e-01 3.03071529e-01 -9.74050760e-01 -1.36686862e-02 -5.19494832e-01 1.65484592e-01 -5.52265108e-01 7.25208044e-01 -3.31563085e-01 -7.63126075e-01 -2.61712492e-01 -6.73523784e-01 6.93820238e-01 -1.69234648e-01 -6.12142861e-01 -1.27315187e+00 4.34943214e-02 7.89586604e-01 -8.73075724e-02 4.06397551e-01 1.21219885e+00 -1.46450067e+00 -4.29899633e-01 -8.11106265e-01 3.92084494e-02 2.82152653e-01 3.79838318e-01 -2.06721827e-01 -1.16833818e+00 -4.09989536e-01 -6.90306127e-02 -3.68194371e-01 4.65787172e-01 -2.83874929e-01 1.08392692e+00 -8.71525168e-01 -5.75078130e-01 5.19077480e-01 1.22206557e+00 -3.11317831e-01 5.86719573e-01 2.60722339e-01 9.69117641e-01 1.00038874e+00 1.65007159e-01 3.98486912e-01 4.71749425e-01 5.96920848e-01 9.55462307e-02 1.22678578e-01 2.41456077e-01 -6.67075753e-01 -1.28957396e-02 3.38891208e-01 3.72178376e-01 -5.85179746e-01 -8.23735237e-01 6.74148679e-01 -1.67637777e+00 -9.95945990e-01 1.36467531e-01 2.53007293e+00 9.82513964e-01 -2.03474686e-01 1.58785611e-01 3.31203900e-02 9.00017917e-01 1.07008182e-01 -3.90756935e-01 -4.36461598e-01 -2.56035537e-01 -1.58398941e-01 1.00018787e+00 7.06294835e-01 -1.10767567e+00 9.06775773e-01 5.82898045e+00 5.65917313e-01 -6.41335666e-01 -1.82943761e-01 6.18038833e-01 7.37124532e-02 -6.13879025e-01 3.72633487e-01 -8.07620645e-01 5.48466027e-01 1.12827516e+00 -7.74750769e-01 4.84129459e-01 6.49853587e-01 -5.51784895e-02 5.65158874e-02 -1.44956589e+00 6.47024751e-01 2.80043572e-01 -1.26388705e+00 3.00207108e-01 2.37861827e-01 6.05589986e-01 -4.08243567e-01 -2.06855610e-01 1.70649029e-02 9.23547685e-01 -9.67456341e-01 4.70872134e-01 5.25440335e-01 7.41866648e-01 -8.79738271e-01 6.57979131e-01 5.67185521e-01 -6.92744970e-01 -3.17161903e-03 -3.10837775e-01 2.89234847e-01 -1.30307242e-01 6.02835894e-01 -1.20239961e+00 8.84473860e-01 5.38694024e-01 4.56283629e-01 -6.94079220e-01 6.04818463e-01 -3.77006471e-01 5.28670192e-01 -3.65897149e-01 1.60431743e-01 -9.77884885e-03 -3.09920341e-01 6.71275556e-01 1.25836301e+00 -3.16930525e-02 1.53967202e-01 1.42674208e-01 7.33298838e-01 -6.03557408e-01 4.45430726e-01 -1.13593674e+00 -4.85251725e-01 5.94037831e-01 1.07216060e+00 -5.85128307e-01 -4.33354110e-01 -4.28620547e-01 1.24016953e+00 6.13137841e-01 5.23672998e-01 -2.19705164e-01 -5.17535210e-01 4.49610800e-01 3.00728887e-01 1.49163082e-01 2.30493382e-01 -4.18582372e-03 -1.49285781e+00 1.68599799e-01 -9.32107031e-01 8.22469175e-01 -5.73593259e-01 -1.23772514e+00 4.26994711e-01 -1.20397247e-01 -8.84861708e-01 -3.52329195e-01 -2.76605189e-01 -3.52743775e-01 1.05319607e+00 -1.02106524e+00 -1.19343615e+00 -1.07344417e-02 6.12403393e-01 -1.24936648e-01 -2.36694440e-01 1.16180742e+00 7.84107372e-02 -6.57406509e-01 8.94061029e-01 2.97408342e-01 5.41238487e-01 7.83118725e-01 -1.37554693e+00 5.65714180e-01 9.88264799e-01 3.02986503e-01 8.31625342e-01 8.19357455e-01 -1.01446319e+00 -1.19730151e+00 -1.30656421e+00 1.28076947e+00 -1.06393528e+00 6.68814421e-01 -6.64891958e-01 -1.20520020e+00 1.30626822e+00 1.85459282e-03 7.73445740e-02 1.11196101e+00 -3.78466472e-02 -6.17678344e-01 5.92139699e-02 -1.60882568e+00 5.33821166e-01 9.60793793e-01 -8.01494956e-01 -7.22347200e-01 5.84538221e-01 5.88742197e-01 -2.67315805e-01 -8.41752887e-01 -3.34778219e-01 2.60961950e-01 -7.15463042e-01 7.56651402e-01 -1.07531846e+00 -1.81278914e-01 -2.91652977e-01 -6.94425181e-02 -1.23155093e+00 1.79933429e-01 -8.86208594e-01 1.89641833e-01 1.70541012e+00 3.93987507e-01 -8.86501491e-01 1.26586354e+00 1.27946508e+00 4.96037304e-01 1.47375122e-01 -9.63534236e-01 -7.11258590e-01 -7.78545509e-04 -9.75961760e-02 8.01107705e-01 1.57109857e+00 2.96614259e-01 3.63085657e-01 -4.68556076e-01 4.57641870e-01 7.35149086e-01 -1.56371623e-01 9.93598342e-01 -1.41820240e+00 -2.35193390e-02 2.38043323e-01 -4.26194698e-01 -4.05856252e-01 7.53407896e-01 -1.30173898e+00 -4.56235260e-01 -1.20040476e+00 1.63145021e-01 -5.37295163e-01 1.78740025e-01 6.35131001e-01 -4.30482626e-02 -7.68542439e-02 -1.58260614e-01 8.01685154e-02 -5.09335518e-01 3.61690730e-01 3.95465583e-01 -1.20006554e-01 -2.26498812e-01 2.43733048e-01 -1.06797719e+00 8.23334992e-01 7.34742403e-01 -7.74966896e-01 -5.46940029e-01 -2.13889286e-01 1.41513139e-01 -1.03000626e-01 4.11361814e-01 -6.02824628e-01 4.75877941e-01 1.58013880e-01 3.21503401e-01 -3.56869459e-01 3.41491662e-02 -1.18551183e+00 3.61456484e-01 2.65080065e-01 -5.97100854e-01 -1.47607699e-01 1.53135911e-01 9.84237134e-01 -4.19659913e-03 -4.45879221e-01 4.88959283e-01 -2.59097993e-01 -3.16100001e-01 1.96977377e-01 -2.54107952e-01 2.77464688e-01 1.02049208e+00 -3.11432421e-01 -4.36272055e-01 -4.50306773e-01 -9.04626310e-01 2.70714849e-01 1.04261589e+00 1.44747451e-01 4.97706085e-02 -1.00039184e+00 -6.07289433e-01 3.04403037e-01 3.61111164e-01 -2.41685554e-01 3.12378518e-02 6.91191703e-02 -3.45589668e-01 3.74320209e-01 -2.34188829e-02 -7.18218610e-02 -1.39865160e+00 9.33596194e-01 1.80252954e-01 -3.36830497e-01 -6.06056094e-01 5.42060971e-01 -2.79308464e-02 -7.55393505e-01 2.68012822e-01 3.96190546e-02 -1.11090355e-01 1.47580743e-01 6.07359886e-01 3.63137662e-01 2.87058800e-01 -8.61305833e-01 -3.06143075e-01 8.10450092e-02 -3.87768835e-01 -1.01168901e-01 1.09970331e+00 -2.93807447e-01 -4.03322905e-01 1.47858202e-01 1.17359424e+00 6.08371496e-01 -7.70373344e-01 -5.92043161e-01 4.91965979e-01 -5.89882135e-01 -4.65288222e-01 -7.47602761e-01 -8.76097023e-01 4.43107992e-01 9.76793393e-02 2.87152976e-01 8.25205266e-01 -1.66809693e-01 3.58380586e-01 8.38176847e-01 3.29857111e-01 -1.03378570e+00 -5.23573697e-01 -2.91729923e-02 5.60596704e-01 -1.06352615e+00 3.60535383e-01 -6.15881920e-01 -7.27414727e-01 8.35105717e-01 2.74671465e-01 4.25880969e-01 4.09743726e-01 6.45734146e-02 2.15182915e-01 -8.76602307e-02 -3.22504252e-01 3.11868727e-01 -9.27184746e-02 9.24241245e-01 -8.60662945e-03 -1.18671037e-01 -1.06016904e-01 6.48344338e-01 -4.31270838e-01 -8.85635614e-02 8.45022798e-01 8.42450738e-01 -5.07523157e-02 -1.32136023e+00 -3.54253590e-01 5.46680212e-01 -7.64060795e-01 -4.09977138e-02 -8.72349620e-01 7.88651049e-01 -1.54501542e-01 7.55472839e-01 -2.72999376e-01 -7.91090280e-02 3.59161735e-01 3.87091786e-01 4.19113189e-02 -7.50951111e-01 -8.26966465e-01 -7.11164176e-01 7.58069098e-01 -2.97952622e-01 -2.85921961e-01 -7.39176095e-01 -8.14072967e-01 -4.53303576e-01 -2.62844175e-01 6.43272281e-01 4.13529575e-01 7.87337065e-01 2.70772070e-01 -1.95442095e-01 3.96678805e-01 -3.28205466e-01 -4.86164987e-01 -4.06477839e-01 -8.04190874e-01 8.77669275e-01 2.15152547e-01 -2.19098534e-02 -4.73193824e-01 6.59876704e-01]
[6.153720855712891, 7.015879154205322]
021b4816-37f9-40ea-81da-32035f5682e4
end-to-end-human-pose-and-mesh-reconstruction
2012.09760
null
https://arxiv.org/abs/2012.09760v3
https://arxiv.org/pdf/2012.09760v3.pdf
End-to-End Human Pose and Mesh Reconstruction with Transformers
We present a new method, called MEsh TRansfOrmer (METRO), to reconstruct 3D human pose and mesh vertices from a single image. Our method uses a transformer encoder to jointly model vertex-vertex and vertex-joint interactions, and outputs 3D joint coordinates and mesh vertices simultaneously. Compared to existing techniques that regress pose and shape parameters, METRO does not rely on any parametric mesh models like SMPL, thus it can be easily extended to other objects such as hands. We further relax the mesh topology and allow the transformer self-attention mechanism to freely attend between any two vertices, making it possible to learn non-local relationships among mesh vertices and joints. With the proposed masked vertex modeling, our method is more robust and effective in handling challenging situations like partial occlusions. METRO generates new state-of-the-art results for human mesh reconstruction on the public Human3.6M and 3DPW datasets. Moreover, we demonstrate the generalizability of METRO to 3D hand reconstruction in the wild, outperforming existing state-of-the-art methods on FreiHAND dataset. Code and pre-trained models are available at https://github.com/microsoft/MeshTransformer.
['Zicheng Liu', 'Lijuan Wang', 'Kevin Lin']
2020-12-17
null
http://openaccess.thecvf.com//content/CVPR2021/html/Lin_End-to-End_Human_Pose_and_Mesh_Reconstruction_with_Transformers_CVPR_2021_paper.html
http://openaccess.thecvf.com//content/CVPR2021/papers/Lin_End-to-End_Human_Pose_and_Mesh_Reconstruction_with_Transformers_CVPR_2021_paper.pdf
cvpr-2021-1
['3d-absolute-human-pose-estimation']
['computer-vision']
[-2.90022820e-01 2.17505172e-01 -1.02115810e-01 -2.38193497e-01 -6.55953526e-01 -2.90008396e-01 5.32729268e-01 -3.09275597e-01 -1.22852124e-01 3.90364259e-01 3.41201782e-01 1.76642358e-01 1.66532010e-01 -6.95324063e-01 -1.19880474e+00 -2.87956208e-01 9.34280753e-02 1.21070468e+00 2.93451428e-01 -1.88049987e-01 -2.38993645e-01 5.76388478e-01 -1.50318432e+00 2.68557996e-01 3.37115109e-01 7.36379504e-01 1.26454487e-01 6.56403184e-01 2.01746941e-01 2.65422910e-01 -1.78274095e-01 -5.12739897e-01 3.08053613e-01 3.26755866e-02 -9.35437202e-01 1.96431100e-01 8.38594496e-01 -5.84891796e-01 -6.63386822e-01 5.06526947e-01 8.45680952e-01 5.45554049e-02 6.40327096e-01 -1.16072667e+00 -6.35504067e-01 2.68333763e-01 -7.59427905e-01 -3.66228372e-01 6.87206328e-01 1.41558588e-01 9.30473924e-01 -1.15823269e+00 1.13398790e+00 1.73304248e+00 9.45515990e-01 5.50782442e-01 -1.45966971e+00 -5.32053590e-01 3.18897903e-01 2.13527866e-03 -1.47722876e+00 -2.64378190e-01 7.09657609e-01 -6.28094673e-01 1.00003719e+00 1.29986241e-01 1.08856285e+00 1.41704595e+00 1.59906790e-01 8.92876923e-01 7.16819525e-01 -3.30146074e-01 -3.58635604e-01 -5.64218760e-01 -2.20379189e-01 1.15545928e+00 -1.03526399e-01 3.28413397e-02 -5.41636467e-01 -4.24763799e-01 1.54457200e+00 5.98370247e-02 -1.24103144e-01 -7.71076381e-01 -1.52561414e+00 5.31070292e-01 4.74791169e-01 -1.96282268e-01 -3.99160475e-01 6.73075140e-01 3.14758271e-01 -7.09314570e-02 5.37916958e-01 -1.43159106e-01 -5.32521844e-01 -2.65968926e-02 -6.41691744e-01 8.34645331e-01 5.61510324e-01 1.14707327e+00 4.76611763e-01 -2.93662459e-01 -1.51503831e-01 6.72675371e-01 4.42267597e-01 6.03693604e-01 -1.78914815e-01 -1.25191998e+00 3.97425026e-01 4.24272746e-01 8.58792663e-02 -9.46697056e-01 -5.48937619e-01 -2.93025792e-01 -5.86194158e-01 3.03127855e-01 3.51649165e-01 1.64959952e-01 -1.25570607e+00 1.77778339e+00 7.41899729e-01 1.95989743e-01 -7.02273071e-01 1.14109421e+00 1.23007429e+00 2.42859513e-01 -3.35323811e-02 5.16305208e-01 1.53347957e+00 -1.10274911e+00 -5.50244391e-01 -7.98825026e-02 1.48674846e-01 -7.24215209e-01 1.15293646e+00 2.40886301e-01 -1.32116532e+00 -5.47366500e-01 -5.24820566e-01 -7.27176428e-01 -1.92247294e-02 1.48490623e-01 7.46552527e-01 -2.97255977e-03 -9.01234746e-01 6.92325950e-01 -1.23976898e+00 -3.95890862e-01 3.89769346e-01 4.87304121e-01 -6.01234376e-01 4.24347222e-02 -9.96864676e-01 9.07484174e-01 -1.78559035e-01 1.88300833e-01 -9.08213854e-01 -7.70660043e-01 -9.61606622e-01 -2.61058033e-01 2.84043700e-01 -1.31434429e+00 1.29240894e+00 -4.65644568e-01 -1.63901699e+00 1.18571556e+00 -4.08857733e-01 4.71956432e-02 8.74322951e-01 -5.20707369e-01 1.89965993e-01 9.04941186e-02 1.27503097e-01 9.43171859e-01 8.26033533e-01 -1.39951944e+00 -8.68579745e-02 -4.27041173e-01 1.37573272e-01 1.26459628e-01 3.72260660e-01 -1.72564864e-01 -1.04987824e+00 -8.62267077e-01 2.43761137e-01 -1.09326029e+00 -1.29991040e-01 5.59937119e-01 -6.66602433e-01 -3.81716639e-01 5.90007603e-01 -9.25719082e-01 7.75952935e-01 -1.66966903e+00 9.60597694e-01 2.41844356e-01 4.24692303e-01 2.68129222e-02 -2.09651873e-01 4.41006571e-01 2.30718609e-02 -1.50139868e-01 -1.67155653e-01 -9.58762050e-01 2.50938416e-01 3.50037783e-01 1.14026800e-01 7.00587869e-01 5.52484244e-02 1.27943444e+00 -6.62560284e-01 -6.77837193e-01 4.89441156e-01 1.19576442e+00 -8.48668635e-01 1.68507472e-01 -5.11286199e-01 9.09987748e-01 -3.19861591e-01 8.68767500e-01 5.81423819e-01 -4.46851790e-01 2.57164370e-02 -4.15687174e-01 1.97884396e-01 1.57603443e-01 -1.20826840e+00 2.18566918e+00 -4.76579845e-01 3.28426540e-01 2.05820113e-01 -5.05952179e-01 4.86213237e-01 4.89357173e-01 6.37794971e-01 -3.20438564e-01 3.11866164e-01 3.54543468e-03 -5.65309346e-01 -3.88841510e-01 1.86186638e-02 1.77292407e-01 9.29453075e-02 2.24395379e-01 2.16190532e-01 1.30266473e-02 -2.45735958e-01 4.43274900e-02 8.01819205e-01 8.92119110e-01 7.08444566e-02 -8.53079632e-02 1.18015647e-01 -3.23566496e-01 4.74241912e-01 1.97450325e-01 2.11267337e-01 9.50759590e-01 3.34526926e-01 -6.53163552e-01 -1.15916610e+00 -1.46061516e+00 -1.97692528e-01 1.06966865e+00 1.26028478e-01 -6.56540811e-01 -8.10630441e-01 -4.34186965e-01 5.18141925e-01 1.09988332e-01 -8.46089780e-01 3.86560440e-01 -8.27934444e-01 -3.69254015e-02 4.58158821e-01 6.22524738e-01 1.75996482e-01 -9.62962627e-01 -3.86297464e-01 1.97950765e-01 -5.14914036e-01 -1.09551001e+00 -8.34246874e-01 -3.79064173e-01 -7.74236798e-01 -1.21095467e+00 -9.86198783e-01 -7.70208180e-01 7.27568626e-01 -9.71978009e-02 1.27769589e+00 3.11703384e-01 -5.66227376e-01 5.95114350e-01 -1.19270481e-01 -1.30688310e-01 -2.16610190e-02 1.60839722e-01 1.70566440e-01 -1.24048464e-01 -1.16978534e-01 -7.82291174e-01 -7.99446464e-01 4.93162155e-01 -3.37697983e-01 3.16910982e-01 2.56288916e-01 7.69012570e-01 1.04318786e+00 -5.54962456e-01 1.44985765e-01 -8.21865082e-01 4.07101139e-02 -2.00219035e-01 -2.90704995e-01 5.66920452e-02 -7.63891935e-02 2.92460397e-02 9.56685618e-02 -5.61215341e-01 -8.72244298e-01 4.72125918e-01 -3.98508132e-01 -8.47999752e-01 3.28530557e-02 4.71541770e-02 -2.88557738e-01 -1.25885531e-01 4.29852962e-01 -2.56083667e-01 2.78353035e-01 -1.02012014e+00 5.51244676e-01 2.81926066e-01 6.56262636e-01 -8.65997314e-01 7.24408269e-01 7.62394428e-01 1.26297846e-01 -6.53911889e-01 -5.46098411e-01 -2.51320660e-01 -9.55209851e-01 -2.25906223e-01 9.86435592e-01 -1.10101271e+00 -9.76392686e-01 4.65538293e-01 -1.46592391e+00 -5.60891330e-01 -1.66232865e-02 2.20248505e-01 -8.37695777e-01 5.03786564e-01 -9.48436916e-01 -4.66522902e-01 -3.10112327e-01 -1.35543704e+00 1.70651543e+00 -3.99089456e-01 -6.35206938e-01 -9.39787567e-01 3.99047248e-02 4.76302356e-01 -4.15550061e-02 7.15655148e-01 6.72097921e-01 1.64188623e-01 -6.10353827e-01 -1.39775172e-01 7.02307075e-02 -1.02196597e-01 5.13203070e-02 3.72349285e-02 -7.18994021e-01 -3.94796163e-01 -6.32305384e-01 -3.00925374e-01 7.95603812e-01 4.98627931e-01 1.14952040e+00 -1.91635653e-01 -6.02282643e-01 7.99596906e-01 8.78908575e-01 -6.79922819e-01 5.77924550e-01 1.03045367e-01 1.25851381e+00 5.76910257e-01 3.40389699e-01 3.96773428e-01 8.51511836e-01 1.17493141e+00 4.97862309e-01 -3.69567722e-01 -5.46263158e-01 -6.18914008e-01 3.37712746e-03 6.03249073e-01 -7.03784943e-01 5.81669528e-03 -8.86995077e-01 3.96067649e-01 -1.97058403e+00 -7.01866329e-01 -1.78036258e-01 1.97354770e+00 8.61614525e-01 -1.39989734e-01 4.51253653e-01 -2.04105958e-01 6.62456214e-01 7.18096271e-02 -5.11174202e-01 9.12557915e-02 1.65795088e-01 4.78131324e-01 3.39343876e-01 8.45888436e-01 -1.05436778e+00 1.14917529e+00 6.24607229e+00 5.89395404e-01 -7.31117189e-01 3.65486771e-01 1.49236312e-02 -4.05769765e-01 -3.34960580e-01 -1.51837736e-01 -6.12920225e-01 2.27736264e-01 1.09027535e-01 5.27896643e-01 6.01350486e-01 6.84085608e-01 9.24881324e-02 3.08171898e-01 -1.29261363e+00 1.18965757e+00 -1.13128521e-01 -1.21494234e+00 2.11477026e-01 1.85711443e-01 4.62817580e-01 6.26028329e-02 -5.17751351e-02 -9.81886685e-02 2.53710836e-01 -1.14171731e+00 1.14163268e+00 7.08188415e-01 1.05593956e+00 -5.72284102e-01 2.40348354e-01 2.10934103e-01 -1.43761027e+00 4.61287290e-01 -1.95311829e-01 -8.82928148e-02 4.46366340e-01 4.87285703e-01 -3.53502095e-01 4.56968665e-01 8.72385800e-01 6.84118152e-01 -3.57012838e-01 6.62098110e-01 -3.35877776e-01 8.19745287e-02 -5.28533816e-01 4.73935038e-01 -3.41771871e-01 1.12717547e-01 6.86618805e-01 9.73780155e-01 -1.38640786e-02 6.11388637e-03 3.67094219e-01 9.00768518e-01 -4.65549640e-02 1.91155281e-02 -3.83513063e-01 3.36618006e-01 4.75456208e-01 8.82606447e-01 -5.76340377e-01 -2.73565471e-01 -2.79526800e-01 1.26279211e+00 6.17862642e-01 4.30143297e-01 -9.87531126e-01 1.51471803e-02 8.77841592e-01 6.70036972e-01 3.43697488e-01 -4.97105181e-01 -1.62193358e-01 -1.25034213e+00 2.80153751e-01 -8.87020051e-01 6.67061061e-02 -6.75884664e-01 -1.28353465e+00 4.75013852e-01 1.61663771e-01 -9.45820510e-01 3.93996743e-04 -4.88756299e-01 -2.50107378e-01 7.10686862e-01 -9.38011229e-01 -1.86860943e+00 -3.04500937e-01 6.88789189e-01 4.26444411e-01 2.83745259e-01 8.58945549e-01 3.19230974e-01 -2.94272631e-01 6.24600470e-01 -4.65367526e-01 2.07133681e-01 7.73023963e-01 -1.11348975e+00 1.00680494e+00 2.74734974e-01 2.55967379e-01 6.61214709e-01 6.92814350e-01 -9.80641186e-01 -1.57679451e+00 -8.67768884e-01 7.80916810e-01 -9.62851286e-01 1.01056419e-01 -8.44437897e-01 -6.52475595e-01 1.24469578e+00 -1.41566679e-01 2.81297922e-01 1.80597052e-01 3.43656957e-01 -6.11435115e-01 3.48879784e-01 -1.06252742e+00 6.95626736e-01 1.87897420e+00 -4.18856949e-01 -5.11062264e-01 6.33826852e-01 7.69771516e-01 -1.19613171e+00 -1.15844369e+00 4.91510898e-01 9.97115910e-01 -7.59951234e-01 1.45913661e+00 -5.85761666e-01 3.90651971e-01 -1.31611466e-01 -1.55996025e-01 -1.13089097e+00 -5.72442055e-01 -6.93089664e-01 -5.01143157e-01 8.46126676e-01 1.05160050e-01 -4.78763551e-01 9.15813267e-01 5.39926171e-01 -5.31127416e-02 -1.11354768e+00 -1.14857101e+00 -5.92471838e-01 7.31735826e-02 -3.04793745e-01 7.15898752e-01 7.30439782e-01 -3.45964253e-01 1.04471087e-01 -6.83456779e-01 2.46859297e-01 9.89578366e-01 2.05418635e-02 1.13834393e+00 -1.44530118e+00 -5.30172944e-01 -1.54246196e-01 -4.19881165e-01 -1.27826262e+00 4.87979203e-01 -9.88477647e-01 -1.03366882e-01 -1.84856427e+00 9.28340256e-02 -3.75229686e-01 2.44441733e-01 7.73410022e-01 -2.69799475e-02 5.87322056e-01 1.86447158e-01 1.40447170e-01 -2.73515970e-01 6.36061788e-01 1.72883475e+00 -6.41489625e-02 -1.22395143e-01 -1.05137087e-01 -1.45425633e-01 9.04431105e-01 4.99867827e-01 -3.58688653e-01 -1.86690792e-01 -7.58669913e-01 3.10901273e-02 1.19866505e-01 9.41406846e-01 -7.32600093e-01 -2.87059769e-02 5.23129739e-02 5.78083277e-01 -6.65438712e-01 6.92522287e-01 -6.21181190e-01 6.56945288e-01 3.23049933e-01 -1.03449151e-02 1.59327716e-01 7.52749965e-02 5.33629239e-01 3.37958694e-01 3.51598054e-01 6.34885073e-01 -3.18210036e-01 -3.03236753e-01 7.90933371e-01 -1.71530712e-02 5.63599877e-02 8.25269938e-01 -1.42985493e-01 1.80826947e-01 -2.85088629e-01 -1.16824830e+00 2.98187643e-01 8.00288320e-01 6.04517579e-01 6.86971545e-01 -1.64280069e+00 -7.49196887e-01 1.56716987e-01 -6.27254173e-02 4.96981889e-01 3.65416735e-01 8.26853812e-01 -7.57183135e-01 1.79940715e-01 -9.50281322e-02 -8.60302210e-01 -1.36948121e+00 4.21331614e-01 4.22358483e-01 5.29049858e-02 -1.37993968e+00 9.13203478e-01 2.32108355e-01 -8.85777473e-01 4.06661242e-01 -3.82178456e-01 4.81700778e-01 -3.49664778e-01 2.50560790e-01 5.09312093e-01 -5.77964820e-02 -9.35658157e-01 -4.82175380e-01 1.19562733e+00 2.61034369e-01 -1.21681333e-01 1.27350616e+00 -2.14165282e-02 -2.23987654e-01 2.63368964e-01 1.21529531e+00 6.95995986e-02 -1.45013452e+00 -3.22172582e-01 -6.02351069e-01 -5.73584437e-01 -1.62297964e-01 -6.31420195e-01 -1.17921984e+00 7.81163633e-01 2.50344843e-01 -5.72971404e-01 5.69127977e-01 4.50504720e-01 1.08137488e+00 1.10791266e-01 6.91892624e-01 -7.73821712e-01 9.99036878e-02 3.77646267e-01 1.41026497e+00 -9.07601237e-01 1.32133737e-01 -8.58118176e-01 -3.63090843e-01 8.44457388e-01 4.72513020e-01 -3.55271429e-01 8.36640239e-01 3.68540257e-01 -5.18802442e-02 -4.36075717e-01 -4.85582888e-01 -9.91007611e-02 5.62757730e-01 6.69878602e-01 5.15251219e-01 2.72088557e-01 -9.38828811e-02 4.63924259e-01 -3.04283351e-01 1.19920924e-01 -1.32306635e-01 9.19653893e-01 3.63233872e-02 -1.38886464e+00 -4.85596061e-01 3.21099877e-01 -2.85497874e-01 1.92900389e-01 -3.53170365e-01 9.72455859e-01 2.50385255e-01 3.97898078e-01 8.72234702e-02 -4.51106906e-01 7.11005509e-01 4.90986975e-03 1.12690389e+00 -6.35942459e-01 -4.03164476e-01 2.82768041e-01 9.28396508e-02 -1.03025329e+00 -3.30621690e-01 -7.19229043e-01 -1.45214236e+00 -5.75770974e-01 -1.41704053e-01 -4.05118465e-01 4.26457465e-01 6.56310916e-01 7.17267036e-01 4.91138607e-01 -8.65412727e-02 -1.73844028e+00 -2.13151798e-01 -8.09054673e-01 -4.42607075e-01 6.25308931e-01 3.61456990e-01 -1.51307464e+00 -1.98369520e-03 4.76535521e-02]
[6.999768257141113, -1.1811882257461548]
c9666b61-9e3c-4799-8fa6-bc848236dadb
interpretable-visualizations-with
2006.06640
null
https://arxiv.org/abs/2006.06640v1
https://arxiv.org/pdf/2006.06640v1.pdf
Interpretable Visualizations with Differentiating Embedding Networks
We present a visualization algorithm based on a novel unsupervised Siamese neural network training regime and loss function, called Differentiating Embedding Networks (DEN). The Siamese neural network finds differentiating or similar features between specific pairs of samples in a dataset, and uses these features to embed the dataset in a lower dimensional space where it can be visualized. Unlike existing visualization algorithms such as UMAP or $t$-SNE, DEN is parametric, meaning it can be interpreted by techniques such as SHAP. To interpret DEN, we create an end-to-end parametric clustering algorithm on top of the visualization, and then leverage SHAP scores to determine which features in the sample space are important for understanding the structures shown in the visualization based on the clusters found. We compare DEN visualizations with existing techniques on a variety of datasets, including image and scRNA-seq data. We then show that our clustering algorithm performs similarly to the state of the art despite not having prior knowledge of the number of clusters, and sets a new state of the art on FashionMNIST. Finally, we demonstrate finding differentiating features of a dataset. Code available at https://github.com/isaacrob/DEN
['Isaac Robinson']
2020-06-11
null
null
null
null
['image-clustering']
['computer-vision']
[-1.19261026e-01 -1.35711610e-01 -1.96410745e-01 -4.09472018e-01 -2.48960868e-01 -8.16228807e-01 4.86789376e-01 8.50884095e-02 -1.98750019e-01 1.47705942e-01 3.74707669e-01 -3.51663470e-01 -5.65012276e-01 -4.84550953e-01 -6.26007140e-01 -8.33867669e-01 -6.12764716e-01 5.88094831e-01 -1.77144334e-01 8.35646614e-02 3.61109853e-01 6.32801473e-01 -1.50558853e+00 3.55150878e-01 6.13328516e-01 4.92780864e-01 -2.74196684e-01 7.21370697e-01 -5.89277409e-02 1.43791541e-01 -8.00905108e-01 1.02931835e-01 4.12044048e-01 -6.25836194e-01 -5.92166722e-01 -4.55751806e-01 5.28554201e-01 4.64570858e-02 -2.65450507e-01 9.17989016e-01 3.42748702e-01 2.43130967e-01 9.96761322e-01 -1.80094802e+00 -8.36991489e-01 6.47798836e-01 -8.01859438e-01 1.67071119e-01 6.92206547e-02 1.68132678e-01 1.01219130e+00 -8.95779192e-01 1.11765623e+00 1.35105526e+00 5.94789147e-01 2.69429117e-01 -1.71566355e+00 -6.88688874e-01 -2.04786807e-01 1.06674626e-01 -1.27040529e+00 -7.73144439e-02 1.09012461e+00 -8.32857490e-01 5.07801294e-01 5.34512460e-01 8.95972610e-01 7.64199436e-01 -1.02966078e-01 8.79225314e-01 9.72081065e-01 -3.43338013e-01 4.07589346e-01 -2.21792683e-02 4.54828948e-01 7.77508557e-01 8.24626386e-02 1.05661914e-01 -7.79974878e-01 -1.49448052e-01 5.96208453e-01 2.11988673e-01 -1.74312964e-01 -1.02622938e+00 -1.48874795e+00 1.06675005e+00 9.26052690e-01 1.77688852e-01 3.20950262e-02 2.11658105e-01 5.76141119e-01 5.03386967e-02 4.32787001e-01 7.63982832e-01 9.59285162e-03 -1.55158505e-01 -1.18946791e+00 1.49224907e-01 6.25256181e-01 4.95366365e-01 7.46382415e-01 -2.56028652e-01 6.88482029e-03 5.84990799e-01 2.50776649e-01 -1.30765438e-01 5.01443326e-01 -1.36215711e+00 -1.75853088e-01 7.82527208e-01 -3.32173914e-01 -1.31125414e+00 -5.97744763e-01 -1.60863072e-01 -7.50442803e-01 8.84260476e-01 5.51443219e-01 4.23300192e-02 -8.97305548e-01 1.61328506e+00 3.58124465e-01 1.15178376e-01 -1.66019097e-01 1.04711115e+00 8.37365448e-01 5.71550667e-01 -3.52286398e-01 4.10730630e-01 1.13398540e+00 -8.07493389e-01 -6.15916848e-01 3.34947020e-01 9.10489440e-01 -2.85529882e-01 1.48660672e+00 3.97999704e-01 -6.83275580e-01 -4.21655148e-01 -1.46524632e+00 -4.03448969e-01 -1.08292115e+00 2.75566280e-01 4.32236075e-01 4.28828120e-01 -1.23681712e+00 8.98204029e-01 -1.10309160e+00 -6.59999073e-01 6.38450742e-01 2.91344970e-01 -2.79837281e-01 3.27490300e-01 -6.77794158e-01 4.47193146e-01 4.72962499e-01 -8.00128654e-02 -5.98084390e-01 -9.87529933e-01 -8.69780838e-01 9.49528962e-02 -4.80809920e-02 -4.60319519e-01 6.34421289e-01 -6.89053953e-01 -1.20918918e+00 7.82127261e-01 -1.72346070e-01 -2.75515050e-01 3.79051298e-01 -7.35496134e-02 -1.81481332e-01 1.78781033e-01 1.35756228e-02 1.10287082e+00 4.20158714e-01 -1.65151525e+00 -3.16340119e-01 -3.46143156e-01 -1.79006740e-01 9.57253650e-02 -3.09085786e-01 2.81942990e-02 -3.39383304e-01 -5.22704720e-01 7.44713172e-02 -8.49955261e-01 -8.34696926e-03 6.43113732e-01 -9.14991379e-01 -1.66360646e-01 1.67168641e+00 -6.77854776e-01 1.11384308e+00 -2.65004683e+00 2.33145729e-01 7.79730976e-01 8.80165100e-01 -6.65127039e-02 -7.32142776e-02 4.14012909e-01 -4.56473917e-01 4.58602130e-01 -5.70758164e-01 -2.48797268e-01 2.72822022e-01 1.40532210e-01 1.17641434e-01 6.16196394e-01 -2.86325384e-02 7.83246875e-01 -8.46981585e-01 -3.38850081e-01 2.21928567e-01 7.21392989e-01 -4.42048937e-01 -2.40141097e-02 -8.76804814e-02 1.73179761e-01 2.90653497e-01 3.36778104e-01 5.44040561e-01 -4.31444913e-01 3.87488753e-01 -2.13911146e-01 -2.36747295e-01 -6.62371665e-02 -9.70192075e-01 1.73347640e+00 2.13447824e-01 1.38946581e+00 9.59393196e-03 -1.02523923e+00 9.85058963e-01 -2.72692859e-01 2.40900964e-01 -2.48381242e-01 -6.33243173e-02 -2.10044697e-01 2.18031853e-01 -4.49519247e-01 4.24061269e-01 3.45400304e-01 6.86770007e-02 7.30341494e-01 -1.78836957e-02 2.98214108e-01 5.04233956e-01 6.91927791e-01 9.62069631e-01 2.13500589e-01 -1.99554134e-02 -4.44682091e-01 -2.87025452e-01 2.18435481e-01 2.22417176e-01 3.51841927e-01 -2.63694208e-02 5.80055892e-01 1.13000059e+00 -5.02102435e-01 -1.24575365e+00 -1.27965713e+00 -7.32842684e-02 1.20756018e+00 -7.77294021e-03 -6.94676757e-01 -7.46771872e-01 -5.89374781e-01 3.35334241e-01 7.27082849e-01 -1.35185766e+00 -1.07147463e-01 -3.88926685e-01 -5.06215334e-01 6.77105904e-01 6.01769030e-01 9.77624878e-02 -9.08878863e-01 -8.84673059e-01 -4.33349699e-01 2.86247700e-01 -2.09391817e-01 -4.22312945e-01 5.04494131e-01 -7.51119852e-01 -1.11730790e+00 -3.63415688e-01 -8.40969980e-01 7.85447180e-01 -5.56530133e-02 1.02872765e+00 -1.14690587e-01 -7.75445044e-01 2.75477730e-02 -5.51574714e-02 -2.22231463e-01 -3.74252141e-01 -5.98834902e-02 7.01611862e-02 -3.32796246e-01 5.13303459e-01 -7.93110192e-01 -7.13185370e-01 1.56534255e-01 -9.56361830e-01 1.14313342e-01 3.49930584e-01 8.09014916e-01 5.51121891e-01 7.16358125e-02 8.23322237e-02 -8.98064017e-01 9.18995321e-01 -5.32631695e-01 -4.94282067e-01 -1.11867040e-02 -5.36705017e-01 2.80127734e-01 7.06229329e-01 -2.86892802e-01 -2.70855814e-01 1.00594096e-01 5.33619761e-01 -8.48460853e-01 -2.05159754e-01 7.33614981e-01 4.96533792e-03 2.64473230e-01 9.33253586e-01 -1.58638641e-01 6.16917729e-01 -5.38142562e-01 8.59651148e-01 3.41568321e-01 7.83288956e-01 -2.75428087e-01 7.33307242e-01 6.05954349e-01 9.04141814e-02 -6.62798464e-01 -4.14057113e-02 -3.63143653e-01 -1.13404238e+00 -2.46290356e-01 1.05922461e+00 -1.63620561e-01 -9.95725572e-01 -1.96407586e-01 -6.60070181e-01 -5.18305779e-01 -5.41149139e-01 2.93620110e-01 -4.17056084e-01 2.37811148e-01 -4.43528801e-01 -5.09714127e-01 -5.39568588e-02 -9.74300385e-01 5.78393281e-01 1.43726349e-01 -7.29695976e-01 -1.16686583e+00 3.37622792e-01 -2.02322021e-01 2.25270465e-01 8.03549170e-01 1.43253338e+00 -7.34916806e-01 -2.40021899e-01 7.31971785e-02 -2.94300735e-01 -2.72470317e-03 4.37017567e-02 8.29297841e-01 -7.07248569e-01 -2.55674899e-01 -7.31390119e-01 7.02854199e-03 8.88194680e-01 6.31836116e-01 1.44928181e+00 -3.44650775e-01 -6.18712246e-01 7.85051227e-01 1.16658151e+00 3.05805564e-01 3.73752445e-01 3.19693297e-01 7.48724699e-01 7.63816953e-01 1.51590511e-01 1.71503708e-01 1.41580582e-01 5.09079039e-01 4.14245486e-01 -5.99816918e-01 -4.00977721e-03 -3.35034043e-01 9.15088356e-02 5.26131868e-01 1.69615269e-01 6.11156598e-02 -1.17572987e+00 5.45866728e-01 -1.94509530e+00 -8.67208838e-01 -2.83697061e-02 2.08344841e+00 5.15109420e-01 -1.81446180e-01 6.40371561e-01 2.35916376e-01 6.11439645e-01 1.74713477e-01 -8.66341949e-01 -5.76151669e-01 -1.11377751e-02 7.48504475e-02 1.67443976e-01 3.13863397e-01 -1.10906339e+00 5.83741426e-01 6.59562922e+00 5.17196953e-01 -1.19180071e+00 -2.57033527e-01 7.16126740e-01 -7.21354306e-01 -2.22513869e-01 -1.58200905e-01 5.62755251e-03 4.46424156e-01 8.79811883e-01 -2.92892575e-01 4.62579459e-01 8.06353152e-01 3.12960565e-01 3.46736908e-02 -1.49766910e+00 9.48447227e-01 6.21575937e-02 -1.60823643e+00 -9.83363762e-02 2.33706027e-01 2.67314792e-01 -1.20760933e-01 5.98430276e-01 -1.11503795e-01 8.17724109e-01 -1.42769599e+00 4.02570844e-01 5.06769717e-01 8.39059472e-01 -9.71479774e-01 3.61427635e-01 -5.32804541e-02 -9.70372736e-01 1.14932522e-01 -1.29684865e-01 1.97788492e-01 -1.27769217e-01 4.29671317e-01 -1.01242507e+00 2.72753090e-01 1.13657153e+00 9.97324526e-01 -8.32496881e-01 9.90292609e-01 -5.82798086e-02 6.27979457e-01 -4.14950222e-01 -2.17124611e-01 2.52728999e-01 -4.25372303e-01 5.50412953e-01 1.38855791e+00 2.92334676e-01 -3.06150526e-01 -2.72794478e-02 1.36645818e+00 -8.15921575e-02 2.14739610e-02 -8.79084289e-01 -2.22349793e-01 5.42915940e-01 1.39184880e+00 -9.84710813e-01 -4.04172868e-01 2.14931488e-01 9.26855326e-01 4.05674160e-01 6.29844129e-01 -4.62508202e-01 -8.70685577e-01 8.05875242e-01 9.66852717e-03 2.75071144e-01 -3.67043227e-01 -5.87546110e-01 -5.84298730e-01 -1.99402466e-01 -7.39575565e-01 6.59599900e-01 -9.60716784e-01 -1.13486981e+00 1.06627658e-01 1.38302654e-01 -1.30498910e+00 -2.22206429e-01 -7.70217001e-01 -8.53396416e-01 6.63052499e-01 -6.08567178e-01 -8.18209469e-01 -2.28801697e-01 3.29576135e-01 1.19917154e-01 -1.41312657e-02 7.40177333e-01 -1.71132252e-01 -8.41448188e-01 6.07200801e-01 6.72691882e-01 5.03051937e-01 6.60201669e-01 -1.70104825e+00 3.59443724e-01 6.93224907e-01 3.41096580e-01 1.02484190e+00 7.51404703e-01 -4.41003650e-01 -1.14477229e+00 -8.19360316e-01 1.65864259e-01 -4.12752151e-01 6.29692376e-01 -8.19355369e-01 -9.69490826e-01 7.82709897e-01 5.05133927e-01 -1.33401811e-01 1.27382886e+00 4.88689542e-01 -6.42788649e-01 2.62176126e-01 -1.19429982e+00 8.63622010e-01 1.08279550e+00 -4.02622283e-01 -2.42944658e-01 2.17721120e-01 6.54876113e-01 -1.85869887e-01 -8.47881854e-01 -5.05474303e-03 8.88436735e-01 -1.05586076e+00 9.44092035e-01 -9.59532559e-01 7.06654251e-01 -6.42036974e-01 4.42892089e-02 -1.73839617e+00 -6.54953718e-01 -3.38664383e-01 -1.62566409e-01 9.06719983e-01 7.61535227e-01 -5.00495553e-01 1.00861692e+00 4.21839744e-01 -4.55518402e-02 -8.32967460e-01 -8.63167346e-01 -7.65672326e-01 2.34872401e-01 8.42894688e-02 5.43868959e-01 1.44566369e+00 3.95922691e-01 1.30017117e-01 -2.86655519e-02 -3.25633287e-02 5.71079075e-01 2.02340633e-01 1.01485395e+00 -1.44404626e+00 -1.96484569e-02 -8.33698332e-01 -6.46804631e-01 -6.10899866e-01 3.29838954e-02 -1.24746764e+00 -1.18541598e-01 -1.60228276e+00 1.54051721e-01 -2.04010934e-01 -2.77827799e-01 7.63000488e-01 7.91146830e-02 3.92324835e-01 2.88102239e-01 3.18412781e-01 -5.18884003e-01 3.75795543e-01 8.19951415e-01 -2.22594738e-01 -4.97130901e-01 -8.17030013e-01 -6.63213730e-01 5.38217187e-01 9.04087842e-01 -4.21325356e-01 -2.78927177e-01 -1.32024631e-01 -7.88171683e-03 -5.19126356e-01 5.45371115e-01 -9.36469913e-01 2.38156766e-01 2.91089922e-01 9.06406581e-01 -8.08719277e-01 3.45892638e-01 -6.46824121e-01 2.96711922e-01 4.61365670e-01 -6.75486863e-01 4.45828140e-01 2.66070575e-01 3.78578961e-01 3.53591796e-03 2.18334749e-01 7.28325427e-01 2.89649785e-01 -4.92664367e-01 -1.51362106e-01 -3.55783761e-01 -2.33851999e-01 1.02207851e+00 -5.01960993e-01 -7.69922018e-01 -4.66209233e-01 -8.88366818e-01 3.58226687e-01 7.66391218e-01 2.29507223e-01 6.31332517e-01 -1.46891224e+00 -4.53476071e-01 2.98195571e-01 5.73074892e-02 3.70605774e-02 1.70945615e-01 7.65971243e-01 -6.89013720e-01 -3.67688090e-02 -5.05086720e-01 -1.02832937e+00 -1.52582979e+00 8.80320728e-01 2.06694156e-01 3.97436656e-02 -7.42046714e-01 6.71968997e-01 2.29530141e-01 -6.25968993e-01 1.54212162e-01 -3.43285710e-01 -1.88676998e-01 2.40805253e-01 3.98498714e-01 5.66764712e-01 -3.72526377e-01 -4.93041813e-01 -3.49541456e-01 4.39538747e-01 -4.83718552e-02 -3.60113829e-01 1.42238843e+00 3.21545511e-01 -2.21793875e-01 8.95235419e-01 1.62212956e+00 -1.34521708e-01 -1.28873551e+00 5.94374724e-02 -3.91746685e-02 -4.62056041e-01 -2.69365370e-01 -8.91248465e-01 -1.11489666e+00 1.03880429e+00 9.10677493e-01 4.28522140e-01 9.60573435e-01 1.81664973e-01 1.80093512e-01 2.68435866e-01 -4.30884987e-01 -1.01721537e+00 2.48973906e-01 8.17745402e-02 8.63105059e-01 -9.30360198e-01 6.21443056e-02 -2.21560165e-01 -5.41176915e-01 1.14962435e+00 4.31082606e-01 -3.10571402e-01 7.42226541e-01 4.66639519e-01 4.61427480e-01 -7.53520906e-01 -4.71310079e-01 2.28499740e-01 3.67951363e-01 7.67775297e-01 4.53744739e-01 1.47063956e-01 -1.05432896e-02 -2.40486637e-02 -7.64522552e-01 -3.38741571e-01 2.59490341e-01 9.79843199e-01 -1.42124936e-01 -6.47503972e-01 -2.79040873e-01 7.52042413e-01 8.53896588e-02 4.97717373e-02 -7.81920135e-01 1.08854425e+00 -1.11115612e-01 4.89669025e-01 5.57509005e-01 -6.34629309e-01 7.65231773e-02 3.47173959e-01 -1.18732676e-01 -2.50455409e-01 -4.32511628e-01 -8.88903141e-02 -1.54662415e-01 -6.73072934e-01 -3.61544311e-01 -7.38875210e-01 -1.38227034e+00 -4.19869095e-01 1.81429297e-01 1.68026417e-01 7.40300596e-01 2.80626416e-01 7.56780863e-01 4.84186053e-01 3.22926193e-01 -1.10504663e+00 1.64199576e-01 -8.85266840e-01 -8.15487981e-01 5.80220282e-01 2.91617125e-01 -6.37522519e-01 -5.08471847e-01 1.77440479e-01]
[7.978028774261475, 4.444721698760986]
58bbaac6-6f01-4219-9892-9638f7219f2d
metacovid-a-siamese-neural-network-framework
null
null
https://reader.elsevier.com/reader/sd/pii/S0031320320305033
https://reader.elsevier.com/reader/sd/pii/S0031320320305033?token=A078019693958EA8EF7D7236196492ACBFED6694FECCCC2FC640FA775670FD4C55081AF13AD614D7920C53BFF7C25DA9&originRegion=eu-west-1&originCreation=20211205131409
MetaCOVID: A Siamese neural network framework with contrastive loss for n-shot diagnosis of COVID-19 patients
Various AI functionalities such as pattern recognition and prediction can effectively be used to diagnose (recognize) and predict coronavirus disease 2019 (COVID-19) infections and propose timely response (remedial action) to minimize the spread and impact of the virus. Motivated by this, an AI system based on deep meta learning has been proposed in this research to accelerate analysis of chest X-ray (CXR) images in automatic detection of COVID-19 cases. We present a synergistic approach to integrate contrastive learning with a fine-tuned pre-trained ConvNet encoder to capture unbiased feature representations and leverage a Siamese network for final classification of COVID-19 cases. We validate the effectiveness of our proposed model using two publicly available datasets comprising images from normal, COVID-19 and other pneumonia infected categories. Our model achieves 95.6% accuracy and AUC of 0.97 in diagnosing COVID-19 from CXR images even with a limited number of training samples.
['M. ShamimHossain', 'Mohammad Shorfuzzaman']
2020-10-17
null
null
null
pattern-recognition-2020-10
['covid-19-detection', 'pneumonia-detection']
['medical', 'medical']
[ 4.24066752e-01 -3.75195414e-01 -5.77707402e-02 -2.65189737e-01 -6.31666064e-01 -4.61164266e-01 3.04083705e-01 6.33597896e-02 -3.19392085e-01 4.34001684e-01 1.53876171e-01 -3.68526369e-01 -4.89754379e-01 -5.28722465e-01 -6.09363437e-01 -5.86216092e-01 -2.03361079e-01 8.42146575e-01 -2.16562673e-01 3.17229867e-01 2.84005012e-02 8.82981479e-01 -9.95923340e-01 7.64687955e-01 9.10289168e-01 7.46579409e-01 5.49421847e-01 1.42847657e+00 3.41487139e-01 1.20823181e+00 -4.90376532e-01 5.23431152e-02 1.23379163e-01 -4.53173250e-01 -7.06537306e-01 -3.25955689e-01 3.95639211e-01 -7.59785116e-01 -3.98140699e-01 5.32450497e-01 6.28260434e-01 -3.12055945e-01 1.26049638e+00 -1.05746388e+00 -3.26210111e-01 -8.93073976e-02 -5.22977233e-01 9.63862240e-01 -9.55719054e-02 4.50076431e-01 5.94272435e-01 -7.28541315e-01 6.32857561e-01 8.05339515e-01 1.06680596e+00 7.10686088e-01 -4.89858419e-01 -4.88348275e-01 -3.42515081e-01 3.63820255e-01 -1.06780946e+00 2.27463141e-01 4.24293369e-01 -7.51933455e-01 1.27007926e+00 2.66935736e-01 6.28958285e-01 1.39663839e+00 5.68867683e-01 8.30742538e-01 6.72204018e-01 1.57613248e-01 2.61208545e-02 -2.07145810e-01 1.38359696e-01 9.08218741e-01 3.34516108e-01 -6.13401644e-03 5.29183960e-03 -6.72030091e-01 5.61760247e-01 1.01071930e+00 -1.42878041e-01 -9.88322645e-02 -1.31527185e+00 9.99387324e-01 6.96557343e-01 1.78330332e-01 -9.92391467e-01 -2.49854736e-02 6.27561212e-01 -7.45915920e-02 2.38385856e-01 6.24174714e-01 -7.34001815e-01 2.14145988e-01 -8.56019020e-01 2.63354778e-01 2.50575423e-01 3.16029072e-01 -2.31058747e-02 6.71763122e-02 -5.39008856e-01 7.69004226e-01 1.62021011e-01 1.28253782e+00 4.31530863e-01 -7.98960328e-01 5.16356528e-01 6.86320901e-01 2.10518241e-02 -8.28376889e-01 -7.69610584e-01 -4.93412882e-01 -1.34220445e+00 -2.02626228e-01 -1.27170354e-01 -5.48028946e-01 -1.08278799e+00 1.36114132e+00 1.95300192e-01 7.25461006e-01 2.31964692e-01 8.26986253e-01 5.49761057e-01 8.28652799e-01 9.17515904e-02 -1.80141538e-01 1.50157821e+00 -9.19321656e-01 -3.58105808e-01 1.18829772e-01 7.54309952e-01 -3.23212355e-01 6.71543956e-01 1.90109566e-01 -7.62576163e-01 -2.00058445e-01 -9.46508944e-01 5.01020730e-01 -4.01109487e-01 3.33002925e-01 3.66820097e-01 1.43206015e-01 -7.58252680e-01 4.99160379e-01 -1.16277313e+00 -4.29971844e-01 1.13123620e+00 3.96997452e-01 -7.75060803e-02 -2.33128995e-01 -8.96580756e-01 9.48354661e-01 1.10443868e-01 8.01832676e-02 -1.30359626e+00 -1.20328748e+00 -2.05548123e-01 9.03712660e-02 -1.01630511e-02 -1.29738474e+00 9.03795898e-01 -4.34981763e-01 -7.03948379e-01 8.97529244e-01 -8.89044404e-02 -7.91151166e-01 4.43912715e-01 -4.42504525e-01 -2.12425187e-01 7.13120222e-01 -6.44487664e-02 5.26088178e-01 9.49553370e-01 -7.99414337e-01 -6.86524749e-01 -4.27501082e-01 -4.68161017e-01 -5.93938604e-02 -1.46438152e-01 2.40729511e-01 3.08420867e-01 -7.89470017e-01 -4.45428997e-01 -1.11225283e+00 -2.20976278e-01 -1.31355628e-01 -2.50642657e-01 -2.84599811e-01 1.24506116e+00 -6.95459425e-01 8.84430289e-01 -1.72379792e+00 -2.05534354e-01 5.53115085e-02 6.79892302e-01 1.05153036e+00 -2.86843628e-01 2.40441650e-01 7.87388310e-02 1.01001509e-01 -4.15919781e-01 1.53684556e-01 -5.58329880e-01 8.96659344e-02 -3.05692941e-01 5.84045768e-01 7.94934750e-01 1.18755853e+00 -9.08924460e-01 -4.70424324e-01 4.58243757e-01 9.11842763e-01 -6.96357250e-01 8.32623482e-01 -3.91869247e-01 4.55431104e-01 -7.03588247e-01 7.33315766e-01 6.73144162e-01 -1.03268826e+00 4.86589372e-02 -1.29887417e-01 4.58587646e-01 -1.40434369e-01 -3.71661663e-01 8.64171207e-01 -4.41458911e-01 4.14804608e-01 -1.82078838e-01 -8.67275596e-01 2.67380834e-01 5.41681409e-01 9.25980985e-01 -2.28494853e-01 3.34392458e-01 5.39395884e-02 8.82126167e-02 -1.04162872e+00 -2.77174503e-01 2.11439610e-01 4.38877255e-01 8.46874356e-01 -1.93250284e-01 1.69570461e-01 -5.33517860e-02 4.69457954e-02 1.60966313e+00 -4.46847141e-01 2.63178766e-01 9.02750269e-02 5.91982782e-01 3.84628296e-01 3.00832629e-01 9.74747419e-01 -5.33546805e-01 7.65948296e-01 -6.81033591e-04 -8.40701163e-01 -1.31669748e+00 -1.19815695e+00 -1.67724922e-01 8.33245039e-01 -4.43690956e-01 2.86999196e-01 -6.95123374e-01 -9.79270637e-01 1.05394134e-02 3.43740016e-01 -8.55808079e-01 -6.79597259e-02 -1.05836678e+00 -9.30967867e-01 6.91828668e-01 9.26839828e-01 3.52035433e-01 -1.42226505e+00 -1.15626049e+00 1.26861513e-01 -1.83453914e-02 -8.50017428e-01 -2.83680022e-01 2.51424849e-01 -8.20512593e-01 -1.45410740e+00 -9.75421727e-01 -9.45419014e-01 6.75151229e-01 2.13796392e-01 7.61962593e-01 3.86530519e-01 -8.67626607e-01 2.92318940e-01 -2.16265619e-01 -5.37939548e-01 -5.43732822e-01 1.37524635e-01 2.75580976e-02 -2.98579693e-01 5.00826001e-01 -1.15491092e-01 -1.10231912e+00 -5.68591282e-02 -7.43488312e-01 -8.16225335e-02 6.37051523e-01 1.08532810e+00 6.71870828e-01 -2.84249991e-01 8.75895858e-01 -8.84904087e-01 6.53394043e-01 -1.02752721e+00 -2.86023587e-01 3.45942318e-01 -5.53838372e-01 -2.86959231e-01 1.00942302e+00 -3.42765838e-01 -8.22938561e-01 3.96962687e-02 5.83988093e-02 -1.07477891e+00 -9.88319423e-03 1.17088474e-01 7.70178020e-01 3.29314619e-01 4.14401531e-01 3.03864688e-01 1.45097196e-01 -3.15867543e-01 -1.97190717e-02 9.89758074e-01 6.33731663e-01 1.81993857e-01 5.62215984e-01 6.21819735e-01 3.23483646e-02 -5.57141721e-01 -1.14306188e+00 -6.69437766e-01 -3.82636249e-01 -4.54760902e-02 1.38067937e+00 -9.67482686e-01 -8.00496042e-01 4.97325301e-01 -1.22153485e+00 -3.16100903e-02 1.32223323e-01 7.68466413e-01 -5.07092178e-01 -5.80970459e-02 -9.09522831e-01 -4.49113965e-01 -1.19893849e+00 -1.02265000e+00 9.86677289e-01 5.31253926e-02 -3.95896345e-01 -8.84989679e-01 5.92294395e-01 6.08422816e-01 7.63506711e-01 4.05143231e-01 1.20859754e+00 -1.05626166e+00 -5.13909280e-01 -1.61507383e-01 -5.93598485e-01 4.37266350e-01 3.79505455e-01 1.87755451e-01 -8.71567011e-01 -5.03357947e-01 1.83487326e-01 -4.66695338e-01 8.08660209e-01 6.80255413e-01 1.23916841e+00 -5.95637679e-01 -5.33755898e-01 7.88966358e-01 1.32224321e+00 5.91839492e-01 9.99853611e-02 -8.33814293e-02 8.56526196e-01 1.06096797e-01 3.72417778e-01 5.59946001e-01 2.43670821e-01 4.21570316e-02 5.16086996e-01 -1.61226973e-01 -1.18691653e-01 1.26633257e-01 -1.14818357e-01 9.31668997e-01 -2.84371637e-02 -4.19187605e-01 -1.26661313e+00 5.84059954e-01 -1.49474537e+00 -1.13900113e+00 2.28195526e-02 1.38665402e+00 4.77592826e-01 -3.33749592e-01 -7.42474347e-02 -2.85190046e-01 6.95057511e-01 7.99262226e-02 -7.95363545e-01 -5.45357347e-01 5.47820528e-04 2.93071777e-01 1.35996118e-01 5.74670881e-02 -1.41386819e+00 2.98103243e-01 6.43921185e+00 2.08210692e-01 -1.48409247e+00 2.12301880e-01 7.50558615e-01 -3.00503403e-01 1.19463652e-01 -9.98004556e-01 -2.22992018e-01 5.19825816e-01 9.21066999e-01 2.22537622e-01 3.56757939e-01 9.10440743e-01 1.59897372e-01 7.06498265e-01 -9.60646272e-01 8.89214814e-01 1.48825303e-01 -1.88096762e+00 1.61771342e-01 -2.72818893e-01 1.12603378e+00 8.53062689e-01 1.46312192e-01 1.09971873e-01 3.76798883e-02 -1.22058785e+00 -2.56383121e-01 7.03369319e-01 8.15840542e-01 -7.59081483e-01 1.17012155e+00 1.88176543e-01 -9.04668450e-01 -1.45531610e-01 -2.07211271e-01 4.91900146e-01 -4.22105938e-02 7.10977018e-02 -1.62531519e+00 2.31895410e-03 6.95656419e-01 5.71041942e-01 -3.84367168e-01 8.39214742e-01 3.06810826e-01 8.90119314e-01 -2.75138497e-01 -3.53665888e-01 3.94213378e-01 2.31887147e-01 5.20229697e-01 1.43325949e+00 1.69444501e-01 3.91435415e-01 -1.83763113e-02 8.49268079e-01 -1.17157891e-01 9.56576020e-02 -8.33962679e-01 -3.63345236e-01 4.23051745e-01 1.04656374e+00 -4.69641715e-01 -7.05893636e-01 -1.55645639e-01 7.26335168e-01 1.83017865e-01 2.28573173e-01 -9.74666178e-01 -3.19877923e-01 5.16741276e-01 -6.64427818e-04 7.77850926e-01 3.72596711e-01 -3.87709022e-01 -9.41936255e-01 -3.60288054e-01 -1.03959334e+00 5.20837486e-01 -9.10856128e-01 -1.42225921e+00 8.94762516e-01 -2.15511262e-01 -1.12334943e+00 -6.09052300e-01 -6.86851144e-01 -9.86387014e-01 4.50085402e-01 -1.42989028e+00 -9.74729598e-01 -3.28573465e-01 4.40930426e-01 4.38650906e-01 -5.77176630e-01 9.58720505e-01 1.31244376e-01 -6.04649782e-01 3.05449516e-01 3.31620455e-01 1.42767102e-01 3.11798863e-02 -1.01986766e+00 3.91198248e-02 6.63406968e-01 -3.71118307e-01 8.13828230e-01 1.61749497e-01 -8.00746560e-01 -1.29333889e+00 -1.66860080e+00 8.21019828e-01 -5.88550448e-01 3.76034617e-01 3.40992324e-02 -9.10220206e-01 6.35439575e-01 1.98876292e-01 1.24335207e-01 6.61903858e-01 -7.25441039e-01 -2.71962315e-01 7.85544068e-02 -1.58213162e+00 4.04462159e-01 7.63679445e-01 -5.16802073e-01 -9.62045133e-01 6.63037241e-01 7.59933949e-01 2.08320953e-02 -8.69376004e-01 9.81281579e-01 4.18168128e-01 -5.97602427e-01 1.13433313e+00 -1.14102685e+00 5.72927713e-01 -7.71504045e-02 -4.02003489e-02 -8.43159616e-01 -3.51275325e-01 -2.12295517e-01 -3.65250558e-01 2.08856076e-01 1.27473027e-01 -5.21868527e-01 8.81553710e-01 -1.08240388e-01 -1.85073186e-02 -1.47919977e+00 -5.66853404e-01 -1.70410231e-01 -7.40468726e-02 -1.83529228e-01 5.45734465e-01 9.99294460e-01 -6.70149505e-01 3.14843170e-02 -3.06597471e-01 3.19292635e-01 4.33022350e-01 2.97611862e-01 3.04383993e-01 -1.14833474e+00 -1.02047138e-01 -3.09614122e-01 -1.75933212e-01 -2.75281876e-01 -6.65512607e-02 -1.05101264e+00 -2.09329292e-01 -1.65906537e+00 7.16851175e-01 -4.48769927e-01 -8.28092158e-01 4.54149932e-01 -2.86233842e-01 3.28738540e-01 1.00765578e-01 4.44456935e-01 -5.13038278e-01 1.91970587e-01 1.29050827e+00 -2.62736261e-01 1.45680513e-02 2.00969875e-01 -2.47768581e-01 7.59165823e-01 9.74617362e-01 -8.54946077e-01 -4.63352531e-01 -4.26231503e-01 1.10470704e-04 1.29259542e-01 5.09641051e-01 -7.94829369e-01 -1.05701253e-01 -3.64130318e-01 6.50563121e-01 -1.14068353e+00 1.20846704e-01 -8.24890435e-01 -6.88964054e-02 1.22269297e+00 -4.25529748e-01 2.51139492e-01 -7.21763596e-02 4.85488266e-01 1.29714951e-01 1.42510995e-01 8.70430708e-01 4.13547233e-02 -7.67108351e-02 6.10723972e-01 -6.49066329e-01 3.48280698e-01 1.19767010e+00 1.37431324e-01 -5.85066676e-01 7.28892982e-02 -2.03513786e-01 1.72854856e-01 3.42287309e-02 3.91828120e-01 9.75612104e-01 -9.05088007e-01 -1.12003386e+00 3.25180441e-01 4.05946672e-02 -9.16324183e-03 4.79479283e-01 9.53773439e-01 -1.29459071e+00 8.61846864e-01 -2.50458449e-01 -8.87128830e-01 -1.29316783e+00 8.15428317e-01 6.10272110e-01 -5.09332120e-01 -7.27749884e-01 8.28265190e-01 2.73906857e-01 -3.63957673e-01 2.43722767e-01 -3.83718938e-01 -1.67407662e-01 -2.39517555e-01 7.38289952e-01 4.21175331e-01 -1.67867560e-02 -3.69316131e-01 -8.92741561e-01 5.43975472e-01 -5.53063273e-01 7.87353754e-01 1.68531883e+00 4.30297196e-01 2.12434512e-02 2.42943019e-02 1.53482783e+00 -3.42791319e-01 -1.03044248e+00 -2.68644877e-02 -3.47619295e-01 7.98283494e-04 -1.98817864e-01 -1.08096409e+00 -1.14936948e+00 9.74971235e-01 1.20674586e+00 -2.79193014e-01 1.02042913e+00 4.11835164e-02 1.16661274e+00 7.71234870e-01 -2.85094976e-01 -6.66543007e-01 4.40252692e-01 4.24805611e-01 7.39232779e-01 -1.41793621e+00 -1.20162383e-01 3.56236815e-01 -8.43716681e-01 8.65160167e-01 4.79732990e-01 -6.03803396e-01 6.42741442e-01 2.82170236e-01 4.72361386e-01 -6.35962963e-01 -1.16455853e+00 3.43703300e-01 2.44433433e-01 7.94524670e-01 1.27785817e-01 2.42422715e-01 1.69807389e-01 4.19856101e-01 2.73514479e-01 1.19258843e-01 1.58772573e-01 9.03418839e-01 -4.03249085e-01 -3.12700331e-01 -8.34666416e-02 1.24493909e+00 -8.98248672e-01 -3.48408729e-01 -2.68357664e-01 4.69417602e-01 1.84910730e-01 5.20440698e-01 2.26503596e-01 -2.77967542e-01 -1.77504793e-01 1.10603301e-02 2.67921537e-01 -3.70804727e-01 -8.33250403e-01 -3.71811926e-01 -3.49289328e-01 -4.12515759e-01 -2.75937706e-01 -4.52068418e-01 -1.45203817e+00 2.42494210e-03 3.56793366e-02 -2.67438948e-01 4.79864299e-01 7.83430398e-01 7.55870700e-01 7.86291122e-01 9.54419732e-01 -2.36760497e-01 -8.38579834e-01 -8.27238858e-01 -2.24381864e-01 4.74441379e-01 8.34549904e-01 -2.73916394e-01 -4.04259264e-01 4.72041965e-02]
[15.546623229980469, -1.7338091135025024]
07f48685-b537-4adc-92c4-0de59e5ef85e
global-proxy-based-hard-mining-for-visual
2302.14217
null
https://arxiv.org/abs/2302.14217v1
https://arxiv.org/pdf/2302.14217v1.pdf
Global Proxy-based Hard Mining for Visual Place Recognition
Learning deep representations for visual place recognition is commonly performed using pairwise or triple loss functions that highly depend on the hardness of the examples sampled at each training iteration. Existing techniques address this by using computationally and memory expensive offline hard mining, which consists of identifying, at each iteration, the hardest samples from the training set. In this paper we introduce a new technique that performs global hard mini-batch sampling based on proxies. To do so, we add a new end-to-end trainable branch to the network, which generates efficient place descriptors (one proxy for each place). These proxy representations are thus used to construct a global index that encompasses the similarities between all places in the dataset, allowing for highly informative mini-batch sampling at each training iteration. Our method can be used in combination with all existing pairwise and triplet loss functions with negligible additional memory and computation cost. We run extensive ablation studies and show that our technique brings new state-of-the-art performance on multiple large-scale benchmarks such as Pittsburgh, Mapillary-SLS and SPED. In particular, our method provides more than 100% relative improvement on the challenging Nordland dataset. Our code is available at https://github.com/amaralibey/GPM
['Philippe Giguère', 'Brahim Chaib-Draa', 'Amar Ali-bey']
2023-02-28
null
null
null
null
['metric-learning', 'image-similarity-search', 'visual-place-recognition', 'metric-learning']
['computer-vision', 'computer-vision', 'computer-vision', 'methodology']
[-1.66154757e-01 -7.62564242e-02 -2.43466780e-01 -4.62448180e-01 -1.42448187e+00 -5.70588231e-01 6.11657977e-01 5.13699174e-01 -6.18019521e-01 7.37995863e-01 -1.82701088e-02 -1.60036832e-02 -1.20021343e-01 -7.47678638e-01 -1.00021064e+00 -6.27290249e-01 -3.10868949e-01 7.96023607e-01 1.97909713e-01 1.19889807e-03 1.97676495e-01 4.47905183e-01 -1.54005134e+00 2.16358453e-01 7.44205058e-01 1.12196374e+00 2.64734924e-01 4.11886990e-01 1.56469107e-01 5.69369018e-01 -3.41787577e-01 -2.13912189e-01 4.64153230e-01 -4.39329445e-02 -6.87757850e-01 -2.14119360e-01 5.61959088e-01 -2.72686660e-01 -2.55742013e-01 8.11724424e-01 6.37759268e-01 3.48258227e-01 5.53915858e-01 -1.13327777e+00 -4.96740490e-01 6.16239548e-01 -6.99638784e-01 1.50381833e-01 1.52467698e-01 1.08000502e-01 1.33512402e+00 -1.09718418e+00 4.53768849e-01 9.68680263e-01 8.44342589e-01 1.91160306e-01 -1.32626235e+00 -6.80308819e-01 2.12791353e-01 4.11866307e-01 -1.64788127e+00 -4.58278060e-01 7.08434463e-01 -1.68698370e-01 1.02449846e+00 2.02341706e-01 4.90561008e-01 9.02282536e-01 -2.55958080e-01 1.00636804e+00 8.36966515e-01 -4.23980206e-01 4.84335303e-01 -8.62493552e-03 9.35160518e-02 8.16197634e-01 1.42073065e-01 -2.58975029e-02 -4.86527950e-01 -2.99020350e-01 3.80364776e-01 6.34880215e-02 -9.29739103e-02 -4.99848485e-01 -8.17539990e-01 8.60877693e-01 8.52039576e-01 5.39130494e-02 -2.62247473e-01 4.08277512e-01 2.54171550e-01 2.46532150e-02 5.16028523e-01 2.80693233e-01 -3.26831967e-01 -2.18955874e-01 -1.14244688e+00 4.43625420e-01 7.42762208e-01 7.36036539e-01 1.15567958e+00 -4.88673925e-01 -2.30905026e-01 1.00472724e+00 2.34946907e-01 2.03408703e-01 3.52312237e-01 -8.62488091e-01 5.71473598e-01 5.73628306e-01 -5.40129319e-02 -9.18171883e-01 -3.51703823e-01 -4.91145015e-01 -4.62516367e-01 1.75803274e-01 3.50539356e-01 1.01234764e-04 -1.09788823e+00 1.73325348e+00 4.74561393e-01 3.64929974e-01 -3.65466893e-01 8.01428914e-01 3.69661838e-01 7.25269914e-01 3.17741223e-02 4.32025284e-01 1.15149856e+00 -1.24570954e+00 -4.71442286e-03 -5.47726154e-01 6.54318452e-01 -5.38135648e-01 1.13465703e+00 3.56604457e-01 -9.91859853e-01 -3.54228467e-01 -1.16354477e+00 -4.71460670e-01 -5.42003751e-01 3.22586179e-01 6.25010252e-01 3.88072878e-01 -1.06863594e+00 8.04253817e-01 -1.07732296e+00 -1.46261483e-01 7.61270761e-01 3.64195704e-01 -3.53339046e-01 -2.53613859e-01 -7.28761911e-01 6.41935825e-01 2.94830441e-01 1.81095079e-01 -1.07213116e+00 -8.57791126e-01 -9.82566833e-01 1.19773909e-01 1.91009611e-01 -5.21545410e-01 1.26618922e+00 -7.54984140e-01 -1.11679196e+00 9.44870472e-01 -2.93197632e-01 -7.60229468e-01 6.34474576e-01 -3.97464573e-01 2.30872869e-01 5.60552254e-02 3.02816451e-01 8.56992364e-01 5.63752055e-01 -1.13479257e+00 -4.73488271e-01 -4.44578946e-01 -1.27709657e-02 2.49290675e-01 -2.15621918e-01 -2.82906801e-01 -7.36830831e-01 -4.89619970e-01 3.10823396e-02 -9.34074521e-01 -3.28648984e-01 2.93539673e-01 -4.22637969e-01 -2.66883969e-01 3.43941092e-01 -5.67726851e-01 7.73086965e-01 -2.33637547e+00 9.43080932e-02 4.10018831e-01 2.31206387e-01 1.13262177e-01 -2.27535605e-01 4.32501107e-01 1.24175109e-01 -2.82832980e-02 -4.83475000e-01 -9.92233038e-01 3.35103869e-01 1.71591043e-01 -2.57358491e-01 7.04439580e-01 1.72090858e-01 8.49423766e-01 -8.10108542e-01 -2.33029664e-01 2.56641686e-01 4.85116959e-01 -8.14449310e-01 5.29218949e-02 -2.99005866e-01 1.26306102e-01 -2.61640370e-01 6.19318604e-01 7.12315202e-01 -3.57236922e-01 1.16891703e-02 2.37183779e-01 1.32132601e-03 5.97377896e-01 -1.18066955e+00 1.85705054e+00 -5.91852129e-01 4.84443724e-01 -1.79162115e-01 -8.38589013e-01 7.51968801e-01 -2.43364736e-01 2.89804757e-01 -8.75608563e-01 -4.33031693e-02 3.66580695e-01 -4.57894385e-01 4.01550084e-02 5.20576954e-01 2.40058869e-01 -1.21678188e-01 1.69342324e-01 7.06225336e-02 2.92836547e-01 2.36115113e-01 -9.73169357e-02 1.10067976e+00 2.13836715e-01 2.53203154e-01 -2.00794682e-01 2.66419053e-01 -8.57008398e-02 5.56906641e-01 7.99147785e-01 -8.42020810e-02 8.96460474e-01 5.41963756e-01 -6.28311634e-01 -1.01243389e+00 -1.08724630e+00 -1.05234578e-01 1.04807854e+00 9.31497067e-02 -5.13657093e-01 -6.27071440e-01 -7.30144441e-01 1.60695001e-01 4.46952522e-01 -7.55915105e-01 -1.58941045e-01 -5.27796745e-01 -6.35999084e-01 4.57396805e-01 5.81205130e-01 4.27810907e-01 -9.75139320e-01 -6.12160087e-01 1.06052592e-01 1.51338996e-02 -8.75455499e-01 -2.42012024e-01 5.48828304e-01 -5.94048381e-01 -8.88685226e-01 -7.11796761e-01 -8.06576371e-01 6.84717178e-01 1.55880898e-01 1.11608791e+00 1.27155185e-02 -4.08982247e-01 -9.47238356e-02 -3.44070494e-01 -1.91423073e-01 3.13281268e-01 6.12057686e-01 -3.07719857e-01 4.15036529e-02 3.03905159e-01 -6.47292018e-01 -8.11662793e-01 3.06527819e-02 -5.59331417e-01 -1.80953175e-01 5.55229068e-01 7.78132558e-01 1.09361184e+00 -3.16149175e-01 1.52614653e-01 -7.58089781e-01 2.43089393e-01 -7.66505003e-01 -8.80089700e-01 1.04103684e-01 -2.64145106e-01 2.92361915e-01 7.31977463e-01 -7.29351565e-02 -4.44311559e-01 2.46401817e-01 -2.55762726e-01 -6.39434934e-01 -1.37438059e-01 5.65462649e-01 2.01213155e-02 -1.95619613e-01 7.34065533e-01 1.94563091e-01 -2.43947119e-01 -7.55730629e-01 2.48130143e-01 4.03318346e-01 4.80385840e-01 -7.20863879e-01 8.02224934e-01 5.63569605e-01 -1.01949297e-01 -5.90405405e-01 -8.50687027e-01 -6.08465254e-01 -3.59228075e-01 2.07177952e-01 5.96242666e-01 -1.10963988e+00 -6.53418660e-01 3.70985419e-01 -7.80042529e-01 -7.58087993e-01 -3.92349601e-01 3.13890517e-01 -4.69966829e-01 2.35664785e-01 -3.89829695e-01 -6.73259735e-01 -2.70008087e-01 -1.06068027e+00 1.43071306e+00 1.82849228e-01 -7.11477995e-02 -8.27515483e-01 4.33892310e-01 2.52743661e-01 2.35744029e-01 3.20901245e-01 6.15492702e-01 -7.35232413e-01 -7.96776652e-01 -3.79828513e-01 -2.24416971e-01 1.75639540e-01 -3.27204525e-01 -2.36566663e-01 -1.21701896e+00 -3.68516326e-01 -4.27337587e-01 -5.90025127e-01 1.30709636e+00 2.46695653e-01 1.62952614e+00 -2.76815802e-01 -2.82527924e-01 1.03122401e+00 1.59884441e+00 -2.45018557e-01 7.01683760e-01 6.43469512e-01 6.87410593e-01 2.40154654e-01 5.40847540e-01 7.12229848e-01 7.03810692e-01 7.78963268e-01 5.42016685e-01 -1.14125989e-01 -5.78486267e-03 -4.42900538e-01 2.08207577e-01 1.80673361e-01 2.03875393e-01 -1.67623729e-01 -1.07705855e+00 8.79530668e-01 -1.92881727e+00 -8.02809715e-01 3.47371221e-01 2.43845963e+00 8.07795644e-01 9.39528272e-02 2.67786771e-01 1.49840057e-01 3.72121006e-01 2.93281645e-01 -4.67362195e-01 -1.61422580e-01 1.57170027e-01 5.05512595e-01 7.76367545e-01 6.41727448e-01 -1.40626502e+00 9.24958825e-01 4.99072647e+00 9.35704052e-01 -1.19403553e+00 1.35195747e-01 8.61122966e-01 -5.06015718e-01 -2.20106900e-01 -5.88329509e-03 -9.60345387e-01 4.76595640e-01 9.07282710e-01 1.15533486e-01 6.92816973e-01 1.13883746e+00 -1.15718897e-02 -2.44573772e-01 -1.20754814e+00 1.16066813e+00 -9.52258427e-03 -1.38763261e+00 -2.43092299e-01 2.08415449e-01 7.01392412e-01 5.81292987e-01 2.31692180e-01 3.30294311e-01 1.81042612e-01 -9.74630952e-01 1.00108171e+00 1.42379537e-01 5.93346477e-01 -9.88657117e-01 5.01596570e-01 2.78463334e-01 -1.16044784e+00 -2.03047872e-01 -4.87676173e-01 -3.37214582e-03 -1.26964092e-01 8.19304705e-01 -9.02997196e-01 4.82247770e-01 7.86238432e-01 6.83500051e-01 -7.89913476e-01 1.55269301e+00 -2.56798059e-01 5.16346574e-01 -7.92317390e-01 -8.32519457e-02 6.29975677e-01 7.11075142e-02 3.02045822e-01 1.06946206e+00 3.47399831e-01 -4.25833404e-01 3.97963107e-01 8.75480056e-01 -3.07476819e-01 -6.48064911e-03 -5.60466111e-01 1.52229398e-01 6.47954822e-01 1.22132659e+00 -6.01021290e-01 -1.41521782e-01 -1.07889444e-01 9.98571098e-01 9.61954236e-01 1.07199870e-01 -9.25676465e-01 -4.87698168e-01 7.88786054e-01 9.94546860e-02 6.81846082e-01 -2.70429283e-01 -4.43079770e-01 -9.64784384e-01 4.77718234e-01 -5.20531535e-01 4.58622098e-01 -3.07351589e-01 -1.14871454e+00 4.60282266e-01 -1.14925966e-01 -1.14037371e+00 -2.29446828e-01 -4.80002314e-01 -6.20411277e-01 8.56821418e-01 -1.88545287e+00 -9.76173937e-01 -3.94291580e-01 4.64371979e-01 3.99358153e-01 2.17065603e-01 7.01676786e-01 5.59404433e-01 -6.66373730e-01 1.01127338e+00 2.42410958e-01 1.87490582e-01 5.24261951e-01 -1.37540853e+00 5.48281908e-01 7.55207777e-01 2.51912355e-01 4.59985226e-01 4.35312539e-01 -2.66481370e-01 -1.16740870e+00 -1.21317399e+00 9.86863732e-01 -3.33755970e-01 4.22602832e-01 -7.00275898e-01 -6.60254002e-01 7.99215794e-01 -4.83022630e-02 3.83183420e-01 5.82258940e-01 3.07671815e-01 -4.34702098e-01 -2.79280335e-01 -1.09751856e+00 5.11134863e-01 9.65686202e-01 -5.62202692e-01 -2.26924509e-01 5.81024230e-01 4.41246092e-01 -5.49407482e-01 -5.39132714e-01 1.80501655e-01 3.38905245e-01 -8.98887157e-01 9.92541790e-01 -2.30273232e-01 4.36908156e-01 -3.12737852e-01 -2.24026456e-01 -1.13783550e+00 -3.27531219e-01 -6.50756478e-01 -1.04755506e-01 9.59562302e-01 6.54921412e-01 -6.58777535e-01 8.70124459e-01 4.10104007e-01 -1.92956179e-01 -1.15846479e+00 -1.04855239e+00 -7.96923697e-01 4.31629755e-02 -4.07381147e-01 5.37909985e-01 6.24121487e-01 -1.77254528e-01 -5.63334208e-03 -1.93892658e-01 3.20773602e-01 8.20680618e-01 1.93822503e-01 7.63030291e-01 -1.01328540e+00 -4.96470332e-01 -3.87643129e-01 -5.23534119e-01 -1.01452088e+00 1.39683545e-01 -1.10257161e+00 2.36212909e-01 -1.59868765e+00 1.43748254e-01 -7.49204218e-01 -3.13436091e-01 8.56634855e-01 -6.14571534e-02 3.85623515e-01 2.49324128e-01 1.61936045e-01 -8.72209847e-01 7.31804729e-01 7.37344623e-01 -1.54662520e-01 -2.36410260e-01 -1.08387142e-01 -6.01745188e-01 5.60938776e-01 8.76493335e-01 -6.02312982e-01 -3.14949572e-01 -5.82411408e-01 2.23889202e-01 -4.36725140e-01 6.68828964e-01 -1.41936290e+00 2.45821387e-01 1.70197278e-01 5.49023509e-01 -5.67340374e-01 6.49825394e-01 -5.97603142e-01 -1.41604632e-01 3.08301926e-01 -4.26543415e-01 3.14338878e-02 3.08059990e-01 5.76290727e-01 -1.58134907e-01 -1.33341029e-01 8.10005069e-01 -1.27118051e-01 -7.88135290e-01 4.50029850e-01 2.97865987e-01 1.73616409e-01 1.07782364e+00 -1.31225158e-02 -2.73599803e-01 -1.05342381e-01 -5.96357882e-01 4.05520022e-01 5.36144972e-01 2.30680600e-01 4.98747468e-01 -1.24909377e+00 -5.90844929e-01 2.31032252e-01 1.74273208e-01 4.20062482e-01 1.16256058e-01 8.13709319e-01 -7.78070390e-01 1.90642849e-01 -1.13684431e-01 -5.54262400e-01 -8.23612034e-01 3.27294350e-01 4.06338453e-01 -3.67417455e-01 -6.71667218e-01 1.07252848e+00 4.59358767e-02 -5.10535717e-01 5.18589616e-01 -3.62691045e-01 1.83660865e-01 1.48474872e-02 6.14910781e-01 2.87513494e-01 3.06925625e-01 -4.01532888e-01 -6.11883581e-01 3.96105111e-01 -2.71921039e-01 -1.46130472e-01 1.66603386e+00 1.83396727e-01 7.16816932e-02 4.09167558e-01 1.63874435e+00 -9.56814736e-02 -1.46096849e+00 -1.14941448e-01 -2.87385192e-05 -5.02488494e-01 1.86723694e-02 -7.42964149e-01 -1.06933331e+00 7.87703753e-01 6.42276883e-01 -1.04551487e-01 9.85982120e-01 8.89871716e-02 8.73771846e-01 5.44947386e-01 4.20260906e-01 -1.08657444e+00 -1.91796064e-01 5.09745777e-01 8.00289810e-01 -1.17882049e+00 -1.38585716e-01 -5.80901653e-02 -4.67711300e-01 7.72906899e-01 4.29598272e-01 -5.64189315e-01 5.59657335e-01 1.38973534e-01 -1.61351323e-01 -1.35697708e-01 -6.42218351e-01 -2.62651145e-01 2.09065244e-01 2.81186044e-01 1.54377103e-01 1.97482064e-01 -6.52619600e-02 3.43642771e-01 -3.14825386e-01 -4.80566248e-02 -2.60953829e-02 1.02761841e+00 -4.67494696e-01 -1.05263436e+00 -1.06899396e-01 3.77029181e-01 -2.50270218e-01 -2.42663503e-01 -3.96423221e-01 4.63704884e-01 6.23329170e-02 4.65721399e-01 6.07369691e-02 -3.59967291e-01 1.70790911e-01 8.84101391e-02 3.78275335e-01 -5.04277408e-01 -5.97104371e-01 -2.28620648e-01 6.50797710e-02 -8.91526937e-01 1.04316592e-01 -8.03245604e-01 -1.29774523e+00 -3.35551560e-01 1.02340117e-01 -5.35544269e-02 6.57698214e-01 7.70342231e-01 6.79210067e-01 2.66561270e-01 6.37727380e-01 -1.39819312e+00 -7.39849508e-01 -6.73734248e-01 -2.30645061e-01 2.54870534e-01 4.70347643e-01 -6.87250793e-01 -3.42025220e-01 -4.55069512e-01]
[8.014788627624512, -1.8310340642929077]
436ea698-06f7-47b2-89b4-4facfa6ddb0d
value-aware-transformers-for-1-5d-data
null
null
https://openreview.net/forum?id=S3qhbZwzq3H
https://openreview.net/pdf?id=S3qhbZwzq3H
Value-aware transformers for 1.5d data
Sparse sequential highly-multivariate data of the form characteristic of hospital in-patient investigation and treatment poses a considerable challenge for representation learning. Such data is neither faithfully reducible to 1d nor dense enough to constitute multivariate series. Conventional models compromise their data by requiring these forms at the point of input. Building on contemporary sequence-modelling architectures we design a value-aware transformer, prompting a reconceptualisation of our data as 1.5-dimensional: a token-value form both respecting its sequential nature and augmenting it with a quantifier. Experiments focused on sequential in-patient laboratory data up to 48hrs after hospital admission show that the value-aware transformer performs favourably versus competitive baselines on in-hospital mortality and length-of-stay prediction within the MIMIC-III dataset.
['Parashkev Nachev', 'Amy Nelson', 'Amy R Tso', 'Timothy J Roberts', 'James F Cann']
2021-09-29
null
null
null
null
['length-of-stay-prediction']
['medical']
[ 4.63479072e-01 -5.28443046e-02 -3.74078900e-01 -4.89892274e-01 -8.00254524e-01 -5.36523700e-01 5.16754568e-01 7.73271620e-01 -5.19508123e-01 7.30557323e-01 6.46694541e-01 -8.97759318e-01 -5.92507064e-01 -6.43317044e-01 -4.95663166e-01 -5.30289352e-01 -7.65811920e-01 9.60904658e-01 -3.72457594e-01 -1.78714335e-01 -1.09930262e-01 6.51512861e-01 -1.14373422e+00 7.85037994e-01 5.53306825e-02 1.10257208e+00 -2.99153209e-01 9.27835882e-01 -9.71708149e-02 1.25183272e+00 -4.05827880e-01 4.30279896e-02 4.37068582e-01 -5.18389940e-01 -8.04924011e-01 2.05922890e-02 2.52197087e-02 -2.49411657e-01 -3.92764539e-01 2.58421600e-01 6.30449235e-01 -4.70146835e-02 9.13230360e-01 -1.02207589e+00 -1.44230992e-01 5.52677631e-01 -4.68969345e-02 5.36599398e-01 4.73436117e-01 2.51053929e-01 9.21815813e-01 -6.50216997e-01 3.65215600e-01 1.04278505e+00 1.13931334e+00 3.51464421e-01 -1.58000588e+00 -4.19256091e-02 -1.17915198e-01 -1.85920790e-01 -1.18346167e+00 -3.48289639e-01 4.98712122e-01 -7.03528523e-01 1.32664192e+00 6.53234363e-01 8.46931100e-01 1.30910933e+00 3.64415199e-01 4.25374925e-01 9.07155871e-01 -1.64153814e-01 3.47756118e-01 -3.86155128e-01 3.51450853e-02 4.53649580e-01 2.01320365e-01 4.67083246e-01 -4.24668849e-01 -6.28766716e-01 9.97749805e-01 7.91178167e-01 -3.33692320e-02 -4.04108942e-01 -1.80244589e+00 7.90082812e-01 -1.56349719e-01 1.84917480e-01 -7.31613696e-01 1.63170278e-01 8.45290959e-01 7.34544992e-01 2.35724688e-01 5.19751966e-01 -9.12513852e-01 -4.43959594e-01 -1.08782005e+00 2.18845040e-01 8.53032768e-01 9.64195788e-01 2.12647468e-01 1.17640063e-01 -3.43273371e-01 5.22684813e-01 -1.41876131e-01 7.02456236e-02 8.38578939e-01 -6.68807149e-01 3.77701163e-01 6.03648841e-01 1.86527502e-02 -5.64435244e-01 -8.45011175e-01 -3.90174508e-01 -1.36926806e+00 -2.54430652e-01 3.71922106e-01 -5.33182621e-02 -8.17021251e-01 1.41962671e+00 3.83817852e-02 4.34000939e-01 5.81600750e-03 4.35775012e-01 6.66777849e-01 3.04413199e-01 2.37228215e-01 -5.80855370e-01 1.39676964e+00 -2.01503471e-01 -4.01056230e-01 1.76219106e-01 1.11446702e+00 -2.52432287e-01 6.94541395e-01 6.22226000e-01 -1.02966499e+00 -3.93907189e-01 -5.32806516e-01 8.82683620e-02 -8.94402862e-02 -4.76219833e-01 7.11295426e-01 4.34411645e-01 -1.04680741e+00 7.67469466e-01 -9.14946556e-01 -6.12878092e-02 4.53338116e-01 4.27421808e-01 -7.45943189e-01 -7.89509621e-03 -9.79806125e-01 8.69708359e-01 2.95342237e-01 -1.33630529e-01 -1.00848126e+00 -1.33531916e+00 -9.07335401e-01 8.34781677e-02 -3.85812260e-02 -1.20181966e+00 1.13504171e+00 -4.29641724e-01 -9.14479494e-01 9.36062515e-01 -4.31296527e-02 -8.32404554e-01 6.39933228e-01 9.08534974e-03 -4.20675784e-01 7.49347508e-02 -2.64869183e-01 -3.54335620e-03 8.47877204e-01 -7.09855795e-01 -4.30344939e-01 -2.43286222e-01 -9.97054726e-02 -9.85400975e-02 -1.07437081e-03 6.81130355e-03 2.84873307e-01 -1.00681746e+00 -5.63306585e-02 -7.38586009e-01 -9.51347888e-01 -2.60352224e-01 -2.52246439e-01 -1.97555814e-02 3.90058428e-01 -4.19512451e-01 1.50778902e+00 -1.97156799e+00 1.94625899e-01 2.51027316e-01 5.92644513e-01 2.87297298e-04 -9.49473828e-02 7.70132303e-01 -7.71175086e-01 2.88686335e-01 -4.89828199e-01 -5.02697349e-01 -2.14067042e-01 4.94482785e-01 -2.22764432e-01 7.61771798e-01 3.40080321e-01 1.00915813e+00 -9.59574282e-01 -6.18406117e-01 3.01734298e-01 3.23250890e-01 -9.21018481e-01 6.67648494e-01 -1.25621352e-02 4.42575693e-01 -4.87479061e-01 7.35964119e-01 1.86303407e-01 -4.14744407e-01 3.30831081e-01 7.28286803e-02 4.52391803e-02 2.01648846e-01 -9.41259444e-01 1.72736824e+00 -3.46594959e-01 1.05013803e-01 -3.21516305e-01 -1.25326288e+00 9.04511988e-01 5.62919438e-01 1.22226989e+00 -5.40093064e-01 3.07237118e-01 7.35299885e-02 3.35907452e-02 -4.64273691e-01 1.80081576e-01 -8.74569297e-01 -2.52691507e-01 4.00133133e-01 -1.92805707e-01 -2.78095871e-01 1.43920183e-01 2.47368693e-01 1.46528959e+00 -2.19120860e-01 8.11222732e-01 -4.14764583e-01 3.46198678e-01 -1.21303312e-01 5.02095580e-01 7.33974993e-01 -3.96192148e-02 9.88661110e-01 8.61058772e-01 -9.81151581e-01 -1.16155899e+00 -1.06428003e+00 -4.03535515e-01 6.96905851e-01 -8.20421934e-01 -6.04463398e-01 -2.72675790e-02 -5.40217400e-01 3.61124963e-01 4.63863045e-01 -9.87323225e-01 -1.96501195e-01 -7.11025894e-01 -9.60332751e-01 5.78625917e-01 7.73004949e-01 -6.79041326e-01 -9.16235685e-01 -9.93270576e-01 7.61151731e-01 -7.59884808e-03 -6.91057980e-01 -2.08559245e-01 8.27658415e-01 -1.31306982e+00 -1.20723021e+00 -7.21946836e-01 -4.94297862e-01 4.55465823e-01 -4.22443300e-01 1.61111784e+00 3.82893905e-02 -4.91649538e-01 4.62916762e-01 -3.11508358e-01 -4.96816874e-01 -6.61689878e-01 -5.05794436e-02 1.39212072e-01 1.29137812e-02 2.71525770e-01 -7.42913485e-01 -7.23880351e-01 -1.99640095e-01 -1.12378120e+00 -1.51784644e-01 5.66047192e-01 1.05699730e+00 7.59770930e-01 -4.28376347e-01 7.21289217e-01 -1.02507353e+00 6.47138536e-01 -1.06831980e+00 -7.18055442e-02 -8.70770514e-02 -6.16409183e-01 1.08702697e-01 8.43440711e-01 -3.92082959e-01 -2.89699763e-01 2.61947453e-01 -8.29700753e-02 -7.19379425e-01 -2.31777430e-01 6.78826630e-01 3.05919379e-01 6.69099510e-01 6.56215549e-01 3.30439180e-01 4.81618822e-01 -6.71910346e-01 1.36828959e-01 3.16680998e-01 5.21786630e-01 -6.67711377e-01 3.96601230e-01 3.60856354e-01 5.55588603e-01 -7.65100598e-01 -2.40302265e-01 -7.54687428e-01 -8.18174422e-01 3.66212845e-01 5.07906973e-01 -1.01056039e+00 -8.83926570e-01 2.21007347e-01 -8.22452366e-01 -3.59617978e-01 -9.39160883e-01 4.25725669e-01 -1.00117004e+00 3.85631293e-01 -7.48054147e-01 -6.07175112e-01 -3.20347905e-01 -7.58319914e-01 9.54601288e-01 -8.36899161e-01 -6.19884849e-01 -1.22793555e+00 3.56902421e-01 -2.71945953e-01 5.53169966e-01 9.53400373e-01 1.40254486e+00 -1.04915404e+00 -8.63798484e-02 -4.97346073e-01 1.39272809e-02 3.86312544e-01 1.54364258e-01 -3.46803188e-01 -5.17521441e-01 -3.37002903e-01 1.80195734e-01 -2.49461100e-01 7.24196672e-01 4.20763075e-01 1.45157969e+00 -4.61095393e-01 -1.08738504e-01 8.12435865e-01 1.43188858e+00 9.02243927e-02 5.68940997e-01 1.74821854e-01 5.45380652e-01 4.49714601e-01 7.79601336e-02 1.34893775e+00 4.01671350e-01 1.03224449e-01 4.24194992e-01 -1.82434231e-01 1.46458060e-01 -1.60745233e-02 -1.24259830e-01 7.98212588e-01 1.72977224e-02 1.38065433e-02 -1.21943986e+00 7.51307070e-01 -1.64691293e+00 -1.11563802e+00 -3.62267375e-01 2.21038842e+00 1.09877801e+00 -2.08908878e-02 4.50707734e-01 4.36967432e-01 -4.82242294e-02 2.07775578e-01 -3.09046179e-01 -6.16273582e-01 -1.12426080e-01 2.91844964e-01 3.75974596e-01 1.39984086e-01 -9.01010573e-01 1.18176498e-01 7.42154503e+00 2.52174199e-01 -9.41364169e-01 -8.09626579e-02 9.27746117e-01 -4.10215616e-01 -3.68698061e-01 -2.99858034e-01 -2.57490277e-01 5.99041462e-01 1.69817281e+00 -2.91190535e-01 1.89548835e-01 7.18383253e-01 3.43489379e-01 2.76629329e-01 -1.76878023e+00 1.26321411e+00 -2.90885538e-01 -1.68473577e+00 2.50672787e-01 -5.10381442e-03 3.87974143e-01 3.77438031e-02 -1.75394956e-02 4.09563005e-01 3.13552946e-01 -1.67471647e+00 6.42413676e-01 7.47521162e-01 1.31305802e+00 -5.72616160e-01 6.76822960e-01 2.82574952e-01 -1.04743934e+00 -2.17252642e-01 -9.44452658e-02 -2.01290220e-01 2.58282065e-01 4.76773232e-01 -1.06817210e+00 6.06377184e-01 3.23183656e-01 9.59956765e-01 -4.04167175e-01 8.58896017e-01 6.60372734e-01 6.63268983e-01 -3.83570820e-01 5.07223010e-01 3.24188232e-01 1.33659333e-01 2.15514719e-01 1.68827486e+00 3.07018548e-01 3.36452335e-01 2.43741438e-01 3.08694333e-01 2.00918779e-01 7.11357966e-02 -9.61445153e-01 -1.82163000e-01 1.02851100e-01 6.26027346e-01 -2.36989722e-01 -6.63602948e-01 -5.61140716e-01 4.53889579e-01 1.05730638e-01 2.37454817e-01 -4.97804880e-01 -1.37614068e-02 7.02632427e-01 6.11244678e-01 3.01877677e-01 -8.37549567e-02 -6.31863952e-01 -1.20797539e+00 -2.93314934e-01 -1.31671739e+00 9.84864473e-01 -1.63693383e-01 -1.46386325e+00 8.08869779e-01 1.36586025e-01 -1.50838709e+00 -7.79979408e-01 -6.09260082e-01 -2.73979187e-01 9.92202103e-01 -1.39993715e+00 -9.65822518e-01 5.99483065e-02 1.01391530e+00 3.63320500e-01 -1.03733152e-01 1.42610407e+00 2.36975417e-01 -2.55065680e-01 3.63216579e-01 1.17975354e-01 7.11348886e-03 2.10624993e-01 -1.12945652e+00 3.92251283e-01 3.29928041e-01 -3.03226531e-01 7.11075842e-01 6.49661064e-01 -3.78695041e-01 -1.77959216e+00 -1.09115100e+00 1.02853262e+00 -1.06574285e+00 8.21257830e-01 -5.05268909e-02 -9.76326168e-01 8.51335466e-01 -1.94489151e-01 3.50894213e-01 1.03946745e+00 9.76756364e-02 -5.88259578e-01 4.06409167e-02 -1.12230575e+00 1.69584885e-01 1.08335674e+00 -6.56837583e-01 -1.01505303e+00 3.15840572e-01 5.87557793e-01 -1.62382960e-01 -1.43857706e+00 5.12137890e-01 4.26460445e-01 -8.82846355e-01 1.38169527e+00 -1.39526927e+00 5.70981741e-01 1.66031823e-01 -3.29855263e-01 -1.14384484e+00 -5.34287870e-01 -8.59042108e-01 -4.41680640e-01 4.27485764e-01 1.13335676e-01 -5.67264438e-01 7.64722943e-01 5.94316065e-01 -2.08000556e-01 -1.11718643e+00 -1.13089323e+00 -8.20499301e-01 3.86866599e-01 -5.91270089e-01 8.52858543e-01 9.93085802e-01 3.17099929e-01 2.67151207e-01 -4.79269356e-01 -2.68189311e-01 5.24131238e-01 7.47268228e-03 4.46307868e-01 -1.47206903e+00 -3.61159623e-01 -5.52280545e-01 -6.01008832e-01 -6.19805098e-01 -2.00416669e-02 -1.03654087e+00 -4.69850957e-01 -1.57713723e+00 2.84113765e-01 -5.87593734e-01 -9.14622247e-01 4.26221162e-01 8.21808949e-02 -1.26125470e-01 -3.62652391e-02 2.42475465e-01 -3.72542351e-01 2.88123250e-01 8.69825661e-01 3.78147960e-02 -1.13847129e-01 -4.04577143e-02 -5.62494993e-01 3.55043977e-01 6.60635233e-01 -6.66992307e-01 -4.93124992e-01 -1.84384793e-01 1.81129113e-01 8.82697344e-01 3.54136944e-01 -7.13037074e-01 -1.64505377e-01 -4.91958767e-01 3.92303199e-01 -6.16350472e-01 3.33677500e-01 -9.88608181e-01 3.75235140e-01 7.20456541e-01 -5.36813498e-01 6.76136851e-01 1.78514302e-01 7.14900970e-01 -1.66392371e-01 2.03682125e-01 7.08997309e-01 -3.65635157e-01 -4.18521404e-01 7.29210913e-01 -5.24812162e-01 4.13666666e-01 8.03668618e-01 -1.96474329e-01 -1.48492642e-02 -9.91362631e-02 -8.21651578e-01 -7.04557747e-02 3.21547896e-01 3.38608116e-01 7.83145428e-01 -1.11418390e+00 -1.18473160e+00 6.63355410e-01 2.69741356e-01 3.10057197e-02 3.63072753e-01 1.04105651e+00 -5.40464997e-01 5.23244143e-01 -2.17330351e-01 -5.19218445e-01 -9.76486862e-01 1.03595924e+00 1.91193938e-01 -5.41484535e-01 -8.55854034e-01 6.17999852e-01 1.35629654e-01 -1.87297821e-01 2.49684453e-01 -7.48712003e-01 2.03443933e-02 3.81861418e-01 6.56032681e-01 1.53954312e-01 2.82158911e-01 -4.14924055e-01 -5.32760620e-01 -1.90926582e-01 4.24836800e-02 1.52366996e-01 1.63348913e+00 2.37439528e-01 -6.13493510e-02 6.44414365e-01 1.64464402e+00 -6.25937283e-01 -9.70104694e-01 -2.47274861e-01 2.40499794e-01 -3.05856526e-01 -2.40092576e-01 -6.27129316e-01 -5.39346397e-01 8.35480928e-01 3.07538182e-01 2.05102459e-01 9.03441072e-01 -2.01070830e-01 5.97191334e-01 4.90459621e-01 3.94627035e-01 -4.45400149e-01 1.70839787e-01 7.08402097e-01 1.11841059e+00 -9.63989615e-01 -6.02445379e-02 3.02462846e-01 -5.42532384e-01 1.02565658e+00 -1.86960265e-01 -2.02385455e-01 9.15378094e-01 4.77152050e-01 2.89625069e-03 -2.51257777e-01 -1.34710050e+00 1.43421724e-01 1.29136518e-01 6.93636239e-01 6.27016842e-01 4.00862724e-01 -1.39503062e-01 7.97538579e-01 -3.11429381e-01 2.47670665e-01 6.45268500e-01 1.23810470e+00 8.70305449e-02 -9.92834032e-01 -2.55775958e-01 9.97153878e-01 -7.37783551e-01 -5.08878767e-01 1.17890447e-01 6.70051217e-01 -2.74630487e-01 6.36234999e-01 1.30198136e-01 -1.61882579e-01 6.36720300e-01 3.77888411e-01 3.10578614e-01 -8.19219053e-01 -1.03942347e+00 -1.34294644e-01 2.31792897e-01 -6.94434285e-01 -1.04344822e-01 -8.12037945e-01 -1.15249574e+00 -4.07155901e-01 4.64375377e-01 1.27746180e-01 3.34691018e-01 4.87133861e-01 4.96636450e-01 7.08781004e-01 5.81142962e-01 -6.87765837e-01 -1.22454941e+00 -8.25756729e-01 -6.96273148e-01 7.29790807e-01 1.01154006e+00 -1.98530450e-01 -3.34545642e-01 3.30060214e-01]
[7.967310428619385, 6.2387166023254395]
7731d09c-202b-4646-914b-a73d5fffd015
multi-color-balance-for-color-constancy
2105.10228
null
https://arxiv.org/abs/2105.10228v1
https://arxiv.org/pdf/2105.10228v1.pdf
Multi-color balance for color constancy
In this paper, we propose a novel multi-color balance adjustment for color constancy. The proposed method, called "n-color balancing," allows us not only to perfectly correct n target colors on the basis of corresponding ground truth colors but also to correct colors other than the n colors. In contrast, although white-balancing can perfectly adjust white, colors other than white are not considered in the framework of white-balancing in general. In an experiment, the proposed multi-color balancing is demonstrated to outperform both conventional white and multi-color balance adjustments including Bradford's model.
['Hitoshi Kiya', 'Yuma Kinoshita', 'Teruaki Akazawa']
2021-05-21
null
null
null
null
['color-constancy']
['computer-vision']
[-2.81228591e-02 -6.30309165e-01 1.19072525e-02 -2.22640201e-01 -3.48667771e-01 -6.31413162e-01 1.57980159e-01 -9.84968990e-02 -2.73315042e-01 8.35905135e-01 -1.65333733e-01 -2.68019348e-01 1.31132886e-01 -6.21493936e-01 -2.85518050e-01 -6.27476394e-01 6.61315203e-01 3.38356644e-01 2.17240021e-01 -5.30136943e-01 4.44068491e-01 6.03297174e-01 -1.50367284e+00 -2.01634914e-01 1.32447636e+00 6.31669343e-01 -1.17752723e-01 8.24613392e-01 -3.25192481e-01 4.94979799e-01 -6.89396679e-01 -5.09487629e-01 5.44172645e-01 -8.64154935e-01 -1.43998846e-01 5.86402789e-02 9.99748826e-01 -1.93605587e-01 4.50869761e-02 1.51456201e+00 5.33901513e-01 3.96727681e-01 2.77131230e-01 -1.59154177e+00 -1.18060350e+00 1.45949304e-01 -1.33945823e+00 -1.23061396e-01 1.57229587e-01 -6.45454600e-02 9.25564706e-01 -6.18433356e-01 3.42502385e-01 1.42434907e+00 3.50307584e-01 7.67604053e-01 -1.28281462e+00 -1.16156852e+00 3.73688012e-01 1.15704879e-01 -1.24576807e+00 -1.51808903e-01 9.28611517e-01 -2.60569215e-01 4.72974963e-03 7.35966861e-01 8.45963538e-01 2.61770755e-01 2.57324696e-01 4.19488907e-01 1.83246434e+00 -8.64607632e-01 2.28314385e-01 -6.54228181e-02 6.51362017e-02 6.86491966e-01 8.55800807e-01 3.23119581e-01 -6.57670736e-01 -3.81998233e-02 7.05853581e-01 -2.56105244e-01 -1.86175510e-01 -4.66674417e-01 -1.22770119e+00 3.58364046e-01 6.86383486e-01 8.89060497e-02 -3.20986435e-02 3.24764431e-01 -1.29053980e-01 2.68049482e-02 6.72269642e-01 3.21094632e-01 -3.66589464e-02 2.90761799e-01 -1.04360175e+00 1.30531505e-01 4.57227260e-01 8.78637135e-01 9.53571081e-01 5.34049332e-01 -2.67131716e-01 6.49606764e-01 1.89864144e-01 1.08143735e+00 2.56335229e-01 -9.17561769e-01 2.26247057e-01 7.00368822e-01 7.28184640e-01 -1.02604377e+00 -7.16912568e-01 -2.09542945e-01 -9.27135289e-01 1.04010522e+00 6.28175735e-01 -2.14198887e-01 -1.30946279e+00 1.82758403e+00 6.07434690e-01 3.07963882e-03 -2.50391662e-01 1.22730911e+00 5.39688349e-01 3.07480633e-01 1.30956754e-01 -1.14379406e-01 1.07289469e+00 -1.01393425e+00 -9.40513372e-01 -2.33689785e-01 -1.98556349e-01 -1.11291170e+00 1.21787775e+00 4.09168690e-01 -1.00675261e+00 -6.41602337e-01 -1.13495970e+00 6.53549656e-02 -5.13976157e-01 1.80002406e-01 7.07209289e-01 1.22977984e+00 -1.34845185e+00 6.27217069e-02 -2.04345420e-01 -3.34301382e-01 -2.01654449e-01 1.96655188e-03 -1.05896577e-01 -5.42842560e-02 -9.51455474e-01 1.12182951e+00 1.65235683e-01 5.67177176e-01 -3.01527590e-01 -5.45305490e-01 -4.31591660e-01 -1.28272250e-01 3.30821484e-01 -4.79091913e-01 9.11213696e-01 -1.28106761e+00 -1.69757450e+00 9.20161426e-01 -4.48450536e-01 2.82863230e-01 1.05645859e+00 -2.79481232e-01 -7.72220731e-01 -7.63778109e-03 -9.35653150e-02 6.26855314e-01 4.76695418e-01 -2.10680914e+00 -6.59811020e-01 -2.28751466e-01 2.75999010e-01 3.48189294e-01 6.26532584e-02 1.22075919e-02 -7.38883913e-01 -5.32221377e-01 4.21962380e-01 -1.13185394e+00 -1.61255702e-01 3.74960274e-01 -8.78863752e-01 4.04456079e-01 4.31827337e-01 -6.21639073e-01 1.05653250e+00 -1.91700506e+00 -1.32910088e-01 6.62761807e-01 4.66452599e-01 -7.22958893e-02 -1.54265881e-01 1.42357749e-04 -2.27897257e-01 -6.07796945e-02 1.17981192e-02 -8.04326981e-02 2.77996182e-01 -5.56483157e-02 9.37245190e-02 6.31150842e-01 -2.85764515e-01 6.58082426e-01 -9.81306076e-01 -4.60537851e-01 4.85740095e-01 2.38327175e-01 -2.45219007e-01 -3.07180174e-02 1.57772034e-01 3.79667282e-01 2.01445475e-01 8.63680899e-01 1.33638632e+00 -2.66585462e-02 2.79649794e-01 -2.76784867e-01 -3.28520477e-01 -8.63556147e-01 -1.51731718e+00 1.02678490e+00 -1.90217108e-01 7.32003629e-01 1.29659459e-01 -1.07033767e-01 1.08595169e+00 -1.49996415e-01 4.46649253e-01 -1.03773844e+00 -5.63357351e-03 1.99776560e-01 1.13358162e-01 1.09227866e-01 1.04449987e+00 -5.00369132e-01 1.95371941e-01 3.45022470e-01 -6.92533731e-01 -4.46996003e-01 5.40307820e-01 9.26087201e-02 1.53729364e-01 3.44802231e-01 2.17138931e-01 -5.17609060e-01 5.30801833e-01 1.26335114e-01 7.88044751e-01 6.05382383e-01 -5.23425341e-01 6.29361451e-01 3.25361133e-01 -2.76006222e-01 -9.62072372e-01 -1.27242482e+00 1.92978770e-01 1.30589354e+00 1.07892382e+00 1.16656862e-01 -7.04995215e-01 -1.25584200e-01 3.48231584e-01 7.14666486e-01 -9.57444906e-01 -1.55529939e-02 -2.34116152e-01 -8.83508146e-01 1.81458607e-01 3.56433988e-01 6.47121787e-01 -6.43510818e-01 -4.92697328e-01 -2.02555671e-01 -1.09761499e-01 -7.00135887e-01 -7.63801754e-01 -1.37809366e-01 -4.35372382e-01 -1.45246243e+00 -8.18629682e-01 -2.17636541e-01 9.05566335e-01 7.53178537e-01 1.08263576e+00 4.61500257e-01 -2.24058330e-01 2.92818606e-01 -3.56712401e-01 -5.74624121e-01 -2.52820313e-01 -5.03745437e-01 -5.21790646e-02 1.46515906e-01 3.86981696e-01 1.42636076e-01 -8.11734915e-01 6.06317282e-01 -8.50391090e-01 4.78850424e-01 3.04639280e-01 4.28963572e-01 5.79307795e-01 -1.15006715e-01 1.95271969e-01 -1.14976454e+00 5.26832879e-01 2.12297633e-01 -9.11592185e-01 9.03979599e-01 -8.60344648e-01 -1.64993376e-01 5.03440261e-01 -4.49922979e-01 -1.23588634e+00 -3.79415184e-01 4.86287445e-01 -1.86410844e-01 2.66827136e-01 -1.52242765e-01 -2.06540570e-01 -6.39660239e-01 4.89726007e-01 -1.77346244e-01 -4.28184956e-01 -3.62553269e-01 7.99818039e-01 9.91556868e-02 6.75083399e-01 -7.02393591e-01 1.14839733e+00 5.93201578e-01 3.06029290e-01 -3.19630355e-01 -4.83071059e-01 -2.50691503e-01 -6.93513930e-01 -7.72430241e-01 8.62113595e-01 -6.53824091e-01 -8.67414355e-01 8.07877243e-01 -6.66543961e-01 -3.73302191e-01 -3.78766051e-03 3.04328620e-01 -2.90343732e-01 2.67834455e-01 -4.86351192e-01 -1.07489586e+00 -6.15071245e-02 -8.61524224e-01 6.97089076e-01 8.98483634e-01 6.32257834e-02 -8.99102747e-01 1.10698238e-01 1.28348306e-01 4.64594543e-01 5.71007311e-01 9.22083020e-01 1.49950281e-01 -3.06579769e-01 -1.20077655e-01 -6.97821021e-01 -5.00824787e-02 4.39444155e-01 7.08808661e-01 -8.30607772e-01 -2.54107565e-01 -5.83874524e-01 9.49636549e-02 6.04864895e-01 5.70616305e-01 7.12810457e-01 3.36018771e-01 -2.80796718e-02 6.44007504e-01 2.08515477e+00 5.09841263e-01 5.50686002e-01 6.39721751e-01 9.28682029e-01 4.45857257e-01 6.36720777e-01 4.57504541e-01 4.28109050e-01 5.67656696e-01 4.44566876e-01 -9.04018402e-01 -3.87839943e-01 -1.52809158e-01 -3.08977574e-01 6.06475413e-01 -3.08308631e-01 -2.39238724e-01 -8.45321000e-01 5.80297969e-02 -1.67460680e+00 -6.84855282e-01 -6.72734797e-01 2.18976831e+00 6.92289770e-01 1.60943773e-02 8.78021717e-02 -4.03277278e-02 1.48846650e+00 -4.64238487e-02 -8.03074658e-01 -4.17055756e-01 -5.17690599e-01 1.45163000e-01 9.18500543e-01 6.69070542e-01 -7.83868790e-01 8.32776010e-01 7.61116362e+00 3.89261067e-01 -1.04704666e+00 3.97321023e-02 5.88026285e-01 -7.41012171e-02 -6.17867410e-01 1.39088973e-01 -4.22026813e-01 5.21568120e-01 1.77729607e-01 -4.29497778e-01 6.86611891e-01 5.57261765e-01 3.86092573e-01 -7.99092889e-01 -4.38840985e-01 1.09020674e+00 3.66698891e-01 -6.79742634e-01 2.51733184e-01 -2.66618401e-01 1.22088039e+00 -7.91782439e-01 1.31546602e-01 -1.40718058e-01 7.81638086e-01 -6.22411311e-01 1.36750901e+00 7.09710956e-01 9.20996964e-01 -6.63881958e-01 2.46346951e-01 -3.94514203e-01 -1.28956151e+00 1.60046637e-01 -5.31649053e-01 8.28292519e-02 7.51810223e-02 6.61673427e-01 1.49129942e-01 7.62743354e-01 5.39952695e-01 7.52007514e-02 -8.17995429e-01 1.48291957e+00 -4.14696753e-01 5.38144335e-02 1.21628694e-01 6.74055740e-02 4.02546972e-02 -5.79564691e-01 4.58434112e-02 9.38484967e-01 3.16428751e-01 3.25175077e-02 1.45749658e-01 7.63276517e-01 2.82709207e-02 1.96050137e-01 2.63880521e-01 3.92531127e-01 4.31806505e-01 1.31759000e+00 -1.16456735e+00 -5.93284249e-01 -3.88821602e-01 9.94718969e-01 -9.15869549e-02 8.36009681e-01 -1.01145709e+00 -4.39386755e-01 6.79086387e-01 -2.05666885e-01 -3.21736991e-01 -3.49321872e-01 -8.10425699e-01 -1.05936241e+00 -4.15355772e-01 -6.50742829e-01 2.88571775e-01 -1.50852633e+00 -1.16679978e+00 3.83941174e-01 -5.26156612e-02 -1.39105499e+00 5.93763769e-01 -7.86802649e-01 -6.02183044e-01 1.29600453e+00 -1.84979200e+00 -1.26517415e+00 -8.95618260e-01 6.85141861e-01 -8.95155966e-03 2.12592095e-01 3.58802319e-01 2.02104166e-01 -6.69192016e-01 5.35643160e-01 5.59508145e-01 -1.81331366e-01 1.31123173e+00 -1.68177366e+00 1.64490074e-01 1.21501863e+00 -1.25327885e-01 4.36251223e-01 1.01340699e+00 -6.76190317e-01 -1.03642428e+00 -5.41009963e-01 3.99999171e-01 1.23517932e-02 5.51535010e-01 -1.83769632e-02 -5.51011920e-01 3.05739671e-01 2.74256885e-01 -2.11279050e-01 8.02750647e-01 5.22811189e-02 -6.09729886e-01 -5.36431849e-01 -1.23785961e+00 7.55972266e-01 7.19028473e-01 -3.16176981e-01 -1.35991827e-01 2.06790455e-02 1.91457346e-01 -5.63851357e-01 -1.69233292e-01 -4.37313914e-02 7.91781425e-01 -1.25181448e+00 8.65839779e-01 -3.97960842e-01 1.36665717e-01 -8.68053615e-01 -1.44605204e-01 -1.52695227e+00 -5.69098651e-01 -4.32437420e-01 5.60572505e-01 1.06295240e+00 1.79956540e-01 -6.31738722e-01 6.12614453e-01 9.89707291e-01 7.66591951e-02 7.85161182e-02 -7.62561262e-01 -6.16128087e-01 7.50737339e-02 -3.32089812e-02 8.65024447e-01 9.63416934e-01 -2.92448312e-01 -3.06694090e-01 -7.55397677e-01 2.74310023e-01 9.12138939e-01 5.48543632e-01 8.11611176e-01 -1.22048938e+00 -3.93170118e-02 -8.28765988e-01 -9.54612866e-02 -4.16582733e-01 1.29126050e-02 -3.12593281e-01 2.53111005e-01 -1.80311596e+00 4.94184524e-01 -4.87534583e-01 -8.28704178e-01 3.50451112e-01 -8.10140848e-01 9.37284827e-01 6.15091860e-01 4.03722599e-02 -4.29729432e-01 1.03180237e-01 1.78789628e+00 -3.69042218e-01 -1.80441573e-01 -3.25396925e-01 -1.03794754e+00 3.44202816e-01 7.39329338e-01 5.18755689e-02 -2.66944408e-01 -4.17024583e-01 4.19532329e-01 -1.87532514e-01 2.05695733e-01 -1.07035744e+00 7.76891932e-02 -9.36439514e-01 7.70472646e-01 -4.28764284e-01 7.18236938e-02 -9.40511346e-01 3.91475856e-01 5.04978776e-01 -1.08779527e-01 4.54408526e-01 4.52216268e-01 3.50843757e-01 1.06812663e-01 2.18898393e-02 1.13473201e+00 -2.45249476e-02 -9.86916661e-01 -1.27934247e-01 -4.28840928e-02 -1.65910468e-01 1.12244081e+00 -5.19483328e-01 -9.20597315e-01 -3.79728854e-01 -5.80007493e-01 8.03606883e-02 1.00940835e+00 2.72317618e-01 2.21215561e-01 -1.73127365e+00 -5.06056488e-01 -2.74405088e-02 2.53450632e-01 -7.69465387e-01 4.98442471e-01 5.85902154e-01 -1.01029146e+00 8.83821994e-02 -7.23860919e-01 -6.24602474e-02 -1.17474163e+00 7.16581404e-01 4.54344630e-01 2.10978255e-01 -2.37739816e-01 6.80047989e-01 4.14788991e-01 -2.05420509e-01 -1.67430684e-01 -2.92645842e-01 1.72200222e-02 -1.31596297e-01 5.26077509e-01 7.09791541e-01 -1.45035580e-01 -8.49856913e-01 -3.55310887e-01 1.13420761e+00 5.81831694e-01 -2.20531523e-01 7.41705239e-01 -4.66830939e-01 -4.21157897e-01 6.25023961e-01 3.11634094e-01 4.68107730e-01 -1.25363946e+00 8.72608274e-02 -2.67974734e-01 -1.23864055e+00 -2.02332541e-01 -1.31466687e+00 -1.40286064e+00 5.84280968e-01 8.85542035e-01 1.48693383e-01 1.47445297e+00 -7.56734431e-01 2.95369714e-01 -1.48959920e-01 4.97504681e-01 -1.32862365e+00 2.49843788e-03 -1.53104020e-02 5.24606943e-01 -1.11582899e+00 2.21776992e-01 -3.57714593e-01 -8.55689049e-01 1.24042153e+00 1.18070686e+00 -1.20095775e-01 2.34261602e-01 1.66633070e-01 7.91800201e-01 1.66815668e-01 -1.99226126e-01 -4.81161773e-01 5.35634756e-01 7.05587149e-01 4.40607995e-01 4.54244673e-01 -4.11509007e-01 -2.09847197e-01 1.25850989e-02 -1.03556886e-01 7.81322420e-01 6.44725978e-01 -6.06757522e-01 -1.04595733e+00 -1.10415316e+00 2.98478246e-01 8.68497863e-02 6.80770203e-02 -5.45362711e-01 9.06048954e-01 2.55761504e-01 1.22734845e+00 1.25708282e-01 -4.14985716e-01 5.28858185e-01 -2.12566093e-01 4.98689085e-01 -7.42827654e-02 -3.95890892e-01 2.22007588e-01 -2.74800479e-01 -4.84540045e-01 -7.02825785e-01 -1.57807410e-01 -9.93801713e-01 -1.14056957e+00 -5.08110762e-01 6.11390211e-02 6.41868770e-01 4.97040868e-01 -4.31122154e-01 6.29741311e-01 5.16159832e-01 -7.04873919e-01 -2.52417959e-02 -7.19936490e-01 -1.15302527e+00 6.78310394e-01 2.09166870e-01 -9.24989700e-01 -3.82248193e-01 1.16744284e-02]
[10.502717971801758, -2.5415666103363037]
23beb4bb-b4ff-48e7-8022-bd55ec668ba1
kernelized-multi-graph-matching
2210.05206
null
https://arxiv.org/abs/2210.05206v1
https://arxiv.org/pdf/2210.05206v1.pdf
Kernelized multi-graph matching
Multigraph matching is a recent variant of the graph matching problem. In this framework, the optimization procedure considers several graphs and enforces the consistency of the matches along the graphs. This constraint can be formalized as a cycle consistency across the pairwise permutation matrices, which implies the definition of a universe of vertex~\citep{pachauri2013solving}. The label of each vertex is encoded by a sparse vector and the dimension of this space corresponds to the rank of the bulk permutation matrix, the matrix built from the aggregation of all the pairwise permutation matrices. The matching problem can then be formulated as a non-convex quadratic optimization problem (QAP) under constraints imposed on the rank and the permutations. In this paper, we introduce a novel kernelized multigraph matching technique that handles vectors of attributes on both the vertices and edges of the graphs, while maintaining a low memory usage. We solve the QAP problem using a projected power optimization approach and propose several projectors leading to improved stability of the results. We provide several experiments showing that our method is competitive against other unsupervised methods.
['S. Takerkart', 'Guillaume Auzias', 'Rohit Yadav', 'François-Xavier Dupé']
2022-10-11
null
null
null
null
['graph-matching']
['graphs']
[ 3.68481427e-01 1.49357766e-01 -1.79185256e-01 -1.48841888e-01 -4.53666508e-01 -6.84118688e-01 6.00827634e-01 3.35488021e-01 -1.57642946e-01 4.11578655e-01 1.02662943e-01 -1.14972191e-03 -7.00737774e-01 -7.60181308e-01 -6.98413670e-01 -8.03299069e-01 -5.85140251e-02 6.64132893e-01 8.33429694e-02 -5.20258695e-02 3.20234686e-01 5.44082940e-01 -1.40212524e+00 1.47135444e-02 7.45093942e-01 6.40589058e-01 1.18849106e-01 4.24544394e-01 1.06734097e-01 1.77579790e-01 1.04799382e-02 -6.97866619e-01 6.31224632e-01 -2.59273589e-01 -7.91754901e-01 2.92682528e-01 8.49535584e-01 3.41458321e-01 -5.57069838e-01 1.38078380e+00 3.01983207e-01 1.73854426e-01 5.76092601e-01 -1.69649553e+00 -3.98599207e-01 6.04281902e-01 -6.98207796e-01 -1.86837003e-01 6.21877313e-01 -4.38561708e-01 1.56035936e+00 -5.40609777e-01 8.70835125e-01 9.72662032e-01 5.57064891e-01 -1.20730266e-01 -1.56311679e+00 -2.99307346e-01 -1.36135900e-02 4.84608620e-01 -1.68383956e+00 -1.74596220e-01 6.62223935e-01 -5.67293286e-01 5.89779377e-01 6.75271690e-01 5.80757618e-01 4.91506010e-01 1.80595368e-01 2.13552922e-01 7.56419301e-01 -5.26410580e-01 2.13660911e-01 1.30198300e-01 2.56671041e-01 8.67020011e-01 6.11863673e-01 -4.04134877e-02 -5.72772384e-01 -6.20059490e-01 3.04931372e-01 -1.71931401e-01 -2.54345596e-01 -1.07429934e+00 -1.23737311e+00 7.99485862e-01 2.33317956e-01 3.29104155e-01 -2.32561156e-01 5.08820601e-02 1.99301973e-01 2.94261396e-01 1.99408665e-01 2.11897999e-01 4.49905135e-02 3.37767035e-01 -7.75267780e-01 1.79571450e-01 1.15187335e+00 1.13320839e+00 1.17362869e+00 -3.90958041e-01 -2.32217591e-02 6.32514179e-01 4.80152220e-01 5.45633137e-01 -1.33896872e-01 -8.64728928e-01 6.54532075e-01 8.56315553e-01 -1.72055081e-01 -1.63902831e+00 -3.93044472e-01 -2.23428965e-01 -9.42657769e-01 -1.75020888e-01 4.54763591e-01 1.67991936e-01 -4.28576410e-01 1.87372184e+00 4.02215481e-01 3.94226462e-01 -2.28833616e-01 5.69379210e-01 5.13163328e-01 5.20414948e-01 -2.82029718e-01 -3.39701056e-01 1.26627076e+00 -7.75725782e-01 -4.89096642e-01 -2.33885273e-02 5.21809459e-01 -5.87387741e-01 3.39586169e-01 1.52802259e-01 -9.05076444e-01 -1.59735277e-01 -1.13810921e+00 2.03031480e-01 -2.54347175e-01 3.38844322e-02 5.73625267e-01 7.49906898e-01 -1.30362451e+00 8.98645699e-01 -3.69152397e-01 -3.73584479e-01 -2.76797950e-01 6.62036955e-01 -9.91690278e-01 -2.25884989e-01 -9.49712038e-01 7.61720657e-01 5.32669842e-01 2.38954514e-01 -2.06118986e-01 -5.34524739e-01 -9.73473787e-01 1.54293790e-01 4.12548333e-01 -6.98254049e-01 3.26539516e-01 -6.58466220e-01 -9.40550268e-01 1.11297476e+00 -1.85318336e-01 -1.68146834e-01 4.05099779e-01 3.39901298e-01 -2.71085918e-01 7.05942884e-02 2.36503363e-01 6.71927854e-02 7.82566071e-01 -1.02181029e+00 -5.46395719e-01 -4.05026227e-01 -1.33556966e-03 1.30705118e-01 -2.30233163e-01 4.69103046e-02 -9.07464266e-01 -3.68956894e-01 3.25452298e-01 -1.42484784e+00 -2.57494241e-01 -3.70596886e-01 -5.69896340e-01 -1.76439404e-01 1.81402683e-01 -6.77978635e-01 1.40695405e+00 -2.25041127e+00 8.11429381e-01 1.00120866e+00 3.30999017e-01 -2.90750295e-01 -3.31504881e-01 8.67973268e-01 -2.82280028e-01 -2.29479179e-01 -4.49507922e-01 -1.98870078e-01 2.60926723e-01 1.82246625e-01 6.13878071e-02 1.06630313e+00 -1.98114157e-01 7.14061439e-01 -8.77174675e-01 -4.29999650e-01 4.13248502e-02 2.44183734e-01 -4.23087478e-01 9.36158374e-02 1.26155108e-01 9.03702974e-02 -2.88077980e-01 3.20440501e-01 9.59011018e-01 -3.38636011e-01 7.52673268e-01 -4.62651789e-01 -6.98704422e-02 -2.49047995e-01 -1.97657955e+00 1.66447735e+00 5.55995442e-02 3.49554926e-01 1.55643299e-01 -1.05890143e+00 6.81837380e-01 1.01053484e-01 8.85081232e-01 -3.21151167e-01 -3.54097411e-02 2.39562392e-01 -4.59088152e-03 -1.72290355e-01 5.19030094e-01 3.25717121e-01 -2.29718909e-01 4.45620894e-01 6.48611337e-02 9.69337821e-02 6.67845190e-01 5.38867950e-01 1.15821707e+00 -1.82839856e-01 3.54537696e-01 -7.23228395e-01 6.98236048e-01 -2.21718997e-01 5.62740803e-01 6.56426013e-01 2.83306718e-01 2.64629543e-01 7.57184446e-01 -2.81729251e-01 -1.11730540e+00 -9.51119065e-01 -1.14358597e-01 7.84832120e-01 2.69040644e-01 -6.68347657e-01 -6.58142388e-01 -3.89670372e-01 4.53120917e-01 5.97040802e-02 -7.02605009e-01 -7.44503811e-02 -5.20647645e-01 -8.70017946e-01 1.39934510e-01 -1.39491819e-02 5.78493392e-03 -5.86361945e-01 4.98852395e-02 5.30383103e-02 -2.41501089e-02 -1.09142029e+00 -7.17295885e-01 8.54096375e-03 -5.96619606e-01 -1.31793237e+00 -2.89683580e-01 -8.96030486e-01 9.32549059e-01 1.43979549e-01 8.46014023e-01 1.31130025e-01 -3.27563703e-01 3.65596890e-01 -2.54418999e-01 2.25511804e-01 -2.13947177e-01 8.19590092e-02 2.59944469e-01 5.98704875e-01 -1.16107548e-02 -6.80258811e-01 -1.03617631e-01 2.82549471e-01 -8.72033656e-01 9.11933705e-02 3.60273570e-01 6.65645003e-01 7.93576121e-01 6.02240190e-02 -3.57269756e-02 -1.22003806e+00 5.03179610e-01 -4.87338632e-01 -8.78691077e-01 6.74581885e-01 -6.24668360e-01 5.63097656e-01 3.09165567e-01 -1.86693147e-01 -4.06827807e-01 4.84888375e-01 4.43667054e-01 -1.68198749e-01 3.74523640e-01 5.11538088e-01 -5.53207457e-01 -7.00250626e-01 1.48835167e-01 1.61627933e-01 -2.02289298e-01 -4.94149268e-01 5.49671948e-01 3.09196472e-01 4.08484608e-01 -5.43003798e-01 1.10979378e+00 5.45508564e-01 6.42702818e-01 -7.26774812e-01 -4.54324931e-01 -8.01058769e-01 -7.82801092e-01 -2.29412064e-01 5.36518157e-01 -6.38888121e-01 -9.34330881e-01 3.21923524e-01 -8.41184974e-01 1.94225147e-01 6.28162250e-02 4.49702710e-01 -5.43214202e-01 9.35183287e-01 -4.16357040e-01 -5.20811260e-01 -1.61283165e-01 -1.04537916e+00 7.58436084e-01 -9.45323408e-02 -1.42888576e-01 -9.42529976e-01 5.31757593e-01 2.03735486e-01 5.66133000e-02 3.39798421e-01 1.10479295e+00 -6.51547432e-01 -7.82228529e-01 -3.83122146e-01 -4.13043857e-01 -1.30932495e-01 -1.73384622e-02 4.42519933e-02 -4.03471589e-01 -6.44690692e-01 -3.67376655e-01 3.65949303e-01 7.63347745e-01 1.70815110e-01 7.34797657e-01 -1.94601923e-01 -5.59542716e-01 7.97357917e-01 1.77466631e+00 -2.01655805e-01 2.52070606e-01 2.26413563e-01 9.90913033e-01 7.43777633e-01 4.11412179e-01 4.29730833e-01 4.66583520e-01 9.81938779e-01 3.61411422e-01 1.74443319e-01 2.13942364e-01 -8.25914368e-02 6.52408004e-02 1.06363106e+00 -3.63501683e-02 -1.86716273e-01 -7.78512061e-01 5.26540279e-01 -2.19193602e+00 -8.82791400e-01 -5.34600019e-01 2.46492290e+00 3.47572088e-01 -4.20449108e-01 4.72628847e-02 -1.04896855e-02 9.88401949e-01 2.05134600e-01 -1.70967653e-01 -3.39404672e-01 -2.73047358e-01 9.40823108e-02 8.31969857e-01 7.87696898e-01 -1.02821994e+00 4.06086713e-01 6.39921808e+00 5.19328296e-01 -4.28218693e-01 -9.40509290e-02 -8.59591439e-02 8.78025070e-02 -5.21966755e-01 3.71559381e-01 -6.98899925e-01 5.06533086e-01 7.63812065e-01 -5.83913147e-01 8.00721884e-01 5.00154674e-01 -2.90981829e-01 -5.67530058e-02 -1.18326354e+00 1.04551184e+00 3.73627990e-01 -1.04184604e+00 -1.53588019e-02 4.14476365e-01 8.84461761e-01 7.61241838e-02 -1.93244189e-01 -1.18604630e-01 1.24871120e-01 -6.45901740e-01 4.18785870e-01 5.75275600e-01 5.84372699e-01 -8.10840786e-01 4.45534438e-01 9.77615826e-03 -1.56560338e+00 1.48678035e-01 -4.28645581e-01 2.94027507e-01 1.72415778e-01 5.60391545e-01 -3.45542431e-01 1.12133741e+00 3.27422976e-01 6.75400853e-01 -4.52474385e-01 1.24770951e+00 -9.03949328e-03 -1.89639889e-02 -3.35637957e-01 1.89451501e-01 2.99106892e-02 -1.00098932e+00 7.89826155e-01 1.10102355e+00 2.69044548e-01 -5.87003455e-02 3.81228238e-01 4.47381854e-01 -1.64958373e-01 5.61069727e-01 -7.15227246e-01 3.13130952e-02 4.22049522e-01 1.37730992e+00 -6.46991193e-01 -1.02934435e-01 -5.18099308e-01 1.07384622e+00 6.92185700e-01 1.73781648e-01 -4.83250320e-01 -3.39531749e-01 6.84475422e-01 -1.64300010e-01 2.69525945e-01 -1.02353916e-01 6.90548271e-02 -1.25197983e+00 3.48608017e-01 -8.67745042e-01 7.02640355e-01 -1.96236670e-01 -1.26424873e+00 4.95075732e-01 2.88076885e-02 -1.03516877e+00 -6.33622929e-02 -6.36418164e-01 -4.21426654e-01 8.54959369e-01 -1.17185295e+00 -9.39050794e-01 -1.03708148e-01 5.99219382e-01 -4.36806738e-01 -9.48103964e-02 8.64211380e-01 5.09733558e-01 -7.93927670e-01 5.16518235e-01 3.72853398e-01 -1.00263253e-01 4.80705649e-01 -1.37203288e+00 3.10299784e-01 1.05042541e+00 4.20576900e-01 5.65122545e-01 7.57959247e-01 -7.55000830e-01 -1.76058710e+00 -8.69018137e-01 1.25199258e+00 -2.34320164e-01 9.12880957e-01 -4.11156923e-01 -7.97526181e-01 7.53216863e-01 2.61971150e-02 -4.82711457e-02 7.69050777e-01 3.04966837e-01 -3.78501028e-01 -1.81161687e-01 -8.32509637e-01 3.71602029e-01 1.16841578e+00 -6.02559149e-01 -1.96355984e-01 5.33512175e-01 3.53484273e-01 -2.16886193e-01 -1.16704285e+00 3.41210574e-01 4.89739567e-01 -6.09855592e-01 8.59929323e-01 -6.58830345e-01 -2.32408702e-01 -5.16909719e-01 -2.67325610e-01 -1.16453147e+00 -8.14001560e-01 -6.95601285e-01 -1.25256762e-01 1.19503450e+00 4.40053016e-01 -5.77709734e-01 7.99943507e-01 7.01262712e-01 2.64426470e-01 -3.97521973e-01 -1.11146069e+00 -7.57428288e-01 -4.83075351e-01 -4.75176796e-02 5.83031476e-01 1.06568646e+00 3.32102150e-01 2.24517167e-01 -6.68271124e-01 5.40745556e-01 1.07207155e+00 5.22378385e-01 6.40418530e-01 -1.37783265e+00 -5.82781255e-01 -2.77131975e-01 -8.62728596e-01 -3.44735682e-01 4.99244034e-01 -1.41599679e+00 -2.08298951e-01 -1.37378871e+00 6.32210314e-01 -3.40761542e-01 -1.45026550e-01 2.68810868e-01 -6.83965310e-02 6.01315685e-02 2.34144345e-01 8.79288614e-02 -6.14170492e-01 2.00154692e-01 7.23048925e-01 -1.98708400e-01 -3.39168422e-02 -3.86330783e-02 -3.38868022e-01 3.14624220e-01 3.68879408e-01 -4.61543888e-01 -1.51604384e-01 -2.97713608e-01 5.72414517e-01 1.32545054e-01 4.86314334e-02 -6.43715620e-01 6.18584037e-01 -1.24275051e-01 -2.69877851e-01 -4.78261352e-01 2.38989606e-01 -9.61835682e-01 1.03176343e+00 4.46757555e-01 -3.12731750e-02 3.02409261e-01 -1.19540326e-01 7.15956092e-01 -1.05777621e-01 -3.61394912e-01 5.72886825e-01 9.76899192e-02 -5.77968538e-01 4.64666545e-01 1.17501631e-01 -8.76007080e-02 1.09503329e+00 -8.74836296e-02 -1.22917451e-01 -4.02380750e-02 -6.99760616e-01 3.86419922e-01 6.85163021e-01 3.25564116e-01 2.94168651e-01 -1.57033610e+00 -8.55313778e-01 2.54696280e-01 2.71749884e-01 -5.44975281e-01 2.73596317e-01 1.01290429e+00 -4.12732780e-01 4.05730218e-01 -2.75770009e-01 -3.58990818e-01 -1.57810128e+00 6.82497084e-01 1.65124655e-01 -4.43391591e-01 -5.23750246e-01 5.92135370e-01 -5.84417991e-02 -4.37095463e-01 1.49743035e-01 2.45906919e-01 -3.22751194e-01 2.61158943e-01 1.76240250e-01 6.26541197e-01 1.77900031e-01 -1.03159046e+00 -5.83515763e-01 7.68031359e-01 2.92446930e-02 1.61697250e-02 1.32598519e+00 1.07404981e-02 -8.13796520e-01 1.01278819e-01 1.34868324e+00 3.38886231e-01 -5.99236608e-01 -4.43621129e-01 2.56479084e-01 -5.97792089e-01 -3.10847938e-01 -8.05197209e-02 -1.04790235e+00 7.31211156e-02 1.65395901e-01 1.20147139e-01 8.48658442e-01 -1.54222492e-02 2.70480037e-01 3.44743341e-01 4.83001232e-01 -9.36880708e-01 -5.19336998e-01 2.99933285e-01 7.39577889e-01 -9.12970006e-01 2.05011517e-01 -8.17902446e-01 -3.16170692e-01 8.87282014e-01 1.99095890e-01 -2.82198071e-01 7.08707750e-01 -5.82959168e-02 -4.12898242e-01 -2.70337313e-01 -4.72152144e-01 -2.36388013e-01 7.72392392e-01 1.99840844e-01 3.27709615e-02 3.30147892e-01 -6.57115221e-01 1.55244485e-01 -1.34182528e-01 -5.58710635e-01 2.14418322e-01 4.67897654e-01 -1.33550093e-01 -1.63284075e+00 -2.00936303e-01 4.66917485e-01 -8.52656811e-02 -1.76210720e-02 -5.67277193e-01 5.10349751e-01 5.85015193e-02 7.82485068e-01 -1.71495348e-01 -4.39214468e-01 4.87764657e-01 -1.76729903e-01 5.75916946e-01 -5.69210827e-01 -4.76425380e-01 -2.18085513e-01 2.07435384e-01 -6.72134936e-01 -3.95173937e-01 -9.19207454e-01 -7.03946233e-01 -4.62766677e-01 -5.25638402e-01 4.54723477e-01 5.43584824e-01 7.96074450e-01 3.02604139e-01 7.38239363e-02 8.00904572e-01 -5.87819159e-01 -4.33160931e-01 -5.65723419e-01 -1.07672143e+00 6.88057005e-01 -1.25071436e-01 -5.10418713e-01 -4.21996504e-01 -3.20151597e-01]
[7.164320468902588, 5.194514274597168]
95fb191a-c49e-43d6-84c5-ad3f1e6a6d7a
divide-and-denoise-learning-from-noisy-labels
null
null
https://openreview.net/forum?id=LJPfn2jgIrW
https://openreview.net/pdf?id=LJPfn2jgIrW
Divide and Denoise: Learning from Noisy Labels in Fine-grained Entity Typing with Cluster-wise Loss Correction
Fine-grained Entity Typing(FET) has witnessed great progress since distant supervision was introduced, but still suffers from label noise. Existing noise control methods applied to FET rely on predicted distribution and deals instances isolately, thus suffers from confirmation bias. In this work, We propose to tackle the two limitations with a cluster based loss correction framework named Feature Cluster Loss Correction(FCLC). FCLC first train a coarse backbone model as feature extractor and noise estimator. Then perform loss correction on each cluster to learn directly from noisy labels. Experimental results on three public datasets show FCLC achieves the best performance over existing competitive systems. Auxiliary experiments further show FCLC is stable to hyper-paramerters and even works with extream scneriaos like no clean data is available.
['Anonymous']
2021-11-16
null
null
null
acl-arr-november-2021-11
['entity-typing']
['natural-language-processing']
[ 2.74540409e-02 -5.05382232e-02 -1.39907792e-01 -5.40180326e-01 -1.14476860e+00 -2.57491142e-01 4.82757390e-01 1.82416603e-01 -8.04873586e-01 1.07978737e+00 1.27137706e-01 1.97078437e-01 3.25195156e-02 -6.85999870e-01 -8.21038425e-01 -7.52921700e-01 1.68069810e-01 6.02972269e-01 2.76316553e-01 1.56599939e-01 -1.86757922e-01 1.55643120e-01 -1.42926347e+00 5.70601106e-01 1.08471692e+00 8.24454844e-01 -4.50867005e-02 4.37124014e-01 -1.26369327e-01 9.73762155e-01 -6.49502099e-01 -7.68354475e-01 1.85983628e-01 -7.54871592e-02 -8.42931032e-01 -6.05297327e-01 3.89596999e-01 -4.86864746e-02 -1.38779506e-01 1.23204553e+00 9.17690933e-01 -6.04903046e-03 5.46530604e-01 -1.45638561e+00 -5.16776383e-01 9.99136984e-01 -4.94617194e-01 6.67120889e-02 2.73513440e-02 -1.43414453e-01 1.19255197e+00 -1.03998077e+00 8.90952647e-01 1.24998248e+00 1.39460099e+00 8.25880885e-01 -1.54932237e+00 -9.75059271e-01 2.80012250e-01 3.29938740e-01 -1.69755828e+00 -2.08423018e-01 3.46436232e-01 -6.22892305e-02 1.05699933e+00 2.44653165e-01 -6.66884333e-02 1.47691989e+00 -1.66072696e-01 1.11928833e+00 1.13960075e+00 -4.06640679e-01 1.48882642e-01 3.72378051e-01 3.26794326e-01 4.81488973e-01 4.73009408e-01 5.63107841e-02 -5.62544346e-01 -3.98134261e-01 9.05959159e-02 3.64584522e-03 -1.65891930e-01 -1.59065515e-01 -9.23487246e-01 5.57088912e-01 3.19487661e-01 2.75788844e-01 -4.72475849e-02 3.39668691e-02 7.39731669e-01 5.39642751e-01 5.60301363e-01 2.23670244e-01 -8.79312754e-01 -2.45481342e-01 -8.38410974e-01 4.13483053e-01 8.36482406e-01 1.15979660e+00 7.55158603e-01 -4.18990612e-01 -4.26810592e-01 9.70483959e-01 -8.34699273e-02 5.27105391e-01 4.63054836e-01 -5.85179389e-01 7.34210730e-01 6.10172272e-01 -3.75931524e-02 -7.93566525e-01 -6.91635311e-01 -5.02511442e-01 -1.04112196e+00 -4.64594513e-02 3.86487842e-01 -3.97382140e-01 -8.05047095e-01 1.84763646e+00 4.72267002e-01 4.81031209e-01 -3.09927016e-01 5.81566095e-01 7.44239509e-01 1.74355671e-01 4.13045257e-01 9.50410813e-02 1.04197919e+00 -7.40335822e-01 -8.70522082e-01 1.72333777e-01 1.15374351e+00 -4.09947336e-01 1.13057494e+00 6.24684453e-01 -6.14505708e-01 -2.68606126e-01 -8.59113276e-01 -6.29210770e-02 -5.41729927e-01 1.13246344e-01 4.26090986e-01 7.05976427e-01 -7.82787800e-01 6.79350853e-01 -1.11059546e+00 -3.16297024e-01 7.04837799e-01 3.92617524e-01 -6.66756570e-01 -3.26585144e-01 -1.24269533e+00 8.26921642e-01 5.13181388e-01 1.35539174e-01 -5.74570298e-01 -1.06796420e+00 -7.14012742e-01 7.78975785e-02 5.01171291e-01 -6.21616364e-01 1.10912168e+00 -6.61399126e-01 -1.06076717e+00 7.17346132e-01 -2.36653268e-01 -5.70907712e-01 8.59205306e-01 -5.09710789e-01 -4.77290124e-01 -5.56026161e-01 3.34109575e-01 2.00005844e-01 3.02342057e-01 -1.23444664e+00 -1.02240968e+00 -2.30138376e-01 -3.10968906e-01 -5.21155484e-02 -2.82019079e-01 1.75163671e-02 -5.17334104e-01 -6.46365881e-01 -9.30071156e-03 -7.73525894e-01 -1.76107064e-01 -4.49442744e-01 -7.89903581e-01 -5.22510469e-01 6.83302104e-01 -3.73689711e-01 1.57795298e+00 -1.88497448e+00 -1.73460379e-01 2.75080830e-01 2.74910778e-01 3.94366562e-01 -2.92969793e-02 3.15687686e-01 5.96619360e-02 1.24220483e-01 -2.30855495e-01 -6.37718797e-01 2.39814401e-01 2.59657264e-01 1.39160812e-01 4.77275282e-01 2.55953938e-01 7.57011533e-01 -8.07753742e-01 -5.71174264e-01 -2.61671275e-01 4.11506325e-01 -9.94361520e-01 2.35674247e-01 -6.03777915e-02 8.13759863e-02 -3.59064847e-01 7.50334561e-01 8.58413696e-01 -2.51539797e-01 2.61760890e-01 -3.82494837e-01 2.09831238e-01 2.23553956e-01 -1.41987467e+00 1.58213198e+00 -3.22850943e-01 2.24175602e-01 9.97980982e-02 -8.51094365e-01 6.54093087e-01 1.14309847e-01 2.77824610e-01 -4.84782100e-01 6.87868744e-02 3.09318632e-01 -3.36963356e-01 -5.93344808e-01 2.39611134e-01 -3.89549077e-01 -3.31641674e-01 7.22682253e-02 2.45277748e-01 7.04417467e-01 -1.45452786e-02 3.09172422e-01 1.49208295e+00 1.90052539e-01 9.04301181e-02 -9.07542035e-02 4.51746821e-01 -1.18460864e-01 1.29293299e+00 1.14465499e+00 -4.42667454e-01 8.39899004e-01 4.28335816e-01 -2.14250475e-01 -8.32814336e-01 -8.99028182e-01 -4.91644621e-01 1.37247920e+00 8.53698477e-02 -6.50250256e-01 -8.60025346e-01 -1.35529530e+00 3.55932266e-01 4.36607420e-01 -8.30458641e-01 -2.80795787e-02 -6.45814538e-01 -1.21226728e+00 1.06339216e+00 7.11338818e-01 3.42728347e-01 -1.06448042e+00 3.00142735e-01 2.67344892e-01 -3.81067723e-01 -1.02851546e+00 -3.08073670e-01 6.03511333e-01 -2.80112654e-01 -1.11859250e+00 -1.73936784e-01 -7.39035964e-01 4.85864282e-01 -1.66840732e-01 1.32432580e+00 2.32696071e-01 -5.42259574e-01 -1.62311956e-01 -4.15239394e-01 -4.55980033e-01 -1.45255834e-01 5.65790057e-01 2.37785310e-01 -6.74057603e-02 8.94077122e-01 -4.50490743e-01 -4.53764379e-01 2.61372030e-01 -6.60610497e-01 -4.29170340e-01 4.57241088e-01 1.23933148e+00 7.27020144e-01 5.13853393e-02 9.44045186e-01 -1.83698678e+00 2.35529944e-01 -7.23234653e-01 -3.70378017e-01 3.70385200e-01 -9.16137874e-01 1.20337903e-01 6.48140371e-01 4.08381410e-02 -1.45993865e+00 -1.34503320e-01 -4.25887257e-01 -1.36372507e-01 -4.71860975e-01 1.80236429e-01 -7.27381229e-01 2.95808822e-01 7.87232399e-01 -2.19682038e-01 -5.20357311e-01 -9.75569606e-01 3.19381177e-01 8.58686149e-01 5.42930007e-01 -7.45320380e-01 7.44378090e-01 4.37810212e-01 -3.64137173e-01 -2.87611276e-01 -1.10676122e+00 -6.78931117e-01 -6.66644275e-01 4.30266231e-01 4.54758734e-01 -1.16620564e+00 -7.82744169e-01 5.39665759e-01 -7.65171707e-01 -1.47034392e-01 -2.62290418e-01 3.54341865e-01 -1.88135803e-01 2.39279434e-01 -8.89124095e-01 -6.09556735e-01 -3.81787270e-01 -7.59150684e-01 1.04624665e+00 1.45297768e-02 -3.42024751e-02 -9.77901042e-01 2.11622894e-01 2.42682740e-01 2.63274103e-01 3.57830852e-01 5.81676841e-01 -9.87955451e-01 -4.47182238e-01 -4.54424888e-01 -4.03425455e-01 4.13433284e-01 -2.96171661e-02 -2.49659941e-01 -1.27089429e+00 -4.36681867e-01 -3.97948354e-01 -4.59073186e-01 1.20151484e+00 8.12743828e-02 1.16116595e+00 -2.00970873e-01 -7.00649500e-01 8.90325189e-01 1.69182122e+00 -4.29356903e-01 4.70522523e-01 5.36642969e-01 1.10195303e+00 2.49604300e-01 7.80560076e-01 4.01442945e-01 6.57561719e-01 5.84089279e-01 9.24993083e-02 -2.90310550e-02 -3.78733784e-01 -4.97244924e-01 2.47743621e-01 8.18197548e-01 1.12059854e-01 -2.63862848e-01 -8.59388173e-01 6.17108047e-01 -2.07834315e+00 -8.98342490e-01 -2.99784541e-01 1.81562161e+00 1.23137772e+00 1.77487791e-01 5.36592714e-02 7.65793994e-02 7.58094788e-01 -4.02944356e-01 -5.66729367e-01 -1.43848201e-02 -5.01976848e-01 1.46713659e-01 6.70226455e-01 2.26939216e-01 -1.23240364e+00 1.06649184e+00 5.61652899e+00 1.27465987e+00 -6.60493970e-01 4.99063998e-01 5.54252028e-01 -2.57625461e-01 -2.04838857e-01 -2.52721101e-01 -1.38479137e+00 6.68389142e-01 7.86871493e-01 1.28255844e-01 9.52767879e-02 8.80538821e-01 -1.49687946e-01 1.63217679e-01 -1.05973971e+00 8.95995021e-01 -1.78930923e-01 -9.38979447e-01 -2.38544583e-01 -3.13789278e-01 1.00916088e+00 4.52298731e-01 -1.99632481e-01 7.22146332e-01 8.01336884e-01 -9.93671358e-01 4.59198415e-01 5.67787826e-01 8.06899548e-01 -1.01156735e+00 1.35598290e+00 5.20073056e-01 -9.55913901e-01 -1.27033889e-02 -6.20857537e-01 2.39241689e-01 -1.77555069e-01 8.76358032e-01 -7.45140135e-01 5.94875216e-01 1.00724530e+00 6.65640771e-01 -5.48447192e-01 1.20305479e+00 -9.62500423e-02 9.28141654e-01 -4.25137132e-01 1.75792977e-01 -7.77158812e-02 3.25440019e-01 2.65240848e-01 1.79853141e+00 8.47629905e-02 -1.00071795e-01 2.48895496e-01 4.87599790e-01 -5.33028483e-01 2.01680690e-01 -1.99974149e-01 5.86879253e-01 8.43234062e-01 1.34697926e+00 -3.81706238e-01 -1.99298859e-01 -5.53070247e-01 1.03589487e+00 9.96971846e-01 2.73963988e-01 -8.36119533e-01 -6.15385354e-01 8.07499051e-01 5.21687791e-02 4.12389934e-01 5.30531764e-01 -5.22376359e-01 -1.46205258e+00 5.31273782e-02 -8.15936029e-01 7.75120020e-01 -1.88310277e-02 -1.99223888e+00 5.53413391e-01 -4.43523288e-01 -1.04622555e+00 2.33404249e-01 -4.17534739e-01 -4.00994033e-01 7.30604827e-01 -1.71689093e+00 -1.33262658e+00 -2.04631016e-01 5.87373435e-01 1.28789335e-01 -1.00785475e-02 8.14235449e-01 8.28024328e-01 -9.06717539e-01 1.59105515e+00 5.41361094e-01 3.13381761e-01 1.23360765e+00 -1.77716529e+00 3.28495592e-01 6.38258159e-01 -1.04584150e-01 5.34059465e-01 6.36259854e-01 -7.17806160e-01 -9.58999157e-01 -1.62243927e+00 1.19792497e+00 -9.16284919e-01 4.28974420e-01 -6.97851717e-01 -1.17975390e+00 7.21810222e-01 -1.95289940e-01 4.43778276e-01 7.61804938e-01 6.65053606e-01 -6.98546112e-01 -3.25005859e-01 -1.52160144e+00 7.74456635e-02 1.35606122e+00 -4.04472053e-01 -2.90841371e-01 2.90517986e-01 7.13901460e-01 -4.40775841e-01 -1.04184771e+00 4.31670517e-01 4.49311793e-01 -8.89657199e-01 7.37178087e-01 -9.13662612e-01 -2.25035623e-02 -3.59830648e-01 -1.72543362e-01 -1.32215214e+00 -4.11726177e-01 -3.37219417e-01 2.07365267e-02 1.88794363e+00 7.63140798e-01 -4.44202453e-01 8.95910323e-01 7.08173275e-01 -5.87669946e-02 -4.79013026e-01 -9.07700956e-01 -9.24189508e-01 2.49560043e-01 -6.04765117e-01 7.81446993e-01 1.36559439e+00 1.28607750e-01 3.09620023e-01 -5.94189942e-01 3.78395259e-01 8.43368828e-01 -2.37508252e-01 7.26658583e-01 -1.31968760e+00 -2.53358364e-01 -2.59128045e-02 -4.39926088e-01 -4.46841121e-01 3.18258375e-01 -1.20365977e+00 2.54396975e-01 -1.15457654e+00 5.57802379e-01 -9.92653131e-01 -7.83342779e-01 7.46430278e-01 -7.83727467e-01 4.79350358e-01 -1.03712054e-02 8.25790465e-02 -1.18217468e+00 4.25925791e-01 5.15671551e-01 4.47031073e-02 1.08659118e-02 9.93646458e-02 -7.58499265e-01 7.30872393e-01 7.31996298e-01 -8.71143341e-01 9.76998825e-04 -2.61863083e-01 2.89889872e-01 -5.98199546e-01 1.72022488e-02 -9.86240506e-01 3.45907956e-01 2.26192281e-01 3.77952278e-01 -5.39295912e-01 -2.95664608e-01 -8.28295052e-01 1.47322908e-01 8.35218579e-02 -4.71312702e-01 -2.84364522e-01 -8.12172517e-02 9.64508176e-01 -2.53963262e-01 -1.32388115e-01 7.42656469e-01 1.00327119e-01 -5.86069524e-01 3.26287866e-01 5.32276221e-02 6.60271823e-01 6.81598544e-01 1.86251760e-01 -3.35209310e-01 3.20955008e-01 -9.25136805e-01 4.67215925e-01 3.33352804e-01 3.19147408e-01 4.60938551e-02 -1.55389333e+00 -9.27460492e-01 1.43068492e-01 3.61251026e-01 1.04996592e-01 2.87872076e-01 7.47862697e-01 -1.28153279e-01 2.28602737e-01 3.31631213e-01 -5.47931314e-01 -1.37022722e+00 4.80050921e-01 3.20230365e-01 -5.89022398e-01 -7.00792730e-01 1.33263934e+00 1.44885495e-01 -1.03059876e+00 6.50443494e-01 -2.38733783e-01 3.43863480e-02 -2.51902873e-03 5.52236319e-01 6.65168881e-01 5.83847106e-01 -4.52358037e-01 -5.04842520e-01 1.77892551e-01 -4.99275267e-01 4.29795891e-01 1.44515431e+00 -2.45077416e-01 5.02288342e-02 2.74089575e-01 1.35594189e+00 -3.44812833e-02 -1.21602774e+00 -7.76108742e-01 5.77783644e-01 -4.17230278e-01 -8.66589993e-02 -1.13521576e+00 -9.17614818e-01 5.76941729e-01 6.92809463e-01 -2.18111146e-02 9.90564108e-01 -1.07289299e-01 8.90950322e-01 4.55040932e-01 4.15136188e-01 -1.28173840e+00 -4.50621635e-01 6.07275844e-01 2.58207679e-01 -1.51278877e+00 -2.10759431e-01 -4.40193504e-01 -6.13088727e-01 4.85064715e-01 8.98445547e-01 3.61749679e-02 9.58923519e-01 6.36603415e-01 1.14098869e-01 -7.08322823e-02 -1.06795168e+00 -2.97284514e-01 1.60014749e-01 8.10907900e-01 6.13597751e-01 1.71561763e-01 -3.13737124e-01 1.35109448e+00 -2.03999672e-02 6.73108101e-02 6.99097216e-02 6.74882472e-01 -2.60793537e-01 -1.40866339e+00 -1.11557402e-01 6.73285604e-01 -8.79407227e-01 -3.25168490e-01 -1.77140728e-01 6.71190560e-01 7.97720015e-01 7.31077552e-01 -3.38415772e-01 -3.72748047e-01 7.34801471e-01 2.66652226e-01 1.32557184e-01 -5.69551766e-01 -1.12328136e+00 -1.68264151e-01 2.60162860e-01 -6.98123574e-01 -1.37132466e-01 -9.09725904e-01 -1.29049182e+00 -3.80702078e-01 -7.65166640e-01 3.03029686e-01 3.26976359e-01 8.21681976e-01 4.67109501e-01 4.04527336e-01 5.93981266e-01 -2.62967616e-01 -5.77236414e-01 -1.05958092e+00 -8.19191217e-01 8.01031649e-01 3.54075760e-01 -5.72165728e-01 -4.82010752e-01 -1.01033062e-01]
[9.509174346923828, 8.79630184173584]
983a34a3-39e8-4125-b93c-ad7f9012e8d9
adversarially-tuned-scene-generation
1701.00405
null
http://arxiv.org/abs/1701.00405v2
http://arxiv.org/pdf/1701.00405v2.pdf
Adversarially Tuned Scene Generation
Generalization performance of trained computer vision systems that use computer graphics (CG) generated data is not yet effective due to the concept of 'domain-shift' between virtual and real data. Although simulated data augmented with a few real world samples has been shown to mitigate domain shift and improve transferability of trained models, guiding or bootstrapping the virtual data generation with the distributions learnt from target real world domain is desired, especially in the fields where annotating even few real images is laborious (such as semantic labeling, and intrinsic images etc.). In order to address this problem in an unsupervised manner, our work combines recent advances in CG (which aims to generate stochastic scene layouts coupled with large collections of 3D object models) and generative adversarial training (which aims train generative models by measuring discrepancy between generated and real data in terms of their separability in the space of a deep discriminatively-trained classifier). Our method uses iterative estimation of the posterior density of prior distributions for a generative graphical model. This is done within a rejection sampling framework. Initially, we assume uniform distributions as priors on the parameters of a scene described by a generative graphical model. As iterations proceed the prior distributions get updated to distributions that are closer to the (unknown) distributions of target data. We demonstrate the utility of adversarially tuned scene generation on two real-world benchmark datasets (CityScapes and CamVid) for traffic scene semantic labeling with a deep convolutional net (DeepLab). We realized performance improvements by 2.28 and 3.14 points (using the IoU metric) between the DeepLab models trained on simulated sets prepared from the scene generation models before and after tuning to CityScapes and CamVid respectively.
['V. S. R. Veeravasarapu', 'Ramesh Visvanathan', 'Constantin Rothkopf']
2017-01-02
adversarially-tuned-scene-generation-1
http://openaccess.thecvf.com/content_cvpr_2017/html/Veeravasarapu_Adversarially_Tuned_Scene_CVPR_2017_paper.html
http://openaccess.thecvf.com/content_cvpr_2017/papers/Veeravasarapu_Adversarially_Tuned_Scene_CVPR_2017_paper.pdf
cvpr-2017-7
['scene-generation']
['computer-vision']
[ 3.70621681e-01 3.33183080e-01 5.86247444e-01 -4.69659418e-01 -8.42766404e-01 -7.37139761e-01 8.56102347e-01 -4.09875095e-01 -2.78133750e-01 8.31264079e-01 -1.24423824e-01 -2.41289422e-01 2.04110995e-01 -1.03328228e+00 -1.09088922e+00 -7.61655688e-01 4.02476639e-01 1.15918112e+00 4.03977811e-01 -3.68457675e-01 1.19881639e-02 8.09198856e-01 -1.65106797e+00 4.38948631e-01 7.89799035e-01 6.42068982e-01 2.83647418e-01 7.54069626e-01 -1.79067791e-01 4.42169398e-01 -9.22391772e-01 -2.60973305e-01 5.69465578e-01 -3.95097315e-01 -5.88733196e-01 5.25374651e-01 7.74551392e-01 -1.01449110e-01 -2.09866270e-01 1.16829765e+00 3.68650854e-01 1.68196738e-01 1.08464396e+00 -1.52597380e+00 -7.29368627e-01 5.23745008e-02 -4.31780010e-01 -1.57450110e-01 1.14872612e-01 5.97940087e-01 2.22623810e-01 -6.14399552e-01 8.76576841e-01 1.51145530e+00 6.51758730e-01 7.19283283e-01 -1.55242109e+00 -6.09686732e-01 -2.32944340e-01 -1.82087705e-01 -1.35656738e+00 -2.05946282e-01 8.73666525e-01 -7.53491521e-01 5.85147917e-01 -3.55510861e-02 3.48184288e-01 1.65548098e+00 8.07103813e-02 4.12767857e-01 1.29813099e+00 -3.81970286e-01 4.89728987e-01 7.38081157e-01 -2.17836991e-01 3.75462890e-01 1.81713670e-01 4.19155478e-01 -1.01869730e-02 -5.30814081e-02 8.15241575e-01 -3.67670298e-01 -1.73583895e-01 -7.59712636e-01 -7.62143075e-01 1.03655899e+00 5.50357223e-01 -6.94575831e-02 -2.64098287e-01 1.67508200e-01 3.38401198e-01 9.88242254e-02 4.77899998e-01 3.88688445e-01 -2.36113101e-01 2.79754668e-01 -7.84253716e-01 4.38647866e-01 7.32747853e-01 1.05558467e+00 8.91014516e-01 5.37375629e-01 -2.09781006e-01 7.32683241e-01 3.53715301e-01 8.01343739e-01 3.94226700e-01 -7.97299623e-01 3.57291400e-01 4.67039615e-01 2.20667154e-01 -8.30500364e-01 -3.38941179e-02 -5.64241350e-01 -7.98598468e-01 7.78589189e-01 8.89845371e-01 -1.98399156e-01 -1.42112029e+00 1.66221809e+00 5.10591805e-01 2.94909149e-01 1.97281271e-01 7.93684185e-01 6.23213828e-01 6.17193222e-01 1.94485188e-01 4.57316995e-01 9.30585206e-01 -6.08087599e-01 -1.25644758e-01 -2.67535299e-01 4.55404788e-01 -7.07072556e-01 1.38172698e+00 2.39986569e-01 -6.36505663e-01 -9.49815512e-01 -8.86062801e-01 3.02724004e-01 -6.75004780e-01 -1.11082502e-01 2.74303555e-01 1.01398039e+00 -1.09812617e+00 4.52447504e-01 -6.29331589e-01 -3.68352115e-01 8.13000560e-01 -8.39571953e-02 -3.42247009e-01 -1.68956026e-01 -1.02530003e+00 7.79461920e-01 6.62947655e-01 -1.79061726e-01 -1.43495667e+00 -7.76358962e-01 -8.61624539e-01 -2.97953039e-01 1.20845824e-01 -7.95126379e-01 7.89763331e-01 -1.24642694e+00 -1.49061811e+00 1.16312742e+00 4.37509120e-01 -6.42045319e-01 1.04070342e+00 1.64609328e-02 -3.54422629e-01 -1.29570916e-01 1.73805013e-01 1.03152514e+00 1.08725452e+00 -1.76431525e+00 -3.28376889e-01 -2.50851989e-01 -7.93463960e-02 5.64704202e-02 2.33452410e-01 -4.70719516e-01 -2.09972695e-01 -5.09377778e-01 -8.99532586e-02 -1.15905309e+00 -3.80876243e-01 -1.10484004e-01 -5.82337856e-01 2.05498442e-01 9.76727545e-01 -5.15213370e-01 1.62168935e-01 -2.06459403e+00 -2.72343636e-01 3.81838918e-01 -7.57036954e-02 4.10255432e-01 -1.60911649e-01 1.89066440e-01 -3.15005511e-01 -2.58058272e-02 -3.69677842e-01 -3.14135134e-01 4.45103869e-02 3.68716002e-01 -6.64653480e-01 4.95125175e-01 2.19443619e-01 7.45277345e-01 -8.15901279e-01 -1.59811229e-01 6.24502063e-01 6.46667182e-01 -5.87012768e-01 3.08850199e-01 -4.86591429e-01 8.08585227e-01 -1.63310632e-01 -2.32016556e-02 1.04701376e+00 9.45662111e-02 -1.11337982e-01 -6.83225393e-02 4.27517056e-01 -1.81453511e-01 -1.27137637e+00 1.57397079e+00 -5.08397698e-01 5.71219981e-01 -3.43807697e-01 -8.71790946e-01 1.21745646e+00 -5.16946875e-02 5.72208390e-02 -4.91572320e-01 1.07434593e-01 -8.19078609e-02 -1.86790466e-01 -2.60807931e-01 4.45997447e-01 -2.01582938e-01 -8.69745240e-02 1.51900068e-01 3.05740237e-01 -7.73027062e-01 -1.44944921e-01 5.06260470e-02 7.95814574e-01 6.04622960e-01 -1.21170178e-01 -2.75538623e-01 2.66005605e-01 3.54407340e-01 1.55556291e-01 9.12193120e-01 -1.16740555e-01 1.09637940e+00 4.64083880e-01 -2.98863441e-01 -1.46447504e+00 -1.50635743e+00 -1.65057525e-01 6.08167529e-01 -3.27245295e-02 2.16825709e-01 -1.04364097e+00 -9.24999475e-01 -1.30308140e-02 1.44670963e+00 -8.25723410e-01 -4.19985265e-01 -3.39034289e-01 -6.57260656e-01 6.35640025e-01 3.69488806e-01 6.13775432e-01 -1.10395479e+00 -6.02365911e-01 8.75980631e-02 2.10875109e-01 -1.22585809e+00 9.44747590e-03 3.25953066e-02 -5.97061157e-01 -1.03184700e+00 -5.32088459e-01 -3.68913114e-01 6.77607059e-01 -1.09703578e-01 1.33451498e+00 -4.24167246e-01 -3.48781228e-01 3.93368781e-01 -2.28359163e-01 -5.26120901e-01 -1.03164124e+00 -2.80457675e-01 -1.91291496e-01 1.29810691e-01 1.16311789e-01 -6.87674701e-01 -3.95704806e-01 3.33749771e-01 -1.20262396e+00 2.13850215e-01 2.84828752e-01 8.50055575e-01 4.38886970e-01 5.20775244e-02 3.73186141e-01 -1.32187665e+00 3.71230811e-01 -5.62491715e-01 -8.01659524e-01 1.70330033e-02 -2.88178802e-01 6.65455014e-02 7.07030654e-01 -4.86521661e-01 -1.38783288e+00 2.05834776e-01 -1.95416093e-01 -9.22885597e-01 -9.18136060e-01 -1.42774224e-01 -3.78484428e-01 1.45194173e-01 1.32920659e+00 1.04410656e-01 -1.28133029e-01 -1.52567178e-01 6.66984260e-01 4.45727646e-01 7.59077072e-01 -7.60753930e-01 1.05832148e+00 6.23223901e-01 5.94822019e-02 -7.01287627e-01 -6.57860875e-01 -1.19635768e-01 -7.04265535e-01 -3.00229818e-01 8.68171096e-01 -7.34498799e-01 -1.45239308e-01 5.15985489e-01 -1.14505041e+00 -5.64918101e-01 -7.52050281e-01 2.49860525e-01 -7.70150125e-01 1.18864797e-01 -1.03354856e-01 -7.93213129e-01 1.03352480e-01 -1.12206924e+00 1.33128607e+00 1.09157309e-01 -6.43681660e-02 -1.13998532e+00 1.93211921e-02 2.75027782e-01 3.57905895e-01 8.43275309e-01 8.73297751e-01 -6.83973312e-01 -3.79791737e-01 -3.38552922e-01 -2.41586223e-01 6.77689672e-01 -3.27062048e-03 4.12452742e-02 -1.34522295e+00 -1.31828904e-01 3.61483879e-02 -3.37210208e-01 5.87642550e-01 3.65228206e-01 1.10473597e+00 -2.20743543e-03 -2.42539793e-01 5.49236834e-01 1.57918370e+00 2.35010877e-01 1.08452356e+00 1.82167679e-01 8.02124083e-01 7.24724293e-01 4.29504931e-01 2.20095560e-01 -1.25730783e-01 6.73078656e-01 6.02145493e-01 -2.28431433e-01 -5.31973898e-01 -5.41292250e-01 1.72577292e-01 -1.47755459e-01 3.21483582e-01 -5.06972194e-01 -1.04000890e+00 6.30972087e-01 -1.50615406e+00 -8.45596254e-01 -3.40653896e-01 2.38787150e+00 5.30887306e-01 5.34510553e-01 7.13708326e-02 -1.21776052e-01 8.04511428e-01 -2.18309790e-01 -5.03826916e-01 -3.16569924e-01 -1.23370931e-01 3.53063613e-01 4.83985126e-01 4.74572539e-01 -1.00873280e+00 1.11190057e+00 5.28530788e+00 9.62809741e-01 -1.11457133e+00 9.82742831e-02 8.66343796e-01 3.51976067e-01 -2.40975738e-01 -4.13462035e-02 -7.34369755e-01 5.27516007e-01 9.37272668e-01 1.01128109e-01 2.07541257e-01 1.22534025e+00 1.30268201e-01 2.15960923e-03 -1.09279811e+00 9.60028529e-01 9.72537883e-03 -1.39342070e+00 2.87379980e-01 1.28214210e-01 1.01728630e+00 9.61207896e-02 2.73888499e-01 3.76863331e-01 9.30255413e-01 -1.18648493e+00 8.52651656e-01 5.13592660e-01 8.39593112e-01 -6.56751037e-01 6.68047130e-01 5.20249307e-01 -6.33717000e-01 3.82984996e-01 -4.70381290e-01 3.02448213e-01 1.00426273e-02 5.56248724e-01 -1.47745419e+00 4.55315590e-01 5.62999845e-01 2.23995760e-01 -6.31227314e-01 7.69842505e-01 -2.01177686e-01 6.40328526e-01 -2.95294762e-01 2.62512505e-01 2.93562829e-01 -3.76565039e-01 6.60123169e-01 9.77178097e-01 1.12308308e-01 -5.02810240e-01 1.75488353e-01 1.40628338e+00 1.93967924e-01 -2.87488222e-01 -9.99424219e-01 4.22953904e-01 1.62669912e-01 1.04204512e+00 -8.68336916e-01 -3.28608930e-01 1.64599672e-01 1.03689885e+00 1.53126508e-01 5.87157965e-01 -1.17034030e+00 -6.81148469e-02 4.77149516e-01 3.45829278e-01 2.18162537e-01 -9.55944732e-02 -3.67317080e-01 -8.27028811e-01 -3.60734403e-01 -7.02657163e-01 1.10365503e-01 -1.15199733e+00 -1.40635800e+00 9.03826416e-01 3.06679696e-01 -1.26839995e+00 -5.82261443e-01 -6.45278096e-01 -5.56954861e-01 1.22536778e+00 -9.67943549e-01 -1.33079040e+00 -5.92140794e-01 6.74987853e-01 5.95582247e-01 -2.51155883e-01 7.75256991e-01 1.07854675e-03 -8.29371735e-02 3.92735243e-01 1.47070765e-01 1.12551481e-01 4.80291992e-01 -1.33826685e+00 7.05590785e-01 7.89548695e-01 3.32749575e-01 1.25070093e-02 1.04074407e+00 -5.38430989e-01 -7.10715890e-01 -1.65395749e+00 8.76202956e-02 -8.13760519e-01 3.11350852e-01 -5.75420856e-01 -1.02818406e+00 6.31299317e-01 7.82868639e-02 1.63456038e-01 1.95753604e-01 -4.57084358e-01 -4.22353595e-01 1.02084853e-01 -1.55791748e+00 7.13338017e-01 1.01601982e+00 -3.86926770e-01 -4.30625826e-01 5.22805214e-01 5.33728778e-01 -5.12875259e-01 -4.32122201e-01 3.15101624e-01 1.00336105e-01 -1.15179837e+00 1.03470945e+00 -8.32701921e-01 2.48150259e-01 -4.27428544e-01 -3.66096020e-01 -1.62183034e+00 -4.81216721e-02 -3.02135050e-01 5.27509689e-01 1.27158070e+00 2.75319606e-01 -6.94457471e-01 1.01951647e+00 4.98984247e-01 -2.44399637e-01 -1.72488153e-01 -7.49506831e-01 -6.96486294e-01 2.94303209e-01 -7.17201769e-01 5.89745164e-01 8.28300238e-01 -1.10618472e+00 3.80991697e-01 9.26416460e-03 2.88507789e-01 7.25331843e-01 -1.25899374e-01 1.37276912e+00 -1.04975843e+00 -4.41822976e-01 -2.11322501e-01 -7.59512901e-01 -5.96705556e-01 3.24792832e-01 -8.61462831e-01 2.03082841e-02 -1.32642686e+00 -1.85796842e-01 -7.11904883e-01 1.88334808e-01 1.39594406e-01 2.27162674e-01 3.21410745e-01 6.35917783e-02 -1.09608904e-01 -1.01958044e-01 6.01738155e-01 1.27297425e+00 -1.82611704e-01 -8.57245028e-02 2.89517604e-02 -3.89863282e-01 7.21323729e-01 8.55613172e-01 -5.18006027e-01 -7.65210748e-01 -1.77549720e-01 -1.88672736e-01 -1.27009898e-01 9.06247616e-01 -1.26795876e+00 -3.12119067e-01 -1.13872290e-01 6.00987494e-01 -4.68252957e-01 4.34611499e-01 -8.21769893e-01 6.86528444e-01 1.68992639e-01 -1.09586783e-01 -4.73072261e-01 4.93708551e-01 6.69677794e-01 9.54185501e-02 -1.20413177e-01 1.14447725e+00 -1.74882203e-01 -7.76121318e-01 8.10899660e-02 -1.09436959e-01 2.52525747e-01 1.16001499e+00 -4.51913863e-01 -2.62589544e-01 -3.84213120e-01 -9.26070452e-01 -3.00278753e-01 6.54705286e-01 3.77201736e-01 5.43346524e-01 -1.30642998e+00 -7.92435646e-01 4.04897571e-01 1.20793946e-01 3.80045980e-01 3.44364464e-01 2.47943565e-01 -6.87710047e-01 4.58287112e-02 -4.19489592e-01 -9.28041458e-01 -9.31448221e-01 6.59022450e-01 6.93055868e-01 -1.15238428e-01 -5.17573059e-01 7.79706955e-01 6.84207678e-01 -7.91258693e-01 -1.27181739e-01 -3.21195573e-01 5.38859256e-02 -4.38795209e-01 -6.87893759e-03 1.65186480e-01 2.25091979e-01 -7.04547405e-01 -4.19706069e-02 1.33283913e-01 2.03628972e-01 -1.73564270e-01 1.05502665e+00 2.71119535e-01 3.20148587e-01 2.65226394e-01 1.32626498e+00 -1.42500073e-01 -1.69687235e+00 5.91791980e-02 -2.86473930e-01 -5.62178314e-01 -1.08112633e-01 -8.11920881e-01 -1.06368506e+00 8.56720746e-01 1.03999543e+00 1.65129870e-01 8.33034992e-01 5.09620048e-02 2.42809772e-01 5.08721545e-02 3.85617584e-01 -8.54722023e-01 9.81988385e-02 3.26825678e-01 1.04847538e+00 -1.16957736e+00 -4.26789731e-01 -5.58223069e-01 -7.56186664e-01 8.77332866e-01 7.58089781e-01 -5.29019833e-01 7.37503648e-01 9.30591375e-02 2.07263187e-01 -1.68128610e-01 -3.38392645e-01 -2.10329816e-01 2.36606076e-01 1.18840945e+00 -1.16268069e-01 2.11368337e-01 5.34911692e-01 1.39468074e-01 -4.53509808e-01 -1.98341057e-01 5.64607441e-01 5.34101546e-01 -1.40984088e-01 -8.87606204e-01 -6.81667209e-01 2.90950179e-01 -6.09658994e-02 5.38673513e-02 -2.18935385e-01 1.03369212e+00 3.14487815e-01 8.46090019e-01 2.34580755e-01 -2.89791733e-01 3.71523410e-01 1.83263123e-02 5.10174811e-01 -6.89641714e-01 -2.31652007e-01 -1.22045144e-01 7.83868432e-02 -3.67445797e-01 -1.52368680e-01 -5.95455229e-01 -1.00588596e+00 -9.54346135e-02 -1.68372374e-02 -5.41660711e-02 8.00066173e-01 8.03270340e-01 2.91096628e-01 6.29272044e-01 5.49799085e-01 -1.00776291e+00 -5.92590749e-01 -9.59209740e-01 -5.95308244e-01 9.74363804e-01 -1.91865444e-01 -8.43886256e-01 -3.15177888e-01 3.74338686e-01]
[9.848800659179688, 1.1870567798614502]
49caba48-ae92-446f-be3f-011d18c0bc29
authorship-verification-average-similarity
null
null
https://aclanthology.org/R15-1012
https://aclanthology.org/R15-1012.pdf
Authorship Verification, Average Similarity Analysis
null
['Rafael Mu{\\~n}oz Guillena', "Mar{\\'\\i}a Pelaez Brioso", 'Daniel Castro Castro', 'Yaritza Adame Arcia']
2015-09-01
authorship-verification-average-similarity-1
https://aclanthology.org/R15-1012
https://aclanthology.org/R15-1012.pdf
ranlp-2015-9
['authorship-verification']
['natural-language-processing']
[-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01 -8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01 -2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00 -3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01 -9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01 7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01 7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00 -1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01 3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01 9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01 5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00 -5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01 1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00 7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01 -1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02 -1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01 1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00 1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01 3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01 8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01 9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01 -9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01 -1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01 -2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01 -8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00 1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01 8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00 -6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01 -6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01 3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01 4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02 4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01 1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01 9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01 -1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01 -7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00 -1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02 1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01 -1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01 -1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01 -4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01 -1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01 6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01 -1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00 -7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01 -1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01 3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01 -1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01 1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01 3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00 -2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01 2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01 -1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01 4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01 2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00 1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01 -3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00 7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01 3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01 -2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01 7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01 -2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01 6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00 7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00 -3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01 -6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00 3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01 1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01 -5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01 -5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01 -2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01 -1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01 5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01 -2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01 -4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01 5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01 -6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00 1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01 8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02 -6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01 -1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01 1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01 8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03 3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01 8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01 -1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01 6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01 -9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01 -1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00 -5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02 -5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01 1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01 -4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01 1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01 6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01 5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01 1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01 1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01 1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01 1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00 -8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01 -1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01 6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01 4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00 -1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00 -7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00 2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01 -4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02 1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01 3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01 9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01 7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01 6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01 1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01 -1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00 3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01 9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01 -4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01 4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00 2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01 5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01 2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01 -2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01 -6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01 -1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01 -5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01 -1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00 -6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01 -1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01 1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02 1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01 5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01 -4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01 -1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01 6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01 7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01 1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01 1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01 8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01 8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00 -4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01 -5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02 5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01 -2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00 -3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01 6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01 5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00 3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01 -8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00 -8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01 3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01 -3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05 -1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01 8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02 6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01 5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00 1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01 2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02 -1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01 -4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02 9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01 -8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02 7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01 -3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01 -1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01 -1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01 -2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01 1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03 6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01 -4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01 9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01 8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01 1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01 -1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01 8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01 5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01 9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01 -4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01 1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01 -9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00 5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01 -7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00 -6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01 -5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01 -5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01 4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01 -9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00 -8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00 -5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01 -2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00 -3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02 2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01]
[-7.279503345489502, 3.6174473762512207]
50d18110-1003-47ea-aaf9-a928eca6f357
nonnegative-low-rank-tensor-completion-via
2305.07976
null
https://arxiv.org/abs/2305.07976v1
https://arxiv.org/pdf/2305.07976v1.pdf
Nonnegative Low-Rank Tensor Completion via Dual Formulation with Applications to Image and Video Completion
Recent approaches to the tensor completion problem have often overlooked the nonnegative structure of the data. We consider the problem of learning a nonnegative low-rank tensor, and using duality theory, we propose a novel factorization of such tensors. The factorization decouples the nonnegative constraints from the low-rank constraints. The resulting problem is an optimization problem on manifolds, and we propose a variant of Riemannian conjugate gradients to solve it. We test the proposed algorithm across various tasks such as colour image inpainting, video completion, and hyperspectral image completion. Experimental results show that the proposed method outperforms many state-of-the-art tensor completion algorithms.
['Pawan Kumar', 'Jayadev Naram', 'Tanmay Kumar Sinha']
2023-05-13
null
null
null
null
['image-inpainting']
['computer-vision']
[ 5.20462021e-02 -4.07652915e-01 7.91085809e-02 -1.66265324e-01 -6.80773258e-01 -6.55442894e-01 3.83603394e-01 -6.07890666e-01 -4.01936114e-01 4.41425562e-01 3.89019758e-01 -3.83135498e-01 -3.48854780e-01 -4.02717292e-02 -5.39820433e-01 -8.76313448e-01 -1.81403026e-01 -3.48950177e-03 -5.15127718e-01 -3.38717312e-01 2.77011991e-01 2.83286870e-01 -9.97048378e-01 2.64346004e-01 8.63159299e-01 6.91330492e-01 1.21093050e-01 8.29769015e-01 1.11299634e-01 9.54277754e-01 1.15231320e-01 -5.65463722e-01 5.36626577e-01 -7.87996799e-02 -9.29965258e-01 7.34430194e-01 6.15436792e-01 -3.75288874e-01 -5.91000199e-01 1.22738445e+00 8.01902190e-02 2.70818144e-01 4.20156211e-01 -1.39109373e+00 -1.18985832e+00 -2.23589335e-02 -9.09339905e-01 9.01727192e-03 3.79777104e-01 -3.49781275e-01 1.34987593e+00 -1.54634953e+00 5.85981727e-01 1.15183282e+00 4.11093056e-01 4.50330377e-01 -1.34776855e+00 -9.11955163e-02 2.13826954e-01 1.36925980e-01 -1.18263721e+00 -3.76374036e-01 9.17351365e-01 -7.83073187e-01 5.09814382e-01 4.44554776e-01 5.34973800e-01 6.31414294e-01 -4.87748682e-02 7.86302269e-01 1.24901521e+00 -1.70946300e-01 3.09043508e-02 -4.41764593e-01 -5.10692671e-02 9.03571248e-01 3.12558532e-01 -2.92110920e-01 -5.14777422e-01 -3.18990707e-01 7.28545547e-01 3.46712589e-01 -3.57594460e-01 -5.55213451e-01 -1.79979658e+00 8.99474263e-01 3.26472223e-01 7.69780278e-02 -4.45088029e-01 2.25658193e-01 1.83507979e-01 3.16504747e-01 6.55723691e-01 2.54648685e-01 -1.72744185e-01 9.00090039e-02 -5.27065933e-01 7.86796212e-03 7.95457542e-01 9.70916092e-01 8.64645243e-01 2.37292558e-01 1.28355548e-01 8.44474494e-01 4.22160625e-01 5.08411407e-01 2.48801745e-02 -1.04113221e+00 8.91656101e-01 2.90060163e-01 3.31352204e-01 -1.20004869e+00 -2.40130410e-01 -4.33584116e-03 -1.08246744e+00 -1.63091302e-01 1.01722181e-01 -1.99231938e-01 -7.04122782e-01 1.40620255e+00 4.02608603e-01 4.29551452e-01 1.03975423e-01 1.41077018e+00 4.66483444e-01 6.10151291e-01 -4.15770829e-01 -2.33018339e-01 1.01230085e+00 -9.77365732e-01 -9.41243529e-01 1.52632535e-01 4.90819752e-01 -1.21939301e+00 8.26886475e-01 4.02162611e-01 -9.54232335e-01 -9.76101086e-02 -9.13535237e-01 -4.62117523e-01 1.71728712e-02 4.67381477e-01 1.31249344e+00 5.20084381e-01 -1.16968203e+00 6.68784976e-01 -7.93068647e-01 -1.48364037e-01 -1.81818739e-01 3.03840935e-01 -7.54996955e-01 -5.76894104e-01 -7.48625517e-01 3.46714079e-01 -9.67836529e-02 8.27863753e-01 -8.99547815e-01 -5.74351549e-01 -7.86055684e-01 -4.72564280e-01 2.52677232e-01 -9.19603646e-01 8.17286372e-01 -8.12517166e-01 -1.48176670e+00 8.35525155e-01 -1.74109042e-01 3.68574768e-01 3.83545458e-01 -3.83433312e-01 -3.01550388e-01 3.35490763e-01 1.69618696e-01 2.18634754e-01 1.49337077e+00 -1.20231116e+00 -2.62115836e-01 -4.14273053e-01 4.32988495e-01 3.72796834e-01 -5.51864803e-01 -1.76866800e-02 -5.57426214e-01 -8.61277759e-01 6.44064665e-01 -1.29104221e+00 -4.45401102e-01 -3.03326659e-02 -4.42803353e-01 1.20017447e-01 7.32103944e-01 -9.67316151e-01 8.30336690e-01 -2.25113726e+00 9.28836465e-01 4.69053015e-02 3.87132019e-01 -3.24164331e-01 -4.21484292e-01 5.78848600e-01 -5.68765223e-01 -5.70018999e-02 -3.96609873e-01 -7.61232376e-01 -8.97456706e-02 5.28039336e-01 -4.65097159e-01 9.05376196e-01 3.13761979e-01 5.20344555e-01 -1.22033525e+00 -1.45054743e-01 7.72168264e-02 6.00381017e-01 -6.84956849e-01 3.00090879e-01 1.54255748e-01 6.23342633e-01 -4.09789503e-01 6.69713378e-01 1.04523993e+00 -1.91134095e-01 2.24101931e-01 -7.04752505e-01 -2.62089133e-01 -8.88174027e-02 -1.45403576e+00 2.20884752e+00 -3.63920599e-01 3.79201531e-01 6.55495286e-01 -1.07920337e+00 4.47242677e-01 2.27214724e-01 9.22983825e-01 -7.68559650e-02 -1.10788539e-01 1.60165071e-01 -1.88341945e-01 -6.66157365e-01 9.31038499e-01 -1.97193637e-01 4.09278154e-01 6.22851610e-01 -9.85740311e-03 -1.64473206e-01 5.01434922e-01 7.34206557e-01 1.05225778e+00 3.25808287e-01 -6.01226568e-01 -3.18530351e-01 5.73564649e-01 -2.75168836e-01 5.26840687e-01 3.75702381e-02 2.41093665e-01 6.92381859e-01 6.10237896e-01 -3.86843443e-01 -1.04119492e+00 -9.06338811e-01 1.93642173e-02 1.14186668e+00 -1.75634533e-01 -6.43736839e-01 -5.39474607e-01 -5.12085497e-01 -5.74689656e-02 -3.77016068e-02 -6.55392349e-01 2.29607821e-01 -3.61533225e-01 -1.12877917e+00 -2.47212853e-02 1.06552705e-01 3.08556855e-01 -1.20803036e-01 4.22892898e-01 -1.31789058e-01 -5.05588114e-01 -1.51902032e+00 -9.75782573e-01 -3.17858160e-01 -1.28736043e+00 -1.10924339e+00 -8.36890399e-01 -7.41356730e-01 1.30985141e+00 1.13142443e+00 8.52293789e-01 1.61895931e-01 -2.35816672e-01 1.03509521e+00 -3.70830297e-01 4.04880285e-01 1.11893471e-02 -3.29092920e-01 2.64817566e-01 9.16460752e-01 -2.78651476e-01 -5.24788797e-01 -5.84676504e-01 2.60259032e-01 -1.42754078e+00 1.77950114e-02 4.94254708e-01 1.06288004e+00 6.69324577e-01 -2.07016021e-01 1.32274896e-01 -7.29883552e-01 8.72559607e-01 -3.33577335e-01 -5.41273475e-01 2.28782177e-01 -3.44105005e-01 3.71321440e-01 4.71669614e-01 -3.78557384e-01 -8.64268601e-01 4.45557356e-01 4.68090415e-01 -8.08384061e-01 7.80280948e-01 7.58066654e-01 -8.19186643e-02 -5.66131592e-01 2.54849732e-01 7.45600015e-02 -3.24340910e-03 -7.78022289e-01 8.30881000e-01 2.92267859e-01 4.38049972e-01 -9.07429814e-01 1.19303751e+00 1.01401877e+00 4.22401428e-01 -1.08014536e+00 -1.01900339e+00 -8.53140354e-01 -8.51982117e-01 -2.94973373e-01 7.87850559e-01 -1.22743034e+00 -6.53747678e-01 2.06180781e-01 -1.21663785e+00 1.50625661e-01 -4.06731591e-02 8.37765992e-01 -5.99796176e-01 1.02829409e+00 -7.66690075e-01 -5.37463784e-01 -5.62696233e-02 -1.01720929e+00 1.12637341e+00 -4.22306359e-01 5.32643676e-01 -1.18050742e+00 3.68329644e-01 6.22444034e-01 -1.88640952e-02 1.89191133e-01 6.49510443e-01 4.78249103e-01 -7.98644781e-01 -2.14059204e-01 -4.44119394e-01 7.47486830e-01 2.41152242e-01 1.16381407e-01 -7.69134104e-01 -6.06314778e-01 1.54908687e-01 -1.80064887e-01 9.21865821e-01 -1.01603180e-01 9.53394711e-01 -2.83324808e-01 2.90303767e-01 1.02935064e+00 1.42768657e+00 -7.23032892e-01 6.48097754e-01 -5.80648370e-02 1.48647499e+00 5.28728843e-01 5.74496746e-01 6.75605118e-01 5.33581436e-01 3.38127762e-01 6.61224425e-01 -3.63974273e-01 1.37258440e-01 -6.62764460e-02 5.70578694e-01 1.60186529e+00 -6.36983514e-01 5.53704143e-01 -6.81862712e-01 5.19315243e-01 -2.16455793e+00 -7.37289011e-01 -5.38548291e-01 2.29515314e+00 5.77475429e-01 -6.39667928e-01 3.72524038e-02 3.01956069e-02 7.66682446e-01 3.29636991e-01 -2.94931978e-01 -3.49975675e-01 -1.42333090e-01 -3.42731066e-02 7.09835947e-01 6.69851124e-01 -1.22517502e+00 8.09310436e-01 7.04136753e+00 1.75898790e-01 -1.02679682e+00 1.88499212e-01 3.31864133e-02 6.79569766e-02 -6.12637162e-01 3.44304800e-01 -1.21729188e-01 -1.23798952e-01 3.56392860e-01 -3.85983847e-03 1.18722510e+00 6.45192444e-01 2.59061515e-01 1.79413617e-01 -1.18742430e+00 1.36947191e+00 4.08620298e-01 -1.07615852e+00 3.29988062e-01 2.42574066e-01 1.13722777e+00 5.73667027e-02 3.37005585e-01 -9.20169652e-02 2.09041163e-01 -6.73079371e-01 5.36692023e-01 6.54090047e-01 6.19703770e-01 -5.46551466e-01 1.40986726e-01 -2.18852848e-01 -1.23911548e+00 4.27715965e-02 -6.96295738e-01 -1.99650019e-01 -2.07867268e-02 9.31413472e-01 -3.42567533e-01 8.99531364e-01 4.32728916e-01 1.30564034e+00 -6.28188610e-01 1.00857151e+00 -3.51758212e-01 3.36161673e-01 -1.04879394e-01 4.34190214e-01 3.65788549e-01 -1.23201478e+00 6.63898468e-01 9.98262882e-01 2.24211305e-01 2.27435172e-01 2.47411922e-01 6.03858113e-01 -3.37640852e-01 2.90032357e-01 -6.15417838e-01 -4.60742891e-01 -4.73699421e-01 1.78538764e+00 -4.64931011e-01 -1.40564460e-02 -8.00830364e-01 1.46018231e+00 3.47905099e-01 8.06737304e-01 -6.35465145e-01 -1.56807423e-01 9.68452990e-01 -4.54262316e-01 2.51215130e-01 -9.54582214e-01 -8.83167684e-02 -1.95871055e+00 5.30718565e-01 -6.79793835e-01 4.25591528e-01 -8.14687729e-01 -1.41651106e+00 2.12903693e-01 -2.90964425e-01 -1.53736377e+00 2.02199906e-01 -9.51281071e-01 -4.32038009e-01 7.81146407e-01 -1.61015022e+00 -1.35359335e+00 -2.04183444e-01 1.14528310e+00 -1.45463049e-02 1.23577878e-01 5.98696530e-01 7.64569342e-01 -7.79703975e-01 -1.54709890e-02 5.78400373e-01 1.68571919e-01 5.91833830e-01 -1.49628854e+00 1.31093353e-01 1.12394619e+00 1.20333046e-01 7.82579958e-01 3.33538830e-01 -4.51663673e-01 -2.61147952e+00 -9.99427319e-01 5.20800114e-01 -3.50169122e-01 1.25354600e+00 -3.23776662e-01 -4.85400379e-01 7.07146168e-01 9.28113908e-02 3.95585835e-01 8.58976722e-01 1.94259644e-01 -7.74928451e-01 -1.60995618e-01 -7.28905439e-01 5.55382431e-01 1.03412843e+00 -9.21700418e-01 -1.68518335e-01 9.79335666e-01 6.32979274e-01 -1.34228274e-01 -9.95030820e-01 -4.01482582e-02 3.61251384e-01 -5.23959696e-01 9.30515826e-01 -1.12063372e+00 3.53643298e-01 -6.02830887e-01 -5.08536458e-01 -1.37149358e+00 -4.59337771e-01 -1.16230726e+00 -8.91572833e-02 9.00708079e-01 2.52024055e-01 -3.86042595e-01 6.52686059e-01 8.81325364e-01 -3.51495802e-01 -3.44968319e-01 -8.56426835e-01 -7.66870439e-01 -2.69169986e-01 -2.83771098e-01 6.61330968e-02 1.39141142e+00 9.54653099e-02 4.92899477e-01 -8.94869089e-01 3.89238566e-01 1.06149316e+00 1.54681459e-01 6.93984032e-01 -8.92343879e-01 -4.07227278e-01 1.55892909e-01 -2.86087334e-01 -1.02986240e+00 1.56677961e-01 -1.17676020e+00 -1.88786551e-01 -1.43422055e+00 2.58099556e-01 -9.54958647e-02 -9.57435295e-02 2.25428343e-01 -1.49305105e-01 2.12052003e-01 4.11989570e-01 3.62186730e-01 -7.05619812e-01 7.90569425e-01 1.69410563e+00 -3.84589434e-01 1.47095909e-02 -3.51139724e-01 -5.65335870e-01 5.18709898e-01 3.99901420e-01 -3.04187179e-01 -3.06630611e-01 -8.84421110e-01 8.73337209e-01 7.04080090e-02 2.14281663e-01 -4.61078644e-01 8.74043852e-02 -3.55604649e-01 7.20977411e-02 -2.25291073e-01 5.22475600e-01 -8.56107295e-01 3.36690471e-02 8.88947546e-02 1.65647075e-01 3.77725959e-01 -2.08470464e-01 7.03637779e-01 -3.40925694e-01 -1.51190788e-01 4.34337914e-01 1.82437263e-02 -5.14184475e-01 6.09840870e-01 -1.14820354e-01 9.15011764e-02 4.91652429e-01 3.11657727e-01 -1.32190570e-01 -3.02765667e-01 -1.00563443e+00 5.59530072e-02 3.13106596e-01 5.02942741e-01 9.33607936e-01 -1.67073309e+00 -8.53936672e-01 6.56999499e-02 1.06328353e-01 -2.60730505e-01 1.57142818e-01 1.31338859e+00 -5.81840694e-01 9.24348980e-02 -1.73442215e-02 -5.99141598e-01 -1.02236617e+00 7.45525718e-01 1.49508297e-01 -7.80863762e-02 -3.95917028e-01 5.26836634e-01 2.26001620e-01 -4.78740871e-01 -6.06438192e-03 -2.93490708e-01 -2.93246731e-02 1.45977303e-01 4.50853407e-01 6.84746206e-01 1.28633782e-01 -9.03308868e-01 -1.27425179e-01 8.28859031e-01 6.41449541e-02 -1.87863737e-01 1.50658393e+00 -2.41004810e-01 -8.98501456e-01 3.34623218e-01 1.45211065e+00 3.18093672e-02 -1.08238864e+00 -1.48757979e-01 -5.01534119e-02 -8.81914973e-01 3.34800601e-01 7.05788434e-02 -1.36170185e+00 8.68142307e-01 2.34063983e-01 1.94260225e-01 1.12364364e+00 -5.53298295e-01 6.14948928e-01 7.49780655e-01 3.43178123e-01 -1.09003854e+00 2.99611241e-01 7.69800007e-01 1.32019138e+00 -1.24236143e+00 3.58638108e-01 -6.89667702e-01 -5.45127153e-01 1.13164401e+00 1.19565055e-01 -3.24141592e-01 8.25430691e-01 -4.51040804e-01 -8.76377001e-02 -1.94569409e-01 -6.02480650e-01 -3.45778227e-01 5.02847791e-01 3.88737172e-01 5.70893288e-01 3.13233227e-01 -4.74475265e-01 -1.44875973e-01 1.85473919e-01 -1.56716347e-01 8.95017743e-01 8.68840337e-01 1.65636629e-01 -1.32866704e+00 -7.83423543e-01 2.44814470e-01 -4.45502758e-01 -2.11297274e-01 -6.10806942e-01 1.23380184e-01 -3.54239941e-01 1.01711130e+00 -4.44161326e-01 -5.94200253e-01 2.53084302e-01 -2.19985366e-01 6.91869497e-01 -5.14792323e-01 -1.43358916e-01 2.70453572e-01 7.37513751e-02 -6.53511703e-01 -8.25293541e-01 -8.26985717e-01 -8.86113167e-01 -4.07354891e-01 -2.76912868e-01 2.58240074e-01 1.00034654e+00 4.98981774e-01 3.46486419e-01 8.70348588e-02 1.25769949e+00 -1.11447024e+00 -5.69199741e-01 -7.70019412e-01 -9.85444188e-01 8.31276774e-01 5.19704878e-01 -6.47737920e-01 -5.87328553e-01 2.71832436e-01]
[7.375734329223633, 4.513314723968506]
ef794dde-a9c1-4553-b880-bd3f93da557d
hdr-chipqa-no-reference-quality-assessment-on
2304.13156
null
https://arxiv.org/abs/2304.13156v1
https://arxiv.org/pdf/2304.13156v1.pdf
HDR-ChipQA: No-Reference Quality Assessment on High Dynamic Range Videos
We present a no-reference video quality model and algorithm that delivers standout performance for High Dynamic Range (HDR) videos, which we call HDR-ChipQA. HDR videos represent wider ranges of luminances, details, and colors than Standard Dynamic Range (SDR) videos. The growing adoption of HDR in massively scaled video networks has driven the need for video quality assessment (VQA) algorithms that better account for distortions on HDR content. In particular, standard VQA models may fail to capture conspicuous distortions at the extreme ends of the dynamic range, because the features that drive them may be dominated by distortions {that pervade the mid-ranges of the signal}. We introduce a new approach whereby a local expansive nonlinearity emphasizes distortions occurring at the higher and lower ends of the {local} luma range, allowing for the definition of additional quality-aware features that are computed along a separate path. These features are not HDR-specific, and also improve VQA on SDR video contents, albeit to a reduced degree. We show that this preprocessing step significantly boosts the power of distortion-sensitive natural video statistics (NVS) features when used to predict the quality of HDR content. In similar manner, we separately compute novel wide-gamut color features using the same nonlinear processing steps. We have found that our model significantly outperforms SDR VQA algorithms on the only publicly available, comprehensive HDR database, while also attaining state-of-the-art performance on SDR content.
['Alan C. Bovik', 'Sriram Sethuraman', 'Hai Wei', 'Yongjun Wu', 'Zaixi Shang', 'Joshua P. Ebenezer']
2023-04-25
null
null
null
null
['video-quality-assessment', 'video-quality-assessment']
['computer-vision', 'time-series']
[ 1.57926098e-01 -7.51388788e-01 -8.16512182e-02 -3.94577920e-01 -7.81613350e-01 -6.23535097e-01 4.27882522e-01 -2.38453910e-01 -1.08356282e-01 4.58130568e-01 4.68862742e-01 -6.60240948e-02 -1.85170561e-01 -7.90496349e-01 -5.24066508e-01 -6.25930905e-01 -4.15477663e-01 -1.89218223e-01 3.96342218e-01 -7.25990474e-01 1.91970170e-01 8.31616521e-01 -1.68526757e+00 5.10902524e-01 6.90164804e-01 1.23493707e+00 -8.44247490e-02 1.13739479e+00 3.22557360e-01 8.83233368e-01 -5.78670621e-01 -3.75391901e-01 6.38794959e-01 -5.03433645e-01 -3.31047684e-01 1.75836354e-01 6.82412446e-01 -8.56378198e-01 -1.07549560e+00 1.00884616e+00 5.10076404e-01 1.15555815e-01 4.43065256e-01 -1.12384069e+00 -8.93872857e-01 1.87587246e-01 -6.74903393e-01 6.23337746e-01 7.00527191e-01 5.44239998e-01 1.19267356e+00 -8.13515007e-01 7.10815609e-01 1.33366263e+00 4.84286368e-01 3.15859169e-01 -1.08470213e+00 -4.58204061e-01 -1.53228268e-01 6.22775733e-01 -1.39431858e+00 -5.25402188e-01 8.52580726e-01 -1.12120189e-01 9.62616384e-01 3.47445011e-01 7.64476180e-01 8.87542903e-01 2.69893974e-01 4.70232338e-01 1.21110260e+00 -5.86197451e-02 1.94123164e-01 -3.55470985e-01 -3.09342057e-01 4.47412640e-01 -5.44487219e-03 4.17995900e-01 -7.18010545e-01 1.65758371e-01 9.99446392e-01 -1.51151031e-01 -6.50175810e-01 -3.75809729e-01 -1.15192223e+00 6.46690130e-01 4.26865160e-01 1.94117069e-01 -2.56762683e-01 1.85247049e-01 3.04635018e-01 8.04078281e-01 3.77282500e-01 1.75382808e-01 -3.15595478e-01 -5.10668993e-01 -1.28102219e+00 2.95693502e-02 4.67084795e-01 9.33948696e-01 7.64562845e-01 2.68783212e-01 -1.52007252e-01 9.43136156e-01 1.75344020e-01 7.16836691e-01 -5.26607633e-02 -1.49452007e+00 3.86729717e-01 1.56909153e-01 -3.28697041e-02 -9.81669307e-01 -1.49239689e-01 -1.90430418e-01 -8.87062073e-01 4.67125028e-01 4.46474612e-01 1.93462908e-01 -8.82718623e-01 1.45593774e+00 -9.16367322e-02 -9.56652835e-02 -4.51569147e-02 1.31105828e+00 7.06455469e-01 9.82887387e-01 -1.19641595e-01 -3.01217407e-01 9.72716212e-01 -4.49050993e-01 -6.07824206e-01 3.08465213e-01 -7.49557316e-02 -8.02469969e-01 1.16427720e+00 6.79660559e-01 -1.47415435e+00 -8.03795874e-01 -1.31893563e+00 -3.49163324e-01 -1.45464316e-01 -3.66809219e-01 2.47920498e-01 7.95255721e-01 -1.62470126e+00 7.69617617e-01 -3.92238289e-01 -1.69044733e-01 2.39497155e-01 1.79801404e-01 -3.04572731e-01 -6.26159549e-01 -1.20586419e+00 6.64785028e-01 -1.44175544e-01 -4.19097990e-02 -1.06691933e+00 -9.57788229e-01 -7.60923028e-01 7.82478154e-02 2.99024671e-01 -3.86701286e-01 5.86685658e-01 -1.24395788e+00 -1.56511712e+00 7.46602356e-01 7.89839327e-02 -1.85299844e-01 6.50326431e-01 -6.39495701e-02 -8.73787522e-01 6.44573331e-01 -4.67216969e-01 6.42195404e-01 1.19108331e+00 -1.39309990e+00 -5.25951087e-01 -2.25784913e-01 2.54990518e-01 1.97468087e-01 -7.47729689e-02 1.82355210e-01 -6.54182553e-01 -8.66757512e-01 8.42091516e-02 -6.22365236e-01 2.17950329e-01 3.58610481e-01 -3.56743149e-02 3.59613359e-01 9.11874056e-01 -8.19087207e-01 1.24572372e+00 -2.30940485e+00 1.26482859e-01 4.84700143e-01 5.60350478e-01 1.99109480e-01 -4.36185986e-01 1.95714355e-01 -1.41371831e-01 1.09948136e-01 5.79990782e-02 3.92337322e-01 -8.24190080e-02 5.60817346e-02 -1.94441274e-01 6.62677348e-01 3.44152480e-01 6.67017639e-01 -7.70971537e-01 -3.09623122e-01 5.10142684e-01 8.10898483e-01 -7.22767353e-01 2.21732095e-01 1.32091627e-01 1.81026176e-01 9.32525396e-02 7.61085510e-01 9.48988676e-01 -2.74737552e-02 -8.82160813e-02 -6.84208512e-01 -2.40779087e-01 -7.61499032e-02 -1.18120635e+00 1.33428097e+00 -1.85932040e-01 1.02292502e+00 -5.67262294e-03 -5.14608622e-01 9.09873664e-01 5.39458096e-02 7.84282267e-01 -1.11866570e+00 -7.24495947e-02 2.87549019e-01 8.66134912e-02 -3.59030962e-01 7.05835223e-01 -3.44971828e-02 2.07140699e-01 -7.27333203e-02 1.99136168e-01 -1.80278763e-01 2.78421849e-01 2.08866149e-01 1.28966355e+00 -7.43161961e-02 3.26284736e-01 -1.33485109e-01 4.88255620e-01 -4.21321571e-01 3.58208239e-01 7.29685664e-01 -6.50343239e-01 9.53072488e-01 5.67906618e-01 -8.38049948e-02 -1.61926520e+00 -1.56824863e+00 -1.67519078e-01 1.21541369e+00 3.66425842e-01 -1.75711125e-01 -3.18402618e-01 -2.51690328e-01 -5.27697653e-02 3.18401396e-01 -4.09489632e-01 -1.66270569e-01 -7.08128035e-01 -5.46961606e-01 3.97425890e-01 2.76094675e-01 6.99746251e-01 -7.03245938e-01 -3.40527028e-01 8.97312313e-02 -1.31067619e-01 -1.24547553e+00 -4.45863813e-01 5.74098490e-02 -6.65589333e-01 -8.06757331e-01 -1.01691842e+00 -2.72442043e-01 -8.99037812e-03 4.33629453e-01 1.48325431e+00 -8.59334767e-02 -2.74722844e-01 5.64193964e-01 -6.20651662e-01 4.26253259e-01 -5.30111670e-01 -5.04066467e-01 -1.21584989e-01 -1.15094036e-01 1.45290107e-01 -5.57424963e-01 -7.95030177e-01 3.98724526e-01 -9.15764749e-01 -2.77168125e-01 5.17786264e-01 5.62486649e-01 6.11265540e-01 1.99381903e-01 4.48973030e-01 -2.78457493e-01 2.58760124e-01 -4.73255008e-01 -4.37265724e-01 8.91771168e-02 -4.76487100e-01 -1.30504772e-01 7.70508468e-01 -3.41419756e-01 -9.69318092e-01 -5.14385462e-01 -4.43723440e-01 -5.65565228e-01 -1.43770784e-01 -1.61792919e-01 -4.32131559e-01 -4.47463512e-01 5.38329124e-01 2.34069526e-01 -1.97712719e-01 -1.93277389e-01 4.93504941e-01 4.75005478e-01 7.14914262e-01 -2.68233120e-01 1.17266953e+00 5.65650165e-01 1.59109354e-01 -9.86712515e-01 -1.61751643e-01 -4.98442531e-01 -2.40870968e-01 -5.18455446e-01 6.55243993e-01 -1.09447396e+00 -4.86175746e-01 4.72174078e-01 -6.18858993e-01 -2.36521885e-01 -5.63977540e-01 3.31303269e-01 -8.61938477e-01 5.16323209e-01 -1.05590940e+00 -4.02365834e-01 -5.60756288e-02 -1.14585853e+00 9.13351595e-01 2.09428705e-02 2.31748879e-01 -6.45022750e-01 7.10912943e-02 8.05208683e-02 5.79123974e-01 8.30976292e-02 1.02228415e+00 -9.88949761e-02 -8.73909175e-01 2.98624605e-01 -7.06664205e-01 6.26341701e-01 -2.62844898e-02 4.19518679e-01 -8.83295596e-01 -4.26875740e-01 -2.07541272e-01 -1.20028764e-01 9.73396003e-01 6.92831755e-01 1.01629603e+00 -4.25891876e-02 4.45060730e-01 1.04854131e+00 1.87310696e+00 2.28333697e-01 1.20470917e+00 2.31032342e-01 6.12323940e-01 1.59437418e-01 6.19904816e-01 6.74560428e-01 1.24140859e-01 8.51374924e-01 4.20225590e-01 -3.74323457e-01 -6.39793158e-01 1.29557759e-01 7.71990061e-01 7.29512036e-01 -3.74666005e-01 -3.76039296e-01 -4.80902046e-01 3.23399782e-01 -1.15486562e+00 -1.14820778e+00 1.23911146e-02 2.00039935e+00 8.28137398e-01 6.50683194e-02 3.06372166e-01 4.20913607e-01 5.61775684e-01 5.11572301e-01 -5.82253158e-01 -5.32267034e-01 -6.40443385e-01 2.55310655e-01 5.83676934e-01 3.47660482e-01 -8.44072580e-01 6.29894376e-01 7.33634615e+00 8.48032355e-01 -1.06416953e+00 -1.05510958e-01 5.80195606e-01 -3.11953336e-01 -4.70009923e-01 -4.04201448e-01 -3.08383644e-01 4.72917616e-01 1.03167796e+00 7.16204345e-02 9.09044623e-01 6.75842106e-01 5.88237286e-01 1.17756138e-02 -1.02375722e+00 1.34803295e+00 1.19489349e-01 -1.13840353e+00 2.46814281e-01 1.30834013e-01 8.81117284e-01 6.20611683e-02 5.97798645e-01 1.12263530e-01 1.98950004e-02 -8.71141195e-01 8.30419123e-01 4.37418133e-01 1.29148531e+00 -9.09835398e-01 5.01655579e-01 -6.07706904e-01 -1.27700925e+00 -3.14430684e-01 -4.64034826e-01 3.23623121e-01 1.49533227e-01 6.72359049e-01 -3.32058519e-02 2.46072233e-01 9.27251935e-01 9.90448415e-01 -8.22201848e-01 1.03439856e+00 1.47866756e-01 2.99195349e-01 5.05862720e-02 5.69581151e-01 2.45333891e-02 -6.76862001e-02 6.71841860e-01 1.28124905e+00 4.70988631e-01 1.57971919e-01 -2.72162646e-01 5.70105314e-01 -2.24698260e-01 -1.04301281e-01 -4.96010721e-01 7.34917074e-02 2.18709990e-01 1.04899025e+00 -3.60513002e-01 -3.57160121e-01 -6.76489532e-01 1.07930911e+00 -2.73579806e-01 6.22142971e-01 -1.09926236e+00 -3.57303172e-01 1.01736009e+00 2.33098567e-01 5.26437163e-01 -3.54400635e-01 -1.23483464e-01 -1.34182155e+00 -1.63074955e-01 -1.34860075e+00 2.50127614e-01 -9.78762090e-01 -1.41160011e+00 4.23849702e-01 -2.41483837e-01 -1.43127143e+00 -2.09145918e-01 -7.40591586e-01 -1.43989444e-01 4.99028325e-01 -1.95622444e+00 -6.41368151e-01 -5.78524411e-01 1.10389829e+00 5.66931248e-01 -3.06740478e-02 2.58254975e-01 6.83922470e-01 -8.27428475e-02 6.07442200e-01 4.43023384e-01 7.36013427e-02 9.24605370e-01 -1.23065317e+00 1.08564168e-01 1.12973106e+00 -7.48815015e-02 1.95286512e-01 8.63085210e-01 -3.79691571e-01 -1.55846977e+00 -9.49355900e-01 1.46451652e-01 -6.89385235e-02 6.04494333e-01 -2.83614434e-02 -7.99080312e-01 3.80522996e-01 7.50810513e-03 2.44713292e-01 5.00020862e-01 -2.84650594e-01 -7.54682600e-01 -5.02168894e-01 -1.39716852e+00 5.22872865e-01 1.18244493e+00 -9.05081809e-01 -1.59031928e-01 -2.49142796e-01 6.59349442e-01 2.49813925e-02 -1.27744234e+00 4.41713393e-01 6.15515888e-01 -1.59821308e+00 1.22634304e+00 1.90485138e-02 7.18218386e-01 -4.53207552e-01 -7.18866229e-01 -1.16965079e+00 -4.68121856e-01 -6.85446203e-01 -5.22858202e-01 1.03275096e+00 -2.20455453e-02 -1.40164375e-01 3.98663789e-01 2.21030265e-01 -1.23989813e-01 -3.52118671e-01 -8.69916677e-01 -7.30684698e-01 -2.02724949e-01 -4.83084410e-01 4.21013027e-01 6.71770751e-01 -1.91741571e-01 -2.40912855e-01 -5.04226625e-01 2.84915641e-02 8.02852988e-01 -1.04445554e-01 3.30753326e-01 -7.54554272e-01 -3.83072138e-01 -4.65180367e-01 -8.56130958e-01 -1.24416268e+00 -3.84919822e-01 -3.77832651e-01 2.83118151e-02 -1.22338736e+00 3.53529990e-01 -3.39854747e-01 -4.61567670e-01 -5.40186390e-02 -7.64892017e-03 7.66074955e-01 4.87139583e-01 1.30813435e-01 -8.47909212e-01 3.75340283e-01 1.42508090e+00 -7.38711581e-02 -5.32789789e-02 -4.73200679e-01 -4.21492070e-01 4.59768295e-01 4.54885811e-01 1.71890017e-02 -2.53452539e-01 -3.48937333e-01 2.68138975e-01 2.00092956e-01 4.81972814e-01 -1.36290157e+00 -2.63349444e-01 -1.20396137e-01 8.45634460e-01 -4.06374097e-01 3.65536064e-01 -8.92873883e-01 2.09184498e-01 1.94856256e-01 -2.84909993e-01 1.28322408e-01 -1.98391810e-01 4.00688469e-01 -3.16623002e-01 3.92086923e-01 1.31430185e+00 4.41514440e-02 -1.20844817e+00 4.83431160e-01 -4.68748033e-01 1.28262654e-01 8.33914638e-01 -4.74122852e-01 -5.73329568e-01 -7.07891226e-01 -5.24779856e-01 -3.80021542e-01 9.64119375e-01 4.57473159e-01 9.25049305e-01 -1.28294218e+00 -7.62196541e-01 2.97293544e-01 -3.05404961e-02 -7.86877215e-01 5.54040492e-01 5.69328547e-01 -8.05267751e-01 4.15907195e-03 -7.08716452e-01 -5.51754594e-01 -1.07421887e+00 7.32920945e-01 3.74577790e-01 -1.14397712e-01 -5.99447668e-01 5.19354403e-01 4.92676422e-02 4.66947019e-01 1.21024484e-02 -2.89864808e-01 -1.17166638e-01 -5.89411557e-02 7.50184894e-01 8.21811795e-01 -2.23545171e-02 -1.03747261e+00 -2.81958282e-01 7.77357519e-01 1.79043353e-01 4.80786115e-02 1.22277009e+00 -7.26252913e-01 1.41600952e-01 3.24472398e-01 1.54686975e+00 1.42371833e-01 -1.70275640e+00 -1.42635107e-02 -4.30715173e-01 -9.45785701e-01 2.39220172e-01 -8.39341879e-01 -1.39303839e+00 8.06738317e-01 1.11235678e+00 3.08073103e-01 1.78251457e+00 -6.45485520e-02 8.15757334e-01 -1.74383894e-01 4.73560244e-01 -1.09238827e+00 4.73917335e-01 3.10210168e-01 7.44167626e-01 -1.11347175e+00 1.22107238e-01 -4.29574519e-01 -5.81364572e-01 1.37195218e+00 2.82703191e-01 -2.10707888e-01 4.00959998e-01 3.48924488e-01 -1.47161763e-02 1.54605269e-01 -6.69868648e-01 -3.41668636e-01 4.11642134e-01 9.21970487e-01 2.46091813e-01 -1.71778455e-01 7.86959678e-02 -1.52001277e-01 7.49054085e-03 -9.09775868e-02 7.65164196e-01 4.69361335e-01 -6.67336464e-01 -6.61111295e-01 -3.84499580e-01 3.27057868e-01 -6.11626923e-01 -3.31802785e-01 1.17256604e-01 7.00444341e-01 2.18926877e-01 1.15452480e+00 1.40239134e-01 -5.56732833e-01 3.02929074e-01 -4.38732296e-01 7.19392419e-01 8.55740160e-02 -3.05053800e-01 7.96326026e-02 -8.44391957e-02 -1.18179286e+00 -3.86181593e-01 -4.00185376e-01 -9.55696583e-01 -1.00553858e+00 1.84754729e-01 -5.40932119e-01 5.02673447e-01 5.66957891e-01 1.37607262e-01 5.27362406e-01 1.00598562e+00 -9.30632472e-01 -3.20861876e-01 -3.96028638e-01 -9.40172255e-01 7.99200714e-01 8.21454346e-01 -4.67708290e-01 -5.48304319e-01 1.40438750e-01]
[11.5588960647583, -1.9111452102661133]
ede036c7-7a8f-4474-ae39-aeae5ce3217c
towards-understanding-distributional
2110.03155
null
https://arxiv.org/abs/2110.03155v4
https://arxiv.org/pdf/2110.03155v4.pdf
Interpreting Distributional Reinforcement Learning: A Regularization Perspective
Distributional reinforcement learning~(RL) is a class of state-of-the-art algorithms that estimate the whole distribution of the total return rather than only its expectation. Despite the remarkable performance of distributional RL, a theoretical understanding of its advantages over expectation-based RL remains elusive. In this paper, we attribute the superiority of distributional RL to its regularization effect in terms of the value distribution information regardless of its expectation. Firstly, by leverage of a variant of the gross error model in robust statistics, we decompose the value distribution into its expectation and the remaining distribution part. As such, the extra benefit of distributional RL compared with expectation-based RL is mainly interpreted as the impact of a \textit{risk-sensitive entropy regularization} within the Neural Fitted Z-Iteration framework. Meanwhile, we establish a bridge between the risk-sensitive entropy regularization of distributional RL and the vanilla entropy in maximum entropy RL, focusing specifically on actor-critic algorithms. It reveals that distributional RL induces a corrected reward function and thus promotes a risk-sensitive exploration against the intrinsic uncertainty of the environment. Finally, extensive experiments corroborate the role of the regularization effect of distributional RL and uncover mutual impacts of different entropy regularization. Our research paves a way towards better interpreting the efficacy of distributional RL algorithms, especially through the lens of regularization.
['Bei Jiang', 'Xiaodong Yan', 'Linglong Kong', 'Yafei Wang', 'Enze Shi', 'Yi Liu', 'Yingnan Zhao', 'Ke Sun']
2021-10-07
null
null
null
null
['distributional-reinforcement-learning']
['methodology']
[-1.77833617e-01 3.98522764e-01 -2.97113299e-01 -1.97060362e-01 -8.33137691e-01 -4.14571047e-01 4.10168499e-01 2.30166182e-01 -7.90522099e-01 8.17084491e-01 5.09367645e-01 -4.03169751e-01 -4.42063242e-01 -7.24911571e-01 -8.29927444e-01 -1.09152567e+00 -1.28674544e-02 8.43800455e-02 -3.73283952e-01 -2.45693788e-01 2.55621433e-01 2.48082533e-01 -1.22852647e+00 -4.33900803e-01 9.32139158e-01 1.06910384e+00 1.50511742e-01 8.39543641e-02 2.27211998e-03 9.47934031e-01 -4.62447077e-01 -5.03466666e-01 4.40565050e-01 -4.07499969e-01 -2.24909946e-01 -2.28456184e-01 -3.15896630e-01 -3.59083205e-01 -2.12147802e-01 1.40892589e+00 6.71177149e-01 4.70542938e-01 8.17705214e-01 -1.08988190e+00 -6.27711296e-01 9.45828915e-01 -6.51920259e-01 5.41578755e-02 -7.62337074e-03 5.60121238e-01 1.48440325e+00 -4.84823048e-01 4.54597771e-01 1.28912711e+00 4.76171970e-01 3.72296780e-01 -1.34884405e+00 -5.70888877e-01 4.05358404e-01 -4.94903624e-01 -1.07246506e+00 -1.09280564e-01 9.08883154e-01 -4.20940548e-01 7.39256203e-01 -1.24993294e-01 5.26615143e-01 1.05826330e+00 2.56797731e-01 1.13520765e+00 1.28899741e+00 -3.41575325e-01 5.08997142e-01 3.07746142e-01 -2.14713722e-01 3.98732632e-01 3.75995576e-01 6.99584603e-01 -3.49667907e-01 -6.20377697e-02 7.26503253e-01 -1.47137821e-01 -2.09691912e-01 -6.21506453e-01 -6.95105493e-01 9.84037936e-01 4.12267089e-01 7.23070204e-02 -5.32687843e-01 5.70143044e-01 5.15140474e-01 2.36373439e-01 7.27785945e-01 6.75524235e-01 -4.45766509e-01 -2.97437608e-01 -7.05144942e-01 2.83670545e-01 5.11175096e-01 5.95416307e-01 6.98448956e-01 4.52577680e-01 -3.55368048e-01 4.73615646e-01 7.47784555e-01 7.60412753e-01 4.03020382e-01 -9.36519980e-01 5.82464933e-01 5.51147580e-01 2.25124180e-01 -8.33724439e-01 -2.80859292e-01 -7.91693270e-01 -5.62945127e-01 3.25862527e-01 4.85264301e-01 -5.50381660e-01 -1.17428020e-01 2.34779501e+00 1.31770089e-01 -8.10516477e-02 1.32886872e-01 9.13707495e-01 9.63488817e-02 3.10712874e-01 1.25721782e-01 -4.75994110e-01 8.42280388e-01 -4.96433616e-01 -6.88535750e-01 -1.05510883e-01 5.28531373e-01 -2.27090847e-02 1.51802015e+00 7.81951174e-02 -1.05889666e+00 3.11454646e-02 -8.55103433e-01 4.19409662e-01 -2.55760886e-02 -1.75149798e-01 6.83313072e-01 6.13996804e-01 -8.40150476e-01 7.48840570e-01 -5.80607772e-01 2.59289294e-01 4.83856648e-01 1.70824662e-01 2.25819990e-01 5.70726991e-01 -1.21821940e+00 8.02623987e-01 2.89678097e-01 6.17375597e-02 -8.65035772e-01 -8.33334029e-01 -7.24847376e-01 3.20971191e-01 7.14605629e-01 -4.63011771e-01 1.19018734e+00 -1.17547679e+00 -1.81797230e+00 2.67778039e-01 2.76389837e-01 -6.91561282e-01 8.57442379e-01 -2.71728039e-01 1.82953864e-01 -7.69917145e-02 -9.26736668e-02 1.87913746e-01 9.12308693e-01 -1.11326897e+00 -1.50971368e-01 -4.09762233e-01 -8.60588998e-02 3.53291363e-01 -2.52726346e-01 -4.43927258e-01 2.11182728e-01 -7.29324222e-01 -5.46245813e-01 -8.22489440e-01 -3.95657927e-01 -4.35321331e-01 -2.43833825e-01 -2.43831232e-01 -7.50647560e-02 -2.73929954e-01 1.47465599e+00 -2.19127440e+00 2.36533016e-01 4.66854095e-01 7.18880668e-02 -1.18132710e-01 -7.17130378e-02 4.19581681e-01 -6.46632388e-02 2.67363340e-01 -2.65491098e-01 -2.56285161e-01 5.37372887e-01 -6.01967983e-02 -7.91183829e-01 7.35639215e-01 8.75397921e-02 9.81506944e-01 -9.52584386e-01 -6.29596785e-02 1.43101364e-01 4.09489244e-01 -7.22857416e-01 1.00892097e-01 -5.50286591e-01 4.39339042e-01 -7.40775347e-01 2.22342476e-01 3.40358496e-01 -2.58814567e-03 1.35954618e-01 1.94161654e-01 -2.50592619e-01 1.66965887e-01 -1.04869604e+00 1.44458961e+00 -4.20579106e-01 3.58263582e-01 1.83912665e-02 -9.73801136e-01 8.56163144e-01 -8.60959571e-03 5.84349453e-01 -6.55515194e-01 4.07791018e-01 2.29095370e-01 -7.72208124e-02 -2.70722538e-01 5.34148932e-01 -3.13954681e-01 -1.55069232e-01 9.39231753e-01 -1.46148711e-01 -1.43453017e-01 -1.40253335e-01 1.09694339e-01 6.71526492e-01 4.49274749e-01 2.68907666e-01 -5.09234011e-01 1.18799947e-01 -5.71535230e-01 5.89341998e-01 8.60055625e-01 -4.09452856e-01 -3.68456915e-02 9.47073936e-01 4.72415239e-03 -8.81863058e-01 -1.10140443e+00 -1.47396713e-01 1.19727945e+00 -4.62353081e-02 -2.76285410e-01 -5.62880516e-01 -8.14264715e-01 4.82159495e-01 1.26089394e+00 -7.76295602e-01 -6.06202662e-01 -2.78812379e-01 -9.30895150e-01 6.94848061e-01 3.77019942e-01 4.77549642e-01 -1.04764915e+00 -1.00202405e+00 -7.73636401e-02 -5.07871509e-02 -3.74541998e-01 -5.16569436e-01 3.81510764e-01 -9.36714888e-01 -7.58885145e-01 -5.29645562e-01 3.50445546e-02 5.04136086e-01 -3.32072854e-01 1.11434841e+00 -4.29722399e-01 1.88511133e-01 7.57906497e-01 -2.75747664e-02 -8.45135272e-01 -1.82617247e-01 -1.86645836e-01 2.70254582e-01 -3.51500958e-02 1.64708704e-01 -5.95386028e-01 -7.18327045e-01 -1.28384843e-01 -7.58851349e-01 -5.90671182e-01 7.08890438e-01 8.24023783e-01 5.74702978e-01 3.06471549e-02 8.84462893e-01 -7.24173009e-01 1.22098768e+00 -6.69333696e-01 -9.11800563e-01 1.99663609e-01 -1.08438599e+00 7.17193127e-01 4.08000410e-01 -5.18107116e-01 -1.38487327e+00 -3.46460313e-01 7.26803541e-02 -4.48078930e-01 4.01276857e-01 5.73784530e-01 4.59132902e-02 3.34816784e-01 4.29164082e-01 2.75611341e-01 2.34224677e-01 -2.12965623e-01 6.29178762e-01 2.24037424e-01 7.41985068e-02 -9.06688690e-01 5.26664972e-01 2.86182076e-01 7.18124211e-02 -5.55959404e-01 -9.58773613e-01 -7.54053472e-03 -1.83453597e-02 -3.67046863e-01 7.99722552e-01 -8.55439723e-01 -1.14537978e+00 5.28745875e-02 -6.12420142e-01 -5.30038655e-01 -1.13267982e+00 6.57358825e-01 -1.01263094e+00 2.14742258e-01 -3.34542662e-01 -1.47002316e+00 -2.79661804e-01 -1.19339287e+00 7.02202201e-01 1.38118625e-01 1.24863103e-01 -1.24308121e+00 2.15817854e-01 -5.81196919e-02 5.80653012e-01 2.60545224e-01 9.30214047e-01 -6.67831421e-01 -3.51701051e-01 7.18364790e-02 2.67032310e-02 4.73752886e-01 -3.02656561e-01 -1.33972540e-01 -8.37207615e-01 -2.81844318e-01 3.96799177e-01 -5.75990140e-01 1.10515618e+00 7.80931592e-01 9.84718561e-01 -3.63235176e-01 2.96397597e-01 5.99086046e-01 1.50392818e+00 1.33392597e-02 3.46787125e-01 4.10937697e-01 2.15232968e-01 5.72792649e-01 6.02241635e-01 1.01199341e+00 2.55547166e-01 2.08077237e-01 7.42512345e-01 2.58474529e-01 4.88796175e-01 -6.60805464e-01 8.06108057e-01 5.36196709e-01 2.31370348e-02 -5.48789911e-02 -5.80921829e-01 1.95551932e-01 -1.91996694e+00 -1.03153980e+00 4.72118825e-01 2.40820456e+00 9.44809437e-01 1.50398985e-01 2.33121991e-01 -2.17369914e-01 4.60996509e-01 4.01425511e-01 -9.83687639e-01 -5.42085707e-01 -1.99817494e-01 -2.09830478e-01 7.42816567e-01 4.74746704e-01 -8.14016998e-01 8.37218523e-01 6.30037785e+00 1.05102360e+00 -9.36054170e-01 -2.70141602e-01 7.38158047e-01 -3.35883588e-01 -8.78729522e-01 -1.19111322e-01 -6.42924905e-01 4.90782201e-01 8.73706341e-01 -4.30179983e-01 7.55963624e-01 9.89742637e-01 2.71394491e-01 -2.94004261e-01 -9.55823481e-01 6.87064171e-01 -1.79231212e-01 -9.54361141e-01 -2.21127495e-01 3.93223286e-01 5.73451102e-01 1.70889780e-01 6.19732916e-01 6.61744833e-01 6.77958786e-01 -1.03390276e+00 1.11835527e+00 6.69050694e-01 5.87985694e-01 -1.08938396e+00 7.36621737e-01 6.07483804e-01 -8.21274221e-01 -4.33657914e-01 -5.15533984e-01 -2.04386711e-01 -1.96446255e-01 7.32234120e-01 -5.43847859e-01 2.64259815e-01 3.37405175e-01 5.22249877e-01 -1.27427652e-01 3.72014999e-01 -4.50046450e-01 6.21919453e-01 -2.03711092e-01 -1.98882416e-01 4.14513886e-01 -7.70520210e-01 7.85545349e-01 1.03408933e+00 4.69098017e-02 -8.22142884e-02 1.03278821e-02 1.36037993e+00 -1.27856195e-01 1.78533524e-01 -8.60689342e-01 -3.17353010e-01 4.03556019e-01 8.54653180e-01 -3.70173663e-01 1.67553887e-01 -1.79814681e-01 2.39777297e-01 6.13330185e-01 4.50547785e-01 -9.65660453e-01 -1.61097929e-01 6.02510810e-01 -2.69535750e-01 2.14125335e-01 4.64742631e-03 -4.90812600e-01 -1.15297949e+00 1.15224712e-01 -7.32583880e-01 2.63893813e-01 -1.82236150e-01 -1.28473437e+00 1.03419237e-01 -5.35897948e-02 -1.00821495e+00 -2.32818946e-01 -2.80961514e-01 -4.24867600e-01 7.14099169e-01 -1.63556898e+00 -4.41141456e-01 4.99948978e-01 5.59774101e-01 2.86067724e-01 -1.81947947e-01 5.47625244e-01 -2.06359088e-01 -8.54209781e-01 7.13753939e-01 5.82023442e-01 -1.13140754e-01 6.35321558e-01 -1.46346593e+00 -2.85112143e-01 4.58068222e-01 -1.90114170e-01 6.78405762e-01 7.76992679e-01 -5.86052477e-01 -1.57502627e+00 -9.26601529e-01 2.39321560e-01 -5.34730196e-01 9.56467152e-01 -1.39825627e-01 -4.42300826e-01 6.36691630e-01 7.16331527e-02 -1.95384085e-01 5.70338607e-01 1.55389816e-01 -4.42053914e-01 -1.34011418e-01 -1.18556261e+00 7.04735816e-01 6.55463457e-01 -5.36147118e-01 -5.77267051e-01 -6.42467514e-02 9.71626043e-01 4.67218608e-02 -8.43364358e-01 2.67953753e-01 4.69847977e-01 -1.11754870e+00 9.03038144e-01 -4.87644136e-01 5.16940057e-01 1.47021621e-01 -3.35790277e-01 -1.52384090e+00 -3.62746641e-02 -6.76511407e-01 -1.31397679e-01 1.17247975e+00 3.10870349e-01 -8.84860039e-01 4.42694455e-01 6.28124833e-01 1.39753342e-01 -9.22165036e-01 -9.23120439e-01 -8.72498035e-01 6.57464683e-01 -6.10721946e-01 3.94436926e-01 7.29172885e-01 2.11178884e-01 2.21231561e-02 -4.03358549e-01 -2.45758697e-01 9.20250893e-01 5.85923456e-02 4.68354493e-01 -1.06970572e+00 -4.94661897e-01 -7.02094316e-01 2.60263741e-01 -8.35094452e-01 7.43230939e-01 -9.98698592e-01 1.90208673e-01 -8.37956607e-01 7.83032328e-02 -4.40788537e-01 -6.34854734e-01 1.63856566e-01 -1.85917333e-01 -4.89571363e-01 5.10518968e-01 2.42202580e-01 -7.40867913e-01 1.15699482e+00 1.18260765e+00 7.97379091e-02 -5.23179471e-01 1.18574008e-01 -1.02713871e+00 9.94180739e-01 8.93998206e-01 -5.19640505e-01 -7.20430911e-01 -2.44380116e-01 7.35798478e-01 2.72293895e-01 1.55728787e-01 -3.24003965e-01 -3.15165147e-02 -4.76164877e-01 4.19586152e-02 -1.80863336e-01 -2.73833051e-03 -7.31032848e-01 -4.01354134e-01 4.22439456e-01 -9.16764259e-01 -4.00525369e-02 1.59248427e-01 9.92196262e-01 3.37245837e-02 -7.90218115e-02 7.36430168e-01 -1.59457207e-01 -9.65892673e-02 2.65013780e-02 -3.97549361e-01 7.03226447e-01 8.97939384e-01 2.25532815e-01 -5.12471236e-02 -5.68003893e-01 -3.57464135e-01 2.62683004e-01 3.35889548e-01 -4.49321680e-02 4.88947272e-01 -1.07915831e+00 -6.49188042e-01 1.37965441e-01 -1.86839506e-01 -1.74809366e-01 3.75045724e-02 1.06116199e+00 2.03149915e-01 3.06438357e-01 8.43675062e-02 -2.56820321e-01 -4.32868212e-01 6.11557126e-01 4.02888507e-01 -6.77649796e-01 -6.22176111e-01 7.88717926e-01 3.84265959e-01 -3.61424476e-01 5.75434506e-01 -3.20416719e-01 9.96355712e-02 1.97161421e-01 2.02885464e-01 5.38776278e-01 -4.67602998e-01 -1.41919523e-01 -2.79709697e-01 3.24724019e-01 3.24373126e-01 -4.92615968e-01 1.38109326e+00 -2.85344839e-01 3.29010636e-02 6.90090895e-01 8.23834956e-01 1.49057761e-01 -1.81623256e+00 -1.82609543e-01 2.68029511e-01 -2.17860460e-01 1.64960861e-01 -8.25397730e-01 -1.18003094e+00 8.14951241e-01 3.31385344e-01 -9.78070199e-02 1.00223374e+00 -2.13823020e-01 2.00708732e-01 4.42601264e-01 3.63587618e-01 -1.45722544e+00 1.81635663e-01 4.69489098e-01 8.12656641e-01 -1.24750698e+00 1.93918765e-01 3.74464899e-01 -1.13814569e+00 7.96840191e-01 2.29096100e-01 -4.75210011e-01 7.20697224e-01 3.35808516e-01 -3.31169188e-01 4.80777323e-02 -9.15017962e-01 -2.49731466e-01 -6.30334811e-03 3.43028665e-01 3.60214174e-01 2.09185749e-01 -4.21154886e-01 9.92382884e-01 -1.81164905e-01 -2.28232056e-01 3.47780257e-01 6.40028656e-01 -3.38309079e-01 -7.53052115e-01 2.87249126e-02 2.89247334e-01 -6.20088100e-01 -2.34952256e-01 -1.99141920e-01 8.14500749e-01 -3.41470063e-01 8.38024616e-01 -6.92781359e-02 -1.93201974e-01 2.51411617e-01 1.24054044e-01 1.96885079e-01 -1.23316839e-01 -8.33484173e-01 2.38429993e-01 -3.17657530e-01 -7.28884637e-01 -2.05754340e-01 -7.93479919e-01 -1.28623497e+00 -9.80174989e-02 -4.39321876e-01 2.69036949e-01 6.64562583e-01 8.67620587e-01 2.90191948e-01 4.80199248e-01 7.90105343e-01 -5.63282967e-01 -1.64802182e+00 -6.14993036e-01 -9.04769897e-01 3.78735214e-01 3.47610235e-01 -7.31519282e-01 -9.62041557e-01 -6.61191046e-01]
[4.111146450042725, 2.535659074783325]
4a8835a9-5ad6-47f8-b64f-cac259699247
an-online-sequence-to-sequence-model-for
1706.06428
null
http://arxiv.org/abs/1706.06428v1
http://arxiv.org/pdf/1706.06428v1.pdf
An online sequence-to-sequence model for noisy speech recognition
Generative models have long been the dominant approach for speech recognition. The success of these models however relies on the use of sophisticated recipes and complicated machinery that is not easily accessible to non-practitioners. Recent innovations in Deep Learning have given rise to an alternative - discriminative models called Sequence-to-Sequence models, that can almost match the accuracy of state of the art generative models. While these models are easy to train as they can be trained end-to-end in a single step, they have a practical limitation that they can only be used for offline recognition. This is because the models require that the entirety of the input sequence be available at the beginning of inference, an assumption that is not valid for instantaneous speech recognition. To address this problem, online sequence-to-sequence models were recently introduced. These models are able to start producing outputs as data arrives, and the model feels confident enough to output partial transcripts. These models, like sequence-to-sequence are causal - the output produced by the model until any time, $t$, affects the features that are computed subsequently. This makes the model inherently more powerful than generative models that are unable to change features that are computed from the data. This paper highlights two main contributions - an improvement to online sequence-to-sequence model training, and its application to noisy settings with mixed speech from two speakers.
['George Tucker', 'Chung-Cheng Chiu', 'Yuping Luo', 'Navdeep Jaitly', 'Kevin Swersky', 'Ilya Sutskever', 'Dieterich Lawson']
2017-06-16
null
null
null
null
['noisy-speech-recognition']
['speech']
[ 3.52104753e-01 2.11301446e-01 7.68530443e-02 -6.17757261e-01 -7.63114870e-01 -8.63573253e-01 6.60355508e-01 -3.07587624e-01 -2.97406465e-01 7.17488348e-01 -2.78102476e-02 -5.06870747e-01 9.12052244e-02 -6.35396957e-01 -8.67781818e-01 -7.55820572e-01 -7.75327533e-02 6.01122558e-01 7.96007216e-02 -3.38453621e-01 -2.54688747e-02 4.14904594e-01 -1.80944622e+00 4.68372852e-01 5.39274752e-01 6.39249384e-01 5.98399341e-01 1.19931030e+00 -4.09060448e-01 7.98741758e-01 -8.49488854e-01 -3.34416538e-01 4.58577760e-02 -8.20147991e-01 -5.41459262e-01 8.80138427e-02 5.07820956e-02 -4.86985803e-01 -3.81415218e-01 7.85487592e-01 5.63886106e-01 1.79363042e-02 4.42548305e-01 -9.66793120e-01 -4.51629847e-01 7.53308833e-01 1.70880124e-01 1.61805570e-01 6.24190032e-01 1.41214654e-01 9.49082434e-01 -8.25824380e-01 6.78075790e-01 1.24560785e+00 6.20290756e-01 8.33576679e-01 -1.19548750e+00 -3.21059287e-01 1.95766002e-01 2.99872845e-01 -1.01153731e+00 -8.57958496e-01 5.18217206e-01 -2.91238040e-01 1.44211638e+00 5.61390877e-01 6.84640884e-01 1.45070529e+00 -1.12883717e-01 9.71725464e-01 9.74729240e-01 -7.37460911e-01 2.38266051e-01 7.61692077e-02 -2.35590085e-01 5.20309627e-01 -4.36151266e-01 5.54111063e-01 -8.63789737e-01 1.46783218e-01 5.94340384e-01 -2.30261907e-01 -2.52198935e-01 6.08449727e-02 -8.41350257e-01 6.93822443e-01 -6.85659423e-02 5.50172031e-01 -3.71948451e-01 1.26634672e-01 2.95789629e-01 7.31802046e-01 3.01063418e-01 1.34252325e-01 -5.26119530e-01 -8.56207013e-01 -1.38652515e+00 3.14827412e-01 1.15177834e+00 9.06946182e-01 5.23072958e-01 4.28787410e-01 4.72317077e-02 8.00121963e-01 4.60947305e-01 5.25619149e-01 6.98355496e-01 -5.15095592e-01 2.86672026e-01 1.09891951e-01 3.51070240e-02 -5.96361458e-01 -3.07986289e-02 -4.57228035e-01 -6.26276076e-01 1.18050493e-01 4.24270689e-01 -1.29925549e-01 -1.24589074e+00 1.73160696e+00 9.63217393e-02 1.04901455e-01 -1.80411451e-02 5.96824050e-01 5.03989637e-01 8.98800194e-01 -2.22791478e-01 -4.16720718e-01 7.61939406e-01 -7.01442301e-01 -9.09921587e-01 -3.71054500e-01 5.16298294e-01 -8.69131684e-01 9.74921763e-01 4.99464333e-01 -1.20812476e+00 -6.40490413e-01 -9.94720459e-01 1.94452956e-01 -5.39303958e-01 -1.13316037e-01 3.44053328e-01 1.00986648e+00 -1.32034433e+00 7.74675429e-01 -1.02593708e+00 -2.64604032e-01 6.36489987e-02 4.57708001e-01 -1.63169846e-01 -1.62725262e-02 -1.26857996e+00 1.07787991e+00 3.03324997e-01 3.39102715e-01 -1.20657420e+00 -3.68551046e-01 -5.99297523e-01 7.10000098e-02 2.64209479e-01 -4.95396107e-01 1.77134442e+00 -1.32885361e+00 -2.07622457e+00 4.43867445e-01 -5.94435990e-01 -4.43912715e-01 6.96625888e-01 -3.00573438e-01 -3.81999701e-01 -4.13801484e-02 -3.71078312e-01 2.67156273e-01 1.05240130e+00 -1.11459887e+00 -5.84618032e-01 -4.53793891e-02 -1.90266654e-01 -4.49330099e-02 7.47160316e-02 5.67351058e-02 -4.06445295e-01 -4.15206045e-01 -4.00935002e-02 -8.00073147e-01 -1.30449414e-01 -4.58481640e-01 -2.83871025e-01 -2.78369397e-01 7.85053432e-01 -8.21260989e-01 1.23934209e+00 -2.07942629e+00 1.64014608e-01 -1.61656514e-02 -3.14065069e-01 8.24229360e-01 -1.06111333e-01 9.54684258e-01 -1.13751262e-01 2.94265538e-01 -1.32824495e-01 -5.57142258e-01 7.98606575e-02 5.26161551e-01 -5.79593301e-01 1.54125109e-01 3.98390621e-01 9.29451108e-01 -8.88010681e-01 -1.80422783e-01 2.98777312e-01 4.72710550e-01 -4.02004808e-01 4.80212331e-01 -3.49796295e-01 3.11711341e-01 -5.92325181e-02 2.91409880e-01 3.95353287e-01 1.03936002e-01 2.82009035e-01 5.04854798e-01 -2.50323743e-01 8.16018283e-01 -1.06247902e+00 1.50115895e+00 -5.91588914e-01 7.04030931e-01 5.10915518e-02 -1.28747022e+00 8.56178880e-01 8.84859622e-01 5.46940565e-02 -4.34533596e-01 -4.97415438e-02 5.03115654e-01 2.45472729e-01 -6.90905809e-01 2.46948138e-01 -4.56835270e-01 1.42195225e-01 4.01532710e-01 2.71082520e-01 -2.70497948e-01 3.80659908e-01 9.68565866e-02 1.07844901e+00 4.06411648e-01 2.40456179e-01 3.23071897e-01 3.50643665e-01 -1.98325440e-01 5.63778400e-01 1.06592083e+00 1.43747643e-01 6.31452501e-01 3.14404756e-01 -2.67014146e-01 -1.25115716e+00 -9.90375876e-01 2.67580062e-01 1.15241385e+00 -5.13607204e-01 -2.65195996e-01 -8.41116846e-01 -5.62339902e-01 -2.98541903e-01 9.18699265e-01 -4.63510990e-01 -7.11030811e-02 -5.42494059e-01 -3.09908330e-01 6.12112939e-01 5.69499254e-01 1.93092246e-02 -1.26971817e+00 -4.34655517e-01 6.90758944e-01 -1.42412826e-01 -8.43221605e-01 -2.58298993e-01 5.20006180e-01 -8.80421937e-01 -5.70857704e-01 -5.59912920e-01 -4.89591420e-01 4.51079726e-01 -1.37777612e-01 1.02591634e+00 1.31175518e-01 -1.43597037e-01 3.92345250e-01 -5.24255991e-01 -5.75479031e-01 -1.08626330e+00 1.77954331e-01 5.91213405e-02 1.00985661e-01 3.29683125e-01 -7.37568498e-01 -3.73509713e-02 1.01774260e-01 -9.09894645e-01 1.05121685e-03 8.64224613e-01 1.09846842e+00 2.67085582e-01 -1.87820658e-01 8.68555844e-01 -8.04722965e-01 4.82399434e-01 -2.57865012e-01 -3.10300082e-01 2.81938463e-01 -6.14886284e-01 1.33341774e-01 9.14486110e-01 -4.96619135e-01 -1.06112373e+00 1.72713682e-01 -7.78589725e-01 -5.13812959e-01 -4.21276927e-01 7.28023887e-01 -2.32726112e-01 5.00740170e-01 4.44760472e-01 7.33675778e-01 2.33241737e-01 -6.19735241e-01 3.96270305e-01 9.45619881e-01 3.80879968e-01 -2.19157547e-01 7.61371195e-01 7.02611282e-02 -2.83832163e-01 -1.13995743e+00 -5.79958618e-01 -5.00075579e-01 -7.52093136e-01 -2.69003481e-01 4.09772962e-01 -5.68774998e-01 -5.10556042e-01 5.38348258e-01 -1.30182457e+00 -4.21652079e-01 -3.74151528e-01 3.57420713e-01 -7.14162588e-01 3.48813385e-01 -5.56014597e-01 -1.34697509e+00 -1.34960696e-01 -9.08439100e-01 7.18817472e-01 9.68596339e-02 -2.89346218e-01 -1.08068562e+00 -9.10093337e-02 5.80507368e-02 5.44452727e-01 -3.00468057e-01 7.10398376e-01 -7.88481116e-01 -5.64417779e-01 -3.35523129e-01 4.38880980e-01 7.40096927e-01 5.46511300e-02 4.19772208e-01 -1.24830139e+00 -2.49117583e-01 4.59260382e-02 -2.27901518e-01 7.04059064e-01 2.54447550e-01 9.12518859e-01 -4.28529382e-01 -2.66857535e-01 3.36017758e-01 1.02778339e+00 5.70477009e-01 7.60106862e-01 -6.03717715e-02 2.34983534e-01 4.11902636e-01 2.79643446e-01 1.88671306e-01 -4.12698910e-02 6.81219757e-01 2.09754303e-01 1.24349624e-01 -2.35545501e-01 -5.97819269e-01 7.63052523e-01 1.33145380e+00 -1.39746621e-01 -5.18510818e-01 -6.83428526e-01 7.05828428e-01 -1.76959550e+00 -1.42182958e+00 -8.75664651e-02 2.22815847e+00 9.11758423e-01 2.23551795e-01 3.51324379e-02 4.38046306e-01 3.76220644e-01 -4.19684015e-02 -3.78002435e-01 -7.92957246e-01 4.44062799e-02 5.87105691e-01 1.93558093e-02 6.66901112e-01 -7.07456589e-01 8.96974444e-01 7.23925495e+00 8.95279825e-01 -1.23140514e+00 2.10051268e-01 2.32102871e-01 -1.30282640e-01 -4.91879463e-01 5.50826378e-02 -9.64923143e-01 5.82727253e-01 1.57109666e+00 2.24107970e-02 5.97731471e-01 8.34212840e-01 4.87845719e-01 -7.95254260e-02 -1.32528794e+00 9.00574386e-01 -5.62950317e-03 -1.15262961e+00 -3.56249288e-02 -6.45535365e-02 3.07742268e-01 -5.91268539e-02 1.58523142e-01 5.30193686e-01 2.82686800e-01 -1.22291648e+00 9.12941337e-01 5.88583887e-01 7.24247932e-01 -7.01885045e-01 6.44238472e-01 7.75133014e-01 -8.33205163e-01 -1.04771584e-01 -2.38553807e-01 -2.37434447e-01 4.52770025e-01 6.21264994e-01 -1.40225708e+00 5.82292736e-01 3.78489047e-01 2.24624217e-01 -1.20621532e-01 8.07853460e-01 -4.64901626e-01 1.13286221e+00 -3.60363513e-01 -3.09588045e-01 3.05379838e-01 4.67492938e-02 5.35833061e-01 1.52322853e+00 5.54219186e-01 -1.44176558e-01 3.80454883e-02 7.93871224e-01 2.51882195e-01 -1.36384308e-01 -7.51773596e-01 -4.46807146e-01 4.80559319e-01 8.44955564e-01 -3.28487128e-01 -3.98600876e-01 -4.73687559e-01 9.96880233e-01 3.21265996e-01 2.99869865e-01 -5.98355830e-01 -3.20854425e-01 4.61489350e-01 1.46839693e-01 5.44583499e-01 -4.91801262e-01 4.54759672e-02 -8.24481010e-01 7.08411122e-03 -1.13201308e+00 -1.26965806e-01 -6.32046521e-01 -1.16701031e+00 5.61657310e-01 -1.92854404e-01 -8.47949922e-01 -1.15410006e+00 -5.55053473e-01 -4.25690740e-01 1.13998055e+00 -1.27190697e+00 -9.69921827e-01 1.57762691e-01 2.88602650e-01 8.35865736e-01 -7.04005081e-03 1.12627792e+00 2.78580487e-01 -2.99966604e-01 5.73045850e-01 2.98428833e-01 1.72420919e-01 4.23569769e-01 -1.22630191e+00 6.60354376e-01 1.02332532e+00 6.04888439e-01 7.95090020e-01 8.50481510e-01 -4.83364046e-01 -1.42672265e+00 -6.58402383e-01 1.35061228e+00 -4.55638766e-01 3.57838243e-01 -6.52286470e-01 -8.70803952e-01 7.66165316e-01 1.65954620e-01 -2.94595182e-01 7.05237031e-01 2.40275070e-01 -1.03434317e-01 -5.59349023e-02 -6.35970056e-01 3.92626673e-01 9.91729081e-01 -7.50567019e-01 -7.72751391e-01 8.87706503e-02 4.60885882e-01 -4.40179795e-01 -3.96247059e-01 7.85464421e-03 6.30585551e-01 -1.18627453e+00 7.09218025e-01 -6.38271928e-01 1.45312682e-01 -1.87174171e-01 -9.62254032e-02 -1.50760221e+00 -6.60922527e-02 -1.15331697e+00 -1.85397640e-01 1.19399166e+00 8.16731215e-01 -6.56164169e-01 5.23592174e-01 4.44348782e-01 -3.94860029e-01 -6.07308745e-01 -1.15110648e+00 -1.16969597e+00 -1.01712368e-01 -7.29517758e-01 5.78070462e-01 6.81831777e-01 -1.32668493e-02 3.12858075e-01 -4.20472354e-01 -3.03812232e-02 9.43643302e-02 -2.27887300e-03 7.69875765e-01 -1.01026177e+00 -7.46313870e-01 -3.06386024e-01 -3.03735465e-01 -1.42007494e+00 -3.90747376e-02 -6.75585866e-01 4.46811676e-01 -1.38669026e+00 -1.94065094e-01 -3.82358760e-01 -2.02633339e-04 5.75062752e-01 -1.55871704e-01 -2.35488415e-01 1.93760648e-01 7.83247724e-02 2.52702683e-02 4.23156351e-01 9.85831439e-01 1.28070608e-01 -2.75834739e-01 2.93561369e-01 -1.62905529e-01 4.36557204e-01 7.99596131e-01 -6.04358733e-01 -4.83153135e-01 -2.56799757e-01 1.91580877e-01 1.55107394e-01 2.42556810e-01 -8.38337541e-01 1.29698843e-01 6.48610294e-02 2.15793937e-01 -6.13116562e-01 5.26165783e-01 -5.37635684e-01 3.75247121e-01 3.91159356e-01 -3.14305246e-01 -1.34388104e-01 5.05625606e-02 5.26640654e-01 -4.10599649e-01 -6.78287804e-01 4.81294811e-01 -3.74096990e-01 -5.21902978e-01 1.02834940e-01 -8.00056815e-01 -2.07633689e-01 7.45177507e-01 -2.48621434e-01 8.43948126e-02 -8.12172592e-01 -9.90088880e-01 -4.35066432e-01 1.05783768e-01 4.82429177e-01 4.58597660e-01 -8.70911658e-01 -6.00375295e-01 3.38020504e-01 -3.18409473e-01 -1.35517508e-01 1.25711530e-01 7.28570938e-01 -2.62359202e-01 6.61130130e-01 2.15655252e-01 -6.93498373e-01 -1.26072848e+00 5.56477427e-01 2.96101123e-01 -2.37303853e-01 -4.58330899e-01 1.01190817e+00 -2.21345872e-01 -4.41181034e-01 1.88996568e-01 -2.91672409e-01 2.60929197e-01 7.26412907e-02 5.43189228e-01 -1.64463203e-02 2.78345287e-01 -4.50004756e-01 -1.72988996e-01 1.03445858e-01 -1.63725898e-01 -6.56923473e-01 1.38081551e+00 -1.46594852e-01 1.65047273e-01 8.91843379e-01 9.08545017e-01 -5.01227714e-02 -1.23292255e+00 -4.79369387e-02 -3.08384243e-02 -4.64576155e-01 -9.57261622e-02 -1.01987422e+00 -6.51579261e-01 1.29227066e+00 3.80935013e-01 6.21429205e-01 8.52373779e-01 -2.17940092e-01 6.61372423e-01 3.10595930e-01 2.59701818e-01 -1.25856137e+00 -8.98872316e-03 8.08147848e-01 8.33792865e-01 -8.95573854e-01 -4.94499773e-01 -1.40272334e-01 -3.86857122e-01 1.28760624e+00 6.43040612e-02 3.04245442e-01 4.34050292e-01 5.70628762e-01 2.27110133e-01 2.09793046e-01 -1.01629329e+00 -3.20762962e-01 5.64678796e-02 8.73785973e-01 6.62084281e-01 6.66989163e-02 -1.41268715e-01 3.72927725e-01 -4.32075620e-01 1.94098264e-01 3.63537878e-01 1.03373790e+00 -4.56495583e-01 -1.52738523e+00 -3.31770480e-01 2.67412931e-01 -3.97257864e-01 -2.00063422e-01 -4.51254189e-01 6.71825051e-01 4.04473096e-02 1.36947238e+00 1.98457837e-02 -3.46792579e-01 1.40218213e-01 8.17452788e-01 5.75228572e-01 -7.45328009e-01 -5.85596085e-01 1.52106479e-01 3.29587519e-01 -3.39243889e-01 -1.94166541e-01 -9.04064655e-01 -1.05420423e+00 -3.03241074e-01 -4.87024724e-01 2.27814987e-01 9.73947942e-01 1.13929033e+00 3.65110576e-01 6.37454808e-01 7.34521449e-01 -7.85091400e-01 -8.65151584e-01 -1.29225910e+00 -4.20563400e-01 1.18531309e-01 4.64111209e-01 -3.63891542e-01 -5.01154006e-01 4.55973744e-01]
[14.482097625732422, 6.815150737762451]
89fd9007-21f5-4f4c-b8d2-6e93f4dabdb0
learning-in-implicit-generative-models
1610.03483
null
http://arxiv.org/abs/1610.03483v4
http://arxiv.org/pdf/1610.03483v4.pdf
Learning in Implicit Generative Models
Generative adversarial networks (GANs) provide an algorithmic framework for constructing generative models with several appealing properties: they do not require a likelihood function to be specified, only a generating procedure; they provide samples that are sharp and compelling; and they allow us to harness our knowledge of building highly accurate neural network classifiers. Here, we develop our understanding of GANs with the aim of forming a rich view of this growing area of machine learning---to build connections to the diverse set of statistical thinking on this topic, of which much can be gained by a mutual exchange of ideas. We frame GANs within the wider landscape of algorithms for learning in implicit generative models--models that only specify a stochastic procedure with which to generate data--and relate these ideas to modelling problems in related fields, such as econometrics and approximate Bayesian computation. We develop likelihood-free inference methods and highlight hypothesis testing as a principle for learning in implicit generative models, using which we are able to derive the objective function used by GANs, and many other related objectives. The testing viewpoint directs our focus to the general problem of density ratio estimation. There are four approaches for density ratio estimation, one of which is a solution using classifiers to distinguish real from generated data. Other approaches such as divergence minimisation and moment matching have also been explored in the GAN literature, and we synthesise these views to form an understanding in terms of the relationships between them and the wider literature, highlighting avenues for future exploration and cross-pollination.
['Shakir Mohamed', 'Balaji Lakshminarayanan']
2016-10-11
null
null
null
null
['density-ratio-estimation']
['methodology']
[ 5.90094507e-01 4.54445571e-01 -2.53017485e-01 -4.31413591e-01 -8.52529943e-01 -6.17370367e-01 8.87712300e-01 -4.61241335e-01 -4.27653193e-02 9.75361407e-01 1.21204779e-01 -5.18318176e-01 -3.50777805e-01 -1.22561300e+00 -7.20450163e-01 -9.93645251e-01 1.76680043e-01 6.93618953e-01 -3.51517856e-01 -4.89673577e-02 1.99256927e-01 4.48336631e-01 -1.59012926e+00 -1.41469598e-01 9.39386368e-01 9.76652980e-01 -2.81290412e-01 6.45923555e-01 -1.36000693e-01 8.40551734e-01 -7.90967345e-01 -1.00772548e+00 -4.41870205e-02 -1.02850890e+00 -4.78371352e-01 1.45984646e-02 1.17509812e-01 -2.44043574e-01 3.53821442e-02 9.26461816e-01 5.50263345e-01 -6.71384931e-02 1.29673707e+00 -1.54488564e+00 -9.52843964e-01 4.61920470e-01 -2.61722654e-01 -2.59922147e-02 1.25719070e-01 1.00409999e-01 9.04158115e-01 -4.64989036e-01 2.49926314e-01 1.09550726e+00 8.24436665e-01 7.84534395e-01 -1.35673845e+00 -5.45522392e-01 -1.60256088e-01 -4.67259698e-02 -1.20107400e+00 -4.41673815e-01 6.48391485e-01 -5.85459709e-01 5.61572969e-01 5.33939838e-01 9.56483245e-01 1.43494654e+00 8.56110379e-02 1.00736606e+00 1.25850368e+00 -5.62863588e-01 5.37512600e-01 1.70356855e-01 -3.27174693e-01 6.04948282e-01 1.35409340e-01 4.29341495e-01 -4.99659359e-01 -8.37789699e-02 1.14619219e+00 -2.18416065e-01 -2.08133891e-01 -3.18189472e-01 -6.76766992e-01 1.40912366e+00 5.04822284e-02 1.98382095e-01 -1.55200839e-01 4.00524348e-01 1.19788758e-01 1.77790180e-01 7.03323007e-01 2.30826959e-01 -5.14106303e-02 -2.38139763e-01 -1.19926739e+00 4.14186865e-01 1.02916300e+00 8.20231974e-01 6.98521972e-01 5.58126986e-01 -2.90296655e-02 7.99440801e-01 5.51976144e-01 5.93794703e-01 3.35762113e-01 -1.03592229e+00 -6.98826686e-02 1.13232262e-01 -2.06455007e-01 -5.28611958e-01 2.00295851e-01 -5.07873237e-01 -9.16931748e-01 2.88555771e-01 3.73529464e-01 -3.12570602e-01 -8.77214849e-01 2.09527397e+00 1.96404099e-01 2.06848264e-01 -1.00134864e-01 4.48493481e-01 5.03725111e-01 5.58038116e-01 -1.50900677e-01 -1.80895209e-01 9.98062909e-01 -3.35466415e-01 -4.71691549e-01 -5.00715785e-02 2.69320697e-01 -5.31997919e-01 7.85006523e-01 4.66677994e-01 -1.23013616e+00 -2.74819046e-01 -9.96313453e-01 1.55136883e-01 -4.03413802e-01 -1.20285600e-01 1.07497823e+00 1.26144338e+00 -1.27633941e+00 6.94320619e-01 -8.91728818e-01 -3.79565626e-01 6.68658674e-01 2.44291916e-01 4.66209687e-02 9.09046084e-02 -1.03406382e+00 9.13654387e-01 2.22795159e-01 1.66884497e-01 -8.41763258e-01 -8.51808667e-01 -8.90129209e-01 -1.72001213e-01 8.66113231e-02 -1.28720427e+00 1.27767265e+00 -1.20595109e+00 -1.72511971e+00 7.99125314e-01 -1.57267243e-01 -4.33620512e-01 6.36063099e-01 -5.51405130e-03 -2.63999790e-01 -2.43436173e-01 -2.96595301e-02 7.15182126e-01 9.40298796e-01 -1.13600338e+00 -3.79425287e-01 -1.78791463e-01 -2.47858360e-01 -9.99132469e-02 7.09727034e-02 -2.33936265e-01 1.69634111e-02 -1.05571425e+00 -1.93352461e-01 -6.74985468e-01 -4.96319681e-02 -1.44929305e-01 -4.30519134e-01 -1.61026075e-01 6.34874165e-01 -5.11937320e-01 9.27853525e-01 -1.64466274e+00 2.61585772e-01 2.63130099e-01 1.57470584e-01 9.55838785e-02 -2.03354694e-02 5.41083336e-01 -4.33533899e-02 2.51750767e-01 -7.48664916e-01 -3.45170557e-01 2.52480328e-01 3.13380569e-01 -6.91436350e-01 4.48966414e-01 6.04985476e-01 1.34111547e+00 -7.20333815e-01 -1.34005234e-01 3.68875444e-01 8.50828171e-01 -5.72487950e-01 2.91885853e-01 -3.14303011e-01 3.55430186e-01 -2.54512161e-01 7.03227937e-01 5.36337316e-01 9.03049018e-03 -1.94562692e-02 1.95469111e-01 1.15067042e-01 2.68755823e-01 -1.01164651e+00 1.28491569e+00 -3.91264409e-01 7.29803264e-01 -4.56130989e-02 -1.36068368e+00 9.71919119e-01 1.47795454e-01 2.34295547e-01 -2.64439255e-01 2.35048637e-01 2.08624437e-01 -2.63030250e-02 3.86158340e-02 8.94427598e-02 -6.53877795e-01 4.64489162e-02 6.29559636e-01 4.10745323e-01 -6.05861247e-01 9.85215232e-02 -6.34187534e-02 8.77110004e-01 3.97688866e-01 2.03214481e-01 -1.37697265e-01 1.11376628e-01 -4.27849442e-01 2.30798915e-01 1.01127601e+00 2.81073362e-01 5.94523072e-01 6.00821912e-01 -2.10801318e-01 -1.12872529e+00 -1.59025264e+00 -3.99955422e-01 6.26920640e-01 -5.20148277e-01 -2.27138877e-01 -1.01429200e+00 -4.41697329e-01 -2.64293738e-02 9.59596753e-01 -9.47132230e-01 -1.24122344e-01 -2.31429726e-01 -1.31780767e+00 5.79137862e-01 7.62328446e-01 3.00450832e-01 -1.15359986e+00 -3.04147184e-01 1.00447871e-02 2.02948991e-02 -5.72316349e-01 9.27299857e-02 4.45785165e-01 -1.08801973e+00 -9.83916521e-01 -1.02055633e+00 -4.75483179e-01 4.19304222e-01 -4.43010569e-01 1.39403415e+00 -1.66135922e-01 -2.29210675e-01 6.90771639e-01 -1.75696701e-01 -7.49492586e-01 -8.58025193e-01 2.31888574e-02 -2.34375671e-01 -1.81548193e-01 5.58840871e-01 -9.77930069e-01 -2.20562249e-01 2.45926261e-01 -1.13359952e+00 1.71617873e-03 7.26526380e-01 9.79868829e-01 4.49947327e-01 -1.44241169e-01 8.38691533e-01 -1.00145876e+00 7.19069660e-01 -6.51038885e-01 -6.56608582e-01 2.27868333e-01 -6.78249955e-01 1.46148801e-01 2.83221900e-01 -2.95063913e-01 -8.86856258e-01 -4.45223957e-01 -6.20787919e-01 -2.01060846e-01 -2.89948046e-01 3.46156180e-01 -4.36360389e-01 5.52069247e-02 4.59495068e-01 5.06916106e-01 1.69532493e-01 -1.88076660e-01 7.52739251e-01 4.10909563e-01 6.14665449e-01 -8.15655112e-01 9.07708466e-01 2.73815572e-01 2.42634580e-01 -8.88381720e-01 -6.33322895e-01 1.50109217e-01 -3.81427437e-01 -7.48187676e-02 7.99257874e-01 -6.60935581e-01 -4.93095815e-01 6.02885664e-01 -7.81451225e-01 -5.93948543e-01 -8.07863116e-01 2.48433903e-01 -1.14960849e+00 6.87878504e-02 -5.68856835e-01 -1.23156643e+00 -1.47610396e-01 -1.10134435e+00 1.09718013e+00 3.79551917e-01 -3.74153197e-01 -1.60187578e+00 2.43603408e-01 1.62688032e-01 5.91556072e-01 3.74606609e-01 1.10028923e+00 -6.01084232e-01 -5.28622150e-01 3.67408693e-02 9.17040184e-02 6.55433059e-01 1.33291930e-01 2.36346990e-01 -1.29774070e+00 7.32111791e-03 2.05156401e-01 -3.42179745e-01 8.48449290e-01 7.91063964e-01 1.20349908e+00 -3.90544623e-01 -2.03295648e-01 8.62958670e-01 1.23158360e+00 1.85507521e-01 1.04010797e+00 -1.02544188e-01 4.89813924e-01 7.31953442e-01 2.17651017e-02 3.04511845e-01 6.62589096e-04 4.38027114e-01 3.37844998e-01 -1.28722399e-01 -2.76197251e-02 -5.55911779e-01 2.53748506e-01 7.47685790e-01 -2.31932148e-01 -4.57119286e-01 -4.42839980e-01 4.23276901e-01 -1.60214114e+00 -1.35215950e+00 1.75291430e-02 2.31469917e+00 8.83838713e-01 2.55792160e-02 4.34847176e-01 3.06829542e-01 7.37970233e-01 1.12578459e-01 -5.40777028e-01 -3.99614066e-01 -1.95554167e-01 8.47966850e-01 1.77134335e-01 5.30999959e-01 -9.21050012e-01 6.89863384e-01 7.71429300e+00 1.12631190e+00 -9.17945445e-01 -1.23753943e-01 1.18382597e+00 3.07904296e-02 -8.55477929e-01 5.17351106e-02 -6.22972608e-01 5.62633276e-01 1.11370969e+00 1.40103344e-02 6.24617755e-01 7.95358062e-01 -2.04500899e-01 -2.12049797e-01 -1.39930916e+00 8.61833394e-01 1.19416408e-01 -1.46064615e+00 4.24577072e-02 4.11197543e-01 8.24242949e-01 -1.62576288e-01 4.04915839e-01 3.22183222e-01 6.59625828e-01 -1.57900941e+00 7.04282641e-01 6.57085896e-01 8.04697990e-01 -8.79816294e-01 6.01282060e-01 3.14149320e-01 -7.24422276e-01 3.75554472e-01 -2.80628204e-01 -1.21126659e-02 2.91992396e-01 8.30757618e-01 -7.38302290e-01 4.32322711e-01 3.43534619e-01 5.32314241e-01 -3.13774198e-01 9.54327166e-01 -4.38442945e-01 9.71347809e-01 -3.43735754e-01 -1.29883394e-01 -1.38150930e-01 -5.45961797e-01 4.42454845e-01 1.09129250e+00 5.94085813e-01 -3.41117412e-01 -5.31898677e-01 1.55276179e+00 8.90855417e-02 -2.62817115e-01 -7.38951623e-01 -2.36464456e-01 4.49372530e-01 1.08708596e+00 -8.07142615e-01 3.28180902e-02 -2.08823711e-01 7.16614842e-01 1.96806371e-01 3.35791618e-01 -8.70138705e-01 -1.91886470e-01 5.76533377e-01 1.59209743e-01 2.81368256e-01 4.16302383e-02 -3.37398499e-01 -9.77221966e-01 -2.56679952e-01 -1.00093412e+00 1.92615449e-01 -7.36507654e-01 -1.68775511e+00 2.32162505e-01 3.42510134e-01 -6.43040895e-01 -9.53564525e-01 -7.92760432e-01 -8.01093102e-01 1.15722620e+00 -1.20010185e+00 -1.15087748e+00 -9.03496295e-02 2.56145477e-01 3.10881168e-01 -2.72358894e-01 9.65119123e-01 8.61668866e-03 -3.13084722e-01 6.76906645e-01 2.19494179e-01 1.18744060e-01 2.76231378e-01 -1.42381597e+00 7.08568931e-01 7.02175677e-01 6.16515100e-01 6.80286527e-01 5.79215348e-01 -6.17801964e-01 -1.28346252e+00 -7.90704668e-01 4.89801019e-01 -7.65366733e-01 5.47299743e-01 -5.12224555e-01 -6.48968279e-01 7.45712638e-01 4.31806520e-02 -2.91873455e-01 1.07024789e+00 1.97782129e-01 -9.53268781e-02 1.78696260e-01 -1.22295117e+00 4.40540463e-01 8.50652695e-01 -5.16410172e-01 -4.16871756e-01 -1.44128380e-02 2.30618641e-01 -2.39811480e-01 -7.67766356e-01 2.79041201e-01 7.20205069e-01 -1.37857127e+00 9.81057167e-01 -4.62244093e-01 5.34655213e-01 6.92626908e-02 -2.68246204e-01 -1.33232307e+00 -4.32285778e-02 -8.52785587e-01 -2.81152099e-01 1.67172885e+00 3.27764690e-01 -1.02748656e+00 9.14122641e-01 4.76154327e-01 -8.59709531e-02 -1.09737635e+00 -9.36630487e-01 -6.73775673e-01 5.21265388e-01 -8.12202990e-01 6.18230700e-01 6.50609016e-01 -4.71910477e-01 2.07300887e-01 -2.30965391e-01 -4.46520388e-01 6.65993989e-01 -2.22189829e-01 8.10946882e-01 -1.42020488e+00 -5.60882032e-01 -6.80706024e-01 -5.38454950e-01 -1.04055357e+00 3.11737776e-01 -9.51068521e-01 -1.70502186e-01 -1.37095881e+00 1.92738742e-01 -4.82484967e-01 1.40656844e-01 1.17438592e-01 4.15728800e-02 4.70982283e-01 -1.22549161e-01 -3.72313336e-02 7.77522996e-02 6.41847312e-01 9.81508851e-01 9.72612724e-02 9.51485708e-02 3.56371164e-01 -1.14356697e+00 7.94014335e-01 6.18846476e-01 -3.78100425e-01 -7.32747078e-01 2.51133591e-02 3.72951210e-01 -7.07408041e-02 7.54118860e-01 -8.77148628e-01 -3.50445062e-01 -1.51082754e-01 8.24503899e-01 -3.80327672e-01 4.05735523e-01 -3.76988441e-01 3.84523660e-01 2.20059961e-01 -2.69884199e-01 4.00681682e-02 1.16753876e-01 4.96251285e-01 -5.73283900e-03 -5.46404898e-01 7.02729762e-01 -1.93620980e-01 -1.99292645e-01 2.70096064e-01 -4.25506890e-01 5.14815271e-01 7.91706026e-01 -3.26880783e-01 -2.75432497e-01 -8.49703431e-01 -5.53092480e-01 -3.34826499e-01 6.05507851e-01 8.69577527e-02 4.54769552e-01 -1.36178863e+00 -7.86421061e-01 3.22303444e-01 -3.56762081e-01 3.05919684e-02 -1.45835215e-02 6.48102701e-01 -3.31343740e-01 2.69118875e-01 -1.79146603e-02 -5.80964029e-01 -5.52273929e-01 3.15607786e-01 4.41832900e-01 -1.17124490e-01 -2.57908791e-01 1.01070273e+00 3.90328288e-01 -3.85771096e-01 5.37218936e-02 -8.95084664e-02 1.63695559e-01 -2.10141041e-03 3.90792996e-01 3.84760767e-01 -5.06868474e-02 -4.42584932e-01 -2.02275842e-01 4.35173273e-01 3.48032206e-01 -4.09590960e-01 1.39216089e+00 2.69115746e-01 -1.16891451e-01 6.26217544e-01 1.00551903e+00 3.31982411e-02 -1.36085641e+00 2.36598685e-01 -3.10882121e-01 -2.91385353e-01 -1.57651111e-01 -9.32090461e-01 -1.02575004e+00 9.98318434e-01 4.29823935e-01 5.24148822e-01 1.11843097e+00 1.80395156e-01 2.74941027e-01 -1.65105402e-01 1.26760542e-01 -7.45457709e-01 7.16257542e-02 2.23892316e-01 7.98088074e-01 -9.26259041e-01 6.97781378e-03 -2.32790411e-01 -3.31758887e-01 1.04172647e+00 1.99408978e-01 -1.94472685e-01 4.96213466e-01 6.60568058e-01 -3.08821082e-01 7.54317865e-02 -5.08408129e-01 -1.71751440e-01 4.89593804e-01 1.38127697e+00 6.53946400e-01 -7.15944394e-02 -1.06757889e-02 4.65038151e-01 -7.39178300e-01 4.57388014e-02 1.84943154e-01 5.58456361e-01 -1.77940920e-01 -1.58155990e+00 -2.66016662e-01 7.44789720e-01 -3.70259196e-01 -2.19947189e-01 -4.14688200e-01 9.42427516e-01 1.00954950e-01 8.23624432e-01 2.35110521e-01 -1.97274461e-01 -3.34217906e-01 4.16417956e-01 9.81048286e-01 -5.00414014e-01 -2.32633129e-01 7.94405416e-02 2.13950202e-02 -1.05112217e-01 -5.63229442e-01 -9.22426522e-01 -3.78958941e-01 -6.19020045e-01 -4.52646226e-01 2.31423795e-01 6.81977928e-01 1.07554007e+00 -9.45573822e-02 5.28652370e-01 2.93904483e-01 -9.36327398e-01 -5.01331389e-01 -8.70425940e-01 -6.96379006e-01 -1.91942915e-01 3.15077752e-02 -5.33966243e-01 -5.64648509e-01 -8.24236944e-02]
[11.568778038024902, -0.043232399970293045]
5b7499c2-114a-4421-94fa-1fba56a3653c
tddiscourse-a-dataset-for-discourse-level
null
null
https://aclanthology.org/W19-5929
https://aclanthology.org/W19-5929.pdf
TDDiscourse: A Dataset for Discourse-Level Temporal Ordering of Events
Prior work on temporal relation classification has focused extensively on event pairs in the same or adjacent sentences (local), paying scant attention to discourse-level (global) pairs. This restricts the ability of systems to learn temporal links between global pairs, since reliance on local syntactic features suffices to achieve reasonable performance on existing datasets. However, systems should be capable of incorporating cues from document-level structure to assign temporal relations. In this work, we take a first step towards discourse-level temporal ordering by creating TDDiscourse, the first dataset focusing specifically on temporal links between event pairs which are more than one sentence apart. We create TDDiscourse by augmenting TimeBank-Dense, a corpus of English news articles, manually annotating global pairs that cannot be inferred automatically from existing annotations. Our annotations double the number of temporal links in TimeBank-Dense, while possessing several desirable properties such as focusing on long-distance pairs and not being automatically inferable. We adapt and benchmark the performance of three state-of-the-art models on TDDiscourse and observe that existing systems indeed find discourse-level temporal ordering harder.
['Luke Breitfeller', 'Aakanksha Naik', 'Carolyn Rose']
2019-09-01
null
null
null
ws-2019-9
['temporal-relation-classification']
['natural-language-processing']
[-4.74180467e-02 4.07696754e-01 -6.41638398e-01 -5.22357166e-01 -9.15457726e-01 -9.70358431e-01 1.26819706e+00 7.57907093e-01 -4.23417926e-01 9.10588086e-01 7.92512715e-01 -6.31039739e-01 -3.10629815e-01 -5.51277936e-01 -4.02805686e-01 -2.46563390e-01 -7.34313011e-01 6.62080765e-01 6.18854463e-01 -4.15754110e-01 8.04050639e-02 9.69151333e-02 -1.09620202e+00 6.89457834e-01 5.57255983e-01 7.34762073e-01 -8.11878070e-02 6.42000496e-01 1.41115934e-01 1.14534223e+00 -4.63637471e-01 -3.16546470e-01 5.92310652e-02 -6.97369754e-01 -1.43536699e+00 -3.30248326e-01 1.77895159e-01 -2.13758767e-01 -6.04340971e-01 3.43875706e-01 1.19447298e-01 3.39666367e-01 6.80714905e-01 -1.00540900e+00 -5.23906350e-01 1.07477438e+00 -2.48358324e-01 1.03958654e+00 6.43934369e-01 -1.75644130e-01 1.70255649e+00 -4.16123569e-01 1.16069067e+00 1.23615789e+00 7.46064901e-01 4.78247330e-02 -1.32235646e+00 -1.17124364e-01 4.66775119e-01 5.50404608e-01 -1.05505788e+00 -7.02268183e-01 6.48620486e-01 -6.43193007e-01 1.75540686e+00 3.24579328e-01 3.90067220e-01 1.05354714e+00 1.82366863e-01 5.79566061e-01 8.88640642e-01 -6.08643591e-01 4.53604721e-02 -6.45898819e-01 4.46628541e-01 3.26048166e-01 -4.91291404e-01 -5.80059998e-02 -7.98289180e-01 5.17376214e-02 4.86446887e-01 -6.56420171e-01 -1.68083593e-01 -1.46298492e-02 -1.59641051e+00 4.67074960e-01 2.33281240e-01 7.08994031e-01 -1.08677194e-01 -1.75897703e-01 9.27183032e-01 7.12702811e-01 7.21535325e-01 5.74596405e-01 -6.58940375e-01 -6.75666153e-01 -7.03856945e-01 3.27986181e-01 8.68223071e-01 8.39043975e-01 3.72172505e-01 -6.62780643e-01 -2.74740785e-01 7.09915519e-01 -1.40681073e-01 -4.60245490e-01 5.19719601e-01 -1.10267055e+00 9.22014058e-01 3.47190589e-01 2.14029938e-01 -1.01699853e+00 -5.72449982e-01 6.70433939e-02 -2.43208468e-01 -5.61033249e-01 7.50621617e-01 -8.23349133e-02 -3.78813148e-01 1.94801342e+00 2.99276471e-01 1.41163230e-01 1.33720309e-01 6.05106175e-01 6.40929699e-01 6.70653045e-01 -3.92788611e-02 -7.45329857e-01 1.39390314e+00 -8.23476374e-01 -9.22258973e-01 -2.42733076e-01 1.32371330e+00 -7.07209527e-01 8.60943496e-01 6.77559525e-02 -1.10943270e+00 -2.21351758e-01 -9.30483937e-01 -3.48262310e-01 -3.35865051e-01 -2.78117090e-01 7.48960674e-01 -8.22836906e-02 -8.46331000e-01 7.64916241e-01 -1.19158745e+00 -5.64505577e-01 -5.97430468e-02 1.10140927e-01 -4.48325694e-01 3.30118388e-01 -1.64862657e+00 1.33203721e+00 6.66307330e-01 5.81536666e-02 -4.92162794e-01 -7.16609478e-01 -1.12500942e+00 -2.99930423e-01 5.89444578e-01 -2.34781221e-01 1.75693405e+00 -4.23503429e-01 -1.15123653e+00 9.87508476e-01 -4.61955994e-01 -8.53323400e-01 3.98629814e-01 -1.62627950e-01 -6.80700243e-01 2.44991347e-01 3.46059293e-01 4.25561756e-01 8.37020874e-02 -5.21046817e-01 -6.78276360e-01 -1.51372477e-01 4.55402404e-01 3.21190774e-01 -3.07906866e-01 5.19942939e-01 -2.62121499e-01 -6.75155938e-01 1.58362895e-01 -8.74766350e-01 1.13072149e-01 -3.73801619e-01 -3.62317860e-01 -9.13747668e-01 8.87080669e-01 -6.50942743e-01 1.75244701e+00 -1.89540195e+00 1.42023966e-01 -3.81324559e-01 -6.64979443e-02 -1.27253219e-01 -2.49603000e-02 8.50936294e-01 -3.11460137e-01 1.47450790e-01 -7.56690130e-02 -3.34523439e-01 1.54884100e-01 4.95961785e-01 -4.34312791e-01 6.00659251e-01 3.32483709e-01 8.69964182e-01 -1.23283589e+00 -7.74388731e-01 -5.87097071e-02 -2.20847920e-01 -3.81328464e-01 3.90169211e-02 -5.31294525e-01 5.17327189e-01 -3.27177465e-01 2.03002766e-01 -1.89050525e-01 -1.64481074e-01 6.13723040e-01 -1.15219049e-01 -4.71277744e-01 1.64386928e+00 -7.57724702e-01 1.83768773e+00 -4.74747211e-01 8.94178510e-01 -4.10084337e-01 -1.13772547e+00 5.69323540e-01 8.38217735e-01 7.27885902e-01 -8.36197078e-01 -3.99444737e-02 9.36299637e-02 4.13560301e-01 -7.51223505e-01 5.63617110e-01 -1.39614254e-01 -4.46120679e-01 7.56143332e-01 3.51704162e-04 1.14720181e-01 8.96571517e-01 3.84076148e-01 1.37863624e+00 2.33860373e-01 5.78203857e-01 -2.35969409e-01 1.42711386e-01 2.09738061e-01 7.88515329e-01 3.47433150e-01 -3.04136604e-01 4.01067555e-01 1.04090714e+00 -4.29060876e-01 -1.11929739e+00 -8.48651588e-01 -3.70502353e-01 1.17166770e+00 -1.22863621e-01 -1.06086612e+00 -1.88657507e-01 -9.04944181e-01 -3.15952331e-01 7.08778977e-01 -8.20770383e-01 2.81164825e-01 -1.17976201e+00 -6.08844519e-01 8.58033180e-01 6.47792339e-01 9.02475789e-02 -7.76446462e-01 -4.50468153e-01 5.42036772e-01 -6.78708851e-01 -1.31533158e+00 -6.13610625e-01 4.44864929e-01 -5.94181001e-01 -1.20956922e+00 -5.92204593e-02 -6.60422862e-01 1.18811965e-01 -2.71300226e-01 1.45594358e+00 -3.11555177e-01 1.54065877e-01 -2.70774383e-02 -5.24701476e-01 -6.66950122e-02 -3.44476223e-01 3.40840071e-01 -1.15663029e-01 -3.68088990e-01 3.91486824e-01 -7.56191015e-01 -2.19226792e-01 2.84176797e-01 -5.21974683e-01 1.11431945e-02 -5.18870987e-02 7.71923304e-01 7.05492944e-02 5.31296320e-02 6.85356796e-01 -8.24444532e-01 4.53339130e-01 -6.57797158e-01 -2.33558685e-01 1.14777021e-01 -7.96713531e-02 1.59314543e-01 4.54731405e-01 -4.10119265e-01 -1.19825065e+00 -3.75440419e-01 2.90243756e-02 1.55077621e-01 -7.97448531e-02 9.84402895e-01 1.17146783e-01 7.37953067e-01 7.48818874e-01 -2.81357884e-01 -3.78043503e-01 -3.46594721e-01 4.95493829e-01 2.10037351e-01 6.69754863e-01 -8.68043244e-01 3.02470773e-01 2.84375995e-01 -9.51267108e-02 -5.95813751e-01 -1.31404710e+00 -5.75207412e-01 -1.11472940e+00 3.35458927e-02 8.38006079e-01 -7.40065098e-01 -5.10742605e-01 4.78687743e-03 -1.33450437e+00 -7.23742664e-01 -1.76586002e-01 5.28107405e-01 -6.73363626e-01 3.14725339e-01 -1.12208879e+00 -5.64574540e-01 2.89570063e-01 -5.55567682e-01 8.22362900e-01 -4.27431345e-01 -1.15825999e+00 -1.42827880e+00 3.24582636e-01 1.65752500e-01 -1.23676404e-01 4.26986337e-01 9.52183366e-01 -8.58057320e-01 -3.03061455e-01 4.76336107e-02 2.09897712e-01 -2.73261458e-01 3.65057468e-01 1.32612750e-01 -6.44881189e-01 1.46343812e-01 -1.54188097e-01 -4.17857617e-01 8.25048923e-01 2.51622170e-01 5.94490886e-01 -5.62867463e-01 -5.81652105e-01 -2.27393173e-02 7.20896423e-01 3.58647019e-01 4.33987319e-01 6.29591107e-01 3.88320118e-01 9.40122247e-01 9.95208800e-01 3.03667307e-01 8.54937434e-01 1.00454056e+00 1.15016885e-01 2.77882248e-01 -9.23900157e-02 -1.54256254e-01 4.15374994e-01 9.60783303e-01 -1.31509081e-02 -4.72491324e-01 -1.15240455e+00 1.25201035e+00 -2.24107552e+00 -1.34532213e+00 -5.27459502e-01 1.84865165e+00 1.53456569e+00 4.86996293e-01 2.17738107e-01 3.11996788e-01 5.45929193e-01 6.16282582e-01 4.05267291e-02 -4.56236064e-01 -1.62044138e-01 -2.65470278e-02 4.79004830e-02 6.97742462e-01 -1.51847816e+00 9.19849992e-01 6.15849113e+00 4.71076339e-01 -7.74920940e-01 2.30107665e-01 4.78630692e-01 -2.87600756e-01 -2.51048684e-01 3.56952697e-01 -7.38480747e-01 4.41270858e-01 1.28920066e+00 -2.48941600e-01 1.25288501e-01 8.43377635e-02 5.23347139e-01 -1.07750058e-01 -1.85634208e+00 4.05858576e-01 -2.99338669e-01 -1.50407398e+00 -5.68730175e-01 -1.81327015e-01 7.29976416e-01 -3.69921662e-02 -3.67978036e-01 2.59518176e-01 4.84255672e-01 -8.66361797e-01 1.05296218e+00 2.79069245e-01 5.13983369e-01 -3.64708573e-01 5.84740281e-01 4.27803397e-01 -1.43740857e+00 1.78524911e-01 2.90108502e-01 -5.71794808e-01 5.71687758e-01 5.40293932e-01 -9.10535932e-01 8.04370522e-01 5.01186192e-01 1.18991685e+00 -3.61517757e-01 6.13560796e-01 -2.87289500e-01 7.06250250e-01 -5.68904579e-01 1.32786930e-01 4.68489587e-01 1.46230593e-01 5.45668185e-01 1.28911293e+00 1.04919523e-01 4.75561678e-01 3.74455959e-01 3.48415554e-01 3.93573754e-02 -2.21608341e-01 -8.31567585e-01 -1.08303934e-01 7.61195660e-01 7.82762825e-01 -6.29378974e-01 -4.05629218e-01 -5.21943927e-01 6.49968565e-01 5.10059059e-01 -1.00387605e-02 -8.90430391e-01 -2.07524002e-01 4.92196441e-01 1.38219267e-01 2.44975030e-01 -6.84800327e-01 -2.20055521e-01 -1.13441336e+00 3.58003736e-01 -5.89239299e-01 9.80271816e-01 -5.06678760e-01 -1.32118547e+00 5.48983932e-01 3.47631186e-01 -1.10030079e+00 -7.65535772e-01 -7.09742308e-02 -6.27434134e-01 6.77427590e-01 -1.12737095e+00 -1.14055264e+00 3.51879835e-01 3.42800587e-01 6.47507846e-01 4.58447844e-01 7.09297061e-01 4.35957998e-01 -4.67308074e-01 4.41048890e-01 -3.99026871e-01 2.54835546e-01 1.16138601e+00 -1.40134025e+00 4.67534482e-01 9.88105655e-01 2.52534032e-01 7.88584054e-01 9.21854675e-01 -6.19136572e-01 -8.22246015e-01 -9.74578142e-01 1.82836568e+00 -8.57343614e-01 1.34735656e+00 -2.91278422e-01 -9.51809883e-01 1.30890083e+00 4.52043653e-01 -4.75158769e-04 4.19041306e-01 8.39966238e-01 -5.55910289e-01 1.02695547e-01 -4.31035280e-01 7.33362675e-01 1.45243180e+00 -1.03016460e+00 -1.20594943e+00 5.42661846e-01 8.70615125e-01 -7.88268864e-01 -1.30262113e+00 4.02138293e-01 1.82192937e-01 -6.83662593e-01 7.74798512e-01 -7.28207290e-01 7.37127364e-01 -3.38456631e-01 3.62619273e-02 -1.08921933e+00 -1.36106730e-01 -8.39037299e-01 -2.87649006e-01 1.77893496e+00 7.42975473e-01 -4.54757541e-01 4.94534671e-01 4.82078791e-01 -6.85892522e-01 -5.42364657e-01 -1.05941033e+00 -1.02943647e+00 1.37008131e-01 -5.03636658e-01 2.77648985e-01 1.51910055e+00 8.47590268e-01 6.22770965e-01 -1.25703871e-01 3.52423005e-02 -3.60812247e-02 1.54370755e-01 3.84925932e-01 -1.04629970e+00 -3.90612930e-01 -4.95010197e-01 -1.14072524e-01 -1.14694953e+00 4.51786518e-01 -9.24634755e-01 1.50245830e-01 -1.38896942e+00 -8.68820474e-02 -6.16732180e-01 -3.84515673e-02 8.05730581e-01 -9.73586366e-02 1.22953638e-01 -1.86847612e-01 3.65245134e-01 -8.11459959e-01 4.84909892e-01 1.12437284e+00 -2.21149605e-02 -3.06171507e-01 -3.85923684e-01 -3.54853749e-01 7.12528169e-01 5.66659451e-01 -5.39702177e-01 -6.09931290e-01 -6.08632028e-01 2.22628444e-01 5.26194453e-01 2.25753754e-01 -4.66661364e-01 4.71945465e-01 -3.34337473e-01 -2.32785612e-01 -6.61712229e-01 3.40311974e-01 -3.73865873e-01 1.43209085e-01 3.48917767e-02 -9.30311143e-01 1.96624994e-01 1.42784342e-01 4.85358894e-01 -6.12058878e-01 -7.07706064e-02 3.06867838e-01 -9.35456622e-03 -7.36020386e-01 -1.87627960e-03 -4.41259772e-01 3.70321661e-01 1.13727677e+00 4.37517948e-02 -7.57508516e-01 -2.14900821e-01 -9.72480536e-01 1.87223688e-01 2.23656848e-01 5.02237916e-01 -1.43278569e-01 -1.30938017e+00 -6.54506624e-01 -5.34088790e-01 2.16303512e-01 5.35733998e-02 -3.38099226e-02 1.19194150e+00 -2.42997155e-01 8.38436961e-01 2.54951239e-01 -5.29193759e-01 -1.26537764e+00 5.23907721e-01 1.45647869e-01 -7.38559008e-01 -8.25561523e-01 8.45510483e-01 -1.07148245e-01 -2.35703975e-01 2.05302928e-02 -5.66020012e-01 -2.90650368e-01 5.16384363e-01 1.90217882e-01 -1.36132790e-02 3.55296195e-01 -5.58813870e-01 -6.00629270e-01 7.16285929e-02 -3.15749884e-01 -5.22354603e-01 1.39734209e+00 -3.18542749e-01 -4.53329474e-01 9.75879431e-01 1.17732835e+00 3.78222391e-02 -1.14920080e+00 -3.73164713e-01 7.64658928e-01 -1.66103780e-01 -1.77194029e-01 -5.65683246e-01 -2.35168621e-01 5.38389325e-01 -4.61152107e-01 8.63483071e-01 8.20205271e-01 4.02081639e-01 8.64685893e-01 2.48549998e-01 2.97114700e-01 -1.04365551e+00 1.89930215e-01 1.02998412e+00 8.24294329e-01 -9.39690948e-01 -4.99573909e-03 -5.95254898e-01 -4.17686671e-01 9.78321612e-01 5.00642240e-01 4.00601625e-02 4.34335887e-01 3.01416665e-01 -1.34940475e-01 -2.49773905e-01 -1.49691248e+00 -1.62453637e-01 4.65389490e-01 2.77295381e-01 1.06203043e+00 -1.05972290e-01 -6.56282842e-01 4.46633518e-01 -3.79530042e-01 -3.96179378e-01 4.62232292e-01 1.17810142e+00 -1.09956162e-02 -1.33733034e+00 -7.96278417e-02 1.77977875e-01 -6.50901914e-01 -2.58548319e-01 -3.75770062e-01 9.64609265e-01 1.42049575e-02 1.12256324e+00 5.54687560e-01 -5.63085452e-02 1.88779995e-01 4.22281958e-02 5.31332910e-01 -7.21642315e-01 -5.01508474e-01 -1.82416048e-02 1.20460224e+00 -5.31128764e-01 -7.51731932e-01 -1.18836927e+00 -1.31561494e+00 -2.73349524e-01 -1.29023986e-02 2.96328783e-01 -1.21399060e-01 1.42850959e+00 1.14727661e-01 8.07736993e-01 2.71338344e-01 -6.32514298e-01 -2.31799874e-02 -9.73408282e-01 -2.05148727e-01 5.15263021e-01 4.29271072e-01 -6.49454057e-01 -1.91819981e-01 4.87230241e-01]
[9.101238250732422, 9.205011367797852]
3d21f9e0-d75f-451d-b648-3f8f141c5d97
m2-net-multi-stages-specular-highlight
2207.09965
null
https://arxiv.org/abs/2207.09965v1
https://arxiv.org/pdf/2207.09965v1.pdf
M2-Net: Multi-stages Specular Highlight Detection and Removal in Multi-scenes
In this paper, we propose a novel uniformity framework for highlight detection and removal in multi-scenes, including synthetic images, face images, natural images, and text images. The framework consists of three main components, highlight feature extractor module, highlight coarse removal module, and highlight refine removal module. Firstly, the highlight feature extractor module can directly separate the highlight feature and non-highlight feature from the original highlight image. Then highlight removal image is obtained using a coarse highlight removal network. To further improve the highlight removal effect, the refined highlight removal image is finally obtained using refine highlight removal module based on contextual highlight attention mechanisms. Extensive experimental results in multiple scenes indicate that the proposed framework can obtain excellent visual effects of highlight removal and achieve state-of-the-art results in several quantitative evaluation metrics. Our algorithm is applied for the first time in video highlight removal with promising results.
['Xingjun Wang', 'Kun Hu', 'Zhaoyangfan Huang']
2022-07-20
null
null
null
null
['highlight-detection', 'highlight-removal']
['computer-vision', 'computer-vision']
[ 5.96418977e-01 -4.31000531e-01 2.62317151e-01 4.23691422e-02 -4.45403486e-01 -3.70637357e-01 4.86714184e-01 1.69452317e-02 -3.41257840e-01 6.73398793e-01 2.41076306e-01 3.20945680e-01 3.91356051e-02 -5.02704203e-01 -5.03011703e-01 -7.77014315e-01 1.52136475e-01 -7.66173601e-01 5.15907645e-01 -1.40496448e-01 5.88443875e-01 5.48321486e-01 -1.68026698e+00 6.10894620e-01 9.47842181e-01 7.12820053e-01 8.62557516e-02 5.89583755e-01 -6.83466420e-02 8.59618306e-01 -7.12494314e-01 -1.58059727e-02 3.05767834e-01 -3.26982349e-01 -9.39753056e-02 3.98902923e-01 5.00238597e-01 -7.84271181e-01 -1.43759072e-01 1.14611316e+00 7.39708662e-01 1.36889428e-01 4.51213330e-01 -1.33551013e+00 -6.73941612e-01 4.35737699e-01 -1.49059200e+00 5.89142919e-01 3.45539838e-01 1.13673486e-01 6.03132486e-01 -1.46565163e+00 5.70412219e-01 1.25721586e+00 3.01976830e-01 2.52838016e-01 -1.01968229e+00 -8.84307265e-01 1.82470724e-01 2.44597867e-01 -1.28044713e+00 -4.89054233e-01 1.23117256e+00 -1.66134015e-01 4.60194975e-01 4.80017662e-01 5.90921223e-01 6.60906911e-01 2.04087928e-01 9.98010516e-01 1.37969804e+00 -4.92716581e-01 -2.93792337e-01 1.92629948e-01 2.58604977e-02 7.00604618e-01 7.37022832e-02 4.57643829e-02 -9.93700445e-01 4.25724275e-02 5.39811194e-01 2.51127362e-01 -5.45162737e-01 -1.72678381e-02 -1.01762331e+00 2.38882124e-01 3.68717551e-01 -3.28353010e-02 -3.65709037e-01 -6.59187660e-02 3.95399153e-01 1.53450683e-01 5.77254593e-01 1.23151084e-02 -1.35088041e-01 1.91318899e-01 -1.32957494e+00 3.97730917e-01 2.00767338e-01 1.02544546e+00 5.68924129e-01 2.09842294e-01 -6.89653754e-01 1.05006218e+00 1.11226223e-01 5.19027531e-01 1.40888199e-01 -4.08888102e-01 3.13229144e-01 5.28631032e-01 2.53322572e-02 -1.25012481e+00 -4.32616711e-01 -4.99740005e-01 -7.38871694e-01 4.96086031e-01 -1.34600788e-01 -2.20989853e-01 -8.61094296e-01 1.23465800e+00 5.16893864e-01 3.72539043e-01 -1.39679208e-01 1.00097930e+00 1.21998000e+00 5.30474603e-01 1.28248513e-01 -4.13715392e-01 1.53060913e+00 -1.33511770e+00 -1.00789475e+00 3.69707905e-02 -8.32397714e-02 -1.34564769e+00 8.97707283e-01 4.36412334e-01 -1.45687330e+00 -7.92766869e-01 -1.24994397e+00 -3.69756043e-01 -3.71049047e-01 7.13904619e-01 4.01437253e-01 4.58601564e-01 -8.11288357e-01 2.42890477e-01 -2.59952903e-01 -2.83406794e-01 5.76253951e-01 -8.04092884e-02 -1.76733956e-01 -7.60279000e-02 -7.01683402e-01 4.44584459e-01 2.95789748e-01 3.14829975e-01 -7.86027431e-01 -9.59809184e-01 -6.44646645e-01 2.21066356e-01 7.97891617e-01 -4.91099328e-01 9.72592473e-01 -1.03470314e+00 -1.24833250e+00 7.35794425e-01 -2.88875759e-01 2.90758293e-02 6.63870156e-01 -6.68291628e-01 -6.43127501e-01 4.39062238e-01 -4.08289619e-02 4.88729328e-01 1.18221295e+00 -1.51507139e+00 -8.78860474e-01 2.90080626e-02 -9.43774208e-02 3.12179774e-01 -3.45762610e-01 6.13521457e-01 -9.07133818e-01 -1.13105381e+00 -8.19065645e-02 -5.86142361e-01 3.16243976e-01 1.33293152e-01 -7.62823224e-01 4.03738588e-01 1.45456982e+00 -7.33789325e-01 1.50411320e+00 -2.29165483e+00 -1.22035369e-01 1.40324458e-01 5.05978942e-01 3.96907091e-01 -2.82531053e-01 5.28544545e-01 -3.38807881e-01 2.34204680e-02 -2.29058433e-02 -2.35047087e-01 -3.82500678e-01 -5.97577751e-01 -4.60593462e-01 2.91827053e-01 8.13080549e-01 7.46731699e-01 -7.17249274e-01 -5.85215211e-01 3.97351503e-01 7.59693503e-01 -3.66309524e-01 -2.73407008e-02 2.64928013e-01 1.78007826e-01 -2.56999224e-01 1.02532780e+00 1.41340184e+00 1.02545537e-01 -1.48128942e-01 -5.77663302e-01 -4.63578075e-01 -6.13940120e-01 -1.29116261e+00 1.18575537e+00 4.96979244e-02 7.51615465e-01 3.64379466e-01 -1.99624658e-01 9.44118857e-01 -1.03769839e-01 2.90141135e-01 -7.14304268e-01 1.99398965e-01 -1.33882672e-01 -7.15832487e-02 -6.95945919e-01 9.82677400e-01 1.41289651e-01 4.00684834e-01 2.71241844e-01 -2.80544311e-01 1.94959775e-01 3.19742769e-01 4.14331198e-01 6.84683621e-01 3.40343982e-01 7.52543435e-02 -3.36655229e-01 9.77959991e-01 -4.26028728e-01 4.46542501e-01 3.59988272e-01 -4.42547917e-01 7.32790649e-01 5.99896967e-01 -4.69009355e-02 -8.37741137e-01 -9.38439548e-01 1.72120184e-01 1.40934527e+00 5.23239374e-01 -6.43060863e-01 -7.73399770e-01 -3.92962337e-01 1.38432056e-01 2.98178047e-01 -7.89926052e-01 -1.11526683e-01 -6.04933023e-01 -5.71413338e-01 3.04151416e-01 4.00981218e-01 8.95533979e-01 -1.26888573e+00 -6.38819873e-01 -1.21107884e-01 2.07010329e-01 -9.46671724e-01 -8.75822961e-01 -1.88564092e-01 -4.87018466e-01 -1.30257189e+00 -9.62678194e-01 -8.66038799e-01 8.48443389e-01 1.07238567e+00 4.38474536e-01 5.04308999e-01 -6.47804320e-01 1.26811773e-01 -2.96426147e-01 -5.02069354e-01 1.94607034e-01 -4.48817551e-01 -4.63266939e-01 2.86430120e-01 -4.88116108e-02 -4.74736035e-01 -9.22090828e-01 1.55465826e-01 -1.15631294e+00 3.85016203e-01 8.08392227e-01 8.57717991e-01 5.37315667e-01 2.37346515e-01 3.77286077e-01 -1.04960823e+00 8.74490261e-01 -1.13284208e-01 -2.33870640e-01 1.94045931e-01 -3.21277589e-01 -4.44201261e-01 5.96116304e-01 -4.84109014e-01 -1.52573729e+00 -2.75458753e-01 5.26282609e-01 -5.06635606e-01 1.10483564e-01 2.77819335e-01 -2.37347484e-01 -3.61968130e-01 3.00415218e-01 2.54334748e-01 -1.88354582e-01 -4.53929514e-01 5.18819749e-01 4.12870437e-01 7.06046343e-01 -3.27897280e-01 9.48316574e-01 8.36138248e-01 -5.09973476e-03 -9.75538552e-01 -5.39472520e-01 -3.81688416e-01 -7.12999940e-01 -6.37880206e-01 4.46591318e-01 -1.07514274e+00 -6.34459555e-01 7.62596190e-01 -9.83780146e-01 1.96331572e-02 1.97488159e-01 2.67833006e-02 6.35980666e-02 5.39031029e-01 -5.83808839e-01 -8.28889310e-01 -7.59429932e-01 -9.85608399e-01 1.32739866e+00 9.51903284e-01 3.76231790e-01 -4.57741261e-01 -5.62366009e-01 5.72298467e-02 5.92632651e-01 6.30378783e-01 6.52284503e-01 -6.52431175e-02 -8.56981277e-01 -1.17917039e-01 -8.64779770e-01 2.72522241e-01 2.27571830e-01 6.97720110e-01 -1.02627242e+00 -3.09043288e-01 -3.88778418e-01 2.02960864e-01 1.24448025e+00 3.09872180e-01 1.12713063e+00 1.23154305e-01 -4.79232520e-01 7.25705445e-01 1.62275338e+00 3.24031413e-01 6.57996595e-01 4.48609978e-01 8.43524456e-01 5.15546501e-01 9.14036274e-01 5.38130224e-01 -2.98069138e-02 4.18743700e-01 8.24708939e-02 -6.81096554e-01 -6.96576357e-01 5.40552847e-02 1.68637931e-01 5.04654706e-01 -3.97298746e-02 -1.04698285e-01 -5.16348124e-01 4.20865089e-01 -1.78917742e+00 -1.06900156e+00 -1.94346562e-01 1.89802384e+00 5.64456105e-01 3.01592112e-01 1.87158138e-01 6.68312982e-02 1.03363419e+00 4.92897898e-01 -4.78302687e-01 -2.92642713e-01 -6.27844095e-01 1.45544484e-01 2.63618946e-01 1.84099942e-01 -1.33831763e+00 1.08688867e+00 5.88242865e+00 1.02154040e+00 -1.14616024e+00 -1.11298315e-01 7.56868362e-01 -5.27166903e-01 7.64124244e-02 -1.48011088e-01 -5.81043482e-01 6.19917870e-01 6.80344105e-02 -4.75158602e-01 1.50416777e-01 6.92018211e-01 6.14352703e-01 -5.11511922e-01 -5.83424628e-01 1.25392306e+00 3.98268789e-01 -1.13339436e+00 1.96929783e-01 -3.38384449e-01 8.11309993e-01 -6.62659705e-01 4.43422586e-01 8.29961672e-02 -3.84248495e-01 -6.67901039e-01 8.10931027e-01 8.65300357e-01 6.97479188e-01 -1.24054217e+00 1.81332275e-01 -2.21174300e-01 -1.50482106e+00 -1.15471058e-01 -1.64113224e-01 1.53898835e-01 2.40705296e-01 7.10583150e-01 -2.51950413e-01 7.60171950e-01 6.36140883e-01 7.96697438e-01 -7.52645254e-01 1.57690704e+00 -3.36202651e-01 1.50317565e-01 1.07434645e-01 1.27150595e-01 2.02486897e-03 -1.70922905e-01 7.27909029e-01 1.70242381e+00 -1.12319015e-01 1.20419011e-01 1.51541382e-01 7.49000132e-01 -1.15907282e-01 3.74944955e-01 -2.76219457e-01 2.87185937e-01 5.04754543e-01 1.84428799e+00 -1.02245116e+00 -4.13118601e-01 -3.89595449e-01 1.14370632e+00 4.28787572e-03 6.60438776e-01 -1.18494785e+00 -8.02537441e-01 2.08023667e-01 -9.15795416e-02 4.93773550e-01 -5.94461635e-02 -1.70991540e-01 -7.92514563e-01 1.88565433e-01 -1.00464559e+00 1.56616643e-01 -1.30627286e+00 -7.84789026e-01 4.71219987e-01 -4.96424139e-02 -1.27977335e+00 5.64922094e-01 -5.27201176e-01 -1.09471393e+00 9.11177039e-01 -2.05770946e+00 -1.35471117e+00 -6.99839115e-01 5.53428411e-01 6.84614182e-01 1.93488095e-02 6.34390861e-02 3.84004921e-01 -9.64100242e-01 5.10314941e-01 2.28432640e-01 -6.45469353e-02 1.10551465e+00 -8.64569843e-01 -1.16661333e-01 1.20765483e+00 -4.18724656e-01 7.92346716e-01 5.00791788e-01 -8.73479784e-01 -1.32192552e+00 -1.10383940e+00 1.74373034e-02 9.67667848e-02 4.43752766e-01 -2.02492654e-01 -8.44397902e-01 1.76381931e-01 4.86404121e-01 -7.34084919e-02 4.74488467e-01 -2.93704867e-01 -4.34796721e-01 -8.97370279e-02 -1.01161730e+00 6.84934616e-01 8.59358013e-01 -5.57653248e-01 -1.91146865e-01 1.44399211e-01 5.39271235e-01 -6.80018067e-01 -3.31177801e-01 4.91670430e-01 8.38498414e-01 -1.17944169e+00 9.68998313e-01 -2.29668602e-01 6.64753020e-01 -6.94714308e-01 3.55329603e-01 -9.54842567e-01 -2.28580996e-01 -8.66135001e-01 -1.61189675e-01 1.68680668e+00 1.40276507e-01 -2.14343056e-01 3.82451773e-01 2.48864681e-01 -2.86253523e-02 -7.87847996e-01 -3.71767551e-01 -2.73035407e-01 -7.01609313e-01 6.88111484e-02 5.59796989e-01 8.97066712e-01 -3.44898731e-01 1.05407953e-01 -6.78495884e-01 1.44982085e-01 6.79216385e-01 3.27350438e-01 9.37057436e-01 -8.85833919e-01 1.16084449e-01 -5.46925604e-01 -5.27638867e-02 -7.39348054e-01 -2.70746350e-01 -3.23100537e-01 -1.09556988e-01 -1.34138680e+00 7.18811750e-01 2.66606718e-01 -5.61673462e-01 2.96216995e-01 -7.20698297e-01 3.44362438e-01 4.62682456e-01 -1.44256130e-02 -7.24130511e-01 5.65898359e-01 1.50941205e+00 -1.20738789e-01 -3.07420015e-01 -4.83822286e-01 -9.21164513e-01 8.12582374e-01 5.78583837e-01 -8.46510157e-02 -2.94954300e-01 -1.96368486e-01 -2.79351681e-01 -4.14415032e-01 4.55205858e-01 -1.08809340e+00 2.93494374e-01 -1.37836859e-01 9.09001887e-01 -9.67649937e-01 3.48822981e-01 -6.03538275e-01 -1.88084856e-01 2.93544739e-01 -1.74658492e-01 1.85794502e-01 8.58840883e-01 5.12408197e-01 -2.36045659e-01 1.60321191e-01 8.65826666e-01 2.17862949e-01 -9.70438421e-01 5.52687384e-02 -3.19389589e-02 -2.31958687e-01 1.13471484e+00 -4.22290027e-01 -1.12863815e+00 -1.87524617e-01 -4.40194279e-01 3.65123361e-01 2.82710701e-01 5.37058830e-01 9.76453006e-01 -1.20633125e+00 -8.02272260e-01 3.33097458e-01 4.83725742e-02 -3.79107028e-01 9.61079121e-01 1.09997797e+00 -5.38426697e-01 9.16414615e-03 -4.55933094e-01 -1.64402321e-01 -1.94270563e+00 8.54600012e-01 -7.77118206e-02 1.25990599e-01 -6.78901672e-01 9.39289987e-01 5.84288120e-01 5.23080051e-01 8.74683857e-02 -3.39620024e-01 -4.24613774e-01 2.34693974e-01 1.04096425e+00 6.43802345e-01 -2.85254061e-01 -9.56335187e-01 -3.02436799e-01 9.63850737e-01 -5.57280123e-01 9.73681137e-02 1.29174066e+00 -3.52165103e-01 -1.98772445e-01 8.92790500e-03 9.29508567e-01 7.01424539e-01 -1.13079941e+00 -7.11489841e-02 -3.37077349e-01 -8.51197898e-01 2.43673939e-02 -7.96425998e-01 -1.29965258e+00 6.93115592e-01 6.70788229e-01 1.72533363e-01 1.70252788e+00 -5.90260983e-01 5.32429218e-01 -1.38731375e-01 -2.81787872e-01 -1.22855914e+00 5.56147039e-01 -3.88814732e-02 1.13807034e+00 -1.13840890e+00 5.43801129e-01 -1.02248085e+00 -5.27618229e-01 1.24581587e+00 8.25831115e-01 -2.92237520e-01 5.57574570e-01 4.27524477e-01 4.81168665e-02 -3.19373280e-01 -5.45424819e-01 -4.02733266e-01 6.92332506e-01 4.41454262e-01 4.55509573e-01 -3.47343802e-01 -5.76390028e-01 4.51069951e-01 2.51434326e-01 -2.42330447e-01 6.78766131e-01 8.25804651e-01 -7.31393516e-01 -6.03332698e-01 -5.00507236e-01 2.71236986e-01 -6.28478467e-01 -4.19236034e-01 -5.88107944e-01 1.00897384e+00 8.40205327e-02 1.12593412e+00 -2.44619116e-01 -4.58882034e-01 5.29787838e-01 -1.67887434e-01 2.44044870e-01 -2.43566230e-01 -7.10453093e-01 7.43550837e-01 -6.54124143e-03 -6.71246290e-01 -5.25206447e-01 -4.82774287e-01 -1.15053034e+00 -1.33637518e-01 -5.92755854e-01 -1.88179791e-01 5.22332072e-01 3.51205468e-01 4.72366631e-01 1.12216818e+00 5.78820407e-01 -8.93423021e-01 2.44026825e-01 -9.27534282e-01 -7.96846032e-01 6.15735352e-01 5.41392982e-01 -7.06427217e-01 -2.67043859e-01 1.97817326e-01]
[10.872462272644043, -2.566481113433838]
4700f97d-d58b-420f-a049-155520892ed5
theoretical-analysis-of-an-xgboost-framework
2112.01566
null
https://arxiv.org/abs/2112.01566v1
https://arxiv.org/pdf/2112.01566v1.pdf
Theoretical Analysis of an XGBoost Framework for Product Cannibalization
This paper is an extension of our work where we presented a three-stage XGBoost algorithm for forecasting sales under product cannibalization scenario. Previously we developed the model based on our intuition and provided empirical evidence on its performance. In this study we would briefly go over the algorithm and then provide mathematical reasoning behind its working.
['Mohammad Bari', 'Gautham Bekal']
2021-12-02
null
null
null
null
['mathematical-reasoning']
['natural-language-processing']
[-1.49131984e-01 1.55529067e-01 -9.96259332e-01 -7.81912982e-01 1.83278114e-01 -5.96090436e-01 6.58787012e-01 4.12000865e-01 -1.64214805e-01 2.84354419e-01 -1.63373441e-01 -9.72134173e-01 -5.87531090e-01 -7.50334024e-01 -6.49517953e-01 -4.31619644e-01 -2.30834797e-01 5.89654624e-01 -1.69603348e-01 -5.05073309e-01 1.05559826e+00 4.06425178e-01 -1.50094223e+00 2.57876903e-01 7.60437787e-01 1.24191666e+00 -1.47038354e-02 8.23896408e-01 -2.47276537e-02 8.64294052e-01 -5.07201314e-01 -9.42385197e-01 8.67084324e-01 -3.81616205e-01 -1.02350318e+00 4.02790755e-01 -2.17709780e-01 -2.52180815e-01 2.15892479e-01 9.08142388e-01 -1.30243450e-01 1.93242222e-01 6.70776904e-01 -1.70010197e+00 -8.75570238e-01 8.20048928e-01 -6.82017863e-01 2.72017598e-01 4.95737582e-01 9.47123487e-03 1.07414138e+00 -1.66308463e-01 6.51990175e-01 1.15488362e+00 6.04415417e-01 6.27331585e-02 -9.37882841e-01 -3.92410785e-01 6.90762222e-01 1.76938757e-01 -9.03795898e-01 2.78829038e-01 8.67385089e-01 -2.46561587e-01 1.07964063e+00 4.70698357e-01 1.02325046e+00 5.07880628e-01 5.71848989e-01 1.27864432e+00 1.59642136e+00 -8.54140401e-01 4.78713870e-01 3.60870570e-01 8.47314477e-01 1.51608586e-01 6.44817114e-01 6.06637955e-01 -4.12525922e-01 2.25905739e-02 4.65583742e-01 -3.84382457e-02 5.67017078e-01 -2.44526163e-01 -1.33504018e-01 1.37635374e+00 5.23470819e-01 1.61837831e-01 -7.63123631e-01 -2.35157043e-01 1.90585494e-01 5.57721972e-01 8.95709336e-01 3.50846618e-01 -6.04632974e-01 1.85025364e-01 -9.36077237e-01 4.68955547e-01 9.68499362e-01 8.53660941e-01 7.23689869e-02 -3.88110757e-01 3.86928320e-01 2.08298504e-01 6.66745186e-01 -1.76612243e-01 6.68756485e-01 -5.16731262e-01 -1.42370865e-01 3.71616989e-01 3.62499624e-01 -8.53341222e-01 -4.04474139e-01 -4.36927289e-01 1.46148801e-01 4.47770387e-01 2.17169598e-01 1.43888285e-02 -1.11183977e+00 6.11441553e-01 1.21705055e-01 -5.09837389e-01 -4.78344820e-02 9.18760061e-01 3.64536077e-01 6.00944996e-01 5.17518580e-01 -6.48695648e-01 1.13488936e+00 -1.34183681e+00 -8.81607533e-01 1.97917491e-01 5.15641928e-01 -1.04073083e+00 5.95808268e-01 1.11524379e+00 -1.08389878e+00 -5.40176570e-01 -1.25261188e+00 2.70632565e-01 -5.72669744e-01 -4.81078535e-01 1.74300098e+00 1.18368042e+00 -7.89285541e-01 9.49310303e-01 -7.95557618e-01 -6.91605330e-01 3.73653769e-02 1.87746719e-01 4.91889268e-01 1.81545779e-01 -8.08501780e-01 1.22571135e+00 5.91297984e-01 3.58478993e-01 -3.73949677e-01 -1.04482267e-02 -3.98669124e-01 -4.32241052e-01 3.15175414e-01 -3.75136584e-01 1.70143080e+00 -1.18344140e+00 -1.60653234e+00 7.65956759e-01 8.26857984e-02 -6.78030133e-01 4.37180758e-01 -2.06239730e-01 -7.55457342e-01 -4.03505534e-01 -1.66962519e-01 3.80979091e-01 3.57357562e-01 -1.53501487e+00 -1.04036820e+00 -6.81203604e-01 2.65083373e-01 -4.45408411e-02 4.18722153e-01 5.14959812e-01 -5.40378271e-03 -7.01740205e-01 3.69627923e-01 -8.52364302e-01 -4.20115411e-01 -9.08312082e-01 -2.06523821e-01 -4.80982393e-01 3.33349779e-02 -3.65770459e-01 1.38489223e+00 -1.51107562e+00 -6.17031336e-01 3.64926517e-01 -2.98680663e-01 5.31316847e-02 2.78677076e-01 1.04023147e+00 -2.80973196e-01 3.10186893e-01 3.61303449e-01 1.23341054e-01 3.32866520e-01 1.05902039e-01 -3.38216275e-01 4.79936600e-01 -6.63562641e-02 9.62720871e-01 -9.30486202e-01 -1.85678601e-01 5.45215458e-02 -2.14128911e-01 2.29950510e-02 -2.42914736e-01 -1.71246052e-01 -5.10104477e-01 -4.66817170e-01 1.33667779e+00 1.04372168e+00 1.57156274e-01 2.30594382e-01 -1.27799585e-01 -5.26968122e-01 3.17634553e-01 -9.06369627e-01 1.03538227e+00 -2.63024028e-02 1.59925625e-01 -7.72085786e-02 -1.22114599e+00 7.89582014e-01 -1.58773527e-01 3.61132622e-01 -7.25330114e-01 6.95129335e-01 2.65462250e-01 -3.07294191e-03 -4.31935370e-01 6.06523812e-01 -6.50381207e-01 7.43836388e-02 6.68314159e-01 -2.20221832e-01 -7.88170006e-03 1.99260712e-01 7.01418240e-03 5.25068581e-01 5.84990025e-01 6.03034317e-01 -6.59239650e-01 3.88693586e-02 8.28737736e-01 1.74027354e-01 8.93140256e-01 -4.11386222e-01 -2.48396486e-01 1.87083229e-01 -9.39082801e-01 -5.60538888e-01 -8.32520306e-01 -3.40788513e-01 1.33660793e+00 4.10529912e-01 -3.48255217e-01 -5.27713418e-01 -8.53976846e-01 3.27116787e-01 1.27451730e+00 -7.28295267e-01 1.80417031e-01 -2.67793834e-01 -1.15387344e+00 -2.01445296e-01 5.36977530e-01 2.91749518e-02 -8.91623497e-01 -6.83121085e-01 1.92465574e-01 6.03754878e-01 -1.42174512e-01 1.43069237e-01 5.35259604e-01 -1.23809636e+00 -1.20766675e+00 -2.23281354e-01 -6.31473660e-01 6.99636996e-01 4.80436444e-01 1.02090931e+00 1.57096013e-01 -2.56549776e-01 -3.03818621e-02 -9.42878842e-01 -1.14974320e+00 -3.29519629e-01 -1.15264930e-01 -4.63583767e-02 -5.22726357e-01 1.20942461e+00 -8.97627696e-03 -7.03196108e-01 5.00068486e-01 -8.52313042e-01 -3.96665782e-01 5.94071925e-01 5.84132135e-01 2.42517114e-01 7.38574326e-01 3.05716008e-01 -1.24911952e+00 1.09972930e+00 -4.89422292e-01 -8.64338219e-01 3.33826512e-01 -1.74955845e+00 -5.11930995e-02 2.60417387e-02 -2.05930740e-01 -1.17325485e+00 3.13949674e-01 -8.39053467e-02 3.65237027e-01 9.32711437e-02 6.67784333e-01 3.71392012e-01 -2.24049576e-02 4.07880634e-01 -3.26833278e-01 8.14078450e-02 -6.92449927e-01 6.64005518e-01 7.95128524e-01 2.01034263e-01 -4.01881546e-01 4.30250287e-01 3.38112384e-01 -2.40444049e-01 -2.46338286e-02 -6.56032562e-01 -5.92957854e-01 -6.79705083e-01 -4.56175864e-01 2.88994998e-01 -3.07957321e-01 -1.26512277e+00 2.35487521e-01 -6.54700577e-01 -1.56443566e-01 -3.34640890e-02 6.35995984e-01 -7.81325400e-01 -5.16845174e-02 -7.51886487e-01 -1.46384323e+00 -2.83161074e-01 -7.36572742e-01 4.50957358e-01 1.34934008e-01 -1.94002360e-01 -1.17697239e+00 9.95008126e-02 5.02570212e-01 2.99315333e-01 1.56127989e-01 5.19345105e-01 -7.01007307e-01 -2.52791584e-01 -4.34359372e-01 2.26063132e-01 5.53757101e-02 -7.91124329e-02 3.41283321e-01 -7.38294065e-01 -2.29512215e-01 5.60714900e-01 -6.96278214e-02 9.16771829e-01 7.60212243e-01 5.56970358e-01 -3.78608316e-01 -8.10545802e-01 9.90348607e-02 1.77728522e+00 8.55777442e-01 3.99341881e-01 1.15362239e+00 -1.19772129e-01 1.33027554e+00 1.63062191e+00 3.10128093e-01 3.33688825e-01 5.10254323e-01 5.56498230e-01 -3.82289253e-02 4.23770010e-01 -4.14330482e-01 9.58356932e-02 3.65552396e-01 -3.18523407e-01 -6.52384043e-01 -4.82391477e-01 3.70458335e-01 -2.04971266e+00 -5.71927011e-01 -2.97377467e-01 1.75509560e+00 3.15262735e-01 5.85513353e-01 1.05050802e+00 1.67035714e-01 6.63559258e-01 -3.39940071e-01 -3.98512751e-01 -1.24802017e+00 2.73541272e-01 1.23790011e-01 1.26968718e+00 5.80115020e-01 -8.52780104e-01 8.92494857e-01 8.86985111e+00 5.17845869e-01 -8.17766249e-01 -2.19500959e-01 9.24348176e-01 -1.84441004e-02 -2.03360736e-01 3.52928102e-01 -7.73978710e-01 2.84202039e-01 9.00672674e-01 -2.92324126e-01 2.01237768e-01 1.03107941e+00 2.36094534e-01 -4.93202478e-01 -1.00745416e+00 2.79005826e-01 1.15745082e-01 -8.03748190e-01 -1.63660720e-01 1.78245515e-01 4.63505954e-01 -4.07771468e-01 4.53522727e-02 2.34153748e-01 1.01063204e+00 -8.50149095e-01 8.43661666e-01 1.34672865e-01 -4.52457309e-01 -8.95712733e-01 8.67503226e-01 2.22884461e-01 -4.46726292e-01 -4.05426741e-01 -3.85785788e-01 -7.97531903e-01 5.36152780e-01 3.52529407e-01 -9.55947220e-01 9.05257940e-01 5.38497269e-01 2.76005983e-01 -4.48773891e-01 1.04787099e+00 -1.31381169e-01 6.34574831e-01 -1.35351956e-01 -3.76968443e-01 5.99093676e-01 -4.30349410e-01 -1.13887392e-01 9.59652424e-01 1.05223931e-01 2.52863884e-01 3.49673927e-02 5.62505126e-01 7.60125339e-01 3.20632070e-01 -2.98421234e-01 -8.49364847e-02 1.76048979e-01 1.02695739e+00 -1.45211661e+00 -1.81146726e-01 -4.57747847e-01 6.17505968e-01 -2.61100441e-01 -7.90007934e-02 -5.72567642e-01 -2.68762708e-01 2.51844347e-01 3.64883304e-01 3.87353331e-01 1.56285673e-01 -9.02067721e-01 -5.24343610e-01 -2.16162518e-01 -5.19503176e-01 7.26916373e-01 -5.18276334e-01 -1.54550946e+00 3.57043386e-01 5.04248917e-01 -9.74347949e-01 -1.61426902e-01 -1.06940365e+00 -6.44368112e-01 5.15682101e-01 -1.20734453e+00 -1.20105040e+00 2.84442902e-01 -1.92006767e-01 6.82743609e-01 1.38861989e-03 3.37020636e-01 -2.11609557e-01 -5.26049316e-01 3.93474042e-01 2.27017283e-01 -5.94706953e-01 2.10188970e-01 -1.15271759e+00 5.64589024e-01 6.54574335e-01 -1.34147257e-01 1.11653030e+00 1.35776448e+00 -1.18029690e+00 -1.28915966e+00 -3.52770239e-01 9.68665183e-01 -3.19875002e-01 9.68110263e-01 -1.44594312e-01 -2.17632681e-01 1.03138316e+00 7.60714054e-01 -9.24872100e-01 7.41097212e-01 6.43064380e-01 2.10507110e-01 -1.99928477e-01 -1.33642030e+00 1.17550164e-01 7.74646878e-01 5.00844896e-01 -7.86460519e-01 5.15337229e-01 2.75842875e-01 -1.05882585e-01 -9.52255905e-01 2.65648335e-01 9.70489979e-01 -1.05444455e+00 6.80433273e-01 -8.94613802e-01 1.61140859e-01 8.93365219e-02 1.22808442e-01 -1.14883602e+00 -6.49213314e-01 -6.58427954e-01 3.25617969e-01 9.32746291e-01 6.05446935e-01 -6.83940768e-01 8.83444607e-01 7.29454696e-01 -1.30073903e-02 -8.26149881e-01 -2.63228416e-01 -1.12854970e+00 1.52197838e-01 -4.92164880e-01 8.10425699e-01 7.78957844e-01 6.39099360e-01 1.99807554e-01 -1.00023776e-01 -3.19331288e-01 5.62604427e-01 6.22940779e-01 3.97694975e-01 -1.15759945e+00 -3.80419612e-01 -7.73664594e-01 -2.74026811e-01 -9.70323920e-01 -1.77644759e-01 -6.95629656e-01 -5.55700846e-02 -1.23177540e+00 2.22385213e-01 -3.05218250e-01 -5.02672613e-01 1.67849660e-01 4.32969294e-02 1.44068018e-01 2.97770262e-01 2.48932898e-01 -3.11999053e-01 -3.07089329e-01 9.19935226e-01 5.09560183e-02 -4.45952863e-01 5.21368921e-01 -1.36326957e+00 6.11826062e-01 1.09357274e+00 -4.08536851e-01 -5.30329704e-01 2.28997935e-02 6.28376126e-01 -2.13539433e-02 4.79733273e-02 6.80265129e-02 -5.10130897e-02 -4.03447688e-01 3.17901582e-01 -1.06753743e+00 -1.94360510e-01 -9.49920177e-01 3.91894788e-01 7.43927836e-01 -5.28328657e-01 4.05495077e-01 1.27208591e-01 5.61681390e-01 -2.01031659e-02 -8.27272654e-01 2.09479749e-01 -2.05040529e-01 -5.95252514e-01 -2.97171474e-01 -3.86262655e-01 -1.19476938e+00 1.40979624e+00 -4.90142316e-01 -3.01825166e-01 -2.61420250e-01 -8.24292243e-01 3.77542168e-01 6.50306880e-01 5.44317484e-01 1.46237195e-01 -1.17962432e+00 -2.20758691e-01 -1.23434350e-01 6.40150905e-02 -9.43038166e-01 -3.09227318e-01 6.94578350e-01 -9.74753320e-01 9.03385401e-01 -5.87693274e-01 9.58156288e-02 -1.09840596e+00 1.41171741e+00 -2.72048384e-01 -2.10685447e-01 -1.18522309e-01 8.38773429e-01 -9.04439464e-02 5.95805719e-02 -3.90727036e-02 -1.87300295e-01 -3.78075182e-01 5.13776504e-02 2.94013828e-01 8.30547690e-01 8.87594521e-02 -5.44402421e-01 -4.34624046e-01 2.83673674e-01 -5.27911425e-01 -1.15206510e-01 1.34462464e+00 -2.99248308e-01 -8.16458985e-02 5.73019683e-01 7.47631431e-01 -4.68865514e-01 -8.54171753e-01 3.24791014e-01 3.88962686e-01 -8.04793060e-01 4.39256072e-01 -1.24851120e+00 -9.17955339e-01 1.03301279e-01 6.71912611e-01 1.02387393e+00 1.37632847e+00 8.23696628e-02 5.46879709e-01 -6.40996099e-02 3.58348727e-01 -1.65424967e+00 -5.83424866e-01 -3.98348570e-01 5.60521841e-01 -1.22842288e+00 6.57897055e-01 -7.52551556e-01 -8.86917651e-01 1.17022681e+00 2.49458402e-01 -3.88493210e-01 9.39477384e-01 -9.24020112e-02 1.04496144e-01 -8.03656101e-01 -6.27970874e-01 -2.84290075e-01 -4.49823821e-03 5.46792328e-01 6.78236485e-01 3.98247033e-01 -1.37031174e+00 7.41754353e-01 -5.58177233e-01 1.82841584e-01 2.44647190e-01 1.42656147e+00 -5.76471210e-01 -1.43125200e+00 -3.05002958e-01 3.56525660e-01 -7.53225863e-01 1.37984693e-01 -8.91085982e-01 1.31303859e+00 9.28208455e-02 1.62650526e+00 -1.10308034e-03 -5.49110055e-01 4.67824727e-01 -1.05562329e-01 7.11564004e-01 -8.14436302e-02 -8.27745080e-01 5.61759472e-01 2.70447105e-01 -2.94566333e-01 -7.69815207e-01 -1.02595818e+00 -7.79152274e-01 -6.63370967e-01 -8.96976471e-01 3.15656066e-01 1.26580334e+00 6.11873388e-01 -1.07392006e-01 6.95081279e-02 1.05349505e+00 -5.38150370e-01 -6.60085917e-01 -1.07737374e+00 -1.17087281e+00 2.66724378e-01 -1.41113728e-01 -8.14176261e-01 -3.08219850e-01 -2.15633828e-02]
[9.402813911437988, 5.818057060241699]
3f6d8a0c-3632-427b-967a-ef2a1b36e06a
lung-nodule-detection-and-classification-from
null
null
https://doi.org/10.1016/j.jksuci.2020.03.013
https://www.sciencedirect.com/science/article/pii/S1319157820303335
Lung nodule detection and classification from Thorax CT-scan using RetinaNet with transfer learning
Lung malignancy is one of the most common causes of death in the world caused by malignant lung nodules which commonly diagnosed radiologically by radiologists. Unfortunately, the continuous flow of medical images in hospitals drives radiologists to prioritize quantity over quality. This work condition allows misinterpretation especially on ambiguous anatomical structures that resemble lung nodule for example enlarged lymph nodes and resulting in decreasing sensitivity and accuracy of malignant lung nodule detections and late diagnosis proven to be fatal to patients. To address the problem, this paper proposed a novel lung nodule detection and classification model using one stage detector called as “I3DR-Net.” The model was formed by combining pre-trained natural images weight of Inflated 3D ConvNet (I3D) backbone with feature pyramid network to multi-scale 3D Thorax Computed tomography scan (CT-scan) dataset. I3DR-Net able to produce remarkable results on lung nodule texture detection task with mAP 49.61% and 22.86%, and area under curve (AUC) 81.84% and 70.36% for public and private dataset. Additionally, I3DR-Net successfully outperform previous state-of-the-art Retina U-Net and U-FRCNN + mean average precision (mAP) by 7.9% and 7.2% (57.71% VS 49.8% VS 50.5%) for malignant nodule detection and classification task.
['Tjeng Wawan Cenggoro', 'Suryadiputra Liawatimena', 'Ivan William Harsono']
2020-04-08
null
null
null
journal-of-king-saud-university-computer-and-2
['lung-nodule-3d-detection', 'lung-nodule-3d-classification', 'lung-nodule-detection', 'lung-nodule-classification']
['computer-vision', 'computer-vision', 'medical', 'medical']
[ 1.39395013e-01 3.34155500e-01 -5.90442605e-02 2.44915560e-01 -6.96459711e-01 -3.48167300e-01 3.96940708e-01 -2.24496111e-01 -4.30081725e-01 4.66616005e-01 9.58525091e-02 -4.91225183e-01 -1.80830657e-01 -8.52733195e-01 -4.10086244e-01 -6.61770284e-01 3.34644355e-02 6.29420340e-01 8.42553496e-01 3.23517531e-01 -1.31230444e-01 7.16715813e-01 -1.01361990e+00 4.69414860e-01 4.33320791e-01 1.01856148e+00 5.03675997e-01 8.43186975e-01 6.98141158e-02 9.20239389e-01 -1.41946152e-01 -3.32758464e-02 3.80416930e-01 -3.12903702e-01 -7.30324745e-01 1.33336172e-01 2.38769203e-01 -3.94980639e-01 -4.31650341e-01 8.64264011e-01 7.00567722e-01 -4.92809534e-01 1.18158770e+00 -6.64513707e-01 -6.61257029e-01 4.09143358e-01 -8.01135242e-01 7.62372017e-01 -1.90162197e-01 2.67255723e-01 4.53875035e-01 -9.97828841e-01 4.65943068e-01 9.08897400e-01 8.72997522e-01 6.86946750e-01 -5.90968251e-01 -4.01466250e-01 -6.65843904e-01 -4.38678004e-02 -1.28117418e+00 3.25253248e-01 -1.32542685e-01 -4.72949445e-01 9.28145111e-01 6.46237850e-01 6.91848934e-01 7.54781425e-01 8.32680702e-01 5.24295509e-01 1.14498246e+00 -1.27015263e-01 -2.75042683e-01 2.08147913e-01 -2.29590937e-01 1.18779159e+00 9.22458470e-01 1.60386443e-01 3.19389641e-01 -2.07260698e-01 1.38867033e+00 2.82252967e-01 -2.54317939e-01 -1.51444241e-01 -1.60024488e+00 6.77766502e-01 8.14708531e-01 7.67752767e-01 -4.96746778e-01 2.12386385e-01 3.92880529e-01 -1.06473967e-01 -5.14219254e-02 2.15718448e-01 -2.15095758e-01 2.92948395e-01 -5.42471766e-01 -8.72826427e-02 5.23487806e-01 6.40586495e-01 -9.06145871e-02 3.79597247e-02 -8.29990089e-01 5.87920010e-01 3.04892242e-01 8.13703477e-01 1.19556844e+00 -2.65361011e-01 -3.58828604e-02 7.10976779e-01 -1.16157934e-01 -6.75912023e-01 -7.52236307e-01 -1.00251675e+00 -1.39051449e+00 -8.14842880e-02 3.24565083e-01 1.45481840e-01 -1.45773292e+00 8.98371100e-01 1.87696949e-01 2.92040445e-02 9.02046263e-02 9.86502528e-01 1.16048610e+00 2.61903346e-01 2.50146925e-01 2.28855796e-02 1.72367799e+00 -1.02187037e+00 -2.88654894e-01 1.52763933e-01 7.69871056e-01 -7.81482399e-01 9.57581103e-01 -7.62995705e-02 -9.42981064e-01 -5.01084924e-01 -6.91744089e-01 4.22261387e-01 3.95920277e-02 6.45223320e-01 2.41470665e-01 6.49426103e-01 -1.13264036e+00 3.16208333e-01 -8.69854867e-01 -7.45210707e-01 6.26975417e-01 6.15052283e-01 -2.30798453e-01 -1.63711812e-02 -8.34309816e-01 1.05170131e+00 4.47285354e-01 -1.32027298e-01 -1.02130163e+00 -8.27589512e-01 -1.32971080e-02 -1.56864207e-02 4.89547879e-01 -1.10165584e+00 1.52335882e+00 -5.71889699e-01 -8.38820636e-01 1.06535339e+00 4.03016299e-01 -5.80128789e-01 9.06639457e-01 5.25214449e-02 -3.55596721e-01 3.14614594e-01 1.06957145e-01 6.62013829e-01 7.35914290e-01 -6.44426644e-01 -9.32973027e-01 -2.68390715e-01 -6.31728768e-01 2.37569124e-01 2.41407426e-03 -2.68514305e-01 -2.65330493e-01 -5.17418087e-01 2.26573825e-01 -9.78130460e-01 -3.74335051e-01 1.86667308e-01 -2.60430336e-01 -3.80704105e-01 1.15555549e+00 -5.96457303e-01 1.13294959e+00 -1.74118817e+00 -5.35618007e-01 2.28468999e-01 6.10115349e-01 6.75446868e-01 2.59278536e-01 -3.21370512e-01 -2.78939903e-02 2.81593084e-01 -1.40468374e-01 7.15924323e-01 -3.87644619e-01 1.28254853e-02 4.27307487e-01 4.57760125e-01 3.04812551e-01 1.36216545e+00 -6.71913207e-01 -9.29411888e-01 4.02980387e-01 6.02769971e-01 -2.70725071e-01 7.43262470e-03 2.65394062e-01 2.72495896e-01 -8.19011569e-01 9.59539175e-01 5.83806932e-01 -7.09114254e-01 -1.59582302e-01 -3.59695882e-01 2.79256869e-02 -3.31528336e-01 -8.19708407e-01 1.01416016e+00 -3.24964672e-01 3.18158776e-01 -2.83826977e-01 -5.02451777e-01 6.62638307e-01 7.77003884e-01 6.33098722e-01 -6.12979829e-01 4.66109753e-01 6.33405745e-01 4.74456042e-01 -8.72557104e-01 -3.46595317e-01 -2.84118772e-01 2.59465516e-01 2.46888489e-01 -1.86270058e-01 -1.35896429e-01 -1.14157856e-01 1.57200713e-02 1.56224692e+00 -4.56551373e-01 8.28533590e-01 -4.21847284e-01 9.16110396e-01 1.52395904e-01 -1.81154460e-02 8.87356222e-01 -5.81370413e-01 7.32385278e-01 3.02646846e-01 -6.11316204e-01 -1.03540337e+00 -1.20634878e+00 -4.70299929e-01 5.03208876e-01 -3.30679744e-01 5.43685973e-01 -5.48294425e-01 -1.06652343e+00 1.50258109e-01 1.32814184e-01 -8.53939354e-01 4.44037020e-02 -6.24680698e-01 -8.88300300e-01 6.44069493e-01 6.50603116e-01 8.83625567e-01 -1.14258492e+00 -9.97067750e-01 1.45013273e-01 1.67389661e-01 -7.68179953e-01 -3.17741394e-01 3.36673200e-01 -1.08593404e+00 -1.39141893e+00 -1.27361739e+00 -9.17795897e-01 7.44444311e-01 3.85169148e-01 1.21450388e+00 2.58976281e-01 -1.02353585e+00 1.88225627e-01 -7.79473037e-02 -3.52527946e-01 -5.69241703e-01 1.20750591e-01 -2.33580425e-01 -4.12605137e-01 3.52767110e-01 -1.08111061e-01 -8.47100079e-01 3.53239000e-01 -9.54191208e-01 -9.37434286e-03 1.72357821e+00 8.64251077e-01 7.45697975e-01 -8.51983577e-02 5.45807600e-01 -1.08339226e+00 4.87413734e-01 -5.67342997e-01 -3.56526881e-01 2.61828095e-01 -5.82299352e-01 -1.93611935e-01 4.54954863e-01 -3.24642450e-01 -7.62280166e-01 1.87302649e-01 8.30521062e-02 -5.51560760e-01 -2.27614745e-01 2.36116033e-02 6.23732865e-01 -2.22495779e-01 9.87217605e-01 2.33384863e-01 -6.33243248e-02 -1.65821742e-02 -2.62641460e-01 8.49507153e-01 6.92323744e-01 2.73402274e-01 9.57194626e-01 4.16403532e-01 5.00989020e-01 -5.83135009e-01 -9.43446457e-01 -7.99369931e-01 -6.54645324e-01 -3.02231044e-01 1.20930886e+00 -9.34023917e-01 -3.45908493e-01 1.32933781e-01 -6.80016637e-01 6.89376965e-02 -3.14738572e-01 7.11305439e-01 -2.19196364e-01 2.70206153e-01 -4.75424677e-01 -4.35653895e-01 -8.34612131e-01 -9.61452961e-01 8.84744942e-01 2.76724935e-01 -7.61394724e-02 -7.82529950e-01 -2.63762772e-01 1.98913872e-01 1.05663753e+00 3.01506013e-01 8.95032942e-01 -8.33778977e-01 -6.26631439e-01 -5.17880142e-01 -9.03473198e-01 1.22578457e-01 4.44846362e-01 -1.25482962e-01 -8.96454096e-01 -1.14079416e-01 4.24529612e-02 -1.61096066e-01 8.40187609e-01 7.34914362e-01 1.29414880e+00 5.67970015e-02 -6.44551873e-01 3.15834969e-01 1.55451870e+00 2.97731787e-01 3.11880678e-01 1.39282599e-01 6.44434690e-01 -8.65334123e-02 2.35090777e-01 2.12297484e-01 -3.36951584e-01 -3.68194655e-02 6.60958171e-01 -1.84268966e-01 -7.16296256e-01 4.97700050e-02 -1.81979910e-01 6.23622835e-01 -1.39143512e-01 -3.19732100e-01 -1.23170948e+00 5.73610783e-01 -1.36987674e+00 -7.77787507e-01 -5.05211055e-01 1.79237747e+00 3.40327114e-01 2.97317177e-01 4.29696068e-02 8.35646875e-03 8.61350060e-01 -2.68940121e-01 -4.93424386e-01 5.78423813e-02 2.11066097e-01 2.81556726e-01 9.68539715e-01 -1.89924285e-01 -1.27489424e+00 3.60683769e-01 5.38168859e+00 9.71306622e-01 -1.26621008e+00 2.55349904e-01 8.78110647e-01 2.23819166e-01 5.95892109e-02 -5.97234547e-01 -4.93339300e-01 1.01844750e-01 5.74994862e-01 -1.27948269e-01 -3.15182000e-01 8.81732166e-01 5.06163724e-02 -3.24974775e-01 -8.08849394e-01 9.46032524e-01 -1.41693443e-01 -1.27638626e+00 3.07343215e-01 8.23457316e-02 7.92378962e-01 3.72026682e-01 2.74886876e-01 2.84429312e-01 -6.20485991e-02 -1.45807672e+00 3.63886133e-02 4.00129020e-01 1.14008808e+00 -4.35239285e-01 1.37346530e+00 4.69276816e-01 -1.23924601e+00 1.68993454e-02 -5.28613150e-01 5.00870466e-01 -4.24885601e-01 4.39282238e-01 -1.67548335e+00 3.41328770e-01 8.04308951e-01 3.64128768e-01 -8.62548172e-01 1.33322310e+00 2.40198985e-01 6.95153356e-01 -4.64366078e-01 -5.36653519e-01 6.81724489e-01 2.95890778e-01 3.46745461e-01 1.28324795e+00 6.69294417e-01 1.88643232e-01 -3.06139905e-02 6.29740834e-01 -1.35191903e-01 2.64885306e-01 -7.12375700e-01 9.99056697e-02 5.55132665e-02 1.54145980e+00 -1.19612503e+00 -4.87823159e-01 -4.28861380e-01 6.07712984e-01 -2.68236130e-01 -2.86935031e-01 -1.06697130e+00 8.93651769e-02 -5.20939112e-01 5.39555371e-01 3.71053815e-01 5.30852020e-01 -7.16751367e-02 -6.76519752e-01 -2.91213185e-01 -4.08158869e-01 4.53210562e-01 -6.14354908e-01 -1.28891182e+00 8.16696882e-01 -2.59453654e-01 -1.43027818e+00 1.14576682e-01 -9.13450837e-01 -6.57449305e-01 6.57024384e-01 -1.50229824e+00 -1.06783354e+00 -7.82702923e-01 4.45031583e-01 5.82361102e-01 -4.27231908e-01 7.04552889e-01 3.64354216e-02 -1.88075632e-01 3.53811800e-01 2.63404138e-02 1.22492194e-01 6.08858883e-01 -1.35371375e+00 6.37680441e-02 2.54667372e-01 -4.56244379e-01 -2.62682676e-01 6.47079423e-02 -7.49389529e-01 -1.22199464e+00 -1.61739123e+00 6.58075333e-01 -4.08995509e-01 5.44103563e-01 3.82099152e-01 -7.74589002e-01 3.11885327e-01 -9.96025465e-03 6.59767151e-01 3.14396441e-01 -1.00458550e+00 1.12747595e-01 3.03185880e-01 -1.56216037e+00 4.80453432e-01 8.42258751e-01 -1.13690021e-02 -3.59975666e-01 5.26523829e-01 3.38501364e-01 -4.83716667e-01 -8.76017153e-01 8.55173767e-01 5.67785680e-01 -1.01042998e+00 9.07501757e-01 -2.27372050e-01 3.43595296e-01 -2.48092651e-01 3.53591926e-02 -5.59518337e-01 -9.06162202e-01 9.91339907e-02 5.48974052e-02 9.66226533e-02 6.09215856e-01 -4.09588188e-01 1.05156791e+00 -6.18768763e-03 -1.52664974e-01 -1.20019293e+00 -9.77149069e-01 -4.68699783e-01 1.28814191e-01 6.77605020e-03 3.99137102e-02 6.35105848e-01 -8.32940698e-01 7.24111721e-02 1.85731694e-01 -4.30023465e-05 5.02291501e-01 -3.15795630e-01 2.16455325e-01 -1.28903246e+00 -1.57815486e-01 -6.48216486e-01 -5.01584351e-01 -4.92642790e-01 -7.36766279e-01 -1.06394446e+00 -3.14036757e-01 -1.83239746e+00 7.02943742e-01 -1.94101006e-01 -5.38755536e-01 3.42688531e-01 -1.71075299e-01 4.18216646e-01 -8.78253579e-02 4.86410469e-01 -3.97739291e-01 -5.36449030e-02 2.08796430e+00 -1.94974076e-02 8.70617256e-02 3.82383615e-01 -3.42614770e-01 7.86377192e-01 9.54410136e-01 -5.48778236e-01 -2.92162299e-01 -6.65079579e-02 -1.81217253e-01 2.91480780e-01 5.68062782e-01 -1.50518239e+00 1.14889592e-01 4.37990809e-03 9.37570870e-01 -1.00463998e+00 -1.61277741e-01 -1.09832001e+00 1.94323242e-01 1.46533573e+00 -6.62663206e-02 1.24287484e-02 9.38444212e-03 7.91269064e-01 -7.39053264e-02 -1.94256961e-01 1.10143125e+00 -6.28090441e-01 -5.69034398e-01 4.57619637e-01 -5.82012534e-01 -1.43066898e-01 1.52321231e+00 -4.26831305e-01 -1.39921308e-01 5.53524494e-02 -6.99423134e-01 -9.38457809e-03 -8.63247216e-02 1.32331818e-01 6.13136709e-01 -1.31426716e+00 -9.63059485e-01 1.29464433e-01 3.81906331e-02 2.75289625e-01 2.54326493e-01 1.37185812e+00 -1.10116112e+00 9.64446068e-01 -1.26689181e-01 -9.66125250e-01 -1.33594871e+00 2.50826836e-01 9.12507176e-01 -7.74390638e-01 -8.07403445e-01 9.67343628e-01 3.40747654e-01 -1.92409575e-01 2.38613352e-01 -8.32833767e-01 -1.81450471e-01 -3.05268019e-01 2.18643639e-02 3.89250219e-01 3.31168294e-01 -4.26524431e-01 -4.39197809e-01 5.84355175e-01 -3.05767059e-01 4.32415485e-01 9.62387264e-01 3.55782002e-01 9.70460847e-02 9.56892967e-02 1.02668333e+00 -3.46186876e-01 -6.79744244e-01 -1.45906195e-01 5.39162755e-02 -9.51686800e-02 1.10568374e-01 -1.28827488e+00 -1.06075203e+00 5.70082307e-01 1.25352716e+00 1.88147843e-01 9.92617190e-01 2.50145435e-01 8.17176461e-01 6.05560124e-01 -9.24059972e-02 -5.78656852e-01 2.68621683e-01 3.77179861e-01 7.78037369e-01 -1.52617204e+00 1.88909918e-01 -4.75306690e-01 -6.73221111e-01 1.30882502e+00 9.11331534e-01 -4.54447120e-01 8.91975105e-01 2.32039064e-01 2.63398975e-01 -4.70153660e-01 -7.98569381e-01 -1.57386377e-01 5.65512955e-01 3.86807948e-01 6.97499037e-01 3.37502569e-01 -3.23674023e-01 4.55319703e-01 2.09429234e-01 2.15510011e-01 3.75888765e-01 1.12386584e+00 -9.85233128e-01 -4.13093448e-01 -6.26619816e-01 1.20370281e+00 -9.50775445e-01 1.20147280e-01 -3.50976408e-01 1.30532479e+00 2.63877004e-01 2.31134295e-01 4.80522811e-02 -9.81497243e-02 2.59702891e-01 6.84651956e-02 3.04578036e-01 -5.76123357e-01 -9.37188804e-01 4.79366809e-01 -2.08515540e-01 -6.50512129e-02 -4.28023815e-01 -3.10142130e-01 -1.36149192e+00 2.18336768e-02 -4.37361926e-01 8.38471428e-02 4.02012408e-01 4.86405700e-01 -2.01240685e-02 1.04458845e+00 3.78951818e-01 -2.79358834e-01 -9.27149951e-01 -1.15986705e+00 -5.92097640e-01 7.32547641e-02 2.19234586e-01 -3.79765928e-01 -3.30920041e-01 -1.02761216e-01]
[15.40157699584961, -2.142075538635254]
fc0a6efc-19df-45fc-bd68-e5bf950b37e9
nonlinear-equivariant-imaging-learning-multi
2211.12786
null
https://arxiv.org/abs/2211.12786v1
https://arxiv.org/pdf/2211.12786v1.pdf
Nonlinear Equivariant Imaging: Learning Multi-Parametric Tissue Mapping without Ground Truth for Compressive Quantitative MRI
Current state-of-the-art reconstruction for quantitative tissue maps from fast, compressive, Magnetic Resonance Fingerprinting (MRF), use supervised deep learning, with the drawback of requiring high-fidelity ground truth tissue map training data which is limited. This paper proposes NonLinear Equivariant Imaging (NLEI), a self-supervised learning approach to eliminate the need for ground truth for deep MRF image reconstruction. NLEI extends the recent Equivariant Imaging framework to nonlinear inverse problems such as MRF. Only fast, compressed-sampled MRF scans are used for training. NLEI learns tissue mapping using spatiotemporal priors: spatial priors are obtained from the invariance of MRF data to a group of geometric image transformations, while temporal priors are obtained from a nonlinear Bloch response model approximated by a pre-trained neural network. Tested retrospectively on two acquisition settings, we observe that NLEI (self-supervised learning) closely approaches the performance of supervised learning, despite not using ground truth during training.
['Mohammad Golbabaee', 'Peter Hall', 'Marion I. Menzel', 'Carolin M. Pirkl', 'Kwai Y. Chau', 'Ketan Fatania']
2022-11-23
null
null
null
null
['magnetic-resonance-fingerprinting']
['medical']
[ 7.93807268e-01 2.94631451e-01 -1.62071899e-01 -4.84006554e-01 -1.00834835e+00 -3.44645202e-01 6.29751623e-01 -2.81378955e-01 -5.67247570e-01 7.04238117e-01 4.64020282e-01 3.20217013e-02 -6.42600417e-01 -3.61200154e-01 -1.02167785e+00 -8.83504152e-01 -3.89449239e-01 6.22499526e-01 -2.25650333e-02 1.43209994e-01 -3.93333808e-02 4.47409093e-01 -8.29749942e-01 3.41403902e-01 6.47919953e-01 7.99739957e-01 5.00906527e-01 5.00260234e-01 4.02976274e-01 1.12298870e+00 2.44759500e-01 2.42985755e-01 1.76736712e-01 -4.10618067e-01 -1.00009680e+00 -1.00387193e-01 8.00513864e-01 -5.77942610e-01 -8.33290398e-01 9.41831291e-01 4.69601363e-01 1.88199040e-02 9.16108966e-01 -6.53649092e-01 -5.64268589e-01 5.68319917e-01 -3.01473230e-01 5.21286070e-01 -9.52906609e-02 -2.94962451e-02 3.46155673e-01 -9.57660079e-01 8.65614533e-01 4.93891269e-01 1.04583883e+00 5.31032920e-01 -1.66445673e+00 -3.49355876e-01 -5.78775942e-01 4.06853966e-02 -1.22240126e+00 -5.47036409e-01 7.20312476e-01 -7.54825890e-01 7.12976336e-01 1.86210677e-01 5.03990889e-01 9.30602491e-01 7.23578990e-01 3.98792475e-01 1.58215725e+00 -3.66655588e-01 1.74125552e-01 -2.77263552e-01 -8.70882440e-03 9.15658474e-01 5.48189841e-02 3.69763017e-01 -6.14583910e-01 -2.94318765e-01 1.27316642e+00 1.19549716e-02 -5.38482010e-01 -6.90295100e-01 -1.58422601e+00 5.33295751e-01 5.53163648e-01 5.81972897e-01 -7.54978180e-01 3.30494940e-01 6.48827553e-02 1.35383412e-01 4.02780116e-01 4.79080111e-01 3.81438695e-02 3.99307787e-01 -1.34213257e+00 -1.03387274e-01 3.67738515e-01 5.00628471e-01 6.59859359e-01 9.45228115e-02 -1.94244578e-01 6.24022543e-01 3.50715071e-01 6.58551097e-01 7.25575805e-01 -1.39615345e+00 -1.70312852e-01 -1.03683800e-01 -6.81230947e-02 -7.48841763e-01 -6.28088057e-01 -8.62927318e-01 -1.07588398e+00 2.39208955e-02 5.01982749e-01 1.63836777e-02 -9.32695985e-01 1.77203679e+00 1.03795156e-01 5.66424847e-01 -1.70640498e-01 1.11181843e+00 4.18451935e-01 3.29285651e-01 -2.05729008e-01 -4.19844925e-01 9.14474487e-01 -6.94502771e-01 -6.92905605e-01 6.15135254e-03 5.73773801e-01 -3.95445704e-01 8.80755782e-01 4.46335763e-01 -1.17589796e+00 -9.25481990e-02 -9.11896229e-01 1.19734056e-01 2.35581100e-01 -7.40233287e-02 6.82731748e-01 3.53836715e-01 -1.22228289e+00 8.36589336e-01 -1.24227285e+00 -3.92646082e-02 4.17838216e-01 5.51418543e-01 -7.49473929e-01 -4.99819726e-01 -1.05628586e+00 9.71672356e-01 1.39557971e-02 1.60091385e-01 -1.26179111e+00 -1.34039164e+00 -7.06912398e-01 -5.10324001e-01 7.24363253e-02 -7.81643689e-01 9.72938597e-01 -7.68296123e-01 -1.39419568e+00 9.93614316e-01 2.06964836e-02 -5.83922744e-01 7.00860322e-01 -1.05093338e-01 -3.02536428e-01 7.86038816e-01 3.32291037e-01 6.32463813e-01 1.00609457e+00 -1.22686863e+00 3.72464508e-01 -4.89649624e-01 -4.30801660e-01 -1.36931939e-03 7.53690377e-02 -4.79282975e-01 5.85287549e-02 -5.74625671e-01 4.79371756e-01 -1.06844771e+00 -3.77364933e-01 1.59263790e-01 -2.18847498e-01 3.92220497e-01 5.41112423e-01 -1.08908975e+00 5.89121222e-01 -1.91017890e+00 1.60803780e-01 4.55542326e-01 4.62849230e-01 -3.38096112e-01 -1.90116495e-01 -1.70979097e-01 -4.51831400e-01 -3.55969846e-01 -6.44197702e-01 -9.00978744e-02 -3.50737065e-01 3.33251119e-01 4.44162637e-02 1.06042743e+00 -2.31560215e-01 9.90419805e-01 -1.16194463e+00 -5.50907016e-01 3.17584097e-01 4.79330927e-01 -5.64594984e-01 2.12686390e-01 1.90615416e-01 1.13157701e+00 -2.24972397e-01 2.66781837e-01 6.12990499e-01 -5.18636525e-01 2.90156871e-01 -7.19262838e-01 -1.86627120e-01 -8.33170339e-02 -5.07058740e-01 2.39256763e+00 -4.97847050e-01 3.90148729e-01 3.98236692e-01 -1.49479389e+00 4.62159425e-01 6.27497673e-01 1.39093387e+00 -7.87919939e-01 -4.70303968e-02 4.15801883e-01 -2.71485578e-02 -6.78986907e-01 -1.69546038e-01 -6.83703542e-01 2.67634064e-01 5.97110093e-01 5.30251026e-01 -2.06090689e-01 -3.17511946e-01 1.83434382e-01 1.26335597e+00 3.12875658e-01 -2.36160502e-01 -8.40490460e-01 3.03097099e-01 7.78690279e-02 2.71279454e-01 9.92012322e-01 -1.36682138e-01 8.54840934e-01 1.84199736e-02 -5.50566733e-01 -1.15427852e+00 -1.42553592e+00 -8.41076255e-01 4.93149757e-01 -6.06105551e-02 2.23869443e-01 -9.25256014e-01 -4.33969855e-01 -2.50163734e-01 4.07487422e-01 -8.19803536e-01 -1.90751076e-01 -8.61629009e-01 -9.44467545e-01 4.64683712e-01 1.60703823e-01 3.89902472e-01 -7.06588507e-01 -4.98386443e-01 4.00251687e-01 -6.45939291e-01 -1.24267876e+00 -5.23889124e-01 3.41586709e-01 -1.23911262e+00 -7.78315723e-01 -1.08387005e+00 -5.56143105e-01 8.25805366e-01 1.21391796e-01 9.46743608e-01 -2.64593422e-01 -4.65483636e-01 7.20645905e-01 4.20989878e-02 2.05987588e-01 -3.24180037e-01 -4.89817671e-02 2.14897066e-01 1.09237313e-01 -2.42944539e-01 -9.25289154e-01 -7.91130006e-01 2.04423785e-01 -9.90825415e-01 2.21637577e-01 9.07899499e-01 1.18019283e+00 1.15921032e+00 -2.83044368e-01 4.18024987e-01 -9.19812083e-01 3.91970668e-03 -6.14628136e-01 -1.71179235e-01 2.92417824e-01 -6.49635971e-01 4.43621755e-01 2.65172035e-01 -4.74042237e-01 -8.86110544e-01 1.49756297e-01 -4.85549532e-02 -5.80442905e-01 -1.90373976e-02 6.63127184e-01 4.19364572e-01 -6.57216430e-01 9.60421920e-01 5.24308503e-01 5.08922637e-01 -2.87497222e-01 1.75145596e-01 -2.05638725e-02 1.15554893e+00 -6.76423430e-01 5.62008202e-01 8.99795473e-01 4.23682392e-01 -8.08558166e-01 -8.58372211e-01 -3.78299743e-01 -9.95272696e-01 -4.89002079e-01 1.04169631e+00 -6.69341326e-01 -2.62114704e-01 2.36272767e-01 -7.73041844e-01 -6.32432282e-01 -5.15897870e-01 1.15049946e+00 -1.07957029e+00 3.84406924e-01 -7.98001111e-01 -2.21378148e-01 -3.62593114e-01 -1.06170321e+00 9.66095865e-01 -4.47660059e-01 6.87685446e-04 -1.24561846e+00 2.37208307e-01 4.80993807e-01 7.91209459e-01 6.01982832e-01 9.26726401e-01 -1.44957565e-02 -6.84055209e-01 -7.47139081e-02 5.96013963e-02 3.01762789e-01 1.01631358e-01 -9.23073113e-01 -8.92893553e-01 -4.27811593e-01 3.50230455e-01 -4.47597265e-01 7.84224629e-01 9.50760424e-01 1.31147599e+00 -3.00406843e-01 -1.50524482e-01 1.11056745e+00 1.43713963e+00 -3.03935647e-01 6.91317737e-01 1.16334930e-02 7.25779951e-01 5.26946187e-01 1.06953448e-02 1.66800097e-01 2.35548630e-01 5.24793088e-01 3.50939870e-01 -4.36572045e-01 -3.75242561e-01 -3.76710236e-01 1.28723130e-01 8.23583245e-01 -3.79131615e-01 4.16697413e-01 -9.93788064e-01 4.06772822e-01 -1.55956984e+00 -8.43931258e-01 -6.26302809e-02 2.19841743e+00 7.67489552e-01 -3.37056696e-01 -2.11997405e-01 -6.52082413e-02 3.62456441e-01 -6.44441321e-02 -7.99592853e-01 5.89704931e-01 -2.71641146e-02 3.94021720e-01 8.13875258e-01 7.54671812e-01 -9.53049242e-01 4.62334484e-01 6.54901695e+00 5.72261274e-01 -1.60477507e+00 8.02565634e-01 6.13504589e-01 5.42080700e-02 -5.62272429e-01 -5.73333576e-02 2.28920102e-01 5.07835224e-02 1.04241550e+00 2.25141764e-01 7.91402519e-01 1.90415308e-01 3.30315292e-01 2.98266318e-02 -1.02166951e+00 1.03920639e+00 1.08939797e-01 -1.66615641e+00 -2.71023840e-01 1.41267017e-01 7.54551411e-01 5.45411587e-01 2.29407266e-01 -1.16563752e-01 1.96879208e-02 -1.17565858e+00 6.62036240e-01 1.08145475e+00 1.23467386e+00 -1.94650471e-01 5.00356078e-01 3.62379193e-01 -5.69067359e-01 4.05825138e-01 -1.03562862e-01 5.34571946e-01 2.41134197e-01 7.49581218e-01 -7.24846184e-01 5.02975404e-01 6.46562457e-01 7.95897543e-01 -1.84937403e-01 6.20045006e-01 2.25588903e-01 7.89677858e-01 -3.29344392e-01 8.07671547e-01 1.84512466e-01 -5.87486960e-02 5.67984283e-01 1.05549335e+00 2.60381430e-01 3.40155452e-01 1.45478114e-01 1.04463863e+00 1.55282319e-01 -7.40377679e-02 -6.04238689e-01 1.17387280e-01 -2.31263727e-01 1.25989270e+00 -6.36763453e-01 -4.45260815e-02 -1.82042301e-01 1.01114857e+00 7.22603649e-02 5.40543497e-01 -5.05260348e-01 2.70667642e-01 -2.05137759e-01 9.49896336e-01 -2.26422176e-01 -4.91287768e-01 -1.36056527e-01 -1.30394554e+00 -1.91503182e-01 -4.27422464e-01 1.81505978e-01 -1.01443434e+00 -1.30204117e+00 4.74292397e-01 2.53015667e-01 -1.00879347e+00 -4.58900809e-01 -6.33491397e-01 -2.78536230e-01 8.42289627e-01 -1.55013287e+00 -1.12897682e+00 -3.70563447e-01 9.00579393e-01 -2.36782417e-01 9.12062451e-02 8.53530288e-01 4.45596814e-01 1.35176465e-01 3.49853665e-01 1.43187195e-01 7.84750506e-02 7.47563124e-01 -1.27283907e+00 -3.47872049e-01 5.53960681e-01 -1.31295428e-01 7.07651138e-01 4.33533102e-01 -5.40569305e-01 -1.73274040e+00 -1.06229079e+00 4.11930382e-01 -1.68415353e-01 6.17494583e-01 -5.96073940e-02 -9.09770548e-01 7.74803638e-01 1.55693516e-02 1.02090728e+00 7.45737910e-01 -2.59290189e-01 -1.18883364e-01 -1.51501194e-01 -1.52501082e+00 5.53041473e-02 9.69929934e-01 -8.42159629e-01 -2.58642316e-01 5.98530769e-01 2.81321347e-01 -4.92539078e-01 -1.45347714e+00 4.75203842e-01 7.94769764e-01 -7.55606353e-01 9.84412730e-01 -2.60190219e-01 3.68755400e-01 -1.71932757e-01 -3.12606663e-01 -1.22066915e+00 -5.81715584e-01 -4.98746395e-01 -3.86324897e-02 2.49528110e-01 1.31543159e-01 -6.30305946e-01 9.09969330e-01 3.41381818e-01 -5.50976932e-01 -5.05136371e-01 -1.15423870e+00 -6.65683329e-01 3.07836711e-01 -2.97026306e-01 3.90181243e-02 1.36395907e+00 -1.34690449e-01 6.38528988e-02 -4.66023207e-01 2.24892050e-01 1.36256683e+00 -1.49142027e-01 3.17742750e-02 -9.91062641e-01 -6.30892754e-01 7.49863163e-02 -2.33460903e-01 -1.01069725e+00 2.39104837e-01 -1.49785423e+00 1.94941327e-01 -1.30722630e+00 3.61109674e-01 -6.61982834e-01 -4.63063389e-01 4.21453953e-01 4.52539384e-01 5.82322061e-01 -1.62164524e-01 6.11322701e-01 -3.26307684e-01 3.03328097e-01 1.71359730e+00 -2.56741732e-01 2.65815914e-01 -4.02806729e-01 -2.43074059e-01 5.47609806e-01 4.74680901e-01 -6.29838467e-01 -4.65380102e-01 -5.61906278e-01 -1.17418148e-01 8.49642336e-01 6.84465289e-01 -1.02409708e+00 3.80697817e-01 5.04740141e-02 5.92758119e-01 -2.12278396e-01 1.42747268e-01 -7.64602423e-01 4.43775147e-01 6.61683381e-01 -4.52503920e-01 -1.69217542e-01 -5.96609451e-02 3.25586289e-01 -2.00284213e-01 -1.01334788e-01 1.00507843e+00 -4.83173996e-01 -2.84637958e-01 7.20867932e-01 -3.95267963e-01 3.37971747e-02 3.36776495e-01 -5.65194041e-02 1.09364204e-01 -3.35495681e-01 -1.31821465e+00 -3.43874216e-01 1.64974868e-01 -2.25389376e-01 6.98642850e-01 -1.31019783e+00 -8.92671525e-01 3.30040157e-01 -1.62638143e-01 -1.43064663e-01 6.56765044e-01 1.55643404e+00 -5.93338072e-01 4.92746979e-01 -3.98100883e-01 -1.07773578e+00 -3.64083111e-01 3.23525578e-01 1.06210232e+00 -4.57659572e-01 -9.28555965e-01 4.96179193e-01 2.57450372e-01 -8.63662660e-01 -2.46526510e-01 -2.84658879e-01 2.32992619e-01 -6.46447420e-01 3.03500295e-01 6.40566722e-02 3.60935688e-01 -7.89478958e-01 -3.06751817e-01 6.93252921e-01 8.22497979e-02 -4.94003415e-01 1.44215488e+00 7.27412626e-02 -1.52352676e-01 2.75923550e-01 1.38454199e+00 -2.51015067e-01 -1.40215218e+00 -6.03143871e-01 -2.06859604e-01 -1.31350502e-01 8.29480827e-01 -9.12690878e-01 -1.06748259e+00 8.18122149e-01 1.20681834e+00 -4.58480984e-01 9.07318592e-01 1.90235134e-02 6.28767610e-01 2.24353805e-01 6.52567208e-01 -5.45539200e-01 2.12554395e-01 3.06066573e-01 1.09339333e+00 -1.16726851e+00 1.83994155e-02 -2.75481820e-01 -1.03809528e-01 1.03729343e+00 9.25779492e-02 -3.89544934e-01 1.01104486e+00 5.68090379e-01 -8.00462812e-02 -3.19216907e-01 -2.25967124e-01 4.18544352e-01 4.11689311e-01 6.91256344e-01 5.49256921e-01 1.71873450e-01 -8.27293023e-02 2.52332598e-01 -8.59290659e-02 3.18048269e-01 2.26689875e-01 7.20492482e-01 -9.29256156e-02 -7.63418317e-01 -2.37742171e-01 5.40028870e-01 -5.79985142e-01 -1.00956470e-01 2.38618061e-01 4.75903839e-01 2.23344103e-01 2.11091250e-01 1.16526365e-01 3.06991264e-02 2.44474597e-02 -2.95989424e-01 1.28612542e+00 -3.67691964e-01 -1.85579211e-01 1.44104272e-01 -4.05244917e-01 -6.84093356e-01 -6.12882435e-01 -9.31355596e-01 -1.43021548e+00 1.19651454e-02 6.77505583e-02 8.35411400e-02 7.23462880e-01 1.10912573e+00 3.10448445e-02 3.97524357e-01 5.42917848e-01 -1.04866409e+00 -4.64003533e-01 -8.87116253e-01 -7.69120395e-01 4.66588348e-01 6.24849498e-01 -6.98244631e-01 -1.87863588e-01 2.57235229e-01]
[13.508493423461914, -2.410276412963867]